Qiu et al., 2019 - Oligogenic Effects of 16p11.2 Copy-Number Variation on Craniofacial Development. Cell Reports   28:3320-3328.e4 Full text @ Cell Rep.

Fig. 1 Differential Effects of 16p11.2 Copy Number on Dimensions of the Frontal, Nasal, Maxillary, and Mandibular Regions

(A) On each 3D facial image, 24 landmarks were placed and two angular measurements were calculated. A description of landmarks is provided in Table S2. After averaging symmetric distances, 156 distance measures were compared between the CNV and control groups.

(B) 18 measures were significant after correction for a FWER <5%. Regression coefficients for duplication versus control (y axis) and deletion versus control (x axis) show that reciprocal CNVs have reciprocal effects on growth of the major craniofacial processes. The category “Other” represents features that span multiple processes. The 14 most informative facial features based on LASSO selection are drawn in (A) and colored by facial region according to the legend. For clarity, some nasal distances are excluded.

(C) Facial features associated with deletion and duplication were visualized as a computer-generated model face in which specific features were adjusted according to the observed effect sizes (from B and Table S2).

(D) The average surface topography was generated from multiple (>5) age-matched subjects with each genotype. Note that subtle differences in BMI are also apparent; however, these effects are controlled for in the statistical analysis and do not influence the feature selection.

Fig. 2 Three-Dimensional Models of Deletion, Control, and Duplication Groups

(A–D) 3D models were generated by averaging of the surface topography of faces from multiple subjects. Separate models were constructed for (A) female children (deletion: n = 7, mean age 9.15 years; control: n = 8, mean age 9.92 years; duplication: n = 5, mean age 12.73 years), (B) female adults (deletion: n = 4, mean age 20.13 years; control: n = 8, mean age 23.71 years; duplication: n = 5, mean age 23.25 years), (C) male children (deletion: n = 7; mean age 8.90 years; control: n = 9, mean age 9.27 years; duplication: n = 9, mean age 9.20 years), and (D) male adults (deletion: n = 5; mean age 25.53 years; control: n = 10, mean age 36.59 years; duplication: n = 5, mean age 36.48 years).

Fig. 3 Classification of 16p11.2 Genotype Based on Facial Features

(A and B) Discriminant coefficients based on features that were significant at FDR <0.05 can distinguish the subjects based on genotype, with better discrimination for younger subjects (age ≤ 20 years). The linear model was controlled for age, head circumference, BMI, sex, and ancestry principal components. Linear discriminant analysis was applied to subjects for which the above demographic information was complete for the full sample (N = 220; 8 had missing predictors; A) and the younger group (N = 107; 6 had missing predictors; B).

Fig. 4 Validation of Mirror Craniofacial Effects in Rat and Mouse Models of 16p11.2 Deletion and Duplication

All pairwise distances were analyzed for nineteen landmarks on the dorsal skull and three on the mandible as shown here and in Table S4. Distances are colored according to craniofacial region using the same scheme as in Figure 1. Distances that span multiple craniofacial processes are denoted as “other.” ML, mandibular length; MW, mandibular width.

(A) In the rat models, 52 individual features differed significantly by genotype. Regression coefficients for the duplication deletion show significant mirror effects.

(B) Informative features were identified by LASSO selection, and features that correspond to a specific facial process in rat are shown.

(C) In the mouse models, 12 craniofacial measures that discriminated mutant and control groups were selected by LASSO. Regression coefficients of these features show mirror effects of deletion and duplication similar to those in human and rat.

(D) Features that correspond to specific facial processes in mouse.

Fig. 5 <italic>In Vivo</italic> Modeling of the 16p11.2 CNV Implicates Single Gene Drivers and Epistatic Effects Influencing Cartilage Structures in the Zebrafish Pharyngeal Skeleton

(A) Representative ventral images of −1.4col1a1:egfp zebrafish larvae at 3 days post-fertilization (dpf). Orientation arrows indicate anterior (A), posterior (P), left (L), and right (R). Scale bar, 300 μm.

(B) Quantitative assessment of the CHA of larvae injected with single human mRNAs for each of the 30 genes located in the 16p11.2 BP4-BP5 region. Images were measured as shown in (A) (angle between dashed lines). Seven transcripts induced a significant reduction in CHA after Tukey’s p value adjustment (adjusted p < 0.01). Dosage is 12.5 pg for KIF22 and PPP4C and 50 pg for all other genes.

(C) Quantitative assessment of the CHA of F0 mutant batches injected with single combinations of each of sez6l2, taok2a, and taok2b gRNAs with or without Cas9. Dosage is 50 pg gRNA and 200 pg Cas9 protein.

(D) Quantitative assessment of the CHA of larvae injected with single or equimolar combinations of human KCTD13, MAPK3, and MVP mRNAs. Dosage is 50 pg.

(E) Quantitative assessment of the CHA of F0 mutant batches injected with single or equimolar combinations of kctd13, mapk3, and mvp gRNAs with or without Cas9. Dosage is 50 pg gRNA and 200 pg Cas9 protein. The number of larvae measured for each condition is indicated at the base of each bar in the graphs. The data are represented as the mean ± SEM; ns, not significant; **p < 0.01, ***p < 0.001, and ****p < 0.0001 versus uninjected controls. Tukey’s multiple comparison tests were applied following a significant one-way ANOVA.

Fig. S3 ZFIN is incorporating published figure images and captions as part of an ongoing project. Figures from some publications have not yet been curated, or are not available for display because of copyright restrictions.

Acknowledgments:
ZFIN wishes to thank the journal Cell Reports for permission to reproduce figures from this article. Please note that this material may be protected by copyright. Full text @ Cell Rep.