Eyes open on stem cells

Abstract

Recently, the murine cornea has reemerged as a robust stem cell (SC) model, allowing individual SC tracing in living animals. The cornea has pioneered seminal discoveries in SC biology and regenerative medicine, from the first corneal transplantation in 1905 to the identification of limbal SCs and their transplantation to successfully restore vision in the early 1990s. Recent experiments have exposed unexpected properties attributed to SCs and progenitors and revealed flexibility in the differentiation program and a key role for the SC niche. Here, we discuss the limbal SC model and its broader relevance to other tissues, disease, and therapy.

Figure 1

Seminal studies that support the traditional LSC model (A and B) Schematic representation of the traditional LSC model (A); cell types and cellular hierarchy are shown in (B). This model considers SCs (green) as rare, immortal, and slow-cycling cells that give rise to abundant, short-lived, and fast-cycling progenitors (blue) that possess a limited replicative lifespan and can proliferate 3–4 times before they terminally differentiate (gray). (C–E) Key evidence for the location of SCs to the limbus. (C) Detection of infrequently dividing cells in the limbus as nucleotide label-retaining cells (LRCs, yellow) by pulse-chase experiments in vivo. (D) Limbal epithelial cells possess superior growth in vitro, and they can form large clones (holoclones) that can be serially passaged and are believed to be derived from a single SC (green), whereas small-size clones are considered to be those derived by a short-lived progenitor (blue). (E) Total LER in rabbits had no detectable effect on most corneas; however, two consecutive corneal debridements led to delayed healing and corneal neovascularization and opacification (Huang and Tseng, 1991). (F) Open questions and concerns. CE, corneal epithelium; D, differentiated cell; P, progenitor cell.

Figure 2

Cell lineage tracing used to identify SC location (A) Schematic illustration of the LSC and CSC models. SC distribution in each model is shown (green). (B) The outcome as predicted by the LSC or CSC models. The LSC model predicts that all of the corneal clones would be extinct by tissue turnover, whereas LSCs would form long-term radial clones that would replace all of the corneal cells by the turnover time point. The CSC model predicts that SCs distributed throughout the cornea would generate long-term clones scattered through the cornea surface. (C–E) Schematic representation of Confetti double transgenic system (C), possible rearrangement of the Brainbow cassette following tamoxifen-induced Cre recombination, and (D) typical image of lineage-traced cornea, (E) 3 months postinduction. (F–H) Illustration of K15-GFP transgene (F), (G) typical image of enucleated eye showing that K15-GFP labels a discrete population of limbal cells, and (H) lineage tracing of triple transgenic (K15-GFP; K14-Cre^ERT2; R26R-Brainbow^2.1) animals showing that radial clones develop from the K15-GFP⁺ limbus regime. Scale bars are (E) 50 μm and (H) 100 μm. CFP, cyan fluorescent protein; YFP, yellow fluorescent protein.

Figure 3

Total LSC depletion in murine is restored by dedifferentiation of corneal epithelial committed cells, if the limbal niche is intact (A and B) Adult triple transgenic K15-GFP; Brainbow^2.1; K14-Cre^ERT2 animals were induced by tamoxifen. Four months later, radial RFP⁺ limbal clones that emerge from the K15-GFP⁺ limbus were imaged (left picture, upper panel). This transgene allows a well-controlled LER (intact niche) and monitoring of the healing by dedifferentiation of corneal committed cells (RFP⁺ clone) that reexpressed the K15-GFP signal within 7–10 days post-LER. The lower panel in (A) is a schematic illustration of the upper panel from Bhattacharya et al. (2023). (B) Schematic illustration of an experiment in which LER was combined with localized damage to the niche by a chemical agent in K15-GFP mice. In this case, the dedifferentiation process did not occur and the cornea became neovascularized and opaque (LSCD). Scale bar, 50 μm (A).

Figure 4

Discrete LSC populations mediate corneal homeostasis and wound healing (A and B) Immunofluorecent staining of whole-mount flattened cornea. (A) Staining of the outer qLSC marker IFITM3 and the inner active LSC marker K15-GFP. (B) Higher magnification of the limbus regime shows the preferential presence of blood vessels in the GPHA2⁺ outer limbus. (C) Four months posttamoxifen injection to Confetti (UBC-Cre^ERT2; R26R-Brainbow^2.1) transgenic, the outer and inner limbal clones were visualized. Although the inner limbus clones developed radially and replenished central cornea, the outer limbus clones displayed a circumferential pattern. Inner limbal clones are long-lived and extremely large in size, strongly suggesting that the cell of origin is an SC, not a transient amplifying cell. Scale bars, 300 μm (A) and 100 μm (B and C).

Figure 5

Two-photon live imaging uncovered a bicompartmentalization of the murine LSC niche Two-photon microscopy technique allows the tracing of the lifetimes of individual LSCs after their laser-induced genetic cell labeling. (A) Following the lineage of individual cells in the basal layer of the inner limbus revealed that this regime contains abundant LSCs that divide frequently and undergo radial movement to renew the corneal epithelium under homeostasis. The basal layer of the outer limbus contains abundant LSCs that divide significantly less frequently and do not contribute to corneal epithelial lineage under homeostasis, but do participate in corneal wound healing. (B) Another approach that allows the detection of division rates is by tracing the dilution of H2B-GFP chimeric protein in transgenic animal. Genetic systems allow the efficient labeling of all limbal and corneal epithelial cells with Histone 2B protein that is fused to the GFP reporter. Then, doxycycline injection shuts off the transgenic expression, and the nuclear GFP labeling is reduced by 50% within every cell division until it becomes undetectable after 6–7 division rounds. This further confirms the detection of label-retaining slow cycling cells in the outer limbus.

Figure 6

Abundant LSC model (A) The abundant LSC model suggests that the inner limbus contains abundant active LSCs that divide every ∼3 days to replenish the cornea and maintain homeostasis. The outer limbus contains abundant qLSCs that participate in wound repair and thus serve as a reservoir. (B) (i) Schematic illustration of the abundant LSC population model. Cell type annotation (ii) and the lineage hierarchy (iii) are shown. (C) Open questions.

Figure 7

Hypotheses for “LSCD” development Example of 3 of the prevailing hypotheses proposed to explain the development of LSCD pathology with different initiating event. (A) LSC boundary failure results in the abnormal presence of conjunctival cells in the limbus and corneal periphery. Conjunctival cells preferentially outcompete LSCs and encourage the development of blood vessels in the underneath stroma. (B) LSC loss causes conjunctival cell penetration into the cornea, which results in the development of blood vessels in the underneath stroma. (C) Loss of antiangiogenic signals of the cornea stimulates the development of blood vessels in the underneath stroma and transdifferentiation. Each hypothesis may be relevant to LSCD induced by environmental or genetic factors. Although scenarios (A) and (B) involve conjunctival cell invasion into the cornea, scenario (C) involves the transdifferentiation of corneal epithelial cells into conjunctival-like cells.