Chemomechanical remodeling of glucocorticoid-sensitive fibroblasts for multimodal synergistic therapy of skin fibrosis

Summary

Challenges and opportunities in multimodal synergistic therapy for skin fibrosis encompass elucidating the mechanisms of synergistic treatment, optimizing and developing highly coupled combinations, and eliminating therapeutic resistance. This study reveals that the excessively deposited extracellular matrix of hypertrophic scar not only forms a physical barrier for local drug delivery but also generates high mechanical stress, which drives glucocorticoid insensitivity by activating the FAK-AKT-HDAC2 axis in fibroblasts. Both mechanical and biological barriers result in poor outcomes of triamcinolone acetonide therapy for hypertrophic scars. To address this, a chemomechanical antifibrotic approach is engineered by integrating a microneedle-based transdermal delivery platform, immobilized enzymes, and long-acting sustained-release microspheres. This strategy significantly sensitizes scar fibroblasts by disrupting the fibrotic extracellular matrix and the resultant mechanics-induced cellular programs for drug resistance, thus notably reversing the fibrotic characteristics. These findings uncover a mechanism of glucocorticoid resistance and present a multimodal, self-administrable therapy against fibrosis.

Graphical abstract

Highlights

• •ECM stiffening attenuates the therapeutic outcomes of GC therapy in patients with HS
• •ECM stiffness drives glucocorticoid resistance via FAK-AKT-HDAC2 axis
• •Collagenase-mediated ECM mechanical remodeling disrupts physical and biological barriers

Lu et al. reveal that ECM stiffness induces glucocorticoid resistance in hypertrophic scars by activating the FAK-AKT-HDAC2 pathway. They develop a chemomechanical microneedle platform that remodels the fibrotic niche and resensitizes fibroblasts to therapy, effectively reversing scar progression.

Introduction

As evolution advances, the healing response favors the replacement of damaged tissue through the rapid deposition of collagenous connective tissue—a process defined as fibrosis.^1,2 When adult mammalian skin sustains any injury that insult dermis, specific fibroblasts, such as En-1-positive fibroblasts, orchestrate tissue regeneration in a fibrotic repair manner for prioritization of rapid repair and restoration of tissue integrity.^3,4 This leads to pathological scar tissue and ultimately impairs tissue function, thereby diminishing the quality of life for hundreds of millions of individuals globally. Nonetheless, advances in developing antifibrotic or antiscarring therapeutics have been slow and remain inadequate to meet the growing demand in medical healthcare.^5,6 Intralesional injection of glucocorticoid (GC) remains the front-line treatment for active pathological scars due to its capacity to suppress inflammation, fibroblast proliferation, and collagen biosynthesis.^5,7 Yet similar to GC resistance observed in inflammatory diseases, both intrinsic and acquired insensitivity often limit the clinical efficacy of GC in scar treatment, leading to unsatisfactory outcomes.^6,8,9,10 The mechanisms underlying GC sensitivity in scars remain unclear. Understanding how fibrotic progression drives GC resistance is essential for developing innovative therapies and improving clinical management of pathological scars.

Fibrosis is a complex, dynamic process that involves multiple signaling feedback and feedforward loops between cells and extracellular matrix (ECM).¹¹ Recent advances have highlighted the central role of ECM mechanical hallmarks in triggering mechanical signaling pathways in cells that activate cellular programs, resulting in cellular phenotypic transformation, fibrosis aggravation, and chemoresistance.^{12,13,14,15,16} As a common skin fibrosis, a hypertrophic scar (HS) is typical of cellular responses to intrinsic accumulated mechanical stress during healing.¹⁷ Consequently, altered ECM architecture and mechanical abnormalities are both outcomes and causes of HS progression.^18,19 In pathological contexts, the imbalance between the biosynthesis and degradation of ECM causes aberrant tissue remodeling, where fibroblast-producing fibrous collagen replaces fibrin-rich ECM to form a denser and stiffer stroma that acts as a natural physical barrier that limits the diffusion and access of therapeutic agents, potentially leading to a moderate clinical efficacy of GC.^13,17,20,21 However, evidence indicates that GC can counteract the cellular mechanoresponses of fibroblasts by destabilizing the mRNAs of the mechanosensors, suggesting a crosstalk between GC-dependent intracellular signal and mechanotransduction pathways.²² Therefore, the high mechanical stiffness mediated by a dense ECM may in turn play an additional role in driving GC resistance by activating specific cellular programs. However, the precise mechanisms of fibroblast GC insensitivity are unknown.

Given the known direct influence of the mechanical properties of ECM on cellular processes, we hypothesize that fibroblasts within scar tissue develop a GC-resistant phenotype by coupling cell-ECM mechanical signaling and biochemical pathways. Using an engineered tunable three-dimensional (3D) matrix that mimics the biochemical ingredients and mechanical properties of scar tissue, we demonstrated that the increase in the mechanical stiffness of the matrix activated a serine/threonine-protein kinase (AKT)-centered biochemical signaling axis in the fibroblasts cultured within it, thereby attenuating GC sensitivity. This stiffness-dependent mechanism drives cellular programs conferring GC resistance, identifying ECM mechanical cues as therapeutic targets to sensitize scar treatment.

Aberrant ECM remodeling, characterized by collagen type I enrichment, is a major contributor to tissue stiffening during scar progression.^23,24 Modulating the collagen architecture could thus mitigate detrimental mechanical cues for improved therapeutic effect.²⁵ To this end, we developed a chemomechanical antifibrotic platform that remodels the mechanical microenvironment and enhances GC delivery. The system employs collagenase-immobilized microspheres (MPs) co-loaded with GC to overcome both physical (stiff ECM) and biological (GC insensitivity) barriers. These sustained-release MPs were packaged into dissolving microneedle (MN) patches for transdermal delivery. This integrated strategy enhanced MP penetration and restored GC sensitivity by mitigating ECM stiffness and associated cellular resistance, leading to significantly improved outcomes in an animal HS model. Our findings support a chemomechanical approach to antiscarring therapy, demonstrating that targeting ECM mechanics can substantially enhance drug efficacy.

Results

ECM stiffening attenuates the therapeutic outcomes of GC treatment in patients with HS

Mechanical abnormalities in tissues are both outcomes and causes of fibrosis progression. ECM-mediated mechanical cues are widely recognized as triggers for the activation of cellular mechanobiological processes.^26,27 It is unclear if the mechanical stiffness of ECM impacts therapeutic resistance to GC in HS. To determine the correlation between HS stiffness and the therapeutic response to GC, we evaluated 90 patients who received GC therapy for HS. Intralesional injection of triamcinolone acetonide (TA) was performed per month, and multidimensional scar assessments were performed for scar progression and therapeutic outcomes (Figure 1A). In both thoracic and cervical scars, GC treatment outcomes were influenced by the initial hardness value of the scar and not its location. Obvious regression was observed in scars with lower stiffness values, whereas poor efficacy, even scar progression, was exhibited by scars with higher stiffness values (Figure 1B). In silico analysis demonstrated no significant correlation between the pretreatment Vancouver scar scale (VSS) score and initial scar hardness values. However, a significant positive correlation emerged between the post-treatment VSS score and hardness values (Figure 1C). To directly assess therapeutic efficiency, we calculated the difference between the pretreatment VSS score and the score after each treatment cycle was calculated as therapeutic efficiency. The subsequent correlation analysis reveals a significant negative correlation between this treatment efficiency and the initial scar hardness value (Figures S1A and S1B). A Shore hardness of 10 was used as the threshold for classification, and we further stratified the cohort into soft (hardness value: 0–10) and stiff (hardness value: 10–20) scars. A significant decrease in the hardness value and VSS score was revealed in the soft scar that proceeded to every cycle of GC treatment, whereas a slight decrease was observed in the stiff scar until the second therapeutic cycle (Figures 1D and S1C). The relative ratios of stiffness values at posttreatment to the initial value, calculated to reflect therapeutic efficacy, suggested the correlation of reduced GC response and advanced HS stiffness (Figure 1E). Together, these data indicate that high ECM stiffness attenuates the therapeutic outcomes of GC in patients with HS.

Figure 1.

Elevated mechanical stiffness informs poor therapeutic efficacy of glucocorticoid (GC) in patients with HS

(A) Schematic of the clinical course and timeline of GC therapy for patients with established HS.

(B) Representative images and ultrasound records of HS with different hardness values at the thoracic and cervical region, before and after GC treatment, respectively. Scale bars, 5 mm.

(C) Correlation analysis of the Vancouver scar scale (VSS) score and initial hardness value of HS pretreatment and posttreatment in patients from Southwest Hospital (n = 90).

(D) Statistical analysis of hardness values of defined groups for soft and stiff scars before and during therapeutic cycles. A Shore hardness of 10 was used as the threshold for classification, and the cohort was stratified into soft (hardness value: 0–10) (n = 45) and stiff scars (hardness value: 10–20) (n = 45).

(E) Quantification of the relative ratios of the hardness value posttreatment to the initial value in soft (n = 45) and stiff (n = 45) scars.

Statistical analysis was performed using linear regression with 95% confidence intervals, two-tailed Student’s t test, or one-way ANOVA. Results are presented as means ± SD.

Mechanical stiffness drives GC resistance in scar fibroblasts within the 3D matrix

As the primary downstream reactors of fibrosis, fibroblasts, which mainly synthesize and assemble ECM, are the central contributors to scarring progression as well as the key cellular targets of GC treatment.^7,28,29 Consequently, the GC sensitivity of fibroblasts may serve as a proxy for the responsiveness of HSs to GC therapy. Given the dominance of collagen in defining the mechanical characteristics of HS ECM (Figures S2A and S2B), we simulated the biomechanical properties and biochemical composition of ECM in vitro using a 3D collagen matrix.³⁰ The tunable stiffness of this matrix, modulated by the content of collagen type I, was varied to investigate the regulatory effects of mechanical stiffness on the intrinsic GC resistance of the scar fibroblasts. Patient-derived fibroblasts were populated within this 3D matrix, which, while simplifying the pathological microenvironment of scarring, effectively mimicked its mechanical features (Figure 2A). The tuning of a collagen concentration of ∼3 mg/mL (low collagen content) to ∼7 mg/mL (high collagen content) altered the stiffness of the 3D matrix, with the Young’s modulus ranging from ∼120 Pa (soft) to ∼380 Pa (stiff) (Figure S3A). A concentration of 50 μg/mL TA was selected as the optimal dose, ensuring minimal cytotoxicity while inhibiting the expression of fibrosis markers (Figure S3B).

Figure 2.

Sensing ECM stiffness activates the FAK-AKT-HDAC2 signaling axis to drive GC resistance for scar fibroblasts

(A) Schematic overview of the experimental setup for scar fibroblasts cultured within the 3D collagen matrix.

(B) Real-time qPCR analysis of the expression of fibrosis-related genes (ACTA2, COL1A1, and CTGF) in patient-derived fibroblasts cultured in soft and stiff 3D matrices (n = 3).

(C and D) Heatmap for transcriptomic profiles (C) and expression level based on the transcripts per million (TPM) method (D) of genes related to the GC response and fibrosis of fibroblasts grown in the 3D matrix treated with or without triamcinolone acetonide (TA, 50 μg/mL). Analyzed using RNA-seq data (n = 3, biologically independent samples for sequencing). The scale bar shows color-coded differential expression, with red indicating higher levels and blue indicating lower levels of expression.

(E) Expression of genes of GR and GR coregulator of fibroblasts grown in a 3D matrix from RNA-seq (n = 3).

(F) Differences in the global transcriptional landscape based on differentially expressed genes (DEGs) induced by stiffness and TA.

(G) Upregulated and downregulated DEGs in scar fibroblasts grown in stiff versus soft 3D matrix.

(H) Kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis using DEGs of fibroblasts grown in stiff versus soft 3D matrix. Circle size corresponds to the number of genes implicated in each pathway. p values were determined by one-sided Fisher’s exact test.

(I) Gene set enrichment analysis (GSEA) of the transcriptional signature from fibroblasts cultured in stiff versus soft 3D matrix. NES, normalized enrichment score. FDR, false discovery rate.

(J) Immunoblot analysis of phosphorylated FAK (Tyr397), FAK, phosphorylated AKT (Ser473), AKT, and HDAC2 in scar fibroblasts with the indicated treatments.

(K) Representative confocal microscopy images of the distribution of NR3C1 cellular localization in scar fibroblasts grown in soft and stiff matrix with the indicated treatment (n = 10). Red, NR3C1; blue, DAPI; green, F-actin. Scale bars, 100 μm.

(L) Cellular localization analysis of NR3C1 via western blot analysis of nuclear and cytoplasmic cell fractions in cell lysates from scar fibroblasts grown in soft and stiff matrix with the indicated treatment. β-actin and histone H3 proteins are shown as controls for cytosolic and nuclear fractions. C, cytoplasmic fraction; N, nuclear fraction.

(M) Representative confocal microscopy images of COL1A1 and HDAC2 staining of normal skin and scar tissue. Violet, COL1A1; green, HDAC2; blue, nuclear (DAPI). Scale bars, 100 μm (top) and 50 μm (bottom). HOO: Shore hardness value. The normal skin and scar samples used here were derived from patient #3, patient #8, and patient #11, respectively, as indicated in Table S1.

Statistical analysis was performed using the two-tailed Student’s t test or one-way ANOVA. Results are presented as means ± SD.

To rigorously assess the in vitro model for patient-derived fibroblasts exposed to TA simulation within the ECM mimic with varying stiffness, we initially compared the transcriptional profiles of fibrosis-related genes in both soft and stiff 3D matrices. Increased matrix deposition corresponded to elevated expression of myofibroblast markers, including α-smooth muscle actin (ACTA2), collagen type I alpha 1 chain (COL1A1), and connective tissue growth factor (CTGF, also known as CCN2) (Figure 2B).²⁴ The findings confirmed that ECM stiffness serves as a critical signal for fibroblasts to initiate a fibrotic response. In the soft matrix, the expression of these genes was pharmacologically suppressed by TA, an effect that was markedly attenuated in the stiff matrix. Taking the percentage of downregulation in the soft matrix as a reference, GC sensitivity was dramatically diminished due to increased stiffness. To further dissect the impact of mechanical stiffness on the TA-induced global transcriptional landscape, we performed RNA sequencing (RNA-seq) of scar fibroblasts within a 3D matrix of varying stiffness treated with or without TA. Both changes in the mechanics and TA treatment led to significantly different transcriptomic profiles (Figures S4A and S4B).

Volcano plots showed the most prominently expressed features according to the empirical Bayes method (fold change >1; false discovery rate adjusted [FDR] < 0.05) (Figures S4C–S4E). The gene-based analysis revealed two distinct clusters related to GC response and fibrosis. A heatmap demonstrated the upregulation of genes associated with GC response, such as FKBP5,³¹ KLF9,³² LPL,³³ and GLI1,³⁴ and the downregulation of genes associated with fibrosis, such as POSTN,²⁸ IL6, THBS1, and ITGA1,¹⁴ after GC treatment. Compared to fibroblasts cultured in the soft matrix, the differences in expression between the control and GC-treated groups were less pronounced than fibroblasts cultured in the stiff matrix (Figure 2C). The quantification of expression level according to the transcripts per million reads method verified this trend (Figure 2D). Collectively, these data suggest that mechanical cues derived by ECM under mechanical stress shift fibroblasts toward an increased GC-resistant phenotype.

Mechanical signaling crosstalk with GC resistance in scar fibroblasts through the activation of the focal adhesion kinase-AKT-HDAC2 signaling axis

Lipophilic GC exerts its functions by binding to the GC receptor (GR), which is a transcription factor encoded by NR3C1. Upon ligand binding, GR undergoes a conformational change, partially dissociates from the chaperone complex, translocates to the nucleus, and interacts with the genome.^35,36 Concurrently, GR recruits a coregulator, which convert the signal-driven allosteric transitions into context-specific transcription regulatory actions, modulating the chromatin structure to either activate or repress gene transcription.^9,36 Given the global differences in the transcriptional program of fibroblasts within varying stiffness matrices in response to GC treatment, we hypothesize that the mechanism of GC resistance, stemming from crosstalk with mechanical cues, can be attributed to universal defects in GC signaling, such as reduced GR expression, impeded nuclear translocation, or modified coregulator expression, rather than specific alterations in downstream signaling.

To identify factors that mediate the attenuation of GC signaling at the transcriptional level, we screened a cluster of genes encoding GR and its coregulators, such as NCOA1, HDAC1, HDAC2, and CARM1,^35,36 from RNA-seq data. Notably, HDAC2, encoding histone deacetylase 2, was among the top downregulated genes in response to mechanical stiffening and exhibited an inverse trend between the soft and stiff matrix groups compared with other coregulators (Figure 2E). HDAC2 was found to mediate GC resistance in the alveolar macrophages of patients with chronic obstructive pulmonary disease,³⁷ indicating a shared mechanism of GC resistance in fibroblasts and immune cells induced by mechanical abnormality and inflammation, respectively. To determine how the mechanical signaling cascades with biochemical signaling counteract the GC effect, further analysis of differentially expressed genes (DEGs) in the absence of GC revealed that many fibrosis-related genes, such as HIF1A, fibronectin (FN1), and CCL2,¹⁹ and mechanical-transmission-related genes, such as ITGA2, TNC,¹⁴ and DPP4,⁴ were upregulated in cells cultured in the stiff matrix, whereas matrix-degradation-related genes, such as LOX,³⁸ DKK1,³⁹ and TIMP3,⁴⁰ were downregulated (Figures 2F and 2G). Gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) analysis were performed to functionally categorize the enriched DEGs (Figures 2H, S5A, and S5B). GO analysis identified upregulated cell adhesion/ECM organization features and downregulated cell-cycle/chromosome organization features in fibroblasts grown in the stiff matrix (Figure S5B). KEGG analysis identified upregulated ECM-receptor interactions and the classical pathway of mechanical sensing (focal adhesion/Hippo signaling pathway) as well as the PI3K-AKT signaling pathway and the downregulated pathways related to cell-cycle arrest in cells within the stiff matrix (Figure 2H). PI3K-AKT was described as a coactivated signaling pathway downstream of focal adhesion kinase (FAK).⁴¹ Gene set enrichment analysis (GSEA) validated that genes associated with focal adhesion and the AKT pathway were markedly enriched in fibroblasts cultured in the stiff matrix relative to those cultured in the soft matrix (Figure 2I). Overactivation of the AKT pathway results in reduced expression of HDAC2 and impaired nuclear translocation of GR.^42,43 Collectively, these data indicate that activation of the AKT pathway facilitates a signaling conversion from mechanical cues to biochemical transduction and plays a central role in driving GC resistance.

We investigated the molecular signaling cascade mediated by the AKT pathway in scar fibroblasts. The results shown in Figures 2J and S6A demonstrate that mechanical stimuli derived from ECM stiffening activate the FAK-AKT signaling pathway. Specifically, stiff matrix conditions promoted the phosphorylation of the mechanosensor FAK at tyrosine 397. And this FAK activation triggered downstream phosphorylation of AKT at serine 473, an event that could be completely abolished by the AKT inhibitor MK2206. To elucidate how the AKT pathway conveys mechanical FAK signaling to suppress the GC response, we also examined the expression of the GR, GR coactivator HDAC2 downstream of AKT, and the fibrosis marker α-SMA. Elevated stiffness significantly repressed HDAC2 expression but exerted no obvious effect on the protein level of GR (Figures 2J and S6A). The disruption of AKT signaling with MK2206 markedly relieved the stiffness-induced inhibition of HDAC2 upon TA treatment. In the stiff matrix, a combination of MK2206 with TA treatment increased the inhibition of both mRNA and protein expression of fibrotic markers, including α-SMA, COL1A1, DPP4, and fibronectin, (Figures S6B and S6C). These data suggest that AKT activation functions in a signaling cascade to promote HDAC2-mediated GC resistance. To further explore the role of AKT signaling in the impairment of GR cytoplasmic-nuclear shuttling, we assessed the nuclear and cytoplasmic distribution of NR3C1 under distinct mechanical stress (Figure 2K). Immunofluorescence revealed effective nuclear relocalization upon TA treatment in the soft matrix, which was inhibited in the stiff matrix. The inhibition of AKT signaling with MK2206 restored the nuclear translocation of NR3C1 following TA treatment (Figure S7), indicating that AKT signaling also reduced GR activity by hindering the GC-induced cytoplasmic-nuclear translocation. Similar results were obtained via western blot analysis of nuclear and cytoplasmic cell fractions in scar fibroblasts, in which increased stiffness impeded the effective cytoplasmic-nuclear shuttling and inhibition of AKT with MK2206 increased the nuclear translocation of the GR following TA treatment (Figure 2L). This effect was further confirmed in scar fibroblasts cultured on a 2D monolayer with the addition of an AKT agonist, SC79 (Figures S8A and S8B). Moreover, we examined the deposition of collagen type I and expression of HDAC2 in soft and stiff scar tissue from patients with HS (Figures 2M and S2). Increased scar stiffness is accompanied by excess deposition of collagen type I and reduced expression of HDAC2, which further confirmed that collagen-type-I-dominated ECM stiffening contributed to GC disability by inhibiting the expression of GR coregulator.

Collectively, these findings demonstrate that increased collagen deposition results in elevated mechanical stress, more cell-matrix interactions, and focal adhesion. The activation of the FAK-AKT-HDAC2 axis attenuates the GR activity of scar fibroblasts, and the AKT pathway performs as a coupling mechanism between mechanical FAK signaling and GC-resistant programs (Figure 3).

Figure 3.

Schematic diagram of mechanical-stiffness-mediated GC resistance

Altering ECM composition and organization are identified as a hallmark of scar. Collagen-dominated mechanical stiffness represents a major impediment to effective medication. At the tissue level, dense ECM acts as a physical barrier that limits the access of therapeutic agents. At the molecular level, the sensing of direct cell-ECM signaling arising from mechanical cues regulates cellular phenotypes, leading to aberrant fibrosis expansion and drug resistance. Increased collagen deposition results in more cell-matrix interactions and focal adhesion. And subsequent FAK-mediated mechanical signaling activates the biochemical AKT pathway, which downregulates the expression of the GR coregulator HDAC2 and simultaneously impedes the nuclear translocation of GR, leading to the impairment of GC efficacy.

Rational design of a chemomechanical platform for synergistic scar management

Dense ECM acts as a physical barrier that limits drug penetration at the tissue level,^44,45 while cell-ECM mechanical signaling regulates cellular phenotypes, promoting fibrosis and drug resistance at the molecular level.^11,46 Since tissue stiffness largely stems from collagen type I deposition,⁴⁷ we hypothesized that disrupting collagen architecture could sensitize HS therapy by normalizing fibrotic niches and blocking mechanosignaling. Collagenases, which selectively degrade collagen fibers and unwind triple helices,²⁵ are promising candidates for modulating fibrotic mechanical cues. To test whether remodeling the mechanical microenvironment alongside GC delivery could overcome ECM-imposed physical and biological barriers, we developed COLA1@PLGA-TA—a multifunctional system composed of PLGA-based microspheres (MPs) for sustained TA release, surface-decorated with collagenase type I (COLA1) for ECM remodeling (Figure 4A). PLGA-TA MPs were prepared via emulsion-solvent evaporation, and COLA1 was immobilized onto PLGA via EDC/NHS chemistry.^48,49 The structure of chemical coupling via amide bonds was confirmed at peaks of 1,660 and 3,100 cm⁻¹ using Fourier transform infrared spectra (Figure S9A). ¹H-nuclear magnetic resonance (¹H-NMR) spectroscopy showed the emergence of characteristic peaks in the aromatic region (7–7.5 ppm), which are attributable to the aromatic amino acid residues presented in the collagenase,⁵⁰ and a significant enhancement of the broad amide proton signal (7.5–8.5 ppm), which is a marker for the peptide bonds within the protein structure (Figure S9B).⁵¹ COLA1@PLGA-TA MPs exhibited a rougher surface and homogeneous nitrogen distribution (Figure 4B), with a slightly larger diameter (14.2 ± 5.7 μm) than PLGA-TA MPs (13.1 ± 4.3 μm; Figure 4C). Each mg of COLA1@PLGA-TA contained 44.8 ± 2.16 μg TA and 10.44 ± 0.18 U active collagenase (Figure 4D). Immobilized collagenase retained >92% activity after 30 days at 4°C, compared to ∼52% at room temperature (Figure 4E). PLGA-TA MPs showed sustained TA release in vitro: ∼14% at 24 h, ∼40% at 14 days, and ∼50% at 35 days (Figures 4Fand 4G). To overcome the skin barrier, functional MPs were further fabricated into a dissolved MN patch for transdermal drug delivery and release (Figure 4H). Referring to our previous study,⁵² the designed MNs were cast based on polyvinyl pyrrolidone and shaped as a pyramid, with a height of 1,000 μm and side length of 300 μm at the base, as an array of 15 × 15 in 1 cm² (Figure S9C). To facilitate visualization, MPs, which were decorated with fluorescein isothiocyanate (FITC)-labeled collagenase and laden with fluorescent Nile red instead of TA, were used as cargoes (Figures 4I and S9D). A total of 1.1 mg MPs was loaded into each MN patch after two steps of centrifugation (Figure 4J). The PLGA MP-loaded MN patch withstood a much higher compressive force (>0.65 N) than the blank ones (Figure 4K), indicating that the addition of MPs elevated the mechanical strength of the MN patch. To test the capacity of transdermal delivery, the dissolved MNs, which were applied on the human scar, displayed prominent degradation within 2 h, with a degradation rate of 100% (Figures 4L and 4M). Cargoes of MNs were completely transmitted into the tissues after detaching the backing layer at 2 h after penetration (Figure S9E).

Figure 4.

Physicochemical characterizations of the designed synergistic platform for scar therapy

(A) Schematic of engineered COLA1@PLGA-TA microspheres (MPs).

(B) Scanning electron microscopy (SEM) images of PLGA-TA and COLA1@PLGA-TA MPs for surface microstructure and element analysis. Red, carbon (C); green, oxygen (O); blue, nitrogen (N). Scale bars, 10 μm.

(C) Particle size distribution of PLGA-TA and COLA1@PLGA-TA MPs (n ≥ 12). Scale bars, 20 μm.

(D) Quantification of the TA- and COLA1-loading capacity of COLA1@PLGA-TA MPs (n = 3).

(E) Long-term tests of bound enzyme activity at room temperature (RT) and 4°C over 30 days (n = 3).

(F and G) Kinetics of cumulative release of TA from PLGA MPs over 36 days (F) and within 24 h (G) (n = 3).

(H) Schematic of the fabrication process of COLA1@PLGA-TA-loaded microneedles (MNs) arrayed patch via mold casting.

(I) Digital photograph and fluorescence images of PVP MNs patch packaging PLGA MPs. MPs encapsulating Nile red and surface modified with FITC-labeled COLA1. Scale bars, 500 and 200 μm (enlarged images).

(J) PLGA MPs loading (mg/patch) of COLA1@PLGA-TA MNs prepared via two centrifugations (n = 3).

(K) Compression tests of blank MNs and MPs loaded with PVP MNs.

(L) Morphological change of the MN patch before and after insertion into isolated human HS at the indicated time points. Scale bars, 1 mm.

(M) Simulated value of a single microneedle and the degradation rate from 3D scanning and reconstruction (n = 12). Results are presented as means ± SD.

Engineered MN platform breaks through both physical and biological barriers

To in vitro simulate the process of drug delivery, artificial cell-free 3D matrix based on rat tail collagen type I was used to model the biochemical components and mechanical properties of scar ECM. The Nile-red-loaded MPs were sprinkled on the top surface of the collagen hydrogel, penetrating downward under gravity. Confocal microscopy revealed that after 24 h, COLA1-decorated PLGA MPs displayed a more spatial distribution than enzyme-free PLGA MPs (Figure 5A). Scanning electron microscopy images showed that PLGA MPs were mainly dispersed on the surface of the hydrogel with a compact morphology, whereas COLA1@PLGA MPs were visible in the deep layers of the hydrogel with a looser structure (Figure 5B). These results suggest that collagenase immobilization enables PLGA MPs to diffuse beyond the original space dimension by modulating the physical barrier. Similarly, when delivered via MNs, COLA1@PLGA MPs spread beyond the application site and disrupted collagen architecture, whereas PLGA MPs remained confined (Figures 5C and S10A). Rheology confirmed that COLA1@PLGA MN treatment reduced storage modulus (<100 Pa) compared to PLGA MN treatment (>300 Pa), indicating mechanical softening (Figure S10B). To investigate the access of therapeutic TA and the following cellular response, primary scar fibroblasts were cultured in a monolayer and covered by a collagen hydrogel to segregate with the superstratum (Figure 5D). COLA1@PLGA-TA MNs significantly reduced cell viability compared to controls or monotherapies, while empty or PLGA-only MNs showed no effect (Figure 5E). Confocal imaging confirmed that enzymatic enhancement promoted MP penetration and stronger suppression of cell proliferation (Figure 5F). These results suggest that integration with collagenase enhances the drug delivery of TA by interfering with the topological and mechanical structure of the matrices.

Figure 5.

Integration of ECM mechanical remodeling and pharmacotherapy sensitization

(A) Schematic and representative confocal microscopy images of the spatial distribution of Nile-red-loaded PLGA and COLA1@PLGA MPs, respectively. Red, Nile red-labeled MPs. Scale bars, 200 μm.

(B) Representative SEM images of collagen matrix morphology and the spatial distribution of PLGA and COLA1@PLGA MPs. Yellow line indicates the boundary of the surface and inner part of the collagen matrix. Scale bars, 500 and 100 μm (enlarged images).

(C) Schematic and representative confocal microscopy images of drug diffusion mediated by MN patches packaged with Nile-red-loaded PLGA and COLA1@PLGA MPs. Scale bars, 500 μm.

(D) Schematic showing evaluation of drug access and cellular response for collagen hydrogel covering scar fibroblasts.

(E) Quantification of cell viability of fibroblasts with the indicated treatments (n = 9).

(F) Representative confocal microscopy images of cell density and MP access to cellular layer after treatment with MNs loaded with the indicated MPs. Blue, DAPI; red, F-actin; orange, MPs. Scale bars, 200 and 50 μm (enlarged images).

(G) Schematic of scar fibroblasts grown in a 3D matrix and treated with the indicated MN patch.

(H and I) Representative images of collagen matrix deformation mediated by fibroblast-generated contraction force (H) and quantification of matrix contraction indexes (CI) (n = 3) (I). Blue circle indicates the initial size of the collagen matrix. Yellow contour indicates the deformed collagen matrix at the end of the experiment.

(J) Immunoblot analysis of phosphorylated FAK (tyr397), FAK, phosphorylated AKT (Ser473), AKT, HDAC2, and α-SMA in scar fibroblasts in the 3D matrix with the indicated treatments.

Statistical analysis was performed using one-way ANOVA. Results are presented as means ± SD.

To elucidate whether the integrated MN system could disrupt the biological barrier of GC insensitivity, we examined the cellular response and molecular pathway variation of cells grown in the stiff 3D matrix (Figure 5G). Scar contracture is the most common scar comorbidity and is largely caused by mechanical force transmission between activated fibroblasts and ECM.^5,14,53 Using a contraction assay, we observed that COLA1@PLGA-loaded MNs attenuated the cell-generated contractile force and the ability to remodel the surrounding ECM environment of activated fibroblasts (Figure 5H). Compared to normal fibroblasts with a contracture index (CI) of ∼0.01, HS-derived fibroblasts showed a drastic contraction response to mechanical stress with a CI of ∼0.91, which was inhibited by both TA (CI = 0.67) and COLA1 monotherapies (CI = 0.78). Nevertheless, combination therapy potentiated a stronger effect (CI = 0.49) than monotherapy. The empty (CI = 0.92) and PLGA-loaded MNs (CI = 0.90) showed no significant inhibition on fibroblasts contraction (Figure 5I). Immunoblotting was performed to measure the markers of the GC resistance program and fibrosis, all of which were more activated in scar fibroblasts than in normal fibroblasts (Figure 5J). The introduction of COLA1 reduced FAK activation and suppressed the phosphorylation of downstream AKT, thereby restoring HDAC2 protein levels. Notably, the AKT signal directing the GC resistance pathway was reversed by COLA1 orchestrating mechanical remodeling. The integration of enzyme-induced mechanical remodeling and TA-induced action exhibited a more pronounced downregulation of fibrosis-related α-SMA than either monotherapy. Altogether, these findings demonstrate that MNs loading the COLA1@PLGA-TA system break through the physical barrier to facilitate drug delivery and prime a low-stress microenvironment to restore the therapeutic vulnerability of TA, eventually supporting the inhibition of fibroblast-mediated fibrosis.

Chemomechanical COLAI@PLGA-TA system accelerates HS regression in vivo

To evaluate the therapeutic effects of combined ECM modulating and enhanced TA delivery, we examined in vivo antifibrosis efficacy by adopting a rabbit ear HS model, which shares more pathological similarity with that of humans.^52,54 The modeling and therapeutic course are illustrated in Figure 6A. Circular full-thickness wounds with a diameter of 1 cm were created on the ventral side of the rabbit ear, and scar modeling lasted for 28 days. The establishment of HS was validated by the formation of a local lesion with a raised red surface characterized by excessive and disordered collagen deposition (Figures S11A and S11B). After application of MN patches to HS, MN tracts penetrated the scar epidermis and extended into the dermis, reaching a depth of approximately 710–742 μm (Figures S11C and S11D). Empty and MN patches carrying MPs of PLGA, PLGA-TA, COLA1@PLGA, or COLA1@PLGA-TA were applied on the formed HS on days 30 and 45. Scar thickness, blood flow, color, and mechanical properties were assessed. After each therapeutic cycle, the gross appearance and blood flow of HS exhibited an obvious difference in the groups receiving different treatments. The mechanical properties of the scar were characterized by measuring Shore hardness (HOO) and ultimate tensile strength. Compared to untreated control, empty and PLGA-loaded MNs exerted no therapeutic effect on thickness, blood perfusion, and mechanical hardness of HS, indicating that these blank carrier systems alone do not produce a therapeutic effect (Figures S12A–S12D). COLA1@PLGA-TA-loaded MNs induced more accelerative effects on both scar regression and blockade of blood perfusion compared with when TA or COLA1 was used alone (Figure 6B). After 30 days, HS thickness in the COLA1@PLGA-TA group decreased by 0.56 mm (∼18.62% reduction), while untreated scars thickened (∼14.37% increase). PLGA-TA and COLA1@PLGA groups showed similar reductions (∼10.10% and ∼10.25%, respectively; Figure 6C). A robust network of capillaries is often present in immature HS, which corresponded to a significant increase of blood flow signals in the untreated scar compared with normal skin.⁵⁵ PLGA-TA, COLA1@PLGA, and COLA1@PLGA-TA treatments showed benefits on the decline of blood perfusion. In particular, the inhibitory effects of both zootherapies were inferior to that of COLA1@PLGA-TA treatment, which resulted in a closest perfusion area and speckle intensity to the normal level (Figures 6B and 6D). Scar color also improved: COLA1@PLGA-TA-treated scars showed lightness (L: 69.22) and redness (A: 3.33) values close to normal skin (L: 69.96, A: 3.87), unlike untreated scars (L: 62.43, A: 12.44; Figure 6E). The COLA1@PLGA-TA-treated scar showed a more compliant hardness (HOO: 12.33) than that of PLGA-TA (HOO: 17.43), COLA1@PLGA (HOO: 15.33), and untreated HS (HOO: 22.83) (Figure 6F). Consistently, the untreated scar displayed the lowest maximum strain (<25%) with the ultimate tensile strength of 6.45 MPa, compared with the maximum strain of ∼41% and the ultimate tensile strength of 13.07 MPa for normal skin (Figure 6G). After the treatments, the ultimate tensile stress in the COLA1@PLGA-TA group reached 9.81 MPa, which was ∼1.5 times of untreated HS and ∼1.3 times of either monotherapy. This result indicated that HS tissues’ extensibility and ability to resist external stress were significantly improved by COLA1@PLGA-TA treatment. These data confirm that COLA1@PLGA-TA MNs synergistically reduce ECM stiffness and enhance drug efficacy, offering a promising HS treatment. Histopathological analysis of major organs revealed no abnormalities, supporting the biosafety of COLA1@PLGA-TA MPs (Figure S13).

Figure 6.

In situ administration of the chemomechanical therapy platform promoted recovery from HS in vivo

(A) Experimental schematic and timeline of the synergistic strategy for the injury-induced HS model on the rabbit ear.

(B) Representative digital and blood flux images of normal skin and scar with varied treatments at the indicated therapeutic time. Color scales indicate the blood flow signal. Scale bars, 5 mm.

(C) Statistical data for thickness changes and reduction rate of scar after the second MN therapy from (A) (n = 6).

(D) Quantification of blood flow of HS from (A) (n = 3).

(E) Color measurement of scar with indicated treatments after second MN therapy (n = 3). L, lightness value. R, redness value.

(F) Statistical data for hardness values of HS after varied treatments (n = 3).

(G) Tensile stress-strain curves and ultimate tensile strength in HS samples from different groups after varied treatments (n = 3).

Statistical analysis was performed using one-way ANOVA. Results are presented as means ± SD.

Synergistic therapy disrupts fibrotic niches

A characteristic pathological abnormality of HS, defined as fibrotic niches, comprises abundant stress-fiber-rich fibroblasts, accumulated fibrous ECM, and profibrotic signaling factors.^11,12 To investigate the therapeutic efficacy of multifunctional MNs at the microlevel, we evaluated the histological features of fibrotic niches in tissues harvested from rabbits that received two cycles of treatments. H&E staining showed thinner dermis and flatter epidermis in the COLA1@PLGA-TA group (Figure 7A). The scar elevation index (SEI) of untreated scar was >3 times that of normal skin. Compared to the untreated scar (SEI: ∼3.34), COLA1@PLGA group (SEI: ∼2.25), and PLGA-TA (SEI: ∼1.84), SEI decreased dramatically to a minimum of 1.04 after COLA1@PLGA-TA MN treatment, even close to the normal level (Figures 7A and 7B). Somatic cell infiltration, ∼10-fold higher in untreated HS, was most reduced by COLA1@PLGA-TA (Figure 7B). Masson trichrome staining showed collagen area decreased to 39.17% with COLA1@PLGA-TA versus 75.01% in controls and ∼60% in monotherapy groups (Figures 7A and 7B). These results indicate that synergistic therapy alleviated the recruitment and infiltration of proliferative fibroblasts, with the immoderate production and alignment of collagen. To further analyze the collagen architecture of scar tissue in detail, the algorithms CT-FIRE, CurveAlign, and MatFiber were utilized.^26,56 Collagen parameters were quantified and sequentially analyzed based on images of picrosirius red staining. There was a much higher ratio of type I/III collagen in control HS tissue (∼6.41) than in normal skin (∼1.07), which was prominently decreased benefited from collagenase cascading with TA (∼1.57) (Figures 7C and 7D). By contrast, monotherapy with collagenase (∼3.47) and TA (∼5.15) exerted inadequate effects (Figure 7D). The untreated scar exhibited an obvious disorganization of the dermal architecture across metrics with increased unidirectional alignment and collagen fiber elongation. By contrast, treatment with COA1@PLGA-TA promoted a more randomly aligned collagen in the dermis layer, which was similar to the representative network of basket-weave-like collagen fibers in normal skin (Figure 7C). The lengths of the collagen fibers decreased significantly after treatment with both COLA1@PLGA and COLA1@PLGA-TA, indicating that collagenase plays a key role in increasing the number of shorter collagen fibers and decreasing overall architectural complexity (Figure 7D). Activation of distinct molecular mechanisms, including cell adhesion, mechanotransduction, and desmoplastic reactions, contributed to the expansion of fibrosis niches.²⁹ The expression of vinculin and Ras homolog family member A (RhoA), which act as a mechanical sensor and transducer, respectively, was analyzed.⁵³ Immunohistochemistry showed strong mechanical signaling activation in HS (p-FAK/FAK, RhoA, vinculin), attenuated by all treatments but most effectively by COLA1@PLGA and COLA1@PLGA-TA (Figures 7E and S14A). The administration of MNs loaded with COLA1@PLGA and COLA@PLGA MPs significantly decreased the accumulation of p-FAK-, RhoA-, and vinculin-positive cells compared with that upon TA monotherapy (Figures 7D and S14B). This result highlighted the potency of collagenase in disrupting the cell-ECM mechanical communication. In the control scar, abundant cells, which expressed myofibroblast-activation-related transforming growth factor β1 (TGF-β1) and myofibroblast marker α-SMA, were distributed in the whole dermis (Figure 7E). Consistently, compared with COLA1 monotherapy and TA monotherapy, synergistic therapy was most efficient in reducing TGF-β1 and α-SMA outcomes, even close to normal levels (Figure 7F). These findings demonstrated that integration of mechanotherapy and pharmacotherapy enhanced therapeutic efficacy for reshaping fibrotic niches, thereby promoting fibrosis regression.

Figure 7.

MN-mediated synergistic therapy disrupts fibrotic niches by remodeling the ECM and inhibiting the profibrotic response

(A) Representative H&E and Masson trichrome staining images of normal skin and scar tissue sections after second treatments. Scale bars, 3 and 200 μm (enlarged images).

(B) Quantitative analysis of the scar elevation index (SEI), somatic cell infiltration from H&E-stained images, and the percentage of collagen from Masson trichrome staining images of HS tissue (n = 5).

(C) Representative picrosirius red staining images and collagen organization analysis by CurveAlign (alignment) and CT-Fire (length) for normal skin and scar tissue. Scale bars, 50 μm.

(D) Statistical data of the ratio of type I/III collagen and quantification of collagen alignment and fiber length from (C) (n = 5).

(E) Immunohistochemistry staining images for phosphorylated FAK, α-SMA, and TGF-β1 in normal skin and scar tissue. Scale bars, 100 μm.

(F) Quantitative analysis of integrated optical density (IOD) of phosphorylated FAK, α-SMA, and TGF-β1. The positive signal (optical density) was semiquantitatively analyzed in five areas throughout tissue sections by measuring average brownness, with optical density = log [maximum intensity/mean intensity] (n = 5).

Statistical analysis was performed using one-way ANOVA. Results are presented as means ± SD.

Discussion

Fibrosis, characterized by the accumulation of fibroblasts and aberrant deposition of fibrous connective tissue, represents the final outcome of injury in nearly every human organ.¹² As the largest and most superficial organ, the skin is especially prone to damage and abnormal scarring.^3,26 There is no optimal method to fully restore skin function. Despite the advances in clinical management, such as pressure or compression therapy, chemotherapy, laser therapy, and radiotherapy, HS exhibits high therapeutic disability and recurrence.^6,57 GC remains the only therapeutic drug in well-accepted, evidence-based, and recommended treatments for HS⁵⁸; however, its clinical efficacy is often exhausted due to intrinsic or acquired resistance.^7,59 In long-term clinical practice, we have learned that the efficacy of GC is inversely correlated with the mechanical stiffness of scar tissues, which is widely recognized as pivotal triggers in HS formation and progression.¹ However, molecular evidence is lacking regarding whether mechanical signaling stimulation impairs sensitivity toward GC in scars. We found that a stiffer, collagen-rich matrix directly interacts with fibroblasts via cell-ECM pathways to reduce GR activity and induce GC resistance. In patient-derived fibroblasts, FAK-mediated mechanical signaling activates the AKT pathway, which downregulates the GR coregulator HDAC2 and impairs GR nuclear translocation, compromising GC efficacy. This resistance was reversed by blocking stiffness-induced AKT signaling, suggesting that targeting mechanical-biochemical crosstalk may resensitize GC-resistant scars. These findings enhance our understanding of how ECM-mediated mechanical cues affect drug response in HS and other fibrotic diseases.

Altering ECM composition and organization is one of the hallmarks of fibrotic diseases as well as solid tumors, and elevated mechanical stiffness represents a major impediment to effective medication.^13,16 Recent studies show that mechanical stress is not merely a physical barrier to drug delivery but also a regulator of cellular phenotypes that promote disease progression and drug resistance.^60,61,62 In cancer, mechanical cues induce chemoresistance through mechanisms such as altered drug efflux, DNA repair, and apoptosis.^13,46 Similarly, ECM-integrin-FAK signaling can drive resistance by upregulating antiapoptotic molecules or modulating drug transporters.^63,64 These mechanisms highlight the broader relevance of ECM-mediated signaling beyond cancer, including in cutaneous fibrosis, where ECM stiffness both results from and promotes fibroblast activation post-wounding.^19,65 This insight underlies a promising avenue for mechanical therapeutic theories and strategies that combine the targeting modulation of cell-ECM interactions with therapeutic agents to improve patient outcomes.

Motivated by the importance of considering the impact of ECM mechanical signaling when developing antifibrosis strategies, we sought for synergistic therapy that could disrupt the resistant mechanism for potentiating the drug response. Therapeutic interventions to directly disrupt mechanical cues in fibrosis niches and indirectly block signal transduction with mechanical-pathway-focused inhibitors are promising approaches for mechanotherapeutics.^21,45 However, in cancer therapy, modest survival benefits have been achieved when chemotherapeutics are combined with mechanical-signaling-targeted inhibitors, with only a few exceptions advancing outcomes.^13,66,67,68 Given that the formed ECM acts as a physical barrier to limit access of all cell-dependent drugs, an ECM-targeted therapeutic strategy may be more effective for cutaneous scars, with a much denser and stiffer ECM. In our study, collagen type I in the ECM was identified as a common therapeutic target to remodel the tissue mechanical microenvironment, enhance fibroblasts sensitivity toward GC, and facilitate spatiotemporal drug delivery. Consequently, we introduce a multimodal synergistic therapy that addresses the physical barriers of the ECM and subsequent biological resistance to GC treatment in HS. Through integrated construction, mechanical and pharmacological therapies were highly coupled to form a chemomechanical antifibrotic scheme. This comprehensive strategy circumvents the shortcomings of conventional HS care, such as secondary injuries, ulcers, poor adherence, and high clinician dependency, offering promising home-based self-management for patients with HS.^7,69,70

Our study provides both clinical and mechanistical evidence, which increased matrix stiffness as a key factor impairing GC efficacy in scar treatment. These results significantly expand the known biological roles of mechanical forces, revealing their capacity to modulate cellular responses to chemical signals and dictating therapeutic outcomes. However, in a broader clinical context, therapeutic outcomes are modulated by a confluence of parameters, such as scar area, depth, and anatomical location.^1,5 The in vitro model isolated the variable of matrix stiffness to establish a clear causal relationship. This reductionist approach was crucial for mechanistic insight but does not fully capture the integrated clinical reality. Therefore, our findings position elevated stiffness as a critical and previously underappreciated contributor to GC resistance, which likely interacts with other patient- and scar-specific factors. Future studies that integrate multi-parameter clinical data with sophisticated in vitro models incorporating additional variables will be essential to build a more holistic predictive model for treatment response and to develop personalized therapeutic strategies for complex scar management.^23,30,62

In conclusion, our study elucidates the mechanisms of GC insensitivity in HS and presents a multimodal synergistic approach to HS therapy that addresses the physical and biological barriers to effective treatment. To sensitize drug therapy for improving clinical outcomes in fibrotic diseases, the chemomechanical strategy holds promise for highly coupled combination therapies that target the source of altered ECM properties. Future research may be required to translate these findings into clinical practice and fully comprehend the long-term implications of this treatment modality.

Limitations of the study

Some limitations necessitate further exploration. Notably, this study is confined to a rabbit ear model, which despite its relevance, does not fully replicate the complexity of human skin fibrosis. Future studies should include clinical trials to validate the safety and efficacy of our treatment modality in human participants. In addition, although we have demonstrated the potential of collagen-targeted degradation for enhanced drug penetration and reduced mechanical stress, the long-term effects of these interventions on skin integrity and function remain unclear. Studies should be conducted to assess the durability of the therapeutic effects and potential side effects over time.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Qing Zhang (zhangqing09@tmmu.edu.cn).

Materials availability

Further information and request for materials should be directed to and will be fulfilled by the lead contact.

Data and code availability

• •Raw data for transcriptomic analysis were deposited under National Center for Biotechnology Information (NCBI) with accession numbers BioProject PRJNA1203251. All data are publicly available (https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA1203251).
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this work is available from the lead contact upon request.

Acknowledgments

This work was financially supported by the National Key R&D Program of China (grant no. 2021YFA1101100), National Natural Science Foundation of China (grant nos. 82530085, 82072188, and 82002036), the State Key Laboratory of Trauma and Chemical Poisoning (grant no. 2025K002), and Natural Science Foundation of Chongqing (grant no. CSTB2025NSCQ-GPX0584), and some of the illustrations were created with reference to pictures in BioRender.com.

Author contributions

Conceptualization, Y.L., J.T., Q.Z., and G.L.; methodology, Y.L., P.Y., W.G., Y.W., L.R., L.S., J.L., N.Z., L.H., Q.Z., and G.L.; investigation, Y.L., L.S. Y.Q., L.R., D.X., M.A., Y.Z., X.Z., and J.T.; writing – original draft, Y.L. and Q.Z.; writing – review & editing, J.Z., J.T., Q.Z., and G.L.; funding acquisition, H.L., J.T., Q.Z., and G.L.; resources, W.G., Y.W., D.X., M.A., X.Z., D.H., and H.L.; supervision, J.Z., J.T., Q.Z., and G.L.

Declaration of interests

The authors declare no competing interests.

STAR★METHODS

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti-β-Actin	Cell Signaling Technology	Cat# 4970; RRID: AB_2223172
anti-phospho-FAK (Tyr397)	Cell Signaling Technology	Cat# 3283; RRID: AB_2173659
anti-total FAK	Cell Signaling Technology	Cat# 3285; RRID: AB_2269034
anti-phospho-AKT (Ser473)	Cell Signaling Technology	Cat# 4060; RRID: AB_2315049
anti-total AKT	Cell Signaling Technology	Cat# 4691; RRID: AB_915783
anti-Glucocorticoid Receptor/NR3C1	Cell Signaling Technology	Cat# 12041; RRID: AB_2631286
anti-Histone H3	Cell Signaling Technology	Cat #9715; RRID: AB_331563
anti-α-Smooth Muscle Actin (α-SMA)	Cell Signaling Technology	Cat# 19245; RRID: AB_2734735
anti-YAP	Cell Signaling Technology	Cat# 14074; RRID: AB_2650491
anti-Fibronectin/FN1	Cell Signaling Technology	Cat# 26836; RRID: AB_2924220
anti-HDAC2	Abcam	Cat# ab32117; RRID: AB_732777
anti-RhoA	Cell Signaling Technology	Cat# 2117; RRID: AB_10693922
anti-Vinculin	Cell Signaling Technology	Cat# 13901; RRID: AB_2728768
anti-COL1A1	Cell Signaling Technology	Cat# 72026; RRID: AB_2904565
Anti-DPP4	Cell Signaling Technology	Cat# 67138; RRID: AB_2728750
Anti-rabbit IgG, HRP-linked Antibody	Cell Signaling Technology	Cat# 7074; RRID: AB_2099233
Alexa Fluor 594-conjugated anti-rabbit IgG	Invitrogen	Cat# A32740; RRID: AB_2762824
Alexa Fluor 488-conjugated anti-rabbit IgG	Invitrogen	Cat# A11034; RRID: AB_2576217
Alexa Fluor 647-conjugated anti-rabbit IgG	Invitrogen	Cat# A21245; RRID: AB_2535813
Biological samples
Human normal skin	Institute of Burn Research, Southwest Hospital of Army Medical University	N/A
Human scar tissues	Institute of Burn Research, Southwest Hospital of Army Medical University	N/A
Chemicals, peptides, and recombinant proteins
Triamcinolone acetonide	Kunming Jida	Cat# H53021604
Collagenase type I	Sigma-Aldrich	Cat# SCR103
Dulbecco’s modified Eagle’s medium	HyClone	Cat# SH30243.01
Fetal Bovine Serum	Gibco	Cat# 10091-148
Penicillin–streptomycin	HyClone	Cat# SV30010
PBS without calcium, magnesium	HyClone	Cat# SH30256.01
0.25% trypsin/EDTA	HyClone	Cat# SH30042.01
Telo-collagen type I	This paper	N/A
MK2206	Selleckchem	Cat# S1078
SC79	Selleckchem	Cat# S7863
FITC-conjugated phalloidin	Thermo Fisher Scientific	Cat# F432
Rhodamine-conjugated phalloidin	Thermo Fisher Scientific	Cat# R415
PLGA	MedChemExpress	Cat# HY-B2247A
Polyvinyl Alcohol	Sigma-Aldrich	Cat# 341584
NaN3	Sigma-Aldrich	Cat# S2002
Polyvinyl Pyrrolidone K90	Sigma-Aldrich	Cat# 81440
Polydimethylsiloxane Sylgard 184	Dow	Cat# DKN-184
Nile Red	Sigma-Aldrich	Cat# 72485
FITC	Sigma-Aldrich	Cat# 46950
Protease Inhibitor Cocktail	Sigma-Aldrich	Cat# P8340
Phosphatase Inhibitor Cocktail	Roche	Cat# 04906837001
Triton X-100	Beyotime	Cat# P0096
Critical commercial assays
Total RNA Extraction Reagent, Trizol	Invitrogen	Cat# 15596018CN
PrimeScript RT Reagent Kit with gDNA Eraser	Takara	Cat# RR047A
TB Green Premix Ex Taq II	Takara	Cat# RR820A
Antifade Mounting Medium containing DAPI	Beyotime	Cat# P0131
Enhanced Immunohistochemical Detection Kit	ZSGB-Bio	Cat# SP-9000
NE-PER Nuclear and Cytoplasmic Extraction Reagents	Thermo Fisher Scientific	Cat# 78835
Cell Counting Kit-8	MedChemExpress	Cat# HY-K0301
Super Signal West Pico Kit	Pierce	Cat# 34577
Tissue Protein Extraction Reagent	CWBIO	Cat# CW0891M
Hematoxylin-Eosin (HE) Stain Kit	Solarbio	Cat# G1120
Modified Masson’s Trichrome Stain Kit	Solarbio	Cat# G1346
Modified Sirius Red Stain Kit	Solarbio	Cat# G1472
Deposited data
Bulk RNA-seq raw data	This paper	PRJNA1203251
Experimental models: Cell lines
Human normal fibroblasts	This paper	N/A
Human scar fibroblasts	This paper	N/A
Experimental models: Organisms/strains
New Zealand rabbits	Animal Center of Army Medical University	N/A
Oligonucleotides
hRPL13A-F: GCCATCGTGGCTAAACAGGTA	BGI Genomics	N/A
hRPL13A-R: GTTGGTGTTCATCCGCTTGC	BGI Genomics	N/A
hACTA2-F: AAAAGACAGCTACGTGGGTGA	BGI Genomics	N/A
hACTA2-R: GCCATGTTCTATCGGGTACTTC	BGI Genomics	N/A
hCOL1A1-F: GAGGGCCAAGACGAAGACATC	BGI Genomics	N/A
hCOL1A1-R: CAGATCACGTCATCGCACAAC	BGI Genomics	N/A
hCTGF-F: CAGCATGGACGTTCGTCTG	BGI Genomics	N/A
hCTGF-R: AACCACGGTTTGGTCCTTGG	BGI Genomics	N/A
hTGF-β1-F: CTAATGGTGGAAACCCACAACG	BGI Genomics	N/A
hTGF-β1-R: TATCGCCAGGAATTGTTGCTG	BGI Genomics	N/A
hFN1-F: CGGTGGCTGTCAGTCAAAG	BGI Genomics	N/A
hFN1-R: AAACCTCGGCTTCCTCCATAA	BGI Genomics	N/A
hDPP4-F: TACAAAAGTGACATGCCTCAGTT	BGI Genomics	N/A
hDPP4-R: TGTGTAGAGTATAGAGGGGCAGA	BGI Genomics	N/A
Software and algorithms
R	R Core Team	version 4.1.3 http://www.r-project.org/; RRID: SCR_001905
Graphpad prism	GraphPad Software	version 8.0.2 https://www.graphpad.com/scientificsoftware/prism/; RRID: SCR_002798
BioRender	BioRender	https://BioRender.com/l98x498
ImageJ	National Institutes of Health	https://ImageJ.net/ij/download.html; RRID: SCR_003070
SPSS	IBM SPSS Statistics	version 23.0.0 https://www.ibm.com/cn-zh/spss; RRID: SCR_002865
CurveAlign	The University of Wisconsin-Madison research labs of Drs. Kevin Eliceiri and Abhishek Kumar	version 5.0. https://loci.wisc.edu/ctfire/
CT-FIRE	The University of Wisconsin-Madison research labs of Drs. Kevin Eliceiri and Abhishek Kumar	version 2.0. https://loci.wisc.edu/curvealign/
MATLAB Compiler Runtime (MCR)	MathWorks	version R2014b MCR https://www.mathworks.com/products/compiler/mcr.html

Experimental model and study participant details

Patient characteristics and human biospecimen collection

A total of 90 patients (male: 31, female: 59) diagnosed with postburn hypertrophic scarring and receiving intralesional GC therapy from Dec 31, 2020, to Apr 11, 2024, at the Institute of Burn Research, Southwest Hospital of Army Medical University, were retrospectively analyzed. Clinical information was collected from the patients’ medical records, including age, sex, injury source, lesion location, clinical course, and scar assessment. Patients with HS for <6 months since wound healing were included. Totally, 0.5–1 mL of TA (40 mg) was percutaneously injected into the scar monthly. Multidimensional scar assessment was performed before the initial administration and after each treatment session using the VSS score, digital images, ultrasound images (Mindray, China), and a hardness tester (Shore, Landtek Instrument, China). A Shore hardness of 10 was used as the threshold for classification, and we further stratified the cohort into soft (hardness value: 0–10, n = 45) and stiff (hardness value: 10–20, n = 45) scars. Gender was determined by self-report or physician report. Statistical analysis indicated that baseline scar hardness and GC treatment outcomes showed no significant correlation with patient age (p = 0.9187) or gender (p = 0.5057) within this cohort, as assessed by linear regression. Tissue specimens of the scar were obtained from eight patients with HS who underwent cicatrectomy, and residually normal skin was obtained from four age- and sex-matched patients who received skin or flap grafting at Institute of Burn Research, Southwest Hospital of Army Medical University. For pathological staining, fresh tissues were fixed with 4% paraformaldehyde and stored at 4°C until sectioning. For primary cell culture, fresh tissues were processed into manual dissociation and digestion for cell isolation. Information on clinical human normal skin and scar samples is listed in Table S1. During this study, approval was granted from the Human Medical Ethics Committees of Southwest Hospital under protocol number KY2020012. Written informed consent was obtained from all participants, and no personal information was disclosed. All procedures were performed according to the principles of the Declaration of Helsinki.

Animal studies of fibrosis

The application of COLA1@PLGA-TA was investigated in the rabbit ear HS model, which was established according to our previously reported method.⁵⁰ Healthy New Zealand rabbits (female, 2 kg, 24 weeks) were purchased from the Animal Center of Army Medical University and housed under controlled environmental conditions with a temperature range of 20°C–26°C, a relative humidity of 30–70%, and a 12-h light-dark cycle. All experimental procedures involving animals were approved by the Laboratory Animal Welfare and Ethics Committee of Army Medical University (AMUWEC20192101). The rabbits were anesthetized with 1% pentobarbital (30 mg/kg). Full-thickness skin of equal size at four sites on the shaved ventral side of the rabbit ear was inflicted with wounds using a sterile 1-cm diameter dermal biopsy punch. The perichondrium was completely removed using a scalpel. In 4 weeks, the wounds were healed and HSs were established. To validate the rabbit ear HS model, random scar tissue was collected for histopathological examination.

Cell isolation and culture

Primary dermal fibroblasts were isolated from fresh tissue specimens of HSs (scar fibroblasts) and healthy skin (normal fibroblasts). The tissue pieces were enzymatically digested with 0.2% (w/v) collagenase type I in 0.25% trypsin/EDTA for 20 min at 37°C. The digested products were filtered using a 100-mm nylon filter and centrifuged at 1,200 rpm for 5 min. The dissociated pellet was resuspended in Dulbecco’s modified Eagle’s medium containing 10% (v/v) fetal bovine serum and 100 U/mL penicillin–streptomycin at 37°C in 5% CO₂.The primary cell lines were found free of mycoplasma contamination and were authenticated by short tandem repeat (STR) profiling. The cells were expanded to a density of 1 × 10⁴ cells cm⁻² for all experiments. Cell viability was measured using Cell Counting Kit-8 after incubation for ∼2 h at OD450 using a SpectraMax M5 microplate reader (Molecular Devices, USA).

Fibroblasts cultured in the 3D collagen matrix

The 3D matrix of collagen type I was prepared as a fibroblast-populated collagen lattice system, referring to our previous description.⁵⁰ In brief, telo-collagen type I was extracted from the rat tail with 5% (v/v) acetic acid, sedimented using 10% (w/v) NaCl, and redissolved in 0.1 M HCL. Collagen solution was neutralized using 1 M NaOH on ice until a pH of 7–7.5 was achieved, and the concentration of collagen was quantified by lyophilization. Collagen was mixed with phosphate buffer saline (PBS) to achieve the desired collagen concentration. A total of 200 μL of cell suspensions with 5 × 10⁵ cells was thoroughly mixed with 800 μL collagen solution on ice and transferred into a 6-well non-TC plate. After incubation for 30 min at 37°C to achieve gelation, isovolumetric medium was added and incubated for 48 h without disturbance to allow the cells to adhere to the collagen scaffolds. The following experiments proceeded with the replacement of fresh medium, which was correspondingly supplemented with TA (50 μg/mL), with or without the AKT inhibitor MK2206 (500 nmol/L) or the AKT agonist SC79 (4 μg/mL).

Method details

RNA isolation and real-time qPCR analysis

For extracting RNA from cells cultured in collagen matrix, the cell-encapsulated collagen matrix was centrifuged at 2000 rpm for 5 min to discard the extra liquid. Total RNA extraction reagent was added with an additional vortex for 5 min, and the cells were further lysed using Tissue Cell-destroyer (AI31000, Abclonal, China). Subsequently, 1 μg of RNA was used to transcribe mRNA to cDNA following the standard protocols of PrimeScript RT Reagent Kit with gDNA Eraser. RT-qPCR experiments were performed using TB Green Premix Ex Taq II in CFX Connect Real-Time PCR Detection System (Bio-Rad, USA). The expression of target genes was normalized using ribosomal protein L13a (RPL13a) and analyzed by calculating threshold values (Ct) and fold changes relative to the control using the 2^−ΔΔCt method. The primers and their corresponding sequences used for real-time qPCR are listed in key resources table.

Western blot analysis

Cells in the collagen matrix were lysed in tissue protein extraction reagent supplemented with Protease Inhibitor Cocktail and a phosphatase inhibitor cocktail, according to the protocol for tissue lysis. The cytoplasmic and nuclear extracts of cultured cell were separated and prepared using NE-PER Nuclear and Cytoplasmic Extraction Reagents. Equal amounts of protein were loaded on a Bis–Tris-buffered gel, separated by MOPS–SDS-PAGE, and transferred to a PVDF membrane (Bio-Rad, USA). The membrane was incubated with the appropriate primary antibody and antirabbit HRP-conjugated secondary antibodies (1:2000). The following antibodies were used: β-actin (1:2000), Histone H3 (1:1000), Phospho-FAK (Tyr397) (1:1000), total FAK (1:1000), Phospho-AKT (Ser473) (1:1000), total-AKT (1:1000), NR3C1 (1:1000), α-SMA (1:1000), DPP4 (1:1000), Fronectin (1:1000), and HDAC2 (1:2000). Blots were visualized using Super Signal West Pico Kit. Quantification of protein bands was performed using the ImageJ software.

RNA-Seq

RNA-Seq was performed on an Illumina NovaSeq X Plus platform (Illumina, USA) with a read length of 2 × 150 bp. After adaptor trimming and quality control using fastp,⁷¹ the reads were mapped to the Ensembl GRCh38.p13 human reference using HISAT2.⁷² The mapped reads of each sample were assembled using StringTie.⁷³ To identify DEGs between samples, the expression level of each transcript was calculated according to the TPM reads method. Differential expression analysis was performed using DESeq2. DEGs with |log2FC| ≥ 1 and FDR <0.05 were identified as significant. Functional enrichment analysis was conducted to determine DEGs that were significantly enriched in GO terms and KEGG pathways compared with the whole-transcriptome background. GO functional enrichment and KEGG pathway analysis were carried out by Goatools and Python scipy software, respectively. GSEA was performed using GSEA v. 3.0. Statistical significance was evaluated using 10,000 random permutations of the gene set with a signal-to-noise metric for ranking genes. An FDR of <0.05 was considered significant. Raw data for transcriptomic analysis were deposited under National Center for Biotechnology Information (NCBI) BioProject PRJNA1203251.

Histological analysis and immunofluorescence staining

Cells cultured in the collagen matrix were fixed with 4% PFA for 30 min and washed with PBS three times. For fixed skin tissue, samples from patients #1–12 (Table S1) and animal tissue were embedded in paraffin for sectioning into 5-μm thick sections. The sections were stained with hematoxylin-eosin (H&E), Masson’s trichrome, and picrosirius red after deparaffinization and rehydration. To retrieve the epitopes, deparaffinized sections were processed for heat-induced antigen repair in 10 mM citrate buffer (pH 6.0) for 10 min at 98°C.

Cells in the collagen matrix or sections were blocked with 10% goat serum for 1 h to block nonspecific binding after permeation with 0.3% Triton X-100 for 10 min; incubated overnight at 4°C with primary antibodies, including anti-NR3C1 (1:200), anti-COL1A1 (1:500), anti-HDAC2 (1:500), anti-Phospho-FAK (1:100), anti-total FAK (1:100), anti-vinculin (1:100), anti-TGF-β1 (1:100), anti-α-SMA (1:100), and anti-RhoA (1:500), washed with PBS; and incubated with secondary antibodies, including goat Alexa Fluor 594-conjugated IgG (1:2000), goat Alexa Fluor 488-conjugated IgG (1:2000) and goat Alexa Fluor 647-conjugated IgG (1:2000), for 60 min at ambient temperature. The sections were mounted with antifade mounting medium containing DAPI after an intense washing step. Immunohistochemistry was performed using an enhanced immunohistochemical detection kit (ZSGB-Bio, China) following the manufacturer’s protocol. The filamentous actin cytoskeleton was visualized with FITC or rhodamine-conjugated phalloidin (1:250). All histological sections were visualized using a virtual slide microscope (Olympus VS200, Japan), and cell images were obtained using a confocal laser scanning microscope (CLSM, Zeiss, Germany). The images were analyzed and quantified using ImageJ. SEI, which is the ratio of total tissue thickness of scar tissue compared with that of normal tissue above the cartilage surface, was calculated to quantify the degree of scarring as follows:

$$SEI=\frac{T_{\text{scar}}}{T_{\text{normal}\text{skin}}}$$

Where T_scar represents the maximum thickness of scar tissue, and T_normal represents the maximum thickness of the normal dermis around scar tissue, which were measured from the top point of the epithelium to the surface of the cartilage in the scar and normal tissue, respectively. Polarizer accessories were used to obtain picrosirius red images, and fiber alignment was analyzed using CurveAlign and MatFiber.⁷⁴ The degree of alignment ranged from 0 (completely randomized fiber alignment) to 1 (completely aligned fibers). The lengths of individual collagen fibers were quantified using CT-FIRE.

Rheological characterizations of the 3D matrix

Rheological characterization of the stiff (high collagen content, 7 mg/mL) and soft (low collagen content, 3 mg/mL) collagen matrix was performed on a Discovery HR-20 rheometer (TA Instrument, USA) using a 40-mm cone plate. Oscillatory rheology in a time-dependent manner was used to measure storage modulus (G′) and loss modulus (G″) after achieving equilibrium. Rheological tests were performed under an oscillatory time-sweep for 60 s at 25°C (10 s/pt, 0.5% strain) at a constant frequency rate (10 rad/s). The matrix was kept hydrated, and gas bubbles were removed to avoid rheological artifacts during the tests.

Preparation and characterization of PLGA-TA MPs

TA-encapsulated PLGA MPs (PLGA-TA) were prepared according to the oil/water double-emulsion–solvent evaporation method,⁴⁸ as described with some modifications. Briefly, TA was completely codissolved in free carboxyl terminal PLGA (50: 50) in methylene chloride under ultrasound at a concentration of 5% (w/w). The mixture (oil phase) was added to five times the volume of 2% (w/v) polyvinyl alcohol (MW = 80 kDa) solution (water phase) and then emulsified for 1 min under ultrasound to form an oil/water suspension. Next, the emulsion was quickly transferred to 10 times the volume of 0.5% polyvinyl alcohol (w/v) solution, and methylene chloride was allowed to completely evaporate by stirring at room temperature for 12 h. Hardened MPs were collected by filtering against a screen (100 mesh) and thoroughly washed with deionized water. The morphology and particle size distribution of the prepared PLGA-TA MPs were characterized using a scanning electron microscope (Philips, Netherlands). LC-20AD high-performance liquid chromatography (Shimadzu, Japan) was used to determine whether TA was loaded onto PLGA-TA MPs. In detail, 5 mg lyophilized PLGA MPs were dissolved in 20-mL acetonitrile. The resulting solution was filtered (0.22 μm) and analyzed via high-performance liquid chromatography with a mobile phase of 60:40 v/v (methanol: water) at a flow rate of 0.5 mL/min. The prepared samples and standards in acetonitrile or PBS were injected onto a C18 column (Ultimate Plus-C18, 5 μg, 4.6 × 150 mm) maintained at 30°C. TA was detected at 240 nm, and drug loading was calculated according to the ratio of the mass of TA to PLGA-TA MPs.

Drug-release behavior of TA

The drug-release behavior of PLGA-TA MPs in vitro was measured in PBS containing 0.02% Tween-80 and 0.05% NaN₃. The lyophilized MPs (∼50 mg) were submerged in 30-mL Dulbecco’s modified Eagle’s medium, with mild shaking at 37°C throughout this experiment. At each time point (2, 4, 6, and 8 h, and 1, 3, 5, 7, 14, 28, and 35 days), the MPs were separated by centrifugation at 4000 rpm (3200 × g) for 5 min, and 100 μL of the supernatant was collected. The concentration of the cumulative released TA was determined via high-performance liquid chromatography as described above.

Preparation and characterization of collagenase-decorated PLGA-TA MPs

Collagenase was immobilized on the exposed free carboxyl of the PLGA polymer via 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC)/N-Hydroxysuccinimide (NHS) chemistry.⁴⁹ In brief, PLGA-TA MPs (10 mg/mL) were suspended in 2-morpholinoethanesulfonic acid buffer (50 mM, pH 6.0) and hydrated for 2 h. Next, 16.4 mg of EDC and 9.8 mg of NHS were added to 5 mL of PLGA-TA suspension and stirred for 4 h at 25°C. Then, 100 mg of collagenase type I was conjugated to activated PLGA at 25°C for 24 h. The products were washed with deionized water and lyophilized to obtain collagenase-decorated PLGA-TA (COLA1@PLGA-TA). COLA1@PLGA-TA MPs were identified by amido linkage using an FTIR-8300 series spectrometer (Shimadzu, Japan) and verified through elemental distribution (C/N/O) using the energy-spectrum scanning function of scanning electron microscopy. An Agilent 400MR spectrometer (Agilent Technologies, USA) was used to obtain ¹H-NMR spectra. The content of immobilized enzyme was calibrated with the enzyme activity unit (U) using a collagenase activity assay kit (ab196999, Abcam, USA), according to the manufacturer’s procedure. The spatiotemporal curve of changes in enzyme activity under different preservation conditions (4°C and room temperature) were measured according to the same method.

Fabrication and characterization of the COLA1@PLGA-TA-encapsulated MN patch

The MPs were blended with 10% polyvinyl pyrrolidone solution (PVP K90, 1300 kDa), and the mixture was cast into a plasma-pretreated PDMS MN mold. Centrifugation at 2000 rpm for 20 min allowed the MPs to enter the tapered microcavities, and excess solution on the surface was scraped away. Then, 20% polyvinyl alcohol was added and incubated at 4°C for 24 h to form the backing layer. The MN patches were demolded and dried at 60°C for 24 h in vacuum oven. The prepared multifunctional MN patches were stored at 4°C, and sterilized with epoxymethanol (55°C and 1 h) before use. To calculate the loading capacity of the MN patch for the MPs, three pieces of the MN patch were redissolved in deionized water, centrifuged to separate the MPs, and lyophilized to weight. To observe the distribution of MPs in MNs, Nile red-loaded PLGA MPs were decorated with FITC-labeled collagenase. The mechanical strength of the MN patches was tested by pressing a stainless-steel plate against the MNs on an MTS E44 universal tester with a 100 N compression load cell, as in our previous work. The blank MN patches without loaded MPs was used as control. The simulated in vivo degradation of the prepared MNs was performed in surgically excised human HS. MN patches were applied to the human scar tissue and removed at the designed time points (10, 60, and 120 min). The changes in the morphology and volume of MN were recorded using a 3D digital microscope (Leica DVM6, Germany).

Enzyme-action for in situ delivery

Fluorescein-labeled ColA1@PLGA MPs were suspended in PBS at a concentration of 1 mg/mL or encapsulated in an MN patch. Both the suspension and MN formulations were applied on the stiff collagen matrix. After the reaction at 37°C for 24 h, in situ drug delivery and penetration capacity of the MPs in the collagen hydrogel were observed using a confocal laser scanning microscope. The morphological architecture and mechanical viscoelasticity of artificial ECM were evaluated via scanning electron microscopy imaging and rheological tests, respectively.

Matrix deformation mediated by cell contraction

The scar fibroblast-populated 3D collagen scaffold was created following the protocol described above.⁵⁰ After an initial preculture period (48 h), empty and MN patches loaded with PLGA, COLA1@PLGA, PLGA-TA, or COLA1@PLGA-TA were subjected to collagen scaffolds and remained for >5 days until the diameter of the matrix no longer changed. Cell viability was measured using Cell Counting Kit-8 after incubation for ∼2 h at OD450 using a SpectraMax M5 microplate reader (Molecular Devices, USA). Diameters were quantified using ImageJ, and the contraction index was calculated as follows:

$$CI=1-{\left(\frac{A}{A_0}\right)}^2$$

Where A represents the quantified area of the collagen matrix at the end of the experiment and A₀ represents the initial area of the collagen matrix.

Evaluation of in vivo synergistic therapy for HS

Modeled rabbits were randomized into five groups. Empty and MN patches loaded with cargoes (PLGA, COLA1@PLGA, PLGA-TA, and COLA1@PLGA-TA) were applied to penetrate HS tissues and maintained in situ overnight. The second cycle of therapy was administered 15 days after initial therapy. Scar assessments, including those of thickness, hardness, blood perfusion, and color, were performed using Vernier calipers, Shore durometer, moorFLPI-2 Laser speckle contrast imaging system (Moor Instruments, UK), and MiniScan XE Plus spectrocolorimeter (HunterLab, USA). Scars from uninjured dermis (normal skin) and scars in the axial axis (tip to posterior of rabbit ear) were collected on day 30 for mechanical tests and pathological staining. The biomechanical property of HS tissue was tested with tensile failure testing. The specimens were shaped to a dumbbell using a scalpel. Specimen length, thickness, and width were measured manually using Vernier calipers (0.01 mm). The ends of the dumbbell-shaped geometry were wrapped in gauze to ensure that the specimens were fixed by clamps and vertically secured in an MTS machine (Meitesi Testing Technology, China). The specimens were stretched at a constant rate of 1 mm/s until failure, at which ultimate tensile strength was recorded.

Quantification and statistical analysis

Data are presented as means ± SD of at least ≥3 independent measurements. Statistical analysis was performed by linear regression with 95% confidence intervals, two-tailed Student’s t test, or one-way Analysis of Variance (ANOVA) using SPSS 23.0. Tukey’s post hoc test was used for multiple post hoc comparisons to determine statistical significance after one-way ANOVA. p < 0.05 was considered statistically significant. Graph analysis was fabricated using GraphPad Prism 8.0.

Published: February 17, 2026