Abstract
A key unresolved issue in molecular evolution is how paralogs diverge after gene duplication. For multifunctional genes, duplication is often followed by subfunctionalization. Subsequently, new or optimized molecular properties may evolve once the protein is no longer constrained to achieve multiple functions. A potential example of this process is the evolution of the yeast heterochromatin protein Sir3, which arose by duplication from the conserved DNA replication protein Orc1. We previously found that Sir3 subfunctionalized after duplication. In this study, we investigated whether Sir3 evolved new or optimized properties after subfunctionalization . This possibility is supported by our observation that nonduplicated Orc1/Sir3 proteins from three species were unable to complement a sir3Δ mutation in Saccharomyces cerevisiae. To identify regions of Sir3 that may have evolved new properties, we created chimeric proteins of ScSir3 and nonduplicated Orc1 from Kluyveromyces lactis. We identified the AAA+ base subdomain of KlOrc1 as insufficient for heterochromatin formation in S. cerevisiae. In Orc1, this subdomain is intimately associated with other ORC subunits, enabling ATP hydrolysis. In Sir3, this subdomain binds Sir4 and perhaps nucleosomes. Our data are inconsistent with the insufficiency of KlOrc1 resulting from its ATPase activity or an inability to bind ScSir4. Thus, once Sir3 was no longer constrained to assemble into the ORC complex, its heterochromatin-forming potential evolved through changes in the AAA+ base subdomain.
GENE duplication is a fundamental mechanism through which genetic diversity increases. A common outcome of gene duplication is subfunctionalization, in which the functions of an ancestral gene are partitioned between the duplicates. Several theoretical models describe subfunctionalization, and differ with respect to whether new or specialized functions emerge after subfunctionalization. Models including Duplication-Degeneration-Complementation (Force et al. 1999) and Constructive Neutral Evolution (Stoltzfus 1999) describe a situation in which no new function arises. In this scenario, the duplicate genes acquire complementary inactivating mutations, such that they each perform a subset of the ancestral functions. Other models, such as Escape from Adaptive Conflict (Hittinger and Carroll 2007) and Gene Sharing (Hughes 1994) describe a situation in which new or enhanced function does arise. In this scenario, the ancestral gene cannot optimize any one of its functions without disrupting the others. However, after duplication and subfunctionalization, the duplicate genes escape this conflict and can acquire optimizing mutations. To gain insight into gene fate after subfunctionalization, we have examined the yeast genes ORC1 and SIR3, a duplicate pair that arose during a whole-genome duplication, and for which there is extensive functional and structural information.
Orc1 is a conserved eukaryotic protein required for DNA replication. It is the largest subunit of the Origin Recognition Complex (ORC), which binds to, and identifies, origins of replication throughout the genome (Li and Stillman 2012). In Saccharomyces cerevisiae, ORC also serves as a silencer-binding protein, which recruits the SIR complex (Silent Information Regulator) to the silenced cryptic mating-type loci, HMRa and HMLα. In particular, the Orc1 subunit binds to Sir1 (Triolo and Sternglanz 1996), which, in turn, recruits other Sir proteins to establish a domain of repressive or “silenced” chromatin. The paralog of Orc1, Sir3, is also involved in silenced chromatin formation, but has a different role. Sir3 binds to deacetylated nucleosomes (Hecht et al. 1995; Onishi et al. 2007), enabling the spreading of the SIR complex along a chromosome (Hoppe et al. 2002; Luo et al. 2002; Rusche et al. 2002). SIR-mediated silencing occurs at the cryptic mating-type loci, whose repression maintains haploid cell identity, and at telomeres, where this chromatin is thought to stabilize the ends of chromosomes.
Orc1 and Sir3 retain homology along their entire lengths and have three major structural domains (Figure 1). The AAA+ domain is a hallmark of the AAA+ superfamily, whose members couple ATP hydrolysis to structural rearrangements of proteins. In this family, ATP binding and hydrolysis requires amino acids from two interacting AAA+ domains. Although both Orc1 and Sir3 contain an AAA+ domain, Sir3 lacks key amino acids required to bind ATP and has no detectable ATPase activity (Bell et al. 1995). Instead, the AAA+-like domain in Sir3 interacts with Sir4 (Chang et al. 2003; King et al. 2006; Ehrentraut et al. 2011), and is also proposed to bind histones (Hecht et al. 1995; Altaf et al. 2007; Ehrentraut et al. 2011). A second domain is a winged helix domain, whose ancestral function is to bind DNA (Gaudier et al. 2007). However, the winged helix domains of both Sir3 and Orc1 in S. cerevisiae lack this activity and instead are involved in homo-dimerization (Oppikofer et al. 2013). The third domain is the N-terminal bromo-adjacent homology (BAH) domain. The BAH domains of both Orc1 and Sir3 bind to nucleosomes (Onishi et al. 2007; Armache et al. 2011; Zhang et al. 2015). The BAH domain of ScOrc1 also binds to Sir1 (Triolo and Sternglanz 1996; Hou et al. 2005). In addition to these three structural domains, there is a rapidly evolving linker that separates the BAH domain from the AAA+ domain.
Structural organization of Orc1/Sir3 proteins. Major structural domains are represented in white for ScSir3 or black for KlOrc1. Each domain is labeled with its molecular function. Amino acid sequence identity (similarity) between ScSir3 and KlOrc1 was calculated for each domain using EMBL-EBI LALIGN Pairwise Sequence alignment (Huang and Miller 1991).
Prior work indicates that Orc1 and Sir3 subfunctionalized after duplication (van Hoof 2005; Hickman and Rusche 2010). In particular, the nonduplicated Orc1 from Kluyveromyces lactis, which serves as a proxy for the ancestral nonduplicated protein, has both replication and silencing functions. KlOrc1 is required for transcriptional repression at heterochromatic loci, and its nucleosome-binding BAH domain is required for spreading across these loci (Hickman and Rusche 2010), as seen for ScSir3. Another indication that the ancestral protein had a silencing function is that the nonduplicated ORC1 gene from Lachancea kluyverii weakly complements a sir3Δ mutation in S. cerevisiae (van Hoof 2005). These results imply that Orc1 was involved in heterochromatin formation before the gene duplication, and, therefore, that subfunctionalization occurred after duplication.
Despite the role of KlOrc1 in forming heterochromatin in its native context, we found that when KlOrc1 is expressed in a S. cerevisiae sir3Δ strain, heterochromatin formation does not occur. This inability of KlOrc1 to complement a sir3Δ mutation suggests that Sir3 acquired new or optimized molecular properties after duplication. If so, Sir3 could be an example of the type of evolutionary change described by the EAC and gene sharing models. To examine this possibility, we identified the sections of KlOrc1 that are insufficient for heterochromatin formation by creating chimeric proteins composed of ScSir3 and KlOrc1. We discovered that the primary insufficiency lies in the AAA+ base subdomain. This portion of Sir3 interacts with Sir4. However, our data are inconsistent with the insufficiency of the KlOrc1 base subdomain resulting from a failure to interact with ScSir4. Moreover, our data do not support the possibility that the presumed ability of KlOrc1 to bind ATP interferes with heterochromatin formation in S. cerevisiae. Therefore, the AAA+ base subdomain may have acquired new or optimized molecular properties after duplication.
Materials and Methods
Yeast strain construction and growth
The yeast strains used in this study (Table 1) were derived from W303-1a. Yeast were grown in YM (0.67% yeast nitrogen base without amino acids, 2% glucose) or CSM-Trp [YM supplemented with a mixture of amino acids and other nutrients but lacking tryptophan (MPBio 4512-522)]. Yeast transformation was performed using lithium acetate and PEG (Schiestl and Gietz 1989). Cells were harvested at an OD600 ∼1, washed twice with TEL (10 mM Tris, pH 7.5; 1 mM EDTA; 100 mM LiOAc), and resuspended in 10 μl TEL/ OD cells. For transformation, 100 μl of cells was added to 0.1 µg plasmid DNA or 0.1 μg of linear DNA plus 30 µg sheared salmon sperm DNA.
To express KlSir4-HA or ScSir4-HA in S. cerevisiae, we first generated tagging cassettes on plasmids, and then used these cassettes to transform strain LRY1098. Plasmid pLR588 contains the ScSIR4 gene with a C-terminal HA tag and downstream HIS3 marker. The original source for the 3xHA tag was pMPY-3XHA (Schneider et al. 1995). This plasmid was altered by site-directed mutagenesis to add a stop codon immediately after the HA tag (…DVPDYAA stop), yielding pLR522. Next, HIS3 from pRS403 (Sikorski and Hieter 1989) was ligated into the SphI and XmaI sites of pLR522 in place of URA3, generating pLR541. The original source for the ScSIR4 gene was pJR2027 (gift from Jasper Rine), based on vector pRS315. The 3x-HA tag plus HIS3 marker from pLR541 was inserted at the 3′ end of the ScSIR4 gene through homologous recombination in yeast, generating pLR588. Finally, plasmid pLR588 was modified by replacing the open reading frame of ScSIR4 with that of KlSIR4 (KLLA0F13420g) to generate pLR1122. The tagging cassettes were isolated from both plasmids and used to transform S. cerevisiae.
Plasmid construction
Plasmids used in this study (Table 2) were derived from pLR911, a yeast CEN/TRP1 plasmid expressing Sir3-V5 from the ScSIR3 promoter. This plasmid was created by replacing the HA tag in pJR2299 (ScSIR3-HA plasmid from Jasper Rine) with a V5 tag amplified from pFA6a–6×GLY–V5–hphMX4 (Funakoshi and Hochstrasser 2009) using homologous recombination. To express Orc1 proteins from nonduplicated species in S. cerevisiae, plasmids were constructed in which the open reading frame of ScSIR3 in pLR911 was replaced with the ORC1 open reading frame from another species. ORC1 from Kluyveromyces lactis was isolated from plasmid pLR824 (Hickman and Rusche 2010), ORC1 from Zygosaccharomyces rouxii was isolated from yeast strain CBS732 (Pribylova and Sychrova 2003) and ORC1 from Lachancea kluyveri was derived from plasmid pLR0649 (van Hoof 2005). To generate chimeric ScSIR3 and KlORC1 genes, portions of KlORC1 were amplified from pLR824 and used to replace the homologous regions of ScSIR3 in plasmid pLR911.
Plasmids were generated either by homologous recombination in yeast or by PCR stitching. For homologous recombination, the parent plasmid, pLR911, was cut with a restriction enzyme within the region to be replaced, and a PCR product was generated containing the desired region of KlORC1 flanked by ∼40 bp of ScSIR3 sequence. A sir3Δ strain (LRY1098) was transformed with both DNA fragments (150 ng of cut plasmid and 300 ng of PCR product), and cells containing recombined plasmids were selected on CSM-Trp medium. Candidate plasmids were recovered from yeast and amplified in Escherichia coli. For PCR stitching, which is a variant of site-directed mutagenesis, a PCR product was created in which the desired region of KlORC1 was flanked by ∼25 bp of ScSIR3 sequence. This PCR product was then used as a primer to amplify and modify pLR911. The cycling parameters were 3 min at 95°, 18 cycles of 40 sec at 95°, 50 sec at 62°, and 27 min at 68°, followed by 20 min at 68°. The newly synthesized plasmid was recovered through E. coli after the template DNA had been digested with DpnI, which cleaves only methylated DNA. All plasmids were confirmed by sequencing.
Mating assay
Semiquantitative mating assays were performed as previously described (Lynch and Rusche 2010) using MATα cells (LRY1098) containing a plasmid expressing the chimeric SIR3/ORC1 genes and a selectable marker TRP1. These cells were grown to midlog phase, resuspended in CSM-Trp at a density of 10 OD/ml, and diluted in a 10-fold series in CSM-Trp. For the plating control, 3 μl of each dilution was spotted on CSM-Trp. For mating, an equal volume of MATa mating partner (LRY1021) was added to each dilution at a constant concentration of 10 OD/ml in YPD (1% yeast extract, 2% Peptone, 2% glucose); 3 μl of that mixture was spotted on YM to select for prototrophic diploids.
Immunoblotting
Immunoblots were performed as previously described (Hickman and Rusche 2010). Cells were grown to an OD600 of 1.0, and fixed with a final concentration of 10% TCA prior to cell lysis; 45 OD equivalents of fixed cells were lysed by vortexing 5 min in the presence of silica beads (0.5 mm dia. #11079105z, BioSpec Products) in 40 μl lysis buffer [10 mM HEPES pH 7.9, 150 mM KCl, 1.5 mM MgCl2, 0.5 mM DTT, 10% glycerol, 5 µg/ml chymostatin, 2 µg/ml pepstatin A, 156 µg/ml benzamidine, 35.2 µg/ml TPCK, 174 µg/ml PMSF, and 1× Complete Protease Inhibitor (Roche)]. Proteins were denatured by the addition of 1/3 volume sample buffer (30% glycerol, 15% β-mercaptoethanol, 6% SDS, 200 mM Tris pH 6.8, 0.08 mg/ml bromophenol blue) and incubation at 95° for 5 min. Finally, samples were clarified by centrifugation. Aliquots (10 µl) of protein extract were resolved on a 7.5% acrylamide gel, transferred to membrane (Amersham 45004008), and probed to detect Sir3/Orc1 chimeric proteins (anti-V5, ab3792; Millipore) or Pgk1 (ab113687; Abcam) as a loading control.
RNA isolation and cDNA synthesis
RNA was extracted in hot phenol (Schmitt et al. 1990) from cells grown to an OD600 of 1.0 in CSM-Trp. For cDNA synthesis, genomic DNA was removed from RNA samples in a total volume of 30 μl containing 3 μg of RNA, DNAse buffer, and 3 units Ambion Turbo DNase (AM2238) at 37° for 30 min. The samples were then extracted with phenol/chloroform and precipitated with ethanol. cDNA was generated using the iScript Advanced cDNA Synthesis Kit (170-8843; Bio-Rad). A 20 μl reaction containing 10 μl of DNase treated RNA, 1× iScript Advanced Mix, and 1 μl iScript reverse transcriptase was incubated for 30 min at 42° and then inactivated for 5 min at 85°.
cDNA was quantified using a Bio-Rad CFX384 real-time PCR machine. The silenced genes (HMRa1 or YFR057W) and a control gene (NTG1) were quantified relative to a standard curve prepared from genomic DNA (LRY1098 containing pLR0911). Primers are listed in Table 3. The ratio of the silenced gene to the control gene was determined for each sample. Two technical replicates (independent RNA isolations) were prepared from each of two biological replicates (independent transformants). The ratios for all four replicate samples were averaged, and this value was normalized to the average ratio in cells expressing ScSir3-V5.
Chromatin immunoprecipitation
Chromatin immunoprecipitation was performed as previously described (Rusche and Rine 2001). Cells were grown in CSM-Trp, harvested at an OD600 of ∼1, and crosslinked for 30 min in 1% formaldehyde. For immunoprecipitation, 2 µl of anti V5 antibody (Millipore ab3792) was used. Two technical replicates (independent immunoprecipitations) were prepared from each of two biological replicates (independent transformants). The immunoprecipitated DNA was analyzed by real-time PCR to quantify the recovery of silenced loci (HMR-E and TelVI-R) and a control locus (ATS1) relative to a standard curve prepared from input DNA. The ratio of the experimental locus to the control locus is represented as relative enrichment. Primers are listed in Table 3.
Yeast immunofluorescence
Immunofluorescence was performed as previously described (Keeling and Miller 2011). The yeast cell wall was digested with Zymolyase (10 mg/ml in 1 M sorbitol/PBS) for 45 min at room temperature. Cells were then blocked with 5 mg/ml BSA in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7.4) for 1 hr, incubated overnight in primary antibody (anti-V5, ab3792; Millipore) diluted 1:100 in PBS, and then incubated 2 hr at room temperature in secondary antibody (111-165-144, Cy3-conjugated AffiniPure Goat Anti-Rabbit IgG; Jackson ImmunoResearch). Finally, the cells were stained with DAPI (1 µg/ml in PBS) for 15 min and visualized as previously described (Chavel et al. 2014) on a Zeiss PLAN-APOCHROMAT 100× objective using the DIC, rhodamine and DAPI channels. Images were captured for multiple fields of view, and the subcellular localization of the Cy3 signal was scored for 150 cells of each genotype.
Data availability
All strains and plasmids are available upon request. All data necessary for evaluating our conclusions are represented within this paper.
Results
Nonduplicated Orc1 proteins did not complement a sir3∆ mutation in S. cerevisiae
Given that Orc1 and Sir3 subfunctionalized (Hickman and Rusche 2010), nonduplicated Orc1 proteins might be expected to complement a sir3∆ mutation in S. cerevisiae. To test this hypothesis, we expressed three nonduplicated Orc1 proteins from the ScSIR3 promoter in an S. cerevisiae sir3∆ strain. To assess silenced chromatin formation, we examined whether the cryptic mating-type locus HMRa was repressed using a semi quantitative mating assay. These MATα cells will only mate if HMRa is silenced. The haploid MATα cells were plated with a MATa tester strain, and prototrophic diploids were selected on minimal medium (Figure 2A). Controls indicated that yeast cells lacking Sir3 did not mate (row 1), whereas cells expressing ScSir3 did mate (rows 2 and 3). In contrast, cells expressing nonduplicated Orc1 proteins from Zygosaccharomyces rouxii, K. lactis, or L. kluyveri did not mate (rows 4–6). The nonduplicated Orc1 proteins were expressed at levels comparable to ScSir3 (Figure 2B), indicating that the failure to complement was not due to lack of expression. This inability of the nonduplicated Orc1 proteins to restore silenced chromatin formation in the absence of ScSir3 is consistent with the hypothesis that Sir3 evolved a new property after duplication.
Nonduplicated Orc1 proteins did not complement a sir3Δ mutation in S. cerevisiae. (A) Mating was assessed for sir3Δ MATα cells (LRY1098) carrying plasmids encoding ScSir3 (pJR2299, pLR911), Z. rouxii Orc1 (pLR1058), K. lactis Orc1 (pLR915), L. kluyverii Orc1 (pLR1056) or an empty vector (pRS314). Cells were diluted 10-fold and spotted on CSM lacking tryptophan (CSM-Trp) as a plating control (right). The same diluted cells were combined with a constant amount of MATa tester cells (LRY1021) and spotted on minimal medium (YM) to detect the prototrophic diploids. The evolutionary relationship of the species (Byrne and Wolfe 2005) is indicated by a cladogram on the left. (B) Expression of Orc1/Sir3 proteins in the cells used in (A) was examined by immunoblotting. Orc1/Sir3 proteins were detected with antibodies against V5. Pgk1 (3-phosphoglycerate kinase) served as a loading control.
These results differ from a previous report that ORC1 from L. kluyveri does complement a sir3Δ mutation in S. cerevisiae (van Hoof 2005). However, the reported complementation is extremely weak, and we have been unable to reproduce this result using the original strains and plasmids (data not shown).
The KlOrc1 AAA+ domain was insufficient for silencing in S. cerevisiae
To delineate the region of KlOrc1 that is insufficient in S. cerevisiae, and thus may have evolved new molecular properties, we created chimeric proteins of ScSir3 and KlOrc1. These chimeric proteins were expressed from a ScSIR3 promoter and had a C-terminal V5 tag. As a starting point, we focused on the four structural divisions of the protein (Figure 1A and Figure 3A) and replaced the BAH, linker, AAA+, and winged helix domains of ScSir3 with the homologous portions of KlOrc1. All these chimeric proteins were expressed at levels comparable to ScSir3 (Figure 3B).
The KlOrc1 AAA+ domain was insufficient for silencing in S. cerevisiae. (A) The chimeric proteins studied are represented by graphics, in which white indicates ScSir3 sequences and black indicates KlOrc1 sequences. (B) Expression of the chimeric proteins described in (A) was assessed by immunoblotting, as for Figure 2. (C) Mating assays were conducted as for Figure 2. S. cerevisiae cells expressed the indicated ScSir3-KlOrc1 chimeric proteins. (D) Levels of a1 mRNA were measured by quantitative RT-PCR for cells expressing the indicated chimeric proteins. For each strain, a1 mRNA was quantified relative to a control locus, NTG1, and then mRNA levels were normalized to cells expressing ScSir3-V5. (E) Levels of YFR057W mRNA were measured as for (D). (F) The association of chimeric proteins with the HMR-E silencer was determined by ChIP. The x-axis represents the relative enrichment of the chimeric protein at the silencer compared to a control locus, ATS1. (G) The association of chimeric proteins with telomere VI-R was determined by ChIP as for (F).
We first tested whether these chimeric proteins were able to silence HMRa using the mating assay (Figure 3C). Cells expressing chimeric proteins containing the BAH, linker, or winged helix domains of KlOrc1 were able to mate as efficiently as full-length ScSir3 (Figure 3C, constructs 5, 6, 8). In contrast, cells expressing a chimeric protein containing the KlOrc1 AAA+ domain did not mate (Figure 3C, construct 7). To directly assess the ability of these chimeric proteins to repress transcription, we measured the levels of a1 mRNA from HMRa as well as YFR057W at telomere VI-R, which is a read-out for telomeric silencing (Figure 3, D and E). In keeping with the mating assay, a1 and YFR057W were no longer repressed in the presence of KlOrc1 or the chimeric protein containing the KlOrc1 AAA+ domain. These results indicate that the primary insufficiency of KlOrc1 is in the AAA+ domain.
To determine whether these chimeric proteins were deficient in recruitment to silenced regions, we preformed chromatin immunoprecipitation (ChIP) (Figure 3, F and G). For this assay, we used an antibody against the V5 tag and examined relative enrichment compared to a control locus, ATS1, which is not associated with Sir proteins. We found that the chimeric proteins containing the KlOrc1 BAH or winged helix domains had a modest decrease in association with HMR-E and telomere VI-R. In contrast, full-length KlOrc1 and the chimeric protein containing the K. lactis AAA+ domain did not localize to HMR-E or telomere VI-R. Thus, the insufficiency in the AAA+ domain is due to a deficiency in recruitment rather than a later step in chromatin assembly or transcriptional repression.
One potential explanation for the lack of function of KlOrc1 and the chimeric protein containing the KlOrc1 AAA+ domain is that they were not properly localized to the nucleus. Therefore, we examined the subcellular localization of these proteins using an immunofluorescence assay (Figure 4). From a population of 150 cells, we determined the numbers of cells with nuclear, cytoplasmic, or nuclear and cytoplasmic localization of the V5 tag. In addition, some cells lacked any signal, presumably because the antibodies failed to enter the cells. All the V5-tagged proteins were primarily nuclear and had similar fractions of cells with each localization pattern. This included the functional ScSir3-V5 and chimeric protein containing KlOrc1 BAH and the nonfunctional KlOrc1-V5 and chimeric protein containing the KlOrc1 AAA+ domain. Thus, improper localization did not account for the lack of function of KlOrc1 or the chimeric protein containing KlOrc1 AAA+.
ScSir3, KlOrc1, and chimeric proteins were predominantly nuclear. (A) The proteins examined are represented by graphics, in which white indicates ScSir3 sequences and black indicates KlOrc1 sequences. (B) Representative images of Cy3-labeled V5-tagged proteins. The first column shows the DIC images merged with both fluorescence channels. The next two columns show Cy3 and DAPI staining, and the last column shows the merged Cy3 and DAPI channels. (C) The subcellular localization of the Cy3 signal was scored for 150 cells. Localization was nuclear (purple), nuclear and cytoplasmic (gray), or cytoplasmic (green). In addition, some cells had no fluorescence (yellow).
Disruption of Walker A box in KlOrc1 did not restore silencing
The AAA+ domain of KlOrc1 is presumed to have ATPase activity, based on the presence of key conserved amino acids and the critical role for this activity in DNA replication. However, the Sir3 AAA+ domain has no detectable ATPase activity. Therefore, one potential explanation for the inability of the KlOrc1 AAA+ domain to function in silenced chromatin formation in S. cerevisiae is that its presumed ATPase activity interferes with the assembly of silenced chromatin. If this were the case, disrupting this activity might enable the protein to generate heterochromatin in S. cerevisiae. Therefore, we mutated the Walker A box, which is a canonical motif required for ATP binding. In particular, we placed the nonfunctional Walker A box of ScSir3 into full-length KlOrc1 or the chimeric protein containing the KlOrc1 AAA+ domain (Figure 5A). However, this substitution did not rescue the ability of KlOrc1 to achieve mating in S. cerevisiae (Figure 5). Therefore, there are likely other regions of the KlOrc1 AAA+ domain that are insufficient for silencing, and it is not simply the presumed ATPase activity of this domain that hinders function in S. cerevisiae.
Disruption of Walker A box in KlOrc1 did not restore silencing. (A) An alignment of Walker A boxes from K. lactis Orc1 and S. cerevisiae Sir3 and Orc1. The bar indicates the amino acids that were replaced in KlOrc1 (pLR1103) or a chimeric protein with the KlOrc1 AAA+ domain (pLR1119). (B) Expression of the proteins with mutated Walker A domains was assessed by immunoblotting. (C) Mating assays were conducted for S. cerevisiae cells expressing the indicated chimeric proteins.
The base subdomain of the KlOrc1 AAA+ domain was insufficient for silencing in S. cerevisiae
To narrow down the region of the AAA+ domain of KlOrc1 that was insufficient in S. cerevisiae, we created additional chimeric proteins. First, the major structural units of the AAA+ domain were replaced with the homologous portions of KlOrc1, namely the pre-AAA+, base, and lid subdomains (Figure 6A). These chimeric proteins were all expressed (Figure 6B). Cells expressing a chimeric protein containing the KlOrc1 AAA+ base subdomain were unable to mate (Figure 6C, construct 12) or repress transcription (Figure 6, D and E). In contrast, the chimeric proteins containing KlOrc1 pre-AAA+ or lid subdomains promoted mating and transcriptional silencing (constructs 11, 13). Moreover, the chimeric protein containing the KlOrc1 AAA+ base subdomain was not associated with silenced loci (Figure 6, F and G). Nevertheless, although it did not function, the chimeric protein containing the KlOrc1 AAA+ base subdomain was properly localized to the nucleus (Figure 4).
The base subdomain of the KlOrc1 AAA+ domain was insufficient for silencing in S. cerevisiae. (A) The structure of the AAA+ domain (PDB-3te6) is colored to show the three subdomains, with a portion of the pre-AAA+ subdomain in green, the base subdomain in red, and the lid subdomain in blue. Only a small fragment of the pre-AAA+ domain was modeled in the crystal structure. The Sir4-interaction loop is colored violet. The graphical representation of chimeric proteins is indicated. (B) Expression of chimeric proteins was assessed by immunoblotting. (C) Mating assays were conducted using cells expressing the indicated chimeric proteins. mRNA levels were measured for a1 (D) or YFR057W (E) using quantitative RT-PCR for cells expressing the indicated proteins. The association of chimeric proteins with the HMR-E silencer (F) or telomere VI-R (G) was determined by ChIP.
We noted that the KlOrc1 pre-AAA+ subdomain also disrupted function at Telomere VI-R but not HMR (construct 11, compare Figure 6, E and G with Figure 6, D and F). The pre-AAA+ subdomain of ScSir3 interacts with Rap1 (Moretti and Shore 2001), and this interaction is especially critical for recruitment to telomeres. Perhaps the pre-AAA+ subdomain of KlOrc1 has lower affinity for ScRap1. Nevertheless, the subdomain does function at HMR-E.
To further delineate the critical region within the base subdomain, we made additional chimeric proteins. The smallest portion of KlOrc1 that was nonfunctional in S. cerevisiae was the base subdomain without its first two α-helices (construct 14). This chimeric protein was unable to silence at HMRa or telomere VI-R and did not associate with these loci (Figure 6). Substitutions of individual helices within the base subdomain did not perturb function (data not shown). Thus, the critical deficiency in KlOrc1 lies within the AAA+ base subdomain.
Providing KlSir4, the natural binding partner of KlOrc1, did not restore silencing
One known property of the AAA+ base domain in ScSir3 is to bind ScSir4 (Ehrentraut et al. 2011). If this interaction surface has evolved over time, species–specific contacts between Sir3/Orc1 and Sir4 might prevent KlOrc1 from associating with ScSir4. As a first test of this possibility, we generated a chimeric protein in which the Sir4-interacting loop of ScSir3 (Ehrentraut et al. 2011) was replaced with the homologous loop from KlOrc1. If this loop from KlOrc1 does not interact with ScSir4, this chimeric protein should be nonfunctional. However, it functioned equivalently to full-length ScSir3 (Figure 6, construct 15).
As a further test of the possibility that KlOrc1 cannot interact with ScSir4, we determined whether the ability of KlOrc1 to form heterochromatin could be restored by expressing its natural partner, Sir4 from K. lactis. To do so, we replaced the genomic copy of ScSIR4 with KlSIR4. Importantly, cells expressing ScSir3 and KlSir4 were able to mate at a low level [Figure 7B, row 3; see also Astrom and Rine (1998)], indicating that KlSir4 does function as part of the S. cerevisiae SIR complex. Nevertheless, KlOrc1 remained unable to restore mating in cells expressing KlSir4 (Figure 7B, row 4). This lack of function was also observed for chimeric proteins containing portions of the KlOrc1 AAA+ domain (rows 5–7). Furthermore, these chimeric proteins were not recruited to HMRa or to telomere VI-R (Figure 7, D and E). Therefore, the insufficiency of the KlOrc1 AAA+ base subdomain is most likely not due to an inability to bind Sir4.
Coexpression of KlSir4 and KlOrc1 did not restore silencing in S. cerevisiae. (A) The graphical representation of chimeric proteins. (B) Mating assays were conducted using cells expressing the indicated chimeric proteins, as well as KlSir4-HA or ScSir4-HA. (C) Expression of chimeric proteins was assessed by immunoblotting. The association of chimeric proteins with the HMR-E silencer (D) or telomere VI-R (E) was determined by ChIP.
The KlOrc1 bridging helix destabilized ScSir3
In addition to the AAA+ domain, a second region of KlOrc1 that disrupted function in ScSir3 was the bridging helix between the AAA+ and winged helix domains. We originally found that a chimeric protein containing the KlOrc1 winged helix domain plus the bridging helix was not functional (Figure 8, construct 17). To localize the portion of this domain that interfered with silencing, we created chimeric proteins containing just the winged helix domain or the bridging helix. A chimeric protein containing the KlOrc1 winged helix domain was expressed and restored mating similarly to ScSir3 (Figure 8, construct 8). In contrast, the chimeric protein containing just the KlOrc1 bridging helix was unable to restore silencing (Figure 8, construct 18). Moreover, this chimeric protein was not expressed (Figure 8B). This lack of expression was not due to a sequence error in the plasmid, and occurred for three independent transformants. In fact, the mRNA for this chimeric protein was present in these transformants (data not shown). Therefore, we considered the possibility that the KlOrc1 bridging helix destabilizes the ScSir3 protein.
The K. lactis bridging helix destabilized ScSir3. (A) The AAA+ (PDB-3te6) and Winged Helix (PDB-3zco) domains are colored to show the bridging helix (black), AAA+ lid subdomain (blue), and winged helix domain (gray). Portions of the bridging helix are modeled in each of the two structures, although some amino acids are not in either structure. The graphical representation of each chimeric protein is indicated to the right. (B) Expression of chimeric proteins was assessed by immunoblotting. (C) Mating assays were conducted using S. cerevisiae cells expressing the indicated ScSir3-KlOrc1 chimeric proteins. mRNA levels were measured for a1 (D) or YFR057W (E) using quantitative RT-PCR. The association of chimeric proteins with the HMR-E silencer (F) or telomere VI-R (G) was determined by ChIP.
This potential instability of the chimeric protein could occur because the KlOrc1 bridging helix does not pack well against the ScSir3 protein. If so, stability might be maintained if the KlOrc1 bridging helix were placed into ScSir3 in a manner that retains appropriate packing interactions. To test this idea, we created two additional chimeric proteins. One chimeric protein contains the last quarter of the KlOrc1 bridging helix and the KlOrc1 winged helix domain against which it packs (Oppikofer et al. 2013), and the second chimeric protein contains the other three-quarters of the bridging helix and the AAA+ lid domain against which it packs (Ehrentraut et al. 2011). These two chimeric proteins were expressed (Figure 8B, constructs 19 and 20), indicating that protein stability was restored. Moreover, these chimeric proteins functioned similarly to ScSir3 (Figure 8, C–G). Given that neither portion of the bridging helix interferes with function when it occurs in the presence of its proper packing partner, we conclude that the lack of function of the chimeric protein containing the full KlOrc1 bridging helix results from poor packing against ScSir3. Thus, the KlOrc1 bridging helix does not lack a property present in ScSir3, and there is no evidence that a new molecular property evolved in either the bridging helix or the winged helix domain.
Discussion
In this study, we identified a discrete portion of the nonduplicated KlOrc1 protein as being insufficient for heterochromatin formation in the duplicated species S. cerevisiae. Specifically, the insufficiency lies in the AAA+ base subdomain minus its first two alpha helices. We hypothesize that this portion of Sir3 acquired new molecular properties after duplication.
The AAA+ base subdomains of duplicated Orc1 and Sir3 have distinct functions, consistent with this region evolving after duplication. The AAA+ base subdomain of Orc1 interlocks with other AAA+ domains in ORC (Chen et al. 2008; Bleichert et al. 2015), and these interactions promote ATP binding and hydrolysis. In contrast, the AAA+ subdomain of Sir3 contacts the heterochromatin protein Sir4 and is also proposed to bind nucleosomes (Ehrentraut et al. 2011). We considered whether the inability of KlOrc1 to function in the Sir complex is due to its potential ATPase activity. However, disrupting the Walker A ATP-binding domain of KlOrc1 did not recover the ability to function in the silencing complex (Figure 5). Therefore, the potential ATPase activity in itself is not the reason that KlOrc1 fails to generate heterochromatin in S. cerevisiae.
In ScSir3, the AAA+ base subdomain does not bind or hydrolyze ATP, but instead binds ScSir4. A similar interaction is thought to occur in K. lactis, with KlOrc1 binding KlSir4. It was therefore possible that KlOrc1 fails to function in S. cerevisiae because it cannot bind ScSir4. Indeed, Sir4 is one of the most rapidly evolving proteins within the Saccharomyces clade (Zill et al. 2010), and thus could theoretically form species-specific interactions with Sir3/Orc1. However, given that KlSir4 complements a Scsir4Δ mutation (Astrom and Rine 1998), it must interact with S. cerevisiae Sir proteins, and in much the same way that ScSir4 does. Moreover, expression of KlSir4 in S. cerevisiae did not restore silencing in the presence of KlOrc1 (Figure 7). Therefore, the insufficiency within the AAA+ base subdomain is most likely not due to its inability to bind ScSir4.
The other proposed function for the AAA+ base subdomain and surrounding regions is to bind nucleosomes. Although the best characterized nucleosome-binding domain of Sir3 is the BAH domain (Onishi et al. 2007; Armache et al. 2011), the existence of a lower affinity nucleosome-binding region in the AAA+ domain is supported by biochemical studies (Hecht et al. 1995; Altaf et al. 2007; Ehrentraut et al. 2011). It is possible that this secondary nucleosome-binding region does not exist in nonduplicated Orc1 proteins, such as KlOrc1, and only evolved after the AAA+ base domain was no longer constrained to assemble into ORC. Further studies will be required to test this possibility.
This study suggests that Sir3 and Orc1 followed the EAC or Gene Sharing model of evolution after duplication (Hughes 1994; Hittinger and Carroll 2007). Even though the nonduplicated KlOrc1 contributes to heterochromatin formation in its native context (Hickman and Rusche 2010), it was unable to complement a sir3Δ mutation in S. cerevisiae (Figure 2). Moreover, nonduplicated Orc1 proteins from two other species also failed to function in S. cerevisiae, suggesting that Sir3 gained new molecular properties after subfunctionalization. A potential “conflict” experienced by the nonduplicated Orc1 might be that because the AAA+ domain must pack into the ORC it cannot evolve surfaces that pack optimally with the Sir chromatin. Perhaps due to this conflict, heterochromatin formation in K. lactis requires additional factors not needed in S. cerevisiae. In particular, the Sum1 DNA-binding protein is required for silencing at cryptic mating-type loci in K. lactis (Hickman and Rusche 2009), but is not needed in S. cerevisiae. The further evolution of Sir3 after duplication may have eliminated the need for Sum1. Thus, functional differences in the proteomes of S. cerevisiae and K. lactis may offset functional differences in ScSir3 and KlOrc1.
We conclude that Sir3 specialized within the AAA+ base subdomain after subfunctionalization. Our data are not consistent with the scenarios that the ATPase of Orc1 interferes with silencing, or that KlOrc1 fails to bind ScSir4. Therefore, we favor the possibility that this region gained the ability to bind nucleosomes.
Acknowledgments
We thank Ambro van Hoof and Hana Sychrová for strains and plasmids. We also thank Gerald Koudelka for help analyzing the structural modules of ScSir3, and Jacky Chow for help with the fluorescence microscopy. This research was supported by startup funds provided to L.N.R. by the College of Arts and Sciences at the State University of New York at Buffalo and National Science Foundation (NSF) grant MCB-1615367.
Footnotes
Communicating editor: O. Rando
- Received January 23, 2017.
- Accepted August 9, 2017.
- Copyright © 2017 by the Genetics Society of America