Abstract
By testing the susceptibility to DNA damaging agents of several Candida albicans mutant strains derived from the commonly used laboratory strain, CAI4, we uncovered sensitivity to methyl methanesulfonate (MMS) in CAI4 and its derivatives, but not in CAF2-1. This sensitivity is not a result of URA3 disruption because the phenotype was not restored after URA3 reintroduction. Rather, we found that homozygosis of a short region of chromosome 3R (Chr3R), which is naturally heterozygous in the MMS-resistant-related strains CAF4-2 and CAF2-1, confers MMS sensitivity and modulates growth polarization in response to MMS. Furthermore, induction of homozygosity in this region in CAF2-1 or CAF4-2 resulted in MMS sensitivity. We identified 11 genes by SNP/comparative genomic hybridization containing only the a alleles in all the MMS-sensitive strains. Four candidate genes, SNF5, POL1, orf19.5854.1, and MBP1, were analyzed by generating hemizygous configurations in CAF2-1 and CAF4-2 for each allele of all four genes. Only hemizygous MBP1a/mbp1b::SAT1-FLIP strains became MMS sensitive, indicating that MBP1a in the homo- or hemizygosis state was sufficient to account for the MMS-sensitive phenotype. In yeast, Mbp1 regulates G1/S genes involved in DNA repair. A second region of homozygosis on Chr2L increased MMS sensitivity in CAI4 (Chr3R homozygous) but not CAF4-2 (Chr3R heterozygous). This is the first example of sign epistasis in C. albicans.
THE opportunistic fungal pathogen Candida albicans is often isolated as a highly heterozygous diploid; the genome of the reference strain SC5314 has 67,500 single nucleotide polymorphisms (SNPs) (Jones et al. 2004; Braun et al. 2005; Muzzey et al. 2013). SNPs found within regulatory regions can affect transcription levels between the alleles (Staib et al. 2002). Even synonymous SNPs residing in open reading frames (ORFS) can result in differences in the rate and efficiency of messenger RNA (mRNA) translation since poorly used codons delay protein synthesis. These delays may lead to misfolding of the nascent protein or formation of mRNA secondary structures (reviewed in Larriba and Calderone 2008). Recent genome-wide analysis of cis elements on gene expression showed that allele-specific effects are often due to mRNA levels and/or translation efficiency (Muzzey et al. 2014).
While the majority of SNPs reside in intergenic regions, more than half of open reading frames contain one or more SNPs. The vast majority (78%) of these are nonsynonymous, implying that a significant fraction of ORFs encode proteins that differ in one or more amino acids (Jones et al. 2004) that may affect crucial properties. Nonconservative amino acid substitutions within an enzyme’s catalytic domain could result in an inactive allele (Gómez-Raja et al. 2008). However, many SNPs will cause only minor or insignificant alterations in protein properties.
Mitotic recombination or less frequently, chromosome truncation or loss, will reveal deleterious allele that were masked by a functional allele in heterozygous diploid organisms like C. albicans. These loss-of-heterozygosity (LOH) events can occur over long ranges of the chromosome (LR-LOH) to homozygose many genes and that could cause new genetic interactions that yield unexpected phenotypes (Weinreich et al. 2005). LOH in regulatory regions can alter gene expression responses to some environmental conditions (Staib et al. 2002).
In animal cells, LR-LOH causes genetic instability, and along with aneuploidy, is associated with human disease and observed in >90% of solid tumors. C. albicans exploits these natural occurrences to generate new phenotypes, including but not limited to auxotrophy (Gómez-Raja et al. 2008), mating proficiency (Magee and Magee 2000), and antifungal drug resistance (Selmecki et al. 2006, 2008; Niimi et al. 2010; Sasse et al. 2012).
In C. albicans, the frequency of spontaneous LOH is 10−4–10−6 (Forche et al. 2009b; Lephart and Magee 2006) and increases significantly under stress conditions in vitro (Forche et al. 2011), in vivo (Forche et al. 2009b), or during molecular manipulations such as construction of modified laboratory strains (Selmecki et al. 2006; Arbour et al. 2009; Bouchonville et al. 2009; Abbey et al. 2011). Importantly, short- or long-range LOH events yield new genotypes and, potentially, new phenotypes that are heritable. Thus, molecular manipulation of strains carries with it the risk of introducing unidentified mutations (i.e., LOH).
CAI4 is a commonly used ura3∆∆ derivative of C. albicans-type strain SC5314 (Fonzi and Irwin 1993) that was derived directly from URA3/ura3∆ strain CAF2-1. Importantly, when genes involved the nonhomologous end-joining (NHEJ) pathway of DNA repair are deleted in CAI4, we detected two new phenotypes: MMS- and temperature sensitivity. Here, we show that both phenotypes result from genetic alterations unrelated to deletion of NHEJ genes. MMS sensitivity results initially from an LOH on the right arm of chromosome 3 (Chr3R) that occurred during the construction of CAI4. This region contains the MBP1 ORF and when one of the alleles (allele a) is present either in a homozygous or hemizygous state it is sufficient to confer MMS sensitivity. Furthermore, we identified an additional LOH event on Chr2L in some derivatives of CAI4. While the LOH of Chr2L is neutral by itself, it enhances MMS sensitivity when found together with the Chr3R homozygosity and suggests sign epistasis. Our results are consistent with previous findings that genetically manipulated laboratory strains of C. albicans contain additional, nontarget genetic alterations that result in unexpected phenotypes. Furthermore, some of these alterations are the underpinnings for genome and protein diversification, and different selection pressures, depending on the environment, will determine the expansion of specific variants (Ford et al. 2015).
Materials and Methods
Strains and growth conditions
C. albicans strains used in this study are listed in Supplemental Material, Table S1.
Cells were routinely grown in solid or liquid YPD (2% glucose, 1% yeast extract, 2% bactopeptone, 25 µg/ml uridine) unless otherwise specified.
C. albicans transformation and selection of transformants
To generate gene disruptions, parental strains were transformed with a SAT1-flipper cassette by electroporation in a MicroPulser Electroporator system (Bio-Rad, Hercules, CA) (Reuss et al. 2004). Nourseothricin-resistant (NouR) colonies were selected on YPD plates supplemented with 200 µg/ml of nourseothricin. SAT1 loss was induced by overnight growth in liquid YPM (2% maltose, 1% yeast extract, 2% bactopeptone) and nourseothricin-sensitive (NouS) derivatives were selected on YPD plates containing 20 µg/ml of nourseothricin.
In order to look at the consequences of long-range LOH, URA3 was integrated into Uri− strains to generate URA3 heterozygotes (Wilson et al. 1999). The URA3 cassettes were PCR amplified from pGEM–URA3 using oligonucleotides complementary to the flanking regions of the selected integration positions on Chr2L or Chr3R (Table S2), and Uri+ transformants were selected on SC plates lacking uridine (0.7% yeast nitrogen base, 2% glucose). To isolate derivatives that had undergone LOH, Uri+ strains were grown in liquid YPD for 16 hr and subsequently 100 µl of the culture was spread onto SC plates containing 0.1% 5-FOA and 25 µg/ml uridine to select for 5-FOAR colonies. Reintegration of IRO1–URA3 in CAI4 and CAI4-L was as described previously (Noble and Johnson 2005). MBP1 integration was generated through transformation of TCS570 with the ApaI–BglI fragment of pTC47 (Figure 7B) and NouR colonies were selected.
Construction of disruption cassettes and gene cloning
Cassettes for disruption of SNF5, POL1, orf19.5854.1, or MBP1:
Upstream and downstream regions of each ORF were PCR amplified from the genomic DNA of CAF2-1 with oligonucleotides listed in Table S2, and each ApaI–XhoI (upstream) and SacII–SacI (downstream) fragment subsequently cloned in the pSFS2A plasmid (provided by J. Morschhäuser) flanking the SAT1-flipper cassette. Each disruption cassette was release by digestion with ApaI and SacI before transformation.
Cloning and integration of MBP1b into its own locus:
The full MBP1 ORF flanked by 276 bp and 243 bp of its upstream and downstream regions, respectively, was PCR amplified from the genomic DNA of TCS1070 (mbp1a∆::SAT1-FLIP/MBP1b) using primers MBP1–ApaI and MBP1–SacI. The resulting blunt-ended fragment was cloned into plasmid pNIM1 (from which CaGFP had been previously released by incubation with SalI and BglII) (Park and Morschhäuser 2005) to yield plasmid pTC46. The 243-bp region downstream of MBP1 was newly amplified from genomic DNA with primers MBP1–SacII and MBP1–SacI and cloned by blunt-end ligation in the MluI position of pTC46 to generate pTC47. The MBP1b integration cassette was released from pTC47 with ApaI and BglI.
MMS and temperature sensitivity assays
MMS and temperature sensitivity was determined using a drop test assay. For MMS, 7-μl aliquots of fivefold serial dilutions from exponential cultures (OD (optical density) ≅ 1) were spotted on YPD plates containing 0.02% or 0.03% (v/v) MMS and incubated for 40 hr at 28° before being photographed. Screenings for the MMS sensitivity of 5-FOAR segregants were routinely carried out by spotting 7-µl aliquots of 25-fold dilutions from exponential cultures (OD ≅ 1) on YPD plates containing 0.02% MMS. YPD plates lacking MMS were used as control. For temperature sensitivity, aliquots were spotted on YPD plates and incubated at 43° and 28° for 40 hr.
Genotyping of polymorphic loci
Genotyping of the indicated polymorphic loci was carried out by either direct sequencing (locus SNF5 and SNP71) or SNP-RFLP analysis of PCR-amplified DNA. For direct sequencing, DNA was PCR amplified for 30 cycles using the Expand High FidelityPLUS PCR System (Roche) following the manufacturer’s instructions for higher accuracy. PCR product was purified using the QIAquick PCR Purification Kit (Qiagen, Valencia, CA) and directly sequenced at DNA sequencing facilities at Universidad de Extremadura. Sequences were assembled, edited, and compared to sequences reported on the Candida Genome Database (CGD) using Lasergene software (DNASTAR). Locus SNF5 was PCR amplified from both CAF2-1 and CAI4 genomic DNAs using a set of primers (Table S2). Most primers used for SNP amplification are described in Forche et al. 2009. Heterozygosity in the locus POL1 was detected by PCR amplification of a 360-bp internal fragment using primers 2021-F and 2380-R (sequences listed in Table S2) followed by digestion with endonuclease HincII (allele a cut, allele b no cut).
SNP/CGH arrays
Genomic DNA was prepared from overnight cultures and labeled with Cy3 or Cy5 thymidine, using Klenow polymerase and added to a solution containing Agilent 10× blocking agent and 22.5 ml of Agilent 2× hybridization buffer as described in Abbey et al. (2011). Each mixture of Cy3- and Cy5-labeled DNA was treated according to Agilent standard protocols with incubation for 24 hr at 65°. After incubation, arrays were washed using wash buffers and acetonitrile solutions per Agilent standard protocols. Microarrays were scanned as 16-bit TIFF images and analyzed using BlueFuse for Microarrays (BlueGnome, Cambridge, UK). Data analysis and visualization were performed as previously described (Abbey et al. 2011).
Data availability
The authors state that all data necessary for confirming the conclusions presented in the article are represented fully within the article.
Results
CAI4 and its derivatives are MMS and temperature sensitive relative to CAF2-1
Proper expression of C. albicans URA3 (orf19.1716 ) is important for virulence and for dimorphic transition (Lay et al. 1998; Brand et al. 2004; Sharkey et al. 2005). As such, phenotypic assays often compare Uri+ versions of mutants, which have been constructed in a CAI4 background, to CAF2-1 (URA3/ura3). However, these observed phenotypic differences may vary when these mutants are compared to CAI4 (ura3/ura3). Here, we illustrate this “anomalous” behavior with mutants defective in NHEJ, such as ku70∆ and lig4∆, and phenotypes associated with MMS sensitivity. A heterozygous ku70Δ/KU70 ura3Δ/URA3 strain (LCD1A) was similarly sensitive to ≥0.02% MMS (Figure 1A and Figure S1A) as the homozygous ku70Δ/ku70Δ ura3Δ/URA3 strain (LCD2A), suggesting that KU70 is haploinsufficient with respect to MMS sensitivity. Alternatively, the KU70 allele a (the only allele present in strain LCD1A) may be inactive or hypofunctional, whereas allele b has wild-type function. To distinguish between these hypotheses, we generated new heterozygous strains and selected two types of transformants: one retaining allele a (LCD1C, isogenic to LCD1A) and the other retaining allele b (LCD1B). Both of these new heterozygous strains were similarly sensitive to 0.03% MMS (Figure S1A), indicating these alleles of KU70 are haploinsufficient. Intriguingly, LCD3A, a reconstituted strain carrying both KU70 alleles, had gene expression levels comparable to that of CAF2-1 (Chico et al. 2011), but did not suppress the MMS sensitivity phenotype (not shown), a result difficult to reconcile with haploinsufficiency.
Sensitivity assay to MMS and temperature of C. albicans strains derived from strain CAI4-L. (A) Fivefold serial dilutions of both heterozygous and homozygous ku70 mutants derived from strain CAI4-L were spotted on YPD plates supplemented or not with 0.03% MMS. Plates were incubated at 28°. For temperature sensitivity YPD plates inoculated with the same strains were incubated at 28° and 43°. Images were taken after 40 hr. (B) MMS and temperature sensitivity of two independent CAI4-L derivatives (TCS570 and TCS571) in which IRO1 and URA3 had been reintegrated in its own locus.
The temperature sensitivity phenotype was similarly inconsistent among these strains. The ku70Δ/ku70Δ ura3Δ/URA3 strain (LCD2A) was only moderately thermosensitive at 43° but unable to grow at 45°, whereas the wild type (CAF2-1, ura3Δ/URA3) grew robustly at both temperatures. Like the ku70 null strain, heterozygous ku70Δ/KU70 ura3Δ/URA3 strains LCD1A and 1B were thermosensitive, in an allele-independent manner (Figure 1A and Figure S1A). Reintegration of one or both KU70 alleles in the null strain failed to restore the temperature-resistant phenotype (not shown).
The anomalous behavior of ku70 mutants, was further complicated by the observation that both heterozygous and null ku70 mutants in a Uri− background (ura3Δ/ura3Δ) did not differ in MMS sensitivity from CAI4-L (the CAI4 strain maintained in our laboratory) (Chico et al. 2011; Figure 1A and Figure S1B). When we directly compared CAI4-L to CAF2-1 for MMS and temperature phenotypes, we observed that CAI4-L was MMS sensitive and thermosensitive (Figure 1). These perplexing results motivated us to investigate the genetic underpinnings of the MMS sensitivity observed in CAI4-L and its derivative strains.
MMS and temperature sensitivity is independent of URA3 expression
To determine whether the lack of URA3 (CAI4-L), or its decreased expression, potentially derived from position effects, accounted for MMS and/or temperature sensitivity, we compared CAI4-L (ura3Δ/ura3Δ) to ku70 null strains that were either Uri+ (ku70Δ::hisG-URA3-hisG/ku70Δ::hisG) or Uri− (ku70Δ::hisG/ku70Δ::hisG) for MMS and temperature sensitivity. Similar sensitivities were observed between the aforementioned strains, all of which were markedly less resistant than CAF2-1 (ura3Δ/URA3) (Figure 1A), implicating URA3 expression effects. While reintegration of URA3 at its endogenous locus in the CAI4-L background (TCS570 and TCS571) restored thermotolerance similar to that of CAF2-1 at 28°, temperature sensitivity at 43° and MMS sensitivity was unchanged (Figure 1B). Similarly, supplementing uridine to the growth media resulted in slight growth improvements at 28° but not at 43° in CAI4-L and its Uri+ derivatives (TSC570 and TSC571) (not shown). Taken together, these results suggest that the MMS- and temperature-sensitive phenotypes of ku70 deletants constructed in a CAI4-L background cannot be attributed to the absence or low expression of URA3.
We next tested the hypothesis that both phenotypes resulted from some unknown genetic event(s) that occurred during the construction of CAI4 or the propagation of CAI4-L in our laboratory. If this were the case, all CAI4-L derivatives should display MMS- and temperature sensitivity regardless of the ORF targeted for deletion. We performed phenotypic analysis of two additional sets of mutants derived from CAI4-L: one set investigating heterozygous and homozygous disruption of LIG4 (orf19.5798 ), a gene also involved in NHEJ, and a second set investigating analogous disruptions in SHE9/MDM33 (orf19.5796 ), whose gene function is unrelated to DNA metabolism (Andaluz et al. 2001; Messerschmitt et al. 2003). Consistent with the hypothesis, we observed sensitive phenotypes similar to those seen in ku70 disrupted strains in both sets of disruption (data not shown).
LOH of Chr3R occurred during CAI4 construction and confers MMS sensitivity
The seminal paper by Fonzi and Irwin (1993) describes construction of three isogenic ura3Δ::imm434/ura3Δ::imm434 strains, CAF3-1, CAF4-2, and CAI4, by disruption/conversion of the remaining URA3 allele in CAF2-1 (ura3Δ::imm434/URA3). Importantly, there are differences in MMS and temperature sensitivity among these three Uri− strains. CAF3-1 and CAF4-2 were MMS resistant, similar to the parental CAF2-1, whereas CAI4 and CAI4-L were sensitive (Figure 2A). This implies that the genetic alteration responsible for MMS sensitivity occurred only in CAI4, but not in CAF3-1 and CAF4-2. Of note, while CAF3-1 and CAF4-2 are MMS resistant, they are extremely thermosensitive at 43°, indicating that the thermo- and MMS sensitivities are unlinked (Figure 2A).
Phenotypic and genomic analysis of strains Uri− derived from CAF2-1. (A) MMS- and temperature sensitivities of strain CAF2-1 and its Uri− derivatives CAF4-2, CAF3-1, CAI4, and CAI4-L (Fonzi and Irwin 1993). The assay was carried out as described in Figure 1. (B) Visualization of SNP/CGH array data of the same strains. Regions that were homozygous in the reference strain or were not informative were not colored. Heterozygous regions are gray colored and homozygous pink or blue colored. Of note, there is not univocal correlation between pink and a or blue and b. For Chr3R, pink is a, and for Chr2L, blue is a. Unambiguous ascription of LOH regions to aa or bb alleles according to current GCD assembly 22 required analysis of individual SNPs and is indicated in the text (see also Andaluz et al. 2011).
The three isogenic Uri− strains were generated independently and through different genetic manipulations. CAF3-1 was obtained by selection on 5-FOA plates (Fonzi and Irwin 1993), and accordingly 5-FOAR resulted from a recombination event that included either: (1) a long-range gene conversion or (2) crossover/break-induced replication followed by cosegregation of the two ura3Δ markers. Both CAF4-2 and CAI4 were products of a second disruption step at the URA3 locus. The gene replacement event disrupting the remaining URA3 allele in CAF4-2 was verified (Fonzi and Irwin 1993), while the molecular mechanism underlying the construction of CAI4 was not reported (Figure S2).
It was recently shown that multiple LOH events occurred and accumulated in the laboratory lineage of C. albicans strains (Abbey et al. 2011). One relevant LOH is ∼64 kb of Chr3R, beginning within 12–15 kb from CEN3 and including the SNP71 marker, which is detected in some stocks of CAI4 and in all of its derivatives (Abbey et al. 2011), including the RM series of strains (Alonso-Monge et al. 2003), BWP17 (Wilson et al. 1999), and the SN series (Noble and Johnson 2005). Furthermore, RM10 harbors a second long-range LOH on Chr2L. MMS sensitivity was evident in all RM derivatives, as in CAI4 (Figure 3).
MMS sensitivity of the indicated Uri− strains derived from CAI4 (Abbey et al. 2011). Assays were carried out as described in Figure 1 using YPD plates supplemented with uridine. Strains CAF2-1 and CAI4-L were used as control.
SNP/CGH analysis of Uri− strains (CAF3-1, CAF4-2, and CAI4) revealed several important insights including: (1) all strains remained diploid; (2) in CAF3-1, homozygosis of a large portion of Chr3L initiated via recombination near CEN3; (3) CAI4 resulted from either short-range gene conversion at URA3 or through gene disruption of the remaining URA3 ORF but not from a long-range recombination event; and (4) no Chr3R LOH events were detected in CAF4-2 and CAF3-1 (Figure 2B). Of note, CAI4-L harbors a second long-range LOH event on Chr2L, similar to that detected in RM10, suggesting that this may be a recombination hotspot. Detailed analysis of individual SNPs in these strains supports the inferred chromosomal perturbations (Figure S2).
Chr3R homozygosity is sufficient to confer MMS sensitivity in CAF4-2
The only feature common between CAI4 and its derivatives is the 65-kb homozygous region on Chr3R, suggesting that it confers MMS sensitivity. To test this hypothesis, we induced homozygosis of the Chr3R region close to CEN3 in CAF4-2, by first inserting URA3 in select intergenic regions of Chr3R (see below and Figure 4) and selected for LOH on 5-FOA. The Uri− derivatives obtained were tested for MMS susceptibility and for genetic rearrangements, including the allelic status of SNP44 and SNF5. T3, a strain in which URA3 was inserted 4.5 kb from CEN3 (between coordinates 831 and 832 kb) was MMS resistant and maintained heterozygosity for SNPs on Chr3L and Chr3R (not shown). However, all the 5-FOAR segregants of T3 exhibited MMS susceptibility. Not only were all the 5-FOAR segregants homozygous for the a allele of SNP44 (Chr3L) but also for an a allele marker within POL1 (Chr3R), suggesting that the LOH event involved loss of the entire Chr3 b homolog.
Effect of Chr3R LR-LOH induction in the MMS sensitivity of strain CAF4-2. (A) Diagram of Chr3 showing the position of the inserted URA3 ORF in the CAF4-2 derivatives T6 and T7 as well as the rearrangements of Chr3 and Chr2 in several 5-FOAR (Uri−) segregants (P21, P27, P29, P32, P41, and P42) as deduced from the subsequent SNP-RFLP analysis. (B) MMS sensitivity assay of the same 5-FOAR segregants. The 7-µl aliquots of 25-fold dilutions from exponentially growing cultures were spotted on YPD plates supplemented or not with 0.02% MMS. Plates were incubated at 28° for 40 hr.
In order to favor LOH events mediated through recombination over those from whole chromosome loss, we inserted URA3 ∼55 kb from CEN3 (between coordinates 882,317 and 882,533 kb), and subsequently analyzed the 5-FOAR segregants from two independent transformants, T6 and T7 (Figure 4 and Figure S3). While the majority of 5-FOAR segregants resulted in the loss of Chr3b (88% and 95% in T7 and T6, respectively), we isolated several 5-FOAR segregants that were heterozygous for SNP44, implying that LOH was due to recombination (Figure 4A and Figure S3B). The 5-FOAR strains P41 and P42 were derived from T6, and P21, P27, P29, and P32 were derived from T7. These recombinant strains fell into two groups: the first group (P27 and P29) was MMS resistant and maintained both alleles of SNF5, whereas the second group (P21, P32, P41, and P42) was MMS sensitive and homozygous for SNF5a (Figure 4B and Figure S3A). From these results, we conclude that segmental Chr3R homozygosis, including SNP71 and SNF5, confers MMS sensitivity in C. albicans.
Candidate Chr3R genes responsible for MMS sensitivity
The Chr3R homozygous region in CAI4 includes 11 ORFs (Figure 5 and Table 1) that are highly polymorphic in SC5314 and its derivatives. MMS-sensitive derivatives of CAF4-2 (P21, P32, and P41) are homozygous for the Chr3R a alleles of these 11 ORFs (Figure 4, Figure 5, Figure S4, and Table 1), similar to what we observed in CAI4. The remaining MMS-resistant strains (P27 and P29) were a mixture of heterozygous and homozygous (allele b) ORFs/DNA tracks (Figure 4, Figure 5B, Figure S4, and Table 1).
Visualization of SNP/CGH array data of Chr3R LOH-induced strains derived from CAF4-2. (A) SNP/CGH profiles of P strains. Regions that were homozygous in the reference strains or were not informative were not colored. As described in Figure 2, for Chr3R, pink is a, and for Chr2L, pale blue is a. (B) High-resolution depiction of the allelic state of markers located between CEN3 and CGD coordinate 910,000. Each strain was analyzed by SNP/CGH array and data visualized as described (Abbey et al. 2011) using the C. albicans genome sequence assembly 21. Regions that were homozygous in the reference strain CAF4-2 or were not informative were not colored. Heterozygous regions are gray colored. Alleles aa are pink colored and alleles bb blue colored. There exists a perfect correlation between our allele assignment and that reported in sequence assembly 22. A snapshot of the raw data is shown in Figure S4.
While our data cannot unambiguously identify the pathways that yielded the P strains, we propose the following mechanisms (Lee et al. 2009; Lee and Petes 2010; St Charles et al. 2012). The allelic configurations in MMS-sensitive strains (P21, P32, and P41) are consistent with interhomolog reciprocal exchange between orf19.5854.1 and CEN3, followed by cosegregation of the chromatids carrying the a alleles into the same daughter cell (Figure 6, A and B). The allelic configurations in MMS-resistant strains (P27 and P29) are consistent with reciprocal crossover and gene conversion resulting from a double-strand break (DSB) during G1 near the URA3 insertion site on the b homolog (Figure 6C). Such an event would not affect ORFs proximal to CEN3, which remained heterozygous, as noted (Figure 6C and Table 1). CEN3 distal sequences in all the P strains have more complicated patterns, in which some polymorphisms are detected, interspersed in the homozygous regions (alleles a for P21, P32, and P41, and alleles b for P27 and P29) and are indicative of complex recombination events or additional mutations (Hicks et al. 2010; Deem et al. 2011). It should be noted that P21, which exhibited the highest MMS sensitivity, also carries a Chr2L LOH event (Figure 4 and Figure S3C; see below).
Proposed mechanisms for the generation of the several P strains (FOAR segregants) derived from URA3 transformants T6 and T7 as deduced from the SNP-CGH results. (A) Strain P41: BIR or reciprocal crossover without associated conversion initiated by a G2 DSB between orf19.5854.1 and CEN3, which marks the transition from heterozygous markers to LOH (alleles a). (B) P21 and P32: reciprocal crossover with associated conversion resulting from a G1 DSB in homolog a (red) between orf19.5854.1 and CEN3, which is repaired using homolog b (blue). This would explain the existence of a short region homozygous for homolog b sequences around the CGD coordinate 837,000 in both strains. A BIR event (instead of reciprocal crossover) is also possible (not shown). (C) P27 and P29: Reciprocal crossover with associated conversion resulting from a G1 DSB in homolog b in the URA3 neighborhood. One chromatid is repaired to yield a reciprocal crossover with conversion and the second one is repaired using gene conversion or BIR. URA3 is indicated with a square. orf19.5854.1 and orf19.5880, which delimit the CAI4 homozygosed region, are indicated with a triangle and a diamond, respectively. The unusual homozygosity of the Chr3R region distal to orf19.5880 (not included in the diagram) prevents the unambiguous identification of the mechanism responsible for the conversion event.
Of the 11 ORFs in the homozygosed region found in P21, P32, and P41, only two carry only synonymous SNPs. Therefore, the a alleles of one or more of the remaining ORFs are likely to be responsible for the MMS-sensitivity phenotype. Based on their reported GO functions in Candida and Saccharomyces (http://www.candidagenome.org/ and http://www.yeastgenome.org/), four of these ORFs, SNF5, POL1, orf19.5854.1, and MBP1, were selected for further study. SNF5 (orf19.5871 ) encodes a component of the Swi/Snf chromatin remodeling complex and null mutants display increased MMS sensitivity in Saccharomyces cerevisiae (Chai et al. 2005) and anomalous biofilm formation and other pleiotropic defects in C. albicans (Finkel et al. 2012). We found that SNF5 homozygous deletants in the CAF2-1 background (TCS1038 and TCS1039) were more sensitive to MMS than its parent, but not as sensitive as CAI4 (Figure S5A). In SC5314 and CAF2-1, electrophoretic analysis of SNF5 PCR products indicated the presence of two bands. However, in CAI4 and CAI4-L only the largest band was detected (see CAF4-2 and CAI4 profiles in Figure S3A) and sequencing of SNF5 from CAF2-1 revealed eight SNPs relative to SNF5 from CAI4. The sequence from CAI4 therefore identifies the haplotype of the larger allele (not shown). The length differences in SNF5 alleles results from a 159-bp indel after nt 227 of the SNF5 coding sequence; consequently the larger allele encodes 53 additional amino acids (Table 2). This larger SNF5 allele has not been annotated in the CGD or in the recent assembly of a phased diploid C. albicans genome (Muzzey et al. 2013).
We next asked if POL1, orf190.5854.1, and MBP1 had significantly altered a alleles that may confer MMS sensitivity when homozygous. POL1 (orf19.5873 ) encodes the DNA polymerase required for replication initiation and carries several nonsynonymous SNPs (CGD; Table 2). The S. cerevisiae ortholog of orf19.5854.1 (orf19.5854.1 ) is YBT1, which encodes an ATP-binding ATPase that transports and sequesters Pb into the vacuole (Sousa et al. 2015). The a allele of orf19.5854.1 carries a nonsense mutation resulting in a 69 amino acid truncation (Table 2). Vacuolar sequestration of MMS may help cell survival in the presence of this toxicant. Finally, MBP1 (orf19.5855 ) encodes a transcription factor that, together with Swi6, forms MBF to regulate genes involved in DNA replication and repair during the cell cycle and following MMS treatment in S. cerevisae (Travesa et al. 2012, and references therein). In C. albicans, MBP1 has six nonsynonymous SNPs, and null mutants have minor growth defects under normal conditions (Hussein et al. 2011). We hypothesized that MMS sensitivity may result from a defective MBP1 allele.
MBP1a is responsible for MMS sensitivity in CAI4
Hemyzygosis of either SNF5 or its adjacent gene, POL1, as well as hemizygosis of both of them together (Figure S5A), did not change MMS sensitivity. Similarly, hemizygosis of orf19.5854.1 did not affect the MMS sensitivity phenotype (Figure S5B). Thus, MMS sensitivity does not appear to be due to hemi- or homozygosis of any of these three genes.
In contrast, hemizygous MBP1a strains were MMS sensitive while hemizygous MBP1b strains were not MMS sensitive (Figure 7A). Furthermore, consistent with the idea that the defects in MBP1a are responsible for the MMS sensitivity, reintegration of the MBP1b allele in a MBP1a/MBP1a strain restored MMS resistance (Figure 7B) and MMS-resistant strains, P27 and P29, retained MBP1 heterozygosity. Thus, at least one of the six nonsynonymous SNPs in MBP1 results in a hypo- or nonfunctional protein, a premise consistent with the fact that both MBP1 alleles are similarly expressed (Muzzey et al. 2014).
MMS sensitivity of CAI4 and derivatives strains is MBP1 allele dependent. (A) MMS susceptibility of MBP1a or MBP1b hemizygous mutants derived from strains CAF2-1, TCS1018 (POL1a/pol1b-Δ), and TCS1031 (SNF5a/snf5b-Δ). (B) Effect of MBP1b integration in the indicated MBP1a homozygous strains. (B.1) Cassette for MBP1b integration into its locus. (B.2) Upper row, MMS-sensitivity assay. Lower row, verification of correct integration of MBP1b in transformants (NATR) TCS1090 and TCS1093 (both derived from strain TCS570) that had reverted the MMS-sensitivity phenotype.
MMS-induced growth polarization
Genotoxic stress, including hydroxyurea and MMS treatments, can trigger growth polarization of wild-type C. albicans, resulting in elongated cells (Shi et al. 2007; Sun et al. 2011; Wang et al. 2012; Loll-Krippleber et al. 2014). To determine if MMS-induced filamentation can be attributed to homozygosis of MBP1a, we compared CAF2-1 (ura3Δ/URA3) and CAI4-L (ura3Δ/ ura3Δ) cell morphologies in response to MMS. CAF2-1 grew in chains of elongated cells with lateral branches but no long filaments were observed, whereas CAI4-L grew predominantly as long filaments. Reintegration of a single URA3 at its endogenous locus (strain TCS570) did not alter its cell morphology (Figure 8). Therefore, the loss of URA3 is not responsible for the enhanced filamentous growth displayed by CAI4-L in response to MMS. However, replacement of one of the MBP1a alleles of TCS570 with MBP1b resulted in a strain (TCS1090) that showed a response to MMS comparable to that of SC5314 and CAF2-1 (Figure 8). Therefore, we conclude that the exacerbated MMS-induced filamentous growth observed in CAI4 and its derivatives is due to homozygosis of MBP1a.
MMS-induced filamentous growth in C. albicans strains CAF2-1 and CAI4-L derivatives. A YPD overnight culture of exponentially growing cells from the indicated strains was refreshed and adjusted to OD600 = 1. Following a further incubation for 2 hr at 30° with shaking, one half was suspended in YEPD supplemented or not (control) with 0.02% MMS. After 16 hr at 30° with gentle shaking, samples were photographed (DIC). Strain TCS570 is a CAI4-L derivative in which URA3 has been reintegrated into its locus. Strain TCS1090 is a TCS570 derivative in which MBP1b has substituted one of the two MBP1a alleles.
Sign epistasis results in increased MMS sensitivity of CAI4-L
We next investigated the role of another LOH region, found on Chr2L in CAI4, RM10 and their derivatives, in MMS susceptibility. There is a slight, but reproducible difference in MMS sensitivity between CAI4-L and CAI4 (Figure 9A). When we compare their SNP/CGH profiles, a homozygous region is evident on Chr2L in CAI4-L, whereas heterozygosity is maintained in CAI4 (Figure 2B) (Abbey et al. 2011; Andaluz et al. 2011). In CAI4-L, this LOH occurred via an interhomolog recombination event at coordinate 1,656,168 and we hypothesized that Chr2L homozygosity may contribute to the MMS sensitivity phenotype. Supporting this idea, P21 (a CAF4-2-derived segregant with Chr3R LOH) was even more MMS sensitive than CAI4, reaching the susceptibility of that seen in CAI4-L (Figure 4B). SNP analysis revealed that P21 acquired Chr2L segmental homozygosity (Figure S3C).
Effect of Chr2L LR-LOH induction in the MMS sensitivity of strain CAI4. (A) Comparison of MMS sensitivities of CAI4-L and CAI4. (B) Diagram of Chr2 showing the position of the inserted URA3 ORF in transformants T3 and T9. Both transformants were PCR verified for the correct insertion of URA3. (C) MMS sensitivity of several FOAR segregants (P strains) derived from transformants T3 and T9. (D) RFLP analysis of SNP60 and SNP56 markers from the indicated strains following digestion of the PCR amplification products with TaqI and EcoRV, respectively (Forche et al. 2009a). (E) Diagram of Chr2 showing the LR-LOH region present in all FOAR segregants exhibiting MMS sensitivity similar to CAI4-L.
To further evaluate the role of this Chr2L LOH in MMS sensitivity, we constructed two strains (T3 and T9) with URA3 insertions within orf19.3148 (between coordinates 1,383,200 and 1,385,782) in a CAI4 background (Figure 9B). Sixteen 5-FOAR segregants from both T3 and T9 were SNP typed and tested for MMS sensitivity. The 5-FOAR segregants heterozygous for the SNP60 and SNP56 markers displayed MMS sensitivity similar to that of CAI4, whereas all segregants homozygous for the SNP60 marker were MMS sensitive to a similar degree as CAI4-L (Figure 9, C and D). Importantly, all MMS-sensitive segregants carried only the SNP60a allele, just like CAI4-L (Forche et al. 2008). Since SNP68 on Chr2R remained heterozygous (not shown), it is likely that Chr2L-homozygous/MMS-sensitive strains resulted from CO or BIR occurring between the CEN2 and orf19.3148. This is reminiscent of the events postulated to have occurred in CAI4-L, although we cannot exclude the possibility of a terminal truncation (Figure 9E). By contrast, the MMS-resistant 5-FOAR segregants were heterozygous for SNP60 (Figure 9, C and D) and SNP68 (not shown) and likely restored orf19.3148 through a GC event using the other allele as a template.
To test if Chr2L LOH is sufficient to confer MMS sensitivity independently of the Chr3R LOH, we constructed URA3 insertions in the same region of Chr2 in a CAF4-2 background so that Chr3R remained heterozygous. About 60 5-FOAR isolates from two independent transformants (20 from T1 and 40 from T24) were tested for MMS susceptibility (Figure S6C). We observed a range of MMS sensitivity with some 5-FOAR segregants that were more sensitive to MMS and others that were similar to the control. Some 5-FOAR segregants had acquired the LR-LOH on Chr2L (indicated by SNP56 and SNP60 homozygosity) (Figure S6B) but conserved the CAF4-2 MMS rsistance (Figure S6C). This is likely a reflection of the intrinsic genomic instability of C. albicans, which has shown to be enhanced by stresses including 5-FOA (Wellington and Rustchenko 2005; Rustchenko 2007; Bouchonville et al. 2009; Gerstein and Berman 2015) and from random LOH or mutational events (Larriba and Calderone 2008). We infer that Chr2L homozygosis is neutral in isolation but can significantly enhance MMS sensitivity in conjunction with Chr3R LOH and is the first report of sign epistasis in C. albicans.
Discussion
In this study, we identified a MMS-sensitive phenotype associated with a Chr3R LOH event that occurred during the construction of CAI4. Besides, a second LR LOH event on Chr2L present in some CAI4 derivatives exacerbates that phenotype, an indicator of sign epistasis. The high degree of heterozygosity of the C. albicans genome, as exemplified by sequencing efforts of SC5314 and WO-1 (Butler et al. 2009), can be irreversibly modified by LOH events. LOH occurs spontaneously with 10−6 events per cell per generation and is elevated by one to two orders of magnitude when cells are exposed to external stressors or undergo transformation protocols utilizing heat shock (Wellington and Rustchenko 2005; Bouchonville et al. 2009; Forche et al. 2011). During the construction of CAI4, cells were subject to two independent heat shocks with counterselections each (Fonzi and Irwin 1993) and subsequent derivatives constructed for additional auxotrophic markers were again exposed to such stresses (Alonso-Monge et al. 2003; Noble and Johnson 2005). Therefore, it is not surprising that these strains acquired tracts of LOH unrelated to their target regions (Abbey et al. 2011). Furthermore, these types of laboratory manipulations can increase the occurrence of aneuploidy (Bouchonville et al. 2009) and it has been shown that some stocks of CAI4 carry trisomies of Chr1 and/or Chr2 (Chen et al. 2004; Selmecki et al. 2005).
The regions of LOH and aneuploidies occurring in commonly used C. albicans strains are easily identifiable by the use of high-resolution SNP/CGH microarrays (Abbey et al. 2011). Relevant to this work is a short region of Chr3R nearby CEN3 that homozygosed during the construction of CAI4 (ura3Δ/ura3Δ) from CAF2-1 (ura3Δ/URA3). Phenotypic analysis of mutants derived from CAI4 is complicated because URA3 expression levels affect filamentation and virulence properties (Bain et al. 2001; Chen et al. 2004; Noble and Johnson 2005; Sharkey et al. 2005; Noble et al. 2010) as well as replication stress (Poulter 1990). Yet we found that, (1) URA3 reintegration into its endogenous locus does not affect MMS sensitivity and (2) supplementing YPD with additional uridine did not modify MMS sensitivity (Figure 1).
LOH is an irreversible process and thus LOH-associated phenotypes are found in all descendant strains. In this study, we identified MMS and thermosensitive phenotypes associated with mutants constructed in the CAI4 background. In the case of KU70 (orf19.1135 ), since it is a key player in NHEJ (Chico et al. 2011), one of the pathways for DSB repair (Critchlow and Jackson 1998; Daley et al. 2005), it would not be surprising that Caku70 null strains are sensitive to high MMS concentrations. Indeed, S. cerevisiae ku70 mutants are sensitive to MMS (Foster et al. 2011); besides, in this background, telomeres become shorter and cause thermosensitivity at 37° (Barnes and Rio 1997). Since C. albicans is intrinsically more thermotolerant than S. cerevisiae, analysis of the thermosensitivity of Caku70 null strains required incubation at higher temperatures (42–43°). We found that both phenotypes are not due to the targeted deletions but inherent to CAI4 strain. Therefore, it is critical to recognize that random, nontargeted LOH can occur during strain construction and result in unexpected phenotypes, which often are mistakenly attributed to the target mutation if the appropriate comparisons are not made. A lesson derived from our results is that phenotypic characterization, in particular growth polarization in response to 0.02/0.03% MMS, limited to comparison of the Uri+ versions of deletants constructed in the CAI4 strain or in some of its descendants to CAF2-1 or SC5314 strains should be revisited.
This study further emphasizes the role of mitotic recombination in modifying and/or altering phenotypes and diversifying diploid descendants during clonal reproduction without having to go through a sexual cycle (Mandegar and Otto 2007; Otto and Gerstein 2008). It is possible that we could detect additional traits, beyond MMS and thermosensitivity, if tested under appropriate selective conditions. Importantly, the diploid state of C. albicans tolerates disruptive mutations in one allele, due to heterozygous masking, similar to that seen in animals and other organisms. Phenotypes caused by the mutant allele will reveal themselves after monosomy or LOH homozygosis for that allele.
The diploid state has been considered a capacitor for evolution (Schoustra et al. 2007) through accumulation of recessive mutations, provided they show sign epistasis, i.e., mutations in isolation are either neutral or deleterious, but are advantageous in combination or with the appropriate genetic background (Weinreich et al. 2005). In this regard, some CAI4 stocks carry additional regions of LOH, such as CAI4-L, which has Chr2L long-range LOH and no detectable aneuploidy (Andaluz et al. 2011). We cannot distinguish whether this long-range LOH was spontaneous or a result of unknown stress. This region of Chr2L homozygosis has been detected in RM10 and SN strains, both of which are derived from CAI4 (Abbey et al. 2011), as well as among parasexual progeny (Forche et al. 2008). The relatively high frequency at which we detect Chr2L homozygosis may be due to a recombination hot spot on Chr2L and/or due to a selective advantage to this event under laboratory growth conditions. Here, we found that Chr2L homozygosis was neutral for MMS sensitivity when in isolation, but in combination with Chr3R homozygosis, it increased MMS sensitivity. This work represents a clear example of sign epistasis, which has a proposed important role in evolution (Weinreich et al. 2005), but has not previously been reported in C. albicans.
Sign epistasis may be more frequently detected in complex traits, like MMS sensitivity, that are influenced by hundreds of genes (Begley et al. 2002, 2004; Horton and Wilson 2007). Repair of MMS lesions involves pathways including base excision repair (Boiteux and Jinks-Robertson 2013), double strand break repair (Symington et al. 2014), chromatin remodeling (Oum et al. 2011) pathways, etc. Mutations in these genes will likely result in altered sensitivity to MMS. Furthermore, it is likely that combination of weak alleles across multiple genes can cause MMS sensitivity, despite low or no sensitivity associated with any individual weak allele. Weak alleles present in deletion backgrounds may enhance the phenotype of the deleted gene of interest. In S. cerevisiae, a mutant allele of RAD5, rad5-G535R, was sensitive to MMS (Fan et al. 1996). However in response to UV light, another complex trait, the interactions between RAD5 and RAD52 alleles are more complex. Strains carrying the rad5-G535R allele were more UV sensitive, particularly at high doses compared to wild type; strains with rad52 rad5-G535R double mutations were more UV sensitive than either mutation in isolation (Fan et al. 1996).
Given that long-range LOH results in homozygosity of many genes, it is predicted that there will be a number of altered phenotypes associated with LOH, particularly in highly heterozygous organisms. Regions of homozygosity (ROH) have recently been implicated with complex traits and diseases in human, specifically height (Yang et al. 2010), schizophrenia, and late-onset Alzheimer’s disease (Nalls et al. 2009). C. albicans, because it is a highly heterozygous diploid, may be a good model for further studies of this nature. A recently identified long-range LOH in clinical isolates of C. albicans has been linked to fluconazole resistance and these types of events were both recurrent and persistent (Ford et al. 2015). ERG11 was included in the region of LOH and in some cases, a driver mutation in ERG11 (i.e., a persistent mutation for a genotype not found in the progenitor strain) was identified. It is likely that the driver mutation arose first in a heterozygous manner and was homozygosed by a subsequent LOH. However, it was not known if this driver mutation was sufficient to cause fluconazole resistance or if other genes within the region of LOH also contributed to the azole resistance phenotype.
Even in the absence of driver mutations, LOH events can confer new phenotypes, since allelic variation occurs within the heterozygous parental strain. This premise is exemplified in our study. Alleles of MBP1 differ in activity, and hemizygous MBP1a strains are more sensitive to MMS than hemizygous MBP1b strains. Importantly, MBP1a lacks an N-terminal low complexity region (LCR) in which two Ser residues are mutated to Pro and Phe, respectively (Table 2). Terminal LCRs are enriched for translation and stress-response-related terms (Coletta et al. 2010). We suggest that substitution of one or both Ser residues (see Results) are likely responsible for differences in Mbp1 activity. In particular, Ser152 (Mbp1b) resides within a Ser-enriched stretch (148–157) and is potentially phosphorylated (NetPhos 2.0 server).
We rule out the possibility that the MMS phenotype is due to a combination of weak alleles within the region of Chr3R LOH: MBP1a homozygosis or hemizygosis is sufficient and the simultaneous homozygosity or hemizygosity of SNF5 and POL1, two other candidate polymorphic genes, had no additional effect. We also considered that MMS sensitivity requires homozygosis and not hemizygosis of SNF5, since some fungal phenotypes need transvection between alleles (Aramayo and Metzenberg 1996). However, MMS resistance was not restored when SNF5 homozygosity in CAI4 was disrupted.
In S. cerevisiae, Mbp1 binds DNA and, together with Swi6, forms the transcription complex MBF to regulate genes involved in DNA replication and repair in G1/S phase; MBF activates these genes during G1 and represses them, with Nrm1, outside of the G1 phase. In response to replication stress, phosphorylation of Nrm1 releases MBF to activate genes involved with DNA repair and replication (Travesa et al. 2012). In S. cerevisiae, POL1 is an Mbp1 target gene induced in response to genotoxic stress caused by hydroxyurea, MMS, or camptothecin (Travesa et al. 2012). It is currently unknown if this regulatory circuit is conserved in C. albicans.
Beyond providing evidence that LOH is associated with novel phenotypes in CAI4 (which raises issues for the lab strains derived from it), we also revealed an important feature of SNF5 that was not previously annotated in the CGD. Taken together with an earlier study reporting 11 new SNPs in HIS4 (Gómez-Raja et al. 2008), we posit that the number of polymorphisms reported for the C. albicans SC5314 genome, including SNPs, indels, and truncated ORFs (Jones et al. 2004; Braun et al. 2005; van het Hoog et al. 2007; Muzzey et al. 2013), is a conservative estimate and could be even higher than documented in the current databases.
Acknowledgments
We thank Belén Hermosa (Universidad de Extremadura) for technical support. This study was supported by Ayuda a Grupos from Junta de Extremadura. A.B. was supported by a fellowship from Junta de Extremadura. J.B. was funded by an award from the Israel Science Foundation (340/13).
Footnotes
Communicating editor: A. P. Mitchell
Supplemental material is available online at www.genetics.org/lookup/suppl/doi:10.1534/genetics.116.189274/-/DC1.
- Received March 22, 2016.
- Accepted May 17, 2016.
- Copyright © 2016 by the Genetics Society of America
