Human limbal biopsy-derived stromal stem cells prevent corneal scarring

Abstract

Conventional allograft therapy for corneal scarring is widespread and successful, but donor tissue is not universally available, and some grafts fail owing to rejection and complications such as endothelial failure. We investigated direct treatment of corneal scarring using autologous stem cells, a therapy that, if successful, could reduce the need for corneal grafts. Mesenchymal cells were expanded from small superficial, clinically replicable limbal biopsies of human cadaveric corneo-scleral rims. Limbal biopsy-derived stromal cells (LBSCs) expanded rapidly in media containing human serum, were highly clonogenic, and could generate spheres expressing stem cell genes (ABCG2, Nestin, NGFR, Oct4, PAX6, and Sox2). Human LBSCs differentiated into keratocytes expressing characteristic marker genes (ALDH3A1, AQP1, KERA, and PTGDS) and organized a thick lamellar stroma-like tissue containing aligned collagen and keratan sulfate proteoglycans when cultured on aligned nanofiber substrata. When engrafted into mouse corneal wounds, LBSCs prevented formation of light-scattering scar tissue containing fibrotic matrix components. The presence of LBSCs induced regeneration of ablated stroma with tissue exhibiting lamellar structure and collagen organization indistinguishable from that of native tissue. Because the limbus can be easily biopsied from either eye of an affected individual and LBSCs capable of corneal stromal remodeling can be expanded under xeno-free autologous conditions, these cells present a potential for autologous stem cell-based treatment of corneal stromal blindness.

Conflict of interest statement

Competing interests: The authors declare that they have no competing interests.

Copyright © 2014, American Association for the Advancement of Science.

Figures

Fig. 1. Ex vivo expansion and clonogenicity of limbal biopsy stromal cells

(A) Phase-contrast images of primary cells cultured from limbal biopsy tissue prepared by digestion with collagenase only. In initial plating (P0), red dashed lines mark islands of epithelial cells. Scale bars, 50 μm. (B) The length of time (in days) required to expand cells from P0 to P3 was compared for LBSCs prepared with dispase and collagenase (DSP + COL) or collagenase only (COL), expanding cells in either HS or FBS. Data are means ± SD (n = 4). P value determined by two-way analysis of variance (ANOVA). (C) Clonal growth of LBSCs in HS and FBS. (D) Percentage of clonogenic cells in P3 cultures. Data are means ± SD (n = 4). (E) Colony size was calculated with Fiji image analysis software. Data are means ± SD (n > 400). P values in (D) and (E) were determined by a two-sided t test. (F) Corneal fibroblasts in FBS did not exhibit clonal growth (n = 4).

Fig. 2. Gene expression during ex vivo differentiation of LBSCs

LBSCs expanded to P3 in FBS or HS were cultured in differentiation conditions on collagen gels for 1 week. mRNA was quantified as copies per nanogram of total cellular RNA, determined by quantitative polymerase chain reaction (qPCR). Data are averages ± SD from four different cell lines, each obtained from a different donor cornea. P values were determined by two-sided t test.

Fig. 3. Generation of a stroma-like three-dimensional matrix ex vivo

ECM produced by LBSCs cultured on aligned nanofiber substrate for 4 weeks was imaged by confocal microscopy capturing optical sections at different z levels above the substratum. (A) Type I collagen fibrils (green) and keratocytes (nuclei, blue; F-actin, red) are shown at different depths of the construct. (B) Thickness of collagenous matrix at 4 weeks in HS or FBS was determined from confocal analysis. Data in (B) show averages ± SD from cell lines from four different donors. Lack of significance (NS; P > 0.05) was determined by a two-sided t test. (C) Cornea-specific keratan sulfate proteoglycan (KSPG) was detected by immunoblotting. Alternate lanes show sensitivity of the heterogeneous KSPG band (130 to 300 kD) to keratanase (KSase). Mr, relative molecular mass.

Fig. 4. LBSC engraftment and stromal matrix synthesis in mouse cornea in vivo

Fluorescent DiO-labeled human LBSCs were transferred to a superficially debrided mouse cornea in a fibrin gel as depicted in fig. S3. (A) One week after wounding, whole-mount staining showed persistence of the human LBSCs (green) in the central corneal region. (B) At 1 month, histological sections immunostained with human-specific antibodies show human keratocan and collagen type I. Omission of primary antibody controlled nonspecific staining. In all images, nuclei are stained with 4′,6-diamidino-2-phenylindole (DAPI) (blue). Anterior of the eye is oriented up in each image in (B), and the corneal epithelium is visible as a dense layer of cells near the top of each image. Ctrl, control.

Fig. 5. LBSCs block deposition of fibrotic matrix in healing murine corneas

(A) Debridement-wounded mouse corneas were treated with fibrin gel only (no cells) or with 50,000 LBSCs in fibrin gel. After 4 weeks of healing, histological sections (epithelium oriented up) were stained for fibrotic markers decorin, biglycan, tenascin C, fibronectin, and hyaluronan. Images are representative of sections from three corneas for each condition. (B) Quantification of SPARC and type III collagen (COL3A1) mRNA pre-operative (Pre-op) and 2 weeks after treatment with LBSCs or no cells. Data are averages ± SD (n = 3). P values were determined by a two-sided t test.

Fig. 6. LBSC treatment influences light transmission properties of ECM deposited after debridement

(A) Macroscopic images of mouse eyes in diffuse lighting reveal opaque scars (arrow) in untreated (no cells) eyes but none in LBSC-treated corneas. (B) OCT imaging shows transverse optical sections of preoperative (Pre-op) eyes and those 2 and 4 weeks after debridement, with scarring visible as bright pixels in the corneal stroma. (C) Thresholding of high-intensity pixels in three-dimensional (3D) OCT images of individual corneas defines scarred region (green) at 2 and 4 weeks after debridement. (D) Light scatter in 3D OCT scans at 2 and 4 weeks was compared with the values in preoperative eyes. Data are means ± SD. The number of eyes is indicated in the graph. P values were determined with unpaired t tests at each time point compared to respective Pre-op values (table S4). (E) Transmission electron micrographs 4 weeks after debridement show the ablated region of the anterior stroma. epi, epithelial cells; bm, basement membrane; k, keratocyte processes; L, amorphous matrix deposit (lake). Insets show magnification of a box (1 μm × 1 μm) from the indicated region containing orthogonal views of collagen fibrils. Scale bar, 2 μm. Images are representative of n = 3 animals.

Fig. 7. Vascularization of debridement wounds

One month after wounding, whole-mount corneas were stained with antibody to CD31 (red) to detect ingrowth of blood vessels. Cell nuclei were imaged with DAPI (blue), and added human LBSCs appear green. (A) Vessels in a healed cornea in which no cells were added. (B) DiO-labeled LBSCs are visible (green), but no vessels were present in the central LBSC-treated wound. (C) A stacked bar graph shows the proportion of vascularized corneas in human LBSC-treated (n = 5) and untreated (n = 5) mouse eyes analyzed by staining as in (A). P value was obtained from a two-tailed χ2 test with a 2 × 2 contingency table.