Differential effects of Hsp90 inhibition on corneal cells in vitro and in vivo

Abstract

The transformation of quiescent keratocytes to activated fibroblasts and myofibroblasts (KFM transformation) largely depends on transforming growth factor beta (TGFβ) signaling. Initiation of the TGFβ signaling cascade results from binding of TGFβ to the labile type I TGFβ receptor (TGFβRI), which is stabilized by the 90 kDa heat shock protein (Hsp90). Since myofibroblast persistence within the corneal stroma can result in stromal haze and corneal fibrosis in patients undergoing keratorefractive therapy, modulation of TGFβ signaling through Hsp90 inhibition would represent a novel approach to prevent myofibroblast persistence. In vitro, rabbit corneal fibroblasts (RCFs) or stratified immortalized human corneal epithelial cells (hTCEpi) were treated with a Hsp90 inhibitor (17AAG) in the presence/absence of TGFβ1. RCFs were cultured either on tissue culture plastic, anisotropically patterned substrates, and hydrogels of varying stiffness. Cellular responses to both cytoactive and variable substrates were assessed by morphologic changes to the cells, and alterations in expression patterns of key keratocyte and myofibroblast proteins using PCR, Western blotting and immunocytochemistry. Transepithelial electrical resistance (TEER) measurements were performed to establish epithelial barrier integrity. In vivo, the corneas of New Zealand White rabbits were wounded by phototherapeutic keratectomy (PTK) and treated with 17AAG (3× or 6× daily) either immediately or 7 days after wounding for 28 days. Rabbits underwent clinical ophthalmic examinations, SPOTS scoring and advanced imaging on days 0, 1, 3, 7, 10, 14, 21 and 28. On day 28, rabbits were euthanized and histopathology/immunohistochemistry was performed. In vitro data demonstrated that 17AAG inhibited KFM transformation with the de-differentiation of spindle shaped myofibroblasts to dendritic keratocyte-like cells accompanied by significant upregulation of corneal crystallins and suppression of myofibroblast markers regardless of TGFβ1 treatment. RCFs cultured on soft hydrogels or patterned substrates exhibited elevated expression of α-smooth muscle actin (αSMA) in the presence of 17AAG. Treatment of hTCEpi cells disrupted zonula occludens 1 (ZO-1) adherens junction formation. In vivo, there were no differences detected in nearly all clinical parameters assessed between treatment groups. However, rabbits treated with 17AAG developed greater stromal haze formation compared with controls, irrespective of frequency of administration. Lastly, there was increased αSMA positive myofibroblasts in the stroma of 17AAG treated animals when compared with controls. Hsp90 inhibition promoted reversion of the myofibroblast to keratocyte phenotype, although this only occurred on rigid substrates. By contrast, in vivo Hsp90 inhibition was detrimental to corneal wound healing likely due to impairment in corneal epithelial closure and barrier function restoration. Collectively, our data demonstrated a strong interplay in vitro between biophysical cues and soluble signaling molecules in determining corneal stromal cell phenotype.

Figure 1:. Hsp90 inhibition alters stromal cell morphology without altering viability.

(A) Phase contrast images demonstrate a change in cell morphology from spindle shaped fibroblast/myofibroblasts to stellate/dendritic keratocyte-like with 500 nM 17AAG treatment in the presence/absence of 10 ng/ml TGFβ1 treatment. Scale bar is 100 μm. (B) Calcein-AM/ethidium homodimer labelling demonstrates that 500 nM 17AAG did not detrimentally affect cell viability. Scale bar is 100 μm. Staurosporine was used as positive control for loss of cell viability while untreated cells were negative control. Images are representative ones observed from three independent experiments. Histograms are data from n =3 independent experiments, with mean ± standard deviation.

Figure 2:. Hsp90 inhibition upregulates keratocyte markers while suppressing myofibroblast markers in vitro.

(A) Relative mRNA expression of keratocan (KERA), collagen 3A1 (COL3A1), aldehyde dehydrogenase 1A1 (ALDH1A1) and αSMA (ACTA2) by rabbit corneal stromal cells treated with DMSO (vehicle control) or 500 nM 17AAG in the presence/absence of 10 ng/ml TGFβ1 for 3 days. (B) Representative fluorescence images of RCFs immunolabelled for expression of αSMA, F-actin, and counterstained with DAPI. Images are representative of three independent experiments. Bar plots represent data expressed as mean ± standard error in mean (SEM) along with individual data points from n = 3 independent experiments. Two-way ANOVA followed by Tukey’s multiple comparisons was performed to establish statistical significance. *p<0.05, **p<0.01, ***p<0.001 compared with DMSO control. #p<0.05 comparing +TGFβ1 with −TGFβ1 cells within the same treatment group (i.e. 17AAG vs DMSO).

Figure 3:. Hsp90 inhibition immediately after PTK wounding does not improve stromal wound healing.

(A) Conjunctivitis score determined by the presence of conjunctival congestion, chemosis and discharge, (B), Uveitis score determined by the presence of aqueous flare, anterior chamber cell and iridal hyperemia (C), percentage epithelial wound healing determined by the quantitative assessment of fluorescein staining in the PTK treated eye, and (D) percentage of rabbits with epithelial re-ulceration over 28 days. Data are mean ± standard deviation from n = 6 animals.

Figure 4:. Hsp90 inhibition immediately after PTK wounding results in increased stromal haze formation.

(A) Representative clinical images demonstrating stromal haze in both control and 17AAG treated groups when compared with baseline, (B) stromal haze scoring using the SPOTs system in rabbits over the course of treatment, (C) average stromal haze score as a function of treatment, and (D) representative long axis lens views of the central cornea taken at baseline and Day 21 in both control and 17AAG treated rabbits demonstrated a thickened band of hyperreflective section in the anterior stroma consistent with clinical stromal haze formation, (E) Quantitative analysis of the stromal haze documented using FD-OCT over time in control and 17AAG treated rabbits, (F) Average stromal haze thickness as a function of treatment group, are illustrated. Data are mean ± standard deviation from n = 6 animals.

Figure 5:. Hsp90 inhibition results in increased immunostaining of αSMA positive stromal cells.

Representative immunohistochemistry images from rabbit cornea on day 21, after PTK wounding, demonstrating the incidence of myofibroblasts (αSMA staining, red) in control and 17AAG treated animals. Nuclei are stained with DAPI (blue) Bar plot demonstrates mean ± standard deviation of relative fluorescence intensity quantified for αSMA immunolabel from n = 6 animals per group. One-way ANOVA followed by Tukey’s multiple comparison test was performed to establish statistical significance. ***p<0.001, ****p<0.0001 compared respective groups indicated.

Figure 6:. Substratum topography modulates cellular response to Hsp90 inhibition and TGFβ1 treatment.

(A) Relative mRNA expression for KERA and ACTA2 by RCFs cultured on either planar or 1400 nm pitch anisotropically patterned substrates and treated with DMSO (vehicle control) or 500 nM 17AAG in the presence/absence of 10 ng/ml TGFβ1 for 3 days. (B) Representative immunocytochemistry image demonstrating immunolabeling for αSMA (green), F-Actin (red), and DAPI (blue) in RCFs under various treatment conditions. Bar plots represent data expressed as mean ± standard error in mean (SEM) along with individual data points from n = 3 independent experiments. Three-way ANOVA followed by multiple comparisons was performed to establish statistical significance. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 compared with DMSO control. #p<0.05, ##p<0.01 comparing +TGFβ1 with −TGFβ1 cells within the same treatment group (i.e. 17AAG vs DMSO), &p<0.05, &&p<0.01, &&&p<0.001 comparing cells on 1400 nm patterned topography with cells on planar substrates for same treatment group (±TGFβ).

Figure 7:. Substratum stiffness modulates cellular response to Hsp90 inhibition and TGFβ1 treatment.

Relative mRNA expression in RCFs, cultured on either tissue culture plastic (TCP) or polyacrylamide hydrogels of various elastic moduli (5 kPa vs. 25 kPa), treated with DMSO (vehicle control) or 500 nM 17AAG in the presence/absence of 10 ng/ml TGFβ1 for 3 days. Bar plots represent data expressed as mean ± standard error in mean (SEM) along with individual data points from n = 3 independent experiments. Three-way ANOVA followed by multiple comparisons was performed to establish statistical significance. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 compared with DMSO control. #p<0.05, ##p<0.01 comparing +TGFβ1 with −TGFβ1 cells within the same treatment group (i.e. 17AAG vs DMSO), &p<0.05, &&p<0.01, &&&p<0.001 comparing cells on hydrogels with cells on tissue culture plastic substrates for same treatment group (±TGFβ).

Figure 8:. Hsp90 inhibition disrupts barrier function of stratified corneal epithelial cells.

Transepithelial electrical resistance (TEER) measured in stratified corneal epithelial cells treated with DMSO (vehicle control) or 17AAG (5 μM, 10 μM) over 24 h. Bar plots represent data expressed as mean ± standard error in mean (SEM) along with individual data points (duplicate) from n = 3 independent experiments. One-way ANOVA followed by Dunnett’s multiple comparison test was performed to establish statistical significance. ***p<0.001 compared with control group.