Experiment
The current study unites newly developed morphometric methods with quantitative trait analysis to quantify the genetic basis of baculum size and shape variation. The novel morphometric approach used three-dimensional micro-CT scans from 369 bacula, representing 75 distinct strains from the BXD panel of mice.
Two genetic models were used:
BXD recombinant inbred lines originated from a cross between two classical inbred strains, C57BL/6J (B6) and DBA/2J (D2). Individuals from each RI line in the BXD panel are essentially genetically identical and can be considered biological replicates of the same genotype. Males were sampled that were from 60 to 200 days old since the baculum is fully developed at that age.
LGxSM advanced intercross lines (AILs) were also analyzed. AILs are maintained thru random breeding of unrelated individuals. Each generation accumulates more recombination events that break up the genomes of the two parental strains while avoiding inbreeding. As part of unrelated research, several hundred individuals of the Chev:LG,SM-G43 (F43) and Chev:LG,SM-G44 (F44) generation were genotyped at thousands of markers. One hundred forty-four of those same F43 and F44 mice were sample here in the current study.
For all specimens the glans penis was cut proximal to the baculum and prepared for micro-CT scanning. Used here, 'baculum' refers to the single bone in the baculum, excluding the distal fibrocartilage.
The baculum lacks any true landmarks, so 802 points (semi-landmarks) were mathematically defined along each bone that corresponded to regionally homologous regions across all specimens. Size was quantified as centroid size, the square root of the sum of squared distances of the 802 semi-landmarks from their centroid. Shape differences were quantified between all possible pairs of bacula in a Generalized Procrustes framework. The goal was to define a single metric that summarized each specimen's shape in the context of the respective parental strains.
QTL mapping included all 73 BXD RI lines and employed the SCANONE function in the R package QTL using Hayley-Knott regression to estimate the location and the effect of QTL (NCBI Build 37). To determine significance phenotypes and genotypes were permuted 1000 times and the 95th quartile of the 1000 maximum LOD scores were used as the empirical significance threshold. Genotypes were downloaded from GeneNetwork and included 3805 markers spaced at a mean 0.54 cM throughout the genome. The 2953 markers that were also part of the updated mouse genetics map (Cox, et al 2009) were used.
QTL analysis of the BXD lines identified 3 significant QTL:
A QTL for baculum size, Bacszq1 (baculum size QTL 1) mapped to Chromosome 1 between 136.3-150.1 Mb with a LOD score of 7.45.
A second QTL for baculum size, Bacszq2 (baculum size QTL 2) mapped to Chromosome 12 between 58.0 and 76.4 Mb with a LOD score of 4.21. The two markers nearest the QTL peaks on Chromosomes 1 and 12 displayed a clear difference in centroid size among BXD RI lines carrying a B6 allele vs a D2 allele. The two QTL were not in linkage disequilibrium.
A QTL for baculum shape, Bacsq1 (baculum shape QTL 1) mapped to Chromosome 2 between 145.7 and 161.0 Mb with a LOD score of 4.32. At the marker closest to the peak of Bacsq1 there was a clear difference in shape between the 33 BXD RI lines with a B6 allele compared with the 40 RI lines with the D2 allele.
To potentially identify candidate genes RNA seq data from 10 5-week-old B6 (N=5) and D2 (N=5) individuals was generated. (B6 bacula data were derived from C57BL/6N).
Potential candidate genes, Table S3:
for Bacszq1 were: Kif14, Ptprc, Aspm, and Cfhr2;
for Bacszq2 were Sec23a, Mgat2, Arf6, 2700049A03Rik, Dact1, Hif1a and Rhoj; Plcg1, Rbl1, Tgm2 and Mafb;
for Bscsq1 were Plcg1, Rbl1, Tgm2 and Mafb.
The LGxSM AILs were analyzed using the R package QTLREL which accounts for background genetic relatedness prior to scanning for QTL. Age and weight were included as covariates. Individuals were genotyped at 4716 markers spaced at a mean 0.36 cM. Significance thresholds were estimated with 1000 permutations. 144 bacula were micro-CT scanned including 36 F43, 89 F44, 10 LG and 9 SM strains. No significant QTL were identified for either baculum size or shape. The results may have been an indication that the sample size was underpowered.
Analysis Tools