WHAT IS ALREADY KNOWN ON THIS TOPIC

Activation of fibroblasts is crucial for the development of all fibrotic processes, in particular in systemic sclerosis. Fibrogenic cytokines and growth factors play an important role in this process; also the integrin-mediated interaction of fibroblasts with the surrounding extracellular matrix environment is a crucial activator. However, the exact molecular mechanisms involved in this activation are still unknown.

WHAT THIS STUDY ADDS

Abrogating the interaction of fibroblasts with collagen, the main structural component in the extracellular environment, induces a major defect in force generation, matrix organisation and YAP activation, thereby compromising fibroblast activation, myofibroblast formation and function. Integrin-mediated collagen interaction is required for challenging situations such as fibrotic processes but are dispensable for (murine) development.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICYIntroduction

Excessive deposition of extracellular matrix (ECM) constituents and their aberrant macromolecular organisation is the hallmark of fibrotic diseases. Activated fibroblasts (myofibroblasts) play a key role in this process by producing and remodelling ECM. They are regulated by a variety of cytokines and growth factors released from different cell types but also by their interaction with the ECM.1 2 The ECM is a complex network of multidomain proteins interacting to form large macromolecular complexes that determine the biomechanical properties of tissues. It is composed of proteins belonging to several families in which collagens constitute the most abundant ECM component.3 4 Besides providing structural support, the collagens and other ECM constituents have critical roles in modulating numerous cellular activities.4–6 They function as reservoir for growth factors, but also directly interact with most cell types through specific cell surface receptors.7 These interactions provide vital cues impacting cellular behaviour and disease processes.8

Integrins form a large group of receptors for ECM constituents. They are transmembrane heterodimers composed of a ligand-specific α subunit in non-covalent association with a β subunit. Integrins lack enzymatic activity but execute bidirectional signalling between the extracellular and the intracellular compartments as their intracellular domains are linked to the actomyosin cytoskeleton, whereas their extracellular domains can bind ECM ligands.7 Integrin signalling thereby integrates information from the ECM and cytoskeleton and influences virtually all cellular functions.9 Through their ability to couple to the contractile actomyosin cytoskeleton, integrin-based adhesions are a major site for sensing, generating and transducing mechanical tension to downstream effectors such as YAP/TAZ,10 11 thereby regulating fibroblast activities.12 Tension-induced signalling from ligand-bound integrins clustered in adhesion sites involves FAK, SRC-family kinases, phosphoinositide-3 kinase, JNK pathway and RHO family GTPases leading to stress fibre formation and actin reorganisation to YAP/TAZ activation and MRTF-SRF (myocardin-related transcription factors, serum response factor)-dependent transcriptional changes.10 12–14

The major integrins facilitating fibroblast-collagen interactions are α1β1, α2β1 and α11β1.15 16 Their distribution across cell types and tissues and ligand preference varies. The collagen-binding α10β1 integrin is restricted to cartilage.17

Much work defining the molecular interactions of integrins with collagen or other ECM components has used cultured cells attaching to extracted ECM, ECM fragments or synthetic peptides. In vivo the situation may be more complex, since the cells encounter the surrounding ECM in a supramolecular organisation. It remains an open question which binding sites are exposed and accessible or masked depending on biological context.18–21

Single and double knockout mouse models were instrumental in delineating regulatory functions of individual collagen-binding integrins; for review, see Gullberg and Eble.22 α1β1, mainly expressed by mesenchymal cells, supports cell proliferation, stimulates tumour angiogenesis and impacts T cell and innate immunity, while α2β1, more widely expressed, regulates platelet and mast cell functions, suppresses tumour angiogenesis and impacts immune responses differently from α1β1.16 23–25 These knockout models have also revealed various regulatory functions of collagen-binding integrins in granulation tissue formation and collagen deposition in in vivo models.26 27 Loss of integrin subunit α11 (ITA11), expressed in a restricted fashion mainly by fibroblasts, decreased collagen abundance due to impaired myofibroblast formation.28 Integrin α11β1 was shown to largely contribute to collagen remodelling through TGFβ mediated JNK signalling26 whereas integrins α1β1 and α2β1 have minor roles in this process.29

Besides integrins, cell–collagen interaction is facilitated by the discoidin domain receptors (DDR)-1 and DDR-2.30 The DDRs are receptor tyrosine kinases that undergo autophosphorylation on collagen binding, triggering downstream signalling pathways that regulate ECM remodelling, proliferation, migration and survival. In addition, cross-talk of DDRs with integrins may either positively31 or negatively32 affect integrin function. DDR2 but not DDR1 is expressed by mesenchymal cells, and binds to fibrillar collagens.30

The studies using mice with either single or double deletion of the α1β1, the α2β1 or the α11β1 integrin revealed only slight phenotypic alterations, which could be due to functional compensation by remaining integrins. To eliminate compensation and to address the question, which role integrin-mediated contact of cells with collagen has in regulating fibroblast activities, we here generated and analysed a mouse model lacking all three integrins that were shown to bind collagens in in vitro systems. We also evaluated the significance of fibroblast-collagen interactions in a challenging situation, an induced fibrotic process that is characterised by excessive production and remodelling of collagen. We focused our analysis on the skin and on fibroblasts obtained from this tissue that is characterised by a collagen-rich ECM in its dermal compartment.

We observed that triple knockout (tKO) mice are viable but smaller than WT. They show attenuated dermal fibrosis and upregulation of DDR2 receptors. tKO fibroblasts in culture have aberrant cytoskeletal and contractile properties, which disable the generation of a strong mechanical response. Our findings reveal that together, the integrins α1β1, α2β1 and α11β1 are essential for regulating cellular force balance required for fibroblast activation and the development of fibrosis. They provide the basis for a potential therapeutic approach to modulate fibroblast functions in fibrosing diseases.

ResultsMice with global ablation of integrins α1β1, α2β1 and α11β1 are viable, with reduced mechanoresilience and increased DDR2 levels in their skin

tKO mice with a full-body depletion of the collagen-binding integrins α1β1, α2β1 and α11β1 were generated by breeding mice lacking α1β1 and α2β125 with α11β1-deficient mice33 (figure 1A and online supplemental figure S1). They are born at expected Mendelian ratio. In the first 3 months of postnatal life, tKO mice are visibly smaller than WT littermates (figure 1B), but indistinguishable from Wild type (WT) mice in later life. Body weight, in contrast, is reduced and does not normalise within 3 months (figure 1C). Histological analysis of the skin of young adult tKO mice showed no major alterations in the different compartments including the dermis (figure 1D). The collagenous ECM, inspected by transmission electron microscopy, failed to show gross differences in ultrastructural organisation of collagen fibrils (figure 1E). In contrast to the unaltered morphology, tensile strength tests revealed that tKO skin ruptures at lower force, is less deformable and stiff (figure 1F–J). However, maximum stress, toughness and elastic (E-) modulus of tKO back skin are slightly but not significantly reduced compared with the skin from sex-matched and age-matched WT mice (figure 1K–M).

Figure 1

Characteristic features of adult mice with constitutive ablation of integrins α1β1, α2β1 and α11β1. (A) Representative image of WT (left) and tKO mice at 6 weeks. (B) Body length and (C) weight of tKO mice are reduced in comparison to WT littermates. N=5. (D) Histological analysis (H&E staining) of the dorsal skin of 3-month-old mice shows no major alterations. (E) Representative electron micrographs of dorsal skin from 3-month-old mice shows no alterations in the collagen fibrillar arrangement. (F–M) Graphs from skin tensile strength tests of back skin from 10-week-old mice reveal significant reduction in (F) maximum force, (G) maximum deformation, (H) stiffness, (I) energy and (J) maximum strain of tKO skin, while (K) maximum stress, (L) elastic modulus and (M) toughness do not differ from WT littermates; N=11 WT/8 tKO. (N) Representative confocal images of back skin of 3-month-old mice show significantly increased DDR2 staining in tKO (O) upper and (P) lower dermis. N=4 per genotype. Quantification in (N) was restricted to dermal cells excluding hair follicles, arrector pili muscle and blood vessels (shown in magnified insets). Red arrows in insets depict DDR2 positive fibroblasts. Each data point represents a separate mouse. P values were calculated using two-way ANOVA with Sidak’s multiple comparisons test for (B, C), Student’s t-test for (F–M) and Mann-Whitney U test for (O, P). Data are presented as mean±SD. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Scale bars indicate 200 µm in (D), 500 nm in (E) and 30 µm in (N). ANOVA, analysis of variance; tKO, triple knockout; WT, Wild type.

Given the mild alterations in unchallenged skin, we asked if the activity of the missing integrins is compensated by other receptors. Contact with collagen in dermal fibroblasts is not restricted to the collagen-binding integrins but can also be mediated by non-integrin DDR2.30 Indeed, increased DDR2 staining in the upper and lower dermis was detected in the skin of tKO mice (figure 1N–P).

Increased levels of active DDR2 on tKO fibroblasts do not mediate spreading on collagen

To assess if ablation of the three integrins affects levels of other relevant cell surface receptors, we investigated the levels of the major fibronectin receptor, integrin α5β1 and of integrin α10β1, a collagen receptor normally expressed by chondrocytes. Flow cytometry did not reveal significant differences in surface levels of integrin subunit α5 ITA5 (figure 2A,B) and β1 (ITB1) (figure 2C,D) in tKO compared with WT fibroblasts seeded on fibronectin coats for 6 hours, indicating that surface levels of the major fibronectin receptor, integrin α5β1, are not significantly altered in tKO fibroblasts. No compensatory de novo expression of integrin α10, normally not present in fibroblasts, was detected in tKO dermal fibroblasts (online supplemental figure S2). However, and in accordance with increased DDR2 signals in tKO skin, increased DDR2 levels were also seen by immunoblotting of proteins from cultured primary dermal tKO fibroblasts grown on coated collagen I substrates (figure 2E–H and online supplemental figure S3). The increase was not only detected in fibroblasts cultured for 6 hours on collagen I coats, but also in fibroblasts released from tissue culture plastic by trypsin treatment (figure 2E; 0 hour), that is, prior to contact with collagen, indicating that its expression is not exclusively induced by fibroblast-collagen contact. Exposure of fibroblasts grown on tissue culture plastic to solubilised collagen (monomers) triggered a significant increase in phosphorylated DDR2 after 6 hours, which was >3 fold higher in tKO than in WT fibroblasts (figure 2I–L). Levels of total DDR2 (figure 2I, K) and the ratio of phosphorylated to total DDR2 (figure 2I, L) tended to be higher in tKO fibroblasts but did not differ significantly from those in WT fibroblasts.

Figure 2

Primary dermal fibroblasts derived from tKO mice display unaltered expression of fibronectin binding integrin α5β1 but increased expression of non-integrin collagen receptor DDR2. (A–D) Flow cytometric analysis of WT and tKO fibroblasts isolated from mouse skin seeded on fibronectin for 6 hours reveals unaltered surface levels of (A, B) integrin subunit α5 (ITA5) and (C, D) integrin subunit β1 (ITB1) in tKO fibroblasts. Representative western blots of primary dermal fibroblasts before (0 hour, trypsinised cells) and after seeding for 6 hours on (E) collagen I coating. Significantly increased levels of (F) phosphorylated DDR2 and (G) total DDR2 are detected in tKO fibroblasts seeded for 6 hours on collagen I, while (H) the ratio of phosphorylated to total DDR2 is comparable. (I) Stimulation of fibroblasts adherent to tissue culture plastic with acid-solubilised collagen I (10 µg/mL in 0.1% acetic acid) for 1-hour results in a (J) significant increase in phosphorylated DDR2 and (K) a slight but not significant increase in total DDR2 and (L) in the ratio of phosphorylated to total DDR2 in tKO fibroblasts. N=4–5 per genotype. Each data point represents fibroblasts from a separate mouse. Data are presented as mean±SD. P values were calculated using Mann-Whitney U test. *p≤0.05, **p≤0.01. DDR2, discoidin domain receptors 2; tKO, triple knockout; WT, Wild type.

To test if increased DDR2 surface levels contribute to adhesion and spreading of tKO fibroblasts, we compared the spreading behaviour and morphology of WT and tKO fibroblasts after 6 hours on collagen I and fibronectin (figure 3). Clearly, less than 5% of tKO fibroblasts had spread on collagen I despite high levels of phosphorylated DDR2 detected at this time point, whereas more than 40% of WT fibroblasts were spread. In contrast, both, WT and tKO fibroblasts appeared to be spread to comparable extent on fibronectin (figure 3A,B). A closer inspection of stained cells revealed, however, distinct differences in actin cytoskeleton architecture and cell size also on fibronectin (figure 3C–F). Whereas WT fibroblasts formed distinct actin microfilaments and stress fibres as well as clearly defined paxillin-positive focal adhesions (FAs) on collagen I, no such organisation was detected in tKO fibroblasts (figure 3C). These results clearly showed that the collagen receptor DDR2, although present at elevated levels in tKO fibroblasts, does not rescue their failure to adhere to and spread on collagen I. On fibronectin, both WT and tKO fibroblasts spread to larger area than on collagen I. Stress fibres and peripherally as well as centrally located adhesions were formed by WT fibroblasts, whereas size, abundance of stress fibres and matrix adhesions were reduced in tKO cells (figure 3D–F, see also figure 5). Clearly, the modified characteristics of the tKO fibroblasts on fibronectin were not on account of reduced levels of integrin α5 chain (figure 2A–D) but due to other cellular changes.

Figure 3

tKO fibroblasts spread to significantly smaller size than WT on collagen and fibronectin. (A) Phase contrast images of WT and tKO fibroblasts seeded for 6 hours on collagen I and fibronectin. (B) Whereas virtually no tKO fibroblasts adhere and spread on collagen I, adhesion to fibronectin is comparable to WT. (C, E) Representative confocal images of fibroblasts seeded for 6 hours on (C) collagen I and (E) fibronectin. (D, F) Quantification of cell size shows significant reduction in tKO spreading on both substrates. Each data point represents cells from a separate mouse. For (B) N=6 mice per genotype and for (D, F) N=4 mice per genotype (≥ 22 cells per mouse) were analysed. Data are presented as mean±SD. P values were calculated using Kruskal-Wallis test followed by Dunn’s test of multiple comparison test for (B) and Mann-Whitney U test for (D). *p≤0.05. Scale bars indicate 100 µm in (A) and 30 µm in (C, E). tKO, triple knockout; WT, Wild type.

tKO fibroblasts produce and arrange an abnormal ECM

We asked if the spreading defect of tKO fibroblasts might be caused by an abnormal matrix deposited by these cells and can be rescued by a WT matrix, and vice versa, if the tKO matrix can induce spreading defects in WT fibroblasts. WT and tKO fibroblasts were, therefore, cultured for 10 days to deposit matrices. These were decellularised and then seeded with either WT or tKO cells to create autologous or heterologous systems, respectively. As expected, tKO fibroblasts on tKO matrix were significantly less spread than WT fibroblasts on WT matrix (figure 4A top and B). In contrast, tKO fibroblasts tended to spread to larger size on heterologous WT matrix, and WT fibroblasts were slightly, although not statistically significantly, smaller on tKO matrix (figure 4A bottom and B). However, differences in ECM assembled by WT and tKO fibroblasts were detected, which revealed distinct differences in collagen I network organisation. Both tKO and WT fibroblasts deposited an extensive extracellular collagen I network, but in contrast to WT, tKO networks featured only few thick collagen fibre bundles, whereas fine fibres were more abundant (figure 4C) and prevailed in particular in tKO matrix after decellularisation (figure 4D).

Figure 4

Differences in deposited ECM and secretome between tKO and WT fibroblasts. (A) Representative confocal images showing staining for collagen I (red) and phalloidin (blue). Fibroblasts were seeded for 2 hours either on autologous or heterologous (bottom) ECM deposited by either WT or tKO primary dermal fibroblasts for 10 days followed by decellularisation. (B) On autologous matrix, tKO fibroblasts spread to significantly smaller size than WT. This spreading defect is partially rescued by seeding tKO fibroblasts on WT matrix. WT fibroblasts show a minor decrease in cell spreading when seeded on tKO matrix. Each data point represents fibroblasts from a separate mouse, N=4 mice per genotype, N≥25 fibroblasts per mouse. (C, D) Representative confocal images of ECM deposited by WT and tKO fibroblasts for 10 days, stained for collagen I (white) and DAPI (blue), show decreased amounts of thick collagen fibres (C) before and more clearly (D) after decellularisation. (E) Volcano plot and (F) list of significantly downregulated (blue) and upregulated (pink) proteins in tKO fibroblast supernatants as determined by SILAC-based secretome analysis. N=4. (G) STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) analysis and (H) Gene ontology (GO) term analysis representing biological processes potentially affected due to the significantly altered proteins in the tKO secretome. (B) Data are presented as mean±SD. P values were calculated using Kruskal-Wallis test followed by Dunn’s multiple comparison test. Scale bars indicate 30 µm in (A, C, D). ECM, extracellular matrix; tKO, triple knockout; WT, Wild type.

To obtain a comprehensive overview of the secretome of tKO fibroblasts, we applied SILAC to label the proteins released by tKO and WT fibroblasts into culture media over a 24-hour period. The list of significantly regulated proteins was surprisingly short and included eight downregulated (in tKO vs WT) and two upregulated candidate proteins (figure 4E,F).

Several of these downregulated proteins are involved in collagen fibrillogenesis such as thrombospondin 2 (TSP2)34 periostin (POSTN),35 procollagen C-endopeptidase enhancer 1 (PCOC1)36 and fibulin 4.37 Insulin-like growth factor binding protein 5 (IBP5)38 and fibulin 4 are known to regulate production of various ECM components including collagen and fibronectin. Additionally, fibulin 4 affects the activation of TGFβ signalling and αSMA production.37 IBP5 is upregulated in the fibrotic skin lesions of SSc patients39 and fibrotic intestine in Crohn’s disease.40 Chemokine ligand 2 (CCL2) recruits macrophages critical in dermal ECM turnover on tissue injury.41 STRING analysis (Search Tool for the Retrieval of Interacting Genes/proteins) indicated an interacting network of ECM proteins including POSTN, matrix metalloproteinase-2 (MMP2) and chemokine (C-C motif) ligand 2 (CCL2) with further connections to PCOC1 and TSP2 (figure 4G). Gene ontology term analysis used to identify the biological processes affected by the ablation of the three integrins revealed that these candidates fall predominantly in the category ‘extracellular structure organisation’ (figure 4H). Together, these results show that the lack of the three integrins on cultured fibroblasts induces quantitative and qualitative changes in the ECM deposited by these cells, which can impact cell spreading and cytoskeletal organisation.

Impaired mechanotransduction in tKO fibroblasts in vitro

To investigate the mechanisms of the cell-intrinsic differences in cell spreading (figure 3), we analysed the ability of the cells to generate traction stresses at FAs, which is required for this process. Traction force microscopy of WT and tKO fibroblasts seeded on polyacrylamide gels coated with collagen I revealed that tKO fibroblasts produced about 50% less traction stress than WT fibroblasts (figure 5A,B). To further substantiate the defect in actomyosin contractility, we quantified the capacity of WT and tKO fibroblasts to contract and remodel collagen I fibrils in floating collagen lattices, a function that relies on cell traction.42 As expected, lattices with tKO fibroblasts remained significantly larger, reflecting severely reduced mechanotransduction needed to remodel collagen I networks (figure 5C,D).

Figure 5

tKO fibroblasts show significantly altered properties essential for efficient mechanotransduction and collagen remodelling. (A, B) Traction stress is significantly reduced in tKO fibroblasts seeded onto collagen I-coated polyacrylamide gels with a nominal stiffness of 8.44 kPa. N=3 mice per genotype, N>12 cells per mouse per experiment. (C) Representative images of floating collagen lattices seeded with fibroblasts for the indicated time reveal (D) significantly reduced capacity of tKO fibroblasts to remodel and contract the lattices. N=fibroblasts from 4 mice per genotype. (E) Western blot of fibroblasts seeded on collagen I for 6 hours showing (F) significantly reduced levels of active YAP, (G) unaltered amount of total YAP and (H) decreased ratio of active YAP: total YAP in tKO fibroblasts. (I) Western blot of fibroblasts seeded on fibronectin for 6 hours showing slightly but not significantly reduced levels of (J) active YAP, (K) total YAP and (L) ratio of active YAP: total YAP in tKO fibroblasts. N=fibroblasts from 5 separate mice per genotype. Western blots of fibroblasts seeded for 6 hours on collagen I (M) or fibronectin (O) showing (N) significantly elevated ratio of G actin: F actin in tKO fibroblasts on collagen but (P) no significant alteration on fibronectin. N=fibroblasts from 4 separate mice per genotype. (Q–T) Representative confocal images showing fibroblasts seeded for 6 hours on (Q, R) collagen I or on (S, T) fibronectin. tKO fibroblasts on collagen show (Q) highly reduced nuclear localisation of YAP and (R) negligible active YAP, accompanied by (Q, R) drastically reduced cell spreading and absence of actin microfilaments (Phalloidin staining) and FAs (Paxillin staining). (S, T) tKO fibroblasts seeded on fibronectin do not show altered (S) nuclear YAP signals and (T) active YAP signals, however, (S, T) altered architecture of the actin cytoskeleton (Phalloidin) and of FAs (Paxillin). (U–W) FA analysis of tKO fibroblasts seeded on collagen I for 6 hours revealed a significant decrease in (U) numbers, (V) percentage of cell area occupied by, and (W) size of FAs. N=fibroblasts from 4 separate mice per genotype. Data are presented as mean±SD. P values were calculated using Mann-Whitney U test. *p<0.05, **p<0.01, Scale bars indicate 10 µm in (A) and 30 µm in (Q–T) . FAs, focal adhesions; tKO, triple knockout; WT, Wild type.

We next analysed potential downstream consequences of the reduced contractility of tKO fibroblasts by analysing YAP signalling, a key mechanosensitive transcriptional coactivator of several prosurvival, cytoskeletal and ECM genes (online supplemental figure S4).14 Adhesion to collagen I for 6 hours induced robust activation of YAP in WT fibroblasts, which was reduced by more than 50% in tKO fibroblasts (figure 5E,F), while levels of total YAP did not differ between WT and tKO fibroblasts (figure 5E,G). On fibronectin, a slight, but insignificant reduction in active YAP was seen in tKO fibroblasts (figure 5I–L). In addition, remodelling of the actin cytoskeleton was impaired in tKO fibroblasts grown on collagen I. The ratio of monomeric G-actin to filamentous F-actin was significantly elevated in tKO, indicating insufficient actin remodelling in comparison to WT fibroblasts (figure 5M,N). A similar tendency was observed in tKO cells on fibronectin, but this difference was also not significant (figure 5O,P). In addition to determining YAP levels by immunoblotting, we assessed its subcellular distribution. Staining using an antibody detecting unphosphorylated (active) nuclear YAP clearly revealed decreased nuclear YAP in the tKO fibroblasts on collagen I, in comparison to WT (figure 5Q) but active, nuclear YAP signals were completely absent in tKO fibroblasts (figure 5R). In contrast, tKO fibroblasts grown on fibronectin showed active YAP in the nuclei (Figure 5S,T), indicating that tKO fibroblasts are capable of perceiving and transducing stress on stimulation of α5β1 integrin receptors. This finding agrees with unaltered levels of active to total YAP detected by immunoblotting (figure 5I–L).

The finding of altered G-actin to F-actin levels led us to study the actin cytoskeleton in more detail. As noted previously (figure 3), the reduced spreading of tKO fibroblasts seeded on collagen I was confirmed by phalloidin staining (figure 5Q,R). In addition, stress fibres were virtually absent in tKO cells, and paxillin-positive FAs were scarce in comparison to WT cells. In contrast, organisation of F-actin stress fibres as well as formation of FAs, clearly visible in WT fibroblasts on fibronectin, were also reduced in tKO fibroblasts on fibronectin (figure 5S,T). In particular, numbers of FAs per tKO fibroblast, area occupied by and average size of FAs were significantly reduced (figure 5U–W). pFAK and pSRC signals43 were reduced in tKO fibroblasts on collagen but normal on fibronectin (online supplemental figure S5) suggesting impairment in adhesion assembly in tKO fibroblasts on collagen but not fibronectin and impaired maturation or disturbance of further downstream signalling on fibronectin.

Absence of collagen-binding integrins leads to a reduced fibrotic response in mice

To delineate the biological relevance of integrin-mediated cell-collagen interactions in vivo under conditions of induced collagen synthesis and remodelling, we subjected tKO and WT mice to intradermal bleomycin injections that result in skin fibrosis.44 Histological inspection of bleomycin lesions clearly showed attenuated fibrosis in tKO skin, reflected by significantly reduced dermal thickness (figure 6A–D) and a strong reduction of collagen content (figure 6H) in comparison to WT controls. These changes in the dermal connective tissue were confirmed by second harmonic generation analysis, showing a significant increase in collagen fluorescence intensity in WT compared with tKO skin lesions (figure 6E–G), together indicating that tKO fibroblasts deposit reduced amounts of collagen into fibrotic lesions. Ultrastructural analysis of the collagen network by transmission electron microscopy revealed that induction of fibrosis resulted in a change in collagen fibril organisation in the upper dermis in WT skin, which displayed a significantly denser packing of collagen fibrils than tKO lesions (figure 6I,J). This difference in collagen fibrillar organisation between tKO and WT skin was not observed on saline injection, serving as solvent control treatment. Histological analysis also did not reveal any obvious changes in the inflammatory infiltrate and vascularisation between tKO and WT in the bleomycin lesions. However, significantly reduced numbers of αSMA-positive myofibroblasts (figure 7D–F) and of pSMAD2-positive nuclei (figure 7A–C) were detected in fibrotic lesions of tKO than WT mice, reflecting attenuated TGFβ signalling and explaining the reduced collagen deposition in tKO fibrotic lesions (figure 6).

Figure 6

Reduced fibrotic response to bleomycin in tKO mice. (A) Representative H&E staining of skin of WT and tKO mice following repeated injections with NaCl (used as solvent control) shows (B) comparable dermal thickness (indicated by black arrow), whereas (C) bleomycin treatment results in a significant reduction in dermal thickness of tKO lesions as also seen by analysing the (D) relative dermal thickness when normalised to NaCl treated skin and (H) collagen content in tKO lesions. Representative images of skin sections (E) acquired using second harmonic generation microscopy showing collagen in green and nuclei (DAPI) in blue. Quantification shows (F) comparable collagen signals in NaCl treated skin of WT and tKO mice, however, (G) significantly lower collagen fluorescence intensity in bleomycin-injected tKO skin. N=4–5 per genotype, each data point represents skin or fibroblasts from a separate mouse. (I) Representative transmission electron micrographs showing more densely packed collagen fibrils in the upper dermis in bleomycin-injected lesions of WT than in tKO lesions. In contrast, collagen fibril arrangement is comparable in WT and tKO skin on NaCl injection. (J) Quantification of area occupied by collagen fibrils (per cent), indicating that the interfibrillar space is larger in tKO than in WT lesions. The red dotted line marks the epidermal-dermal basement membrane. Data are presented as mean±SD. P values were calculated using Mann Whitney U test for (B–D, F–H). *p<0.05, **p<0.01. Scale bars indicate 500 µm in (A), 100 µm in (E) and 500 nm in (I). tKO, triple knockout; WT, Wild type.