Paracrine Signaling by Distinct Adipose Tissue Depots Regulate Fibroblast Mechanobiology and Functional Heterogeneity

Abstract

Purpose

Adipose tissues have long been recognized for their diverse endocrine functions that serve to regulate tissue homeostasis. Although adipose depot-specific secretory profiles can differentially regulate fibroblast fate and fibrotic tissue remodeling, systematic investigation of how adipose depots of varying phenotypes influence fibroblast mechanobiology remains lacking

Methods

By integrating secretome profiling with functional and mechanobiological assays in a controlled adipose-fibroblast coculture platform, we mechanistically investigate how canonical brown (BAT), white (WAT), and beige adipose tissues (thoracic and abdominal perivascular adipose tissue; T-PVAT and A-PVAT, respectively) govern fibroblast-to-myofibroblast transition (FMT) through depot-specific paracrine signaling.

Results

Thermogenic depots (BAT and T-PVAT) enhance fibroblast proliferation, collagen deposition, and stiffness (1.9-fold increase in single cell modulus; p < 0.0001) alongside specific adipokines that regulate metabolism (ANGPTL3, FGF1) and inhibit calcification (Fetuin A). In contrast, lipid-storing depots (WAT and A-PVAT) promote fibroblast migration (79% scratch wound closure at 12 h; p < 0.01) and matrix degradation through upregulated pro-inflammatory cytokines (CCL2, IL-6) and fibrogenic mediators (PAI-1, TIMP-1). Mechanistically, BAT and T-PVAT secretomes induce pro-contractile and matrix-producing fibroblast phenotypes, while A-PVAT and WAT promote migratory and matrix-remodeling behaviors. Further, T-PVAT uniquely combines BAT-like genetic identity with elevated RAGE-DPP4 signaling, resulting in excessive fibroblast activation, stiffening, and collagen production that can be attenuated via pharmacological inhibition of DPP4.

Conclusions

Our findings establish adipose depots as distinct modulators of fibroblast phenotype, wherein depot-specific mediators promote mechanobiological alterations that balance reparative and pathological remodeling processes related to fibrosis. Differential biological responses by fibroblasts exposed to diverse adipose phenotypes underscores the potential for adipose-driven stromal crosstalk to mitigate fibrotic remodeling alongside metabolic and cardiovascular disease comorbidities.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12195-025-00864-z.

Introduction

Adipose tissue was once regarded primarily as a reservoir for energy homeostasis and storage but is now recognized as a dynamic endocrine organ with depot-specific functional diversity. In particular, distinct anatomical adipose depots defined based on adipocyte phenotypes—namely, brown adipose tissue (BAT), white adipose tissue (WAT), and beige adipose tissue (BeAT)—exhibit unique immune, metabolic, and secretory profiles that influence their interactions with neighboring stromal cells and organs [28, 59]. Traditionally, BAT is characterized by high mitochondrial content and its ability to drive thermogenesis through uncoupled mitochondrial respiration, whereas WAT primarily functions in excess lipid and caloric energy storage; BeAT displays an intermediate phenotype, adopting both BAT- and WAT-like characteristics based on canonical gene expression, cellular morphology, and biological function [42, 59]. Perivascular adipose tissue (PVAT) is a specialized subcategory of BeAT that surrounds blood vessels throughout the body. Aortic PVAT has been shown to exhibit regional heterogeneities in adipose composition and function which has been associated with regional presentation of cardiovascular disease [32]. Namely, thoracic PVAT (T-PVAT) shares molecular and morphological similarities with BAT, including elevated uncoupling protein 1 (UCP1) expression and vasodilatory properties; whereas abdominal PVAT (A-PVAT) functionally aligns more with WAT, promoting excess inflammation and vascular dysfunction [6, 55]. These spatial diversities in adipose phenotype underscore the critical, yet understudied, role of adipose tissue in modulating vascular and stromal cell biology to promote soft tissue remodeling and fibrosis.

Fibrosis is a persistent condition characterized by excessive extracellular matrix (ECM) deposition that remains a leading cause of organ dysfunction. Despite extensive study of this pathological condition in multiple soft tissue diseases, there remains no effective therapies to target and remove established fibrotic lesions [60]. Thus, targeting the underlying biological process central to fibrosis prior to the development of excess ECM deposition remains the preferred treatment course of action. Central to fibrotic remodeling is the fibroblast-to-myofibroblast transition (FMT), a process marked by heightened contractility, α-smooth muscle actin (αSMA) expression, and ECM synthesis by activated fibroblasts [6, 20, 64]. Emerging evidence highlights adipose-derived paracrine factors as key regulators of FMT, though their effects vary dramatically across adipose depots [42, 57]. For instance, thermogenic depots (BAT and T-PVAT) secrete metabolites which have been shown to modulate fibroblast proliferation and ECM turnover (irisin, FGF21, etc.), playing a potential protective or reparative role against pathological fibrosis [25, 30, 52]. On the other hand, pro-inflammatory cytokines secreted by WAT and A-PVAT (TNF-α, IL-6, etc.), amplify collagen degradation and inflammatory responses in stromal cell populations to promote a pro-fibrotic phenotype [27, 46, 58, 62]. Previous studies have also highlighted the role of PVAT in promoting regional vascular fibrosis via TGF-β1 signaling [29] and have demonstrated that adipose-derived mediators can either promote tissue repair or exacerbate fibrotic remodeling depending on the depot of origin [6, 55].

Advances in experimental in vitro modeling techniques, including the use of multicellular coculture platforms and conditioned media-based approaches, enable precise investigation of biological mechanisms associated with paracrine interactions between cells and tissues [14, 15]. Despite growing evidence of the importance of adipose-fibroblast crosstalk in soft tissue remodeling and fibrosis, systematic comparisons of the effect of canonical adipose signaling on fibroblast mechanobiology and biological function remain lacking. By integrating morphological, genetic, and mechanobiological analyses with secretome profiling of adipose-conditioned media, this study comprehensively explores how adipose depots influence fibroblast fate. We demonstrate that thermogenic adipose depots (BAT and T-PVAT) promote myofibroblast activation through distinct secretory profiles that enhance ECM synthesis and contractility, while lipid-storing depots (A-PVAT and WAT) drive inflammatory matrix degradation independent of overt myofibroblast activation. These findings not only advance our understanding of adipocyte–fibroblast crosstalk in the context of FMT but also underscore the therapeutic potential of targeting depot-specific adipose signals to mitigate fibrotic tissue remodeling alongside metabolic and cardiovascular comorbidities [4, 41, 49].

Methods

Mice

Male wild type C57BL/6 J mice (The Jackson Laboratory; ID:000664) between 8 and 12 weeks of age were included in the study. All mice were housed in pathogen-free environmental conditions at standard room temperature and 47% humidity with a 12 h light/dark cycle and regular chow and water available ad libitum. All animal procedures were approved by the Washington University in St. Louis Institutional Animal Care and Use Committee (IACUC).

Adipose Tissue Isolation and Culture

Prior to adipose tissue isolation, experimental mice were anesthetized by isoflurane inhalation. Following confirmation of anesthesia, canonical BAT (interscapular adipose tissue), WAT (epidydimal adipose tissue), and PVAT (aortic perivascular adipose tissue) depots were gently dissected. To investigate regional differences in PVAT, adipose tissue surrounding the thoracic aorta (from aortic arch to diaphragm; T-PVAT) was separated from that surrounding the abdominal aorta (from diaphragm to iliac bifurcation; A-PVAT). Isolated adipose depots were immediately stored on ice in sterile biopsy storage media composed of complete Dulbecco’s Modified Eagle Medium (DMEM; Corning, MT10013CV) supplemented with 1% antibiotic-antimycotic solution (Sigma-Aldrich, A5955). To standardize the amount of adipose tissue used in subsequent experiments, each depot was sectioned into 5 ± 0.25 mg pieces. After weighing, trimmed adipose tissues were quickly rinsed in warmed 70% ethanol (Fisher Bioreagents, BP28184) and placed in sequential rinses of Dulbecco’s Phosphate-Buffered Saline Solution (1x DPBS; Corning, MT21031CV) and biopsy storage media prior to placement in low serum tissue culture media comprised of complete DMEM (Corning, MT17205CV) supplemented with 0.5% fetal bovine serum (FBS; Gibco, A5256801), 1% penicillin/streptomycin (PS; Corning, 30002Cl) and 1% L-Glutamine (Sigma-Aldrich, G7513). Each piece of trimmed adipose tissue was transferred to individual wells of a 24-well plate (Fisherbrand, FB012929) and maintained in 2 mL of low serum tissue culture media. After 48 h, the conditioned media from each cultured piece of adipose tissue was collected, aliquoted, and stored at −80 °C for future experiments.

Gene Expression Analysis

Relative gene expression focused on adipose tissue identity was measured from each adipose depot using quantitative reverse transcription polymerase chain reaction (RT-qPCR). Briefly, total RNA was extracted separately from each adipose depot using the RNeasy Micro Kit (Qiagen, 74004). Following extraction, RNA purity and quality was evaluated with spectrophotometer-based absorbance measurements. Once quality was confirmed, total mRNA was reverse transcribed into cDNA using the SuperScript IV VILO Master Mix (Invitrogen, 11766050). Real-time qPCR was performed on the QuantStudio3 Real-Time PCR system based on detection of PowerTrack SYBR Green Master Mix (Invitrogen, A46012). Relative gene expression between adipose depots was calculated using the comparative CT method (2^-ΔΔCT) [33]. Raw CT values for each gene of interest were normalized to glyceraldehyde 3-phosphate dehydrogenase (Gapdh; ΔCT) as the endogenous reference housekeeping gene and fold changes in gene expression were determined as 2^-ΔΔCT; ΔΔCT values were calculated in Microsoft Excel. Amplification of each gene of interest was performed in duplicate across a minimum of three independent biological replicates. Primers for RT-qPCR were synthesized by Integrated DNA Technologies. Primer specificity was confirmed by single‐product amplification by melt‐curve analysis and by BLAST analysis against the target genome to ensure no off-target binding sites were present. Forward and reverse primer sequences for selected genes of interest related to adipose tissue identity are shown in Table 1.

Table 1.

Primers used for RT-qPCR analysis of adipose tissue-specific genes

Gene	Forward (5’ → 3’)	Reverse (5’ → 3’)
Ucp1	AGGCTTCCAGTACCATTAGGT	CTGAGTGAGGCAAAGCTGATTT
Ppargc1a	TATGGAGTGACATAGAGTGTGCT	CCACTTCAATCCACCCAGAAAG
Pparg	TCGCTGATGCACTGCCTATG	GAGAGGTCCACAGAGCTGATT
Lep	ATGTGCCCTTCCGATATACAACC	CGTGTCATCCACTAATCTTCTGG
Adipoq	CAAGGCCGTTCTCTTCACCT	CCCCATACACTTGGAGCCAG
Gapdh	AGGTCGGTGTGAACGGATTTG	TGTAGACCATGTAGTTGAGGTCA

Total Protein Quantification

To quantify the amount of total protein secreted by different adipose depots, conditioned media from each culture condition was analyzed with a commercial kit based on the use of bicinchoninic acid (BCA) for colorimetric protein detection (Pierce BCA Protein Assay Kit; ThermoFisher Scientific, 23227). Briefly, bovine serum albumin (BSA) was diluted in low serum tissue culture media to prepare a ladder of diluted standards. As per the manufacturer instructions, working reagent (WR) was prepared by mixing BCA Reagent A with BCA Reagent B at a ratio of 50:1, %v/v. Using a 96-well plate (Corning, 3596), 25 μL of either protein standard or warmed (room temperature) conditioned media were pipetted into individual wells in duplicate. 200 μL of WR was added to each well followed by gently mixing the plate on a wave motion rocker for 30 sec. The plate was then incubated at 37 °C and 5% CO₂ for 30 min. After incubation, the plate was cooled to room temperature and optical densities in each well were read at 562 nm by a multimodal cell imaging plate reader (BioTek, Cytation 1). Low serum tissue culture media was used as a blank control to offset background signal from optical readings of experimental wells. A standard curve was generated from known protein concentrations of the BSA ladder (in μg/mL) and the corrected optical readings for each standard well. Total secreted protein concentration in adipose-conditioned media was calculated by interpolating corrected optical density measurements onto the standard curve.

Mouse Adipokine Array

A selection of adipose-specific secreted factors was detected in conditioned media using a Mouse Adipokine Array Kit (R&D Systems, ARY013), according to the manufacturer’s protocol. Briefly, conditioned media samples from each adipose depot were incubated overnight at 4 °C with membranes pre-spotted with capture antibodies against 38 adipokine-related proteins in duplicate. After washing, the membranes were incubated with a cocktail of biotinylated detection antibodies for 1 h at room temperature, followed by plasmonic-fluors with an 800CW infrared dye [34]. Fluorescent intensities of the captured protein at each spot on the membrane were captured using a Licor Odyssey CLx Infrared Imaging System and ImageStudio software. Pixel densities at each spot were quantified and extracted using custom image processing software developed in MATLAB (R2023a). The relative expression of each detected adipokine was normalized to positive control reference spots provided on the membrane with all values reported as a normalized fluorescence intensity. A minimum of three independent biological replicates were analyzed for each adipose depot.

Cell Culture

NIH/3T3 mouse fibroblasts were maintained in a high serum tissue culture media during expansion and passaging prior to performing any cell-related experiments. High serum tissue culture media was comprised of complete DMEM (Corning, MT17205CV) containing 10% FBS (Gibco, A5256801), 1% PS (Corning, 30002Cl) and 1% L-Glutamine (Sigma-Aldrich, G7513). Cells were transferred to low serum tissue culture media overnight prior to use in any experimental procedure. Cells were incubated at 37 °C and 5% CO₂ throughout the entire period of study.

Cell Proliferation Assay

In vitro proliferation rate of fibroblasts maintained in either low serum media or conditioned media from each adipose depot was assessed using a tetrazolium-based colorimetric MTT assay (Cell Proliferation Kit I; Roche, 11465007001), as per the manufacturer’s instructions. Fibroblasts were initially seeded in high serum tissue culture media at 10,000 cells/well in a 96-well plate. After 24 h of culture, media was replaced by 100 μL of either low serum tissue culture media (control) or adipose-conditioned media. Cells were cultured for 48 h, after which 10 μL of MTT labeling reagent (final concentration 0.5 mg/mL) was added to each well. After 4 h of incubation, 100 µL of solubilization solution was added to each well to dissolve any formazan crystals that had formed. Plates were incubated overnight at 37 °C and absorbance was measured at wavelengths of 550 nm and 700 nm using a multimodal plate reader (BioTek, Cytation 1). The final optical density in each well was calculated by subtracting the 700 nm reference absorbance from the 550 nm measurement absorbance to account for non-specific background staining of the colorimetric reagents. For each biological replicate in the conditioned media groups, proliferation measurements were performed in triplicate.

Scratch Wound Healing Assay

In vitro cell migration behavior was evaluated using a scratch wound healing assay. Briefly, fibroblasts were seeded in a 24-well plate at 65,000 cells/well in high serum tissue culture media to promote attachment and expansion. When cell monolayers reached 80-90% confluence (approximately 24 h post-seeding), the culture medium was replaced with either low serum tissue culture media (control) or conditioned media from each adipose depot. A uniform cell-free zone was created by manually scratching the monolayer with a sterile 100 µl pipette tip to generate a linear wound. Closure rates were monitored at 3-h intervals from 0 to 15 h post-scratch using an inverted DIC microscope (LAXCO, NA 0.30 WD72) equipped with a 4× magnification objective (Lplan). Wound closure was quantified by measuring the cell-free area at each post-scratch time point using ImageJ software (Version 1.53, National Institutes of Health). The extent of wound closure was expressed as an areal percentage calculated as ($\left(A_o-A_t\right)\left./\phantom{\left(A_o-A_t\right)A_o}\right)A_o\times 100\%$ where $A_o$ denotes the initial wound area and $A_t$ denotes the wound area at time $t$. All wound healing experiments were performed in duplicate for each biological replicate in the conditioned media groups.

Collagen Deposition Assay

To examine the extent of in vitro collagen deposition by cultured fibroblasts, picrosirius red (PSR) staining and quantification was performed. Briefly, cells were plated in a 96-well plate at 10,000 cells/well in high serum tissue culture media to promote attachment. After 24 h, the media was replaced by 100 μL of either low serum tissue culture media (control) or adipose-conditioned media. After 48 h of culture, the media was removed and cells were fixed in-place with 10% neutral buffered formalin (Sigma-Aldrich, HT501128) for 10 min. After fixation, cells were washed in DPBS on a plate rocker for 2 × 15 min prior to adding 50 μL of PSR stain (Electron Microscopy Science, 26357-02) per well. Stained cells were incubated at room temperature for 1 h followed by rinsing in three changes of 0.5% acetic acid in distilled water (dIH₂O) (Sigma-Aldrich, A6283). To elute the collagen-bound PSR stain for optical quantification, 100 μL of 0.1 N sodium hydroxide (Sigma-Aldrich, 2105) was added to each well and the plate was incubated at room temperature for 1 h. Optical densities were read at 540 nm (BioTek, Cytation 1). Collagen deposition measurements were performed in triplicate for each biological replicate in the conditioned media groups.

Immunofluorescence

To characterize the distinct morphological features of each adipose depot, freshly isolated adipose tissues were immersion-fixed in 4% paraformaldehyde for 24 h at 4 °C. Following fixation, tissues were thoroughly rinsed in DPBS to remove residual fixative prior to embedding in optimal cutting temperature compound (OCT; Fisher Scientific, 2373057). Serial tissue sections at 8 μm thickness were obtained using a cryostat maintained at −39 °C (Thermo Scientific, CryoStar NX50) and mounted on Superfrost Plus microscope slides (Fisher Scientific, 1255015) for staining. Slides were gently washed with DPBS to remove residual OCT and stained with Prolong Gold with DAPI (Invitrogen, P36935) to visualize cell nuclei while mounting; native adipose tissue autofluorescence was used to delineate tissue morphology. Excitation at 488 nm was used to visualize autofluorescence of ECM components (reticular, collagen, and elastin fibers) and lipofuscin deposits [9, 54]. This signal delineated adipocyte boundaries (defined by non-autofluorescent intracellular spaces) from autofluorescent stromal architecture [10]. Nuclei density was quantified by normalizing the count of DAPI-stained nuclei by total adipose tissue area including both autofluorescent and non-autofluorescent regions (see Fig. S1). Adipocyte size (perimeter and area) was quantified using Cellpose, an open-source image segmentation algorithm, with a custom model trained on autofluorescence images from BAT and WAT depots (see Fig. S6) [43, 53]. A detailed description of the Cellpose segmentation model, accuracy, and image quantification procedures can be found in the Supplementary Materials document.

To confirm adipocyte identity based on canonical protein expression, tissue sections were blocked for 1 h at room temperature in DPBS containing 5% BSA (Sigma-Aldrich, A3059) and 3% goat serum (Sigma-Aldrich, G9023) following fixation/permeabilization in a solution of 10% neutral buffered formalin (Millipore Sigma, HT501128) and 0.1% Triton X-100 (Sigma-Aldrich, x100). Sections were then incubated overnight at 4 °C with rabbit polyclonal anti-uncoupling protein 1 (UCP1) antibody (1:500; Abcam, ab10983). After 2 x 15 min washes in DPBS, sections were incubated for 1 h at room temperature with Cy5-conjugated goat anti-mouse IgG secondary antibody (1:300; Life Technologies, A10524). After extensive washing with DPBS for 30 min, stained coverslips were mounted using Prolong Gold with DAPI.

In vitro myofibroblast activation was assessed by immunofluorescent detection of selected contractile and matrix adhesion proteins—α-smooth muscle actin (αSMA) and vinculin, respectively. Briefly, fibroblasts were seeded in a 24-well plate at 25,000 cells/well on sterile cover glass (12 mm diameter; Electron Microscopy Science, 72230-01) and maintained in high serum tissue culture media to promote attachment. After 24 h, the media was replaced by 1 mL of either low serum tissue culture media (control) or adipose-conditioned media. After 48 h of culture, cells were washed with DPBS and fixed/permeabilized for 10 min at room temperature using 10% neutral buffered formalin containing 0.1% Triton X-100 (Sigma-Aldrich, X100). After washing with DPBS for 2 × 15 min on an orbital shaker, samples were blocked for 1 h at room temperature in DPBS containing 5% BSA and 3% goat serum. Cells were then incubated with mouse monoclonal anti-vinculin antibody (1:200; Sigma-Aldrich, V9131) overnight at 4 °C. Stained samples were washed in DPBS 2 × 15 min and incubated with Cy5-conjugated goat anti-mouse IgG secondary antibody (1:300; Life Technologies, A10524) and Cy3-conjugated anti-αSMA antibody (1:300; Sigma-Aldrich, C6198) for 1 h at room temperature. After extensive washing with DPBS for 30 min, stained coverslips were mounted using Prolong Gold with DAPI.

Fluorescence imaging of stained cells and tissues was performed using an Olympus BX53 microscope equipped with a dual-sensor DP80 digital camera (Olympus). Images were acquired using consistent magnification and exposure settings across all experimental conditions to allow for quantitative image processing and analysis.

DPP4 Inhibition Assay

To investigate the role of the RAGE-DPP4 signaling axis in promoting fibroblast activation, we performed targeted inhibition of DPP4 signaling via in vitro treatment with the highly selective inhibitor sitagliptin phosphate monohydrate (sitagliptin; Abcam, ab142101). A 10 mM stock solution of sitagliptin was prepared in sterile DPBS and stored at −20 °C protected from light until use. For experiments, the sitagliptin stock was thawed and diluted in either low serum tissue culture media (media control) or adipose-conditioned media (e.g., T-PVAT) to a final concentration of 20 nM. Vehicle (treatment control) groups received the same volume of DPBS as the treatment groups, but without sitagliptin.

Traction Force Microscopy

To assess single cell contractile force generation, a traction force microscopy (TFM) protocol was implemented, as previously described [5, 23, 26]. Briefly, polyacrylamide (PAA) gels of an intermediate-to-hard stiffness (7.32 kPa) containing embedded fluorescent microspheres (excitation/emission: 540/560 nm; ThermoFisher Scientific, F8809) were fabricated in glass-bottom 35 mm culture plates (Mattek, P35G014C). Following gel fabrication and quality control, fibroblasts were seeded at 5,000 cells/gel in high serum tissue culture media to promote attachment to the PAA gel. After 24 h, the media was replaced by 2 mL of either low serum tissue culture media (control) or adipose-conditioned media. After 48 h, gels were placed inside the incubation chamber of a wide-field Zeiss LSM 880 inverted confocal microscope at 37 °C and 5% CO₂ and individual cells were located and imaged under a 20x objective. For each cell, a phase contrast image was taken to define the cell boundary and two images of the fluorescent microspheres just below the gel surface were taken under stressed (attached) and relaxed (detached; after 10% Triton-X100 solution treatment) conditions to determine the local bead displacement and corresponding traction forces using a modified MATLAB script developed from an open-source regularized TFM analysis [18].

Following manual tracing of the cell boundary in phase contrast images, the cell area $\left(CSI=4\pi \left(A/P^2\right);\text{measure}\text{of}\text{circularity}\right)$ were calculated. Additionally, comparisons between fluorescent bead locations in “stressed” and “relaxed” images were performed using a local cross-correlation-based particle image velocimetry technique [22]. Using the identified bead displacement field $u(x,y)$, where $u$ is the displacement vector at any pixel location $(x,y)$ contained in the image, an optimal L1-regularized Fourier Transform traction cytometry (FTTC) approach was used to inversely identify corresponding cell-generated traction vectors assuming PAA gel incompressibility $(\nu =0.5)$ and known stiffness $(\text{E}=7.32\text{kP}a)$ [7]. Following identification of an optimal traction vector $T(x,y)$ at each pixel location, the total force generated by the cell $\left|F\right|$ was calculated as the summation of the traction vector magnitude at each pixel within the cell boundary [45],

$$\left|F\right|=\iint \sqrt{T_x{\left(x,y\right)}^2+T_y{\left(x,y\right)}^2}dxdy$$ 1

where subscripts $x$ and $y$ denote components of the traction vector $T$ in the $x$-and $y$-direction, respectively. Additionally, the strain energy density $W$—a scalar measure of elastic deformation of the PAA gel and thus a proxy for cell contractility—was calculated as one-half the summation of the dot product between the traction vector $T(x,y)$ and displacement vector $u(x,y)$ at each pixel within the cell boundary [36],

$$W=\frac{1}{2}\iint T\left(x,y\right)·u\left(x,y\right)dxdy$$ 2

A more detailed methodological description of each aspect of the TFM procedure can be found in the Supplementary Materials document.

Adipose Tissue–Fibroblast Coculture

In a subset of experiments, indirect cocultures between adipose tissues and fibroblasts were performed. Briefly, fibroblasts were seeded in a 24-well plate at 25,000 cells/well in high serum tissue culture media. After 24 h, the cell culture media was replaced with 2 mL of fresh high serum media. Trimmed samples from each adipose depot were carefully transferred to each well and floated on the surface of the media. Cells were maintained in coculture for 48 h prior to subsequent analysis; the use of high serum media here aids in the generation of a single cell suspension prior to testing.

Micropipette Aspiration

For single cell stiffness and viscoelasticity measurements, micropipette aspiration was performed as previously described [24, 56]. Briefly, custom micropipettes were pulled from borosilicate glass capillaries (World Precision Instruments, TW100-6) using a micropipette puller (Sutter Instrument Company, P-97) at the standard settings (pressure = 500, heat = 550, pull = 55, velocity = 75) and shaped (10 μm diameter, 30-degree tip angle) using a microforge (Narishige, MF-900).

On the day of mechanical testing, adherent cells were enzymatically detached from the 24-well plate using 1x TrypLE Select (Gibco, 12563-011) which was neutralized with high serum tissue culture media prior to pelleting via centrifugation (300×g, 5 min, 4 °C). Cell suspensions were resuspended in 1 mL of ice-cold 1 × Hanks’ Balanced Salt Solution (HBSS; Corning, MT21022CM) and maintained on ice for ≤ 30 min to preserve membrane integrity. Resuspended cells were then aliquoted and subjected to either elastic (linear pressurization at 4 Pa/s for 150 s) or viscoelastic (step pressurization to 600 Pa held for 150 s) mechanical characterization. Following aspiration, video data was collected and processed to reveal the aspiration distance of each cell into the micropipette in response to controlled pressurization.

Single cell bulk stiffness (i.e., Young’s modulus E) was calculated from the elastic testing protocol under the simplifying assumption of an infinite elastic half-space [21], wherein

$$\mathrm{\Delta }P=\frac{2\pi }{3}E\frac{L_p}{R_p}\mathrm{\Phi }$$ 3

Here, $\mathrm{\Delta }P$ denotes the applied suction pressure, $L_p$ is the aspirated length of the cell into the pipette, $R_p$ is the pipette radius, and $\mathrm{\Phi }$ = 2.1 is a constant that accounts for pipette wall thickness effects. Thus, the slope of the best fit linear relationship between $\mathrm{\Delta }P$ and the ratio $L_p/R_p$ is proportional to the stiffness $E$. Viscoelastic parameters, such as the apparent cell viscosity ($\mu$), were derived from the standard linear solid model [17],

$$L\left(t\right)=\frac{\mathrm{\Phi }{R}_c\mathrm{\Delta }P}{\pi k_1}\left[1-\frac{k_2}{k_1+k_2}{e}^{-\frac{t}{\tau }}\right]$$ 4

Here, $k_1$ and $k_2$ represent elastic constants, $R_c$ is the cell radius, $\tau$ is the relaxation time constant, $t$ is the aspiration time, and again Φ = 2.1. Following identification of elastic and time constants via optimization, the viscosity $\mu$ was calculated as $\mu =\tau k_1{k}_2/(k_1+k_2)$. For all single cell biophysical measurements, outlier detection was performed in MATLAB based on the interquartile range and excluded from further data analysis.

Statistical Analysis

Data throughout the manuscript are presented as mean ± standard error of the mean (SEM) from at least three independent biological replicates. Sample sizes are denoted by individual data points in each plot. Statistical significance was determined using either one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test for multiple pairwise comparisons or ordinary two-way ANOVA followed by Tukey’s multiple comparisons test for comparisons with two factors, based on the experimental design. One-way ANOVA followed by Dunnett’s multiple comparisons test was used when comparing treatment groups to a single control group. Differences were considered statistically significant at p < 0.05. All statistical tests were two-tailed unless otherwise specified. Statistical analyses were performed using GraphPad Prism software (version 10.0, GraphPad Software, San Diego, CA).

Results

Characterization of Adipose Tissue Identity: Morphological and Genetic Profiling

To confirm the identities of the canonical adipose tissues included in this study—intrascapular adipose tissue (BAT), perivascular adipose tissue (BeAT; T-PVAT and A-PVAT), and epididymal adipose tissue (WAT)—we first evaluated their morphological and genetic hallmarks. Autofluorescence imaging revealed distinct tissue composition and cellular morphologies across adipose tissue subtypes. When imaged at the same magnification, brown adipocytes exhibited the smallest cell size, while adipocyte diameter progressively increased in tissues with a whiter macroscopic appearance (Fig. 1a). Consistent with this observation, quantitative analysis of nuclei density (e.g., number of nuclei per tissue area) demonstrated that BAT harbored the highest nuclei density, followed by T-PVAT (p > 0.05), A-PVAT (p < 0.001), and WAT (p < 0.001; Fig. 1b and Fig. S1, each data point represents average values from one biological sample). Although T-PVAT exhibited a lower nuclei density than BAT on average, this difference was not statistically significant, indicating overlapping morphological features between these depots. Quantitative analysis revealed average adipocyte perimeters of 60.3 ± 3.6 μm for BAT, 59.5 ± 3.0 μm for T-PVAT, 94.0 ± 6.5 μm for A-PVAT, and 151.0 ± 8.7 μm for WAT, confirming a progressive increase in adipocyte size from brown to white depots (p < 0.0001, one-way ANOVA; Fig. 1c, Fig. S6 and S7).

Fig. 1.

Morphological and genetic characterization of adipose tissue depots. a Representative immunofluorescence images of autofluorescence signal (green) from BAT, T-PVAT, A-PVAT, and WAT depots under 20x magnification. Autofluorescence originates from ECM components and lipofuscin outlining adipocyte boundaries. Nuclei are labeled with DAPI (blue). Scale bar = 100 µm. b Quantification of nuclei density in adipose tissue sections calculated as number of nuclei per mm² of tissue. c Quantification of adipocyte perimeters (μm) in adipose tissue sections (see Fig. S7 for representative images). n = 6 mice/group. d Relative mRNA expression of adipose-specific genes canonically enriched in either brown adipocytes (Ucp1, Ppargc1a) or white adipocytes (Pparg, Lep, Adipoq). n = 3 mice/group. e Quantification of UCP1 median fluorescence intensity. f Representative immunofluorescent images of UCP1 (red), and DAPI (blue) from each adipose depot. n = 6 mice/group. All data shown as mean ± SEM. Statistical significance from one-way ANOVA with post-hoc Tukey test for multiple comparisons denoted by *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, or ****p ≤ 0.0001

Gene expression analysis of adipose phenotype markers further differentiated these four adipose tissue depots (p < 0.005 for all genes; Fig. 1d). Namely, BAT and BeAT showed pronounced enrichment of brown adipocyte-specific genes Ucp1 (protein: UCP1) and Ppargc1a (protein: PGC-1α). Indeed, Ucp1 expression was significantly reduced in WAT compared to BAT and both BeAT depots (p < 0.0001). Ppargc1a levels in BAT and BeAT were similarly elevated compared to WAT, albeit less than Ucp1, thus reinforcing the thermogenic potential of BAT, T-PVAT, and A-PVAT. In contrast, WAT-enriched genes including Pparg (protein: PPARγ), Lep (protein: Leptin), and Adipoq (protein: adiponectin) were found to be highest in WAT but also further distinguished T-PVAT from A-PVAT. Specifically, T-PVAT and BAT shared indistinguishable expression levels of Pparg, Lep, and Adipoq, suggesting a genetic profile in T-PVAT that aligns more closely with BAT than with WAT. However, A-PVAT displayed more pronounced Lep and Adipoq mRNA levels, consistent with its WAT-like phenotype. UCP1 immunostaining further confirmed adipocyte identity at the protein level. Immunofluorescence analysis revealed a progressive decrease in UCP1 expression from thermogenic to lipid-storing depots (BAT > T-PVAT > A-PVAT > WAT; Fig. 1e,f). BAT exhibited the highest median UCP1 intensity (50.22 ± 8.71 AU), followed by T-PVAT (43.83 ± 2.83 AU), A-PVAT (35.39 ± 2.93 AU), and WAT (29.44 ± 0.56 AU). One-way ANOVA confirmed significant differences across depots (p = 0.031), with post hoc testing revealing markedly elevated UCP1 in BAT versus WAT (p < 0.05). This protein-level validation aligns with transcriptional profiling (Fig. 1d) and reinforces observed differences in canonical BAT and WAT and similarities between BAT and T-PVAT.

Together, these data establish a continuum of adipose tissue identities: BAT and T-PVAT cluster morphologically and genetically as thermogenically-poised depots, whereas A-PVAT retains some Ucp1 expression despite aligning more closely with lipid-storing WAT.

Distinct Secretory Profiles of Adipose Depots Reveal Functional Specialization

In addition to tissue morphology, we investigated functional differences in adipose depot-specific paracrine signaling by analyzing secreted proteins from BAT, T-PVAT, A-PVAT, and WAT conditioned media. Quantification of total secreted proteins via BCA assay revealed that BAT secreted significantly higher protein amounts per 5 mg of tissue than all other adipose depots (~1.5-fold increase; Fig. 2a). Notably, T-PVAT, A-PVAT, and WAT exhibited similar levels of total protein secretion. The increased protein concentration in BAT-conditioned media suggests a heightened secretory activity in canonically thermogenic adipose depots.

Fig. 2.

Secretory profiles of adipose depots reveal functional specialization. a Quantification of total secreted protein in conditioned media from adipose tissue cultures (in µg/mL). Results are shown for BAT, T-PVAT, A-PVAT, and WAT depots. n = 11 mice/group. b–d Relative protein expression from 38-plex adipokine antibody array. Results for each protein are normalized to individual membrane positive controls for each adipose conditioned media (see Fig. S3 for representative membrane images). Data has been stratified by functional/biological category and results are shown as b growth factors and metabolic regulators, c pro-inflammatory cytokines and mediators, and d ECM and tissue remodeling factors. n = 3 mice/group; see Fig. S2 for all adipokine array results and Table S1 for a full list of protein names, abbreviations, and categories. All data shown as mean ± SEM. Statistical significance from one-way ANOVA with post-hoc Tukey test for multiple comparisons denoted by *p ≤ 0.05, **p ≤ 0.01, or ***p ≤ 0.001

To further reveal the composition of proteins secreted by each adipose depot, we utilized a 38-plex patterned adipokine array as an unbiased screening tool (see Fig. S2 and S3, Table S1). Using this approach, we identified depot-specific secretory signatures that could be stratified into three functional categories: (1) growth factors and metabolic regulators (Fig. 2b), (2) pro-inflammatory cytokines/mediators (Fig. 2c), and (3) ECM remodeling/tissue structural proteins (Fig. 2d). The brown adipocyte-rich depots (BAT and T-PVAT) preferentially secreted metabolic regulators—including ANGPTL3, FGF1, and RBP4—in a manner that was significantly greater than WAT secretion (p < 0.05, 0.01, 0.001; Fig. 2b). WAT and A-PVAT showed enrichment of various adipogenic and angiogenic factors including IGFBP-3, IGFBP-5, and VEGF. Although the trends are similar (e.g., elevated expression in white adipocyte-rich depots, WAT and A-PVAT), significant increases in secretion were only observed for IGFBP-3 (p < 0.05, 0.001; Fig. 2b).

Potent pro-inflammatory mediators—including CCL2, M-CSF, IL-6 and PTX3—were found to be predominantly enriched in WAT and A-PVAT secretomes (Fig. 2c). In particular, A-PVAT exhibited significant increases in CCL2, IL-6, and PTX3 over both BAT and T-PVAT (p < 0.05, 0.01) whereas only CCL2 secretion by WAT was significantly elevated over thermogenic depots (p < 0.05). This observation is consistent with the anti-inflammatory properties previously associated with brown adipocytes. Several inflammatory mediators—including CRP and ICAM-1—were however found to be elevated in BAT and T-PVAT secretomes over both A-PVAT and WAT (p < 0.05, 0.01, 0.001). It should also be noted that T-PVAT uniquely expressed elevated levels of both RAGE (Fig. 2c) and DPP4 (Fig. 2d) over canonical brown and white adipose depots (p < 0.05, 0.01). Taken together, this data suggests a depot-specific regulation of immune and fibrotic mediators which may provide insight into specific remodeling mechanisms, such as RAGE-DPP4 signaling by T-PVAT. Secretion of factors regulating ECM remodeling and tissue structure were found to distinguish depots more clearly. Namely, WAT and A-PVAT secretomes exhibited elevated levels of PAI-1 and TIMP-1 (albeit not significant), which is consistent with the potential for increased fibrotic remodeling; BAT and T-PVAT both secreted higher levels of the free fatty acid carrier protein Fetuin A (p < 0.05, 0.01, 0.001; Fig. 2d).

Findings from our secretome profiling analysis aligns with the adipose depot identities previously established by morphology and gene expression: BAT and T-PVAT share thermogenic-associated secretion profiles with low levels of inflammation, whereas WAT and A-PVAT converge on inflammatory and fibrotic mediators. The unique enrichment of the RAGE-DPP4 axis in T-PVAT, despite its genetic similarity to BAT, highlights a divergence in adipose-derived cues between depots which may contribute regional variations in fibrotic tissue remodeling [3].

Adipose-Derived Secretomes Differentially Regulate Fibroblast Proliferation and Migration

To assess functional differences in adipose depot-specific secretomes, we exposed fibroblasts to either conditioned media (CM) from BAT, T-PVAT, A-PVAT, or WAT cultures, or low serum tissue culture media controls. Standard in vitro cell proliferation (MTT) and cell migration (scratch wound) mechanobiological assays were performed under different CM conditions to identify how secreted factors from each adipose depot distinctly influence fibroblast behaviors.

To this end, fibroblast proliferation was significantly altered by CM phenotype (p < 0.001, one-way ANOVA). Specifically, CM from thermogenic depots (BAT and T-PVAT) significantly enhanced fibroblast proliferation to a similar degree relative to control (~1.4-fold increase; p < 0.001), whereas A-PVAT and WAT CM showed no difference in proliferation from control (Fig. 3a). Similarly, scratch wound assays revealed depot-specific effects on fibroblast migration. While all groups exhibited significant wound closure over time (p < 0.0001, Two-way ANOVA; Fig. 3c), fibroblasts in WAT CM closed 79.2 ± 5.9% of the wound area by 12 h post-scratch, which was significantly faster than those in BAT CM (55.9 ± 9.6%; p < 0.01) or T-PVAT CM (55.2 ± 9.3%; p < 0.05) (Fig. 3b–d). Wound closure by fibroblasts in WAT CM remained significantly greater than BAT and T-PVAT CM at 15 h post-scratch (Fig. 3b,e). At all time points, migration of fibroblasts in A-PVAT CM mirrored that of control (A-PVAT: 82.98 ± 3.76% vs. control: 83.42 ± 2.77% at t = 15 h; p > 0.05), while BAT CM and T-PVAT CM suppressed fibroblast migration and wound closure to a similar degree, relative to control (BAT vs. T-PVAT: p > 0.99; Fig. 3c-e). Together, these functional assays appear to be consistent with the secretory profiles from each adipose depot (cf. Fig. 2): BAT’s metabolically active factors promote fibroblast proliferation, while WAT’s pro-inflammatory and angiogenic mediators accelerate migration.

Fig. 3.

Adipose secretomes differentially regulates fibroblast proliferation and migration. a Quantitative measurement of fibroblast proliferation after 48 h of growth based on measures of optical density from the MTT assay. Proliferating fibroblasts were cultured in either conditioned media from BAT, T-PVAT, A-PVAT, or WAT or a low serum media (control). b Representative in vitro scratch wound healing assay images at 0, 9 and 12 hours. Wound boundary denoted by magenta lines; scale bar: 1 mm. c Time-dependent quantification of wound closure (%) over 15 hours after scratch placement. n = 4 mice/group. Green stars indicate times with statistical differences between groups. Detailed comparison of percent wound closure at d 12 h and e 15 h post-scratch. All data shown as mean ± SEM. Statistical significance from either a one-way ANOVA or c–e two-way ANOVA with post-hoc Tukey test for multiple comparisons denoted by *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, or ****p ≤ 0.0001

Adipose Secretomes Modulate ECM Remodeling and Myofibroblast Differentiation

To examine adipose-induced structural and phenotypic changes in fibroblasts, we quantified collagen deposition and expression of myofibroblast markers following exposure to CM culture conditions. Picrosirius red (PSR) staining for fibrillar collagens was significantly affected by adipose depot conditioned media (p < 0.01; one-way ANOVA). Further investigation revealed that fibroblasts cultured in BAT CM acquired a more synthetic phenotype and deposited significantly more collagen than fibroblasts cultured in either A-PVAT or WAT CM conditions (p < 0.05; Fig. 4a,b). T-PVAT CM induced a similar trend as BAT CM, albeit not statistically significant, with a ~ 1.2-fold increase in collagen deposition compared to A-PVAT and WAT CM conditions. Notably, both the A-PVAT CM and WAT CM conditions produced collagen levels that were comparable to the low serum control condition (Fig.4a–b).

Fig. 4.

Adipose secretomes modulate ECM remodeling and myofibroblast differentiation. a Representative images of in vitro picrosirius red (PSR) staining showing collagen deposition in fibroblasts cultured in either conditioned media from BAT, T-PVAT, A-PVAT, or WAT or a low serum media (control). b Quantification of collagen deposition by optical density measurements after elution of bound PSR stain. n = 6 mice/group. Quantification of average intensity from c α-smooth muscle actin (αSMA) immunofluorescence and d vinculin immunofluorescence. n = 4 mice/group. e Representative co-stained immunofluorescent images of fibroblasts with αSMA (yellow), vinculin (red), and DAPI (blue) from adipose conditioned media and control groups. All data shown as mean ± SEM. Statistical significance from one-way ANOVA with post-hoc Tukey test for multiple comparisons denoted by *p ≤ 0.05

Immunofluorescence staining and quantitative image analysis (see representative images in Fig. 4e) further identified myofibroblast differentiation and substrate adhesion dynamics in adipose-treated fibroblasts. Consistent with our observations of increased in vitro collagen deposition, fibroblasts cultured in BAT CM exhibited a significantly enhanced αSMA expression—a key cytoskeletal hallmark of myofibroblast activation—compared to A-PVAT CM and WAT CM conditions (p < 0.05: Fig. 4c,e). Again, fibroblasts cultured in T-PVAT CM showed a similar non-significant trend toward increasing αSMA expression over A-PVAT CM and WAT CM as those maintained in BAT CM. Using a similar approach, vinculin staining intensity was used as a proxy for ECM attachment and focal adhesion maturation. In general, vinculin staining showed similar patterns as αSMA—higher expression in BAT and T-PVAT groups compared to A-PVAT and WAT. However, the effect size for vinculin expression was smaller than that of αSMA expression and statistical significance was not achieved between any experimental groups (p > 0.05; Fig. 4d,e).

Together, these findings reveal the existence of adipose-specific regulation of fibroblast function, wherein BAT appears to promote a pro-fibrogenic phenotype characterized by enhanced proliferation and ECM synthesis. While T-PVAT shares genetic and morphological features with BAT, its attenuated browning capacity results in more modest effects. Conversely, WAT and A-PVAT appear to promote a matrix-degradative phenotype despite their pro-migratory behaviors.

Adipose Secretomes Reprogram Fibroblast Mechanobiology and Traction Dynamics

To better understand how distinct adipose depots modulate cell biophysical properties, we first performed micropipette aspiration on fibroblasts following coculture with adipose tissue. Elastic (Young’s modulus) and viscous (apparent viscosity) behaviors were quantified to assess single cell bulk biophysical properties of stiffness and fluid-like behavior. Notably, fibroblasts in each coculture condition exhibited a significant decrease in cell radius when in suspension (p < 0.0001, one-way ANOVA; Fig. S4a). Additionally, fibroblasts cocultured with T-PVAT exhibited a ~1.9-fold increase in Young’s modulus compared to control (565.1 ± 38.7 Pa vs. 304.0 ± 14.2 Pa; p < 0.0001, one-way ANOVA). Indeed, T-PVAT CM induced greater increases in fibroblast stiffness than BAT CM (296.2 ± 30.9 Pa; p < 0.0001), A-PVAT CM (342.9 ± 25.2 Pa; p < 0.0001), and WAT CM conditions (361.1 ± 27.9 Pa; p < 0.0001) (Fig. 5a). When considering viscous cell behaviors, BAT-conditioned fibroblasts displayed elevated viscosity, relative to control (17501 ± 2744 Pa·s vs. 3097 ± 552 Pa·s; p < 0.0001), A-PVAT CM (10429 ± 1905 Pa·s; p < 0.05), and WAT CM (5800 ± 923 Pa·s; p < 0.0001) conditions (Fig. 5b). Together, results from micropipette analysis suggest a transition toward stiffer, more solid-like biophysical properties of fibroblasts treated with media from thermogenic adipose depots (BAT and T-PVAT).

Fig. 5.

Adipose secretomes reprogram fibroblast mechanobiology and traction dynamics. Micropipette aspiration measurements of a bulk stiffness (Young’s Modulus, $E$ in Pa) and b apparent viscosity ($\mu$ in Pa s) of individual fibroblasts following coculture with samples of either BAT, T-PVAT, A-PVAT, or WAT or a low serum media (control). n = 19-74 cells/group with adipose tissue from at least 3 mice. Traction force microscopy allows for quantification of c cell boundaries and d contractile force generation (in nN). Morphological analysis of individual e cell attachment area (in µm²) and f cell shape index (CSI) as a measure of circularity. Following inverse identification of an optimal traction field, g total force generation (in nN) and h strain energy density (in pJ) was calculated for each cell. n = 18-52 cells/group with adipose tissue from at least 3 mice. All data shown as mean ± SEM. Statistical significance from one-way ANOVA with post-hoc Tukey test for multiple comparisons denoted by *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, or ****p ≤ 0.0001

Complementary traction force microscopy (TFM) analysis provided further insights into the adipose-induced changes in single cell mechanobiology. Fibroblasts treated with BAT, T-PVAT, and A-PVAT CM exhibited significant morphological changes when plated on PAA gels characterized by a transition from elongated and spindle-shaped to a more rounded morphology (cf. Fig. 5c). This change in cell shape was captured by measurable increases in cell area (Fig. 5e) and a higher cell shape index (CSI, circularity) (Fig. 5f; p < 0.0001, one-way ANOVA) in the BAT CM group (area: 3964.39 ± 439.58 μm²), compared to both control (area: 1922.38 ± 200.91 μm²) and WAT CM (area: 1675.73 ± 98.42 μm²) groups; PVAT CM (area: T-PVAT: 4317.76 ± 237.13 μm², A-PVAT: 3494.75 ± 429.37 μm²) and WAT CM groups had similar values of CSI as control, on average. Notably, unlike BAT and PVAT CM groups, treatment with WAT CM produced minimal morphological changes in adherent cells, as seen by cell area and CSI measurements that are nearly identical to control (Fig. 5e,f).

TFM revealed the contractile force generation by individual fibroblasts in each experimental condition (Fig. 5d). Indeed, the addition of adipose CM significantly altered the contractile force produced by single cells (control: 15.48 ± 1.78 nN, BAT: 60.72 ± 11.72 nN, T-PVAT: 63.74 ± 6.82, A-PVAT, 72.56 ± 11.60 nN, WAT: 26.80 ± 3.75 nN; p < 0.0001, one-way ANOVA; Fig. 5g). While all CM groups had higher forces than control, on average, only the WAT CM group (p > 0.05) did not reach statistical significance (p < 0.05 for BAT; p < 0.01 for T-PVAT and A-PVAT, respectively; Fig. 5g). No significant correlation was observed between average force generation and average αSMA expression across groups (Fig. S5). Similarly, strain energy density—a measure of the energy stored in the PAA gels due to cell-generated forces—was significantly elevated as a result of adipose secretory profiles (control: 0.0006 ± 0.00009 pJ, BAT: 0.003 ± 0.00059 pJ, T-PVAT: 0.002 ± 0.00022 pJ, A-PVAT: 0.004 ± 0.00081 pJ, WAT: 0.0009 ± 0.00013 pJ; p < 0.0001, one-way ANOVA; Fig. 5h), with BAT, T-PVAT, A-PVAT secretomes again eliciting a more pronounced effect than WAT (p > 0.05), relative to control (p < 0.05 for BAT and T-PVAT; p < 0.001 for A-PVAT, respectively). Given the similar trends in cell area and total force, and prior observations that force generation scales with cell attachment area, we also normalized each contractility measure to cell area. When normalized to cell area, there remained a significant effect of adipose CM on both normalized force (p = 0.0075, one-way ANOVA; Fig.S4b) and normalized strain energy (p = 0.0012, one-way ANOVA; Fig.S4c), however differences between groups were diminished, suggesting a compensatory scaling with cell area.

Collectively, these results demonstrate that secretomes from different adipose depots exert distinct effects on fibroblast mechanobiology. Specifically, BAT and T-PVAT CM robustly enhanced cell stiffness, viscous resistance, and force generation while promoting a rounded morphology. In contrast to the pro-migratory behavior, our observation of mechanical inertness by A-PVAT- and WAT-treated fibroblasts highlights a decoupling of motility from cytoskeletal rigidity, possibly mediated by inflammatory mediators (e.g., CCL2, IL-6), that prioritize matrix degradation over structural integrity. Together, such mechanobiological adaptations may underlie the depot-specific fibroblast functional heterogeneities observed after exposure to adipose-conditioned media.

DPP4 Inhibition Attenuates T-PVAT-Induced Fibroblast Activation

Given our observations of enhanced fibroblast activation and increased RAGE-DPP4 protein secretion by T-PVAT, we investigated if targeted interruption of the DPP4 signaling axis via treatment with sitagliptin could inhibit T-PVAT-induced changes in fibroblast behaviors. Consistent with our prior observations (cf. Fig. 3a), fibroblast proliferation increased by 75.2% in T-PVAT CM versus control media (p < 0.001, two-way ANOVA); however, sitagliptin did not affect the proliferation rate with values remaining 72.7% higher in T-PVAT CM with sitagliptin compared to control with sitagliptin (p < 0.001, two-way ANOVA; Fig. 6a). No significant differences were observed between vehicle and sitagliptin treatments within each media (p > 0.05, two-way ANOVA; Fig. 6a). Collagen deposition, as measured by PSR staining (cf. Fig. 4a,b), rose 1.28-fold in fibroblasts exposed to T-PVAT CM relative to control media (p < 0.05, two-way ANOVA; Fig. 6b). Sitagliptin treatment reduced collagen deposition by 31.5% in T-PVAT CM (T-PVAT + sitagliptin vs. T-PVAT + vehicle, p < 0.01; Fig. 6b), but did not affect collagen levels in control media (control + sitagliptin vs. control + vehicle, p > 0.05). Immunofluorescence staining revealed a 1.24-fold increase in αSMA expression (cf. Fig. 4c) by fibroblasts cultured in T-PVAT CM versus control media (p < 0.01, two-way ANOVA; Fig. 6c). Sitagliptin treatment reduced αSMA intensity by 21.0% in T-PVAT CM (T-PVAT + sitagliptin vs. T-PVAT + vehicle, p < 0.01; Fig. 6c) without affecting αSMA levels in control media (control + sitagliptin vs. control + vehicle, p > 0.05). Vinculin expression exhibited a similar, albeit non-significant, trend as αSMA with an increase in the T-PVAT CM condition and a slight reduction with sitagliptin treatment (p > 0.05, two-way ANOVA; Fig. 6d).

Fig. 6.

DPP4 inhibition attenuates T-PVAT-induced fibrogenic activation. In all experiments, cultured fibroblasts were treated with either T-PVAT conditioned media (T-PVAT) ± sitagliptin (20 nM; DPP4 inhibitor) or low serum media (Ctrl) ± sitagliptin. a Quantitative measurement of fibroblast proliferation over 48 h of growth based on measures of optical density from the MTT assay. b Quantification of collagen deposition by optical density measurements after elution of bound PSR stain. Quantification of average intensity from c α-smooth muscle actin (αSMA) immunofluorescence and d vinculin immunofluorescence. e Representative co-stained immunofluorescent images of αSMA (yellow), vinculin (red), and DAPI (blue) from T-PVAT ± sitagliptin (bottomr row) and Ctrl ± sitagliptin (top row) groups. n = 3 mice/group for all measurements. All data shown as mean ± SEM. Statistical significance from ordinary two-way ANOVA followed by Tukey’s multiple comparisons denoted by *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001

Together, these data demonstrate that DPP4 inhibition by sitagliptin attenuates T-PVAT-induced collagen synthesis and myofibroblast differentiation, thereby validating a functional role for the RAGE–DPP4 signaling axis in PVAT-driven fibrogenic alterations to fibroblast behaviors.

Discussion

Our study demonstrates that adipose tissue exhibits distinct secretory profiles, and that these depot-specific secretomes have profound influences on fibroblast behavior. By integrating morphological, genetic, and mechanobiological analysis, we have identified that thermogenic depots such as BAT and T-PVAT not only enhance fibroblast proliferation and ECM synthesis but also secrete metabolic factors such as ANGPTL3, FGF1 and Fetuin A, which protect against vascular calcification and inflammation [25, 37, 38, 50, 57]. On the other hand, lipid-storing depots such as A-PVAT and WAT preferentially secrete pro-inflammatory cytokines (e.g., CCL2, IL-6) and fibrogenic mediators (e.g., PAI-1, TIMP-1) that promote fibroblast migration and matrix degradation which aligns with their roles in metabolic inflammation and perivascular fibrosis [12, 27, 42]. These compartmentalized secretory profiles underscore the divergent physiological roles of adipose tissues: BAT and T-PVAT support energy dissipation and tissue stabilization during homeostasis, while A-PVAT and WAT drive matrix turnover and tissue remodeling [9, 15, 42, 47, 50].

While many of our findings are consistent with prior observations of canonical adipose depot function, we have also identified a few key discrepancies related to fibroblast mechanobiology. Namely, our findings suggest that secreted factors from BAT and T-PVAT, despite being less inflammatory and more metabolically driven, have a stronger impact on myofibroblast activation than signaling by WAT and A-PVAT depots. Indeed, fibroblasts cultured in conditioned media from thermogenic adipose depots enhance ECM production and decrease cell migration (Fig. 4), increase cell attachment and spreading (Fig. 5), increase intracellular contractile protein expression (Fig. 4), and enhance cell stiffness and biophysical properties (Fig. 5); all of these characteristics are hallmarks of activated fibroblasts and FMT [20, 35, 64]. These data help to provide insight into current contradictory reports regarding adipose regulation of myofibroblast conversion by establishing adipose depot identity as a critical determinant. Namely, some studies suggest adipose-derived factors suppress fibroblast activation [14], whereas others demonstrate pro-fibrotic effects [42]. This depot-specific paradigm explains how adipose tissues can simultaneously exhibit anti-fibrotic (WAT/A-PVAT) and pro-remodeling (BAT/T-PVAT) functions within the same organism.

The discrepancy between secreted proteins and cellular responses from BAT and T-PVAT conditions may suggest additional bioactive factors not identified through our adipokine array contribute to fibroblast heterogeneity. Targeted exogenous protein treatments, rather than complete conditioned media experiments, may be an approach to better define mechanisms underlying thermogenic adipose-induced fibroblast activation. Additionally, unbiased high-throughput proteomics approaches—such as liquid chromatography-mass spectrometry (LC-MS)—could be instructive in identifying more robust compositional differences in adipose tissue secretomes [14, 48]. That said, the uniquely elevated levels of RAGE and DPP4 in T-PVAT may indicate a specialized paracrine axis that exacerbates vascular dysfunction, despite its genetic similarity to BAT. Indeed, our functional validation with sitagliptin treatment to selectively inhibit DPP4 signaling establishes a causal role for this pathway in T-PVAT-driven fibroblast activation. Sitagliptin treatment attenuated collagen deposition and αSMA expression—two key hallmarks of FMT and myofibroblast differentiation—by fibroblasts exposed to T-PVAT conditioned media (p < 0.01; Fig. 6b,c,e), while proliferation remained unchanged (Fig. 6a). These findings are consistent with prior studies implicating RAGE signaling in diabetic vasculopathy [8, 44] and DPP4 signaling in impaired vascular repair [61]. Given that tissues were harvested from healthy mice in the current study, elevated RAGE-DPP4 levels represent a physiological baseline for T-PVAT. Thus, we propose this axis supports vascular homeostasis under normal conditions but may be pathologically amplified in metabolic disease to drive fibrosis through mechanotransductive pathways (e.g., Hippo/YAP-mediated cytoskeletal remodeling [13, 51]).

Critically, our inhibition data provide insight into the apparent paradox between T-PVAT’s BAT-like genetic identity and its functional divergence. The RAGE-DPP4 axis operates as a T-PVAT-specific modulator that decouples ECM synthesis/pro-contractile signaling from proliferative responses (Fig. 6a-d). This compartmentalization suggests therapeutic targeting of DPP4 (e.g., via sitagliptin) could mitigate dysregulated fibroblast stiffening and vascular fibrosis [40, 63] without disrupting BAT-like metabolic repair functions. Future studies should explore whether RAGE-DPP4 amplification underlies regional susceptibility to fibrotic cardiovascular diseases, such as thoracic aortic remodeling in hypertension [2, 3].

The functional divergence among adipose depots becomes apparent when considering their effects on fibroblast mechanobiology. Conditioned media from BAT and T-PVAT robustly enhanced fibroblast proliferation and ECM deposition, which suggests BAT and T-PVAT secretomes reduce inflammatory cytokines while promoting tissue repair and thermogenic niche maintenance [6, 16, 19, 52, 55]. However, their induction of fibroblast stiffening and αSMA expression (Fig. 4-5) suggests a capacity to promote FMT and excess fibrosis if dysregulated. In contrast, A-PVAT and WAT secretomes accelerate fibroblast migration, likely through NF-κB and JAK/STAT pathway activation via pro-inflammatory cytokines, which prioritizes matrix degradation and remodeling over tissue structural integrity [1, 31, 46, 62]. This is consistent with our observations of elevated PAI-1 and TIMP-1 in white adipocyte-rich depots (see Fig. 2d). Functional differences between depots also extend to fibroblast biophysical properties. Even though BAT and T-PVAT secretomes enhance fibroblast stiffness and viscous resistance, mechano-phenotyping further demonstrated that RAGE-DPP4 interactions in T-PVAT may engage novel mechanotransduction pathways to induce excessive fibroblast activation and stiffening [13, 39, 51]. Meanwhile, the decoupling of motility from cytoskeletal rigidity in WAT-conditioned fibroblasts (high migration rates with low stiffness) may indicate pro-inflammatory signaling promotes matrix degradation and turnover over preserving tissue structural integrity. The lack of correlation between average αSMA expression and traction force generation across groups (cf Fig. S5) may suggest myofibroblast differentiation and contractile force generation represent distinct, temporally dynamic cellular processes that are potentially independently regulated by adipose-mediated factors (e.g., cytokines, growth factors, and bioactive molecules, etc.) and depend upon regulatory mechanisms other than αSMA (e.g., RhoA/ROCK, myosin light chain (MLC) phosphorylation, etc.). These heterogeneous adipose functions are not only consistent with the secretory, genetic, and biomechanical profiles of the respective depots but also highlight a complex balance between reparative and pathological processes that regulate soft tissue remodeling and fibrosis.

While our findings provide critical insights into adipose-fibroblast crosstalk, several limitations need to be acknowledged and taken into consideration. First, the use of 3T3 fibroblasts, albeit a widely used cell culture model for studying fibroblast mechanobiology, may not fully recapitulate the behavior of primary fibroblasts in vivo. This becomes particularly relevant in disease-specific contexts such as vascular fibrosis or metabolic syndrome. For example, mechanobiological responses to PVAT secretomes may be different in primary aortic adventitial fibroblasts (same anatomical location in vivo) versus immortalized cell lines such as 3T3s. The use of immortalized cell lines limits our ability to assess patient- or tissue-specific heterogeneity and interspecies differences (e.g., murine adipose vs. human fibroblast responses) remain unexplored. Second, our conditioned media approach is intended to isolate the effects of depot-specific paracrine factors but excludes direct cell-cell contact and bidirectional communication inherent to physical coculture systems. This may potentially underestimate the role of matrix-bound factors or dynamic cellular interactions in shaping fibroblast behavior in vivo. Furthermore, the stability and kinetics of secreted factors in conditioned media were not assessed, limiting insights into whether observed fibroblast responses reflect acute or chronic exposure to adipose-derived signals. Third, secretome profiling of conditioned media from intact adipose tissue cultures aggregates biological signals from adipocytes, stromal vascular cells (e.g., pre-adipocytes, fibroblasts), and immune populations (e.g., macrophages) [11, 27, 28], rather than adipocytes alone. While holistic, this approach may obscure the precise cellular origins of depot-specific signaling factors and mediators. For instance, we see increased RAGE-DPP4 secretion by T-PVAT, but this enrichment could originate from either adipocytes, macrophages, or perivascular stromal cells. Future high resolution (single cell) secretome analyses will be necessary to clarify this cell-specific regulation in more detail. Finally, while our in vitro models enabled precise mechanistic dissection, they lack the systemic complexity of in vivo microenvironments, where neural, hormonal, and hemodynamic inputs may modulate adipose-fibroblast crosstalk. Our study focused on male mice as adipose donors to align with prior work on adipose-vascular crosstalk, but we acknowledge that sex hormones influence adipose biology and fibroblast responses, thus limiting generalizability of the current findings to females. Thoracic and abdominal PVAT were selected due to their distinct phenotypic profiles and established roles in regional vascular inflammation and remodeling [29, 32]. While these depots provide critical insights into adipose heterogeneity, future studies should explore sex-specific responses and additional depots to fully characterize adipose-driven fibrosis. Addressing these limitations through primary cell validation, spatial co-culture systems, and single-cell secretomics will refine our understanding of adipose-driven fibrosis and its potential for therapeutic targeting.

Conclusions

In conclusion, our comprehensive analysis positions adipose depots as key regulators of fibroblast function through spatially organized paracrine networks. Particularly, our findings offer new insights into the adipose-driven regulation of fibroblast-to-myofibroblast transition, a process central to the development of fibrosis. The distinct paracrine signals from BAT, T-PVAT, A-PVAT, and WAT converge to determine fibroblast fate, balancing tissue repair with the potential for pathological fibrosis. The RAGE–DPP4 axis in T-PVAT emerges as a physiologically relevant, therapeutic target governing fibroblast contractility and ECM production. BAT’s pro-fibrogenic effects further reveal its dual role in metabolism and stromal remodeling. These findings highlight the potential for adipose signaling to influence stromal cell behavior, which may contribute to metabolic and cardiovascular comorbidities in contexts where adipocyte-fibroblast crosstalk is dysregulated. Future studies exploring adipose-derived proteins, extracellular vesicles, or metabolites as mediators of stromal reprogramming—particularly in vivo or using patient-derived models—are needed to establish physiological relevance and clarify mechanistic links between adipose heterogeneity and tissue fibrosis. While our work identifies depot-specific fibroblast responses in vitro, the exclusive use of NIH/3T3 cells limits direct translation to human pathophysiology. Further validation in primary fibroblasts, preclinical models, or clinical cohorts will be critical to assess the therapeutic applicability of targeting adipose-fibroblast interactions.

Supplementary Information

Below is the link to the electronic supplementary material.

Matthew R. Bersi

is an Assistant Professor in the Department of Mechanical Engineering & Materials Science with an affiliate faculty appointment in the Department of Biomedical Engineering and the Division of Biology and Biomedical Sciences at Washington University in St. Louis. He received his B.S. from Texas A&M University and Ph.D. from Yale University, both in Biomedical Engineering. He was also a postdoctoral fellow in Biomedical Engineering and a member of the Cardiovascular Research Training Program at Vanderbilt University and Vanderbilt University Medical Center. Dr. Bersi’s interdisciplinary research is centered around experimental and computational approaches for characterizing the multiscale relationship between biomechanics and inflammation in soft tissue remodeling with applications ranging from cardiovascular disease to reproductive biomechanics and women’s health. He is also a recipient of the NIH K99/R00 Pathway to Independence Award (2019), the AHA Career Development Award (2022), and the NSF CAREER Award (2024).

Author Contributions

Individual author contributions to the final manuscript as outlined by the CRediT taxonomy are as follows: Conceptualization: YX, MRB; Data curation: YX, NK; Formal analysis: YX, NK, TAT, BKO, JAE; Funding acquisition: MRB; Investigation: YX, NK, TAT, JAE; Methodology: YX, NK, TAT, JAE, JCS; Project administration: MRB; Resources: MRB; Software: YX, JAE, MRB; Supervision: YX, MRB; Validation: YX, NK, TAT, JAE; Visualization: YX, NK, TAT; Writing—original draft: YX, NK, TAT, JAE, MRB; Writing—review & editing: YX, NK, TAT, BKO, JAE, JCS, MRB.

Funding

This work was supported, in-part, by grants from the National Institutes of Health (R00 HL146951 to MRB and T32 HL170959 to BKO), the American Heart Association (941138 to MRB), and the National Science Foundation (2340666 to MRB).

Data Availability

The original data presented in the study are included in figures throughout the article and supplemental material. All data and analysis tools will be published to the Washington University Research Data Repository at the time of publication. Further reasonable inquiries regarding data access can be directed to the corresponding author.

Declarations

Competing Interests

The authors declare no conflicts of interest, financial or otherwise.

Ethical Approval

All animal procedures were approved by the Washington University in St. Louis Institutional Animal Care and Use Committee (IACUC) and performed in accordance with the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines 2.0.

Consent to Participate

Not applicable.