Evidence for a Retinal Progenitor Cell in the Postnatal and Adult Mouse

Abstract

Progress in cell therapy for retinal disorders has been challenging. Recognized retinal progenitors are a heterogeneous population of cells that lack surface markers for the isolation of live cells for clinical implementation. In the present application, our objective was to use the stem cell factor receptor c-Kit (CD117), a surface marker, to isolate and evaluate a distinct progenitor cell population from retinas of postnatal and adult mice. Here we report that, by combining traditional methods with fate mapping, we have identified a c-Kit-positive (c-Kit⁺) retinal progenitor cell (RPC) that is self-renewing and clonogenic in vitro, and capable of generating many cell types in vitro and in vivo. Based on cell lineage tracing, significant subpopulations of photoreceptors in the outer nuclear layer and bipolar, horizontal, amacrine and Müller cells in the inner nuclear layer are the progeny of c-Kit⁺ cells in vivo. The RPC progeny contributes to retinal neurons and glial cells, which are responsible for the conversion of light into visual signals. The ability to isolate and expand in vitro live c-Kit⁺ RPCs makes them a future therapeutic option for retinal diseases.

1. INTRODUCTION

Visional signals are mediated by photoreceptors, bipolar cells and ganglion cells, and the loss of these neurons leads to irreversible blindness (de Jong, 2006; Friedman et al., 2004; Hartong et al., 2006; Quigley and Broman, 2006; Ramsden et al., 2013; Thylefors et al., 1995). Photoreceptors are the light sensor of the outer nuclear layer (ONL) of the retina, and degeneration and death of these cells occur with retinitis pigmentosa, a disease that affects 1/3000~4000 individuals younger than 60 years of age (Hartong et al., 2006). Age-related macular degeneration (AMD), a major cause of blindness in people over 50 years, is characterized by dysfunction of the retinal pigment epithelium (RPE) followed by apoptosis of photoreceptors (de Jong, 2006; Friedman et al., 2004). The high ocular pressure of glaucoma (Quigley and Broman, 2006), or the ischemic retinopathy of diabetes (Antonetti et al., 2012), also results in the loss of functional retinal cells leading to impaired vision. Bipolar cells of the retinal inner nuclear layer (INL) transmit the visional signal from the photoreceptors to the ganglion cells and then into the brain. In addition to neurons, RPE and Müller glial cells support the nutrition and homeostasis of the retina (Goldman, 2014; Ramsden et al., 2013; Reichenbach and Bringmann, 2013).

Embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) have been used to generate RPE for the treatment of AMD, preventing its progression by protecting the underlying photoreceptors(Mead et al., 2015). However, the photoreceptors that have been lost in AMD still require replacement. The use of retinal progenitor cells (RPCs) for neuronal regeneration is an active area of investigation; although progress has been challenging due to the lack of specific surface markers for isolation of RPCs, and the fact that previously identified RPCs are a heterogeneous population of cells(Ahmad et al., 2000; Cepko, 2014; Gualdoni et al., 2010; MacLaren et al., 2006; Tropepe et al., 2000). Thus, the differentiation potential of individual progenitor cells has been difficult to evaluate. ESCs and iPSCs are capable of generating precursors of photoreceptors in vitro, but transplantation of these precursors in vivo has had limited success (Binder, 2011; Mead et al., 2015; Schwartz et al., 2012). Moreover, ESCs and iPSCs need to be differentiated prior to implantation in vivo, to avoid the risk of tumorigenesis(Cui et al., 2013; Shirai et al., 2016; West et al., 2012). Following lineage specification, ESCs and iPSCs lose their integration capacity and their multipotent phenotype, which limits their therapeutic potential(West et al., 2012). To our knowledge, a population of cells that expresses a stem/progenitor cell antigen and maintains the self-renewing and multipotent characteristics of a stem/progenitor cell in vivo is critical. The lack of a specific surface marker that allows for isolation and expansion of live progenitor cells from the eye, has been an issue preventing identification of a primitive cell capable of regulating physiologic cell renewal and organ reconstitution following injury.

The stem cell factor receptor c-Kit, also known as tyrosine-protein kinase Kit or CD117, is a protein involved in the development, maturation, and survival of neurons (Hirata et al., 1993; Jin et al., 2002). Both c-Kit and Kit ligand, stem cell factor, are present on cell surface membranes of neuronal cells in the central nervous system, including retinas of mice and humans (Das et al., 2004; Hasegawa et al., 2008; Koso et al., 2007; Mochizuki et al., 2014; Morii et al., 1994; Zhou et al., 2015), and the peripheral nervous system (Goldstein et al., 2015; Guijarro et al., 2013; Sachewsky and Morshead, 2014). c-Kit-positive (c-Kit⁺) cells have also recently been identified from the retinal neuroblast layer of human eyes (embryonic weeks 12~14), and are being proposed as RPCs with the potential for application in retinal degeneration without tumorigenesis (Chen et al., 2016; Zhou et al., 2015). However, it is not known whether c-Kit⁺ cells with progenitor cell properties exist in the postnatal or adult retina, and whether progeny of these cells contribute to the architecture of the retina. The expression of c-Kit has been employed previously for the identification and characterization of hematopoietic, cardiac, and lung stem/progenitor cells (Bolli et al., 2011; Itkin et al., 2012; Kajstura et al., 2011), suggesting that the presence of c-Kit may uncover a pool of resident RPCs critical for the maintenance of neuronal cells responsible for vision.

Here, we report for the first time that the mouse eye possesses a primitive c-Kit⁺ cell that is self-renewing, clonogenic and multipotent, the three critical identifiers of tissue specific stem/progenitor cells (Weissman, 2000). In addition, lineage tracing techniques in vivo demonstrate that the major cell types in the ONL and INL of the adult retina are progeny of c-Kit⁺ cells. The identification of this class of resident progenitor cells in the postnatal and adult eye will help to advance our understanding of neuronal regeneration and tissue repair in disorders of the retina.

2. MATERIALS and METHODS

2.1. c-Kit Lineage Tracing

A lineage tracing model in mice, cKit^CreERT2+, was utilized for genetic fate mapping studies as previously described (Goss et al., 2016; Hatzistergos et al., 2015). In brief, cKit^CreERT2+ mice contain a Cre-ERT2 construct inserted in the first exon of c-Kit. Upon transcription from the c-Kit locus, Cre-ERT2 is expressed and remains in the cytoplasm. In the presence of tamoxifen (TAM), the ERT2 receptor is activated and Cre is translocated to the nucleus, where it promotes recombination. The cKit^CreERT2+ mice were bred to IRG reporter mice (De Gasperi et al., 2008). The mice were maintained in an Association for Assessment and Accreditation of Laboratory Animal Care-approved animal facility at the University of Miami, Miller School of Medicine, and procedures were performed using Institutional Animal Care and Use Committee-approved protocols according to NIH standards.

2.2. Tamoxifen Administration

For lineage tracing studies, Cre-ERT2 was activated by intraperitoneal injections of 100 μL of TAM (Sigma-Aldrich), dissolved in peanut oil (Sigma-Aldrich) at a concentration of 20 mg/mL, or 400 mg/kg in food at designated time points, as described previously (Goldstein et al., 2015; Hatzistergos et al., 2015). In the 4-day treatment group, the 6-month-old mice received a daily injection of TAM for 4 consecutive days and the eyes were harvested on day 10 from the initial injection. For treatment groups of 1 and 3.5 months, the 6-week-old mice received TAM from food followed by the harvesting of eyes. One eye was used for flow cytometry assays and the other was for immunohistochemical staining. Mice carrying only the cKit^CreERT2+ and cKit^+/+IRG alleles, receiving the same TAM treatment as cKit^CreERT2+IRG mice, and the cKit^CreERT2+IRG mice without TAM treatment were used as controls.

2.3. Isolation and In Vitro Culture of c-Kit⁺ Cells

Cell culture was performed as previously described (Klassen et al., 2004; Li et al., 2013), with modifications. Mice were sacrificed on postnatal day 1 (P1), The retinas were dissociated in PBS containing collagenase I (10 mg/ml) and collagenase II (25 mg/ml, Worthington Biochemical). The dissociated cells were cultured in dishes pre-coated with laminin (20µg/ml, Sigma-Aldrich), in growth medium, containing DMEM/F12 medium (Lonza) supplemented with murine basic fibroblast growth factor (bFGF, 20 ng/ml, PeproTech), murine epidermal growth factor (EGF, 20 ng/ml, PeproTech), B27 (1:50, Gibco), N2 (1:100, Gibco), insulin/transferrin/sodium selenite (1:500, Lonza) and leukemia inhibitor factor (LIF, 10 ng/ml, Chemicon). After expansion of the cells, c-Kit⁺ cells were sorted using anti-mouse c-kit (CD117) MicroBeads (Miltenyi Biotec, Supp Table 1).

2.4. Differentiation of c-Kit⁺ RPCs

Cell differentiation assays were performed as previously described (Kelley et al., 1994; Li et al., 2013), with modifications. c-Kit⁺ cells were cultured in differentiation conditions using DMEM/F12 medium supplemented with bFGF (10ng/ml) and B27 (1:50) for the next 2 days. To promote neural cell differentiation, including Müller cells and bipolar cells, the cells were next switched to the differentiation medium plus N2 (1:100) for another 6 days. For amacrine cell differentiation, cells were cultured in the differentiation medium plus JAG1 (40 nM, AnaSpec, Inc.) for 6 days. For horizontal cell differentiation, cells were cultured in the differentiation medium plus nerve growth factor (NGF, 10 ng/ml, Sigma-Aldrich) and insulin-like growth factor 1 (IGF-1, 10 ng/ml, Sigma-Aldrich) for 6 days. For photoreceptor differentiation, cells were cultured in the differentiation medium plus N2 (1:100), Docosahexaenoic acid (DHA, 50 nM, Sigma-Aldrich), Retinoic acid (RA, 2 μM, Sigma-Aldrich), γ-secretase inhibitor (DAPT, 10 μM, Sigma-Aldrich) for 2 days, and then changed to medium containing DMEM/F12 with B27 (1:50), NGF (10 ng/ml), IGF-1 (10 ng/ml) and brain-derived neurotrophic factor (BDNF, 10 ng/ml, Sigma-Aldrich) for another 4–6 days.

2.5. Limiting Dilution and Clone Formation

Limiting dilution for clone formation was performed as we previously described (Liu et al., 2015). Briefly, 100 c-Kit⁺ cells were plated in 100 mm diameter dish (the density, ≈ 1 cell / 60 mm²) to obtain multicellular clones derived from a single founder cell. The clones formed over 2 – 3 weeks in growth medium (see Isolation and in vitro culture of c-Kit⁺ cells), were then imaged by phase contrast microscopy and fluorescent microscopy for c-Kit immunostaining (see immunocytochemistry staining).

2.6. Flow Cytometry

For the surface marker, cells dissociated from retinal tissue or cells detached from culture dishes were blocked with CD32/16 for 15 minutes at room temperature (RT). Then, the cell suspensions were incubated with either primary antibody conjugated with fluorescence at 4 °C for 30 minutes. For intracellular and nuclear markers, cells were exposed to fixation and permeability buffer (BD Bioscience), and blocked with donkey serum. The cells were then incubated with primary antibodies at 4°C for 30 minutes followed by a second antibody conjugated with fluorescence at 4°C for 30 minutes. The cells were then assessed using BD FACS Canto II and data was analyzed using FlowJo software (TreeStar). Further details of antibodies are provided in the Supp Table 1.

2.7. Immunofluorescent Staining

Immunohistochemistry: The eyeballs were pre-fixed in fixation buffer (5% acetic acid, 0.4% paraformaldehyde, 0.315% saline, and 37.5% ethanol) at RT for 30 minutes. The cornea and lens were removed and the eye shells remained in pre-fixation buffer for another 2 hours at RT followed by fixation in 4% paraformaldehyde at 4 °C overnight. The eye shells were then processed and embedded in paraffin. The sections (5 μm or 30 μm) were deparaffinized, rehydrated and then boiled in 10 mM citrate buffer. They were blocked with 10% donkey serum and then incubated with primary antibodies at 4 °C overnight. Finally, species-matched secondary antibodies conjugated with fluorescence were applied for 1 hour at 37 °C. Nuclei were stained with 4’, 6-diamidino-2-phenylindole (DAPI).

Immunocytochemistry: Cells were fixed with 4% paraformaldehyde at RT for 10 minutes, and then incubated with 10% donkey serum for 30 minutes at RT. Next, the cells were incubated with primary antibodies at 4 °C overnight, followed by a second antibody conjugated with fluorescence for 1 hour at 37 °C. Nuclei were stained with DAPI. Both cells and tissue sections were analyzed using a confocal microscopy system (Olympus). Further details of antibodies are listed in Supp. Table 1.

2.8. Western Blotting

Total protein was extracted from retinas of wild-type mice at different time points. An equal amount of protein for each condition was separated by 7.5% Mini-PROTEAN^® TGX™ precast gels (BIO-RAD) and transferred onto polyvinylidene fluoride membranes, following incubation with the primary antibody at 4 °C overnight, and then a secondary antibody conjugated with peroxidase for 1 hour. Chemiluminescent results were obtained by exposure to x-ray films. Further details of antibodies are listed in Supp. Table 1.

2.9. Statistical Analysis

For comparisons between groups in Fig. 1A, a one-way analysis of variance, followed by Bonferroni’s multiple comparison test was performed. Statistical significance for comparisons was accepted at P<0.05. The numbers of samples per group are specified in the figure legend.

Figure 1.

c-Kit⁺ cells are present in mouse retinas after birth. (A) Flow cytometry assessment of c-Kit⁺ (APC) cells as a % of total retina cells at postnatal day (P)1, P7, P14 and postnatal week (W)4, W13, W57. Representative flow cytometry scatter plots (left) and quantitation (right). Data are represented as mean ± SEM (n=3 for each time point). *P<0.0001 versus P1 and P7. (B) Representative image of an entire 4 week retina (5 μm section) with immunostaining for c-Kit (green, white arrowheads) and 4’,6-diamidino-2-phenylindole (DAPI, blue). INL: inner nuclear layer (yellow arrow); CB: ciliary body (red arrow). (C) Representative confocal imaging of a retina (30 μm section) immunostaining for c-Kit (green, white arrowheads) and DAPI. The area of white box is shown in higher magnification (bottom panel). Red arrowheads point to the polarized projections of c-Kit⁺ cells. IPL: inner plexiform layer; GCL: ganglion cell layer. (D) Representative confocal imaging of immunofluorescence co-staining for c-Kit (green, white arrowheads) and markers (red) of neurons for PKCα (protein kinase C alpha), Calb (calcium-binding protein calbindin), GAD (glutamic acid decarboxylase 65&67), ChAT (choline acetyltransferase; from first to fourth rows, respectively) or Müller cells for GS (glutamine synthetase, fifth row), as well as DAPI. The areas of white boxes are shown in higher magnification (second to fourth columns).

3. RESULTS

3.1. c-Kit⁺ Cells are Present in Postnatal and Adult Retinas.

Previous studies have shown that the expression of c-Kit in the retina is high during late fetal development and in the neonatal period, but it decreases markedly by 13 days after birth (Koso et al., 2007; Morii et al., 1994). We confirmed this finding by Western blotting of c-Kit expression in retinas collected at postnatal day (P) 1, P7, and P14 (Supp. Fig. 1A). Next, we harvested intact retinas from mice at P1, P7, P14 and week (W) 4, W13 and W57 after birth. The tissue was dissociated into single cell suspensions, and the presence of c-Kit⁺ cells was established by flow cytometry (Fig. 1A). Consistent with the results by Western blotting, the percentage of c-Kit⁺ cells in the entire retina was notably high at P1 and P7, and markedly reduced at P14. Importantly, c-Kit expression persisted at low levels in retinas of adult mice, up to 57 weeks of age (Fig. 1A and Supp. Fig. 1B).

Sections of the retinas from W4 (Fig.1B) and W15 (Supp. Fig. 1C) mice were labeled with c-Kit antibody, and c-Kit⁺ cells (green, white arrowheads) were identified. The c-Kit⁺ cells were distributed throughout the entire retina, from the central to the peripheral regions, but not in the ciliary bodies (red arrow, Fig. 1B). They were located in the basal side of the INL, where RPCs are proposed to reside (in zebrafish and goldfish) and clusters of proliferating cells are observed (Raymond and Hitchcock, 2000; Wu et al., 2001), with a non-overlapping distribution, typical of neuronal self-avoidance (Zipursky and Grueber, 2013). This pattern of c-Kit⁺ cells was also noted in the retinas of mice at W15 (Supp. Fig. 1C).

To better characterize the morphology of the c-Kit⁺ cells in the INL, higher power images were obtained from thicker sections of retinas (30 μm) from W4 mice, and stained for c-Kit (Fig. 1C). The c-Kit⁺ cells possessed polarized projections (Fig. 1C, lower panel; red arrowheads), which originated from their cell body within the INL and subsequently crossed the inner plexiform layer extending to the nearby ganglion cell layer (GCL). Moreover, these projections appear to have limited branching.

Immunolabeling for c-Kit, and for proteins restricted to neurons and Müller cells, was implemented to determine the phenotype of cells expressing c-Kit. Cells positive for c-Kit did not co-express with protein kinase C alpha (PKCα) (Fig. 1D, first row), which is present in rod bipolar cells (Klooster and Kamermans, 2016). Similarly, c-Kit⁺ cells were negative for the calcium-binding protein calbindin (Calb) (Fig. 1D, second row) commonly found in horizontal cells and amacrine cells, glutamic acid decarboxylase 65&67 (GAD; Fig. 1D, third row) and choline acetyltransferase (ChAT; Fig. 1D, fourth row) that typify major populations of GABAergic amacrine cells (Balasubramanian and Gan, 2014; Klooster and Kamermans, 2016). Glutamine synthetase (GS; Fig. 1D, fifth row) that identifies Müller cells (Klooster and Kamermans, 2016) was also not detected in c-Kit⁺ cells. Finally, c-Kit⁺ cells did not co-localize with staining for ionized calcium binding adaptor molecule 1 (Iba1), a marker of microglial calls (Supp Fig. 1D). Collectively, these findings indicate that c-Kit⁺ cells are a population of cells that do not express established markers of either neurons or Müller cells in the INL of the retina. Although, we can not exclude all subpopulations of amacrine or bipolar cells by the markers used for immunostaining.

3.2. Fate Mapping Demonstrates Quiescent c-Kit⁺GFP⁺ Cells in Adult Retinas of Mice Exposed to Tamoxifen for Short Periods of Time.

To characterize c-Kit⁺ cells in adult retinas in vivo, we used the Kit^CreERT2+ transgenic mouse model crossed-bred with the two-color IRG reporter line to identify Cre-mediated recombination (De Gasperi et al., 2008; Goldstein et al., 2015; Hatzistergos et al., 2016; Hatzistergos et al., 2015). There was no evidence for leakiness of green fluorescence protein (GFP) expression in the entire retinas of cKit^CreERT2+,cKit^+/+IRG, and cKit^CreERT2+IRG mice in the absence of tamoxifen (TAM; Supp. Fig. 2A). Retinal tissue was collected from cKit^CreERT2+IRG mice exposed to TAM for 4 days and then harvested 10 days post initial administration of TAM. Flow cytometry detected native (n) GFP of 0.68% and 1.55% in the total population of retinal cells from two different lineage tracing mice (Fig. 2A). There was no evidence of nGFP in cKit^CreERT2+ or cKit^+/+IRG mice receiving the same dose of TAM. Additional immunofluorescence staining for c-Kit⁺ and GFP using anti-GFP antibody (abGFP) identified 11 c-Kit⁺ cells and 8 of them were GFP⁺ cells in the INL (from 4 slides, two mice), ≈ 73% c-Kit⁺ cells were labeled with GFP in the retinas. abGFP⁺ cells were 9 in total (from 4 slides, two mice), and 8 of 9 abGFP⁺ cells were c-Kit⁺ (Fig. 2B and 2C). No GFP⁺ cells were PCNA⁺ (data not shown), a marker for cell proliferation, consistent with a quiescent state of the adult central nervous system under basal conditions (Bonaguidi et al., 2011; Seri et al., 2001).

Figure 2.

Genetic fate mapping demonstrates GFP labeled c-Kit⁺ cells, present in adult retinas after short-term exposed to tamoxifen. Mice received tamoxifen for 4 days and then harvested 10 days post administration of tamoxifen. (A) Flow cytometry scatter plots show native GFP⁺ (nGFP) cells (FITC⁺PE⁺) as a % of total retinal cells from cKit^CreERT2+(top left panel), cKit^+/+IRG (top right panel), and cKit^CreERT2+IRG (bottom panels) mice. (B) Representative confocal imaging of GFP staining (abGFP, white) and DAPI in the retinal inner nuclear layer (INL). (C) Representative confocal imaging of co-staining for c-Kit (green, left) and abGFP (white, middle), and merged with DAPI (right) in higher magnification.

3.3. Fate Mapping Demonstrates that c-Kit-derived GFP Accumulated in Adult Retinas of Mice Exposed to Tamoxifen for Longer Periods of Time.

Because the retina is a low turnover organ (Moshiri et al., 2004; Ooto et al., 2004), and neural stem/progenitor cells in central nerve system are believed to be largely quiescent, we hypothesized that the evaluation of RPC progeny by lineage tracing may require an extended period of TAM exposure. Thus, cKit^CreERT2+, cKit^+/+IRG, and cKit^CreERT2+IRG mice received TAM up to 3.5 months post initial administration of TAM at 6 weeks of age. To screen for false expression of the report protein, retinas were immunostained for GFP using three different GFP antibodies, and there was no evidence of abGFP in the retinas of cKit^CreERT2+ and cKit^+/+IRG mice receiving TAM, and no abGFP in cKit^CreERT2+IRG mouse receiving no TAM (Supp. Fig. 2B). Additionally, the retinas were stained for c-Kit, and there was no evidence of c-Kit immunofluorescence other than the basal side of the INL (Supp. Fig. 2C, left panel). Additional real time PCR assessment of retinas from cKit^CreERT2+ mice receiving no TAM and cKit^+/+IRG mice receiving TAM for 1 month demonstrated the level of c-Kit message in retinas was comparable in both mice (Supp. Fig. 2C, right panel). Thus, prolonged TAM exposure did not lead to expression of c-Kit mRNA or protein in mature cells of the retina. Flow cytometry of single cell suspensions from fresh retinal tissue demonstrate that quantitatively GFP⁺ cells composed 1.12±0.62%, 4.1±1.8%, and 14.6±4.2% of the entire retina from mice exposed to TAM for 4 days and then harvested at day 10 after initial administration of TAM, or for 1 month or for 3.5 months, respectively (Fig. 3). Due to the increased number of nGFP cells at longer periods of TAM exposure, we assessed the retinas for cell death and proliferation. Supp. Fig. 3 demonstrates evidence of dying cells, at longer periods of TAM exposure in the ONL. Moreover, there was increased PCNA staining in the retinal INL of mice receiving long-term TAM, consistent with cell proliferation, compared with mice receiving no TAM or short-term TAM exposure. These data suggest that longer-term TAM exposure is associated with retinal injury (Cho et al., 2012; Toler, 2004) and a subsequent repair response leading to proliferation.

Figure 3.

Genetic fate mapping demonstrates increased GFP labeled cells in adult retinas exposed to tamoxifen over time. cKit^CreERT2+, cKit^+/+IRG and cKit^CreERT2+IRG mice received tamoxifen (TAM) for 4 days and then harvested 10 days post administration (dpa) of TAM, or for 1 month or for 3.5 months post administration (mpa) of TAM. (A) Representative flow cytometry scatter plots show native GFP⁺ (nGFP) cells (FITC⁺PE⁺ and FITC⁺) as % of total retinal cells at 1 (top panels) and 3.5 (bottom panels) mpa. (B) Quantitation of flow cytometry show % of nGFP⁺ cells in retinas of cKit^CreERT2+IRG mice at 10 dpa, 1 and 3.5 mpa, respectively. Data are represented as mean ± SEM (2 independent experiments for each time point).

3.4. Progeny of c-Kit⁺ RPCs Comprise Cells of Retinal INL and ONL in Fate Mapping Adult Mice, including Neurons and Müller Cells.

To investigate the distribution of c-Kit⁺ cell’s progeny, mice received TAM for 3.5 month. nGFP was apparent in freshly isolated retinas from TAM-treated cKit^CreERT2+IRG mice (Fig. 4A, bottom row), while no nGFP was observed in the retinas of TAM-treated cKit^CreERT2+ and cKit^+/+IRG mice (Fig. 4A; top two rows). Low power imaging revealed that nGFP was dispersed throughout the retinas of cKit^CreERT2+IRG mice (Fig. 4B; first row, left and right merged panels), which was confirmed by abGFP staining (Fig. 4B, first row, middle and right merged panels). nGFP was further visualized in the ONL and INL of retinas at an intermediate power of imaging (Fig. 4B; second row). Finally, high power imaging demonstrated nGFP and abGFP were co-localized in the ONL and INL (Fig. 4B third and forth rows). However, nGFP was not detected in the GCL (Supp. Fig. 4; left panel).

Figure 4.

Genetic fate mapping demonstrates that c-Kit⁺ cells generate progeny in the retina outer and inner nuclear layers after long-term exposed to tamoxifen. cKit^CreERT2+, cKit^+/+IRG and cKit^CreERT2+IRG mice received tamoxifen for 3.5 months. (A) Representative images of native GFP (nGFP, green) and DsRed (red) of fresh retinas. (B) Representative images of co-localized nGFP (green, first row, left), GFP immunostaining (abGFP, white, first row, middle) and merged image with DAPI in the retina. The confocal imaging of nGFP (second row left) and merged image with DAPI (second row right) shows in an intermediate magnification. The higher magnification represents confocal imaging for co-localization of nGFP and abGFP in both the outer and inner nuclear layers (ONL, INL; third and forth rows, respectively) of retinas.

To investigate the retinal cell types that evolved from cells of c-Kit origin, paraffin-embedded tissue sections, collected from cKit^CreERT2+IRG mice exposed to TAM for 3.5 months starting at 6 weeks of age, were stained for GFP and specific lineage markers. In the ONL, retinal cells expressing GFP were positive for rhodopsin (Rho) (Fig. 5A, first row), which is found in the specialized photoreceptors termed rods (Klooster and Kamermans, 2016). Moreover, bipolar cells expressing PKCα, horizontal cells and amacrine cells expressing Calb, amacrine cells positive for GAD, and Müller cells positive for GS, were found to co-express with GFP in the INL (Fig. 5A, second through fifth rows, respectively). In contrast, cells positive for nGFP were not present in the GCL, and did not co-localize with cells expressing the ganglion cell marker tubulin beta 3 (Tuj-1, Supp. Fig. 4; left and middle panels, respectively).

Figure 5.

c-Kit-derived cells give rise to neurons and Müller cells in adult retinas. cKit^CreERT2+IRG mice received tamoxifen for 3.5 months. (A) Representative confocal imaging of retinas GFP immunostained (abGFP, white), markers (red) for neurons or Müller cells, and DAPI. abGFP co-stained with Rho (rhodopsin) in outer nuclear layer (ONL; first row), PKCα, Calb, GAD (second to fourth rows, respectively), or GS (fifth row) in inner nuclear layer (INL). The areas in the white boxes are shown in higher magnification (second to fourth columns). (B) Scatter plots of flow cytometry showing nGFP⁺ cells (FITC⁺ and FITC⁺PE⁺; first column) and % of nGFP⁺Rho⁺ (second column), nGFP⁺ PKCα⁺ (third column) and nGFP⁺GS⁺ cells (fourth column). Quantitation of flow cytometry is shown in the bottom pie graph (average of 2 independent experiments).

To quantitate the progeny derived from c-Kit⁺ cells, after mice were exposed to TAM for 3.5 months, the retinas were enzymatically dissociated and the single cell suspensions were analyzed by flow cytometry. Control mice similarly received TAM, and the retinas were processed in an identical manner. A percentage of cells double positive for nGFP and for markers of neurons or Müller cells were assessed in the nGFP⁺ population of cells (Fig. 5B). In the total population of nGFP⁺ cells, Rho⁺ cells of 7.19%, PKCα⁺ cells of 28.9% and GS⁺ cells of 28.6% were detected, which represent photoreceptors, bipolar cells and Müller cells, respectively. Retinas from mice receiving TAM for 1 month, were freshly harvested for flow cytometry assessment. In the total nGFP⁺ population of cells, there were GAD⁺ cells of 25.1% and Calb⁺ cells of 3.7%, which are markers for amacrine cells and horizontal cells (Supp. Fig. 5).

Collectively, these data demonstrate that c-Kit⁺ cells generate photoreceptors, bipolar cells, horizontal cells, amacrine cells and Müller cells – the major cell types of the adult mouse retina. Thus, retinal c-Kit⁺ cells are multipotent cells able to contribute to the generation of different types of neurons and supporting glial cells in the retina.

3.5. Retinal c-Kit⁺ Cells Possess Progenitor Cell Properties In Vitro

To determine whether c-Kit⁺ cells possessed a profile that included the epitopes identified in established RPCs (Irie et al., 2015; Kim et al., 2009; Li et al., 2013; Surzenko et al., 2013), retinas from mouse pups at postnatal day 1 were harvested and after expansion, c-Kit⁺ cells were sorted with immunomagnetic beads. Based on flow cytometric analysis, these c-Kit⁺ cells expressed minimal levels of hematopoietic stem cell antigens, CD34 and CD133, and few cells expressed epitopes for bone marrow-derived immune cells such as CD45, CD11b and F4/80 (Supp. Fig. 6). The endothelial cell marker, Flk1, was also rarely found. A high percentage of c-Kit⁺ cells expressed the stem cell antigen Sca1, and a small subpopulation of cells stained for CD105, CD90.2 and CD140b, but few cells expressed CD73 (Supp. Fig. 6). These data suggest that c-Kit⁺ cells are not of hematopoietic or vascular origin, and do not fulfill the criteria of mesenchymal stem/stromal cells (Goldstein et al., 2015; Liu et al., 2015).

To determine whether retinal c-Kit⁺ cells were able to generate multicellular clones and self-renew, individual cells were plated at limiting dilution (Goldstein et al., 2015; Liu et al., 2015), at a density of one c-Kit⁺ cell every 60 mm². Over a period of 2–3 weeks, multicellular clones were observed (Fig. 6A, left panel). Cells had the expected spindle shape and were positive for c-Kit by immunostaining (Fig. 6A, middle panel). The expression of c-Kit in the large majority of clonal cells (90.2%) was confirmed by flow cytometry (Fig. 6A, right panel). The presence of c-Kit expression in the majority of cells in the clones demonstrates the ability to self-renew.

Figure 6.

Progenitor cell properties of isolated c-Kit⁺ cells in vitro. (A) A phase contrast image of a representative single cell-derived clone after 14 days in culture (left) and a representative image of a clone stained for c-Kit⁺ (green) and DAPI (blue, middle). The flow cytometry histogram of cells stained for c-Kit (APC) showing % of c-Kit⁺ cells in the clone (right, green; grey line-APC isotype control). (B) Representative images of immunofluorescence co-staining for c-Kit (green) and markers (red) for nestin, Chx10 (Visual System Homeobox 2), Pax6 (paired box protein Pax-6), Sox2 (sex determining region Y-box 2), and Rax (retina homeobox protein Rx; first to fifth rows, respectively). The areas in the white boxes are shown at higher magnification (fourth column). (C) Representative scatter plots (top panels) and bar graph quantitation (bottom panel) of flow cytometry showing expression of c-Kit (APC) versus markers of retina progenitors (PE) for Nestin, Chx10, Pax6, Sox2, and Rax. Data are represented as mean ± SEM (n=3~4 for each marker).

The expression of several markers of RPCs were tested in c-Kit⁺ cells by immunostaining and then performing confocal microscopy. The majority of c-Kit⁺ cells expressed the neuroectodermal stem cell marker, nestin, the Visual System Homeobox 2 (VSX2 or Chx10), the paired box protein Pax-6 (Pax6), the SRY sex determining region Y-box 2 (Sox2), and the retina homeobox protein Rx (Rax; Fig. 6B, first to fifth rows, respectively). Additionally, by flow cytometry, we found that the stem cell marker c-Kit was detected in a minimum of 73.0% to a maximum of 84.9% of nestin-, Chx10-, Pax6-, Sox2- and Rax-positive cells (Fig. 6C). Thus, the surface marker c-Kit identifies a compartment of resident RPCs that comprises the majority of previously characterized RPCs. In addition, the presence of Sox2 in c-Kit⁺ RPCs, which is also expressed in primitive, multipotent stem/progenitor cells such as ESCs and iPSCs (Schwartz et al., 2012), supports the concept that c-Kit⁺ RPCs are multipotent cells with a significant degree of plasticity.

To determine the multipotency of clonal c-Kit⁺ cells, these cells were cultured in different types of specialized media to promote differentiation into photoreceptors, other neurons, and Müller glial cells (Li et al., 2013). After a culture period of 8–10 days in each type of specialized media, the presence of recoverin (Rec), PKCα, Calb, GAD and GS were detected; these proteins are typically found in photoreceptors (Fig. 7A, first row), bipolar cells (Fig. 7A, second row), horizontal cells (Fig. 7A, third row) and amacrine cells (Fig. 7A, third and fourth rows), and Müller cells (Fig. 7A, fifth row), respectively. Furthermore, c-Kit⁺ cells placed in medium to promote differentiation into ganglion cells led to expression of Tuj-1 in vitro (Supp. Fig. 4; right panel), even though nGFP was not evident in Tuj-1⁺ cells from cKit^CreERT2+IRG mice exposed to TAM. Additionally, flow cytometry was performed to assess the differentiation potential of c-Kit⁺ cells (Fig. 7B). After 8–10 days in each type of specialized medium, subpopulation of c-Kit⁺ cells expressed Rec, Rho, PKCα, Calb, GAD, ChAT, GS and glial fibrillary acidic protein (GFAP) indicating that the differentiated cells had acquired, respectively, the photoreceptor, bipolar cell, horizontal cell, amacrine cell and Müller cell phenotype. Thus, retinal c-Kit⁺ cells possess the biological properties of progenitor cells; they are self-renewing, clonogenic and multipotent giving rise to photoreceptors, bipolar cells, horizontal cells, amacrine cells, and the supporting Müller cells.

Figure 7.

Multipotency of isolated c-Kit⁺ cells in vitro. Cells from c-Kit⁺ clones cultured in differentiation medium for 8~10 days and stained for markers (red) of neurons or Müller cells, and DAPI. (A) Representative images of immunostaining for Rec (recoverin), PKCα, Calb, GAD or GS (first to fifth rows, respectively). The areas in the white boxes are shown at higher magnification (third and fourth columns). (B) Representative scatter plots (left) and bar graph quantitation (right) of flow cytometry assays showing % of differentiated (diff) cells expressing markers of neurons for Rec, Rho, PKCα, Calb, GAD, ChAT, or Müller cells for GS and GFAP (glial fibrillary acidic protein). Data are represented as mean ± SEM (n=2~4 for each marker).

4. DISCUSSION

Previously, a well-defined population of tissue specific progenitor cells that could be isolated from retinas after birth was difficult to identify, limiting our understanding of neuroregeneration and impeding our advancement of progenitor cell therapy for retinal degeneration disorders, a leading cause of blindness. The results of the present study indicate that the mouse retina possesses a distinct population of resident progenitor cells after birth, and the progeny of these RPCs contribute to vital cell types in the retina. Traditional methods have been combined with a fate mapping strategy to identify c-Kit⁺ RPCs that have properties comparable to those of other tissue specific adult stem/progenitor cells (Bolli et al., 2011; Itkin et al., 2012; Kajstura et al., 2011), i.e. self-renewing, clonogenic and multipotent in vitro, and as demonstrated by lineage tracing studies, their progeny differentiate into crucial cell lineages of the retina. By targeting the surface receptor c-Kit, we distinguished a homogeneous cell population that differentiates in vitro and in vivo into many of the cells composing the retina, including photoreceptors, bipolar cells, horizontal cells, amacrine cells and Müller cells.

It has been decades since postnatal mammalian neurogenesis in the retina was discovered, with the recognition of a heterogeneous pool of progenitor cells capable of differentiating into neurons and Müller cells in vitro and photoreceptors in vivo (Cepko, 2014; Cui et al., 2013; Li et al., 2013; Ramsden et al., 2013; Reichenbach and Bringmann, 2013). Additionally, various lineage tracing methods including retroviral marking in mice and chickens, live imaging in zebrafish, and more recently genetic tracing in mice have been employed to identify RPCs (Cepko, 2014). These protocols have suggested that specific RPCs produce specific daughter cells in terminal divisions (Cepko, 2014). However, the markers used to distinguish these RPCs are intracellular proteins, which preclude the isolation, characterization and in vitro expansion of a uniform pool of live cells. These limitations have been overcome in the current study.

The receptor tyrosine kinase c-Kit defines hematopoietic, cardiac and lung stem/progenitor cells (Bolli et al., 2011; Itkin et al., 2012; Kajstura et al., 2011) and, as shown here, RPCs. These stem/progenitor cell classes share a common surface marker, c-Kit, that was originally thought to be restricted to hematopoietic stem cells and mast-cells (Lennartsson and Ronnstrand, 2012). Importantly, c-Kit expression does not condition the differentiation of these stem cells into a common specialized progeny, which remains specific for each organ. Based on these results, we propose that the expression of c-Kit helps to identify a tissue specific progenitor cell in the retina where cell renewal is currently believed to be mediated in part by dedifferentiation of Müller cells, which reacquire a primitive state, replicate and then reassume a specialized cell phenotype (Cepko, 2014; Ramachandran et al., 2010).

Consistent with previous observations in the bone marrow and in the myocardium (Bolli et al., 2011; Liu et al., 2015), RPCs were most prominent during fetal development and shortly after birth, and decreased in the adult mouse. The expansion of specialized cell populations during postnatal maturation of organs results in a progressive reduction in the numerical density of stem/progenitor cells. In the adult mouse, c-Kit⁺ RPCs were located in the basal side of the INL, with a non-overlapping distribution characteristic of neuronal self-avoidance (Zipursky and Grueber, 2013), which is critical for proper neuronal development and wiring. Interestingly, the cell bodies of c-Kit⁺ RPCs exhibited polarized projections extending to the GCL, which is a source of stem cell factor (data not shown, and (Morii et al., 1994)), suggesting that c-Kit⁺ cells may communicate with ganglion cells, one of the earliest born cell types, for the determination of cell fate in the mouse retina (Cepko, 2014).

c-Kit⁺ RPCs are distributed from the central to the peripheral regions of the retina, but not in the ciliary bodies, which were previously proposed to house retinal progenitors (Cicero et al., 2009; Martinez-Navarrete et al., 2008). These data suggest that c-Kit⁺ RPCs are a distinct population of cells from previously described progenitors. One other location suggested to contain RPCs (in zebrafish and goldfish) is the basal side of the retinal INL (Raymond and Hitchcock, 2000; Wu et al., 2001), which is mainly composed of Müller cells and amacrine cells. c-Kit⁺ cells were previously identified in the INL of the mouse retina (Morii et al., 1994), but the majority of dividing c-Kit⁺ cells did not co-express markers of amacrine cells (Koso et al., 2007). Our current data demonstrate that c-Kit⁺ cells do not co-stain with Calb, ChAT or GAD, which label most GABAergic amacrine cells. However, to date there are 33 different subtypes of amacrine cells that have been identified in the rodent retina (Balasubramanian and Gan, 2014; Cherry et al., 2009; MacNeil and Masland, 1998). They are divided into two major subpopulations, GABAergic and glycinergic types, and a small population of neither GABAergic nor glycinergic amacrine cells. In the present study, we cannot exclude the possibility that c-Kit⁺ cells may generate a subpopulation of amacrine cells. In the future, it will be interesting to further investigate whether amacrine cells are downstream of c-Kit⁺ cell differentiation, as Koso and colleagues have hypothesized that a population of non-dividing c-Kit⁺ cells may represent a subset of amacrine cells during development (Koso et al., 2007). It is also noteworthy that Chen and colleagues recently reported that amacrine cells expressing leucine-rich repeat-containing G-protein-coupled receptor 5 (Lgr5) possess regeneration potential in the adult mouse retina (Chen et al., 2015).

Importantly, the c-Kit-specific fate mapping studies indicate that significant subpopulations of photoreceptors in the ONL, and bipolar cells, horizontal cells, amacrine cells and Müller cells in the INL, are the progeny of c-Kit⁺ cells in vivo. The magnitude of c-Kit-derived cells in the retinas of mice receiving TAM over a 3.5 month period, and the evidence of cell death in the ONL and proliferation in the INL, may suggest there is a level of injury amplifying the number of c-Kit-derived cells. However, we have no evidence that TAM is inducing expression of c-Kit at the mRNA or protein level in mature cells.

Beyond the possibility of isolating retinal c-Kit⁺ cells for cell therapy, it will be critical to further explore disease models producing damage in the retinas of these lineage tracing mice to determine the potential role of endogenous c-Kit⁺ RPCs in retinal repair after injury in vivo. Furthermore, another important issue raised by these studies is the origin of c-Kit-derived progenitor cells in the mouse retina. Elegant studies have been performed by Hatzistergos and colleagues on the origin of c-Kit-expressing progenitor cells in the heart, with evidence that the cells originate, in part, from cardiac neural crest cells(Hatzistergos et al., 2015). From our analysis of the c-Kit⁺ retinal cells by flow cytometry, they are not of hematopoietic or vascular origin, and do not fulfill the criteria of mesenchymal stem/stromal cells (Supp. Fig. 6). However, additional studies will need to be performed to understand the specific origin of c-Kit⁺ cells in the mouse retina.

5. CONCLUSIONS

Taken together, our findings verify the presence of c-Kit⁺ RPCs in postnatal and adult mice, and these cells are self-renewing, clonogenic, and multipotent. Our data also advocate that c-Kit⁺ cells support the homeostasis of cells in the retina. The progeny of c-Kit⁺ cells contribute to retinal neurons and glial cells responsible for transmission of visual signals from the ONL, through the INL and GCL into the brain. Moreover, we propose that the ability to specifically isolate a homogeneous population of cells using c-Kit as a marker, and expanding these cells in vitro, will provide future cell therapeutic strategies for retinal diseases.