Control of telomere length in yeast by SUMOylated PCNA and the Elg1 PCNA unloader

Telomeres cap and protect the linear eukaryotic chromosomes. Telomere length is determined by an equilibrium between positive and negative regulators of telomerase activity. A systematic screen for yeast mutants that affect telomere length maintenance in the yeast Saccharomyces cerevisiae revealed that mutations in any of ~500 genes affects telomere length. One of the genes that, when mutated, causes telomere elongation is ELG1, which encodes an unloader of PCNA, the processivity factor for replicative DNA polymerases. PCNA can undergo SUMOylation on two conserved residues, K164 and K127, or ubiquitination at lysine 164. These modifications have already been implicated in genome stability processes. We report that SUMOylated PCNA acts as a signal that positively regulates telomerase activity. We also uncovered physical interactions between Elg1 and the CST (Cdc13-Stn1-Ten) complex and addressed the mechanism by which Elg1 and Stn1 negatively regulates telomere elongation, coordinated by SUMO. We discuss these results with respect to how chromosomal replication and telomere elongation are coordinated.


Introduction
Telomeres cap the ends of linear eukaryotic chromosomes and safeguard them against genome instability. Telomeric DNA consists of G/C-rich DNA repeats, with the G-rich strand extending to form a 3' single-stranded overhang. Telomeres are primarily involved in counteracting the gradual shortening of the telomeric DNA caused by conventional DNA replication, thereby solving the 'end-replication problem' (Bonnell et al., 2021;Soudet et al., 2014). Another important function of telomeres is to protect chromosome ends from improper recognition as a DNA double-strand break (DSB) (reviewed in de Lange, 2018).
An important, conserved protein complex involved in the regulation of telomere length is the CST, which in Saccharomyces cerevisiae is composed of Cdc13, Stn1, and Ten1 (Chen and Lingner, 2013;Ge et al., 2020;Lim et al., 2020;Puglisi et al., 2008;Rice and Skordalakes, 2016). The CST has similarity to the ssDNA binding complex RPA and localizes specifically to the single-stranded telomeric DNA, where it is involved in chromosome end capping and telomere length regulation (Hughes et al., 2000;Lin and Zakian, 1996;Mersaoui and Wellinger, 2019). The CST facilitates

Results
The long telomeres phenotype of elg1Δ strains requires PCNA SUMOylation Previous results suggested that the Elg1 protein preferentially interacts with the SUMO machinery and with SUMOylated PCNA (Parnas et al., 2011;Parnas et al., 2010;Shemesh et al., 2017). To test whether PCNA modifications play any role in the elongated telomere phenotype of elg1Δ, we combined the ELG1 deletion with mutations in the POL30 gene, encoding PCNA. We mutated either lysine 127, lysine 164, or both (hereafter referred as pol30-K127R, pol30-K164R, and pol30-RR, respectively) in the elg1Δ background. Figure 1 shows that the mutations in POL30 have little effect by themselves. Importantly, whereas the single mutants did not affect the long telomeres of elg1Δ, mutating both lysine residues of each PCNA subunit completely abrogated the long telomere phenotype. Lysine 164 can be modified by both SUMO and ubiquitin; mutating only this residue prevents ubiquitination, but not SUMOylation. Figure 1 demonstrates that this mutant shows no effect. To ensure that SUMO, and not ubiquitin, is the modification responsible for the long telomeres of elg1Δ, we deleted RAD18, encoding the E3 ubiquitin ligase required to ubiquitinate PCNA (Hoege et al., (Teloblot) showing that lack of SUMOylation of PCNA prevents telomere elongation. Independently created colonies were passaged ten times, its DNA extracted, digested with XhoI, and run in an agarose gel. The DNA was then transferred to a nitrocellulose membrane, which was incubated with a radioactive probe that detects telomeres, and a size marker. (B) Teloblot showing that overexpression of wild-type PCNA or Pol30-RR-SUMO fusion, but not Pol30-ubiquitin fusion or Pol30-RR causes telomere elongation.  Stelter and Ulrich, 2003). Figure 1 shows that lack of Rad18 has no effect on the elongated telomeres of elg1Δ. (Deletion of the SUMO E3 ligases SIZ1 and SIZ2 [to prevent SUMOylation] by itself alters telomere length [Hang et al., 2011] and thus they could not be used here.) We conclude that SUMOylation of PCNA is required for the long telomeres phenotype of elg1Δ mutants.
These results suggest that SUMOylated PCNA may play a central role in determining telomere length also in cells with wild-type Elg1 activity. We thus next asked whether artificially overexpressing modified and unmodified PCNA may mimic the long telomere phenotype of elg1Δ. Since it is hard to force ubiquitination or SUMOylation of PCNA in vivo, we used covalent fusions between PCNA and either ubiquitin or SUMO. These have been shown in the past to mimic naturally modified PCNA molecules (Parker et al., 2007;Takahashi et al., 2020). We overexpressed either wild-type Pol30, Pol30-RR (which cannot be modified) ,or Pol30-RR fused either to ubiquitin or to SUMO, in pol30-RR cells, unable to carry out modifications of the PCNA encoded by the genome. Figure 1B shows that telomeres become elongated upon overexpression of the wild-type protein (which can be modified), or the Pol30-RR version fused to SUMO, but do not elongate upon overexpression of the unmodifiable Pol30-RR version, or the one fused to ubiquitin. These results, together with those of Figure 1A, show that accumulation of SUMOylated PCNA, and not unmodified or ubiquitinated PCNA, is both necessary and sufficient for telomere elongation in the presence or in the absence of Elg1.

Genetic and physical interactions between Elg1 and Stn1
The elongated telomere phenotype of Elg1 was identified in systematic screens of the yeast knockout collection (Askree et al., 2004;Gatbonton et al., 2006), together with ~200 additional mutants of similar phenotypes. To identify the pathway of telomere length maintenance in which Elg1 participates, we systematically combined the elg1Δ mutant with all the other mutants with long telomeres. Most double mutants exhibited telomeres longer than each of the single mutants, indicating that they affect different pathways. One of the mutants showing clear epistasis (the double mutant was not longer than the single mutants) carried a deletion of RRP8, a gene involved in modification of the ribosomal RNA. Further analysis demonstrated that the knockout of RRP8 exhibited the 'neighboring gene effect,' and in fact the deletion of RRP8 caused a decrease in the expression of its neighboring gene, STN1 (Ben-Shitrit et al., 2012). We confirmed the genetic interactions between ELG1 and STN1 alleles by combining elg1Δ with the stn1-13 and stn1-164 alleles (Grandin et al., 1997), which exhibit long telomeres. The telomeres of the double mutants are not longer than those of the single mutants ( Figure 2A), indicating that Elg1 works in the same TLM pathway as Stn1.
Given the epistatic relations between the ELG1 and STN1 genes, we addressed, by co-immunoprecipitation (IP) and yeast two-hybrid (YTH) assay, whether the proteins in the CST complex interact with Elg1. IP of epitope-tagged Elg1 pulled down Stn1 but not the other two members of the CST complex, Cdc13 or Ten1 ( Figure 2B). For our YTH experiments, the Elg1 protein was divided into three main functional domains: an N-terminal domain (amino acids 1-234), a central AAA domain (amino acids 235-514), and a C-terminal domain (amino acids 515-791) (Itzkovich et al., 2023; Figure 3A). The AAA and C-terminal domains contain the conserved RFC boxes and are required for the interactions with the small RFC subunits, shared by all RFC-like domains. The N-terminal domain (hereafter referred to as NTD) is unique to Elg1 (Arbel et al., 2021). We thus concentrated on this domain and used the C-terminus as a negative control as it does not interact with any of the proteins tested here.
The NTD contains SUMO-interacting motifs (SIMs), which were previously shown to interact with SUMO and the SUMO machinery (Parnas et al., 2011;Parnas et al., 2010;Figure 3A). We used the YTH technique to test potential interactions between Elg1 and the CST components, and their dependency on SUMO. The N-terminus of Elg1 interacted strongly with Cdc13 and Stn1, and weakly with Ten1 ( Figure 3B, C, Figure 3-figure supplement 1). In contrast, no interaction was detected with Elg1's CTD. We could also detect interactions of Cdc13, Stn1, and Elg1's NTD with SUMO ( Figure 3). These were used as positive controls in all our experiments. Since the results with Ten1 were much weaker than those with Cdc13 and Stn1, we concentrated on these two last proteins. Below, we The online version of this article includes the following source data for figure 1: Source data 1. Source data is the original Southern blot.

Figure 1 continued
dissect, using YTH in various genetic backgrounds, the interactions between Elg1, Stn1, Cdc13, and PCNA. The results are also summarized in Table 1.
Interestingly, the interaction of Elg1's NTD with Stn1, but not with Cdc13, was abolished when the SIM motifs (amino acids I27A, I93K, I121A, and I122A) were mutated ( Figure 3B and C). We confirmed these results by co-immunoprecipitation (co-IP) ( Figure 2B). These results suggest that the interaction with Stn1 is mediated by SUMOylation of a protein, whereas the interaction of Cdc13 with Elg1-NTD is independent of the SIM in the plasmid-borne copy of Elg1. We confirmed this result by deleting the genes encoding the SUMO-specific E3 enzymes, SIZ1 and SIZ2 in the genome (a double deletion is necessary, as many times each protein can compensate for the lack of the other). Figure 3D shows that indeed deletion of the genes encoding these enzymes abolishes the interaction of Stn1 and Cdc13 with Elg1's NTD. We conclude that SUMO plays a role in mediating the Elg1-CST interactions. Since Elg1 preferentially interacts with SUMOylated PCNA (Parnas et al., 2010) and Stn1 can bind SUMO noncovalently in YTH ( Figure 3C), we reckoned that SUMOylated PCNA may mediate the interaction between Stn1 and Elg1. We therefore tested whether abolishing PCNA's SUMOylation The online version of this article includes the following source data for figure 2: Source data 1. The source data contains the Southern blots and the Western blots used to make the figure.
sites in the genome (pol30-RR) had an effect on the Stn1-Elg1 interaction. We found, however, that Elg1 and Stn1 could still interact in the pol30-RR strain ( Figure 3D), despite the fact that PCNA cannot become SUMOylated. Consequently, a different SUMOylated target seems to mediate the interaction between Elg1 and Stn1, or else either one of the proteins is SUMOylated. Despite numerous attempts, we were unsuccessful in observing SUMOylation of either Stn1 or Elg1.
We also noted that, despite the fact that mutation in the plasmid-borne NTD's SIM had no effect on the interaction with Cdc13 ( Figure 3B), deletion of SIZ1 and SIZ2, and mutation in the SUMOylation sites in PCNA completely abolished the interactions between Elg1's NTD and Cdc13 ( Figure 3D). This implies that PCNA SUMOylation is necessary for the interaction between Elg1 and Cdc13. We confirmed by co-IP that Cdc13 indeed physically interacts with PCNA ( Figure 3E). The interaction The online version of this article includes the following source data and figure supplement(s) for figure 3: Source data 1. The Source data contains the Western blots used in Figure 3E.  between Cdc13 and PCNA was almost completely abolished when PCNA could not be SUMOylated ( Figure 3-figure supplement 2), In summary, the interaction between Cdc13 and both PCNA and Elg1, but not that of Elg1 with Stn1, is dependent on SUMOylation of PCNA.

Elg1's functional activity is essential for its interaction with Cdc13
When we repeated the YTH assays in a strain in which the ELG1 gene is deleted from the genome, we were surprised to see that now Cdc13 failed to interact with the plasmid-borne Elg1-NTD ( Figure 4A, compare to Figure 3B). In contrast, the interaction between Stn1 and Elg1-NTD was not affected by the deletion of the genomic ELG1. Taken together, these results imply that whereas Stn1 interacts directly with the plasmid-borne Elg1-NTD, via its SIM-mediated SUMO-binding, the interaction with Cdc13 require both PCNA SUMOylation and the genomic ELG1 gene. We can envision two formal possibilities for the latter requirement: either (1) the interaction with Cdc13 requires parts of the Elg1 protein not present in the plasmid-borne NTD (e.g., the AAA or CTD domain) or (2) Elg1's activity (e.g., PCNA unloading) may be required to facilitate the physical interaction of the NTD with Cdc13.
To distinguish between these two possibilities, we carried out YTH experiments with strains carrying different alleles of ELG1 in the genome. Three mutations were tested first: elg1-sim (mutated in amino acids I27A, I93K, I121A I122A), which does not interact with SUMO; elg1-TT386/387DD, hereafter referred to as elg1-DD, which affects the interface Elg1-PCNA ; and elg1-sim+DD, an allele that combines both mutations. Whereas the SIM mutation has a very mild effect on the ability of cells to unload PCNA, and displays no phenotypes, the elg1-DD is more impaired and has a stronger phenotype, and the elg1-sim+DD allele has essentially a null phenotype, similar to that of elg1Δ, but expressing the protein at normal levels (Itzkovich et al., 2023;Shemesh et al., 2017).
We found that in strains carrying genomic elg1-DD and elg1-sim+DD alleles, Cdc13 failed to interact with plasmid borne Elg1-NTD, whereas in the elg1-sim strain the interaction was weak but clearly seen ( Figure 4B-D). This observation indicates that the PCNA unloading activity of Elg1 is essential to allow the interaction between Cdc13 and the plasmid borne Elg1-NTD: this activity is high in elg1-sim mutants, intermediate in the elg1-DD strain, and is abolished by the elg1-sim+DD mutations. Note that the presence of a functional SIM at the plasmid-borne NTD is sufficient to ensure a mild interaction with Cdc13, but only if the genomic copy retains PCNA unloading activity (as in the elg1-sim mutant). On the other hand, the presence of a functional SIM in the inactive elg1-DD allele does not warrant an interaction of Elg1's NTD (which has SIM motifs) with Cdc13. We confirmed this idea by analyzing the interaction between Cdc13 and Elg1's NTD in another SIM-containing mutant strain devoid of PCNA unloading activity, elg1-WalkerA (KK343/4DD), which lacks ATPase and PCNA unloading activity (Itzkovich et al., 2023). This mutant expresses wild-type levels of protein, and the mutant has essentially the same phenotype of an elg1Δ allele. Figure 4E shows that no interaction between the plasmid-borne Elg1 NTD and Cdc13 could be detected despite the fact that functional SIMs are present both at the NTD and in the genome-encoded Elg1-WalkerA protein. All the mutants expressed the Elg1 protein at wild-type levels ( Figure 4-figure supplement 1). We conclude that whereas the interaction with Stn1 is direct and mediated by the SUMOylation of a protein (different from PCNA), the interaction between Elg1's NTD and Cdc13 is indirect and can only take place after PCNA has been SUMOylated and unloaded. This suggests a model for the coordination of telomerase activity and DNA replication: the presence of SUMOylated PCNA at telomeres may serve as a positive signal for telomerase activity. Once Elg1 unloads PCNA, however, Stn1-Elg1 can bind Cdc13, ending telomerase activity. In this context, it is possible to understand the results seen in Figure 3D, showing that mutations in the SUMO E3 enzymes or in the sites of SUMOylation of PCNA, both of which prevent PCNA SUMOylation, show the same effect as the sim+DD or walkerA mutations in the genomic ELG1 copy, which prevent PCNA unloading activity.  Figure supplement 2. SUMOylation of the genomic copy of Cdc13 has no effect on the interactions of a wt copy with Stn1 or Elg1.

SUMOylation of Cdc13 is needed for interaction with Elg1
Cdc13 undergoes SUMOylation, and this modification plays an essential role in the negative regulation of telomere length (Hang et al., 2011). Thus, Cdc13 is a good candidate for the target of SUMOylation required for the interaction between Elg1 and Stn1. Accordingly, mutations of the Cdc13 SUMOylation sites (cdc13-snm, 'SUMO no more') weaken its interaction with Stn1 and lead to elongated telomere phenotype (Hang et al., 2011). We therefore checked whether the cdc13-snm mutation affects also the interaction with Elg1. Figure 4F shows that indeed preventing SUMOylation of Cdc13 abolishes the interaction between this protein and Elg1. Note that the strain used expresses a wild type genomic copy of CDC13. Thus, SUMOylation of the plasmid-borne version of Cdc13 (i.e., in cis) is required for these interactions in the YTH assay. No effect was observed when the YTH plasmid carried a wild-type copy of CDC13, and the genomic copy was the cdc13-snm allele (Figure 4-figure supplement 2A). In summary, the N-terminus of Elg1 interacts with Cdc13 only if (1) Cdc13 can be SUMOylated ( Figure 4F), (2) PCNA can be SUMOylated ( Figure 3D), and (3) PCNA can be unloaded ( Figure 4A, C-E).
The interaction of Elg1 with Stn1 takes place only at late S-phase The fact that the interactions of the Elg1 NTD with Cdc13 are dependent on PCNA modification and unloading, whereas those with Stn1 are not, suggests that the interaction with Cdc13 may be mediated by, or dependent on, the interaction of Elg1 with Stn1. To further dissect this point, we divided the Stn1 protein into an NTD (first 281 amino acids) and a CTD (amino acids 282-495). The later has been shown to be the region that interacts with Cdc13 (Petreaca et al., 2007). Figure 4-figure supplement 2B shows that the NTD of Elg1 also interacts with Stn1 via its C-terminal domain, both in WT and elg1Δ strains.
Having established Stn1 as the main interactor of Elg1 in the CST complex, we next examined the interaction between these proteins throughout the cell cycle phases. We arrested a strain with tagged Elg1, Stn1, and DNA Polymerase Delta in G1 with alpha factor. Cells were released into the cell cycle and samples were taken at intervals during two cell cycles. This allowed us to map the timing of interaction between Elg1 and Stn1 proteins, and whether this interaction is codependent on the movement of replication fork. Figure 5-figure supplement 1 shows that the total level of these proteins does not change throughout the cell cycle. We immunoprecipitated Elg1 and measured the level of the other two proteins by western blot (Figure 5A-C). Pol3 was co-IPed with Elg1 strongly only during the S-phase (40-70 min, 120-140 min), whereas Stn1 was only detected in late S-phase (60-70 min, 140-160 min) for both cell cycles and was barely detected in G1 ( Figure 5A-C). This pattern coincides with telomerase activity at telomeres, which is low in early to mid S-phase and peaks in late S-phase (Puglisi et al., 2008;Taggart et al., 2002).
To monitor the arrival of the replication fork to telomeres, we performed chromatin immunoprecipitation at telomeres (Telo-ChIP) in synchronized cells. Figure 5D shows that the timing of arrival of PCNA to the telomeres (60-70 min into the cell cycle) coincides with the timing of the co-IP between Elg1 and Stn1 ( Figure 5A-C). Thus, our results are consistent with the idea that Elg1 moves with the replication fork, and interacts with Stn1 at the end of S-phase, when telomeres are replicated and telomerase is activated.

The interaction between Cdc13 and Stn1 is dependent on Elg1
Our YTH and IP data suggest that unlike the relationship between Cdc13 and Elg1, the interaction between Stn1 and Elg1 is direct (Figures 2 and 3). Since both Elg1 and Cdc13 bind Stn1 at its C-terminus (Figure 4-figure supplement 1 and Petreaca et al., 2007), it is possible that the absence of Elg1 may affect the interaction between the two CST members. We thus measured the co-IP of Cdc13 and Stn1 in a wild-type or an elg1Δ strain. Cells were cell-cycle synchronized in S-phase as the interaction between these two proteins was barely seen when immunoprecipitated in an asynchronous culture. When Elg1 was absent, Stn1 exhibited a strong reduction in its co-precipitation with Cdc13 ( Figure 5E and F). This suggests that the interaction between Stn1 and Cdc13 is, at least partly, dependent on Elg1.
Model: Elg1 negatively regulates the telomere length by forming an interaction with the CST complex Taken together, our results suggest the following model ( Figure 6). Following firing of the origin proximal to the telomeres at late S-phase, Elg1 moves with the fork, unloading PCNA from the lagging strand as the fork progresses (Kubota et al., 2015;Kubota et al., 2013;Shemesh et al., 2017; Figure 6A). SUMOylation of PCNA acts as a signal for the activation of telomerase ( Figures 1A, B and 6B). Once Elg1 reaches the telomeres, it interacts with Stn1 ( Figure 5). This interaction requires SUMOylation of Stn1 or of some other protein, which is not PCNA ( Figure 3D) nor Cdc13 (Figure 4-figure supplement 1A) and could be Stn1 itself or another telomeric protein (Hang et al., 2011). The Elg1-Stn1 complex then unloads the SUMOylated PCNA at the telomeres, and Cdc13 becomes SUMOylated ( Figure 6C), now interacting with Elg1-Stn1. SUMOylation of Cdc13 allows its interaction with Stn1, and presumably with Ten1 (Hang et al., 2011), creating the complete CST complex, and preventing further telomerase activity ( Figure 6D). The fact that co-IP experiments detect a strong interaction with Stn1 but fail to detect an interaction with Cdc13 ( Figure 2B) suggests that indeed the interaction Elg1-Cdc13 is very transient, and Elg1 leaves chromatin immediately, leaving the C-terminus of Stn1 free to interact with Cdc13. It is possible that occupancy. (E) Interaction between Stn1 and Cdc13 in a wild type and a elg1Δ strain. (F) Quantitation of three independent biological repeats of the experiment shown in (E). **p<0.001.
The online version of this article includes the following source data and figure supplement(s) for figure 5: Source data 1. The Source Data contains the original Western blots used to make the figure.   a single 'wave' of SUMOylation at the telomeres coordinates PCNA unloading with the interaction between Elg1 and Cdc13. Alternatively, unloading of PCNA by Elg1 may allow the recruitment of SUMO ligases to modify Cdc13. In support for this idea, YTH interactions between Stn1 or Cdc13 with SUMO are abolished in siz1Δ siz2Δ strains ( Figure 3D).

Discussion
We have presented evidence for the fact that two negative regulators of telomerase activity, Elg1 and Stn1, work in the same pathway to coordinate chromosomal DNA replication and telomere elongation. We present a model ( Figure 6) in which the unloading of SUMOylated PCNA by Elg1 at telomeres facilitates the interaction between Stn1 and Cdc13 that negatively regulate telomerase activity. This model is supported by several observations. First, the double mutants of Elg1 and Stn1 do not show increased telomere length phenotype in comparison to the single mutants, indicating that both proteins work in the same pathway that negatively regulates telomere length (Figure 2A). In addition, the interaction between Elg1 and Stn1 seems direct and occurs exclusively during the late S-phase ( Figures 2B and 5A-C). This interaction is neither dependent on SUMOylation of PCNA ( Figure 3D) nor on Cdc13 (Figure 4-figure supplement 1A). In contrast, the interaction between Cdc13 and Elg1 is dependent on both the SUMOylation of PCNA ( Figure 3D) and of Cdc13 itself ( Figure 4F). We suggest that Elg1 participates in the negative regulation of telomere length by unloading SUMOylated PCNA from the telomeres. This eliminates a positive signal for telomerase activity and causes the exchange of Est1 by Stn1 as interactor of Cdc13, effectively terminating telomerase activity (Chandra et al., 2001;Gao et al., 2007;Ge et al., 2020;Hang et al., 2011;Puglisi et al., 2008). As expected from the model, the interaction Elg1-Cdc13 is transient and cannot be detected by co-IP.
We present evidence for the fact that SUMOylated PCNA serves as a positive signal for telomere elongation (Figure 1). Preventing its unloading (by mutations in the ELG1 gene) leads to elongated telomeres, and so does PCNA overexpression, provided the protein is able to be SUMOylated, or is fused to SUMO (thus mimicking SUMOylated PCNA) (Figure 1). ELG1 is the only PCNA unloader in eukaryotes; however, it is not an essential gene, implying that there are alternative ways in which PCNA can get unloaded (either spontaneously or with the help of RFC) (Arbel et al., 2021). Once this step does take place, Stn1 can replace Est1 as Cdc13's partner and stop the activity of telomerase, but at a longer telomere steady-state equilibrium point. This explains both the long telomere phenotype of elg1 mutants, and the fact that it is less severe than that of stn1 alleles (Figure 2). The fact that pol30-RR mutants have stable, normal-sized telomeres (Figure 1) implies that despite lacking PCNA SUMOylation, they are able to solve the 'end replication problem.' Our results support the notion (Maicher et al., 2017) that there are two alternative modes of telomerase: one carrying 'basal activity,' with minimal genetic requirements, able to maintain most telomeres at normal size, and another, 'sustained activity,' required for increased elongating activity (Maicher et al., 2017). Previous results have shown that the first mode does not require Rnr1, the large subunit of the ribonucleotide reductase enzyme, whereas the second depends on it (Maicher et al., 2017). Similarly, SUMOylated PCNA is only needed for the second mode of telomerase. In our experiments, the phenotype that we monitored was elongated telomeres (e.g., by deletion of ELG1 or by overexpression of PCNA) which required sustained telomerase activity. To solve the 'end replication problem' and maintain normal telomere length, very little telomerase activity might be needed, and even unmodified PCNA or telomeric proteins might be able to provide that basal level.
Many proteins in the cell become SUMOylated at one point or another. This modification is usually transient due to the activity of de-SUMOylases (which eliminate the modification) and STUbLs (SUMOdependent ubiquitin ligases, which usually send the protein to degradation). It has been proposed that SUMOylation often works by targeting many physically interacting proteins together. This may be executed by the local recruitment of SUMO ligases to preassembled, co-localized protein groups. SUMOylation thus may increase their interactions, acting to increase their local concentration or effective interaction (Jentsch and Psakhye, 2013). SUMOylation of telomeric proteins seems to be one of those cases (Garg et al., 2014;Hang et al., 2011;Matmati et al., 2018;Miyagawa et al., 2014). In our model, PCNA, Cdc13, and possibly Stn1 undergo SUMOylation to ensure the proper sequence of events in the coordination between telomerase activity and chromosomal replication. Cdc13 binds Pol12, a subunit of Polymerase Alpha (Grossi et al., 2004;Petreaca et al., 2006;Puglisi et al., 2008), allowing the synthesis of the lagging strand on the newly synthesized ssDNA. Thus, the arrival of the replisome to a short telomere at the end of S-phase brings PCNA, whose SUMOylation can elicit both telomerase activity, and its subsequent repression when PCNA is unloaded by Elg1. This in turn primes synthesis of the complementary strand, thus coordinating telomere elongation with chromosomal replication. Yeast strains, plasmids, and media

Materials and methods
All yeast strains used are described in Table 2. All plasmids are described in Table 3. Yeast extractpeptone-dextrose (YPD) medium was prepared with a ready-to-use mixture (Formedium). Synthetic complete (SC) minimal medium was prepared with 2% dextrose (Formedium), yeast nitrogen base without amino acids (Difco), and all necessary amino acids. A 2% agar (Difco) was added for solid medium. Standard yeast genetic procedures were used to create single and double mutants. Unless stated otherwise, all experiments were carried out at 30 o C.

PCNA overexpression
The high copy number pGAD424 plasmid alone, or carrying the wild-type POL30, the mutant pol30-KK127,164RR (pol30-RR), or the pol30-RR fused to ubiquitin or to SUMO (Parker et al., 2007;Takahashi et al., 2020) were transformed into pol30-RR strains and grown under selective conditions.

Southern teloblots
For each genotype, at least three independently created strains were tested after 10 consecutive passages on YPD plates (Harari and Kupiec, 2018;Harari et al., 2013;Harari et al., 2017;Rubinstein et al., 2014). Genomic DNA was isolated by pelleting 15 ml saturated overnight yeast culture grown in YPD, then beating the cells using 0.5 mm glass beads in 500 μl lysis buffer and 500 μl phenol chloroform. Cells were spun down for 10 min at 14,000 rpm, and ∼500 μl of the top clear solution, containing the genomic DNA, was carefully taken out. DNA was precipitated using 1000 μl of 95% ethanol and then pelleted. The supernatant was discarded, and 500 μl of 70% ethanol was added.

Yeast two-hybrid
The yeast two-hybrid assays were performed using PJ69-4A (James et al., 1996) strains cotransformed with a LEU2-marked plasmid containing genes fused to the GAL4 activating domain (pACT or pGAD424) and a URA3-marked plasmid containing genes fused to the GAL4 DNA binding domain (pGBU9). Strains containing the test plasmids were grown for 24 hr at 30°C in SC-Ura-Leu liquid medium and were plated on the reporter maker SC-His medium.

Protein extraction and immunoprecipitation (IP) assays
Cells were grown to mid-logarithmic phase, washed once with water, and resuspended in lysis buffer (20 mm Tris-HCl, pH 7.5, 0.5 mm EGTA, 0.5 mm EDTA, 1 mm DTT, 125 mm potassium acetate, 12.5% glycerol, 0.1% Triton X-100, protease inhibitor mixture, and 1 mm phenylmethylsulfonyl fluoride). Cells were broken for 45 min with glass beads, centrifuged for 10 min at 10,000 × g, and the supernatant was collected. 20-30 μg of total protein extract was resolved on SDS-PAGE using 10% acrylamide gels. For immunoprecipitation, 1000 μg of proteins were prepared and pre-cleared with 20 μl of protein A-Sepharose and protein G-Sepharose beads mixture (GE Healthcare). 2 μl of (HA, Santa Cruz Biotechnology [sc7392; 1:1000] or MYC, Santa Cruz Biotechnology [9E10, SC-40; 1:1000]) antibodies were added to the cleared extract and incubated overnight at 4°C. The beads were washed once with lysis buffer, once with lysis buffer containing 0.5 m NaCl, and twice with buffer A (50 mm Tris-HCl, pH 7.5, 0.1 mm EGTA, 0.1% β-mercaptoethanol). The resulting immunoprecipitates were used for in vitro kinase assays.

Western blotting
Cells were collected by centrifugation, resuspended in 600 ul of phosphate-buffered saline with 1% Triton X-100 (PBST), supplemented with a protease inhibitor cocktail (Roche), and subjected to mechanical rupture using glass beads. The cell debris were removed by centrifugation, and the supernatants were applied onto 0. Primary antibody was added for 12 hr at 4°C. The blot was washed 3 × 5 min with TBST (Tris-buffered saline Tween-20) and secondary antibody was added for 1 hr. The blot was washed 3 × 5 min with TBST and subjected to electro-chemiluminiscence.

Cell cycle synchronization
To assay Elg1 and Stn1 interaction during the cell cycle, the triple-tagged strain was grown in YPD to OD 600 0.6-0.8, followed by the addition of 500 ng/ml α factor and was grown until ∼90% of the cells appeared unbudded or exhibit Shmoo formation (for ∼2 hr). The α-factor was removed by centrifugation and washing cells 2-3 times with warm YPD. Cells were released into YPD at OD 600 0.6-0.8 with addition of pronase at a final concentration of 0.1 mg/ml. Cells were collected for both FACS and western blot analysis at different time points after release. Cells for FACS analysis were fixed in 70% ethanol, digested with RNase overnight, washed again and stained with propidium iodide (15 μg/ml), and analyzed by flow cytometry.

Flow cytometry
200 µl of a logarithmic cell culture (0.6 OD 600 ) were harvested, resuspended in 60 µl of 50 mM Tris pH 7.5, and 140 µl of ethanol was added; cells were then kept overnight at 4°C. Fixed cells were centrifuged and washed once in 200 µl of 50 mM Tris pH 7.5 buffer and resuspended in 100 µl RNAse (0.2 mg/ml in 50 mM Tris pH 7.5) for 2 hr at 37°C. In addition, proteinase-K (0.2 mg/ml in 50 mM Tris pH 7.5) was added to each tube and cells were incubated for 60 additional minutes at 50°C. 20 µl of the sample was taken into a new tube and a 180 µl of 18 µg/ml propidium iodide 50 mM Tris pH 7.5 was added. The samples were kept in the dark at 4°C overnight, sonicated twice at low setting (20% power) for 3-5 s, and stored in the dark at 4°C. The flow cytometry MACSQuant system was used for reading. Results were analyzed using either the Flowing Software or the FlowJo program.
Chromatin immunoprecipitation 50 ml of each strain were grown to OD 600 ≈ 1 in YPD. 1.5 ml formaldehyde (37% solution) was added for 15 min, and the formaldehyde was quenched with 2.5 ml of 2.5 M glycine for 5 min. Cells were harvested, washed once with 15 ml cold PBS, and broken down for 10 min with glass beads in 600 μl lysis buffer (50 mM HEPES-KOH pH 7.5, 140 mM NaCl, 1 mM EDTA, 1% Triton X100, 0.1% Na-deoxycholic acid). The supernatant (lysate) was removed to a new tube. The glass beads were washed with 500 μl lysis buffer, centrifuged, and the supernatant was added to the lysate. The lysate was sonicated 6-8 times for 10-15 s at 80% amplitude with 1 min on ice between each time. The sonicated material was centrifuged for 20 min at 2500 rpm. The supernatant was used for immunoprecipitations (IP). The sonicated proteins were pre-cleared with a 25 μl protein A sepharose and protein G sepharose beads mixture (GE Healthcare), and the appropriate antibodies were added to the cleared extract and incubated overnight at 4°C. PCNA were immunoprecipitated with 2-5 μg of anti-PCNA antibody (Sigma). A total of 10% of the extract was saved as input. The beads after the IP were washed once with lysis buffer, once with lysis buffer with 360 mM NaCl, once with washing buffer (10 mM Tris/HCl pH 8, 0.25 M LiCl, 0.5% NP40, 0.5% Na-deoxycholic acid, 1 mM EDTA), and once with TE (10 mM Tris/ HCl pH 8 and 10 mM EDTA). The washed beads and the input were treated with elution buffer (50 mM Tris/HCl pH 8, 10 mM EDTA, 1% SDS) overnight at 65°C. The DNA was precipitated, resuspended in water, and used for PCR real-time analysis (ABI StepOnePlus Real-Time PCR System); primer concentration and cycles number were calibrated individually for each reaction. All experiments are plotted as the average of at least three independent biological repeats, and each biological repeat is the average of three technical PCR repeats. The oligonucleotides used are Y′-element 5′-GGCT TGAT TTGG CAAA CGTT -3′ 5′-GTGA ACCG CTAC CATC AGCA T-3′

Materials availability
All materials in this article can be available upon contact with the corresponding author.