Wnt Activation in Mature Dermal Adipocytes Leads to Lipodystrophy and Skin Fibrosis via ATGL‐Dependent Lipolysis

ABSTRACT

Accumulation of extracellular matrix (ECM) and dermal adipocyte lipodystrophy occurs during skin fibrosis, which compromises the skin's flexibility and function. Sustained Wnt activation in dermal progenitor cells leads to fibrotic ECM thickening in the dermis and lipodystrophy of dermal white adipose tissue (DWAT). Mouse genetic models with lineage tracing demonstrate that Wnt activation in mature dermal adipocytes is sufficient to induce adipocyte lipodystrophy and fibrotic ECM remodeling in the skin. Upon withdrawal of adipocyte‐restricted Wnt activation, lipodystrophy and fibrosis are reversed. Mechanistically, Wnt activation stimulates the Adipose Triglyceride Lipase (ATGL)‐mediated lipolysis pathway, and lipolysis is an early event in the skin of Systemic Sclerosis patients. Atgl in dermal adipocytes is functionally required for Wnt‐induced lipodystrophy in the DWAT and fibrotic remodeling. Collectively, this study demonstrates that Wnt activation in dermal adipocytes promotes lipolysis, suggesting a therapeutic avenue for the treatment of lipodystrophy and skin fibrosis.

We report the first tractable and reversible Wnt activation model in dermal adipocytes in mouse skin. Our findings demonstrate that sustained Wnt activation in mature dermal adipocytes leads to lipodystrophy, which is sufficient to induce skin fibrosis in vivo. Notably, this process is reversible, as withdrawal of Wnt activation promotes recovery of lipodystrophy and fibrosis. Using a dermal adipocyte‐specific Adipose triglyceride lipase (Atgl) knockout model, we show that Wnt‐induced lipodystrophy and skin fibrosis are mediated by lipolysis. (Created with BioRender.com).

1. Introduction

In chronic tissue fibrosis, extracellular matrix proteins such as collagens accumulate, leading to an increase in tissue stiffness and a‐cellularity that can cause organ failure and morbidity [1]. Fibrosis can occur in many tissues including skin, lung, liver, heart, intestine, and kidney [1, 2]. Despite its devastating impact on nearly 1 in 4 people and an annualized incidence of major fibrosis‐related conditions in 1 in 20 people globally, no effective treatment exists to reverse fibrosis [2]. During fibrosis, several tissues such as skin, liver, and lung have a loss of lipid‐filled cells leading to lipodystrophy [3, 4, 5, 6, 7]. However, the signals and mechanisms underlying fibrotic lipodystrophy are not fully understood.

In the skin, the dermal adipocytes form a distinct dermal white adipose tissue (DWAT) layer that sits beneath the ECM‐rich dermis, making the skin an ideal system to understand the physiological mechanisms underlying lipodystrophy. Dermal fibroblasts are the key producers of ECM proteins such as collagens and proteoglycans, which provide structural integrity to the skin [8, 9]. DWAT contributes to thermoregulation and produces antimicrobial peptides to modulate immune response upon injury [10, 11]. The dynamic change in size of DWAT in wound healing and during homeostatic conditions occurs during adaptive thermogenesis and the hair follicle cycle [7, 10, 12]. Skin fibrosis is marked by the buildup of extracellular matrix (ECM) due to heightened activation and proliferation of dermal fibroblasts, along with lipodystrophy of the DWAT. These changes collectively lead to tissue stiffening and impaired function. ECM accumulation in chronic skin fibrosis occurs in diseases such as systemic sclerosis (SSc), atopic dermatitis, keloids, and psoriasis affecting 100 million people globally each year [13]. DWAT lipodystrophy is a hallmark of acute fibrosis in mouse wound healing and chronic fibrosis in human systemic scleroderma skin (SSc) and chemical and genetic models of fibrosis, but the underlying mechanisms and impact of lipodystrophy on the whole skin are unclear [14, 15, 16, 17]. To identify new strategies for preventing and reversing established fibrosis, we need to define the inducing signals that cause fibrotic lipodystrophy using the skin as a model.

Of the many profibrotic pathways, the canonical Wnt/β‐catenin signaling pathway stands out because it is a conserved profibrotic pathway across tissues and anti‐adipogenic [18, 19]. The Wnt signaling pathway is elevated in human fibrotic skin and lung tissues and elevated in chemical induced mouse models of skin and lung fibrosis [20, 21, 22, 23]. Deletion or inhibition of Wnt/β‐catenin attenuates chemical induced skin and lung fibrosis [20, 24]. Wnt/β‐catenin signaling activation can upregulate profibrotic genes such as connective tissue growth factor (CTGF), TGF‐β signaling components, smooth muscle actin (ACTA2), and matrix genes such as COL1A1 to cause ECM accumulation [21, 25, 26]. In contrast, Wnt/β‐catenin signaling activation leads to epigenetic silencing of Ppar‐γ and Cebp‐α which are the central regulators of adipocyte fate [27, 28]. Wnt signaling pathway activation in the epidermis promotes dermal adipogenesis and the Wnt pathway preserves mesenchymal progenitor multipotency in the white adipocyte depot [29, 30]. Previously, our lab has shown that Wnt/β‐catenin signaling activation in Engrailed1 ⁺ fibro‐adipocyte progenitors and their derivatives can cause ECM accumulation in the dermis which causes an increase in dermal thickness and lipid depletion of the DWAT that leads to a decrease in size of the adipocytes and DWAT thickness [31]. How does Wnt/β‐catenin signaling activation in mature adipocytes differ mechanistically from its effects on progenitors and epidermal cells? In mature adipocytes, lipid homeostasis is a balance of lipid accumulation and lipid breakdown. Lipolysis is the physiological pathway that adipocytes use to break down triacylglycerol (TAG) stored in the lipid droplet. It involves a sequence of lipases starting with Adipose Triglyceride Lipase (ATGL) which results in the release of fatty acids and glycerol [17, 32]. Emerging evidence suggests that fatty acids from adipocytes can contribute to the fibrotic transformation and tissue dysfunction [2]. Whether Wnt/β‐catenin signaling activation in mature dermal adipocytes is sufficient to cause ATGL‐dependent lipodystrophy and its impact on ECM accumulation and remodeling in the skin is unclear.

To determine if Wnt/β‐catenin signaling activation in mature dermal adipocytes is sufficient to cause lipodystrophy or skin fibrosis, we constructed an inducible and reversible genetic mouse model to activate the Wnt signaling pathway in Adiponectin + mature dermal adipocytes with indelible genetic lineage tracing. We found Wnt/β‐catenin signaling activation in mature dermal adipocytes was sufficient to cause dermal ECM remodeling, elevated proliferation of multiple cell types in the skin, and lipodystrophy of mature dermal adipocytes without perturbing cell survival. These phenotypic changes require the ATGL‐dependent lipolysis pathway and can also be rescued upon withdrawal from Wnt/β‐catenin signaling activation. Collectively, these data show that skin fibrosis‐associated lipodystrophy and ATGL‐dependent lipolysis may be a new therapeutic target in fibrosis treatment.

2. Materials and Methods

AdiponectinCreER [33] (Jax stock 024671); Rosa26mTmG [34] (Jax stock 007676, respectively); Rosa26rtTA‐EGFP [35] (Jax stock 005572); TetO‐deltaN89β‐catenin [36]; Atgl^flox/flox [37] (Jax stock 024278) mice were used for generating the dermal‐adipocyte restricted, inducible, and reversible Wnt signaling activation and conditional deletion of lipolysis mouse model. Case Western Reserve Institutional Animal Care and Use Committee approved all animal procedures in accordance with American Veterinary Medical Association guidelines protocol 2013–0156, approved 21 Dec. 2021, Animal Assurance No. A3145‐01 at Case Western. Mice are maintained and bred on a mixed genetic background in the Animal Resource Center (ARC) at Case Western Reserve University. Males and females are included in all experiments, and similar findings are reported for both sexes. Phenotypic analysis was based on at least two to five litters of mice with litter‐matched mutants and controls.

To spatially and temporally restrict activation of the Wnt signaling pathway in mature Adiponectin‐CreER dermal adipocytes in mouse dorsal skin, 100 μL of 5 mg/mL tamoxifen in 100% ethanol (ApexBio B5965) was applied topically to shaved dorsal skin of weaned mice at 21 days (P21) and 22 days old (P22) [7, 38]. The Adiponectin‐CreER line was used to recombine both Rosa26 membrane‐targeted tandem dimer Tomato/membrane‐targeted green fluorescent protein (R26mT/mG) and Rosa26 reverse tetracycline regulator transactivator (R26rtTA). For induction of β‐catenin‐myc tagged transgene expression, the key transducer of the canonical Wnt signaling pathway, in the Adiponectin‐CreER/+; R26rtTA/R26rtTA; R26mTmG/+; TetO‐deltaN89β‐catenin/+ mice (Adipo‐β‐cat^istab), 23‐day‐old (P23) mice were given 6 g/kg doxycycline chow (Envigo‐Harlan) and 2 mg/mL doxycycline (Fisher Scientific 446 061 000)water daily for 5 (P23‐28) and 10 days (P23‐33). In reversal experiments after Wnt activation regimen, mice were switched to 10 days of regular chow and water (P33‐P43). To specifically and genetically inhibit lipolysis in mature dermal adipocytes on mouse dorsal skin, the Adiponectin‐CreER line was used to delete Atgl, ^fl/fl as previously shown [7, 39, 40, 41]. Atgl ^fl/fl mice were bred into the Adipo‐β‐cat^istab background, and the Wnt activation regimen was performed with doxycycline in chow and water to Adipo‐β‐cat^istab; Atgl ^fl/fl mice for 5 or 10 days, and skin was analyzed at specific time points at P28, P33. Control groups were CreER negative or bigenic with similar treatments as the experimental group.

2.1. Human RNA Sequencing

Skin biopsies from 33 healthy controls and 48 early‐stage (mean disease duration of 1.3 years) SSc patients were analyzed. Patient selection and sample treatments were as described [42] in the PRESS cohort. Briefly, 3‐ or 4‐mm forearm skin samples for each patient were collected and immersed in RNAlater solution, then flash frozen and shipped to University of Texas Houston (UTH) study site on dry ice. miRNeasy Mini kits (Qiagen) were used for RNA extraction and Agilent 2100 Bioanalyzer (Agilent Technologies) was used for RNA integrity testing. Illumina TruSeq stranded Total RNA Library Prep Gold kit was used for cDNA library preparation according to the manufacturer's protocol. Agilent 2200 TapeStation (Agilent Technologies) was used for cDNA quality testing and KAPA Library Quantification Kit (KAPA Biosystems) was used for cDNA quantification before sequencing. 10 pM concentration of the libraries was loaded on cBot (Illumina) and used HiSeq 2500 (Illumina) for a 2 × 76 bp paired‐end sequencing. For each sample, 50 million reads were generated [42]. These data are publicly available in the NCBI GEO database under the accession numbers GSE130995.

2.2. Body Weight and Fat Mass

Control and Adipo‐β‐cat^istab mice were weighed after 10 days of doxycycline food and water treatment. Then, inguinal and gonadal fat were taken and weighed.

2.3. Histological Staining

Tissue was fixed in 10% buffered formalin (Fisher Scientific SF100‐4) for 1 h at 4°C, embedded in paraffin, and sectioned at 7 μm. Paraffin sections were stained with Masson's trichrome according to standard protocols. Paraffin sections were also stained with 1.6% w/v Picrosirius red (26357–02; Electron Microscopy Science) for 20 min, then rinsed for 5 min in acidified water (1% acetic acid in tap water). Polarized light microscopy allows for better viewing of collagen fibers due to its birefringent nature. The Olympus BX microscope and Cell Sens entry software were used to take brightfield images. For Alcian Blue staining, paraffin sections were deparaffinized by successive ethanol solutions, then washed in distilled water. For sulfated proteoglycan staining (Ph = 2.5), place the slides in 3% acetic acid for 5 min, then stain with Alcian blue pH = 2.5 (1 g alcian blue in 100 mL 3% acetic acid solution) for 30 min. For high acidic sulfated proteoglycans, stain with Alcian blue pH = 1.0 (1 g alcian blue in 100 mL 0.1 M HCl) for 30 min. Quantification of skin extracellular matrix represents the average of three different locations in 3 sections/mouse.

2.4. Immunohistochemistry and Immunofluorescence

Paraffin tissue was fixed in 10% formalin for 1 h at 4°C, embedded in paraffin, and sectioned at 7 μm. Paraffin sections were deparaffinized by successive ethanol solutions of 99%, 96%, 90%, 80%, 70%, and 50%, then washed in 1× PBS. Then slides underwent heat‐based antigen retrieval in a citrate buffer (10 mM Tri‐Sodium Citrate dihydrate, pH = 6) at 90°C–94°C for 15 min in a water bath. Slides were then cooled for 50 min to room temperature and blocked with 10% goat serum (Thermo Fisher Scientific 16 210 064), 1% BSA (BP1600100; Fisher Scientific), and 0.05% Tween 20 (BioWorld 42 030 016–1) for 85 min at room temperature in a humid chamber. Tissues were incubated with proper primary antibodies diluted to a certain ratio with blocking buffer in a humid chamber at 4°C overnight. The following primary antibodies were used for bright field immunohistochemistry and immunofluorescence as described previously [26, 31, 43]. Rabbit anti‐Myc Tag (ab9106, 1:500; Abcam), Rabbit anti‐phospho‐HSL (Ser 565) (4137, 1:1600; Cell Signaling), Rabbit anti‐alpha SMA (ab124964, 1:1000; Abcam). For Myc‐tag and p‐HSL, after antigen retrieval by citric buffer, 0.3% H₂O₂ for 10 min, 5 min TBS wash, and 85 min blocking buffer (1% goat serum+1% BSA+ 0.1% Tween 20 in TBS) incubation, Goat anti‐Rabbit Biotinylated secondary antibody (1:250; Vector Labs, San Francisco, CA) was used for 30 min at room temperature. After two 5 min washes in TBS, 4 μg/mL neutravidin‐HRP (31 030; Thermo Scientific) diluted in a blocking buffer for 30 min was applied, followed by DAB development (Myc‐tag is 6.5 min, p‐HSL is 10 min). For alpha‐SMA, after antigen retrieval by citric buffer, 5 min TBS wash, and 1 h blocking buffer (1% donkey serum+1% BSA+ 0.1% Tween 20 in TBS) incubation, Donkey anti‐rabbit Alexa 647 (A31573,1:500; Invitrogen) secondary antibody was applied at room temperature for 30 min. Nuclei were counterstained with hematoxylin before mounting in Permount Mounting Medium (SP15‐100; Fisher Chemical) or counterstained with DAPI (D9542,1:2000; Sigma‐Aldrich) before mounting in Fluoroshield (Sigma‐Aldrich). Positive and negative controls were used for specificity detection of the antibodies.

Flash‐frozen dorsal skin is embedded in OCT (Tissue‐Tek) and sectioned at 14 μm at −20°C. Cryo‐sectioned samples were fixed by 4% PFA for 15 min at room temperature and washed in PBS twice. Slides were blocked with 10% Donkey serum (50 413 115; Fisher Scientific) and 0.25% Triton X‐100 (BP151100; Fisher Scientific) for 30 min at room temperature in a humid chamber. Tissues were incubated with proper primary antibodies diluted in block buffer to a certain ratio in a humid chamber at 4°C overnight. The following primary antibodies were used for immunofluorescence as described previously [7, 43]: Chicken anti GFP (ab13970, 1:1000; Abcam), Rabbit anti Perilipin (ab3526, 1:500; Abcam), Rabbit anti Ki67 (ab15580, 1:500; Abcam). After two 5‐min PBS washes, Donkey anti Rabbit Alexa 647 (A31573,1:500; Thermo Fisher Scientific), Donkey anti Chicken Alexa 488 (111 545 003, 1:500; Jackson ImmunoResearch), and Donkey anti Rabbit Alexa 488 (A32790, 1:500; Thermo Scientific) secondary antibodies were applied at room temperature for 30 min. TUNEL staining was performed on Cryosections by TUNEL TMR red kit (11 767 291 910, 5% TUNEL enzyme; Roche Diagnostic Crop). Nuclei were counterstained with DAPI and then mounted in Fluoroshield. Positive and negative controls were used for specificity detection of the antibodies and kit.

2.5. B‐CHP Staining

Paraffin sections were deparaffinized by successive ethanol solutions, then washed in PBS. Slides were then blocked with 5% goat serum in PBS at room temperature for 20 min. Apply 3–4 drops per slide of biotin block (SP 2001; Vector Laboratories) and incubate the slides in the humid chamber for 15 min at room temperature. Stain with a biotin‐conjugated collagen hybridizing peptide (B‐CHP, BIO300, 3Helix) at a concentration of 2 μM in block buffer overnight. 4 μg/mL neutravidin‐HRP diluted in a blocking buffer was applied for 30 min, followed by 4 min of DAB (SK‐4100; Vector Laboratories) development.

2.6. Imaging

Masson's Trichrome, Myc‐Tag, p‐HSL, Alcian Blue, and Picrosirius red images were taken at room temperature by an Olympus BX60 microscope with a digital camera (DP70, Olympus). Masson's trichrome staining images were taken at 4× objective (Olympus UPIanFI 4×/0.13) by Cell Sens Entry software (Version 1.5, Olympus Corporation 2011). Myc‐Tag, p‐HSL, and Alcian Blue images were taken at 10× and 40× objectives. Picrosirius red images were taken at 40× under polarized light produced by microscope U‐ANT (U‐P115, Olympus) and U‐POT (U‐P110, Olympus). Exposure between controls and mutants was the same. Immunofluorescence images were taken by an inverted wide‐field Leica Dmi8 microscope at 20× (506 243 Germany) and 40× (506 243 Germany) magnification. Images were analyzed in Fiji/ImageJ and processed by Adobe Photoshop and Adobe InDesign.

2.7. Atomic Force Microscopy on Skin Sections

Tissue samples used for mechanical properties were 50 μm cryosections of flash frozen mouse dorsal skin in the sagittal plane. Mechanical properties of mouse skin were analyzed using a high‐performance MFP‐3D‐Bio atomic force microscope (AFM; Oxford Instruments, Santa Barbara, CA, USA). Tip‐less AFM cantilevers (ARROW‐TL1Au, Nanoworld, nominal spring constant: 0.03 N/m) were modified by gluing a 25‐μm polystyrene bead as we described earlier [44]. At least 75 force‐distance curves were collected from randomly selected locations on the lower epidermis of each tissue section. Each force‐distance curve was conducted at a setpoint of 5 nN at a constant scan rate of 0.3 Hz to determine the stiffness of control and mutant samples. The Hertz model was used to calculate Young's modulus (E_Y) from curves at an indentation depth of around 500 nm. Three non‐overlapping images were taken from 3 sections separated by 50 μm to attain spatial heterogeneity per animal.

2.8. Morphometric and Image Analysis Pipelines

2.8.1. Adipocyte Area Measurement

20× immunofluorescence images of GFP and PLN1 double staining were used for individual adipocyte area measurement. Approximately 30 adipocytes per mouse of control and GFP⁺ PLN1⁺ mutant were measured from non‐overlapping fields and analyzed using the segmentation editor tool in Fiji/ImageJ. The areas of adipocytes were binned into a histogram using GraphPad Prism (GraphPad Software, San Diego, CA).

2.8.2. Collagen Amount Measurement

10× Masson's Trichrome stained images were used for total collagen analysis of the DWAT. The blue that stains for collagen in the Masson's Trichrome staining was thresholded by ImageJ with the HSB parameters (Hsu: 135–195; Saturation: 25–255; Brightness: 20–255) [45]. The white color in the mask image after HSB threshold stands for the blue stained collagen. For each animal, 4–6 images were used, and for each image, three 40 pixels × 30 pixels ROIs were taken at the bottom of the DWAT and the middle of the DWAT. The area of the white color in the ROI was the collagen amount.

2.8.3. Cell Proliferation Analysis

For quantification of Ki67 positive cells in epidermis and dermis, 10× immunofluorescence images of Ki67 and GFP double staining were used. Measurements were made in rectangular ROIs (170 × 144 px) in the left, middle, and right of the dermis or epidermis of each image. The number of nuclei with overlapping DAPI and Ki67 nuclei was counted with Fiji's cell counter tool. The ratio of overlapping Ki67 and DAPI nuclei to all DAPI stained nuclei was taken. For each animal (controls and mutants), 3 images were used from the left, middle, and right of the tissue section, and on each image, one ROI on the left, middle, and right of the picture was used. For quantification of Ki67 positive cell number in DWAT, 40× immunofluorescence images of Ki67 and GFP double staining were used. Fiji cell counter was used to count GFP⁺ adipocytes and Ki67⁺ nuclei within GFP⁺ adipocytes. Ki67 nuclei to total GFP positive adipocytes was taken as the ratio of Ki67⁺ in the DWAT. 3 images were used from the left, middle, and right of the tissue section for each mouse (controls and mutants). The results were graphed as a scatter plot using GraphPAD Prism.

2.8.4. Biotin Conjugate Collagen Hybridizing Peptide (B‐CHP) Analysis

40× magnification images were used for quantification. Images were all white‐balanced in Adobe Photoshop. One 350 × 200 px ROI was taken per 40× image. Images were deconvoluted using Fiji. All images were thresholded to 1150. Percent area in brown in the ROI was taken and averaged per animal. Each animal is represented by 9 ROIs in total, 3 ROIs per each location: upper dermis, lower dermis, and DWAT.

2.8.5. Myc‐Tag Expressing Adipocytes

40× brightfield images were analyzed for Wnt activation efficiency in mature dermal adipocytes. One image from the left, middle, and right of the dorsal skin was analyzed per mouse. The cell counter tool of Fiji was used to count the total number of adipocytes in the DWAT area and the number of Myc‐Tag positive adipocyte nuclei (brown) inside the fixed field. The number of Myc‐Tag positive adipocytes was divided by the total number of adipocytes in the DWAT area to get the Wnt activation efficiency in the mature dermal adipocytes and graphed by GraphPAD Prism.

2.8.6. P‐HSL Positive Adipocyte Counting

40× brightfield images were analyzed for lipolysis pathway activation. One image from the left, middle, and right of the dorsal skin was analyzed per mouse. The cell counter tool of Fiji was used to count the total number of adipocytes in the DWAT area and the number of the p‐HSL positive adipocyte cytoplasm (brown) inside the fixed field. The number of p‐HSL positive adipocytes was divided by the total number of adipocytes in the DWAT area to get the p‐HSL positive ratio in the mature dermal adipocytes and graphed by Graph PAD Prism.

2.8.7. Tamoxifen Recombination Specificity

40× images of immunofluorescence taken by Leica Dmi8 microscope were analyzed for tamoxifen specificity. Images of sections in the DWAT, subcutaneous, gonadal, and inguinal fat were taken after topical tamoxifen application on dorsal skin and doxycycline treatment. Images were taken in the left, middle, and right side across three sections on a slide and analyzed. The number of GFP positive adipocytes and the total number of adipocytes in the whole image were counted by Fiji (dimension 1360 × 1024). The Tamoxifen recombination efficiency is the ratio of GFP positive adipocytes to the total number of adipocytes and was graphed using Graph PAD Prism.

2.8.8. Alpha‐SMA Analysis

20× images of immunofluorescence taken by Leica Dmi8 microscope were analyzed for alpha‐SMA expression. The image was divided into three 1424 pixels × 438 pixels ROIs, with locations standing for upper dermis, lower dermis, and DWAT. The number of nuclei surrounding the alpha‐SMA signal is counted. For each animal, 3 images were taken and 3 ROIs for each image.

2.8.9. TWOMBLI (The Work Flow of Matrix BioLogy Informatics) Analysis of Matrix

PSR‐stained mouse dorsal skin sections were imaged using polarized light microscopy and used for TWOMBLI analysis. For each animal, 3 non‐overlapping 40× images from the left, middle, and right of the dorsal skin were taken at the dermis immediately above the DWAT in controls (P33 or P43) and from the remodeled dermis (very close to the remaining DWAT) above the panniculus carnosus muscle layer in 2 days Tamoxifen painting (P21‐P22) 10 days doxycycline chow (P23‐p33) and 2 days Tamoxifen painting (P21‐P22) 10 days doxycycline chow (P23‐p33) 10 days reversal (P33‐P43). Two fixed sizes of ROIs (376 × 195 pix) were taken from the 40 images. The ROIs from these pictures were run through TWOMBLI in Fiji/ImageJ [46]. The parameters were set as the following list: Contrast Saturation: 0.35, Minimum Line Width: 5, Maximum Line Width: 10, Minimum Curvature Window: 20, Maximum Curvature Window: 70, Minimum Branch Length: 10, Maximum Display HDM: 220, Minimum Gap Diameter: 9. Line masks of the matrix network were generated by the Fiji Ridge Detection tool. According to the line mask, the ECM metrics including curvature, fractal dissension, number of endpoints, lacunarity, and alignment were calculated on each ROI and the average of the total of 6 ROIs per animal.

2.8.10. AFT (Alignment by Fourier Transform) Analysis of Collagen Alignment

The same 40× images of PSR stained mouse dorsal skin by polarized light microscopy were used to take new ROIs for AFT. These new ROIs were then converted to 16‐bit grayscale. 40× non‐overlapping images were used from left, middle, and right from the sections. Two fixed sizes of ROIs (376 × 195 pix) were taken from the dermis immediately above the DWAT in controls (P33 or P43) and from the remodeled dermis (very close to the remaining DWAT) above the panniculus carnosus muscle layer in 2 days Tamoxifen painting (P21‐P22) 10 days doxycycline chow (P23‐p33) and 2 days Tamoxifen painting (P21‐P22) 10 days doxycycline chow (P23‐p33) 10 days reversal (P33‐P43). The ROIs were converted into 16‐bit grayscale by Fiji and input into Alignment by Fourier Transform (AFT) [47] by MATLAB (Mathworks, v2023b). Median order parameter and heatmap of AFT vector orientation were output using the following parameters: Window Size: 30 pixels, Window Overlap: 50%, Neighborhood Radius: 2× vectors. Local masking and filtering were not applied. The Median Order Parameter has a range of 0 to 1 where 0 is indicative of unorganized alignment (isotropy) and 1 is complete alignment (anisotropy). The median order parameters were averaged across 6 ROIs per animal and graphed by Graph PAD Prism.

2.8.11. Proteoglycan Area Measurement

Alcian Blue pH 2.5 stained images were used for proteoglycan area analysis. 40× images from the upper dermis, lower dermis, and DWAT were used, and two fixed‐sized ROIs were taken per 40× image. The ROIs were then run through FIJI/ImageJ with the following color threshold parameters: Hue: 0–255, Saturation: 75–255, Brightness: 0–200. The remaining thresholded area was then measured. The results were averaged across 6 ROIs per animal and graphed in Graph PAD Prism. Alcian Blue pH 1.0 images were analyzed with full 40× images instead of using ROIs in order to decrease bias and variability. Color threshold parameters: Hue: 105–175, Saturation: 70–255, Brightness: 0–255 were used. The subsequent steps remained the same between Alcian pH 2.5 and 1.0.

2.8.12. PCA (Principal Component Analysis)

PCA was comprised of TWOMBLI output and the covariance between variables was checked in RStudio. Each data point is one ROI with outliers removed before analysis calculated by Microsoft Excel. The curvature metrics output by TWOMBLI were not included in the PCA analysis as the result of low covariance values. PCA biplot was generated by principal components 1 and principal component 2.

2.9. RNA Extraction and qPCR Analysis

For mature dermal adipocyte qPCR, DWAT tissue was manually cut from 5 mm biopsy punches (9 033 515, Premier Uni‐Punch) from the mouse dorsal skin after 2 days Tamoxifen painting (P21‐P22) and 5 days (P23‐P28) or 10 days doxycycline chow (P23‐P33). RNA was extracted from flash frozen DWAT tissue by Trizol reagent (15 596 026; Thermo Fisher Scientific). 4 ng cDNA was used for qPCR after RT reaction as previously described [26, 48] Adiponectin and Axin2 mRNA were quantified relative to Hprt using Taqman master mix (4 304 437, Thermo Fisher Scientific) by the probes (Mm00456425_m1, Mm00443610_m1, Mm03024075_m1; Thermo Fisher Scientific).

For the whole skin qPCR, flash frozen mouse whole dorsal skins were ground down to a fine powder in the liquid nitrogen. 350 μL of Trizol (15 596 026, Invitrogen) was added for each 0.5 cm × 0.5 cm skin sample, then used the RNeasy Mini Kit (74 104, Qiagen) to extract RNA. 40 ng cDNA was used for qPCR after RT reaction. Adiponectin and Perilipin mRNA were quantified relative to ACTB using TaqMan master mix (4 304 437, Thermo Fisher Scientific) by the probes (Mm00456425_m1, Mm00558672_m1, Hs01060665_g1, Thermo Fisher Scientific).

The qPCR was performed on an Applied Quantstudios Biosystem 3 PCR System. A relative quantity of fold change of Ct values controls was expressed by the $2^{-\mathrm{\Delta \Delta }{C}_{\mathrm{t}}}$ formula. qPCR data was presented in univariate scatter plots as previously described [49] in Graph PAD Prism.

2.10. Cell Culture

3T3‐L1 cell lines and primary intradermal adipocyte progenitors were differentiated, stained with Oil Red O, and a free glycerol assay was performed as previously described [31]. 3T3‐L1 adipocytes were treated with differentiation media or treated with differentiation media followed by 7 μM Wnt agonist, LiCl₂, or treated with differentiation media, LiCl₂, and 40 μM ATGL inhibitor, atglistatin (Sigma Aldrich, SML1075) for 6 days and stained with Oil Red O as previously described [31]. Following differentiation, primary intradermal adipocyte progenitor cells were treated with differentiation media followed by 7 μM Wnt agonist, CHIR99021, or treated with differentiation media, CHIR, and 40 μM ATGL inhibitor, atglistatin, or treated with differentiation media, followed by 40 μM atglistatin. Cultures were stained with Oil Red O and the average Oil Red O area (2 fields per well, 4 biological replicates, and 2–4 technical replicates per sample per treatment) was obtained using Image J/Fiji. Free glycerol release in conditioned media was quantified using manufacturer instructions (Sigma Aldrich F6428) (2 technical replicates and 4 biological replicates/sample).

2.11. Statistical Analysis

Power analysis was performed to determine the sample size with power = 90. Each individual point in the figures was the average value of 3–15 technical replicates per animal according to different experiments (indicated within the methods of each assay).

3. Results

3.1. Wnt Signaling Activation in Mature Dermal Adipocyte Is Sufficient to Induce Fibrotic Remodeling in the Whole Skin

To determine if Wnt activation in the mature dermal adipocytes in the mouse dorsal skin is sufficient to induce skin fibrosis, we generated a new lineage‐traceable mouse model combined with an inducible and reversible expression of stabilized β‐catenin‐Myc tag (Adipo‐β‐cat^istab) (Figure 1A,B) [7, 33, 35, 36]. Lineage tracing of mature dermal adipocytes by topical tamoxifen inducible Adiponectin‐CreER; R26mT/mG reporter revealed that 70% of adipocytes in the DWAT layer were GFP positive in the dorsal skin and 20%–30% of adipocytes were GFP positive in other fat depots, which is consistent with previous reports (Figure S1A–C) [7]. After 5 days of Wnt activation in the Adipo‐β‐cat^istab, we found nuclear expression of β‐catenin‐Myc‐tag in at least 45%–55% of adipocytes in a fixed field in the DWAT layer (Figure S2A–C). qPCR analysis of DWAT tissue showed a three‐fold elevated expression of Axin2 mRNA, a transcriptional target of the canonical Wnt signaling pathway (Figure S2D). Body weight and fat mass of gonadal and inguinal pads after 10 days Doxycycline chow are comparable between P33 control and Adipo‐β‐cat^istab mice after 10 days of Wnt activation (Figure S2E–G). Together these data indicate that Adiponectin‐CreER /+ can recombine R26mTmG and activate Wnt signaling in a subset of DWAT of tamoxifen‐treated Adipo‐β‐cat^istab mice. By 10 days of Wnt activation in mature dermal adipocytes with dietary doxycycline, we observed the earliest significant decrease of DWAT thickness, as well as an increase in collagen amount in DWAT, accompanied by dermal ECM expansion (Figure 1C–F). Although Wnt activation occurred in a subset of mature dermal adipocytes of Adipo‐β‐cat^istab mice, we still found a significant tissue level decrease in DWAT thickness and an increase in dermal thickness. At the cellular level, we measured the size of individual Adiponectin‐ CreER; R26mT / mG lineage traced GFP ⁺/PLIN1 ⁺ adipocytes after 10 days Wnt activation (Figure 1G,H). Compared to the size of PLN1 ⁺ adipocytes in control mice, the size of the lineage‐marked GFP ⁺ dermal adipocytes was consistently decreased in Adipo‐β‐cat^istab mice and the adipocytes between hair follicles were flat in morphology and located adjacent to the hair shaft (Figure 1I, Figure S1D). Furthermore, as an additional internal control, the size of GFP⁺ PLN1⁺ + adipocytes in the 10 days Adipo‐β‐cat^istab mice was also significantly decreased compared to the GFP ^‐ PLN1 ⁺ adipocytes (Figure 1 I). Next, we investigated if reversal of Adipo‐β‐cat^istab tissue level phenotypes occurs upon subsequent withdrawal from Wnt activation (Figure 2A). After 10 days of withdrawal from Wnt activation, we observed a rescue of ECM expansion in the dermis and lipodystrophy as well as the collagen amount abundance in the DWAT in Adipo‐β‐cat^istab mice (Figure 2B–E). Subsequent withdrawal from Wnt activation in Adipo‐β‐cat^istab mice also led to recovery in size of lineage marked lipid depleted mGFP ⁺ adipocytes (Figure 2F–H). Thus, our genetic inducible and reversible model demonstrates that sustained activation of the Wnt signaling pathway in the mature dermal adipocytes of the dorsal skin contributes to the two main characteristics of skin fibrosis: ECM expansion of dermis and lipodystrophy of DWAT.

FIGURE 1.

Adipo‐β‐cat^istab leads to dermal thickness increase, DWAT thickness decrease and mature dermal adipocyte size decrease. (A) Experimental design: 21‐d‐old (P21) control and Adipo‐β‐cat^istab mice were treated with topical tamoxifen on dorsal skin and fed doxycycline containing chow and water for inducible Wnt activation in mature DWAT cells. Skin was harvested after 10 days (P33). (B) Gene map of Adipo‐β‐cat^istab mice. (C) Masson's trichrome staining of mouse dorsal skin cut in sagittal plane from (P33) control, 10 days Adipo‐β‐cat^istab. (n = 6–8). (D, E) Quantification of mouse dorsal dermal thickness and DWAT thickness in (C) (n = 5–8). (F) Quantification of area percentage of trichrome in Masson's Trichrome staining of DWAT for (P33) control and 10 days Adipo‐β‐cat^istab. (n = 5–6). Scale Bar= 200μm(G) Schematic for tamoxifen induced R26mTmG recombined mature adipocyte lineage labeling during Wnt activation. (H, I) GFP and Perilipin 1 immunofluorescence staining and quantification of PLN1⁺ vesicles in control (P33) and GFP⁻ and GFP+ vesicles in 10 days Adipo‐β‐cat^istab with low magnification on the left (n = 5–6). Scale Bar = 100 μm. Black and blue bars are the dermal (n = 5–8) and DWAT thickness. p values were calculated with unpaired, two‐tailed t‐test with Welch's correction and a p value of < 0.05 is considered significant. Data are mean ± S.D.

FIGURE 2.

Fibrotic changes in dermis, DWAT and adipocyte size caused by Adipo‐β‐cat^istab are reversible by removing Wnt signaling pathway. (A) Experimental design: 21‐d‐old (P21) control and Adipo‐β‐cat^istab mice were treated with topical tamoxifen on dorsal skin and fed doxycycline containing chow and water for inducible‐reversible Wnt activation in mature DWAT cells. Skin was harvested after 10 days doxycycline treatment and subsequent to 10 days of withdrawal from Wnt activation (P43). (B) Masson's trichrome staining of mouse dorsal skin for (P43) control and 10 days Adipo‐β‐cat^istab; 10 days reversal. (C, D) Quantification of mouse dorsal dermal thickness (n = 6–7) and DWAT thickness (n = 5–6) in (B). (E) Quantification of area percentage of stained by Masson's Trichrome staining of DWAT for (P43) (n = 5–6). Scale Bar= 200μm (F) Schematic for tamoxifen induced R26mTmG recombined mature adipocyte lineage labeling during Wnt activation and after subsequent withdrawal from Wnt activation. (G, H) GFP and Perilipin 1 immunofluorescence and quantification of area of PLN1⁺ adipocytes in control (P43) and GFP⁺ vesicles in 10 days Adipo‐β‐cat^istab; 10 days reversal (n = 7–9). Scale Bar = 100 μm. Black and blue bars are the dermal (n = 5–8) and DWAT thickness. p values were calculated with unpaired, two‐tailed t‐test with Welch's correction and a p value of < 0.05 is considered significant. Data are mean ± S.D.

3.2. Wnt Signaling Activation in the Mature Dermal Adipocytes Leads to Increase in Cell Proliferation Throughout the Skin

Next, we tested the impact of Wnt activation on cell survival of mature dermal adipocytes in dorsal skin by Terminal deoxynucleotidyl transferase dUTP nick‐end labeling (TUNEL) assay. Cell death marker by TUNEL assay in the DWAT region was comparable in controls and Adipo‐β‐cat^istab skin (Figure S3A–D). In homeostatic conditions, the proliferation index of cells in the dermis of adult skin was very low and mature dermal adipocytes are terminally differentiated post‐mitotic cells [50]. We took advantage of lineage tracing to determine if mature dermal adipocytes dedifferentiated after Wnt activation induced lipodystrophy and re‐entered the mitotic cycle (Figure 3). Compared to the morphology of mature lipid‐filled dermal adipocytes in the control skin, the DWAT layer in Adipo‐β‐cat^istab skin had mGFP⁺ dermal adipocyte cells that had shrunken and clustered around hair follicles with some being Ki67⁺(Figure 3A, Figure S3E). Next, we measured the proliferation index of cells in different parts of the skin by calculating the percent of Ki67⁺ cells in a fixed area. As expected, the proliferation index was low in control skin dermis and DWAT regions and visible in distal hair matrix cells in the anagen hair follicles [51] (Figure 3B). In contrast, Ki67⁺ nuclei were visible inside mGFP⁺ lineage‐marked cells in Adipo‐β‐cat^istab skin and quantification showed a significant increase of Ki67⁺ mGFP⁺ adipocyte number in the DWAT (Figure 3A,B). We also found elevated expression of Ki67⁺ cells in the dermal layer and the epidermis (Figure 3C–E). These data suggest the difference in the number and pattern of Ki67⁺ cells in the hair follicles and different compartments of the 10d Adipo‐β‐cat^istab skin may likely be in response to the changes in the extracellular environment [52]. Finally, we demonstrated a significant increase of α‐SMA expression in the DWAT layer and a significant decrease in the mRNA expression of mature adipocyte markers, Adiponectin and Perilipin1 in 10d Adipo‐β‐cat^istab mouse dorsal skin compared to control (Figure S4). Together, these data suggest that Wnt activation in mature dermal adipocytes leads to tissue‐wide changes in the skin.

FIGURE 3.

Adipo‐β‐cat^istab leads to whole skin cell proliferation response. (A, B) Ki67 (pseudo‐color Red) and GFP immunostaining on P33 control and 10 days Adipo‐β‐cat^istab in the DWAT and quantification Ki67⁺ in GFP⁺ adipocytes per fixed field (n = 5–8). (C–E) Images with insets of Ki67 and GFP immunostaining in the skin and quantification of percent of Ki67⁺ cells in a fixed field within the epidermal and dermal layers (n = 6–8). DAPI = blue, GFP = green, Ki67 = red, PLN1 = purple. Scale bar = 100 μm. Yellow arrow stands for Ki67⁺ cell in epidermis and white arrow stands for Ki67⁺ cell in dermis. p values were calculated with unpaired, two‐tailed t‐test with Welch's correction and a p value of < 0.05 is considered significant. Data are mean ± S.D.

3.3. Wnt Activation in Mature Dermal Adipocytes Results in Stiff Skin, Remodeling, and Topographical Changes in Dermal Collagen and Elevated Proteoglycans in the Dermis

Since we observed a significant increase in the area occupied by collagen in DWAT and an increase in dermal thickness in Adipo‐β‐cat^istab mice (Figure 1D,F), we measured the mechanical stiffness of the skin as a functional indicator of skin fibrosis. Using atomic force microscopy on skin sections, we found that the Young's modulus of the lower dermis was significantly higher in 10d Adipo‐β‐cat^istab (Figure 4A,B) mice which indicates an increase in stiffness. Since the amount of collagen can contribute to the stiffness of the skin, we analyzed collagen remodeling by visualizing the location of unfolded collagen chains with Biotin Conjugated Collagen Hybridizing Peptide (B‐CHP) [53]. Compared to control skin, we found that B‐CHP significantly increased in the DWAT and dermis in the 10 d Adipo‐β‐cat^istab mice (Figure 4C–F). Topography, the surface characteristics of collagen fibers, is a good indicator of fibrotic changes in the skin [54]. Based on our recent studies, we focused on the lower dermis above the DWAT and it is also a comparable location across genotypes [54]. Principal Component Analysis (PCA) was performed on the results of The Workflow Of Matrix BioLogy Informatics (TWOMBLI) algorithm on polarized light images of Picrosirius Red stained collagen fibers. PCA analysis showed that fiber characteristics of 10‐day Adipo‐β‐cat^istab mice were very different from the control mice at the tissue level and these characteristics were reversed after 10 days of reversal (Figure S5). In particular, the high‐density matrix (HDM) was significantly higher, and alignment (coherency) was significantly lower in 10d Adipo‐β‐cat^istab mice (Figure 4G–I). To further verify the alignment of the fibrotic collagen in 10d Adipo‐β‐cat^istab mice, the Alignment by Fourier Transform (AFT) algorithm was applied (Figure S6A). The heat map generated from the vector map showed that the orientation of the collagen was more disorganized, further supported by the median order parameter being significantly decreased in the 10d Adipo‐β‐cat^istab mice (Figure S6B,C). We also investigated the expression of sulfated proteoglycans, which are dysregulated in skin fibrosis [9, 26, 31]. In comparison to control, sulfated proteoglycans and highly acidic sulfated proteoglycans were increased in the dermal and DWAT layers in 10 days Adipo‐β‐cat^istab skin (Figure 4J–M, Figure S7).

FIGURE 4.

Adipo‐β‐cat^istab leads to topographical changes, collagen remodeling and proteoglycan expression in dermis and DWAT. (A) Experimental design: 21‐d‐old (P21) control and Adipo‐β‐cat^istab mice were treated with topical tamoxifen on dorsal skin and fed doxycycline containing chow and water for inducible Wnt activation in mature DWAT cells. Skin was harvested after 10 days (P33). (B) Quantification of skin stiffness on (P33) control, 10 days Adipo‐β‐cat^istab. Dotted line demarcates the DWAT from the dermis. Each blue cross represents the location of force curve that was performed on the tissue (n = 4). (C) Collagen remodeling was visualized with collagen hybridizing peptide conjugated to biotin (B‐CHP, brown) on (P33) control, 10 days Adipo‐β‐cat^istab. (D–F) Representative views with insets of upper‐dermis, lower‐dermis and DWAT quantification of percent area B‐CHP = brown in a fixed field in upper‐dermis, lower dermis, and DWAT in (C), (n = 5–6). (G) Collagen was stained with picrosirius red and imaged with polarized light microscopy (PSR), High Density Matrix (HDM) and line masks generated by TWOMBLI algorithm plugin in ImageJ/FIJI to quantify a broad range of collagen fiber characteristics. (H, I) Quantification of collagen fiber alignment and HDM (n = 6–7) based on TWOMBLI. (J) Proteoglycans were stained with Alcian blue pH = 2.5 and imaged under bright field. Representative images and location of insets (black box) of genotypes in (C). (K–M) Quantification of proteoglycan area per ROI in upper‐dermis, lower dermis, and DWAT in (J), (n = 7). Scale bar =100 μm. p values were calculated with unpaired, two‐tailed t‐test with Welch's correction and a p value of < 0.05 is considered significant. Data are mean ± S.D.

Next, we investigated if collagen remodeling, topographical changes, and proteoglycan increasing were reversible upon subsequent withdrawal from Wnt activation in Adipo‐β‐cat^istab mice (Figure 5A). We first measured the stiffness of the skin to gauge the functional recovery at the tissue level and found the Young's modulus was comparable between the controls and reversal skin, indicating recovery from the fibrotic state (Figure 5B). Also, after 10 days of withdrawal from Wnt activation in mature dermal adipocytes, the B‐CHP signal was rescued and comparable to the control DWAT and dermis (Figure 5C–F). Collagen remodeling and topography were reversible after 10 days of withdrawal from Wnt activation in the mature dermal adipocytes (Figure 5G–I; Figure S6D,E). The sulfated proteoglycans and highly acidic sulfated proteoglycans were restored to control levels after 10 days of withdrawal from Wnt activation (Figure 5J–M, Figure S7). These results indicate that the fibrotic remodeling of the ECM, topographical, and proteoglycan expression patterns in the whole skin are dependent on sustained Wnt activation in the mature dermal adipocytes. Altogether, Wnt activation in the mature dermal adipocytes results in elevated collagen anisotropy, remodeling, proteoglycan expression, and profibrotic collagen structure in the dermal compartment, which are also reversible.

FIGURE 5.

Fibrotic changes in topography, collagen remodeling and proteoglycan expression in dermis and DWAT caused by Adipo‐β‐cat^istab are reversible by removing Wnt signaling pathway. (A) Experimental design: 21‐d‐old (P21) control and Adipo‐β‐cat^istab mice were treated with topical tamoxifen on dorsal skin and fed doxycycline containing chow and water for inducible‐reversible Wnt activation in mature DWAT cells. Skin was harvested after 10 days doxycycline treatment and subsequent to 10 days of withdrawal from Wnt activation (P43). (B) Quantification of skin stiffness on P43 control and 10 days Adipo‐β‐cat^istab;10 days reversal. (n = 4). (C) Collagen remodeling was visualized with collagen hybridizing peptide conjugated to biotin (B‐CHP, brown) on P43 control and 10 days Adipo‐β‐cat^istab;10 days reversal. (D–F) Representative views with insets of upper‐dermis, lower‐dermis and DWAT quantification of percent area B‐CHP = brown in a fixed field in upper‐dermis, lower dermis, and DWAT in (C), (n = 4–5). (G) Collagen was stained with picrosirius red and imaged with polarized light microscopy (PSR), High Density Matrix (HDM) and line masks generated by TWOMBLI algorithm plugin in ImageJ/FIJI to quantify a broad range of collagen fiber characteristics. (H, I) Quantification of collagen fiber alignment and HDM (n = 5–7) based on TWOMBLI. (J) Proteoglycans were stained with Alcian blue pH = 2.5 and imaged under bright field. Representative images and location of insets (black box) of genotypes in (C). (K–M) Quantification of proteoglycan area per ROI in upper‐dermis, lower dermis, and DWAT in (J), (n = 6). Scale bar =100 μm. p values were calculated with unpaired, two‐tailed t‐test with Welch's correction and a p value of < 0.05 is considered significant. Data are mean ± S.D.

3.4. ATGL‐Dependent Lipolysis Pathway Is Activated in Early Adipo‐β‐cat^istab Mice and Elevated Expression in Human SSc

The lipid content in adipocytes is a balance between lipogenesis and lipolysis. The lipolysis pathway, which starts with hydrolysis of triglyceride (TAG) catalyzed by ATGL to release glycerol and fatty acids, is a homeostatic pathway in adipocytes for lipid catabolism and energy generation. First, we investigated if the lipolysis pathway is required for Wnt signaling induced lipid depletion in a cell‐autonomous manner. We found Wnt activation with LiCl₂ in 3T3L1 cell differentiated adipocytes and with CHIR99209 in mouse primary intradermal adipocyte cultures led to lipid depletion and an increase in free glycerol (Figure S8). This lipid depletion effect was attenuated by the ATGL inhibitor, Atglistatin, suggesting that the lipid depletion occurs through an ATGL‐dependent pathway (Figure S8). Next, we queried the activation of the lipolysis pathway by visualizing the expression of phosphorylated‐Hormone Sensitive Lipase (p‐HSL) in Adipo‐β‐cat^istab mice (Figure 6A,B). In the control skin, we found 11%–13% adipocytes have p‐HSL expression per fixed field (Figure 6C,D). After 5 days of Wnt activation in the DWAT layer of Adipo‐β‐cat^istab, we found 65%–68% of adipocytes express p‐HSL after 5 days of Wnt activation in the DWAT layer (Figure 6C,D). To understand the translational potential, we investigated if ATGL‐pathway lipolysis activation is an early event in human SSc. We queried mRNA expression of the core neutral lipases and associated genes in the lipolysis pathway in skin biopsies from a cohort of patients with early‐stage (mean disease duration of 1.3 years) SSC (n = 48) and healthy controls (n = 33) in the Prospective Registry for Early Systemic Sclerosis (PRESS) Cohort [42]. We found that Pnpla2 (ATGL), Lipe (HSL), Mgll (MGL), Abhd5 (CGI‐58), and Perilipin1 (PLN1) were all significantly increased in the skin within the first 2 years of the SSc disease compared to the healthy skin (Figure 6E–I) (GSE130955). The lipolysis pathway genes are comparable to control skin in the third year of the SSc disease (Figure S9). This result shows that lipolysis is an early event in human SSc and modulating adipocyte lipid handling homeostasis could be a potential target for SSc therapy.

FIGURE 6.

ATGL‐dependent lipolysis pathway is activated in early Adipo‐β‐cat^istab mice and also increase expression in human SSc. (A) Experimental design: 21‐d‐old (P21) control and Adipo‐β‐cat^istab mice were treated with topical tamoxifen on dorsal skin and fed doxycycline containing chow and water for inducible Wnt activation in mature DWAT cells. Skin was harvested after 5 days (P28). (B) Schematic of lipolysis pathway. (C) Representative images with insets of immunohistochemistry staining of phosphorylated‐HSL (brown) in control (P28) and 5 days Adipo‐β‐cat^istab. Scale bar = 100 μm. (D) Quantification of %p‐HSL+ adipocyte in the fixed field on skin sections in (C) (n = 3). (E–I) Scatter plots of selected genes (Pnpla2, Lipe, Mgll, Abhd5, Plin1) involved in lipolysis pathway from human normal or SSc (GEO: GSE130995) (Normal = 33; SSc = 58). p values were calculated with unpaired, two‐tailed t‐test with Welch's correction and a p value of < 0.05 is considered significant. Data are mean ± S.D.

3.5. Wnt Signaling Activation in Mature Dermal Adipocyte Causes Lipodystrophy and Fibrotic Changes of the Skin via ATGL‐Dependent Lipolysis

To determine if the lipolysis pathway is functionally required for DWAT lipodystrophy in Adipo‐β‐cat^istab mice, we conditionally deleted Atgl with the tamoxifen inducible Adiponectin‐CreER line in mature dorsal dermal adipocytes as previously described [7, 40, 41, 55] (Figure 7A,B). We found nearly 45%–54% of Myc‐tag‐β‐catenin⁺ adipocytes were in Adipo‐β‐cat^istab; Atgl ^fl/fl (Figure S10). After 10 days of Wnt activation in the mature dermal adipocyte, the DWAT and dermal thickness and total amount of collagen in DWAT were comparable to controls, showing that Atgl‐dependent lipolysis is required for Wnt activated lipodystrophy of the DWAT (Figure 7C–F). The size of the individual adipocytes in Adipo‐β‐cat^istab; Atgl ^fl/fl mice is comparable to the controls, showing the adipocytes are protected from lipodystrophy (Figure 7G,H). Finally, we investigated the impact of Wnt‐activation‐caused‐DWAT lipolysis on the ECM remodeling and collagen topography. The B‐CHP expression showed comparable levels of ECM remodeling in the control and 10 days of Wnt activation in Adipo‐β‐cat^istab; Atgl ^fl/fl mice (Figure 7I–L). Meanwhile, TWOMBLI and AFT analysis of the collagen topography showed that the HDM, alignment, and median order parameters in the 10 days Adipo‐β‐cat^istab; Atgl ^fl/fl mice were all comparable to the control (Figure 7M–O, Figure S11). The expression of sulfated proteoglycan and high acidic sulfated proteoglycans were also comparable to the control in the 10 days Adipo‐β‐cat^istab; Atgl ^fl/fl mice (Figure 7P–S, Figure S12).

FIGURE 7.

Adipo‐β‐cat^istab leads to lipodystrophy and fibrotic skin via ATGL‐dependent lipolysis pathway. (A) Experimental design: 21‐d‐old (P21) control CreER negative; Atgl ^fl/fl and Adipo‐β‐cat^istab; Atgl ^fl/fl mice were treated with topical tamoxifen on dorsal skin and fed doxycycline containing chow and water for inducible Wnt activation in mature DWAT cells. Skin was harvested after 10 days (P33). (B) Gene map of Adipo‐β‐cat^istab; Atgl ^fl/fl mice. (C) Masson's trichrome staining of mouse dorsal skin for P33 controls CreER negative; 10 days Adipo‐β‐cat^istab, Atgl ^+l+ ; 10 days Adipo‐β‐cat^istab, Atgl ^fl/fl . Scale bar = 200 μm, black and blue bars are the dermal and DWAT thickness. (D, E) Quantification of dorsal dermal thickness and DWAT thickness in (C) (n = 6–7). (F) Quantification of area percentage of trichrome in Masson's Trichrome staining of DWAT for P33 controls CreER negative, Atgl ^fl/fl ;10 days Adipo‐β‐cat^istab, Atgl ^fl/fl . (n = 8) (G, H) GFP and Perilipin 1 immunofluorescence staining and quantification of adipocyte area of GFP⁺ lipid droplets in skin sections of genotypes in (G) (n = 4). DAPI = blue, GFP = green, PLN1 = purple, scale bar = 100 μm. (I–L) B‐CHP staining with representative views with regions of interest (as shown in Figure 4B) and quantitation of collagen remodeling staining with B‐CHP (brown) in a fixed area on skin sections of genotypes in (I). Scale bar = 100 μm, (n = 5). (M) Polarized light images of dorsal dermal collagen stained with picrosirius red (PSR). High Density Matrix (HDM) and line masks images generated by FIJI on skin sections of genotypes in (I). (N, O) Quantification of collagen fiber alignment and HDM based on TWOMBLI algorithm on skin sections of genotypes in (M), (n = 5–6). (P) Proteoglycans were stained with Alcian blue pH = 2.5 and imaged under bright field. Representative images and location of insets (black box) of genotypes in (I). (Q–S) Quantification of proteoglycan area per ROI in (P), (n = 6). p values were calculated with unpaired, two‐tailed t‐test with Welch's correction and a p value of < 0.05 is considered significant. Data are mean ± S.D.

Altogether, the activation of Wnt signaling in the mature dermal adipocytes results in lipodystrophy, ECM remodeling, and the formation of profibrotic collagen, a process facilitated by the ATGL‐dependent lipolysis pathway (Figure 8).

FIGURE 8.

Summary model showing that lipodystrophy and ECM remodeling in fibrosis are dependent on sustained Wnt activation in dermal adipocytes via the ATGL‐mediated lipolysis pathway.

4. Discussion

Wnt signaling pathway is anti‐adipogenic, and novel gain‐of‐function mutations in human CTTNB1 (β‐catenin) are associated with body fat distribution [56]. Wnt activation in the other skin compartments, such as epidermis, promotes adipogenesis and maintains the multipotency of mesenchymal progenitors in adipocyte depots [29, 30]. Here, we tested the functional role of canonical Wnt signaling pathway activation in mature dermal adipocytes, especially in the context of the onset and recovery of fibrosis and lipodystrophy. By using an inducible and reversible model of Wnt activation combined with lineage tracing in mouse mature dermal adipocytes, we demonstrated that lipodystrophy and associated ECM fibrotic remodeling are dependent on sustained Wnt activation in dermal adipocytes. Our results also show that the lipolysis pathway is activated in the first 2 years of SSc patients, which makes the investigation of this mechanism more relevant for clinical treatment. Recently, Wnt signaling pathway activation led to an increase in ATGL expression during mouse skin wound healing, but the function of ATGL‐dependent lipolysis was not tested [57]. Using genetic tools, we determined that ATGL‐dependent lipolysis in the mature dermal adipocytes in vivo is required for both Wnt‐induced lipodystrophy and the associated ECM remodeling in the dermis. Altogether, our study is the first to restrict Wnt activation and deletion of Atgl in mature dermal adipocytes in vivo to demonstrate the impact on the DWAT and the adjacent tissues, such as the ECM‐rich dermis.

By combining lineage tracing with the inducible and reversible model of Wnt activation on hair follicle stage matched skin, we were able to determine which aspects of Wnt‐induced fibrosis and lipodystrophy are reversible in mouse skin. First, we found that Wnt induced lipodystrophy in mature dermal adipocytes is a reversible state. We found lineage marked GFP⁺ mature dermal adipocytes can recover in size within 10 days after withdrawal from Wnt activation. Recovery of lineage labeled dermal adipocyte cell size and DWAT area at the tissue level suggests refilling of the lipid‐depleted adipocytes. The mechanism of how the adipocytes replenish the lipid content after withdrawal from Wnt activation is unclear. Either reuptake of fatty acids or use of the de novo lipogenesis pathway to accumulate lipid content could allow adipocytes to store fatty acids after Wnt activation is abrogated [58]. Future studies could focus on physiological pathways that facilitate the recovery of lipodystrophic adipocytes. Second, our inducible and reversible Wnt activation in dermal adipocytes revealed that profibrotic ECM remodeling in the DWAT and dermal fibroblast layer is also a reversible state. Our model demonstrates that Wnt activation in a subset of mature dermal adipocytes not only has a cell autonomous effect, but also impacts neighboring cells and the dermal layer. This interpretation is supported by the evidence of an increase in cell proliferation, ECM accumulation, and collagen remodeling in the DWAT and dermal layers in Adipo‐β‐cat^istab dorsal skin. It is possible that the free fatty acid and glycerol released by Wnt‐induced lipodystrophy can activate the quiescent dermal fibroblasts into profibrotic fibroblasts. Studies have shown that perturbation of fatty acid oxidation to glycolysis or expression of CD36, fatty acid transporter, leads to accumulation of ECM in fibroblasts [59]. Analysis of single cell RNA‐seq of SSc revealed a signature for fatty acid transport and metabolism in subsets of fibroblasts [55]. Thus, our inducible and reversible model of Wnt activation in dermal adipocytes highlights the importance of the lipid depletion event in the DWAT layer as a new player in promoting skin fibrosis.

The source of the profibrotic cell that mediates the ECM remodeling and accumulation in most fibrosis models and our model is unclear. There is controversy in the field about the transition of mature adipocytes to activated fibroblasts at the functional level in acute and chronic fibrosis models [60, 61, 62, 63, 64]. In the standard bleomycin fibrosis model, a small number of non‐inducible Adiponectin‐Cre lineage labeled adipocyte cells appeared in the dermal layer with fibroblast morphology, suggesting trans‐differentiation [60]. Recently, it was shown that adipocytes can dedifferentiate during wound healing, indicated by α‐SMA expression for myofibroblast identity, which was recently shown to be insufficient for the function of myofibroblasts in skin [57, 63, 65]. Adiponectin, a signature gene of the mature adipocytes, decreases in expression during adipocyte de‐differentiation [66]. Adiponectin receptor signaling can downregulate collagen and α‐smooth muscle actin mRNA expression in dermal fibroblasts in vitro [67]. In late‐stage wounds, macrophages promote fibrosis by phagocytosing the dermal Wnt inhibitor SFRP4, sustaining chronic Wnt activity and preventing regenerative repair in the wound‐induced hair neogenesis model [68]. Interestingly, DWAT thickness is highest during P32 late anagen stage of the murine hair cycle [69]; we found that sustained Wnt activation in a subset of mature dermal adipocytes was sufficient to cause a significant decrease in the DWAT layer thickness and ECM expansion and remodeling, compared to age and hair follicle stage matched control skin. Future studies can investigate the interplay between skin fibrosis associated with lipodystrophy, ECM remodeling, and hair follicles. Altogether, our data support a model where Wnt activation leads to lipodystrophy in dermal adipocytes that may allow them to de‐differentiate and contribute to the profibrotic process. Future research will take advantage of single nuclei RNA seq analysis to track the cell trajectory of GFP⁺ lineage marked cells and other cell types in the whole skin of Adipo‐β‐cat^istab mice.

Human SSc skin and limited scleroderma subtype that does not have a proliferative and immune activation has a decrease in lipid handling genes for fatty acid metabolism, lipid biosynthesis expression [70]. Here, we found core lipolysis genes were significantly elevated in the first 2 years of disease duration in human SSc (Figure S9). The mechanisms of DWAT lipodystrophy and its impact on ECM in skin are emerging from functional studies of the ATGL‐lipolysis pathway. In the short‐term bleomycin injection skin fibrosis model (Adipo nectin‐CreER; Atgl ^fl/fl ) or with pharmacological inhibition with Atglistatin on wild type dorsal skin, it leads to dermal expansion [55, 71]. In both these models, bleomycin exposure and Atglistatin treatment is on the whole skin, thereby affecting multiple cell types. Given that Adiponectin‐CreER is expressed in other mature white adipose tissue depots and we get low levels of recombination in visceral fat depots, we cannot exclude the possibility that Wnt induced lipolysis in remote fat depots may contribute to the fibrotic changes in the dorsal skin. Here, we restricted Wnt activation and Atgl deletion to mature dermal adipocytes (Adipo‐β‐cat^istab; Atgl ^fl/fl ) and rescued the lipodystrophy and ECM accumulation in skin. To the best of our knowledge, our model in vivo is the first to demonstrate that lipolysis from mature dermal adipocytes is functionally required and sufficient for ECM accumulation in the DWAT and dermal layers. Recently, lipochondrocytes in ear skin were found to have protected stable lipid vacuoles that lack lipolysis signature genes such as Perilipin1 and Lipe/HSL and have increased expression of Wnt antagonists [72, 73]. Our tissue restricted model will allow us to further dissect the molecular mediators that promote ECM accumulation and translate our findings to use lipolysis pathway inhibition as a new therapeutic target for modulating ECM remodeling and adipocyte lipid handling in fibrosis.

The mechanism of how Wnt signaling activation in dermal adipocytes activates ATGL‐dependent lipolysis is unclear. The ENCODE consortium study, using the HCT116 human colon cancer cell line that has elevated Wnt signaling, shows intronic regulatory elements that are active (by H3K27Ac) and enriched for TCF7L2 between exon1 and 2 in human PNPLA2 and LIPE (GSM782123) [74, 75]. The link between Wnt and ATGL activation could be mediated by Wnt effectors of lipodystrophy such as Dipeptidyl peptidase‐4 (DPP4) and its substrates [31] or physiological processes such as Reactive Oxygen Species (ROS) [76]. Further investigation is necessary to determine the role and mechanism of these factors in the activation of the lipolysis axis in our model. Our genetic model reveals that Wnt activation in mature dermal adipocytes is sufficient to cause ATGL‐dependent lipolysis in the DWAT and a profibrotic response in the skin. Furthermore, our analysis shows that the lipolysis genes are significantly elevated in the first 2 years of the disease in SSc patients, which indicates a close relationship between adipocyte lipid handling homeostasis and the clinical disease. Thus, recovery from the lipodystrophy during skin fibrosis could possibly bring back the functions of the DWAT layer in immune response, hair follicle cycle regulation, thermoregulation, and vascularization and restore ECM homeostasis [7, 11, 12].

Author Contributions

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Atit had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study Conception and Design: Atit and Ma. Acquisition of Data: Ma, Segal, Reynolds, Madhavan, Gregory, Wyetzner, Montegut, Jussila, and Ertugral. Analysis and Interpretation of Data: Atit, Ma, Segal, Montegut, Madhavan, Jussila, Wyetzner, Ertugral, and Kothapalli. Writing and editing: Atit and Ma wrote the manuscript and were edited by all the authors. All the authors have approved the submitted version.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1.

Ma Q., Segal E. X., Montegut M. A., et al., “Wnt Activation in Mature Dermal Adipocytes Leads to Lipodystrophy and Skin Fibrosis via ATGL‐Dependent Lipolysis,” The FASEB Journal 39, no. 13 (2025): e70830, 10.1096/fj.202501380R.

Data Availability Statement

Data will be made available after publication and request.