Molecular Cell, Vol. 20, 563–573, November 23, 2005, Copyright ª2005 by Elsevier Inc. DOI 10.1016/j.molcel.2005.09.021
s
from the programmed induction of DSBs at the lepto- lated, and mechanistically and functionally related. Un-tene stage of meiotic prophase that are repaired by in-
teraction with a chromatid from the homolog through
various intermediate steps that span from the zygotene
to the end of pachytene (Keeney, 2001; Smith and
like CRs, NCRs, as products of recombination, are not
thought to play any role during meiotic prophase. Inter-
estingly, it is rather the intermediates leading to their for-
mation that have been proposed to participate to pairing
between homologs at leptotene/zygotene (Carpenter,
1987; Zickler and Kleckner, 1998). This contribution of*Correspondence: bdemassy@igh.cnrs.frCrossover and Noncrossov
Pathways in Mouse Meiosi
He´le`ne Guillon,
1
Fre´de´ric Baudat,
1
Corinne Grey,
1
R. Michael Liskay,
2
and Bernard de Massy
1,
*
1
Institute of Human Genetics
Centre National de la Recherche Scientifique
UPR1142
141 rue de la Cardonille
34396 Montpellier Cedex 5
France
2
Molecular and Medical Genetics
Oregon Health and Science University
3181 Southwest Sam Jackson Park Road
Portland, Oregon 97239
Summary
During meiosis, recombination between homologous
chromosomes generates crossover (CR) and non-
crossover (NCR) products. CRs establish connections
between homologs, whereas intermediates leading to
NCRs have been proposed to participate in homolo-
gous pairing. How these events are differentiated
and regulated remains to be determined. We have de-
veloped a strategy to detect, quantify, and map NCRs
in parallel to CRs, at the Psmb9 meiotic recombination
hot spot, in male and female mouse germ lines. Our re-
sults report direct molecular evidence for distinct CR
and NCR pathways of DNA double-strand break (DSB)
repair in mouse meiosis based on three observations:
both CRs and NCRs require Spo11, NCR products
have shorter conversion tracts than CRs, and only
CRs require the MutL homolog Mlh1. We show that
both products are formed from middle to late pachy-
tene of meiotic prophase and provide evidence for an
Mlh1-independent CR pathway, where mismatch re-
pair does not require Mlh1.
Introduction
Meiosis is a specialized cell cycle that allows halving
of the chromosome content. This is achieved by a
modified cell cycle composed of an S phase followed
by two divisions. During the first division, termed reduc-
tional, homologous chromosomes segregate, whereas
sister-chromatids separate at the second division. In
most organisms, the reductional segregation requires
the establishment of connections between homologs vi-
sualized as chiasmata and detected genetically as CRs,
the reciprocal products of recombination between ho-
mologous chromosomes (Petronczki et al., 2003). Both
cytological and genetic analyses indeed show the for-
mation of at least one CR per homolog during meiosis.
Studies in S. cerevisiae have shown that CRs resulter
Nicolas, 1998). The repair of DSBs generates another
type of recombination product but without exchange
of flanking markers and with a local nonreciprocal trans-
fer of genetic information: gene conversion without CR
or NCR products. The observation that gene conversion
may or may not be associated with exchange of flanking
markers has suggested a mechanistic link between CRs
and NCRs that has been incorporated into various mod-
els of meiotic recombination (Holliday, 1964; Meselson
and Radding, 1975; Szostak et al., 1983).
However, recent studies in S. cerevisae have given rise
to a different view where CRs and NCRs, which both de-
pend on Spo11-induced DSBs, result from two distinct
repair pathways that do not share the same intermedi-
ates and require different functions. The differentiation
between these pathways takes place early in the pro-
cess of recombination, and NCRs and CRs are not alter-
native products of a common double Holliday junction
(dHJ) intermediate as proposed (Szostak et al., 1983).
These conclusions were established on the basis of the
identification of mutants that specifically reduce the
number of CRs, or allow NCR formation in the absence
of CR, together with the analysis of intermediates of
NCRs and CRs (Allers and Lichten, 2001; Hunter and
Kleckner, 2001). The commitment of DSB repair toward
CR is defined at the time or soon after DSB formation
at leptotene/zygotene and mediated by the ZIP1, ZIP2,
ZIP3, MSH4, MSH5, and MER3 genes (ZMM) (Borner
et al., 2004). Strains carrying mutations in one of these
genes have a reduction of CRs, whereas NCRs are not
affected. Formation of CR in this pathway also requires
Mlh1 and Mlh3 proteins, thought to act at a late step
in the recombination reaction (Abdullah et al., 2004;
Argueso et al., 2004; Hunter and Borts, 1997; Wang et al.,
1999). An additional level of complexity in the repair
pathway of DSBs is due to the regulation of CRs by
positive interference, which leads to the nonrandom dis-
tribution of CRs that are more widely spaced than ex-
pected (Jones, 1984). In S. cerevisiae, CRs can be di-
vided into two classes: interfering and noninterfering.
Interfering CRs are those dependent on the ZMM genes
and are products of a pathway going through dHJ forma-
tion, where dHJs are preferentially resolved toward CRs.
Noninterfering CRs depend on the Mus81-Mms4 pro-
teins (De Los Santos et al., 2003), which act on a pathway
that might involve dHJ or other intermediates suggested
on the basis of the enzymatic activities of these proteins
(Hollingsworth and Brill, 2004).
Thus, at the time or soon after DSB formation, a com-
plex regulation operates to channel the repair into at
least three possible outcomes: NCR, interfering CR, or
noninterfering CR. This raises several questions about
how these three types of events are generated, regu-

Molecular CellNCRs to the pairing process varies between organisms,
suggesting alternative balances between the various
DSB repair pathways (Gerton and Hawley, 2005). These
different contributions of NCRs can be deduced from
the phenotypes of spo11 mutants. In spo11 mutants of
S. cerevisiae, S. macrospora, C. cinereus, A. thaliana,
and M. musculus, homologous pairing and synapsis
are defective, whereas they occur normally in C. elegans
and D. melanogaster (Keeney, 2001; Lichten, 2001). This
suggests that early recombination intermediates, those
leading to both NCRs and CRs or to NCRs only, can play
an essential role in homolog pairing.
In M. musculus, genetic and cytological studies have
defined some components of the meiotic recombination
pathway. The mechanistic conservation of recombina-
tion initiation by DSBs has been shown with the analysis
of the mouse Spo11 (Baudat et al., 2000; Mahadevaiah
et al., 2001; Romanienko and Camerini-Otero, 2000).
Meiotic-specific DNA breaks were indeed detected at
the H2-Ea recombination hot spot (Qin et al., 2004).
The number of DSB repair events, NCRs and CRs, is es-
timated on the basis of the cytological detection of
RAD51/DMC1 proteins, which are thought to represent
early intermediates of recombination (Ashley et al.,
1995; Barlow et al., 1997; Moens et al., 1997). These pro-
teins localize as foci mostly along unsynapsed axes of
chromosomes, and their number per nucleus peaks at
about 250 in leptotene (Moens et al., 2002). CRs are cy-
tologically identified by the presence of MLH1 and
MLH3: both proteins colocalize on chromosome axes
of pachytene bivalents at sites corresponding to chias-
mata, whose number ranges from 23 to 27 per nucleus
(Anderson et al., 1999; Baker et al., 1996; Lipkin et al.,
2002; Marcon and Moens, 2003). This localization is con-
sistent with the reduction in chiasma formation ob-
served in Mlh1
2/2
(Baker et al., 1996; Edelmann et al.,
1996) and Mlh3
2/2
(Lipkin et al., 2002) mice. Therefore,
if 250 recombination events are initiated, most are not
processed toward CRs and are thus expected to be re-
paired either as NCRs between homologs or by ex-
changes between sister chromatids. With respect to
the noninterfering CR pathway, although the mouse
Mus81 gene has been identified, it remains to be deter-
mined if this gene plays any role during meiosis. If it
does, the Mus81 pathway should only contribute to a mi-
nor fraction of CR events, as Mus81
2/2
mice are fertile
and therefore not expected to have a strong reduction
in CR frequency (McPherson et al., 2004).
Our goal was to identify the common and specific
properties of CRs and NCRs in mouse meiosis and to
address the issue of the conservation of the phenomena
as described in S. cerevisiae. Following the strategy we
used to analyze CRs during mouse spermatogenesis
(Guillon and de Massy, 2002), we have extended our ap-
proach to analyze in parallel CRs and NCRs both in oo-
genesis and spermatogenesis in mice. Our analysis was
developed at the Psmb9 hot spot on chromosome 17.
Here, we report the measurements in wild-type (wt) mei-
osis of the frequencies and timing of NCRs and CRs and
the mapping of exchange points in recombinant mole-
cules. Furthermore, we tested whether both events,
564NCRs and CRs, were Spo11 dependent and determined
the consequences of Mlh1 deficiency on NCR and CR
formation.Results
A Method for the Parallel Detection
of CRs and NCRs
The 2 kb region at the 3
0
end of the Psmb9 gene in the ma-
jor histocompatibility complex on chromosome 17 is
a CR hot spot in mouse hybrids where one parent carries
the wm7 haplotype (derived from Mus musculus molos-
sinus). The CR activity, measured genetically, is 2%–3%
per gamete both in males and females of B10.A(R209) 3
B10 hybrids in which B10.A(R209), hereafter named
R209, carries the wm7 haplotype (Shiroishi et al., 1991).
We have recently developed a direct method for the
analysis of meiotic CRs at the Psmb9 hot spot by
allele-specific PCR on genomic DNA extracted from
sperm or testis cells (Guillon and de Massy, 2002). We
have now extended this molecular approach to analyze
both CRs and NCRs in the male and female germ lines.
The strategy, outlined in Figure 1A, is similar to the one
used for the analysis of gene conversion in humans (Jef-
freys and May, 2004). NCRs were selected at the BsrFI
site, where we previously detected gene conversion
products without CR in sperm DNA (Guillon and de
Massy, 2002).
An example of such a CR/NCR detection assay is
shown in Figure 1B, where 48 independent pools, con-
taining 191 amplifiable genomes each, were tested. Thir-
teen pools are positive for PCR P3/P7 and thus contain
a CR, and six are positive for PCR P5/P6 and thus con-
tain an NCR. Given the number of amplifiable genomes
per pool, the frequencies of B6-R209 CRs (representing
half of total CRs) and of NCRs at BsrFI on B6 chromo-
some are 0.17% 6 0.1% and 0.07% 6 0.05%, respec-
tively (see Experimental Procedures). Among the ten
pools positive for P3/P4, we found as expected the six
pools positive for NCR. The other four pools are positive
for CR and have therefore an exchange point to the left
of BsrFI. One pool (#2) was positive in all three PCRs
and thus contains both CR and NCR molecules. The fre-
quencies of CRs can also be independently determined
by direct selection of CRs after two rounds of allele-
specific PCRs (Guillon and de Massy, 2002). It is impor-
tant to note that all frequencies reported refer to geno-
mic DNA extracted from tissues containing meiotic
and mitotic cells (i.e., testis and ovary).
Both CRs and NCRs Appear during Middle
to Late Pachytene in Spermatogenesis
Taking advantage of the first synchronous wave of sper-
matogenesis, we have shown previously that CRs are
formed between 11 and 16 days postpartum, the period
corresponding to progression through the pachytene
stage of meiotic prophase I (Guillon and de Massy,
2002). Here, we further defined the timing of CR prod-
ucts and compared it to that of NCRs by taking samples
from 11 to 21 days postpartum. Our results show that
CRs were not detected at 11 days (48,000 molecules
tested) and that most of them are formed between 14
and 17 days when the cells from the first wave of meiosis
(Figure 2), initiated at approximately day 8, progress
from middle to the end of pachytene (Goetz et al.,1984). The further increase, between 18 and 21 days, is
probably due to entry of spermatogonia into meiosis af-
ter 8 days. Parallel measurements of NCRs at BsrFI