Collagen-specific molecular chaperone Hsp47 in inguinal white adipose tissue promotes high-fat diet-induced inflammatory gene expression in male mice

Abstract

A chronic high-fat diet (HFD) leads to obesity through a process involving adipocyte hypertrophy, hyperplasia, induction of pro-inflammatory gene expression, and ultimately fibrosis that impairs systemic health. Adipocytes are surrounded by an extracellular matrix (ECM) containing collagen. The interaction between adipocytes and the ECM regulates adipocyte turnover under both physiological and HFD conditions. Previous studies have shown that the collagen-specific molecular chaperone Hsp47 (SerpinH1) increases with overfeeding and decreases with short-term fasting. Hsp47 affects adipocyte size through collagen-integrin interactions. However, the long-term contribution of Hsp47 under chronic HFD conditions remains unclear, particularly whether there are depot-specific differences between subcutaneous inguinal white adipose tissue (ingWAT) and visceral epididymal white adipose tissue (epiWAT). In this study, mice fed an HFD for 12 weeks exhibited increased Hsp47 mRNA and protein specifically in adipose tissue. Adipocyte-specific Hsp47 knockout (aKO) mice exhibited reduced adipocyte size and decreased fibrosis markers and pro-inflammatory gene expression, specifically in ingWAT, with minimal impact in epiWAT. Serum analysis revealed that the metabolic deterioration induced by an HFD was alleviated in Hsp47 aKO mice. These results suggest that increased Hsp47 in ingWAT during an HFD contributes to adipose tissue dysfunction by promoting pro-inflammatory gene expression. These findings highlight Hsp47 as a potential therapeutic target for obesity-related metabolic disorders.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-45003-4.

Introduction

In recent decades, metabolic disorders have emerged as one of the most pressing public health issues worldwide. Among them, obesity is a major global health challenge. From 1990 to 2021, the global prevalence of adult obesity (BMI ≥ 30 kg/m²) has increased substantially, rising from 10% to 20.8% in women and from 5.8% to 14.8% in men. Prevalence rates rose steadily across both developed and developing countries, highlighting obesity as a widespread public health concern¹. Obesity is tightly connected to cardiovascular disease, hypertension, and several types of cancer². Therefore, it is crucial to further improve our understanding of the cellular and molecular mechanisms underlying obesity pathogenesis. This will ultimately help identify innovative therapeutic strategies.

Adipocytes store lipids as an energy reserve under normal conditions; however, excess energy intake causes obesity, which is characterized by adipocyte hypertrophy (increased cell size) and hyperplasia (increased cell number). While adipose tissue expansion is initially an adaptive response, chronic overnutrition can lead to pathological changes, including inflammation, fibrosis, and systemic metabolic dysfunction. High-fat diet (HFD) feeding in mice is a widely used model of diet-induced obesity and recapitulates these features. Adipocytes under HFD conditions exhibit a fibroblast-like transcriptional signature, characterized by the upregulation of extracellular matrix (ECM) genes, including various collagens (e.g., Col1a1 and Col3a1) and fibronectin³. HFD induces a time-dependent progression of adipose tissue pathology. In the early stages, HFD triggers inflammation, marked by elevated inflammatory cytokines and, in many settings, increased immune cell infiltration. Prolonged HFD feeding leads to later stages, including fibrosis, characterized by excessive collagen deposition and ECM remodeling⁴. This exacerbates adipose tissue dysfunction and systemic metabolic derangements⁵.

Collagen deposition and ECM remodeling are central to the development and function of adipose tissue⁶. Under physiological conditions, the ECM provides a compliant scaffold that maintains tissue integrity and allows for dynamic changes in adipocyte size and number. In particular, the mechanical properties of collagen networks, such as stiffness and fiber organization, are known to directly influence adipocyte hypertrophy and proliferation^7,8. Several cell types contribute to ECM production in adipose tissue, including mature adipocytes and stromal vascular fraction cells such as preadipocytes and fibroblasts. However, their relative contributions and regulation are not fully understood⁹. Under HFD conditions, pro-fibrotic cytokines (e.g., TGF-β, IL-13) are induced and immune cell infiltration increases, which together drive collagen synthesis and ECM remodeling¹⁰. The resulting excess deposition of collagen, particularly types I and VI, stiffens adipose tissue, promotes local hypoxia, and sustains a feed-forward loop of inflammation and fibrosis, ultimately compromising adipose plasticity and systemic metabolic homeostasis¹¹.

Heat shock protein 47 (Hsp47; SerpinH1) is a collagen-specific molecular chaperone located in the endoplasmic reticulum (ER). Hsp47 is essential for correct procollagen folding and for the secretion of procollagen out of the cells¹². Hsp47 binds to triple-helical procollagen and prevents its aggregation by inhibiting aberrant interactions between procollagen molecules. Hsp47 plays a crucial role in collagen biology involved in physiological processes such as early development, bone formation, and wound healing. However, in fibrotic diseases, Hsp47 acts as a pathogenic factor that promotes abnormal collagen accumulation. From a therapeutic perspective, Hsp47 is recognized as a therapeutic target for treating various fibrotic diseases, including those affecting the liver, lungs, heart, kidneys, and intestines¹³. Knockdown of Hsp47 in hepatic stellate cells using specific siRNA delivery resolved liver fibrosis¹⁴. Recently, Shin et al. demonstrated that Hsp47 regulates the protein levels of peroxisome proliferator-activated receptor gamma (PPARγ), which is a key determinant of adipocyte differentiation and function. Hsp47 expression in adipocytes increases under conditions of overnutrition and obesity but decreases during fasting and caloric restriction¹⁵. Furthermore, adipocyte-specific knockout or pharmacological inhibition of Hsp47 reduces fat mass and adipocyte size by modulating collagen-integrin interactions and activating focal adhesion kinase signaling. These findings suggest that Hsp47 contributes to the regulation of adipocyte size and adipose tissue expansion during short-term metabolic changes. Despite these insights, the role of Hsp47 during chronic HFD exposure, particularly its depot-specific contribution in subcutaneous inguinal white adipose tissue (ingWAT) versus visceral epididymal white adipose tissue (epiWAT), remains unclear.

Given the central role of Hsp47 in collagen biosynthesis and its connection to inflammation and fibrosis, we hypothesized that Hsp47 exacerbates adipose tissue fibrosis and inflammation during prolonged HFD feeding, thereby worsening metabolic dysfunction. In this study, we examined the long-term impact of Hsp47 in adipose tissue under chronic HFD, focusing on depot-specific effects and the potential therapeutic benefits of targeting Hsp47 in obesity-related metabolic diseases.

Results

High-fat diet induces the expression of collagen chaperone Hsp47 in adipose tissue

To investigate the effect of Hsp47 on chronic metabolic stress, C57BL/6J mice were fed either ND or HFD for 12 weeks starting at 8 weeks of age. This duration is sufficient to trigger early inflammatory transcriptional responses to HFD⁴. Western blot analysis revealed that Hsp47 protein levels were significantly elevated in epiWAT, ingWAT, and brown adipose tissue (BAT) (Fig. 1a, b, Supplementary Fig. 1a), whereas kidney and lung showed no significant change (Supplementary Fig. 1b, c). Quantitative PCR further detected the transcriptional up-regulation of the Hsp47 (Serpinh1) relative to GAPDH expression (Fig. 1c). These findings showed that HFD transcriptionally induces Hsp47 at the mRNA level, resulting in increased Hsp47 protein levels specifically in adipose tissues.

Fig. 1.

High-fat diet induces Hsp47 expression in adipose tissue. (a) Immunoblot of Hsp47 and GAPDH in epiWAT (epididymal white adipose tissue) and ingWAT (inguinal white adipose tissue) from mice fed a normal diet (ND) or high-fat diet (HFD) for 12 weeks. (b) Quantification of Hsp47 protein (normalized to GAPDH) in epiWAT and ingWAT. (c) Quantification of Hsp47 mRNA (normalized to Gapdh) in epiWAT and ingWAT. (d) Immunoblots confirming adipose-specific Hsp47 deletion. Samples: ND flox (ND-fed Hsp47 flox/flox), ND aKO (ND-fed Adipoq-Cre; Hsp47 flox/flox), HFD flox (HFD-fed Hsp47 flox/flox), and HFD aKO (HFD-fed Adipoq-Cre; Hsp47 flox/flox). Blots shown for Hsp47, the ER chaperone Grp94, and loading controls (β-actin) in epiWAT and lung. (e) Quantification of Hsp47 protein level in epiWAT, (f) Quantification of Hsp47 protein level in lung tissue, (g) Quantification of Grp94 (ER stress marker) protein level in epiWAT. Data are presented as mean ± SD. Group comparisons were assessed by two-tailed t-tests. Variance equality was evaluated using the Fligner–Killeen test; when group variances were not significantly different (P > 0.05), Student’s t-test was applied, otherwise Welch’s t-test was used. *, **, *** indicate P < 0.05, 0.01, 0.001, respectively; n.s., not significant. Exact P values are provided in Supplementary Table S1.

To elucidate the functional role of Hsp47 in adipose tissue, we generated adipose tissue-specific Hsp47 knockout (aKO) mice by crossing Hsp47 flox/flox mice with Adipoq-Cre mice¹⁶. Western blot analysis confirmed the efficient knockout of Hsp47 specifically in adipose tissues (epiWAT and ingWAT) in both ND and HFD conditions (Fig. 1d, e, Supplementary Fig. 2a, b), while Hsp47 levels remained unchanged in the lung and kidney (Fig. 1f, Supplementary Fig. 1b, c). In previous studies of highly collagen-expressing cells including chondrocytes and hepatic stellate cells, Hsp47 knockout led to ER stress due to the accumulation of misfolded procollagen in the ER^16,17. To assess ER stress, we examined the protein levels of the ER chaperone Grp94, an ER stress marker. No significant increase in Grp94 was observed in the adipose tissues of Hsp47 aKO mice (Fig. 1g), indicating that Hsp47 deletion did not induce ER stress. Similarly, HFD did not elicit an ER stress response.

Body weight was monitored across groups. Under ND conditions, Hsp47 knockout did not significantly affect body weight (Supplementary Fig. 1e). Under HFD, an analysis of body weight over time using a linear mixed-effects model revealed a significant interaction between genotype (flox and aKO) and time (weeks) (p = 0.0039) (Supplementary Fig. 1f). This indicates that aKO mice gained weight more slowly than flox mice over the course of the HFD period. In response to the HFD, the weights of the epiWAT and ingWAT increased in both the aKO and flox mice (Supplementary Fig. 1g, h). Although differences in depot mass between flox and aKO did not reach statistical significance, ingWAT mass tended to be lower in aKO mice than in flox controls.

Adipocyte-specific deletion of Hsp47 selectively limits adipocyte hypertrophy and fibrosis in inguinal white adipose tissue under high-fat diet

To examine depot-specific morphological changes in adipose tissue, we analyzed sections of epiWAT and ingWAT from Hsp47 flox/flox (flox) and adipocyte-specific Hsp47 knockout (aKO) mice after 12 weeks on ND or HFD (Fig. 2a). In flox mice, HFD produced a clear rightward shift of adipocyte size distributions in both depots (two-sample Kolmogorov–Smirnov tests on mouse-level empirical cumulative distribution functions (ECDFs) with 10,000 permutations: epiWAT D = 0.270, p = 0.0022; ingWAT D = 0.195, p = 0.0020; Fig. 2b, c). In aKO mice, HFD still shifted the distribution in epiWAT (D = 0.262, p = 0.0038; Fig. 2d), but not in ingWAT (D = 0.057, p = 0.159; Fig. 2e), indicating that adipocyte Hsp47 is required for HFD-induced hypertrophy specifically in ingWAT. Cross-genotype comparisons supported this pattern (Supplementary Fig. 3): among all depot-by-diet combinations, a significant difference between flox and aKO was observed only in ingWAT under HFD (Supplementary Fig. 3d).

Fig. 2.

Adipocyte Hsp47 drives collagen deposition and adipocyte hypertrophy in inguinal white adipose tissue under a high-fat diet.

(a) Representative Sirius Red–stained sections of epididymal (epiWAT) and inguinal (ingWAT) white adipose tissue from Hsp47 flox/flox (flox) and adipocyte-specific Hsp47 knockout (aKO; Adipoq-Cre; Hsp47 flox/flox) mice after 12 weeks on a normal diet (ND) or high-fat diet (HFD). Scale bars, 100 μm.

Collagen deposition was quantified by Sirius Red staining. Under ND, the percentage of Sirius Red-positive area per microscopic field did not differ between flox and aKO, indicating similar baseline collagen per tissue area. In flox mice, 12 weeks of HFD did not change collagen-positive area in either depot (Fig. 2a, f). Despite the robust increase in adipocyte size with HFD (Fig. 2b, c), this suggests that at this time point the tissue had not yet entered a phase of overt pathological collagen accumulation. In aKO mice, HFD did not alter collagen deposition in epiWAT when compared with aKO under ND, whereas in ingWAT the collagen-positive area under HFD was significantly lower in aKO than in flox (Fig. 2a, f). These data indicate that, relative to flox controls, collagen accumulation (reflecting altered synthesis and/or turnover) is selectively blunted in ingWAT of aKO mice. Together with the depot-restricted attenuation of adipocyte hypertrophy (Fig. 2b–e), these findings support a depot-specific role for adipocyte Hsp47: it promotes adipocyte hypertrophy and collagen deposition in subcutaneous ingWAT, whereas remodeling of visceral epiWAT proceeds largely independently of adipocyte-derived Hsp47.

Hsp47 deletion suppresses fibrosis- and inflammatory gene expression specifically in inguinal white adipose tissue

To define depot-specific transcriptional effects of adipocyte Hsp47 loss, we quantified mRNA levels of extracellular-matrix and inflammatory genes in epiWAT and ingWAT (Fig. 3). As expected, HFD increased the expression of fibrillar collagen genes including Col1a1, Col3a1, and Col6a1 in both depots. In epiWAT, adipocyte-specific Hsp47 knockout (aKO) did not measurably alter collagen gene expression relative to flox controls. In ingWAT, aKO mice showed a modest, directionally consistent reduction in Col1a1, Col3a1, and Col6a1 compared with flox under HFD. Although these changes were small, they align with the reduced collagen deposition detected histologically (Fig. 2).

Fig. 3.

Adipose Hsp47 knockout attenuates HFD-induced inflammatory and fibrotic gene expression in ingWAT. Relative mRNA expression of Col1a1, Col3a1, Col6a1, Fn1 (fibronectin), Il6 (interleukin-6), Tnfa (tumor necrosis factor-α), and Ccl2 (monocyte chemoattractant protein-1) in epididymal (epiWAT) and inguinal (ingWAT) white adipose tissue from mice fed a normal diet (ND) or high-fat diet (HFD) for 12 weeks, measured by qPCR (normalized to Gapdh). Data are presented as mean ± SD. Group comparisons were assessed by two-tailed t-tests. Variance equality was evaluated using the Fligner–Killeen test; when group variances were not significantly different (P > 0.05), Student’s t-test was applied, otherwise Welch’s t-test was used. *, **, *** indicate P < 0.05, 0.01, 0.001, respectively; n.s., not significant. Exact P values are provided in Supplementary Table S1.

Stronger effects were observed for inflammatory and matrix-associated genes. In ingWAT from HFD-fed aKO mice, Fn1 (fibronectin) as well as key cytokines/chemokines, Il6 (interleukin-6), Tnfa (tumor necrosis factor-α), and Ccl2 (monocyte chemoattractant protein-1), were lower than those in HFD-fed flox mice. In contrast, epiWAT showed no consistent Hsp47 knockout effect on these transcripts. Because IL-6, TNFα, and CCL2 are canonical drivers of adipose inflammatory signaling, their selective reduction in ingWAT is consistent with the depot-specific limitation of fibrosis and adipocyte hypertrophy observed in Fig. 2. As a coarse assessment of macrophage abundance, we also examined the macrophage marker F4/80 in bulk ingWAT lysates by immunoblotting (Supplementary Fig. 4). In this 12-week HFD model, F4/80 protein levels did not significantly increase with HFD, and no significant difference was detected between HFD-flox and HFD-aKO groups. Together, these data indicate that adipocyte Hsp47 preferentially supports a pro-fibrotic and pro-inflammatory gene program in subcutaneous ingWAT, with minimal impact on visceral epiWAT.

Hsp47 deletion alleviates high-fat diet-induced systemic metabolic disturbances and renal stress-related changes

We next assessed whether loss of adipocyte Hsp47 influences systemic physiology (Fig. 4). Under ND, blood chemistry values were comparable between flox and aKO mice, indicating that Hsp47 deficiency does not perturb basal metabolic status.

Fig. 4.

Adipose-specific Hsp47 knockout mitigates HFD-associated systemic and renal stress markers. Serum analytes: glucose, HDL-C (high-density lipoprotein cholesterol), inorganic phosphate (IP), calcium (Ca), and albumin; Urine analytes: urea nitrogen (U-UN), creatinine (U-CRE), uric acid (U-UA), potassium (U-K), and inorganic phosphate (U-IP). Data are presented as mean ± SD. Group comparisons were assessed by two-tailed t-tests. Variance equality was evaluated using the Fligner–Killeen test; when group variances were not significantly different (P > 0.05), Student’s t-test was applied, otherwise Welch’s t-test was used. *, **, *** indicate P < 0.05, 0.01, 0.001, respectively; n.s., not significant. Exact P values are provided in Supplementary Table S1.

In flox mice, HFD produced the expected systemic changes, including higher glucose and an altered HDL profile (HDL-C). In addition, HFD was associated with increases in serum inorganic phosphate and total calcium, disturbances that have been reported in rodent models of diet-induced metabolic stress and can precede overt renal dysfunction¹⁷. In aKO mice, serum inorganic phosphate and calcium were significantly lower than in HFD-flox and showed partial normalization toward ND levels, whereas glucose and HDL-C were not significantly different from HFD-flox.

To evaluate kidney involvement, we examined a panel of serum and urine markers related to nitrogen and mineral handling, including urea nitrogen, creatinine-related indices, uric acid, potassium, phosphate, and protein. In HFD-fed flox mice, several of these markers shifted in a direction consistent with early renal strain^18,19, whereas values in aKO mice remained similar to ND. Although these assays do not establish clinical kidney disease, the overall pattern suggests that loss of adipocyte Hsp47 mitigates systemic and renal stress signals triggered by HFD.

Integrating these findings with the ingWAT-restricted reduction of Il6, Tnfa, and Ccl2 (Fig. 3), a plausible explanation is that dampened subcutaneous adipose inflammatory gene expression would contribute to the improved systemic profile in aKO mice. Collectively, the data point to an adipose-to-systemic axis in which adipocyte Hsp47 promotes HFD-induced metabolic and mineral dysregulation, with ingWAT making a disproportionate contribution to these systemic changes.

Discussion

In this study, we investigated the role of Hsp47 in white adipose tissue under normal and high-fat diet (HFD) conditions. Under HFD, adipocyte hypertrophy, fibrosis, and inflammatory gene expression were markedly attenuated in ingWAT of Hsp47 adipocyte-specific knockout (aKO) mice. However, comparable effects were not observed in epiWAT. By contrast, under normal diet, Hsp47 deficiency produced no overt adipose phenotype. Mechanistically, suppression of Hsp47 reduces the amount and quality of properly folded collagen in the extracellular matrix (ECM)^20,21. In HFD models, lysyl oxidase (LOX)-dependent collagen crosslinking promotes ECM stiffening and amplifies fibrosis and inflammatory gene expression²², whereas reducing ECM crosslinking and stiffness attenuates inflammatory gene expression²³. Thus, by reducing the supply of mature collagen that serves as the crosslinking substrate, Hsp47 suppression would reduce inflammatory gene expression and fibrosis under HFD. The findings in this study indicate that Hsp47 plays a role in amplifying pathological ECM remodeling in obesity and that adipocyte-specific suppression of Hsp47 can mitigate inflammatory gene expression and fibrosis at the tissue level.

This study demonstrates depot-specific phenotypic differences between ingWAT and epiWAT. One possible cause of these differences is their divergent developmental origins. As epiWAT contains a mesothelial component derived from WT1 (Wilms tumor 1)-expressing progenitors, whereas ingWAT arises from distinct mesodermal lineages, the two types of adipose tissue would have unique transcriptional and stress responses²⁴. A second reason is the different cellular composition of each depot. The composition and activity of the stromal vascular fraction cells, including preadipocytes, endothelial cells, fibroblasts and immune cells, differ depending on the depot. These differences influence ECM remodeling in adipose tissue²⁵. In this experiment, the Hsp47 knockout was restricted to adipocytes. Indeed, Western blotting of adipose tissue from aKO mice on a high-fat diet revealed a residual Hsp47 signal (see Fig. 2 and Supplementary Fig. 2). This signal likely originates from non-adipocyte compartments. Furthermore, ingWAT generally exhibits reduced mobilization of new precursor cells, instead tending towards the hypertrophy of existing adipocytes²⁶. In contrast, epiWAT shows more pronounced proliferation accompanied by ECM stiffening. This stiffening inhibits further hypertrophy while exacerbating fibrosis and inflammatory gene expression. The combination of these characteristics of lineage and microenvironment likely explains why the three observed outcomes in Hsp47 aKO mice (reduced collagen accumulation, decreased inflammatory markers such as Il6, Tnfa, and Ccl2, and suppressed increases in adipocyte size) were specific to ingWAT. This depot-specific difference in Hsp47 involvement advances our understanding of the mechanisms underlying adipose tissue-specific stress responses in the context of a high-fat diet.

Importantly, serum analyses revealed that Hsp47 deficiency partially mitigated systemic metabolic impairments caused by HFD. IngWAT has traditionally been considered to have a protective metabolic role, buffering against ectopic lipid deposition and preserving systemic homeostasis. However, our data suggest that ingWAT can become maladaptive under an HFD, contributing to systemic metabolic dysregulation. ECM remodeling and adipocyte hypertrophy in ingWAT may influence systemic lipid handling and inflammatory signaling and may be associated with improved systemic markers when attenuated in Hsp47 KO mice. Notably, this study raises the possibility that ingWAT remodeling under HFD is associated with systemic changes related to renal stress. Chronic HFD can provoke renal injury through oxidative stress, inflammation, and lipid accumulation in the kidneys^27,28 and also contribute to dysfunction in multiple organs in part due to ectopic fat accumulation²⁹. In our model, attenuation of ingWAT inflammatory gene expression/signaling coincided with partial normalization of serum inorganic phosphate and calcium. However, in qPCR analyses of a limited set of renal fibrosis and inflammatory marker genes (Supplementary Fig. 5), we did not detect significant differences between HFD-flox and HFD-aKO mice. These data do not provide direct evidence of renal protection. It is still possible that renal changes, if they are present, are not detected by these markers. Future studies using renal histology and broader molecular and functional measurements will be needed to clarify whether a fat–kidney axis operates in this model.

By contrast, under ND conditions Hsp47 deficiency produced no overt adipose phenotype, unlike tissues with high obligatory collagen output. For example, global or cartilage-specific Hsp47 deletion causes severe collagen defects and perinatal lethality, underscoring the essential role of Hsp47 in skeletal development²¹. Thus, in adipocytes Hsp47 appears dispensable at baseline but acts as a pathological amplifier during diet-induced remodeling, thereby promoting ECM deposition and inflammatory gene expression when the microenvironment demands high collagen throughput. This view aligns with evidence that adipose fibrosis and ECM stiffening accompany metabolic deterioration in obesity.

Finally, our results have important implications for therapeutic strategies targeting Hsp47. We previously reported the small-molecule Hsp47 inhibitor Col003, which disrupts the collagen-Hsp47 interaction and suppresses collagen accumulation in fibroblasts²¹ and is active in mouse models^15,30. Although Col003 has only moderate potency, more active derivatives such as Col051 − 053 have recently been developed^31,32. Our findings suggest that pharmacological inhibition of Hsp47 may attenuate pathological adipose tissue remodeling in obesity. The absence of a phenotype under normal diet conditions further implies that Hsp47 inhibition could selectively act under pathological stress without impairing physiological adipose tissue function, representing a potentially safe and effective therapeutic strategy.

In summary, we identify Hsp47 as a key regulator of pathological ECM remodeling in adipose tissue under HFD conditions. Its effects are depot-specific, most prominently in subcutaneous ingWAT, and extend to the regulation of systemic metabolism. These results provide mechanistic insight into adipose tissue remodeling and establish a foundation for therapeutic targeting of Hsp47 in obesity-associated metabolic diseases.

Limitations of this study include the use of only male mice; therefore, our findings should be validated in females. In addition, inflammatory gene expression in inguinal WAT was assessed mainly by qPCR, and we did not perform cell-type–resolved analyses (e.g., flow cytometry or immunohistochemistry) to quantify immune cell infiltration or macrophage subsets. Accordingly, our supplemental F4/80 immunoblot provides only a coarse measure of macrophage abundance. Because Adipoq-Cre targets mature adipocytes, residual Hsp47 expression may derive from stromal vascular cells; however, adipocyte/SVF fractionation was not feasible with the available materials. Finally, kidney and liver assessments were limited to transcript markers in frozen tissues without histopathology; therefore, organ-specific interpretation of serum changes (e.g., albumin) is limited in the present study.

Methods

Mouse

C57BL/6J male mice (SHIMIZU Laboratory Supplies Co) were housed in groups and fed a normal diet (Rodent Diet with 10 kcal% Fat, D12450J, ResearchDiets.com) or a high-fat diet (Rodent Diet with 60 kcal% Fat, D12492, ResearchDiets.com) for 12 weeks starting at 8 weeks of age. We used male mice to reduce variability associated with the estrous cycle in metabolic and inflammatory readouts and to align with commonly used HFD paradigms in which male mice show robust adipose tissue remodeling under chronic HFD.

Adipose tissue-specific Hsp47 conditional-knockout (aKO) mice were generated by crossing Hsp47 floxed mice³⁰ and Adipoq-Cre (adipose tissue-specific promoter) transgenic animals (B6;FVB-Tg(Adipoq-Cre)1Evdr/J, 010803, The Jackson Laboratory)³³. Littermates homozygous for the floxed allele with Adipoq-Cre (aKO; Adipoq-Cre; Hsp47 flox/flox) and Cre-negative floxed controls (flox; Hsp47 flox/flox) were used for experiments. Genotyping was performed by PCR using the following primer pairs. For floxed Hsp47 alleles and the Cre transgene: 5ʹ-GAGTGGGCTGAGCCCTCTCAAGAAAATCC-3ʹ, 5ʹ-CTTCGGTCAGGCCCAGTCCTGCCAGATG-3ʹ; and 5ʹ-ACGGACAGAAGCATTTTCCA-3ʹ, 5ʹ-GGATGTGCCATGTGAGTCTG-3ʹ, respectively. No anaesthesia was used in this study. Mice were euthanized by CO₂ inhalation in accordance with institutional guidelines. Tissues were rinsed in saline to remove blood and either snap-frozen in liquid nitrogen for biochemical analysis or fixed in zinc formalin for histological analyses.

All mouse experiments were carried out with the approval of the Kyoto Sangyo University Committee for Animal Care and Welfare (2018-38, 2019-37, 2020-07, 2021-05, and 2022-05) and all procedures were performed in accordance with relevant guidelines and regulations. All blood and urine tests were performed at Oriental Yeast Co.,Ltd. This study is reported in accordance with the ARRIVE 2.0 guidelines.

Exact sample sizes (n per group for each experiment) are listed in Supplementary Table S1 (and in the figure legends where applicable).

Immunoblotting

Frozen tissues (50–100 mg) were minced on ice in a Petri dish and homogenized in RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, protease inhibitor cocktail; 25955-11, Nacalai Tesque) using a TissueRuptor II (QIAGEN). Lysates were cleared by centrifugation (20,000 × g, 20 min, 4 °C). Protein concentration was determined with the BCA Protein Assay Kit (23225, Thermo Scientific). Primary antibodies: anti-Hsp47 (ADI-SPA-470, Enzo Life Sciences), anti-Grp94 (ADI-SPA-850, Enzo Life Sciences), anti-GAPDH (5G4-6C5, HyTest), anti-β-actin (clone C4; MAB1501, Merck) and anti-F4/80 (#70076, Cell Signaling Technology).

Blot images were prepared in accordance with the Nature Research digital image and integrity policies. When non-adjacent lanes or bands from different membranes are shown in one panel, the junction is indicated by white space. Only linear, whole-image brightness/contrast adjustments were applied uniformly; no selective enhancement, background subtraction, or splicing was performed. Uncropped scans for all cropped blots are provided in Supplementary Fig. 6.

Quantitative PCR (qPCR)

RNA from adipose tissue was isolated with RNeasy Lipid Tissue Mini Kit (74804, QIAGEN) using TissueRuptor II. RNA from kidney and liver tissue was extracted with TRIzol (Thermo Fisher). cDNA was synthesized with ReverTraAce qPCR RT Master Mix with gDNA Remover (FSQ-301, TOYOBO). qPCR used THUNDERBIRD™ Next SYBR^® qPCR Mix (QPX-201, TOYOBO) with the following primers.

Hsp47(SerpinH1): TGAGAAGTTTCACCACAAGATGG, CTTCTCGTCGTCATAGTAGTTGT,

Gapdh: AGGTCGGTGTGAACGGATTTG, TGTAGACCATGTAGTTGAGGTCA, Col1a1: GCTCCTCTTAGGGGCCACT, CCACGTCTCACCATTGGGG,

col3a1: CTGTAACATGGAAACTGGGGAAA, CCATAGCTGAACTGAAAACCACC,

col6a1: CTGCTGCTACAAGCCTGCT, CCCCATAAGGTTTCAGCCTCA,

fn1: GATGTCCGAACAGCTATTTACCA, CCTTGCGACTTCAGCCACT,

il6: CCCCAATTTCCAATGCTCTCC, GGATGGTCTTGGTCCTTAGCC,

Tnfa: ATGGCCTCCCTCTCATCAGT, TTTGCTACGACGTGGGCTAC,

Ccl2: CTTCTGGGCCTGCTGTTC, GTGGGGCGTTAACTGCAT.

Histology, Sirius Red staining, and image analysis

The adipose tissues fixed in Zinc Formalin Fixative, pH 6.25 (21516 − 3.75, Polyscience, Inc.) were paraffin-embedded, sectioned at 10 μm, and stained with Sirius Red³⁴. Images were captured on a BZ-X100 microscope (Keyence).

Adipocyte size and cell-area distributions were quantified in CellProfiler v4.2.5 using a modified “ExampleNeighbors” pipeline. Otsu thresholding was used; the smoothing filter size was set to 7. Per-image measurements were exported to SQLite, aggregated per mouse, and analyzed in R. Distributions are displayed as relative frequency (mean ± SD) in bins of 0.5 × 10³ µm².

Sirius Red–positive area. For each image, total tissue area and thresholded positive area were obtained using “the Yen method”; the mean threshold across images in a sample set was applied, and percent positive area was computed per mouse.

Statistics

All analyses used the mouse as the unit of replication. Unless stated otherwise, values are presented as mean ± SD, and tests were two-sided with a significance threshold of α = 0.05. Exact sample sizes (n per group for each experiment/contrast) are reported in the figure legends and in Supplementary Table S1.

Adipocyte size distributions. For each mouse, adipocyte cross-sectional areas were binned (bin width = 0.5 × 10³ µm²). Per-mouse histograms were normalized to sum to 1 (probabilities) and converted to empirical cumulative distribution functions (ECDFs). Group differences in the entire distribution were tested with the two-sample Kolmogorov–Smirnov (KS) statistic, using 10,000 mouse-label permutations to obtain permutation P values (unit of replication = mouse). To control multiplicity, contrasts were organized a priori into families and Benjamini–Hochberg false discovery rate (FDR) correction was applied within each family, with adjusted q-values reported: Primary family (2 tests): aKO ND vs. HFD in ingWAT and epiWAT. Secondary family (2 tests): flox ND vs. HFD in ingWAT and epiWAT (positive-control comparisons). Exploratory family (4 tests): flox vs. aKO within ND and within HFD in both depots.

Scalar outcomes: For single-number endpoints (e.g., Sirius Red–positive area), groups were compared with two-tailed t-tests. Equality of variance was assessed with the Fligner–Killeen test; Student’s t was used when variances did not differ (P > 0.05), otherwise Welch’s t was applied. Exact P and FDR-adjusted q values for all analyses are provided in Supplementary Table S2.

Software: Analyses were performed in R 4.3.3 (Angel Food Cake) on macOS (aarch64-apple-darwin20) using the packages dplyr, tidyr, readr, and ggplot2. Distribution tests used custom R scripts implementing the mouse-label permutation KS procedure (10,000 permutations). Image quantification used CellProfiler 4.2.5, and additional plots were generated in GraphPad Prism 7.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

S.I. and R.K. designed the experiments. S.I., R.K., and A.K. performed the experiments. S.I. analyzed the data. S.I. and K.N. wrote the manuscript.

Funding

This work was supported by a Grant-in-Aid for Scientific Research (S; 24227009) from the Japan Society for the Promotion of Science (JSPS) (to K.N.); a Grant-in-Aid for Young Scientists from the JSPS (grant number JP18K14393 to S.I.). This work was partially supported by the Acceleration of Transformative Research for Medical Innovation (ACT-M) Set-up Scheme and ACT-M from the Japan Agency for Medical Research and Development.

(b–e) Adipocyte cross-sectional area distributions plotted as relative frequency (mean ± SD) by size bin (bin width = 0.5 × 10³ µm²). Panels show ND vs. HFD within genotype and depot: (b) epiWAT, flox; (c) ingWAT, flox; (d) epiWAT, aKO; (e) ingWAT, aKO. For inference on distributional shifts, mouse-level empirical cumulative distribution functions (ECDFs) were compared with two-sample Kolmogorov–Smirnov tests using 10,000 mouse-label permutations (unit of replication = mouse); D statistics and permutation P values (and BH-adjusted q values within the prespecified families) are reported in the Results and Supplementary Table S3.

(f) Quantification of Sirius Red–positive area (%) relative to total tissue area. Bars show mean ± SD. Group comparisons used two-tailed t-tests (variance assessed by the Fligner–Killeen test; Student’s t applied when variances were equal, otherwise Welch’s t). P < 0.05, P < 0.01, P < 0.001; n.s., not significant. Exact P values are provided in Supplementary Table S1.

Data availability

All data supporting the findings of this study are included in the article and its Supplementary Figures and tables. Scans of all immunoblots are provided in Supplementary Fig. 6. Additional raw image data, including histology whole-slide images, as well as the underlying numerical data for graphs, are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.