**5. In humans and rodents, defects in HRR enzymes lead to infertility, as would be expected if removal of DNA damages is an essential function of meiosis**

About 15% of all couples in the US are infertile, and an important cause of male infertility appears to be oxidative stress during gametogenesis (Makker et al., 2009). During spermatogenesis in the mouse, DNA repair capability declines after meiosis is complete, allowing accumulation of DNA damage (Marchetti & Wyrobek, 2008). Lewis & Aitken (2005) reviewed evidence that DNA damages in the germ line of men are associated with poor semen quality, low fertilization rates, impaired pre-implantation development, increased abortion, and elevated incidence of disease in the offspring including childhood cancer. They noted that the natural causes of this DNA damage are uncertain, but the major candidate is oxidative stress. On the hypothesis that meiosis is an adaptation for DNA repair, it is expected that loss of ability to repair DNA damages during meiosis would have adverse effects, including infertility. Although the finding of such adverse effects is expected on the hypothesis that meiosis is an adaptation for repairing naturally caused DNA

Meiosis as an Evolutionary Adaptation for DNA Repair 363

major classes of meiotic recombination. If, during recombination, the chromosome arms on opposite sides of a DSB exchange partners, the recombination event is referred to as a crossover (CO). If the original configuration of chromosome arms is maintained, the recombination event is referred to as a non-crossover (NCO) (see Figure 1). The relative occurrence of NCO or CO recombination events is relevant to evolutionary theories of meiosis which assume producing genetic variation is the function of meiosis. NCO events have little effect on linkage disequilibrium (the statistical association of genes at different loci) and so produce very little genetic variation in terms of new combinations of genes.

Data based on tetrad analysis from several species of fungi indicates that the majority (about 2/3) of recombination events during meiosis are NCOs [see Whitehouse (1982), Tables 19 and 38, for summaries of data from *S. cerevisiae*, *Podospora anserine*, *Sordaria fimicola* and *Sordaria brevicollis*]. More recent work also supports a bias towards NCOs during meiosis. In mouse meiosis there are > 10-fold more DSBs than CO recombinants (Moens et al., 2002), suggesting that most DSBs are repaired by NCO recombination. In *D. melanogaster* there is at least a 3:1 ratio of NCOs to COs (Mehrotra & McKim, 2006). These observations indicate that the majority of recombination events are NCOs. These NCOs involve informational exchange between two homologs but not physical exchange, and little genetic variation is created. Thus explanations for the adaptive function of meiosis that focus exclusively on

Andersen & Sekelsky (2010) have argued that a common mechanism called "synthesis dependent strand annealing" (see section 7, below) is employed in both meiotic HRR of the NCO type and mitotic HRR (which is largely of the NCO type), and thus meiotic and mitotic NCOs probably have a similar function. Substantial evidence indicates that HRR during mitosis is an adaptation to repair DNA damages that originate from diverse endogenous and exogenous sources (e.g. endogenous ROS from oxidative metabolism and exogenous Xrays, UV, chemical carcinogens) (see examples in Table 1; also Lisby & Rothstein, 2009). Thus NCO recombination during meiosis, as in mitosis, likely functions to repair of DNA

**7. NCO recombination likely occurs by synthesis-dependent strand annealing**  Molecular models of meiotic recombination have evolved over the years as relevant evidence accumulated. The model that has been most influential in recent decades has been the Double-Strand Break Repair model (Szostak et al. 1983). By this model, during each recombination event two Holliday Junctions (HJs) are formed and resolved (see Figure 1). Thus the Double-Strand Break Repair model can also be referred to as the Double Holliday Junction (DHJ) model. The DHJ model was considered to provide an explanation for both CO and NCO types of recombination events. However, Allers & Lichten (2001) showed that, although CO recombinants are likely formed by a pathway involving resolution of Holliday junctions, NCO recombinants arise by a different pathway that acts earlier in meiosis. Allers & Lichten (2001), McMahill et al. (2007) and Andersen & Sekelsky (2010) have presented evidence that NCO recombinants are generated during meiosis by an HRR repair process referred to as "Synthesis-Dependent Strand Annealing" or "SDSA" (see Figure 1). During SDSA the invading strand from a chromosome with a DSB is displaced from the D-loop structure of an intact chromosome and its newly synthesized sequence anneals to the other side of the break on the chromosome with the original DSB. This process can accurately

However, CO and NCO events are equivalent from the point of view of HRR.

crossing over are inadequate to explain the majority of recombination events.

damages from diverse sources.

damages, this finding does not prove the hypothesis. Another possibility is that during meiosis damages are introduced in a programmed fashion, leading to HRR. Such HRR may be necessary for proper pairing and segregation of chromosomes, and this process may be required for fertility (see section 8 below).

Inherited mutations in genes that specify proteins necessary for HRR cause infertility (Table 2) indicating that production of functional gametes depends on HRR. Genes *brca1, atm*, and *mlh1* are expressed in mitosis, but at a higher level in meiosis, and gene *dmc1* is expressed exclusively in meiosis (Table 2).


Table 2. Mutant genes defective in HRR that cause infertility in human and/or mouse

Brca1 functions during both meiotic and mitotic recombination. The inheritance of a mutant *brca1* allele substantially increases a woman's lifetime risk for developing breast or ovarian cancer due to a deficiency in HRR of DNA DSBs in somatic cells. Male *brca1* defective mice are infertile due to meiotic failure during spermatogenesis (Table 2), indicating that HRR is necessary during meiosis.

The Atm protein acts during both meiotic and mitotic recombination in detection and signaling of DSBs, and is necessary for fertility of females and males in both humans and mice (Table 2). Gametogenesis is severely disrupted in Atm-deficient mice as early as the leptonema stage of prophase I, resulting in apoptotic degeneration (Barlow et al., 1998).

Mismatch repair protein Mlh1 (homolog of *E. coli* MutL) is necessary for meiotic recombination (Wei et al., 2002). Mutation in the *mlh1* gene causes blockage at the pachytene stage of meiosis and female and male infertility (Table 2).

*Dmc1* is a meiosis specific gene. Dmc1 protein (a homolog of *E. coli* RecA protein) functions during meiotic recombination to promote recognition of homologous DNA and to catalyze strand exchange. *Dmc1* deficient female and male mice are infertile due to arrest of gametes in meiotic prophase (Table 2).

The evidence reviewed in this section indicates that defective HRR of DNA damages during meiosis causes infertility.

#### **6. Non-crossover (NCO) recombination during meiosis is likely an adaptation for DNA repair**

Meiotic recombination appears to be a near universal feature of meiosis [although it may be absent in some situations, such as in *Drosophila* males (Chovnick et al., 1970)]. There are two

damages, this finding does not prove the hypothesis. Another possibility is that during meiosis damages are introduced in a programmed fashion, leading to HRR. Such HRR may be necessary for proper pairing and segregation of chromosomes, and this process may be

Inherited mutations in genes that specify proteins necessary for HRR cause infertility (Table 2) indicating that production of functional gametes depends on HRR. Genes *brca1, atm*, and *mlh1* are expressed in mitosis, but at a higher level in meiosis, and gene *dmc1* is

*brca1* Mouse 3× male mice are infertile Galetzka et al., 2007;

are infertile

are infertile

are infertile

Brca1 functions during both meiotic and mitotic recombination. The inheritance of a mutant *brca1* allele substantially increases a woman's lifetime risk for developing breast or ovarian cancer due to a deficiency in HRR of DNA DSBs in somatic cells. Male *brca1* defective mice are infertile due to meiotic failure during spermatogenesis (Table 2), indicating that HRR is

The Atm protein acts during both meiotic and mitotic recombination in detection and signaling of DSBs, and is necessary for fertility of females and males in both humans and mice (Table 2). Gametogenesis is severely disrupted in Atm-deficient mice as early as the leptonema stage of prophase I, resulting in apoptotic degeneration (Barlow et al., 1998). Mismatch repair protein Mlh1 (homolog of *E. coli* MutL) is necessary for meiotic recombination (Wei et al., 2002). Mutation in the *mlh1* gene causes blockage at the pachytene

*Dmc1* is a meiosis specific gene. Dmc1 protein (a homolog of *E. coli* RecA protein) functions during meiotic recombination to promote recognition of homologous DNA and to catalyze strand exchange. *Dmc1* deficient female and male mice are infertile due to arrest of gametes

The evidence reviewed in this section indicates that defective HRR of DNA damages during

**6. Non-crossover (NCO) recombination during meiosis is likely an adaptation** 

Meiotic recombination appears to be a near universal feature of meiosis [although it may be absent in some situations, such as in *Drosophila* males (Chovnick et al., 1970)]. There are two

4× females and males in

Table 2. Mutant genes defective in HRR that cause infertility in human and/or mouse

Infertility in mutant females/males

both humans and mice

female and male mice

References

Cressman et al., 1999

Galetzka et al., 2007; Barlow et al., 1998

Galetzka et al., 2007; Wei et al., 2002

Pittman et al., 1998

required for fertility (see section 8 below).

expressed exclusively in meiosis (Table 2).

expression in testes vs. somatic

*mlh1* Mouse 1.7× female and male mice

meiotic cells

stage of meiosis and female and male infertility (Table 2).

cells

Gene Species Fold-increased

*atm* human,

mouse

necessary during meiosis.

in meiotic prophase (Table 2).

meiosis causes infertility.

**for DNA repair** 

*dmc1* Mouse specific for

major classes of meiotic recombination. If, during recombination, the chromosome arms on opposite sides of a DSB exchange partners, the recombination event is referred to as a crossover (CO). If the original configuration of chromosome arms is maintained, the recombination event is referred to as a non-crossover (NCO) (see Figure 1). The relative occurrence of NCO or CO recombination events is relevant to evolutionary theories of meiosis which assume producing genetic variation is the function of meiosis. NCO events have little effect on linkage disequilibrium (the statistical association of genes at different loci) and so produce very little genetic variation in terms of new combinations of genes. However, CO and NCO events are equivalent from the point of view of HRR.

Data based on tetrad analysis from several species of fungi indicates that the majority (about 2/3) of recombination events during meiosis are NCOs [see Whitehouse (1982), Tables 19 and 38, for summaries of data from *S. cerevisiae*, *Podospora anserine*, *Sordaria fimicola* and *Sordaria brevicollis*]. More recent work also supports a bias towards NCOs during meiosis. In mouse meiosis there are > 10-fold more DSBs than CO recombinants (Moens et al., 2002), suggesting that most DSBs are repaired by NCO recombination. In *D. melanogaster* there is at least a 3:1 ratio of NCOs to COs (Mehrotra & McKim, 2006). These observations indicate that the majority of recombination events are NCOs. These NCOs involve informational exchange between two homologs but not physical exchange, and little genetic variation is created. Thus explanations for the adaptive function of meiosis that focus exclusively on crossing over are inadequate to explain the majority of recombination events.

Andersen & Sekelsky (2010) have argued that a common mechanism called "synthesis dependent strand annealing" (see section 7, below) is employed in both meiotic HRR of the NCO type and mitotic HRR (which is largely of the NCO type), and thus meiotic and mitotic NCOs probably have a similar function. Substantial evidence indicates that HRR during mitosis is an adaptation to repair DNA damages that originate from diverse endogenous and exogenous sources (e.g. endogenous ROS from oxidative metabolism and exogenous Xrays, UV, chemical carcinogens) (see examples in Table 1; also Lisby & Rothstein, 2009). Thus NCO recombination during meiosis, as in mitosis, likely functions to repair of DNA damages from diverse sources.
