HSP47 levels determine the degree of body adiposity

Abstract

Adiposity varies among individuals with the influence of diverse physiological, pathological, environmental, hormonal, and genetic factors, but a unified molecular basis remains elusive. Here, we identify HSP47, a collagen-specific chaperone, as a key determinant of body adiposity. HSP47 expression is abundant in adipose tissue; increased with feeding, overeating, and obesity; decreased with fasting, exercise, calorie restriction, bariatric surgery, and cachexia; and correlated with fat mass, BMI, waist, and hip circumferences. Insulin and glucocorticoids, respectively, up- and down-regulate HSP47 expression. In humans, the increase of HSP47 gene expression by its intron or synonymous variants is associated with higher body adiposity traits. In mice, the adipose-specific knockout or pharmacological inhibition of HSP47 leads to lower body adiposity compared to the control. Mechanistically, HSP47 promotes collagen dynamics in the folding, secretion, and interaction with integrin, which activates FAK signaling and preserves PPARγ protein from proteasomal degradation, partly related to MDM2. The study highlights the significance of HSP47 in determining the amount of body fat individually and under various circumstances.

Fig. 1. The association of HSP47 expression with human body adiposity.

a Transcript enrichment analysis of human adipose tissue compared with other various tissues and cells (GTEx). Gene ontology (b) and KEGG pathway (c) analyses of adipose-enriched genes. d Schematic diagram of determinant screening for body adiposity (Human adipose tissue-enriched genes GTEx: Adipose-enrichment top-ranked 500 genes from TPM > 100 in human adipose tissue; Feeding induced genes in human subcutaneous adipose tissue (GSE154612: fold change > 1.1, p value < 0.05); Overeating-induced genes in human adipose tissue (GSE28005: fold change > 1.1, p value < 0.15); Obesity-induced gene in human subcutaneous adipocytes (GSE80654: fold change > 1.1, p value < 0.15);Body adiposity-associated genes in GWAS. e HSP47 gene expressions in human tissues (GTEx). f HSP47 protein intensity extracted from Proteomics DB. g Human subcutaneous adipose tissue from healthy subjects under fasting and feeding conditions (GSE154612, n = 11, 2 h or 24 h after standard meal; p = 0.0013). h Human subcutaneous adipose tissue before and after over-eating (GSE28005, n = 12, high-fat diet for 56 days). i Human subcutaneous adipocytes from non-diabetic lean subjects and subjects with obesity (lean, n = 10; obese, n = 10; p = 0.0071; GDS3602). j Human subcutaneous adipose tissue from monozygotic twin-pairs discordant for body adiposity (GSE92405, n = 25, intrapair difference in BMI > 3 kg/m²; p = 0.024). k Human subcutaneous adipose tissue from healthy young male subjects before and after 12 weeks of exercise training (GSE116801, n = 10, 60–80 min cycling/day, 5 days/week; p = 0.0114). l Human subcutaneous adipose tissue from overweight and obese before and after low-calorie diet interventions (GSE77962; before, n = 51; after, n = 51; 1250 kcal/d diet for 12 weeks or 500 kcal/d diet for 5 weeks; p = 0.0039). m Human subcutaneous adipose tissue from obese female subjects before and 1 year after bariatric surgery (GSE72158, n = 42 each, Roux-en-Y gastric bypass [RYGB]; p = 0.000521). n Human subcutaneous adipose tissue from gastrointestinal cancer patients with (n = 13) or without (n = 14) cachexia (GSE20571; p = 0.000276). Correlation analyses between HSP47 gene expression in human subcutaneous adipose tissue and body adiposity traits, including fat mass (o), BMI (p), waist circumference (q), hip circumference (r), lean mass (s; fat-free mass) in a cohort METSIM study (GSE70353). Data represent the mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.001; n refers to sample size. Statistical significance was determined by Two-tailed paired (g, h, j, k, m), unpaired (i, l, n) t-test, and pearson correlation test (o–s). Source data are provided as a Source Data file.

Fig. 2. Hormonal regulation of HSP47 by insulin and corticosteroids.

a Relative protein expression of HSP47 in 3T3-L1 adipocytes after insulin (100 nM; p < 0.0001) or dexamethasone (Dex;100 nM; p < 0.0052) treatments for 48 h (n = 4 each). b, Relative protein expression of HSP47 in ex vivo cultured mouse adipose tissue after insulin (100 nM; p < 0.0001) or dexamethasone (100 nM; p < 0.0001) treatments for 48 h (n = 4 each). c Relative protein expression of HSP47 in 3T3-L1 adipocytes after insulin (Ins; 100 nM; ctrl vs Insulin p = 0.0013) treatment with/without insulin receptor inhibitor (OSI-906; IRi; 0.2 µM; Insulin vs Ins.+ IRi, p = 0.0019) for 48 h (n = 3 each). d Relative protein expression of HSP47 in 3T3-L1 adipocytes after dexamethasone (100 nM; p = 0.0003) treatments with/without glucocorticoid receptor antagonist (RU486; GRi; 10 µM; Dex vs Dex+GRi, p = 0.0005) for 48 h (n = 3 each). e Relative protein expression of HSP47 in mouse adipose tissue after insulin receptor inhibitor (OSI-906; IRi; 100 mg/kg) treatment for 24 h (n = 3 each; p = 0.0045). f Relative protein expression of HSP47 in mouse adipose tissue of control (flox/flox) or adipose-specific GR knockout (AGRKO) mice after 2 weeks of corticosterone treatment (n = 8; p < 0.001). g Relative gene expression of Hsp47 in chicken adipose tissue after short-term 5 h fasting (Fast.) or ad libitum feeding (Feed.) with/without insulin neutralization by porcine anti-insulin serum (GSE35581; n = 5 each; Fast. vs Feed. p = 0.0194; Feed. vs anti-Insulin, p = 0.0215). h Relative gene expression of HSP47 in ex vivo cultured human omental adipose tissue with dexamethasone (0, 1, 10, 1000 nM; n = 3; 10 nM, p = 0.0014; 1000 nM, p = 0.0002) or cortisol (200 nM; n = 1) (GSE88966). Data represent the mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.001; n refers to sample size. Statistical significance was determined by Tukey–Kramer test (a–d, g, h) and two-tailed unpaired t-test (e and f). Source data are provided as a Source Data file.

Fig. 3. Genetic association of HSP47 gene expression with body adiposity traits.

a Schematic summary of genetic variants in HSP47 gene (NM_001235.5) and protein (NP_001226.2) loci associated with body adiposity traits (rs606452, rs668347, rs645935, and rs584961), BMI-adjusted waist and/or hip circumferences in GWAS catalog, or common variants (rs584961 and rs605040) without any reported traits. eQTL analyses of genetic variants rs606452 (b; genotype CC n = 338; AC n = 127; AA n = 18), rs668347 (c; genotype GG n = 339; TG n = 127; TT n = 17), rs645935 (d; genotype CC n = 347; TC n = 124; TT n = 12), rs584961 (e; genotype GG n = 390; AG n = 84; AA n = 9), rs605040 (f; genotype TT n = 116; TC n = 246; CC n = 121), and rs646474 (g; genotype TT n = 132; TC n = 235; CC n = 116) in cultured human fibroblast cells. eQTL analyses of genetic variants rs606452 (h; genotype CC n = 315; AC n = 133; AA n = 21), rs668347 (i; genotype GG n = 316; TG n = 132; TT n = 21), rs645935 (j; genotype CC n = 325; TC n = 130; TT = 14), rs584961 (k; genotype GG n = 370; AG n = 89; AA n = 10), rs605040 (l; genotype TT n = 129; TC n = 240; CC n = 100), and rs646474 (m; genotype TT n = 140; TC n = 231; CC n = 98) in human omental adipose tissue. Allele frequency of rs606452 (n), rs668347 (o), rs645935 (p), and rs584961 (q) variants from 1000 genome projects. Statistical p-value is from a t-test that compares observed normalized enrichment score (NES) from single-tissue eQTL analysis to a null NES of 0. Box plots are shown as median and 25th and 75th percentiles; points are displayed as outliers if they are above or below 1.5 times the interquartile range; n refers to sample size. Source data are provided as a Source Data file.

Fig. 4. HSP47 ablations reduce the level of body adiposity and adipocyte size.

a Schematic summary of adipose-specific Hsp47 knockout (AdHSP47KO: KO) mouse model. Body weight (b), food intake (c), visceral adipose tissue (VAT; epididymal fat) mass (d; p = 0.0396), subcutaneous adipose tissue (SAT; inguinal fat) mass (e; p = 0.0015), fat (total of VAT, SAT, and BAT) mass (f; p = 0.004) and lean mass (g) of littermate control and AdHSP47KO mice under normal diet (Control n = 17; AdHSP47KO n = 12). h Schematic summary of HSP47 inhibition (HSP47i) mouse model under fasting and refeeding condition. Body weight (i), food intake (j), VAT mass (k; p = 0.0109), SAT mass (l; p = 0.00039), FAT mass (m; p = 0.000842), and lean mass (n) of control and HSP47i mice (n = 5 each). o Representative whole-mount confocal microscopy image (scale bar indicates 50 µm) of lipid staining in control and AdHSP47KO adipose tissue (epididymal fat). p Quantification of adipocyte size in the adipose tissue (epididymal fat) of control and AdHSP47KO mice (n = 3 each; 0.2–0.4, p = 0.027; 0.8–1.2, p = 0.0102; 1.6–2.0; p = 0.0184; 2.0–2.4, p = 0.00053; 2.4–2.8, p = 0.017). q Representative hematoxylin and eosin (H&E) staining image (scale bar 50 µm; grayscale) of control and HSP47i adipose tissue (epididymal fat). r Quantification of adipocyte size in the adipose tissue (epididymal fat) of control and HSP47i mice (n = 3 each; 0.15–0.3, p = 0.037; 0.3–0.5, p = 0.0168; 0.5–1, p = 0.00052; 2–2.5, p = 0.01557; 2.5–3, p = 0.00012; 3–3.5, p = 0.013; 3.5–4.5, p = 0.010). Data represent the mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.001; n refers to sample size. Statistical significance was determined by two-tailed unpaired t-test. Source data are provided as a Source Data file.

Fig. 5. HSP47 ablations decrease FAK signal and PPARγ protein.

Confocal microscopy image (a) of extracellular collagen protein (anti-Collagen VI) in the epididymal adipose tissue of control and AdHSP47KO (n = 4 each); the arithmetic mean fluorescence intensity for collagen (b; p = 0.0073). Western blot image (c) of HSP47, focal adhesion kinase (FAK) signal, and PPARγ proteins in the epididymal adipose tissue of control and AdHSP47KO (n = 6); the densitometry of HSP47 (d; p = 2.96e–12), phosphorylated FAK (e; Phospho-FAK; pFAK; p = 0.000466), and PPARγ (f; p = 2.19e–6) proteins. Western blot image (g) of FAK signal and PPARγ proteins in the epididymal adipose tissue of control and HSP47i (n = 4 each); the densitometry of pFAK (h; p = 0.011) and PPARγ (I; p = 0.00074) proteins. Western blot image (j) of HSP47, FAK signal and PPARγ proteins in 3T3-L1 adipocytes after control or Hsp47 siRNA (n = 3 each); the densitometry of HSP47 (k; p = 6.59e–6), pFAK (l; p = 0.0239), and PPARγ (m; p = 0.0121) proteins. Western blot image (n) of Collagen multimer forms (monomer, dimer, and trimer; non-reduced; anti-Collagen VI) in 3T3-L1 adipocytes after control or HSP47 siRNA (n = 3 each); the densitometry (o) of monomer (p = 0.0139), dimer (p = 0.00043), and trimer (p = 0.003) forms of collagen protein. Western blot image (p) of Integrin-bound Collagen (IP; anti-Integrin β1, IB; anti-Collagen VI), FAK signal (Input), and PPARγ (Input) proteins in 3T3-L1 adipocytes after HSP47i for 3 h (n = 3 each); the densitometry of Collagen (q; anti-Collagen VI; p = 0.00783), pFAK (r; p = 0.0058) and PPARγ (s; p = 1.59e–5) proteins. Western blot image (t) of FAK signal and PPARγ proteins in 3T3-L1 adipocytes after RGD peptide treatment (0, 1, 4 mM) for 16 h (n = 3 each); the densitometry of pFAK (u; p = 1.74e−5) and PPARγ (v; p = 0.0064) proteins (control vs RGD 4 mM). Western blot image (w) of Integrin β1, FAK signal, and PPARγ proteins in 3T3-L1 adipocytes after Itgb1 siRNA for 48 hours (n = 3 each); the densitometry of Integrin β1 (x; p = 2.53e−5), pFAK (y; p = 0.00035) and PPARγ (z; p = 0.000361) proteins. Data represent the mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.001; n refers to sample size. Statistical significance was determined by two-tailed unpaired t-test. Source data are provided as a Source Data file.

Fig. 6. The low body adiposity of HSP47 ablations is attributed to the decreased stability and activity of PPARγ.

Western blot image (a) of FAK signal and PPARγ proteins in 3T3-L1 adipocytes treated with FAK inhibitor (PF-573228; FAKi; 0.5 and 1 µM; n = 3 each) for 3 h; the densitometry graph (b; control vs 0.5 µM, p = 0.0063; control vs 1 µM, p < 0.0001) of PPARγ protein. Western blot image (c) of focal adhesion kinase (FAK) signal in 3T3-L1 adipocytes after FAKi (1 µM) treatment with cycloheximide (50 ug/ml) for 0, 30, 60, and 90 min; the densitometry graph (d) of PPARγ protein. e Western blot image of ubiquitinated PPARγ protein in FLAG-PPARγ and HA-Ub overexpressed HEK293T cells after FAKi (10 µM) treatment with/without MG132 (10 µM) for 3 h. f Western blot image of FAK signal and PPARγ proteins in 3T3-L1 adipocytes after FAKi (1 µM) treatment with/without MG132 (10 µM) for 0, 2 and 3 h. Western blot image (g) of FAK signal and PPARγ proteins in 3T3-L1 adipocytes after FAKi (1 µM) treatment with/without MDM2i (MI-773; 0, 1, 2, and 4 µM) for 3 hours; the densitometry graph (h; Control vs FAKi, p < 0.0001; FAKi vs FAKi + MDM2i 4 µM, p = 0.0289) of PPARγ protein. i Western blot image of ubiquitinated PPARγ protein in FLAG-PPARγ and HA-Ub overexpressed HEK293T cells after HSP47i (200 µM) treatment with/without MG132 (10 µM) for 3 h. j Western blot image of FAK signal and PPARγ proteins in 3T3-L1 adipocytes after HSP47i (200 µM) treatment with/without MG132 (10 µM) for 0, 2 and 3 h. Western blot image (k) of FAK signal and PPARγ proteins in 3T3-L1 adipocytes after HSP47i (200 µM) treatment with/without MDM2i (MI-773; 0, 2, and 4 µM) for 3 h; the densitometry graph (l; Control vs HSP47i, p = 0.0002; HSP47i vs HSP47i + MDM2i 4 µM, p = 0.0073) of PPARγ protein. m Realtime qPCR data of PPARγ and the target genes (Adipoq, p = 0.01045; Fabp4, p = 0.007259; Cd36, p = 0.048212; Lpl, p = 0.051; Scd1, p = 0.016136) in 3T3-L1 adipocytes after HSP47i (200 µM) treatment for 3 h (n = 3 each). n Realtime qPCR data of PPARγ and the target genes (Adipoq, p = 0.0065, Cd36, p = 0.017; Lpl, p = 0.0196; Scd1, p = 0.000272) in in vivo mouse adipose tissue (epididymal fat) after HSP47i (100 mg/kg; single shot followed by 24 h fasting and 24 h refeeding) treatment (n = 4 each). White adipose tissue (WAT) mass (o the total mass of epididymal and inguinal fat; p WAT percentage per body weight) of control (Ctrl) and AdHSP47KO (KO) with/without pioglitazone (pio; 15 mg/kg; every other day for a week) treatment (Ctrl n = 17; KO n = 12; KO + Pio n = 10). The total mass of WAT, Ctrl vs KO, p = 0.0096; KO vs KO + Pio, p = 0.0068. WAT percentage, Ctrl vs KO, p = 0.0395. q, r White adipose tissue (WAT) mass (q; total mass of WAT, r; WAT percentage) of control (Ctrl) and HSP47i (100 mg/kg) with/without pioglitazone (pio; 30 mg/kg) treatment (n = 5 each). WAT total mass, Ctrl vs HSP47i, p = 0.0122; HSP47i vs HSP47i + Pio, p = 0.0122. WAT percentage, Ctrl vs HSP47i, p < 0.0001; HSP47i vs HSP47i + Pio, p = 0.0189. Data represent the mean ± SEM. *p < 0.05, **p < 0.01, and ***p < 0.001; n refers to sample size. Statistical significance was determined by Tukey–Kramer test (h and l), two-tailed unpaired t-test (m and n), and Wilcoxon test (o–r). Source data are provided as a Source Data file.

Fig. 7. Graphical summary.

The level of body fat in individuals is subject to considerable variation and influenced by numerous factors, such as physiological, pathological, environmental, hormonal, and genetic conditions. In this study, we have identified HSP47, a collagen-specific chaperone, as a key determinant of body adiposity that is abundant in fat tissues and strongly correlated with human conditions of high or low body adiposity. HSP47 expression is increased by feeding, overeating, and obesity, while it is decreased by fasting, exercise, calorie restriction, bariatric surgery, and cachexia. It is also significantly correlated with several body adiposity traits, including body fat mass, BMI, waist, and hip circumferences. Additionally, insulin and glucocorticoids, endocrine hormones that are associated with systemic nutritional states and body fat storage/loss, respectively up- and down-regulate the expression of HSP47. In humans, increased HSP47 gene expression resulting from intron or synonymous variants is associated with higher body adiposity traits, such as hip and waist circumferences. Similarly, in mice, the ablation of HSP47 results in significantly lower body adiposity compared to the control group. Mechanistically, HSP47 plays a critical role in promoting collagen protein dynamics, including folding, secretion, and interaction with integrin. This, in turn, activates FAK signaling and preserves PPARγ protein from proteasomal degradation, partly related to MDM2. Our findings emphasize the critical role of HSP47 in determining body adiposity, providing valuable insights into the individual variability and differences in body adiposity traits observed in diverse circumstances.