Identification of Recessive Lethal Alleles in the Diploid Genome of a Candida albicans Laboratory Strain Unveils a Potential Role of Repetitive Sequences in Buffering Their Deleterious Impact

Candida albicans is a major fungal pathogen, whose mode of reproduction is mainly clonal. Its genome is highly tolerant to rearrangements, in particular loss of heterozygosity events, known to unmask recessive lethal and deleterious alleles in heterozygous diploid organisms such as C. albicans. By combining a site-specific DSB-inducing system and mining genome sequencing data of 182 C. albicans isolates, we were able to ascribe the chromosome 7 homozygosis bias of the C. albicans laboratory strain SC5314 to an heterozygous SNP introducing a premature STOP codon in the MTR4 gene. We have also proposed genome-wide candidates for new recessive lethal alleles. We additionally observed that the major repeat sequences (MRS) on chromosome 7 acted as hot spots for interhomolog recombination. Maintaining MRS in C. albicans could favor haplotype exchange, of vital importance to LOH events, leading to homozygosis of recessive lethal or deleterious alleles that inevitably accumulate upon clonality.

KEYWORDS Candida albicans, homologous recombination, loss of heterozygosity, major repeat sequences, recessive lethal alleles I n diploid genomes, new mutations are heterozygous, and their effect is generally masked by the presence of the ancestral allele. As claimed in Haldane's sieve, only mutations that confer a fitness advantage as heterozygotes can invade the population. Although true, it does not make specific prediction about the fitness of the mutant homozygotes. Recent studies of Saccharomyces cerevisiae have observed maintenance of genetic variation due to heterozygote advantage, as a result of overdominance of mutated alleles (1). In addition, Gerstein and colleagues used the model organism S. cerevisiae to show that recessive beneficial mutations can avoid Haldane's sieve in clonal organisms, through rapid loss of heterozygosity (LOH), and thus contribute to rapid evolutionary adaptation (1). Similarly, in Candida albicans, mutations followed by genomic rearrangements such as LOH events and isochromosome formation have been associated with the acquisition of antifungal resistance (2,3), bringing forth the idea that mechanisms favoring genome plasticity could contribute to C. albicans fitness within the host and upon exposure to antifungal agents. C. albicans is a frequent human commensal yeast responsible for both mucosal fungal infections and the majority of life-threatening nosocomial fungal infections (4). Its diploid genome displays a high degree of plasticity that includes, in particular, LOH events. Despite frequent LOH, overall heterozygosity is maintained in the C. albicans population, as illustrated by various studies that highlighted the elevated levels of natural heterozygosity, with a heterozygous position every ϳ200 to 250 bp on average (5)(6)(7). Furthermore, several studies revealed that genome heterozygosity showed a significant correlation with higher growth rates (6,8,9).
Essentially, mutations can be categorized as beneficial, harmful, or neutral and can be differently assigned depending on the organism's environment. Because the mode of reproduction of C. albicans is mainly clonal, and therefore mimics inbreeding in higher eukaryotes, an increased number of recessive lethal alleles (RLA) in the C. albicans genome is expected compared to other eukaryotes that undergo true sexual reproduction. Various types of mutations can impact the functionality of alleles and render them inactive; however, mutations introducing premature STOP codons would convey the most obvious effect. Within the C. albicans laboratory strain SC5314, Muzzey et al. (10) reported almost 200 genes for which one of the alleles contains a single nucleotide polymorphism (SNP) that introduces a premature STOP codon. Functional differences have already been reported for the two alleles of a heterozygous gene, and in all instances, the effect of the recessive mutation was visible only upon homozygosis toward the mutated allele (HIS4 [11], MBP1 [12], GPI16/MRF2 [13]). Moreover, SNPs in promoter regions have been shown to alter expression regulation of two alleles (14). Of interest, LOH is pervasive in C. albicans isolates, as homozygous regions can be found in all sequenced isolates and affect all of the chromosomes. These LOH events vary in size: they can be limited to a single chromosomal region, affect an entire chromosomal arm, or even cover the entire chromosome (6,7,9).
Recently, a combination of molecular tools has been developed to study genome dynamics in C. albicans. First, an LOH reporter system takes advantage of fluorescent markers at an artificial heterozygous locus containing the BFP and GFP genes (15). Consequently, the appearance of spontaneous LOH events for the given locus can be monitored by the fluorescent status of cells using flow cytometry (15). Second, LOH events are often a result of DNA double-strand breaks (DSB) (16) resolved by means of various DNA repair mechanisms which can either be independent or dependent of homologous recombination. Feri et al. (13) developed an inducible, locus-specific DNA DSB system that uses the I-SceI meganuclease from S. cerevisiae. When coupled to the BFP/GFP LOH reporter system, this system can be used to study the consequence of a targeted DNA DSB on the appearance of LOH events. Indeed, Feri et al. (13) have shown that I-SceI-induced DNA breaks are predominantly repaired by gene conversion result-ing in limited LOH. Nevertheless, various patterns of long-range LOH can also be obtained. Of note, the engineered system, alongside sequence resources, helped identify a RLA on Chr4B of C. albicans strain SC5314 (13). This RLA is the consequence of a nonsense mutation in the GPI16 gene that encodes an essential component of the glycosylphosphatidylinositol (GPI) anchor biosynthetic machinery and explains why Chr4B is never observed in the homozygous state in C. albicans strains SC5314. Notably, this RLA appeared unique to strain SC5314 (13).
Although LOH can be observed on all eight chromosome pairs, prior studies conducting haplotype characterization of (i) progeny from the parasexual life cycle (17), (ii) homozygous diploid isolates derived from RAD52 double knockout mutants (18), and (iii) haploid strains of C. albicans (8) showed a chromosome homozygosis bias in the C. albicans laboratory strain SC5314. This suggests that mutations, potentially RLAs, could apply constraints on the directionality of LOH events. Indeed, the homozygosis state of some chromosomes was observed only for a given homolog while recurrently absent for the other homolog. This is the case for chromosomes 1, 4, 6, as well as chromosome 7 (Chr7) for which homozygosis of haplotype B (HapB) is never observed, while haplotype A homozygosis is, suggesting the presence of RLAs on Chr7 HapB (Chr7B).
In this study, we aimed to identify the RLA(s) on Chr7B using an approach similar to that developed by Feri et al. (13) when searching for RLAs on Chr4B. This approach also allowed addressing the role that repetitive sequences, such as the major repeat sequences (MRS), might play on the overall genome dynamics of C. albicans. MRS are unique to C. albicans and Candida dubliniensis, and are found throughout their genomes. MRS are composed of three subregions: RB2 which contains the FRG8 gene, the RPS region which varies in repeat numbers (and thus, in size), and the HOK region. C. albicans possesses eight MRS, one on each chromosome with the exceptions of Chr7 where the presence of one MRS on each arm is observed and of Chr3 where an incomplete MRS is located (19). MRS expansion and contraction have previously been shown to be involved in chromosome loss where the chromosome copy containing the shorter MRS region is spontaneously lost (20). Furthermore, MRS have also been shown to be involved in chromosome translocation (19), when two different chromosomes exchange large regions of an arm. Results presented below confirm that homozygosis of Chr7B is not recovered and that this is the consequence of a premature STOP codon in the Chr7B-borne allele of the RNA helicase-encoding gene MTR4. Furthermore, we highlight that repeat regions such as MRS are hot spots for interhomolog recombination upon DNA repair and play a role in LOH dynamics in C. albicans.

RESULTS
Strain engineering to promote and detect long-range LOH on Chr7. Genome analysis revealed that the left arm of Chr7 carries only 9 heterozygous SNPs in 3 ORFs, while the right arm of Chr7 carries 784 heterozygous SNPs in 105 ORFs. Because RLAs are more likely to be found in heterozygous regions, we focused on the right arm of Chr7 to understand Chr7 homozygosis bias. To efficiently screen for the presence of RLAs on Chr7 right arm, we engineered strains carrying an artificial heterozygous BFP/GFP LOH reporter system (15) associated with an I-SceI DNA DSB-inducing system (13) (Fig. 1A). Because we chose to insert the I-SceI target site (TS) associated with the auxotrophic marker URA3 conferring uridine prototrophy in the most mrs-7b proximal, gene-free region found on the right arm of Chr7, this setup allows rendering a maximum number of alleles homozygous on this arm while avoiding mrs-7b. Integration of the URA3-I-SceI TS cassette can occur on either Chr7 homologs (Chr7A or Chr7B); thus, transformants were screened by SNP-RFLP to identify which Chr7 haplotype was targeted (see Fig. S1 in the supplemental material). Using the heterozygous SNP at position 727,328, we showed that 28/51 C. albicans transformants had integrated the I-SceI TS on Chr7A (55%) and 23/51 on Chr7B (45%), demonstrating the absence of integration bias for this locus. Two independent transformants that had integrated the I-SceI TS on Chr7 HapA (CEC5061) or HapB (CEC5062), were selected and used in subsequent analysis (Fig. 1A).
Strains CEC5061 and CEC5062 underwent preliminary characterization regarding their fluorescence status, as well as their growth rate to ensure that the successive transformation steps did not significantly alter their fitness. The fluorescence status of the intermediate and final strains was verified by flow cytometry. The flow cytometry outputs clearly displayed (i) the absence of fluorescence signals in the parental strain CEC4591, (ii) a shift toward the mono-GFP gate upon integration of the P TDH3 -GFP-ARG4 cassette in CEC4679, (iii) a shift toward the double-fluorescent BFP/GFP gate upon integration of the P TDH3 -BFP-HIS1 cassette in CEC4685, and (iv) the double-fluorescence status of the population upon integration of the URA3-I-SceI-TS cassette in strains CEC5061 and CEC5062 (Fig. S2A).
Growth curves performed demonstrated that only the insertion of the URA3-I-SceI TS cassette in strains CEC5061 and CEC5062 resulted in a higher growth rate, almost certainly due to the uridine prototrophy in these strains, as URA3 deletion has been shown to result in significant decreases in C. albicans growth rate even when ura3⌬ strains are grown in rich medium (21) (Fig. S2B).
Validation of the I-SceI DNA DSB induction system on Chr7 by 5-FOA counterselection. As the I-SceI TS is associated with the genetic auxotrophic marker URA3, we could assess the frequency of cells that have survived an I-SceI-induced DNA DSB at the TS by monitoring the frequency of appearance of 5-fluoroorotic acid-resistant (5-FOA R ) colonies upon I-SceI induction. Indeed, 5-FOA R colonies should have lost the URA3 genetic marker (uridine auxotrophy) and are likely to represent cells that have sustained an I-SceI-dependent DNA DSB through a LOH event, even though point mutations in the URA3 gene cannot be excluded. We obtained 501 times more 5-FOA R colonies after I-SceI induction when the I-SceI TS was located on Chr7A and 547 times more 5-FOA R colonies when the I-SceI TS was located on Chr7B (Fig. 1B). These data confirmed the efficiency of the I-SceI-dependent DNA DSB induction system on Chr7. The majority of I-SceI DNA DSB-induced 5-FOA R colonies (90 to 100%) resulted from DNA DSB repair by gene conversion, as suggested by fluorescence and auxotrophy profiles of 32 Ura Ϫ colonies (Fig. 1C). Similar to what has been observed for Chr4 (13), DNA DSBs by I-SceI on Chr7 are predominantly repaired by gene conversion repair mechanisms, resulting in short-range LOH events.
An I-SceI-induced DNA DSB on Chr7B leads to viable cells homozygous for the right arm of Chr7A. Although the 5-FOA assays yield information on the overall occurrence of LOH events encompassing the URA3 gene, it does not allow us to efficiently study the underrepresented long-range LOH events. Thus, we also investigated LOH frequency upon I-SceI expression using flow cytometry, an assay that specifically detects long-range LOH events. As expected, upon induction of I-SceI in strain CEC5062, possessing the I-SceI TS on the GFP-bearing Chr7B, we observed a 12-fold increase in the appearance of mono-BFP cells (Fig. 1D). We also observed a fivefold increase in the appearance of mono-GFP cells (Fig. 1D). A subset of each population was recovered by fluorescence-activated cell sorting (FACS) and further characterized. We observed that, while the majority of the mono-BFP population included true mono-BFP cells displaying complete Chr7A homozygosis distal to the I-SceI TS, 100% of the rare true mono-GFP cells displayed only partial homozygosis of Chr7B. From a mechanistic point of view, the mono-BFP cells resulted most likely from the repair of the DNA DSBs by mechanisms of break-induced replication or mitotic crossover, while the mono-GFP cells could be one of the possible outcomes of DNA DSB repair by gene conversion with crossover during the G 2 phase of the cell cycle (Fig. 1E).
Absence of recovery of cells being fully homozygous for the right arm of Chr7B. Unlike targeting the right arm of Chr7B, a DNA DSB on the right arm of Chr7A in strain CEC5061 should lead to a higher increase in frequency of the mono-GFP cells compared to the mono-BFP cells (Fig. 1D). Although an augmentation in frequency of both mono-BFP and mono-GFP cells was obtained, the mono-BFP cells still appeared at a higher frequency, 8 ϫ 10 Ϫ3 (Ϯ 8 ϫ 10 Ϫ4 ), compared to 1 ϫ 10 Ϫ4 (Ϯ 3 ϫ 10 Ϫ5 ) for the mono-GFP cells in the induced condition (Fig. 1D). Characterization of a subset of FACS-sorted mono-BFP cells confirmed that the true mono-BFP cells had arisen from a DNA DSB repaired by gene conversion with crossover during G 2 . Further characterization of FACS-sorted mono-GFP cells (corresponding to the expected fluorescence) revealed that the I-SceI-induced homozygosis of Chr7B was only partial in the targeted region. Thus, rather than having experienced break-induced replication or mitotic crossover events extending from the I-SceI site to the BFP/GFP locus, the rare mono-GFP individuals were likely to have resulted from DNA DSB repair by gene conversion with crossover during G 2 (Fig. 1E). In conclusion, we were unable to recover a single individual that had undergone complete homozygosis of the right arm of Chr7B distal to the I-SceI TS. This suggests that complete homozygosis of the right arm of Chr7B is associated with lethality in the SC5314 genetic background, thus confirming the chromosome homozygosis bias previously observed and validating the hypothesis that the presence of RLA(s) in this region could be a cause.
A data mining strategy identifies a heterozygous mutation in the MTR4 gene as a possible cause of the homozygosis bias of Chr7. Genome sequence data obtained from a collection of 182 C. albicans isolates (7), including the reference strain SC5314, was used to compile all heterozygous SNPs within ORFs of Chr7 and search for SNPs (i) generating a premature STOP codon, (ii) showing a heterozygous genotype in SC5314, (iii) never observed in the homozygous state in the collection of 182 genomes, and (iv) located in a coding region never found to be dispensable in C. albicans. Only one such SNP was identified, located at position 746,359 on Chr7B. In C. albicans strain SC5314, this SNP causes a change from AAA (lysine) on Chr7A to TAA (STOP) on Chr7B in the C7_03400C gene. This gene is the orthologue of S. cerevisiae MTR4 that encodes an essential ATP-dependent RNA helicase involved in RNA processing in S. cerevisiae (22)(23)(24). In C. albicans SC5314, the HapA allele of MTR4 encodes a full-length Mtr4 protein, while the HapB allele carrying the STOP-introducing SNP encodes a truncated Mtr4 K880* protein that misses a C-terminal DSHCT domain, common to DEAD box helicases ( Fig. 2A).
The mtr4 K880* allele is responsible for the Chr7 homozygosis bias. To validate that the identified RLA candidate mtr4 K880* is truly responsible for the Chr7 homozygosis bias, the full-length MTR4 ORF was placed ectopically under the control of the P TDH3 constitutive promoter in strain CEC5061, giving rise to strain CEC5075 (Fig. 2B). Results shown in Fig. 1B and C confirmed that the DNA DSB-inducing system was functional in this strain and that the I-SceI-induced DNA DSB was predominantly repaired by gene conversion as observed in other instances. Strikingly, flow cytometry analysis revealed a significant elevation in the fold increase of mono-GFP cells obtained upon induction of I-SceI in strain CEC5075 by comparison to CEC5061 (Fig. 2C). Upon cell sorting and characterization of these mono-GFP cells, the majority (93%) appeared as true mono-GFP individuals that had become fully homozygous for Chr7B from the I-SceI TS to the BFP/GFP LOH reporter system. These cells were likely to have resulted from the repair of the induced DNA DSB by break-induced replication or mitotic crossover, indicating that complete homozygosis of the right arm of Chr7B is compatible with viability upon addition of a functional MTR4 allele (Fig. 2D). Similar to what was observed with the CEC5061 parental strain, I-SceI induction in the CEC5075 MTR4complemented strain resulted in an augmentation of mono-BFP cells (Fig. 2C). Eightyeight percent of these mono-BFP cells corresponded to cells where the DNA DSB had been repaired by gene conversion with crossover during G 2 (Fig. 2D).
Upon DNA DSB, major repeat sequences are a source of interhomolog recombination. Our initial strategy for unveiling RLAs was to induce a DNA DSB downstream of mrs-7b (the MRS located on the right arm of Chr7 [ Fig. 3A]) in order to ensure that the repeat sequences would not interfere with the DNA repair mechanisms. To seek validation of this initial assumption, the I-SceI TS was moved upstream of mrs-7b, between the centromere and mrs-7b, on either HapA or HapB and in the presence of an ectopic copy of MTR4 (Fig. 3B). By plating on 5-FOA-containing medium, we showed While no significant difference in augmentation of both populations is detected when the I-SceI TS is placed upstream of mrs-7b (CEC5072), a significant difference is observed when the I-SceI TS is placed downstream of mrs-7b (CEC5078). Significance was determined using a bilateral t test (P value). that the inducible I-SceI DNA DSB system was functional in the new location and that gene conversion was the preferred mechanism of repair, independently of the targeted haplotype ( Fig. 1B and C). Notably, when the I-SceI TS was localized on HapB upstream of mrs-7b in an MTR4-complemented strain (CEC5072), induction of I-SceI expression resulted in almost equal increases in the mono-GFP and mono-BFP cell populations relative to noninduced conditions (Fig. 3C). This contrasted to what was observed when the I-SceI TS was localized on HapB downstream of mrs-7b in an MTR4-complemented strain (CEC5078), whereby a greater increase in the mono-BFP cell population than mono-GFP cell population was observed upon I-SceI induction (Fig. 3C). This observation could be explained by interhomolog recombination at mrs-7b, which would link the I-SceI TS to either the BFP or GFP markers when the TS is located upstream of mrs-7b (Fig. 3B). This would result in equal proportions of mono-BFP and mono-GFP cells upon repair of the I-SceI-induced DNA DSB, keeping in mind that break-induced replication and mitotic crossover are the predominant molecular mechanisms leading to longrange LOH (Fig. 3C). In contrast, when the I-SceI TS is located downstream of mrs-7b, it remains linked to the GFP marker, regardless of interhomolog recombination at mrs-7b, thus predominantly yielding mono-BFP cells upon I-SceI-induced DNA DSB repair ( Fig. 3B and C). Overall, our data suggest that the MRS is a hot spot for interhomolog recombination upon DNA repair on the right arm of Chr7.
Of interest, C. albicans Chr7 is characterized by the occurrence of a second MRS on the left arm, namely, mrs-7a (Fig. 3A). Strains possessing the I-SceI TS between the centromere and mrs-7a and the BFP/GFP LOH reporter system on the left arm of Chr7 also exhibited an equal rate in mono-BFP and mono-GFP cells upon I-SceI induction, whatever the location of the I-SceI TS on HapA or HapB (data not shown), suggesting an increase in the number of cells that have undergone crossover at mrs-7a, linking the I-SceI TS to either the BFP or GFP marker. Thus, mrs-7a also appears to be a hot spot for interhomolog recombination on Chr7.

DISCUSSION
Previous studies have shown that C. albicans strains harbor recessive lethal alleles (RLAs) that are responsible for a bias upon homozygosis of certain chromosomes, whereby only one of the two homologs can be retained in the homozygous state (Table 1). Yet, the nature of these RLAs is generally unknown. In this report, we have ascribed the Chr7 homozygosis bias of the C. albicans laboratory strain SC5314 to a heterozygous SNP introducing a premature STOP codon in the MTR4 gene. Furthermore, we have unveiled the contribution of the major repeat sequence (MRS) to interhomolog recombination and hence, chromosome dynamics in C. albicans.
Our work focused on the homozygous bias observed for Chr7, suggesting the presence of at least one RLA on Chr7B (Table 1). Using a fluorescence-based LOH reporter system and an I-SceI-dependent DNA DSB-inducing system, we have shown that while long-range homozygosis of Chr7A does not affect cell viability, long-range homozygosis of Chr7B is nonviable in C. albicans strain SC5314. A library of SNPs compiled from 182 clinical C. albicans genomes (7) was searched for SNPs that would  Fig. 2A]). In S. cerevisiae, MTR4 encodes an ATP-dependent RNA helicase whose deletion results in nuclear accumulation of unprocessed RNAs (27), and reduction of function results in increased sensitivity to both benomyl and nocodazole (28). Importantly, it has been shown that the C-terminal domain of other fungal RNA helicases is critical for their proper RNAunwinding function (29), consistent with mtr4 K880* being a loss-of-function allele and homozygosity of this allele being lethal.
Our study and that of Feri et al. (13) indicate that the mining of the genomes of a large panel of C. albicans isolates for premature STOP-introducing SNPs is a suitable approach to identify RLAs responsible for chromosome homozygosis bias in C. albicans strains. As a reference, we provide a list of 70 genes that harbor a STOP-introducing SNP in one of the two alleles in the laboratory strain SC5314 in Table S1 in the supplemental material, including 12 alleles ( Table 2) that were never found in the homozygous state in a collection of 182 genome-sequenced isolates representative of the C. albicans population (7). The locations of these 12 candidate RLAs across the C. albicans genome are shown in Fig. 4. Subsequent Sanger sequencing will be necessary to confirm these alleles. Our reduced number of genes with SNPs introducing a premature STOP codon in strain SC5314 compared to Muzzey et al. (10) could be explained by the stringency of our analysis. Indeed, only positions with high SNP quality and coverage depth of Ͼ20 for all 182 strains of our collection were considered for further analysis.
A promising candidate to explain the homozygosis bias observed for Chr1 is the gcd6* allele (C1_08600C_B) that carries a SNP introducing a premature STOP codon, truncating the protein from the 17 C-terminal amino acids. In S. cerevisiae, GCD6 encodes the catalytic epsilon subunit of the translation initiation factor eIF2B. The truncated domain is found at the C terminus of several translation initiation factors and is important for mediating protein-protein interactions. GCD6 is essential in S. cerevisiae, as null mutants are not viable. The same is likely to be true in C. albicans, since heterozygous mutants are viable, but null mutants have not been reported in the literature. Therefore, it can be hypothesized that homozygosis of gcd6*, located on Chr1B, could result in cell death and be responsible for the homozygosis bias observed on Chr1 (Table 1).
Interestingly, a recent transposon mutagenesis screen in haploid C. albicans also argues that CaGCD6 is essential (30). Among the 12 candidate RLAs presented in Table 2, the same screen argues for essentiality of CaGPI16 (previously identified by Feri et al. [13]) and CaMTR4 (identified in this study). Although the 10 remaining genes in Table 2 were defined as nonessential genes by Segal et al. (30), it is puzzling to see that the mutated alleles with a premature STOP codon were never observed in the homozygous state in the natural population of 182 C. albicans isolates. This observation highlights the complementarity of the study by Segal et al. (30) and our work, given that gene essentiality can vary depending on growth conditions and in vitro assessment of gene essentiality does not necessarily correlate with in vivo essentiality.
From a mechanistic point of view, our work revealed that most of the cells that have undergone an I-SceI-mediated DNA DSB on Chr7 use gene conversion as a repair mechanism, thus limiting LOH extent and the loss of genetic information. Breakinduced replication, mitotic crossover, gene conversion with crossover and chromosome truncation that lead to long-range LOH events in decreasing order are less frequently utilized. Our results are consistent with those of Feri et al. (13) on Chr4, arguing that our observation is not locus specific but can be applied genome-wide in C. albicans. However, we cannot exclude the possibility that the relative usage of these repair mechanisms is specific to I-SceI-induced DNA DSB, i.e., DNA DSBs induced by the CRISPR-Cas9 RNA-guided endonuclease could be preferentially handled by another mechanism of DNA repair. In the course of this study, we also addressed the role that repetitive sequences, such as MRS, might play on the overall genome dynamics of C. albicans. This was achieved by studying how MRS position affects the outcome of I-SceI-induced LOH. Even though most C. albicans chromosomes possess a unique MRS region, the presence of two MRS regions, one on each arm, is unique to Chr7 (19). We observed that the presence of mrs-7b or mrs-7a between the I-SceI TS and the telomere-proximal LOH reporter system on Chr7 results in equal augmentation of the mono-BFP and mono-GFP populations upon I-SceI-dependent DNA DSB (Fig. 3). This contrasts with what is observed when the I-SceI TS is located downstream of the MRS, whereby the monofluorescent population arising by break-induced replication or mitotic crossover is increased relative to the monofluorescent population arising by gene conversion with crossover in the G 2 phase of the cell cycle (a scarce molecular mechanism), unless a RLA is present. This suggests that the MRS could be a hot spot for interhomolog mitotic crossover on Chr7. Indeed, upon recombination events at the MRS, DSBs repaired by break-induced replication or mitotic crossover would result in a relatively equal appearance of both monofluorescent populations when I-SceI-TS is located between the centromere and the MRS (Fig. 3B and C).
Mitotic crossovers at MRS could be an intrinsic feature of these repeat regions or be triggered by either (i) stress resulting from I-SceI overexpression or (ii) the physical I-SceI-induced DNA DSB. The former would imply that increased mitotic crossovers at MRS should be expected on all chromosomes remaining to be tested. The latter would imply that increased mitotic crossovers at Chr7 MRS are observed only upon repair of a DNA DSB upstream of these repeated regions. These latter hypotheses are consistent with a role for stress in the enhancement of the recombination frequency near the MRS or in general, as already suggested by Lephart et al. (20). Importantly, our results confirm the original proposal of Pujol et al. (31) that MRS on Chr7 are hot spots for recombination. Concretely, MRS would allow switching C. albicans haplotypes by generating a new combination of alleles. Information regarding the biological importance of MRS remains scarce despite the positive selection on MRS leading to the maintenance of these large and unique repeats in the C. albicans genome. Such recombination events provide C. albicans with increased opportunities to survive DNA DSBs whose repair can lead to homozygosis of recessive lethal or deleterious alleles. This might explain the maintenance of MRS in this species.  Table S1 using the assigned numbers in this figure.
Strains possessing the BFP/GFP LOH reporter system on the left arm of Chr7 were generated by positioning the heterozygous BFP/GFP locus close to the left telomere (position 108,458 to 108,838) and the I-SceI TS at position 367,060 to 367,145 between mrs-7a (position 228,342 to 242,083) and the centromere (position 425,808 to 428,708).
Following each transformation step, junction PCRs were conducted to ensure proper integrations. Additional controls were conducted such as fluorescence check (flow cytometry of 20,000 cells) and auxotrophy testing of strains (by spotting on SD supplemented with either Arg, His, Ura, or MPA). To ensure that the strains generated in the course of this study did not display growth defects resulting from transformation events, growth curves were generated using a TECAN Sunrise with OD readings (620 nm) every 10 min for a period of 48 h. Doubling times were outputted from the generated OD values and analyzed using GraphPad Prism 5.
Induction of the Tet-On system. In order to activate the Tet-On promoter and achieve I-SceI protein production, single colonies were precultured in liquid SC-His-Arg medium at 30°C. After overnight growth, induction was conducted in YPD plus anhydrotetracycline (ATc) (3 g/ml) (Thermofisher ACROS Organics) for 8 h at 30°C, followed by an overnight recovery in YPD.

5-Fluoroorotic acid selection.
Following the I-SceI induction protocol, as seen above, three different cell dilutions of cultures (20,000 cells, 2,000 cells, and 200 cells) grown in the presence (induced) or absence (noninduced; control) of ATc were plated on 5-fluoroorotic acid (5-FOA)-containing plates in triplicates. Dilutions were verified by plating a volume corresponding to 100 cells on YPD plates. The plates were incubated at 30°C for 3 days before analysis.
Cell preparation for flow cytometry and analysis. All flow cytometry analyses were conducted on the MACSQuant analyzer (Miltenyi Biotec) where BFP is detected with a 405-nm laser and 425-to 475-nm filters and GFP is detected with a 488-nm laser and 500-to 550-nm filters. Data for a maximum of 10 6 cells were analyzed using the FlowJo V10.1 software. The gates to determine the LOH frequencies were arbitrarily selected but conserved throughout sample analysis.
Cell sorting. Induced and noninduced cultures were filtered using BD Falcon Cell strainers in order to remove large debris and filamentous cells that could obstruct the tubing system of the cytometer. The MoFlo Astrios flow cytometer was used to analyze and sort the cells of interest. For each sorted gate, 1,000 cells were recovered in 400 l of liquid YPD medium, plated immediately after cell sorting on four YPD petri plates, and incubated at 30°C for 48 h before collection of results.
SNP-RFLP. In silico identification of heterozygous SNPs affecting a restriction site on only one haplotype of Chr7 was used for haplotype characterization of strains. First, a nucleotide multiple sequence alignment by MUSCLE was conducted using the Chr7A and Chr7B sequences from the reference strain SC5314 (22). Heterozygous SNPs were selected using the following criteria. (i) It interrupts a commonly known restriction site on one haplotype. (ii) The selected enzyme does not cut again within a range of 1 kb. (iii) The heterozygous SNP is present in most strains of the collection of 182 C. albicans clinical strains (7). Second, primer pairs were designed to result in PCR products with different digestion profiles and used to verify the presence of the heterozygous SNP in our strain of interest.
These SNPs were used for two distinct purposes: (i) to assign the Chr7 homolog targeted by I-SceI TS integration and (ii) to monitor the heterozygous status of the left and right arms of Chr7 upon DNA DSB repair. Regions of roughly 2.5 kb surrounding the heterozygous SNPs were amplified by PCR and digested with the appropriate restriction enzyme overnight. C. albicans gDNA extractions, PCRs, and amplicon verifications were conducted following the methods of Feri et al. (13), while a list of the primers used can be found in Table S3. The SNPs at positions 444,929 (BglII cutting HapA) and 727,328 (HpaI cutting HapA) were used to identify the I-SceI-targeted haplotype of the right arm of Chr7, when the I-SceI TS is located upstream and downstream of the mrs-7b, respectively. The 2.2-kb and 2.16-kb regions around the heterozygous SNPs located at positions 414,508 (left arm) and 744,964 (right arm) utilizing AttI (cutting HapB) and HaeII (cutting HapB) enzymes, respectively, were used to assess the heterozygous status of both Chr7 arms. All SNP-RFLP sites and the enzyme-sensitive haplotypes are summarized in Fig. S1.