Contribution of HMGB1 to keratinocyte inflammation in recessive dystrophic epidermolysis bullosa

Abstract

Recessive dystrophic epidermolysis bullosa is an inherited skin disorder characterized by fragile skin, blistering, and chronic wounds. Keratinocytes, the primary cells in the epidermis, are directly affected by persistent injury in recessive dystrophic epidermolysis bullosa, contributing to chronic inflammation. HMGB1 (high mobility group box 1) is elevated in the serum of individuals with recessive dystrophic epidermolysis bullosa. However, its role in keratinocyte inflammation remains unclear. In this study, we report an increase in HMGB1 expression in keratinocytes at chronic wound sites compared with that on matched nonwounded skin from an individual with recessive dystrophic epidermolysis bullosa, suggesting a potential link to the upregulation of local proinflammatory stimuli. Pharmacologic inhibition of HMGB1 using inflachromene reduced lipopolysaccharide-induced secretion of proinflammatory cytokines in keratinocytes, supporting a role for keratinocyte-specific HMGB1 in inflammatory response. Surprisingly, deletion of HMGB1 alone or together with its paralog HMGB2 did not suppress the release of proinflammatory cytokines in response to lipopolysaccharide. Furthermore, inflachromene still reduced the secretion of proinflammatory cytokines in HMGB1- and HMGB2-knockout cells. This unexpected discrepancy between genetic deletion and pharmacologic inhibition points to a more complex role for HMGB1 or off-target effects of the compound. These findings suggest that HMGB1 may contribute to proinflammatory signaling in keratinocytes; however, its exact function needs further investigation.

Introduction

Recessive dystrophic epidermolysis bullosa (RDEB) is a rare genetic skin disease caused by genetic alterations in COL7A1, the gene coding for type VII collagen. Type VII collagen is an extracellular basement membrane protein that anchors the epidermis to the underlying dermis and plays a key role in maintaining skin integrity. Biallelic variants in the COL7A1 gene lead to skin fragility, severe blistering at the dermal–epidermal junction, and chronic wound formation (Cianfarani et al, 2017). Although inflammation initially promotes healing, prolonged activation of the inflammatory response by repeated wounding in RDEB further impairs wound resolution (Cianfarani et al, 2017Cianfarani et al, 2017). If not resolved, wound chronicity predisposes tissues to the development of metastatic squamous cell carcinoma (SCC), the leading cause of morbidity and mortality in RDEB (Fine et al, 2009). Wound resolution and prevention of carcinogenesis represent an unmet need in the treatment of RDEB, requiring a better understanding of the local inflammatory response within chronic wounds.

Chronic inflammation emerged as a central feature in the pathogenesis of RDEB after a seminal study of a monozygotic twin pair with discordant RDEB phenotypes (Odorisio et al, 2014). This previously published study demonstrates the importance of inflammatory signaling, specifically through the TGF-β axis, and provides genetic evidence linking the inflammatory response to disease etiology in RDEB (Anderson-Crannage et al, 2023; Riedl et al, 2022). Evaluation of the microenvironment of chronic skin lesions in RDEB reveals the presence of immune cells, suggesting a proinflammatory microenvironment (Anderson-Crannage et al, 2023; Riedl et al, 2022). Transcriptomic analysis of skin from 3 patients with RDEB compared with healthy controls revealed upregulation of genes associated with immune activation, most notably IL-8– and IFN-α–regulated genes (Breitenbach et al, 2015). Proinflammatory cytokines are also elevated in the blood of patients with RDEB compared with that of healthy controls (Esposito et al, 2016). In particular, elevated IL-6 serum levels correlate with disease severity (Esposito et al, 2016). Although the upregulation of proinflammatory signals in RDEB has been clearly demonstrated, cell type–specific contributions to the inflammatory milieu found in the wounds of patients with RDEB is not yet fully understood. As the cell of origin for SCC, a better understanding of the specific role keratinocytes play in creating a proinflammatory and tumor-permissive environment will enhance our understanding of the pathogenesis of RDEB and the development of SCC.

HMGB1 (high mobility group box 1) is ubiquitously expressed across different cell types. In the nucleus, HMGB1 regulates chromatin structure and remodeling, transcription, and DNA replication and repair (Lange and Vasquez, 2009). In response to environmental stresses, HMGB1 is released from the cell and acts as a damage-associated molecular pattern to trigger the innate immune response and recruit inflammatory cells to damaged sites (Kim et al, 2020, 2018; Kwak et al, 2020). HMGB1 can be actively secreted in response to inflammatory stimuli or passively released from apoptotic cells (Nguyen et al, 2017). In patients with RDEB, HMGB1 is elevated locally in wounded skin and systemically in serum relative to those of healthy controls (Hoste et al, 2015; Petrof et al, 2013; Tamai et al, 2011). Elevated levels of HMGB1 protein have been associated with poor outcomes in multiple inflammatory diseases and malignancies, including RDEB and RDEB-associated SCC (Pandolfi et al, 2016). Keratinocyte-specific HMGB1 primes neutrophils to form neutrophil extracellular traps in skin wounds, causing a delay in wound healing in vivo (Hoste et al, 2019). Neutrophil extracellular trap formation promotes wound-induced tumorigenesis through TNF and RIPK1 kinase activity. Murine studies demonstrate increased keratinocyte HMGB1 expression at wound margins and promotion of keratinocyte and fibroblast activation, angiogenesis, and scar formation by HMGB1 (Dardenne et al, 2013). Despite a link between HMGB1 levels, inflammation, and RDEB disease, the molecular mechanisms of HMGB1 production, secretion, and modulation in keratinocytes remain poorly understood.

In this study, we investigated how HMGB1 contributes to proinflammatory responses in keratinocytes. We found that HMGB1 expression is significantly higher in keratinocytes at wounded regions than in matched nonwounded regions of an individual with RDEB. Inhibition of HMGB1 using inflachromene (ICM), an inhibitor that targets HMGB1, reduced lipopolysaccharide (LPS)-induced inflammatory cytokine secretion in a keratinocyte cell line. Surprisingly, ICM treatment was still able to suppress LPS-stimulated cytokine release in HMGB1-knockout (HMGB1^KO) cells. Our findings demonstrate unknown targets of ICM beyond HMGB1 protein and warrant further investigation of HMGB1’s exact function in keratinocyte-specific signaling through the release of proinflammatory cytokines.

Results

RDEB keratinocytes display increased HMGB1 expression

To determine the contribution of HMGB1 to wound signaling in RDEB, we analyzed single-cell RNA-sequencing datasets from wounded and matched nonwounded skin biopsies of a patient with RDEB who had undergone bone marrow transplantation (Riedl et al, 2022). We focused on 4 cell clusters—basal keratinocytes, nonbasal keratinocytes, myofibroblasts, and fibroblasts (Figure 1a)—defined according to the expression of cell cluster–specific genes (details are provided in Materials and Methods). In both basal and nonbasal keratinocytes, HMGB1 expression was significantly higher in wounded than in nonwounded skin (Figure 1b–d). In contrast, HMGB1 expression in fibroblasts was modestly reduced in wounded relative to that in nonwounded skin and remained unchanged in myofibroblasts regardless of wounding (Figure 1e and f). These results are consistent with the observed increase in HMGB1 in lesional skin from patients with RDEB compared with that in normal human skin (Hoste et al, 2015), suggesting a potential link between keratinocyte-specific response to chronic wounding in RDEB and HMGB1 expression.

Figure 1.

Increased HMGB1 expression in keratinocyte cells at a wounded site of a patient with RDEB. (a, b) Single-cell RNA-sequencing analysis of keratinocyte and fibroblast cells from wounded and nonwounded skin obtained from a patient with RDEB. Identification of epidermal cell clusters by gene expression (see methods) is shown in a. Expression of HMGB1 in each cluster is shown in b. (c–f) Violin plots of HMGB1 expression in indicated cell types from wounded and nonwounded skin from a patient with RDEB. Each dot represents HMGB1 expression per nucleus. Statistical analysis was performed using unpaired Student’s t-test. RDEB, recessive dystrophic epidermolysis bullosa.

Pharmacologic inhibition of HMGB1 suppresses LPS-induced secretion of proinflammatory cytokines and chemokines in keratinocytes

The increase in HMGB1 expression in keratinocytes prompted us to investigate how HMGB1 regulates proinflammatory signaling in this cell population. To address this, we treated the human keratinocyte cell line N/TERT-2G cells with LPS for 24 hours in the presence or absence of ICM, an HMGB1 inhibitor. Cell supernatants were collected and assessed for the release of cytokines and chemokines. Of the cytokines and chemokines tested, IL-6, TNFα, GROa (CXCL1), and IL-8 (CXCL8) were significantly elevated upon LPS stimulation and were suppressed by ICM treatment (Figure 2a and Supplementary Table S1). Although PDGF-AA/AB isoforms, monokine-induced by gamma IFN (CXCL9), IL-1a, granulocyte colony-stimulating factor, soluble CD40 ligand, and IL-18 were also significantly induced by LPS, they were not affected by HMGB1 inhibition (Figure 2a and Supplementary Table S1). We independently validated LPS-induced IL-6 release in both N/TERT-2G and a second human keratinocyte cell line, HEK001 (Figure 2b and c), as a marker of proinflammatory cytokine release in keratinocytes. Notably, ICM treatment suppressed LPS-induced IL-6 in both cell lines. Taken together, these results suggest that HMGB1 affects the secretion of a subset of proinflammatory cytokines and chemokines in response to LPS stimulation in keratinocytes.

Figure 2.

Pharmacologic inhibition of HMGB1 suppresses LPS-induced release of proinflammatory cytokines and chemokines in keratinocytes. (a) N/TERT-2G cells were pretreated with 5 μM ICM or DMSO for 3 hours, followed by treatment with either vehicle or 100 μg/ml LPS for an additional 24 hours. Top panel: Heatmap of average fold changes in the indicated cytokines/chemokines after LPS treatment in the presence or absence of ICM based on 3 biological replicates and normalized to DMSO-treated samples. Bottom panels: Representative cytokines/chemokines that were sensitive to ICM treatment are shown (mean ± SEM, n = 3). Adjusted P-values (q-values) <0.05 after multiple testing correction using the Benjamini–Hochberg FDR method are shown. Supplementary Table S1 provides the raw data; log2 fold changes; and statistical results, including q-values. (b, c) ELISA analysis of supernatant IL-6 concentration in (b) N/TERT-2G and (c) HEK001 cells using IL-6 ELISA kit (mean ± SEM, n = 3 biological replicates). Cells were treated with 10 μM (N/TERT-2G) or 5 μM (HEK001) ICM or DMSO for 24 hours prior to vehicle or 20 μg/ml LPS treatment for an additional 24 hours. One-way ANOVA statistical analyses were performed. FDR, false discovery rate; ICM, inflachromene; LPS, lipopolysaccharide.

HMGB1 pharmacologic inhibition but not HMGB1 deletion suppresses LPS-induced IL-6 stimulation in keratinocytes

To further assess the regulation of HMGB1-dependent cytokine release in keratinocytes, we assessed the impact of an HMGB1 gene deletion using the CRISPR/Cas9 strategy. Successful HMGB1 deletion, both in pooled populations and single-cell clones, was confirmed by TIDE (Tracking of Indels by Decomposition) analysis (Brinkman et al, 2014) (Figure 3a). HMGB1 protein levels were confirmed by immunoblotting and immunofluorescence using an HMGB1-specific antibody (Figure 3b and c). Karyotype analyses of HMGB1^KO single clones and an HMGB1^WT clone that had undergone the same CRISPR/Cas9 editing revealed similar karyotypes between HMGB1^WT and HMGB1^KO cells (Figure 3d). Both wild-type and knockout HMGB1 cells had gained 1 copy each of chromosomes 7 and 20 at the end of the clone isolation process.

Figure 3.

HMGB1 regulates IL-6 release at basal conditions. (a–c) Generation of HMGB1^KO N/TERT-2G cells. Validation of HMGB1^KO N/TERT-2G cells by TIDE analysis in a, immunoblot in b, and immunofluorescence in c was performed. (d) Representative karyotype images of HMGB1^WT and HMGB1^KO N/TERT-2G cells. (e) Immunoblot analysis of endogenous and exogenous Flag-HMGB1 expression in wild-type and HMGB1^{KO 1-G5} cells. (f) ELISA analysis of supernatant IL-6 concentration in HMGB1^WT and 2 independent HMGB1^KO cells (denoted by circle and diamond symbols) at basal levels (mean ± SEM, n = 3). TIDE, Tracking of Indels by Decomposition.

To validate the role of HMGB1 in the secretion of keratinocyte proinflammatory cytokines, we first measured IL-6 release in isogenic HMGB1^WT and HMGB1^KO cells. Basal IL-6 release was modestly lower in HMGB1^KO cells than in HMGB1^WT cells (Figure 3e and f). Re-expression of a 3-Flag, N-terminal-tagged HMGB1 (Flag-HMGB1) in HMGB1^KO cells significantly increased IL-6 release relative to HMGB1^KO cells, restoring secretion to a level comparable to that observed in HMGB1^WT cells (Figure 3f). Together, these data support the role of HMGB1 in promoting IL-6 secretion in keratinocytes.

Next, we evaluated the role of HMGB1 under LPS-stimulated conditions. In contrast to the basal state, HMGB1^KO cells exhibited elevated IL-6 release after LPS treatment, comparable to that of HMGB1^WT cells (Figure 4a; compare lanes 2 with 5). Moreover, treatment with ICM still effectively suppressed LPS-induced IL-6 release in both HMGB1^WT and HMGB1^KO cells (Figure 4a; compare lanes 2 with 3 and lanes 5 with 6). These results indicate that the LPS-induced IL-6 release is independent of HMGB1 yet remains sensitive to ICM treatment.

Figure 4.

Pharmacologic inhibition of HMGB1 but not HMGB1 knockout suppresses LPS-induced IL-6 release. (a) Asynchronous cells were pretreated with 5 μM ICM or vehicle for 3 hours, followed by 100 μg/ml LPS or vehicle treatment for an additional 24 hours. Supernatants were collected for IL-6 release analysis by ELISA (mean ± SD, n = 3). (b) Immunoblot analysis of HMGB1 and HMGB2 levels in the indicated cell lines. ICM, inflachromene; LPS, lipopolysaccharide.

The ICM compound inhibits both HMGB1 and HMGB2 (Lee et al, 2014). We hypothesized that HMGB2 may compensate for the loss of HMGB1. To test this, we deleted the HMGB2 gene in both HMGB1^WT and HMGB1^KO cells and confirmed successful knockout by western blot (Figure 4b). In contrast to HMGB1^KO, HMGB2^KO showed a modest reduction in IL-6 secretion compared with LPS-treated HMGB1^WT or HMGB1^KO cells (Figure 4a; lanes 2 and 5 vs lane 7). ICM still significantly suppressed LPS-stimulated IL-6 secretion in HMGB2^KO cells (Figure 4a; lane 9). Similar to HMGB1^KO, HMGB1/2 double knockout cells (HMGB1/2^KO) also maintained IL-6 responsiveness upon LPS treatment, and ICM continued to suppress this response (Figure 4a). Taken together, our results demonstrate that ICM treatment but not genetic deletion of HMGB1 or HMGB2 suppresses LPS-induced IL-6 release, raising the intriguing possibility that ICM may target additional factors beyond HMGB1/2 proteins to regulate proinflammatory cytokine release in keratinocytes.

Discussion

In this study, we investigated the role of HMGB1 in regulating IL-6 secretion in keratinocytes, aiming to clarify its function in the context of RDEB. Single-cell RNA sequencing of wounded and nonwounded skin regions from a patient with RDEB revealed increased HMGB1 expression, specifically in keratinocytes, suggesting a role in RDEB-associated skin injury and innate immune activation (Riedl et al, 2022). By profiling a panel of cytokines and chemokines, we found that pharmacologic inhibition using ICM suppressed a subset of LPS-stimulated proinflammatory cytokine/chemokine release, including IL-6, IL-8 (CXCL8), GROα (CXCL1), and TNFα. These factors have been previously shown to be upregulated in RDEB and play critical roles in the acute phase of wound healing, including recruitment of leukocytes and stimulation of keratinocyte migration, proliferation, and epithelial-to-mesenchymal transition (Alexeev et al, 2017; Anderson-Crannage et al, 2023; Engelhardt et al, 1998). IL-6 is predominantly featured as a biomarker of disease severity in RDEB, correlating with wound body surface area (Karakioulaki et al, 2026; Reimer-Taschenbrecker et al, 2025). Our findings suggest that ICM selectively mediates the release of certain proinflammatory cytokines in keratinocytes and may have therapeutic value in treatment of RDEB. However, the mechanism underlying its function is unclear, as HMGB1^KO keratinocyte cells retained a functional proinflammatory cytokine response to LPS stimulation that could be suppressed by ICM. The discrepancy between pharmacologic inhibition and genetic deletion warrants further investigation to precisely define the role of HMGB1 in keratinocyte-specific proinflammatory cytokine secretion.

Although HMGB1 levels are elevated in both the serum and wounded skin regions of patients with RDEB with chronic blistering (Hoste et al, 2015; Petrof et al, 2013; Tamai et al, 2011), its molecular function in keratinocytes remains unclear. A previous study showed that keratinocyte-specific deletion of HMGB1 delayed wound healing and promoted SCC tumorigenesis in vivo (Hoste et al, 2019), highlighting a potential role for HMGB1 in the progression from chronic wounds to carcinogenesis. Importantly, this did not apply to myeloid cells. Our observed increase in HMGB1 expression in keratinocytes at wounded sites in a patient with RDEB also suggests a keratinocyte-specific, intrinsic proinflammatory response that may contribute to disease pathogenesis. In addition to this keratinocyte-specific response, single-cell RNA sequencing of the same patient also identified distinct immune cell populations within the wounded microenvironment (Riedl et al, 2022), indicating that both intrinsic (keratinocyte-driven) and extrinsic (immune cell–derived) inflammatory signals may contribute to RDEB pathogenesis. Future studies investigating the interplay between these inflammatory components will be critical for developing new therapeutic strategies.

HMGB1 is constitutively expressed across many cell types. Although its role in immune cell–mediated inflammation is well-established, its function in keratinocytes is less well-understood. Our finding that HMGB1 expression is upregulated in keratinocytes from wounded skin compared with that from nonwounded skin suggests a potential role for both context- and cell type–specific HMGB1 function in RDEB disease etiology and warrants further investigation. Owing to the higher expression of HMGB1 in fibroblast cells, the differential HMGB1 expression may not be as drastic compared with keratinocyte cells. Although ICM has been used as an HMGB1 inhibitor, we observed that it suppresses the secretion of proinflammatory cytokines even in HMGB1/2^KO cells, implying that ICM may target additional inflammatory mediators beyond HMGB1 and HMGB2. A recent study identified that ICM also targets KEAP1 (Kelch-like ECH-associated protein 1), a negative regulator of NRF2 in an antioxidant response pathway (Yim et al, 2024). It will be interesting to determine how KEAP1 contributes to a keratinocyte-specific cytokine release in response to inflammatory signals. Identifying ICM-specific targets in keratinocytes may open new therapeutic strategies to dampen inflammation as a potential treatment for patients with RDEB. Finally, the HMGB1^KO and HMGB2^KO keratinocyte lines developed in this study provide a valuable platform for dissecting keratinocyte-specific HMGB1/2-mediated inflammatory signaling and validating future HMGB1/2-targeted pharmacologic inhibitors.

Materials and Methods

Cell lines

N/TERT-2G keratinocyte cells were obtained from James G. Rheinwald through the laboratory of Ellen van den Bogaard at Radboud University Medical Centre (Nijmegen, The Netherlands) (Dickson et al, 2000; Smits et al, 2017). HEK001 human epidermal keratinocyte cells were obtained from ATCC (CRL-2404) (Sugerman and Bigby, 2000). Both cell lines were grown in Keratinocyte Serum-Free Medium (Gibco, 17005-042) with 50 units/ml penicillin and 50 μg/ml streptomycin (Gibco, 15070-063). N/TERT-2G cell media were supplemented with epithelial GF (Gibco, 10450-013, 0.2 ng/ml), bovine pituitary extract (Gibco, 13028-014, 25 μg/ml), and 0.4 mM final concentration of calcium chloride (Teknova, C0477). HEK001 cell media were supplemented with epithelial GF (Gibco, 10450-013, 5 ng/ml). N/TERT-2G cells were passaged using TrypLE Express Enzyme (Gibco, 12604-210), and HEK001 cells were passaged using 0.25% Trypsin-EDTA (Gibco, 25200-056). Cells were pelleted at 300–350g for 5 minutes, resuspended into a single-cell suspension, and counted using a Vi-Cell BLU cell viability analyzer/automated cell counter. Cells were replated at ∼10,000–17,000 cells/cm² to maintain optimal cell growth. Media were changed every 2 days, and cells were passaged every 3–4 days (<70% confluent) to maintain cells in logarithmic growth phase. Mycoplasma detection was performed every 6–12 months using a commercially available detection kit (EZ-PCR Mycoplasma Detection Kit, Sartorius 20-700-20, or Venor GeM Mycoplasma Detection Kit, MilliporeSigma, MP0025).

For Flag-HMGB1-expressing cells, supernatant containing virions was collected 72 hours after transfection and added to N/TERT-2G HMGB1^KO cells in the presence of polybrene (10 μg/ml, Millipore, TR-1003-G) by the spinoculation method. Cells expressing Flag-HMGB1 were selected in media containing 16 μg/ml blasticidin (A1113903) for 4 days. Flag-HMGB1 expression was confirmed by immunoblotting. Flag-HMGB1-expressing cells were maintained in media containing 4 μg/ml blasticidin.

Generation of knockout cell lines by CRISPR/Cas9

Synthetic single-guide RNAs targeting HMGB1 (GAUACUCACGGAGGCCUCUU) and HMGB2 (AAAAAUUACGUUCCUCCCAA) genes were designed using Synthego CRISPR design tool. Single-guide RNAs and Cas9 mRNA were purchased from Synthego. Cas9 mRNA and single-guide RNA were nucleofected into low-passage N/TERT-2G cells by electroporation (Neon transfection system) using optimized settings for N/TERT-2G cells (1500 V, 10 ms, 3 pulses) and for HEK001 cells (1400 V, 20 ms, 2 pulses). To increase cell survival during single-cell cloning, 96-well tissue culture plates were coated with VitroCol type I collagen (Advanced Biomatrix, 5007), according to the manufacturer’s instructions. Cells were plated at 1–5 cells per well in “conditioned media” (media from cells cultured 24–48 hours at optimal density in logarithmic growth phase, sterile filtered, and mixed 1:1 with fresh media). Cell stocks of edited clones were frozen in freezing media (50% Keratinocyte Serum-Free Medium, 20% DMEM/Ham’s F12, 20% fetal bovine serum, 10% DMSO).

To determine the efficiency of CRISPR-induced insertions and deletions, genomic regions flanking CRISPR/Cas9 cut sites were amplified by PCR and sequenced by Sanger sequencing. Primers used to screen HMGB1-KO at exon 3 are forward 5′-ATTCAGAGCAGACTCGGGCGGA-3′ and reverse 5′-TGTGATGCATTGGACAGGGTGC-3′. The resulting amplicons are analyzed using a decomposition algorithm called TIDE that identifies insertions and deletions present in the cell population given a known cut site. P-values generated during TIDE analysis were used to determine the significance of the detected insertions and deletions (Brinkman et al, 2014).

HMGB1 cDNA design and lentiviral production

The wild-type HMGB1 coding sequence with 3xFlag-N-terminal tag (HMGB1-Flag) flanked by attB1/2 sequences was synthesized and cloned into pUC57 cloning vector by Gene Universal. The cDNA was serially cloned into pDONOR221 and subsequently swapped into pLenti6.2/V5-DEST Lentiviral expression vector by Gateway cloning. Successful cloning of the 3xFlag-HMGB1 cDNA insert was confirmed by Sanger sequencing.

For lentiviral production, human epidermal keratinocyte 293T cells were cotransfected with 10 μg of each plasmid—pLenti6.2-Flag-HMGB1, pCMV-dR8.2 dVPR (Addgene, number 8455), and pCMV-VSV-G (Addgene, number 8454)—using the calcium phosphate-mediated ProFection Mammalian Transfection System (Promega, E1200) for 72 hours.

Karyotyping

N/TERT-2G cell lines were verified approximately every year by karyotyping. After 3.0 colcemid treatment, 20 metaphase spreads were analyzed, and karyotypes were generated according to standard cytogenetic protocol.

Cell plating and treatments

Logarithmically growing cells were used for all experiments. A total of 75–100 × 10³ cells/well were plated in a 24-well plate in Keratinocyte Serum-Free Medium. After allowing cell attachment overnight, cells were replenished with new Keratinocyte Serum-Free Medium and pretreated with either 5 μM (HEK001) or 10 μM (N/TERT-2G) of ICM (Sigma-Aldrich, 533060, Cayman Chemicals, 17006) or DMSO followed by 100 μg/ml LPS from E Coli strain O55:B5 (Sigma-Aldrich, L6529) or O111:B4 (Sigma-Aldrich, L2630; lot 0000369272) for 24 hours prior to collection. Different durations of ICM treatment were noted for specific experiments.

ELISA

Supernatants were collected, centrifuged at >10,000g, and used immediately (or stored at −80 °C for ≤3 months). The ELISA kit was purchased commercially (ELISA MAX Deluxe Set Human IL-6, BioLegend, 430516). The protocol was followed according to the manufacturer’s instructions except for IL-6 protein standards, which were resuspended in cell culture media, and supernatants were incubated overnight at 4 °C to optimize signal. Supernatant samples were run in duplicate and read using an M1000 (Tecan). Analysis was performed in GraphPad Prism using a second-order polynomial (quadratic) nonlinear regression to interpolate a standard curve.

The 71-Plex array of human cytokine/chemokine as shown in Figure 2 was performed by Eve Technologies (catalog number HD71). Supernatants were collected in biological triplicates. Raw cytokines/chemokines calculations are reported in Supplementary Table S1. For Figure 2a, the log₂ fold-change difference in LPS treatment alone or combined with ICM was normalized to respective vehicle-treated samples.

Immunoblots

Whole-cell lysates were prepared in 1% SDS lysis buffer or radioimmunoprecipitation assay buffer, separated on Bolt 4–12% Bis-Tris gradient gels (Bio-Rad Laboratories, NW04122), and transferred onto 0.45 μM polyvinylidene fluoride membranes. Clarity Western enhanced chemiluminescence (Bio-Rad Laboratories, 1705060) or WesternBright Quantum (VWR International, 103254-878) chemiluminescent substrate was used to detect bound alkaline phosphatase. Signal was detected using a ChemiDoc Imaging System (Bio-Rad Laboratories). Antibodies used were HMGB1 (EPR3507) (Abcam, number ab79823), HMGB2 (D1P9V, Cell Signaling Technology, number 14163), FLAG M2 (Sigma-Aldrich, number F1804), α-tubulin (DM1A, Sigma-Aldrich, number T9026), Ku70 (Genetex, number GTX70271), and antimouse (VWR International, 102646-160), and antirabbit (VWR International, 102645-182) IgG F(c) γ Goat Polyclonal Antibody-conjugated horseradish peroxidase.

Immunofluorescence

Cells were cultured on coverslides and fixed using 3% sucrose/2% paraformaldehyde for 15 minutes at room temperature and subsequently permeabilized using 0.25% Triton-X100 for 5 minutes on ice. Slides were incubated in blocking buffer (10% milk/3% BSA in PBS containing 0.1% Triton-X100) for 1 hour at room temperature prior to staining overnight at 4 °C with HMGB1 antibody (EPR3507, Abcam, number ab79823) diluted 1:500 in blocking buffer. Slides were washed 3 times with PBS containing 0.1% Tritan-X100 pre and postsecondary antibody staining for 1 hour at room temperature using Alexa Fluor 488 Donkey Anti-Rabbit IgG (H+L) (Jackson ImmunoResearch, 711-545-152) diluted 1:1000 in blocking buffer. Nuclei were counterstained using DAPI (Sigma-Aldrich, D9542) and mounted using ProLong Gold Antifade Mountant (Invitrogen, P36930) prior to imaging using the ×60 objective on a Leica DMi8 fluorescence microscope.

Statistical analysis

Data were analyzed by t-test (for comparison of 2 conditions), 1-way ANOVA (for single-variable datasets with more than 2 conditions), or 2-way ANOVA (for multiple-variable datasets) followed by Tukey's Honest Significant Difference post hoc test to adjust for multiple comparisons in ELISA datasets. P < .05 was considered statistically significant for all ELISA analyses.

Cytokine concentrations from the 71-plex array were analyzed using linear modeling implemented in the limma package (version 3.64.3) in R (version 4.5.3) (Ritchie et al, 2015; Smyth, 2004). Missing or extrapolated values were excluded, resulting in 49 analytes being retained for downstream analysis. Raw concentration values were log₂-transformed to stabilize variance and approximate normality. Linear models were fit independently for each cytokine using a design matrix specifying experimental condition (DMSO, LPS, and LPS + ICM). Differential cytokine and chemokine release was assessed using predefined contrasts comparing LPS with DMSO, LPS + ICM with LPS, and LPS + ICM with DMSO. Log₂ fold changes were calculated as the differences in mean log₂-transformed expression between groups. Standard errors were moderated using empirical Bayes shrinkage implemented through the eBayes function in limma. P-values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate method, and adjusted P-values (q-values) <0.05 were considered statistically significant.

Single-cell RNA-sequencing analysis

Single-cell RNA-sequencing dataset was previously reported (Riedl et al, 2022). The workflow for obtaining single-cell RNA-sequencing gene expression results was implemented using R, version 4.4.2 (R Foundation for Statistical Computing), and the Seurat package, version 5.2.1 (the Satija Lab at the New York Genome Center; Hao et al, 2024). Initial quality control filtering was applied to each single-cell dataset, retaining only cells with >200 and <2500 detected RNA features (genes) and mitochondrial gene expression percentages <20%. After filtering, each dataset underwent a standardized preprocessing pipeline, including data normalization, identification of highly variable features, scaling, and dimensionality reduction through principal component analysis with 19 principal components. Subsequently, clustering was performed using a resolution of 0.7 (Riedl et al, 2022), and cells were visualized through Uniform Manifold Approximation and Projection on the basis of these principal components. Cell identities were determined manually on the basis of established marker gene sets: keratinocytes were subdivided into basal and nonbasal populations, distinguished by differential expression of keratin and epidermal differentiation markers (basal keratinocyte: K15+, K1+, KDAP+, K10+, S100A7+; nonbasal keratinocyte: K15−, K1+, KDAP+, K10+, S100A7+). Fibroblasts were identified by the expression of extracellular matrix and mesenchymal markers ("COL1A1," "COL1A2," "COL3A1," "COL6A1," "COL6A2," "VIM," "DCN," "LUM," "FBLN1," "PDGFRA") (Solé-Boldo et al, 2020), whereas myofibroblasts were defined through expression of smooth muscle–related and matrix genes ("ACTA2," "TAGLN," "MYH11," "MYL9," "CNN1," "COL1A1," "COL3A1," "PDGFRB") (Buono et al, 2023; Guerrero-Juarez et al, 2019; Gur et al, 2022; Valenzi et al, 2019).

Ethics Statement

In accordance with the Declaration of Helsinki, written informed consent was obtained from subjects or guardians in case of minors for molecular analysis of skin biopsies on a University of Minnesota Institutional Review Board protocol (ClinicalTrial.gov NCT02670837).

Data Availability Statement

All relevant data generated or analyzed during this study are included in the published article. Further information and requests for resources and reagents should be directed to HDN (hdnguyen@umn.edu).

ORCIDs

Anja-Katrin Bielinsky: http://orcid.org/0000-0003-1783-619X

Kacey Guenther Bui: http://orcid.org/0000-0002-7384-8286

Ya-Chu Chang: http://orcid.org/0000-0002-2058-3827

Wannasiri Chiraphapphaiboon: http://orcid.org/0000-0002-4244-4895

Christen L. Ebens: http://orcid.org/0000-0003-2430-911X

Hai Dang Nguyen: http://orcid.org/0000-0002-4200-8778

Jakub Tolar: http://orcid.org/0000-0002-0957-4380

Jianfeng Wang: http://orcid.org/0009-0005-8483-0654

Conflict of Interest

The authors state no conflict of interest.

accepted manuscript published online XXX; corrected proof published online XXX