Duplication and Functional Specialization of the Telomere-capping Protein Cdc13 in Candida Species*

Background: Cdc13 in Candida species is small and missing conserved domains. Results: A second Cdc13 homologue (Cdc13B) was shown to form heterodimers with Cdc13 and to mediate overlapping functions in telomere protection. Conclusion: Cdc13 has undergone duplication and functional specialization in a branch of Saccharomycotina yeast. Significance: Understanding the assembly and mechanisms of fungal CST complex provides insights on the evolution and function of this crucial telomere complex. The budding yeast G-tail binding complex CST (Cdc13-Stn1-Ten1) is crucial for both telomere protection and replication. Previous studies revealed a family of Cdc13 orthologues (Cdc13A) in Candida species that are unusually small but are nevertheless responsible for G-tail binding and the regulation of telomere lengths and structures. Here we report the identification and characterization of a second family of Cdc13-like proteins in the Candida clade, named Cdc13B. Phylogenetic analysis and sequence alignment indicate that Cdc13B probably arose through gene duplication prior to Candida speciation. Like Cdc13A, Cdc13B appears to be essential. Deleting one copy each of the CDC13A and CDC13B genes caused a synergistic effect on aberrant telomere elongation and t-circle accumulation, suggesting that the two paralogues mediate overlapping and nonredundant functions in telomere regulation. Interestingly, Cdc13B utilizes its C-terminal OB-fold domain (OB4) to mediate self-association and binding to Cdc13A. Moreover, the stability of the heterodimer is evidently greater than that of either homodimer. Both the Cdc13 A/A homodimer and A/B heterodimer, but not the B/B homodimer, recognized the telomere G-tail repeat with high affinity and sequence specificity. Our results reveal novel evolutionary elaborations of the G-tail-binding protein in Saccharomycotina yeast, suggesting a drastic remodeling of CDC13 that entails gene duplication, fusion, and functional specialization. The repeated and independent duplication of G-tail-binding proteins such as Cdc13 and Pot1 hints at the evolutionary advantage of having multiple G-tail-binding proteins.

Interestingly, recent analysis of Cdc13 homologues reveals a high degree of evolutionary divergence. In particular, we and others have identified and characterized a family of Cdc13 homologues in Candida species that are noticeably smaller than ScCdc13. These homologues lack the N-terminal half of the prototypical Cdc13 and contain just two OB-fold domains: DBD and OB4 (Fig. 1A) (11,16,17). Despite this difference, we showed that the Candida albicans Cdc13 (CaCdc13) indeed associates with telomere DNA in vivo and is required for normal telomere length regulation (18). We further demonstrated high affinity telomere DNA binding by the Candida tropicalis Cdc13 (CtCdc13) in vitro and found that this binding activity requires dimerization of CtCdc13 through its OB4 domain (18). We therefore concluded that the Candida Cdc13 proteins are indeed orthologous to the C-terminal half of ScCdc13 (i.e. they share a common ancestor and perform similar functions). Nevertheless, the lack of the N-terminal half of Cdc13 poses a conundrum, indicating that the functions mediated by the missing domains such as Pol1 binding and telomerase recruitment must either be absent or taken over by other proteins in Candida species. In the present report, we have identified and characterized a second family of Cdc13-like proteins in the Candida species named Cdc13B. We show here that Cdc13B is required for telomere regulation but is functionally distinct from the initially characterized Candida Cdc13 (henceforth referred to as Cdc13A). Like Cdc13A, Cdc13B utilizes the OB4 domain for binding to itself and Stn1. However, unlike Cdc13A, Cdc13B binds telomere G-strand with low affinity and moderate sequence specificity. Surprisingly, Cdc13A and Cdc13B can form heterodimers, and each protein exhibits a preference for heterodimerization over homodimerization. The Cdc13A/ Cdc13B dimer (A/B dimer) binds telomere G-tails with high affinity and sequence specificity in vitro, suggesting that it could constitute the predominant form of Cdc13 in vivo. We propose an evolutionary model to account for the presence of two small Cdc13 paralogues in the Candida clade and that of just one large Cdc13 in Saccharomyces. Our findings provide insights on the assembly and DNA binding mechanisms of CST as well as its unusual elaboration in a fungal branch.
Sequence Analysis-Cdc13, Cdc13A, and Cdc13B homologues from Saccharomycotina yeast were identified in the NCBI and Broad Institute databases by BLAST or psi-BLAST searches. The multiple sequence alignment was generated using the PROMALS server and displayed using BoxShade.
Telomere Analyses-The telomere length analysis and the two-dimensional gel analysis of circular and linear telomeric DNA were performed as described previously (24).
Chromatin Immunoprecipitation-The assays were performed as described previously (18) except that the enrichment of telomeric DNA was measured by PCR amplification (95°C for 30 s, 55°C for 30 s, and 72°C for 30 s for 25 cycles) of a 300-bp subtelomeric fragment on chromosome 3. The forward and reverse primers were Chr3-F and Chr3-R, respectively (Table 2). Pilot titration experiments indicated that for the range of DNA concentrations analyzed in the assays, the PCR signals were roughly proportional to the amount of starting DNAs.
Gel Electrophoretic Mobility Shift Analysis-To analyze the DNA binding activity of Cdc13B, the entire ORF (amino acids 1-561) was cloned in between the SacI and XhoI sites of the pSMT3 vector to enable the expression of the His 6 -SUMO-Cdc13B fusion protein. Because of the atypical translation of the CUG codon in Candida species, the CTG triplets encoding amino acids 175 and 383 of CaCdc13B were mutated to TCG to enable the expression of wild type proteins in Escherichia coli (25). Following induction, extracts were prepared and the fusion proteins purified with Ni-NTA chromatography as described previously (24). In some assays, the fusion protein was first cleaved by the ULP1 protease and then subjected to DNA binding analysis. To analyze the activity of the A/A, A/B, and B/B complexes, we co-expressed appropriate His 6 -SUMO and GST-FLAG fusion proteins in E. coli and then purified the respective complexes by Ni-NTA and M2 affinity chromatography. Binding reactions contained 50 mM Tris-HCl, pH 7.5, 2 mM MgCl 2 , 50 mM NaCl, 0.1 mM EDTA, 1 mM DTT, 40 ng/l poly(dI-dC), 10% glycerol, and specified concentrations of probe and proteins. Following incubation at 25°C for 20 min, the reaction mixtures were subjected to gel electrophoresis through a 5% nondenaturing polyacrylamide gel (acrylamide: bis ϭ 44:1) to resolve the free probe from the DNA-protein complex.
Co-expression and Pulldown Assays-The DNAs encoding full-length CaCdc13A, CaCdc13B, and CaStn1, as well as various domains (see Table 1 for the amino acids included in each expression construct) were amplified by PCR and cloned into the pSMT3 vector (26) and the pGEX4T-2 vector (GE Healthcare) to enable their expression as His 6 -SUMO and GST-FLAG fusion proteins, respectively. Each His 6 -SUMO fusion protein was expressed alone or co-expressed with a GST-FLAG fusion protein in E. coli BL21(DE3). The growth and induction protocols as well as the extract preparation procedures were as described previously (18). For anti-FLAG pulldown assays, ϳ500 l of 10 mg/ml extract was incubated with 20 l of M2-agarose beads (Sigma) in FLAG(250) buffer (50 mM Tris-HCl, pH 7.5, 250 mM NaCl, 10% glycerol, 0.1% Nonidet P-40, 2.5 mM MgCl 2 , and 1 mM DTT). Following incubation with constant mixing on a rotator at 4°C for 2 h, the beads were washed five times with 0.5 ml of FLAG(150) buffer (same as FLAG(250) except that it contained 150 mM NaCl), and then the M2-bound proteins were eluted with 60 l of FLAG(150) containing 0.2 mg/ml 3ϫ FLAG peptide. The eluates were analyzed by SDSpolyacrylamide gel electrophoresis followed by staining with Coomassie Brilliant Blue R-250 or Western blotting. The GST pulldown assays were carried out using glutathione-Sepharose (GE Healthcare). The binding and washing buffers were identical to those of the M2 pulldown assays, and the elution buffer consisted of FLAG(150) supplemented with 15 mM reduced glutathione.

Candida Species Contain a Second Family of Cdc13-like
Proteins-Iterative psi-BLAST searches at the NCBI database uncovered a second family of Cdc13-like proteins in Candida species, which we named Cdc13B. In sequence analysis, most Cdc13B align well with the DBD and OB4 domains of Candida Cdc13A as well as Saccharomyces Cdc13 (Fig. 1, supplemental Fig. 1, and data not shown). The only exception is Candida guilliermondi Cdc13B, which evidently lacks the OB4 domain (supplemental Fig. 1). Overall, the ␣-helices and ␤-strands of the two OB folds constitute the most conserved regions in the sequence alignments, implying that these domains in all three protein families are structurally similar (supplemental Fig. 1 and data not shown). Conversely, the region in between the two OB folds is poorly conserved and variable in length (e.g. ϳ80 amino acids long in Candida lusitaniae Cdc13B and ϳ460 amino acids long in Pichia stipitis Cdc13B) and may thus function partly as a flexible linker. Notably, a CDC13B ORF is present in each of the fully sequenced Candida genomes. Moreover, within this clade, the CDC13A and CDC13B genes form well defined monophyletic groups, consistent with the duplication of an ancestral gene prior to Candida speciation. By contrast, we were unable to detect any CDC13B-like genes in the Saccharomyces and Kluyveromyces clades, suggesting that either the duplication did not take place or that the duplicated gene was lost in these lineages.
Cdc13B and Cdc13A Mediate Overlapping but Nonredundant Functions in Telomere Length Regulation-To address the function of Cdc13B, we attempted to generate a homozygous null strain by performing two rounds of URA-Blaster-mediated gene replacement. Although we were able to obtain several

Duplication and Functional Specialization of Cdc13
OCTOBER 4, 2013 • VOLUME 288 • NUMBER 40 heterozygous clones, none of the more than 30 transformants screened after the second round of gene replacement was a homozygous null mutant, suggesting that CDC13B is essential for cell viability, just like CDC13A (18). However, we were able to construct and characterize a CDC13A ϩ/Ϫ CDC13B ϩ/Ϫ combination mutant. We subjected the individual and compound heterozygous deletion strains to both growth and telomere length analyses. Upon repeated passages in YPD medium, none of the mutants exhibited an obvious growth defects or senescence (data not shown). However, although neither the CDC13A ϩ/Ϫ nor the CDC13B ϩ/Ϫ strain exhibited changes in telomere lengths, the combination mutant was found to contain longer and more heterogeneous telomeres (Fig. 2, A and B). The synergistic effect of deleting one copy each of CDC13A and CDC13B suggests that these genes mediate overlapping functions in telomere length regulation.
Longer and more heterogeneous telomeres are often associated with the accumulation of extrachromosomal telomere circles (t-circles) (24,27). Consistent with this notion, we found elevated levels of t-circles only in the compound heterozygous strain ( Fig. 2C and data not shown). The amount of t-circles in this strain is similar to that found in a strain carrying a single copy of CDC13A-TAP, which was also reported previously to contain long and heterogeneous telomeres ( Fig. 2C and Ref. 18).
To determine whether Cdc13B functions directly at telomeres, we tagged the intact CDC13B locus of the CDC13B ϩ/Ϫ strain with a TAP tag. Western analysis indicated that the tagged protein was expressed at appreciable levels (data not shown). Chromatin immunoprecipitation assays revealed significant association of Cdc13B-TAP with telomere DNA, to the same extent as the previously characterized Cdc13A-TAP, suggesting that both proteins act directly at telomeres (Fig. 2D).
Cdc13B Binds Telomere G-strand with Low Affinity and Sequence Specificity-The presence of a DBD-like domain in Cdc13B (supplemental Fig. 1) suggests that this protein may be capable of binding to G-tails, just like Cdc13A (18). We expressed Cdc13B as a His 6 -SUMO fusion protein and subjected the purified protein to electrophoretic mobility shift assays (EMSA). Despite using a variety of assay conditions and buffers, we were able to observe only low affinity binding of the fusion protein to G-tails (Fig. 3, K d Ͼ Ͼ 1 M). As expected, removing the His 6 -SUMO tag from the fusion protein by ULP1 treatment resulted in a shift in the mobility of the DNA-protein complex, indicating that the complex was due to Cdc13B rather than a contaminant.
Cdc13B Binds Stn1 and Can Form a Homo-oligomer as Well as a Hetero-oligomer with Cdc13A-Next we analyzed the ability of Cdc13B to interact with Stn1, which is an established  . Low affinity DNA binding by C. albicans Cdc13B. A, the purified His 6 -SUMO-CaCdc13B fusion protein was analyzed by SDS-PAGE and Coomassie staining. The identity of the protein was confirmed by ULP1 cleavage, which as predicted released the His 6 -SUMO tag. B, the intact His 6 -SUMO-CaCdc13B fusion protein and the ULP1-treated protein (1 M) were subjected to EMSA using single-stranded oligonucleotide probes (7.5 nM) that consisted of two copies of the C. albicans telomere G-and C-strand repeats (CaC2, (CATCCGTACACCAAGAAGTTAGA) 2 ; CaG2, (TCTAACTTCTTGGTGTACGGATG) 2 ). Note that the fraction of probe in the complexes represents less than 1% of the total probe used in these assays.
activity of Cdc13A (13). His 6 -SUMO-Stn1 was expressed alone or co-expressed with GST-Cdc13B-FLAG in E. coli (Fig. 4A), and the extracts were subjected to affinity purification using M2 (anti-FLAG)-agarose. A significant amount of Stn1 fusion was detected in the affinity eluate only when it was co-expressed with the Cdc13B fusion, indicating that the two proteins can form a complex, just like Cdc13A and Stn1 (Fig. 4B,  both panels, lanes 1 and 2).
Another established property of both the Cdc13 proteins in Saccharomyces and the small Cdc13A proteins in Candida species is dimerization (5)(6)(7)18). To assess the oligomerization property of Cdc13B, we co-expressed various combinations of His 6 -SUMO and GST-FLAG fusion proteins containing either Cdc13A or Cdc13B in E. coli and subjected the extracts to M2 (anti-FLAG) affinity purification and Western analysis. As shown in Fig. 4B, both GST-Cdc13A-FLAG and GST-Cdc13B-FLAG associated with substantial amounts of His 6 -SUMO-Cdc13B, indicating that Cdc13B can bind to itself as well as the Cdc13A paralogue (lanes 9 and 10). Moreover, C-terminal truncation of Cdc13B appears to abolish these interactions, as the majority of His 6 -SUMO-Cdc13B proteolytic fragments (marked by a bracket) were not recovered by M2 affinity purification (Fig. 4B, lanes 9 and 10). This observation implicates the C-terminal OB4 domain of Cdc13B in Cdc13A/Cdc13B (A/B) and Cdc13B/Cdc13B (B/B) complex formation.
To address the role of the OB4 domain more directly, we performed two additional sets of co-expression/purification analysis. In one set, full-length Cdc13B (His 6 -SUMO-Cdc13B) was synthesized alone or together with the OB4 domain of Cdc13A or Cdc13B (GST-Cdc13A OB4 -FLAG or GST-Cdc13B OB4 -FLAG) (Fig. 4C, lanes 4 -6 and 10 -12, and Table  1). In another set, just the OB4 domain of Cdc13B (His 6 -SUMO-Cdc13B OB4 ) was co-expressed with the same domain of Cdc13A or Cdc13B (GST-Cdc13A OB4 -FLAG or GST-Cdc13B OB4 -FLAG) (Fig. 4C, lanes 1-3 and 7-9). As predicted, both the A/A and A/B complexes can be detected in the two sets of assays, implying that the OB4 domains are sufficient for homo-and hetero-oligomer formation (Fig. 4C, lanes 8 -9 and  11-12). More surprisingly, the levels of the A/B complex were reproducibly and substantially higher than those of the B/B

Duplication and Functional Specialization of Cdc13
OCTOBER 4, 2013 • VOLUME 288 • NUMBER 40 complex in the M2 elution, suggesting that Cdc13B is more efficient at forming hetero-oligomers than homo-oligomers.
We then carried out a "reciprocal" experiment in which GST-tagged Cdc13A OB4 or Cdc13B OB4 was used to pull down His 6 -SUMO-tagged Cdc13A OB4 (Fig. 5A). Both the bait and target proteins also contained the FLAG tag, allowing us to determine simultaneously their expression levels, which were found to be comparable (Fig. 5B). Consistent with the previous experiment, the efficiency of the A/B complex formation was again higher than that of the A/A complex (Fig. 5C, lanes 2 and  3). Taken together, our results indicate that C. albicans Cdc13A and Cdc13B have a greater propensity to form hetero-oligomers than homo-oligomers.
To assess the stoichiometry of the complex, we performed glycerol gradient sedimentation analyses. Because the A/B complex appears to be more stable, we first isolated the complex comprising full-length His 6 -SUMO-Cdc13A and GST-Cdc13B-FLAG by sequential Ni-NTA, M2, and glutathione affinity chromatography and then characterized the sedimentation behavior of the complex. Each protein exhibited a peak signal at ϳ160 kDa, consistent with the predicted size for a mixed dimer (158 kDa, Fig. 6A). The analysis of the "full-length" A/B complex was performed three times using materials obtained by different combinations of affinity steps, and the ϳ160-kDa peaks for both subunits were reproducibly observed (data not shown). We then tested a number of A/B complexes bearing truncated Cdc13A or Cdc13B and found that the complex comprising GST-Cdc13A OB4 -FLAG and His 6 -SUMO-Cdc13B behaved homogeneously as a ϳ140-kDa complex (Fig.  6B). This estimated size was again close to that predicted for a heterodimer (125 kDa). Coomassie staining of the two proteins in the putative dimer peak fraction was also consistent with the two proteins being at a ratio of 1:1 (Fig. 6C). Taken together, our observations suggest that the simplest form of the A/B complex is likely to be a heterodimer.
The Cdc13A/B Dimer Binds the Cognate Telomere G-strand with High Affinity and Sequence Specificity-The propensity of C. albicans Cdc13A and Cdc13B to form heterodimers prompted us to compare the DNA binding properties of the homodimers with that of the heterodimer. For this analysis, we co-expressed appropriate combinations of His 6 -SUMO and GST-FLAG fusion proteins and isolated the A/A, A/B, and B/B complexes by sequential Ni-NTA and M2 affinity chromatography. As expected, each preparation contained an approximately equal molar amount of the two fusion proteins that reacted with the relevant antibodies in Western analysis (Fig.  7A). In gel mobility shift assays, both the A/A and A/B complex substantially reduced the levels of free probes (46-nt oligonucleotides with two copies of the C. albicans telomere G-strand repeat) with an ϳK d of 40 and 55 nM, respectively (Fig. 7B, lanes  1-9, and 7C). Both the A/A and A/B complex appeared to experience aggregation; the majority of shifted probes for both the A/A and A/B complexes migrated minimally into the gel. In contrast, the B/B complex interacted weakly with the same probe (Fig. 7B, lanes 10 and 11). Inspection of the pattern of shifted probes revealed an interesting difference between the A/A and A/B samples; although a significant proportion of the A/A-DNA complex migrated as a heterogeneous smear throughout the lane, little smear was observed in the A/B-DNA samples. Although the reason for the discrepancy is not clear, one possible explanation is that the A/A-DNA complex may be unstable and dissociate during electrophoresis.
We than analyzed the DNA binding specificity of the A/B complex by competition assays. As shown Fig. 7D, the complex exhibited substantial preference for the cognate telomere repeat in comparison with two non-telomeric oligonucleotides (Non-Telo 1 and Non-Telo 2). More specifically, the concentrations of the nontelomeric oligos needed to inhibit 50% complex formation are at least 50-fold higher than that of the telomere oligo (Fig. 7D). Taken together with previous results on the C. tropicalis Cdc13A/A complex, these observations indicate that both the A/A and A/B complex probably have the requisite affinity and sequence specificity to interact with the telomere G-strand in vivo.

DISCUSSION
We have uncovered a second Cdc13 homologue in Candida species and shown that the two paralogues perform overlapping and nonredundant functions in telomere regulation. Unexpectedly, the two paralogues manifest a propensity to form heterodimers, and the heterodimers interact with the cognate telomere repeat with high affinity and sequence specificity. Candida species may thus carry a variant CST complex with distinct composition and properties. The evolutionary and mechanistic implications of these findings are discussed below.
Potential Functions of Cdc13B-The discovery of Cdc13B proteins in Candida spp. provides a possible resolution to the conundrum posed by the small size of Cdc13A proteins in this group of fungi. That is, the functions served by the N-terminal half of Cdc13 in Saccharomyces and Kluyveromyces yeast could conceivably be performed by Cdc13B. These functions include, among others, dimerization, binding to Pol1, and binding to Est1 (5,9,28). Disruption of these interactions in Saccharomyces causes complex and disparate effects on telomere lengths, making it difficult to gauge potential conservation of functions from the existing telomere length data on our Cdc13B mutants. Thus, whether Cdc13B can indeed execute some or all of the functions ascribed to the N-terminal half of large Cdc13 is a question that can only be resolved by detailed investigation of more discrete mutations.
Homo-dimerization and Hetero-dimerization of Cdc13A and Cdc13B through the OB4 Domain-We have shown earlier that the OB4 domain of Candida glabrata Cdc13 and those of C. albicans and C. tropicalis Cdc13A can self-associate to form dimers or other higher order structures (18). In keeping with this observation, we now demonstrate that Cdc13B can also self-associate through its OB4 domain. Perhaps more surprisingly, both Cdc13A and Cdc13B exhibited a greater propensity to form heterodimers than homodimers, suggesting that the heterodimer constitutes the predominant form of Candida Cdc13 in vivo. Because the A/B dimer binds telomere G-strand with high affinity and sequence specificity, it has the requisite biochemical property to localize to telomeres. By contrast, the low DNA binding affinity of the B/B complex suggests that it may not represent a physiologically relevant form of Cdc13. Likewise, even though the A/A complex can certainly bind telomere DNA, several observations suggest that it may be unstable and less capable of functioning at telomeres than the A/B complex. In particular, the A/A-DNA complex appears to disassemble more quickly in gel mobility shift assays, resulting in substantial heterogeneity in the migration of the probe. This can be rationalized by the dissociation of the A/A dimer during electrophoresis and by the inability of the monomers to bind DNA stably (18). Clearly, more studies are necessary to ascertain the physiologically relevant form of Cdc13 in vivo. Moreover, the Cdc13 dimer is presumably just a part of a larger CST complex, and how the unusual A/B dimer in Candida interacts with Stn1 and Ten1 is clearly an interesting question for future investigation.
Evolution of Cdc13 Homologues in Saccharomycotina Yeast-The discovery and characterization of Cdc13B in the current report suggests a plausible model for Cdc13 evolution in the Saccharomycotina subphylum of budding yeast (Fig. 8). In this FIGURE 6. Glycerol gradient sedimentation analysis of the Cdc13A/Cdc13B complexes. A, the complex comprising His 6 -SUMO-Cdc13A and GST-Cdc13B-FLAG was purified by sequential Ni-NTA, M2-agarose, and glutathione-Sepharose affinity chromatography and then subjected to glycerol gradient analysis. The distribution of each protein across the gradient fractions was determined by Western analysis (bottom) and plotted (top). The peak positions for the thyroglubulin (670 kDa), aldolase (158 kDa), and BSA (66 kDa) standards as determined in a parallel gradient are marked by arrows. B, the same as A except that the complex comprising GST-Cdc13A OB4 -FLAG and His 6 -SUMO-Cdc13B was purified by Ni-NTA and M2-agarose and analyzed. Note that the standards in this experiment migrate to slightly different positions from those in A. C, the peak fraction from the gradient shown in B (as marked by an arrow at bottom) was subjected to SDS-PAGE and Coomassie staining. Two different amounts of the fraction were analyzed. Assuming that the staining was proportional to the size of the protein, the ratio of the two proteins is ϳ1:1.
model, a single, 2-OB-fold CDC13 gene in the ancestral budding yeast is presumed to undergo gene duplication followed by fixation to yield two paralogues. In the Candida clade, subfunctionalization of the two paralogues resulted in a pair of telo-mere-capping proteins that share overlapping but nonredundant activities. By contrast, in the common ancestor of Saccharomyces and Kluyveromyces yeast, the two paralogues underwent gene fusion to yield a 4-OB-fold Cdc13, which in turn evolved a specialized function for its individual domains. This scenario is motivated by several key findings in the current study as well as previous observations. First, the existence of two Cdc13-like proteins with common biochemical properties (e.g. binding to each other and Stn1) in the Candida clade argues strongly for duplication of an ancestral gene. Second, the large Cdc13 proteins in Saccharomyces appear to result from the fusion of a pair of duplicated genes; the high resolution structure of ScCdc13 OB1 bears unmistakable resemblances to its DBD, and the OB2 structure is most similar to OB4 of C. glabrata Cdc13 (5, 7). The notion of gene fusion could also account for the absence of Cdc13B in Saccharomyces, assuming that the aforementioned gene duplication transpired in the common precursor of the Saccharomyces and Candida lineages. Although it is possible to imagine alternative evolutionary pathways, our model is relatively parsimonious and invokes just two major events to account for the disparate collection of present day Cdc13 in Saccharomycotina yeast. More importantly, the model makes testable predictions about the structure of Cdc13 orthologues in other related Saccharomycotina yeast. Specifically, if the contention that the ancestral yeast har- were derived from three independent sets of assays. D, gel mobility shift assays were performed using the CaG2 probe (7.5 nM) and the A/B complex (70 nM). Increasing concentrations of unlabeled CaG2 and two non-telomeric oligos were added to the assays as competitors. The fractions of probes bound in the assays were determined and plotted against the competitor/probe ratios. Data (averages Ϯ S.D.) were derived from three independent sets of assays. In the proposed model, a 2-OB-fold Cdc13 was duplicated, and the resulting paralogues were both retained in the descendants. In the Candida clade, the two genes underwent functional specialization (neofunctionalization or subfunctionalization) and became the present day CDC13A and CDC13B. In the Saccharomyces and Kluyveromyces lineages, the duplicated CDC13 became fused to each other, resulting in gene products that are twice the size of the ancestral protein. Again, neofunctionalization or subfunctionalization of individual OB-fold domains is presumed to be responsible for the disparate functions of these domains in the present day CDC13. See "Discussion" for more details.
bors a 2-OB-fold Cdc13 is true, then other early branching species in the subphylum, such as Yarrowia lipolytica, are also more likely to harbor such an orthologue rather than the large 4-OB-fold Cdc13 found in Saccharomyces and Kluyveromyces. Further studies will be necessary to test the validity of this and related hypotheses.
Recurrent Duplication of G-tail-binding Proteins-Fixation of duplicated G-tail-binding proteins (GTBP) appears to be a remarkably common evolutionary event. Previous studies have uncovered recurrent duplication of the POT1 gene in diverse organisms including plants, mammals, and worms (29 -31). Our discovery of CDC13 gene duplication reinforces the impression that the existence of multiple GTBP can confer evolutionary advantages to an organism. What could be the potential benefits? All of the known G-tail-binding proteins are modular polypeptides that interact with a multiplicity of factors to confer telomere protection and regulate both telomere G-and C-strand synthesis. If one surface of the polypeptide is responsible for binding several target proteins, it may be difficult to fine-tune the affinity of interactions to achieve optimal functions. The duplication of a GTBP followed by subfunctionalization may enable each GTBP to achieve optimal binding to its specific interaction partner, thus enhancing organismal fitness. In this regard, it will be interesting to determine whether Candida Cdc13A and Cdc13B use similar surfaces to bind distinct target proteins.