Linkage maps constructed from genetic analysis of gene order and crossover frequency provide few clues to the basis of the genomewide distribution of meiotic recombination, such as chromosome structure, that influences meiotic recombination. To bridge this gap, we have generated the first cytological recombination map that identifies individual autosomes in the male mouse. We prepared meiotic chromosome (synaptonemal complex [SC]) spreads from 110 mouse spermatocytes, identified each autosome by multicolor fluorescence in situ hybridization of chromosome- specific DNA libraries, and mapped 12,000 sites of recombination along individual autosomes, using immunolocalization of MLH1, a mismatch repair protein that marks crossover sites. We show that SC length is strongly correlated with crossover frequency and distribution. Although the length of most SCs corresponds to that predicted from their mitotic chromosome length rank, several SCs are longer or shorter than expected, with corresponding increases and decreases in MLH1 frequency. Although all bivalents share certain general recombination features, such as few crossovers near the centromeres and a high rate of distal recombination, individual bivalents have unique patterns of crossover distribution along their length. In addition to SC length, other, as-yet-unidentified, factors influence crossover distribution leading to hot regions on individual chromosomes, with recombination frequencies as much as six times higher than average, as well as cold spots with no recombination. By reprobing the SC spreads with genetically mapped BACs, we demonstrate a robust strategy for integrating genetic linkage and physical contig maps with mitotic and meiotic chromosome structure.

SC Spreads and Immunostaining

Three juvenile (20–21 d old) C57BL/6J mice (the same line analyzed by the Mouse Genome Sequencing Project) were used to prepare and immunolabel the SC spreads, as described elsewhere (Anderson et al. 1999). Complete sets of SCs in which the SCs were well separated but not obviously stretched or broken and that had ≥ 19 MLH1 foci were selected for analysis. Three fluorescent images (4, 6-diamino-2-phyenylindole [DAPI], SCP3, and MLH1) were captured for each SC set.

mFISH

After image acquisition of the immunofluorescence signals, the spermatocyte preparations were subjected to two or three rounds of denaturation and FISH. To identify each autosome, chromosome-specific painting probes (Rabbitts et al. 1995) were combinatorially labeled with fluorescein isothiocyanate (FITC)–2-deoxyuridine 5-tri phosphate (dUTP), Cy5-dUTP (both from Amersham), or 6-carboxytetramethylrhodamine (TAMRA)-dUTP (Applied Biosystems) and were combined to form two different probe pools (table 1). The probes were fluorescently labeled by the incorporation of FITC-dUTP, Cy5-dUTP (both from Amersham), or TAMRA-dUTP (Applied Biosystems) during the PCR reaction. For labeling, the 2- deoxythymidine 5-triphosphate (dTTP) concentration in the PCR reaction mixture was reduced to 120 mM (2- deoxyadenosine 5-triphosphate [dATP], 2-deoxycytidine 5_-triphosphate [dCTP], and 2-deoxyguanosine 5-triphosphate [dGTP] 160 mMeach), and either 30 mMFITCdUTP, 30 mM Cy5-dUTP (both from Amersham), or 10 mMTAMRA-dUTP was added. The painting probes were precipitated together with 20 mg of mouse Cot-1 DNA (Invitrogen/Life Technologies) and were dissolved in 15 ml of hybridization solution (50% formamide; 2#sodium saline citrate (SSC), pH 7.0; and 10% dextran sulfate). Five genetically mapped BAC clones (Osoegawa et al. 2000; Genetic and Physical Maps of the Mouse Chromosome Web site; Roswell Park Mouse Screening Project Web site) were hybridized to SC spreads in a third round of FISH, after chromosomes were identified using the two chromosome-painting hybridizations described above. The BAC DNA was labeled by nick translation with Cy5-dUTP (Invitrogen/Life Technologies). For each BAC, 400 ng of labeled DNA and 4 mg of mouse Cot- 1 DNA were precipitated together and dissolved in 15 ml of hybridization solution. The BACs were mapped to mitotic metaphase chromosomes, using the same hybridization solutions and standard procedures. To allow FAL analysis and sequential hybridizations on the same nuclei, the FISH protocol (Muller et al. 2002) was modified as follows: After imaging the antibody- labeled cells, the antifade solution was removed by washing the slides twice in 4# SCC/0.1% Tween-20 at 37C for 15 min, and the slides were dehydrated in an alcohol series. The first round of denaturation was 7 min, and in situ hybridization was performed for 48 h at 37C. After separate image acquisition of the three different colors of FISH signals, the next round of FISH (including antifade removal, dehydration, denaturation, and hybridization) was performed, as described above, except that the denaturation time was reduced to 2 min.

Fluorescence Microscopy, Image Acquisition,

and Analysis

The location of each imaged, immunolabeled SC spread was recorded using a software-controlled automatic stage (Maerzhaeuser) so that it could be accurately relocated on the slide for mFISH. Digital images were obtained using a cooled CCD camera, Quantix series (Photometrics), coupled to a Zeiss Axioplan II microscope. Each color signal was acquired as a black-and white image, using appropriate filter sets (Chroma Technologies), and was merged with SmartCaptureVP software (DigitalScientific).

Analysis of MLH1 Distribution on SCs

After the identity of each bivalent was determined by use of mFISH, the images of the corresponding SC spreads were analyzed for MLH1 distribution. For each SC, the position of each MLH1 focus was recorded as a relative distance (percentage of total length) from the centromere, using MicroMeasure (Reeves 2001). The centromeric end of each SC was identified by the surrounding DAPI-bright AT-rich heterochromatin. MLH1 foci were mapped if the MLH1 and SCP3 signals overlapped. A total of 110 sets were analyzed, and 70 SCs (3% of the total of 2,090 autosomal SCs) were eliminated from the analysis because of SC overlap near an MLH1 focus, possible SC stretching, or inconclusive FISH identification. To obtain an average absolute length for each identified SC, the average relative length of each SC was determined, then multiplied by the average absolute length of a complete set of mouse SCs (164.5 µm = 163.7 ÷ 0.995, to correct for rounding errors) (table 2) (Anderson et al. 1999). The relative position of each focus was multiplied by the average absolute length for the appropriate SC to obtain the absolute (micrometer) position of each focus. The data for each of the autosomal SCs were pooled and graphed in histogram form to demonstrate the pattern of MLH1 distribution for each SC. The 0.2-µm intervals used for graphing correspond to ∼3.44 Mb of DNA (calculated from the size, in megabases, of the autosomal fraction of the mouse genome; National Center for Biotechnology Information). The interference distance between two foci on the same SC was calculated as a percentage of the euchromatic length of the SC, as described by Anderson et al. (1999). To convert MLH1 foci to distances in centimorgans, the number of MLH1 foci in each 0.2-µm SC interval was divided by the total number of SCs observed, then multiplied by 50 map units per MLH1 focus (one crossover p 50 cM). The average number of MLH1 foci per 0.2-µm interval was determined for each autosomal SC in our sample. For the larger SCs, the average was about three; for the shorter SCs this number was slightly higher. To identify potential “hot” and “cold” regions of recombination, we added or subtracted one focus to identify regions of high (>14 for longer SCs, as high as 6 for the shorter SCs) and low (<2 for longer SCs and <4 for shorter SCs) recombination frequency, respectively.