UmPot1s in Basidiomycetes harbor an extra N-terminal OB fold that exhibits no nucleic acid-binding activity

Like the well characterized fission yeast and primate orthologs, UmPot1 possesses three conserved OB fold domains (OB1, OB2 and OB3) and exhibits high affinity binding to the canonical G-strand repeats [26]. However, UmPot1 also contains a long N-terminal extension (~ 350 amino acids before OB1) of unknown function (Fig 1A). To gauge the extent of conservation for this region, we used BLASTp or PSI-BLAST to identify additional fungal Pot1 homologs both within and beyond the Basidiomycota phylum. Notably, the majority of basidiomycete Pot1s contain an equivalent N-terminal extension (~200–350 amino acids (a.a.)) before OB1 (Fig 1A). Sequence alignment coupled with secondary structure prediction revealed a moderately conserved 120 a.a. domain exhibiting structural features of an OB-fold, i.e., five β strands with an intervening α helix (Fig 1B) [29]. Thus, most basidiomycete Pot1s appear to harbor four OB-fold domains (designated OB-N, OB1, OB2 and OB3). Interestingly, Myler et al. recently showed that this is true for the majority of metazoan POT1s as well [28]. In support of common ancestry, we found that the two groups of OB-Ns from basidiomycetes and metazoan Pot1s align well to each other (S1A Fig). In addition, homology modeling of a basidiomycete Pot1 OB-N (from PaPot1) yielded a structure that is superimposable on the OB-A domain from human RPA70 (Fig 1C). Similarities between metazoan and basidiomycete Pot1s are further underscored by the conservation of zinc-binding motifs in the OB3 of these homologs (S1B and S1C Fig). Interestingly, in contrast to basidiomycetes, the Pot1 homologs in other fungal phyla (e.g., Ascomycota, Chytridiomycota, Zoopagomycota, etc.) contain just three OB folds (Fig 1A). Because phylogenomic analysis indicates that Chytridiomycota and Zoopagomycota represent deeper fungal lineages than Basidiomycota, it is possible that OB-N might have been independently lost in multiple fungal branches [30], just as it was lost in selected metazoans [28]. Alternatively, basidiomycete Pot1 might have acquired OB-N through the duplication of another OB fold in the protein, such OB1. However, phylogenetic comparison suggests that the OB-N of basidiomycetes is more closely related to the OB-N of metazoans than to the OB1 of basidiomycetes (S2A Fig). Likewise, the same analysis failed to uncover a strong kinship between the OB-N and OB2 of basidiomycete Pot1 (S2B Fig). In addition, given the identical domain organization for the 4-OB Pot1 and RPA1, we assessed the relationship between the OB-N of Pot1 and RPA70 (S2C Fig). Again, the OB-N of metazoan and basidiomycete Pot1 are more similar to each other than to the OB-N of RPA70. Therefore, we favor the notion of a common origin between metazoan and basidiomycete Pot1 OB-N. However, we cannot rule out the possibility that the OB-N of basidiomycete Pot1 was acquired independently and came to resemble its metazoan counterpart through convergent evolution.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. U maydis Pot1 and other basidiomycetes harbor an extra N-terminal OB fold domain.

A. Domain organization of Pot1 homologs from major fungal branches. B. Multiple sequence alignment of the N-terminal region of Basidiomycete Pot1 homologs. The alignment and secondary structure predictions were generated using PROMALS3D and displayed using MView. C. A homology model of the putative OB-N domain from Paxillus ammoniavirescens (PaPot1) (generated from SWISS-MODEL) is superimposed on the OB-A domain of human RPA70. Note that the five beta strands (labeled as S1 to S5) in the two structures are spatially well aligned. D. EMSA analysis of the DNA-binding activities of UmPot1 truncation variants. The domains included in each variant are illustrated at the top and the binding assays shown at the bottom. 5’-labeled G4 oligonucleotide was used as the probe.

https://doi.org/10.1371/journal.pgen.1010182.g001

In contrast to OB-N, the OB3 domain is evidently well conserved in all fungal phyla, except for the loss of Zinc-binding residues in ascomycetes Pot1 (S1C Fig). We conclude that relative to other fungal Pot1s, basidiomycete Pot1 may be structurally closer to the majority of metazoan orthologs.

To assess the contribution of OB-N as well as other domains of UmPot1 to DNA binding, we purified truncation variants of Pot1 and analyzed their G-strand binding activity (Figs 1D and S3A and S3B). Consistent with results for fission yeast and mammalian Pot1 orthologs, we found that high affinity G-strand-binding was mediated exclusively by the OB1 and OB2 of U. maydis Pot1. In contrast, neither the OB-N nor OB3 exhibited any appreciable binding activity (Fig 1D). These results suggest that OB-N may mediate its function through protein-protein interactions.

UmPot1 physically interacts with Rad51 and stimulates the strand exchange activity of Rad51

Since our overall objective in studying U. maydis telomeres was to dissect the interplay between telomeres and repair proteins, we screened for potential interactions between Pot1 and DNA repair proteins. Using FLAG-tagged Pot1 in affinity pull down assays, we detected substantial binding of Pot1 to purified Rad51, but not Brh2 (Fig 2A). This binding is probably direct (i.e., not through contaminating nucleic acids) because the addition of benzonase had no effect (Fig 2B). We then examined the Rad51-binding activities of different domains of Pot1 and detected significant activity for OB-N and OB12 of Pot1, but minimal binding for OB3 (S3C Fig). A Pot1 fragment that spans the N-terminus and OB12 showed stronger binding to Rad51 than either domain alone, further supporting the contributions of both regions to Rad51 interaction.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. U maydis Pot1 binds to Rad51 and regulate the strand-exchange activity of Rad51.

A. Anti-FLAG agarose resin was pre-bound with Pot1-FLAG and tested for interaction with purified Rad51 or MBP-Brh2/Dss1 complex. The bound proteins were eluted with FLAG₃ peptides and the levels of Rad51, Brh2 and Pot1 in the eluates analyzed using anti-Rad51, anti-MBP, and anti-FLAG antibodies. B. The interaction between Pot1-FLAG and Rad51 was analyzed in the absence or presence of Benzonase. C. The ability of Rad51 to mediate strand exchange between a 70-mer ssDNA and a labeled 39-mer dsDNA was analyzed in the absence or presence of ATP. The substrates and products are marked by an open and a closed arrowhead, respectively. Asterisks designate the positions of the 5’ ³²P label. D. The effects of Pot1 and Tay1 on strand exchange were analyzed in the absence or presence of Rad51. The substrates and products are marked by an open and a closed arrowhead, respectively. Asterisks designate the positions of the 5’ ³²P label. An additional band is detected in Pot1-containing reactions (marked by a filled circle). While the identity of this is unclear, it is not Rad51-dependent and unlikely to be relevant to the strand-exchange reaction.

https://doi.org/10.1371/journal.pgen.1010182.g002

To determine if Pot1 can modulate the recombinase activity of Rad51, we employed an oligonucleotide-based strand exchange assay [31,32]. As expected, Rad51 alone catalyzed strand exchange between the duplex and ssDNA substrates in an ATP-dependent manner (Fig 2C). Interestingly, while Pot1 alone exhibited no strand exchange activity, it stimulated strand exchange by ~ 2-fold in the presence of Rad51 (Fig 2D). By contrast, the duplex telomere protein Tay1, which does not bind Rad51 [27], had no effect on strand exchange either in the absence or presence of Rad51 (Fig 2D). We further examined the ability of Pot1 truncation derivatives to stimulate strand exchange, and found that both the N-terminus and OB12 domains were required for stimulation (S3D Fig). Together, these results implicate direct Pot1-Rad51 interaction in the stimulatory effect of Pot1. It is worth noting that the substrates in the strand exchange assays do not harbor telomere repeats (S2 Table) and are not expected to bind Pot1, again suggesting that the effect of Pot1 is mediated through protein-protein interaction rather than DNA-binding.

UmPot1 promotes telomere maintenance

The Pot1-Rad51 interaction suggests that the telomere maintenance function of Rad51 [24,26] might be regulated by Pot1. To test this idea, we sought to determine the consequence of disrupting this interaction. Because mutating the OB12 domain will likely abrogate DNA-binding and trigger telomere deprotection (making it difficult to interpret the phenotypes), we deleted just the OB-N of Pot1 (pot1ΔN) and used this allele to replace genomic pot1 through homologous recombination. The pot1ΔN mutant did not exhibit obvious growth defect or UV sensitivity (Fig 3A), suggesting that telomere protection is intact and Pot1 does not affect the genome-wide repair function of Rad51. In addition, pot1ΔN did not exhibit growth defects or evidence of senescence even following extensive passage on plates (Fig 3B). We then examined telomere maintenance of pot1ΔN using Southern analysis of telomere restriction fragments (TRF), and detected a moderate telomere length reduction in later passages of the mutant in comparison to the wild type strain (Fig 3C, streaks 6 and 10). To obtain a more quantitative assessment of telomere lengths, we used STELA (single telomere length analysis) to characterize the subpopulations of telomeres that harbor the UT4 subtelomere sequence [22]. Consistent with Southern analysis of TRF lengths, the STELA products in pot1ΔN were ~ 200 bp shorter than wild type at streak 10, supporting the notion that Pot1 promotes telomere maintenance (Fig 3D). To further test the notion that Pot1-Rad51 interaction underpins the replication function of Pot1, we compared the telomeres of the rad51Δ pot1ΔN double mutant to those of rad51Δ and pot1ΔN single mutants (S4A Fig). No significant differences were observed, supporting the idea that Rad51 acts in the same pathway as the OB-N of Pot1 to promote telomere maintenance. We also explored the effect of replication stress on the telomere phenotype of pot1ΔN. However, even low concentrations of hydroxyurea induced substantial telomere length heterogeneity in the wild type strain and there was no obvious exacerbation of telomere defects in pot1ΔN (S4B Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Pot1 N-terminal deletion impairs telomere maintenance.

A. Serial dilutions of the indicated U. maydis strains were directly spotted onto a YPD plate or first subjected to UV irradiation prior to spotting. The plates were incubated for 2 days (-UV) or 3 days (+UV) and the results photographed. B. Serial dilutions of cultures of UCM350 and pot1ΔN that had been passaged for the indicated numbers of streaks were spotted onto a YPD plate and incubated for 2 days. C. Chromosomal DNAs from two independently propagated clonal cultures of UCM350 and pot1ΔN were subjected to TRF Southern analysis at the indicated passages (i.e., Streak 3, 6 and 10). Mean TRF lengths from the pairs of assays were calculated and plotted at the bottom. P-values were calculated using two-tailed Student’s t tests and displayed. D. Chromosomal DNAs from two independently propagated clonal cultures of UCM350 and pot1ΔN strains were subjected to STELA analysis at Streak 10. Mean STELA lengths were calculated from 3 independent assays of each DNA sample and shown at the bottom.

https://doi.org/10.1371/journal.pgen.1010182.g003

Transcriptional repression of Umpot1 triggers severe growth defects and multiple telomere aberrations

We next interrogated the telomere protection function of U. maydis pot1. Because deleting this gene is likely to be lethal, we generated a conditional mutant (pot1^crg1) by replacing the endogenous pot1 promoter with that from crg1, which supports elevated transcription in arabinose-containing media (YPA) but is strongly repressed in glucose-containing media (YPD). To assess the OB-N of Pot1 in telomere protection, we also substituted the pot1ΔN promoter with the crg1 promoter (pot1ΔN^crg1). Both mutants exhibited normal growth in YPA, but suffered dramatic growth defects in YPD, consistent with telomere deprotection (Figs 4A and S5A and S5B). Indeed, both mutants became growth arrested in glucose, with pot1ΔN^crg1 yielding especially small colonies consistent with accelerated arrest. Analysis of cell morphology revealed cells of varying sizes and shapes (S5 Fig), suggesting the arrest can occur in different stages of the cell cycle. This makes pot1-deficient cells distinct from several other telomere protein mutants such as ku70 and trf2-deficient cells that show predominantly elongated morphologies consistent with G2/M arrest [23,27].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. pot1 deficiency triggers growth arrest and multiple telomere aberrations.

A. Serial dilutions of the indicated strains were grown on YPA (crg1 promoter on) and YPD (crg1 promoter off) plates for 2 and 3 days, and the results photographed. B. Chromosomal DNAs from three independently generated pot1^crg1 and pot1ΔN^crg1 strains grown in YPA and YPD were subjected to telomere restriction fragment analysis. The overall telomere contents of the samples were quantified and plotted on the right. P-values were calculated using two-tailed Student’s t tests. C. Chromosomal DNAs from three independently generated pot1^crg1 and pot1ΔN^crg1 strains grown in YPA and YPD were subjected to C-circle and G-circle assays. The relative levels of C- and G-circles from the three mutants were quantified and plotted at the bottom. P-values were calculated using two-tailed Student’s t tests. D. Chromosomal DNAs from three independently generated pot1^crg1 and pot1ΔN^crg1 strains grown in YPA and YPD were subjected to in-gel hybridization analysis of ssDNA on both the G- and the C-strand. E. The levels of ssDNA from the in-gel hybridization assays were quantified (N = 3) and plotted. P-values were calculated using two-tailed Student’s t test.

https://doi.org/10.1371/journal.pgen.1010182.g004

Analysis of telomere repeat DNAs revealed a multiplicity of defects in both the pot1^crg1 and pot1ΔN^crg1 mutants grown in YPD, but with some notable differences (Fig 4). First, while the lengths of telomeres (as determined by TRF Southern analysis) became quite heterogenous in both mutants, the overall telomere content was substantially higher in pot1ΔN^crg1 (Fig 4B). The heterogeneous telomere fragments of pot1^crg1 were detected in the absence of PstI digestion, indicating that they exist as extrachromosomal telomere repeat (ECTR) (S5C Fig). Second, while C-circles (a marker of telomere recombination) were found to be elevated in both mutants, the pot1ΔN^crg1 mutant also harbored high levels of G-circles (Fig 4C). Third, while pot1^crg1 exhibited a modest increase in ssDNA on the C-strand (~10 fold), pot1ΔN^crg1 showed instead extremely high levels of ssDNA on the G-strand (Fig 4D and 4E). Interestingly, the G-strand and C-strand ssDNA each showed some sensitivity to both 3’-overhang and 5’-overhang nucleases (i.e., Exonuclease I and RecJ_f) (S5D Fig). This can only be explained by ssDNA on linear ECTRs, which is consistent with the high levels of ECTRs detected in Southern analysis. Together, these observations suggest that loss of pot1 triggers high levels of both telomere recombination and resection, leading to increases in both ds and ss telomere DNA, with much of the DNA in the form of ECTR. In addition, the differences between the telomere aberrations in pot1^crg1 and pot1ΔN^crg1 mutants suggest that even at low levels, the residual Pot1 can still alter aberrant DNA repair reaction at telomeres. For example, if Pot1 OB-N interacts with DNA repair factors such as Rad51 to influence DNA processing, the absence of this domain in pot1ΔN^crg1 could result in differential generation of ssDNA. In support of residual Pot1 expression, RT-qPCR analysis revealed significant pot1 RNA in the pot1^crg1 mutant grown in YPD, at ~1/2 to 1/3 of the normal pot1 RNA level in the parental strain (S5E Fig). In contrast to the drastic increases in abnormal structures associated with telomere recombination and resection, PCR analysis of telomere-telomere fusions revealed only low levels of fusions in the pot1^crg1 and pot1ΔN^crg1 mutants grown either in YPA or YPD (S6 Fig). Hence, non-homologous end joining does not appear to be strongly activated by pot1 deficiency.

Interestingly, even when grown in YPA, pot1ΔN^crg1 exhibited greater telomere length heterogeneity and a modest increase in C-circle levels (Fig 4A and 4B), suggesting mild telomere de-protection. Because this phenotype was not observed in pot1ΔN, it was presumably due to the over-expression of this truncated allele. While the underlying mechanism is unclear, it is again consistent with the notion that the Pot1ΔN may interact differently with DNA repair factors than Pot1. While Pot1ΔN could retain some interaction with Rad51 through its OB1 and OB2 domains, the overall affinity is likely to be weaker than full length Pot1. There could also be as yet unidentified interactions between Pot1 and other DNA repair factors that are altered by removal of OB-N. High levels of Pot1ΔN may therefore cause an imbalance of DNA repair factors at telomeres to trigger telomere deprotection.

The growth and telomere defects of pot1-deficient U. maydis are greatly suppressed by deletions of rad51 and brh2

Recent studies in mammalian cells suggest that pot1 plays an especially prominent role in suppressing homology-directed repair at telomeres [7,9]. However, this function does not appear to be well conserved in fission yeast [13]. To examine the roles of HDR factors in pot1-deficient U. maydis, we analyzed the phenotypes of the pot1^crg1 rad51Δ and pot1^crg1 brh2Δ double mutants. Remarkably, both double mutants showed substantially improved growth and manifested more normal morphology in comparison to pot1^crg when grown in YPD (S7 Fig). Thus, Rad51 and Brh2 were evidently required for the toxicity and growth arrest triggered by pot1 deficiency, i.e., by mediating aberrant repair. Indeed, analysis of telomere repeat DNAs in the double mutants revealed substantial albeit incomplete suppression of the telomere abnormalities observed in pot1^crg1 when the strains are grown in YPD (Fig 5). For example, both the abnormally long and short telomere fragments detected in pot1^crg1 are greatly reduced (but not eliminated) in pot1^crg1 rad51Δ and pot1^crg1 brh2Δ (Fig 5A). Similarly, the extremely elevated C-strand ssDNA levels are suppressed by ~10 to 20-fold in the double mutants (Fig 5B). Finally, the levels of C-circles (a marker of telomere recombination) in pot1^crg1 are also reduced by more than 10-fold when either rad51 or brh2 is deleted (Fig 5C). Hence, the two core HDR factors appear to play a pivotal role in generating aberrant telomere structures in the absence of pot1. These results are like those observed in pot1-deficient mammalian cells [7], supporting the utility of interrogating the U. maydis mutants in dissecting the underlying mechanisms.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Telomere aberrations in pot1-deficient U. maydis are greatly suppressed by rad51Δ and brh2Δ.

A. Chromosomal DNAs from the indicated wild type and mutant strains grown in YPA and YPD were subjected to telomere restriction fragment analysis. Two independently generated pot1^crg1 strains were analyzed in all the assays in this figure. B. Chromosomal DNAs from the indicated wild type and mutant strains grown in YPA and YPD were subjected to in-gel hybridization analysis of ssDNA on both the G- and the C-strand (left). The levels of ssDNA from two independent sets of assays were quantified and plotted (right). P-values were calculated using two-tailed Student’s t tests and displayed. C. Chromosomal DNAs from the indicated wild type and mutant strains grown in YPA and YPD were subjected to C-circle assays (left). The relative levels of C-circle signals from these strains were quantified and plotted (right). Data are from two independent sets of assays. P-values were calculated using two-tailed Student’s t tests and displayed.

https://doi.org/10.1371/journal.pgen.1010182.g005

It is worth noting that like rad51Δ and consistent with our previous report [24], the brh2Δ single mutant also exhibited significant telomere shortening. Therefore, even though there is no detectable physical interaction between Pot1 and Brh2 (Fig 2A), the latter may nevertheless cooperate with Rad51 to facilitate telomere maintenance, as has been reported for mammalian cells [33]. Our results suggest that this pathway could be modulated by Pot1-rad51 interaction. One other notable result, from the ssDNA analysis, was the frequent detection of prominent ssDNA signals (both G and C-strand) in the wells for strains bearing either rad51Δ or brh2Δ mutations regardless of the pot1 status (Fig 5B, lanes marked with asterisks). This could be due to the accumulation of replication or recombination intermediates, which are known to manifest slow mobility, and is consistent with a role for Rad51 and Brh2 in promoting telomere replication [24].

Upregulation of telomere-repeat containing RNAs is a cardinal feature of pot1-deficient cells

While the observations above implicate Rad51 and Brh2 in promoting aberrant telomere repair in pot1-deficient U. maydis, the mechanisms by which the various telomere aberrations are produced and how they trigger cell cycle arrest remain incompletely understood. In this regard, it is worth noting that both Pot1 and Rad51 have been linked to the regulation of telomere repeat-containing RNA (TERRA), which is proposed to be a critical determinant of telomere maintenance and stability [34]. It should be noted that TERRA generally refers to RNAs synthesized by Pol II from subtelomeric promoters and elongated into the G-strand repeats; most studies to date addressed the roles of this non-coding RNA species at telomeres [35,36]. However, RNAs that carry the telomere C-strand repeat, named ARIA, have also been reported in S. pombe [37,38]. In support of the nexus linking pot1, rad51 and TERRA, POT1 deficiency in human cells was shown to increase TERRA-DNA hybrid formation [7], whereas RAD51 is proposed to directly promote such hybrids through its strand exchange activity [39]. These precedents argue that TERRA may participate in the pathways mediated by pot1-deficiency and rad51/brh2 deletions.

To characterize the potential involvement of telomere-encoded RNA in our mutant strains, we first analyzed the existence of such RNAs by Northern analysis and found little RNA in wild type cells but high levels of transcripts in pot1-deficient mutants. Using strand-specific probes, we showed that RNAs corresponding to both G- and C-strands of telomeres were strongly up-regulated in the pot1 deficient mutant (Fig 6A and 6B). These transcripts were heterogeneous in size but predominantly short (< 1000 nt). RT-PCR analysis confirmed the upregulation of both G- and C-strand RNAs and further suggests that these RNAs include both the subtelomeric region and telomere repeats (S8 Fig). For example, the UT6 subtelomeric region can be amplified using cDNA generated by the CCCTAA₄ primer, consistent with the existence of G-strand TERRA that spans the subtelomeric-telomeric junction (S8 Fig). Likewise, heterogeneous PCR products were generated when C-strand RNA was reverse transcribed with a UT6 primer and then amplified using forward and reverse primers that anneal to UT6 and telomere repeats, respectively (S8 Fig). This again implies the existence of C-strand RNAs that span the subtelomeric-telomeric junction. Notably, our data do not exclude the presence of C-strand RNAs that comprise exclusively of just the telomere repeats, called ARIA [38]. However, to avoid confusion, we will refer to our C-strand subtelomere-telomere RNAs as C-TERRA. In addition, our data are consistent with the existence of C-strand RNAs that comprise exclusively of subtelomere sequence, named ARRET (S8 Fig). Altogether, the Northern and RT-PCR analyses demonstrate that abnormally high levels of both G- and C-strand RNAs are present in pot1-deficient cells, due either to increased transcription or to increased RNA stability. Although the end-point PCR assays could not provide quantitative assessment of RNA levels, the relative signals suggest that C-TERRA was present at higher levels in the mutant (see below for additional evidence).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Telomere repeat-containing RNAs are elevated by pot1-deficiency and suppressed by rad51Δ and brh2Δ.

A. RNAs from the indicated wild type and mutant strains grown in YPD were isolated by acid phenol extraction and subjected to Northern analysis using sequentially probes that detect G-strand repeats, C-strand repeats, and 26S rRNA. B. The ratios of G- and C-strand RNA signals to the 26 rRNA signals from Northern analysis were calculated, and then normalized against the ratio of the pot1^crg1 mutant and plotted. The normalized values for the pot1^crg1 G- and C-strand RNA levels were each set to 100. C. RNAs from the indicated wild type and mutant strains grown in YPD were isolated by acid phenol extraction, treated with Turbo DNase (Thermo Fisher Inc.), and subjected to dot blot analysis using sequentially probes that detect the G-strand repeats, C-strand repeats, and 26S rRNA. D. The ratios of G- and C-strand signals to the 26 rRNA signals from dot blot analysis (n = 2 or 3) were calculated, and then normalized against the ratio of the pot1^crg1 mutant and plotted. P-values were calculated using two-tailed Student’s t tests and displayed.

https://doi.org/10.1371/journal.pgen.1010182.g006

We next compared the levels of telomere repeat RNAs in the pot1 single and double mutants by both Northern and dot blots (Fig 6A–6D). Both analyses showed that while there were drastic increases in TERRA and C-TERRA (by > 20-fold) in the pot1 single mutant, these increases were strongly suppressed by rad51 and brh2 deletions (by > 5-fold). Thus, aberrant transcription of telomere repeats from both G- and C-strands is also regulated by HDR factors in pot1-deficient U. maydis. Lastly, we used specific nuclease treatments to assess the potential existence of RNA-DNA hybrids in the acid phenol-extracted nucleic acid preparations (prior to DNase I treatment), which should be enriched for both RNA and RNA-DNA hybrid (S9 Fig). Interestingly, we found that the G-strand hybridization signals in the pot1^crg1, pot1^crg1 rad51Δ, and pot1^crg1 brh2Δ mutants were all substantially reduced by pre-treatment with DNase I, implying that the G-strand DNA: C-TERRA hybrid (hybrid I) represents a substantial fraction of the signals. The signals were also partially sensitive to RNase H cleavage, indicating the presence of the alternative hybrid (C-strand DNA: TERRA; hybrid II). In contrast, the C-strand signals in the dot blot assays were mostly insensitive to either nuclease, suggesting that they were predominantly in the form of ssRNA or RNA-RNA hybrid. Notably, the C-strand RNA in hybrid I should be degraded by RNase H and should render the C-strand dot blot signal RNase H-sensitive if hybrid I is abundant relative to free C-strand RNAs. This not being the case, we surmise that telomere C-strand RNA may be substantially more abundant than G-strand TERRA in the pot1^crg1 mutant, implying differences between these two RNA species with respect to their production or stability.

U. maydis Pot1 suppresses the toxic activities of core HDR factors at telomeres

Previous analysis of S. pombe Pot1 (SpPot1) suggests that fungal Pot1 may act differently than the mammalian orthologs in telomere protection. To wit, rather than aberrant recombination, loss or inactivation of SpPot1 leads to rapid telomere resection/loss followed by frequent circularization of chromosomes to stabilize the genome and sustain cell viability [12,13]. In contrast, we showed that the phenotypes of UmPot1-deficient cells are very similar to those of comparable mammalian cells, suggesting that suppression of HDR activities may be a widely conserved function of Pot1 in evolution. Importantly, the loss of viability triggered by U. maydis pot1-deficiency is almost completely suppressed by rad51Δ or brh2Δ, confirming the toxicity of HDR factors in the context of telomere dysfunction. This toxicity is at least somewhat specific given that rad51Δ fails to ameliorate the telomere toxicity triggered by ku70 deficiency [23]. Instead, the ku70 deficiency-induced abnormalities are completely suppressed by concurrent blm deletion [25]. Thus, there is a clear division of labor among telomere proteins in managing aberrant repair. The mechanistic underpinnings for this remain unclear and demand further investigation.

Viewed from another perspective, it is striking that in U. maydis, ku70 and pot1 become dispensable for cell viability when specific DNA repair factors are deleted. Such examples of telomere protein dispensability in the context of DNA repair deficiency reify the concept that telomeres have evolved to restrain aberrant repair. It is worth noting that in mammalian cells, there are few examples where the loss of DNA repair factors has been shown to restore the viability of cells missing a key telomere protein. This is probably because mammalian DNA repair factors are often essential for viability in normal mitotic cells. The reduced dependency of U. maydis on DNA repair makes it possible to firmly establish the causal connections between telomere deprotection, aberrant telomere DNA repair, and loss of cell viability.

U. maydis Pot1 promotes telomere maintenance by enhancing the function of Rad51

POT1 has been proposed to promote telomere maintenance in mammalian cells through multiple mechanisms. First, POT1 unwinds G-quadruplexes in vitro, and this biochemical activity is thought to facilitate telomerase activity and telomere replication [40,41]. Second, POT1 binds to and acts in conjunction with the CST complex to facilitate telomere replication and C-strand fill-in synthesis [14,18]. Our analysis of U. maydis Pot1 suggests an additional, non-mutually exclusive mechanism, namely through the regulation of Rad51 activity. Notably, while we showed that Pot1 can stimulate the strand exchange activity of Rad51 through direct protein-protein interaction, how this in vitro regulation reflects the in vivo mechanism of Rad51 at telomeres remains unclear. If Pot1 enhances the formation or stability Rad51 filaments, this could promote complete replication by stabilizing stalled forks. However, it is also clear that rampant strand exchange by Rad51 at telomeres is toxic to the cell and is restrained by Pot1 (see the section above). Thus, it is possible that Pot1-Rad51 interaction may positively regulate some aspects of Rad51 activity while suppressing others in a context dependent manner, e.g., through modulation by other telomere and DNA-repair factors. More studies will be necessary to disentangle the molecular basis of Pot1-mediated regulation of Rad51 in U. maydis.

Notably, mammalian RAD51 is also implicated in telomere replication [33], but a direct POT1-RAD51 connection has not been reported. Instead, there is substantial evidence that RAD51 is regulated by CST, the complex that acts in parallel with shelterin to promote telomere maintenance. For example, CST associates with Rad51, and CST deficiency impairs the recruitment of RAD51 to genomic sites of replication stress [42]. In addition, CTC1 disease mutations that compromise telomere maintenance often disrupt its association with RAD51 [43]. Whether POT1 acts in conjunction or in parallel with CST to regulate RAD51 function at mammalian telomeres is an interesting question for future analysis. Conversely, it will be worthwhile to determine if U. maydis CST subunits regulate Rad51 function. However, it should be noted that only Stn1 and Ten1 homologs are readily discernible in the U. maydis genome (just like S. pombe), suggesting that some fungal lineages may have undergone significant remodeling of the CST complex [24,44].

Aberrant transcription of telomere repeats at de-protected telomeres

There is growing evidence that while TERRA is crucial for normal telomere regulation [34,45], it is often aberrantly expressed from damaged telomeres and can promote abnormal telomere metabolism by forming RNA-DNA hybrids or by interacting with a multiplicity of factors [46–49]. Of more direct relevance to the current study, the depletion or knockout of shelterin components in both mammals and fungi has been shown to trigger the production of telomere repeat RNAs as well as RNA-DNA hybrids. For example, depleting human TRF2 leads to rampant up-regulation of TERRA, which in turn alters histone methylation and promotes end-to-end fusion [50]. Similarly, knocking out human POT1 triggers both abnormal recombination as well as the accumulation of telomere RNA-DNA hybrid [7]. Consistent with these precedents, we observed dramatic upregulation of telomere repeat containing RNAs in pot1-deficient U. maydis. However, both the structure of the RNAs and their relative expression manifest notable features.

One notable feature is the structure of the C-strand RNA. In mammalian cells, this RNA species is evidently expressed at very low levels (if at all) from normal telomeres, and its structure has not been investigated in detail [35,36]. In S. pombe, the C-strand RNAs are synthesized to some degree from normal telomeres and are dramatically up-regulated in a number of mutants, including rap1Δ, taz1Δ, and poz1Δ [38,51]. However, unlike C-TERRA in U. maydis, which encompasses both subtelomere and telomere repeats, the S. pombe C-strand RNAs are reported to comprise just the telomere repeat (ARIA) or subtelomere regions (ARRET). This discrepancy indicates either there are species-specific differences in the biogenesis of C-strand RNA, or the S. pombe ARIA and ARRET could arise through post transcriptional processing. An unresolved conundrum concerning C-strand repeat RNA (ARIA or C-TERRA) is that transcription must initiate within the telomere repeat region, which (unlike subtelomeric Pol II promoters responsible for TERRA transcription) does not contain promoter- or initiator-like elements. Thus, an unusual, promoter-independent mode of RNA synthesis is probably involved. A potential precedent for this is provided by a recent report of MRN-dependent transcription by RNA Polymerase II at DSBs [52]. Regardless of the potential mechanisms, our observation suggests that C-strand RNA synthesis may be relatively common at deprotected telomeres and is worthy of further investigation. Conversely, the growing evidence for promiscuous transcription at de-protected telomeres suggests that a key function of telomere proteins is in fact to suppress aberrant transcription.

Another unexpected finding is the greater abundance of C-TERRA relative to TERRA in the pot1 mutants. As described before, both the RT-PCR and dot blot analyses support the idea that C-TERRA is present at higher levels when pot1 is depleted (S8 and S9 Figs). Whether this is due to enhanced synthesis or greater stability of the C-strand RNA is unclear. The high levels of C-TERRA may partially account for the formation of hybrid I (G-strand DNA: C-TERRA), which was readily detected in the mutant. Most studies of telomere RNA-DNA hybrid have focused on the roles of hybrid II (C-strand DNA: TERRA) and the associated R-loops, and have implicated these structures in telomere abnormalities. Our identification of hybrid I uncovers yet another structure that could be responsible for exacerbating telomere damage.

Implications for the evolution of telomere protein complexes

The discoveries that basidiomycete Pot1 has four OB-fold domains and that UmPot1 interacts with Rad51 have evolutionary implications. Based on structural and functional considerations, it has been argued that RPA, CST and POT1-TPP1 may represent ancient paralogs (i.e., RPA1 ≈ CTC1 ≈ POT1 and RPA2 ≈ STN1 ≈ TPP1) and accordingly may share more structural and functional similarities than hitherto realized [17]. Our analysis, coupled with that by Myler et al. [28], indicates that the common ancestor of fungal and metazoan POT1 probably contained four OB-folds and utilized the two central OBs (OB1 and OB2) for high affinity DNA-binding, just like RPA1. This notion reinforces the growing evidence for structural similarities–and hence evolutionary kinship–between these OB-fold-rich complexes (S10 Fig) [53]. Also, as discussed above, the regulation of RAD51 by CST in mammals and its regulation by Pot1 in U. maydis underscore functional similarities between CST and POT1-TPP1, albeit in different organisms. Further analysis of RPA, CST and POT1-TPP1 in fungi, metazoan and deeper branches of eukaryotes should allow us to adjudicate the evolutionary relationships between these complexes.

Pot1 homolog identification, alignment, structural prediction, and phylogenetic analysis

Fungal homologs of UmPot1 were identified using BLASTp or PSI-BLAST with appropriate taxonomic restrictions. Similar numbers of homologs that belong to the Basidiomycota, Ascomycota, and Chytridiomycota/Zoopagomycotina lineages were collected for further analysis. The IDs for these homologs are listed in S3 Table. Homologs were aligned using either PSI-Coffee (http://tcoffee.crg.cat/) or PROMALS3D (http://prodata.swmed.edu/promals3d/promals3d.php), and the output displayed using MView (https://www.ebi.ac.uk/Tools/msa/mview/; set to default parameters except coloring by group). The secondary structure predictions associated with the alignments were obtained from the PROMALS3D output and added to the MView displays. Homology modeling was carried out using either PHYRE2 (http://www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index) or SWISS-MODEL (https://swissmodel.expasy.org/) and the predicted structures displayed using PYMOL. To test the hypothesis that basidiomycete OB-N may have arisen through duplication of OB1, we assessed the similarity of basidiomycete OB1 to both basidiomycete OB-N and metazoan OB-N. Briefly, a multiple sequence alignment between the three groups of OB folds was generated using T-Coffee. The Phylip alignment was then used to infer phylogeny via FastME 2.0 (http://www.atgc-montpellier.fr/fastme/) [54] and the results plotted using FigTree [55]. The same strategy was used to analyze basidiomycete Pot1 OB2 and RPA1 OB-N as possible sources of basidiomycete Pot1 OB-N.

Ustilago maydis strain construction and growth conditions

Standard protocols were employed for the genetic manipulation of U. maydis [56–58]. All U. maydis strains used in this study were haploid and were derived from the UCM350 background [57,59] and are listed in S1 Table.

The pot1^crg1 strain was constructed by integrating a linearized pRU12-pot1 plasmid containing fragments that correspond to (1) 700 bp of the pot1 promoter and (2) the first 700 bp of the pot1 ORF into the pot1 genomic locus in UCM350. Briefly, the two pot1 fragments were generated by PCR using appropriate oligos with flanking restriction enzyme sites (S2 Table), and then cloned together in between the NdeI and EcoRI sites of pRU12 [60]. Correct recombination of XbaI treated pRU12-pot1 into the genomic locus results in the insertion of the cbx^R marker and the crg1 promoter precisely adjacent to and upstream of the pot1 ATG initiation codon. Following transformation, the correct mutant strains (pot1^crg1) were selected on YPA supplemented with carboxin (4 μg/ml) and confirmed by PCR. The pot1ΔN^crg1 was generated in identical fashion except that nucleotide (nt) 1126 to 1825 of the pot1 ORF was used in place of the nt 1 to 700 fragment in the cloning of the pRU12-pot1ΔN disruption plasmid. The pot1ΔN strain was constructed by integrating linearized pCbx-pot1ΔN, a derivative of pRU12-pot1ΔN that contains (i) a 5’ homology region with nt -1400 to -700 of the pot1 promoter, and (ii) a 3’ homology region with nt -700 to -1 of the pot1 promoter fused to nt 1226 to 1825 of the ORF. The rad51Δ and brh2Δ disruption cassettes have been described before [24]. The pot1^crg1 rad51Δ and pot1^crg1 brh2Δ mutants were generated by transforming pot1^crg1 with the same rad51 and brh2 knockout cassettes as those used for the derivation of the single mutants [24]. The pot1^crg1 and pot1ΔN^crg1 strains were grown in YPA (1% yeast extract, 2% peptone, 1.5% arabinose) or YPD (1% yeast extract, 2% peptone, 2% dextrose) to induce or repress pot1 expression, respectively.

For growth analysis of the mutants, liquid cultures of U. maydis strains were diluted to OD₆₀₀ of 0.2 (corresponding to ~ 1.5 x 10⁶ cells/ml). Four microliters of 6-fold serial dilutions of the cultures were then spotted onto YPA or YPD plates and grown at 30°C for 2 to 3 days.

RT-qPCR analysis of pot1 RNA levels

For RT-qPCR analysis of pot1 expressions levels, RNAs were extracted and purified from log-phase U. maydis cultures using acid phenol extraction [61]. The samples were treated with TURBO DNase (Thermo Fisher) and then reverse transcribed in 20 μl reactions that contained 1x SSIV Buffer, 12.5 μM primer (Pot1-PCR-3020R), 1 mM dNTP, 5 mM DTT, 20 U NxGen RNase Inhibitor (Lucigen), and 40 U Superscript IV RT (Thermo Fisher). The reaction mixtures were incubated at 55°C for 40 min and then heated at 70°C for 15 min to inactivate the reverse transcriptase. qPCR was carried out in Chai Open qPCR Dual Channel thermocycler using Chai Green Master mix with 8 μM primers (Pot1-PCR-2821F and Pot1-PCR-3020R) and 1 μl of cDNA. Cycling parameters were as follows: 45 cycles of 95°C for 30 sec, 53°C for 20 sec, 72°C for 40 sec.

Telomere length and structural analyses

Southern analysis of telomere restriction fragments (TRF Southern) was performed using DNA treated with PstI as described previously [24,26]. The blots were hybridized to a cloned restriction fragment containing 82 copies of the TTAGGG repeats. For quantitation, we used the formula ‘mean TRF length = Σ (OD_i × L_i) / Σ (OD_i)’ because the majority of TRF length variation stems from differences in subtelomeric lengths [62,63]. The in-gel hybridization assays for detecting ssDNA on the G- and C-strand have also been described [23]. STELA for telomeres bearing UT4 subtelomeric elements was performed essentially as reported before [22]. To detect telomere-telomere fusions, chromosomal DNA was subjected to PCR using two subtelomeric primers (UT4-F and UT6-F) that are extended by DNA polymerase towards the chromosome ends. The PCR reactions (20 μl) contained 1x Failsafe PreMix H, 0.1 μM each of UT4-F and UT6-F, and 2 U Failsafe polymerase; the thermocycling consisted of 36 cycles of 94°C for 30 sec, 63°C for 30 sec, and 72°C for 2 min. The fusion PCR products were detected by Southern using either a UT4 or UT6 subtelomeric probe [27]. C-circle and G-circles assays were performed as before except that NxGen phi29 DNA Polymerase (Lucigen Corp.) were used for rolling circle amplifications [26].

Analysis of telomere repeat-containing RNAs

RNAs were purified from log-phase U. maydis cultures using acid phenol extraction [61]. For Northern analysis, 3 μg of total RNA were resolved by electrophoresis through a 1.0% agarose/formaldehyde gel [64]. After photography under UV light to visualize ethidium bromide-stained RNAs, the gel contents were transferred to a Hybond-N membrane (GE Healthcare) and hybridization was performed in Church mix using a ³²P-labeled single-strand oligonucleotide complementary to the G-strand repeats (CCCTAA₈). Following exposure to a phosphor screen, the blot was stripped and first re-probed for the C-strand telomere repeat RNA using ³²P-labeled TTAGGG₈, and subsequently for 26S rRNA using a PCR fragment (amplified with the UmrRNA_26S_121F and UmrRNA_26S_642R primers (S2 Table)). For dot blot analysis, 0.2 μg of RNAs (untreated or pre-treated with various nucleases) in 10 μl were diluted with 30 μl of RNA incubation/denaturation solution (65.7% formamide, 7.7% formaldehyde, 1.3x MOPS buffer), heated at 65°C for 5 min, and then cooled on ices. After adding 40 μl of 20x SSC, the entire samples were applied to Hybond-N using a Bio-Dot Microfiltration Apparatus (Bio-Rad Laboratories, Inc.). After UV crosslinking, the blot was hybridized sequentially to the same probes as those for Northern analysis. For end-point RT-PCR, the RNA samples from acid phenol extraction were first treated with Turbo DNase (Thermo Fisher Inc.). Fifty ng of RNAs were subjected to reverse transcription using Superscript III RT at 55°C for 40 min, followed by heating at 70°C for 15 min to inactivate RT, and then subjected to PCR (33–37 cycles of 95°C for 15 sec, 66°C for 15 sec, and 72°C for 30 sec) with Q5 polymerase (NEB Inc.) and appropriate primers (S2 Table).

Purification of Pot1-FG and truncation variants

To express full length and deletion mutants of U. maydis Pot1, we amplified the pot1 open reading frame carrying a C-terminal FLAG tag from genomic DNA by PCR amplification using suitable primer pairs (S2 Table). The PCR fragments were cloned into the pSMT3 vector to enable the expression of His₆-SUMO-Pot1-FG fusion proteins. BL21 codon plus strains bearing the expression plasmids were grown in LB supplemented with kanamycin and induced by IPTG as previously described [26]. Each fusion protein was purified by (i) Ni-NTA chromatography, (ii) ULP1 cleavage and, and (iii) anti-FLAG affinity chromatography as previously described.

EMSA assays

For DNA binding assays, purified Pot1-FG was incubated with 1–10 nM P³²-labeled probe and 0.5 μg/μl BSA in binding buffer (25 mM HEPES-KOH, pH 7.5, 5% glycerol, 3 mM MgCl₂, 0.1 mM EDTA, 1 mM DTT) at 22°C for 15 min, and then subjected to electrophoresis at 0.8 V/cm through a 6% non-denaturing polyacrylamide gel in 1x TBE, followed by PhosphorImager analysis.

Pull down assays for Pot1-Rad51 and Pot1-Brh2 interactions

Pot1-FLAG was immobilized on FLAG affinity resin and incubated with 1 μM purified Rad51 or MBP-Brh2/DSS1 complex at 4°C in FLAG(65) buffer (50 mM Tris-HCl, pH 7.5, 65 mM NaCl, 10% glycerol, 0.1% NP-40, 2.5 mM MgCl₂, 1 mM DTT). Following incubation with mixing on a rotator at 4°C for 1 h, the beads were washed 4 times with 0.5 ml of the FLAG(100) buffer (same as FLAG(65) except that NaCl is at 100 mM), and then the bound proteins eluted with 60 μl FLAG(150) (same as FLAG(65) except that NaCl is at 150 mM) containing 0.2 mg/ml FLAG₃ peptide. The eluates were analyzed by Western Blot using anti-Rad51 or anti-MBP antibodies.

Rad51 strand exchange assays

The assays were modified from published protocols [31,32] and were performed in 15 μl of SE buffer (25 mM Tris.HCl, pH 7.5, 25 mM KCl, 2.4 mM MgCl₂, 1 mM DTT, 2 mM ATP) containing 0.25 μM labeled dsDNA (39-mer with the P³²-labeled top strand), 0.75 μM unlabeled ssDNA (70-mer complementary to the top strand of the 39-mer), and indicated amounts of Rad51 and Pot1-FG. After 20 min of incubation at 35°C, the reactions were terminated by the addition of 5 μl stop solution (80 mM EDTA, 4% SDS) containing 0.5 μg/μl proteinase K, and incubated for another 20 min at 35°C. Following the addition of 4 μl loading dye (40% glycerol, 0.25% bromophenol blue, 0.25% xylene cyanol), the samples were analyzed by electrophoresis at 0.8 V/cm through a 10% polyacrylamide gel in 1x TBE, and visualized by PhosphorImager scanning.