Consequences of recombination for the evolution of the mating type locus in Chlamydomonas reinhardtii

Rationale Recombination suppression in sex chromosomes and mating type loci can lead to degeneration due to reduced selection efficacy and Muller’s ratchet effects. However, genetic exchange in the form of non-crossover gene conversions may still take place within crossover-suppressed regions. Recent work has found evidence that gene conversion may explain the low levels of allelic differentiation in the dimorphic mating type locus (MT) of the isogamous alga Chlamydomonas reinhardtii. However, no one has tested whether gene conversion is sufficient to avoid the degeneration of functional sequence within MT. Methods Here, we calculate levels of linkage disequilibrium (LD) across MT as a proxy for recombination rate and investigate its relationship to patterns of population genetic variation and the efficacy of selection in the region. Results We find that levels of LD predict selection efficacy across MT, and that purifying selection is stronger in shared genes than MT-limited genes to the point of being equivalent to that of autosomal genes. Conclusions We argue that isogamous systems without secondary sexual characteristics exhibit reduced selective pressure to differentiate sex chromosomes, and that recombination via gene conversion plays an important role in both reducing differentiation and preventing degeneration of crossover suppressed mating type loci.


Introduction
MT-gene mid that determines mating type specificity and the MT+ gene fus1 of the extent of recombination across the MT locus has been conducted, nor the 137 consequences of varying recombination across the region. 138 Here, we investigate the occurrence and extent of linkage disequilibrium across 139 the entire C. reinhardtii mating type locus and report an analysis of its effects on the 140 population genetics and molecular evolution of the shared and MT-limited regions. 141 Specifically, we address the following questions: 1) What are the spatial patterns  (Table S1). We used the C. reinhardtii v5.3 reference genome (Merchant et al., 2007) 156 but because it is derived from an MT+ individual and does not include organelles, 1 kb of one another. We only considered windows with a minimum of 30 polymorphic 191 sites. Next, we averaged these r 2 values with Z nS (Kelly, 1997)  to Z nS scores over the mating type locus using the R package depmixS4 (Visser and 200 Speekenbrink, 2010).

201
Identification, alignment, and analysis of coding sequence polymorphism. We 202 identified pairs or shared genes from the MT locus alleles using a multi-faceted 203 approach. Firstly, we used the gene IDs present in the genome annotation of C.
204 reinhardtii reference v5.3 (Merchant et al., 2007)  based on their translated amino acid sequence using MUSCLE v3.8.31 (Edgar, 2004) and then back translated each to its original nucleotide sequence.  (Lewontin, 1964)  of D between loci, such that a value of 1 always represents complete linkage.

258
In addition to organelle-to-mating type comparisons, intra-region D' was calcu-

266
All scripts used in this work can be found at https://github.com/aays/ 267 2019-mt-locus. All data analysis and visualization was performed in R (R Core 268 Team, 2018) using the tidyverse suite of packages (Wickham, 2017).   Robertson, 1966, Felsenstein, 1974). Hill-Robertson effects result 293 in a net reduction in linked neutral diversity due to either background selection or 294 selective sweeps. Thus, we expect nucleotide diversity (π) in shared genes to be 295 higher than that of MT -limited genes, which we found to be the case (π in shared 296 genes = 0.0168, pi in MT+ genes = 0.0037, π in MT-genes = 0.0022). We also find 297 that π in each mating type correlates negatively with Z nS , suggesting that genes that

302
To then test whether genes with higher recombination rates undergo more efficient 303 purifying selection, we calculated both the ratio of nonsynonymous to synonymous 304 diversity (π N /π S ) as well as Z nS in the coding regions of both shared and MT -limited 305 genes in the MT locus. π N /π S almost always ranges from 0 to 1, with lower values 306 suggesting more efficient selection; values above 1 are rare, and instead suggest that 307 balancing selection may be acting to preserve nonsynonymous diversity. Here, weighting by the number of sites in each gene, we find a weakly positive 309 relationship between Z nS and π N /π S , suggesting that regions of higher recombination 310 also show more efficient selection (Fig. 3b, R 2 = 0.106, p = 0.043). This result 311 excludes two genes, MT0796 and 522915, which showed extreme values of π N /π S

312
(1.85 and 3.35 respectively); including these outlier genes makes the fit marginally 313 nonsignificant (R 2 = 0.094, p = 0.051) but the direction of the relationship remains 314 the same. 315 We next examined whether recombination rate affected GC content in shared 316 genes. Either or both of GC-biased gene conversion or selection on base composition 317 are expected to drive increases in GC content in regions of higher recombination.

318
Here, we calculated GC content at intronic, intergenic, and four fold degenerate sites 319 (GC4) in 1 kb windows across shared regions, and found a significant but very weakly 320 negative relationship between GC content and Z nS (R 2 = 0.023, p = 0.00028, slope 321 = −0.048).   for selection on MT genes. We show that recombination rate predicts the degree to 374 which shared regions have differentiated across MT haplotypes, and that genes with 375 higher recombination rates exhibit greater selection efficacy against deleterious alleles.

376
Furthermore, we use patterns of LD breakdown to show evidence for biparental 377 inheritance of the chloroplast genome but not the mitochondrial genome.

378
Over the MT locus, we observe strongly elevated LD in the R domain, while LD 379 in the flanking T and C domains resemble autosomal levels (Fig. 1). The pattern we 380 show is expected, given that the T and C domains are highly syntenic while the R 381 domain contains inversions and autosomal translocations (Ferris and Goodenough, 382 1994, Ferris et al., 2002, De Hoff et al., 2013, resulting in recombination suppression 383 (Charlesworth, 1994, Wright et al., 2016. Although LD is elevated in the R domain, results (Boynton et al., 1987, Nakamura, 2010, Ness et al., 2016. In contrast, no such 398 evidence exists for the MT-genes and the mtDNA, supporting strict uniparental 399 inheritance of the mitochondria even over long evolutionary timescales. until allele-specific expression data across more shared genes are obtained. Finally, gene conversion across the MT locus is also expected to affect local GC content.

414
Recent work on the strength of gBGC in C. reinhardtii found a significant GC bias 415 in noncrossover gene conversions (Liu et al., 2017). Consistent with predictions from 416 gBGC, we found that GC content was highest in shared MT genes, and that the 417 MT -limited genes MID and FUS1 had extremely low GC content. 418 We predicted that selection against deleterious mutations across the MT locus 419 would be weaker in regions with less recombination due to interference between 420 selected sites (Hill and Robertson, 1966, Felsenstein, 1974, Comeron et al., 2008.

421
Indeed, we found that diversity levels were lower in shared genes with low recombi- correlation between π and Z nS is still present in both alleles (Fig. 3a). Furthermore,

428
MT regions with higher LD exhibited higher π N /π S ratios, indicative of reduced 429 purifying selection (Fig. 3b). Interestingly, the overall level of π N /π S in shared genes 430 was not significantly different from those of autosomal genes, suggesting that gene

445
Theoretical models predict reduced recombination between haploid sex chromo-446 somes if different sexes undergo differing selective pressures (Immler and Otto, 2015), 447 such as in anisogamous systems. The MT locus of V. carteri, for instance, is five times 448 larger than C. reinhardtii MT, more differentiated, and also carries more sex-specific 449 genes (Ferris et al., 2010). Differentiation between the V. carteri MT alleles likely 450 occurred after the transition to anisogamy (Hiraide et al., 2013, Hamaji et al., 2018, 451 suggesting that selection for recombination suppression followed the appearance of 452 selective pressure on sex-specific traits. Thus, given our findings that gene conversion 453 is a strong predictor of both selection efficacy and differentiation levels across C.   Table S1: Field strains of C. reinhardtii used in this study. All strains were obtained from the Chlamydomonas Resource Center (chlamycollection.org). Mating types of MT-strains CC-3059 and CC-3062 are mislabelled as MT+ on the Resource Center website, and are instead MT-individuals (R.J. Craig, personal communication)