Permalink
Cannot retrieve contributors at this time
Fetching contributors…
Cannot retrieve contributors at this time
| <?xml version="1.0" encoding="UTF-8"?><!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.1d1 20130915//EN" "JATS-archivearticle1.dtd"><article article-type="research-article" dtd-version="1.1d1" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><front><journal-meta><journal-id journal-id-type="nlm-ta">elife</journal-id><journal-id journal-id-type="hwp">eLife</journal-id><journal-id journal-id-type="publisher-id">eLife</journal-id><journal-title-group><journal-title>eLife</journal-title></journal-title-group><issn publication-format="electronic">2050-084X</issn><publisher><publisher-name>eLife Sciences Publications, Ltd</publisher-name></publisher></journal-meta><article-meta><article-id pub-id-type="publisher-id">01236</article-id><article-id pub-id-type="doi">10.7554/eLife.01236</article-id><article-categories><subj-group subj-group-type="display-channel"><subject>Research article</subject></subj-group><subj-group subj-group-type="heading"><subject>Biochemistry</subject></subj-group></article-categories><title-group><article-title>Global cellular response to chemotherapy-induced apoptosis</article-title></title-group><contrib-group><contrib contrib-type="author" id="author-6509"><name><surname>Wiita</surname><given-names>Arun P</given-names></name><xref ref-type="aff" rid="aff1"/><xref ref-type="aff" rid="aff2"/><xref ref-type="other" rid="par-3"/><xref ref-type="fn" rid="con1"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><contrib contrib-type="author" id="author-6528"><name><surname>Ziv</surname><given-names>Etay</given-names></name><xref ref-type="aff" rid="aff3"/><xref ref-type="aff" rid="aff4"/><xref ref-type="fn" rid="con2"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><contrib contrib-type="author" id="author-6529"><name><surname>Wiita</surname><given-names>Paul J</given-names></name><xref ref-type="aff" rid="aff5"/><xref ref-type="fn" rid="con3"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><contrib contrib-type="author" id="author-6530"><name><surname>Urisman</surname><given-names>Anatoly</given-names></name><xref ref-type="aff" rid="aff1"/><xref ref-type="aff" rid="aff6"/><xref ref-type="fn" rid="con5"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><contrib contrib-type="author" id="author-6531"><name><surname>Julien</surname><given-names>Olivier</given-names></name><xref ref-type="aff" rid="aff1"/><xref ref-type="fn" rid="con4"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><contrib contrib-type="author" id="author-4366"><name><surname>Burlingame</surname><given-names>Alma L</given-names></name><xref ref-type="aff" rid="aff1"/><xref ref-type="fn" rid="con6"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><contrib contrib-type="author" id="author-6320"><name><surname>Weissman</surname><given-names>Jonathan S</given-names></name><xref ref-type="aff" rid="aff3"/><xref ref-type="aff" rid="aff7"/><xref ref-type="other" rid="par-2"/><xref ref-type="fn" rid="con7"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><contrib contrib-type="author" corresp="yes" id="author-6510"><name><surname>Wells</surname><given-names>James A</given-names></name><xref ref-type="aff" rid="aff1"/><xref ref-type="aff" rid="aff3"/><xref ref-type="corresp" rid="cor1">*</xref><xref ref-type="other" rid="par-1"/><xref ref-type="fn" rid="con8"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><aff id="aff1"><institution content-type="dept">Department of Pharmaceutical Chemistry</institution>, <institution>University of California, San Francisco</institution>, <addr-line><named-content content-type="city">San Francisco</named-content></addr-line>, <country>United States</country></aff><aff id="aff2"><institution content-type="dept">Department of Laboratory Medicine</institution>, <institution>University of California, San Francisco</institution>, <addr-line><named-content content-type="city">San Francisco</named-content></addr-line>, <country>United States</country></aff><aff id="aff3"><institution content-type="dept">Department of Cellular and Molecular Pharmacology</institution>, <institution>University of California, San Francisco</institution>, <addr-line><named-content content-type="city">San Francisco</named-content></addr-line>, <country>United States</country></aff><aff id="aff4"><institution content-type="dept">Department of Radiology</institution>, <institution>University of California, San Francisco</institution>, <addr-line><named-content content-type="city">San Francisco</named-content></addr-line>, <country>United States</country></aff><aff id="aff5"><institution content-type="dept">Department of Physics</institution>, <institution>The College of New Jersey</institution>, <addr-line><named-content content-type="city">Ewing</named-content></addr-line>, <country>United States</country></aff><aff id="aff6"><institution>Department of Pathology, University of California, San Francisco</institution>, <addr-line><named-content content-type="city">San Francisco</named-content></addr-line>, <country>United States</country></aff><aff id="aff7"><institution>Howard Hughes Medical Institute, University of California, San Francisco</institution>, <addr-line><named-content content-type="city">San Francisco</named-content></addr-line>, <country>United States</country></aff></contrib-group><contrib-group content-type="section"><contrib contrib-type="editor"><name><surname>Cravatt</surname><given-names>Ben</given-names></name><role>Reviewing editor</role><aff><institution>The Scripps Research Institute</institution>, <country>United States</country></aff></contrib></contrib-group><author-notes><corresp id="cor1"><label>*</label>For correspondence: <email>jim.wells@ucsf.edu</email></corresp></author-notes><pub-date date-type="pub" publication-format="electronic"><day>29</day><month>10</month><year>2013</year></pub-date><pub-date pub-type="collection"><year>2013</year></pub-date><volume>2</volume><elocation-id>e01236</elocation-id><history><date date-type="received"><day>16</day><month>07</month><year>2013</year></date><date date-type="accepted"><day>23</day><month>09</month><year>2013</year></date></history><permissions><copyright-statement>© 2013, Wiita et al</copyright-statement><copyright-year>2013</copyright-year><copyright-holder>Wiita et al</copyright-holder><license xlink:href="http://creativecommons.org/licenses/by/3.0/"><license-p>This article is distributed under the terms of the <ext-link ext-link-type="uri" xlink:href="http://creativecommons.org/licenses/by/3.0/">Creative Commons Attribution License</ext-link>, which permits unrestricted use and redistribution provided that the original author and source are credited.</license-p></license></permissions><self-uri content-type="pdf" xlink:href="elife01236.pdf"/><related-article ext-link-type="doi" id="ra1" related-article-type="commentary" xlink:href="10.7554/eLife.01587"/><abstract><object-id pub-id-type="doi">10.7554/eLife.01236.001</object-id><p>How cancer cells globally struggle with a chemotherapeutic insult before succumbing to apoptosis is largely unknown. Here we use an integrated systems-level examination of transcription, translation, and proteolysis to understand these events central to cancer treatment. As a model we study myeloma cells exposed to the proteasome inhibitor bortezomib, a first-line therapy. Despite robust transcriptional changes, unbiased quantitative proteomics detects production of only a few critical anti-apoptotic proteins against a background of general translation inhibition. Simultaneous ribosome profiling further reveals potential translational regulation of stress response genes. Once the apoptotic machinery is engaged, degradation by caspases is largely independent of upstream bortezomib effects. Moreover, previously uncharacterized non-caspase proteolytic events also participate in cellular deconstruction. Our systems-level data also support co-targeting the anti-apoptotic regulator HSF1 to promote cell death by bortezomib. This integrated approach offers unique, in-depth insight into apoptotic dynamics that may prove important to preclinical evaluation of any anti-cancer compound.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.001">http://dx.doi.org/10.7554/eLife.01236.001</ext-link></p></abstract><abstract abstract-type="executive-summary"><object-id pub-id-type="doi">10.7554/eLife.01236.002</object-id><title>eLife digest</title><p>Many cancer treatments work by causing cancer cells to enter an advanced stage of a process known as programmed cell death or apoptosis. When a cell begins apoptosis, it takes a series of metabolic steps–such as fragmenting its DNA or reducing its volume–that eventually kills it. The cancer cells in tumours are able to grow because they are able to avoid apoptosis.</p><p>When cancer cells are treated with cytotoxic drugs they do not die immediately but try to stave off the effect of the drug. However, we still know relatively little about what happens at the molecular levels as cancer cells struggle to avoid apoptosis.</p><p>Now Wiita et al. have combined two methods for studying cancer cells–deep sequencing of RNA and quantitative proteomics–to simultaneously observe a variety of processes, including the transcription of genes to produce messenger RNA (mRNA) molecules, the translation of these mRNA molecules to produce proteins, and the proteolysis (or breakdown) of these proteins when the cells were subjected to chemotherapy.</p><p>Wiita et al. studied how human myeloma cells responded to bortezomib, a drug that is used to treat various blood cancers, and found that ribosomes–the complex molecular machines that perform the translation step– reacted to the chemotherapy by preferentially translating certain mRNA molecules in order to produce a set of proteins that protect the cell. Developing drugs to inhibit the effects of these stress-response proteins could make the cancer cells more responsive to existing anticancer drugs. When this effort to stay alive is ultimately unsuccessful, the destruction of proteins appears surprisingly unrelated to the previous attempts that were made to protect the cell.</p><p>With further work the “global cellular response” approach developed by Wiita et al. could lead to the discovery of new drug targets, improve our understanding of drug resistance in chemotherapy, and provide new ways to monitor how patients respond to treatment.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.002">http://dx.doi.org/10.7554/eLife.01236.002</ext-link></p></abstract><kwd-group kwd-group-type="author-keywords"><title>Author keywords</title><kwd>apoptosis</kwd><kwd>proteomics</kwd><kwd>ribosome profiling</kwd><kwd>caspase</kwd><kwd>myeloma</kwd></kwd-group><kwd-group kwd-group-type="research-organism"><title>Research organism</title><kwd>Human</kwd></kwd-group><funding-group><award-group id="par-1"><funding-source><institution-wrap><institution>National Institutes of Health</institution></institution-wrap></funding-source><award-id>R01GM081051</award-id><principal-award-recipient><name><surname>Wells</surname><given-names>James A</given-names></name></principal-award-recipient></award-group><award-group id="par-2"><funding-source><institution-wrap><institution>Howard Hughes Medical Institute</institution></institution-wrap></funding-source><principal-award-recipient><name><surname>Weissman</surname><given-names>Jonathan S</given-names></name></principal-award-recipient></award-group><award-group id="par-3"><funding-source><institution-wrap><institution>Damon Runyon Cancer Research Foundation Post-Doctoral Fellowship (DRG 111-12)</institution></institution-wrap></funding-source><principal-award-recipient><name><surname>Wiita</surname><given-names>Arun P</given-names></name></principal-award-recipient></award-group><funding-statement>The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.</funding-statement></funding-group><custom-meta-group><custom-meta><meta-name>elife-xml-version</meta-name><meta-value>2</meta-value></custom-meta><custom-meta specific-use="meta-only"><meta-name>Author impact statement</meta-name><meta-value>The simultaneous monitoring of transcription, translational control, protein production and proteolysis in the same system can improve our understanding of how cancer cells respond to chemotherapy.</meta-value></custom-meta></custom-meta-group></article-meta></front><body><sec id="s1" sec-type="intro"><title>Introduction</title><p>Most cytotoxic chemotherapeutics eliminate tumor cells by activating the intrinsic apoptotic pathway (<xref ref-type="bibr" rid="bib27">Kaufmann and Earnshaw, 2000</xref>). The final stages leading to mitochondrial pore formation and caspase activation have been well-described at the molecular level (<xref ref-type="bibr" rid="bib18">Gonzalvez and Ashkenazi, 2010</xref>; <xref ref-type="bibr" rid="bib60">Spencer and Sorger, 2011</xref>; <xref ref-type="bibr" rid="bib51">Parrish et al., 2013</xref>). However, prior to these terminal stages of apoptosis, it is becoming clear that cells fight to resist the pharmacological insult in ways that appear unique to the treatment applied (<xref ref-type="bibr" rid="bib17">Geva-Zatorsky et al., 2010</xref>). Existing genome-wide studies of cellular response to chemotherapeutic treatment have primarily relied on changes at the transcriptional level (<xref ref-type="bibr" rid="bib30">Lamb et al., 2006</xref>). Elegant single cell studies have tracked a subset of proteins in response to various chemotherapeutics (<xref ref-type="bibr" rid="bib10">Cohen et al., 2008</xref>; <xref ref-type="bibr" rid="bib17">Geva-Zatorsky et al., 2010</xref>). Recent work has identified hundreds of caspase cleavage substrates in apoptotic cells treated with chemotherapeutics (<xref ref-type="bibr" rid="bib59">Shimbo et al., 2012</xref>).</p><p>However, each of these approaches only captures a segment of the functional reaction to a chemotherapeutic insult and does not tell the full story of how cancer responds to apoptosis-inducing drug treatment. It is known that cells undergoing apoptosis show strong suppression of protein translation (<xref ref-type="bibr" rid="bib8">Bushell et al., 2004</xref>). While a few specific transcripts are known to escape this translational suppression (<xref ref-type="bibr" rid="bib61">Spriggs et al., 2010</xref>), the general link between transcriptional and translational changes during apoptosis is not well understood. Furthermore, different chemotherapeutics produce distinct quantitative signatures of caspase cleavage substrates following apoptosis (<xref ref-type="bibr" rid="bib59">Shimbo et al., 2012</xref>), and it is unclear how the cellular response to chemotherapy prior to apoptotic induction may influence the later deconstruction of cellular protein contents.</p><p>Here we examine in depth the response of a multiple myeloma cell culture line (MM1.S) exposed to the proteasome inhibitor bortezomib, a clinically relevant model system. Multiple myeloma, one of the most common blood cancers, is an aggressive malignancy of clonal plasma cells. Proteasome inhibition has become an effective first-line therapy for this disease, though myeloma currently has no known cure (<xref ref-type="bibr" rid="bib36">Lonial et al., 2011</xref>). We use a suite of emerging systems-level approaches to globally examine the dynamic interplay between transcription, translation, and proteolytic degradation during chemotherapy-induced apoptosis in this system (<xref ref-type="fig" rid="fig1">Figure 1A</xref>). Via ribosome profiling, we identify preferential translation and translational regulation of genes expected to reduce unfolded protein stress after bortezomib treatment. Surprisingly, despite these changes in translational control, during rapid apoptosis only a few critical pro-survival proteins are detectably increased in proteomic studies. Furthermore, we find that cleavage patterns by both caspase and non-caspase proteases during apoptosis are largely independent of drug response at the transcriptional and translational level prior to cell death. We also use this integrated data to examine the potential of small molecule therapeutics to work in concert with bortezomib in myeloma. Such a global examination of how cells struggle at all levels with a chemotherapeutic insult provides important insights into mechanisms of therapeutic resistance and novel methods to assess tumor response to chemotherapy.<fig-group><fig id="fig1" position="float"><object-id pub-id-type="doi">10.7554/eLife.01236.003</object-id><label>Figure 1.</label><caption><title>Experimental design and paired mRNA-seq/ribosome profiling data.</title><p>(<bold>A</bold>) MM1.S cells were prepared in six separate flasks each containing 3 × 10<sup>8</sup> cells. All time points past 0 hr were exposed to 20 nM bortezomib. Cells were harvested and analyzed using four systems-level technologies shown. (<bold>B</bold>) Cells undergo rapid apoptosis with initial increase in caspase activation seen at 6 hr and <10% cell viability by 12 hr (duplicate, mean ± SD). Total RNA and mRNA were isolated and measured by spectrophotometry; total protein was measured by BCA assay (measured in duplicate, isolated from 10<sup>6</sup> cells; mean ± SD). G(<italic>t</italic>), describing global mRNA degradation, is derived from a sigmoid fit to the total mRNA data. (<bold>C</bold>) Heat map showing log<sub>2</sub> expression compared to 0 hr for normalized mRNA and ribosome footprint read density (reads per kilobase million, RPKM) for 5680 well-expressed transcripts (<xref ref-type="supplementary-material" rid="SD1-data">Figure 1—source data 1</xref>). Also shown is calculated translational efficiency (TE) relative to 0 hr, calculated as the ratio of (footprint RPKM)/(mRNA RPKM) per transcript at each time point. We used unsupervised hierarchical clustering to define five broad groups of transcripts with relation to mRNA, footprint, and translational efficiency changes.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.003">http://dx.doi.org/10.7554/eLife.01236.003</ext-link></p><p><supplementary-material id="SD1-data"><object-id pub-id-type="doi">10.7554/eLife.01236.004</object-id><label>Figure 1—source data 1.</label><caption><title>Deep sequencing data, translational efficiency, and gene clusters.</title><p>(<bold>A</bold>) mRNA-seq and (<bold>B</bold>) ribosome profiling deep sequencing data with assigned UCSC hg19 identifier, associated UniProt accession number, and gene identifier as defined by kgXref feature in UCSC genome browser. For samples at each time point, we display raw reads that uniquely mapped to canonical transcript as well as mapped reads normalized to total read number and transcript length (reads per kilobase million, RPKM) for comparison across samples. These 5680 tracked transcripts had RPKM ≥1 across all 12 samples (six mRNA-seq, six ribosome footprint). (<bold>C</bold>) Translational efficiency (TE) for all tracked transcripts as measured by (RPKM of footprint sample/RPKM of mRNA-seq sample) at each time point. Table is ordered by descending TE at 0 hr. The histone genes are artifactually elevated as these mRNA's do not contain poly (<bold>A</bold>) tails and are therefore under-represented in the mRNA-seq data, as previously noted in <xref ref-type="bibr" rid="bib25">Ingolia et al. (2011)</xref>. (<bold>D</bold>) List of transcripts included in each Cluster (as in <xref ref-type="fig" rid="fig1">Figure 1C</xref>) as uploaded to Ingenuity Pathway Analysis server for analysis (results in <xref ref-type="table" rid="tbl1">Table 1</xref>).</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.004">http://dx.doi.org/10.7554/eLife.01236.004</ext-link></p></caption><media mime-subtype="xlsx" mimetype="application" xlink:href="elife01236s001.xlsx"/></supplementary-material></p></caption><graphic xlink:href="elife01236f001"/></fig><fig id="fig1s1" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.005</object-id><label>Figure 1—figure supplement 1.</label><caption><title>Sample comparison and biological subgroups.</title><p>(<bold>A</bold>) mRNA and (<bold>B</bold>) footprint samples prepared and sequenced independently at early time points show strong correlation in read density, demonstrating consistency in experimental methods. In footprint samples the mean difference between read density at 1.5 hr and 0 hr across transcripts = 28%, reflecting both experimental error and biological variation. (<bold>C</bold>) Translational efficiency spans >10-fold across all tracked transcripts without any large shifts across time points.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.005">http://dx.doi.org/10.7554/eLife.01236.005</ext-link></p></caption><graphic xlink:href="elife01236fs001"/></fig><fig id="fig1s2" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.006</object-id><label>Figure 1—figure supplement 2.</label><caption><title>Immunoblots.</title><p>Western blotting for proteins related to ER stress/UPR (ATF4, CHOP, XBP1s [‘s’ for spliced, active form]) and apoptosis (Bid, XIAP).</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.006">http://dx.doi.org/10.7554/eLife.01236.006</ext-link></p></caption><graphic xlink:href="elife01236fs002"/></fig><fig id="fig1s3" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.007</object-id><label>Figure 1—figure supplement 3.</label><caption><title>Biological subgroups.</title><p>Genes curated from the literature related to ER stress (left) and apoptosis (right) that were found in among the ∼5700 tracked transcripts. Genes highlighted with ribosome footprint analysis are highlighted with red arrows (<italic>DDIT3</italic> = CHOP; <italic>DDIT4</italic> = REDD1; <italic>PPP1R15A</italic> = GADD34). All data are shown as Log<sub>2</sub> expression ratio vs 0 hr.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.007">http://dx.doi.org/10.7554/eLife.01236.007</ext-link></p></caption><graphic xlink:href="elife01236fs003"/></fig></fig-group></p></sec><sec id="s2" sec-type="results"><title>Results</title><sec id="s2-1"><title>Simultaneous monitoring of multiple systems-level processes during apoptosis</title><p>We designed our experiment to monitor four systems-level processes simultaneously in the same cellular population, from transcription through protein production and proteolysis (<xref ref-type="fig" rid="fig1">Figure 1A</xref>). Deep sequencing of mRNA (mRNA-seq) examines the detailed transcriptional response of myeloma cells to bortezomib-induced stress. We pair this data with ribosome profiling, the deep sequencing of ribosome-protected mRNA fragments during active translation; this approach offers significant insight into proteome dynamics and translational regulation not available by monitoring transcript expression alone (<xref ref-type="bibr" rid="bib24">Ingolia et al., 2009</xref>, <xref ref-type="bibr" rid="bib25">2011</xref>; <xref ref-type="bibr" rid="bib62">Stern-Ginossar et al., 2012</xref>). To directly measure changes at the protein level after bortezomib perturbation, we use isobaric labeling by iTRAQ (<xref ref-type="bibr" rid="bib43">Mertins et al., 2012</xref>) and tandem mass spectrometry to measure in an unbiased fashion the relative abundance of ∼2600 proteins. Finally, using enzymatically-driven labeling of free protein α-amines (N-terminomics) (<xref ref-type="bibr" rid="bib39">Mahrus et al., 2008</xref>; <xref ref-type="bibr" rid="bib1">Agard et al., 2012</xref>), we develop a quantitative mass spectrometry assay to monitor to the relative kinetics of both caspase and non-caspase cleavage events during apoptotic cellular deconstruction. To begin, we prepared separate flasks of MM1.S cells and harvested cells at five time points after induction of apoptosis with 20 nM bortezomib. Similar to previous studies (<xref ref-type="bibr" rid="bib59">Shimbo et al., 2012</xref>), caspase activity was first detected at 6 hr and the majority of cells were non-viable by 12 hr (<xref ref-type="fig" rid="fig1">Figure 1B</xref>). mRNA levels were also markedly decreased prior to loss of cell viability, as previously seen across a range of apoptotic conditions (<xref ref-type="bibr" rid="bib13">Del Prete et al., 2002</xref>; <xref ref-type="bibr" rid="bib8">Bushell et al., 2004</xref>). Total RNA levels, primarily reflecting ribosomal RNA abundance, and total protein levels were relatively stable.</p></sec><sec id="s2-2"><title>Ribosome occupancy generally tracks with transcript production</title><p>To understand how myeloma cells were struggling after bortezomib exposure, we determined the relative changes in transcription and translation within the remaining mRNA pool at each time point. We normalized mRNA-seq and ribosome profiling sample using reads per kilobase million (RPKM) and focused our analysis on 5680 well-expressed transcripts (<xref ref-type="fig" rid="fig1">Figure 1C</xref>). We compared log<sub>2</sub> expression vs the untreated sample and organized the data using unsupervised hierarchical clustering. This approach allowed us to center on relative expression changes found in multiple independent samples. We generally found that changes in ribosome footprints tracked with changes in transcript abundance. The transcripts could be sub-divided into five clusters (<xref ref-type="fig" rid="fig1">Figure 1C</xref>). Clusters ‘Upreg’ and ‘Downreg’ encompass transcripts that are generally increased or decreased, respectively, at both the level of mRNA and footprint reads (<xref ref-type="fig" rid="fig2">Figure 2A,B</xref>). Cluster ‘Stable’ includes transcripts that demonstrate very mild changes at both the mRNA and footprint level. We calculated the translational efficiency (TE) from the ratio of ribosome footprint to mRNA-seq read density for each transcript (<xref ref-type="bibr" rid="bib25">Ingolia et al., 2011</xref>). This enabled us to characterize two other groups that are particularly interesting, Cluster ‘TE Up’ and ‘TE Down’. These groups showed little change in the level of mRNA, but large increases or decreases, respectively, in relative ribosome footprint density. This suggests that genes in these two clusters are regulated at the level of translation.<fig-group><fig id="fig2" position="float"><object-id pub-id-type="doi">10.7554/eLife.01236.008</object-id><label>Figure 2.</label><caption><title>Translational efficiency (TE) during bortezomib-induced apoptosis.</title><p>(<bold>A</bold> and <bold>B</bold>) Paired mRNA and ribosome footprint reads for <italic>HSPB1</italic> (Cluster Upreg) and <italic>HMGB1</italic> (Cluster Downreg) showing that protein translation generally tracks with transcript abundance. Displayed read counts are median-normalized to total aligned reads in 0 hr sample for comparison across the time course. Thick green bars = protein coding sequence (CDS); medium bars = untranslated regions (UTR); thin lines = intronic or intergenic regions. Read counts are inverted depending on coding direction. (<bold>C</bold>) <italic>ATF4</italic> shows increased footprint read density in the absence of mRNA increase at 6 hr. (<bold>D</bold>) <italic>SEC61B</italic>, a downstream target of XBP1, shows increased footprints and decreased mRNA reads, leading to increased TE. (<bold>E</bold>) Many ribosome structural proteins, including <italic>RPL7A</italic>, are found in Cluster TE down. (<bold>F</bold>) Log<sub>2</sub> changes in transcript TE from Cluster TE up and Cluster TE down at 12 hr vs 0 hr demonstrate statistically significant changes in the overall TE distribution (p<0.0001, Mann-Whitney test). (<bold>G</bold>) Mean (±SEM) changes in TE during the apoptotic time course generally do not exceed twofold even at later time points.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.008">http://dx.doi.org/10.7554/eLife.01236.008</ext-link></p></caption><graphic xlink:href="elife01236f002"/></fig><fig id="fig2s1" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.009</object-id><label>Figure 2—figure supplement 1.</label><caption><title>Subgroup analysis from biological pathways showing changes in translation efficiency from Ingenuity Pathway Analysis.</title><p>Genes included in IPA as XBP1 downstream targets (left) and Genes included as part of the eIF2 signaling pathway in IPA (right). Green boxed area indicates numerous ribosome structural proteins with decreased footprint reads relative to mRNA. <italic>SEC61B</italic> and <italic>RPL7A</italic> (shown in <xref ref-type="fig" rid="fig2">Figure 2</xref>) are highlighted with red arrows, as is PERK (<italic>EIF2AK3</italic>). All data are shown as Log<sub>2</sub> expression ratio vs 0 hr.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.009">http://dx.doi.org/10.7554/eLife.01236.009</ext-link></p></caption><graphic xlink:href="elife01236fs004"/></fig></fig-group></p></sec><sec id="s2-3"><title>A subset of stress response genes show changes in translational efficiency</title><p>Bortezomib is known to be an inducer of the unfolded protein response (UPR) and endoplasmic reticulum (ER) stress (<xref ref-type="bibr" rid="bib2">Aronson and Davies, 2012</xref>). Myeloma plasma cells are particularly sensitive to proteasome inhibition in vivo due to extremely high rates of immunoglobulin production, potentially leading to apoptosis via the UPR (<xref ref-type="bibr" rid="bib47">Obeng et al., 2006</xref>; <xref ref-type="bibr" rid="bib64">Walter and Ron, 2011</xref>). In <xref ref-type="fig" rid="fig2">Figure 2C</xref> we display reads for one of the critical factors in the UPR, <italic>ATF4,</italic> a transcription factor known to be under translational control during ER stress (<xref ref-type="bibr" rid="bib37">Lu et al., 2004</xref>). Consistent with this prior work, we find that <italic>ATF4</italic> does not increase in transcript abundance but shows a nearly threefold increase of ribosome occupancy (see heat map in <xref ref-type="fig" rid="fig1s3">Figure 1—figure supplement 3</xref>).</p><p>To better assess the biological implications of these results we turned to Ingenuity Pathway Analysis (IPA) (Ingenuity Systems, <ext-link ext-link-type="uri" xlink:href="http://www.ingenuity.com/">www.ingenuity.com</ext-link>) (<xref ref-type="table" rid="tbl1">Table 1</xref>). Cluster Upreg is enriched for genes related to protein ubiquitination (p=1.50 × 10<sup>−34</sup>), protein degradation (9.63 × 10<sup>−9</sup>), chaperones (4.15 × 10<sup>−11</sup>), and hypoxic response (2.90 × 10<sup>−10</sup>). Cluster Downreg includes genes important in cellular proliferation (p=3.61 × 10<sup>−11</sup>) and DNA repair (3.81 × 10<sup>−9</sup>). These findings are consistent with mRNA microarray studies of bortezomib response (<xref ref-type="bibr" rid="bib44">Mitsiades et al., 2002</xref>). We examined in more detail a subset of genes related to cellular apoptosis and both ER stress and hypoxic response (<xref ref-type="fig" rid="fig1s3">Figure 1—figure supplement 3</xref>). Notably, we found little change in expression or translation of canonical apoptosis players.<table-wrap id="tbl1" position="float"><object-id pub-id-type="doi">10.7554/eLife.01236.010</object-id><label>Table 1.</label><caption><title>Biological relevance of findings from Ingenuity Pathway Analysis (IPA)</title><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.010">http://dx.doi.org/10.7554/eLife.01236.010</ext-link></p></caption><table frame="hsides" rules="groups"><thead><tr><th>Cluster</th><th>Molecular and cellular function</th><th><italic>p</italic> value</th><th>Canonical pathway</th><th><italic>p</italic> value</th><th>Upstream regulator</th><th><italic>p</italic> value</th></tr></thead><tbody><tr><td rowspan="5">Cluster Upreg</td><td>Post-translational modification</td><td>4.15 e-11</td><td>Protein ubiquitination pathway</td><td>1.50e-34</td><td>NFE2L2 (NRF2)</td><td>7.77e-16</td></tr><tr><td>Protein Folding</td><td>4.15 e-11</td><td>NRF2-mediated oxidative stress response</td><td>2.90e-10</td><td>HSF1</td><td>1.70e-14</td></tr><tr><td>Cell death and survival</td><td>2.47 e-10</td><td/><td/><td>EIF2AK3 (PERK)</td><td>6.47e-10</td></tr><tr><td>Protein synthesis</td><td>8.82 e-09</td><td/><td/><td/><td/></tr><tr><td>Protein degradation</td><td>9.63 e-09</td><td/><td/><td/><td/></tr><tr><td rowspan="3">Cluster Stable</td><td>RNA post-transcriptional modification</td><td>1.44e-09</td><td>–</td><td/><td>–</td><td/></tr><tr><td>Molecular transport</td><td>8.87 e-09</td><td/><td/><td/><td/></tr><tr><td>Protein trafficking</td><td>8.87 e-09</td><td/><td/><td/><td/></tr><tr><td>Cluster TE Up</td><td>–</td><td/><td>Mitochondrial dysfunction</td><td>6.04e-10</td><td>XBP1</td><td>8.79e-10</td></tr><tr><td rowspan="5">Cluster Downreg</td><td>Cell death and survival</td><td>7.56e-12</td><td>–</td><td/><td>TP53</td><td>2.67e-15</td></tr><tr><td>Cellular growth and proliferation</td><td>3.61e-11</td><td/><td/><td>MYC</td><td>1.53e-10</td></tr><tr><td>DNA replication, recombination and repair</td><td>3.81e-09</td><td/><td/><td>XBP1</td><td>1.65e-10</td></tr><tr><td>RNA post-transcriptional modification</td><td>8.91e-08</td><td/><td/><td>INSR</td><td>5.94e-09</td></tr><tr><td/><td/><td/><td/><td>E2F4</td><td>6.20e-09</td></tr><tr><td rowspan="2">Cluster TE Down</td><td>RNA post-transcriptional modification</td><td>8.88e-11</td><td>EIF2 signaling</td><td>2.35e-17</td><td>–</td><td/></tr><tr><td>Gene expression</td><td>3.24e-09</td><td>Regulation of eIF4 and p70S6K signaling</td><td>1.40e-08</td><td/><td/></tr><tr><td>5% transcripts highest TE at 0 hr</td><td>–</td><td/><td>Mitchondrial dysfunction</td><td>1.52e-26</td><td>–</td><td/></tr><tr><td>Increased 5′ UTR translation</td><td>–</td><td/><td>EIF2 signaling</td><td>2.45 e-17</td><td>MYC</td><td>1.09e-08</td></tr><tr><td/><td/><td/><td/><td/><td>HSF1</td><td>8.11e-08</td></tr><tr><td>Decreased 5′ UTR translation</td><td>–</td><td/><td>–</td><td/><td>–</td><td/></tr></tbody></table><table-wrap-foot><fn><p>Clusters defined by hierarchical clustering of mRNA, footprint, and translational efficiency data as shown in <xref ref-type="fig" rid="fig1">Figure 1B</xref> (included genes listed in <xref ref-type="supplementary-material" rid="SD1-data">Figure 1—source data 1</xref>). We also analyzed the top 5% of transcripts by translational efficiency at 0 hr as calculated by (footprint RPKM)/(mRNA RPKM) per transcript. p values are as calculated using Fisher′s exact test by Ingenuity Pathway Analysis software. We report findings here with a p<10<sup>−7</sup>. Results from each category (Molecular and cellular function; Canonical pathway; Upstream regulator) are listed in order of decreasing p, independent of the other categories.</p></fn></table-wrap-foot></table-wrap></p><p>Surprisingly, IPA showed that Cluster TE Up included many genes regulated by XBP1 (p=8.79 × 10<sup>−10</sup>) (<xref ref-type="table" rid="tbl1">Table 1</xref>), an important component of the UPR (<xref ref-type="bibr" rid="bib64">Walter and Ron, 2011</xref>). The only known function of XBP1 is as a transcription factor (<xref ref-type="bibr" rid="bib1a">Acosta-Alvear et al., 2007</xref>) without known direct effects on translation. Although we do see a slight increase in the active XBP1 protein (<xref ref-type="fig" rid="fig1s2">Figure 1—figure supplement 2</xref>), most downstream XBP1 targets do not respond by increases in mRNA abundance. Instead we see increased translation efficiency (<xref ref-type="fig" rid="fig2s1">Figure 2—figure supplement 1</xref>). Notably, transcripts related to the ER transport machinery (including <italic>SEC61B</italic>, <xref ref-type="fig" rid="fig2">Figure 2D</xref>) demonstrate 1.5- to 2-fold increases in translation vs transcription. Therefore, changes in translation of XBP1downstream targets may reflect the activity of a parallel UPR mechanism to favor adaptation to cellular stress.</p><p>Cluster TE Down includes many transcripts involved in the translational machinery itself, particularly ribosomal structural proteins (<xref ref-type="fig" rid="fig2">Figure 2E</xref>, <xref ref-type="fig" rid="fig2s1">Figure 2—figure supplement 1</xref>) of which the most significant pathway affected involved eIF2 signaling (p=2.35 × 10<sup>−17</sup>) (<xref ref-type="table" rid="tbl1">Table 1</xref>). In parallel we examined the 5% of genes with the highest translational efficiency in the untreated sample. The pathway ‘mitochondrial dysfunction’ was highly enriched (p=1.52 × 10<sup>−26</sup>) (<xref ref-type="table" rid="tbl1">Table 1</xref>), including numerous cytochrome <italic>c</italic> oxidase and NADH dehydrogenase subunits (<xref ref-type="supplementary-material" rid="SD1-data">Figure 1—source data 1</xref>). This finding suggests a favoring of translation of aerobic metabolism components in myeloma cells at baseline. In <xref ref-type="fig" rid="fig2">Figure 2F</xref> we display the distribution of log<sub>2</sub>-fold changes in translational efficiency for Clusters TE Up and TE Down, which both show a significant difference between 12 hr vs 1.5 hr (p<0.0001, Mann-Whitney test). These changes contrast with the large majority of transcripts which show little change in TE (<xref ref-type="fig" rid="fig2">Figure 2G</xref>).</p></sec><sec id="s2-4"><title>Differential 5′ UTR translation</title><p>Taking advantage of the nucleotide resolution of ribosome profiling, we examined whether there were any large-scale changes in ribosome occupancy along mRNA during apoptosis. We performed a metagene analysis, where all footprint profiles are averaged and then aligned based on the midpoint of the protected reads (<xref ref-type="fig" rid="fig3">Figure 3A</xref>). We find a strong peak of ribosome occupancy at both the 5′ and 3′ ends of annotated coding sequence (CDS) and a 3-nucleotide offset resulting from position of the ribosome P site, as described previously (<xref ref-type="bibr" rid="bib24">Ingolia et al., 2009</xref>). We also note peaks appearing every three nucleotides across the averaged reads, consistent with the triplet nucleotide coding sequence. Averaged across all transcripts, we did not find any large changes in footprint read distribution across mRNAs at different time points.<fig-group><fig id="fig3" position="float"><object-id pub-id-type="doi">10.7554/eLife.01236.011</object-id><label>Figure 3.</label><caption><title>Nucleotide resolution of ribosome profiling reveals changes in 5′ UTR translation.</title><p>(<bold>A</bold>) Metagene analysis, the alignment of footprint reads across all transcripts, to both the 5′ CDS start site and 3′ termination site (<xref ref-type="bibr" rid="bib24">Ingolia et al., 2009</xref>). Reads counts are median-normalized to total aligned reads per sample. No significant changes are noted in overall read alignments during apoptosis. (<bold>B</bold>) Relative translation of 5′ UTR measured by the ratio of 5′ UTR reads to CDS reads per transcript. The log<sub>2</sub> change in this ratio is compared to the 5′ UTR/CDS ratio at 0 hr. (<bold>C</bold> and <bold>D</bold>) Examples of genes with most increased or decreased relative 5′ UTR translation include <italic>HSPA8</italic> (Hsc70 protein) and <italic>CYCS</italic> (cytochrome <italic>c</italic>), respectively. (<bold>E</bold>–<bold>H</bold>) Genes known to be involved in ER stress response show increased read density at 5′ UTR upstream open reading frames (uORFs). <italic>ATF4</italic> shows density at known uORFs as well as a single upstream AUG codon. Putative uORF translation with noted initiation codon is also identified based on areas of increased read density for <italic>DDIT4</italic> (REDD1), <italic>PPP1R15B</italic>, and <italic>TXNIP</italic> (see detailed sequence information in <xref ref-type="fig" rid="fig3s2">Figure 3—figure supplement 2</xref>).</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.011">http://dx.doi.org/10.7554/eLife.01236.011</ext-link></p><p><supplementary-material id="SD2-data"><object-id pub-id-type="doi">10.7554/eLife.01236.012</object-id><label>Figure 3—source data 1.</label><caption><title>5′ UTR translation.</title><p>(<bold>A</bold>) RPKM values of reads uniquely mapped to either 5′ UTR or CDS of transcripts (same list of transcripts as in <xref ref-type="supplementary-material" rid="SD1-data">Figure 1—source data 1</xref>). Sequence and length of 5′ UTR and CDS (for RPKM calculation) are defined based on the canonical transcript isoform annotated in hg19. (<bold>B</bold>) Transcripts with 5′ UTR/CDS translation ratio increased ≥twofold at a minimum of three time points compared to transcript 5′ UTR/CDS translation ratio at 0 hr. <bold>(C)</bold> as in <bold>B</bold> but for transcripts with decreased 5′ UTR to CDS translation ratio.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.012">http://dx.doi.org/10.7554/eLife.01236.012</ext-link></p></caption><media mime-subtype="xlsx" mimetype="application" xlink:href="elife01236s002.xlsx"/></supplementary-material></p></caption><graphic xlink:href="elife01236f003"/></fig><fig id="fig3s1" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.013</object-id><label>Figure 3—figure supplement 1.</label><caption><title>5′ UTR length.</title><p>For genes found to have most increased (274 genes) and most decreased (219 genes) relative 5′ UTR translation during the time course, there is no significant difference in the distribution of annotated 5′ UTR length (p=0.94, Mann-Whitney U test).</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.013">http://dx.doi.org/10.7554/eLife.01236.013</ext-link></p></caption><graphic xlink:href="elife01236fs005"/></fig><fig id="fig3s2" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.014</object-id><label>Figure 3—figure supplement 2.</label><caption><title>Ribosome profiling demonstrates strong footprint read density in known uORFs of <italic>DDIT3</italic> (CHOP) and <italic>PPP1R15A</italic> (GADD34).</title><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.014">http://dx.doi.org/10.7554/eLife.01236.014</ext-link></p></caption><graphic xlink:href="elife01236fs006"/></fig><fig id="fig3s3" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.015</object-id><label>Figure 3—figure supplement 3.</label><caption><title>Sequence data indicating uORFs suggested by increased footprint read density in <italic>REDD1</italic>, <italic>TXNIP</italic>, and <italic>PPP1R15B</italic> 5′ UTRs.</title><p>Putative <italic>REDD1</italic> uORFs are translated out of frame with the CDS; <italic>TXNIP</italic> uORFs 1 and 2 are out of frame, while uORF3 is in frame; both <italic>PPP1R15B</italic> uORFs are in frame with the CDS.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.015">http://dx.doi.org/10.7554/eLife.01236.015</ext-link></p></caption><graphic xlink:href="elife01236fs007"/></fig></fig-group></p><p>We next investigated whether there were general changes in proteome control by differing ribosome occupancy of the mRNA 5′ untranslated region (UTR). Ribosome occupancy in this region may indicate translation of short regulatory polypeptides in upstream open reading frames (uORFs) or production of alternate N-terminal isoforms of canonically translated proteins (<xref ref-type="bibr" rid="bib25">Ingolia et al., 2011</xref>; <xref ref-type="bibr" rid="bib33">Lee et al., 2012</xref>; <xref ref-type="bibr" rid="bib62">Stern-Ginossar et al., 2012</xref>). Others have shown that yeast responding to oxidative stress produce large increases in 5′ UTR translation (<xref ref-type="bibr" rid="bib16">Gerashchenko et al., 2012</xref>). A general decrease in 5′ UTR translation was also seen in differentiating mouse embryonic stem cells (<xref ref-type="bibr" rid="bib25">Ingolia et al., 2011</xref>). In contrast, we identified no overall trend toward altered 5′ UTR translation relative to CDS translation in our system (<xref ref-type="fig" rid="fig3">Figure 3B</xref>). However, these frequency distributions are broad and some individual transcripts do have large changes in the relative translation of the 5′ UTR during apoptosis. We examined transcripts with >twofold change in 5′ UTR translation relative to CDS translation when compared to untreated sample in at least three time points after drug exposure (<xref ref-type="supplementary-material" rid="SD2-data">Figure 3—source data 1</xref>). By χ<sup>2</sup> analysis, genes in Cluster Upreg were significantly over-represented among the 274 genes in the increased UTR translation group (p=0.033). Both Clusters Upreg and Downreg were over-represented among the 219 genes in the decreased UTR translation group (p=0.0006 and p=0.014, respectively). No other Clusters showed significant over- or under-representation. We did not find any difference in 5′ UTR length between the groups with most increased and most decreased relative 5′ UTR translation (<xref ref-type="fig" rid="fig3s1">Figure 3—figure supplement 1</xref>). IPA (<xref ref-type="table" rid="tbl1">Table 1</xref>) and inspection of genes in the increased UTR translation group revealed numerous translation elongation and initiation factors, ribosome structural proteins, chaperones (i.e., <italic>DNAJA1</italic>, <italic>HSP90AB1</italic>, <italic>CCT5</italic>, <italic>HSPA8</italic>; <xref ref-type="fig" rid="fig3">Figure 3C</xref>), and proteasomal subunits. The decreased UTR translation group does not show as strong a biological pattern although we do find ubiquitin (<italic>UBB</italic>), the co-chaperone BAG3, and cytochrome <italic>c</italic> (<italic>CYCS</italic>; <xref ref-type="fig" rid="fig3">Figure 3D</xref>), that are relevant to bortezomib-induced apoptosis.</p><p>The production of ATF4 is thought to be governed by the relative translation of two uORFs in the 5′ UTR (<xref ref-type="bibr" rid="bib37">Lu et al., 2004</xref>). We find that changes in the footprint density across both of these uORFs correspond with increased translation of the <italic>ATF4</italic> CDS (<xref ref-type="fig" rid="fig1s3">Figure 1—figure supplement 3</xref>) and detection of the protein product by immunoblot (<xref ref-type="fig" rid="fig1s2">Figure 1—figure supplement 2</xref>). We do also note strong footprint read density of a single amino acid uORF 22 nucleotides from the 5′ end of the ATF4 transcript (<xref ref-type="fig" rid="fig3">Figure 3E</xref>). Ribosome occupancy of this single methionine is of unclear significance, though it may also play an uncharacterized role in regulating ATF4 translation in myeloma.</p><p>We note that in these samples translation elongation was inhibited by cycloheximide. This treatment does not identify uORFs with as high precision as inhibition of initiation complexes by harringtonine or lactimidomycin (<xref ref-type="bibr" rid="bib25">Ingolia et al., 2011</xref>; <xref ref-type="bibr" rid="bib33">Lee et al., 2012</xref>; <xref ref-type="bibr" rid="bib62">Stern-Ginossar et al., 2012</xref>). Thus, we cannot confidently assign 5′ UTR read density across all genes to specific start codons, particularly non-AUG uORFs. However, we did focus on a subset of genes important in the ER stress response. We first looked at <italic>DDIT3</italic> (CHOP) and <italic>PPP1R15A</italic> (GADD34), where in both genes we found strong footprint read density in uORFs known to regulate CDS translation (<xref ref-type="fig" rid="fig3s2">Figure 3—figure supplement 2</xref>) (<xref ref-type="bibr" rid="bib26">Jousse et al., 2001</xref>; <xref ref-type="bibr" rid="bib34">Lee et al., 2009</xref>). We next examined other genes with a role in the ER stress response but without known uORF translation. These include <italic>DDIT4</italic> (REDD1), a repressor of mTOR signaling potentially under control of <italic>ATF4</italic> (<xref ref-type="bibr" rid="bib66">Whitney et al., 2009</xref>); <italic>TXNIP</italic>, recently found to be a key mediator of apoptosis under the UPR (<xref ref-type="bibr" rid="bib35">Lerner et al., 2012</xref>; <xref ref-type="bibr" rid="bib48">Oslowski et al., 2012</xref>); and <italic>PPP1R15B</italic>, which also plays a role in the dephosphorylation of eIF2α (<xref ref-type="bibr" rid="bib20">Harding et al., 2009</xref>). Intriguingly, all three genes showed apparent uORF translation, identified as areas of increased 5′ UTR read density with a peak corresponding to AUG or near-AUG initiation codons (<xref ref-type="bibr" rid="bib25">Ingolia et al., 2011</xref>) and bounded by a stop codon on the 3′ end (<xref ref-type="fig" rid="fig3">Figure 3F–H</xref>, <xref ref-type="fig" rid="fig3s3">Figure 3—figure supplement 3</xref>).</p></sec><sec id="s2-5"><title>Quantitative proteomics detects only a small subset of anti-apoptotic proteins notably produced</title><p>We next explored how the changes at the level of transcription and translation compare to the changes in protein levels during bortezomib-induced apoptosis. At each time point we isolated total protein and employed 6-plex iTRAQ labeling with a different isobaric mass tag (<xref ref-type="bibr" rid="bib43">Mertins et al., 2012</xref>) so that data from all time points can be compared simultaneously in a single mass spectrometry (MS) experiment. After analysis on two different mass spectrometers, a total of 2686 proteins were identified with at least two unique peptides and quantifiable iTRAQ reporter intensities. Typical MS spectra are shown in <xref ref-type="fig" rid="fig4">Figure 4A</xref>.<fig-group><fig id="fig4" position="float"><object-id pub-id-type="doi">10.7554/eLife.01236.016</object-id><label>Figure 4.</label><caption><title>Unbiased and targeted quantitative proteomics reveals abundance changes in a small subset of proteins.</title><p>(<bold>A</bold>) Example mass spectra demonstrating both <italic>m/z</italic> peaks used for peptide sequencing and iTRAQ reporter ion signal to measure relative abundance across time points. (<bold>B</bold>) Hierarchical clustering heat map of paired mRNA and ribosome footprint relative read density vs 0 hr with relative iTRAQ protein abundance for 2572 proteins. (<bold>C</bold>) Inset of heat map for proteins increased in relative abundance by >50% at ≥2 time points shows few proteins are measurably produced during bortezomib-induced apoptosis. (<bold>D</bold>) Targeted selected reaction monitoring (SRM) assays orthogonally validate iTRAQ data for 152 proteins. In this representative data, each colored trace monitors the intensity of a given parent and fragment ion pair, as demarcated in the peptide sequence; multiple co-eluting peaks positively identify a targeted peptide. (<bold>E</bold>) Protein abundance measured by both SRM and iTRAQ demonstrate strong correlation across time points (<italic>r</italic> = 0.80).</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.016">http://dx.doi.org/10.7554/eLife.01236.016</ext-link></p><p><supplementary-material id="SD3-data"><object-id pub-id-type="doi">10.7554/eLife.01236.017</object-id><label>Figure 4—source data 1.</label><caption><title>Proteomic data.</title><p>iTRAQ proteomic data organized at the protein (<bold>A</bold>) and peptide (<bold>B</bold>) level for 2686 proteins. All proteins included had a minimum of two unique peptides mapping to the protein; all peptides showed minimum iTRAQ reporter ion intensity of 300 cps at 0 hr. (<bold>C</bold>) iTRAQ proteomic data matched by UniProt Accession number to genes tracked in mRNA and footprint data across the time course. (<bold>D</bold>) List of proteins identified by iTRAQ proteomics but not present in analyzed genes from RNA deep sequencing. (<bold>E</bold>) List of parent and fragment ion transitions used for all peptides in SRM validation of iTRAQ data. (<bold>F</bold>) Protein and (<bold>G</bold>) peptide intensity data for targets tracked in SRM assay.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.017">http://dx.doi.org/10.7554/eLife.01236.017</ext-link></p></caption><media mime-subtype="xlsx" mimetype="application" xlink:href="elife01236s003.xlsx"/></supplementary-material></p></caption><graphic xlink:href="elife01236f004"/></fig><fig id="fig4s1" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.018</object-id><label>Figure 4—figure supplement 1.</label><caption><title>Protein abundance comparison.</title><p>Frequency distribution of protein abundance monitored by iTRAQ proteomics (∼2700 proteins) and ribosome profiling (∼5700 proteins) compared to average cellular protein abundance as estimated in PaxDB (<ext-link ext-link-type="uri" xlink:href="http://www.pax-db.org/">www.pax-db.org</ext-link>). We note that both distributions examine similar portions of the proteome, though the iTRAQ demonstrates greater coverage of higher abundance proteins.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.018">http://dx.doi.org/10.7554/eLife.01236.018</ext-link></p></caption><graphic xlink:href="elife01236fs008"/></fig><fig id="fig4s2" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.019</object-id><label>Figure 4—figure supplement 2.</label><caption><title>Comparison of baseline (0 hr) read density of mRNA transcripts vs number of identified peptides mapping to each protein in iTRAQ proteomics.</title><p>We find a weak but positive correlation (<italic>R</italic> = 0.27, p<0.0001). This distribution is comparable to the relationship between absolute transcript and protein copies at baseline in <xref ref-type="fig" rid="fig5s2">Figure 5—figure supplement 2</xref>.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.019">http://dx.doi.org/10.7554/eLife.01236.019</ext-link></p></caption><graphic xlink:href="elife01236fs009"/></fig><fig id="fig4s3" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.020</object-id><label>Figure 4—figure supplement 3.</label><caption><title>Relative 5′ UTR translation across the time course for upregulated genes.</title><p>Relative 5′ UTR translation (measured as normalized footprint read density in 5′ UTR vs normalized footprint read density in coding sequence) over the time course for selected genes (<bold>A</bold>) upregulated at the mRNA and footprint level and also increased by iTRAQ proteomics and (<bold>B</bold>) upregulated at the mRNA and footprint level with no detected change by iTRAQ. We note that for both sets of genes relative 5′ UTR translation stays relatively constant during bortezomib-induced apoptosis, consistent with genome-wide analysis in <xref ref-type="fig" rid="fig3">Figure 3B</xref>.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.020">http://dx.doi.org/10.7554/eLife.01236.020</ext-link></p></caption><graphic xlink:href="elife01236fs010"/></fig><fig id="fig4s4" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.021</object-id><label>Figure 4—figure supplement 4.</label><caption><title>Unbiased and targeted proteomics comparison to deep sequencing data.</title><p>Data from 150 proteins targeted in SRM assay to validate iTRAQ data. Comparison heat map of mRNA and footprint read density with iTRAQ and SRM relative protein abundance.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.021">http://dx.doi.org/10.7554/eLife.01236.021</ext-link></p></caption><graphic xlink:href="elife01236fs011"/></fig></fig-group></p><p>We compared the proteins identified by MS as well as those tracked by ribosome profiling to a database of cellular protein abundance (<ext-link ext-link-type="uri" xlink:href="http://www.pax-db.org/">www.pax-db.org</ext-link>; <xref ref-type="fig" rid="fig4s1">Figure 4—figure supplement 1</xref>; <xref ref-type="bibr" rid="bib65">Wang et al., 2012</xref>). There was significant overlap between the protein abundance range monitored both by ribosome profiling and iTRAQ proteomics, demonstrating that these two techniques are probing a similar portion of the proteome. We concentrated on 2572 of these proteins that also had corresponding mRNA and ribosome footprint data (<xref ref-type="supplementary-material" rid="SD3-data">Figure 4—source data 1</xref>). We noted a slight positive correlation (<italic>R</italic> = 0.27) between baseline transcript read density and number of peptides identified by iTRAQ proteomics (<xref ref-type="fig" rid="fig4s2">Figure 4—figure supplement 2</xref>). Strikingly, for the vast majority of proteins there is no distinct change in relative protein abundance while large changes are observed in relative mRNA and footprint reads (<xref ref-type="fig" rid="fig4">Figure 4B</xref>).</p><p>Nonetheless, we did find a subset of 12 proteins (0.47% of total) with >50% increase in protein abundance across at least two time points (<xref ref-type="fig" rid="fig4">Figure 4C</xref>). The majority of these proteins show strong upregulation at the mRNA and footprint level and have potential functions mediating either the folding (<italic>HSPA1A, DNAJB1, BAG3, HSPB1, SERPINH1, CLU</italic>) or degradation (<italic>SQSTM1</italic>) of unfolded proteins. These proteins are key players in an anti-apoptotic response that reduce cellular stress after proteasome inhibition by bortezomib. Interestingly, we also find ferritin light chain (<italic>FTL</italic>) increased (<xref ref-type="fig" rid="fig4">Figure 4C</xref>). Another group has recently found that depleting ferritin light chain enhances myeloma cell sensitivity to bortezomib (<xref ref-type="bibr" rid="bib9">Campanella et al., 2012</xref>), suggesting it also plays an anti-apoptotic role here. Of note, we did not observe any distinct changes in relative 5′ UTR translation for the few proteins increased by iTRAQ (<xref ref-type="fig" rid="fig4s3">Figure 4—figure supplement 3</xref>).</p><p>To further validate the iTRAQ findings, we employed an orthogonal method of label-free, targeted, quantitative mass spectrometry termed Selected reaction monitoring (SRM). This method is analogous to Western blotting without the need for dozens of antibodies and potential improvement in performance characteristics (<xref ref-type="bibr" rid="bib40">Maiolica et al., 2012</xref>). We prepared an independent biological time course of MM1.S cells treated with 20 nM bortezomib and isolated tryptic peptides from total cellular lysate. We tracked the relative abundance of one to three peptides from 152 proteins. The SRM peak area intensity data was compared to the iTRAQ reporter signal (e.g., in <xref ref-type="fig" rid="fig4">Figure 4A–D</xref>, respectively; heat map in <xref ref-type="fig" rid="fig4s4">Figure 4—figure supplement 4</xref>). Comparing across all time points in <xref ref-type="fig" rid="fig4">Figure 4E</xref> we found a strong correlation between SRM and iTRAQ data (<italic>R</italic> = 0.80). For the few increased proteins the abundance changes detected by SRM are greater than those found by iTRAQ, which is consistent with the greater suppression of iTRAQ reporter ions often seen in complex samples (<xref ref-type="bibr" rid="bib49">Ow et al., 2009</xref>). The label-free SRM data do not suffer from this same limitation and more likely reflect the true change in protein abundance. This independent MS assay confirms that only a small subset of proteins show measurable increases in abundance during apoptosis.</p></sec><sec id="s2-6"><title>Global translational shutdown during apoptosis</title><p>Our quantitative proteomic results are strongly consistent with polysome analysis (<xref ref-type="fig" rid="fig5s1">Figure 5—figure supplement 1</xref>) showing a significant decrease in translation well before any loss in cell viability. We wished to better understand the mechanism leading to this translational shutdown during apoptosis. During the UPR, it is well-known that phosphorylation of the initiation factor eIF2α by the ER-resident sensor kinase PERK is associated with general translation inhibition (<xref ref-type="bibr" rid="bib64">Walter and Ron, 2011</xref>). At early time points we find that PERK appears phosphorylated, as previously reported (<xref ref-type="bibr" rid="bib3">Atkins et al., 2013</xref>), and at later time points it undergoes proteolysis (<xref ref-type="fig" rid="fig5">Figure 5A</xref>). In contrast to other drugs that induce ER stress, we find levels of eIF2α phosphorylation actually decrease after bortezomib exposure (<xref ref-type="fig" rid="fig5">Figure 5A</xref>). This finding suggests that there must be other drivers of translation inhibition in this system.<fig-group><fig id="fig5" position="float"><object-id pub-id-type="doi">10.7554/eLife.01236.022</object-id><label>Figure 5.</label><caption><title>Translational shutdown during apoptosis and quantitative modeling reconciles deep sequencing and proteomic data.</title><p>(<bold>A</bold>) Western blotting tracked UPR proteins important in modulating translation via ER stress. We detect the known caspase cleavage of PERK and 4E-BP1. (<bold>B</bold>) Differential rate equations describing protein production. R(<italic>t</italic>) = mRNA abundance; P(<italic>t</italic>) = protein abundance; <italic>k</italic><sub>sp</sub> = protein translation rate constant; <italic>k</italic><sub>dp</sub> = protein degradation rate constant. G(<italic>t</italic>) describes global mRNA degradation (<xref ref-type="fig" rid="fig1">Figure 1B</xref>). (<bold>C</bold>) Absolute copies per cell at 0 hr of both mRNA (∼5700 transcripts) and protein measured by iBAQ (∼3400 proteins) demonstrates distributions of same order of magnitude as those seen in mouse NIH3T3 cells (<xref ref-type="bibr" rid="bib56">Schwanhausser et al., 2011</xref>). (<bold>D</bold>) Extrapolation of absolute mRNA copy numbers per cell based on RPKM, not accounting for mRNA degradation, is well-fit by either a sigmoid or quadratic function to define R(<italic>t</italic>). (<bold>E</bold>) Absolute protein copies per cell were extrapolated from iBAQ results based on SRM assay intensity. The modified mass-action model incorporating mRNA degradation was able to well-fit the subset of proteins that were detectably increased and also explained why others increased at the transcriptional level did not show protein increases.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.022">http://dx.doi.org/10.7554/eLife.01236.022</ext-link></p><p><supplementary-material id="SD4-data"><object-id pub-id-type="doi">10.7554/eLife.01236.023</object-id><label>Figure 5—source data 1.</label><caption><title>mRNA and protein absolute abundance.</title><p>(<bold>A</bold>) Absolute copies of mRNA and (<bold>B</bold>) protein per MM1.S cell at 0 hr as determined using mRNA read density and iBAQ calculation, respectively, as described in the ‘Materials and methods’. (<bold>C</bold>) Relative 5′ UTR translation at baseline vs absolute protein copies per cell by iBAQ. (<bold>D</bold>) Extrapolated mRNA and protein copies per cell for 13 genes as used for model fitting as shown in <xref ref-type="fig" rid="fig5s4">Figure 5—figure supplement 4</xref>.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.023">http://dx.doi.org/10.7554/eLife.01236.023</ext-link></p></caption><media mime-subtype="xlsx" mimetype="application" xlink:href="elife01236s004.xlsx"/></supplementary-material></p></caption><graphic xlink:href="elife01236f005"/></fig><fig id="fig5s1" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.024</object-id><label>Figure 5—figure supplement 1.</label><caption><title>Polysome profiles.</title><p>Polysome profiles by sucrose gradient during the apoptotic time course show decrease in polysome peaks over time during during apoptosis. Translation is largely suppressed well before any loss in cell viability.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.024">http://dx.doi.org/10.7554/eLife.01236.024</ext-link></p></caption><graphic xlink:href="elife01236fs012"/></fig><fig id="fig5s2" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.025</object-id><label>Figure 5—figure supplement 2.</label><caption><title>Comparison between deep sequencing and proteomic data.</title><p>Correlation of mRNA amount (<bold>A</bold>) and ribosome footprint (<bold>B</bold>) read density (in RPKM) with absolute protein abundance by iBAQ MS analysis at 0 hr in MM1.S cells demonstrates limited correlation. (<bold>C</bold>) Strong correlation of ribosome footprint and mRNA read density at 0 hr. All correlations in <bold>A</bold>–<bold>C</bold> with p<0.0001. (<bold>D</bold>) Relative 5′ UTR translation at baseline vs protein abundance. We note a weak but non-significant correlation (p=0.35) suggesting that increased 5′ UTR translation may relate to lower steady-state protein abundance.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.025">http://dx.doi.org/10.7554/eLife.01236.025</ext-link></p></caption><graphic xlink:href="elife01236fs013"/></fig><fig id="fig5s3" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.026</object-id><label>Figure 5—figure supplement 3.</label><caption><title>mRNA-seq data scaled by global mRNA degradation.</title><p>Relative mRNA expression data from <xref ref-type="fig" rid="fig4">Figure 4B</xref> scaled to reflect global mRNA degradation at later time points during apoptosis. By this scaling (scaling factors for mRNA based on fit G(<italic>t</italic>) function = 1.0 at 0 hr, 0.93 at 1.5 hr, 0.91 at 3 hr, 0.48 at 6 hr, 0.17 at 9 hr, 0.12 at 12 hr), assuming that all mRNA are degraded at the same rate, the absolute mRNA abundance can be estimated across all cells in the population. Upper inset demonstrate that transcripts with absolute mRNA abundance changes across the entire time course are more likely to have increases in protein level, whereas those with transient increases in mRNA generally do not result in detectable protein production.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.026">http://dx.doi.org/10.7554/eLife.01236.026</ext-link></p></caption><graphic xlink:href="elife01236fs014"/></fig><fig id="fig5s4" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.027</object-id><label>Figure 5—figure supplement 4.</label><caption><title>Model fits to 13 selected genes.</title><p>We used our quantitative model to describe a subset of genes increased at both the transcript and protein level (<bold>A</bold> and <bold>B</bold>) and those where relative transcript levels increased but protein levels were unchanged (<bold>C</bold> and <bold>D</bold>). Solid lines in (<bold>A</bold> and <bold>C</bold>) represent quadratic fits (FTL, SQSTM1, CTSD) or four-parameter sigmoid fits (all others) to relative mRNA abundance data to describe R(<italic>t</italic>), as in <xref ref-type="fig" rid="fig5">Figure 5E</xref>. Solid lines in (<bold>B</bold> and <bold>D</bold>) represent fits to modified mass-action to describe P(<italic>t</italic>) as shown in <xref ref-type="fig" rid="fig5">Figure 5F</xref>.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.027">http://dx.doi.org/10.7554/eLife.01236.027</ext-link></p></caption><graphic xlink:href="elife01236fs015"/></fig><fig id="fig5s5" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.028</object-id><label>Figure 5—figure supplement 5.</label><caption><title>Predicted protein changes from mRNA changes.</title><p>Model predictions demonstrating that even during apoptosis, for the same increase in relative transcript levels low-abundance proteins (500 copies per cell at baseline) will have substantial relative protein increases, whereas high abundance proteins (100,000 copies per cell at baseline) will have much smaller changes in relative protein abundance. This is even true when the absolute transcript abundance change is much greater for the high abundance protein and including <italic>k</italic><sub>sp</sub>, the rate constant describing proteins produced per mRNA per hour, as greater for the high abundance protein based on the results of <xref ref-type="bibr" rid="bib56">Schwanhausser et al. 2011</xref>.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.028">http://dx.doi.org/10.7554/eLife.01236.028</ext-link></p></caption><graphic xlink:href="elife01236fs016"/></fig></fig-group></p><p>Phosphorylation and dephosphorylation of proteins in the eIF4 complex may also inhibit mRNA cap-dependent translation, including hypophosphorylation of 4E-BP1 (<xref ref-type="bibr" rid="bib61">Spriggs et al., 2010</xref>). Indeed, we observed 4E-BP1 dephosphorylation at the 6 hr time point. This is followed by rapid caspase degradation of 4E-BP1 during cell death (<xref ref-type="fig" rid="fig5">Figure 5A</xref>). It is also known that caspases target other proteins in the eIF4 complex (<xref ref-type="bibr" rid="bib8">Bushell et al., 2004</xref>). Using N-terminomics (described in more detail below) we found that proteolysis of initiation factors begins at the 6 hr time point and accelerates later in apoptosis (<xref ref-type="fig" rid="fig6s1">Figure 6—figure supplement 1</xref>). Another potentially important mechanism is the global degradation of mRNA during apoptosis (<xref ref-type="bibr" rid="bib13">Del Prete et al., 2002</xref>) (<xref ref-type="fig" rid="fig1">Figure 1B</xref>), depriving ribosomes of substrates for translation. Together, these results demonstrate that well before the loss of cellular viability, translation is inhibited both by shutting down translation initiation and destruction of mRNA.</p></sec><sec id="s2-7"><title>A quantitative mass-action model describes protein changes</title><p>We sought a quantitative explanation for the limited changes in relative protein level we observed despite large changes in relative transcript and ribosome footprints. Recently developed systems-level technologies, as we use here, enable genome-wide quantitative assessment of the efficiency for decoding of mRNA to protein (<xref ref-type="bibr" rid="bib63">Vogel and Marcotte, 2012</xref>). In this apoptotic system, proteins are relatively stable due to several factors: at 20 nM bortezomib in MM1.S cells, proteasomal activity is almost completely inhibited (<xref ref-type="bibr" rid="bib6">Berkers et al., 2005</xref>). The proteasome is also extensively cut by caspases leading to loss of activity (<xref ref-type="bibr" rid="bib19">Gray et al., 2010</xref>; <xref ref-type="fig" rid="fig6s1">Figure 6—figure supplement 1</xref>). Moreover, we find that endoproteolysis during apoptosis cuts fewer than 20% of the proteins in the cell and often only once or twice per protein leading to stable domains (<xref ref-type="bibr" rid="bib14">Dix et al., 2008</xref>; <xref ref-type="bibr" rid="bib39">Mahrus et al., 2008</xref>; <xref ref-type="bibr" rid="bib12">Crawford et al., 2013</xref>). These domains would be indistinguishable from intact proteins by iTRAQ. Therefore, we expect that detectable protein degradation is extremely limited and changes by iTRAQ largely reflect protein production alone. In contrast, mRNAs with very different stabilities (steady-state half-lives on the order of minutes to hours) are rapidly degraded with similar kinetics under various apoptotic inducers even prior to cell death (<xref ref-type="bibr" rid="bib13">Del Prete et al., 2002</xref>).</p><p>Mass-action models, as shown in <xref ref-type="fig" rid="fig5">Figure 5B</xref>, can describe the dynamic relationship between protein and mRNA but require absolute abundance measurements per cell. As described in (<xref ref-type="bibr" rid="bib56">Schwanhausser et al., 2011</xref>), we used measurements of total mRNA combined with mRNA-seq read density to estimate the absolute abundance of each transcript per cell in untreated MM1.S cells. For absolute protein abundance we used a recently described method of label-free quantitation termed intensity based absolute quantitation (iBAQ) (<xref ref-type="bibr" rid="bib56">Schwanhausser et al., 2011</xref>). From untreated MM1.S cells we generated absolute abundance estimates for 3369 proteins (<xref ref-type="supplementary-material" rid="SD4-data">Figure 5—source data 1</xref>). Compared to baseline protein abundance, we found a slightly stronger correlation for footprint read density than mRNA (<xref ref-type="fig" rid="fig5s2">Figure 5—figure supplement 2</xref>). Interestingly, we also found that increased 5′ UTR translation at baseline may correlate with decreased steady-state protein abundance (<xref ref-type="fig" rid="fig5s2">Figure 5—figure supplement 2</xref>). Overall distributions of protein and mRNA copies per MM1.S cell, with a median of 11 transcript copies and ∼16,000 protein copies (<xref ref-type="fig" rid="fig5">Figure 5C</xref>), were similar to those previously found in mouse NIH3T3 cells (<xref ref-type="bibr" rid="bib56">Schwanhausser et al., 2011</xref>).</p><p>We used these absolute abundance measurements to prepare an approximate quantitative model in our system. We built a mass-action model similar to that in (<xref ref-type="bibr" rid="bib56">Schwanhausser et al., 2011</xref>) but modified to incorporate a term to account for global mRNA degradation (<xref ref-type="fig" rid="fig1 fig5">Figures 1B and 5B</xref>) (see ‘Materials and methods’ for details of modeling). This model describes the change in protein copies per cell as a function of transcript copies per cell, the rate of translation, and the rate of protein degradation. We analyzed a subset of 13 genes with transcriptional increase but non-uniform changes in protein levels (<xref ref-type="fig" rid="fig5">Figure 5D</xref>, <xref ref-type="fig" rid="fig5s4">Figure 5—figure supplement 4</xref>). This model was able to fit the protein abundance data, primarily by varying the term <italic>k</italic><sub>sp<italic>,</italic></sub> describing the number of proteins produced per transcript per hour (<xref ref-type="fig" rid="fig5">Figure 5E</xref>, <xref ref-type="fig" rid="fig5s4">Figure 5—figure supplement 4</xref>). <italic>k</italic><sub>sp</sub> ranged from 10 to 270, consistent with the range found previously (<xref ref-type="bibr" rid="bib56">Schwanhausser et al., 2011</xref>).</p><p>Based on this model, we can begin to reconcile our deep sequencing and proteomic data. During the early stages of apoptosis, the group of transcripts with multi-fold increases in relative abundance do not all lead to measurable protein increases. This is because the absolute number of these transcripts per cell in the sample remains fairly stable as transcription and mRNA degradation offset. This is illustrated by scaling the relative mRNA data shown in <xref ref-type="fig" rid="fig4">Figure 4A</xref> by a factor reflecting global mRNA degradation (<xref ref-type="fig" rid="fig5s3">Figure 5—figure supplement 3</xref>). Notably, the few proteins that are increased tend to demonstrate increases in absolute transcript level at later time points. Our model further shows that with the same fold-increase in transcript level, proteins with low baseline abundance will have more dramatic increases in relative protein concentration when compared to proteins with high baseline abundance (<xref ref-type="fig" rid="fig5s5">Figure 5—figure supplement 5</xref>).</p></sec><sec id="s2-8"><title>Cellular struggle with chemotherapy does not affect apoptotic proteolysis</title><p>It is unknown how the cellular response to chemotherapeutic treatment influences cellular deconstruction by proteases during apoptosis, including both caspase and non-caspase proteases (<xref ref-type="bibr" rid="bib52">Pop and Salvesen, 2009</xref>; <xref ref-type="bibr" rid="bib45">Moffitt et al., 2010</xref>). Our laboratory has developed an N-terminomics technology using the engineered enzyme subtiligase to specifically label and enrich for free protein N-termini generated by proteolysis in complex biological samples (<xref ref-type="bibr" rid="bib39">Mahrus et al., 2008</xref>). Others have also identified caspase cleavage substrates after treatment with chemotherapeutics with different N-terminomics approaches (<xref ref-type="bibr" rid="bib15">Gausdal et al., 2008</xref>; <xref ref-type="bibr" rid="bib23">Impens et al., 2008</xref>). By MS analysis in a variety of cell lines under both apoptotic and non-apoptotic conditions our laboratory has identified >8000 protein cleavage sites, including >1700 putative caspase substrates with an Asp immediately N-terminal to cleavage site (P1 position). We have compiled these cleavage sites into the ‘DegraBase’ (<xref ref-type="bibr" rid="bib12">Crawford et al., 2013</xref>); <ext-link ext-link-type="uri" xlink:href="http://wellslab.ucsf.edu/degrabase">wellslab.ucsf.edu/degrabase</ext-link>).</p><p>We have previously found that different apoptotic inducers result in different patterns of caspase substrate cleavage, though the reasons for these differences remain unknown (<xref ref-type="bibr" rid="bib59">Shimbo et al., 2012</xref>). We can now directly compare whether the cellular response to chemotherapy at the transcriptional and translational level influences the cleavage of substrates during terminal apoptosis. For each Cluster (<xref ref-type="fig" rid="fig1">Figure 1C</xref>) we queried the DegraBase to investigate if a potential substrate was identified in the DegraBase, if it was cut by caspases, and, if so, the relative frequency of caspase cleavage. If the cellular response to bortezomib induced common anti-apoptotic pathways targeted by the caspases, it was possible that genes in Cluster Upreg would be more prominently featured as substrates in the DegraBase. However, we found no evidence that genes in any Cluster were significantly enriched (or de-enriched) among identified caspase substrates (<xref ref-type="fig" rid="fig6">Figure 6A</xref>).<fig-group><fig id="fig6" position="float"><object-id pub-id-type="doi">10.7554/eLife.01236.029</object-id><label>Figure 6.</label><caption><title>N-terminomics tracks proteolytic cleavage substrates.</title><p>(<bold>A</bold>) We compared defined hierarchical clusters (<xref ref-type="fig" rid="fig1">Figure 1C</xref>) to all proteolytic cleavage substrates present in the DegraBase, a database of over 8000 cleavage events (<xref ref-type="bibr" rid="bib12">Crawford et al., 2013</xref>). D at P1 suggests a putative caspase substrate. We find no significant variation in levels of proteolysis by χ<sup>2</sup> analysis of each cluster compared to the set of all substrates. (<bold>B</bold>) Representative SRM intensity for appearance of proteolytically-processed peptides, both caspase- and non-caspase-cleaved (<xref ref-type="supplementary-material" rid="SD5-data">Figure 6—source data 1</xref>). Each trace represents a single parent ion/fragment ion transition; co-elution of multiple targeted transitions confirms peptide. (<bold>C</bold>) Heat map showing relative SRM intensity of 252 proteolytic peptides monitored in this system compared to 0 hr baseline. Five groups defined by unsupervised hierarchical clustering representing relative cleavage kinetics show caspase proteolysis generally precedes other apoptotic proteolytic events.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.029">http://dx.doi.org/10.7554/eLife.01236.029</ext-link></p><p><supplementary-material id="SD5-data"><object-id pub-id-type="doi">10.7554/eLife.01236.030</object-id><label>Figure 6—source data 1.</label><caption><title>Degradomics SRM assay.</title><p>(<bold>A</bold>) SRM intensity data for peptides included in targeted degradomics assay. P1 residue refers to residue immediately N-terminal to identified proteolytic cleavage site based on protein sequence. (<bold>B</bold>) List of parent and fragment ion transitions tracked for targeted degradomic SRM assay.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.030">http://dx.doi.org/10.7554/eLife.01236.030</ext-link></p></caption><media mime-subtype="xlsx" mimetype="application" xlink:href="elife01236s005.xlsx"/></supplementary-material></p></caption><graphic xlink:href="elife01236f006"/></fig><fig id="fig6s1" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.031</object-id><label>Figure 6—figure supplement 1.</label><caption><title>N-terminomics biological data subgroups.</title><p>Log<sub>2</sub> expression of N-terminomics SRM data across apoptotic time course. N-terminomics data divided by biological function show that caspase proteolysis proceeds with similar rates across biological subdivisions. Proteins increased at the transcriptional and ribosome footprint level after bortezomib treatment are primarily included in the ‘chaperones and cell stress’ and ‘proteasome and ubiquitination’ groups. The ‘translation’ group demonstrates cleavage of many translation initiation factors.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.031">http://dx.doi.org/10.7554/eLife.01236.031</ext-link></p></caption><graphic xlink:href="elife01236fs017"/></fig><fig id="fig6s2" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.01236.032</object-id><label>Figure 6—figure supplement 2.</label><caption><title>N-terminomics data with caspase and non-caspase cleavage sites in same substrate.</title><p>These data demonstrate that even in the same protein substrates, caspase cleavage tends to proceed more rapidly during apoptosis than non-caspase cleavage.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.032">http://dx.doi.org/10.7554/eLife.01236.032</ext-link></p></caption><graphic xlink:href="elife01236fs018"/></fig></fig-group></p><p>These findings suggest that protease substrates are not targeted based on changes in their abundance. We further explored this hypothesis by monitoring the kinetic appearance of proteolytic substrates with a quantitative SRM assay (<xref ref-type="bibr" rid="bib1">Agard et al., 2012</xref>; <xref ref-type="bibr" rid="bib59">Shimbo et al., 2012</xref>). We included both non-caspase and caspase-cleaved peptides in our SRM assay with a preference for those found in multiple experiments under apoptotic conditions. Interestingly, many of the chaperone and co-chaperone proteins, relatively increased at the transcriptional and translational levels, had only non-caspase proteolytic peptides identified in the DegraBase. These included Hsp70-1, Hsp40, Hsp27, BiP (<italic>HSPA5</italic>), Hsc70 (<italic>HSPA8</italic>), and BAG3. The highly abundant and constitutively expressed chaperone Hsp90-β had numerous non-caspase cleavages and a single caspase-cleaved peptide.</p><p>We used this SRM assay to explore whether proteins upregulated in response to bortezomib were more rapidly cleaved during apoptotic proteolysis. In <xref ref-type="fig" rid="fig6">Figure 6B</xref> we show example SRM intensity data while <xref ref-type="fig" rid="fig6">Figure 6C</xref> displays a heat map of similar SRM intensity data compared to the untreated baseline across all 252 peptides targeted. Caspase-cleaved peptides begin to appear at the 6 hr time point, when caspase activity was also detected biochemically (<xref ref-type="fig" rid="fig1">Figure 1B</xref>). In <xref ref-type="fig" rid="fig6s1">Figure 6—figure supplement 1</xref> we subdivide the monitored peptides based on biological function. We do not find that stress response proteins are cleaved more rapidly than those in other groups. However, we note rapid cleavage of ATF4 (as seen before, [<xref ref-type="bibr" rid="bib59">Shimbo et al., 2012</xref>]), BAG3, and Hsp90-β. It is possible that these particular proteins represent important nodes in deconstructing the cellular stress response after bortezomib. Generally, our results suggest that drug-specific response at the transcriptional and translational level does not broadly affect protease dynamics once apoptosis ensues.</p></sec><sec id="s2-9"><title>Non-caspase proteolytic events with caspase-like kinetics</title><p>Our prior studies examining quantitative appearance of proteolytic substrates only focused on caspase substrates (<xref ref-type="bibr" rid="bib1">Agard et al., 2012</xref>; <xref ref-type="bibr" rid="bib59">Shimbo et al., 2012</xref>). Here, we also monitored the kinetics of appearance of non-caspase proteolytic events during apoptosis. We tracked 163 caspase-cleaved peptides (D at P1) and 89 peptides derived from non-caspase proteolysis. The list includes 52 proteins that were cleaved by both caspase- and non-caspase proteases to examine the relative kinetics of different proteolytic events in the same substrate (<xref ref-type="fig" rid="fig6s2">Figure 6—figure supplement 2</xref>).</p><p>Unsupervised hierarchical clustering was used to rank the rates of proteolytic cleavage events (<xref ref-type="fig" rid="fig6">Figure 6C</xref>). We defined five broad groups ranging from the Fastest cleavages, with robust substrate appearance at 6 hr, to Slowest/No Change, with no increase in cleavage even at late apoptosis. Interestingly, the large majority of non-caspase-cleaved substrates appear in the Slow and Slowest/No Change groups. While this only represents a small portion of the >3,000 non-caspase endoproteolytic cleavages in the DegraBase, this finding suggests that most non-caspase proteolytic events in apoptotic cells are a consequence of and not a driver of apoptosis itself.</p><p>However, we did find 16 non-caspase-cleaved peptides appearing in either the Fastest, Faster, or Mid categories (<xref ref-type="table" rid="tbl2">Table 2</xref>). Of the 89 non-caspase peptides monitored, 24 were tryptic-like (K or R at P1). Yet none of the 16 most rapidly cleaved peptides had a tryptic-like cleavage and only one has been previously annotated in MEROPS, the largest current database of proteolytic cleavage (<xref ref-type="bibr" rid="bib53">Rawlings et al., 2012</xref>). These 16 peptides were cleaved with kinetics similar to caspases, suggesting that they may play roles in driving apoptosis. Among these we noted a relative preponderance of Pro (3 peptides), Gln (4 peptides), and Glu (4 peptides) residues at the P1 site (<xref ref-type="table" rid="tbl2">Table 2</xref>). While caspases may occasionally cleave at Glu (<xref ref-type="bibr" rid="bib28">Krippner-Heidenreich et al., 2001</xref>), none of the other classic proteases thought to be active during MM1.S cell apoptosis, including the cathepsins or calpains, have a preference for Pro, Gln, or Glu at the P1 site (as described in MEROPS). Our findings suggest that, at least in a limited set of substrates, non-Asp cleavages may be mediated by uncharacterized proteases acting with similar kinetics as the caspases during apoptosis.<table-wrap id="tbl2" position="float"><object-id pub-id-type="doi">10.7554/eLife.01236.033</object-id><label>Table 2.</label><caption><title>Comparison of caspase- and non-caspase cleavages during apoptotic time course as monitored by N-terminomics and SRM</title><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.033">http://dx.doi.org/10.7554/eLife.01236.033</ext-link></p></caption><table frame="hsides" rules="groups"><tbody><tr><td>Total D = P1 peptides</td><td>163</td><td/></tr><tr><td rowspan="3">Total non-D = P1 peptides</td><td>89</td><td/></tr><tr><td>Tryptic-like (K or R at P1)</td><td>24</td></tr><tr><td>Non-tryptic (all other residues at P1)</td><td>65</td></tr></tbody></table><table frame="hsides" rules="groups"><thead><tr><th/><th>Cluster</th><th>D at P1</th><th>Non-D at P1</th></tr></thead><tbody><tr><td/><td><italic>Slowest/no change</italic></td><td>6</td><td>36</td></tr><tr><td/><td><italic>Slow</italic></td><td>23</td><td>28</td></tr><tr><td/><td><italic>Mid</italic></td><td>50</td><td>10</td></tr><tr><td/><td><italic>Faster</italic></td><td>37</td><td>2</td></tr><tr><td/><td><italic>Fastest</italic></td><td>38</td><td>4</td></tr></tbody></table><table frame="hsides" rules="groups"><thead><tr><th colspan="2">Non-D peptides in fastest three clusters</th><th>UniProt</th><th>Gene</th><th>P1</th><th>Sequence Position</th><th>MEROPS annotated?</th></tr></thead><tbody><tr><td><italic>Fastest</italic></td><td>(C)GQVAAAAAAQPPASHGPER</td><td>O95817</td><td>BAG3</td><td>C</td><td>151</td><td>N</td></tr><tr><td/><td>(S)AVGFNEMEAPTTAYK</td><td>P14317</td><td>HCLS1</td><td>S</td><td>208</td><td>N</td></tr><tr><td/><td>(P)GHGSGWAETPR</td><td>O75533</td><td>SF3B1</td><td>P</td><td>304</td><td>N</td></tr><tr><td/><td>(Q)VLTVPATDIAEETVISEEPPAKR</td><td>Q06547</td><td>GABP1</td><td>Q</td><td>306</td><td>N</td></tr><tr><td><italic>Fast</italic></td><td>(Q)ALKEEPQTVPEMPGETPPLSPIDMESQER</td><td>P05412</td><td>JUN</td><td>Q</td><td>223</td><td>N</td></tr><tr><td/><td>(P)AVNGATGHSSSLDAR</td><td>Q07817</td><td>BCL2L1</td><td>P</td><td>63</td><td>N</td></tr><tr><td><italic>Mid</italic></td><td>(E)AAGATGDAIEPAPPSQGAEAK</td><td>P49006</td><td>MARCKSL1</td><td>E</td><td>56</td><td>N</td></tr><tr><td/><td>(Q)AASGDVQTYQIR</td><td>P16220</td><td>CREB1</td><td>Q</td><td>243</td><td>N</td></tr><tr><td/><td>(E)GGIDMDAFQER</td><td>P29083</td><td>GTF2E1</td><td>E</td><td>297</td><td>N</td></tr><tr><td/><td>(A)SIFGGAKPVDTAAR</td><td>P23588</td><td>EIF4B</td><td>A</td><td>358</td><td>Y–meprin alpha (<xref ref-type="bibr" rid="bib5">Becker-Pauly et al., 2011</xref>)</td></tr><tr><td/><td>(E)AIQNFSFR</td><td>O75122</td><td>CLASP2</td><td>E</td><td>946</td><td>N</td></tr><tr><td/><td>(P)HFEPVVPLPDKIEVK</td><td>P49792</td><td>RANBP2</td><td>P</td><td>1170</td><td>N</td></tr><tr><td/><td>(N)SWFENAEEDLTDPVR</td><td>Q13813</td><td>SPTAN1</td><td>N</td><td>2104</td><td>N</td></tr><tr><td/><td>(L)AFSEQEEHELPVLSR</td><td>O75995</td><td>SASH3</td><td>L</td><td>128</td><td>N</td></tr><tr><td/><td>(Q)AIMEMGAVAADKGKK</td><td>O95817</td><td>BAG3</td><td>Q</td><td>522</td><td>N</td></tr><tr><td/><td>(E)AILEDEQTQR</td><td>Q9P2E9</td><td>RRBP1</td><td>E</td><td>1298</td><td>N</td></tr></tbody></table><table-wrap-foot><fn><p>By Fisher′s exact test there is a significant absence of tryptic-like cleavages in the 16 rapidly cleaved peptides (p=0.005).</p></fn></table-wrap-foot></table-wrap></p></sec><sec id="s2-10"><title>An integrated approach informs combination therapeutic regimens</title><p>Genes upregulated at both the transcriptional, translational, and protein level may represent the cellular protective response to avoid apoptosis. We thus hypothesized that by inhibiting these ‘first responders’ to the chemotherapeutic treatment we may be able to drive the malignant cell toward apoptosis and reduce therapeutic resistance. To this end we identified the most prominent upstream regulators of the genes in Cluster Upreg (<xref ref-type="table" rid="tbl1">Table 1</xref>). These include NRF2, a transcription factor regulating the hypoxic response; HSF1, a transcription factor regulating the heat shock response; and PERK, the ER-resident kinase. HSF1 in particular regulates the transcription of many genes we also found increased by proteomics experiments (<xref ref-type="bibr" rid="bib41">Mendillo et al., 2012</xref>).</p><p>We therefore explored whether HSF1 inhibition would enhance cell death by bortezomib in MM1.S cells. We used a recently described small molecule inhibitor of HSF1, KRIBB11 (<xref ref-type="bibr" rid="bib67">Yoon et al., 2011</xref>). This drug was added with and without a low dose of 2.5 nM bortezomib, which alone led to ∼50% cell death over 48 hr. At the 24 hr time point, the combination of KRIBB11 and bortezomib caused significant cell death, while neither compound alone decreased cell viability (<xref ref-type="fig" rid="fig7">Figure 7A</xref>).<fig id="fig7" position="float"><object-id pub-id-type="doi">10.7554/eLife.01236.034</object-id><label>Figure 7.</label><caption><title>Systems-level analysis suggests potential chemotherapeutic combinations with bortezomib in myeloma.</title><p>(<bold>A</bold>) Bort = bortezomib. HSF1 inhibitor used was KRIBB11 (<xref ref-type="bibr" rid="bib67">Yoon et al., 2011</xref>) and shows additive effect with bortezomib. (<bold>B</bold>) PERK inhibitor was GSK2606414 and shows additive effect with bortezomib (<xref ref-type="bibr" rid="bib4">Axten et al., 2012</xref>). All measurements in 24-well plates in triplicate; mean ± SD shown.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.01236.034">http://dx.doi.org/10.7554/eLife.01236.034</ext-link></p></caption><graphic xlink:href="elife01236f007"/></fig></p><p>There are no known small molecules which specifically inhibit NRF2, but we could test whether PERK inhibition could sensitize cells to death using a newly described small molecule PERK inhibitor, GSK2606414 (<xref ref-type="bibr" rid="bib4">Axten et al., 2012</xref>). Indeed, PERK inhibition did sensitize cells to death by bortezomib (<xref ref-type="fig" rid="fig7">Figure 7B</xref>). For both the HSF1 and PERK inhibitors the most robust effects were seen at concentrations in excess of the in vitro IC<sub>50</sub> (∼1.2 µM for KRIBB11, ∼1 nM for GSK2606414). Notwithstanding potential off-target effects, these results would support the notion that integrated systems-level data can inform potential chemotherapeutic combinations.</p></sec></sec><sec id="s3" sec-type="discussion"><title>Discussion</title><p>Here we present a systems-level study of how malignant cells respond to apoptosis induced by a clinically relevant chemotherapeutic. Our results show how myeloma cells initially struggle to adapt to the proteasome inhibitor bortezomib. A general inhibition of translation coupled with mRNA degradation leads to surprisingly little change in overall protein levels despite relative transcriptional increases in many genes. Furthermore, global protein levels are maintained due to inhibition of proteasomal activity by bortezomib and extensive cleavage of proteasomal subunits by caspases (<xref ref-type="bibr" rid="bib19">Gray et al., 2010</xref>). The cell also modulates translational control of stress responsive proteins related to bortezomib treatment. Co-targeting key regulators of these processes such as HSF1 and PERK can further accelerate apoptosis. Myeloma cells ultimately succumb to the pharmacologic insult, leading to activation of the caspases as well as one or more non-caspase proteases that have yet to be characterized. While pieces of this story have been told by other experiments, the approach we present here, examining the integrated regulation of the transcriptional, translational, and proteolytic machinery, gives a global and in-depth view of the apoptotic process.</p><p>Translational regulation during apoptosis has only been examined in a limited context. Bushell and coworkers examined the relative changes in translational efficiency during early TRAIL-induced apoptosis by polysome profiling (microarray analysis of mRNA associated with translating ribosomes) (<xref ref-type="bibr" rid="bib7">Bushell et al., 2006</xref>). Genes with the greatest degree of decrease in TE included many translation-related proteins, similar to our findings. Those with the largest increase, however, did not overlap with our results, suggesting that positive changes in TE may be specific to the apoptotic inducer or induction of the intrinsic vs extrinsic apoptotic pathways.</p><p>Importantly, after proteasomal inhibition in myeloma cells we found that translational reprogramming favors proteins which help the cell adapt to unfolded protein stress. Possible mechanisms for the alteration of translation efficiency involve preferential translation via internal ribosome entry sites (IRES) (<xref ref-type="bibr" rid="bib7">Bushell et al., 2006</xref>), partitioning of transcripts toward ER-bound ribosomes under stress (<xref ref-type="bibr" rid="bib54">Reid and Nicchitta, 2012</xref>), or other uncharacterized regulatory programs governing ribosomal-transcript association. We also identified uORF translation in genes related to the ER stress response, <italic>TXNIP</italic>, <italic>PPP1R15B</italic>, and <italic>DDIT3</italic> (REDD1), that had not previously been characterized. The exact interplay of uORF and CDS translation for these genes remains to be investigated, but it is possible there is a regulatory function similar to that already known for ATF4, CHOP, and GADD34. We also found that differential 5′ UTR translation appears common in the cellular response to bortezomib but relates variably to increased CDS translation. It will be interesting to compare the effects on translation regulation in different cell types exposed to chemotherapeutics with different mechanisms of action to find whether these changes are unique to bortezomib. Incorporation of translation initiation inhibitors in ribosome profiling and paired-end reads in mRNA-seq can also provide novel insight into translation regulation through generation of custom proteomic databases (<xref ref-type="bibr" rid="bib62">Stern-Ginossar et al., 2012</xref>; <xref ref-type="bibr" rid="bib42">Menschaert et al., 2013</xref>; <xref ref-type="bibr" rid="bib58">Sheynkman et al., 2013</xref>).</p><p>Surprisingly, we found that eIF2α phosphorylation did not play a major role in translation inhibition in this apoptotic system involving the UPR. Others have also observed decreased eIF2α phosphorylation in MM1.S cells after proteasome inhibition (<xref ref-type="bibr" rid="bib50">Parlati et al., 2009</xref>). However, eIF2α phosphorylation is thought to be necessary for increased ATF4 translation (<xref ref-type="bibr" rid="bib37">Lu et al., 2004</xref>). The mechanism leading to the induction of ATF4 in this system (<xref ref-type="fig" rid="fig1s2">Figure 1—figure supplement 2</xref>) is therefore unclear, but may relate to translation of the additional uORF we identified (<xref ref-type="fig" rid="fig3">Figure 3E</xref>).</p><p>Our paired transcriptomic and proteomic data stands in stark contrast to studies of heat and osmolar shock in yeast, where increases at the transcript level are well-correlated with increases in protein (<xref ref-type="bibr" rid="bib32">Lee et al., 2011</xref>; <xref ref-type="bibr" rid="bib29">Lackner et al., 2012</xref>). We used a quantitative model incorporating global mRNA degradation to explain this contrast with non-apoptotic systems. While our proteomic analysis is quite extensive, this mass-action model suggests that low abundance proteins not detected by MS, such as CHOP or XBP1 (<xref ref-type="fig" rid="fig1s2">Figure 1—figure supplement 2</xref>), are more likely to be increased in relative protein abundance after transcriptional changes. Future advances in proteomic techniques may allow us to detect more biologically relevant protein changes. The mass-action model also suggests the kinetics of protein production from mRNA transcripts largely governs protein abundance (<xref ref-type="bibr" rid="bib56">Schwanhausser et al., 2011</xref>). In a system undergoing less rapid apoptosis, such as those explored at the single cell level (<xref ref-type="bibr" rid="bib10">Cohen et al., 2008</xref>; <xref ref-type="bibr" rid="bib17">Geva-Zatorsky et al., 2010</xref>), there may be greater opportunity for proteins to accumulate and reveal relationships between ribosome footprint occupancy and protein production.</p><p>Previous work showed that caspase substrates were largely conserved at the identification level after different chemotherapeutic treatments, though different quantitative signatures of caspase proteolysis could be observed (<xref ref-type="bibr" rid="bib59">Shimbo et al., 2012</xref>). Here we find that in the setting of rapid apoptosis, proteins with changes at the transcriptional and translational level do not appear to be targeted more efficiently by proteases during apoptosis. Under conditions of more prolonged apoptosis, however, changes in the proteome may influence the appearance of proteolytic substrates and this remains to be investigated. Furthermore, our identification of non-caspase proteolytic events occurring with similar kinetics to caspase cleavage suggests other proteases are being activated during apoptosis. The biological role of these proteases during apoptosis awaits exploration.</p><p>Based on our systems-level analyses, we identified potential small molecules which may be effective in combination with bortezomib in myeloma therapy. Inhibition of HSF1 and PERK have also recently been proposed by others as promising therapeutic targets in myeloma (<xref ref-type="bibr" rid="bib22">Heimberger et al., 2012</xref>; <xref ref-type="bibr" rid="bib3">Atkins et al., 2013</xref>) and represents a proof of principle for our approach. Our deep sequencing and proteomic results suggest in particular that the combination of bortezomib with HSF1 inhibition could be beneficial in myeloma therapy.</p><p>We hypothesize that using a similar systems-based approach with less well-studied therapeutics will reveal novel means by which cells attempt to evade apoptosis, not just at the transcriptional level but also at the translational and post-translational levels. This analysis will likely complement information gained by genome-wide RNA knockdown approaches (<xref ref-type="bibr" rid="bib68">Zhu et al., 2011</xref>). By targeting these early response pathways identified at multiple systems levels cancer cell death can be maximized and resistance avoided. Furthermore, changes in specific proteins revealed by these analyses may be monitored after treatment to establish novel biomarkers of chemotherapeutic efficacy or drug resistance. We anticipate that integrated, systems-level approaches, such as that presented here, will help to form the future basis of evaluating preclinical chemotherapeutic effects.</p></sec><sec id="s4" sec-type="materials|methods"><title>Materials and methods</title><sec id="s4-1"><title>Cell culture and drug treatment</title><p>MM1.S cells (obtained from ATCC, Manassas, VA, USA) were grown in suspension to 1 × 10<sup>6</sup> cells/ml in RPMI-1640 media with 10% FBS. Bortezomib (LC Laboratories, Woburn, MA, USA) 20 µM stock solution in sterile-filtered phosphate buffered saline (PBS) was simultaneously added to a final concentration of 20 nM to flasks each containing 300 × 10<sup>6</sup> cells (PBS only added to control sample). At the indicated time point cells were separated into aliquots for each experimental approach (15 × 10<sup>6</sup> cells in duplicate for each of mRNAseq, iTRAQ proteomics, and ribosome footprinting; 200 × 10<sup>6</sup> cells for N-terminomics). Cells for ribosome footprinting alone were incubated at 37°C for 1 min with 100 µg/mL cycloheximide (Sigma-Aldrich, St. Louis, MO, USA) from 50 mg/ml stock in 100% EtOH. All cells were washed in PBS (PBS + 100 µg/ml cycloheximide for ribosome footprint samples), pelleted by centrifugation, and flash frozen in liquid N<sub>2</sub>, then stored at −80°C. Cell viability and caspase activity were assessed by Cell-Titer Glo and Caspase-Glo (Promega, Madison, WI, USA) assays per manufacturer protocol, respectively. KRIBB11 and GSK2606414 were purchased from EMD Millipore (Billerica, MA, USA) and 10 mM stock solutions prepared in DMSO and stored in single-use aliquots at −80°C. These drugs and bortezomib were added as indicated in triplicate in 24-well plates at an MM1.S cell density of 0.5 × 10<sup>6</sup> cells/ml.</p></sec><sec id="s4-2"><title>Measurement of total RNA and mRNA</title><p>Duplicate samples of 4 × 10<sup>6</sup> MM1.S cells at each time point after 20 nM bortezomib treatment were pelleted at 800×<italic>g</italic> for 5 min, supernatant aspirated, resuspended in ice-cold PBS, pelleted again, and supernatant aspirated. Total RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, CA, USA) following manufacturer’s protocol; lysis was performed by passing 20x through a 22½ gauge needle. Total RNA was resuspended in 100 µl RNAse-free water and further purified using RNAeasy kit (Qiagen, Germantown, MD) with on-column DNAse I treatment per manufacturer’s protocol. Purified RNA pellet was resuspended in 100 µl RNAse-free water. Total RNA concentration (as shown in <xref ref-type="fig" rid="fig1">Figure 1B</xref>) was measured spectrophotometrically using a NanoDrop ND-1000 UV-Vis spectrophotometer (Thermo Fisher, Waltham, MA, USA).</p><p>mRNA was further purified by poly(A) separation using Oligo (dT)<sub>25</sub> Magnetic Beads kit (New England BioLabs, Ipswich, MA, USA). Total RNA samples were diluted with 500 µl of kit Lysis/Binding buffer and bound to of equilibrated beads (100 µl bead slurry used). Beads were washed once in 500 µl Wash Buffer one followed by one wash in 500 µl Wash Buffer two. mRNA was eluted in 40 µl Elution Buffer and concentration measured by NanoDrop.</p></sec><sec id="s4-3"><title>Immunoblotting and protein measurements</title><p>A separate time course of MM1.S cells was prepared and treated with 20 nM bortezomib as described above 10 × 10<sup>6</sup> MM1.S cells harvested at indicated time points were lysed in RIPA buffer (50 mM Tris-HCl, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, pH 8.0) supplemented with 5 mM EDTA and 1x protease inhibitor cocktail (Cell Signaling Technology, Danvers, MA, USA). Protein concentration from cell lysates was measured by BCA assay (Thermo Fisher, Waltham, MA, USA). ∼25 µg of total protein at each time point was separated on NuPAGE 4–12% Bis-Tris polyacrylmide gels (Invitrogen) and transferred to 0.45 µM PVDF membrane. Primary antibodies used for immunoblotting were obtained from Cell Signaling Technology (α-CHOP, α-XIAP, α-Bid, α-PERK, α-4E-BP1, α-phospho(Thr37/46)-4E-BP1, α-eIF2α, α-phospho(Ser51)-eIF2α), Santa Cruz Biotechnology (Santa Cruz, CA, USA) (α-CREB2 [ATF4]), BioLegend (San Diego, CA, USA) (α-XBP1s), and Sigma (α-β-actin).</p></sec><sec id="s4-4"><title>Ribosome profiling and mRNAseq sample preparation and library generation</title><p>Harvested cell pellets for ribosome profiling were suspended and lysed in 500 µl ice-cold polysome lysis buffer (20 mM Tris, pH 7.4, 250 mM NaCl, 15 mM MgCl<sub>2</sub>, 1 mM dithiothreitol, 0.5% Triton X-100, 24 U/ml Turbo DNase (Ambion, Austin, TX, USA), and 100 µg/ml cycloheximide) by repeated pipetting. Lysate was clarified by centrifugation for 10 min at 20,000×<italic>g</italic> at 4°C. 3 µl of RNase I 100 U/μl (Ambion) was added and sample incubated for 45 min at room temperature. The digestion was stopped by the addition of 10 μl SuperaseIn 20 U/μl (Ambion). Digested samples was then loaded onto a 1 M sucrose cushion, prepared in polysome buffer plus 0.1 U/μl SuperaseIn. Ribosomes were pelleted by centrifugation for 4 hr at 70,000 rpm, 4°C in a TLA-110 rotor. The liquid was removed and the pellet was resuspended in 600 μl 10 mM Tris (pH 7), followed by the immediate addition of 40 μl 20% SDS. The sample was heated to 65°C and RNA was extracted using two rounds of acid phenol/chloroform followed by chloroform alone. RNA was precipitated from the aqueous phase by adding sodium acetate to a final concentration of 300 mM followed by at least one volume of isopropanol. Precipitation was carried out at −20°C overnight and RNA was then pelleted by centrifugation for 30 min at 20,000×<italic>g</italic>, 4°C. The supernatant was discarded, the pellet was air-dried, and the RNA was resuspended in 25 μl Tris (pH 7). RNA was separated by gel electrophoresis on a 15% TBE-Urea gel (Invitrogen) and gel fragments extracted corresponding to ∼25–35 nt in size. RNA was extracted from gel as in <xref ref-type="bibr" rid="bib25">Ingolia et al. (2011)</xref>.</p><p>Harvested cell pellets for mRNAseq were lysed by repeated pipetting in Trizol (Invitrogen) and total RNA isolated per manufacturer protocol. Poly(A) mRNA was purified from the total RNA sample using (dT)<sub>25</sub> DynaBeads (Invitrogen) per manufacturer protocol. mRNA was fragmented in high pH buffer (50 mM NaCO<sub>3</sub>, pH 9.2) for 20 min at 95°C, then precipitated and separated by gel electrophoresis as above. mRNA fragments of 50–90 nt were extracted. Footprint and mRNA sample purity and fragment size were verified by Bioanalyzer 2100 (Agilent, Santa Clara, CA, USA).</p><p>RNA samples were dephosphorylated, ligated to linker, and separated by gel electrophoresis as described previously (<xref ref-type="bibr" rid="bib25">Ingolia et al., 2011</xref>). Ribosome footprint samples were enriched by subtractive hybridization of contaminating rRNA sequences. We used biotin-modified oligonucletoides and capture on streptavidin-coated beads; oligonucleotide sequences complimentary to rRNA were identical to <xref ref-type="bibr" rid="bib62">Stern-Ginossar et al. (2012)</xref>. Reverse transcription and cDNA library preparation were completed as in <xref ref-type="bibr" rid="bib25">Ingolia et al. (2011)</xref>.</p></sec><sec id="s4-5"><title>Analysis of deep sequencing data</title><p>Sequencing was performed on an Illumina HiSeq 2000 using single end, 50-bp reads. Before alignment, linker and poly(A) sequences were computationally removed from the 3′ ends of raw sequencing reads. Bowtie v0.12.8 (<xref ref-type="bibr" rid="bib31">Langmead et al., 2009</xref>) (allowing up to two mismatches) was used to perform the alignments with up to two mismatches allowed. For footprint data only reads of length 26–36 nt (footprint length with cycloheximide (<xref ref-type="bibr" rid="bib25">Ingolia et al., 2011</xref>)) were used for alignment. Reads were first aligned vs human rRNA and tRNA sequences; aligned reads were discarded. All remaining reads were next aligned to known canonical transcripts in hg19 (downloaded from <ext-link ext-link-type="uri" xlink:href="http://genome.ucsc.edu/">genome.ucsc.edu</ext-link> 23 May 2012). Remaining unaligned reads were aligned to the hg19 genome. We used in-house C++ scripts to assign and count unique reads mapping to canonical hg19 transcripts. Only uniquely mapping reads were used for further analysis. The midpoint of footprint reads was used to assign a unique nucleotide location for that read. mRNA-seq reads were assigned a unique nucleotide position at the 5′ end of the read. mRNA and footprint read density were calculated in units of reads per kilobase million (RPKM) to normalize for gene length and total reads per sequencing run. Raw sequencing data is available in the GEO repository with accession number GSE48785.</p><p>Selected sequencing reads were visualized through Integrative Genomics Viewer 2.0 (<xref ref-type="bibr" rid="bib55">Robinson et al., 2011</xref>). Metagene analysis was performed as in <xref ref-type="bibr" rid="bib24">Ingolia et al. (2009)</xref>. 5′ UTR read assignments were based on canonical transcript splice isoform as in hg19. For UTR analysis normalized read counts (in RPKM) were used to compare relative 5′ UTR vs CDS translation across time points. For <xref ref-type="fig" rid="fig1">Figure 1C</xref>, unsupervised hierarchical clustering was performed using centroid linkage across mRNA, footprint, and translation efficiency data with Spearman-Rank correlation in Cluster 3.0 and visualized in TreeView.</p><p>All aligned transcripts were initially assigned to a unique UCSC gene ID. Conversion tables available from UCSC through the kgXref function were used to convert UCSC gene ID to HUGO gene nomenclature and to UniProt accession number for further analysis. Where indicated, gene lists were analyzed by Ingenuity Pathway Analysis (Ingenuity Systems, Redwood City, CA, USA) using default settings for included genes and interaction networks.</p></sec><sec id="s4-6"><title>iTRAQ proteomics—sample preparation</title><p>iTRAQ sample preparation was performed in a similar manner to <xref ref-type="bibr" rid="bib43">Mertins et al. (2012)</xref> with some modifications. 15 × 10<sup>6</sup> cells harvested at each time point (as above) were thawed and lysed by sonication in 250 µl lysis buffer (7M Guanidine HCl, 75 mM NaCl, 100 mM bicine pH 8.0) supplemented with protease inhibitors (500 µM 4-(2-Aminoethyl) benzenesulfonyl fluoride HCl, 1 mM E-64, 1 mM phenylmethylsulfonyl fluoride, and 1 mM EDTA). Lysates were cleared by centrifugation at 16,500×<italic>g</italic> for 10 min and protein concentration was measured by BCA assay. Lysate containing 250 µg of protein was diluted to 100 µl with lysis buffer. Disulfide bonds were reduced with 5 mM dithiothreitol and cysteines alkylated with 10 mM iodoacetimide. Lysate was diluted 1:8 with trypsin dilution buffer (100 mM bicine pH 8.0, 1 mM CaCl<sub>2</sub>, 75 mM NaCl). Sequencing grade modified trypsin (Promega) was added at an enzyme-to-substrate ratio of 1:25. Trypsin digestion took place for 16 hr with agitation at room temperature. Samples were acidified with trifluoroacetic acid to a final concentration of 0.5% to halt digestion. Tryptic peptides were desalted on Waters SepPak C18 (Milford, MA, USA) columns and evaporated to dryness on a vacuum concentrator.</p><p>Peptide samples were resuspended in 40 µl 0.5 M TEAB pH 8.5. Peptide concentration was determined by BCA assay. Labeling with 8-plex iTRAQ reagent was performed per manufacturer instructions (AB Sciex, Framingham, MA, USA): briefly, 100 µg of peptides at each time point was diluted to 30 µl with 0.5 M TEAB; 6-plex iTRAQ reagent was resuspended in 70 µl EtOH. Peptides were mixed with reagent and incubated for 1 hr at room temperature. The reaction was halted using 50 mM Tris/HCl pH 7.5. The labeled reaction products for each sample were combined and evaporated to dryness. Combined labeled peptides were desalted using Waters SepPak C18 cartridge. 0 hr samples were labeled with 113 mass tag; 1.5 hr, 115; 3 hr, 116; 6 hr, 117; 9 hr, 118; 12 hr, 119.</p></sec><sec id="s4-7"><title>iTRAQ proteomics–LC-MS/MS</title><p>Peptides were resuspended in 0.1% formic acid and then separated into 30 fractions via reverse phase high-pH fractionation using an XBridge C18 column (1.0 × 100 mm, 3.5 μm; Waters) on a Waters 2796 BioSeparations Module HPLC. We used a 70 min gradient with a linear increase from 2% to 38% acetonitrile in water and constant 10% ammonium bicarbonate, pH 10.5; flow rate was 50 µl/min. Fractions were evaporated to dryness and stored at −80°C, then resuspended in 0.1% formic acid prior to mass spectrometry analysis.</p><p>The combined iTRAQ sample was then analyzed using two separate mass spectrometers. The first was a QSTAR Elite QqTOF mass spectrometer (AB Sciex) coupled in-line to an Eksigent NLC-1DV-500 HPLC and in-house packed 75-µm × 15-cm C<sub>18</sub> column. For each fraction in-line LC was performed in Buffer A (0.1% formic acid in water) and Buffer B (0.1% formic acid in acetonitrile) at a flow rate of 250 nl/min. The full method extended over 125 min with a linear gradient from 3% to 40% Buffer B over 90 min, increased to 75% B over 5 min and held for 15 min, then re-equilibration at 3% B for 15 min.</p><p>Data-dependent acquisition was performed using Analyst v.2.0 software (AB Sciex) across an <italic>m</italic>/<italic>z</italic> range of 385–1200. We used an isolation window of 100 mDa for ions selected for MS/MS with a dynamic exclusion window of 60 s after acquisition. The two most intense ions in each MS1 scan were chosen for CID fragmentation and sequencing. The ‘iTRAQ’ option was chosen within the instrument software to enhance MS2 collision energy and increase intensity of iTRAQ reporter signal.</p><p>Each fraction of the combined iTRAQ sample was also analyzed on an LTQ Orbitrap Velos mass spectrometer (Thermo Fisher Scientific) coupled in-line to a nanoAcquity UPLC system (Waters). Injected samples were trapped on a Symmetry C18 Column (0.18 × 20 mm, 5 μm; Waters) for 5 min at 1% of Buffer B before starting the gradient; an analytical BEH130 C18 column (0.075 × 200 mm column, 1.7 μm; Waters) was used with a flow rate of 600 nl/min. A linear gradient to 38% Buffer B was run over 100 min, then increased to 75% B over 10 min, increased to 90% B over 5 min, then decreased to 2% B and re-equilibrated for 15 min (total method 130 min). Data-dependent acquisition was performed using the Xcalibur 2.1 software in positive ion mode at a spray voltage of 2.5 kV. MS1 survey spectra were acquired in the Orbitrap with a resolution of 60,000 and a mass range from 300 to 1400 <italic>m</italic>/<italic>z</italic>. All iTRAQ data was acquired by fragmenting the eight most intense ions per cycle in higher-energy collisional dissociation (HCD) mode. Collision energy was set to 45, maximum inject time was 250 ms and maximum ion count was 1 × 10<sup>5</sup> counts. We used an isolation window of 2.3 Th for ions selected for MS/MS. Ions selected for MS/MS were dynamically excluded for 60 s after acquisition.</p></sec><sec id="s4-8"><title>iTRAQ proteomics—peptide identification and protein quantification</title><p>Peptide identification was performed using Protein Prospector (v. 5.9) (University of California, San Francisco). All spectra were searched vs the full human SwissProt database (downloaded 21 March 2012) with reverse sequence database as implemented in Protein Prospector for decoy matches. Search parameters included: Fixed modifications carbamidomethyl (cysteine), iTRAQ-8plex (N-terminus), iTRAQ-8plex (lysine); Variable modifications methione-loss (N-terminus) and methionine oxidation; two missed tryptic cleavages allowed; for Orbitrap Velos data, parent and fragment mass tolerance of 20 ppm was used; for QSTAR Elite data, parent mass tolerance was 50 ppm and fragment mass tolerance was 150 ppm. Expectation value thresholds for peptide identification were modified to maintain a false discovery rate <1% based on the number of spectra matching to reverse decoy sequences.</p><p>iTRAQ signal was quantified using the Search Compare function of Protein Prospector, which extracts the peak intensity of the MS2 signal in the iTRAQ label range. Only peptides with an iTRAQ signal >300 cps at the 0 hr time point were included for further analysis to reduce noise in quantification. Only peptides uniquely matching to a single entry in the human SwissProt database were included. Proteins were only included with a minimum of two unique peptides matched. Protein quantification was performed by summation of iTRAQ signal across all peptides assigned to that protein at each time point. Total log<sub>2</sub> iTRAQ signal intensity across all peptides at a time point was median-normalized for comparison across time points.</p></sec><sec id="s4-9"><title>SRM assay for iTRAQ validation–method development</title><p>We attempted to develop orthogonal, targeted, label-free Selected reaction monitoring (SRM) assays for 250 proteins that were identified in iTRAQ experiments. We primarily concentrated on proteins with relative increases in mRNA expression as found in deep sequencing experiments. Based on prior yeast studies (<xref ref-type="bibr" rid="bib32">Lee et al., 2011</xref>), these would be expected to most likely result in protein changes after perturbation.</p><p>We used the open-source software Skyline (v1.3) (<xref ref-type="bibr" rid="bib38">MacLean et al., 2010</xref>) to first build an unscheduled SRM method. Building these methods requires a library of MS/MS spectra for inclusion of the most intense transition pairs (<italic>m</italic>/<italic>z</italic> data for parent ion and sequence fragment ion) for a given peptide. We used a spectral database of tryptic peptides matched to >8,000 proteins from human cancer cell lines, as identified by extensive fractionation and analysis in HCD mode on the LTQ Orbitrap Velos (AU, J Oses-Prieto, ALB, unpublished data). We selected a minimum of two and up to the seven most intense peptides present in the spectral library for each protein target. For each peptide we used the most intense seven transitions (<italic>y</italic>- and <italic>b</italic>-series ions) for initial measurement by unscheduled SRM. Selected peptides were chosen independently of those found by iTRAQ analysis.</p><p>All SRM analysis was carried out on an AB Sciex QTRAP 5500 triple quadrupole mass spectrometer interfaced in-line with a nanoAcquity UPLC system (Waters) identical to that on the LTQ Orbitrap Velos (Trapping column: Symmetry C18 Column (0.18 × 20 mm, 5 μm; Waters); Analytical column: BEH130 C18 column (0.075 × 200 mm column, 1.7 μm; Waters)). We injected ∼500 ng of tryptic peptides from MM1.S cells on to the mass spectrometer with the following conditions: Trapping for 4 min in 1% buffer B at 5 µl/min, then after injection a linear gradient from 3%–35% B over 80 min, an increase to 90% B over 5 min, then held for 5 min, then a decrease to 3% B for 10 min (total run time 100 min). Unit resolution was used at Q1 and Q3. A three second cycle time was used for all runs. For unscheduled runs a 10 ms acquisition time was used per transition. Multiple injections were used to test for all targeted peptides. Using data analysis in SkyLine software, peptides were selected for further method development based on (1) the signal detection (above baseline) of at least 5 of 7 co-eluting transitions and (2) a retention time within 7 min of that acquired in the initial spectral library (acquired under the same chromatographic conditions).</p><p>Peptides chosen for further development were then limited to the four most intense transitions as found in unscheduled runs. A scheduled SRM method was developed with a retention time window of ±5 min. The four most intense transitions were first subject to collision energy (CE) optimization in ±4 V steps to enhance intensity of detected transitions. CE-optimization data from QTRAP 5500 on the same samples was again analyzed in SkyLine. Peptides were confidently assigned based on the identification of four out of four targeted transitions co-eluting at the appropriate retention time. Ultimately, our final SRM method included from one to three confidently-assigned peptides matching to 152 proteins (all transitions listed in <xref ref-type="supplementary-material" rid="SD3-data">Figure 4—source data 1</xref>) with the CE for each transition used which maximized detected intensity.</p></sec><sec id="s4-10"><title>SRM assay for iTRAQ validation—sample analysis</title><p>We prepared an independent time course of MM1.S cells exposed to 20 nM bortezomib and harvested at 0 hr, 1.5 hr, 3 hr, 6 hr, 9 hr, and 12 hr. Cell viability and caspase activity assays were similar to that shown in <xref ref-type="fig" rid="fig1">Figure 1B</xref>. We prepared tryptic peptides from 15 × 10<sup>6</sup> cells at each time point (as described above prior to iTRAQ labeling). ∼500 ng of tryptic peptides were injected in triplicate and analyzed using the final scheduled SRM assay.</p><p>Peptide intensity in each sample was measured as the sum of all transition peak areas for that peptide (as measured by analysis in SkyLine). For peptides not clearly identified at all time points, a retention time window was used identical to that in other injections with background signal integrated and used for analysis. To normalize peptide concentration across samples, we used peptides derived from a set of high abundance proteins not expected to significantly change during the time course (actin, tubulin, and filamin-A). We derived an index based on the geometric mean intensity of peptides from these ‘housekeeping’ proteins and scaled SRM intensity of all peptides in each sample based on the median value of this index. Corrected peptide intensity was averaged across injections for each sample. Protein intensity was measured as the sum of peptide intensity for each sample.</p></sec><sec id="s4-11"><title>Estimation of absolute mRNA and protein copies per cell</title><p>For mRNA copy number estimation we used an approach similar to that used in <xref ref-type="bibr" rid="bib56">Schwannhausser et al. (2011)</xref>, described initially by <xref ref-type="bibr" rid="bib46">Mortazavi et al. (2008)</xref>. Total mRNA measured in duplicate by NanoDrop in the untreated MM1.S sample (0.88 ± 0.09 µg) was divided by 4 × 10<sup>6</sup> cells to estimate the mRNA yield per cell (2.19 × 10<sup>−13</sup> g). mRNA sequencing reads were approximately evenly distributed across A:C:G:T; the average molecular weight of a RNA monophosphate nucleotide (averaged across AMP, CMP, GMP, and UMP) is 339.5 g/mol. We can thus calculate the total number of mRNA nucleotides per cell (<italic>T</italic>; 6.40 × 10<sup>−15</sup> mol/cell). The copy numbers of individual mRNAs (<italic>c</italic>) can be calculated by the total number of sequencing reads which mapped uniquely to a given transcript (<italic>r</italic><sub><italic>transcript</italic></sub>), the total reads (<italic>r</italic><sub><italic>total</italic></sub>; 8,744,024 reads uniquely aligning to hg19 transcripts for the 0 hr MM1.S sample), and the transcript length (<italic>L</italic>) by the equation:<disp-formula id="equ1"><mml:math id="m1"><mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>r</mml:mi><mml:mi mathvariant="italic">transcript</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>r</mml:mi><mml:mi mathvariant="italic">total</mml:mi></mml:msub></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:mi>c</mml:mi><mml:mo>×</mml:mo><mml:mi>L</mml:mi></mml:mrow><mml:mi>T</mml:mi></mml:mfrac><mml:mo>.</mml:mo></mml:mrow></mml:math></disp-formula></p><p>For protein copy number estimation, we harvested and prepared tryptic peptides (as described above prior to iTRAQ labeling) from 15 × 10<sup>6</sup> untreated MM1.S cells. Peptides were separated into 20 fractions by reverse-phase high pH chromatography (same protocol as above for iTRAQ peptides). Each fraction was analyzed separately in HCD mode on the LTQ Orbitrap Velos Mass Spectrometer and nanoAcquity UPLC as above (gradient: 60 min linear gradient from 3% to 30% buffer B, 5 min gradient to 90% B, held 10 min, re-equilibrate 15 min at 3% B; mass spectrometer settings same as for iTRAQ analysis except collision energy at 35% and ion accumulation at 1 × 10<sup>6</sup>). We analyzed the MS data using the MaxQuant software package (v.1.3.0.5) (<xref ref-type="bibr" rid="bib11">Cox and Mann, 2008</xref>) with Carbamidomethyl cysteine as a fixed modification; Methionine oxidation and N-terminal acetylation as variable modifications; initial parent and fragment MS/MS tolerance of 20 ppm; a minimum peptide length of seven; one missed tryptic site; and razor peptides used for iBAQ quantification. Peptides were searched against the full Uniprot Human Proteome database (downloaded November 10, 2012) and contaminants enabled in MaxQuant; maximum allowable peptide and protein false discovery rate was 1% as searched against reverse sequence library based on Uniprot database. In MaxQuant peptide matches were assigned to protein groups (a cluster of a base protein plus additional proteins matching to a subset of the same peptides). Protein groups matching the reverse database or contaminants were discarded.</p><p>iBAQ quantification in MaxQuant evaluates the intensity of each protein in the sample as the sum of all the peptide intensities in the MS1 scan for peptides which matched to that protein group. iBAQ analysis and database searching resulted in assignment of intensity to 3390 protein groups with a minimum of one assigned peptide. We assigned each protein group to an individual protein based on the primary UniProt ID associated with the analysis output. We used the total iBAQ signal in the sample (<italic>I</italic><sub><italic>total</italic></sub>), across all matched peptides, as a measure of the total cellular protein signal as measured on the mass spectrometer. We divided the measurement of total protein in the sample (from BCA assay) by 10 × 10<sup>6</sup> cells to obtain an estimate of total protein per cell (<italic>g</italic><sub><italic>total</italic></sub>; 5.28 × 10<sup>−11</sup> g). We then calculated the grams of protein in the cell comprised of an individual protein (<italic>g</italic><sub><italic>protein</italic></sub>) based on the iBAQ signal assigned to that protein (<italic>I</italic><sub><italic>protein</italic></sub>) as a fraction of total iBAQ signal.<disp-formula id="equ2"><mml:math id="m2"><mml:mrow><mml:mfrac><mml:mrow><mml:msub><mml:mi>g</mml:mi><mml:mi mathvariant="italic">protein</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>g</mml:mi><mml:mi mathvariant="italic">total</mml:mi></mml:msub></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:mfrac><mml:mrow><mml:msub><mml:mi>I</mml:mi><mml:mi mathvariant="italic">protein</mml:mi></mml:msub></mml:mrow><mml:mrow><mml:msub><mml:mi>I</mml:mi><mml:mi mathvariant="italic">total</mml:mi></mml:msub></mml:mrow></mml:mfrac></mml:mrow></mml:math></disp-formula>Using the grams of an individual protein per cell, the molecular weight of the assigned protein in g/mol, and Avagadro’s number, we assign an estimated number of copies per protein per cell.</p><p>These measures of absolute copies of mRNA transcripts and proteins per cell are clearly rough approximations. In an important initial comparison, the median of the distributions for both mRNA and protein copy per cell are of the same order of magnitude as, though somewhat lower than, those found in <xref ref-type="bibr" rid="bib57">Schwanhausser et al. (2013)</xref>. This may reflect differences in the cell line (mouse NIH3T3 vs human MM1.S) as well as differences in measurement. However, the overall similarity of the distributions found here to those found by <xref ref-type="bibr" rid="bib57">Schwanhausser et al. (2013)</xref> indicates that our data are sufficient as a guide for our mass-action model (below). We further emphasize that our data are not intended as a definitive measure of mRNA or protein copy numbers in MM1.S cells. Even if there are systematic errors in the quantitation (i.e., if true protein copies per cell are all threefold higher), the fitting parameters for our mass-action model (see below) will be uniformly scaled but the overall interpretation of the model will be unchanged. This issue of scaling was illustrated in the corrigendum to the original manuscript by <xref ref-type="bibr" rid="bib57">Schwanhausser et al. (2013)</xref>. The original paper (<xref ref-type="bibr" rid="bib56">Schwanhausser et al., 2011</xref>) used the incorrect standard for protein abundance estimation, leading to threefold in changes in the protein distribution; the corrected scale led to an increase in the fit parameter <italic>k</italic><sub>sp</sub> but no changes to the model or conclusions of the paper.</p></sec><sec id="s4-12"><title>Quantitative modeling of protein translation</title><p>To model the conversion of mRNA to protein in this system we used a mass-action model of translation, as originally described in <xref ref-type="bibr" rid="bib21">Hargrove and Schmidt (1989)</xref>, and further used by <xref ref-type="bibr" rid="bib57">Schwanhausser et al. (2011)</xref>, as an ordinary differential equation:<disp-formula id="equ3"><mml:math id="m3"><mml:mrow><mml:mfrac><mml:mrow><mml:mi>d</mml:mi><mml:mi>P</mml:mi></mml:mrow><mml:mrow><mml:mi>d</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:msub><mml:mi>k</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mi>p</mml:mi></mml:mrow></mml:msub><mml:mi>R</mml:mi><mml:mrow><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>)</mml:mo></mml:mrow><mml:mo>−</mml:mo><mml:msub><mml:mi>k</mml:mi><mml:mrow><mml:mi>d</mml:mi><mml:mi>p</mml:mi></mml:mrow></mml:msub><mml:mi>P</mml:mi><mml:mrow><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:mrow></mml:math></disp-formula>Here the number of protein copies (P(<italic>t</italic>)) changes over time as a function of the number of mRNA transcripts (R(<italic>t</italic>)), the translation rate constant <italic>k</italic><sub>sp</sub>, and the protein degradation rate constant <italic>k</italic><sub>dp</sub>.</p><p>In our system, we first modeled R(<italic>t</italic>) using the normalized sequencing reads (RPKM) at later time points and taking the ratio to the RPKM at 0 hr. We then multiplied this RPKM ratio at each time point by the absolute number of mRNA transcripts at 0 hr as measured above. We focused on analyzing only a subset of 13 transcripts with increased relative mRNA abundance but with variable detected changes at the protein level (i.e., some with increased protein, some with no protein change). We plot this data in <xref ref-type="fig" rid="fig5">Figure 5D</xref>, <xref ref-type="fig" rid="fig5s4">Figure 5—figure supplement 4A,C</xref>. We found that for this subset of transcripts, this mRNA transcript data could be well-approximated by fitting either a quadratic (FTL, SQSTM1, CTSD) or 4-parameter sigmoid (all others). This fit function (fitting performed in GraphPad Prism software) describes R(<italic>t</italic>).</p><p>However, as discussed in the main text, this function R(<italic>t</italic>) is based on <italic>relative</italic> transcript expression at different time points. This would be an effective measure of mRNA transcripts per cell in the setting of constant total mRNA concentration. In this system, though, we find that mRNA is significantly degraded at later time points (<xref ref-type="fig" rid="fig1">Figure 1B</xref>). Therefore, we fit an additional function G(<italic>t</italic>) to this mRNA data as a four-parameter sigmoid fit to describe this global mRNA degradation. By incorporating G(<italic>t</italic>) into the mass-action model above, assuming similar degradation rates of each transcript during apoptosis (as described by <xref ref-type="bibr" rid="bib13">Del Prete et al. (2002)</xref>), the combined term R(<italic>t</italic>)G(<italic>t</italic>) now represents the <italic>absolute</italic> number of mRNA transcripts per cell in our system:<disp-formula id="equ4"><mml:math id="m4"><mml:mrow><mml:mfrac><mml:mrow><mml:mi>d</mml:mi><mml:mi>P</mml:mi></mml:mrow><mml:mrow><mml:mi>d</mml:mi><mml:mi>t</mml:mi></mml:mrow></mml:mfrac><mml:mo>=</mml:mo><mml:msub><mml:mi>k</mml:mi><mml:mrow><mml:mi>s</mml:mi><mml:mi>p</mml:mi></mml:mrow></mml:msub><mml:mi>R</mml:mi><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mi>G</mml:mi><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mo>−</mml:mo><mml:msub><mml:mi>k</mml:mi><mml:mrow><mml:mi>d</mml:mi><mml:mi>p</mml:mi></mml:mrow></mml:msub><mml:mi>P</mml:mi><mml:mrow><mml:mrow><mml:mo>(</mml:mo><mml:mi>t</mml:mi><mml:mo>)</mml:mo></mml:mrow></mml:mrow><mml:mo>.</mml:mo></mml:mrow></mml:math></disp-formula></p><p>To estimate absolute protein copies per cell at later time points, we used the label-free SRM intensity data (as this does not suffer from compression ratio artifacts as seen in iTRAQ data). We took the ratio of SRM intensity at later time points to that at 0 hr, then multiplied by the absolute copies protein copies per untreated MM1.S cell as measured by iBAQ. We scaled the 12 hr SRM data to reflect the decreased protein concentration in the sample.</p><p>We used Mathematica (v. 9) software to numerically solve the differential equation above for each of the 13 genes chosen for analysis. We fit this differential equation to the plotted protein data (as shown in <xref ref-type="fig" rid="fig5">Figure 5E</xref>, <xref ref-type="fig" rid="fig5s4">Figure 5—figure supplement 4B,D</xref>). This equation has three free fitting parameters to describe the six protein data points: protein copies at 0 hr, <italic>k</italic><sub>sp</sub>, and <italic>k</italic><sub>dp</sub>. Protein copies at 0 hr was varied a maximum of 10% from the measured protein copies per cell by iBAQ. <italic>k</italic><sub>dp</sub> was assumed to be relatively constant for all proteins in the background of proteasomal blockade by bortezomib. We only varied the range between 0.010–0.015 hr<sup>−1</sup>, similar to the corresponding average protein half-life found in <xref ref-type="bibr" rid="bib56">Schwanhausser et al. (2011)</xref>. Therefore, we primarily varied the translation rate constant <italic>k</italic><sub>sp</sub> in order to fit this model to the protein data. We performed the fits by manually minimizing the least-squares difference between model fit and protein data in Mathematica. We found that the model could indeed well-describe the protein data; in addition, the fit <italic>k</italic><sub>sp</sub> parameter ranged from 10 to 270 proteins/transcript/hr, consistent with the range in <xref ref-type="bibr" rid="bib56">Schwanhausser et al. (2011)</xref>.</p></sec><sec id="s4-13"><title>SRM assay for degradomics</title><p>As described in <xref ref-type="bibr" rid="bib59">Shimbo et al. (2012)</xref>, cell pellets were lysed by sonication in 4.0% SDS, 400 mM Bicine (pH8.0), and supplemented with protease inhibitors 0.1 mM z-VAD-fmk, 0.1 mM E-64, 1 mM AEBSF, 1 mM PMSF, and 5 mM EDTA. Spike-in internal standards were added to each sample after lysis: 50 µg each of bovine catalase and yeast alcohol dehydrogenase (Sigma). Proteins were reduced by TCEP, cysteines alklyated by iodoacetamide, and free N-termini biotinylated via the reaction of TEVest4 ester and subtiligase enzyme as previously described (<xref ref-type="bibr" rid="bib59">Shimbo et al., 2012</xref>). Biotinylated peptides were precipitated, resuspended in 5.3 M Guanidine HCl and captured on NeutraAvidin agarose beads (Thermo). The beads were extensively washed in 5 M Guanidine HCl followed by on-bead trypsin digestion overnight at room temperature with agitation. As previously described, captured peptides were released from beads by incubation with TEV protease (<xref ref-type="bibr" rid="bib59">Shimbo et al., 2012</xref>). Peptides were desalted using C<sub>18</sub> ZipTips (Millipore), evaporated to dryness, and stored at −80°C for analysis.</p><p>SRM assays were developed in a similar manner as described above, again using SkyLine software. In this case, however, targeted parent/fragment ion transitions were extracted from MS/MS spectra in the DegraBase. This publicly available database lists N-terminally labeled peptides found in human cell culture under both apoptotic and non-apoptotic conditions (<ext-link ext-link-type="uri" xlink:href="http://wellslab.ucsf.edu/degrabase">wellslab.ucsf.edu/degrabase</ext-link>). <italic>b</italic>-ion fragment <italic>m</italic>/<italic>z</italic> was modified as necessary to reflect the Abu- tag at the N-terminus of the peptide (<xref ref-type="bibr" rid="bib59">Shimbo et al., 2012</xref>). We initially chose 400 peptides for development of targeted SRM assays in SkyLine. We used the same chromatography and instrument settings as described for the earlier SRM assays. We initially examined seven transitions per peptide; we moved forward with peptides with a minimum of five co-eluting transitions and a retention time within 7 min of that predicted based on the peptide retention time prediction algorithm (SSRCalc 3.0) implemented in SkyLine. Ultimately, we included 252 proteolytic peptides in the final SRM assay.</p><p>We corrected samples for labeling efficiency and overall peptide concentration using five proteolytic peptides derived from the spike-in standards. Corrected peptide intensities were averaged across duplicate injections at each time point in the 20 nM bortezomib time course. These mean intensity values were used for analysis.</p></sec></sec></body><back><sec sec-type="additional-information"><title>Additional information</title><fn-group content-type="competing-interest"><title>Competing interests</title><fn fn-type="conflict" id="conf1"><p>The authors declare that no competing interests exist.</p></fn></fn-group><fn-group content-type="author-contribution"><title>Author contributions</title><fn fn-type="con" id="con1"><p>APW, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article</p></fn><fn fn-type="con" id="con2"><p>EZ, Acquisition of data, Analysis and interpretation of data</p></fn><fn fn-type="con" id="con3"><p>PJW, Analysis and interpretation of data, Drafting or revising the article</p></fn><fn fn-type="con" id="con4"><p>OJ, Analysis and interpretation of data, Drafting or revising the article</p></fn><fn fn-type="con" id="con5"><p>AU, Conception and design, Contributed unpublished essential data or reagents</p></fn><fn fn-type="con" id="con6"><p>ALB, Drafting or revising the article, Contributed unpublished essential data or reagents</p></fn><fn fn-type="con" id="con7"><p>JSW, Conception and design, Drafting or revising the article</p></fn><fn fn-type="con" id="con8"><p>JAW, Conception and design, Drafting or revising the article</p></fn></fn-group></sec><sec sec-type="supplementary-material"><title>Additional files</title><sec sec-type="datasets"><title>Major dataset</title><p>The following dataset was generated:</p><p><related-object content-type="generated-dataset" document-id="Dataset ID and/or url" document-id-type="dataset" document-type="data" id="dataro1"><name><surname>Wiita</surname><given-names>AP</given-names></name>, <name><surname>Ziv</surname><given-names>E</given-names></name>, <name><surname>Wiita</surname><given-names>PJ</given-names></name>, <name><surname>Urisman</surname><given-names>A</given-names></name>, <name><surname>Julien</surname><given-names>O</given-names></name>, <name><surname>Burlingame</surname><given-names>AL</given-names></name>, <etal/>, <year>2013</year><x>, </x><source>Global response to chemotherapy-induced apoptosis</source><x>, </x><object-id pub-id-type="art-access-id">GSE48785</object-id><x>; </x><ext-link ext-link-type="uri" xlink:href="http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE48785">http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE48785</ext-link><x>, </x><comment>Publicly available at GEO (<ext-link ext-link-type="uri" xlink:href="http://www.ncbi.nlm.nih.gov/geo/">http://www.ncbi.nlm.nih.gov/geo/</ext-link>).</comment></related-object></p></sec></sec><ref-list><title>References</title><ref id="bib1a"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Acosta-Alvear</surname><given-names>D</given-names></name><name><surname>Zhou</surname><given-names>Y</given-names></name><name><surname>Blais</surname><given-names>A</given-names></name><name><surname>Tsikitis</surname><given-names>M</given-names></name><name><surname>Lents</surname><given-names>NH</given-names></name><name><surname>Arias</surname><given-names>C</given-names></name><etal/></person-group><year>2007</year><article-title>XBP1 controls diverse cell type- and condition-specific transcriptional regulatory networks</article-title><source>Mol Cell</source><volume>27</volume><fpage>53</fpage><lpage>66</lpage><pub-id pub-id-type="doi">10.1016/j.molcel.2007.06.011</pub-id></element-citation></ref><ref id="bib1"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Agard</surname><given-names>NJ</given-names></name><name><surname>Mahrus</surname><given-names>S</given-names></name><name><surname>Trinidad</surname><given-names>JC</given-names></name><name><surname>Lynn</surname><given-names>A</given-names></name><name><surname>Burlingame</surname><given-names>AL</given-names></name><name><surname>Wells</surname><given-names>JA</given-names></name></person-group><year>2012</year><article-title>Global kinetic analysis of proteolysis via quantitative targeted proteomics</article-title><source>Proc Natl Acad Sci USA</source><volume>109</volume><fpage>1913</fpage><lpage>8</lpage><pub-id pub-id-type="doi">10.1073/pnas.1117158109</pub-id></element-citation></ref><ref id="bib2"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Aronson</surname><given-names>LI</given-names></name><name><surname>Davies</surname><given-names>FE</given-names></name></person-group><year>2012</year><article-title>DangER: protein ovERload. Targeting protein degradation to treat myeloma</article-title><source>Haematologica</source><volume>97</volume><fpage>1119</fpage><lpage>30</lpage><pub-id pub-id-type="doi">10.3324/haematol.2012.064923</pub-id></element-citation></ref><ref id="bib3"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Atkins</surname><given-names>C</given-names></name><name><surname>Liu</surname><given-names>Q</given-names></name><name><surname>Minthorn</surname><given-names>E</given-names></name><name><surname>Zhang</surname><given-names>SY</given-names></name><name><surname>Figueroa</surname><given-names>DJ</given-names></name><name><surname>Moss</surname><given-names>K</given-names></name><etal/></person-group><year>2013</year><article-title>Characterization of a novel PERK kinase inhibitor with antitumor and antiangiogenic activity</article-title><source>Cancer Res</source><volume>73</volume><fpage>1993</fpage><lpage>2002</lpage><pub-id pub-id-type="doi">10.1158/0008-5472.CAN-12-3109</pub-id></element-citation></ref><ref id="bib4"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Axten</surname><given-names>JM</given-names></name><name><surname>Medina</surname><given-names>JR</given-names></name><name><surname>Feng</surname><given-names>Y</given-names></name><name><surname>Shu</surname><given-names>A</given-names></name><name><surname>Romeril</surname><given-names>SP</given-names></name><name><surname>Grant</surname><given-names>SW</given-names></name><etal/></person-group><year>2012</year><article-title>Discovery of 7-methyl-5-(1-{[3-(trifluoromethyl)phenyl]acetyl}-2,3-dihydro-1H-indol-5-yl)-7H-p yrrolo[2,3-d]pyrimidin-4-amine (GSK2606414), a potent and selective first-in-class inhibitor of protein kinase R (PKR)-like endoplasmic reticulum kinase (PERK)</article-title><source>J Med Chem</source><volume>55</volume><fpage>7193</fpage><lpage>207</lpage><pub-id pub-id-type="doi">10.1021/jm300713s</pub-id></element-citation></ref><ref id="bib5"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Becker-Pauly</surname><given-names>C</given-names></name><name><surname>Barre</surname><given-names>O</given-names></name><name><surname>Schilling</surname><given-names>O</given-names></name><name><surname>Auf dem Keller</surname><given-names>U</given-names></name><name><surname>Ohler</surname><given-names>A</given-names></name><name><surname>Broder</surname><given-names>C</given-names></name><etal/></person-group><year>2011</year><article-title>Proteomic analyses reveal an acidic prime side specificity for the astacin metalloprotease family reflected by physiological substrates</article-title><source>Mol Cell Proteomics</source><volume>10</volume><fpage>M111</fpage><pub-id pub-id-type="doi">10.1074/mcp.M111.009233</pub-id></element-citation></ref><ref id="bib6"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Berkers</surname><given-names>CR</given-names></name><name><surname>Verdoes</surname><given-names>M</given-names></name><name><surname>Lichtman</surname><given-names>E</given-names></name><name><surname>Fiebiger</surname><given-names>E</given-names></name><name><surname>Kessler</surname><given-names>BM</given-names></name><name><surname>Anderson</surname><given-names>KC</given-names></name><etal/></person-group><year>2005</year><article-title>Activity probe for in vivo profiling of the specificity of proteasome inhibitor bortezomib</article-title><source>Nat Methods</source><volume>2</volume><fpage>357</fpage><lpage>62</lpage><pub-id pub-id-type="doi">10.1038/nmeth759</pub-id></element-citation></ref><ref id="bib7"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Bushell</surname><given-names>M</given-names></name><name><surname>Stoneley</surname><given-names>M</given-names></name><name><surname>Kong</surname><given-names>YW</given-names></name><name><surname>Hamilton</surname><given-names>TL</given-names></name><name><surname>Spriggs</surname><given-names>KA</given-names></name><name><surname>Dobbyn</surname><given-names>HC</given-names></name><etal/></person-group><year>2006</year><article-title>Polypyrimidine tract binding protein regulates IRES-mediated gene expression during apoptosis</article-title><source>Mol Cell</source><volume>23</volume><fpage>401</fpage><lpage>12</lpage><pub-id pub-id-type="doi">10.1016/j.molcel.2006.06.012</pub-id></element-citation></ref><ref id="bib8"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Bushell</surname><given-names>M</given-names></name><name><surname>Stoneley</surname><given-names>M</given-names></name><name><surname>Sarnow</surname><given-names>P</given-names></name><name><surname>Willis</surname><given-names>AE</given-names></name></person-group><year>2004</year><article-title>Translation inhibition during the induction of apoptosis: RNA or protein degradation?</article-title><source>Biochem Soc Trans</source><volume>32</volume><fpage>606</fpage><lpage>10</lpage><pub-id pub-id-type="doi">10.1042/BST0320606</pub-id></element-citation></ref><ref id="bib9"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Campanella</surname><given-names>A</given-names></name><name><surname>Santambrogio</surname><given-names>P</given-names></name><name><surname>Fontana</surname><given-names>F</given-names></name><name><surname>Frenquelli</surname><given-names>M</given-names></name><name><surname>Cenci</surname><given-names>S</given-names></name><name><surname>Marcatti</surname><given-names>M</given-names></name><etal/></person-group><year>2012</year><article-title>Iron increases the susceptibility of multiple myeloma cells to bortezomib</article-title><source>Haematologica</source><volume>98</volume><fpage>971</fpage><lpage>9</lpage><pub-id pub-id-type="doi">10.3324/haematol.2012.074872</pub-id></element-citation></ref><ref id="bib10"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Cohen</surname><given-names>AA</given-names></name><name><surname>Geva-Zatorsky</surname><given-names>N</given-names></name><name><surname>Eden</surname><given-names>E</given-names></name><name><surname>Frenkel-Morgenstern</surname><given-names>M</given-names></name><name><surname>Issaeva</surname><given-names>I</given-names></name><name><surname>Sigal</surname><given-names>A</given-names></name><etal/></person-group><year>2008</year><article-title>Dynamic proteomics of individual cancer cells in response to a drug</article-title><source>Science</source><volume>322</volume><fpage>1511</fpage><lpage>6</lpage><pub-id pub-id-type="doi">10.1126/science.1160165</pub-id></element-citation></ref><ref id="bib11"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Cox</surname><given-names>J</given-names></name><name><surname>Mann</surname><given-names>M</given-names></name></person-group><year>2008</year><article-title>MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification</article-title><source>Nat Biotechnol</source><volume>26</volume><fpage>1367</fpage><lpage>72</lpage><pub-id pub-id-type="doi">10.1038/nbt.1511</pub-id></element-citation></ref><ref id="bib12"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Crawford</surname><given-names>ED</given-names></name><name><surname>Seaman</surname><given-names>JE</given-names></name><name><surname>Agard</surname><given-names>N</given-names></name><name><surname>Hsu</surname><given-names>GW</given-names></name><name><surname>Julien</surname><given-names>O</given-names></name><name><surname>Mahrus</surname><given-names>S</given-names></name><etal/></person-group><year>2013</year><article-title>The DegraBase: a database of proteolysis in healthy and apoptotic human cells</article-title><source>Mol Cell Proteomics</source><volume>12</volume><fpage>813</fpage><lpage>24</lpage><pub-id pub-id-type="doi">10.1074/mcp.O112.024372</pub-id></element-citation></ref><ref id="bib13"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Del Prete</surname><given-names>MJ</given-names></name><name><surname>Robles</surname><given-names>MS</given-names></name><name><surname>Guao</surname><given-names>A</given-names></name><name><surname>Martinez</surname><given-names>AC</given-names></name><name><surname>Izquierdo</surname><given-names>M</given-names></name><name><surname>Garcia-Sanz</surname><given-names>JA</given-names></name></person-group><year>2002</year><article-title>Degradation of cellular mRNA is a general early apoptosis-induced event</article-title><source>FASEB J</source><volume>16</volume><fpage>2003</fpage><lpage>5</lpage><pub-id pub-id-type="doi">10.1096/fj.02-0392fje</pub-id></element-citation></ref><ref id="bib14"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Dix</surname><given-names>MM</given-names></name><name><surname>Simon</surname><given-names>GM</given-names></name><name><surname>Cravatt</surname><given-names>BF</given-names></name></person-group><year>2008</year><article-title>Global mapping of the topography and magnitude of proteolytic events in apoptosis</article-title><source>Cell</source><volume>134</volume><fpage>679</fpage><lpage>91</lpage><pub-id pub-id-type="doi">10.1016/j.cell.2008.06.038</pub-id></element-citation></ref><ref id="bib15"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Gausdal</surname><given-names>G</given-names></name><name><surname>Gjertsen</surname><given-names>BT</given-names></name><name><surname>McCormack</surname><given-names>E</given-names></name><name><surname>Van Damme</surname><given-names>P</given-names></name><name><surname>Hovland</surname><given-names>R</given-names></name><name><surname>Krakstad</surname><given-names>C</given-names></name><etal/></person-group><year>2008</year><article-title>Abolition of stress-induced protein synthesis sensitizes leukemia cells to anthracycline-induced death</article-title><source>Blood</source><volume>111</volume><fpage>2866</fpage><lpage>77</lpage><pub-id pub-id-type="doi">10.1182/blood-2007-07-103242</pub-id></element-citation></ref><ref id="bib16"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Gerashchenko</surname><given-names>MV</given-names></name><name><surname>Lobanov</surname><given-names>AV</given-names></name><name><surname>Gladyshev</surname><given-names>VN</given-names></name></person-group><year>2012</year><article-title>Genome-wide ribosome profiling reveals complex translational regulation in response to oxidative stress</article-title><source>Proc Natl Acad Sci USA</source><volume>109</volume><fpage>17394</fpage><lpage>9</lpage><pub-id pub-id-type="doi">10.1073/pnas.1120799109</pub-id></element-citation></ref><ref id="bib17"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Geva-Zatorsky</surname><given-names>N</given-names></name><name><surname>Dekel</surname><given-names>E</given-names></name><name><surname>Cohen</surname><given-names>AA</given-names></name><name><surname>Danon</surname><given-names>T</given-names></name><name><surname>Cohen</surname><given-names>L</given-names></name><name><surname>Alon</surname><given-names>U</given-names></name></person-group><year>2010</year><article-title>Protein dynamics in drug combinations: a linear superposition of individual-drug responses</article-title><source>Cell</source><volume>140</volume><fpage>643</fpage><lpage>51</lpage><pub-id pub-id-type="doi">10.1016/j.cell.2010.02.011</pub-id></element-citation></ref><ref id="bib18"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Gonzalvez</surname><given-names>F</given-names></name><name><surname>Ashkenazi</surname><given-names>A</given-names></name></person-group><year>2010</year><article-title>New insights into apoptosis signaling by Apo2L/TRAIL</article-title><source>Oncogene</source><volume>29</volume><fpage>4752</fpage><lpage>65</lpage><pub-id pub-id-type="doi">10.1038/onc.2010.221</pub-id></element-citation></ref><ref id="bib19"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Gray</surname><given-names>DC</given-names></name><name><surname>Mahrus</surname><given-names>S</given-names></name><name><surname>Wells</surname><given-names>JA</given-names></name></person-group><year>2010</year><article-title>Activation of specific apoptotic caspases with an engineered small-molecule-activated protease</article-title><source>Cell</source><volume>142</volume><fpage>637</fpage><lpage>46</lpage><pub-id pub-id-type="doi">10.1016/j.cell.2010.07.014</pub-id></element-citation></ref><ref id="bib20"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Harding</surname><given-names>HP</given-names></name><name><surname>Zhang</surname><given-names>Y</given-names></name><name><surname>Scheuner</surname><given-names>D</given-names></name><name><surname>Chen</surname><given-names>JJ</given-names></name><name><surname>Kaufman</surname><given-names>RJ</given-names></name><name><surname>Ron</surname><given-names>D</given-names></name></person-group><year>2009</year><article-title>Ppp1r15 gene knockout reveals an essential role for translation initiation factor 2 alpha (eIF2alpha) dephosphorylation in mammalian development</article-title><source>Proc Natl Acad Sci USA</source><volume>106</volume><fpage>1832</fpage><lpage>7</lpage><pub-id pub-id-type="doi">10.1073/pnas.0809632106</pub-id></element-citation></ref><ref id="bib21"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Hargrove</surname><given-names>JL</given-names></name><name><surname>Schmidt</surname><given-names>FH</given-names></name></person-group><year>1989</year><article-title>The role of mRNA and protein stability in gene expression</article-title><source>FASEB J</source><volume>3</volume><fpage>2360</fpage><lpage>70</lpage></element-citation></ref><ref id="bib22"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Heimberger</surname><given-names>T</given-names></name><name><surname>Andrulis</surname><given-names>M</given-names></name><name><surname>Riedel</surname><given-names>S</given-names></name><name><surname>Stuhmer</surname><given-names>T</given-names></name><name><surname>Schraud</surname><given-names>H</given-names></name><name><surname>Beilhack</surname><given-names>A</given-names></name><etal/></person-group><year>2012</year><article-title>The heat shock transcription factor 1 as a potential new therapeutic target in multiple myeloma</article-title><source>Br J Haematol</source><volume>160</volume><fpage>465</fpage><lpage>76</lpage><pub-id pub-id-type="doi">10.1111/bjh.12164</pub-id></element-citation></ref><ref id="bib23"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Impens</surname><given-names>F</given-names></name><name><surname>Van Damme</surname><given-names>P</given-names></name><name><surname>Demol</surname><given-names>H</given-names></name><name><surname>Van Damme</surname><given-names>J</given-names></name><name><surname>Vandekerckhove</surname><given-names>J</given-names></name><name><surname>Gevaert</surname><given-names>K</given-names></name></person-group><year>2008</year><article-title>Mechanistic insight into taxol-induced cell death</article-title><source>Oncogene</source><volume>27</volume><fpage>4580</fpage><lpage>91</lpage><pub-id pub-id-type="doi">10.1038/onc.2008.96</pub-id></element-citation></ref><ref id="bib24"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Ingolia</surname><given-names>NT</given-names></name><name><surname>Ghaemmaghami</surname><given-names>S</given-names></name><name><surname>Newman</surname><given-names>JR</given-names></name><name><surname>Weissman</surname><given-names>JS</given-names></name></person-group><year>2009</year><article-title>Genome-wide analysis in vivo of translation with nucleotide resolution using ribosome profiling</article-title><source>Science</source><volume>324</volume><fpage>218</fpage><lpage>23</lpage><pub-id pub-id-type="doi">10.1126/science.1168978</pub-id></element-citation></ref><ref id="bib25"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Ingolia</surname><given-names>NT</given-names></name><name><surname>Lareau</surname><given-names>LF</given-names></name><name><surname>Weissman</surname><given-names>JS</given-names></name></person-group><year>2011</year><article-title>Ribosome profiling of mouse embryonic stem cells reveals the complexity and dynamics of mammalian proteomes</article-title><source>Cell</source><volume>147</volume><fpage>789</fpage><lpage>802</lpage><pub-id pub-id-type="doi">10.1016/j.cell.2011.10.002</pub-id></element-citation></ref><ref id="bib26"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Jousse</surname><given-names>C</given-names></name><name><surname>Bruhat</surname><given-names>A</given-names></name><name><surname>Carraro</surname><given-names>V</given-names></name><name><surname>Urano</surname><given-names>F</given-names></name><name><surname>Ferrara</surname><given-names>M</given-names></name><name><surname>Ron</surname><given-names>D</given-names></name><etal/></person-group><year>2001</year><article-title>Inhibition of CHOP translation by a peptide encoded by an open reading frame localized in the chop 5’UTR</article-title><source>Nucleic Acids Res</source><volume>29</volume><fpage>4341</fpage><lpage>51</lpage><pub-id pub-id-type="doi">10.1093/nar/29.21.4341</pub-id></element-citation></ref><ref id="bib27"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Kaufmann</surname><given-names>SH</given-names></name><name><surname>Earnshaw</surname><given-names>WC</given-names></name></person-group><year>2000</year><article-title>Induction of apoptosis by cancer chemotherapy</article-title><source>Exp Cell Res</source><volume>256</volume><fpage>42</fpage><lpage>9</lpage><pub-id pub-id-type="doi">10.1006/excr.2000.4838</pub-id></element-citation></ref><ref id="bib28"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Krippner-Heidenreich</surname><given-names>A</given-names></name><name><surname>Talanian</surname><given-names>RV</given-names></name><name><surname>Sekul</surname><given-names>R</given-names></name><name><surname>Kraft</surname><given-names>R</given-names></name><name><surname>Thole</surname><given-names>H</given-names></name><name><surname>Ottleben</surname><given-names>H</given-names></name><etal/></person-group><year>2001</year><article-title>Targeting of the transcription factor max during apoptosis: phosphorylation-regulated cleavage by caspase-5 at an unusual glutamic acid residue in position P1</article-title><source>Biochem J</source><volume>358</volume><fpage>705</fpage><lpage>15</lpage><pub-id pub-id-type="doi">10.1042/0264-6021:3580705</pub-id></element-citation></ref><ref id="bib29"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lackner</surname><given-names>DH</given-names></name><name><surname>Schmidt</surname><given-names>MW</given-names></name><name><surname>Wu</surname><given-names>S</given-names></name><name><surname>Wolf</surname><given-names>DA</given-names></name><name><surname>Bahler</surname><given-names>J</given-names></name></person-group><year>2012</year><article-title>Regulation of transcriptome, translation, and proteome in response to environmental stress in fission yeast</article-title><source>Genome Biol</source><volume>13</volume><fpage>R25</fpage><pub-id pub-id-type="doi">10.1186/gb-2012-13-4-r25</pub-id></element-citation></ref><ref id="bib30"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lamb</surname><given-names>J</given-names></name><name><surname>Crawford</surname><given-names>ED</given-names></name><name><surname>Peck</surname><given-names>D</given-names></name><name><surname>Modell</surname><given-names>JW</given-names></name><name><surname>Blat</surname><given-names>IC</given-names></name><name><surname>Wrobel</surname><given-names>MJ</given-names></name><etal/></person-group><year>2006</year><article-title>The connectivity map: using gene-expression signatures to connect small molecules, genes, and disease</article-title><source>Science</source><volume>313</volume><fpage>1929</fpage><lpage>35</lpage><pub-id pub-id-type="doi">10.1126/science.1132939</pub-id></element-citation></ref><ref id="bib31"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Langmead</surname><given-names>B</given-names></name><name><surname>Trapnell</surname><given-names>C</given-names></name><name><surname>Pop</surname><given-names>M</given-names></name><name><surname>Salzberg</surname><given-names>SL</given-names></name></person-group><year>2009</year><article-title>Ultrafast and memory-efficient alignment of short DNA sequences to the human genome</article-title><source>Genome Biol</source><volume>10</volume><fpage>R25</fpage><pub-id pub-id-type="doi">10.1186/gb-2009-10-3-r25</pub-id></element-citation></ref><ref id="bib32"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lee</surname><given-names>MV</given-names></name><name><surname>Topper</surname><given-names>SE</given-names></name><name><surname>Hubler</surname><given-names>SL</given-names></name><name><surname>Hose</surname><given-names>J</given-names></name><name><surname>Wenger</surname><given-names>CD</given-names></name><name><surname>Coon</surname><given-names>JJ</given-names></name><etal/></person-group><year>2011</year><article-title>A dynamic model of proteome changes reveals new roles for transcript alteration in yeast</article-title><source>Mol Syst Biol</source><volume>7</volume><fpage>514</fpage><pub-id pub-id-type="doi">10.1038/msb.2011.48</pub-id></element-citation></ref><ref id="bib33"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lee</surname><given-names>S</given-names></name><name><surname>Liu</surname><given-names>B</given-names></name><name><surname>Huang</surname><given-names>SX</given-names></name><name><surname>Shen</surname><given-names>B</given-names></name><name><surname>Qian</surname><given-names>SB</given-names></name></person-group><year>2012</year><article-title>Global mapping of translation initiation sites in mammalian cells at single-nucleotide resolution</article-title><source>Proc Natl Acad Sci USA</source><volume>109</volume><fpage>E2424</fpage><lpage>32</lpage><pub-id pub-id-type="doi">10.1073/pnas.1207846109</pub-id></element-citation></ref><ref id="bib34"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lee</surname><given-names>YY</given-names></name><name><surname>Cevallos</surname><given-names>RC</given-names></name><name><surname>Jan</surname><given-names>E</given-names></name></person-group><year>2009</year><article-title>An upstream open reading frame regulates translation of GADD34 during cellular stresses that induce eIF2alpha phosphorylation</article-title><source>J Biol Chem</source><volume>284</volume><fpage>6661</fpage><lpage>73</lpage><pub-id pub-id-type="doi">10.1074/jbc.M806735200</pub-id></element-citation></ref><ref id="bib35"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lerner</surname><given-names>AG</given-names></name><name><surname>Upton</surname><given-names>JP</given-names></name><name><surname>Praveen</surname><given-names>PV</given-names></name><name><surname>Ghosh</surname><given-names>R</given-names></name><name><surname>Nakagawa</surname><given-names>Y</given-names></name><name><surname>Igbaria</surname><given-names>A</given-names></name><etal/></person-group><year>2012</year><article-title>IRE1alpha induces thioredoxin-interacting protein to activate the NLRP3 inflammasome and promote programmed cell death under irremediable ER stress</article-title><source>Cell Metab</source><volume>16</volume><fpage>250</fpage><lpage>64</lpage><pub-id pub-id-type="doi">10.1016/j.cmet.2012.07.007</pub-id></element-citation></ref><ref id="bib36"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lonial</surname><given-names>S</given-names></name><name><surname>Mitsiades</surname><given-names>CS</given-names></name><name><surname>Richardson</surname><given-names>PG</given-names></name></person-group><year>2011</year><article-title>Treatment options for relapsed and refractory multiple myeloma</article-title><source>Clin Cancer Res</source><volume>17</volume><fpage>1264</fpage><lpage>77</lpage><pub-id pub-id-type="doi">10.1158/1078-0432.CCR-10-1805</pub-id></element-citation></ref><ref id="bib37"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lu</surname><given-names>PD</given-names></name><name><surname>Harding</surname><given-names>HP</given-names></name><name><surname>Ron</surname><given-names>D</given-names></name></person-group><year>2004</year><article-title>Translation reinitiation at alternative open reading frames regulates gene expression in an integrated stress response</article-title><source>J Cell Biol</source><volume>167</volume><fpage>27</fpage><lpage>33</lpage><pub-id pub-id-type="doi">10.1083/jcb.200408003</pub-id></element-citation></ref><ref id="bib38"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>MacLean</surname><given-names>B</given-names></name><name><surname>Tomazela</surname><given-names>DM</given-names></name><name><surname>Shulman</surname><given-names>N</given-names></name><name><surname>Chambers</surname><given-names>M</given-names></name><name><surname>Finney</surname><given-names>GL</given-names></name><name><surname>Frewen</surname><given-names>B</given-names></name><etal/></person-group><year>2010</year><article-title>Skyline: an open source document editor for creating and analyzing targeted proteomics experiments</article-title><source>Bioinformatics</source><volume>26</volume><fpage>966</fpage><lpage>8</lpage><pub-id pub-id-type="doi">10.1093/bioinformatics/btq054</pub-id></element-citation></ref><ref id="bib39"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Mahrus</surname><given-names>S</given-names></name><name><surname>Trinidad</surname><given-names>JC</given-names></name><name><surname>Barkan</surname><given-names>DT</given-names></name><name><surname>Sali</surname><given-names>A</given-names></name><name><surname>Burlingame</surname><given-names>AL</given-names></name><name><surname>Wells</surname><given-names>JA</given-names></name></person-group><year>2008</year><article-title>Global sequencing of proteolytic cleavage sites in apoptosis by specific labeling of protein N termini</article-title><source>Cell</source><volume>134</volume><fpage>866</fpage><lpage>76</lpage><pub-id pub-id-type="doi">10.1016/j.cell.2008.08.012</pub-id></element-citation></ref><ref id="bib40"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Maiolica</surname><given-names>A</given-names></name><name><surname>Junger</surname><given-names>MA</given-names></name><name><surname>Ezkurdia</surname><given-names>I</given-names></name><name><surname>Aebersold</surname><given-names>R</given-names></name></person-group><year>2012</year><article-title>Targeted proteome investigation via selected reaction monitoring mass spectrometry</article-title><source>J Proteomics</source><volume>75</volume><fpage>3495</fpage><lpage>513</lpage><pub-id pub-id-type="doi">10.1016/j.jprot.2012.04.048</pub-id></element-citation></ref><ref id="bib41"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Mendillo</surname><given-names>ML</given-names></name><name><surname>Santagata</surname><given-names>S</given-names></name><name><surname>Koeva</surname><given-names>M</given-names></name><name><surname>Bell</surname><given-names>GW</given-names></name><name><surname>Hu</surname><given-names>R</given-names></name><name><surname>Tamimi</surname><given-names>RM</given-names></name><etal/></person-group><year>2012</year><article-title>HSF1 drives a transcriptional program distinct from heat shock to support highly malignant human cancers</article-title><source>Cell</source><volume>150</volume><fpage>549</fpage><lpage>62</lpage><pub-id pub-id-type="doi">10.1016/j.cell.2012.06.031</pub-id></element-citation></ref><ref id="bib42"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Menschaert</surname><given-names>G</given-names></name><name><surname>Van Criekinge</surname><given-names>W</given-names></name><name><surname>Notelaers</surname><given-names>T</given-names></name><name><surname>Koch</surname><given-names>A</given-names></name><name><surname>Crappe</surname><given-names>J</given-names></name><name><surname>Gevaert</surname><given-names>K</given-names></name><etal/></person-group><year>2013</year><article-title>Deep proteome coverage based on ribosome profiling aids mass spectrometry-based protein and peptide discovery and provides evidence of alternative translation products and near-cognate translation initiation events</article-title><source>Mol Cell Proteomics</source><volume>12</volume><fpage>1780</fpage><lpage>90</lpage><pub-id pub-id-type="doi">10.1074/mcp.M113.027540</pub-id></element-citation></ref><ref id="bib43"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Mertins</surname><given-names>P</given-names></name><name><surname>Udeshi</surname><given-names>ND</given-names></name><name><surname>Clauser</surname><given-names>KR</given-names></name><name><surname>Mani</surname><given-names>DR</given-names></name><name><surname>Patel</surname><given-names>J</given-names></name><name><surname>Ong</surname><given-names>SE</given-names></name><etal/></person-group><year>2012</year><article-title>iTRAQ labeling is superior to mTRAQ for quantitative global proteomics and phosphoproteomics</article-title><source>Mol Cell Proteomics</source><volume>11</volume><fpage>M111</fpage><pub-id pub-id-type="doi">10.1074/mcp.M111.014423</pub-id></element-citation></ref><ref id="bib44"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Mitsiades</surname><given-names>N</given-names></name><name><surname>Mitsiades</surname><given-names>CS</given-names></name><name><surname>Poulaki</surname><given-names>V</given-names></name><name><surname>Chauhan</surname><given-names>D</given-names></name><name><surname>Fanourakis</surname><given-names>G</given-names></name><name><surname>Gu</surname><given-names>X</given-names></name><etal/></person-group><year>2002</year><article-title>Molecular sequelae of proteasome inhibition in human multiple myeloma cells</article-title><source>Proc Natl Acad Sci USA</source><volume>99</volume><fpage>14374</fpage><lpage>9</lpage><pub-id pub-id-type="doi">10.1073/pnas.202445099</pub-id></element-citation></ref><ref id="bib45"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Moffitt</surname><given-names>KL</given-names></name><name><surname>Martin</surname><given-names>SL</given-names></name><name><surname>Walker</surname><given-names>B</given-names></name></person-group><year>2010</year><article-title>Proteases implicated in apoptosis: old and new</article-title><source>J Pharm Pharmacol</source><volume>62</volume><fpage>563</fpage><lpage>76</lpage><pub-id pub-id-type="doi">10.1211/jpp/62.05.0002</pub-id></element-citation></ref><ref id="bib46"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Mortazavi</surname><given-names>A</given-names></name><name><surname>Williams</surname><given-names>BA</given-names></name><name><surname>McCue</surname><given-names>K</given-names></name><name><surname>Schaeffer</surname><given-names>L</given-names></name><name><surname>Wold</surname><given-names>B</given-names></name></person-group><year>2008</year><article-title>Mapping and quantifying mammalian transcriptomes by RNA-Seq</article-title><source>Nat Methods</source><volume>5</volume><fpage>621</fpage><lpage>8</lpage><pub-id pub-id-type="doi">10.1038/nmeth.1226</pub-id></element-citation></ref><ref id="bib47"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Obeng</surname><given-names>EA</given-names></name><name><surname>Carlson</surname><given-names>LM</given-names></name><name><surname>Gutman</surname><given-names>DM</given-names></name><name><surname>Harrington</surname><given-names>WJ</given-names><suffix>Jnr</suffix></name><name><surname>Lee</surname><given-names>KP</given-names></name><name><surname>Boise</surname><given-names>LH</given-names></name></person-group><year>2006</year><article-title>Proteasome inhibitors induce a terminal unfolded protein response in multiple myeloma cells</article-title><source>Blood</source><volume>107</volume><fpage>4907</fpage><lpage>16</lpage><pub-id pub-id-type="doi">10.1182/blood-2005-08-3531</pub-id></element-citation></ref><ref id="bib48"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Oslowski</surname><given-names>CM</given-names></name><name><surname>Hara</surname><given-names>T</given-names></name><name><surname>O’Sullivan-Murphy</surname><given-names>B</given-names></name><name><surname>Kanekura</surname><given-names>K</given-names></name><name><surname>Lu</surname><given-names>S</given-names></name><name><surname>Hara</surname><given-names>M</given-names></name><etal/></person-group><year>2012</year><article-title>Thioredoxin-interacting protein mediates ER stress-induced beta cell death through initiation of the inflammasome</article-title><source>Cell Metab</source><volume>16</volume><fpage>265</fpage><lpage>73</lpage><pub-id pub-id-type="doi">10.1016/j.cmet.2012.07.005</pub-id></element-citation></ref><ref id="bib49"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Ow</surname><given-names>SY</given-names></name><name><surname>Salim</surname><given-names>M</given-names></name><name><surname>Noirel</surname><given-names>J</given-names></name><name><surname>Evans</surname><given-names>C</given-names></name><name><surname>Rehman</surname><given-names>I</given-names></name><name><surname>Wright</surname><given-names>PC</given-names></name></person-group><year>2009</year><article-title>iTRAQ underestimation in simple and complex mixtures: “the good, the bad and the ugly”</article-title><source>J Proteome Res</source><volume>8</volume><fpage>5347</fpage><lpage>55</lpage><pub-id pub-id-type="doi">10.1021/pr900634c</pub-id></element-citation></ref><ref id="bib50"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Parlati</surname><given-names>F</given-names></name><name><surname>Lee</surname><given-names>SJ</given-names></name><name><surname>Aujay</surname><given-names>M</given-names></name><name><surname>Suzuki</surname><given-names>E</given-names></name><name><surname>Levitsky</surname><given-names>K</given-names></name><name><surname>Lorens</surname><given-names>JB</given-names></name><etal/></person-group><year>2009</year><article-title>Carfilzomib can induce tumor cell death through selective inhibition of the chymotrypsin-like activity of the proteasome</article-title><source>Blood</source><volume>114</volume><fpage>3439</fpage><lpage>47</lpage><pub-id pub-id-type="doi">10.1182/blood-2009-05-223677</pub-id></element-citation></ref><ref id="bib51"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Parrish</surname><given-names>AB</given-names></name><name><surname>Freel</surname><given-names>CD</given-names></name><name><surname>Kornbluth</surname><given-names>S</given-names></name></person-group><year>2013</year><article-title>Cellular mechanisms controlling caspase activation and function</article-title><source>Cold Spring Harb Perspect Biol</source><volume>5</volume><fpage>a008672</fpage><pub-id pub-id-type="doi">10.1101/cshperspect.a008672</pub-id></element-citation></ref><ref id="bib52"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Pop</surname><given-names>C</given-names></name><name><surname>Salvesen</surname><given-names>GS</given-names></name></person-group><year>2009</year><article-title>Human caspases: activation, specificity, and regulation</article-title><source>J Biol Chem</source><volume>284</volume><fpage>21777</fpage><lpage>81</lpage><pub-id pub-id-type="doi">10.1074/jbc.R800084200</pub-id></element-citation></ref><ref id="bib53"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Rawlings</surname><given-names>ND</given-names></name><name><surname>Barrett</surname><given-names>AJ</given-names></name><name><surname>Bateman</surname><given-names>A</given-names></name></person-group><year>2012</year><article-title>MEROPS: the database of proteolytic enzymes, their substrates and inhibitors</article-title><source>Nucleic Acids Res</source><volume>40</volume><fpage>D343</fpage><lpage>50</lpage><pub-id pub-id-type="doi">10.1093/nar/gkr987</pub-id></element-citation></ref><ref id="bib54"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Reid</surname><given-names>DW</given-names></name><name><surname>Nicchitta</surname><given-names>CV</given-names></name></person-group><year>2012</year><article-title>Primary role for endoplasmic reticulum-bound ribosomes in cellular translation identified by ribosome profiling</article-title><source>J Biol Chem</source><volume>287</volume><fpage>5518</fpage><lpage>27</lpage><pub-id pub-id-type="doi">10.1074/jbc.M111.312280</pub-id></element-citation></ref><ref id="bib55"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Robinson</surname><given-names>JT</given-names></name><name><surname>Thorvaldsdottir</surname><given-names>H</given-names></name><name><surname>Winckler</surname><given-names>W</given-names></name><name><surname>Guttman</surname><given-names>M</given-names></name><name><surname>Lander</surname><given-names>ES</given-names></name><name><surname>Getz</surname><given-names>G</given-names></name><etal/></person-group><year>2011</year><article-title>Integrative genomics viewer</article-title><source>Nat Biotechnol</source><volume>29</volume><fpage>24</fpage><lpage>6</lpage><pub-id pub-id-type="doi">10.1038/nbt.1754</pub-id></element-citation></ref><ref id="bib56"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Schwanhausser</surname><given-names>B</given-names></name><name><surname>Busse</surname><given-names>D</given-names></name><name><surname>Li</surname><given-names>N</given-names></name><name><surname>Dittmar</surname><given-names>G</given-names></name><name><surname>Schuchhardt</surname><given-names>J</given-names></name><name><surname>Wolf</surname><given-names>J</given-names></name><etal/></person-group><year>2011</year><article-title>Global quantification of mammalian gene expression control</article-title><source>Nature</source><volume>473</volume><fpage>337</fpage><lpage>42</lpage><pub-id pub-id-type="doi">10.1038/nature10098</pub-id></element-citation></ref><ref id="bib57"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Schwanhausser</surname><given-names>B</given-names></name><name><surname>Busse</surname><given-names>D</given-names></name><name><surname>Li</surname><given-names>N</given-names></name><name><surname>Dittmar</surname><given-names>G</given-names></name><name><surname>Schuchhardt</surname><given-names>J</given-names></name><name><surname>Wolf</surname><given-names>J</given-names></name><etal/></person-group><year>2013</year><article-title>Corrigendum: global quantification of mammalian gene expression control</article-title><source>Nature</source><volume>495</volume><fpage>126</fpage><lpage>7</lpage><pub-id pub-id-type="doi">10.1038/nature11848</pub-id></element-citation></ref><ref id="bib58"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Sheynkman</surname><given-names>GM</given-names></name><name><surname>Shortreed</surname><given-names>MR</given-names></name><name><surname>Frey</surname><given-names>BL</given-names></name><name><surname>Smith</surname><given-names>LM</given-names></name></person-group><year>2013</year><article-title>Discovery and mass spectrometric analysis of novel splice-junction peptides using RNA-seq</article-title><source>Mol Cell Proteomics</source><volume>12</volume><fpage>2341</fpage><lpage>53</lpage><pub-id pub-id-type="doi">10.1074/mcp.O113.028142</pub-id></element-citation></ref><ref id="bib59"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Shimbo</surname><given-names>K</given-names></name><name><surname>Hsu</surname><given-names>GW</given-names></name><name><surname>Nguyen</surname><given-names>H</given-names></name><name><surname>Mahrus</surname><given-names>S</given-names></name><name><surname>Trinidad</surname><given-names>JC</given-names></name><name><surname>Burlingame</surname><given-names>AL</given-names></name><etal/></person-group><year>2012</year><article-title>Quantitative profiling of caspase-cleaved substrates reveals different drug-induced and cell-type patterns in apoptosis</article-title><source>Proc Natl Acad Sci USA</source><volume>109</volume><fpage>12432</fpage><lpage>7</lpage><pub-id pub-id-type="doi">10.1073/pnas.1208616109</pub-id></element-citation></ref><ref id="bib60"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Spencer</surname><given-names>SL</given-names></name><name><surname>Sorger</surname><given-names>PK</given-names></name></person-group><year>2011</year><article-title>Measuring and modeling apoptosis in single cells</article-title><source>Cell</source><volume>144</volume><fpage>926</fpage><lpage>39</lpage><pub-id pub-id-type="doi">10.1016/j.cell.2011.03.002</pub-id></element-citation></ref><ref id="bib61"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Spriggs</surname><given-names>KA</given-names></name><name><surname>Bushell</surname><given-names>M</given-names></name><name><surname>Willis</surname><given-names>AE</given-names></name></person-group><year>2010</year><article-title>Translational regulation of gene expression during conditions of cell stress</article-title><source>Mol Cell</source><volume>40</volume><fpage>228</fpage><lpage>37</lpage><pub-id pub-id-type="doi">10.1016/j.molcel.2010.09.028</pub-id></element-citation></ref><ref id="bib62"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Stern-Ginossar</surname><given-names>N</given-names></name><name><surname>Weisburd</surname><given-names>B</given-names></name><name><surname>Michalski</surname><given-names>A</given-names></name><name><surname>Le</surname><given-names>VT</given-names></name><name><surname>Hein</surname><given-names>MY</given-names></name><name><surname>Huang</surname><given-names>SX</given-names></name><etal/></person-group><year>2012</year><article-title>Decoding human cytomegalovirus</article-title><source>Science</source><volume>338</volume><fpage>1088</fpage><lpage>93</lpage><pub-id pub-id-type="doi">10.1126/science.1227919</pub-id></element-citation></ref><ref id="bib63"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Vogel</surname><given-names>C</given-names></name><name><surname>Marcotte</surname><given-names>EM</given-names></name></person-group><year>2012</year><article-title>Insights into the regulation of protein abundance from proteomic and transcriptomic analyses</article-title><source>Nat Rev Genet</source><volume>13</volume><fpage>227</fpage><lpage>32</lpage><pub-id pub-id-type="doi">10.1038/nrg3185</pub-id></element-citation></ref><ref id="bib64"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Walter</surname><given-names>P</given-names></name><name><surname>Ron</surname><given-names>D</given-names></name></person-group><year>2011</year><article-title>The unfolded protein response: from stress pathway to homeostatic regulation</article-title><source>Science</source><volume>334</volume><fpage>1081</fpage><lpage>6</lpage><pub-id pub-id-type="doi">10.1126/science.1209038</pub-id></element-citation></ref><ref id="bib65"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Wang</surname><given-names>M</given-names></name><name><surname>Weiss</surname><given-names>M</given-names></name><name><surname>Simonovic</surname><given-names>M</given-names></name><name><surname>Haertinger</surname><given-names>G</given-names></name><name><surname>Schrimpf</surname><given-names>SP</given-names></name><name><surname>Hengartner</surname><given-names>MO</given-names></name><etal/></person-group><year>2012</year><article-title>PaxDb, a database of protein abundance averages across all three domains of life</article-title><source>Mol Cell Proteomics</source><volume>11</volume><fpage>492</fpage><lpage>500</lpage><pub-id pub-id-type="doi">10.1074/mcp.O111.014704</pub-id></element-citation></ref><ref id="bib66"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Whitney</surname><given-names>ML</given-names></name><name><surname>Jefferson</surname><given-names>LS</given-names></name><name><surname>Kimball</surname><given-names>SR</given-names></name></person-group><year>2009</year><article-title>ATF4 is necessary and sufficient for ER stress-induced upregulation of REDD1 expression</article-title><source>Biochem Biophys Res Commun</source><volume>379</volume><fpage>451</fpage><lpage>5</lpage><pub-id pub-id-type="doi">10.1016/j.bbrc.2008.12.079</pub-id></element-citation></ref><ref id="bib67"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Yoon</surname><given-names>YJ</given-names></name><name><surname>Kim</surname><given-names>JA</given-names></name><name><surname>Shin</surname><given-names>KD</given-names></name><name><surname>Shin</surname><given-names>DS</given-names></name><name><surname>Han</surname><given-names>YM</given-names></name><name><surname>Lee</surname><given-names>YJ</given-names></name><etal/></person-group><year>2011</year><article-title>KRIBB11 inhibits HSP70 synthesis through inhibition of heat shock factor 1 function by impairing the recruitment of positive transcription elongation factor b to the hsp70 promoter</article-title><source>J Biol Chem</source><volume>286</volume><fpage>1737</fpage><lpage>47</lpage><pub-id pub-id-type="doi">10.1074/jbc.M110.179440</pub-id></element-citation></ref><ref id="bib68"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Zhu</surname><given-names>YX</given-names></name><name><surname>Tiedemann</surname><given-names>R</given-names></name><name><surname>Shi</surname><given-names>CX</given-names></name><name><surname>Yin</surname><given-names>H</given-names></name><name><surname>Schmidt</surname><given-names>JE</given-names></name><name><surname>Bruins</surname><given-names>LA</given-names></name><etal/></person-group><year>2011</year><article-title>RNAi screen of the druggable genome identifies modulators of proteasome inhibitor sensitivity in myeloma including CDK5</article-title><source>Blood</source><volume>117</volume><fpage>3847</fpage><lpage>57</lpage><pub-id pub-id-type="doi">10.1182/blood-2010-08-304022</pub-id></element-citation></ref></ref-list></back><sub-article article-type="article-commentary" id="SA1"><front-stub><article-id pub-id-type="doi">10.7554/eLife.01236.035</article-id><title-group><article-title>Decision letter</article-title></title-group><contrib-group content-type="section"><contrib contrib-type="editor"><name><surname>Cravatt</surname><given-names>Ben</given-names></name><role>Reviewing editor</role><aff><institution>The Scripps Research Institute</institution>, <country>United States</country></aff></contrib></contrib-group></front-stub><body><boxed-text><p>eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see <ext-link ext-link-type="uri" xlink:href="http://elife.elifesciences.org/review-process">review process</ext-link>). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.</p></boxed-text><p>[Editors’ note: although it is not typical of the review process at <italic>eLife</italic>, in this case the editors decided to include the reviews in their entirety for the authors’ consideration as they prepared their revised submission.]</p><p>Thank you for sending your work entitled “Global cellular response to chemotherapy-induced apoptosis” for consideration at <italic>eLife</italic>. Your article has been favorably evaluated by a Senior editor, a Reviewing editor, and 2 reviewers.</p><p>The Reviewing editor and the two reviewers discussed their comments before we reached this decision, and the Reviewing editor has assembled the following comments to help you prepare a revised submission.</p><p>Both reviewers were enthusiastic about your manuscript and its findings. However, the reviewers have also recommended some revisions to strengthen the paper. Specifically, questions were raised about whether a more thorough bioinformatics analysis, especially one that better integrates the ribosome profiling and proteomics data sets, might identify important examples where protein levels were controlled primarily by proteolytic degradation (or strongly affected by unconventional 5' UTRs). To the extent that you and your co-authors can address these interesting questions, we believe that the revised manuscript would be further improved.</p><p>In this case we think it would be beneficial to also include the full reviews of your article as you prepare your revision.</p><p><italic>Reviewer 1</italic>:</p><p>Overall, this is an interesting study that uses multiple different types of profiling technologies to establish how multiple myeloma cell lines response to bortezomib, the common anti-multiple myeloma treatment.</p><p>The authors characterize a variety of different transcriptional and post transcriptional responses, including changes in transcript levels, changes in ribosome profiling read density, protein levels, and protease cleavage.</p><p>Overall, the manuscript is interesting because of the multiple different approaches.</p><p>If there is a flaw, it is that many of the bioinformatics analyses seem to be performed without a strong rationale (or at least a stated rationale). The data seems to be available, so the data is analyzed. As such, in many cases, no clear hypotheses are generated.</p><p>Here are a few examples:</p><p>1) The authors perform both ribosome profiling and proteomics (iTRAQ). But the authors say that the purpose of this experiment is to measure “the production of proteins during bortezomib-induced apoptosis.” I don't think the authors are properly considering their iTRAQ data, especially with respect the to ribosome profiling data. iTRAQ does not speak to protein production, it speaks to overall protein levels, which is a combination of protein synthesis and degradation. Ribosome profiling gives protein production. The proteomic data would be more valuable if the authors would contrast this with the ribosome profiling data, and then see if the differences are due to proteolytic degradation. For instance, some proteins may be found at a lower level than predicted by the ribosome profiling. Are these proteins that are targeted for degradation?</p><p>2) The authors devote a section to changes in 5'UTR reads. However, they don’t define what these 5'UTR reads correspond to. I think they correspond to uORFs and unconventional in-frame start sites upstream of the canonical ATG. If these are the only two types of 5'UTRs, it seems that these should be analyzed separately since they might have different forms of regulation.</p><p>3) Along the lines of the 5'UTR reads: It seems to me that the model is that increased 5'UTR reads might lead to lower CDS reads, and thereby contribute to the overall protein translation level seen by ribosome profiling and overall expression level seen by iTRAQ. But, this doesn’t seem to be bioinformatically tested. It seems that the 5'UTR reads are analyzed without seeing how these are influencing overall protein levels.</p><p>4) I am not sure if I can tell if the N-terminalomic data gives insights into whether the cleavage events are rare or at a high stoichiometry. Perhaps the authors can comment on this. It seems to me that the authors could extrapolate the important proteolytic cleavage events by looking at the ribosome profiling and iTRAQ datasets (as discussed in point 1 above) to possibly address this issue.</p><p><italic>Reviewer 2</italic>:</p><p>The study of Wiita et al. reports the systems-level analyses of transcription, translation and proteolysis in the multiple myeloma cancer cell line treated with a proteasomal inhibitor bortezomib. This study represent an unprecedented integrative study and even though the conclusive findings are not world shocking, the use of state-of-the-art technologies, the experimental design and data integration really makes this a reference work for future integrative studies and puts this study in the class of pioneering integrative -omics studies.</p><p>Further, the authors highlight some of the pitfalls, such as the increased ionization suppression of iTRAQ reporter ions as compared to SRM approaches in mass spectrometry, which are typically overlooked when solely making use of one -omics approach.</p><p>Further since (regulated) protein proteolysis/degradation cannot be studied by means of ribosome profiling, the combined use of ribosome profiling and shotgun/N-terminomics approaches additionally enables the study of protein degradation; in this case mainly reflected by caspase-mediated proteolysis.</p><p>Since this study represents a comprehensive and integrative -omics story, and viewing the fact that RNA-Seq and proteomics data from the same model system was obtained, the creation of a customized database could have been considered for the comprehensive identification/discovery of database non-annotated peptides (i.e., splice-junction peptides, products of uORF translation, translation at near-cognate start codons, ...). The authors could try to implement this or at least discuss some of the advantages using such an approach and refer to some previous studies making use of customized databases for proteomics (<xref ref-type="bibr" rid="bib62">Stern-Ginossar et al., 2012</xref>; <xref ref-type="bibr" rid="bib42">Menschaert et al., 2013</xref>; <xref ref-type="bibr" rid="bib58">Sheynkman et al., 2013</xref>).</p><p>Overall, this study is well designed and the manuscript well written and this reviewer strongly supports publication of this important work.</p></body></sub-article><sub-article article-type="reply" id="SA2"><front-stub><article-id pub-id-type="doi">10.7554/eLife.01236.036</article-id><title-group><article-title>Author response</article-title></title-group></front-stub><body><p>Response to Reviewer 1:</p><p><italic>1) The authors perform both ribosome profiling and proteomics (iTRAQ). But the authors say that the purpose of this experiment is to measure “the production of proteins during bortezomib-induced apoptosis.” I don't think the authors are properly considering their iTRAQ data, especially with respect the to ribosome profiling data. iTRAQ does not speak to protein production, it speaks to overall protein levels, which is a combination of protein synthesis and degradation. Ribosome profiling gives protein production. The proteomic data would be more valuable if the authors would contrast this with the ribosome profiling data, and then see if the differences are due to proteolytic degradation. For instance, some proteins may be found at a lower level than predicted by the ribosome profiling. Are these proteins that are targeted for degradation</italic>?</p><p>We thank the reviewer for the astute comment. It is absolutely true that the iTRAQ data reflects overall protein levels and, in general, not synthesis alone. In our experimental system, however, treatment with a high dose of bortezomib effectively blocks proteasomal degradation in this system. Bortezomib’s IC<sub>50</sub> for proteasome inhibition in MM1.S cells is known to be ∼5 nM (Chauhan D et al., <italic>Cancer Cell</italic> (2005) 8, 407) and the 20 nM dose has previously been shown to lead to essentially complete proteasomal blockade (Berkers CR et al., <italic>Nat Methods</italic> (2005) 2, 357). Conversely, other endoproteases (such as caspases) appear to cleave protein substrates in a limited number of sites generating large protein fragments (Crawford et al. <italic>Mol Cell Proteomics</italic> (2013) 12, 813; Dix et al <italic>Cell</italic> (2008) 134, 679; Mahrus et al. <italic>Cell</italic> (2008), 134, 866). In the absence of proteasomal activity, these fragments would be expected to remain stable in the cell as large polypeptides (whether retaining any function or not). By iTRAQ mass spectrometry, based on trypsinization of the total protein sample, these proteolytically cleaved fragments would be indistinguishable from intact proteins. As a result iTRAQ provides no information on caspase (or non-caspase) proteolysis which does not degrade proteins below the domain level. Therefore, in this system we believe that degradation as measured by iTRAQ would both be close to zero for all proteins, and relative iTRAQ data across time points largely reflects protein production.</p><p>To clarify this point we have changed the line referred to by the reviewer to “changes in protein levels during bortezomib-induced apoptosis”. We have also modified another statement to state “at 20 nM bortezomib in MM1.S cells, proteasomal activity is almost completely inhibited (<xref ref-type="bibr" rid="bib6">Berkers et al., 2005</xref>)” and added a statement later in the same paragraph: “These domains would be indistinguishable from intact proteins by iTRAQ. Therefore, we expect that protein degradation detectable by iTRAQ is extremely limited and changes by iTRAQ largely reflect protein production alone.”</p><p>The reviewer further makes an excellent point that ribosome profiling data could be used as a predictor for protein production. One could subsequently examine discrepancies between predicted protein production and that detected by proteomics. For the quantitative model presented in <xref ref-type="fig" rid="fig5">Figure 5</xref>, the ribosome profiling data is implicitly included as none of the genes analyzed had large changes in translational efficiency. Hence, any protein changes predicted by ribosome profiling would be equivalent to changes predicated by mRNA-seq data. As shown in <xref ref-type="fig" rid="fig5s4">Figure 5–figure supplement 4</xref>, this model well-describes the analyzed genes and does not lead to any notable discrepancies.</p><p>However, a different and interesting question is inferring protein abundance changes directly from ribosome profiling data (i.e., X number of increased ribosome footprints = Y number of new proteins). We previously attempted to incorporate this analysis as another element of our quantitative model, but theoretical studies suggest that the predicted protein output of a transcript is not a straightforward function of measured ribosome footprint read density (Tuller T. et al. <italic>Cell</italic> (2010) 141, 344; Cannarozzi G. et al. <italic>Cell</italic> (2010) 141, 355). In our current study under conditions of rapid apoptosis, we were unable to test or develop any new predictions for converting ribosome occupancy to protein synthesis as so few proteins show detectable changes in iTRAQ data.</p><p><italic>2) The authors devote a section to changes in 5'UTR reads. However, they don’t define what these 5'UTR reads correspond to. I think they correspond to uORFs and unconventional in-frame start sites upstream of the canonical ATG. If these are the only two types of 5'UTRs, it seems that these should be analyzed separately since they might have different forms of regulation</italic>.</p><p>The reviewer is correct that footprint reads present in 5’ UTRs could apply to either translated upstream open reading frames (uORFs) generating short polypeptides separate from the coding sequence or N-terminal extensions leading to different isoforms of the canonical protein. Indeed we did not clearly define this in the main text. As we mention in the text, stalling translation initiation complexes with the drugs harringtonine or lactimidomycin is necessary to definitively identify start sites. Using the translation elongation inhibitor cycloheximide, as we do here, we cannot readily define these different types of 5’ UTR translation across all transcripts and we therefore must analyze them together. However, to the reviewer’s point, we have added a line to the text stating that “Ribosome occupancy in this region may indicate translation of short regulatory polypeptides in upstream open reading frames (uORFs) or alternate production of N-terminal isoforms of canonically translated proteins...” In the few cases we examine in detail (<xref ref-type="fig" rid="fig3 fig3s1 fig3s2 fig3s3">Figure 3 and supplements</xref>), we characterize these as uORFs because all of the analyzed translation products, whether in-frame or out of frame, have a stop codon prior to the canonical coding sequence; N-terminal extensions would both be in-frame with the canonical coding sequence and have no intervening stop codon.</p><p><italic>3) Along the lines of the 5'UTR reads: It seems to me that the model is that increased 5'UTR reads might lead to lower CDS reads, and thereby contribute to the overall protein translation level seen by ribosome profiling and overall expression level seen by iTRAQ. But, this doesn’t seem to be bioinformatically tested. It seems that the 5'UTR reads are analyzed without seeing how these are influencing overall protein levels</italic>.</p><p>We thank the reviewer for this suggestion to integrate our 5’ UTR analysis with proteomic data. We point out in <xref ref-type="fig" rid="fig3">Figure 3B</xref> that our current analysis compares the ratio of read density in the 5’ UTR to the protein coding sequence (CDS) across the time course. We did not see a strong relationship between changes in UTR translation affecting CDS translation. In support of our result, Geraschenko et al. (<italic>PNAS</italic> (2012) 109, 17394) noted large increases in 5’ UTR translation in yeast after oxidative stress but noted a similarly complex relationship to translation of the protein coding sequence. So from this analysis it does not appear that increased 5’ UTR translation always leads to decreased CDS translation.</p><p>To further explore this idea, we have performed two additional pieces of analysis. We first compared whether relative 5’ UTR translation in untreated cells shows a relationship to steady-state protein abundance by iBAQ. In this analysis, incorporating 2266 genes with 5’ UTR reads and iBAQ data, we visually observe a negative relationship: genes with high levels of relative 5’ UTR to CDS translation appear to show lower levels of protein abundance (new <xref ref-type="fig" rid="fig5s2">Figure 5–figure supplement 2D</xref>; new source data in <xref ref-type="supplementary-material" rid="SD4-data">Figure 5–source data 1</xref>). However, statistical analysis demonstrates a non-significant negative correlation (Pearson <italic>R</italic> = -.02, <italic>p</italic> = 0.35). Therefore, we cannot definitively conclude that increased 5’ UTR translation at baseline leads to lower protein abundance. We have now included a statement: “Interestingly, we also found that increased 5’ UTR translation at baseline may correlate with decreased steady-state protein abundance (<xref ref-type="fig" rid="fig5s2">Figure 5–supplementary figure 2D</xref>).”</p><p>We further examined the changes in relative 5’ UTR read density across the few proteins that are found to be increased by iTRAQ analysis across a time course. We also compared these results to a set of genes with increased read density at the mRNA and ribosome footprint data but no detected change by iTRAQ (new <xref ref-type="fig" rid="fig4s3">Figure 4–figure supplement 3</xref>). The 5’ UTR to CDS read density remained relatively constant for all genes across the time course. We did not, for instance, note consistent decreases in 5’ UTR translation in the genes with increased protein abundance. Instead, this ratio showed little divergence from the value at 0h, consistent with our analysis of all genes in <xref ref-type="fig" rid="fig3">Figure 3B</xref>. However, as mentioned above, this analysis is necessarily limited by the very few proteins for which we saw abundance changes by iTRAQ. We have added a statement: “Of note, we did not observe any distinct changes in relative 5’ UTR translation for the few proteins increased by iTRAQ (<xref ref-type="fig" rid="fig4s3">Figure 4–figure supplement 3</xref>).”</p><p><italic>4) I am not sure if I can tell if the N-terminalomic data gives insights into whether the cleavage events are rare or at a high stoichiometry. Perhaps the authors can comment on this. It seems to me that the authors could extrapolate the important proteolytic cleavage events by looking at the ribosome profiling and iTRAQ datasets (as discussed in point 1 above) to possibly address this issue</italic>.</p><p>The reviewer is correct that the N-terminomic data over a time course, while providing useful information on relative substrate cleavage, does not determine the absolute amount of substrate cleavage relative to the total protein present. As mentioned above in the response to point 1, the iTRAQ data unfortunately also does not offer much insight into protein degradation in this system. This is further evidenced by the fact that so few proteins show signatures of decreased abundance by iTRAQ (<xref ref-type="fig" rid="fig4">Figure 4</xref>) whereas many appear to be proteolytically cleaved by SRM (<xref ref-type="fig" rid="fig6">Figure 6</xref>). We gain some insight into stoichiometry based on Western blotting results (as in <xref ref-type="fig" rid="fig5">Figure 5A</xref> for PERK and 4E-BP1 and <xref ref-type="fig" rid="fig1s2">Figure 1– figure supplement 2</xref> for XIAP and ATF-4), suggesting that peptides which show strong proteolytic cleavage by SRM indicate ∼50 to >90% of total protein cleaved. This extent of cleavage has also been seen by Western blotting in our prior studies incorporating SRM (Agard NJ et al. <italic>PNAS</italic> (2012) 109, 1913; Shimbo K et al <italic>PNAS</italic> (2012) 109, 12432).</p><p>Response to Reviewer 2:</p><p><italic>Since this study represents a comprehensive and integrative -omics story, and viewing the fact that RNA-Seq and proteomics data from the same model system was obtained, the creation of a customized database could have been considered for the comprehensive identification/discovery of database non-annotated peptides (i.e., splice-junction peptides, products of uORF translation, translation at near-cognate start codons, ...). The authors could try to implement this or at least discuss some of the advantages using such an approach and refer to some previous studies making use of customized databases for proteomics (<xref ref-type="bibr" rid="bib62">Stern-Ginossar et al., 2012</xref>; <xref ref-type="bibr" rid="bib42">Menschaert et al., 2013</xref>; <xref ref-type="bibr" rid="bib58">Sheynkman et al., 2013</xref>)</italic>.</p><p>The reviewer makes an excellent point that customized databases incorporating ribosome profiling data for proteomic searches are likely to have a significant impact on future studies. We are aware of the references cited but unfortunately our current data does not lend itself to similar analysis as we cannot definitively identify uORF and alternate start sites in our data, as discussed above. In addition, regarding prediction of splice-junction peptides, these data are greatly informed by the use of long reads and paired-end mRNA-seq to enhance alignment across splice junctions. Our data are currently composed of short reads (50 bp) and single-end sequencing which cannot lead to as comprehensive or accurate splice junction assignment. Therefore, we do not believe that our current data set is appropriate for the creation of such a reference database, though it is clear that such databases are certainly important. We have added a comment to the Discussion relevant to the reviewer’s point: “Incorporation of translation initiation inhibitors in ribosome profiling and paired-end reads in mRNA-seq can also provide novel insight into translation regulation through generation of custom proteomic databases (<xref ref-type="bibr" rid="bib62">Stern-Ginossar et al., 2012</xref>; <xref ref-type="bibr" rid="bib42">Menschaert et al., 2013</xref>; <xref ref-type="bibr" rid="bib58">Sheynkman et al., 2013</xref>).”</p></body></sub-article></article> |