Phenotypic characterization of Adig null mice suggests roles for adipogenin in the regulation of fat mass accrual and leptin secretion

Summary Adipogenin (Adig) is an adipocyte-enriched transmembrane protein. Its expression is induced during adipogenesis in rodent cells, and a recent genome-wide association study associated body mass index (BMI)-adjusted leptin levels with the ADIG locus. In order to begin to understand the biological function of Adig, we studied adipogenesis in Adig-deficient cultured adipocytes and phenotyped Adig null (Adig−/−) mice. Data from Adig-deficient cells suggest that Adig is required for adipogenesis. In vivo, Adig−/− mice are leaner than wild-type mice when fed a high-fat diet and when crossed with Ob/Ob hyperphagic mice. In addition to the impact on fat mass accrual, Adig deficiency also reduces fat-mass-adjusted plasma leptin levels and impairs leptin secretion from adipose explants, suggesting an additional impact on the regulation of leptin secretion.

(C) Body weights recorded weekly from WT and Adig -/female mice (aged 5 weeks) fed a CD or HFD for 24 weeks (n=6-12). (D) Tissue weights measured at the end of the experiment (n= 3-6).
Data is expressed as mean ± SEM and was analysed by 2-way ANOVA (A and C p value obtained by comparing genotypes and diets) or 1-way ANOVA (B and D compared to WT CD) with Bonferroni multiple comparison posthoc testing. *p <0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. White adipose tissues -Inguinal (IngWAT), epididymal (EpiWAT), mesenteric (MesWAT), and retroperitoneal (RetroWAT). Brown adipose tissue (BAT). Data is presented as scatter plot and ANCOVA shown as p value for slope difference between genotypes and diets. incubation. Data is expressed relative to the concentrations secreted in cells exposed to the scrambled siRNA (set as 100) (n=4-7).
(E-H) Knockdown of Adig was performed in 3T3-L1 adipocytes (from day 5 onwards) and repeated every 2 days until D8 of adipocyte differentiation when media was collected and cells were harvested for analysis, as described In 3T3-L1 cells, Adig expression was effectively knocked down (KD) by repeated treatment (alternate days) with Adig siRNA (S3C, S3G). When the KD was initiated prior to the addition of differentiation cocktail, adipocyte differentiation was globally impaired as reflected by reduced lipid accumulation and reduced expression of several well-established markers of adipocyte differentiation ( Figure S3A-C). This made it difficult to ascertain an independent effect of Adig KD on leptin expression or secretion so we also compared leptin expression in cells only exposed to Adig siRNA 5 days after the induction of differentiation. In this setting lipid accumulation and mRNA expression of typical adipocyte markers (Pparγ2, Glut4 and Plin1) was unaffected whereas leptin mRNA was again significantly lower than in the control cells ( Figure S3E-G). Furthermore, the levels of leptin detected in the media were lower in the KD cells ( Figure S3D, S3H). Notably, secreted adiponectin levels, as well as mRNA expression, were also reduced in the Adig siRNA ( Figure S3C, S3D and S3G, S3H), arguing against a leptinspecific effect in this experimental paradigm. In order to confirm the 3T3-L1 results, we also isolated primary preadipocytes from the stromovascular fraction (SVF) of IngWAT derived from WT and Adig -/mice. In these cells, lipid accumulation and gene expression analysis suggested that Adig was required for normal differentiation and that leptin expression was reduced in Adig null cells in keeping with this defect ( Figure S3I-K). This defect could be effectively overcome by incubating the cells with rosiglitazone, a PPARγ agonist ( Figures S3I-K). The mRNA expression of Pparγ2, Plin1, leptin (Lep) and adiponectin (Adipoq) showed a clear reduction in the Adig null derived SVF differentiated adipocytes in line with the lipid accumulation ( Figure S3K).
These data collectively suggest that adipogenin deficiency impairs adipogenesis in cultured adipocytes and in adipocyte precursors derived from the SVF. Leptin expression is then also reduced in these cells but this may simply reflect the impairment in adipocyte differentiation rather than an additional direct effect.  Figure S4. Adig gene, protein, phylogeny and membrane topology (Related to STAR methods section: Gene, protein, phylogeny and membrane topology (Bioinformatics)) Why adipogenin deficiency is associated with the phenotypes reported herein remains unclear, so to begin to address this question we performed a bioinformatics analysis of its evolution and amino acid sequence.
Homology searches in genomic databases identified ADIG paralogues in Mammalia and Sauria but not in other vertebrates. ADIG is transcribed into two splice variants ( Figure S4A). Variant 1 is the only one investigated experimentally thus far. It is translated into a short protein, the existence of which has been confirmed by immunoblotting (Ren et al., 2016b). Mammalian sequences of variant 1 are aligned in Figure S4B. Variant 2 gives a much larger transcript (confirmed by several cDNAs in transcriptomic databases) but the putative protein (197 amino acids) has not been investigated. Only the larger variant is transcribed from saurian genes. The gene structure of representative saurian ADIG and of the two mammalian variants is compared in Figure S4A. The shorter human variant is obtained through intron retention and almost immediate termination. Although the gene structures of the mammalian and saurian larger variants are similar, the homology exists only in the first exon.
The rest of the larger mammalian variant diverged freely suggesting no or disappearing function. Examples of mammalian variants that could be translated into proteins are compared in Figure S4C. Others contain deletions leading to a premature termination. In sharp contrast, the homologous part in exon 1 is exceptionally strongly evolutionarily constrained indicating an important biological function. The saurian protein is constrained over the whole sequence with the N-terminus (first exon) showing the strongest conservation ( Figure S4D).
The predicted structure and membrane topology of human and saurian ADIG is shown in Figure S4E. A signal peptide is not apparent and both mammalian variants have only one predicted transmembrane (TM) helix nearly identical to the saurian first TM helix ( Figure S4E). Variant 1 only contains a short intra-and extracellular segment. The protein could be localized in the endoplasmic reticulum or lysosome according to the Localisation Signal Database (Negi et al., 2015).
Homology searches found one remote protein homologue Smlr1 (Small leucine-rich protein) of as yet unknown function predominantly expressed in adipocytes (proteinatlas.org). The membrane topology of the mammalian Adig is very similar to channel regulatory proteins phospholamban and sarcolipin (Shaikh et al., 2016). Saurian Adig shows a distant similarity to several transporters or channels. These initial observations tentatively hint at the potential function/s of Adig, without offering any clear insights into the observed physiological phenotypes. Figure S4. Adig gene, protein, phylogeny and membrane topology (Related to STAR methods section: Gene, protein, phylogeny and membrane topology (Bioinformatics))