Differentiation of Retinal Ganglion Cells and Photoreceptor Precursors from Mouse Induced Pluripotent Stem Cells Carrying an Atoh7/Math5 Lineage Reporter

The neural retina is a critical component of the visual system, which provides the majority of sensory input in humans. Various retinal degenerative diseases can result in the permanent loss of retinal neurons, especially the light-sensing photoreceptors and the centrally projecting retinal ganglion cells (RGCs). The replenishment of lost RGCs and the repair of optic nerve damage are particularly challenging, as both RGC specification and their subsequent axonal growth and projection involve complex and precise regulation. To explore the developmental potential of pluripotent stem cell-derived neural progenitors, we have established mouse iPS cells that allow cell lineage tracing of progenitors that have expressed Atoh7/Math5, a bHLH transcription factor required for RGC production. These Atoh7 lineage reporter iPS cells encode Cre to replace one copy of the endogenous Atoh7 gene and a Cre-dependent YFP reporter in the ROSA locus. In addition, they express pluripotent markers and are capable of generating teratomas in vivo. Under anterior neural induction and neurogenic conditions in vitro, the Atoh7-Cre/ROSA-YFP iPS cells differentiate into neurons that co-express various RGC markers and YFP, indicating that these neurons are derived from Atoh7-expressing progenitors. Consistent with previous in vivo cell lineage studies, the Atoh7-Cre/ROSA-YFP iPS cells also give rise to a subset of Crx-positive photoreceptor precursors. Furthermore, inhibition of Notch signaling in the iPSC cultures results in a significant increase of YFP-positive RGCs and photoreceptor precursors. Together, these results show that Atoh7-Cre/ROSA-YFP iPS cells can be used to monitor the development and survival of RGCs and photoreceptors from pluripotent stem cells.


Introduction
The neural retina is a component of the central nervous system and plays an essential role in the acquisition and processing of visual information. The mature retina consists of distinct neuronal cell types derived from a common pool of neural progenitor cells during development [1][2][3]. Many blinding diseases, including the prevalent age-related macular degeneration (AMD) and glaucoma, involve the permanent loss of retinal cells, especially the lightsensing photoreceptors or the centrally projecting retinal ganglion cells (RGCs). Despite the evolutionarily conserved neurogenic process and anatomic structure of vertebrate retinas, the spontaneous repair and regenerative capacity of the mammalian retina appears limited compared to teleosts and amphibians [4][5][6][7]. Therefore, de novo production of distinct retinal neurons, especially photoreceptor cells and RGCs, for the purposes of repairing damaged retinas as well as enabling disease mechanism studies remains a high priority.
RGCs are the first neuronal cell type to emerge in the developing vertebrate retina, and remain a minor cell population [8,9]. The production of RGCs from the retinal primordium is stringently controlled by cell-intrinsic transcription factors and influenced by cell-extrinsic signals. The basic-helix-loop-helix (bHLH) transcription factor Atoh7/Math5 plays a critical role in RGC fate specification. In the absence of Atoh7, the majority of RGCs fail to develop in the mouse retina [10,11]. The subsequent differentiation of postmitotic RGCs requires the high-mobilitygroup (HMG) domain transcription factors Sox4 and Sox11 [12], and the POU-domain transcription factor Pou4f/Brn3 [13][14][15]. In the early neurogenic retina, Atoh7 mRNA is expressed by a subset of progenitors [16]. The homeobox gene Pax6, which participates in eye primordium determination and controls the pluripotency of retinal progenitor cells, positively regulates Atoh7 transcription [17][18][19]. In addition, Atoh7 expression and its activity are influenced by the bHLH factors Neurog2 and Hes1 [20,21]. Cell lineage tracing studies have shown that the progeny of Atoh7expressing progenitors also give rise to a subset of early born cone photoreceptor cells [22,23]. In the vertebrate retina, RGC production is negatively regulated by Notch signaling and a number of RGC-derived secreted factors, including GDF11, Sonic Hedgehog (Shh), and VEGF [24][25][26][27][28][29][30]. Consistently, genetic disruption of Notch1, GDF11, or Shh signaling increases Atoh7 expression and enhances RGC and cone photoreceptor genesis [26,28,29,[31][32][33].
In recent years, several groups have established protocols that permit the development of retinal neurons from embryonic stem (ES) cell or induced pluripotent stem (iPS) cell cultures [34][35][36][37]. As a strategy to track stem cell differentiation, fluorescent reporters have been used to monitor ocular tissue and retinal neuron development from the pluripotent cell state. Mouse ES cells with a GFP reporter replacing the retinal specific homeobox gene Rax/ Rx have allowed the enrichment of retinal progenitor cells [36][37][38], and have been used successfully to demonstrate the selforganizing property of the developing optic cup and the differentiation potential of the neural retina [39]. In addition, a virally encoded fluorescent reporter driven by a photoreceptorspecific promoter has been used to enrich photoreceptors derived from human iPS cells [35]. Here, we report the establishment of mouse iPS cells that encode Cre in place of one copy of the Atoh7 gene and a Cre-dependent fluorescent reporter inserted into the ROSA locus. We show that these iPS cells express the fluorescent reporter under conditions promoting ocular tissue induction and neural differentiation, and develop into RGCs and photoreceptor precursors in vitro. These Atoh7 reporter iPS cells can thus be used to monitor a subset of retinal neuronal lineages derived from pluripotent stem cells, and to study the development and survival of RGCs and photoreceptors.

Animals
The Atoh7/Math5-Cre knock-in mouse was described previously [22] and the ROSA-YFP Cre reporter mouse [40] was obtained from the Jackson Laboratory (stock number 003310). We crossed these two mouse lines to generate double heterozygous offspring with the Atoh7 CreKI/+ ; ROSA YFP/+ genotype. The PCR primers used for genotyping were reported previously [26] and shown in Table S1. The animal protocols used were approved by the Animal Research Committee of University of California Los Angeles.

Establishing mouse iPS cells
Mouse embryonic fibroblasts (MEFs) were isolated from embryonic day 13.5 (E13.5) Atoh7 CreKI/+ ; Rosa YFP/+ embryos and expanded in DMEM with 10% fetal bovine serum (FBS). Two lentiviruses, TetO-FUW-OSKM expressing reprogramming factors and FUW-M2rtTA [41] were prepared as described previously [42]. Equal volumes of media containing TetO-FUW-OSKM and FUW-M2rtTA viruses were used to perform four consecutive infections of MEFs over a period of 48 hours in the presence of 6 mg/ml of polybrene (Sigma). The culture medium was changed 12 hours after the last infection. Five days after transduction, MEFs were dissociated using trypsin and re-plated at a density of 1610 5 cells per10 cm-dish on irradiated CF1 mouse feeder cells seeded on a gelatin-coated surface. After 48 hours, the culture medium was replaced by a mouse embryonic stem cell (ESC) medium that contained Glasgow Minimum Essential Medium (GMEM), 10% KnockOut Serum Replacement (KSR), 1% FBS, 0.01 mM non-essential amino acids (NEAA), 1 mM sodium pyruvate, 16 antibiotic-antimycotic (100 units/ml of penicillin, 100 mg/ml of streptomycin, and 0.25 mg/ml of Fungizone), 0.1 mM 2-Mercaptoethanol, 10 3 units/ml leukemia inhibitory factor (LIF), supplemented with 2 mg/ml doxycycline (Sigma). Mouse induced pluripotent stem cell (iPSC) colonies were picked manually based on morphology between 4 and 8 weeks after doxycycline induction and passaged more than ten generations under mouse ES cell culture conditions. Stably passaged Atoh7-Cre/ROSA-YFP iPS cells were further characterized as described in the text and stored in liquid nitrogen.

Differentiation of iPS cells
Atoh7-Cre/ROSA-YFP iPS cell cultures were incubated for 5 minutes with ES dissociation solution containing 0.025% trypsin, 1 mg/ml type IV collagenase, 20% KSR, 1 mM CaCl 2 in PBS. This step was repeated 2-3 times to lift most of the iPS cells from the dish. Floating ESC clusters were collected and plated on culture dishes coated with 0.2% gelatin in ESC medium for 30 minutes at 37uC to remove residual feeder cells. After pre-plating, the floating ESC clusters were further dissociated into single cell suspension with 0.05% trypsin [39].

Alkaline phosphatase histochemistry
Cells were fixed in 4% paraformaldehyde/PBS for 2 minutes at room temperature followed by washes with PBS. The iPS cells were then incubated with alkaline phosphatase detection buffer containing 0.1 mg/ml 5-bromo-4-chloro-3-indolyl phosphate (X-Phos) and 0.25 mg/ml nitroblue tetrazolium (NBT) in the dark at room temperature as previously described [44].

RT-PCR
Total RNA was extracted following the RNAzol B (Tel-Test, Inc) extraction procedure, and Superscript III (Invitrogen) was then used with random priming to synthesize cDNA from 1 mg of total RNA. PCR primers used are shown in Table S1. Each PCR reaction was performed for 35 cycles and the reaction products were analyzed using 2% agarose gels.

Teratoma formation
Dissociated Atoh7-Cre/ROSA-YFP iPS cells (1610 5 cells) were injected subcutaneously into each of six SCID mice. Tumors ranging from 5-10 mm in diameter were harvested after 3 weeks and fixed in 4% paraformaldehyde/PBS. A standard Hematoxlin and Eosin procedure was used to stain paraffin sections of tumor tissues.

Flow Cytometry
Flow cytometry was carried out as described [26] with the following modifications. After trypsin dissociation, single cell suspensions were incubated with 6.2 mM live cell dye (eBioscience) for 20 minute, followed by fixation with 0.25% paraformaldelhyde/PBS for 30 minutes. After washing, the cells were permeablized with 0.1% Triton X100 and incubated sequentially with primary and secondary antibodies to detect YFP and retinal cell type markers before FACS analyses using an LSRII flowcytometer (BD Biosciences). To detect YFP signals, after binding to primary rabbit anti-GFP antibodies, cells were incubated with biotinylated goat anti-rabbit antibodies (1:200, Vector Laboratory) followed by Alexa 488-conjugated streptavidin (1:500, Invitrogen). The cell population positively labeled for the live cell dye was analyzed using FlowJo software (Tree Star). All FACS analyses include ''negative controls'' in which secondary antibodies alone were used to establish the base lines of fluorescent signals. The size of the samples ranged from 50,000 to 100,000 cells each. Data from a minimum of three independent iPS cell cultures were expressed as mean 6 SE. Pairwise comparisons were performed using the Student t-test. Values of p,0.05 were considered statistically significant.

Generation of Atoh7-Cre/ROSA-YFP iPS cells
Since Atoh7 is transiently expressed during retinogenesis and is no longer expressed by differentiated RGCs, we sought to establish iPS cells that will allow us to monitor Atoh7 expression as well as trace the Atoh7 lineage, including mature RGCs. We crossed heterozygous mice in which one allele of Atoh7 had been replaced by Cre (Atoh7 Cre-KI ) [22] with homozygous ROSA-YFP reporter mice [40], which encodes a Cre-dependent YFP-expression cassette, to obtain Atoh7 Cre-KI/+ ; ROSA YFP-KI/+ double heterozygotes. As expected, at embryonic day 13.5 (E13.5) the double heterozygous retinas showed intense YFP signals in the inner retina, where postmitotic RGCs resided (Fig. S1). In addition, a subset of cells in the neuroblast layer expressed a lower level of YFP (Fig. S1). The control retinas with Atoh7 +/+ ; ROSA YFP-KI/+ or Atoh7 Cre-KI/+ ; ROSA +/+ genotypes did not show detectable YFP signals by immunocytochemistry (data not shown). Fibroblasts of Atoh7 Cre-KI/+ ; ROSA YFP-KI/+ double heterozygous embryos were harvested and co-infected with a Tet-inducible lentivirus (tetO-FUW-OSKM) expressing the four Yamanaka reprogramming factors Oct4, Sox2, Klf4, and cMyc, and a lentivirus expressing rtTA [41]. After transduction with the viruses, doxycycline was added to the medium to induce the expression of the reprogramming factors, and cell colonies that acquired mouse embryonic stem (ES) cell morphology were selected and passaged under mouse ES culture conditions (Fig. 1A). These established cell lines, referred to herein as Atoh7-Cre/ROSA-YFP iPS cells, expressed a number of ES cell markers similar to those observed in the control mouse J1 ES cell line, including alkaline phosphatase, Nanog and Oct4 (Fig. 1B-F). In addition, these Atoh7-Cre/ROSA-YFP iPS cells underwent repeated freeze-and-thaw cycles after storage in liquid nitrogen, and were passaged multiple times without changing ES cell morphology and markers when cultured. To further examine the pluripotency of these MEF-derived iPS cells, we implanted the Atoh7-Cre/ROSA-YFP iPS cells into immune-deficient SCID mice. All transplantations (N = 6) into SCID mice resulted in the formation of teratomas (Fig. 1G-K), demonstrating the pluripotent nature of these iPS cells.

Production of retinal progenitors from Atoh7-Cre/ROSA-YFP iPS cells
To evaluate the developmental potential of the Atoh7-Cre/ ROSA-YFP iPS cells, we used the previously described anterior neuro-induction approaches that involve manipulation of Wnt and TGFb/BMP signals [36]. The iPS cells were allowed to spontaneously aggregate into embryoid bodies (EBs) and cultured in the presence of the TGFb inhibitor SB431542 and the Wnt inhibitor ''casein kinase inhibitor-7'' (CKI-7) to promote the anterior ventral neural fate, thus enhancing formation of the eye field. The EBs were then attached to coated surfaces and cultured further in a retinal differentiation medium [43]. At day 7 from the onset of iPS cell-derived EB formation, transcripts of several homeobox genes normally expressed by retinal progenitor cells, including Rax/Rx, Pax6, Vsx2, Otx2, were detected ( Fig. 2A). In addition, numerous Pax6-positive cells were observed in the EB cultures; however, only a subset of the Pax6-positive cells coexpressed Sox2, which is also expressed by retinal progenitor cells (Fig. 2B, C). The Atoh7-Cre/ROSA-YFP iPS cells-derived EBs also expressed the neurogenic bHLH genes Neurog2 and Atoh7 by day 7 (Fig. 2A), indicating the onset of neurogenesis at this time point in culture. Furthermore, we detected both YFP transcript and protein by day 7 (Fig. 2A, B), indicating expression of the YFP reporter under the control of Cre from the Atoh7 locus.
Since Notch signaling is known to influence retinal progenitor cell proliferation and differentiation, we tested whether the csecretase inhibitor DAPT, which inhibits Notch signaling, affected production of YFP-positive cells in the Atoh7-Cre/ROSA-YFP iPS EB cultures. Quantitative analyses using flow cytometry showed that DAPT treatment significantly increased the percentage of YFP-positive cells from 19.063.8% to 36.962.7% by day 14 (Fig. 3), indicating that the production of cells from the Atoh7-expressing lineage was enhanced when Notch signaling was inhibited.

Differentiation of RGCs from Atoh7-Cre/ROSA-YFP iPS cells
We next examined whether Atoh7-Cre/ROSA-YFP iPS cellsderived EBs could produce RGCs. On day 12 of the EB culture, immunocytochemistry and confocal imaging revealed YFP-positive cells with extensive neurites and intense nuclear labeling for Pou4f1/Brn3a, a transcription factor involved in RGC differentiation (Fig. 4), demonstrating the production of authentic RGCs from Atoh7-expressing progenitor cells. In addition, we detected co-labeling of YFP and the low molecular weight neurofilament (NF68), another marker expressed by postmitotic RGCs (Fig. 5A). Furthermore, we observed co-expression of several other RGC markers, including Pou4f1 and the intermediate molecular weight neurofilament NF145, Tubb3/b-tubulin and Islet-1, as well as NF68 and Pax6 (Fig. 5B-D; Fig. S2). These data demonstrate that RGCs are produced in Atoh7-Cre/ROSA-YFP iPS cells-derived EBs, especially from the Atoh7-expressing lineage.
Previous studies have shown that inhibiting Notch signaling can enhance RGC production from early retinal progenitor cells [31][32][33]. We thus tested whether inclusion of DAPT would enhance RGC development in Atoh7-Cre/ROSA-YFP iPS cell cultures. Quantification using flow cytometry showed that the addition of DAPT at EB culture day 7 significantly increased the number of Pou4f1 + cells from 2.060.2% to 3.460.4% at day 10, but did not affect Pou4f1 and YFP double positive cells (Fig. 6A, B). By day 14  (Fig. 6C, D). Among these populations, Pou4f1 high YFP high cells likely represented authentic RGCs derived from Atoh7-expressing progenitors. DAPT treatment resulted in a statistically significant increase of Pou4f1 high YFP high cells from 8.160.8% to 12.560.9% at day 14 (Fig. 6D, E). We also detected a significant increase in the Pou4f1 low YFP low population, which may reflect the influence of DAPT on Atoh7-expressing progenitors during the 7-day DAPT treatment to generate other retinal cell types including early born photoreceptors (see below). Similar to our observations in the day 10 cultures (Fig. 6A, B), DAPT also significantly increased production of the minor Pou4f1 high YFP low cell population in the day 14 cultures (Fig. 6E), which likely represented non-retinal neurons. These quantitative analyses, together with cell marker labeling and imaging, demonstrate that Atoh7-Cre/ROSA-YFP iPS cells can give rise to RGCs in vitro, and that DAPT treatment enhances RGC production in iPS cellderived EB cultures.

Production of photoreceptor precursors
The Atoh7-expressing retinal progenitors are heavily biased toward generating RGCs as well as the early born cone photoreceptors in vivo [22,23]. We thus examined if the established Atoh7-Cre/ROSA-YFP iPS cells could also produce photoreceptors in vitro under our culture conditions. In day 12 EB cultures, we detected strong nuclear immunofluorescent signals for the cone-rod homeobox protein Crx, which is expressed by postmitotic photoreceptor precursors (Fig. 7A, B). Confocal imaging also showed that the Crx protein expression correlated with localization of the photoreceptor protein Recoverin and cytoplasmic YFP.
To confirm that Atoh7-Cre/ROSA-YFP iPS cells can indeed give rise to Crx + photoreceptor precursors, we used flow cytometry to analyze dissociated EB cells. Since the homeobox protein Otx2 promotes Crx expression [45,46] and Notch signals suppress cone photoreceptor differentiation [31][32][33], we quantified the number of Otx2 + and Crx + cells cultured either in the presence or absence of DAPT. We observed a minor but not statistically significant increase of Crx + cells with DAPT treatment at day 12 of EB   (Fig. 7C, D). Neither Otx2 + cells nor Otx2 + Crx + cells were significantly increased with DAPT treatment at day 12 (Fig. 7D). However, by day 14, DAPT treated cultures showed significantly increased Crx + cells from 7.661.3% to 14.161.7%. Quantification showed that most of the Crx + cells were also YFP + , and DAPT enhanced the Crx + YFP + population from 7.361.0 to 13.761.4% (Fig. 7E, F). Our results thus suggest that in these iPS EB cultures, the Atoh7-expressing lineage also generate Crxexpressing photoreceptor precursors, which can be expanded by inhibiting the Notch signal.

Discussion
Repairing vision impairment associated with axonal damage and/or the loss of RGCs is particularly challenging, as RGCs emerge during early retinogenesis and not only connect to retinal interneurons but also extend centrally projecting axons along complex trajectory routes to establish functional visual circuitry. Moreover, the mechanisms underlying vision-impairing diseases like glaucoma and optic nerve neuropathy remain poorly understood, and have been difficult to study in vivo. It is therefore highly advantageous to establish an in vitro RGC culture system that would facilitate the discovery of disease etiologies and therapeutic treatments. In recent years, significant progress has been made in deriving retinal pigment epithelium and photoreceptors from pluripotent stem cells for therapeutic purposes [47]. However, few studies have focused on RGC differentiation [48,49]. In this study, we have taken advantage of the available mouse genetic and cell reprogramming tools to create iPS cells that carry a molecular reporter for neurons derived from progenitors expressing Atoh7, a retinal specific gene critical for RGC development. Our results show that under favorable neural induction conditions, mouse Atoh7-Cre/ROSA-YFP iPS cells can activate the Cre reporter and recapitulate the developmental potential of the endogenous Atoh7-expressing cell lineage. These iPS cells can thus serve as a useful stem cell source for mechanistic studies of RGC and photoreceptor genesis and differentiation, as well as for studies of RGC axon trajectory, guidance, and survival.
Since we used heterozygous Atoh7 Cre-KI/+ ; Rosa YFP-KI/+ MEFs to derive the Atoh7-Cre/ROSA-YFP iPS cells, expression of the knock-in Cre should reflect expression of the endogenous Atoh7, without affecting RGC specification. Furthermore, although Atoh7 expression in the developing mouse retina is transient, the YFP reporter expression from the ROSA locus is constitutive once the Cre-loxP recombination has occurred. Thus, the YFP reporter should mark all progeny derived from cells that have expressed Atoh7 and thus enable cell lineage tracing and monitoring. These Atoh7-Cre/ROSA-YFP iPS cells are particularly useful, as currently there are no reliable commercial anti-Atoh7/Math5 antibodies that can specifically detect Atoh7 protein in immunocytochemical labeling assays. Our characterization has shown that the YFP reporter is expressed as early as day 7 in the EB culture, correlating with the expression of a number of homeobox genes, including Rax, Pax6 and Vsx2, which are normally expressed in the eye domain and retinal primordium. In addition, the YFP reporter expression also coincides with the transcript expression of two early neurogenic bHLH factors Neurog2 and Atoh7. These temporal gene expression results are consistent with previous studies using mouse Rax GFP-KI ES cells under similar culture conditions [36].
The specification of RGCs in vivo is negatively regulated by Notch signaling and a number of secreted cell extrinsic cues [24,26,27,29,30,33]. We show that the c-secretase inhibitor DAPT can significantly expand production of the Atoh7 lineage cells from 19% to 37% by day 14, when the majority of the YFP-labeled cells represent postmitotic neurons. This suggests that DAPT can effectively relieve the suppression of Notch and possibly other signals on RGC specification and thereby increase the number of progeny derived from Atoh7-expressing progenitors in the iPS cell EB cultures. Indeed, FACS quantification data show that DAPT treatment results in a 50% increase of the Pou4f1 and YFP double positive RGCs (from 8% to 12% by day 14). These Pou4f1 and YFP double positive cell are likely authentic RGCs as the fluorescent reporter reflects the developmental history of Atoh7 expression from its endogenous locus. Not surprisingly, DAPT treatment also enhanced a minor population (,1%) of Pou4f1positive cells that did not express high levels of the YFP reporter. These likely represent non-retinal neurons produced in the culture, possibly cortical neurons. Cell lineage tracing studies in vivo have shown that Atoh7expressing progenitors also give rise to other retinal cell types, among which a large portion becomes cone photoreceptors [22,23]. In agreement with the in vivo lineage analyses, we have also detected Crx and YFP double positive cells in the iPSC cultures, indicating that certain Atoh7 lineage-derived cells have adopted a distinct fate from RGC and become photoreceptor precursors. Interestingly, although Otx2 is an upstream positive regulator of Crx [45,46], we did not detect any significant increase in the number of Otx2, or Otx2 and Crx double positive cells with DAPT treatment. This may reflect that Notch signaling leads to only transient enhancement of Otx2 expression in a selective population of retinal progenitors, and these cells then quickly become post-mitotic Crx-expressing photoreceptor precursors at day 12. However, DAPT treatment indeed resulted in enhanced production of Crx and YFP double positive cells (from 7% to 13% by day 14), consistent with the phenotypes of Notch1 and RBPj deletions in the retina [31][32][33]. As expected from in vivo cell lineage tracing studies [22,23], we also observed Crx-positive photoreceptor precursors that do not express YFP, and thus were not derived from the Atoh7 lineage. Although we did not pursue the terminal differentiation of the Crx-positive photoreceptor Deriving Retinal Ganglion Cells from iPS Cells PLOS ONE | www.plosone.org precursors in this study, this population of Atoh7 lineage cells likely contains cone cell precursors, as their production period in vivo overlaps extensively with RGC genesis.
In summary, we have established mouse iPS cells that carry a reporter marking an important sub-neuronal cell lineage of the retina. These pluripotent Atoh7-Cre/ROSA-YFP iPS cells will be useful tools in future investigations to study ocular tissue differentiation, the onset of retinal neurogenesis, cell fate specification, as well as neuronal differentiation and survival.