Nanomaterial-Mediated Reprogramming of Macrophages to Inhibit Refractory Muscle Fibrosis

Abstract

Orofacial muscles are particularly prone to refractory fibrosis after injury, leading to a negative effect on patient’s quality of life and limited therapeutic options. Gaining insights into innate inflammatory response-fibrogenesis homeostasis can aid in the development of new therapeutic strategies for muscle fibrosis. In this study, we identify the crucial role of macrophages in the regulation of orofacial muscle fibrogenesis after injury. Hypothesizing that orchestrating macrophage polarization and functions will be beneficial for fibrosis treatment, we engineer nanomaterials with polyethylenimine functionalization to regulate the macrophage phenotype by capturing negatively charged cell-free nucleic acids (cfNAs). This cationic nanomaterial reduces macrophage-related inflammation in vitro and demonstrates excellent efficacy in preventing orofacial muscle fibrosis in vivo. Single-cell RNA sequencing reveals that the cationic nanomaterial reduces the proportion of profibrotic Gal3⁺ macrophages through the cfNA-mediated TLR7/9-NF-κB signaling pathway, resulting in a shift in profibrotic fibro-adipogenic progenitors (FAPs) from the matrix-producing Fabp4⁺ subcluster to the matrix-degrading Igf1⁺ subcluster. Our study highlights a strategy to target innate inflammatory response-fibrogenesis homeostasis and suggests that cationic nanomaterials can be exploited for treating refractory fibrosis.

1. Introduction

Orofacial muscles are critical for basic life activities such as speech, mastication and respiration.^[1] Unlike most skeletal muscles, which can completely regenerate after injury,^[2] orofacial muscles are particularly prone to refractory fibrosis following maxillofacial trauma or surgery.^{[1, 3]} This specific behavior seriously hampers stomatognathic function, such as velopharyngeal function and speech after cleft palate repair, as well as facial morphology post-lip surgery,^[4] negatively affecting patients’ quality of life.^[5] Key contributors to refractory fibrosis in orofacial muscles include uncontrolled inflammation and excessive extracellular matrix production, for which no effective drug therapy currently exists.^[6] Hence, there is an urgent need for a new therapeutic strategy that minimizes refractory fibrosis while promoting orofacial muscle regeneration.

Muscle regeneration after injury is initiated by the innate inflammatory response induced by damage-associated molecular patterns (DAMPs), represented by cell-free nucleic acid (cfNA).^[7] The success of muscle regeneration hinges on innate inflammatory response-fibrogenesis homeostasis, which involves the intricate coordination of multiple cell types,^[8] including macrophages for responding to DAMPs,^[9] and fibro-adipogenic progenitors (FAPs) for extracellular matrix production.^[10] Macrophages play prominent roles in fibrogenesis by regulating FAPs to reduce the size of the extracellular matrix^[8]. In this context, differently polarized macrophages modulate the function of FAPs, affecting muscle regeneration after injury.^{[6, 8]} With these findings in mind, we hypothesize that orchestrating macrophage polarization and functions could offer a viable therapeutic strategy for refractory orofacial muscle fibrosis.

cfNA capturing strategies using cationic polymer-based biomaterials have shown therapeutic efficacy in various inflammatory diseases,^[11] which can orchestrate macrophage polarization and help disease alleviation.^{[11b, 11c, 11e, 11g, 12]} In this study, we utilize polyethyleneimine-functionalized diselenide-bridged mesoporous silica nanoparticles (MSN-PEI) as cationic nanomaterials to treat orofacial muscle fibrosis, leveraging their ability to regulate macrophage phenotype by capturing negatively charged cfNAs. ^{[11g, 13]} First, we establish that the macrophage phenotypic switch is closely correlated with regeneration and fibrosis in human orofacial muscles after injury. Next, we investigate the therapeutic effect of MSN-PEI in mouse models of orofacial muscle fibrosis, and we explore how macrophage modulation affects FAP-mediated fibrosis. These cationic nanomaterials, with their capacity to orchestrate fibrosis through reciprocal macrophage-FAP crosstalk, show significant therapeutic benefits in treating refractory orofacial muscle fibrosis.

2. Results

2.1. Macrophage phenotype is correlated with orofacial muscle regeneration and fibrosis after injury

Compared to that of limb muscles, the regeneration of facial muscles is inherently impaired by fibrosis.^{[3c, 14]} Following injury, facial muscles exhibit extensive deposition of fibrotic plaques and infiltration of PDGFRα⁺ cells (Figure S1, Supporting Information). This study focused specifically on orofacial muscles. We obtained samples of human lip muscles postsurgical trauma (Postsurgical time, 17.38 ± 10.69 years) (Table S1) and evaluated the correlations between muscle fibrosis (blue areas stained by Masson staining), muscle regeneration (Desmin⁺ areas), and the number of macrophages (F4/80 labeling all macrophages, CD86 labeling proinflammatory macrophages, and CD206 labeling anti-inflammatory macrophages) (Figure 1a). Correlation analyses pertaining to fibrosis and muscle regeneration were conducted on each independent fascicle within these muscles. In regions of regeneration, few cells were positive for F4/80, CD86, and CD206, whereas in fibrotic areas, the number of F4/80-, CD86-, and CD206-positive cells was notably greater (Figure 1b). Overall, regeneration exhibited a negative correlation with fibrogenesis (Figure 1c), and the presence of macrophages was positively associated with orofacial muscle fibrosis after damage (Figure 1d).

Figure 1. Macrophage phenotypic switch is correlated with regeneration and fibrosis in orofacial muscle after injury, with cfNA linked to macrophage phenotypic switching.

(a) Representative images of Masson trichrome staining, Desmin immunohistochemistry, and F4/80, CD86, CD206 immunofluorescence staining of fibrotic and regenerative areas from human lip muscle biopsy. Arrows point to positive cells. (Scale bar, 75 μm.) (b) Scatter plots of different macrophage populations in fibrotic and regenerative areas. Each dot represents an independent fascicle; the solid line is the fitted regression line, and the dashed line shading represents the 95% confidence interval. (n = 63 fascicles for fibrosis area, n = 63 fascicles for regenerative area.) (c) Scatter plots of fibrotic and regenerative areas. Each dot represents an independent fascicle; the solid line is the fitted regression line, and the dashed line shading represents the 95% confidence interval. (n = 63 fascicles for fibrosis area, n = 63 fascicles for regenerative area.) (d) Heatmap of Desmin⁺, ECM⁺ area, and number of F4/80⁺, CD86⁺, CD206⁺ cells between regenerative and fibrotic areas. (n = 63 fascicles for fibrosis area, n = 63 fascicles for regenerative area.) (e) Temporal dynamics of pro-inflammatory macrophages (M1, CD45⁺, Ly6C⁺, CX3CR1⁺) and anti-inflammatory macrophages (M2, CD45⁺, Ly6C⁻, CX3CR1⁺) at 0, 2, 4, and 7 days post-injury (dpi) in a mouse masseter freezing-induced (FI) injury model. (n = 3 per group. Data are presented as mean ± SD.) (f) Temporal dynamics of cfDNA concentration in the mouse masseter FI injury model. (n = 3 or 4 per group. Data are presented as mean ± SD.) (g) Schematic illustration showing that macrophage phenotypic switch is correlated with regeneration and fibrosis in orofacial muscle after injury.

To investigate the relationship between orofacial muscle fibrosis and macrophages, we established a mouse model of orofacial muscle freezing-induced injury.^[3c] The orofacial muscles of mice, represented by the masseter muscle, also undergo macrophage recruitment and polarization after damage (Figure 1e; Figure S2, Supporting Information). Given that cell-free DNA (cfDNA), a classic DAMP released after tissue damage, can act as a danger signal to activate intrinsic immunity and promote macrophage recruitment and polarization,^[15] the possible association between cfDNA and macrophage phenotypic switch was assessed. We found that the peak cfDNA levels post-injury were associated with the number of proinflammatory macrophages (Figure 1f), suggesting that cfDNA might be a promising target for macrophage modulation.

2.2. cfNA-capturing MSN-PEI reduces macrophage-induced inflammation in vitro

Based on the aforementioned correlation between cfNA, macrophage phenotype and orofacial muscle fibrosis (Figure 1g),^{[7b, 7c]} we used cationic nanomaterials to modulate macrophage phenotypic switch by capturing cfNA.^{[11e, 11g]} We prepared diselenide-bridged mesoporous silica nanoparticles (MSNs) with an average diameter of approximately 50 nm (Figure 2 a–b; Figure S3a, Supporting Information). The MSNs had a high surface area (625.3 m² g⁻¹) and large pore volume (1.14 cm³ g⁻¹) with a narrow pore size distribution centered at 7.3 nm (Figure S3 b–c, Supporting Information). The MSN-PEI nanoparticles were produced by functionalizing PEI-25k (18% wt), imparting a positively charged surface without altering morphology or size (Figure 2b–c; Figure S3d, Supporting Information). Free PEI and MSN-PEI exhibited a high binding affinity for calf thymus DNA (ct-DNA), whereas unmodified MSNs did not (Figure 2d). Though its DNA binding affinity was slightly lower than free PEI, MSN-PEI caused significantly less cytotoxicity in RAW 264.7 cells (Figure 2e). Additionally, MSN-PEI exhibited H₂O₂-responsive degradation (Figure S3e, Supporting Information), enabling its ROS-scavenging activity in lipopolysaccharide (LPS)-challenged RAW 264.7 cells (Figure 2f).

Figure 2. Characterization and anti-inflammatory ability of MSN-PEI for cfNA capturing in vitro.

(a) Schematic illustration of the fabrication of a biodegradable MSN-PEI by conjugating PEI onto the MSN. (Scale bar, 100 nm.) (b) Representative TEM images of MSN and MSN-PEI. (c) Zeta potential of MSN and MSN-PEI. (n = 3 per group. Data are presented as mean ± SD.) (d) DNA binding efficiency of MSN, PEI, and MSN-PEI at different nanoparticle:DNA mass ratios at 37°C. (e) Viability of RAW264.7 cells treated for 24 hours with various concentrations of MSN, PEI, and MSN-PEI. (f) Relative fluorescence intensity of oxidized DCF in RAW264.7 cells after incubation with different formulations in the presence or absence of LPS, MSN, and MSN-PEI for 4 hours. (n = 3 per group. Data are presented as mean ± SD. ****p < 0.0001 by one-way ANOVA with Tukey’s multiple comparison test.) (g) Activation of HEK-TLR reporter cells by poly(I:C), ORN 06, and CpG DNA in the absence or presence of MSN, PEI, and MSN-PEI for 24 hours. The corresponding SEAP activity in supernatants from each group is determined with a QUANTI-Blue assay at OD620. (n = 3 per group. Data are presented as mean ± SD. ***p < 0.001, ****p < 0.0001 by one-way ANOVA with Tukey’s multiple comparison test.)

We evaluated whether MSN-PEI could capture cfNA and inhibit TLR activation. Both MSN-PEI and free PEI significantly inhibited poly(I:C) dsRNA-, ORN 06 ssRNA-, and CpG DNA-induced TLR activation, reducing TNF-α and IL-6 secretion by macrophages (Figure 2g; Figure S3f, Supporting Information). Unmodified MSNs with ROS-scavenging ability also inhibited TLR activation and cytokine secretion, although not as effectively as PEI and MSN-PEI. Intracellular trafficking studies in RAW 264.7 cells showed that CpG DNA was primarily located in lysosomes after four hours when no cationic nanomaterials were used. With MSN-PEI, both the nanomaterial and CpG DNA were found primarily in lysosomes and cytoplasm (Figure S4; Supporting Information). Overall, MSN-PEI effectively captured cfNA and scavenged ROS, inhibiting TLR activation and inflammation in vitro.

2.3. MSN-PEI prevents muscle fibrosis and enhances muscle functional recovery in vivo

Cationic nanomaterials have shown therapeutic effects on various inflammatory diseases.^{[11e–g, 13b, 16]} Muscle injury is aseptic inflammation that can be initiated by the release of cfNA.^[7b] Thus, we investigated the efficacy of MSN-PEI in treating masseter muscle injury. In pilot studies, we induced freezing-induced injury and cutting injury in masseter muscles of C57BL/6 mice, finding optimal therapeutic outcomes followed by local injection of MSN-PEI with suitable concentration and suitable administration time (Figure S5–S8; Supporting Information). Based on these findings, we administered MSN-PEI at half-hour, day two, day four, and day six post-injury, followed by histological examination and functional testing at four weeks post-injury (Figure 3a). We also established a blank group and a fully regenerative control group in which we injected barium chloride into the masseter muscles of the mice.

Figure 3. MSN-PEI treatment prevents muscle fibrosis and enhances muscle functional recovery in a mouse masseter injury model.

(a) Experimental schedule of the in vivo study, including: Control group without any treatments; BaCl₂ group receiving 25 μl of 1.2% BaCl₂ intramuscularly injected into the masseter muscle on 0 dpi; FI group using dry ice to freeze the masseter muscle on 0 dpi; and FI+MSN-PEI group receiving MSN-PEI (0.1 mg ml⁻¹, 50 μl) intramuscularly injected into the masseter muscle after FI injury on days 0, 2, 4, and 6. Masticatory function analysis is performed at 28 dpi, and mice are sacrificed for subsequent histological analysis. FI: freezing injury. (b) Representative gross view of the masseter muscle in different groups. The shortest scale of the ruler is 0.5 mm. (c) Representative images of H&E, Sirius Red, and Laminin immunofluorescence staining in the masseter muscle of different groups. (Scale bar: 1.5 mm for whole sample and 75 μm for magnified fields.) (d, e) Quantification of fibrotic area (d) and myofiber cross-sectional area (e). (n = 3 per group. Data are presented as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by one-way ANOVA with Tukey’s multiple comparison test.) (f) Quantification of overall mouse weight at different time points after injury. (n = 6 per group.) (g) Illustration of biting force measurement. (h-i) Quantification of mouse biting force. (n = 6 per group. Data are presented as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by one-way ANOVA with Tukey’s multiple comparison test.) (j-l) Quantification of mouse eating duration (j), chewing frequency (k), and food intake (l). (n = 6 per group. Data are presented as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by one-way ANOVA with Tukey’s multiple comparison test for normally distributed data, and Kruskal-Wallis test with Dunn’s multiple comparison test for non-normally distributed data.) (m) Schematic illustration of MSN-PEI treatment restoring mouse masticatory function after freezing-induced injury.

In situ injection of MSN-PEI significantly improved the muscular morphology of the masseter muscle. Irregular healing defects were observed on the surfaces of the masseter muscles after freezing injury, while MSN-PEI treatment nearly restored complete regeneration (Figure 3b). In the freezing injury group, sirius red staining showed extensive fibrotic deposition, whereas the MSN-PEI group exhibited much less (Figure 3c). Quantitative analysis revealed that the percentage of fibrotic area in the freezing injury group was approximately 14%, significantly greater than in the fully regenerated group (4%) and blank control group (2.5%). MSN-PEI treatment significantly reduced the fibrotic area to 5.4% (Figure 3d). Moreover, both the regenerative control group and the MSN-PEI group exhibited a large amount of centrally nucleated myofibers. The average cross-sectional area (CSA) of myofibers decreased significantly to 328.4 μm² post-injury, representing regenerative defects, whereas after MSN-PEI treatment, the average CSA increased to 626.2 μm², close to the 714.6 μm² found in the regenerative control group (Figure 3e).

Improved muscle regeneration was correlated with restored masticatory function. Using an occlusal force device, we measured the mice’s bite force (Figure 3 g–h), finding that MSN-PEI treatment significantly reversed injury-induced biting loss (Figure 3f). Additionally, treated mice exhibited improved eating duration, chew rate, and food intake (Figure 3 j–l), which was associated with increased body weight (Figure 3f). No significant toxicity differences were observed between treated and control mice, based on biochemical parameters and histopathological examination of the heart, liver, spleen, lung, and kidney (Figure S9–S10, Supporting Information). Overall, MSN-PEI muscle injection showed efficacy with no observed toxicity in a mouse model of masseter muscle injury (Figure 3m).

2.4. MSN-PEI ameliorates cfNA-TLR7/9-mediated macrophage inflammation in orofacial muscle after injury

To elucidate the mechanisms by which MSN-PEI regulates inflammation and reduces muscle fibrosis, we conducted single-cell sequencing on muscle tissues from the control, injury, and MSN-PEI groups 7 days post-injury. Toll-like receptors (TLRs) in macrophages can recognize DAMPs and drive inflammatory processes. TLR1, TLR2, TLR3, TLR4, TLR7, TLR8, and TLR9 are expressed in muscle.^[17] MSN-PEI exerts its therapeutic effects by capturing cfNA, which might inhibit TLR-mediated macrophage activation. Our analysis revealed significant increases in TLR7/9 expression in fibrotic regions of human lip muscles compared to myogenic areas, and in injured mouse masseter muscles compared to controls (Figure S11, Supporting Information; Figure 4a–c). These results indicate a close association between high TLR7/9 expression and orofacial muscle fibrosis. TLR7/9 expression was predominantly found in macrophages and significantly decreased after MSN-PEI treatment (Figure 4c). Immunofluorescence staining confirmed reduced TLR7 expression after seven days of treatment, while TLR9 decreased significantly after three days (Figure 4d–e). Although the decrease in TLR7/9 expression did not correlate with a decrease in fibrosis area, there was abundant regeneration of centrally nucleated myofibers in the MSN-PEI group at 7 days post-injury (Figure 4d–e), indicating enhanced muscle regeneration.

Figure 4. MSN-PEI ameliorates cfNA-TLR7/9-mediated macrophage inflammation in orofacial muscle after injury by capturing cfNA.

(a) Representative immunohistochemistry images of TLR7 and TLR9 in human lip muscle biopsy and quantification of the percentage of TLR7/9-positive cells. Each dot represents an independent fascicle. (n = 11 fascicles for fibrosis area, n = 12 fascicles for regeneration area. Data are presented as mean ± SD. **p < 0.01, ****p < 0.0001 by Student’s t-test.) (b) scRNA-seq identified 10 cell clusters in the integrated analysis of Control, FI (freezing injury), and FI+MSN-PEI groups at 7 dpi (n = 3 samples per group). UMAP dimensionality reduction shows that TLR7 and TLR9 are mainly expressed in the mononuclear phagocyte (MP) cluster. (c) Feature plot of the expression levels of TLR7 and TLR9 in Control, FI, and FI+MSN-PEI groups. (d) Representative immunofluorescence and H&E images in mice masseter muscle at 3 days and 7 days post-injury. TLR7 (red), TLR9 (green). (Scale bar, 75 μm.) (e) Quantification of TLR7-positive cells per field, TLR9-positive cells per field, and fibrotic area in Control, FI, and FI+MSN-PEI groups. (n = 3 samples per group) (f) Temporal dynamics of cfDNA and cfRNA concentrations in Control, FI, and FI+MSN-PEI groups at 3, 7, and 14 dpi. (n = 3 per group. Data are presented as mean ± SD. *p < 0.05 by Student’s t-test.) (g) Relative expression of Tlr7, Tlr9, Myd88, and Rela in mice masseter muscle in Control, FI, and FI+MSN-PEI groups at 3 and 7 dpi. (n = 3 per group. Data are presented as mean ± SD. *p < 0.05 by Student’s t-test.) (h) Schematic illustration of cfNA-TLR7/9 activation in orofacial muscle macrophages: MSN-PEI captures cfNA, inhibits TLR7/9 expression, and dampens NF-κB pathway activity.

TLR7 primarily recognizes cfRNA, while TLR9 targets cfDNA.^[18] In the MSN-PEI group, cfDNA concentrations significantly decreased at days three and seven post-injury, and cfRNA concentrations decreased at day three (Figure 4f), providing evidence that MSN-PEI could reduce the activation of TLR7/9 in vivo by capturing cfNA. TLR7/9 interaction with the MyD88 complex leads to NF-κB pathway activation.^[19] MSN-PEI significantly reduced TLR7/9 and downstream MyD88 gene expression post-injury, inhibiting the NF-κB pathway (Figure 4g). MSN-PEI also reduced TLR7/9 activation in response to muscle injury-related DAMPs in orofacial muscle extracellular fluid (Figure S12a–b, Supporting Information). CpG DNA-ssRNA stimulation induced a proinflammatory phenotype in macrophages, which MSN-PEI reversed (Figure S12c, Supporting Information). ROS depletion may also contribute to reduced ROS-related proinflammatory activation (Figure S13). Overall, MSN-PEI attenuated cfNA-TLR7/9 pathway activation in proinflammatory macrophages, inhibiting inflammation after orofacial muscle injury (Figure 4h).

2.5. MSN-PEI suppresses profibrotic macrophage subclusters through the TLR7/9-NF-κB signaling pathway

Following MSN-PEI treatment, the expression of components of proinflammatory pathways was downregulated in macrophages, suggesting an anti-inflammatory phenotypic switch (Figure S14, Supporting Information). A macrophage phenotypic switch can be found in the masseter muscles of mice, as a significant infiltration of proinflammatory macrophages was observed in the area of muscle injury at 3 days post-injury, while a large infiltration of anti-inflammatory macrophages was observed at 7 days post-injury (Figure 5a). Quantitative analysis confirmed reduced proinflammatory macrophages after MSN-PEI treatment at day three, with more anti-inflammatory macrophages at day 7 (Figure 5b). TNF-α levels decreased, and IL-10 levels increased, indicating a phenotypic shift toward an anti-inflammatory state (Figure 5c). Flow cytometry further verified the macrophage switch. The proportion and number of proinflammatory macrophages (CD45⁺, Ly6C⁺, CX3CR1⁺) per mg of tissue were lower in the MSN-PEI-treated group than in the injury group at 3 days post-injury. However, at 7 days post-injury, the proportion and number of anti-inflammatory macrophages (CD45⁺, Ly6C⁻, CX3CR1⁺) were also reduced in the MSN-PEI-treated group compared to the injury group (Figure 5 d–e). The combined immunofluorescence and FACS data indicate a phenotypic switch in macrophages following injury that differs from typical patterns.

Figure 5. The pro-fibrotic macrophage subcluster 4 is the main effector cell bridging the upstream cfNA-TLR7/9 signaling and the downstream fibrotic process.

(a) Representative immunofluorescence images of F4/80⁺ CD86⁺ pro-inflammatory and F4/80⁺ CD206⁺ anti-inflammatory macrophages in control, FI, and FI+MSN-PEI groups. (Scale bar, 25 μm.) (b) Quantification of pro-inflammatory and anti-inflammatory macrophages in control, FI, and FI+MSN-PEI groups. (n = 3 per group. Data are presented as mean ± SD. **p < 0.01 by Student’s t-test.) (c) TNF-α and IL-10 levels in mouse masseter muscle of the control group, FI, and FI+MSN-PEI groups at 3 and 7 dpi. (n = 3 per group. Data are presented as mean ± SD. *p < 0.05, **p < 0.01 by Student’s t-test.) (d) Comparison of temporal dynamics of pro-inflammatory and anti-inflammatory macrophages in mouse masseter muscle of FI and FI+MSN-PEI groups at 0, 3, and 7 dpi. Numbers of cells per mg tissue are calculated by dividing the total number of cells by the weight of tissue used for analysis. (n = 3 per group. Data are presented as mean ± SD.) (e) Flow cytometry of CD45⁺ Ly6C⁺ CX3CR1⁺ pro-inflammatory and CD45⁺ Ly6C⁻ CX3CR1⁺ anti-inflammatory macrophages in FI and FI+MSN-PEI groups at 3 and 7 dpi. Numbers represent the percentage in each gate of CD45⁺ cells. (f) UMAP dimensionality reduction of macrophages. (g) Changes in the proportion of different macrophage subclusters in control, FI, and FI+MSN-PEI groups. (h) Monocle analysis of macrophage subclusters in control, FI, and FI+MSN-PEI groups. (i) GSVA analysis of pro-fibrotic genes in macrophage subclusters. (j) Venn diagram of macrophage subclusters enrich in TLR signaling and ECM signaling. (k) Monocle analysis of macrophage subcluster 4 in control, FI, and FI+MSN-PEI groups. (l) Immunofluorescent staining of F4/80 (green) and Gal3 (red) in control, FI, and FI+MSN-PEI groups at 7 dpi. (Scale bar, 50 μm.) (m) Quantification of total and Gal3⁺ macrophages. (n = 3 per group. Data are presented as mean ± SD. *p < 0.05 by one-way ANOVA with Tukey’s multiple comparison test.) (n) Schematic illustration of macrophage phenotypic switch in response to MSN-PEI treatment.

Masseter muscle cells were classified into 10 major populations based on signature genes at the single-cell level (Figure S15a, Supporting Information). Mononuclear phagocyte (MP) numbers were lower in the MSN-PEI group than in the injury group (Figure S15b–d, Supporting Information), aligning with flow cytometry data. MPs were divided into six subclusters, while macrophages were categorized into seven subclusters (Figure S16a–c, Supporting Information). Of these, subclusters 2, 3, and 5 increased significantly with MSN-PEI treatment, while subclusters 1, 4, and 6 decreased (Figure 5 f–g). Traditional M1/M2 markers didn’t fully match these subclusters (Figure S16d, Supporting Information). Macrophages in the injury group appeared functionally distinct from those in the MSN-PEI group (Figure 5h).

We examined the gene profiles of macrophage subclusters associated with skeletal muscle fibrosis to identify the macrophage subclusters.^[7c] Subcluster 4 expressed profibrotic genes represented by Galectin-3 (Gal3⁺) and Spp1,^[7c] and scored significantly higher than the other subclusters for the extracellular matrix remodeling pathway (Figure 5i; Figure S16e–l, Supporting Information). A previous study showed that Gal3⁺ macrophages are chronically activated in atrophied muscle and that they promote tissue fibrosis by interacting with stromal cells through the secretion of the cytokines Spp1 and Gal3.^{[7c, 20]} Thus, we defined macrophage subcluster 4 as a key cellular subcluster of profibrotic macrophages (Figure 5j). We further clarified the function of the Gal3⁺ macrophages by GSEA. KEGG enrichment analysis revealed that only macrophage subclusters 3 and 4 were enriched for the TLR signaling pathway and the NF-κB signaling pathway (Figure S1 a, Supporting Information), whereas fibrosis-associated pathways were only enriched in macrophage subclusters 1 and 4 (Figure S17b, Supporting Information), suggesting that macrophage subcluster 4 is a key cell subcluster involved in the cfNA-TLR7/9-mediated fibrogenesis response (Figure 5j). MSN-PEI treatment delayed the differentiation process of macrophage subcluster 4 (Figure 5k), as evidenced by the significant downregulation of its extracellular matrix formation pathway (Figure S1 a–b, Supporting Information).

Immunofluorescence staining confirmed the trend of a decrease in the proportion of Gal3⁺ macrophages after MSN-PEI administration (Figure 5 l–m). At the total macrophage level, the fibrosis-related pathways were less enriched in the MSN-PEI-treated group than in the control group (Figure S18c, Supporting Information). Overall, MSN-PEI treatment promoted muscle regeneration by attenuating TLR7/9-NF-κB signaling pathway-mediated inflammation in macrophages (Figure 5n).

2.6. MSN-PEI modulates profibrotic FAPs from the matrix-producing Fabp4⁺ subcluster to the matrix-degrading Igf1⁺ subcluster

FAPs are directly associated with skeletal muscle fibrosis,^{[3c, 21]} and their behavior is affected by the paracrine effects of macrophages.^[8] The area of FAP infiltration (PDGFRα⁺) decreased significantly, as did the number of proliferating FAPs (PDGFRα⁺ and Ki67⁺), while apoptotic FAPs (PDGFRα⁺ and caspase-3⁺) increased (Figure 6 a–b). These results suggest that sequential proliferation and apoptosis of FAPs occurred after MSN-PEI treatment. The expression of two fibrosis-related genes, Tgfβ1 and Col1a1,^{[21b, 22]} in the masseter muscle was significantly elevated in the injury group at 7 days and 14 days post-injury but decreased after MSN-PEI treatment (Figure 6c), indicating an attenuated fibrotic response.

Figure 6. MSN-PEI inhibits muscle fibrosis by modulating FAPs’ phenotypic switch.

(a) Representative immunofluorescence images of proliferative (Ki67⁺) and apoptotic (Caspase⁺) FAPs (PDGFRα⁺) in the control group, FI, and FI+MSN-PEI groups at 7, 14, and 28 dpi. (Scale bar, 75 μm.) (b) Quantification of PDGFRα⁺ area, proportion of proliferative and apoptotic FAPs in the control group, FI, and FI+MSN-PEI groups at 7, 14, and 28 dpi. (n = 3 per group. Data are presented as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001 by Student’s t-test.) (c) Relative expression of fibrosis-related genes in mouse masseter muscle of control, FI, and FI+MSN-PEI groups at 7 and 14 dpi. (n = 3 per group. Data are presented as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by Student’s t-test for comparison between two groups and one-way ANOVA with Tukey’s multiple comparison test for comparison of multiple groups.) (d) Co-culture of untreated (control group), stimulated plus vehicle (vehicle group), and stimulated plus MSN-PEI (MSN-PEI group) treated RAW264.7 cells and FAPs to investigate the proliferation, apoptosis, and differentiation of FAPs. (e) Representative immunofluorescence images of proliferative (Ki67⁺), apoptotic (Caspase⁺), and differentiating (αSMA⁺) FAPs in control, vehicle, and MSN-PEI groups. (Scale bar, 25 μm.) (f) Quantification of proliferative, apoptotic, and differentiating FAPs in control, vehicle, and MSN-PEI groups. (n = 3 per group. Data are presented as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001 by one-way ANOVA with Tukey’s multiple comparison test.) (g) UMAP dimensionality reduction of FAPs in the integrated group and in control, FI, and FI+MSN-PEI groups, respectively. (h) Changes in the proportion of FAP subclusters in control, FI, and FI+MSN-PEI groups. (i) Mapplot of GO pathway enrichment in Fabp4⁺ FAPs. (j) Expression levels of fibrosis-related genes Cthrc1, Tagln, and Acta2 in different FAP subclusters. (k) Inflammation-related score in different FAP subclusters. (l) KEGG pathway enrichment in Igf1⁺ FAPs. (m) Immunofluorescent analysis and quantification of Fabp4⁺ FAPs in the control, FI, and FI+MSN-PEI groups. (Scale bar, 50 μm.) (n = 3 per group. Data are presented as mean ± SD. *p < 0.05 by one-way ANOVA with Tukey’s multiple comparison test.) (n) Representative images of tyramide signal amplification staining on F4/80 and Gal3 (upper panel), PDGFRα and Fabp4 (lower panel). *Represents Fabp4⁺ FAPs in the adjacent section. ^#Represents Gal3⁺ macrophages in the adjacent section. (Scale bar, 20 μm.) (o) Schematic illustration of phenotypic changes in the FAP population in response to MSN-PEI-mediated macrophage phenotypic switch.

Transwell experiments showed that MSN-PEI inhibited FAP differentiation induced by CpG-stimulated macrophages while increasing FAP apoptosis (Figure 6d–f). Single-cell analysis classified FAPs into six subclusters (Figure S19a, Supporting Information). The injury group was dominated by the Fabp4⁺ subcluster, while the Igf1⁺ subcluster prevailed after MSN-PEI treatment (Figure 6g–h). Fabp4⁺ FAPs were linked to myofibroblast-related pathways and expressed elevated levels of fibrosis-associated genes Cthrc1, Tagln, and Acta2 (Figure 6i–j; Figure S19b–c, Supporting Information), while Igf1⁺ FAPs expressed inflammation-related genes (Figure 6k). Enrichment analysis revealed matrix degradation pathways, including protein and glycan degradation, were upregulated in Igf1⁺ FAPs (Figure 6l). The increase in the abundance of the Igf1⁺ population might contribute to the enhancement of inflammation-related pathways (Figure S19d, Supporting Information) in macrophages via a positive feedback loop,⁴⁴ leading to heightened inflammation in macrophages and FAPs (Figure S20, Supporting Information). Immunofluorescence staining further verified the decrease in Fabp4⁺ FAPs after MSN-PEI administration (Figure 6m), and two adjacent sections revealed the colocalization of Gal3⁺ macrophages and Fabp4⁺ FAPs, providing further evidence for their potential interaction (Figure 6n). Enhanced TNF ligand-receptor interactions between macrophages and FAPs in the MSN-PEI group corresponded with increased Fabp4⁺ FAP apoptosis (Figure S21, Supporting Information). Thus, MSN-PEI modulates FAP phenotypic switch from the matrix-producing Fabp4⁺ subcluster to the matrix-degrading Igf1⁺ subcluster (Figure 6o) by promoting macrophage phenotypic switch through cfNA-TLR7/9 inhibition.

3. Discussion

The orofacial muscles play a critical role in basic life activities, but they have a reduced capacity to regenerate after injury. The mechanism of refractory muscle fibrosis remains unclear, and effective treatment modalities are limited. Since uncontrolled inflammation is a primary contributor to tissue fibrosis,^[6] our investigation began by exploring the microenvironment of the orofacial muscle post-injury.

This study is the first to identify persistent inflammation in orofacial muscles after injury, potentially revealing why orofacial muscle regeneration outcomes differ from limb muscle regeneration. We detected both pro- and anti-inflammatory macrophages even several months after surgical lip repair, which is intriguing. This indicates that orofacial muscles suffer chronic inflammation after a single injury, resulting in impaired muscle regeneration with fibrotic and adipogenic tissue deposition.^[23] The dysfunctional dystrophin, a protein that links intracellular and extracellular structural proteins,^[24] may be a key factor in this impaired regeneration process. A deficiency in dystrophin weakens the cell membrane, leading to disrupted homeostasis and cell death during muscle regeneration. We and Shah’s team^[3a] demonstrated that lip and palate muscle fibers do not ubiquitously express the dystrophin protein (Figure S22, Supporting Information). This dystrophin expression pattern in orofacial muscles might increase their susceptibility to injury and susceptibility to chronic waves of injury after a single surgical stimulus.

We found that chronic persistent inflammation and the macrophage phenotypic switch in the postsurgical fibrotic area of the orbicularis oris muscles were linked to changes in the local cfNA level post-surgery. Chronic inflammation may arise and persist through the activation of DAMPs and pattern recognition receptors (PRRs).^[25] TLR7 and TLR9, endosomal PRRs for cfNA (cfRNA and cfDNA), play vital roles in macrophage phenotypic switch and muscle fibrosis^{[7b, 26]}. Previous studies have demonstrated that in dystrophic muscles characterized by extensive muscle fibrosis, increased expression of TLR7 and TLR9 aggravates the inflammatory process and perpetuates disease progression.^[27]

Although no drug therapy specifically targets skeletal muscle fibrosis, strategies focusing on cfNA-TLR7/9 and downstream NF-κB signaling pathways show promise in reducing muscle inflammation and ameliorating the fibrotic muscle phenotype.^[26–28] Hydroxychloroquine sulfate (HCQ), a small molecule inhibitor of TLR7/8/9, can reduce skeletal muscle fibrosis but can cause side effects such as rheumatoid arthritis, lupus erythematosus and vacuolar myopathy.^[27] Edasalonexent and CAT-1041, both NF-κB inhibitors, significantly reduce fibrotic tissue formation in dystrophic mice and dog muscles,^[29] but are ineffective in patients.^[30] Infliximab, an antibody against TNFα, has been shown to reduce fibrosis and enhance muscle strength in mice with muscular dystrophy, yet it also has a detrimental effect on cardiac function.^[31] The antioxidant N-acetyl-cysteine effectively reduces fibrosis by lowering TNF-α levels and immune cell infiltration,^[32] but it also induces muscle mass loss and suppresses body weight gain.^[30] Given these limitations, we sought a novel strategy to attenuate cfNA-TLR7/9 activation and regulate the macrophage phenotype to combat refractory orofacial muscle fibrosis.

Cationic nanomaterials regulate macrophage phenotypic switch by capturing negatively charged cfNA, providing therapeutic efficacy in inflammatory diseases.^{[11, 13, 33]} This cfNA capturing strategy, utilizing cationic polymers such as the classic third-generation polyamidoamine dendrimer (PAMAM-G3), can mitigate TLR activation in target immune cells. However, PAMAM-G3 exhibits notable in vivo toxicity,^[34] which limits its clinical application. Additionally, the variability between batches of commercially available PAMAM-G3 presents further challenges.^[11g] Consequently, we selected PEI for this study, given its controllable molecular weight and potential for clinical translation,^[35] as it supports scalable manufacturing for viral vector production.^[36] We chose MSNs for their high surface area, with the diselenide-bridged MSN enhancing potential clinical translation.^[37] Meanwhile, stable PEI-functionalized MSNs have shown broad utility in gene delivery without significant in vivo toxicity.^[38]Thus, as this strategy targets the source of inflammation while minimizing side effects, it offers a safer and more efficient treatment approach for refractory orofacial muscle fibrosis. In this study, we investigated this biodegradable cationic nanomaterial MSN-PEI, known for its ability to alleviate inflammation in the digestive and circulatory systems.^{[11g, 13b]} Our results demonstrated that MSN-PEI alleviated refractory orofacial muscle fibrosis by capturing cfNAs and blunting TLR7/9 activation.

Interestingly, these cationic nanomaterials can also influence macrophage phenotypic switch in tissue-specific ways. Our findings showed that the conventional M1-M2 paradigm doesn’t fully capture these macrophage clusters, which could be better classified into categories such as tissue-resident macrophages, monocyte-derived macrophages (MDMs), and profibrotic macrophages. MSN-PEI reduced the proportion of Gal3⁺ macrophages, reversing their transition state and lowering fibrosis-related gene expression levels. Gal3⁺ macrophages were the only cluster where toll-like receptor signaling intersected with extracellular matrix (ECM) deposition, suggesting a connection between inflammatory and fibrotic processes. This aligns with recent findings that identified Gal3⁺ macrophages as key contributors to muscle fibrosis in both acute and chronic injuries.^{[11e, 11g, 13b]}

Since macrophages do not directly deposit matrix, we identified downstream effector cells, fibro-adipogenic progenitors (FAPs).^[8] Our results showed that MSN-PEI application led to conspicuous changes in the FAP population, with an increase in the Igf1⁺ immune-interacting cluster (proinflammatory) and a decrease in the Fabp4⁺ tissue-remodeling cluster (profibrotic). Although it is unclear whether Fabp4⁺ FAPs directly transitioned into Igf1⁺ FAPs or underwent multiple steps, their polarization contributed to the antifibrotic effects of cationic nanomaterials in orofacial muscle. In line with previous work on skin scar formation and rheumatoid arthritis, the increase in the population of proinflammatory fibroblasts prevented their conversion to myofibroblasts and resulted in less fibrosis.^[39]

A “positive feedback loop” has been reported in tissue injury in which immune cell-derived factors activate fibroblasts to produce proinflammatory cytokines, which initiate immune cell recruitment.^[40] Therefore, we speculated that the drastic increase in the proportion of the Igf1⁺ FAP cluster could somehow recruit more MDMs. Although the proportion of MDMs might be reduced after inhibiting cfNA-TLR7/9 activation, the net effect could result in an expanded population of MDMs. Moreover, a decrease in the proportion of Fabp4⁺ cells resulted in a decrease of extracellular matrix deposition. It has been extensively investigated in skeletal muscle regeneration that pro-inflammatory macrophages could exert a pro-apoptotic effect on FAP,^[41] thus keeping muscle fibrosis at bay. But the exchange of inflammatory signals between these two cell types was seldomly studied. It will offer new perspective to further probe into the reciprocal action between Gal3⁺ macrophages and Fabp4⁺ FAPs.

Fatty acid-binding protein 4 (FABP4) is a member of the FABP family and mainly participates in lipid metabolism.^[42] The functional role of FABP4 in skeletal muscle fibrosis has not been well-studied. It is highly expressed in adipocytes and macrophages.^[43] Aberrantly high expression of FABP4 can lead to fibrosis in lung and kidney disease, while inhibition of FABP4 can attenuate fibrosis.^[43–44] Our work provides the first evidence that Fabp4⁺ FAPs are the effector cells responsible for muscle fibrosis and provides insight into the role of the Gal3⁺ macrophage-Fabp4⁺ FAP interaction in the progression of fibrosis, which could serve as a therapeutic target for combating skeletal muscle fibrosis. The colocalization of Gal3⁺ macrophages and Fabp4⁺ FAPs further supported this intercellular communication.

There are several limitations of this study. The therapeutic effects of intramuscular injection with different distances and sites, and systemic drug administration methods, such as intravenous and intraperitoneal injection, should be evaluated in future studies. Meanwhile, although MSN-PEI demonstrated significant ameliorating effects in freezing-induced orofacial muscle fibrosis in mice, which mimicked the clinical scenario, the results in the cutting injury model were less satisfactory. The severity of muscle injury could have an impact on muscle regeneration outcome, rendering the scheme of nanoparticle application context-dependent. Thus, validating the cationic nanomaterial strategy in other muscle fibrosis models, as well as comparing the cationic nanomaterial strategy with other potential pharmacotherapy, should also be carried out in future studies. Next, scRNA-seq and immunofluorescent analysis indicated that Gal3⁺ macrophage and Fabp4⁺ FAPs were the key effector cells, but their exact roles in muscle regeneration and fibrosis still need gain- and loss-of-function studies to validate.

4. Conclusion

Our research reveals that macrophage phenotypic switch and overactivation of TLR7/9 signaling contribute to refractory orofacial muscle fibrosis. Cationic nanomaterials offer a promising approach to alleviate this fibrosis by neutralizing TLR7/9 activity through MSN-PEI-mediated cfNA capture. Our study also demonstrates the importance of the Gal3⁺ macrophage-Fabp4⁺ FAP axis in counteracting fibrosis by coordinating TLR signaling and fibrotic processes. Understanding the reciprocal macrophage-FAP interaction in these processes is crucial for developing new antifibrotic treatment approach (Figure 7).

Figure 7. Gal3⁺ macrophage-Fabp4⁺ FAPs axis can be mediated by cationic nanomaterial through cfNA-TLR7/9 signaling pathway to combat refractory orofacial muscle fibrosis.

The cfNA-TLR7/9 signaling pathway plays a major role in orofacial muscle fibrosis after injury, and cationic nanomaterials can reduce fibrosis by modulating this pathway. Cationic nanomaterials can decrease the proportion of pro-fibrotic Gal3⁺ macrophages, leading to a reduction in pro-fibrotic Fabp4⁺ myofibroblasts, which contributes to attenuated muscle fibrosis. Manipulating the Gal3⁺ macrophage-Fabp4⁺ FAPs axis could be a promising strategy for combating refractory orofacial muscle fibrosis.

5. Experimental Methods

Materials and reagents

Dulbecco’s modified Eagle’s medium (DMEM) (11995065), Penicillin-Streptomycin (P/S) (10,000 U mL⁻¹) (15140122), and trypsin-EDTA (0.25%) (25200056) were purchased from Gibco (Carlsbad, CA, U.S.A.). Fetal bovine serum (FBS) (A5669701) and Horse serum (16050122) were purchased from GIBCO BRL (Grand Island, NY, USA). Poly(I:C) (tlrl-picw), ORN06/LyoVec (tlrl-orn6), CpG2006 (tlrl-2006), and QUANTI-Blue^™ solution (rep-qbs) were purchased from InvivoGen (San Diego, CA, U.S.A). H&E staining kit (G1120) and Masson’s trichrome staining kit (G1346) were purchased from Solarbio (Beijing, China). Paraformaldehyde (4%) (BL539A) was purchased from Biosharp (Hefei, China). ELISA kit for mouse IL-10 (MM-0176M1) was purchased from Jiangsu Meimian (Jiangsu, China). DAB Horseradish Peroxidase Color Development Kit (P0202) was purchased from Beyotime (Shanghai, China). Quant-iT PicoGreen dsDNA assay kit (P11495), DyLight^™ 488 (40603), TRIzol (15596018) and anti-F4/80 (MF48000), sodium pyruvate (11840014), LysoTrackerTM Red DND-99 (L7528), and ELISA kit for mouse TNF-α and IL-6 (BMS607–3, BMS603–2) were purchased from Thermo Scientific (Waltham, Massachusetts, U.S.A). PrimeScript^™ FAST RT reagent Kit with gDNA Eraser (RR092A) and TB Green^® Premix Ex Taq^™ (Tli RNaseH Plus) (RR420A) were purchased from Takara Bio (Beijing, China). Anti-CD16/CD32 antibodies (101319), CD45 (103108), Ly6C (128008), and CX3CR1 (149008) were purchased from Biolegend (San Diego, CA, U.S.A). Anti-CD86 (ER1906–01), anti-TLR7 (ER30606), anti-αSMA (ET1607–53), and HRP Conjugated Goat anti-Rabbit IgG antibody (HA1001) were purchased from Huabio (Jiangsu, China). Anti-CD206 (ab300621), anti-Ki67 (ab15580), anti-desmin (ab32362), and sulforhodamine B assay kit were purchased from Abcam (Cambridge, UK). Barium chloride (202738), Triton X-100 (X-100), donkey serum (S30-M), goat serum (S26F-MLSG), anti-laminin (L9393), collagenase II (C2-BIOC), branched polyethylenimine (PEI, Mw 25 kDa) (408727), ammonium nitrate (1.01187), tetraethyl orthosilicate (8.00658), (3-glycidyloxypropyl)trimethoxysilane (440167), bis.^{[3-(triethoxysilyl)propyl]}diselenide (15200), triethanolamine (8.22341), and cetyltrimethylammonium tosylate (CTAT) (8.14692) were purchased from Sigma-Aldrich (St. Louis, MO, U.S.A.). Anti-TLR9 (BC05301367) was purchased from Bioss (Beijing, China). Anti-PDGFRα (AF1062) was purchased from R&D Systems (Minneapolis, Massachusetts, U.S.A). Anti-embryonic MyHC (BF-G6) and anti-MyH4 (MF20) were purchased from Developmental Studies Hybridoma Bank (Iowa City, Iowa, U.S.A). Anti-cleaved caspase-3 (#9661) was purchased from Cell Signaling Technology (Danvers, Massachusetts, U.S.A). Dispase (04942078001) and pronase (PRON-RO) were purchased from Roche (Shanghai, China). Chicken embryo extract (092850145) was purchased from MP Biomedical (Santa Ana, CA, U.S.A.). Goat Anti-Rat IgG (A23240), IFKine^™ Red Donkey Anti-Rabbit IgG (A24421), IFKine^™ Green Donkey Anti-Rabbit IgG (A24221), IFKine^™ Green Donkey Anti-Goat IgG (A24231), and IFKine^™ Red Donkey Anti-Mouse (A24411) were purchased from Abbkine (Wuhan, China).

Patient samples

Lip muscle samples with the size of 3×3×2 mm from the resected tissue were collected from 29 patients who underwent secondary cleft lip repair surgery at the West China Hospital of Stomatology, Sichuan University (Table. S1). Muscle samples were stored in saline solution at 4°C for subsequent laboratory analysis. These samples were collected with permission from the West China Hospital of Stomatology’s Ethical Committee, Sichuan University (WCHSIRB-CT-2022–342), and with the informed consent of the patients involved.

Synthesis of diselenide-bridged MSN and MSN-PEI

Diselenide-bridged MSNs were prepared as our protocol reported previously.^[45] Cetyltrimethylammonium tosylate (CTAT), triethanolamine, bis.^{[3-(triethoxysilyl)propyl]}diselenide (BTESePD), tetraethyl orthosilicate were gradually dissolved in deionized water and stirring at 80°C. Following centrifugation, the products were collected and washed with ethanol and refluxed in a solution of ethanol of ammonium nitrate (1% w/v) for 12 h. The diselenide-bridged MSN were gathered, rinsed, and refluxed in ethanol to dissolve CTAT.

For synthesizing MSN-PEI, MSN (1.0 g) was dispersed in toluene (250 ml), followed by the addition of (3-glycidyloxypropyl)trimethoxysilane (1.5 ml). The mixture was then refluxed at 80°C for 24 h to yield epoxysilane-functionalized MSN. Following purification, the epoxysilane-functionalized MSN (500 mg) were dispersed in PEI 25K solution (250 mL, 1 mg mL⁻¹) and stirred at room temperature for 24 h. The diselenide-bridged MSN-PEI were subsequently gathered, rinsed, and dried for future applications.

Characterization of MSN and MSN-PEI

The structural features of MSN and MSN-PEI were visualized via transmission electron microscopy (TEM, JEOL, Ltd., Japan) and scanning electron microscopy (SEM, FEI Quanta 200F). Zeta potential and hydrodynamic diameter were characterized using a Zetasizer (Nano ZS90, Malvern Panalytical). The pore size distribution and specific surface area were determined by Barrett-Joyner-Halenda (BJH) and Brunauer-Emmett-Teller (BET) analyses. The PEI content within MSN-PEI was quantified by thermogravimetric analysis (TGA, PerkinElmer, U.S.A.). MSN-PEI degradation was assessed by incubating a 100 μg mL⁻¹ concentration in a H₂O₂ solution (100 μM) at 37°C under constant rotation. Samples were collected at 0, 1, and 2 days for TEM analysis.

DNA binding assay

The binding ability of MSN, PEI, and MSN-PEI with ct-DNA was evaluated as described previously.^[46] ct-DNA solution, PicoGreen were mixed in a 96-well plate. Followed by shaking and adding different concentrations of MSN, PEI, or MSN-PEI solutions. After 1-hour incubation at 37°C, the fluorescence intensity was measured to determine binding ability.

In vitro TLR activation assay

To assess TLR activation, HEK-Blue reporter cells expressing TLR3, TLR7, and TLR9 were purchased from InvivoGen (San Diego, CA, U.S.A.). Cells were maintained in a growth medium consisting of DMEM supplemented with 10% FBS and incubated at 37°C in a humidified atmosphere with 5% CO₂.

To test cfNA capturing ability of cationic nanomaterials, poly(I:C) as dsRNA, ORN 06/LyoVec as ssRNA, and CpG 2006 as CpG DNA were used for respective HEK-Blue TLR cells. Prior to treatment, specific quantities of HEK-Blue reporter cells were seeded in 96-well plates with basal DMEM and cultured for 12 h: 5 × 10⁴ cells per well for TLR3, 4 × 10⁴ cells per well for TLR7, and 8 × 10⁴ cells per well for TLR9. Each cell type was subsequently stimulated with its designated agonist: 1 μg mL⁻¹ low molecular weight poly (I:C) for TLR3, 0.5 μg ml⁻¹ ORN06/LyoVec for TLR7, and 1 μg mL⁻¹ CpG 2006 for TLR9. To test whether cationic nanomaterials can reduce TRL7 and TLR9 activations caused by orofacial muscle fibrosis-related DAMPs, interstitial fluid, and DAMPs medium from C2C12 or 3T3-L1 were prepared. The cells were stimulated with interstitial fluid (10μl) and DAMPs medium from C2C12 or 3T3-L1 (10μl). In the materials-treated groups, MSN, PEI, or MSN-PEI (10 μg mL⁻¹) were introduced 30 min before agonist. Following a 24-hour incubation period, supernatants were collected, and the activity of secreted embryonic alkaline phosphatase (SEAP) was measured with the QUANTI-Blue assay kit.

Anti-inflammatory and ROS assay in vitro

RAW 264.7 cells were cultured in growth medium at 37°C with 5% CO₂. Cells were seeded at a density of 2 × 10⁴ cells per well for anti-inflammatory assay and 1 × 10⁴ cells per well for ROS assay.

For anti-inflammatory assay, cells were pre-treated with CpG 2006 (1 μg mL⁻¹) for 30 min. Subsequently, MSN, PEI, and MSN-PEI (10 μg mL⁻¹) were introduced to the treated well. Then adjusted the final volume to 200 μl. Following a 24-hour incubation, supernatants were collected, and levels of TNF-α and IL-6 in the supernatants were quantified.

For ROS assay, cells were gradually treated with lipopolysaccharide (1 μg mL⁻¹) for 4 h and materials for 24 h. Following treatment, intracellular ROS generation was measured using 2′–7′ dichlorofluorescin diacetate (DCFH-DA).

Cytotoxicity assay in vitro

RAW 264.7 cells were seeded at a density of 1 × 10⁴ per well in 96-well plates. After a 12-hour incubation, varying concentrations of MSN, PEI, or MSN-PEI were added, followed by a 24-hour incubation. Cell viability was then assessed using the sulforhodamine B (SRB) assay.

Fluorescent labeling of the intracellular uptake of MSN-PEI

Dye-labeled MSN-PEI was synthesized following our previously reported protocol.^[11g] To prepare dye-labeled MSN-PEI, Cy7-NHS and FITC were utilized. Briefly, 1 mg of fluorescent label was combined with 10 mg of amino-functionalized MSN-PEI and gently agitated overnight at 4°C. Unbound dye molecules were then removed by centrifugation, yielding the purified Cy7-MEN-PEI and FITC-MSN-PEI for further use.

RAW 264.7 cells were seeded and cultured for 12 h on cover glasses. FITC-CpG (1 μg mL⁻¹) and Cy7-MSN-PEI (2 μg mL⁻¹) were added to the cells for a 4-hour incubation. Cells were sequentially stained with LysoTracker Red DND-99 and DAPI. Confocal laser scanning microscopy (CLSM) was used for visualization of the cells.

Animal studies

All the mice were bought from Dashuo Biological Technology Company, Chengdu, China. They were 6-week-old male mice on C57BL6/J background maintained in a specific pathogen-free (SPF) facility and were randomly assigned to groups. Orofacial muscle injury was performed following established protocol.^{[3c, 47]} Mice were anesthetized with isoflurane and administered meloxicam at 1mg kg⁻¹. Hair was removed by depilatory paste. The skin was incised to expose the masseter muscle. To induce regenerative muscle injury, 50 μl barium chloride (BaCl₂, 1.2%) was injected into the masseter muscle. For the freezing injury, a piece of dry ice was then put on the surface of the masseter muscle for 5 seconds, followed by the closure of the incision with 7–0 absorbable sutures. After the sutures are finished, put the mice in the thermally supported cage while monitoring them constantly until all of their righting reflexes have returned. For the nanomaterial treatment, injections of MSN-PEI solution (50 μl, 0.1 mg ml⁻¹) were placed directly into the masseter muscle after the freezing injury procedure. Subsequently, the injection was administered every other day, totaling four administrations. The weight of mice was measured every other day. The study procedures were approved by the Ethical Committee of the West China Hospital of Stomatology, Sichuan University (WCHSIRB-D-2022–467).

For the toxicity test, PEI (50 μl, 0.1 mg ml⁻¹), MSN (50 μl, 0.1 mg ml⁻¹), and MSN-PEI (50 μl, 0.1 mg ml⁻¹) were injected directly into the masseter muscle after the freezing injury follow the pattern described above. Mice were sacrificed at 28 days post-injury by cervical dislocation for tissue, interstitial fluid, and plasma collection.

In vivo imaging

Mice received local injections of FITC-MSN-PEI (1 mg mL⁻¹) into the masseter muscle. At specified timepoints, in vivo imaging was conducted using IMAGING 200 (Raycision Medical Technology, Hefei, China). Following imaging, the mice were sacrificed, and organ fluorescence signals were assessed with the same system.

FAPs isolation and culture

Mice masseter muscles were carefully dissected according to methods previously described.^[47] Muscle pieces were cut in slurries and digested in Collagenase II (700–800 U ml⁻¹) for 60 min at 37°C with gentle shaking. Next, the supernatant was aspirated down to 10 mL, and Collagenase II (1 mL, 1000 U ml⁻¹) along with Dispase (1 mL, 11 U ml⁻¹) were added. The mixture was incubated for an additional 30 min at 37°C with gentle shaking. Digested mixture was passed several times through an 18G needle to fully release the cells, then filtered through a 40-μm cell strainer. Subsequently, plate the cells into a T75 flask for 3 h. After 3-hour plating, FAPs will attach to the flask. The supernatant containing nonadherent cells was then discarded, and new growth medium was added.

Freshly isolated primary mouse FAPs were cultured in growth medium. For fibrogenic differentiation, FAPs were seeded at a density of 7,000 cells cm⁻² and cultured with FAPs differentiation medium (high-glucose DMEM supplemented with 2.5% FBS and 1% P/S) for a 5-day culturing.

Histological and Immunofluorescent staining of mouse muscle samples

Masseter muscle samples from mice were embedded in O.C.T. compound, and frozen in isopentane at −150°C. Cryosections (10 μm) were made and fixed in acetone, then stained for histological analysis. For cell slides, fixation was performed with paraformaldehyde (PFA) before staining. For H&E staining, sections were stained using a staining kit. For immunofluorescence staining, sections were processed at room temperature for 1 h in blocking buffer (5% bovine serum albumin, 5% donkey/goat serum, and 0.3% Triton-X-100) at room temperature, followed by overnight incubation at 4°C with primary antibodies. The following primary antibodies were used: for muscle tissue sections: anti-F4/80 (1:300), anti-CD86 (1:80), anti-CD206 (1:500), anti-laminin (1:600), anti-TLR9 (1:200), anti-TLR7 (1:200), anti-PDGFRα (1:40), anti-embryonic MyHC (1:10). For cell slides: anti-cleaved caspase-3 (1:200), anti-Ki67 (1:200), anti-αSMA (1:750), anti-MyH4 (1:20). Subsequently, sections and slides underwent 1-hour incubation at room temperature with secondary antibodies: DyLight^™ 488 (1:200), Goat Anti-Rat IgG (1:200), IFKine^™ Red Donkey Anti-Rabbit IgG (1:200), IFKine^™ Green Donkey Anti-Rabbit IgG (1:200), IFKine^™ Green Donkey Anti-Goat IgG (1:200), IFKine^™ Red Donkey Anti-Mouse (1:200). Nuclei were visualized with DAPI.

Histological and Immunofluorescent staining of human muscle samples

The patient’s lip muscle tissues were fixed in 4% PFA overnight, then underwent dehydration in 70%, 80%, 90%, 95%, 100%, and 100% ethanol solutions, with 40 min of dehydration at each level of ethanol solution. The tissues were immersed in xylene for 3h and processed for paraffin embedding. 5-μm paraffin sections were made and dewaxed 3 times with xylene. Then, the sections were dehydrated with gradient ethanol. Masson’s trichrome staining kit was used on sections for staining. For immunofluorescent staining, paraffin sections were fully submerged and heated in sodium citrate buffer (10 mM sodium citrate, 0.05% Tween 20, pH 6.0), and cooled naturally for antigen retrieval. Subsequent staining procedures were the same as described for mice. Primary antibodies used for human muscle sample staining: anti-F4/80 (1:300), anti-CD86 (1:80), anti-CD206 (1:500), anti-desmin (1:200), anti-TLR9 (1:200), and anti-TLR7 (1:200). Either enzyme-conjugated or fluorescent secondary antibody was used: IFKine^™ Green Donkey Anti-Rabbit IgG (1:200), DyLight^™ 488 (1:200), Goat Anti-Rat IgG (1:200), and HRP Conjugated Goat anti-Rabbit IgG antibody (1:200). Nuclei were visualized with hematoxylin and DAPI respectively. All stained sections were scanned with Olympus VS200 fluorescence microscope (Olympus Corporation, Tokyo, Japan).

Tyramide signal amplification for immunofluorescent colocalization

Two adjacent frozen sections were stained for Gal3 and F4/80, Fabp4 and Pdgfra, respectively. Tyramide signal amplification technique was applied to enhance the immunohistochemical detection according to the methods previously described.^[48] The steps for section fixation and antigen retrieval are the same as above. After 15 min of permeabilization with TritonX-100 (0.1%), block the sections for 30 min at room temperature using goat serum (10%). Add the first primary antibody F4/80 (1:50), or Pdgfra (1:100), to incubate at 4°C overnight. Wash the sections and incubate them at 37°C for 45 min with Goat anti-Rat IgG horseradish peroxidase (1:100) or Rabbit Anti-Goat IgG horseradish peroxidase (1:100). Wash again and incubate sections at room temperature for 10 min with FITC working solution (1ml FITC in 300 ml, 0.003% H₂O₂). Then, the sections were subjected to a second round of antigen retrieval and block using donkey serum (10%). Add the second primary antibody, Gal3 (1:50), or Fabp4 (1:50), and incubate at 4°C overnight. Wash with TBST and incubate with Alexa Fluor^®594 donkey anti-rat lgG (H+L) (1:400) or Alexa Fluor^®594 donkey anti-rabbit lgG(H+L) (1:600) at 37°C for 45 min. Finally, visualize nuclei with DAPI.

Experimental Assessment of Masticatory Function

Bite force measurements were conducted using a FlexiForce sensor in conjunction with the Economic Load and Force (ELF) system (Tekscan, Boston, USA). Following the established protocol for bite force testing,^[49] mice underwent a fasting period overnight, and their bite force was measured in plastic cages (Figure 3g).

Mastication rate measurement was conducted as described before.^[50] Briefly, mice were fasted overnight before being presented with a single food pellet. The eating behavior was recorded with a video camera and analyzed by two independent reviewers.

Spontaneous eating behavior was assessed by video recording mice over a 15-hour period. As the mice consumed food, the timing and frequency of their eating were analyzed by two independent reviewers.

Cellular DAMP isolation

C2C12 cells (cat. #CRL-1772) and 3T3-L1 cells (cat. #CL-173) were purchased from American Type Culture Collection (ATCC) (Manassas, VA) and cultured following ATCC protocols before harvesting for DAMP collection. 1 × 10⁶ cells were resuspended in cryogenic tubes and underwent two freeze-thaw cycles. Freeze-thaw cycles were performed by putting cryogenic tubes in liquid nitrogen for 10 min and then in 37°C water bath for 1 min.

Co-culture of RAW264.7 and FAPs

For the co-culture experiment, cell culture inserts with 0.4-μm pores were placed in 24-well culture plates, with RAW264.7 cells cultured in the inserts and FAPs cultured separately in the wells. RAW264.7 cells were maintained in high-glucose DMEM supplemented with 2.5% FBS and 1% P/S; FAPs were cultured in their growth medium for attaching and differentiation medium for co-culturing. For the treatment of the insert, RAW264.7 were seeded in the inserts at 21,000 cells cm⁻² for the co-culture with FAPs. After 12 h, RAW264.7 were stimulated with CpG 2006 (1 μg ml⁻¹) and ssRNA (1000 nM) 30 min prior to the administration of MSN-PEI (10 μg ml⁻¹). For the co-culture of RAW264.7 and FAPs, seed FAPs in growth medium at 7,000 cells cm⁻² in 24 well plate and let them adhere to the culture support for 4 h. Switch the growth medium in FAPs culture to the differentiation medium and transfer the inserts containing pretreated RAW264.7 into the FAPs-cultured plates. Proliferation and apoptosis assays were conducted after 24 h of co-culture, while fibrogenic differentiation was assessed after 5 days of co-culture.

RNA Extraction and Quantitative Real-time PCR

RNA from masseter muscle and cells was extracted using TRIzol. RNA concentration and purity were determined via NanoDrop One. The extracted RNA was reverse transcribed to cDNA by PrimeScript^™ FAST RT reagent Kit with gDNA Eraser. qRT-PCR analysis was conducted with TB Green^® Premix Ex Taq^™ (Tli RNaseH Plus) using primers listed in Table S2. Relative expression levels were calculated using the delta-delta CT method.

Interstitial fluid extraction and quantification of cfDNA and cfRNA

The isolated masseter muscle was placed into a 2ml centrifuge tube with a perforated bottom, created using a 21G needle. For interstitial fluid collection, sodium chloride solution (20 μl, 0.9%) was added to the tube, and centrifugated at 100 g for 10 min at 4°C. PicoGreen was used to measure the concentration of cfDNA. NanoDrop One was used to measure the concentration of cfRNA, and Acclaro Sample Intelligence technology was used to calibrate it.

Enzyme-Linked Immunosorbent Assay (ELISA)

Masseter muscle interstitial fluid samples were extracted as described above. TNF-α and IL-10 ELISA kits were used to measure the fluid’s level of TNF-α and IL-10.

Flow Cytometry

Cell suspension was obtained with methods described above in cell isolation and was incubated with anti-CD16/CD32 antibodies to block Fc gamma receptors prior to staining. Cell suspension was incubated with primary antibodies targeting cell surface markers (CD45, Ly6C, CX3CR1) for 40 min at 4°C in FACS buffer at a concentration of approximately 1 × 10⁷ cells per milliliter. Details of antibody clones, sources, and dilutions used in flow cytometry are provided in Table S3. Cell analysis was performed using Attune NxT Flow Cytometers (Thermo Scientific). FlowJo software 10.8.1 was used to process the data. The gating strategies used for flow cytometry plots are shown in Figure S2a (Supporting Information).

Single-cell preparation and analysis

Single-cell isolation from orofacial muscle followed a published procedure.^[51] For scRNA-seq, 8000 live cells per sample were loaded into a Chromium Controller (10x Genomics) using Single Cell 3’ v2/v3 Reagent Kits for GEM formation, cDNA synthesis, and library preparation, and sequenced on an Illumina HiSeq 4000 (~50,000 reads per cell). Sequencing reads were processed with 10x Genomics Cell Ranger 2.1.0 and gene expression matrices were generated using CeleScope v1.15.0, STAR v2.6.1a, and FeatureCounts v2.0.1. Quality control, dimensionality reduction and clustering were conducted with Scanpy v1.8.2, and clusters visualized with Uniform Manifold Approximation and Projection (UMAP),^[52] while batch effects were removed by Harmony v1.0.^[53] Cell types were identified via SynEcoSysTM reference data.^[7c] Macrophages and FAPs were reclustered for detailed analysis.^[54] DecontX was used for RNA contamination correction.^[55] Differentially expressed genes (DEGs) were identified with Scanpy, while Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) were used to explore macrophage and FAPs functions.^[56] GSEA was performed on inflammation and fibrosis-related genes in macrophage clusters.^[57] Cell-cell interaction (CCI) was predicted using Cellphone DB (v4.0.0) version.^[58] Macrophage differentiation trajectory was reconstructed with the Monocle2 v 2.22.0.^[59] Gene set scoring was performed with UCell v 2.2.0.^[60]

Statistical analysis

All values were expressed as mean ± standard error of the mean, with a minimum of 3 biological replicates. All data were tested for normality, for non-normally distributed data, nonparametric tests were performed. For normally distributed data, the Student’s t-test was applied to compare mean values between two groups, while a one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparisons test was used for comparisons among multiple groups. Correlation analyses were performed using the Spearman correlation coefficient (r). P-values of <0.05 were considered statistically significant. All experiments were conducted at least twice, yielding similar results, and representative data are presented in the paper. Statistical analyses were performed using GraphPad Prism 9, with no animals excluded from the analyses. Quantitative analyses of images from the Olympus VS200 microscope were conducted using ImageJ.

Data availability statement:

The data supporting this study’s findings are available from the corresponding author upon reasonable request.