To Repeat or Not to Repeat: Repetitive Sequences Regulate Genome Stability in Candida albicans

Genome instability often leads to cell death but can also give rise to innovative genotypic and phenotypic variation through mutation and structural rearrangements. Repetitive sequences and chromatin architecture in particular are critical modulators of recombination and mutability. In Candida albicans, four major classes of repeats exist in the genome: telomeres, subtelomeres, the major repeat sequence (MRS), and the ribosomal DNA (rDNA) locus. Characterization of these loci has revealed how their structure contributes to recombination and either promotes or restricts sequence evolution. The mechanisms of recombination that give rise to genome instability are known for some of these regions, whereas others are generally unexplored. More recent work has revealed additional repetitive elements, including expanded gene families and centromeric repeats that facilitate recombination and genetic innovation. Together, the repeats facilitate C. albicans evolution through construction of novel genotypes that underlie C. albicans adaptive potential and promote persistence across its human host.


Introduction
Considerable variability in genome organization exists across the tree of life, ranging from single circular chromosomes in bacterial species to thousands of linear chromosomes in some ferns [1]. These changes in chromosome complement arise through underlying processes of genome instability, giving way to structural reorganization and locus shuffling. Emergence of novel genetic and karyotypic arrangements can increase phenotypic diversity within a population and provide adaptive solutions to emerging selective pressures. Yet, extreme forms of genome plasticity can lead to DNA fragmentation, chromosome loss, and cell death. Thus, a balance between prohibitive and unrestricted genome instability is required to maintain DNA integrity while allowing evolution to produce novel and potentially advantageous genotypes.
Repetitive regions of the genome often serve as hotspots of genome rearrangements and evolutionary innovation. These elements can exist as single or multi-copy loci encoding tandemly-repeated identical or near-identical sequences of variable length ranging from one to thousands of nucleotides. By their very nature, repetitive loci are prone to recombination, producing insertion/deletions (indels) and translocations between both adjacent and distal repeats. Recombination within repeats alters copy number variants in addition to promoting de novo mutations due to errors in DNA exchange and repair [2]. In contrast, recombination between distal, non-allelic elements can result in large-scale chromosomal aberrations such as fusion events, truncations, and translocations. Genetic exchange between repeats does not require perfect sequence identity and commonly occurs between imperfect repetitive sequences [3]. Repetitive loci are also particularly prone to replication fork collapse and environmental damaging agents, including oxidative stress and ultraviolet irradiation, that promote further mutation and recombination [4][5][6]. Condensed heterochromatin across repetitive regions of the genome is thought to help shield these regions from

Telomeres and Recombination
C. albicans maintains the ends of its linear chromosomes through the use of telomeres that are composed of repeating subunits produced by the enzyme telomerase. The telomeric repeats of most eukaryotes are built from a repeating 6 bp GT motif, but many fungal species including the closely-related Saccharomycotina yeasts use repeats of diverse lengths and complexity [33]. In C. albicans, the telomeric repeat is 23 bp (5'-ACTTCTTGGTGTACGGATGTCTA-3'), and has diverged substantially from other Candida species by increasing in both length and complexity [34]. Other Candida CUG paraphyletic species, such as Candida guillermondii and Debaryomyces hansenii, decode the "CUG" codon as serine instead of leucine and use shorter telomeric repeats of 5'-ACTGGTGT-3' and 5'-ATGTTGAGGTGTAGGG-3', respectively [35,36]. Telomerase reverse transcriptase (TERT) and the telomerase RNA component (TERC) work together to extend telomeres by adding single 23 bp repeats to nascent chromosome ends to form the full telomere, which can range in size from 500 bp to 5 kilobases (kb) in C. albicans laboratory isolates [37,38].

Telomere Structure and Maintenance
The telomerase complex, which includes TERT and TERC, forms the basic functional unit required for telomere repeat production and maintenance [39][40][41]. Both TERC and TERT, encoded by TER1 and EST2, respectively, perform repeated rounds of reverse transcription to build G overhangs as part of the end-protective T-loops that prevent telomere attrition [42]. More specifically, Est2 uses the TER1 RNA as a template to add telomeric repeat subunits to the chromosome end in order to maintain replicative capacity and avoid cell senescence. The dual roles of end protection and telomere repeat addition by Est2 are separable. A catalytically inactive Est2 retains the ability to suppress excessive G-strand accumulation despite being unable to add telomeric repeats [42]. Two additional telomerase protein subunits, Est1 and Est3, also contribute to appropriate telomere structural maintenance by suppressing excessive recombination and preventing telomere loss although their precise molecular mechanisms remain obscure [43].
Maintenance of telomeric repeats in C. albicans requires the heterotrimeric CST complex comprised of Cdc13, Stn1, and Ten1. Loss of any of these factors results in dysregulation of telomere length, resulting in either overextended or shrunken telomeres [37]. This contrasts with inactivation of the telomerase complex that only leads to iterative telomere shortening due to the inability to add telomeric repeats. The canonical Ku70/Ku80 complex also controls telomere length and loss of either gene produces heterogeneous telomere lengths indicative of dysfunctional telomerase activity [44,45]. Rap1 (repressor/activator protein 1), an essential protein in telomere maintenance in other budding yeasts, is surprisingly not essential in C. albicans despite also functioning in telomere length regulation [46,47].

Telomere Replication and Recombination
Telomere maintenance through recombination serves a critical function in chromosome stability but can also contribute directly to genome instability during telomere length shortening. As such, the actions of telomere length control and protection from overhang accumulation must be tightly controlled. Telomere length can be regulated through addition of telomeric repeats following their loss during DNA replication or via recombination within single telomeres or between chromosomes [48]. Alternative lengthening of telomeres (ALT), a process that is telomerase-independent but recombination-dependent, adds telomeric repeats to chromosome ends by using pre-formed telomeres as a template [49]. The 3' overhang of the extending telomere invades other telomeric DNA that can be either: within the same telomere through telomere looping, another chromosome's telomere, or an extrachromosomal telomere circle. Telomere circles (t-circles), autonomous plasmids composed entirely of telomere repeat subunits, can be produced as a byproduct of telomere length regulation via intra-chromosomal recombination events [50]. In C. albicans, t-circle frequency increased following loss of Ku70, suggesting these proteins suppress intra-telomeric recombination [44]. However, no direct studies of ALT have been performed in C. albicans despite the presence of homologous genes that appear capable of conducting these functions (e.g., Rad52 [45,49,51]).
Conserved telomeric proteins are required for telomere length regulation. For example, Rap1 likely plays a role in this process, as its deletion produced aberrant telomere repeat-containing DNA structures and t-circles [47]. Yet, all observed nuclear telomere lengthening events in C. albicans have occurred through concatenation of linear elements by TER1 [52], leaving the function of Rap1 in t-circle production and ALT unclear. Our current understanding of the recombination machinery (e.g., Rad52, etc.,) and precise mechanisms for telomere maintenance are severely deficient in C. albicans as little direct experimental work has been performed in recent decades. Investigations into C. albicans telomere biology, in particular, stands out for its potential scientific gains given the paucity of existing data combined with the common involvement of telomeres in contributing to genome plasticity. Newly developed linear plasmids containing in C. albicans share similar mechanisms of linear telomere formation and maintenance for their retention and propagation and may present as a simplified model to study telomere dynamics in C. albicans [53].

Candida albicans Subtelomeres Are Hotspots of Genomic Rearrangements
Subtelomeres are defined as the genomic regions adjacent to telomeric repeats that are enriched for repetitive genetic elements [54]. Repetitive sequences within the subtelomeres include intact transposable elements and expanded gene family members as well as remnants of these functional units following gene disruption or inactivation. Formation of heterochromatin conducive to epigenetic silencing is also a common feature of subtelomeres, although it may be interspersed with euchromatic regions surrounding intact genes. In C. albicans, the histone deacetylase Sir2 maintains the heterochromatin within subtelomeres [55,56]. Surprisingly, Sir2 appears to perform these roles without other Sir complex proteins typically required for subtelomeric silencing through "telomere position effect", as they are absent from the C. albicans genome [57]. The SIR2 paralog, HST1, also regulates subtelomeric gene expression, although these effects are less uniform across C. albicans subtelomeres with only some loci being affected in a ∆/∆hst1 background [58]. These combined effects of heterochromatin formation and gene silencing in C. albicans were recently shown to be strongest within the proximal 10-15 kb of the telomere repeats [59]. As such, we choose to define the C. albicans subtelomeric region as the most telomere-proximal 15 kb of each chromosome arm.

Subtelomeric Gene Functions
C. albicans subtelomeres are relatively gene poor compared to chromosome-internal regions of the genome. Approximately 50 protein-coding genes are annotated within C. albicans subtelomeres in the genome reference strain SC5314 based on the above definition from the Candida Genome Database [60], but most genes remain largely uncharacterized ( Table 2). Most of the subtelomeric genes with functional assignments are either associated with biofilm formation (23 of 55), growth in Spider or other hyphae-inducing media (22 of 55), or have predicted roles in metabolism (21 of 55; Table 2). Coincidently, many of these genes are also assigned roles in virulence either by directly promoting pathogenicity or through adaptive responses promoting persistence across host niches. Genes that may be used in niche adaptation include multiple transporters, cell wall proteins, and metabolite utilization genes.   Approximately one quarter of C. albicans subtelomeric genes (13 of 55) belong to the telomere-associated (TLO) gene family (Figure 2). TLO genes underwent a recent lineage-specific expansion from a single gene in most Candida species to 14 paralogs in C. albicans and are the only gene family with a significant presence within the subtelomeres [61][62][63]. Each TLO contains a MED2 domain, indicating their function as homologs of the Med2 subunit in Mediator, the major eukaryotic transcriptional regulatory complex [63]. TLOs participate in regulating a variety of virulence traits, including growth, resistance to stressors, and biofilm formation [64,65]. Strikingly, individual TLO genes regulate distinct virulence properties despite sharing >98% nucleotide identity between individual paralogs.

Repetitive DNA Elements in the Subtelomere
Repetitive elements commonly cluster in subtelomeric regions where they can buffer gene-rich chromosomal interiors from the detrimental effects of telomere length variation [67]. In particular, retrotransposons are often enriched within subtelomeres as new insertions are expected to incur less of a fitness cost in these gene-poor regions and can provide selective advantages by rescuing telomerase-deficient cells [68,69]. Over 350 unique retrotransposon insertions by 34 distinct families of transposable elements have been identified in the C. albicans SC5314 genome [70], which can be distinguished by the unique sequences of flanking long terminal repeats (LTRs). Of these 34 families, seven LTR families are found within the C. albicans subtelomeres, including an intact copy of Zorro2, the only complete non-LTR retrotransposon in the subtelomeres (Table 3) [70]. Thirteen individual LTRs are annotated in the subtelomeres,~2.5x higher than the genome average. Importantly, LTR sequences incorporated as part of functional genes (e.g., TLOs) are not included among annotated repetitive sequences, suggesting these frequencies are likely an underestimate. Most evidence for transposon integration events in the subtelomeres comes from abandoned LTR repeats that mark previous insertions which subsequently reactivated the intervening transposase to excise itself and reintegrate elsewhere in the genome. In addition to inducing genotypic variation via continual transposon movement, highly abundant non-LTR retrotransposons in C. albicans can generate genetic diversity through recombination between dispersed LTR sequences [71]. Additional sequence elements, detected through molecular investigations, exist in C. albicans subtelomeres. These elements, the Bermuda Triangle sequence (BTS) and the TLO recombination element (TRE), lie immediately centromeric to TLO genes ( Figure 2). The BTS is defined by a 50 bp sequences which share~88% sequence similarity across the 11 BTS-containing subtelomeres [72]. Each BTS is encompassed within the longer TLO recombination element (TRE). The TRE stretches~300 bp, beginning immediately following the TLO coding sequence and extending towards the centromere [59]. While no clear phylogenetic hierarchy is observed among the BTS sequences, TREs can be organized into three groups based on near complete sequence identity within any single cluster. Both the BTS and TRE overlap with a member of the tui-class family of LTR sequences which are annotated in some subtelomeres but missing in others (Figure 2). Given the high sequence homology across the BTS and TRE, it is likely this LTR element is present next to all TLO genes with the exception of ChrRR and Chr7R, which both lack the BTS and TRE [56].

Subtelomeres Are Prone to Mutation
C. albicans subtelomeres experience high rates of recombination, including frequent loss of heterozygosity (LOH). LOH events result from non-reciprocal crossing over or chromosome loss and reduplication of the lost region from the remaining chromosome homolog. Rates of LOH increase along all chromosomes towards the telomeres of C. albicans isolates, indicating a general relationship between the distance from the centromere and allelic homozygosis mediated by recombination [73] (Figure 3). Yet, subtelomeric LOH rates increase an order of magnitude compared to centromere-proximal regions interior to the TRE and loss of the TRE in subtelomeres results in a significant decrease in LOH [56]. Integration of the TRE into an exogenous locus increased LOH rates of an adjacent selectable marker [56]. Thus, the TRE sequence is crucial to the elevated recombination rates in C. albicans subtelomeres where it suppresses mitotic recombination additively with Sir2 silencing. However, some stressors, including fluconazole, promote such a strong genome instability phenotype that the protective effects of Sir2 can be masked in their presence [56]. Rates of loss of heterozygosity (LOH) increase towards chromosome ends. An average LOH rate for each genomic region is depicted for chromosome internal sequences (black), the subtelomeres (yellow), and telomeres (orange) based on published (chromosome internal and subtelomeric) [28,56] and unpublished results (telomeric). All data is derived from LOH assays in which a URA3 marker was inserted within different genomic regions and its location determined by either sequencing or contour clamped homogenous electric field (CHEF) gel analysis and Southern blotting. Centromeres are indicated by grey circles.
Phosphorylation of serine 129 on histone H2A, commonly known as γH2A, is a common marker for heterochromatin in S. cerevisiae [74]. Consistent with localizing to heterochromatic regions, γH2A is enriched at subtelomeres, but is also found at origins of replication and convergent genes where it labels double-strand DNA breaks. More specifically, γH2A abundance is statistically enriched on 13 of the 16 C. albicans subtelomeres (all except ChrRR, 1R, and 7R). Its conspicuous absence in select subtelomeres is thought to be due to incomplete assembly in the current genome assembly and not the absence of γH2A [75]. These γ-sites mark putative recombination-prone genomic loci that are intrinsically more fragile and susceptible to DNA damage [75].
Subtelomeric recombination directly affects their genic repertoire. Continuous passaging of SC5314-derived strains for 30 months led to the gain and loss of specific TLO genes and all telomere-proximal sequences through non-reciprocal recombination events among different chromosome arms [72]. These non-reciprocal exchanges occurred once every 5000 generations on average, with some chromosome arms being favored as either DNA donors or recipients. Bias in amplification and loss of specific subtelomeric sequences due to non-reciprocal DNA exchange suggests that selection favored expansion or loss of certain genomic regions during passaging. Approximately half of the detected recombination events occurred within the BTS element with two additional events initiating in the TRE. Two recombination events took place within TLO genes and produced novel TLO sequences, highlighting the potential for significant sequence diversity to arise via subtelomeric recombination that can contribute to phenotypic plasticity [72].
Genes within the subtelomeres are subject to not only increased rates of recombination, but also increased formation of indels [73]. TLOs can be separated into three clades (α, β, and γ) based on sequence diversity that primarily clusters in the 3' end of the coding sequence. The TLOβ-clade is unique in containing two major indel events in this 3' region that distinguish it from the other TLO clades. The 3' end of TLOγ-clade genes uniformly contains a rho LTR sequence that likely disrupted an ancestral TLOα-clade homolog which then expanded across the subtelomeres through non-reciprocal exchange of this newly-formed TLOγ-clade gene [62]. Further demonstrating the evolutionary potential of continual retrotransposon activity, a subsequent psi LTR-containing retrotransposon insertion disrupted TLOγ4 in the SC5314 background, producing a truncated TLOγ4 gene, which is actively transcribed, and a TLOψ4 pseudogene with no detectable transcript [62,72].

The C. albicans Major Repeat Sequence
Seven of the eight C. albicans chromosomes, the exception being Chr3, contain large tracts of repetitive DNA known as the MRS. These regions are comprised of large, variable arrays of the 2 kb repetitive sequence (RPS) unit that are often flanked by the non-repetitive RB2 (6 kb) and HOK (8 kb) elements [21,76]. RPS-1, the main unit of the MRS, is itself comprised of smaller repeat segments of about 80-170 bp, which are then assembled into two families of ordered segments, REP1 and REP3, which both contain the 29 bp COM29 common sequence (5'-CAAAAAAGGCCGTTTTGGCCATAGTTAAG-3') [77]. Altogether, C. albicans possesses nine complete MRS loci, 14 RB2 elements, and two HOK elements ( Figure 4). MRS loci contain rare 8-bp SfiI restriction sites that have been used extensively to separate chromosome arms for mapping genetic loci prior to construction of the genome reference sequence and for detection of chromosomal rearrangements by contour clamped homogenous electric field (CHEF) electrophoresis [78].

Impacts of the MRS on Chromosome Loss and Recombination
Despite the highly nested repeat structure of the MRS sites, recombination frequencies in the MRS are equivalent to the genome average rates of recombination. Thus, the MRS is unlikely to function as a recombination hotspot, but instead often contributes to translocations between heterologous chromosomes [79]. Insertion of a URA3 selectable marker into the RB2 element of various MRS sites allowed translocation events to be tracked that involved multiple different chromosomes as well as truncations of Chr7 in a few instances [80]. Broad karyotypic surveys across clinical isolates have repeatedly identified chromosomal translocations involving the MRS of different chromosomes, suggesting that these chromosomal rearrangements are rare but stably maintained over longer evolutionary time frames by C. albicans isolates [81][82][83]. In particular, the WO-1 strain, in which the white-opaque cell state switch was first described, contains three chromosomal translocations that are coincident with the MRS of different chromosomes [84,85].
Intrachromosomal recombination also occurs within the MRS. The XhoI restriction enzyme cuts chromosomal DNA on either side of the MRS that can be used to define MRS size by Southern blotting with a RPS-specific probe [79]. Changes in MRS size can also be observed by CHEF electrophoresis of chromosome homologs that resolve from each other due to repeat expansion or contraction. Detection of these events is more readily observable for smaller chromosomes (e.g., Chr7) that have greater length resolution [83]. Loss of the MRS through targeted deletion suggests that these repeats are not required for gross viability in C. albicans [86]. Yet, chromosome loss due to nondisjunction was directly related to MRS size based on selection for loss of Chr5 by growth in sorbose [86]. Chr5 homologs encoding the smaller MRS were preferentially retained compared to homologs with longer MRS regions [79,86].

The rDNA locus
C. albicans rDNA repeats are encoded as an array on the right arm of ChrR at the RDN1 locus. The rDNA locus encompasses the 18s, 5.8s, 25s, and 5s rRNAs that are organized as tandem repeating units. This locus ranges in size from 11.6 kb to 12.5 kb and shifts between 21 and 176 copies of repeating rDNA units across strains and growth conditions [87]. Consequently, the full RDN1 locus can vary between 244 and 2200 kb or approximately 10% of the total size of the C. albicans genome. The presence of the rDNA locus and such massive shifts in chromosome size underlie the distinct name of the encoding chromosome, ChrR [88]. Naming of this chromosome breaks from the numbering system based on length used for all other nuclear chromosomes, as changes in rDNA copy number can greatly alter its size.

Recombination between rDNA Repeats
Unequal intrachromosomal recombination events within the rDNA repeats are frequent and produce the large-scale shifts in RDN1 size. Changes in rDNA repeats contribute to approximately 92% of C. albicans chromosomal variation and instability in colony morphology mutants [87]. The frequency of these recombination events lacks precise quantification but can be observed within five cell divisions by next-generation sequencing (unpublished data). Extrachromosomal rDNA circles and linear rDNA plasmids can also be released during recombination [89,90]. The monopolin complex, shown to be important for regulating rDNA and telomeric repeats in S. cerevisiae [91], also contributes to C. albicans rDNA maintenance. Strains lacking monopolin encoded shorter and more variably-sized rDNA, implicating monopolin in specifically maintaining rDNA length [92].

Additional Repetitive Sequences Shape Candida albicans Genomic Stability
Although telomeres, subtelomeres, the MRS, and the rDNA locus are all clearly defined genomic regions in C. albicans that play important roles in genome stability, multiple additional repetitive sequences throughout the genome promote recombination and shape genome evolution.

The Agglutinin-Like Sequence (ALS) Family of Adhesins
Critical to C. albicans' success within the human host is its ability to adhere to a wide range of substrates. Colonization and adherence are mediated, in part, by the ALS gene family of adhesins. ALS genes family members are encoded by eight distinct loci that each produce a glycoprotein capable of forming amyloid-like aggregates [93,94]. Within each ALS locus, a conserved Ser/Thr-rich domain provides a platform for intergenic recombination between ALS homologs and production of new genetic variants [93][94][95][96]. Indeed, ALS51 is a product of recombination between the ALS1 and the ALS5 loci that is present in multiple clinical isolates of C. albicans. Vast allelic differences can also arise through intragenic recombination. Hypermutability of the ALS7 ORF has produced 60 ALS7 alleles identified across a collection of 66 strains [97]. Variant alleles arise via frequent recombination within tandem repeats and conserved 3' repetitive domains. Mutation and LOH rates within this gene family are also remarkably high, comparable to that within the MRS, subtelomeric regions, and the TLO gene family [24]. Thus, recombination within and among these cell surface antigens allows for rapid gene turnover and adaptation via cellular adherence to environmental substrates that can promote colonization and persistence across body sites [98].

Genome Evolution through Centromere-Associated Repeats
Centromeres in C. albicans are unique, gene devoid stretches of DNA typically flanked by inverted repeats [60,99]. Recombination between these flanking repeats can lead to the formation of isochromosomes, which are chromosomes comprised of two identical chromosome arms joined by a functional centromere. Formation of an isochromosome of Chr5L (i5L) commonly follows C. albicans exposure to azole-class antifungal drugs [29]. Amplification of the left arm of Chr5 confers azole resistance through the acquisition of extra copies of ERG11 and TAC1, encoding the canonical azole drug target and the transcriptional activator of efflux pumps, respectively [100]. Recombination between the centromeric repeats flanking either side of the Chr5 centromere is responsible for producing i5L [29]. Similarly, clinical isolate P78042, which was isolated with a Chr4 trisomy [25], produced an isochromosome of the right arm of Chr4 during passage in the presence of FLC for 100 generations [101]. These examples highlight the importance of recombination and isochromosome formation through centromere-associated repeats for adaptation of C. albicans to antifungal drugs [101].

Recombination Facilitated by Cryptic Long Repeats
A recent scan of the C. albicans SC5314 genome identified 1974 long repeat sequences, excluding the rDNA, the MRS, subtelomeric repeats, and previously characterized complex tandem repeats [101].
Called repeats were required to have sequence matches of 20 bp or longer, which occur more than once in the genome and accounted for 2.87% of the haploid reference genome altogether. Repeat location did not correlate with GC content or ORF density and many repeats encompassed complete and actively transcribed genes [101]. Recombination at these long repeats produced strains harboring copy number variations, LOH, and chromosomal inversions. Consequently, repetitive sequences scattered throughout the genome must be included as additional arbiters of genome instability [101].

Conclusions
Thus, repetitive sequences in the C. albicans genome promote genome instability through various mechanisms that can range from minor changes in sequence length to massive chromosomal rearrangements. Previous work has focused on defining the characteristics of four major repetitive regions (telomeres, subtelomeres, the MRS, and the rDNA locus), but their contributions to genetic and phenotypic diversity still remain generally unresolved. Furthermore, the importance of less well-understood repetitive sequences to genome evolution and karyotypic innovation are just beginning to emerge. What is clear is that repetitive elements within the C. albicans genome produce novel genotypes through changes in DNA sequences or karyotypes that often prove to be beneficial. Some of these changes, such as isochromosome formation during antifungal drug exposure and changes to gene family copy number, are more intuitive, whereas the function of others involving changes in MRS repeat length and rDNA copy number are less easy to grasp. It could be that repeat length modulates the probability of recombination occurring or that the presence of repeats is itself sufficient to promote genome instability. Association of additional repetitive sequences at major regions within the genome (e.g., telomeres, the MRS) suggest that these sites are likely under relaxed selection for new insertions that increase the likelihood of additional recombination events. In contrast, repeats scattered throughout the genome beyond these defined regions may function as back-up sites for genome instability. These cryptic repetitive sequences throughout the C. albicans genome may not only promote recombination but reduce the associated fitness costs of novel genotypes by increasing the resolution of affected sequences to only the locus (loci) that is under selection. Regardless, these repetitive sequences collectively construct a dynamic environment of common and infrequent sites of genomic instability to generate genotypic and phenotypic variants that allow C. albicans adaptation to its diverse niches across the human host and ultimately promote its success as a commensal and pathogen.