Chronic inflammation imposes aberrant cell fate in regenerating epithelia through mechanotransduction

Abstract

Chronic inflammation is associated with a variety of pathological conditions in epithelial tissues, including cancer, metaplasia and aberrant wound healing. In relation to this, a significant body of evidence suggests that aberration of epithelial stem and progenitor cell function is a contributing factor in inflammation-related disease, although the underlying cellular and molecular mechanisms remain to be fully elucidated. In this study, we have delineated the effect of chronic inflammation on epithelial stem/progenitor cells using the corneal epithelium as a model tissue. Using a combination of mouse genetics, pharmacological approaches and in vitro assays, we demonstrate that chronic inflammation elicits aberrant mechanotransduction in the regenerating corneal epithelium. As a consequence, a YAP-TAZ/β-catenin cascade is triggered, resulting in the induction of epidermal differentiation on the ocular surface. Collectively, the results of this study demonstrate that chronic inflammation and mechanotransduction are linked and act to elicit pathological responses in regenerating epithelia.

Figure 1. CSCM in Notch1^Δ mice is associated with an augmented and chronic inflammatory response.

(a) Schematic depiction of the experimental strategy. (b) Histology of WT (Notch1^lox/lox) and Notch1^lox/lox corneal tissue after repeated injury (representative of data from 16 WT corneas and 20 Notch1^Δ corneas). Upper panels = H&E staining, lower panels = immunofluorescence for K12, K1 and CD45. Large panels are low magnification tiled images. Insets on the upper left corner of the H&E images show gross phenotype on the ocular surface. Black (H&E) and white (immunofluorescence) outlined insets show high magnification images of the indicated regions. (c-d) Quantification of the proportion of CD45⁺ cells in the cornea of WT and Notch1^Δ corneas 24 hours after a single injury (c) and 21 days after repeated corneal injury (d). Proportions were measured by performing flow cytometry on dissociated corneas (n = 6 biological replicates for each genotype over three independent experiments. Each replicate consists of cells pooled from 4 corneas isolated from 2 mice of each genotype). (e) QRT-PCR analysis for the indicated cytokines in WT (Notch1^lox/lox) and Notch1^Δ corneal epithelial cells 24 hours after a single corneal injury. Data are expressed relative to the expression in WT unwounded corneal epithelial cells (n = 6 biological replicates for each genotype over three independent experiments. Each replicate consists of corneal epithelial tissue pooled from 6 corneas isolated from 3 mice of each genotype). Scale bars represent 500μm on tiled images and 5μm on all other histological images. Scale bars on images showing gross morphology of corneas represent 1mm. St – Stroma. * P <0.01, ** P<0.05 (unpaired, two tailed t-tests). Error bars represent standard deviation.

Figure 2. Chronic inflammation is necessary and sufficient to induce CSCM.

(a) Schematic depiction of the experimental strategy used to determine if chronic inflammation is necessary for the induction of CSCM in Notch1^Δ mice. (b) Immunofluorescent staining for K12, K1 and CD45 on Notch1^Δ corneas treated with ophthalmic gel (control) or the anti-inflammatory gel Tobradex. Data are representative of 6 corneas per treatment over three independent experiments. Large panels are low magnification tiled images. White outlined insets show high magnification images of the indicated regions. (c) Schematic depiction of the experimental strategy used to determine if chronic inflammation is sufficient to induce CSCM. (d) Immunofluorescent staining for K12, K1 and CD45 on WT (non-transgenic littermates) and K14TSLPTg corneas after the procedure shown in (e). Data are representative of 6 corneas per genotype over three independent experiments. Large panels are low magnification tiled images. White outlined insets show high magnification images of the indicated regions. Scale bars represent 500μm on tiled images and 5μm on all other images. St – Stroma.

Figure 3. CSCM is induced in limbal and peripheral cells during repair.

(a) Immunofluorescent staining for K12 and K1 in WT (Notch1^lox/lox) and Notch1^Δ corneas at 0, 6 and 24 hours after a second corneal injury. Data are representative of 6 corneas per genotype over two independent experiments for each timepoint analysed. Large panels are low magnification tiled images. Insets outlined in green, red and yellow show high magnification images of the limbus, peripheral cornea and central cornea respectively. (b) XY scatter plots showing K12 and K1 expression in the limbus, peripheral cornea and central cornea 24 hours after a second corneal injury. Each data point represents mean fluorescence intensity measured from an individual cornea. Grey boxes = WT (Notch1^lox/lox), black boxes = Notch1^Δ (n = 6 corneas for each genotype over three independent experiments). (c) Model predicting the cellular origin of CSCM. Following injury to the central cornea, stem/progenitor cells in the peripheral cornea become activated and proliferate to generate daughter cells that mediate wound closure. In Notch1^Δ mutants, chronic inflammation promotes epidermal differentiation of activated stem/progenitor cells or their immediate progeny. This model therefore predicts that all epidermal lineage cells in Notch1^Δ corneas will be derived from peripheral stem/progenitor cells and therefore be continuous with the peripheral cornea. (d) Immunofluorescent staining for K12 and K1 on wholemount corneal epithelial tissue from WT (Notch1^lox/lox) and Notch1^Δ mice after repeated injury. Data are representative of 12 corneal wholemounts per genotype over four independent experiments. Images shown are low magnification tiled images. Scale bars represent 500μm on tiled images and 5μm on all other images.

Figure 4. Chronic inflammation promotes CSCM via elevated β-catenin signalling.

(a) Immunohistochemistry for β-catenin on WT (Notch^lox/lox) and Notch1^Δ corneas after repeated injury. Data are representative of 8 corneas per genotype over three independent experiments. (b) Immunofluorescent staining for β-catenin in limbus, peripheral cornea and central cornea of WT (Notch^lox/lox) and Notch1^Δ corneas 24 hours after the second corneal injury. Data are representative of 6 corneas isolated over three independent experiments. (c) Quantification of relative β-catenin expression in limbus, peripheral cornea and central cornea 24 hours after the second corneal injury. Black bars = WT (Notch1^lox/lox), grey bars = Notch1^Δ (n = 6 corneas for each genotype over three independent experiments). Values for expression levels are relative values normalised to the expression level in the conjunctiva of each sample, determined by mean fluorescence intensity. (d) Schematic depiction of the experimental strategy used to determine if β-catenin is necessary for the induction of CSCM. (e) Immunofluorescent staining for K12, K1 and CD45 on Notch1^Δ and Notch1^Δ:Ctnnb1^Δ corneas after the procedure outlined in (d). Data are representative of 8 corneas per genotype over three independent experiments. Large panels are low magnification tiled images. White outlined insets are high magnification images of the indicated regions. (f) Schematic depiction of the experimental strategy used to determine if elevated β-catenin is sufficient to induce CSCM. (g) Immunofluorescent staining for K12, K1 and CD45 on WT (Ctnnb1^{lox(ex3)/lox(ex3)}) and Ctnnb1^ΔEx3 corneas 7 days after a single corneal injury. Data are representative of 8 WT corneas and 10 Ctnnb1^ΔEx3 corneas over four independent experiments. Large panels are low magnification tiled images. White outlined insets are high magnification images of the indicated regions. Scale bars represent 500μm on tiled images and 5μm on all other images. St – Stroma, Limb – limbus, Per – Periphery, Cen – Centre, NS – Not Significant. * P <0.01, ** P<0.05 (unpaired, two tailed t-tests). Error bars represent standard deviation.

Figure 5. Increased ECM deposition in the corneal stroma in response to aberrant inflammation.

(a) Immunofluorescent staining for K14 and Tenascin C (TenC) in WT (Notch1^lox/lox) and Notch1^Δ corneas. Upper panels show uninjured cornea, middle panels show corneal tissue 24 hours after the first injury, lower panels show corneal tissue 24 hours after the second injury. Data are representative of 6 corneas per genotype for each timepoint analysed over three independent experiments. (b) Relative quantification of Tenascin C expression in the limbus, peripheral cornea and central cornea. Upper panel shows uninjured cornea, middle panel shows corneal tissue 24 hours after the first injury, lower panel show corneal tissue 24 hours after the second injury. Values for expression levels are relative values normalised to the expression level in the conjunctival stroma of each sample, determined by mean fluorescence intensity. Black bars = WT (Notch1^lox/lox), grey bars = Notch1^Δ (n = 6 corneas for each genotype at each time point over three independent experiments). (c) Immunofluorescent staining for K14 and Periostin (POSTN) in WT (Notch1^lox/lox) and Notch1^Δ corneas. Upper panels show uninjured cornea, middle panels show corneal tissue 24 hours after the first injury, lower panels show corneal tissue 24 hours after the second injury. Data are representative of 6 corneas per genotype for each timepoint analysed over three independent experiments. (d) Relative quantification of Periostin expression in the limbus, peripheral cornea and central cornea in uninjured tissue (upper panel), corneal tissue 24 hours after the first injury (middle panel) and corneal tissue 24 hours after the second injury (lower panel). Values for expression levels are relative values normalised to the expression level in the conjunctival stroma of each sample, determined by mean fluorescence intensity. Black bars = WT (Notch1^lox/lox), grey bars = Notch1^Δ (n = 6 corneas for each genotype at each time point over three independent experiments). Scale bars on tiled images represent 500μm. St – Stroma, Limb – limbus, Per – Periphery, Cen – Centre, NS – Not Significant. * P <0.01, ** P<0.05 (unpaired, two tailed t-tests). Error bars represent standard deviation.

Figure 6. Activation of mechanotransduction in the CE in response to aberrant inflammation.

(a) Schematic depiction of key molecular mediators and/or sensors of mechanotransduction in epithelial cells. (b-d) Immunofluorescent staining for pFAK (b), ROCK2 (c) and YAP/TAZ (d) in the limbus, peripheral cornea and central cornea of WT (Notch1^lox/lox) and Notch1^Δ mice 24 hours after the second corneal injury. In (d) images without DAPI are shown to enable clearer visualisation of nuclear YAP:TAZ. Data are representative of 6 corneas per genotype over three independent experiments. (e) Quantification of FAK phosphorylation in the limbus, peripheral cornea and central cornea 24 hours after the second injury. Values for expression levels are relative values normalised to the expression level in the conjunctiva of each sample, determined by mean fluorescence intensity. Black bars = WT (Notch1^lox/lox), grey bars = Notch1^Δ (n = 6 corneas for each genotype over three independent experiments). (f,g) Quantification of nuclear:cytoplasmic ratio of ROCK2 (f) and YAP/TAZ (g) in the limbus, peripheral cornea and central cornea 24 hours after the second injury. Black bars = WT (Notch1^lox/lox), grey bars = Notch1^Δ (n = 6 corneas for each genotype over three independent experiments). St – Stroma, Limb – limbus, Per – Periphery, Cen – Centre, NS – Not Significant. * P <0.01, ** P<0.05 (unpaired, two tailed t-tests). Error bars represent standard deviation. Scale bars represent 5μm.

Figure 7. CSCM is associated with increased tissue stiffness and mechanical stimuli.

(a) Immunofluorescence for K14 and Tenascin C (upper panels) and corresponding Atomic Force Microscopy (AFM) nanomechanical property measurement (lower panels) of limbus, peripheral cornea and central cornea after repeated injury. Elastic modulus was determined using AFM force volume mode. Data are representative of 4 corneas per genotype over 4 independent experiments. (b) Quantification of the elastic modulus (kPa) of stromal tissue from the limbus, peripheral cornea and central cornea of the regions shown in (a). Red lines in each box represent median elastic modulus value of 256 force volume measurements in the regions shown in (a). Red lines in each whisker represent outliers. Boxes represent the middle 50% of the data. (c,d) Immunfluorescence for β-catenin, K12 and K1 (c) or YAP/TAZ, K12 and K1 (d) on PCESCs cultured on soft (upper panels) or stiff (lower panels) substrates. Data are representative of 6 individual cultures over 2 independent experiments. (e-g) Quantification of β-catenin expression (e), YAP/TAZ nuclear:cytoplasmic ratio (f) and the proportion of K12⁺K1^- and K12^-K1⁺ cells (g) in PCESCs cultured on soft or stiff substrates (n = 6 for each condition, where one replicate represents quantification from a single culture over 2 independent experiments). For (e), β-catenin expression is determined by mean fluorescence intensity. Scale bars in (a) represent 15μm. Scale bars in (c) and (d) represent 20 μm. Limb – limbus, Per – Periphery, Cen – Centre, * P <0.01, *** P<0.1 (unpaired, two tailed t-tests). Error bars represent standard deviation (e-g) or maximal values/1.5x interquartile range (b).

Figure 8. Manipulation of mechanotransduction affects corneal cell fate

(a) Schematic depiction of the experimental strategy used to determine if inhibition of mechanotransduction prevents CSCM. (b) Immunofluorescent staining for K12, K1 and CD45 on Notch1^Δ corneas treated with vehicle (DMSO) or the FAK inhibitor PF562271 during the procedure outlined in (a). Data are representative of 8 corneas per treatment over four independent experiments. Large panels are low magnification tiled images. White outlined insets show high magnification images of the indicated regions. (c) Quantification of the proportion of Notch1^Δ corneas exhibiting corneal or epidermal differentiation during wound closure following treatment with vehicle or PF562271 (n = 8 corneas for each treatment over four independent experiments). (d) Immunofluorescent staining for K12, K1 and CD45 on Notch1^Δ corneas treated with vehicle (PBS) or the ROCK inhibitor Y27632 after the procedure outlined in (a). Data are representative of 10 corneas per treatment over four independent experiments. Large panels are low magnification tiled images. White outlined insets show high magnification images of the indicated regions. (e) Quantification of the proportion of Notch1^Δ corneas exhibiting corneal or epidermal differentiation during wound closure following treatment with Vehicle or Y27632 (n = 10 corneas for each treatment over four independent experiments). (f) Schematic depiction of the experimental strategy used to determine if ablation of YAP/TAZ is sufficient to induce CSCM during repair. (g) Immunofluorescent staining for K12, K1 and CD45 on WT (Yap^+/+:Taz^+/+:K14Cre^ER) and YAP^Δ:TAZ^Δ corneas after the procedure outlined in (f). Data are representative of 6 corneas for WT and 7 corneas for YAP^Δ:TAZ^Δ isolated over two independent experiments. Large panels are low magnification tiled images. White outlined insets show high magnification images of the indicated regions. (h-i) Immunofluorescent staining for YAP/TAZ, K12 and K1 (h) and β-catenin, K12 and K1 (i) in WT and YAP^Δ:TAZ^Δ corneas as indicated. Data are representative of 6 corneas for WT and 7 corneas for YAP^Δ:TAZ^Δ isolated over two independent experiments. (j) Model depicting how chronic inflammation imposes aberrant cell fate on corneal epithelial stem cells via increased tissue stiffness and mechanotransduction. Scale bars represent 500μm on tiled images and 5μm on all other images. St – Stroma.