SSU72 phosphatase is a telomere replication terminator

Telomeres, the protective ends of eukaryotic chromosomes, are replicated through concerted actions by conventional DNA polymerases and telomerase, though the regulation of this process is not fully understood. Telomere replication requires (C)-Stn1-Ten1, a telomere ssDNA-binding complex that is homologous to RPA. Here, we show that the evolutionarily conserved phosphatase Ssu72 is responsible for terminating the cycle of telomere replication in fission yeast. Ssu72 controls the recruitment of Stn1 to telomeres by regulating Stn1 phosphorylation at S74, a residue that lies within the conserved OB fold domain. Consequently, ssu72Δ mutants are defective in telomere replication and exhibit long 3’ overhangs, which are indicative of defective lagging strand DNA synthesis. We also show that hSSU72 regulates telomerase activation in human cells by controlling the recruitment of hSTN1 to telomeres. Thus, in this study, we demonstrate a previously unknown yet conserved role for the phosphatase SSU72, whereby this enzyme controls telomere homeostasis by activating lagging strand DNA synthesis, thus terminating the cycle of telomere replication.


31
Telomeres are protein-DNA complexes that form the ends of eukaryotic 32 chromosomes (reviewed in 1 ). Telomeres predominantly function to prevent the loss 33 of genetic information and to inhibit DNA repair at the chromosome termini, thus 34 maintaining telomere protection and genome stability. Loss of telomere regulation 35 has been linked to two main hallmarks of cancer: replicative immortality and genome 36 instability 2 . Telomeres face an additional challenge: DNA replication. Due to G-rich 37 repetitive DNA sequence and protective structures, telomeres represent a natural 38 obstacle for passing replication forks 3 . Replication fork collapse can lead to the loss 39 of whole telomere tracts. To counteract these effects, telomerase (Trt1 in S. pombe 40 and TERT in mammals) is responsible for adding specific repetitive sequences to 41 telomeres, compensating for the cell's inability to replicate chromosome ends 4 . 42 However, it is currently not understood how telomerase activity is regulated and how 43 the telomerase cycle is coupled to telomeric DNA replication. Intriguingly, several 44 DNA replication proteins are required for proper telomere elongation 5 . Conversely, 45 specific telomere components are themselves required for proper telomere 46 replication and telomere length regulation 6,7 , suggesting that there is a very thin line 47 separating telomere replication and telomerase activity. Primase recruitment, and therefore lagging-strand synthesis, is delayed at 54 chromosome ends, leading to the accumulation of ssDNA at telomeres. This event 55 results in the activation of the major checkpoint protein kinase Rad3 ATR and the 56 subsequent phosphorylation of telomeric Ccq1-T93, a step required for telomerase 57 activation. Thus, as a consequence of delayed Polα-Primase recruitment to short 58 telomeres and the subsequent accumulation of ssDNA, Rad3 ATR is transiently 59 activated leading to telomerase recruitment and telomere elongation. 60 Another complex known as CST (Cdc13/Stn1/Ten1 in S. cerevisiae and 61 CTC1/STN1/TEN1 in mammals), is known to control telomere replication. This 62 complex is responsible for both protection from 5' strand nucleolytic degradation and 63 recruitment of the Polα-primase complex to telomeres, thus promoting telomere 64 lagging-strand DNA synthesis (Grossi et al., 2004;Lin and Zakian, 1996). Notably, 65 CST is not only required to recruit Polα-primase but is also responsible for the switch 66 from primase to polymerase activity, which is required for gap-less DNA replication 67 11 . In humans, in addition to its role in telomere replication 12 , the CST complex also 68 functions as a telomerase activity terminator 13 by inhibiting telomerase activity 69 through primer confiscation and direct interaction with the POT1-TPP1 dimer. 70 However, the mechanism regulating these CST functions remains unknown. In 71 fission yeast, although stn1 + and ten1 + homologs exist, no Cdc13/CTC1 homolog has 72 been found to date 14 . Recent studies have revealed that Stn1 is required for 73 telomere and subtelomere replication 15 and 16 , supporting the conserved role of 74 fission yeast (C)ST in DNA replication. 75 In agreement with the replication model proposed by 17 and reviewed in 18 , the 76 telomere-binding protein, Rif1 was shown to regulate telomere DNA replication 77 timing by recruiting Glc7 phosphatase to origins of replication and inhibiting Cdc7 78 activities in budding yeast ( Hiraga et al., 2014;Mattarocci et al., 2014). Notably, this 79 role is conserved in other organisms such as fission yeast 22 and human cells 23 . 80 Importantly, rif1 mutants display long telomeres; this effect is suggested to be a 81 result of origin firing dysregulation 18 . However, how telomere replication is 82 terminated and how this is coupled with the regulation of telomere length remains 83 unknown. Here, we report that the phosphatase family member Ssu72 displays a 84 conserved role as a telomere replication terminator. Ssu72 was previously identified 85 as an RNA polymerase II C-terminal domain phosphatase and is highly conserved

97
Ssu72 is a negative regulator of telomere elongation 98 We carried out a genome-wide screen for regulators of telomere homeostasis 99 in S. pombe using a commercially available whole-genome deletion library (Bioneer 100 corporation). This library allowed us to identify new non-essential genes involved in 101 telomere homeostasis in fission yeast ( Figure 1A). Of the genes identified from the 102 screen, we selected the highly conserved phosphatase ssu72+ (SPAC3G9.04) as 103 the most promising candidate for further characterization. We generated a deletion 104 mutant (ssu72∆) as well as a point mutant devoid of phosphatase activity (ssu72-105 C13S) and found that these two mutants possess longer telomeres ( Figure 1B). 106 Additionally, we found that Ssu72 localized to telomeres in a cell cycle-dependent 107 manner. We performed cell cycle synchronization using a cdc25-22 block-release 108 method in a ssu72-myc tagged strain and measured Ssu72 binding to telomeres by 109 chromatin immunoprecipitation (ChIP). Cell cycle phases and synchronization 110 efficiency were measured using the cell septation index. Ssu72-myc is recruited to 111 telomeres in late S phase and declines later in the cell cycle ( Figure 1C). 112 Interestingly, Ssu72 is recruited to telomeres at approximately the same time as the 113 arrival of the lagging strand machinery at chromosome ends 17 . 114 ssu72∆ cells displayed increased (~1 Kb) telomere lengths compared to wild-115 type telomeres (~300 bp) ( Figure 1B). We set out to understand the nature of 116 telomere elongation in the ssu72 mutant background. To test if the telomere 117 elongation was dependent on telomerase, trt1∆ (deletion mutant for the catalytic 118 subunit of telomerase) and ssu72∆ double heterozygous diploids were sporulated.

119
Of the resulting tetrads, trt1∆ and trt1∆ ssu72∆ double mutants were selected and 120 streaked for several generations in order to facilitate telomere shortening in the 121 absence of telomerase. While ssu72 mutant cells displayed long telomeres, ssu72∆ 122 trt1∆ double mutant and trt1∆ single mutant cells displayed similarly shortened 123 telomeres ( Figure 1D). ChIP experiments consistently demonstrated an 124 accumulation of Trt1-myc at ssu72∆ telomeres compared to wt cells ( Figure 1E). 125 Thus, the longer telomeres exhibited by ssu72∆ mutants were a consequence of 126 telomerase deregulation.

127
Two independent studies 27,28 showed that Ccq1 phosphorylation at Thr93 is 128 required for telomerase-mediated telomere elongation in fission yeast. Using

129
Western blot shift analysis, we observed that Ccq1 was phosphorylated in ssu72∆ 130 cells when we compared to those of WT strains ( Figure 1F). To further confirm that 131 telomere elongation was telomerase-dependent, we repeated the previous 132 experiment using a phosphorylation-resistant mutant version of Ccq1 27 . We 133 germinated a double heterozygous ccq1-T93A/+ ssu72∆/+ mutant and analyzed its 134 progeny. As expected, ccq1-T93A ssu72∆ double mutants displayed a similar 135 telomere-shortening rate to that of the ccq1-T93A single mutants ( Figure S1). In 136 agreement with these results, we further showed that telomere length in ssu72∆ 137 mutants was dependent on Rad3, the kinase responsible for Ccq1-T93  In fission yeast, the presence of telomeric ssDNA results in Rad3 activation 146 and telomere elongation 27 . Thus, we investigated whether ssu72∆ mutants 147 accumulated telomeric ssDNA. We carried out in-gel hybridization assays using a C-148 rich probe to measure the accumulation of G-rich DNA at telomeres. Notably, the 149 ssu72∆ mutant strain showed an almost 6-fold increase in G-rich telomere 150 sequences (Figure 2A). We observed that the accumulation of ssDNA at telomeres 151 is increased in ssu72∆ mutants compared to rif1∆ mutants, though both strains have 152 similar telomere lengths. Further, we consistently detected Rad11 RPA -GFP 153 localization at telomeres, as measured by live imaging in ssu72∆ mutant cells 154 ( Figure 2B). 155 Recently, the telomere-binding protein Rif1 was found to control DNA 156 resection and origin firing by recruiting PP1A phosphatase to double strand breaks 157 and origins of replication [20][21][22]29 . We wondered if Rif1 was also responsible for 158 recruiting the Ssu72 phosphatase to telomeres. To test this hypothesis, we 159 combined ssu72∆ and ssu72-C13S (catalytically dead) mutants with rif1∆ and 160 carried out of telomere length epistasis analyses. While single mutants displayed 161 telomere lengths of 1 Kb, ssu72∆ rif1∆ and ssu72-C13S rif1∆ double mutants 162 displayed telomeres that were longer than 3 Kb ( Figure 2C). Thus, our data show 163 that Rif1-mediated regulation of telomere length is independent of Ssu72 in fission 164 yeast.

165
Ssu72 controls telomere length through the Stn1-Ten1 complex 166 A second highly conserved protein complex regulates telomere length and 167 telomerase activity. The budding yeast CST complex (Cdc13 CTC1 , Stn1 and Ten1) 168 plays opposing roles at the telomeres. Cdc13 is required for telomerase recruitment 169 and is activated through its interaction with Est1, a subunit of telomerase 30 . This 170 interaction is promoted by the phosphorylation of Cdc13 at T308 by Cdk1(Cdc28). In 171 contrast, the Siz1/2-mediated SUMOylation of Cdc13 at Lys908 promotes its 172 interaction with Stn1 31 . This interaction is required, with Ten1, for polymerase alpha 173 complex recruitment and telomere lagging-strand DNA synthesis 8 . However, the 174 regulatory mechanism underlying these two opposite functions remains unknown.

175
Despite the lack of Cdc13 CTC1 homologs in fission yeast, the Stn1-Ten1 complex 176 appears to play similar roles to those found in budding yeast and mammals 177 (reviewed in 32 ). Consequently, we hypothesized that Ssu72 controls telomere length  Stn1-myc was recruited to telomeres in S/G2 cells 36 . We observed that the 195 recruitment of Stn1 to telomeres was severely impaired in the absence of Ssu72 196 ( Figure 3A). Thus, our results indicate that Ssu72 functionality is required for Stn1 197 recruitment to telomeres.

198
Based on these findings, we asked whether DNA replication dynamics were 199 affected at ssu72∆ mutant chromosome ends. Genomic DNA derived from WT and 200 ssu72∆ cells was isolated, subjected to NsiI digestion, and analyzed on 2D-gels.

201
Chromosome ends were revealed by Southern blotting using a telomere-proximal 202 STE1 probe 6,37 . In the first dimension, we observed three distinct bands for the wt 203 parental strain but only one thick, smeared band for the ssu72∆ strain. As expected,  Given that Ssu72 phosphatase activity is required to regulate telomere length 215 and that Ssu72 is recruited to telomeres during the S/G2 phases, we hypothesized 216 that Ssu72 might regulate Stn1 phosphorylation in a cell cycle-dependent manner.  Figure S3B) and budding yeast but also throughout higher eukaryotes, including 227 humans and mice ( Figure 3D). Therefore, we decided to mutate the Serine-74 228 residue to aspartic acid (Stn1-S74D), a phosphomimetic amino acid replacement.

229
Cells harboring the stn1-S74D mutation exhibited long telomeres (~1 Kb) 230 similar to those found in ssu72∆ cells ( Figure 3C). We hypothesized that telomere 231 elongation in the stn1-S74D strain was telomerase-dependent. Consistent with this 232 hypothesis, stn1-S74D trt1∆ double mutant telomeres become shorter after 233 sequential streaks ( Figure S4A). Importantly, stn1-S74D ssu72∆ double mutants 234 displayed similar telomere lengths to those in stn1-S74D single mutants. In addition, 235 we performed ChIP experiments in strains expressing Stn1-S74D-myc in order to 236 analyze the recruitment of Stn1 to telomeres. Similar our observations in mutants 237 lacking ssu72 phosphatase, Stn1-S74D-myc was not efficiently recruited to 238 telomeres ( Figure 3E). Taken together, our data suggest that fission yeast Stn1 is 239 phosphorylated at Serine-74 to enable its efficient recruitment to telomeres and, 240 consequently, efficient DNA replication and telomerase regulation. it has been proposed to be a terminator of telomerase activity due to its higher 257 affinity for telomeric single stranded DNA formed after telomerase activation 13 .

258
Further, it has been suggested to promote the restart of stalled replication forks 40,41 .

259
In fission yeast, the ST complex also exhibits this dual function. First, the binding of Ssu72 controls lagging strand synthesis at telomeres ( Figure S5A). 278 We next investigated if Stn1 overexpression was sufficient to rescue the 279 telomere defects observed in ssu72∆ mutants. To test this hypothesis, we replaced 280 the stn1 + endogenous promoter with inducible Thiamine-regulated nmt promoters 43 . 281 We observed that none of the promoters used to overexpress Stn1 rescued the  The results of the previous experiment suggested the hypothesis that Ssu72 300 is required to activate DNA polymerase α at telomeres. To test this, we 301 overexpressed the catalytic subunit of polymerase α (pol1 + ) in cells lacking Ssu72.

302
Previous studies have shown that overexpression of pol1 + was sufficient to rescue 303 strains with lagging strand synthesis defects 5 . Remarkably, using pol1 + expression 304 from multicopy plasmids, we showed that overexpression of pol1 + in ssu72∆ mutants 305 is sufficient to rescue telomere defects ( Figure 4D). Thus, we propose that Ssu72  315 Because Ssu72 is a highly conserved phosphatase and CST has similar 316 functions in different species, we tested if telomere regulation by SSU72 was 317 conserved in human cells. We were not able to produce human cell lines lacking 318 SSU72 using conventional CRISPR/Cas9 technology, suggesting that SSU72 is 319 essential in humans. In contrast to fission yeast, SSU72 is an essential gene both in 320 budding yeast 24,44 and mice 45 . Therefore, we decided to use short hairpin RNAs to 321 downregulate SSU72 protein levels. This approach has been previously used in 322 human cells to study the role of SSU72 in mammals 25 .

323
Our results show that, similar to fission yeast, knockdown of SSU72 in human 324 cells causes telomere dysfunction. We downregulated SSU72 levels using two

361
As expected, SSU72 downregulation in HT1080 cells also resulted in 362 telomere induced foci (TIF), as measured by the localization of 53BP1 to telomeres 363 ( Figure 6A). Importantly, TIF formation was not cell line dependent, as we also 364 observed TIFs in HeLa cells treated with an siRNA against SSU72 (Figure S6B). 365 Our data suggest that the downregulation of SSU72 in human cells mimics 366 previous results obtained in STN1 downregulated cells. Thus, we tested whether 367 cells lacking SSU72 were defective for STN1 recruitment to telomeres. We 368 expressed FLAG-tagged STN1 in HT1080 cells and infected these cells with 369 lentiviral particles expressing an shRNA against either SSU72 or Luciferase. We 370 then carried out telomeric ChIP experiments using FLAG antibodies ( Figure 6B). 371 Upon downregulation of SSU72, we observed a 40% reduction in STN1 binding to 372 telomeres compared to shLuciferase-treated cells. Together, these data indicate an 373 evolutionarily conserved role for SSU72 phosphatase in controlling STN1 recruitment 374 to telomeres and, therefore, in regulating lagging strand syntheses at telomeres.  The SSU72 phosphatase appears to be conserved throughout evolution. The                           Figure S5. A) ssu72∆ telomere length is epistatic with polymerase alpha complex subunits. Genomic DNA of single mutants of pol1-13, spp1-9, spp2-9, ssu72∆ and wt or double mutants spp1-9 ssu72∆ and spp2-9 ssu72∆ was isolated and telomere length was measured carrying out a Southern blot in ApaI digested genomic DNA using a telomeric probe. Temperature sensitive strains were grown at semi-permissive temperature by several generations and DNA was collected to carry out Southern Blot analysis. B) Stn1 overexpression doesn't rescue telomere defect in ssu72∆. We expressed stn1 under 3x (stronger), 41x and 81x (weaker) nmt1 promoter in wt or ssu72∆ background and telomere length was measured in ApaI digested genomic DNA.  Metaphases were collected and FISH was carried out using a PNA-telomeric probe. Quantification of MTS in SSU72 downregulated cell was carried out n = 2; **p ≤0.01 based on a two-tailed Student's t-test to control sample. Error bars represent mean ±standard error of the mean. B) SSU72 downregulation induces 53BP1 foci at telomeres in Hela human cell line. Cells were transfected with two independent siRNAs against human SSU72 using a non-targeting siRNA as a control. After 3 days cells were fixed and IF-FISH was carried out using a 53BP1 antibody and PNA-telomeric probe. Quantification of Telomeric induce foci (TIF) in SSU72 downregulated cell was carried out n = 3; *p ≤0.05 based on a two-tailed Student's t-test to control sample. Error bars represent mean ±standard error of the mean.