Inflammation drives TGFβ1 activation via the αvβ6 integrin-mechanotransduction pathway in human skin
Xuewei Jiang, Sihem Sellami, Jérémy Kessler, Michael Bachmann, Fanny Noulet, Barbara Russo, Nicolo C. Brembilla, Guerkan Kaya, Andrei I. Ivanov, Maria S. Shutova, Bernhard Wehrle-Haller, Wolf-Henning Boehncke
TL;DR
Inflammation in human skin activates TGFβ1 through the αvβ6 integrin pathway, but the effect is limited to nearby cells and involves a mechanical feedback loop.
Contribution
The study reveals a mechanosensitive feedback loop linking inflammation, TGFβ1 activation, and keratinocyte signaling in human skin.
Findings
Inflammation increases αvβ6 integrin and TGFβ1 latency-associated peptide binding.
TGFβ1 activation remains paracrine in psoriatic keratinocytes.
A mechanodependent feedback loop enhances TGFβ1 signaling in the epidermis.
Abstract
Transforming growth factor β (TGFβ) is a critical regulator of skin homeostasis and inflammation, including psoriasis. Multiple mechanisms can regulate TGFβ signaling, such as the activation of latent TGFβ1 through integrin-dependent pathways. We investigated the molecular mechanisms of TGFβ1 activation during cutaneous inflammation using reconstructed human epidermis treated with inflammatory cytokine cocktail. We report αvβ6 integrin upregulation and enhanced binding of TGFβ1 latency-associated peptide (LAP), which, however, did not result in active TGFβ1 release, restricting the TGFβ1 activation to paracrine signaling in keratinocytes. Further, we identify a mechanosensitive positive feedback loop, that occurs downstream of inflammatory stimuli and involves TGFβ-mediated activation of mechanotransduction, further promoting mechanodependent, αvβ6-mediated TGFβ1 activation in…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsCell Adhesion Molecules Research · Hair Growth and Disorders · Skin and Cellular Biology Research
Introduction
Psoriasis, a chronic inflammatory skin disease affecting 2–3% of the population, represents a state of dysregulated epidermal homeostasis, where continuous interactions between immune cells and keratinocytes drive persistent inflammation. Characterized by hyperproliferation and aberrant differentiation of keratinocytes, psoriasis manifests as red, scaly plaques, severely impacting patients’ quality of life.1^,^2^,^3
Psoriatic inflammation is driven by a self-sustaining loop, where T helper 17 (Th17) cells release inflammatory mediators, disrupting homeostasis of keratinocytes, which in turn amplify the immune response, sustaining chronic inflammation. A less explored but pivotal player in this cascade is transforming growth factor β (TGFβ), known for its paradoxical roles in immune suppression and inflammation, regulating Th17 and regulatory T cell (Treg) differentiation in a context-dependent manner.4^,^5^,^6^,^7^,^8 Given its dysregulation in many pathologies including psoriasis,9^,^10^,^11 TGFβ signaling is particularly relevant to the Th17/Treg imbalance in this skin disorder, where it has been proposed to drive the differentiation of interleukin (IL)-17A-producing Th17 cells, a process crucial for psoriasis development.12
Increased levels of TGFβ1 have been reported in skin of psoriasis patients.13^,^14^,^15^,^16 Similarly, transgenic mice overexpressing TGFβ1 in the epidermis spontaneously develop psoriasis-like disease.17^,^18 Increasing evidence indicates that epidermal keratinocytes take an active role in the regulated production of TGFβ1 alongside immune cells and dermal fibroblasts; however, the underlying mechanisms of the epidermal regulation of TGFβ signaling during inflammation are not well explored. It has been proposed that at the disease onset, TGFβ1 expression in psoriatic keratinocytes increases in response to T cell-derived IL-26.12^,^19^,^20 At the same time, the expression of two other isoforms, TGFβ2 and TGFβ3, does not appear to be upregulated in lesional psoriatic skin (Skin Science Foundation bioinformatics hub).13^,^21 This implies that TGFβ1 may act as a disease driver rather than a suppressor in the psoriatic microenvironment.
Whereas the transcriptional regulation is a major mechanism for production of functional cytokines, the regulation of TGFβ release is more intricate. TGFβ is secreted from the cells as a latent inactive form, non-covalently bound to the latency-associated peptide (LAP) forming the small latent complex (SLC). SLC can be further bound to cell membrane-anchored glycoprotein A repetitions predominant (GARP)22 or to latent TGFβ-binding proteins (LTBPs) and, further, to fibrillin-containing extracellular matrix (ECM), allowing for a build-up of a large stock of inactive TGFβ in the tissue.23
A major mechanism for TGFβ1 activation implies the binding of the RGD peptide-containing domain of LAP1 to adhesion receptors αvβ6 and αvβ8 integrins, which break the non-covalent interactions between LAP and TGFβ, allowing TGFβ release and subsequent binding to its receptors on the cell plasma membrane.24^,^25^,^26 A key feature of αvβ6-dependent TGFβ1 activation is the requirement of intra- or extracellular force applied to the integrin-SLC complex.27 In contrast, αvβ8 integrin immobilizes latent TGFβ at the cell surface and supports cleavage by membrane-associated proteases28 or a direct interaction with TGFβ receptor.29^,^30
Epithelial cells typically express integrin αvβ6, whereas αvβ8 was also reported in distinct keratinocyte subsets in mice.31 Depletion of αvβ6 and αvβ8 integrins in mice recapitulates TGFβ1-null phenotypes,32 further underlying the essential role of integrins in TGFβ1 activation. Increased expression of TGFβ-activating integrins has been observed in many inflammatory conditions, such as liver and pulmonary fibrosis, inflammatory bowel disease, and various cancers,33^,^34^,^35^,^36 suggesting that latent TGFβ expression is not the primary determinant to control TGFβ availability in tissues.6
TGFβ initiates canonical Smad signaling pathways and several Smad-independent pathways, such as MAPK, PI3K/Akt, NF-κB, Ras-, Rac-, Rho-GTPases, which are closely interconnected with mechanosignaling.37^,^38 Accordingly, TGFβ can induce partial epithelial-to-mesenchymal transition and wound-healing phenotype in keratinocytes, characteristic for psoriasis pathogenesis.23^,^39
The altered intracellular or extracellular mechanics can in principle become a key trigger for fine-tuning of the TGFβ signaling through αvβ6-dependent activation. At the same time, many immunological studies do not account for the TGFβ activation process, and little is known on how the mechanobiology of cells can affect the outcomes of inflammatory processes. Currently, an interplay between inflammatory mediators and TGFβ signaling in inflamed tissues is poorly understood and represents a major knowledge gap.
Our previous work demonstrated that inflammatory keratinocytes exhibit increased intracellular contractility and mechanotransduction signaling.40 Here, we propose that these mechanical alterations drive TGFβ1 activation via αvβ6 integrin and thereby trigger TGFβ1-dependent signaling cascades. We explore the modulation of the integrin function in human keratinocytes and Reconstructed Human Epidermis (RHE) by inducing a psoriatic phenotype in a cytokine stimulation model.41 We uncover a specific molecular mechanism for the aberrant regulation of the αvβ6 integrin-dependent TGFβ1 activation in human epidermis under psoriasis-like conditions that involves a mechanodependent positive feedback loop. Furthermore, the keratinocytes “keep it to themselves”: this process does not result in the increase of the active TGFβ in the extracellular space that could be eventually available to the T cells infiltrating the epidermis, suggesting that there are alternative, αvβ8 integrin-dependent mechanisms for TGFβ signaling activation in immune cells.30^,^42^,^43^,^44
Results
TGFβ signaling is upregulated in human psoriatic epidermis
To assess the status of TGFβ signaling in psoriatic epidermis, we first examined TGFB1 mRNA expression in human skin. Our analysis revealed that TGFB1 expression was elevated in lesional epidermis compared to healthy controls (Figures 1A and 1B). This finding was further supported by microarray data from the publicly available dataset in the Skin Science Foundation bioinformatics hub (Figure 1C). Next, we analyzed the single-cell RNA sequencing data from,45 which confirmed that TGFB1 mRNA expression, but not that of TGFB2 and TGFB3, was elevated in the keratinocyte cluster in psoriatic skin (Figure 1D), which aligns with recently published data.12Figure 1TGFβ signaling in psoriatic epidermis(A) TGFB1 mRNA expression in the skin revealed by RNAscope. Collagen VII marks the basement membrane. Scale bars, 50 μm.(B) Quantification of RNA signal from (A). One data point corresponds to one donor. Mean ± SD. Welch’s t test.(C) Microarray data on TGFB1, TGFB2, and TGFB3 expression in patient skin (source: Skin Science Foundations Bioinformatics Hub database). Box-and-whiskers plots with median (center line), interquartile range (box), and 1.5 IQR (whiskers). One data point corresponds to one biopsy. The data on lesional psoriatic skin are circled in red. Abbreviations: Hidradenitis suppurativa (HS), Lichen planus (LP), atopic dermatitis (AD).(D) UMAP, annotated data used from.45 Keratinocyte cluster (KC) is circled in red. Other major clusters are labeled as Fib (fibroblasts), VE (vascular endothelium), APC (antigen-presenting cells), LE (lymphatic endothelium).(E) Immunofluorescence staining of pSmad2 in human skin. Scale bars, 100 μm.(F) pSmad2 nuclear intensity in indicated conditions, from (E). One data point corresponds to one donor. Mean ± SD. Unpaired Student’s t test, ∗p < 0.05.
To determine whether this increase in TGFB1 mRNA expression translates into upregulated TGFβ signaling, we analyzed the levels of phosphorylated Smad2 (pSmad2), a downstream effector of TGFβ signaling, in patients’ skin. Immunostaining revealed that a higher number of nuclei became positive for pSmad2 staining in lesional epidermis compared with healthy controls (Figure 1E) and there was a trend for increased nuclear signal intensity (Figure 1F). Thus, we conclude that TGFβ1 signaling is activated in lesional psoriatic epidermis, in agreement with previous reports.
M5 cytokines activate TGFβ/Smad2 signaling in the reconstructed human epidermis
To determine whether TGFβ signaling can be induced in our in vitro system, we generated RHE using N/TERT-1 immortalized normal human keratinocytes.46^,^47 The cultures were then stimulated with a specific inflammatory cytokine cocktail M5 (containing IL-17A, IL-22, Oncostatin M, TNFα, and IL-1α) as previously described40^,^41 (Figure 2A). We detected a slight but not significant increase of TGFB1 mRNA expression in monolayer of undifferentiated N/TERT-1 and in RHE cultures (Figure 2B). At the same time, Smad2 phosphorylation levels were significantly elevated after 24 h of M5 treatment (Figures 2C and 2D), indicating activation of the TGFβ/Smad2 signaling pathway. Thus, our 3D epidermis model successfully recapitulates the activation of TGFβ/Smad2 signaling observed in clinical psoriatic samples.Figure 2. Regulation of TGFβ signaling and TGFβ-activating integrins in skin inflammation(A) Experimental model of in vitro Reconstructed Human Epidermis (RHE).(B) TGFB1 mRNA expression relative to GAPDH in cultures stimulated with M5 overnight, fold change compared with untreated control. One data point corresponds to one independent experiment. Mean ± SD. Paired Student’s t test.(C) Immunofluorescence staining of pSmad2 in RHE cultures stimulated with M5 for 24 h. Scale bars, 50 μm.(D) pSmad2 nuclear intensity from (C). One data point corresponds to one independent experiment. Mean ± SD. Paired Student’s t test, ∗p < 0.05.(E) Microarray data on the integrin expression in patient skin (source: Skin Science Foundations Bioinformatics Hub database). Box-and-whiskers plots with median (center line), interquartile range (box), and 1.5 IQR (whiskers). One data point corresponds to one biopsy. The data for lesional psoriatic skin are circled in red.(F) Immunofluorescence staining of αv integrin and αvβ6 complex in human skin. Scale bars, 100 μm.(G) Protein expression of integrins in RHEs stimulated with M5 for 48 h, analyzed by western blotting.(H) Quantification of protein expression from (G). One data point corresponds to one independent experiment. Mean ± SD. One-sample t test, ∗p < 0.05.(I) Integrin surface expression in cells stimulated with M5 for 24 h, analyzed by flow cytometry. One data point corresponds to one independent experiment. Geometric mean ± SD. One-sample t test, ∗p < 0.05.(J) Immunofluorescence staining of αvβ6 complex in monolayer cultures stimulated with M5 for 24 h. Scale bars, 50 μm.(K) Immunofluorescence staining of αvβ6 integrin complex in RHE cultures stimulated with M5 for 48 h. Scale bars, 100 μm.
Integrin αvβ6 is upregulated in keratinocytes under inflammatory conditions
Since TGFβ is secreted as a latent protein, we next investigated whether inflammation-induced TGFβ signaling upregulation could be driven by enhanced TGFβ activation. To explore this, we analyzed the expression of αvβ6 and αvβ8 integrins, which are known to activate TGFβ1.
Bulk skin microarray data showed a global upregulation of ITGB6, but not ITGAV or ITGB8 in psoriatic skin (Figure 2E). Consistently, we found an upregulation of protein expression of αv integrin and αvβ6 complex in the epidermis of psoriatic skin (Figure 2F).
Based on bulk RNA-sequencing analysis of RHE under control conditions (Database: GEO accession number [GSE298207](GSE298207)), ITGB6 expression averaged 11′978 ± 8′540 raw counts, whereas ITGB8 averaged only 1′983 ± 830, indicating that in keratinocytes, β6 integrin is predominantly expressed, while β8 is expressed at significantly lower levels. In our experimental system, the RHEs stimulated with M5 inflammatory cytokines showed a moderate but significant increase in integrin β6 protein expression as detected by Western blotting (Figures 2G and 2H). Flow cytometry (Flow Cyt) further confirmed that M5 stimulation enhanced surface expression of αv and β6 integrins (Figure 2I). Immunofluorescence analysis of M5-stimulated keratinocyte cultures also revealed an increase in αvβ6 (Figures 2J and 2K). Interestingly, αvβ6 integrin in keratinocytes did not localize at focal adhesions as is typical for many other integrins, but rather at the cell-cell borders, intercalating between adherens junctions (Figures S1A and S1B), consistent with previous findings.48
Inflammation increases LAP binding by αvβ6 integrin complex
To evaluate the functional consequences of altered integrin expression, we assessed the ability of keratinocytes to bind synthetic LAP1-GFP-IgG, henceforth referred to as LAP1-GFP (Figure 3A).49 LAP1-GFP mimics the interaction between the full-length SLC of LAP1-TGFβ1 and the integrins, making it a valuable tool for detecting integrin-mediated binding in various assays. The binding of synthetic LAP1-GFP by the integrin receptors does not release any TGFβ, allowing to isolate the integrin-binding step from downstream TGFβ1 activation events.Figure 3. Regulation of LAP complex binding by keratinocytes under M5 stimulation(A) LAP1-GFP construct.49(B) Immunofluorescence staining with anti-GFP antibody revealing the LAP1-GFP binding in RHE cultures stimulated with M5 for 24 h. Scale bars 100 μm.(C) LAP1-GFP fluorescence intensity quantification for indicated conditions in RHE cultures, from (B). One data point corresponds to one independent experiment. Mean fold change ±SD. One-sample t test.(D) LAP1-GFP fluorescence quantification in monolayer cultures, mean intensity per cell. One data point corresponds to one independent experiment. Mean fold change ±SD. One-sample t test, ∗∗p < 0.01.(E) LAP1-GFP surface binding by the cells in suspension. Fluorescence intensity fold change from control, assessed by Flow Cyt. M5 stimulation was performed for 24 h. One data point corresponds to one independent experiment. Geometric mean ± SD. One-sample t test, ∗∗p < 0.01.(F) Fluorescence quantification of LAP1-GFP bingeing in monolayer cultures stimulated with M5 for 24 h, in the presence of β6 and β8 minibinders. One data point corresponds to one independent experiment. Mean ± SD. Paired Student’s t test, ∗p < 0.05, ∗∗p < 0.01.(G) Scheme of the GARP and LAP1-TGFβ1 construct expression by BOSC-23 cells, presented to keratinocytes.(H) Release of active TGFβ1 by M5-stimulated N/TERT-1 keratinocytes in the presence of integrin β6 and integrin β8 minibinders, detected by TGFβ-reporter cell line MFB-F11 and quantified by chemiluminescent assay. Activation values were normalized to the respective cell-surface levels of GARP-presented TGFβ, quantified by Flow Cyt. A negative control (N/TERT-1 + GARP without TGFβ ± minibinders) showed no change relative to baseline. One data point corresponds to one independent experiment. Mean ± SD. two-way ANOVA, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.(I) Immunofluorescence staining for ALFA-tag (LAP1-TGFβ1 expression) from a BOSC-23 cell, revealing the uptake of latent TGFβ by M5-and minibinder-treated keratinocytes. Scale bars, 50 μm.(J) Fluorescent particle number per contacting BOSC-23 cell in indicated conditions, from (I). One data point corresponds to one cell; one independent experiment of three is shown. Mean ± SD. Unpaired Student’s t test, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.(K) Fluorescent particle size per condition, from (I). Median ± 95% CI. One data point corresponds to one particle (individual data not shown); one independent experiment of three is shown. Mann-Whitney test, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.
We found that the soluble LAP1-GFP added to the culture medium showed a 1.7-fold increase in binding by RHEs under M5-stimulated conditions compared to untreated control (Figures 3B and 3C). Not differentiated keratinocytes showed a 2-fold increase in LAP1-GFP binding in monolayer cultures (Figure 3D) and only 1.3-fold in suspension (Figure 3E). Bound LAP1-GFP colocalized with αvβ6 integrin; it was enriched at the cell-cell borders and partially internalized by the cells (Figure S1C).
To test whether the M5-induced increase in LAP1-GFP binding was mediated by αvβ6 or αvβ8 integrins, we pre-treated the monolayer cultures with integrin-selective synthetic αvβ6 and αvβ8 minibinders, which compete with LAP1 for the integrin binding.50 Selective inhibition of αvβ6 integrin significantly reduced LAP1-GFP binding, while selective αvβ8 inhibition had a weaker effect (Figure 3F). Their combined application did not result in a synergistic effect, emphasizing the selective and specific role of the αvβ6 minibinder.
We conclude that M5 stimulation mimicking psoriatic cytokine milieu predominantly promotes αvβ6 integrin expression and enhanced LAP1 binding by keratinocytes, which reflects the increased activation of latent TGFβ during skin inflammation, enhancing pSmad2 signaling.
Integrin αvβ6 activates TGFβ1 to induce paracrine signaling in keratinocytes
LAP1 binding by the integrin receptors is an important step in the TGFβ1 activation; however, it does not signify a release of active TGFβ1 into the extracellular space. To detect released TGFβ1, we used a chemiluminescent assay with reporter cells. First, adherent keratinocytes were presented to exogenous GARP-bound TGFβ1 (or empty GARP as control) expressed on the surface of transiently transfected BOSC-23 cells (Figure 3G). Selective blocking of αvβ6 or αvβ8 integrins on keratinocytes using minibinders resulted in a significant reduction in released active TGFβ1, particularly with αvβ6 inhibition, similar to when both minibinders were combined (Figure 3H). This signifies that the release of active TGFβ1 by keratinocytes is largely integrin-dependent, and that αvβ6 is the predominantly responsible integrin.
Surprisingly, M5 stimulation did not enhance the release of active TGFβ1 into the medium upon keratinocyte co-culture with BOSC-23 cells expressing GARP-bound TGFβ1, as shown by the chemiluminescent assay in the reporter cells (Figure 3H). That was unexpected because our previous assays consistently showed increased binding of soluble LAP1-GFP upon M5 stimulation. Considering our recent observation that αvβ6-LAP1 complexes are very stable and will cause an αvβ6-dependent internalization of LAP1-GFP-particles from LAP1-GFP-coated glass, we speculated that entire αvβ6 integrin-LAP1-TGFβ1-GARP complexes were increasingly internalized by the M5-stimulated keratinocytes. Using immunofluorescence, we detected the ALFA-tag on the latent TGFβ1 complex presented on the surface of BOSC-23 cells that were coming in contact with keratinocytes. Indeed, M5 cytokines induced a stronger internalization of the complexes by keratinocytes, which was blocked by the αvβ6 minibinder but not αvβ8 minibinder (Figures 3J and 3K). These data suggest that inflammatory cytokines promote the αvβ6-mediated TGFβ1 uptake, resulting in the TGFβ1/Smad2 pathway upregulation in keratinocytes detected in our previous experiments.
LAP1 binding by keratinocytes is mechanodependent
We have previously shown that mechanotransduction is increased in psoriatic epidermis.40 In our experimental system, M5 cytokine stimulation increased intracellular contractility in keratinocytes by promoting phosphorylation of the nonmuscle myosin II light chain (MLC). Notably, the highest increase in LAP1-GFP binding following M5 stimulation was observed in monolayers of adherent cells cultured on glass coverslips, where the high substrate stiffness allows the cells to generate greater intracellular forces.
We hypothesized that the increased intracellular forces might play a functional role in αvβ6 integrin-dependent TGFβ activation. To test this, we modulated mechanical input in keratinocytes using two approaches.
First, we inhibited cell contractility using blebbistatin, a small-molecule inhibitor of nonmuscle myosin II motor activity.51^,^52 Blebbistatin suppressed the M5-induced increase in LAP1-GFP binding in adherent monolayer cultures (2-fold increase) (Figures 4A and 4B), and in RHEs (1.5-fold increase) (Figure 4C), but it had lower effects on cells in suspension (about 20% increase) (Figure 4D). In a second approach, we stimulated the keratinocytes mechanically, by applying uniaxial stretch to cells cultured on elastic silicone plates, consistent with the contractility-dependent effect of the M5 stimulation. The stretched cultures had increased LAP binding compared to non-stretched controls, whereas the effect of stretch and M5 stimulation was not cumulative (Figure 4E). Thus, we conclude that increased mechanical input promotes αvβ6 integrin-dependent LAP binding, an effect that can be abrogated by the inhibition of intracellular contractile forces.Figure 4. Mechanoregulation of LAP binding by keratinocytes(A) Immunofluorescence staining with anti-GFP antibody revealing LAP1-GFP binding by monolayer cultures, counterstained for F-actin. Keratinocytes were stimulated with M5 for 24 h, with or without myosin II inhibition by blebbistatin (BS), applied for 1 h before fixation. Scale bars, 50 μm.(B) LAP1-GFP binding, quantification of fluorescence intensity from (A). One data point corresponds to one independent experiment. Mean fold change ±SD. Paired Student’s t test, ∗p < 0.05, ∗∗p < 0.01.(C) Immunofluorescence staining with anti-GFP antibody revealing the LAP1-GFP bingeing in RHE cultures stimulated with M5 for 24 h with or without myosin II inhibition by blebbistatin. Scale bars, 50 μm.(D) LAP1-GFP surface binding by the cells in suspension. Fluorescence intensity fold change from control, assessed by flow cytometry. M5 stimulation was performed for 24 h, with or without myosin II inhibition by blebbistatin, applied for the last 1 h. One data point corresponds to one independent experiment. Mean fold change ±SD. Paired Student’s t test, ∗p < 0.05.(E) LAP1-GFP fluorescence intensity fold change in monolayer cultures stimulated with M5 for 24 h, with or without cyclic stretch. One data point corresponds to one independent experiment. Mean fold change ±SD. One-sample t test, ∗p < 0.05.
TGFβ1-dependent mechanoregulation creates a positive feedback loop for further TGFβ1 activation
To explore potential feedback mechanisms linking TGFβ1 activation and mechanotransduction, we assessed the effects of TGFβ1 stimulation on the mechanosignaling in keratinocytes. Cell stimulation with pre-activated recombinant TGFβ1 promoted MLC phosphorylation (pMLC), enlargement of cell-substrate adhesions and nuclear translocation of mechanoactivated transcription co-activator yes-associated protein 1 (YAP) in both monolayer cultures (Figures 5A–5F) and RHEs (Figures 5G and 5H) indicating an increase in the intracellular contractility and mechanosignaling. These results recapitulated the effects of M5 treatment on myosin II activation and YAP signaling that we reported in our previous study.40 We then performed RNA sequencing analysis of RHEs (Database: GEO accession number [GSE298207](GSE298207)) and revealed that TGFβ1 treatment significantly altered the expression of 239 genes (FDR <0.05) compared to untreated controls (Tables S1 and S2). The differentially expressed genes (DEGs) were subjected to enrichment analysis using gene ontology cellular components, REACTOME, and KEGG databases, focusing on terms associated with mechanotransduction. Our analysis revealed significant enrichment (adjusted p value <0.05) for the pathways associated with Rho GTPase activation, cytoskeleton and ECM reorganization and interaction, and focal adhesions (Figure 5H). A full list of significantly enriched pathways is provided in Table S1. Notably, we observed an upregulation of expression for a number of genes encoding ECM proteins and ECM remodeling enzymes, αv and β6 integrins, and inflammatory cytokines/chemokines in response to TGFβ1 stimulation in stratified RHEs (Table S2). While M5 cytokine stimulation disrupted cell-cell adhesions (adherens junctions) in keratinocytes,40 TGFβ1 did not decrease the levels of adherens junction proteins (Figure S2).Figure 5. The effect of TGFβ1 on keratinocyte cultures(A) Immunofluorescence staining for mono- or double phosphorylated MLC (pMLC and ppMLC), counterstained for F-actin in monolayer cultures stimulated for 24 h. Scale bars, 20 μm.(B) pMLC and ppMLC fluorescence intensity per cell from conditions in (A). One data point corresponds to one independent experiment. Mean fold change ±SD. One-sample t test, ∗p < 0.05.(C) Immunofluorescence staining for vinculin, a focal adhesion marker, in monolayer cultures stimulated for 24 h. Scale bars, 20 μm.(D) Quantification of focal adhesion size. One data point corresponds to one individual adhesion. One representative experiment from three is shown. Mann-Whitney test, ∗∗∗∗p < 0.0001.(E) Immunofluorescence staining for YAP in monolayer cultures stimulated for 24 h.(F) Quantification of YAP nucleus/cytoplasm ratio. One data point corresponds to one independent experiment. Mean ± SD. Paired Student’s t test, ∗∗p < 0.01.(G) Immunofluorescence staining for pSmad2, pMLC and ppMLC in RHE cultures stimulated with TGFβ1 for 3 days. Scale bars, 50um.(H) Bubble plot depicting biological pathways significantly enriched in epidermal equivalents following stimulation with TGFβ1. The size of each circle represents the number of genes associated with that pathway. The color gradient indicates the adjusted p-value (AdjP-value), with yellow representing more statistically significant enrichment (lower Adj p value).
These data suggest a positive feedback loop mechanism, where TGFβ1 signaling enhances mechanotransduction, as well as αvβ6 expression and inflammatory cytokine production, leading to further TGFβ1 activation in keratinocytes under inflammatory conditions. To test this, we inhibited the TGFβ1 receptor TGFBR1 using small molecule inhibitor SB431542 in M5-stimulated cultures. TGFBR1 inhibition suppressed M5-induced Smad2 (Figures 6A–6D) and MLC (Figures 6E and 6F) phosphorylation, showing that endogenous TGFβ1 is activated in keratinocytes downstream of M5-induced signaling cascades and plays a key role in mediating the effects of the psoriatic cytokine milieu on cell contractility.Figure 6. The effect of TGFBR1 inhibition in the M5-stimulated keratinocyte cultures(A) Immunofluorescence staining for pSmad2 in monolayer cultures treated with M5 and/or SB431542 (SB) for 24 h. Scale bars, 50 μm.(B) pSmad2 fluorescence intensity per nucleus from conditions in (A). One data point corresponds to one independent experiment. Mean fold change ±SD. One-sample t test.(C) Immunofluorescence staining for pSmad2 in RHEs treated with M5 and/or SB431542 for 24 h. Scale bars, 50 μm.(D) pSmad2 fluorescence intensity per nucleus from conditions in (C). One data point corresponds to one independent experiment. Mean fold change ±SD. One-sample t test.(E) Immunofluorescence staining for ppMLC in monolayer cultures treated with M5 and/or SB431542 for 24 h. Scale bars, 50 μm.(F) ppMLC fluorescence intensity per cell. One data point corresponds to one independent experiment. Mean fold change ±SD. Paired Student’s t test, ∗∗p < 0.01. N.s. for other panels.
Discussion
This study bridges the fields of epidermal mechanobiology and immunology and sheds light on the mechanisms of global TGFβ1 regulation in the skin. We demonstrate that the upregulation of TGFβ1 signaling during psoriatic inflammation is mediated by increased integrin αvβ6-dependent TGFβ1 activation within the epidermis. This process is induced by inflammatory cytokines, is mechanodependent, and is reinforced by a positive feedback mechanism. Furthermore, we show that keratinocytes are not only major producers of functional TGFβ1, but also its key effector cells in inflammatory conditions.
Clinical data reveal a significant increase in TGFB1 gene expression specifically in the psoriatic epidermis, consistent with previous reports.12 However, our in vitro inflammation model of M5 cocktail stimulation did not detect a significant TGFB1 mRNA upregulation. This discrepancy might be attributed to the absence of additional cytokines that regulate its transcription in vivo, or differences in gene expression kinetics between RHE model and skin.
On the other hand, the transcriptional regulation of latent protein expression might not be that crucial, as it is superseded by multiple obligatory activation mechanisms at post-transcriptional and post-translational levels. Notably, our model successfully recapitulates the αvβ6 integrin upregulation and binding of latent TGFβ complex, emphasizing its reliability in capturing key elements of the psoriasis pathogenesis.
Indeed, activation of TGFβ1 signaling can be partially explained by the enhanced expression of the latent form. However, studies indicate that TGFβ1 is predominantly regulated not through the protein synthesis, but through the controlled release of active TGFβ1 molecules from the inactive LAP1-bound form, secreted to and anchored within the extracellular environment.6 Multiple cell-type-specific accessory proteins are responsible for anchoring and activating LAP1-TGFβ1 complexes, and many of them are found to be dysregulated in various pathologies, such as fibrosis and cancer.53^,^54^,^55^,^56^,^57 The process of latent TGFβ1 activation in principle allows for numerous possibilities for multi-level signaling modulation, shifting the paradigm for TGFβ1 pathway from a linear non-amplified cascade, to a more complex, environment-dependent signaling cascade, which might explain the variety of TGFβ isoform-specific and cell- and context-specific responses.23
Here, we describe αvβ6 integrin as the major regulator of TGFβ1 activation in the context of psoriatic epidermis. First, its expression is significantly higher than at steady state. Second, we show that increased expression and activation status of αvβ6 (assessed by the ability to bind LAP1) is induced in keratinocytes downstream of the inflammatory cytokines. Third, αvβ6 takes part in the mechanodependent feedback mechanisms and can be directly induced by TGFβ1 signaling. The specific role of the alternative TGFβ1-activating integrin αvβ8, which lacks an actin-binding site and is not mechanosensitive, but is predominantly expressed in dermal and immune cells, should be further explored in future studies. Integrin αvβ8 can be implicated in alternative, mechanoindependent mechanisms of TGFβ1 activation in response to other specific stimuli.30^,^44
We recently showed that psoriatic inflammatory cytokines upregulate Rho GTPase pathway and intracellular contractility forces in keratinocytes leading to the internal activation of mechanosignaling in the epidermis.40 Here, we explore the consequences of this increased mechanotransduction for the keratinocyte homeostasis and suggest that the alterations in the mechanosensitive activation mechanisms can be a major factor affecting TGFβ1 signaling in skin inflammation. Inhibition of intracellular contractility abrogated the M5-induced αvβ6 integrin activation and LAP1 binding in keratinocytes, whereas external mechanostimulation of the cells promoted it. Interestingly, no additive effect of stretch to the M5 stimulation was observed, suggesting that the cytokine-induced upregulation of intracellular contractile forces can mimic external mechanical forces for the integrin activation and integrin-dependent TGFβ1 activation. Next, we established the role for the positive feedback mechanism of TGFβ1 activation—upregulation of Rho GTPase—elevated cell contractility—further TGFβ1 activation, showing that breaking this loop can significantly restrain ongoing TGFβ1 signaling in keratinocytes. Further, our data show that TGFβ1 signaling in keratinocytes promoted expression of TGFβ1-activating αvβ6 (but not αvβ8) integrin and a number of genes encoding RGD-containing ECM proteins such as fibronectin, fibrillin and tenascin C (Table S2) - integrin ligands characteristically elevated in psoriatic skin. It highlights epidermal keratinocytes as active producers of ECM components within the epidermis beyond the basement membrane and as important regulators of skin mechanics.
It is noteworthy that the two models used in the study—monolayer cultures of undifferentiated keratinocytes and differentiated keratinocytes within stratified epidermis (RHE)—represent mechanically distinct systems and serve complementary purposes. Monolayer cultures grown on rigid substrate enabled dissection of precise molecular mechanisms of mechanotransduction-dependent TGFβ regulation at the single-cell resolution. In contrast, RHE models more closely reproduced real epidermal architecture, force distribution, differentiation-dependent patterns of integrin expression, and spatial organization of the integrin-ECM interactions, validating the relevance of uncovered mechanisms in 3D environment.
Further, our data suggest that upon the inflammatory cytokine stimulation, TGFβ1 activation and signaling might be predominantly occurring in keratinocytes intracellularly, potentially leading to localized autocrine or paracrine effects. We did not detect an increase in free extracellular TGFβ by the chemiluminescent reporter assay despite enhanced LAP1-GFP binding and TGFβ1/Smad2 signaling upregulation in M5-stimulated keratinocytes. A plausible mechanism for this phenomenon is that the enhanced internalization of αvβ6 integrin-LAP1-TGFβ1 complexes facilitates pH-dependent activation of TGFβ1 within intracellular compartments and promotes autocrine signaling through endosomes.58
Thus, we propose that TGFβ1 activation and signaling is controlled by the type of the integrin involved. Immune cells mainly and strongly express GARP-TGFβ1 complex, as well as present αvβ8 integrin on their surface. In the presence of αvβ8, GARP-bound latent TGFβ1 is activated in a cell-cell contact-dependent manner with a minimal release of soluble TGFβ1, causing anergy in the cells expressing the GARP-TGFβ1.30^,^59
The presence of αvβ6 integrin on keratinocytes changes this setup completely, offering a mechanoregulatory control of TGFβ1 binding and activation. Upon increased mechanotransduction and αvβ6 activation, keratinocytes would actively clear latent TGFβ1 from the epidermal intercellular space, depriving infiltrating T cells of a potentially tolerogenic signal. This would lead to a paradoxical effect: instead of becoming tolerant to autoantigens, peripheral T cells would remain reactive, ultimately attacking keratinocytes and driving the unchecked production of autoimmune antibodies. That could also explain the paradox of high levels of epidermal TGFβ1 detected in psoriasis, a disease characterized by strong inflammatory component, downregulation of TGFβ-dependent Tregs, and the absence of fibrosis, that would be otherwise induced by TGFβ1 available for fibroblast activation.
Therefore, we demonstrate an important role of the epidermis in the regulation of TGFβ signaling through TGFβ1 mechanoactivation during cutaneous inflammation. Mechanical stress has long been recognized as a trigger for skin inflammation.60^,^61 The Koebner phenomenon, in which psoriatic lesions appear on previously healthy sites following mechanical trauma,62 has been so far explained by several potential molecular mechanisms. Previously primed resident memory T cells can be activated by the alarmin release during physical micro-injury;63 in addition, the activation of YAP-dependent mechanotransduction pathways can reprogram keratinocytes contributing to the inflammation development.60^,^64 Our study indicates that application of physical forces, such as stretch, might locally activate mechanosensitive epidermal αvβ6 integrin, the uptake and activation of TGFβ1 in keratinocytes, and the downstream signaling cascades in the epidermis of predisposed skin toward Koebnerization.
Currently, treatment of moderate-to-severe psoriasis targets the central involved cytokines IL-23 and IL-17 using selective biologics.65^,^66 Yet mild-to-moderate disease lacks innovative therapies, and new topical treatments have not been approved in over a decade.65^,^66 While direct inhibition of TGFβ1 itself is a challenging therapeutic strategy due to safety concerns,67^,^68 understanding the context-specific TGFβ1 activation mechanisms may guide the development of more targeted and safer interventions, specifically for topical application. In this regard, our study provides clinically relevant insights into the psoriasis pathogenesis and the possible mechanisms underlying the Koebner phenomenon. Our data suggest that targeting αvβ6 integrin-mediated TGFβ1 activation in keratinocytes could be a promising strategy, potentially leading to fewer side effects compared with direct TGFβ inhibition. Additionally, exploring mechanotransduction pathways as a therapeutic target presents intriguing possibilities for novel much-needed innovative topical applications.
Limitations of the study
This in vitro study presents certain limitations. First, we can not conclude any association of the sex with the observed effects: the clinical samples have mixed origin and their number is limited; cell line N/TERT-1 is of male origin. Second, even though the M5 cytokine cocktail reproduces key inflammatory features of psoriatic epidermis, it does not fully recapitulate the complexity of the in vivo cytokine milieu, which might explain the absence of transcriptional TGFβ1 upregulation as observed in clinical samples. Our system includes one cell type and does not explore the complexity of the immune and dermal cell interactions. In particular, cells that, unlike keratinocytes, highly express αvβ8 integrin can contribute to alternative mechanoindependent modes of TGFβ1 activation. Third, the study did not fully reproduce the in vivo mechanical conditions of the human skin, in part due to the absence of dermal ECM. Finally, although our data suggest intracellular activation of TGFβ1 downstream of αvβ6 internalization, direct visualization of the intracellular activation sites and the precise trafficking dynamics remains to be established.
Resource availability
Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Maria Shutova ([email protected]).
Materials availability
The constructs generated in this study are available on request.
Data and code availability
- •The data supporting the findings are available within the article. The generated RNA-sequencing datasets are deposited under GEO accession number [GSE298207](GSE298207), and are available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE298207.
- •This paper does not report original code.
- •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.
Acknowledgments
This work was supported by the GRAPPA (Group for Research and Assessment of Psoriasis and Psoriatic Arthritis) Pilot Research Grant and 10.13039/501100007636Fondation Ernst et Lucie Schmidheiny to M.S.S.; by 10.13039/100000002National Institutes of Health grants R01DK126702 and R01DK137822 to A.I.I.; by 10.13039/501100001711Swiss National Science Foundation grants 310030-185261 and 320030-228395 to B.W.-H, and by the Swiss Foundation for Research on Myopathies (FSRMM) to M.B.
We thank Bernard Foglia (University of Geneva) for technical help, Sylvain Lemeille (University of Geneva) for the single cell RNA sequencing analysis, Nicolas Liaudet and the Bioimaging Facility (University of Geneva), Dr. David Tarusso and Immune Landscape laboratory (ILL), Center of Experimental Therapeutics, Department of Oncology UNIL CHUV, Lausanne for performing the RNAscope ISH-IHF staining.
Author contributions
Conceptualization, S.S., A.I.I., M.S.S., B.W.-H., and W.-H.B.; data curation, B.R.; formal analysis, X.J., S.S., J.K., M.B., and B.R.; funding acquisition, M.B., A.I.I., M.S.S., and B.W.-H.; investigation, X.J., S.S., J.K., M.B., F.N., and M.S.S.; methodology, J.K., M.B., M.S.S., and B.W.-H.; project administration, M.S.S., B.W.-H., and W.-H.B.; resources, G.K., B.W.-H., and W.-H.B.; supervision, M.S.S., B.W.-H., and W.-H.B.; validation, N.C.B., A.I.I., M.S.S., B.W.-H., and W.-H.B.; visualization, X.J., S.S., J.K., M.B., B.R., and M.S.S.; writing – original draft, M.S.S.; writing – review and editing, all authors.
Declaration of interests
The authors declare no competing interests.
STAR★Methods
Key resources table
REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesActinAbcamCat#ab3280;RRID:AB_303668ALFA-tagGeneva Antibody FacilityRabbit recombinant ABCD_AL626;RRID: N/ACollagen VIISigma-AldrichCat#C6805;RRID:AB_476860E-cadherin (IF)AbcamCat#ab40772;RRID:AB_731493E-cadherin (WB)Cell SignalingCat#14472;RRID:AB_2728770GFPAbcamCat#ab6556;RRID:AB_305564Integrin αv (IF, IHF, WB)AbcamCat#ab179475;RRID:AB_2716738Integrin αv (Flow cyt)AbcamCat# ab16821;RRID:AB_443484Integrin αvβ6 (IF, IHF, Flow Cyt)AbcamCat# ab97588;RRID:AB_10715984Integrin αvβ6 (IF, IHF)Custom (Wehrle-Haller lab)N/AIntegrin β6ThermoFisher ScientificCat# PA5-47309;RRID:AB_2576254Integrin β8AbcamCat# ab243023;RRID: N/APhospho Myosin Light Chain 2 (Ser19)Cell Signaling TechnologyCat#3671;RRID:AB_330248Phospho-Myosin Light Chain 2 (Thr18/Ser19)Cell Signaling TechnologyCat#3674;RRID:AB_2147464Phospho-Smad2 (Ser465/467)Cell SignalingCat#3108S;RRID:AB_490941TubulinAbcamCat# ab7291;RRID:AB_2241126VincilinSigma-AldrichCat#V9131;RRID:AB_477629YAP1ProteintechCat#13584-1-AP;RRID:AB_2218915α-cateninThermoFisher ScientificCat#13-9700;RRID:AB_2533044β-cateninAbcamCat#ab32572;RRID:AB_725966Biological samplesSkin biopsies (psoriasis patients)Department of Dermatology of the Geneva University HospitalsN/ASkin biopsies (healthy donors)Department of Plastic and Reconstructive Surgery and Division of Clinical Pathology of the Geneva University HospitalsN/AChemicals, peptides, and recombinant proteinsIL-17AR&D SystemsCat#317-ILBIL-1αR&D SystemsCat#200-LAOncostatin MR&D SystemsCat#295-OMIL-22MACS Miltenyl BiotecCat#130-096-295PeproTechCat#300-01ABlebbistatinEMD MilliporeCat#203391TGFβ1PeprotechCat#100-21Integrin β6 minibinderCustom (Wehrle-Haller lab)N/AIntegrin β8 minibinderCustom (Wehrle-Haller lab)N/AGFP-LAP1Custom (Wehrle-Haller lab)N/ACritical commercial assaysRNAscope™ Multiplex Fluorescent Reagent Kit v2Advanced Cell DiagnosticsCat#323100Phospha-lightTM systemThermo FisherCat#T1017Deposited dataRNA-sequencing dataNCBI Gene Expression Omnibus depository[GSE298207](GSE298207)Experimental models: Cell linesN/TERT-1 immortalized normal human keratinocytesProf. E.H. van den Bogaard, Radboud University Medical Center, Nijmegen, The NetherlandsN/ABOSC-23Beat A Imhof, University of Geneva, SV40-transformed HEK293 lineage formN/AMFB-F11Dr. Ina Tesseur, Department of Neurology and Neurological Sciences, Stanford University School of Medicine, StanfordN/AOligonucleotidesGAPDH forward primer 5’-TCG GAG TCA ACG GAT TTG GT-3’Microsynth (Schützenstrasse, Balgach, Switzerland)N/AGAPDH reverse primer 5’-TGA AGG GGT CAT TGA TGG CA-3’MicrosynthN/AITGAV forward primer 5’- CTC CCG CTT CTT CTC TCG GG -3’MicrosynthN/AITGAV reverse primer 5’- AAG AAA CAT CCG GGA AGA CGC -3’MicrosynthN/AITGB6 forward primer 5′- CCG GAA ACA TTC TCC AGC TG -3′MicrosynthN/AITGB6 reverse primer 5′- GCT GAA GGA AGC TGT GTC TC -3′MicrosynthN/AITGB8 forward primer 5′- ACC AGG AGA AGT GTC TAT CCA G -3′MicrosynthN/AITGB8 reverse primer 5′- CCA AGA CGA AAG TCA CGG GA -3′MicrosynthN/ATGFB1 forward primer 5′- TGA CAA GTT CAA GCA GAG TAC ACA CA -3′MicrosynthN/ATGFB1 reverse primer 5′- AGA GCA ACA CGG GTT CAG GTA -3′MicrosynthN/ASoftware and algorithmsImageJSchneider et al.71https://imagej.nih.gov/ij/GraphPad Prism 8GraphPadhttps://www.graphpad.com/scientific-software/prism/R/Bioconductor package edgeR v. 3.18.1.Robinson et al., 2010R/Bioconductor package edgeR v. 3.18.1.Cluster 3.0de Hoon et al., 2004https://www.encodeproject.org/software/cluster/gProfilerRaudvere et al., 2019https://biit.cs.ut.ee/gprofiler/gostPathMe viewerDomingo-Fernandez et al., 2019https://pathme.scai.fraunhofer.de/CytoscapeShannon et al., 2003https://cytoscape.org/OtherKBM-Gold Keratinocyte Growth Medium BulletKitLonzaCat#00192060Keratinocyte-SFM Medium (Kit) with L-glutamine, EGF, and BPEGibcoCat#17005075CnT-Prime Epithelial Culture MediumCellnTecCat#CnT-PRCnT-Prime 3D Barrier Culture MediumCellnTecCat# CnT-PR-3DThincert chambers for RHE modelsGreiner Bio-OneCat# 665640
Experimental model and study participant details
Human skin samples
Psoriatic skin biopsies were taken from untreated adult male and female patients presenting at the Department of Dermatology of the Geneva University Hospitals in Switzerland. Healthy skin biopsies were taken from adult female patients presenting at the Department of Plastic and Reconstructive Surgery and Division of Clinical Pathology of the Geneva University Hospitals in Switzerland. This study was conducted according to the Declaration of Helsinki and approved by the local ethical committee of the University Hospitals of Geneva, Switzerland (2020-0058). Written informed consent was obtained for each individual.
Monolayer cell culture
The human male keratinocyte cell line N/TERT-1 was obtained from Prof. Ellen H. van den Bogaard Laboratory (Radboud University Medical Center, Nijmegen, The Netherlands) thanks to the courtesy of the J. Rheinwald laboratory (Harvard Medical School, Boston, USA). N/TERT-1 cells were cultured in keratinocyte-serum free medium (K-SFM, Gibco), supplemented with 25 μg/ml bovine pituitary extract (BPE) (Invitrogen), 0.2 ng/ml EGF (Invitrogen), 300 μM CaCl_2_ (Sigma-Aldrich) and antibiotics in a humidified atmosphere with 5% CO_2_ at 37°C and passaged at 30-40% confluency, as described previously.40 The time of stimulation with cytokines, TGFβ and inhibitors is indicated in the Figure legends for each experiment.
Reconstructed human epidermis (RHE) model
N/TERT-1 keratinocytes were plated at 250’000 in 0.5ml CnT PRIME medium (CELLnTEC) in the thincert chambers (Greiner Bio-One), placed into 12 well plate, with the wells containing 1ml of CnT PRIME medium outside the chamber. The medium inside the chamber was refreshed the following day to remove not attached cells. After two more days in culture, the medium was replaced with CnT PRIME 3D barrier medium (CELLnTEC) in both inside and outside chambers for 24 hours. Then the thinsert chambers were transferred into deep well plates (Greiner, Cat# 665110) with 4ml CnT PRIME 3D barrier medium in the bottom, whereas the liquid above the cell layer was dried (counted as Day 0 of airlift). The medium was refreshed on Days 3, 5 and 7. At Day 10 of culturing at the air-liquid interface, RHEs were stimulated for indicated time and harvested.
Complementary cell lines
BOSC-23 cells were used as presenters of GARP/TGFβ1 constructs in the activation assays and as producers of the integrin minibinders. They were cultured in DMEM supplemented with 10% FBS at 37°C under 10% CO_2_. MFB-F11 reporter cells were used for chemiluminescent TGFβ reporter assay69; they were cultured with 10% FBS at 37°C under 5% CO_2_.
Method details
Cytokine stimulation
The M5 cocktail comprised the following cytokines at 10ng/ml each: IL-17A, IL-1α, Oncostatin M (all from R&D Systems), IL-22 (MACS Miltenyl Biotec), TNFα (PeproTech London, UK). TGFβ1 (Peprotech or R&D Systems) was used at 2.5ng/ml.
Small molecule inhibitors
The following inhibitors were used: 50μM (-)-Blebbistatin (Calbiochem) and 2μM SB431542 (Tocris). The inhibitors were diluted in DMSO, with the final DMSO concentration in the culture medium of 0.5% for blebbistatin experiments and 0.1% for SB431542 experiments. All controls in the experiments involving inhibitors were treated with DMSO at the corresponding concentration.
Integrin β6 and β8 minibinders
De novo designed αvβ6 and αvβ8 RGD-specific inhibitors with picomolar affinities were produced as small proteins in BOSC-23 cells using the calcium phosphate (Ca_3_(PO_4_)2) transfection method.50 These engineered tagged peptides, referred to as minibinders, were quantified, and their optimal working concentrations were determined by Flow Cyt through a dose-response assay, assessing their binding to the respective integrins (1.5 μg/ml).
BOSC-23 transfection, co-culture for binding and activation assay
BOSC-23 cells were co-transfected with two plasmids encoding respectively TGFβ1 and GARP (JetPrime Polyplus transfection) to generate TGFβ surface-presenting cells. The human TGFβ1 sequence was obtained from Addgene (Cat.#52185) and expressed in pcDNA3 backbone with an N-terminal 6×His- and ALFA-tag. The human GARP (LRRC32) sequence in pcDNA3 backbone contained an N-terminal HA-tag and was obtained from.70 Co-transfected BOSC-23 cells were used 48 h post-transfection. The surface levels of TGFβ were quantified by Flow Cyt using ALFA-tag staining for experimental normalization.
In co-culture experiments, N/TERT-1 keratinocytes were seeded on glass coverslips for interaction assays or on plastic well for activation assays. After reaching at least a monolayer of 60% confluence, cells were pre-incubated 1h with β6-minibinders or/and β8-minibinders at 1.5 μg/ml or Mock medium before exposing them to TGFβ-presenting BOSC-23 cells. BOSC-23 cells transfected only with GARP were used as basal control for endogenous TGFβ activation.
For integrin-TGFβ interaction assays, co-cultures were maintained for 2.5 hours. Non-adherent TGFβ-presenting BOSC-23 cells were then removed, and the remaining adherent cells were fixed with 4% paraformaldehyde, followed by permeabilization and immunostaining (see immunofluorescence staining and imaging of cells).
For paracrine release and activation of TGFβ, N-TERT1 keratinocytes were seeded in 96-wells plate at 15’000 per well. The following day, the medium was removed, and the cells were washed with PBS to remove residual soluble factors. A total of 30,000 TGFβ-presenting BOSC-23 cells were then added onto the keratinocyte monolayers in specific media (DMEM or DMEM + M5 cocktail) to a final volume of 100 μL. Co-cultures were maintained for 24 hours at 37°C with 5% CO_2_ before proceeding with the chemiluminescent reporter assay.
Chemiluminescent TGFβ reporter assay
For chemiluminescent TGFβ reporter assay, 30’000 MFB-F11 reporter cells were seeded in 96-well plates. The following day, cells were washed with PBS, and 50 μL of serum-free medium was added. After 4 hours of incubation, 50 μL of the N/TERT-1 and TGFβ-presenting BOSC-23 co-culture supernatant was added into those same serum free wells. This mix was incubated for 24 hours on the MFB-F11, allowing active TGFβ, if present, to induce Smad signaling, which drives alkaline phosphatase (SBE-SEAP) expression. Resulting enzyme secretion was quantified using Phospha-lightTM system AB applied biosystems kit (ThermoFisher Scientific). 50 μL of culture supernatant was transferred into a 96-well plate and heated at 65°C for 30 minutes to inactivate endogenous phosphatases while preserving SEAP enzymatic activity. Samples were cooled to room temperature before sequential addition of “assay buffer” and “chemiluminescent phosphate substrate”. The assay buffer equilibrates the reaction conditions while the chemiluminescent phosphate substrate is hydrolyzed by SEAP to generate a luminescent signal. Luminescence was measured using a microplate reader with an integration time of 10s seconds per well. Signal intensity is proportional to the level of active TGFβ present in the original sample. The validated minimal linear detection range is ≥ 50 pg/mL of active TGFβ.
LAP binding assays
For the assay in monolayer cultures, cells were seeded at 10’000 per well on glass coverslips in 24 well plates and 3 days after plating they were stimulated with cytokines for 24 hours followed by the small molecule inhibitors for 1 hour. Then LAP1-GFP stock solution (33.4 ng/μl in water) was directly added to the media at final concentration of 670 ng/ml. In the experiments with β6 and β8 minibinders, they were added to the cultures 30 minutes before LAP application.
The RHEs were cultured as described above until day 10 of airlift. After 24 hours of M5 stimulation, LAP1-GFP was added directly to the media at the bottom of the inserts at final concentration of 670 ng/ml. The cultures were incubated for 1 hour at 37°C and 5% CO2. Then RHEs were fixed in 4% formaldehyde overnight and embedded in paraffin. 5μm sections were cut, deparaffinized and stained with anti-GFP antibody as described below.
Mechanostimulation
The silicon culture plates (Cellscale, Cat.# CS-MCFX-424) were sterilized, treated with a mixture of 20% 3-aminopropyltriethoxysilane (APTES) (Thermo Fisher Scientific) in absolute ethanol for 5 minutes, washed three times in absolute ethanol and left to dry. The sialinized plates were then coated with 0.2mg/ml collagen IV (Sigma, Cat.# C5533) in PBS 1 hour at 37°C and washed in PBS 3 times. The keratinocytes were seeded at 8000 cells per well (0.64 cm2), in 0.3 ml modified K-SFM. After they formed islands in 2-3 days, one plate was inserted into the MCFX stretch device (Cellscale) and cultured overnight undergoing cyclic stretching at 10% extension. The other silicone plate was left rested in the same incubator.
Flow cytometry
The cells were seeded at 80’000 in T25 flasks and cultured for 48 hours and then stimulated with M5 cocktail. After 24 hours, they were treated with 50uM of blebbistatin or the equivalent final concentration of DMSO (0.5%) for 1 hour at 37°C and 5% CO2. Then, cells were harvested and 200’000 per condition and incubated with 100uL of respective integrin antibodies (key resources table) or LAP1-GFP-human IgG diluted in K-SFM at 3.3μg/ml supplemented with 1% BSA for 30 minutes on ice by mixing gently every 10 minutes. After 2 washes with ice-cold PBS (Ca2+/Mg2+), cells were incubated with PE-anti human IgG diluted 1/1000 in K-SFM-1% BSA or respectively, anti-mouse-AF555, anti-goat-AF568 or anti-rat-AF555 (Invitrogen) diluted 1/500 in K-SFM-1% BSA. After 2 washes and resuspension with PBS (Ca2+/Mg2+), DraQ7 (ThermoFisher) was added to the cells to test cell viability. The data were acquired on a CytoFLEX (Beckman Coulter), analyzed using FlowJo software and presented as geometric mean.
Immunofluorescence staining and imaging of cells
Coverslips with cells were fixed with 4% paraformaldehyde in PBS at room temperature for 15 minutes, washed with PBS 3 times 5 minutes each, and permeabilized with 0.1% Triton X-100 (Sigma-Aldrich) in PBS for 5 minutes and washed again with PBS 3 times 5 minutes each.
The samples were stained with the primary antibodies (key resources table) for 1 hour at room temperature, washed in PBS 3 times, and then incubated for 1 hour at room temperature with the respective Alexa-fluor conjugated secondary antibodies (Invitrogen) and Alexa Fluor-488 or -647 phalloidin (Invitrogen). The coverslips were washed in PBS 3 times and mounted on the slides using DAPI Fluoromount-G mounting medium (SouthernBiotech). Z-stack confocal images were acquired using the LSM800 Airyscan confocal microscope (Zeiss) at 20x, 40x or 63x magnification.
For the immunostaining of the cells cultured on the silicone plates, all the steps were performed inside the wells of the silicone plate. The wells were then cut out from the silicone plate using scalpel and mounted on the glass slides with the cell side up, covered with a 12mm cover glass.
Immunofluorescence staining and imaging of histological samples
Skin tissue samples or RHEs were fixed in 4% formaldehyde overnight, embedded in paraffin, and 5μm sections were cut and deparaffinized. Antigen retrieval was performed in 10 mM citrate buffer pH 6.0 for 30 minutes at 95°C. For immunofluorescence detection, sections were blocked with 3% BSA and 5% normal goat serum in PBS for 1 hour after the antigen retrieval. Next, sections were incubated for two hours with primary antibodies (key resources table) followed by 1 hour incubation with Alexa Fluor 488-conjugated donkey anti-rabbit and Alexa Fluor 555-conjugated donkey anti-mouse IgG secondary antibodies (Invitrogen), all in PBS with 1% Bovine Serum Albumin (BSA) (Sigma-Aldrich) and 0.1% Tween-20 (Sigma-Aldrich) and mounted with DAPI Fluoromount-G (SouthernBiotech). Z-stack confocal images of fluorescent samples were acquired using the LSM800 Airyscan confocal microscope (Zeiss).
RNA in situ hybridization (ISH) combined with immunofluorescence
For the combined RNA-ISH-IF (RNAscope®) analysis, 5μm thick sections from paraffin blocks were used. In situ hybridization with co-immunostaining (IF) was performed following the manufacturer’s protocol of the RNAscope Multiplex Fluorescent Reagent Kit v2 assay (Cat.# 323100, Advanced Cell Diagnostics). Standard conditions were used: 15 minutes incubation for the Antigen retrieval step and 15 minutes for Protease Plus treatment. Combined ISH and antibody staining with Opal dyes (Akoya Biosciences) was performed manually for ISH and employing an automated Ventana Discovery Ultra Staining module (Ventana, Roche) for IF. For the individual experiments, the following combinations of RNAscope® probes (all from Biotechne) and antibodies were used. All Opal dyes were purchased from Perkin Elmer AG: Probe-Hs-TGFβ1 probe (Cat.# 400881) with opal570 (Cat.# FP1488001KT), and anti-collagen VII (Cat# C6805, Sigma, diluted 1/100) with opal690 (Cat.# FP1497001KT).
Nuclei were visualized by final incubation with Spectral DAPI (1/10, Cat# FP1490, Akoya Biosciences). The slides were mounted with fluorescence mounting medium (Cat# S3023, Dako) and stored in the dark at 4°C until scanned within 48 h. Images were acquired on the Vectra Polaris automated imaging system at 40X magnification (Akoya Biosciences, Marlborough, USA), allowing the unmixing of spectrally overlapping fluorophores and tissue autofluorescence of whole slide scans using Phenochart (Akoya Biosciences).
Western blotting
The RHE cultures were lysed in 100μl ice-cold RIPA lysis buffer containing Phosphatase Inhibitor Cocktails 2 and 3 and Protease Inhibitor Cocktail (Sigma-Aldrich). After centrifugation at 13’000 g for 10 minutes at 4°C, supernatants were collected and protein concentration was measured using Bradford assay kit (Bio-Rad) or BCA assay (Fisher Scientific). The samples were mixed with NuPAGE LDS sample buffer and reducing agent (Invitrogen) and heated at 70°C for 10 min. 10μg total protein was loaded onto a 4-12% NuPAGE BT gradient gel (Invitrogen), resolved by SDS-PAGE, and transferred to a 0.45μm polyvinylidene fluoride membrane in NuPAGE Transfer Buffer at 20V for 1 hour using a Trans-Blot SD Semi-Dry Transfer Cell (Bio-Rad). The membrane was blocked for 30 minutes with 5% Bovine Serum Albumin (BSA) and 0.1% Tween-20 in PBS buffer, incubated with primary antibodies in 0.1% Tween-20 in PBS overnight at 4°C, washed in 0.1% Tween-20 in PBS 3 times, incubated with appropriate HRP-conjugated secondary antibodies for 1 hour, and developed with SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific) and GE ImageQuant LAS4000 (GE Healthcare).
RNA extraction and RT-qPCR
Total RNA was isolated using the Quick-RNA MiniPrep kit (Zymo Research), according to the manufacturer's protocol. cDNA was synthesized using the ImPromII Reverse Transcriptase (Promega) and random hexamer primers (Promega). Genes of interest were quantified with PowerUp SYBR green Master Mix (Applied Biosystem) on QuantStudio 6 Pro instrument (Applied Biosystems) and normalized to GAPDH mRNA using a comparative method (2-ΔΔCt). Non-reverse-transcribed RNA samples and RT mix without RNA were included as negative controls. The primers were from Microsynth (Schützenstrasse, Balgach, Switzerland) and are listed in the key resources table.
RNA sequencing
RNA was isolated from the RHEs treated with 2.5 ng/mL of recombinant human TGFβ1 for 24 hours in three independent experiments with a Qiagen RNAeasy micro-kit. Integrity and quantity of RNA were assessed with Bioanalyzer (Agilent Technologies) (RIN ranging from 7.5 to 10). cDNA libraries were constructed using the Illumina Stranded mRNA lig stranded Kit according to the manufacturer’s protocol. Reads were generated on an Illumina NovaSeq 6000 sequencer. FastQ reads were mapped to the ENSEMBL reference genome (GRCh38) using STAR, version v.2.7.10b, with standard settings. The transcriptome metrics were evaluated with the Picard tools v. 2.21.6. The table of counts with the number of reads mapping to each gene feature of the UCSC human hg38 reference was prepared with HTSeq v.0.11.3. The differential expression analysis was performed using R/Bioconductor package edgeR v.3.18.1. Briefly, the counts were normalized according to the library size and filtered. The genes were filtered on expression levels. As required by the experimental design, a paired t-test was used to assess the differentially expressed genes (DEG). The DEG p values were corrected for multiple testing errors with a 5% false discovery rate (FDR) according to the correction by Benjamini-Hochberg (BH). Library preparation, sequencing, and read mapping of the reference genome were performed by the Genomics Platform at the University of Geneva, Switzerland.
Quantification and statistical analysis
All image analyses were performed using ImageJ software.71 Statistical analyses were performed using GraphPad Prism 8 software and appear in the figure legends. One data point corresponds to one independent experiment, unless stated otherwise. Statistical significance is labelled as follows: ∗p<0.05, ∗∗p<0.01. ∗∗∗p<0.001, ∗∗∗∗p<0.0001. Statistical tests are specified in the corresponding figure legends.
Quantification of LAP-TGFβ1 internalization by keratinocytes
Confocal images acquired at 63x (around 10 images/cells per condition) were presented as maximum projections, and Gaussian blur (radius 1) was applied in ImageJ. To detect the particles, the threshold method using same settings across conditions was applied to the channel corresponding to the ALFA-tag labelling. The particle sizes and their numbers per cell were quantified automatically after the thresholding.
Quantification of pMLC, ppMLC and LAP1-GFP fluorescence intensity
For the monolayer cultures, confocal images acquired at 40x were presented as maximum projections and the total intensities were measured from each field of view after background subtraction and normalized by the number of cells according to the nuclei count.
For the RHE cultures, confocal images were presented as maximum projections and the integrated intensities from several fields of view were measured after background subtraction and normalized by the total area of ROIs. The top cornified layer consisting of dead cells was excluded from ROIs.
Quantification of pSmad2 fluorescence intensity in the nuclei
For the skin biopsies, confocal images obtained at 20x magnification were presented as maximum projections. The regions of interest (ROI) corresponding to the epidermal nuclei were selected based on the threshold in the DAPI channel. The mean intensity within the selection was measured in pSmad2 channel after background subtraction using ImageJ software.
For the monolayer cultures, confocal images obtained at 40x magnification were presented as maximum projections. The ROIs corresponding to the cell nuclei were selected based on the threshold in the DAPI channel. The total fluorescence intensity within ROI of several fields of view was measured and normalized by the number of cell nuclei, with at least 80-100 cells measured per condition.
Quantification of YAP nucleus/cytoplasm ratio
Confocal images obtained at 20x magnification were presented as maximum projections. The ROIs corresponding to the cell nuclei were selected based on the threshold in the DAPI channel after applying Gaussian Blur filter with Sigma radius 2. After the background subtraction, the integrated fluorescence intensity in the YAP channel in the whole images and within nuclei ROIs was measured and summed from several fields of view. The integrated cytoplasm intensity was calculated as the integrated intensity of the whole images minus integrated intensity of the nuclei. The nuclei and the cytoplasm intensities were normalized by the respective area, and the Nucleus/Cytoplasm ratio was calculated.
Quantification of focal adhesion size
Confocal images of vinculin immunofluorescence staining were processed in ImageJ software prior to the analysis. Images were presented as maximum projection from 3 brightest slices in the vinculin channel, then Gaussian Blur filter with Sigma radius 1 was applied. The files in the TIFF format were uploaded to the Focal Adhesion Analysis Server https://faas.bme.unc.edu/,72 with the following parameters: Detection Threshold 2, Min FA size 8, Max FA size –, FAAI ratio 1. Results were extracted from subfolders: Individual pictures – Raw data – Area.
Quantification of band intensity in western blotting
The black-and-white digital images of the developed Western blots were presented in 16bit inverted contrast mode. The local background near the bands of interest was subtracted. The total band intensity was measured and then normalized to the total intensity of the corresponding band of the loading control (actin or tubulin).
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Boehncke W.H.Dressel D.Zollner T.M.Kaufmann R.Pulling the trigger on psoriasis Nature 379199677710.1038/379777 a 08587600 · doi ↗ · pubmed ↗
- 2Eder L.Gladman D.D.Psoriatic arthritis: phenotypic variance and nosology Curr. Rheumatol. Rep.15201331610.1007/s 11926-013-0316-423371481 · doi ↗ · pubmed ↗
- 3Boehncke W.-H.Schön M.P.Psoriasis Lancet 386201598399410.1016/S 0140-6736(14)61909-726025581 · doi ↗ · pubmed ↗
- 4Gutcher I.Donkor M.K.Ma Q.Rudensky A.Y.Flavell R.A.Li M.O.Autocrine transforming growth factor-β1 promotes in vivo Th 17 cell differentiation Immunity 34201139640810.1016/j.immuni.2011.03.00521435587 PMC 3690311 · doi ↗ · pubmed ↗
- 5Lee Y.Awasthi A.Yosef N.Quintana F.J.Xiao S.Peters A.Wu C.Kleinewietfeld M.Kunder S.Hafler D.A.Induction and molecular signature of pathogenic TH 17 cells Nat. Immunol.13201299199910.1038/ni.241622961052 PMC 3459594 · doi ↗ · pubmed ↗
- 6Lafyatis R.Transforming growth factor β--at the centre of systemic sclerosis Nat. Rev. Rheumatol.10201470671910.1038/nrrheum.2014.13725136781 · doi ↗ · pubmed ↗
- 7Choi G.Park Y.J.Cho M.Moon H.Kim D.Kang C.Y.Chung Y.Kim B.S.A critical role for Th 17 cell-derived TGF-β1 in regulating the stability and pathogenicity of autoimmune Th 17 cells Exp. Mol. Med.532021993100410.1038/s 12276-021-00632-934050263 PMC 8178381 · doi ↗ · pubmed ↗
- 8Wang J.Zhao X.Wan Y.Y.Intricacies of TGF-β signaling in Treg and Th 17 cell biology Cell. Mol. Immunol.2020231002102210.1038/s 41423-023-01036-737217798 PMC 10468540 · doi ↗ · pubmed ↗