Experiment
High resolution 3D micro-computed tomography (microCT), together with 3D landmarks and geometric morphometrics are used in the current study to map QTL responsible for variation in skull shape and size. A new QTL analysis package for R statistical software that is more suitable for the mapping of shape as a multivariate feature is also used.
One hundred and sixty three offspring (84 females and 79 males) from an A/J x (C57BL/6J x A/J)F1 backcross and 270 (128 female and 142 males) from the reciprocal (C57BL/6J x A/J) x A/J F1 cross were produced. All 433 animals were sacrificed at postnatal day 28 and their heads were preserved for imaging. Liver tissue was also collected from each animal for DNA extraction. A panel of 1449 SNPs selected from the Welcome-CTC Mouse Strain SNP Genotype Set was used to provide uniform genome distribution at a density of app 3 SNPs per 5 Mb across the genome.
All animals were imaged at the Small Animal Tomographic Analysis (SANTA) Facility at Seattle Children's Research Institute using high-resolution microcomputed tomography (model 1076) employing a standardized imaging protocol. Reconstructed image stacks were loaded into 3D Slicer and rendered in 3D. A random subset of 50 samples was landmarked twice using an initial set of 55 skull landmarks. Figure 1 shows the final set of landmarks used in the study. The centroid size, the square root of the sum of squared Euclidean distances from each landmark to their own centroid, was used as a proxy for overall skull size.
All computations for shape QTL mapping were conducted in R/Shape QTL package written by one of the authors (NN) and is available by request. The log-transformed centroid size was analyzed in a similar manner to shape using Haley-Knott regression using R/qtl v1.28-19. QTL effects were estimated conditional on all covariates (size, direction of the cross and sex).
Interval mapping identified seven QTL responsible for variation in skull centroid size,
Table 1:
QTL Skcsq1, skull centroid size QTL 1, mapped to Chromosome 2 with a LOD score of 3.33 at 60.37 cM near marker rs3658927. The confidence interval spanned 52,795,574-141,957,105 bp.
QTL Skcsq2, skull centroid size QTL 2, mapped to Chromosome 4 with a LOD score of 4.85 at 68.08 cM near marker rs13478002. The confidence interval spanned 129,338,356-142,748,609 bp.
QTL Skcsq3, skull centroid size QTL 3, mapped to Chromosome 5 with a LOD score of 4.71 at 72.26 cM near marker rs13478540. The confidence interval spanned 117,927,219-137,110,565 bp.
QTL Skcsq4, skull centroid size QTL 4, mapped to Chromosome 6 with a LOD score of 5.42 at 34.74 cM near marker rs13478841. The confidence interval spanned 51,455,318-87,816,657 bp.
QTL Skcsq5, skull centroid size QTL 5, mapped to Chromosome 11 with a LOD score of 4.41 at 48.5 cM near marker rs13481127. The confidence interval spanned 45,970,896-91,694,861 bp.
QTL Skcsq6, skull centroid size QTL 6, mapped to Chromosome 13 with a LOD score of 7.81 at 44.75 cM near marker mCV24625340. The confidence interval spanned 53,171,098-94,326,861 bp.
QTL Skcsq7, skull centroid size QTL 7, mapped to Chromosome 16 with a LOD score of 4.6 at 4.43 cM near marker rs4160288. The confidence interval spanned 4,326,565-16,886,506 bp.
QTL Skcsq3 was the only QTL of the 7 whose effect was to reduce skull size. Skcsq7 had the largest effect on increased skull size. Both sex and direction of the cross, but not their interactions, had statistically significant effects on the skull size. Male mice in general and N2 individuals derived from the F1 female x A/J male had larger skulls.
Interval mapping identified 30 QTL responsible for variation in skull shape, Table 2:
QTL Skshq1, skull shape QTL 1, mapped to Chromosome 1 with a LOD score of 12.0 at 10.5 cM near marker mCV24784983. The confidence interval spanned 5.920,984-25,974,921 bp. Candidate genes are Col9a1 and Eya1.
QTL Skshq2, skull shape QTL 2, mapped to Chromosome 1 with a LOD score of 28.3 at 52.6 cM near marker rs3678634. The confidence interval spanned 115,819,089-127,021,793 bp. Candidate genes are Fcgr2b and Gli2.
QTL Skshq3, skull shape QTL 3, mapped to Chromosome 1 with a LOD score of 17.7 at 78.6 cM near marker rs13466711. The confidence interval spanned 168,066,210-175,710,316 bp.
QTL Skshq4, skull shape QTL 4, mapped to Chromosome 10 with a LOD score of 18.3 at 52.9 cM near marker rs13480734. The confidence interval spanned 89,335,908-105,325,377bp. Alx1 is a candidate gene.
QTL Skshq5, skull shape QTL 5, mapped to Chromosome 11 with a LOD score of 27.3 at 48.5 cM near marker rs13481127. The confidence interval spanned 80,139,712-81,911,051 bp.
QTL Skshq6, skull shape QTL 6, mapped to Chromosome 11 with a LOD score of 24.8 at 77.5 cM near marker rs3672597. The confidence interval spanned 112,611,769-114,027,932 bp. Sox9 is a candidate gene.
QTL Skshq7, skull shape QTL 7, mapped to Chromosome 12 with a LOD score of 32.5 at 8.8 cM near marker rs3717860. The confidence interval spanned 16,866,101-28,237,591 bp. Taf1b is a candidate gene.
QTL Skshq8, skull shape QTL 8, mapped to Chromosome 13 with a LOD score of 19.5 at 13.4 cM near marker rs3710348. The confidence interval spanned 30,941,216-35,713,142 bp. Itga2 is a candidate gene.
QTL Skshq9, skull shape QTL 9, mapped to Chromosome 13 with a LOD score of 13.4 at 62.8 cM near marker gnf13.115.241. The confidence interval spanned from 110,281,304-118,066,332 bp.
QTL Skshq10, skull shape QTL 10, mapped to Chromosome 14 with a LOD score of 28.1 at 20.1 cM near marker rs6396829. The confidence interval spanned 30,970,692-38,186,621 bp.
QTL Skshq11, skull shape QTL 11, mapped to Chromosome 15 with a LOD score of 15.0 at 16.8 cM near marker CEL-15_43206205. The confidence interval spanned 39,391,647-47,392,759 bp.
QTL Skshq12, skull shape QTL 12, mapped to Chromosome 16 with a LOD score of 13.9 at 35.4 cM near marker rs4191367. The confidence interval spanned 54,657,678-66,961,706 bp. Candidate genes are Epha3 and Arhgap31.
QTL Skshq13, skull shape QTL 13, mapped to Chromosome 17 with a LOD score of 23.8 at 19.2 cM near marker rs6298471. The confidence interval spanned 36,252,692-43,916,321 bp.
QTL Skshq14, skull shape QTL 14, mapped to Chromosome 18 with a LOD score of 11.4 at 31.3 cM near marker rs6328845. The confidence interval spanned 56,230,871-70,306,352 bp. Ppargc1b, Slc26a2 and Tcof1 are candidate genes.
QTL Skshq15, skull shape QTL 15, mapped to Chromosome 19 with a LOD score of 17.0 at 40.1 cM near marker rs3023496. The confidence interval spanned 46,867,039-49,396,956 bp.
QTL Skshq16, skull shape QTL 16, mapped to Chromosome 2 with a LOD score of 20.6 at 44.3 cM near marker rs13476580. The confidence interval spanned 72,637,926-76,844,869 bp.
QTL Skshq17, skull shape QTL 17, mapped to Chromosome 2 with a LOD score of 16.5 at 73.3 cM near marker rs6209325. The confidence interval spanned 146,817,118-155,755,548 bp. Pax1, Cd93 and Asxl1 are candidate genes.
QTL Skshq18, skull shape QTL 18, mapped to Chromosome 3 with a LOD score of 14.6 at 18.0 cM near marker rs6246699. The confidence interval spanned 34,942,416-39,449,337 bp.
QTL Skshq19, skull shape QTL 19, mapped to Chromosome 3 with a LOD score of 13.4 at 44.0 cM near marker rs4138887. The confidence interval spanned 96,754,992-130,100,724 bp. Alx3, Col11a1 and Gnai3 are candidate genes.
QTL Skshq20, skull shape QTL 20, mapped to Chromosome 4 with a LOD score of 9.7 at 3.1 cM near marker rs3660863. The confidence interval spanned 3,867,296-11,352,972 bp. Plag1, Chd7 and Gdf6 are candidate genes.
QTL Skshq21, skull shape QTL 21, mapped to Chromosome 4 with a LOD score of 20.5 at 40.1 cM near marker rs3711477. The confidence interval spanned 81,877,569-105,794,559 bp. Frem1 is a candidate gene.
QTL Skshq22, skull shape QTL 22, mapped to Chromosome 4 with a LOD score of 23.2 at 65.1 cM near marker rs3663950. The confidence interval spanned 126,198,546-134,724,460 bp. Thrap3, Col16a1 and Arid1a are candidate genes.
QTL Skshq23, skull shape QTL 23, mapped to Chromosome 5 with a LOD score of 24.8 at 17.3 cM near marker rs13459085. The confidence interval spanned 20,284,721-35,523,469 bp. Cad, Whsc1, Fgfr3, Shh, Hmx1 and Sh3pb2 are candidate genes.
QTL Skshq24, skull shape QTL 24, mapped to Chromosome 5 with a LOD score of 14.7 at 59.3 cM near marker CEL-5_117374791. The confidence interval spanned 117,729,579-119,88,175 bp. Tbx3 is a candidate gene.
QTL Skshq25, skull shape QTL 25, mapped to Chromosome 6 with a LOD score of 34.8 at 35.5 cM near marker rs6181382. The confidence interval spanned 78,232,602-84,781,463 bp. Dok1 and Gcfc2 are candidate genes.
QTL Skshq26, skull shape QTL 26, mapped to Chromosome 6 with a LOD score of 27.3 at 78.2 cM near marker rs6265387. The confidence interval spanned 144,446,852-148,260,112 bp. Arntl2 is a candidate gene.
QTL Skshq27, skull shape QTL 27, mapped to Chromosome 7 with a LOD score of 16.4 at 49.5 cM near marker rs13479395. The confidence interval spanned 78,470,627-90,617,547 bp. Kif7 is a candidate gene.
QTL Skshq28, skull shape QTL 28, mapped to Chromosome 8 with a LOD score of 23.2 at 29.7 cM near marker rs13479776. The confidence interval spanned 47,753,313-78,304,228 bp.
QTL Skshq29, skull shape QTL 29, mapped to Chromosome 9 with a LOD score of 11.3 at 22.5 cM near marker rs3174664. The confidence interval spanned 39,523,343-46,544,750 bp. Pvrl1 is a candidate gene.
QTL Skshq30, skull shape QTL 30, mapped to Chromosome 9 with a LOD score of 16.0 at 49.5 cM near marker rs13480351. The confidence interval spanned 90,683,187-99,809,852 bp. Foxl2 is a candidate gene.
Some of the candidate genes identified for skull shape QTL have recently been implicated in some craniofacial disorders, or have known expression patterns in the developing mouse craniofacial region.
Analysis Tools