Assessment of corneal substrate biomechanics and its effect on epithelial stem cell maintenance and differentiation

Abstract

Whilst demonstrated extensively in vitro, the control of cell behaviour via modulation of substrate compliance in live tissues has not been accomplished to date. Here we propose that stem cells can be regulated solely through in situ modulation of tissue biomechanics. By first establishing, via high-resolution Brillouin spectro-microscopy, that the outer edge (limbus) of live human corneas has a substantially lower bulk modulus compared to their centre, we then demonstrate that this difference is associated with limbal epithelial stem cell (LESC) residence and YAP-dependent mechanotransduction. This phenotype-through-biomechanics correlation is further explored in vivo using a rabbit alkali burn model. Specifically, we show that treating the burnt surface of the cornea with collagenase effectively restores the tissue's mechanical properties and its capacity to support LESCs through mechanisms involving YAP suppression. Overall, these findings have extended implications for understanding stem cell niche biomechanics and its impact on tissue regeneration.

Fig. 1

The corneal limbus has distinct mechanic properties. a Schematic representation of the Brillouin spectro-microscope (DPSS laser diode-pumped solid-state laser; PBS polarising beam-splitter; QWP quarter-wave plate; PM-SMF polarisation-maintaining single-mode fibre), showing the confocal microscope, elastic scattering filter, VIPA spectrometer, and the sample in its immersion medium. Whole human corneas maintained intact after enucleation were kept immersed in Carry-C to preserve the tissue’s natural thickness, hydration, and transparency state (inset) during Brillouin spectro-microscopy (BSM). b Representative organ-wide X–Z scans of Brillouin frequency shifts from healthy intact human corneas (n = 3). Brillouin spectra were acquired with a sample spacing of 20 µm, over a 12 × 3 mm (600 × 150 = 9 × 10⁴ points) X–Z transverse section, corresponding to the full corneal width and depth, respectively. These high-resolution scans revealed transversal striæ with high-Brillouin frequency shifts, mostly in the posterior stroma and extending obliquely to the mid or anterior region of the tissue, which probably corresponded to lamellar undulations thought to protect the stromal ultrastructure and shape from external mechanical shocks and the subsequent increase in intraocular pressure. c Representative Y–Z scan of Brillouin frequency shifts of central cornea performed every 2.5 µm, showing a distinct epithelium (Ep; depth = 0–50 µm), sub-epithelial layer (Sub; 50–65 µm), and stroma (St; >65 µm), as well as the location of the Bowman’s layer (arrowhead). d Representative Y–Z-scan of Brillouin frequency shifts of corneal limbus performed every 5 µm, showing the epithelium (depth = 0–50/60 µm), sub-epithelial layer (60–120 µm), and stroma (>120 µm). e BSM measurements performed through the anterior region of the central cornea and limbus were statistically analysed (two-way ANOVA; 100 individual measurements per area, per experiment), along with the average (centre line) ± S.D. values (whiskers) from three independent experiments (n = 3; **p < 0.01 and ***p < 0.001). Shift value distribution was consistently similar between individual corneas but distinct in central vs. limbus

Fig. 2

LESC residency corresponds to limbal regions with distinctly softer mechanical properties. Representative confocal immunofluorescence micrographs of corneal epithelial cell markers and ECM components were used to reconstruct in 3D the limbus and central cornea. Expression of limbal markers ABCG2 (a), CK15 (b), and ΔNp63 (c), were expressed by limbal epithelial cells supported by collagen-I-positive/collagen-V-negative matrix, but not by central corneal epithelium (a–c). Markers such as CK3 (a), β-catenin (b), and integrin-α3β1 (c) showed higher expression in the central corneal epithelium. Histochemical distinction between central cornea and limbus was further evidenced by the basement membrane markers and corresponding receptors. The limbus showed a discontinuous distribution of laminin-1 compared to the central cornea, and the specific expression of laminin-γ3 and integrin-α9 (d). Conversely, the central corneal epithelium was positive for CK3+12 and collagen-VII (e). Cell nuclei were detected using DAPI. f Marker expression was quantified and represented as average ± S.D. from all three independent experiments (n = 3; *** corresponds to p < 0.001 after one-way ANOVA). Source data are provided as a Source Data file. Scale bars, 100 µm

Fig. 3

LESC-like phenotype can be controlled through fine modulation of the mechanical properties of collagen substrates. a Schematic representation of the collagenase treatment method used to modulate the stiffness of compressed collagen gels. High-density, plastic-compressed collagen gels were softened with collagenase type-I solution in well-defined areas (ring-shaped, semi-circular, or entire gel surface) for up to 60 min. b Analysis of compressed collagen gel density after collagenase treatment. The regions corresponding to collagenase-treated collagen gels showed increased transparency under bright-field imaging (upper left panel; scale bar, 5 mm) and lower collagen density compared to untreated regions, as indicated by the lower collagen-I detection by immunofluorescence confocal microscopy (upper right panel; scale bar, 50 µm). The confocal Z-scans (lower panel) demonstrated that the difference in signal intensity between treated and untreated areas was not restricted to the surface, indicating that the collagenase in solution acted through the entire depth of the compressed collagen maintaining a defined treatment zone. c Average frequency ± S.D. of the elastic modulus, E (MPa), of treated (orange) and untreated gels (blue bars) calculated from three independent experiments using force–distance spectroscopy (n = 3). The frequency histograms of treated and untreated gels were used to calculate Gaussian curves by non-linear regression (orange and blue areas, respectively), with corresponding E = 0.7 ± 0.4 and 4.8 ± 3.5 MPa. d Effects of substrate stiffness on the expression of CK3 (differentiation) and CK15 (LESC protein marker) in cells grown for 4 weeks on treated and untreated regions of collagen gels (green staining) after normalisation for total cell number (red staining), and represented as average ± S.D. from three independent experiments (n = 3; ** and *** corresponds to p < 0.01 and 0.001 after one-way ANOVA, respectively). Source data are provided as a Source Data file. e Creation of a pseudo-limbus. Cells growing on ring-shaped softened areas (week 0) expressed higher levels of CK15 compared to the high CK3-positive cells growing on the untreated (stiffer) central region of the collagen gels, up to 4 weeks in culture (scale bar, 5 mm)

Fig. 4

Softening of corneal tissue with collagenase increases expression of limbal markers during ex vivo re-epithelialisation. a Representative whole-cornea X–Z scan of Brillouin frequency shifts measured at a sample spacing of 20 µm from healthy intact human corneas after collagenase treatment (n = 3). Insets correspond to (b) the regions of the central cornea softened with collagenase (treated) or left untreated (control) analysed at very high-resolution scanning, using a sampling distance of 2.5 µm, showing a distinct epithelium (Ep), sub-epithelial layer (Sub), and stroma (St), as well as the location of the Bowman’s layer (white arrowhead). This detailed analysis evidenced the loss of the stiffer Bowman’s layer (black arrowhead) and reduced Brillouin frequency shifts in both epithelium and stroma resulting from collagenase treatment. c Brillouin frequency shifts from the anterior part of both treated (white) and control regions (grey bars) of the central cornea and limbus. The plot represents average ± S.D. of measurements taken from the epithelium, sub-epithelium, and stroma from three independent experiments (100 individual measurements per each individual area, per experiment; n = 3; ns corresponds to p > 0.05 and *, **, and *** to p < 0.05, 0.01, and 0.001 after two-way ANOVA, respectively). Source data are provided as a Source Data file. d Representative confocal immunofluorescence micrographs (3D reconstruction) of laminin-1 (green) and collagen-VII distribution (red staining) in the central cornea, before (upper) and after collagenase treatment (central panel), and after re-epithelialisation (lower panel). e Representative confocal immunofluorescence micrographs (3D reconstruction) of collagenase-softened central cornea after re-epithelialisation. Epithelial cells repopulating softened corneas expressed ABCG2 and CK15 (limbal markers; green) while showing lower CK3+12 and β-catenin expression (differentiation markers; red staining) compared to cells growing on the stiffer, untreated corneas (control). f Epithelial cells expressed integrin-α9 and deposited laminin-γ3 when grown on collagenase-softened corneal substrates, but not on untreated central cornea (control). Cell nuclei were detected using DAPI in all three independent experiments. Scale bars, 50 µm

Fig. 5

Softening of central corneal tissue with collagenase increases expression of limbal markers in vivo. a Schematic representation of the collagenase treatment method used to soften the central region of intact corneas in live rabbits. Clinical observation and slit-lamp examination was performed 1 and 5 days post-intervention, and compared to pre-intervention results (baseline). The central corneal epithelium was also analysed by confocal immunofluorescence (3D reconstruction) 5 days after collagenase treatment (softened), and the corresponding marker expression was quantified. Cells on softened corneas expressed (b) higher levels of ABCG2, CK15, ΔNp63, integrin-α9, and laminin-γ3 (limbal markers) and lower levels of CK3+12 and integrin-α3β1 (differentiation markers) compared to cells growing on the stiffer, untreated corneas (control). Collagen-IV and collagen-VII were also detected in collagenase-treated (softened) corneas, albeit at lower levels compared to control. c The expression of the mechanotransduction marker YAP was also significantly lower in the CK15-positive cells on softened corneas, where it mostly presented a non-nuclear (inactive) form compared to that in CK3-positive cells on untreated corneas (control). Cell nuclei were detected using DAPI. Scale bars, 50 µm. Marker expression was represented as average ± S.D. from three independent experiments (n = 3; *, **, and *** corresponds to p < 0.05, 0.01, and 0.001 after one-way ANOVA, respectively). Source data are provided as a Source Data file

Fig. 6

Softening of alkali-burned corneal tissue with collagenase restores expression of limbal markers ex vivo. a Representative topography of limbal sub-epithelial matrix after application of PBS (control), 0.5 M NaOH (alkali), or NaOH followed by collagenase softening (treated) of whole human corneas, analysed by atomic force microscopy (AFM). AFM scans from three independent experiments showed clearance of ECM components other than collagen fibrils in tissue subjected to alkali burn, and reduced collagen fibril density after collagenase treatment. False colour depth, 500 nm. b Force–distance spectroscopy analysis showed that the burn (alkali) significantly stiffened the limbal sub-epithelial matrix, and that subsequent collagenase softening (treated) restored the mechanical properties of the original tissue (control). The graph represents the distribution of calculated values of elastic modulus, E (MPa), and corresponding average (centre line) ± S.D. (whiskers) from three independent experiments (n = 3; ***p < 0.001). c Representative confocal immunofluorescence micrographs (3D reconstruction) of re-epithelialised limbus after alkali burn and collagenase treatment, with (d) corresponding marker expression quantification. Epithelial cells repopulating the alkali-burned limbal tissue ex vivo expressed significantly lower levels of limbal markers ABCG2, ΔNp63, CK15, and integrin-α9 while showing higher expression of CK3 differentiation markers compared to cells growing on control tissue. e The mechanotransduction marker YAP also changed significantly in tissues made stiffer by alkali, with repopulating cells showing increased YAP expression and a predominant nuclear localisation. However, collagenase treatment successfully restored the ability of burned limbal tissues to support cells expressing LESC, differentiation, and mechanotransduction markers at control levels. Cell nuclei were detected using DAPI. Marker expression was represented as average ± S.D. from three independent experiments (n = 3; *, **, and *** corresponds to p < 0.05, 0.01, and 0.001 after one-way ANOVA, respectively). Source data are provided as a Source Data file. Scale bars, 1 µm (a), 50 µm (c, e)

Fig. 7

Treatment of alkali-burned limbus with collagenase restores expression of limbal markers in vivo. a Representative images of chemically burned rabbit corneas (alkali) and of alkali-burned corneas receiving collagenase treatment (treated), before (baseline) and after burn (day 0), as well as at day 2 and 7 post-burn. The haze resulting from the burn was delimited (traced line), and analysed for recovery (insets), and covered area (b) at day 0 (initial wound), 2 (dark blue) and 7 (light blue bars). Values corresponded to haze area average ± S.D. from three independent experiments (n = 3). c Representative confocal immunofluorescence micrographs (3D reconstruction) of chemically burned (alkali) and collagenase-treated burned limbus (treated) at day 7, with (d) corresponding marker expression quantification. Cell nuclei were detected using DAPI. Epithelial cells repopulating the alkali-burned limbus in vivo expressed significantly lower levels of limbal markers CK15, ΔNp63, ABCG2, and integrin-α9 while showing higher expression of the CK3 differentiation marker compared to the undamaged limbus (control) tissue. However, collagenase treatment successfully restored the ability of the burned limbus to support cells as in control tissues. e The expression of the mechanotransduction marker YAP was also significantly lower in the ΔNp63-positive, CK3-negative cells on treated and control limbus tissues, where it predominantly presented a non-nuclear, inactive form. In contrast, CK3-positive, ΔNp63-negative cells on the alkali-burned tissues showed significantly higher, and mostly nuclear, YAP expression. The expression of markers in d, e was represented as average ± S.D. from three independent experiments (n = 3; *, **, and *** corresponds to p < 0.05, 0.01, and 0.001 after one-way ANOVA, respectively). Source data are provided as a Source Data file. Scale bars, 50 µm

Fig. 8

Consistency of overall results from the multiple experimental model systems under study. Epithelial cells (re)populating collagenase-softened (treated) substrates expressed significantly higher levels of limbal markers and lower levels of differentiation and active mechanotransduction markers compared to the untreated (control) tissues, independently of the system’s degree of complexity (i.e., topography, biochemical composition). In addition, the maintenance of limbal cell morphology and failure to elicit major pro-inflammatory responses after tissue softening were similarly observed for all in vitro, ex vivo, and in vivo experiments. The comparable results obtained on both simple and complex substrates supported the notion that the cellular effects were derived from the modulation of tissue biomechanics, and not due to tissue topography or composition (i.e., exposure to cryptic biochemical cues or growth factors)

Fig. 9

Schematic representation of overall results from in vivo experiments and proposed mechanism of action. Briefly, stiffening of the limbus caused by alkali burn induces the activation and nuclear translocation of YAP, a key regulator of mechanotransduction. The nuclear YAP then acts as a transcription factor, promoting Wtn/β-catenin and suppressing Sox9 signalling, which in turn induce cell differentiation and lead to loss of LESCs. By softening the burned limbus with collagenase, repopulating stem/progenitor cells retain a mostly inactive YAP, with corresponding high Sox9 and low Wtn/β-catenin levels promoting LESC phenotype maintenance