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: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">03640</article-id><article-id pub-id-type="doi">10.7554/eLife.03640</article-id><article-categories><subj-group subj-group-type="display-channel"><subject>Research article</subject></subj-group><subj-group subj-group-type="heading"><subject>Cell biology</subject></subj-group><subj-group subj-group-type="heading"><subject>Human biology and medicine</subject></subj-group></article-categories><title-group><article-title>Peroxisomal lactate dehydrogenase is generated by translational readthrough in mammals</article-title></title-group><contrib-group><contrib contrib-type="author" equal-contrib="yes" id="author-15185"><name><surname>Schueren</surname><given-names>Fabian</given-names></name><xref ref-type="aff" rid="aff1"/><xref ref-type="fn" rid="equal-contrib">†</xref><xref ref-type="fn" rid="con1"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><contrib contrib-type="author" equal-contrib="yes" id="author-15186"><name><surname>Lingner</surname><given-names>Thomas</given-names></name><xref ref-type="aff" rid="aff2"/><xref ref-type="fn" rid="equal-contrib">†</xref><xref ref-type="other" rid="par-2"/><xref ref-type="fn" rid="con5"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><contrib contrib-type="author" equal-contrib="yes" id="author-15187"><name><surname>George</surname><given-names>Rosemol</given-names></name><xref ref-type="aff" rid="aff1"/><xref ref-type="fn" rid="equal-contrib">†</xref><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-17238"><name><surname>Hofhuis</surname><given-names>Julia</given-names></name><xref ref-type="aff" rid="aff1"/><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-15188"><name><surname>Dickel</surname><given-names>Corinna</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" corresp="yes" id="author-15191"><name><surname>Gärtner</surname><given-names>Jutta</given-names></name><xref ref-type="aff" rid="aff1"/><xref ref-type="corresp" rid="cor1">*</xref><xref ref-type="other" rid="par-3"/><xref ref-type="fn" rid="con6"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><contrib contrib-type="author" corresp="yes" id="author-14918"><name><surname>Thoms</surname><given-names>Sven</given-names></name><xref ref-type="aff" rid="aff1"/><xref ref-type="corresp" rid="cor2">*</xref><xref ref-type="other" rid="par-1"/><xref ref-type="fn" rid="con7"/><xref ref-type="fn" rid="conf1"/><xref ref-type="other" rid="dataro1"/></contrib><aff id="aff1"><institution content-type="dept">Department of Pediatrics and Adolescent Medicine</institution>, <institution>University Medical Center, Georg-August-University Göttingen</institution>, <addr-line><named-content content-type="city">Göttingen</named-content></addr-line>, <country>Germany</country></aff><aff id="aff2"><institution content-type="dept">Department of Bioinformatics</institution>, <institution>Institute for Microbiology and Genetics, Georg-August-University Göttingen</institution>, <addr-line><named-content content-type="city">Göttingen</named-content></addr-line>, <country>Germany</country></aff></contrib-group><contrib-group content-type="section"><contrib contrib-type="editor"><name><surname>Sonenberg</surname><given-names>Nahum</given-names></name><role>Reviewing editor</role><aff><institution>McGill University</institution>, <country>Canada</country></aff></contrib></contrib-group><author-notes><corresp id="cor1"><label>*</label>For correspondence: <email>gaertnj@med.uni-goettingen.de</email> (JG);</corresp><corresp id="cor2"><label>*</label>For correspondence: <email>sven.thoms@med.uni-goettingen.de</email> (ST)</corresp><fn fn-type="con" id="equal-contrib"><label>†</label><p>These authors contributed equally to this work</p></fn></author-notes><pub-date date-type="pub" publication-format="electronic"><day>23</day><month>09</month><year>2014</year></pub-date><pub-date pub-type="collection"><year>2014</year></pub-date><volume>3</volume><elocation-id>e03640</elocation-id><history><date date-type="received"><day>09</day><month>06</month><year>2014</year></date><date date-type="accepted"><day>22</day><month>09</month><year>2014</year></date></history><permissions><copyright-statement>© 2014, Schueren et al</copyright-statement><copyright-year>2014</copyright-year><copyright-holder>Schueren et al</copyright-holder><license xlink:href="http://creativecommons.org/licenses/by/4.0/"><license-p>This article is distributed under the terms of the <ext-link ext-link-type="uri" xlink:href="http://creativecommons.org/licenses/by/4.0/">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="elife03640.pdf"/><abstract><object-id pub-id-type="doi">10.7554/eLife.03640.001</object-id><p>Translational readthrough gives rise to low abundance proteins with C-terminal extensions beyond the stop codon. To identify functional translational readthrough, we estimated the readthrough propensity (RTP) of all stop codon contexts of the human genome by a new regression model in silico, identified a nucleotide consensus motif for high RTP by using this model, and analyzed all readthrough extensions in silico with a new predictor for peroxisomal targeting signal type 1 (PTS1). Lactate dehydrogenase B (LDHB) showed the highest combined RTP and PTS1 probability. Experimentally we show that at least 1.6% of the total cellular LDHB is targeted to the peroxisome by a conserved hidden PTS1. The readthrough-extended lactate dehydrogenase subunit LDHBx can also co-import LDHA, the other LDH subunit, into peroxisomes. Peroxisomal LDH is conserved in mammals and likely contributes to redox equivalent regeneration in peroxisomes.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.001">http://dx.doi.org/10.7554/eLife.03640.001</ext-link></p></abstract><abstract abstract-type="executive-summary"><object-id pub-id-type="doi">10.7554/eLife.03640.002</object-id><title>eLife digest</title><p>Amino acids are the building blocks of proteins, and the order of the amino acids in a protein is determined by the order in which ‘codons’ appear in a messenger RNA molecule. Most codons represent a specific amino acid, but there are also three stop codons that are used to mark the end of a protein.</p><p>When the cellular machinery that ‘translates’ the messenger RNA molecule into a protein encounters a stop codon, it stops and releases the completed protein. Sometimes, however, the stop codon is not interpreted as a stop signal, and the translation of the messenger RNA molecule continues until another stop codon is encountered. This process is known as readthrough.</p><p>Some organisms, in particular viruses and fungi, use readthrough to produce a wider range of proteins than their genomes would otherwise allow. While readthrough also occurs in higher organisms such as mammals, it is not known if the resulting proteins perform extra functions that the original protein does not perform.</p><p>A number of factors affect whether readthrough occurs when an mRNA template is being translated. For example, each of the three stop codons has a different likelihood of having its stop signal misinterpreted, and the mRNA sequence that surrounds the stop codon can also affect the likelihood of readthrough.</p><p>Schueren et al. have developed a computational model that estimates how common this form of translational readthrough is in the human genome. The model was based on the identity of the stop codons themselves and the surrounding mRNA sequence. This model was then combined with another model that identifies proteins that are targeted to a structure inside a cell called the peroxisome, which is where a number of essential energy-releasing reactions take place. The combined model enabled Schueren et al. to identify proteins that both perform functions in the peroxisome and are likely to be formed by readthrough.</p><p>The combined model suggested a protein that is a part of lactate dehydrogenase: an enzyme that speeds up chemical reactions that are important for the cell to produce energy. Low levels of lactate dehydrogenase had previously been found in the peroxisome, despite it apparently lacking a specific sequence of amino acids that proteins need to have to enter the peroxisome. However, Schueren et al. confirmed experimentally that readthrough does occur for the lactate dehydrogenase component identified by the model, revealing that it contains a ‘hidden’ peroxisome-targeting region. Furthermore, when more translational readthrough occurred, more lactate dehydrogenase was found in the peroxisomes.</p><p>This unusual way that lactate dehydrogenase enters the peroxisome is an example of how the cell optimizes the used of the genetic information encoded in the genome and in messenger RNA. Translational readthrough always ensures that a certain proportion of lactate dehydrogenase will be brought to the peroxisome. The computational model developed here will be a valuable tool to identify other such proteins produced from genomes, including the human genome and those of other species.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.002">http://dx.doi.org/10.7554/eLife.03640.002</ext-link></p></abstract><kwd-group kwd-group-type="author-keywords"><title>Author keywords</title><kwd>translational readthrough</kwd><kwd>stop codon suppression</kwd><kwd>lactate dehydrogenase</kwd><kwd>gene regulation</kwd><kwd>genome-wide in silico screen</kwd><kwd>peroxisomal targeting signal</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-id institution-id-type="FundRef">http://dx.doi.org/10.13039/501100003385</institution-id><institution>Georg-August-Universität Göttingen</institution></institution-wrap></funding-source><award-id>Research Program, Faculty of Medicine,</award-id><principal-award-recipient><name><surname>Thoms</surname><given-names>Sven</given-names></name></principal-award-recipient></award-group><award-group id="par-2"><funding-source><institution-wrap><institution-id institution-id-type="FundRef">http://dx.doi.org/10.13039/501100001659</institution-id><institution>Deutsche Forschungsgemeinschaft</institution></institution-wrap></funding-source><award-id>LI2050/1-1</award-id><principal-award-recipient><name><surname>Lingner</surname><given-names>Thomas</given-names></name></principal-award-recipient></award-group><award-group id="par-3"><funding-source><institution-wrap><institution-id institution-id-type="FundRef">http://dx.doi.org/10.13039/501100001659</institution-id><institution>Deutsche Forschungsgemeinschaft</institution></institution-wrap></funding-source><award-id>GA354/7-1</award-id><principal-award-recipient><name><surname>Gärtner</surname><given-names>Jutta</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>Whole-genome in silico screening for stop codon contexts that produce elevated translational readthrough and peroxisome targeting reveals that in mammals, the amount of lactate dehydrogenase in the peroxisome is controlled by translational readthrough.</meta-value></custom-meta></custom-meta-group></article-meta></front><body><sec id="s1" sec-type="intro"><title>Introduction</title><p>Translation of genetic information encoded in mRNAs into proteins is carried out by ribosomes. When a stop codon enters the ribosomal A site, release factors bind the stop codon, hydrolyze the peptidyl-tRNA bond, and trigger the release of the polypeptide from the ribosome. If instead of release factor 1 (eRF1), a near-cognate aminoacyl-tRNA pairs with the stop codon in the ribosomal A site, the stop signal is suppressed. Such decoding of a stop codon as a sense codon is known as translational readthrough. As a consequence, translation continues to the next stop codon resulting in the synthesis of C-terminally extended proteins (<xref ref-type="bibr" rid="bib1">Baranov et al., 2002</xref>; <xref ref-type="bibr" rid="bib44">Namy et al., 2004</xref>; <xref ref-type="bibr" rid="bib13">Firth and Brierley, 2012</xref>). Mutant tRNAs are classic stop codon suppressors, but termination also occurs with less than 100% efficiency in normal physiology.</p><p>A number of cis-elements on the mRNA, typically 3′ of the stop codon together with trans-acting factors, are known to influence stop codon readthrough (<xref ref-type="bibr" rid="bib14">Firth et al., 2011</xref>). A case of translational readthrough dependent on RNA cis-elements has recently been found and termed programmed translational readthrough (PTR) (<xref ref-type="bibr" rid="bib11">Eswarappa et al., 2014</xref>). But it is also known that the stop codon itself and the nucleotides before and after the stop codon affect readthrough. The three stop codons differ in their tendency to be suppressed. In human, UAA is least and UGA is most likely to allow readthrough (<xref ref-type="bibr" rid="bib3">Beier and Grimm, 2001</xref>; <xref ref-type="bibr" rid="bib1">Baranov et al., 2002</xref>). Studies also show that the nucleotide immediately downstream of the stop codon is biased and can strongly influence readthrough (<xref ref-type="bibr" rid="bib39">McCaughan et al., 1995</xref>). We here define translational readthrough that is entirely dependent on the stop codon and the nucleotides in its immediate vicinity as basal translational readthrough (BTR). Thus BTR is independent of cis-acting elements and also differs from pharmacologically induced readthrough. Induction of readthrough, most prominently by aminoglycoside antibiotics, is an attractive strategy in the treatment of the large number of genetic disorders caused by premature stop codons (<xref ref-type="bibr" rid="bib4">Bidou et al., 2012</xref>; <xref ref-type="bibr" rid="bib28">Keeling et al., 2014</xref>).</p><p>In viruses, readthrough optimizes the coding capacity of compact genomes (<xref ref-type="bibr" rid="bib13">Firth and Brierley, 2012</xref>). In the yeast <italic>Saccharomyces cerevisiae</italic>, the eukaryotic release factor eRF3 can form prion-like polymers, which introduces a level of epigenetic regulation not found in other eukaryotes (<xref ref-type="bibr" rid="bib54">Tuite and Cox, 2003</xref>). In fungi, translational readthrough extends cytosolic glycolytic enzymes by a cryptic peroxisomal targeting signal (<xref ref-type="bibr" rid="bib17">Freitag et al., 2012</xref>). In <italic>Drosophila</italic>, readthrough is known to affect between 200 and 300 proteins (<xref ref-type="bibr" rid="bib27">Jungreis et al., 2011</xref>; <xref ref-type="bibr" rid="bib10">Dunn et al., 2013</xref>), and in mammals readthrough has been described for more than 50 individual transcripts (<xref ref-type="bibr" rid="bib18">Geller and Rich, 1980</xref>; <xref ref-type="bibr" rid="bib9">Chittum et al., 1998</xref>; <xref ref-type="bibr" rid="bib59">Yamaguchi et al., 2012</xref>; <xref ref-type="bibr" rid="bib10">Dunn et al., 2013</xref>; <xref ref-type="bibr" rid="bib11">Eswarappa et al., 2014</xref>; <xref ref-type="bibr" rid="bib36">Loughran et al., 2014</xref>). Ribosome profiling and phylogenetic approaches provide powerful methods for the systematic identification of readthrough in mammals (<xref ref-type="bibr" rid="bib27">Jungreis et al., 2011</xref>; <xref ref-type="bibr" rid="bib10">Dunn et al., 2013</xref>; <xref ref-type="bibr" rid="bib11">Eswarappa et al., 2014</xref>; <xref ref-type="bibr" rid="bib36">Loughran et al., 2014</xref>).</p><p>We wanted to find a physiological role for translational readthrough in humans by identifying C-terminal extensions with targeting signals that would create a functional difference between the normal and the readthrough-extended form. To achieve this aim, we concentrated on proteins deriving from BTR. Based on experimental data, we assigned regression coefficients to all possible nucleotides in the stop codon context (SCC) and, using those regression coefficients, estimated the readthrough propensity (RTP) of all stop codons in the human genome or transcriptome. We were able to formally derive a new nucleotide consensus for high RTP from the regression coefficients of our model. Then we screened all predicted C-terminal extensions for peroxisomal targeting signals because peroxisomes import most of their matrix proteins through a short targeting signal (PTS1) at the very C-terminus (<xref ref-type="bibr" rid="bib51">Smith and Aitchison, 2013</xref>). We here show that lactate dehydrogenase B (LDHB) combines a very high translational readthrough with a hidden, yet functional and evolutionarily conserved, PTS1. This peroxisomal isoform of LDH, containing the readthrough-extended LDHBx subunit, is likely to be involved in the regeneration of redox equivalents for peroxisomal β-oxidation.</p></sec><sec id="s2" sec-type="results"><title>Results</title><sec id="s2-1"><title>Genome-wide in silico analysis of basal translational readthrough</title><p>In order to develop a computational method to assess the RTP of all human SCCs that would allow the identification of genes with high BTR, we focused on SCCs comprising 15 nucleotides including and surrounding the stop codon (nucleotides −6 to +9, stop codon at positions 1 to 3). In order to calculate linear regression between the SCCs and their experimental BTR values, we formalized SCCs using a binary vector that represented the stop context in a multi-dimensional vector space (<xref ref-type="fig" rid="fig1">Figure 1A</xref> and <xref ref-type="fig" rid="fig1s1">Figure 1—figure supplement 1</xref>). The three stop codons were condensed into one position, so that the binary vector required 51 dimensions, for the four possible nucleotides in the six positions before and after the stop codon, and for the three stop codons (12 × 4 + 3). The vector was combined with experimentally accessible BTR frequencies. For the first approximation model (LIN), we used 66 sequences derived from human nonsense mutations (<xref ref-type="bibr" rid="bib16">Floquet et al., 2012</xref>). The nucleotide sequences of these stop contexts show no bias with respect to RTP, because the contexts and the stop codons evolved independently, and therefore the context nucleotides are random in relation to the stop codon. We calculated a linear regression model for these SCCs and used only the experimental BTR values that had been measured in the absence of aminoglycosides. The model assigns regression coefficients to all possible nucleotides in the stop context (<xref ref-type="fig" rid="fig1s1">Figure 1—figure supplement 1</xref>).<fig-group><fig id="fig1" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.003</object-id><label>Figure 1.</label><caption><title>Genome-wide in silico analysis of basal translational readthrough (BTR) in humans.</title><p>(<bold>A</bold>) Schematic representation of the readthrough propensity (RTP) predictor algorithm. Binary vector representations of stop codon contexts and their experimental readthrough values are used to determine the coefficients of a linear regression model. For prediction of RTP for a given stop codon context, the position-specific regression coefficients are added up. (<bold>B</bold>) RTP for selected human transcripts. LIN denotes first-pass RTP calculations, LINiter iterative improvement of RTP scoring, and LINfs3 and LINfs5 the reduced models. The RTP of all human transcripts can be found in Dataset 1 (<xref ref-type="bibr" rid="bib49">Schueren et al., 2014</xref>). (<bold>C</bold>) Experimental readthrough by dual reporter assay in HeLa cells. Readthrough is expressed as luciferase per Venus signal. The red line marks the background readthrough level obtained from a construct containing two contiguous UAA stop codons separating the Venus and the hRluc. The aminoglycoside geneticin (100 µg/ml) induces translational readthrough. SCC: stop codon context; hRluc: humanized Renilla luciferase. Error bars, SD.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.003">http://dx.doi.org/10.7554/eLife.03640.003</ext-link></p></caption><graphic xlink:href="elife03640f001"/></fig><fig id="fig1s1" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.03640.004</object-id><label>Figure 1—figure supplement 1.</label><caption><title>Schematic representation of the readthrough propensity (RTP) prediction procedure.</title><p>This scheme summarizes how regression coefficients were extracted from experimental basal readthrough (BTR) data. (<bold>1</bold>) Stop codon contexts (SCC) (positions −6 to +9, stop codon at position 1 to 3) with known experimental BTR values are formalized as binary vectors in 51-dimensional vector space. (<bold>2</bold>) The binary vector reserves four entries for the four possible bases in each position (4 × 12) and three for the stop codon. (<bold>3</bold>) In combination with their corresponding experimental readthrough values (in %), they are used to determine the (<bold>4</bold>) coefficients of a linear regression model. (<bold>5</bold>) For RTP calculation for a given SCC, the position-specific regression coefficients are added up. The values used in the example are from the LINiter model. The algorithm is used to calculate the RTP of SCCs of 42,000 unique 3′ transcript termini listed in Dataset 1 (<xref ref-type="bibr" rid="bib49">Schueren et al., 2014</xref>).</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.004">http://dx.doi.org/10.7554/eLife.03640.004</ext-link></p></caption><graphic xlink:href="elife03640fs001"/></fig><fig id="fig1s2" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.03640.005</object-id><label>Figure 1—figure supplement 2.</label><caption><title>Correlation of RTP and BTR in the LINiter model.</title><p>Scatter plot indicating the correlation between readthrough propensity (RTP) and experimental basal translational readthrough (BTR). RTP was obtained by leave-one-out cross-validation. Pearson correlation coefficient 0.34 (p = 0.002).</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.005">http://dx.doi.org/10.7554/eLife.03640.005</ext-link></p></caption><graphic xlink:href="elife03640fs002"/></fig></fig-group></p><p>For a first round of whole-genome RTP prediction, we extracted the SCCs for each transcript from the Ensembl database and calculated RTP by adding up the regression coefficients of all relevant positions. An outline of this algorithm is shown in <xref ref-type="fig" rid="fig1">Figure 1A</xref> and in more detail in <xref ref-type="fig" rid="fig1s1">Figure 1—figure supplement 1</xref>. A sortable list of LIN RTP values for all human transcripts is contained in Dataset 1 (<xref ref-type="bibr" rid="bib49">Schueren et al., 2014</xref>).</p><p>To expand the data basis of the RTP algorithm and to obtain evidence that the algorithm indeed predicts BTR values, we selected candidate transcripts with high, intermediate, and low RTP and tested them using a dual reporter assay (<xref ref-type="fig" rid="fig1">Figure 1B</xref> and <xref ref-type="table" rid="tbl1">Table 1</xref>). For experimental analysis, SCCs spanning 10 nucleotides upstream and downstream of the stop codon were expressed with a 5′/N-terminal yellow fluorescent protein (Venus) and a 3′/C-terminal humanized Renilla luciferase (hRluc) tag. Stop suppression leads to the expression of hRluc, and Venus served as an internal expression control. Readthrough is expressed as luciferase activity per Venus fluorescence. This approach excludes introns and exon junction complexes and, due to the relatively short stretch of variable nucleotides between the reporters, also does not allow for extensive RNA structures that could modulate readthrough. Consequently, this form of the dual reporter assay focuses on the assessment of BTR not influenced by specific cis-elements. The additional candidates tested showed BTR between 0.10% (±0.006%) and 2.91% (±0.15%) relative to the 100% readthrough control expressing the Venus-hRluc fusion protein without an intervening stop codon region (<xref ref-type="fig" rid="fig1">Figure 1C</xref> and <xref ref-type="table" rid="tbl1">Table 1</xref>). The aminoglycoside antibiotic geneticin (G418) increased readthrough by between 3.25 (±0.41) and 40.38 (±5.33)-fold (<xref ref-type="fig" rid="fig1">Figure 1C</xref>). Geneticin could only increase the luciferase-per-Venus signal when a stop codon separated Venus and luciferase, indicating that our dual reporter assay faithfully reports readthrough. The finding that experimental readthrough could be increased by treatment with aminoglycosides also excludes alternative mechanisms such as RNA editing or splicing that might explain the relative increase of the luciferase over the Venus signal. The highest levels of induction can only be reached when basal readthrough is low, and, vice versa, a high BTR somewhat limits the maximum induction factor (<xref ref-type="fig" rid="fig1">Figure 1C</xref>), suggesting that maximal BTR readthrough is limited to levels below 15%.<table-wrap id="tbl1" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.006</object-id><label>Table 1.</label><caption><p>Additional experimental dual reporter readthrough data of stop codon context constructs used for the LINiter model</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.006">http://dx.doi.org/10.7554/eLife.03640.006</ext-link></p></caption><table frame="hsides" rules="groups"><thead><tr><th>Gene symbol</th><th>Stop codon context</th><th>Readthrough (%) (SD)</th></tr></thead><tbody><tr><td><italic>ZNF-574</italic></td><td>GATCAGTGGC TGA CTCTGCCCGA</td><td>0.31 (0.020)</td></tr><tr><td><italic>LDHB</italic></td><td>AAAAGACCTG TGA CTAGTGAGCT</td><td>1.55 (0.087)</td></tr><tr><td><italic>PPP1R3F</italic></td><td>ATTCTCCCAA TAA AGCTTTACAG</td><td>0.18 (0.009)</td></tr><tr><td><italic>LDHB [TGAT]</italic></td><td>AAAAGACCTG TGA TTAGTGAGCT</td><td>0.17 (0.009)</td></tr><tr><td><italic>LDHB [TAA]</italic></td><td>AAAAGACCTG TAA CTAGTGAGCT</td><td>0.20 (0.009)</td></tr><tr><td><italic>LDHB [TAAT]</italic></td><td>AAAAGACCTG TAA TTAGTGAGCT</td><td>0.17 (0.009)</td></tr><tr><td><italic>LENG1</italic></td><td>CCTTACTCAC TGA CTCCTGAGGG</td><td>0.26 (0.009)</td></tr><tr><td><italic>VASN</italic></td><td>GCCCTACATC TAA GCCAGAGAGA</td><td>0.12 (0.004)</td></tr><tr><td><italic>MDH1</italic></td><td>TTCCTCTGCC TGA CTAGACAATG</td><td>2.91 (0.147)</td></tr><tr><td><italic>PRDM10</italic></td><td>CACCAAACCA TGA CTTCCACCCT</td><td>0.13 (0.005)</td></tr><tr><td><italic>FBXL20</italic></td><td>CATCATCCTA TGA CAATGGAGGT</td><td>0.10 (0.006)</td></tr><tr><td><italic>THG1L</italic></td><td>AGCCAGGCTT TGA CGGAAGAGTC</td><td>0.15 (0.006)</td></tr><tr><td><italic>EDEM3</italic></td><td>GGATGAGCTA TGA CTTGCTAAAC</td><td>0.66 (0.027)</td></tr><tr><td><italic>EDN1</italic></td><td>AGCACATTGG TGA CAGACCTTCG</td><td>0.25 (0.008)</td></tr><tr><td><italic>UBQLN1</italic></td><td>CCAGCCATCA TAG CAGCATTTCT</td><td>0.13 (0.009)</td></tr><tr><td><italic>IRAK3</italic></td><td>CAAAAAAGAA TAA ATTCTACCAG</td><td>0.10 (0.007)</td></tr><tr><td><italic>SLC3A1</italic></td><td>TACCTCGTGT TAG GCACCTTTAT</td><td>0.18 (0.008)</td></tr><tr><td><italic>LEPRE1</italic></td><td>GGATGAGCTA TGA CAGCGTCCAG</td><td>0.27 (0.010)</td></tr></tbody></table><table-wrap-foot><fn><p>Stop codon constructs expressing plus/minus 10 nucleotides were analyzed in HeLa cells.</p></fn></table-wrap-foot></table-wrap></p><p>Next we added our candidate sequences and their experimentally determined readthrough levels to obtain an iterative and extended RTP model (LINiter). Again, we applied this model to all human transcripts (see <xref ref-type="bibr" rid="bib49">Schueren et al., 2014</xref> for Dataset 1; model parameters are shown in <xref ref-type="table" rid="tbl2">Table 2</xref>). We measured the correlation of RTP and experimental BTR by leave-one-out cross-validation during computation of the regression coefficients. For the LINiter model, we obtained a weak but significant Pearson correlation coefficient of 0.34 (p = 0.002) (<xref ref-type="fig" rid="fig1s2">Figure 1—figure supplement 2</xref>). To determine the origin of the apparently strong non-linear contribution to RTP, we analyzed the regression coefficients of the LINiter model. Nucleotide positions associated with coefficients of large absolute value contribute most to RTP. The relative contribution of nucleotides within the SCC to the readthrough prediction is shown in <xref ref-type="fig" rid="fig2">Figure 2A</xref>.<table-wrap id="tbl2" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.007</object-id><label>Table 2.</label><caption><p>Regression factors of the LINiter and LINfs3 models</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.007">http://dx.doi.org/10.7554/eLife.03640.007</ext-link></p></caption><table frame="hsides" rules="groups"><thead><tr><th align="center" colspan="8">LINiter model (stop codon context position −6 to +9)</th></tr><tr><th align="center">Base/position</th><th align="center">−6</th><th align="center">−5</th><th align="center">−4</th><th align="center">−3</th><th align="center">−2</th><th align="center">−1</th><th align="center">4</th></tr></thead><tbody><tr><td align="center">A</td><td align="center">−0.00041</td><td align="center">0.00130</td><td align="center">−0.00028</td><td align="center">−0.00073</td><td align="center">−0.00071</td><td align="center">0.00016</td><td align="center">−0.00037</td></tr><tr><td align="center">C</td><td align="center">−0.00105</td><td align="center">0.00164</td><td align="center">0.00075</td><td align="center">−0.00004</td><td align="center">0.00133</td><td align="center">0.00109</td><td align="center">0.00375</td></tr><tr><td align="center">G</td><td align="center">0.00060</td><td align="center">−0.00077</td><td align="center">−0.00041</td><td align="center">0.00193</td><td align="center">−0.00048</td><td align="center">0.00043</td><td align="center">−0.00156</td></tr><tr><td align="center">U/T</td><td align="center">0.00200</td><td align="center">−0.00103</td><td align="center">0.00108</td><td align="center">−0.00002</td><td align="center">0.00100</td><td align="center">−0.00054</td><td align="center">−0.00067</td></tr></tbody></table><table frame="hsides" rules="groups"><thead><tr><th align="center">Base/position</th><th align="center">5</th><th align="center">6</th><th align="center">7</th><th align="center">8</th><th align="center">9</th><th align="center">Stop</th><th/></tr></thead><tbody><tr><td align="center">A</td><td align="center">−0.00068</td><td align="center">0.00276</td><td align="center">−0.00020</td><td align="center">0.00105</td><td align="center">−0.00081</td><td align="center">−0.00026</td><td align="center">TAA</td></tr><tr><td align="center">C</td><td align="center">−0.00097</td><td align="center">−0.00026</td><td align="center">−0.00062</td><td align="center">−0.00017</td><td align="center">0.00148</td><td align="center">−0.00103</td><td align="center">TAG</td></tr><tr><td align="center">G</td><td align="center">−0.00008</td><td align="center">−0.00059</td><td align="center">0.00245</td><td align="center">−0.00058</td><td align="center">0.00014</td><td align="center">0.00243</td><td align="center">TGA</td></tr><tr><td align="center">U/T</td><td align="center">0.00287</td><td align="center">−0.00076</td><td align="center">−0.00049</td><td align="center">0.00084</td><td align="center">0.00032</td><td/><td/></tr></tbody></table><table frame="hsides" rules="groups"><thead><tr><th align="center" colspan="6">LINfs3 model (Stop and position +4 to +6)</th></tr><tr><th align="center">Base/position</th><th align="center">4</th><th align="center">5</th><th align="center">6</th><th align="center">Stop</th><th/></tr></thead><tbody><tr><td align="center">A</td><td align="center">0.00006</td><td align="center">−0.00071</td><td align="center">0.00306</td><td align="center">0.00005</td><td align="center">TAA</td></tr><tr><td align="center">C</td><td align="center">0.00351</td><td align="center">−0.00056</td><td align="center">0.00021</td><td align="center">−0.00052</td><td align="center">TAG</td></tr><tr><td align="center">G</td><td align="center">−0.00111</td><td align="center">0.00010</td><td align="center">−0.00093</td><td align="center">0.00229</td><td align="center">TGA</td></tr><tr><td align="center">U/T</td><td align="center">−0.00064</td><td align="center">0.00299</td><td align="center">−0.00053</td><td/><td/></tr></tbody></table><table-wrap-foot><fn><p>These model weights are ‘raw’, that is as obtained from the ridge regression procedure. For prediction of RTP, the weights associated with nucleotides within the stop codon context and the corresponding stop codon have to be added up. For calculation of our RTP score, we normalized the model weight vectors (i.e., the complete stack of weights) to Euclidean unit sum which corresponds to a division of weights by 0.0088 (LINiter) and 0.0063 (LINfs3), respectively. Furthermore, the sequence feature vectors were normalized to Euclidean unit sum which corresponds to a division by the square root of the length (3.6 and 2, respectively). As a shortcut to this, the sum of raw scores can be divided by 0.0317 and 0.0126, respectively.</p></fn></table-wrap-foot></table-wrap><fig-group><fig id="fig2" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.008</object-id><label>Figure 2.</label><caption><title>Characterization of basal translational readthrough (BTR): consensus and candidates.</title><p>(<bold>A</bold>) Sequence logo plot of regression coefficients of stop codon contexts (SCCs) in the LINiter model. Character size corresponds to regression coefficients. The model treats stop codons as one nucleotide position. Filled/upside-down letters correspond to positive/negative coefficients, respectively. (<bold>B</bold>) Consensus motif for high readthrough propensity (RTP) derived from the predictive model. The stop codon together with the nucleotide triplet following the stop codon provides the best predictor for RTP. The consensus was derived by feature selection: starting from LINiter, positions with the least contribution to prediction were successively eliminated as indicated by the gray arrow. Nucleotide positions on the x-axis mark the removed positions upon transition to a reduced model. LINfs3 (<underline>UGA</underline> CUA, stop codon underlined) represents the global minimum of regression error (filled circle). The model LINfs5, corresponding to a local minimum, additionally encompasses positions +7 and −6, indicating that these positions could also contribute to high BTR. (<bold>C</bold>) BTR determination of candidates from the genome-wide in silico screen. Dual reporter assays with Venus and humanized Renilla luciferase containing SCCs from AQP4 (UGA CUA G), SYTL2 (UGA CUA G), CACNA2D4 (UGA CUA T), and DHX38 (UGA CUU G). AQP4, SYTL2, and CACNA2D4 reveal high BTR in all tissues tested. HT1080, human fibrosarcoma cell line; U373, glioblastoma cell line. HEK, human embryonic kidney cells. Error bars, SD.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.008">http://dx.doi.org/10.7554/eLife.03640.008</ext-link></p></caption><graphic xlink:href="elife03640f002"/></fig><fig id="fig2s1" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.03640.009</object-id><label>Figure 2—figure supplement 1.</label><caption><title>Correlation of RTP and BTR in the LINfs3 model.</title><p>Scatter plot indicating the correlation between readthrough propensity (RTP) and experimental basal translational readthrough (BTR). RTP was obtained by leave-one-out cross-validation. Pearson correlation coefficient 0.41 (p = 0.0001).</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.009">http://dx.doi.org/10.7554/eLife.03640.009</ext-link></p></caption><graphic xlink:href="elife03640fs003"/></fig><fig id="fig2s2" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.03640.010</object-id><label>Figure 2—figure supplement 2.</label><caption><title>Translational readthrough in humans.</title><p>Venn diagram indicating experimentally verified human genes and stop codon contexts (SCCs) associated with above-average translational readthrough. Genes were identified by ribosome profiling (<xref ref-type="bibr" rid="bib10">Dunn et al., 2013</xref>), by phylogenetic approaches (<xref ref-type="bibr" rid="bib27">Jungreis et al., 2011</xref>; <xref ref-type="bibr" rid="bib36">Loughran et al., 2014</xref>), and by in silico profiling (this study). Gene products marked in boldface (black) correspond to sequences carrying the consensus motif UGA CUA (G) identified in this study and by <xref ref-type="bibr" rid="bib36">Loughran et al., (2014)</xref>. The human genome contains 144 [30] transcripts with the high-RTP motifs UGA CUA [G]. Different experimental strategies lead to the identification of genes with high physiological readthrough rates, but the molecular mechanisms underlying readthrough are likely to vary.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.010">http://dx.doi.org/10.7554/eLife.03640.010</ext-link></p></caption><graphic xlink:href="elife03640fs004"/></fig></fig-group></p></sec><sec id="s2-2"><title>A consensus for high readthrough in humans</title><p>The sequence-logo representation of regression factors in <xref ref-type="fig" rid="fig2">Figure 2A</xref> indicates that the three or four nucleotides following the stop codon contribute to readthrough. The quantitative manner in which we derived LINiter values allowed us to rationally derive a nucleotide motif permitting high readthrough in humans. We identified the nucleotide positions with the strongest influence on BTR in humans by feature selection, that is by successively eliminating those positions that contribute least to the prediction (<xref ref-type="fig" rid="fig2">Figure 2B</xref>). One by one the nucleotide positions with the smallest sum of squared regression coefficients were removed from the model. We find that two reduced models improve the prediction. Models with either five or three relevant context positions in addition to the stop codon correspond to the local and global residual error minimum, respectively. LINfs5 comprises nucleotide position −6, the stop codon, and positions +4 to +7, and LINfs3 comprises only the stop codon and positions +4 to +6, that is the codon following the stop (<xref ref-type="fig" rid="fig2">Figure 2B</xref>). The results of this analysis indicate that in humans the stop codon and the three nucleotides immediately downstream of the stop codon have the largest influence on BTR (LINfs3). The corresponding consensus is <underline>UGA</underline> CUA (stop codon underlined). Possibly also the nucleotides at positions +7 (the fourth position after the stop) and −6 contribute to BTR. The RTP-BTR correlation associated with LINfs3 was 0.41 (p = 0.0001) (<xref ref-type="fig" rid="fig2s1">Figure 2—figure supplement 1</xref>). To test if the LINfs3 consensus indeed confers high BTR, we analyzed four additional candidate SCCs. Three high-RTP SCCs were derived from AQP4, SYTL2, and CACNA2D4, and DHX38 was used as a control with a low RTP. AQP4, SYTL2, and CACNA2D4 conform with the LINfs3 consensus, whereas DHX38 does not. AQP4, SYTL2, and CACNA2D4 showed 2.29% (±0.09%), 0.99% (±0.06%), and 0.61% (±0.02%) readthrough in HeLa cells, whereas for DHX38 readthrough was only 0.27% (±0.04%) (<xref ref-type="fig" rid="fig2">Figure 2C</xref>), confirming that LINfs3 SCC indeed allows a very high rate of stop suppression. Next we wanted to test if these conclusions obtained in HeLa cells can be extended to other cell types. We therefore performed dual reporter experiments using the HT1080 fibrosarcoma cell line, the human embryonic kidney cell line (HEK), and the U373 cell line. In all these experiments, the relative distribution of BTR values remained the same, with AQP4 showing the highest and DHX38 the lowest BTR (<xref ref-type="fig" rid="fig2">Figure 2C</xref>). The finding that readthrough is lower in CACNA2D4 than in AQP4 and SYTL2 can also be taken as evidence that SCC position +7 (fourth after the stop) makes a contribution. Taken together, these experiments show that BTR is indeed a property of the respective SCC, and that readthrough may be differently regulated in different tissues.</p><p>The linear approximation underlying the LINiter and the LINfs3 models led to the identification of the UGA CUA (LINfs3) consensus conferring high BTR. A partially overlapping set of genes with this consensus was recently tested (<xref ref-type="bibr" rid="bib36">Loughran et al., 2014</xref>). An overview of all experimentally confirmed cases of translational readthrough shown in <xref ref-type="fig" rid="fig2s2">Figure 2—figure supplement 2</xref> reveals that ribosome profiling, phylogenetic approaches, and RTP screening are complementary approaches. For example, only one of the 42 readthrough genes found by ribosome profiling in foreskin fibroblasts (<xref ref-type="bibr" rid="bib10">Dunn et al., 2013</xref>) contains the UGA CUA consensus. The widely varying levels and sequence requirements for efficient stop codon suppression suggest that multiple molecular mechanisms can cause readthrough in mammals.</p></sec><sec id="s2-3"><title>Identification of peroxisomal targeting signals in readthrough extensions</title><p>The genome-wide in silico analysis of RTP provides the basis for the identification of the physiological functions of a readthrough protein. We have therefore screened the extensions for possible elements that could confer functional differences between the normal and the extended form of the protein. We screened the extensions for possible transmembrane domains (<xref ref-type="bibr" rid="bib29">Krogh et al., 2001</xref>), for prenylation sites (<xref ref-type="bibr" rid="bib61">Zhang and Casey, 1996</xref>), for endoplasmic retention signals (<xref ref-type="bibr" rid="bib60">Zerangue et al., 2001</xref>; <xref ref-type="bibr" rid="bib52">Stornaiuolo et al., 2003</xref>), and for glycosylation sites (<xref ref-type="bibr" rid="bib62">Zielinska et al., 2010</xref>; <xref ref-type="bibr" rid="bib50">Schwarz and Aebi, 2011</xref>).</p><p>To identify genes with a high BTR and a readthrough extension conferring a biological function, we decided to focus on the detection of proteins carrying a hidden peroxisomal targeting signal type 1 (PTS1) in the extension. This targeting mechanism had been shown to divert a small fraction of cytosolic glycolytic proteins to peroxisomes in fungi (<xref ref-type="bibr" rid="bib17">Freitag et al., 2012</xref>). PTS1 cover more than 90% of the targeting motifs of peroxisomal matrix proteins. The alternative PTS2 is found in only very few matrix proteins, and has even been lost in some organisms (<xref ref-type="bibr" rid="bib31">Lanyon-Hogg et al., 2010</xref>). PTS1 is localized at the very C-terminus of a substrate protein. However, the quintessential PTS1, Ser-Lys-Leu (SKL), is neither necessary nor sufficient to support matrix protein import into peroxisomes. Variations exist, and amino acids upstream of the terminal tripeptide also contribute to targeting (<xref ref-type="bibr" rid="bib7">Brocard and Hartig, 2006</xref>). Moreover, PTS1 does not confer a binary decision (to import or not to import), but is likely to determine an equilibrium between cytosolic and peroxisomal localization. This is best exemplified by the peroxisomal marker protein catalase, a considerable amount of which is not imported into peroxisomes due to an inherently weak PTS1 which is associated with low affinity to the cytosolic PTS1-receptor PEX5 (<xref ref-type="bibr" rid="bib38">Maynard et al., 2004</xref>). We took advantage of these scalable properties of PTS1 and adapted to human PTS1 a prediction algorithm that we had previously developed for plants (<xref ref-type="bibr" rid="bib34">Lingner et al., 2011</xref>). This machine learning-based method has been shown to accurately predict proteins with canonical and non-canonical PTS1 peptides and provides evidence for peroxisome targeting in terms of a posterior probability (<xref ref-type="bibr" rid="bib34">Lingner et al., 2011</xref>).</p><p>To program the human PTS1 prediction algorithm, we conducted orthology searches on 24 known human PTS1 sequences in metazoa using BLAST against protein and EST databases. The resulting dataset and several thousand metazoan sequences without peroxisomal association were used as positive and negative examples in a discriminative machine learning setup. Here, the sequences were represented by binary vectors encoding the presence or absence of up to 15 C-terminal amino acids. Models were trained and validated using regularized least squares classifiers (RLSC) and fivefold cross-validation. A more detailed description of the human PTS1 scoring can be found in the ‘Materials and methods’ section. We calculated the PTS1 posterior probabilities of all predicted C-terminal readthrough extensions derived from the human transcriptome (see <xref ref-type="bibr" rid="bib49">Schueren et al., 2014</xref> for Dataset 1).</p></sec><sec id="s2-4"><title>LDHB is extended by translational readthrough</title><p>Based on the assumption that a protein is more likely to target to peroxisomes by a cryptic PTS1 when the RTP and the extension's PTS1 scores are high, we used the product of RTP LINiter scores and PTS1 posterior probabilities as a predictor of functional peroxisomal targeting by a hidden PTS1 in the extension (see <xref ref-type="bibr" rid="bib49">Schueren et al., 2014</xref> for Dataset 1). To avoid negative product scores, we scaled RTP between 0 and 1 before multiplication (now designated RTP<sup>+</sup>).</p><p>We identified LDHB, one of the two human lactate dehydrogenase (LDH) subunits, at the top (position 1 of 42,069 entries) of our sorted list of combined RTP<sup>+</sup> and PTS1 scores (see <xref ref-type="bibr" rid="bib49">Schueren et al., 2014</xref> for Dataset 1). The distribution of RTP<sup>+</sup> × PTS1 product scores over all human transcripts indicates that other candidates must have considerably lower RTPs and/or targeting efficiencies, because the score drops by 50% over the first 40 of 42,069 transcripts (<xref ref-type="fig" rid="fig3">Figure 3A</xref>).<fig-group><fig id="fig3" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.011</object-id><label>Figure 3.</label><caption><title>LDHB is extended by translational readthrough.</title><p>(<bold>A</bold>) Genomic distribution of RTP<sup>+</sup> × PTS1 product scores. Product scores are 0 for rank 5015 to 42069. Green cross: 50% of maximum score. LDHB has the highest product score, exceeding rank 2 by 24%. RTP<sup>+</sup> denotes positively scaled LINiter values. (<bold>B</bold> and <bold>C</bold>) Venus/hRluc dual reporter assay with LDHB wild-type and mutant stop codon contexts. Error bars, SD. (<bold>B</bold>) Wild-type LDHB stop context shows high basal translational readthrough (BTR). Mutational analysis of the LINfs3 consensus of LDHB. Replacement of the stop codon and mutations in positions +4 to +6 reduce readthrough. (<bold>C</bold>) LDHB readthrough induction by the aminoglycoside geneticin. (<bold>D</bold>) Full-length LDHB is extended by readthrough. Western blot of dual tag assay with LDHBx with N-terminal HA- and C-terminal Myc-tag. Molecular mass marker in kDa.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.011">http://dx.doi.org/10.7554/eLife.03640.011</ext-link></p></caption><graphic xlink:href="elife03640f003"/></fig><fig id="fig3s1" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.03640.012</object-id><label>Figure 3—figure supplement 1.</label><caption><title>The LDHB stop context favors readthrough (Western blot).</title><p>HeLa cells transfected with Venus/hRluc dual reporter constructs were analyzed by Western blot. Wild-type LDHB stop context (UGA CUA (stop underlined)) allows stop codon readthrough. Mutation of the stop codon (UAA CUA) and the downstream bases (UAA UUA) reduces readthrough. Geneticin treatment (100 µg/ml) induces translational readthrough in all contexts. Molecular mass marker in kDa.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.012">http://dx.doi.org/10.7554/eLife.03640.012</ext-link></p></caption><graphic xlink:href="elife03640fs005"/></fig></fig-group></p><p>To experimentally confirm high BTR, we expressed the human LDHB SCC in the Venus/hRluc dual reporter assay. Readthrough was 1.55% (±0.09%) and mutation of the stop codon and/or the consecutive nucleotide strongly suppressed readthrough (<xref ref-type="fig" rid="fig3">Figure 3B</xref> and <xref ref-type="fig" rid="fig3s1">Figure 3—figure supplement 1</xref>). Treatment with geneticin increased readthrough to 4.38% (±0.42%) (compare with induction factors in <xref ref-type="fig" rid="fig3">Figure 3C</xref>).</p><p>To establish that the full-length protein is extended by stop suppression, LDHB including the extension (designated LDHBx for ‘extended’) and mutants were expressed with N-terminal HA- and C-terminal Myc-tags and analyzed by Western blotting. Full-length LDHB showed aminoglycoside-inducible readthrough, and the loss of readthrough upon exchange of the stop codon or the nucleotide following the stop codon confirms the special function of the LDHB SCC in stimulating translational readthrough (<xref ref-type="fig" rid="fig3">Figure 3D</xref>).</p></sec><sec id="s2-5"><title>Peroxisomal localization of LDHB depends on translational readthrough</title><p>The identification of LDHB as virtually the only human protein with a high combined readthrough and peroxisomal targeting probability is surprising, because a peroxisomal readthrough-extended LDHBx entails at least one new LDH isoform. On the other hand, LDH activity and isoforms inside peroxisomes were known for more than four decades (<xref ref-type="bibr" rid="bib41">McGroarty et al., 1974</xref>; <xref ref-type="bibr" rid="bib45">Osmundsen, 1982</xref>; <xref ref-type="bibr" rid="bib56">Völkl and Fahimi, 1985</xref>; <xref ref-type="bibr" rid="bib2">Baumgart et al., 1996</xref>; <xref ref-type="bibr" rid="bib40">McClelland et al., 2003</xref>; <xref ref-type="bibr" rid="bib21">Gronemeyer et al., 2013</xref>). In the apparent absence of known targeting signals, however, it has not been possible to explain how the protein can enter the peroxisome. Therefore we conducted an investigation to determine whether the extended human LDHBx protein and the predicted PTS1 therein lead to peroxisomal localization. We expressed LDHBx as a fusion protein with an N-terminal enhanced yellow fluorescent protein (YFP) and co-labeled cells by immunofluorescence with the peroxisomal marker PEX14, a peroxisomal membrane protein. YFP-LDHB showed the expected cytosolic localization (<xref ref-type="fig" rid="fig4">Figures 4A</xref> and <xref ref-type="fig" rid="fig5">5A</xref>). We hypothesized that a large excess of cytosolic YFP-LDHB masks the peroxisomal localization of LDHBx. To remove cytosolic YFP-LDHB, we permeabilized cells by digitonin before fixation and washed out the cytosol using phosphate-buffered saline (PBS). In agreement with peroxisomal targeting through the cryptic PTS1, LDHBx is found localized in peroxisomes after removal of the cytosol (<xref ref-type="fig" rid="fig4 fig5">Figures 4B and 5B</xref>). In control experiments, we show complete removal of cytosolically expressed YFP by cytosol wash-out (<xref ref-type="fig" rid="fig5s1">Figure 5—figure supplement 1</xref>) and peroxisomal localization of a YFP variant fused to PTS1 of the peroxisomal matrix protein ACOX3 (<xref ref-type="fig" rid="fig5s2">Figure 5—figure supplement 2</xref>). To confirm that LDHB targeting to peroxisomes is dependent on the putative PTS1 in the readthrough extension, we changed the SRL terminus (PTS1 probability 94.3%) to SSI (0.002%) and to SR (ΔL, 0.00001%). These mutations blocked YFP-LDHBx targeting to the peroxisome (<xref ref-type="fig" rid="fig4 fig5">Figures 4C and 5C–F</xref>). Remarkably, exchange of the leaky UGA stop with the tighter UAA reduced peroxisomal localization of YFP-LDHB (<xref ref-type="fig" rid="fig6">Figure 6A,B</xref>). Our results show that the high-RTP SCCs as well as the PTS1 in the extension after the stop codon are needed for peroxisome targeting. The extension must be accessible to ribosomal translation and contain a functional PTS1. It is known that PTS1-dependent targeting guides proteins into peroxisomes and not only to the membrane. The dependence of LDH targeting on the hidden PTS1 and on the nature of the stop codon thus confirms that the protein is indeed inside the peroxisome. As expected, replacing the stop codon by tryptophan-encoding UGG renders LDHBx entirely dependent on the PTS1 (<xref ref-type="fig" rid="fig6">Figure 6C,D</xref>).<fig id="fig4" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.013</object-id><label>Figure 4.</label><caption><title>LDHBx targets to the peroxisome by translational readthrough and a hidden peroxisomal targeting signal type 1 (PTS1) in the 3′ extension.</title><p>(<bold>A</bold>–<bold>C</bold>) Direct fluorescence microscopy of transfected HeLa cells. Immunofluorescence with the peroxisome marker anti-PEX14 (red). (<bold>A</bold>) YFP-LDHB (green) mainly localizes to the cytosol. The strong fluorescence signal in the cytosol prevents detection of LDHB in other cellular compartments. (<bold>B</bold>) Upon plasma membrane permeabilization and removal of cytosol (-CYT), a small fraction of LDHB remains co-localized with the peroxisome marker. (<bold>C</bold>) Peroxisomal targeting of LDHB is dependent on the cryptic PTS1 Ser-Arg-Leu (SRL) in the extension. Deletion of the L in SRL blocks import into peroxisomes. (<bold>D</bold> and <bold>E</bold>) Endogenous LDHB is localized to peroxisomes in untransfected wild-type cells. Immunofluorescence with anti-LDHB (green) and anti-PEX14 (red) antibodies. (<bold>D</bold>) Endogenous LDHB is cytosolic. (<bold>E</bold>) Removal of cytosol (-CYT) reveals co-localization with PEX14. Bar 10 µm.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.013">http://dx.doi.org/10.7554/eLife.03640.013</ext-link></p></caption><graphic xlink:href="elife03640f004"/></fig><fig-group><fig id="fig5" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.014</object-id><label>Figure 5.</label><caption><title>Peroxisome targeting of LDHBx is dependent on a hidden peroxisomal targeting signal in the readthrough extension.</title><p>Combined direct fluorescence and immunofluorescence in HeLa cells. (<bold>A</bold>) YFP-LDHBx expression: LDHBx is mainly cytosolic. (<bold>B</bold>) LDHBx targets to the peroxisome. Cells were permeabilized with digitonin, and cytosol was removed by washing with phosphate-buffered saline. (<bold>C</bold>–<bold>F</bold>) Mutation of the cryptic PTS1 in the extension blocks peroxisomal targeting of LDHBx. (<bold>C</bold> and <bold>D</bold>) Deletion of the amino acid L of the SRL in the PTS1 readthrough extension gives a wild-type cytosolic localization of LDHB and blocks import into the peroxisome completely. (<bold>E</bold> and <bold>F</bold>) Similarly, the SRL-to-SSI substitution does not interfere with cytosolic expression of the LDHB but completely blocks peroxisomal localization of LDHBx[SSI]. Bar 10 µm.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.014">http://dx.doi.org/10.7554/eLife.03640.014</ext-link></p></caption><graphic xlink:href="elife03640f005"/></fig><fig id="fig5s1" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.03640.015</object-id><label>Figure 5—figure supplement 1.</label><caption><title>Permeabilization by digitonin allows complete removal of cytosol.</title><p>Combined direct fluorescence with anti-PEX14 immunofluorescence. HeLa cells were transfected with the empty vector expressing YFP in the cytosol. (<bold>A</bold>) Cytosolic expression of YFP. (<bold>B</bold>) Complete removal of cytosolic after cell permeabilization and washing with phosphate-buffered saline. Bar 10 µm.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.015">http://dx.doi.org/10.7554/eLife.03640.015</ext-link></p></caption><graphic xlink:href="elife03640fs006"/></fig><fig id="fig5s2" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.03640.016</object-id><label>Figure 5—figure supplement 2.</label><caption><title>Cell permeabilization and removal of cytosol maintains peroxisomal integrity and co-localization of peroxisome marker (positive control).</title><p>Combined direct fluorescence with anti-PEX14 immunofluorescence in HeLa cells. (<bold>A</bold> and <bold>B</bold>) Cells were transfected with a construct expressing the PTS1 of ACOX3 fused to the C-terminus of the YFP variant Venus. (<bold>B</bold>) Co-localization of PTS1 and PEX14 after removal of cytosol following cell permeabilization. Bar 10 µm.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.016">http://dx.doi.org/10.7554/eLife.03640.016</ext-link></p></caption><graphic xlink:href="elife03640fs007"/></fig></fig-group><fig id="fig6" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.017</object-id><label>Figure 6.</label><caption><title>Peroxisome targeting of LDHBx is dependent on the stop codon.</title><p>Combined direct fluorescence and immunofluorescence in HeLa cells. (<bold>A</bold> and <bold>B</bold>) Exchange of UGA stop codon with the tighter stop UAA (YFP-LDHBx[TAA]) reduces peroxisomal localization of LDHB. (<bold>C</bold> and <bold>D</bold>) When UGA is replaced by tryptophan-coding UGG (LDHBx[TGG]), a larger proportion of LDHB is targeted to the peroxisome, and peroxisome localization becomes obvious without removal of the cytosol. (<bold>B</bold>, <bold>D</bold>) Cytosol was removed after cell permeabilization with digitonin. Bar 10 µm.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.017">http://dx.doi.org/10.7554/eLife.03640.017</ext-link></p></caption><graphic xlink:href="elife03640f006"/></fig></p><p>To obtain more direct evidence for the readthrough-dependent low abundance targeting of human LDHB to peroxisomes, we analyzed untransfected wild-type cells by immunofluorescence with anti-LDHB and anti-PEX14 antibodies. LDHB appears distributed in the cytosol (<xref ref-type="fig" rid="fig4">Figure 4D</xref>). After cytosol depletion, however, the remaining LDHB signal is mainly peroxisomal (<xref ref-type="fig" rid="fig4">Figure 4E</xref>). A small portion of LDHB may localize to other cellular locations protected against cytosol removal. We confirmed these results in human skin fibroblasts, COS-7 cells (monkey kidney fibroblast line), the human glioblastoma cell line U118, and freshly prepared rat cardiomyocytes (<xref ref-type="fig" rid="fig7">Figure 7</xref>). Our data are in agreement with readthrough-dependent targeting of about 1.6% of the LDHB to peroxisomes mediated by the cryptic PTS1 in the extension. Remarkably, treatment of untransfected wild-type HeLa cells with geneticin increased LDHBx levels in the peroxisome (induction factor 1.89, n = 28, <italic>t</italic> test p < 0.0001) suggesting elevated peroxisomal LDHBx levels as a general pharmacological consequence of aminoglycoside treatment.<fig id="fig7" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.018</object-id><label>Figure 7.</label><caption><title>Endogenous LDHB is localized to peroxisomes in wild-type cells.</title><p>Immunofluorescence in wild-type cultured cells (<bold>A</bold>–<bold>F</bold>) or freshly prepared (<bold>G</bold> and <bold>H</bold>) cells with antibodies recognizing LDHB (secondary antibody Alexa488-coupled) and the peroxisome marker PEX14 (secondary antibody Cy3-coupled). (<bold>A</bold> and <bold>B</bold>) COS-7 cells, (<bold>C</bold> and <bold>D</bold>) human skin fibroblasts, (<bold>E</bold> and <bold>F</bold>) U118 glioblastome, and (<bold>G</bold> and <bold>H</bold>) primary rat cardiomyocytes. (<bold>B</bold>, <bold>D</bold>, <bold>F</bold>, <bold>H</bold>) Cytosol was removed after permeabilization with digitonin (-CYT). Bar 10 µm.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.018">http://dx.doi.org/10.7554/eLife.03640.018</ext-link></p></caption><graphic xlink:href="elife03640f007"/></fig></p><p>Next we wanted to test if there is evidence for differential regulation of translational readthrough of LDHB in different cell types. We expressed LDHB and mutant dual reporter constructs in COS-7 cells, U118 cells, and HEK cells. Readthrough of LDHB ranged between 1.55% (±0.09%) in HEK and HeLa and 1.88% (±0.14%) in COS-7. Surprisingly, in U118 cells LDHB readthrough is increased to 5.09% (±1.03%) (<xref ref-type="fig" rid="fig8">Figure 8</xref>). Geneticin induced readthrough by factors ranging between 1.32 (±0.09) and 2.82 (±0.27) (<xref ref-type="fig" rid="fig8">Figure 8</xref>). LDHB stop suppression is thus not restricted to special tissues, and may be differently regulated in different cell types.<fig id="fig8" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.019</object-id><label>Figure 8.</label><caption><title>Evidence for regulation of readthrough.</title><p>LDHB stop codon readthrough in various mammalian cell types. COS-7, HEK, and U118 cells were transfected with LDHB and mutant dual reporter constructs and analyzed by Venus fluorescence and luciferase assays. Readthrough is expressed as hRLuc/Venus signal. Readthrough is induced by 100 µg/ml geneticin.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.019">http://dx.doi.org/10.7554/eLife.03640.019</ext-link></p></caption><graphic xlink:href="elife03640f008"/></fig></p><p>Analysis of animal LDHB orthologs in vertebrates shows that PTS1 in the extension is exclusively and strictly conserved in mammals, supporting the notion of a functional extension in these proteins and an evolutionarily conserved targeting of LDHBx to peroxisomes in mammals (<xref ref-type="fig" rid="fig9">Figure 9</xref>).<fig id="fig9" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.020</object-id><label>Figure 9.</label><caption><title>LDHBx extensions including hidden PTS1 are strictly conserved in mammals.</title><p>Alignments of LDHBx termini from mammals and non-mammalian vertebrates. PTS1 extension is boxed. The conserved readthrough PTS1 extension is found exclusively in mammals and marks the mammalian–non-mammalian border in vertebrates. <italic>Alligator mi.</italic>: <italic>Alligator mississippiensis</italic>; <italic>Canis lupus famil.: Canis lupus familiaris</italic>; <italic>Meleagris gall.</italic>: <italic>Meleagris gallopavo</italic>; <italic>Oryctolagus cun.</italic>: <italic>Oryctolagus cuniculus</italic>; <italic>Taeniopygia gut.: Taeniopygia guttata</italic>.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.020">http://dx.doi.org/10.7554/eLife.03640.020</ext-link></p></caption><graphic xlink:href="elife03640f009"/></fig></p></sec><sec id="s2-6"><title>Piggy-back co-import of LDHA with LDHB</title><p>LDHB together with lactate dehydrogenase A (LDHA) can form five tetrameric LDH isoforms, of which two are homotetramers and three are heterotetramers (<xref ref-type="bibr" rid="bib5">Boyer et al., 1963</xref>; <xref ref-type="bibr" rid="bib37">Markert, 1963</xref>), and peroxisomes have the unusual ability to import folded and even oligomeric proteins (<xref ref-type="bibr" rid="bib42">McNew and Goodman, 1996</xref>; <xref ref-type="bibr" rid="bib31">Lanyon-Hogg et al., 2010</xref>). We therefore wanted to test if peroxisomal LDHBx piggy-backs LDHA into peroxisomes. For this purpose we adapted a two-hybrid assay previously used to analyze co-import of subunits of the dimeric peroxisomal hydrolase Lpx1 in a heterologous system (<xref ref-type="bibr" rid="bib53">Thoms et al., 2011</xref>). When LDHA was expressed as a fusion protein with N-terminal YFP without co-expression of any form of LDHB, the protein localized to the cytosol as expected (<xref ref-type="fig" rid="fig10">Figure 10A</xref>). However, when we co-expressed YFP-LDHA with CFP-LDHBx[TGG], that is cyan fluorescent protein (CFP) fused to the readthrough form of LDHB, we found YFP-LDHA in peroxisomes (<xref ref-type="fig" rid="fig10">Figure 10B</xref>). This experiment shows that the readthrough form of LDHB, LDHBx, can interact with LDHA, and that LDHBx is capable of carrying LDHA into the peroxisome. To show that co-import of LDHA is dependent on the hidden targeting signal in LDHBx, we mutated the targeting signal to SSI, or we deleted the terminal leucine. Either LDHBx PTS1 mutation blocked co-import of LDHA (<xref ref-type="fig" rid="fig10s1">Figure 10—figure supplement 1</xref>). The peroxisome is thus accessible to all four new LDH isoforms containing LDHBx. To support our data on LDHBx-LDHA co-import, we drew a structural model of the LDH-1 tetramer, the fundamental all-B isoform of LDH (<xref ref-type="fig" rid="fig10s2">Figure 10—figure supplement 2</xref>). The C-terminal amino acid leucine is extended by three amino acids not resolved in the structure, and, in LDHBx, by an additional seven amino acids. The model shows that this extension protrudes from the tetramer and is located distal to the protomer-interaction site, confirming that oligomerization is not hampered by the extension. The protruding LDHBx extension carrying the PTS1 is also accessible on the tetramer surface for PEX5 binding and import into the peroxisome.<fig-group><fig id="fig10" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.021</object-id><label>Figure 10.</label><caption><title>Piggy-back co-import of LDHA by LDHBx into peroxisomes.</title><p>Direct fluorescence of YFP-labeled LDHA (green) in the absence or presence of CFP-labeled LDHBx[TGG] (red) combined with immunofluorescence with a peroxisome marker (blue). (<bold>A</bold>) YFP-LDHA localization is mainly in the cytosol when expressed in the absence of LDHBx. (<bold>B</bold>) LDHA is imported into peroxisomes when co-expressed with LDHBx[TGG]. Cytosol was removed after permeabilization with digitonin. Bar 10 µm.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.021">http://dx.doi.org/10.7554/eLife.03640.021</ext-link></p></caption><graphic xlink:href="elife03640f010"/></fig><fig id="fig10s1" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.03640.022</object-id><label>Figure 10—figure supplement 1.</label><caption><title>Mutation of the cryptic targeting signal SRL in LDHBx blocks co-import of LDHA into peroxisomes.</title><p>PTS1 mutated by deletion of the leucine (ΔL) or substitution of RL by SI (SSI) blocks co-import of LDHA into peroxisomes. (<bold>A</bold> and <bold>B</bold>) Co-expression of YFP-LDHA with CFP-LDHBx[TGG, ΔL]. (<bold>C</bold> and <bold>D</bold>) Co-expression of YFP-LDHA with CFP-LDHBx[TGG, SSI]. (<bold>B</bold> and <bold>D</bold>) Cytosol was removed after permeabilization with digitonin. Bar 10 µm.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.022">http://dx.doi.org/10.7554/eLife.03640.022</ext-link></p></caption><graphic xlink:href="elife03640fs008"/></fig><fig id="fig10s2" position="float" specific-use="child-fig"><object-id pub-id-type="doi">10.7554/eLife.03640.023</object-id><label>Figure 10—figure supplement 2.</label><caption><title>Tetrameric lactate dehydrogenase (space fill model) from human heart (LDH-1, all-B isoform).</title><p>The individual subunits are shown in different colors, and the last resolved amino acid at the C-terminus (Leu in position 331) of all subunits is shown in yellow. This structural model shows that the termini protrude from the compact tetramer. The readthrough LDH contains at least one subunit LDHBx that is extended by seven amino acids containing the PTS1. For import into the peroxisome, at least one C-terminus has to bind to the soluble PTS1 receptor PEX5. The structure also shows that the added PTS1 is unlikely to block oligomerization of the protein because the C-termini are far away from the interaction surface of the protomers. The PTS1 has to be protruding from the compact oligomer in an unstructured manner to be buried in the TPR domain pocket of PEX5. Therefore, it is also unlikely that the PTS1 extensions generated by readthrough fold back onto the protein to induce a conformational change that would interfere with the subunit interaction. The structural model is derived from structure 1IOZ (<xref ref-type="bibr" rid="bib47">Read et al., 2001</xref>) in the International Protein Database (<ext-link ext-link-type="uri" xlink:href="http://www.pdb.org/">www.pdb.org</ext-link><italic>)</italic> and was rendered using Jmol, an open-source Java viewer for chemical structures in 3D (<ext-link ext-link-type="uri" xlink:href="http://www.jmol.org/">http://www.jmol.org/</ext-link>).</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.023">http://dx.doi.org/10.7554/eLife.03640.023</ext-link></p></caption><graphic xlink:href="elife03640fs009"/></fig></fig-group></p></sec></sec><sec id="s3" sec-type="discussion"><title>Discussion</title><p>The study of translational readthrough goes back to the origins of molecular biology, but mammalian genes undergoing readthrough have only recently come into focus and are being identified by systemic approaches (<xref ref-type="bibr" rid="bib27">Jungreis et al., 2011</xref>; <xref ref-type="bibr" rid="bib10">Dunn et al., 2013</xref>; <xref ref-type="bibr" rid="bib11">Eswarappa et al., 2014</xref>; <xref ref-type="bibr" rid="bib36">Loughran et al., 2014</xref>). Translational readthrough can be controlled by cis-acting elements, RNA structures of the transcript, that, often mediated by trans-factors, influence the termination process (<xref ref-type="bibr" rid="bib14">Firth et al., 2011</xref>; <xref ref-type="bibr" rid="bib11">Eswarappa et al., 2014</xref>). This mechanism has been termed programmed translational readthrough (PTR) (<xref ref-type="bibr" rid="bib11">Eswarappa et al., 2014</xref>). It is known, however, that the stop codon together with the preceding and immediately following nucleotides (SCC) also influence translational readthrough. We have termed this process basal translational readthrough (BTR) to distinguish it from PTR in general, and also from pharmacologically induced readthrough. In this study we derive a motif conferring high BTR from a linear regression model of SCCs and show that LDHBx undergoes BTR, which in turn affects the intracellular distribution of LDH.</p><sec id="s3-1"><title>A new LDH subunit</title><p>LDH is an enzyme with several isoforms, which has also been instrumental in devising the enzyme isoform concept per se. The identification of the classic muscle and heart subunits LDH-M (LDHA) and LDH-H (LDHB) in the late 1950s was followed by the identification of a testes-specific LDHA variant, LDHC (<xref ref-type="bibr" rid="bib5">Boyer et al., 1963</xref>; <xref ref-type="bibr" rid="bib20">Goldberg et al., 2009</xref>). Now we find that readthrough-extended LDHBx is encoded by the well-known <italic>LDHB</italic> gene by translational stop suppression and can give rise to new isoforms. Peroxisomal LDH is a novel isoform of LDH containing at least one readthrough-extended LDHBx subunit. LDHB readthrough and readthrough-dependent peroxisomal localization are evident in various human cell types, suggesting that the LDHBx subunit is expressed and localized to peroxisomes in all tissues that express LDHB. LDHBx exemplifies a new mechanism of post-transcriptional diversification of the genome's coding potential in mammals.</p><p>The 1.6% LDHBx stop codon readthrough that we find in our experiments corresponds to the 1.5–2% LDH activity found in association with peroxisomes (<xref ref-type="bibr" rid="bib41">McGroarty et al., 1974</xref>; <xref ref-type="bibr" rid="bib45">Osmundsen, 1982</xref>; <xref ref-type="bibr" rid="bib2">Baumgart et al., 1996</xref>), suggesting that cellular suppression of the stop codon is the only pathway for LDHB into peroxisomes. Assuming that peroxisomes fill approximately 1–2% of the cell volume, translational readthrough ensures almost equal concentration of LDH in cytosol and in peroxisomes.</p></sec><sec id="s3-2"><title>A role for peroxisomal LDH</title><p>Fatty acid β-oxidation reactions are the hallmark of peroxisomes in most cell types and organisms. In mammalian peroxisomes, β-oxidation is involved in the degradation of very long chain fatty acids (VLCFA) and biogenetic reactions such as the synthesis of bile acids (<xref ref-type="bibr" rid="bib35">Lodhi and Semenkovich, 2014</xref>). Therefore patients with peroxisomal disorders accumulate VLCFA and bile acid intermediates (<xref ref-type="bibr" rid="bib6">Braverman et al., 2013</xref>). During fatty acid oxidation and other peroxisomal processes, nicotinamide adenine dinucleotide (NAD<sup>+</sup>) is reduced to NADH. However, the pathway of NAD<sup>+</sup> regeneration inside peroxisomes is not clear (<xref ref-type="bibr" rid="bib30">Kunze and Hartig, 2013</xref>). For efficient β-oxidation to occur, it is necessary that a redox shuttle system exists for NAD<sup>+</sup> regeneration, because peroxisomes are impermeable to NAD<sup>+</sup>/NADH (<xref ref-type="bibr" rid="bib55">Visser et al., 2007</xref>). The identification of LDH inside the peroxisome suggested the existence of a lactate/pyruvate shuttle involved in the regeneration of redox equivalents (<xref ref-type="bibr" rid="bib2">Baumgart et al., 1996</xref>; <xref ref-type="bibr" rid="bib40">McClelland et al., 2003</xref>; <xref ref-type="bibr" rid="bib19">Gladden, 2004</xref>). In the absence of a peroxisomal targeting signal, however, peroxisomal LDH was not universally accepted by researchers.</p><p>Lactate/pyruvate shuttling could either occur directly through the peroxisomal membrane (<xref ref-type="bibr" rid="bib55">Visser et al., 2007</xref>) or make use of monocarboxylate shuttles in the peroxisomal membrane (<xref ref-type="bibr" rid="bib40">McClelland et al., 2003</xref>). Generally, functional LDHBx targeting to peroxisomes highlights the role of intracellular lactate shuttle mechanisms (<xref ref-type="bibr" rid="bib8">Brooks, 2009</xref>). In liver peroxisomes, pyruvate production is catalyzed by alanine-glyoxylate aminotransferase, an important enzyme in glyoxylate detoxification. Glyoxylate, however, is itself a substrate of LDH (<xref ref-type="bibr" rid="bib48">Salido et al., 2012</xref>). Therefore, peroxisomal LDH may also be involved in peroxisomal glyoxylate metabolism.</p><p>Peroxisomal LDH is not the first glycolytic enzyme found in peroxisomes. Trypanosomes have sequestered the full set of glycolytic enzymes in specialized peroxisomes called glycosomes (<xref ref-type="bibr" rid="bib22">Gualdrón-López et al., 2012</xref>). And recently, in fungi, part of the glycolytic pathway upstream of pyruvate including glyceraldehyde-3-phosphate dehydrogenase and 3-phosphoglycerate kinase, was shown to be localized to peroxisomes by alternative splicing and/or translational readthrough (<xref ref-type="bibr" rid="bib17">Freitag et al., 2012</xref>). It is compelling that fungi as well as mammals use stop codon suppression to localize a small fraction of glycolytic enzymes to peroxisomes. We hypothesize that both translational readthrough as well as PTS1 evolve easily, and so can divert a low and steady amount of these enzymes to peroxisomes.</p><p>A small fraction of cytosolic LDHB is imported into peroxisomes. This fraction is likely to be constant with respect to the overall LDHB expression levels in given tissue. We speculate that the peroxisomal LDHB shunt helps to coordinate redox processes between the cytosol and the peroxisome. Importantly, our study reveals a new pharmacological effect of readthrough-inducing drugs such as the commonly prescribed aminoglycosides, as they will increase LDHB readthrough and peroxisome import of LDHBx.</p><p>It is not known at the moment whether translational readthrough is regulated in humans. The very high readthrough of approximately 5% in a glioblastoma cell line suggests that readthrough is differentially regulated in different tissues. Future experiments will show if the increased LDHB readthrough we find in this cell line are a cancer-associated dysregulation linked to the Warburg effect (<xref ref-type="bibr" rid="bib25">Hsu and Sabatini, 2008</xref>), or if it just matches a higher abundance of peroxisomes in these cells to ensure an equal concentration of LDH in cytosol and peroxisomes in these cells as suggested above. It is also possible that glial cells generally have a higher demand for peroxisomal LDH that could be involved in neuronal/glial lactate metabolism.</p></sec><sec id="s3-3"><title>A rational approach to translational readthrough</title><p>The first mammalian readthrough proteins were identified by chance (<xref ref-type="bibr" rid="bib18">Geller and Rich, 1980</xref>; <xref ref-type="bibr" rid="bib9">Chittum et al., 1998</xref>; <xref ref-type="bibr" rid="bib59">Yamaguchi et al., 2012</xref>). Recently, two powerful and complementary methods have been employed in the genome-wide identification of readthrough-extended proteins. Ribosome profiling can recognize translating ribosomes in 3′UTRs and thereby identify readthrough and other recoding events outside known coding regions (<xref ref-type="bibr" rid="bib26">Ingolia et al., 2011</xref>; <xref ref-type="bibr" rid="bib10">Dunn et al., 2013</xref>). Phylogenetic approaches such as those implemented in PhyloCSF (<xref ref-type="bibr" rid="bib33">Lin et al., 2011</xref>) evaluate the coding potential of sequences before and after the stop codon to help predict readthrough and are particularly powerful when genome sequences from closely related species are available (<xref ref-type="bibr" rid="bib27">Jungreis et al., 2011</xref>; <xref ref-type="bibr" rid="bib36">Loughran et al., 2014</xref>). Ribosome profiling, however, depends on gene expression, and can identify readthrough events only when the cell type in question is actually analyzed. Ribosome profiling may also fail to identify short readthrough extensions. Phylogenetic approaches, on the other hand, may miss readthrough when it is not conserved in the given dataset or when sufficiently dense datasets are not available, or when the extensions are too short to provide a basis for phylogenetic comparison.</p><p>Our approach to systems-level identification of translational readthrough is based on the formalization of SCCs and a linear regression model with experimental readthrough values. The majority of the input sequences have been derived from patient nonsense mutations. In consequence, these sequences are biased neither by preselection by any pre-determined RTP or experimental readthrough levels, or by the SCCs, because the contexts did not evolve together with the respective stop codons. The algorithm we develop in this paper is limited to six nucleotide positions before and after the stop codon. This approach excludes the identification of extended RNA secondary structures involved in PTR and other recoding events (<xref ref-type="bibr" rid="bib1">Baranov et al., 2002</xref>; <xref ref-type="bibr" rid="bib13">Firth and Brierley, 2012</xref>; <xref ref-type="bibr" rid="bib11">Eswarappa et al., 2014</xref>; <xref ref-type="bibr" rid="bib36">Loughran et al., 2014</xref>). The identification of the LINfs3 consensus and the human genes associated with this consensus justifies this approach. The LINfs3 motif, derived by feature selection, encompasses the stop codon and the first codon after the stop: UGA CUA. Our analysis suggests that positions +7 and −6 might also further contribute to readthrough. We have tested five of the 144 candidates in the genome with the UGA CUA motif and confirmed their high BTR. Highest BTR appears to correlate with a G in position +7 (UGA CUA G) within the LINfs5 consensus. This motif is found 30 times in the human genome and has recently been shown to support high translational readthrough (<xref ref-type="bibr" rid="bib36">Loughran et al., 2014</xref>). The motifs for high BTR are distinct from the consensus UGA CAR YYA (R = A/G, Y = C/U) found in some viruses and yeast (<xref ref-type="bibr" rid="bib43">Namy et al., 2001</xref>; <xref ref-type="bibr" rid="bib23">Harrell et al., 2002</xref>) but resembles the alphavirus-like high readthrough stop codon context (<xref ref-type="bibr" rid="bib32">Li and Rice, 1993</xref>). Interestingly, the same stop suppression context in the <italic>LAMA3</italic> gene has been shown to alleviate the disease severity of an otherwise fatal nonsense mutation in a patient with junctional epidermolysis bullosa, the major and most devastating form of epidermolysis bullosa (<xref ref-type="bibr" rid="bib46">Pacho et al., 2011</xref>).</p><p>The existence of the consensus motif UGA CUA is the origin of the non-linear contribution to RTP in our models. This is supported by the finding that correlation of BTR and RTP for LINfs3 is higher than for the LINiter model, so that the reduced number of parameters in LINfs3 provides a better model fit. This finding implies that with the currently small dataset, compact linear models should be preferred over non-linear models with many parameters. The identification of the few relevant nucleotide positions will help to create datasets with fully specified BTR for a wide range of SCCs and cell types. A larger training set of sequences with verified readthrough rates will allow the development of non-linear approximation models.</p><p>LDHBx shows an unusually high readthrough of 1.6%, and its stop context <underline>UGA</underline> CUA G (stop codon underlined) matches the LINfs3 consensus. The 18-nucleotide extension in LDHBx is unlikely to contain a extensive secondary structure that would suggest a combined effect of BTR and PTR. The identification of LDHBx and the recently discovered readthrough form of vascular endothelial growth factor A, VEGF-Ax (<xref ref-type="bibr" rid="bib11">Eswarappa et al., 2014</xref>), thus mark two extreme and separable cases of physiological stop suppression: LDHBx appears independent of cis-factors beyond the SCC and marks a prototypical example of BTR. In contrast, the readthrough of VEGF-Ax is relatively independent of its SCC but instead requires a more distantly located cis-element (<xref ref-type="bibr" rid="bib11">Eswarappa et al., 2014</xref>). The distinction between PTR and BTR, however, is not exclusive. A thorough analysis of readthrough in OPRK1 and OPRL1 indicates that readthrough levels of more than 30% can be obtained by a combination of cis-elements and UGA CUA-based BTR (<xref ref-type="bibr" rid="bib36">Loughran et al., 2014</xref>).</p><p>The era of systematic analysis of translational readthrough in humans is only beginning. We expect that a combination of in silico modeling and screening, ribosome profiling, phylogenetic methods, and mass spectrometry will help to identify the ‘extensome’, the complete set of readthrough-extended proteins in mammals.</p></sec></sec><sec id="s4" sec-type="materials|methods"><title>Materials and methods</title><sec id="s4-1"><title>RTP calculation algorithm</title><p>To predict the RTP of gene transcripts, we developed a linear regression model based on the SCCs and their experimentally determined basal readthrough values. The SCC comprises the stop codon itself (positions +1 to +3) and the nucleotide sequences surrounding the stop codon (−6 to +9). For the first-pass model (LIN), we re-analyzed 66 SCCs with known experimental basal readthrough values (<xref ref-type="bibr" rid="bib16">Floquet et al., 2012</xref>). The stop codons evolved independently of their contexts (<xref ref-type="table" rid="tbl3">Table 3</xref>). Nucleotide sequences were represented by indicator vector coding. Here, 12 × 4 binary vector entries are used to indicate the presence [1] or absence [0] of a nucleotide (A, C, G, or U) at a particular position (−6 to −1, +4 to +9) surrounding the stop codon. Three further entries are reserved to indicate the type of stop codon (UAA, UAG, or UGA, positions +1, +2, or +3). The resulting feature vectors of all sequences were normalized to Euclidean unit length.<table-wrap id="tbl3" position="float"><object-id pub-id-type="doi">10.7554/eLife.03640.024</object-id><label>Table 3.</label><caption><p>Nucleotide frequencies in each position of the stop codon context</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.024">http://dx.doi.org/10.7554/eLife.03640.024</ext-link></p></caption><table frame="hsides" rules="groups"><thead><tr><th align="center">Nucleotide</th><th align="center">A</th><th align="center">C</th><th align="center">G</th><th align="center">U</th></tr></thead><tbody><tr><td align="center">Position</td><td/><td/><td/><td/></tr><tr><td align="center">−6</td><td align="center">0.2892</td><td align="center">0.2530</td><td align="center">0.2651</td><td align="center">0.1928</td></tr><tr><td align="center">−5</td><td align="center">0.3253</td><td align="center">0.2651</td><td align="center">0.1446</td><td align="center">0.2651</td></tr><tr><td align="center">−4</td><td align="center">0.1566</td><td align="center">0.2289</td><td align="center">0.3494</td><td align="center">0.2651</td></tr><tr><td align="center">−3</td><td align="center">0.2410</td><td align="center">0.3373</td><td align="center">0.2410</td><td align="center">0.1807</td></tr><tr><td align="center">−2</td><td align="center">0.2651</td><td align="center">0.1807</td><td align="center">0.2048</td><td align="center">0.3494</td></tr><tr><td align="center">−1</td><td align="center">0.2410</td><td align="center">0.2530</td><td align="center">0.2651</td><td align="center">0.2410</td></tr><tr><td align="center">4</td><td align="center">0.2289</td><td align="center">0.3133</td><td align="center">0.3373</td><td align="center">0.1205</td></tr><tr><td align="center">5</td><td align="center">0.2651</td><td align="center">0.2530</td><td align="center">0.1446</td><td align="center">0.3373</td></tr><tr><td align="center">6</td><td align="center">0.2771</td><td align="center">0.2169</td><td align="center">0.2530</td><td align="center">0.2530</td></tr><tr><td align="center">7</td><td align="center">0.2530</td><td align="center">0.3133</td><td align="center">0.2892</td><td align="center">0.1446</td></tr><tr><td align="center">8</td><td align="center">0.3253</td><td align="center">0.1687</td><td align="center">0.2410</td><td align="center">0.2651</td></tr><tr><td align="center">9</td><td align="center">0.1807</td><td align="center">0.2771</td><td align="center">0.2771</td><td align="center">0.2651</td></tr><tr><td align="center">Stop codons</td><td align="center">UAA</td><td align="center">UAG</td><td align="center">UGA</td><td/></tr><tr><td align="center">1 to 3</td><td align="center">0.1928</td><td align="center">0.3373</td><td align="center">0.4699</td><td/></tr></tbody></table><table-wrap-foot><fn><p>The nucleotide and stop codon frequencies for positions −6 to −1 and 4 to 9 were calculated for the 81 sequences used in the RTP predictor (LINiter model).</p></fn></table-wrap-foot></table-wrap></p><p>For the estimation of the regression model coefficients, we performed a regularized least-squares (‘rigde’) regression (<xref ref-type="bibr" rid="bib24">Hoerl and Kennard, 1970</xref>). Let <bold>X</bold> be the <italic>n</italic> × <italic>d</italic> matrix of <italic>n</italic> sequence feature vectors with dimensionality <italic>d</italic> and <bold>y</bold> be the (<italic>n</italic>-dimensional) vector of readthrough values associated with the sequences. Then the weight vector <bold>w</bold> = (<bold>X</bold><sup>T</sup><bold>X</bold> + <italic>k</italic> × <bold>I</bold>)<sup>−1</sup> × <bold>X</bold><sup>T</sup><bold>y</bold> represents the solution of the linear least-squares problem and <italic>y</italic> = <bold>w</bold><sup>T</sup><bold>x</bold> corresponds to the RTP value <italic>y</italic> for a sequence feature vector <bold>x</bold>. To evaluate the influence of the regularization parameter <italic>k</italic>, we performed a leave-one-out cross-validation (loo-cv) with <italic>k</italic> = {10<sup>i</sup>|i = −3, −2.7,...0,...,3} for all model types. The minimum loo-cv error in terms of the sum of squared deviations of predictions from known readthrough values was 4.75 × 10<sup>−7</sup> for <italic>k</italic> = 10<sup>0.3</sup> (approximately 1.995).</p><p>For genome-wide prediction of readthrough propensities for human transcripts, we downloaded all 215,621 coding sequences from the Ensembl BioMart (<xref ref-type="bibr" rid="bib15">Flicek et al., 2012</xref>) using the Homo sapiens Genes v74 section (GRCh37.p13) plus 300 nucleotides downstream of the CDS end (<ext-link ext-link-type="uri" xlink:href="http://ensembl.org">ensembl.org</ext-link>, November 2013). Transcripts corresponding to identical protein products, short sequences (<15aa protein-coding) and incomplete (e.g., missing or mislocated stop codon) or insufficiently sequenced (i.e., undetermined nucleotides) DNA were removed. Sequences with identical 3′/C-termini (nucleotide positions −45 to +303) were aggregated to one representative sequence, resulting in 42,069 unique transcripts. ORF extensions were identified by detection of an in-frame stop codon within 300 nucleotides downstream of the annotated stop codon.</p></sec><sec id="s4-2"><title>Iterative model refinement and feature selection</title><p>To obtain a more comprehensive model for RTP prediction, we included 15 sequences and their corresponding experimentally determined readthrough values from this study in the prediction model (see <xref ref-type="bibr" rid="bib49">Schueren et al., 2014</xref> for Dataset 1). The regression coefficients for the iterative model considering all 12 stop context positions (LINiter) were computed as described in the previous section. The minimum regression error was 6.24 × 10<sup>−6</sup> at <italic>k</italic> = 10<sup>0.3</sup>. A sequence logo representation of the regression coefficients for this model is displayed in <xref ref-type="fig" rid="fig2">Figure 2A</xref>. The sequence logo was created using the enoLOGOS web server (<xref ref-type="bibr" rid="bib58">Workman et al., 2005</xref>).</p><p>Furthermore, we evaluated reduced model sizes by stepwise elimination of context positions carrying no or little information for RTP prediction (feature selection). Starting from the complete mode (LIN), we removed the position corresponding to the minimum sum of squared regression coefficients. Regression error and coefficients were then calculated for the remaining positions (including the stop codon) as described above. This procedure was repeated until only the stop codon position was left. <xref ref-type="fig" rid="fig2">Figure 2B</xref> shows the development of the regression error for reduced model sizes by stepwise elimination of positions. Here, a first local minimum can be identified for model LINfs5 with five positions remaining (−6, stop, +4 to +7) and the global minimum corresponds to model LINfs3 with three positions besides the stop codon (stop, +4 to +6).</p></sec><sec id="s4-3"><title>PTS1 prediction algorithm</title><p>To identify cryptic peroxisomal localization signals in readthrough extensions, we adapted a peroxisomal targeting signal type 1 (PTS1) detection algorithm that was previously developed for plant proteins (<xref ref-type="bibr" rid="bib34">Lingner et al., 2011</xref>). For this purpose, we used 24 known human PTS1 proteins (ACOT4, ACOX1, ACOX2, ACOX3, AGXT, AMACR, BAAT, CRAT, DAO, EHHADH, GNPAT, HAO1, HAO2, HSD17B4, IDE, MLYCD, PRDX5, ACOT8, CROT, PECI, ECH1, LONP2, PECR, and PIPOX) and performed orthology searches on metazoan protein and EST sequences using a bidirectional best BLAST hit strategy. Starting from each human protein sequence, we identified significant BLAST hits (e-value < 10<sup>−10</sup>) to metazoan sequences within the ‘nr’ and ‘dbEST’ database. Then, the best hit of each organism was searched against the human proteome and sequences not re-identifying the starting sequence were removed. Afterwards, the starting sequences and putative orthologs were pooled and sequences with uncommon PTS1 tripeptides, that is tripeptides which occurred less than three times, were removed from the set. The resulting set of sequences was used as positive examples for training machine learning models as previously published (<xref ref-type="bibr" rid="bib34">Lingner et al., 2011</xref>). Briefly, a regularized least-square classification algorithm was trained using indicator vector representations of up to 15 C-terminal amino acids of positive and negative example sequences. A set of negative example sequences was created by extracting all metazoan sequences without peroxisomal association from the Swiss-Prot section of UniProt (<ext-link ext-link-type="uri" xlink:href="http://www.uniprot.org/">http://www.uniprot.org</ext-link>) in November 2011. The best model (15 C-terminal amino acids) was determined by fivefold cross-validation and yielded a prediction accuracy of 0.996 and 0.863 in terms of the area under the receiver operating characteristic (ROC) curve (auROC) and the area under the precision/recall curve (auPRC), respectively. When a stop codon was considered in the PTS1 prediction, the stop codon was scored as an undefined amino acid (‘X’) without a contribution to the PTS1 posterior probability.</p></sec><sec id="s4-4"><title>Multiple alignment analysis</title><p>The multiple alignment of genomic sequences for the LDHB SCC (position −36 to +48) was downloaded from the Ensembl database (<ext-link ext-link-type="uri" xlink:href="http://www.ensembl.org">www.ensembl.org</ext-link>) in November 2013. The ‘21 amniota vertebrates’ alignment was used and split into mammalian and non-mammalian species. Sequences without residues in the extension region were deleted and the non-mammalian alignment was augmented by LDHB sequences from the NCBI nucleotide database (<ext-link ext-link-type="uri" xlink:href="http://www.ncbi.nlm.nih.gov/nuccore">http://www.ncbi.nlm.nih.gov/nuccore</ext-link>) in November 2013. In total, the alignments comprise 13 mammals and nine non-mammalian vertebrates: <italic>Homo sapiens</italic> (human), <italic>Mus musculus</italic> (mouse), <italic>Rattus norvegicus</italic> (rat), <italic>Oryctolagus cuniculus</italic> (rabbit), <italic>Pan troglodytes</italic> (chimpanzee), <italic>Gorilla gorilla</italic> (gorilla), <italic>Pongo abelii</italic> (orangutan), <italic>Macaca mulatta</italic> (rhesus macaque), <italic>Felis catus</italic> (cat), <italic>Canis familiaris</italic> (dog), <italic>Equus caballus</italic> (horse), <italic>Bos taurus</italic> (cow), <italic>Ovis aries</italic> (sheep), <italic>Xenopus tropicalis</italic> (western clawed frog), <italic>Anolis carolinensis</italic> (anole lizard), <italic>Ficedula albicollis</italic> (flycatcher), <italic>Taeniopygia guttata</italic> (zebra finch), <italic>Gallus gallus</italic> (chicken), <italic>Meleagris gallopavo</italic> (turkey), <italic>Alligator mississippiensis</italic> (American alligator), <italic>Salmo salar</italic> (salmon), and <italic>Danio rerio</italic> (zebrafish).</p><p>The genomic sequences were translated into amino acid sequences using the ‘EMBOSS Transeq’ web server (<ext-link ext-link-type="uri" xlink:href="http://www.ebi.ac.uk/Tools/st/emboss_transeq/">http://www.ebi.ac.uk/Tools/st/emboss_transeq/</ext-link>). Species trees were obtained from the Interactive Tree Of Life (iTOL) website (<ext-link ext-link-type="uri" xlink:href="http://itol.embl.de/">http://itol.embl.de/</ext-link>) and visualized with the Phylip package (<xref ref-type="bibr" rid="bib12">Felsenstein, 1989</xref>). JalView software (<xref ref-type="bibr" rid="bib57">Waterhouse et al., 2009</xref>) was used to visualize the alignments and to compute alignment quality and consensus. Here, the quality score of an alignment column is inversely proportional to the average cost of all pairs of mutations in terms of BLOSUM 62 substitution scores and the consensus reflects the fraction of the most frequent residue for each column of the alignment.</p></sec><sec id="s4-5"><title>DNA cloning</title><p>Plasmids used in this study are listed in the table in <xref ref-type="supplementary-material" rid="SD1-data">Supplementary file 1</xref>. Oligonucleotides used in this study are listed in the table in <xref ref-type="supplementary-material" rid="SD2-data">Supplementary file 2</xref>.</p><p>The dual reporter vector pDRVL (PST1360) encoding an N-terminal Venus tag and a C-terminal hRluc tag was derived from pEXP-Venus-hRluc (a gift from Ania Muntau and Sören Gersting) by introducing a short multicloning site (MCS) containing BstEII, ClaI, BspEI, and BsiWI restriction sites. pDRVL was created by ligating pre-annealed oligonucleotides OST963 and OST964 into the XhoI site of pEXP Venus-hRluc. Dual reporter constructs PST1384–1385, 1387, 1393–1396, 1418–1426, 1430, 1435, 1437, 1493, 1494, 1497, 1504, and PST1444 were derived from pDRVL by insertion of pre-annealed oligonucleotides OST1081–1084, 1086–1087, 1117–1124, 1144–1145, 1148–1157, 1160–1165, 1158–1159, 1190–1191, 1198–1199, 1229–1230, JH59–60, JH61–62, JH67–68, and JH81–82 into BspEI and BstEII sites, as listed in <xref ref-type="supplementary-material" rid="SD2-data">Supplementary file 2</xref>.</p><p>For cloning of pEYFP-LDHBx (PST1388), the LDHB open reading frame including the stop codon and the 18-nucleotide 3′ extension, was PCR-amplified from pOTB7-LDHB using primers OST1053 and 1054 and inserted into EcoRI and XbaI sites of pEYFP-C1.</p><p>The stop codon variants pEYFP-LDHBx[TGG] (PST1389), pECFP-LDHBx[TGG] (PST1440), pEYFP-LDHBx[TAA] (PST1410), pEYFP-LDHBx[TAAT] (PST1411), and pEYFP-LDHBx[TGAT] (PST1409) were created by amplifying LDHBx using primer OST1053 with reverse primers OST1055, 1127, 1128, and 1129, respectively. Similarly, the PTS1 mutation variants pEYFP-LDHBx[ΔL] (PST1407), pECFP-LDHBx[TGG, ΔL] (PST1512) (deletion of the last amino acid in the cryptic PTS1 SRL), and pEYFP-LDHBx[SSI] (PST1408), pECFP-LDHBx[TGG, SSI] (PST1513) (substitution of the PTS1 SRL by SSI) were created using forward primer OST1053 and reverse primers OST1125, 1263, 1126, and 1264, respectively. LDHA was amplified from human cDNA using primers OST1130 and 1131 and cloned into EcoRI and XbaI sites of pEYFP-C1 to yield pEYFP-LDHA (PST1434).</p><p>For cloning of pEXP Venus-PTS1 (PST1209), primers OST801 and 802 (encoding the PTS1 of ACOX3) were annealed and inserted into pENTR-TOPO-D. Then the PTS1 tag was transferred to pEXP-N-Venus using LR clonase II (Invitrogen, Carlsbad, California).</p><p>Full-length dual reporter constructs pcDNA3.1-HA-LDHBx-Myc and variants were cloned by amplifying LDHB and stop codon variants from PST1388 (LDHB wt), PST1389 (LDHB [TGG]), PST1409 (LDHB [TGAT]), PST1410 (LDHB [TAA]), and PST1411 (LDHB [TAAT]), using primers OST1202 and 1203 and cloning into NheI and BamHI restriction sites of pcDNA3.1/Myc-His(−)A. All plasmids were confirmed by DNA sequencing.</p></sec><sec id="s4-6"><title>Cell culture and transfection</title><p>HeLa cells and human skin fibroblasts were maintained in low glucose Dulbecco's minimal essential medium (DMEM), HEK cells, HT1080, U118, U373 and COS-7 cells in high glucose DMEM. Culture media were supplemented with 1% (wt/vol) glutamine, 5–10% (vol/vol) heat inactivated fetal calf serum (FCS), 100 units/ml penicillin, and 100 µg/ml streptomycin. For U118 cells, 1% non-essential amino acids and 1% pyruvate were added to the media.</p><p>Cells were transfected using Effectene transfection reagent (Qiagen, Germany) as described by the manufacturer. Plasmids were diluted in Buffer EC and Enhancer and incubated for 5 min at room temperature. Effectene was added and incubated for 10 min at room temperature. Prewarmed medium was added to the HeLa cells and to the transfection mixture which was then added to cells and incubated at 37°C in a humidified 5% CO<sub>2</sub> incubator for 24 hr. Then, 6 hr after transfection, transfection reagent was removed, and, where indicated, geneticin (G418) was added at a concentration of 100 µg/ml.</p></sec><sec id="s4-7"><title>Dual reporter assays and readthrough calculation</title><p>Cells were washed with PBS and lysed by Renilla Luciferase Assay Lysis Buffer (Promega, Madison, Wisconsin) according to the manufacturer's manual. Cells were spun down (14 krpm, 2 min, 4°C) and supernatants were stored at −80°C. For Venus fluorescence measurement, cell lysates were diluted 1:25 in PBS and analyzed at 485 nm excitation, 530 nm emission (sensitivity: 130) using a Synergy Mx plate reader (Biotek, Winooski, Vermont). PBS was used as a blank control for fluorescence measurements.</p><p>Undiluted lysates (20 µl) were used to measure hRluc luminescence by the Renilla Luciferase Assay System (Promega) and the Synergy Mx plate reader (Biotek). An automated injector was used to add 100 µl Renilla Luciferase Assay Reagent. Luminescence was read 2 s after injection and integrated over 10 s (sensitivity: 150). Renilla Luciferase Assay Reagent was used as a blank control for hRluc luminescence measurements. Each construct was analyzed in three to seven biological replicates and each biological sample was measured in triplets.</p><p>To obtain readthrough rates, the ratio of hRluc/Venus fluorescence was calculated, and the readthrough of pDRVL was set to 100%. The ratio (y) and standard deviation of fluorescence (x<sub>1</sub>) and luminescence (x<sub>2</sub>) signal for each replicate were calculated using uncertainty propagation (σ<sub>y</sub> = [σ<sup>2</sup><sub>x1</sub> × (dy/dx<sub>1</sub>)<sup>2</sup> + σ<sup>2</sup><sub>x2</sub> × (dy/dx<sub>2</sub>)<sup>2</sup>]<sup>0.5</sup>). Let w<sub>i</sub> = 1/σ<sub>i</sub><sup>2</sup> be the weight of a readthrough value from replicate i with σ<sub>i</sub> being the error of the ratios. Then the weighted mean <bold>x</bold><sub><bold>m</bold></sub> of the replicates and its error σ<sub>xm</sub> were calculated according to <bold>x</bold><sub><bold>m</bold></sub> = (Σ<sub>i</sub>(x<sub>i</sub>w<sub>i</sub>)/Σ<sub>i</sub>w<sub>i</sub>) and σ<sub>xm</sub> = (Σ<sub>i</sub>w<sub>i</sub>)<sup>−0.5</sup>.</p></sec><sec id="s4-8"><title>Immunofluorescence, microscopy, and quantification</title><p>Transfected LDHB and LDHA fusion constructs were detected in HeLa cells by combined direct fluorescence and immunofluorescence experiments. Endogenous LDHB was analyzed in HeLa, U118, and COS-7 cells, and in primary rat cardiomyocytes by immunofluorescence. Approximately 1 × 10<sup>5</sup> cells were seeded on cover slips or on laminin-coated (Sigma, St. Louis, Missouri) glass slides for HEK cells and cardiomyocytes and transfected as indicated. For removal of cytosol, cells were treated with 0.02% (wt/vol) digitonin (Invitrogen) for 5 min at room temperature. Cells were fixed with 10% (wt/vol) formaldehyde for 20 min, and permeabilized with 0.5% Triton X-100 for 5 min. After blocking for 20 min at 37°C with 10% BSA, antigens were labeled with primary antibodies at 37°C for 1 hr. Antibody dilutions were 1:200 for anti-PEX14 rabbit polyclonal antibodies (ProteinTech, Chicago, Illinois) and 1:500 for anti-LDHB mouse monoclonal antibodies (Abnova, Taiwan). Secondary antibody labeling (1:200) was done for 1 hr with antibodies labeled with Cy3 and/or Alexa647 (Jackson Immuno Research, West Grove, Pennsylvania) and/or Alexa488 (MoBiTech, Germany). Cover slips were mounted with Mowiol containing 0.01 mg/ml 4′,6-diamidino-2-phenylindole (DAPI). DAPI was omitted in cases where cells had been transfected with CFP-expressing plasmids.</p><p>Fluorescence microscopy was done using a 100× oil objective (1.3 NA) with a Zeiss Imager M1 fluorescence wide field scope equipped with the Zeiss Axiocam HRm Camera and Zeiss Axiovision 4.8 acquisition software. z-Stacks with 30 images and 0.25 µm spacing were recorded and subjected to deconvolution. Where necessary, linear contrast enhancements were applied (Axiovision).</p><p>To quantify induction of endogenous LDHB by geneticin, fluorescence images from samples prepared with anti-LDHB and anti-PEX14 antibodies were recorded under identical conditions and subjected to deconvolution. The LDHB/PEX14 intensities were measured, and the same threshold ratios were applied to all channel pairs (ImageJ). Induction is expressed as the ratio of LDHB/PEX14 ratios with and without geneticin treatment, respectively.</p></sec><sec id="s4-9"><title>Western blot analysis</title><p>Cells were lysed in RIPA lysis buffer (20 mM Tris–HCl, pH 7.4, 150 mM sodium chloride, 2 mM EDTA, 1% NP40, 1 mM DTT, 0.1 mM PMSF, Complete protease inhibitors [Roche, Switzerland]) 24 hr after transfection. Proteins were separated by SDS-PAGE on a 12% gel, transferred to a nitrocellulose membrane, and probed with primary and secondary antibodies. The following antibodies were used: anti-HA rabbit polyclonal (Abcam, UK), anti-Myc mouse monoclonal (Cell Signaling, UK), anti-luciferase mouse monoclonal (Millipore), anti-GFP mouse monoclonal (Living Colors, Mountain View, California), and anti-actin mouse monoclonal (Sigma). HRP-conjugated goat anti-rabbit IgG and donkey anti-mouse IgG (Jackson Immuno Research) were used as secondary antibodies. We also used 1:1000 dilutions of primary antibody and 1:5000 dilutions of secondary antibody. Reactive bands were revealed with Lumi-light and Lumi-light plus Western blotting substrate (Roche). Images were scanned using Luminescent image analyzer LAS 4000.</p></sec><sec id="s4-10"><title>Data availability</title><p>Dataset 1. Spreadsheet containing predicted readthrough extensions, RTP scores (LIN, LINiter, LINfs5, LINfs3), PTS1 scores, predictions of ER retentions signals, glycosylation motifs, transmembrane domains, and transmembrane topology, and the LINiter<sup>+</sup> × PTS1 product scores for all human transcript termini. Publicly available at the Dryad Digital Repository with the <ext-link ext-link-type="uri" xlink:href="http://dx.doi.org/10.5061/dryad.j2n18">doi 10.5061/dryad.j2n18</ext-link> (<xref ref-type="bibr" rid="bib49">Schueren et al., 2014</xref>).</p></sec></sec></body><back><ack id="ack"><title>Acknowledgements</title><p>We thank Heiner Klingenberg for help with orthology searches of human PTS1 proteins, Ania Muntau and Sören Gersting for plasmids, and Kristina Gamper and Viacheslav Nikolaev for the rat cardiomyocytes. We are grateful to Ellen Krämer and Tanja Wilke for technical assistance, and to Cindy Krause, Peter Meinicke, Olaf Jahn, Johannes Freitag, and Michael Bölker for discussions. We thank Blanche Schwappach, Heinz Neumann, Heike Krebber, and Maya Schuldiner for comments on the manuscript.</p></ack><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>FS, Acquisition of data, Analysis and interpretation of data</p></fn><fn fn-type="con" id="con2"><p>RG, Acquisition of data, Analysis and interpretation of data</p></fn><fn fn-type="con" id="con3"><p>JH, Acquisition of data, Analysis and interpretation of data</p></fn><fn fn-type="con" id="con4"><p>CD, Acquisition of data, Analysis and interpretation of data</p></fn><fn fn-type="con" id="con5"><p>TL, Conception and design, Acquisition of data, Analysis and interpretation of data</p></fn><fn fn-type="con" id="con6"><p>JG, Drafting or revising the article</p></fn><fn fn-type="con" id="con7"><p>ST, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article</p></fn></fn-group></sec><sec sec-type="supplementary-material"><title>Additional files</title><supplementary-material id="SD1-data"><object-id pub-id-type="doi">10.7554/eLife.03640.025</object-id><label>Supplementary file 1.</label><caption><p>Plasmids used in this study.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.025">http://dx.doi.org/10.7554/eLife.03640.025</ext-link></p></caption><media mime-subtype="docx" mimetype="application" xlink:href="elife03640s001.docx"/></supplementary-material><supplementary-material id="SD2-data"><object-id pub-id-type="doi">10.7554/eLife.03640.026</object-id><label>Supplementary file 2.</label><caption><p>Oligonucleotides used in this study.</p><p><bold>DOI:</bold> <ext-link ext-link-type="doi" xlink:href="10.7554/eLife.03640.026">http://dx.doi.org/10.7554/eLife.03640.026</ext-link></p></caption><media mime-subtype="docx" mimetype="application" xlink:href="elife03640s002.docx"/></supplementary-material><sec sec-type="datasets"><title>Major dataset</title><p>The following dataset was generated</p><p><related-object content-type="generated-dataset" source-id="http://dx.doi.org/10.5061/dryad.j2n18" source-id-type="uri" id="dataro1"><collab collab-type="author">Schueren F</collab>, <collab collab-type="author">Lingner T</collab>, <collab collab-type="author">George R</collab>, <collab collab-type="author">Hofhuis J</collab>, <collab collab-type="author">Dickel C</collab>, <collab collab-type="author">Gärtner J</collab>, <collab collab-type="author">Thoms S</collab>, <year>2014</year><x>, </x><source>Data from: Peroxisomal lactate dehydrogenase is generated by translational readthrough in mammals</source><x>, </x><ext-link ext-link-type="uri" xlink:href="http://dx.doi.org/10.5061/dryad.j2n18">http://dx.doi.org/10.5061/dryad.j2n18</ext-link><x>, </x><comment>Available at Dryad Digital Repository under a CC0 Public Domain Dedication.</comment></related-object></p></sec></sec><ref-list><title>References</title><ref id="bib1"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Baranov</surname><given-names>PV</given-names></name><name><surname>Gesteland</surname><given-names>RF</given-names></name><name><surname>Atkins</surname><given-names>JF</given-names></name></person-group><year>2002</year><article-title>Recoding: translational bifurcations in gene expression</article-title><source>Gene</source><volume>286</volume><fpage>187</fpage><lpage>201</lpage><pub-id pub-id-type="doi">10.1016/S0378-1119(02)00423-7</pub-id></element-citation></ref><ref id="bib2"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Baumgart</surname><given-names>E</given-names></name><name><surname>Fahimi</surname><given-names>HD</given-names></name><name><surname>Stich</surname><given-names>A</given-names></name><name><surname>Völkl</surname><given-names>A</given-names></name></person-group><year>1996</year><article-title>L-lactate dehydrogenase A4- and A3B isoforms are bona fide peroxisomal enzymes in rat liver. Evidence for involvement in intraperoxisomal NADH reoxidation</article-title><source>The Journal of Biological Chemistry</source><volume>271</volume><fpage>3846</fpage><lpage>3855</lpage><pub-id pub-id-type="doi">10.1074/jbc.271.7.3846</pub-id></element-citation></ref><ref id="bib3"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Beier</surname><given-names>H</given-names></name><name><surname>Grimm</surname><given-names>M</given-names></name></person-group><year>2001</year><article-title>Misreading of termination codons in eukaryotes by natural nonsense suppressor tRNAs</article-title><source>Nucleic Acids Research</source><volume>29</volume><fpage>4767</fpage><lpage>4782</lpage><pub-id pub-id-type="doi">10.1093/nar/29.23.4767</pub-id></element-citation></ref><ref id="bib4"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Bidou</surname><given-names>L</given-names></name><name><surname>Allamand</surname><given-names>V</given-names></name><name><surname>Rousset</surname><given-names>JP</given-names></name><name><surname>Namy</surname><given-names>O</given-names></name></person-group><year>2012</year><article-title>Sense from nonsense: therapies for premature stop codon diseases</article-title><source>Trends in Molecular Medicine</source><volume>18</volume><fpage>679</fpage><lpage>688</lpage><pub-id pub-id-type="doi">10.1016/j.molmed.2012.09.008</pub-id></element-citation></ref><ref id="bib5"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Boyer</surname><given-names>SH</given-names></name><name><surname>Fainer</surname><given-names>DC</given-names></name><name><surname>Watson-Williams</surname><given-names>EJ</given-names></name></person-group><year>1963</year><article-title>Lactate dehydrogenase variant from human blood: evidence for molecular subunits</article-title><source>Science</source><volume>141</volume><fpage>642</fpage><lpage>643</lpage><pub-id pub-id-type="doi">10.1126/science.141.3581.642</pub-id></element-citation></ref><ref id="bib6"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Braverman</surname><given-names>NE</given-names></name><name><surname>D'Agostino</surname><given-names>MD</given-names></name><name><surname>Maclean</surname><given-names>GE</given-names></name></person-group><year>2013</year><article-title>Peroxisome biogenesis disorders: Biological, clinical and pathophysiological perspectives</article-title><source>Developmental Disabilities Research Reviews</source><volume>17</volume><fpage>187</fpage><lpage>196</lpage><pub-id pub-id-type="doi">10.1002/ddrr.1113</pub-id></element-citation></ref><ref id="bib7"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Brocard</surname><given-names>C</given-names></name><name><surname>Hartig</surname><given-names>A</given-names></name></person-group><year>2006</year><article-title>Peroxisome targeting signal 1: is it really a simple tripeptide?</article-title><source>Biochimica et Biophysica Acta</source><volume>1763</volume><fpage>1565</fpage><lpage>1573</lpage><pub-id pub-id-type="doi">10.1016/j.bbamcr.2006.08.022</pub-id></element-citation></ref><ref id="bib8"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Brooks</surname><given-names>GA</given-names></name></person-group><year>2009</year><article-title>Cell-cell and intracellular lactate shuttles</article-title><source>The Journal of Physiology</source><volume>587</volume><fpage>5591</fpage><lpage>5600</lpage><pub-id pub-id-type="doi">10.1113/jphysiol.2009.178350</pub-id></element-citation></ref><ref id="bib9"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Chittum</surname><given-names>HS</given-names></name><name><surname>Lane</surname><given-names>WS</given-names></name><name><surname>Carlson</surname><given-names>BA</given-names></name><name><surname>Roller</surname><given-names>PP</given-names></name><name><surname>Lung</surname><given-names>FD</given-names></name><name><surname>Lee</surname><given-names>BJ</given-names></name><name><surname>Hatfield</surname><given-names>DL</given-names></name></person-group><year>1998</year><article-title>Rabbit beta-globin is extended beyond its UGA stop codon by multiple suppressions and translational reading gaps</article-title><source>Biochemistry</source><volume>37</volume><fpage>10866</fpage><lpage>10870</lpage><pub-id pub-id-type="doi">10.1021/bi981042r</pub-id></element-citation></ref><ref id="bib10"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Dunn</surname><given-names>JG</given-names></name><name><surname>Foo</surname><given-names>CK</given-names></name><name><surname>Belletier</surname><given-names>NG</given-names></name><name><surname>Gavis</surname><given-names>ER</given-names></name><name><surname>Weissman</surname><given-names>JS</given-names></name></person-group><year>2013</year><article-title>Ribosome profiling reveals pervasive and regulated stop codon readthrough in <italic>Drosophila melanogaster</italic></article-title><source>eLife</source><volume>2</volume><fpage>e01179</fpage><pub-id pub-id-type="doi">10.7554/eLife.01179</pub-id></element-citation></ref><ref id="bib11"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Eswarappa</surname><given-names>SM</given-names></name><name><surname>Potdar</surname><given-names>AA</given-names></name><name><surname>Koch</surname><given-names>WJ</given-names></name><name><surname>Fan</surname><given-names>Y</given-names></name><name><surname>Vasu</surname><given-names>K</given-names></name><name><surname>Lindner</surname><given-names>D</given-names></name><name><surname>Willard</surname><given-names>B</given-names></name><name><surname>Graham</surname><given-names>LM</given-names></name><name><surname>DiCorleto</surname><given-names>PE</given-names></name><name><surname>Fox</surname><given-names>PL</given-names></name></person-group><year>2014</year><article-title>Programmed translational readthrough generates antiangiogenic VEGF-Ax</article-title><source>Cell</source><volume>157</volume><fpage>1605</fpage><lpage>1618</lpage><pub-id pub-id-type="doi">10.1016/j.cell.2014.04.033</pub-id></element-citation></ref><ref id="bib12"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Felsenstein</surname><given-names>J</given-names></name></person-group><year>1989</year><article-title>PHYLIP - Phylogeny Inference Package (Version 3.2)</article-title><source>Cladistics</source><volume>5</volume><fpage>164</fpage><lpage>166</lpage></element-citation></ref><ref id="bib13"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Firth</surname><given-names>AE</given-names></name><name><surname>Brierley</surname><given-names>I</given-names></name></person-group><year>2012</year><article-title>Non-canonical translation in RNA viruses</article-title><source>The Journal of General Virology</source><volume>93</volume><fpage>1385</fpage><lpage>1409</lpage><pub-id pub-id-type="doi">10.1099/vir.0.042499-0</pub-id></element-citation></ref><ref id="bib14"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Firth</surname><given-names>AE</given-names></name><name><surname>Wills</surname><given-names>NM</given-names></name><name><surname>Gesteland</surname><given-names>RF</given-names></name><name><surname>Atkins</surname><given-names>JF</given-names></name></person-group><year>2011</year><article-title>Stimulation of stop codon readthrough: frequent presence of an extended 3′ RNA structural element</article-title><source>Nucleic Acids Research</source><volume>39</volume><fpage>6679</fpage><lpage>6691</lpage><pub-id pub-id-type="doi">10.1093/nar/gkr224</pub-id></element-citation></ref><ref id="bib15"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Flicek</surname><given-names>P</given-names></name><name><surname>Ahmed</surname><given-names>I</given-names></name><name><surname>Amode</surname><given-names>MR</given-names></name><name><surname>Barrell</surname><given-names>D</given-names></name><name><surname>Beal</surname><given-names>K</given-names></name><name><surname>Brent</surname><given-names>S</given-names></name><name><surname>Carvalho-Silva</surname><given-names>D</given-names></name><name><surname>Clapham</surname><given-names>P</given-names></name><name><surname>Coates</surname><given-names>G</given-names></name><name><surname>Fairley</surname><given-names>S</given-names></name><name><surname>Fitzgerald</surname><given-names>S</given-names></name><name><surname>Gil</surname><given-names>L</given-names></name><name><surname>Garcia-Giron</surname><given-names>C</given-names></name><name><surname>Gordon</surname><given-names>L</given-names></name><name><surname>Hourlier</surname><given-names>T</given-names></name><name><surname>Hunt</surname><given-names>S</given-names></name><name><surname>Juettemann</surname><given-names>T</given-names></name><name><surname>Kahari</surname><given-names>AK</given-names></name><name><surname>Keenan</surname><given-names>S</given-names></name><name><surname>Komorowska</surname><given-names>M</given-names></name><name><surname>Kulesha</surname><given-names>E</given-names></name><name><surname>Longden</surname><given-names>I</given-names></name><name><surname>Maurel</surname><given-names>T</given-names></name><name><surname>McLaren</surname><given-names>WM</given-names></name><name><surname>Muffato</surname><given-names>M</given-names></name><name><surname>Nag</surname><given-names>R</given-names></name><name><surname>Overduin</surname><given-names>B</given-names></name><name><surname>Pignatelli</surname><given-names>M</given-names></name><name><surname>Pritchard</surname><given-names>B</given-names></name><name><surname>Pritchard</surname><given-names>E</given-names></name><name><surname>Riat</surname><given-names>HS</given-names></name><name><surname>Ritchie</surname><given-names>GR</given-names></name><name><surname>Ruffier</surname><given-names>M</given-names></name><name><surname>Schuster</surname><given-names>M</given-names></name><name><surname>Sheppard</surname><given-names>D</given-names></name><name><surname>Sobral</surname><given-names>D</given-names></name><name><surname>Taylor</surname><given-names>K</given-names></name><name><surname>Thormann</surname><given-names>A</given-names></name><name><surname>Trevanion</surname><given-names>S</given-names></name><name><surname>White</surname><given-names>S</given-names></name><name><surname>Wilder</surname><given-names>SP</given-names></name><name><surname>Aken</surname><given-names>BL</given-names></name><name><surname>Birney</surname><given-names>E</given-names></name><name><surname>Cunningham</surname><given-names>F</given-names></name><name><surname>Dunham</surname><given-names>I</given-names></name><name><surname>Harrow</surname><given-names>J</given-names></name><name><surname>Herrero</surname><given-names>J</given-names></name><name><surname>Hubbard</surname><given-names>TJ</given-names></name><name><surname>Johnson</surname><given-names>N</given-names></name><name><surname>Kinsella</surname><given-names>R</given-names></name><name><surname>Parker</surname><given-names>A</given-names></name><name><surname>Spudich</surname><given-names>G</given-names></name><name><surname>Yates</surname><given-names>A</given-names></name><name><surname>Zadissa</surname><given-names>A</given-names></name><name><surname>Searle</surname><given-names>SM</given-names></name></person-group><year>2012</year><article-title>Ensembl 2013</article-title><source>Nucleic Acids Research</source><volume>41</volume><fpage>D48</fpage><lpage>D55</lpage><pub-id pub-id-type="doi">10.1093/nar/gks1236</pub-id></element-citation></ref><ref id="bib16"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Floquet</surname><given-names>C</given-names></name><name><surname>Hatin</surname><given-names>I</given-names></name><name><surname>Rousset</surname><given-names>JP</given-names></name><name><surname>Bidou</surname><given-names>L</given-names></name></person-group><year>2012</year><article-title>Statistical analysis of readthrough levels for nonsense mutations in mammalian cells reveals a major determinant of response to gentamicin</article-title><source>PLOS Genetics</source><volume>8</volume><fpage>e1002608</fpage><pub-id pub-id-type="doi">10.1371/journal.pgen.1002608</pub-id></element-citation></ref><ref id="bib17"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Freitag</surname><given-names>J</given-names></name><name><surname>Ast</surname><given-names>J</given-names></name><name><surname>Bölker</surname><given-names>M</given-names></name></person-group><year>2012</year><article-title>Cryptic peroxisomal targeting via alternative splicing and stop codon read-through in fungi</article-title><source>Nature</source><volume>485</volume><fpage>522</fpage><lpage>525</lpage><pub-id pub-id-type="doi">10.1038/nature11051</pub-id></element-citation></ref><ref id="bib18"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Geller</surname><given-names>AI</given-names></name><name><surname>Rich</surname><given-names>A</given-names></name></person-group><year>1980</year><article-title>A UGA termination suppression tRNATrp active in rabbit reticulocytes</article-title><source>Nature</source><volume>283</volume><fpage>41</fpage><lpage>46</lpage><pub-id pub-id-type="doi">10.1038/283041a0</pub-id></element-citation></ref><ref id="bib19"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Gladden</surname><given-names>LB</given-names></name></person-group><year>2004</year><article-title>Lactate metabolism: a new paradigm for the third millennium</article-title><source>The Journal of Physiology</source><volume>558</volume><fpage>5</fpage><lpage>30</lpage><pub-id pub-id-type="doi">10.1113/jphysiol.2003.058701</pub-id></element-citation></ref><ref id="bib20"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Goldberg</surname><given-names>E</given-names></name><name><surname>Eddy</surname><given-names>EM</given-names></name><name><surname>Duan</surname><given-names>C</given-names></name><name><surname>Odet</surname><given-names>F</given-names></name></person-group><year>2009</year><article-title>LDHC: the ultimate testis-specific gene</article-title><source>Journal of Andrology</source><volume>31</volume><fpage>86</fpage><lpage>94</lpage><pub-id pub-id-type="doi">10.2164/jandrol.109.008367</pub-id></element-citation></ref><ref id="bib21"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Gronemeyer</surname><given-names>T</given-names></name><name><surname>Wiese</surname><given-names>S</given-names></name><name><surname>Ofman</surname><given-names>R</given-names></name><name><surname>Bunse</surname><given-names>C</given-names></name><name><surname>Pawlas</surname><given-names>M</given-names></name><name><surname>Hayen</surname><given-names>H</given-names></name><name><surname>Eisenacher</surname><given-names>M</given-names></name><name><surname>Stephan</surname><given-names>C</given-names></name><name><surname>Meyer</surname><given-names>HE</given-names></name><name><surname>Waterham</surname><given-names>HR</given-names></name><name><surname>Erdmann</surname><given-names>R</given-names></name><name><surname>Wanders</surname><given-names>RJ</given-names></name><name><surname>Warscheid</surname><given-names>B</given-names></name></person-group><year>2013</year><article-title>The proteome of human liver peroxisomes: identification of five new peroxisomal constituents by a label-free quantitative proteomics survey</article-title><source>PLOS ONE</source><volume>8</volume><fpage>e57395</fpage><pub-id pub-id-type="doi">10.1371/journal.pone.0057395</pub-id></element-citation></ref><ref id="bib22"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Gualdrón-López</surname><given-names>M</given-names></name><name><surname>Brennand</surname><given-names>A</given-names></name><name><surname>Hannaert</surname><given-names>V</given-names></name><name><surname>Quiñones</surname><given-names>W</given-names></name><name><surname>Cáceres</surname><given-names>AJ</given-names></name><name><surname>Bringaud</surname><given-names>F</given-names></name><name><surname>Concepción</surname><given-names>JL</given-names></name><name><surname>Michels</surname><given-names>PAM</given-names></name></person-group><year>2012</year><article-title>When, how and why glycolysis became compartmentalised in the Kinetoplastea. A new look at an ancient organelle</article-title><source>International Journal for Parasitology</source><volume>42</volume><fpage>1</fpage><lpage>20</lpage><pub-id pub-id-type="doi">10.1016/j.ijpara.2011.10.007</pub-id></element-citation></ref><ref id="bib23"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Harrell</surname><given-names>L</given-names></name><name><surname>Melcher</surname><given-names>U</given-names></name><name><surname>Atkins</surname><given-names>JF</given-names></name></person-group><year>2002</year><article-title>Predominance of six different hexanucleotide recoding signals 3′ of read-through stop codons</article-title><source>Nucleic Acids Research</source><volume>30</volume><fpage>2011</fpage><lpage>2017</lpage><pub-id pub-id-type="doi">10.1093/nar/30.9.2011</pub-id></element-citation></ref><ref id="bib24"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Hoerl</surname><given-names>AE</given-names></name><name><surname>Kennard</surname><given-names>RW</given-names></name></person-group><year>1970</year><article-title>Ridge regression: biased estimation for nonorthogonal problems</article-title><source>Technometrics</source><volume>12</volume><fpage>55</fpage><lpage>67</lpage><pub-id pub-id-type="doi">10.1080/00401706.1970.10488634</pub-id></element-citation></ref><ref id="bib25"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Hsu</surname><given-names>PP</given-names></name><name><surname>Sabatini</surname><given-names>DM</given-names></name></person-group><year>2008</year><article-title>Cancer cell metabolism: Warburg and beyond</article-title><source>Cell</source><volume>134</volume><fpage>703</fpage><lpage>707</lpage><pub-id pub-id-type="doi">10.1016/j.cell.2008.08.021</pub-id></element-citation></ref><ref id="bib26"><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="bib27"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Jungreis</surname><given-names>I</given-names></name><name><surname>Lin</surname><given-names>MF</given-names></name><name><surname>Spokony</surname><given-names>R</given-names></name><name><surname>Chan</surname><given-names>CS</given-names></name><name><surname>Negre</surname><given-names>N</given-names></name><name><surname>Victorsen</surname><given-names>A</given-names></name><name><surname>White</surname><given-names>KP</given-names></name><name><surname>Kellis</surname><given-names>M</given-names></name></person-group><year>2011</year><article-title>Evidence of abundant stop codon readthrough in <italic>Drosophila</italic> and other metazoa</article-title><source>Genome Research</source><volume>21</volume><fpage>2096</fpage><lpage>2113</lpage><pub-id pub-id-type="doi">10.1101/gr.119974.110</pub-id></element-citation></ref><ref id="bib28"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Keeling</surname><given-names>KM</given-names></name><name><surname>Xue</surname><given-names>X</given-names></name><name><surname>Gunn</surname><given-names>G</given-names></name><name><surname>Bedwell</surname><given-names>DM</given-names></name></person-group><year>2014</year><article-title>Therapeutics based on stop codon readthrough</article-title><source>Annual Review of Genomics and Human Genetics</source><volume>15</volume><fpage>371</fpage><lpage>394</lpage><pub-id pub-id-type="doi">10.1146/annurev-genom-091212-153527</pub-id></element-citation></ref><ref id="bib29"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Krogh</surname><given-names>A</given-names></name><name><surname>Larsson</surname><given-names>B</given-names></name><name><surname>von Heijne</surname><given-names>G</given-names></name><name><surname>Sonnhammer</surname><given-names>EL</given-names></name></person-group><year>2001</year><article-title>Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes</article-title><source>Journal of Molecular Biology</source><volume>305</volume><fpage>567</fpage><lpage>580</lpage><pub-id pub-id-type="doi">10.1006/jmbi.2000.4315</pub-id></element-citation></ref><ref id="bib30"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Kunze</surname><given-names>M</given-names></name><name><surname>Hartig</surname><given-names>A</given-names></name></person-group><year>2013</year><article-title>Permeability of the peroxisomal membrane: lessons from the glyoxylate cycle</article-title><source>Frontiers in Physiology</source><volume>4</volume><fpage>204</fpage><pub-id pub-id-type="doi">10.3389/fphys.2013.00204</pub-id></element-citation></ref><ref id="bib31"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lanyon-Hogg</surname><given-names>T</given-names></name><name><surname>Warriner</surname><given-names>SL</given-names></name><name><surname>Baker</surname><given-names>A</given-names></name></person-group><year>2010</year><article-title>Getting a camel through the eye of a needle: the import of folded proteins by peroxisomes</article-title><source>Biology of the cell/under the auspices of the European Cell Biology Organization</source><volume>102</volume><fpage>245</fpage><lpage>263</lpage><pub-id pub-id-type="doi">10.1042/BC20090159</pub-id></element-citation></ref><ref id="bib32"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Li</surname><given-names>G</given-names></name><name><surname>Rice</surname><given-names>CM</given-names></name></person-group><year>1993</year><article-title>The signal for translational readthrough of a UGA codon in Sindbis virus RNA involves a single cytidine residue immediately downstream of the termination codon</article-title><source>Journal of Virology</source><volume>67</volume><fpage>5062</fpage><lpage>5067</lpage></element-citation></ref><ref id="bib33"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lin</surname><given-names>MF</given-names></name><name><surname>Jungreis</surname><given-names>I</given-names></name><name><surname>Kellis</surname><given-names>M</given-names></name></person-group><year>2011</year><article-title>PhyloCSF: a comparative genomics method to distinguish protein coding and non-coding regions</article-title><source>Bioinformatics</source><volume>27</volume><fpage>i275</fpage><lpage>i282</lpage><pub-id pub-id-type="doi">10.1093/bioinformatics/btr209</pub-id></element-citation></ref><ref id="bib34"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lingner</surname><given-names>T</given-names></name><name><surname>Kataya</surname><given-names>AR</given-names></name><name><surname>Antonicelli</surname><given-names>GE</given-names></name><name><surname>Benichou</surname><given-names>A</given-names></name><name><surname>Nilssen</surname><given-names>K</given-names></name><name><surname>Chen</surname><given-names>XY</given-names></name><name><surname>Siemsen</surname><given-names>T</given-names></name><name><surname>Morgenstern</surname><given-names>B</given-names></name><name><surname>Meinicke</surname><given-names>P</given-names></name><name><surname>Reumann</surname><given-names>S</given-names></name></person-group><year>2011</year><article-title>Identification of novel plant peroxisomal targeting signals by a combination of machine learning methods and in vivo subcellular targeting analyses</article-title><source>The Plant Cell</source><volume>23</volume><fpage>1556</fpage><lpage>1572</lpage><pub-id pub-id-type="doi">10.1105/tpc.111.084095</pub-id></element-citation></ref><ref id="bib35"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Lodhi</surname><given-names>IJ</given-names></name><name><surname>Semenkovich</surname><given-names>CF</given-names></name></person-group><year>2014</year><article-title>Peroxisomes: a nexus for lipid metabolism and cellular signaling</article-title><source>Cell Metabolism</source><volume>19</volume><fpage>380</fpage><lpage>392</lpage><pub-id pub-id-type="doi">10.1016/j.cmet.2014.01.002</pub-id></element-citation></ref><ref id="bib36"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Loughran</surname><given-names>G</given-names></name><name><surname>Chou</surname><given-names>MY</given-names></name><name><surname>Ivanov</surname><given-names>IP</given-names></name><name><surname>Jungreis</surname><given-names>I</given-names></name><name><surname>Kellis</surname><given-names>M</given-names></name><name><surname>Kiran</surname><given-names>AM</given-names></name><name><surname>Baranov</surname><given-names>PV</given-names></name><name><surname>Atkins</surname><given-names>JF</given-names></name></person-group><year>2014</year><article-title>Evidence of efficient stop codon readthrough in four mammalian genes</article-title><source>Nucleic Acids Research</source><volume>42</volume><fpage>8928</fpage><lpage>8938</lpage><pub-id pub-id-type="doi">10.1093/nar/gku608</pub-id></element-citation></ref><ref id="bib37"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Markert</surname><given-names>CL</given-names></name></person-group><year>1963</year><article-title>Lactate dehydrogenase isozymes: dissociation and recombination of subunits</article-title><source>Science</source><volume>140</volume><fpage>1329</fpage><lpage>1330</lpage><pub-id pub-id-type="doi">10.1126/science.140.3573.1329</pub-id></element-citation></ref><ref id="bib38"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Maynard</surname><given-names>EL</given-names></name><name><surname>Gatto</surname><given-names>GJ</given-names><suffix>Jnr</suffix></name><name><surname>Berg</surname><given-names>JM</given-names></name></person-group><year>2004</year><article-title>Pex5p binding affinities for canonical and noncanonical PTS1 peptides</article-title><source>Proteins</source><volume>55</volume><fpage>856</fpage><lpage>861</lpage><pub-id pub-id-type="doi">10.1002/prot.20112</pub-id></element-citation></ref><ref id="bib39"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>McCaughan</surname><given-names>KK</given-names></name><name><surname>Brown</surname><given-names>CM</given-names></name><name><surname>Dalphin</surname><given-names>ME</given-names></name><name><surname>Berry</surname><given-names>MJ</given-names></name><name><surname>Tate</surname><given-names>WP</given-names></name></person-group><year>1995</year><article-title>Translational termination efficiency in mammals is influenced by the base following the stop codon</article-title><source>Proceedings of the National Academy of Sciences of USA</source><volume>92</volume><fpage>5431</fpage><lpage>5435</lpage><pub-id pub-id-type="doi">10.1073/pnas.92.12.5431</pub-id></element-citation></ref><ref id="bib40"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>McClelland</surname><given-names>GB</given-names></name><name><surname>Khanna</surname><given-names>S</given-names></name><name><surname>González</surname><given-names>GF</given-names></name><name><surname>Eric Butz</surname><given-names>C</given-names></name><name><surname>Brooks</surname><given-names>GA</given-names></name></person-group><year>2003</year><article-title>Peroxisomal membrane monocarboxylate transporters: evidence for a redox shuttle system?</article-title><source>Biochemical and Biophysical Research Communications</source><volume>304</volume><fpage>130</fpage><lpage>135</lpage><pub-id pub-id-type="doi">10.1016/S0006-291X(03)00550-3</pub-id></element-citation></ref><ref id="bib41"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>McGroarty</surname><given-names>E</given-names></name><name><surname>Hsieh</surname><given-names>B</given-names></name><name><surname>Wied</surname><given-names>DM</given-names></name><name><surname>Gee</surname><given-names>R</given-names></name><name><surname>Tolbert</surname><given-names>NE</given-names></name></person-group><year>1974</year><article-title>Alpha hydroxy acid oxidation by peroxisomes</article-title><source>Archives of Biochemistry and Biophysics</source><volume>161</volume><fpage>194</fpage><lpage>210</lpage><pub-id pub-id-type="doi">10.1016/0003-9861(74)90251-3</pub-id></element-citation></ref><ref id="bib42"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>McNew</surname><given-names>JA</given-names></name><name><surname>Goodman</surname><given-names>JM</given-names></name></person-group><year>1996</year><article-title>The targeting and assembly of peroxisomal proteins: some old rules do not apply</article-title><source>Trends in Biochemical Sciences</source><volume>21</volume><fpage>54</fpage><lpage>58</lpage><pub-id pub-id-type="doi">10.1016/0968-0004(96)80866-8</pub-id></element-citation></ref><ref id="bib43"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Namy</surname><given-names>O</given-names></name><name><surname>Hatin</surname><given-names>I</given-names></name><name><surname>Rousset</surname><given-names>JP</given-names></name></person-group><year>2001</year><article-title>Impact of the six nucleotides downstream of the stop codon on translation termination</article-title><source>EMBO Reports</source><volume>2</volume><fpage>787</fpage><lpage>793</lpage><pub-id pub-id-type="doi">10.1093/embo-reports/kve176</pub-id></element-citation></ref><ref id="bib44"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Namy</surname><given-names>O</given-names></name><name><surname>Rousset</surname><given-names>JP</given-names></name><name><surname>Napthine</surname><given-names>S</given-names></name><name><surname>Brierley</surname><given-names>I</given-names></name></person-group><year>2004</year><article-title>Reprogrammed genetic decoding in cellular gene expression</article-title><source>Molecular Cell</source><volume>13</volume><fpage>157</fpage><lpage>168</lpage><pub-id pub-id-type="doi">10.1016/S1097-2765(04)00031-0</pub-id></element-citation></ref><ref id="bib45"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Osmundsen</surname><given-names>H</given-names></name></person-group><year>1982</year><article-title>Factors which can influence beta-oxidation by peroxisomes isolated from livers of clofibrate treated rats. Some properties of peroxisomal fractions isolated in a self-generated Percoll gradient by vertical rotor centrifugation</article-title><source>The International Journal of Biochemistry</source><volume>14</volume><fpage>905</fpage><lpage>914</lpage><pub-id pub-id-type="doi">10.1016/0020-711X(82)90074-X</pub-id></element-citation></ref><ref id="bib46"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Pacho</surname><given-names>F</given-names></name><name><surname>Zambruno</surname><given-names>G</given-names></name><name><surname>Calabresi</surname><given-names>V</given-names></name><name><surname>Kiritsi</surname><given-names>D</given-names></name><name><surname>Schneider</surname><given-names>H</given-names></name></person-group><year>2011</year><article-title>Efficiency of translation termination in humans is highly dependent upon nucleotides in the neighbourhood of a (premature) termination codon</article-title><source>Journal of Medical Genetics</source><volume>48</volume><fpage>640</fpage><lpage>644</lpage><pub-id pub-id-type="doi">10.1136/jmg.2011.089615</pub-id></element-citation></ref><ref id="bib47"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Read</surname><given-names>JA</given-names></name><name><surname>Winter</surname><given-names>VJ</given-names></name><name><surname>Eszes</surname><given-names>CM</given-names></name><name><surname>Sessions</surname><given-names>RB</given-names></name><name><surname>Brady</surname><given-names>RL</given-names></name></person-group><year>2001</year><article-title>Structural basis for altered activity of M- and H-isozyme forms of human lactate dehydrogenase</article-title><source>Proteins</source><volume>43</volume><fpage>175</fpage><lpage>185</lpage><pub-id pub-id-type="doi">10.1002/1097-0134(20010501)43:23.0.CO;2-#</pub-id></element-citation></ref><ref id="bib48"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Salido</surname><given-names>E</given-names></name><name><surname>Pey</surname><given-names>AL</given-names></name><name><surname>Rodriguez</surname><given-names>R</given-names></name><name><surname>Lorenzo</surname><given-names>V</given-names></name></person-group><year>2012</year><article-title>Primary hyperoxalurias: disorders of glyoxylate detoxification</article-title><source>Biochimica et Biophysica Acta</source><volume>1822</volume><fpage>1453</fpage><lpage>1464</lpage><pub-id pub-id-type="doi">10.1016/j.bbadis.2012.03.004</pub-id></element-citation></ref><ref id="bib49"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Schueren</surname><given-names>F</given-names></name><name><surname>Lingner</surname><given-names>T</given-names></name><name><surname>George</surname><given-names>R</given-names></name><name><surname>Hofhuis</surname><given-names>J</given-names></name><name><surname>Dickel</surname><given-names>C</given-names></name><name><surname>Gärtner</surname><given-names>J</given-names></name><name><surname>Thoms</surname><given-names>S</given-names></name></person-group><year>2014</year><article-title>Data from: Peroxisomal lactate dehydrogenase is generated by translational readthrough in mammals</article-title><source>Dryad Digital Repository</source><pub-id pub-id-type="doi">10.5061/dryad.j2n18</pub-id></element-citation></ref><ref id="bib50"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Schwarz</surname><given-names>F</given-names></name><name><surname>Aebi</surname><given-names>M</given-names></name></person-group><year>2011</year><article-title>Mechanisms and principles of N-linked protein glycosylation</article-title><source>Current Opinion in Structural Biology</source><volume>21</volume><fpage>576</fpage><lpage>582</lpage><pub-id pub-id-type="doi">10.1016/j.sbi.2011.08.005</pub-id></element-citation></ref><ref id="bib51"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Smith</surname><given-names>JJ</given-names></name><name><surname>Aitchison</surname><given-names>JD</given-names></name></person-group><year>2013</year><article-title>Peroxisomes take shape</article-title><source>Nature Reviews Molecular Cell Biology</source><volume>14</volume><fpage>803</fpage><lpage>817</lpage><pub-id pub-id-type="doi">10.1038/nrm3700</pub-id></element-citation></ref><ref id="bib52"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Stornaiuolo</surname><given-names>M</given-names></name><name><surname>Lotti</surname><given-names>LV</given-names></name><name><surname>Borgese</surname><given-names>N</given-names></name><name><surname>Torrisi</surname><given-names>MR</given-names></name><name><surname>Mottola</surname><given-names>G</given-names></name><name><surname>Martire</surname><given-names>G</given-names></name><name><surname>Bonatti</surname><given-names>S</given-names></name></person-group><year>2003</year><article-title>KDEL and KKXX retrieval signals appended to the same reporter protein determine different trafficking between endoplasmic reticulum, intermediate compartment, and Golgi complex</article-title><source>Molecular Biology of the Cell</source><volume>14</volume><fpage>889</fpage><lpage>902</lpage><pub-id pub-id-type="doi">10.1091/mbc.E02-08-0468</pub-id></element-citation></ref><ref id="bib53"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Thoms</surname><given-names>S</given-names></name><name><surname>Hofhuis</surname><given-names>J</given-names></name><name><surname>Thöing</surname><given-names>C</given-names></name><name><surname>Gärtner</surname><given-names>J</given-names></name><name><surname>Niemann</surname><given-names>HH</given-names></name></person-group><year>2011</year><article-title>The unusual extended C-terminal helix of the peroxisomal α/β-hydrolase Lpx1 is involved in dimer contacts but dispensable for dimerization</article-title><source>Journal of Structural Biology</source><volume>175</volume><fpage>362</fpage><lpage>371</lpage><pub-id pub-id-type="doi">10.1016/j.jsb.2011.06.008</pub-id></element-citation></ref><ref id="bib54"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Tuite</surname><given-names>MF</given-names></name><name><surname>Cox</surname><given-names>BS</given-names></name></person-group><year>2003</year><article-title>Propagation of yeast prions</article-title><source>Nature Reviews Molecular Cell Biology</source><volume>4</volume><fpage>878</fpage><lpage>890</lpage><pub-id pub-id-type="doi">10.1038/nrm1247</pub-id></element-citation></ref><ref id="bib55"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Visser</surname><given-names>WF</given-names></name><name><surname>van Roermund</surname><given-names>CWT</given-names></name><name><surname>Ijlst</surname><given-names>L</given-names></name><name><surname>Waterham</surname><given-names>HR</given-names></name><name><surname>Wanders</surname><given-names>RJ</given-names></name></person-group><year>2007</year><article-title>Metabolite transport across the peroxisomal membrane</article-title><source>The Biochemical Journal</source><volume>401</volume><fpage>365</fpage><lpage>375</lpage><pub-id pub-id-type="doi">10.1042/BJ20061352</pub-id></element-citation></ref><ref id="bib56"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Völkl</surname><given-names>A</given-names></name><name><surname>Fahimi</surname><given-names>HD</given-names></name></person-group><year>1985</year><article-title>Isolation and characterization of peroxisomes from the liver of normal untreated rats</article-title><source>European Journal of Biochemistry</source><volume>149</volume><fpage>257</fpage><lpage>265</lpage><pub-id pub-id-type="doi">10.1111/j.1432-1033.1985.tb08920.x</pub-id></element-citation></ref><ref id="bib57"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Waterhouse</surname><given-names>AM</given-names></name><name><surname>Procter</surname><given-names>JB</given-names></name><name><surname>Martin</surname><given-names>DM</given-names></name><name><surname>Clamp</surname><given-names>M</given-names></name><name><surname>Barton</surname><given-names>GJ</given-names></name></person-group><year>2009</year><article-title>Jalview Version 2–a multiple sequence alignment editor and analysis workbench</article-title><source>Bioinformatics</source><volume>25</volume><fpage>1189</fpage><lpage>1191</lpage><pub-id pub-id-type="doi">10.1093/bioinformatics/btp033</pub-id></element-citation></ref><ref id="bib58"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Workman</surname><given-names>CT</given-names></name><name><surname>Yin</surname><given-names>Y</given-names></name><name><surname>Corcoran</surname><given-names>DL</given-names></name><name><surname>Ideker</surname><given-names>T</given-names></name><name><surname>Stormo</surname><given-names>GD</given-names></name><name><surname>Benos</surname><given-names>PV</given-names></name></person-group><year>2005</year><article-title>enoLOGOS: a versatile web tool for energy normalized sequence logos</article-title><source>Nucleic Acids Research</source><volume>33</volume><fpage>W389</fpage><lpage>W392</lpage><pub-id pub-id-type="doi">10.1093/nar/gki439</pub-id></element-citation></ref><ref id="bib59"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Yamaguchi</surname><given-names>Y</given-names></name><name><surname>Hayashi</surname><given-names>A</given-names></name><name><surname>Campagnoni</surname><given-names>CW</given-names></name><name><surname>Kimura</surname><given-names>A</given-names></name><name><surname>Inuzuka</surname><given-names>T</given-names></name><name><surname>Baba</surname><given-names>H</given-names></name></person-group><year>2012</year><article-title>L-mpz, a novel isoform of myelin P0, is produced by stop codon readthrough</article-title><source>The Journal of Biological Chemistry</source><volume>287</volume><fpage>17765</fpage><lpage>17776</lpage><pub-id pub-id-type="doi">10.1074/jbc.M111.314468</pub-id></element-citation></ref><ref id="bib60"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Zerangue</surname><given-names>N</given-names></name><name><surname>Malan</surname><given-names>MJ</given-names></name><name><surname>Fried</surname><given-names>SR</given-names></name><name><surname>Dazin</surname><given-names>PF</given-names></name><name><surname>Jan</surname><given-names>YN</given-names></name><name><surname>Jan</surname><given-names>LY</given-names></name><name><surname>Schwappach</surname><given-names>B</given-names></name></person-group><year>2001</year><article-title>Analysis of endoplasmic reticulum trafficking signals by combinatorial screening in mammalian cells</article-title><source>Proceedings of the National Academy of Sciences of USA</source><volume>98</volume><fpage>2431</fpage><lpage>2436</lpage><pub-id pub-id-type="doi">10.1073/pnas.051630198</pub-id></element-citation></ref><ref id="bib61"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Zhang</surname><given-names>FL</given-names></name><name><surname>Casey</surname><given-names>PJ</given-names></name></person-group><year>1996</year><article-title>Protein prenylation: molecular mechanisms and functional consequences</article-title><source>Annual Review of Biochemistry</source><volume>65</volume><fpage>241</fpage><lpage>269</lpage><pub-id pub-id-type="doi">10.1146/annurev.bi.65.070196.001325</pub-id></element-citation></ref><ref id="bib62"><element-citation publication-type="journal"><person-group person-group-type="author"><name><surname>Zielinska</surname><given-names>DF</given-names></name><name><surname>Gnad</surname><given-names>F</given-names></name><name><surname>Wiśniewski</surname><given-names>JR</given-names></name><name><surname>Mann</surname><given-names>M</given-names></name></person-group><year>2010</year><article-title>Precision mapping of an in vivo N-glycoproteome reveals rigid topological and sequence constraints</article-title><source>Cell</source><volume>141</volume><fpage>897</fpage><lpage>907</lpage><pub-id pub-id-type="doi">10.1016/j.cell.2010.04.012</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.03640.027</article-id><title-group><article-title>Decision letter</article-title></title-group><contrib-group content-type="section"><contrib contrib-type="editor"><name><surname>Sonenberg</surname><given-names>Nahum</given-names></name><role>Reviewing editor</role><aff><institution>McGill University</institution>, <country>Canada</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://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>Thank you for choosing to send your work entitled “Functional translational readthrough in humans” for consideration at <italic>eLife.</italic> Your full submission has been evaluated by James Manley (Senior editor), a Reviewing editor, and 3 peer reviewers, and the decision was reached after discussions between the reviewers. We regret to inform you that the extra work we think will be needed falls outside the scope of a resubmission, but if you can address the concerns below we would be happy to review a new submission at a later date.</p><p>You describe the derivation of a new algorithm to deduce the propensity for nonsense codon readthrough. The resulting quantitative predictor, called the RTP score, was initially derived from published data on aminoglycoside-promoted readthrough efficiency and further developed by the use of dual reporter readthrough assays. You find provocative correlations of expression and readthrough, i.e., genes that are highly expressed have a lower RTP score, whereas genes that are expressed at low levels have higher RTP scores. By coupling the RTP predictor with a peroxisomal targeting sequence (PTS) prediction algorithm you sought genes that have high scores for RTP as well as PTS. This approach identified a new isoform of LDHB that arises from the readthrough of the normal termination codon resulting in translation of a C-terminally extended protein product harboring the peroxisomal targeting sequence. Monitoring the subcellular localization of the appropriate fluorescent fusion proteins validated this conclusion.</p><p>Overall, this is an interesting manuscript, the experiments are well executed, and many of the conclusions are well supported by the data. However, there are major concerns with the paper. One problem is the fact that the function for LDHBP+ in the peroxisome is only postulated, not formally demonstrated (for example, by a loss-of-function experiment or by clinical data). Another major problem is the molecular mechanism involved (please see below). Another concern relates to the repeated claims that the RTP models are “a quantitative predictor for readthrough propensity in the human genome,” which are not supported by a clear, quantitative analysis of RTP performance. Because RTP is a major focus of the manuscript, you should provide additional analyses to substantiate these claims. These are critical questions you must address in any future new submission to <italic>eLife.</italic></p><p>1) The overall rationale of the authors' algorithms needs to be explained in more elaborate, general terms. Further, there needs to be some indication of “public” availability of the algorithm.</p><p>2) Dual reporter assays play a key role in the assessment of nonsense codon contexts that are prone to readthrough. However, it appears that all of these assays used by the authors have the traditional format that excludes any introns. This may be a significant problem since deposition of EJC proteins appears to play a key role in translational efficiency (Wiegand et al [2003] Proc Natl Acad Sci USA 100: 11327-11332; Gudikote et al. [2005] Nature Struct Mol Biol 12: 801-819; Nott et al. [2004] Genes Dev 18: 210-222). The latter experiments raise the possibility that the lack of introns/EJC proteins may affect readthrough efficacy. Hence, at a minimum, the authors need to acknowledge this shortcoming of their experimental approach.</p><p>3) The authors make a strong point of being the “first” to observe that nonsense suppression allows two mammalian proteins to be derived from the same mRNA. However, this point is also apparent from the supplementary data presented in Welch et al. (2007) (Nature 447: 87-91). Again, this should be acknowledged.</p><p>4) The authors do not discuss/compare the consensus sequence identified from their analyses with any of the stop codon sequence contexts that have been reported in the literature. A comparison with known sequence contexts would be helpful for the readers and may exemplify the strength of the RTP predictor.</p><p>5) The authors use one specific case study, the peroxisomal targeting sequence, to show physiological functions of a readthrough protein. However, adding one more example, other than the PTS, would greatly enhance the applicability of the RTP predictor combined with other in silico analyses. Experimental validation of another case study might be beyond the scope of the paper, but it would show a wider applicability of the method described.</p><p>6) In the experimental validation of the role of readthrough in the synthesis of LDHB+P, the authors should demonstrate the validity of their RTP predictor by doing mutational analyses of the stop codon context according to their consensus models.</p><p>7) Although the authors have an intriguing model for the co-import of LDHA with LDHB, It would be important to show two-hybrid data to support it, or address it biochemically using co-IP analyses of the LDHB+P and LDHA.</p><p>8) Although they clearly demonstrate that the extended protein is addressed to the peroxysome it is still not clear what is the function of this protein in this organelle. There is no clear demonstration of a physiological function for this protein. Is there any genetic disease linked either to the absence of the PTS1 motif in LDHB or to the modification of the readthrough consensus motif identified by the authors?</p><p>9) Concerning the molecular mechanism involved. The fact that this readthrough is UGA specific is very intriguing. Indeed up to now the context is known to modify readthrough efficiency but not in such extend. We would expect variations but not a total absence of readthrough product with the two other stop codons. This is reminiscent to the insertion of Sel that is also specific of the UGA codon. It seems that this possibility can be excluded due to the absence of a SECIS element, but can you clearly exclude an alternative mechanism such an editing or an alternative splicing? Supplementary data answering these questions will be important to provide.</p><p>10) It is unclear whether RTP identifies true readthrough candidates, because there are no obvious negative controls to determine whether the readthrough level is “programmed”.</p><p>The authors select fifteen stop codon contexts scored by the lin model, and measure their readthrough rates using an eYFP-stop-luciferase reporter. The observed readthrough rates – with the exceptions of MDH1 (2.91% readthrough) and LDHB (1.55%) are quite low, ranging from 0.1-0.66%. This is a much lower amount of readthrough than is observed for bovine VEGF-Ax (10%, <xref ref-type="bibr" rid="bib11">Eswarappa et al., 2014</xref>) or various human genes (ranging from 0.7 to 60%, <xref ref-type="bibr" rid="bib10">Dunn et al., 2013</xref>). It is unclear whether this amount of readthrough represents a basal level of readthrough that could be obtained with more or less any stop codon context, or an elevated level of readthrough due to something special about the nucleotide contexts selected by the lin model.</p><p>To address this, the authors should include in this assay a set of negative controls, including a construct lacking luciferase, and a set of contexts with very low RTP scores, to establish a background readthrough rate. The authors should then discuss how far above background, if at all, their readthrough candidates are.</p><p>11) The authors claim that RTP is a genome-wide predictor of readthrough, but they do not substantiate this claim with genome-wide data. The authors should provide a genome-wide examination of readthrough using some sort of experimental measure. For example, they could estimate readthrough rates genome-wide using published ribosome profiling data, and compare these to the predictions made by RTP.</p><p>12) The authors claim that RTP is quantitative, but provide no analysis of its quantitative performance. The authors should provide an explicit analysis demonstrating that RTP is quantitative, or a discussion of why it might not be, especially in light of the fact that they acknowledge that “...there is non-linear contribution to RTP in the sense that other high-RTP genes are likely to show readthrough levels higher than predicted.” Such an analysis might be a quantitative comparison (e.g., a scatter plot and a Pearson correlation coefficient) of predicted and observed readthrough rates for all stop codon contexts tested, or an ROC curve describing the sensitivity and specificity of RTP at different score thresholds. Finally, because RTP is a new metric, descriptive statistics on the RTP score distribution should be provided to facilitate interpretation of individual scores.</p><p>13) As mentioned above, the readthrough rates observed for the vast majority of candidates identified in this work are very low compared to known examples. It therefore seems likely that stop codons undergoing higher levels of readthrough were either missed by RTP, or not tested by the experimenters. In addition, certain cases of readthrough are regulated in rodents (<xref ref-type="bibr" rid="bib59">Yamaguchi et al., 2012</xref>) and flies (<xref ref-type="bibr" rid="bib10">Dunn et al., 2013</xref>; Robinson & Cooley, 1997) or differ between different wild-type yeast strains (Torabi & Kruglyak, 2011). Consistent with these observations, the authors themselves note that LDHB undergoes far more readthrough in human U118 cells (∼5%) compared to human fibroblasts and HEK cells (1-2%). It is therefore possible that these models could be overfit to the cell type in which the experiment was performed, in addition to the specific contexts on which the models were trained. In light of these facts, the authors must establish the scope of their claims.</p></body></sub-article><sub-article article-type="reply" id="SA2"><front-stub><article-id pub-id-type="doi">10.7554/eLife.03640.028</article-id><title-group><article-title>Author response</article-title></title-group></front-stub><body><p>We are now submitting a completely revised manuscript. In this paper, we have addressed all of the concerns raised by the reviewers. The changes we made cover both the experimental work and the written material in all sections of the manuscript. Nearly all figures were revised and new figures and figure panels were included (<xref ref-type="fig" rid="fig2">Figure 2C</xref>, <xref ref-type="fig" rid="fig1s2">Figure 1–figure supplement 2</xref>, <xref ref-type="fig" rid="fig2s1 fig2s2">Figure 2–figure supplement 1 and 2</xref>, and <xref ref-type="fig" rid="fig10s1">Figure 10–figure supplement 1</xref>). We have also restructured the distribution of data between the main and supplementary figures in <xref ref-type="fig" rid="fig5 fig10">Figures 5 and 10</xref> to achieve a more logical succession of arguments. For the same reason, one figure was split into two (now <xref ref-type="fig" rid="fig5 fig6">Figure 5 and 6</xref>), and, following one reviewer’s suggestion, we have deleted one figure (formerly Figure 3). Lastly, we have amended the supplementary database by more <italic>in silico</italic> analyses and an “applet” (see below). Julia Hofhuis has been involved in some of the additional experiments and is now added as a co-author.</p><p>While our manuscript was under review, Eswarappa et al. [Cell 157, 1405-18 (2014)] reported programmed translational readthrough (PTR) in the vascular endothelial growth factor A. This study gave us the opportunity to explain our intentions and the focus of our work in more detail. In their study, Eswarappa et al. investigate a case of programmed translational readthrough, which is dependent on a 63 nucleotide cis-element in the transcript. Remarkably, this case of PTR is relatively independent on the stop codon and its context. In contrast, our work focuses on the ability of the stop codon and its context (stop codon context, SCC) to stimulate translational readthrough. To distinguish this type of readthrough from PTR, we now define this SCC-dependent readthrough as basal translational readthrough (BTR). BTR differs also from pharmacologically induced readthrough. The distinction between PTR and BTR might help to explain why the experimental readthrough values vary over a large range with PTR reaching levels of 10, 30, or even 60% for some genes, while BTR does not seem to exceed 1.5 to 5%. In the latter case, readthrough is entirely dependent on a natural stop codon and the nucleotides in its immediate vicinity, whereas PTR involves cis-elements and trans-factors that can enhance readthrough to a large extend. 1.5% BTR may appear low from the perspective of a PTR-regulated gene, but considering that stop codons usually permit less than 0.13% readthrough, this is an at least 10-fold increase. And for the case of LDHB+P (which we now term LDHBx) we provide evidence that it has a biological function. The distinction between PTR and BTR is not to imply that both are mutually exclusive. Another recent study by Loughran et al. [NAR; <ext-link ext-link-type="uri" xlink:href="http://dx.doi.org/10.1093/nar/gku608">doi 10.1093/nar/gku608</ext-link> (2014)] investigates cases of what is probably a combination of both. To allude to what is the main focus of the paper, we have changed the title to “Peroxisomal lactate dehydrogenase is generated by translational readthrough in mammals”.</p><p><italic>1) The overall rationale of the authors' algorithms needs to be explained in more elaborate, general terms. Further, there needs to be some indication of “public” availability of the algorithm</italic>.</p><p>In the revised version of the manuscript we elaborate in detail the rationale of our <italic>in silico</italic>-approach (see Introduction, Results, and Discussion sections). We also extended the explanations in the Figure legends. We have taken special care to point out that our approach focuses on the nucleotides of the stop codon context (SCC) and can therefore only assess readthrough that is dependent on this relatively short stretch close to the stop codon. We have included the RTP values for all human stop codons (Dataset 1). To further increase the accessibility, we have added the regression coefficients (<xref ref-type="table" rid="tbl2">Table 2</xref>), and we add an ‘applet’ to the dataset (Excel sheet containing a set of cell functions) that allows RTP-calculation of user-entered SCCs.</p><p><italic>2) Dual reporter assays play a key role in the assessment of nonsense codon contexts that are prone to readthrough. However, it appears that all of these assays used by the authors have the traditional format that excludes any introns. This may be a significant problem since deposition of EJC proteins appears to play a key role in translational efficiency (Wiegand et al [2003] Proc Natl Acad Sci USA 100: 11327-11332; Gudikote et al. [2005] Nature Struct Mol Biol 12: 801-819; Nott et al. [2004] Genes Dev 18: 210-222). The latter experiments raise the possibility that the lack of introns/EJC proteins may affect readthrough efficacy. Hence, at a minimum, the authors need to acknowledge this shortcoming of their experimental approach</italic>.</p><p>We would like to thank the reviewers for pointing this out, and helping us clarify the scope of our study. Indeed, we analyzed SCCs, and all of the RTP calculations are entirely based on the regression coefficients of the few nucleotides of the SCC. We do not take into account that cis-sequences or secondary structures within the extension could contribute to readthrough. As stated above, Eswarappa et al. [Cell 157, 1405-1418 (2014)] recently published complementary work by analyzing one example of programmed translational readthrough (PTR), dependent on a 63 nucleotide cis-element 3’ of the stop codon. In contrast, our work concentrates on basal translational readthrough (BTR), which is defined here as being independent of genetic elements outside of the nucleotides surrounding the stop codon. We have identified an SCC motif that leads to high basal translational readthrough (BTR). We also provide evidence that this element acts independently of cis-acting factors in the case of the dual reporter constructs (absence of EJC, responsiveness to aminoglycosides, mutational analysis). The RTP calculation in our study is not based on “aminoglycoside-promoted readthrough efficiency”. We only used “basal” readthrough efficiency, which is non-induced readthrough by definition. Also the BTR levels measured by us for inclusion in the iterative model are non-induced (but inducible!), so they were recorded in the absence of readthrough-inducing drugs, indicating that they are entirely dependent on the SCC. While the inducibility of the signal in the assay sufficiently proves that it is real readthrough (because e.g. splicing is not induced by aminoglycosides), we cannot exclude that some of the genes with high BTR are additionally influenced in translational efficiency by introns/EJC. For the case of LDHBx, however, this is unlikely, due to the quantitative correspondence of LDH activity in the peroxisome with readthrough propensity and its very short extension.</p><p>We do not think that focusing on BTR is a shortcoming of our approach. The genome-wide screens by <xref ref-type="bibr" rid="bib11">Eswarappa et al. 2014</xref> and <xref ref-type="bibr" rid="bib10">Dunn et al. 2013</xref> do not identify LDH and none (Eswarappa) or only 1/42 (Dunn) of the other candidates that we have identified. This is also not a shortcoming of their work. It just means that so far no experimental or <italic>in silico</italic> approach is capable of identifying all cases of translational readthrough in mammals. It also indicates that the detailed molecular mechanisms underlying PTR and BTP are likely different. We believe that a combination of our new approach with ribosome profiling and analysis of conservation of ‘non-coding’ regions together with mass spectrometry will identify the ‘extensome’ that is the full set of proteins with above-average readthrough in the proteome.</p><p><italic>3) The authors make a strong point of being the “first” to observe that nonsense suppression allows two mammalian proteins to be derived from the same mRNA. However, this point is also apparent from the supplementary data presented in Welch et al. (2007) (Nature 447: 87-91). Again, this should be acknowledged</italic>.</p><p>Thank you for pointing this out! We have now deleted a somewhat misleading line from the Abstract and the beginning of the Discussion. We do not wish to claim, we are ‘the “first” to observe that nonsense suppression allows two mammalian proteins to be derived from the same mRNA’. Throughout the manuscript, we are quoting several papers together reporting more than 50 proteins that have shown this before [<xref ref-type="bibr" rid="bib10">Dunn et al. 2013</xref>, <xref ref-type="bibr" rid="bib27">Jungreis et al. 2011</xref>, <xref ref-type="bibr" rid="bib18">Geller and Rich 1980</xref>, and now additionally <xref ref-type="bibr" rid="bib9">Chittum et al. 1998</xref>, <xref ref-type="bibr" rid="bib11">Eswarappa et al. 2014</xref>, and <xref ref-type="bibr" rid="bib36">Loughran et al. 2014</xref>]. We have now included a diagram (<xref ref-type="fig" rid="fig2s2">Figure 2–figure supplement 2</xref>) that gives an overview on the experimentally confirmed mammalian readthrough genes. Together, with <xref ref-type="bibr" rid="bib11">Eswarappa et al. 2014</xref> our study is the first reporting functional translational readthrough in humans, defined as a readthrough event, in which the normal and the extended form have distinct physiological roles and/or localizations. Nonetheless, we have toned down all reference to being “first” in the manuscript.</p><p>In our study, we have used the pharmacological induction of readthrough only to support the idea that what we are measuring is indeed readthrough (as opposed to splicing, or RNA editing, see below). We have not included the genes identified by Welch et al. because in our study we only focus on genes undergoing detectable levels of readthrough without drug treatment.</p><p><italic>4) The authors do not discuss/compare the consensus sequence identified from their analyses with any of the stop codon sequence contexts that have been reported in the literature. A comparison with known sequence contexts would be helpful for the readers and may exemplify the strength of the RTP predictor</italic>.</p><p>We have now expanded the Discussion section and compare the SCCs with already published readthrough motifs to point out the strength of the RTP predictor. In addition, we have now included a diagram (<xref ref-type="fig" rid="fig2s2">Figure 2–figure supplement 2</xref>) that displays the experimentally confirmed mammalian readthrough genes in context.</p><p><italic>5) The authors use one specific case study, the peroxisomal targeting sequence, to show physiological functions of a readthrough protein. However, adding one more example, other than the PTS, would greatly enhance the applicability of the RTP predictor combined with other in silico analyses. Experimental validation of another case study might be beyond the scope of the paper, but it would show a wider applicability of the method described</italic>.</p><p>To enhance the applicability of the RTP algorithm, we now include more <italic>in silico</italic> analysis of the readthrough extensions, containing potential endoplasmic reticulum retention signals, glycosylation sites, transmembrane domains, and possible farnesylation sites (Database 1).</p><p><italic>6) In the experimental validation of the role of readthrough in the synthesis of LDHB+P, the authors should demonstrate the validity of their RTP predictor by doing mutational analyses of the stop codon context according to their consensus models</italic>.</p><p>We already were working on an extended mutational analysis of the LDHB+P (now LDHBx) stop codon context according to our consensus model while the paper was under review. In the revised manuscript, we show in <xref ref-type="fig" rid="fig3">Figure 3B</xref> the result of our analysis. In addition we have measured more candidates with the LINfs3 consensus (<xref ref-type="fig" rid="fig2">Figure 2C</xref>).</p><p><italic>7) Although the authors have an intriguing model for the co-import of LDHA with LDHB, It would be important to show two-hybrid data to support it, or address it biochemically using co-IP analyses of the LDHB+P and LDHA</italic>.</p><p>We use a two-hybrid (albeit not a yeast two-hybrid) experiment that shows the interaction of LDHBx with LDHA. Numerous descriptions in the literature indicate that any protein can enter peroxisomes when it interacts with a protein that is imported into peroxisomes [eg. McNew et al. TiBS 21, 54-58 (1996), Thoms et al. J Struct Biol 175, 362-371 (2011)]. The two-hybrid assay is based on this co-import and uses LDHA fused with the yellow fluorescent protein (YFP) in combination with LDHBx fused to the C-terminus of the cyan fluorescent protein (CFP). We show that overexpressed YFP-LDHA can only enter the peroxisome, when CFP-LDHBx is co-expressed (<xref ref-type="fig" rid="fig10">Figure 10</xref>). When we mutate the targeting signal of LDHBx in the extension, LDHA cannot enter the peroxisome any more (<xref ref-type="fig" rid="fig10s1">Figure 10–figure supplement 1</xref>). We checked if the amino acid extension present in LDHBx could potentially interfere with the interaction of LDHB with LDHA. To answer this question, we now show the structure of tetrameric LDH in <xref ref-type="fig" rid="fig10s2">Figure 10–figure supplement 2</xref>. The C-termini of LDHB are at the very surface of LDH with maximum distance from the interaction surface of the protomers. We can conclude the following from this analysis: (1) The (extended) C-terminus of LDHBx is unlikely to interfere with the oligomerization in the tetramer, and (2) the C-terminal extension is easily accessible to the peroxisomal import receptor PEX5 that must bind the short PTS1 contained in the additional amino acids. If the extension was buried inside the protein, it could affect conformation of LDHBx to interfere with LDHA binding, and the extension would be inaccessible to PEX5, and thereby LDHBx-containing oligomers could not be targeted to the peroxisome.</p><p><italic>8) Although they clearly demonstrate that the extended protein is addressed to the peroxysome it is still not clear what is the function of this protein in this organelle. There is no clear demonstration of a physiological function for this protein. Is there any genetic disease linked either to the absence of the PTS1 motif in LDHB or to the modification of the readthrough consensus motif identified by the authors?</italic></p><p>At least six previous studies found LDH activity and/or LDH protein(s) in the peroxisome (<xref ref-type="bibr" rid="bib41">McGroarty et al., 1974</xref>; <xref ref-type="bibr" rid="bib45">Osmundsen, 1982</xref>, <xref ref-type="bibr" rid="bib56">Völkl and Fahimi, 1985</xref>; <xref ref-type="bibr" rid="bib2">Baumgart et al., 1996</xref>, <xref ref-type="bibr" rid="bib40">McClelland et al., 2003</xref>; <xref ref-type="bibr" rid="bib21">Gronemeyer et al., 2013</xref>). One of them (<xref ref-type="bibr" rid="bib2">Baumgart et al., 1996</xref>) experimentally addresses the function of the protein in this organelle and suggested a role in NAD<sup>+</sup> regeneration. In spite of all these reports of peroxisomal LDH association, it has never been universally accepted that LDH indeed enters the peroxisome and indeed exerts a function in peroxisomes, because a targeting signal or targeting mechanism could not be identified. In the second, cell biological part of our paper we therefore chose to study the readthrough-dependent targeting (as a function of the readthrough-extension) of LDHBx to peroxisomes. We not only identify a (cryptic) peroxisomal targeting signal, and peroxisomal localization of a protein, but also provide evidence for readthrough-dependent targeting to the peroxisome. We show that targeting (1) occurs in untransfected cells; (2) is dependent on the stop codon: UAA, a tighter stop codon than UGA, abrogates targeting, whereas UGG, a sense-mutation of the stop, improved targeting; and (3) that aminoglycoside –treatment increases the amount of LDH in the peroxisome. Our work provides an answer to this long-standing question, of how LDH enters the peroxisome.</p><p>At the moment, the LDHBx extension is expected to be in the 3’UTR, so it is not under scrutiny when it comes to the identification of genetic diseases by exon sequencing. We are however, discussing what is probably the only known disease associated with the LINfs3 consensus (<xref ref-type="bibr" rid="bib46">Pacho et al. 2011</xref>).</p><p><italic>9) Concerning the molecular mechanism involved. The fact that this readthrough is UGA specific is very intriguing. Indeed up to now the context is known to modify readthrough efficiency but not in such extend. We would expect variations but not a total absence of readthrough product with the two other stop codons. This is reminiscent to the insertion of Sel that is also specific of the UGA codon. It seems that this possibility can be excluded due to the absence of a SECIS element, but can you clearly exclude an alternative mechanism such an editing or an alternative splicing? Supplementary data answering these questions will be important to provide</italic>.</p><p>We now show in <xref ref-type="fig" rid="fig3">Figure 3B</xref> that the readthrough for LDHBx is indeed UGA specific, because we find absence of readthrough (or only background levels) with the other stop codons or mutations in the LINfs3 consensus. We agree about the absence of SECIS element. However, we analyzed LDHB and the other experimentally tested transcripts regarding potential A-to-I editing sites using the RADAR database (<ext-link ext-link-type="uri" xlink:href="http://rnaedit.com">http://rnaedit.com</ext-link>), and based on these results we can exclude RNA editing in these transcripts. We are certain that alternative splicing does not occur that close to the stop codon, because alternatively spliced transcripts would be included in the Ensembl database and as such would have been subject of our genome-wide <italic>in silico</italic> screen. Perhaps the most convincing argument ruling out RNA editing or splicing is our finding that the apparent readthrough is strongly induced (in some cases more the 40-fold) by aminoglycosides in the SCC as well as in the full-length experiments. Neither RNA editing nor splicing are known to be stimulated by these drugs.</p><p><italic>10) It is unclear whether RTP identifies true readthrough candidates, because there are no obvious negative controls to determine whether the readthrough level is “programmed”</italic>.</p><p><italic>The authors select fifteen stop codon contexts scored by the lin model, and measure their readthrough rates using an eYFP-stop-luciferase reporter. The observed readthrough rates</italic> – <italic>with the exceptions of MDH1 (2.91% readthrough) and LDHB (1.55%) are quite low, ranging from 0.1-0.66%. This is a much lower amount of readthrough than is observed for bovine VEGF-Ax (10%,</italic> <xref ref-type="bibr" rid="bib11"><italic>Eswarappa et al., 2014</italic></xref><italic>) or various human genes (ranging from 0.7 to 60%,</italic> <xref ref-type="bibr" rid="bib10"><italic>Dunn et al., 2013</italic></xref><italic>). It is unclear whether this amount of readthrough represents a basal level of readthrough that could be obtained with more or less any stop codon context, or an elevated level of readthrough due to something special about the nucleotide contexts selected by the lin model</italic>.</p><p><italic>To address this, the authors should include in this assay a set of negative controls, including a construct lacking luciferase, and a set of contexts with very low RTP scores, to establish a background readthrough rate. The authors should then discuss how far above background, if at all, their readthrough candidates are</italic>.</p><p>Our data and experiments contain negative controls in the sense that we have included many genes with low RTP and therefore a low BTR, for example PPP1R3F (0.18%), or PRDM10 (0.13%), or THG1L (0.15%) (<xref ref-type="fig" rid="fig1">Figure 1B,C</xref> and <xref ref-type="table" rid="tbl1">Table 1</xref>). Following the reviewers’ suggestion, we have now included a new negative control that is even better suited than a luciferase-less control to establish the background levels: We use a construct that has two successive stop codons. This construct shows a BTR of 0.13%, indicating that this is the background level below we cannot distinguish between readthrough and experimental noise. As we neither want to obscure this fact, nor do we want to artificially scale our data, we decided not to subtract this value in the BTR measurement. Instead we indicate the background level in <xref ref-type="fig" rid="fig1">Figure 1C</xref> by a red line. As stated above, due to the specific form of the dual reporter assay we are using, cis-element dependent readthrough cannot be the origin of the readthrough levels we are measuring. The high-RTP consensus emerged from data that was in no way primed to readthrough, so we are confident (even if we cannot prove it), that readthrough of more than 5%, depending on the cell type, cannot be obtained by BTR. For PTR (<xref ref-type="bibr" rid="bib11">Eswarappa et al., 2014</xref>, and probably also Dunn et al., even though the latter study does not analyze the mechanism leading to readthrough) there seem to be no upper limit of readthrough levels, depending on the type of cis-elements and trans-acting factors, even 100% readthrough seem possible. Again, we are very thankful that the reviewers’ comments have given us the opportunity to distinguish between cis-element dependent readthrough (PTR) and SCC-dependent readthrough (BTR). It is clear that BTR can (Loughran et al.) but need not (Eswarappa et al.) be modulated by PTR. And, as stated above and below, the BTR of approx. 1.5% for LDHBx is not low, because, as we explain in the Discussion, it leads to a roughly equal distribution of LDH in the peroxisome and in the cytosol. With higher readthrough, the peroxisomal LDH activity would exceed the cytosolic! Regarding our prediction of readthrough for all human transcripts we are aware that many of the listed RTP values correspond to what could be interpreted as the ‘baseline’ of measurable BTR. We are also aware that negative RTP values are somehow counter-intuitive, as they would suggest a negative BTR, which is not possible. Here, a calibration of RTP values to BRT probabilities by means of, e.g., analysis of the distribution of RTP scores could help to assess more intuitively interpretable RTP values. However, with the limited dataset and the apparent nonlinear dependence of BTR on the SCC a calibration is not possible at the moment. A workaround would be to just include the first 144 (consensus) sequences in Dataset 1 (or trim the list according to the first negative control or an arbitrary threshold), but we feel that readers could be interested in the complete list. (Another possibility for avoiding negative RTP would have been to use “constrained” estimation methods, i.e. methods that avoid negative coefficients. However, for our first approach we wanted to use the simplest methods possible to take into account the limited number of sequences.) We hope that extended datasets of BTR measurements and nonlinear methods in the future will provide improved RTP values.</p><p><italic>11) The authors claim that RTP is a genome-wide predictor of readthrough, but they do not substantiate this claim with genome-wide data. The authors should provide a genome-wide examination of readthrough using some sort of experimental measure. For example, they could estimate readthrough rates genome-wide using published ribosome profiling data, and compare these to the predictions made by RTP</italic>.</p><p>We have now linked our genome-wide <italic>in silico</italic> screen better to existing experimental data on translational readthrough. We have analyzed more candidates to validate the LINfs3 (<xref ref-type="fig" rid="fig2">Figure 2C</xref>). We have extended the Discussion on the hits found in several studies comparing them in more detail. We have also added a figure (supplement to <xref ref-type="fig" rid="fig2">Figure 2</xref>) describing the overlap between our data and the previously identified readthrough proteins. This includes a search for existing readthrough proteins for the consensus described in our paper and an extended the discussion on why there is relatively little overlap between the different approaches applied so far. And we have extended the discussion on the regression between measured BTR and RTP.</p><p><italic>12) The authors claim that RTP is quantitative, but provide no analysis of its quantitative performance. The authors should provide an explicit analysis demonstrating that RTP is quantitative, or a discussion of why it might not be, especially in light of the fact that they acknowledge that “...there is non-linear contribution to RTP in the sense that other high-RTP genes are likely to show readthrough levels higher than predicted.” Such an analysis might be a quantitative comparison (e.g. a scatter plot and a Pearson correlation coefficient) of predicted and observed readthrough rates for all stop codon contexts tested, or an ROC curve describing the sensitivity and specificity of RTP at different score thresholds. Finally, because RTP is a new metric, descriptive statistics on the RTP score distribution should be provided to facilitate interpretation of individual scores</italic>.</p><p>We added scatter plots showing the correlation of BTR and RTP for 81 sequences for the LINiter as well as the LINfs3 model (Figure supplements to <xref ref-type="fig" rid="fig1 fig2">Figures 1 and 2</xref>). Here, the RTP values have been obtained from predictions within the leave-one-out cross-validation and are therefore an indicator of the generalization capacity of our approach. The plots show a nonlinear dependency of BTR and RTP for both models, indicating that nonlinear regression models could provide a better fit. The Pearson correlation for both models is significant but weak with the LINfs3 model showing a slightly higher correlation (0.41, p=0.0001) than the LINiter model (0.34, p=0.0022). This suggests that the LINfs3 model with its only 15 parameters is more suitable for our purpose than the LINiter model with 51 parameters. Due to the lack of true negative readthrough (readthrough 0%) examples a computation of classification performance indices such as sensitivity and specificity is not possible. We do not claim RTP to be a new metric; we rather introduce this notion to keep the formulation throughout the manuscript as concise as possible.</p><p><italic>13) As mentioned above, the readthrough rates observed for the vast majority of candidates identified in this work are very low compared to known examples. It therefore seems likely that stop codons undergoing higher levels of readthrough were either missed by RTP, or not tested by the experimenters. In addition, certain cases of readthrough are regulated in rodents (</italic><xref ref-type="bibr" rid="bib59"><italic>Yamaguchi et al., 2012</italic></xref><italic>) and flies (</italic><xref ref-type="bibr" rid="bib10"><italic>Dunn et al., 2013</italic></xref><italic>; Robinson & Cooley, 1997) or differ between different wild-type yeast strains (Torabi & Kruglyak, 2011). Consistent with these observations, the authors themselves note that LDHB undergoes far more readthrough in human U118 cells (∼5%) compared to human fibroblasts and HEK cells (1-2%). It is therefore possible that these models could be overfit to the cell type in which the experiment was performed, in addition to the specific contexts on which the models were trained. In light of these facts, the authors must establish the scope of their claims</italic>.</p><p>The BTR readthrough rates are in the low percent range. As detailed above, we provide evidence that these rates are entirely dependent on the SCC and do not require cis-elements or trans-factors. Considering our finding that readthrough of some of the LINfs3 consensus candidates are at least 10-fold above the physiological background, we do not think that, from the perspective of a stop codon and its’ usual function to stop translation, these are low rates. It is clear that the RTP approach would miss any cases of readthrough that is independent of the stop codon, and we now state this more clearly in the Introduction, the Results and in the Discussion. Overfitting: we are aware that the number of 66/81 sequences represents the lower limit for a multivariate regression on 51 dimensions. However, with that knowledge in mind we deliberately restricted our analysis to linear models, applied a rigorous model validation using a leave-one-out strategy and performed a feature selection procedure to further reduce the number of model parameters (to 15 in the case of LINfs3). The fact that we could identify the relevant position of the SCC as well as the consensus motif for high RTP assured us that our model generalizes as good as possible for linear approaches. Furthermore, we pave the way for future extended datasets that now can be built by concentrating on the relevant SCC positions. This reduces the number of experiments for fully specified BTR assessment based on the SCC considerably (from 4^12*3 ∼ 50,000,000 to 4^3*3=192) and will allow the application of nonlinear models for RTP prediction. The additional experiments we have added to test the LINfs3 consensus (<xref ref-type="fig" rid="fig2">Figure 2C</xref>) suggest that the predictions derived from one cell type in principle can be applied to another, although the overall level of BTR may differ between cell types.</p></body></sub-article></article> |