Project overview.

Depicting steps for candidate gene identification based on human transcriptomic data (1), enriching for genes expressed within adipose stem and progenitor cells, (3) GO term enrichment analysis of candidate genes to focus on development and differentiation processes, (4) prioritisation of candidate genes based on conservation to zebrafish and expression dynamics during adipocyte progenitor differentiation and (5) an in vivo CRISPR screen in zebrafish to functionally characterise genes.

Prioritisation of candidate adipose morphology genes based on conservation to zebrafish and expression dynamics during progenitor differentiation.

A. GO term enrichment analysis of the 102 candidate morphology genes enriched in adipose stem and progenitor cells. 54 of these genes were enriched in four GO terms related to development and differentiation. B. Venn diagram showing gene intersection between GO terms. Cell differentiation (GO:0030154) and cellular developmental process (GO:0048869) had 100% overlap. C. Grey bars denote % protein similarity between human and zebrafish orthologs. Dot plots below show scaled mean expression (colour) and % of cells expressing (dot size) each candidate genes within four cell types identified during progenitor differentiation to adipocytes. Cell types are: non-induced progenitors, MGP+ SWAT cells, ADIPOQ+ adipogenic cells and others, based on Yang Loureiro et al., 2023. D. PCA projection of single cells according to cluster designation from C. E. Example candidate morphology genes and their expression during adipocyte differentiation.

Zebrafish gene targets and adipose morphology statistics.

Adjusted p values after Benjamini-Hochberg FDR correction for 23 statistical tests. Small, medium and large correspond to effect sizes (Cohen’s d). KS = Kolmogorov-Smirnov test. LMM = linear mixed model with experiment as random effect. Values in bold indicate statistical significance. All genes were targeted in two independent replicates (except for aspa and lamb2 which only had a single replicate).

Spatial dynamics of subcutaneous adipose growth in zebrafish.

A. Nile Red stained zebrafish at 11.6 mm standard length (SL). Black signal denotes Nile Red+ neutral lipid within adipose. The magenta dotted box denotes the area shown in B. B. Higher magnification image showing region of interest. Magenta outlined lipid droplets (LDs) highlight the subcutaneous adipose tissue. C. Zoomed in view of subcutaneous adipose LDs. White dots are melanosome pigment granules. D. Magenta liens outline the LD segmentation performed on LDs from C. E. Segmented subcutaneous adipose LDs colour-coded according to LD diameter from four representative zebrafish at different sizes. Fish sizes are shown in mm SL. Strata were defined in 200 μm intervals from the most anterior (to the right) LD. F. Number of LDs in each strata according to size of the zebrafish (SL). Zebrafish were categorised into 5 groups according to size. G. Mean diameter of LDs in each strata according to size of zebrafish.

Calculation of adipose morphology value based on relationship between lipid droplet size and overall adipose depot size.

A. Diameter of subcutaneous adipose LDs plotted relative to depot area. Mean LD diameters per fish are dark grey, individual LD diameters are light grey. Line was fitted using a generalised additive model. B. Morphology values were calculated as the deviation of mean LD diameter from the fitted model in A. Note the normal distribution of morphology values around zero. C. Inverse relationship between morphology value and number of LDs. A linear model was fitted to show the inverse relationship.

Identification of mutants with hypertrophic or hyperplastic adipose morphology.

A. Morphology values for individual mutants (blue) with respect to their Cas-9 only control (grey). Dotted vertical line represents a morphology value of zero. B. Cumulative probability functions showing shift towards hypertrophic (foxp1b, txnipa and mmp14b) or hyperplastic (ptenb, cxcl14 and srpx) morphology. C. Segmented subcutaneous adipose LDs from each respective mutant colour-coded according to LD diameter.

Stable foxp1b zebrafish mutants have hypertrophic adipose but are unable to undergo hypertrophic expansion in response to a high-fat diet.

A. Phylogenetic tree showing relatedness of zebrafish Foxp1a and Foxp1b amino acid sequences to human, mouse, opossum and ceolocanth Focp1. Bar indicates substitutions per X. B. Overview of human Foxp1 domain structure showing a polyQ, coiled-coil and forkhead domain. Zoomed view of the DNA-binding forkhead domain showing structural features including helix and S. Essential amino acids of DNA binding are highlighted in orange, amino acids essential for DNA binding are highlighted in blue. Zebrafish wild-type Foxp1a and Foxp1b sequences are shown aligned to human, along with the ed116 foxp1a and ed125 foxp1b mutant alleles. Grey boxes show the addition of nonsense peptide followed by premature stop codon. C. Western blot showing the reduction of Foxp1 protein in double foxp1a;foxp1b zebrafish mutants. B-actin is used as a loading control. Asterisk indicates the Foxp1 protein band. D. Nile Red fluorescence images showing adipose lipid distribution (black signal) in wild-type, foxp1aed116, foxp1bed125and double foxp1aed116;foxp1bed125 zebrafish mutants. E indicates the eye. Asterisk indicates lipid accumulation within the liver in double mutants. Scale bar is 1 mm. E. Violin plots showing fish size (standard length, mm) of foxp1aed116, foxp1bed125 and double foxp1aed116;foxp1bed125 mutants compared to wild-type siblings. F. Violin plots showing Nile Red-positive adipose area in foxp1aed116, foxp1bed125 and double foxp1aed116;foxp1bed125 mutants compared to wild-type siblings. G. Violin plots showing average LD diameter in foxp1aed116, foxp1bed125 and double foxp1aed116;foxp1bed125mutants compared to wild-type siblings. H. LipidTox stained LDs within subcutaneous adipose of foxp1aed116, foxp1bed125, double foxp1aed116;foxp1bed125 mutants and wild-type sibling. I. Overview of high-fat diet feeding experiment where zebrafish were genotyped and Nile Red imaged at 35 dpf, before undergoing a 14-day high-fat diet immersion (5% chicken egg yolk) for two hours daily in addition to normal feeding regimen. Control diet was two hour immersion in system water daily. Post-diet Nile Red imaging was performed at 49 dpf. J. Violin plots showing effects of diet and genotype on average LD diameter. Statistical tests were one-way ANOVA followed by a Tukey’s HSD post-hoc test.