Small-molecule–directed, efficient generation of retinal pigment epithelium from human pluripotent stem cells

Significance

Cell-based approaches utilizing retinal pigment epithelial (RPE)-like cells derived from human pluripotent stem cells (hPSCs) are being developed for the treatment of retinal degeneration. In most research published to date, the choice of the factors used to induce RPE differentiation is based on data from developmental studies. Here, we developed an unbiased approach directed at identifying novel RPE differentiation-promoting factors using a high-throughput quantitative PCR screen complemented by a novel orthogonal human induced pluripotent stem cell (hiPSC)-based RPE reporter assay. We identified chetomin, a dimeric epidithiodiketopiperazine, as a strong inducer of RPE; combination with nicotinamide resulted in efficient RPE differentiation. Single passage of the whole culture yielded a highly pure hPSC-RPE cell population that displayed many of the morphological, molecular, and functional characteristics of native RPE.

Abstract

Age-related macular degeneration (AMD) is associated with dysfunction and death of retinal pigment epithelial (RPE) cells. Cell-based approaches using RPE-like cells derived from human pluripotent stem cells (hPSCs) are being developed for AMD treatment. However, most efficient RPE differentiation protocols rely on complex, stepwise treatments and addition of growth factors, whereas small-molecule–only approaches developed to date display reduced yields. To identify new compounds that promote RPE differentiation, we developed and performed a high-throughput quantitative PCR screen complemented by a novel orthogonal human induced pluripotent stem cell (hiPSC)-based RPE reporter assay. Chetomin, an inhibitor of hypoxia-inducible factors, was found to strongly increase RPE differentiation; combination with nicotinamide resulted in conversion of over one-half of the differentiating cells into RPE. Single passage of the whole culture yielded a highly pure hPSC-RPE cell population that displayed many of the morphological, molecular, and functional characteristics of native RPE.

Age-related macular degeneration (AMD) is the leading cause of irreversible vision loss and blindness among the elderly in industrialized countries. Dysfunction of the retinal pigment epithelium (RPE) is an early event associated with AMD. The RPE, a monolayer of pigmented cells directly abutting the photoreceptor cell layer, plays many important roles in vision and in maintaining the health and integrity of the retina (1). As the RPE deteriorates, there is progressive degeneration of photoreceptor cells.

Successful antiangiogenesis treatments have been developed for the neovascular, or “wet,” form of AMD. However, there are no Food and Drug Administration-approved treatment options available for the majority of AMD patients, who suffer from the more common nonneovascular, or “dry,” form of the disease. In the past few years, however, transplantation of human pluripotent stem cell-derived RPE (hPSC-RPE) has emerged as a promising new therapy for dry AMD. A Phase I clinical trial of human embryonic stem (hES)-derived RPE cells recently reported some preliminary encouraging results (2). Additionally, a Phase I trial that will use RPE cells generated from human induced pluripotent stem cells (hiPSCs) reprogrammed from the patients’ own skin cells recently injected their first patient (3).

If the promise of hiPSC-based approaches for AMD is to be translated into the clinic, each patient would require individualized generation of RPE cells from his or her stem cells, thereby necessitating the development of simple, efficient, safe, and affordable protocols for RPE generation. Although highly efficient protocols have been established, they rely upon mixtures of growth factors (4–6) with the use of complex biologics derived from animal cells or bacteria, presenting potential clinical challenges. As an alternative approach, use of nonbiological products such as small molecules would limit the risk of infection or immune rejection (7). In addition, they offer a cost-effective alternative with less lot-to-lot variability. As a consequence, the development of small-molecule–based protocols has been the focus of extensive research (8). To date, however, the small-molecule–only protocols that have been developed for RPE differentiation have demonstrated only limited efficiency (9).

We report here development of a highly efficient, small-molecule–based protocol for the production of RPE from hPSC that is suitable for clinical application. We performed a quantitative real-time PCR (qPCR)-based high-throughput screen (HTS) for molecules that promote RPE differentiation, and complemented the screen with a novel orthogonal hiPSC-based RPE reporter assay for compound validation. Using this strategy, we identified chetomin (CTM) as a potent promoter of RPE differentiation. Its use in combination with a previously known neural inducer, nicotinamide (NIC) (10, 11), provides a one-step treatment for differentiation of a wide range of hPSC lines. Following a single whole-dish passage, a pure and functional monolayer of RPE cells is obtained. This protocol should prove useful for the cost-efficient production of RPE cells.

Results

An RPE-Specific Tyrosinase Enhancer Drives Green Fluorescent Protein Expression in RPE Cells During hiPSC Differentiation into RPE.

To follow RPE differentiation, we used a reporter hiPS cell line that expresses green fluorescent protein (GFP) specifically when the cells are differentiated into the RPE lineage (12). A GFP cassette was put under the control of an RPE-specific tyrosinase enhancer (Fig. 1A) (13) following a previously described approach (14). In mouse, we verified that this construct provided spatially and temporally restricted expression of GFP to RPE cells (Fig. 1 B and C). The tyrosinase enhancer–GFP construct (Tyr-GFP) was cloned into a lentiviral vector containing a red fluorescent protein (RFP) controlled by a constitutive Ubiquitin-C promoter (Fig. 1D). The construct was transduced into B42, a fibroblast-derived hiPS cell line (15), to generate the reporter line 3D1. In their undifferentiated, pluripotent state, 3D1 cells constitutively expressed RFP without evidence of GFP expression (Fig. 1E). After about 30 d of spontaneous differentiation, GFP expression was observed in colonies composed of pigmented cells whose morphology was reminiscent of RPE cells (Fig. 1F). To validate that GFP expression was indeed associated with RPE differentiation, fluorescence-activated cell sorting (FACS) was performed on the GFP-positive and -negative populations. Upon reseeding, FACS-sorted GFP-positive cells formed a confluent culture of GFP+/RFP+ cells, with a polygonal and pigmented morphology characteristic of RPE cells (Fig. 1 G and H). They also displayed higher expression of mature RPE markers compared with the GFP-negative cells (Fig. 1I). Together, these data show that GFP expression is specifically activated in cells committed to RPE fate during in vitro differentiation of Tyr-GFP 3D1 hiPSC.

Fig. 1.

Specific expression of a Tyr-GFP reporter in vivo, and in vitro during hiPSC differentiation. (A) Schematic of the GFP cassette under control of an RPE-specific enhancer (Tyr-GFP). (B and C) Tyr-GFP expression in vivo in the mouse eye. (D) Schematic of the Tyr-GFP:Ub-c-RFP construct. (E) Microscope images of 3D1 hiPSC in the pluripotent state. (F) Microscope images of 3D1 hiPSC during spontaneous RPE differentiation. (G) Morphology of Tyr-GFP 3D1 hiPSC after FACS and further culture for 1 mo. (H) Microscope images of 3D1 hiPSC-RPE. (I) RPE mature marker expression in sorted 3D1 hiPSC-RPE as measured by qPCR. Expression levels of GFP-positive sorted cells were normalized by expression levels of the GFP-negative cells. Error bars represent SD of biological replicates. BEST1, bestrophin 1; DCT, dopachrome tautomerase; PMEL17, premelanosome protein; RPE65, retinal pigment epithelium-specific protein, 65 kDa; TYR, tyrosinase; TYRP1, tyrosinase-related protein 1. [Scale bar: 40 μm (for all images in Fig. 1).]

High-Throughput qPCR Screen Identifies CTM as a Potent Inducer of RPE Differentiation.

As our primary screen for RPE differentiation-promoting small molecules, we directly measured endogenous RPE marker expression using high-throughput qPCR (HT qPCR). Tyr-GFP 3D1 hiPSCs were seeded in 384-well plates and treated for 10 d with four different concentrations of molecules from a focused library of 303 compounds (StemSelect; Millipore) (SI Materials and Methods and Table S1). Cells were then cultured for 5 d in differentiation medium (DM) to allow for higher marker expression levels to be achieved, and then expression of three RPE markers [microphthalmia-associated transcription factor (MITF), orthodenticle homeobox 2 (OTX2), and premelanosome protein (PMEL17)] was assessed by qPCR (Fig. 2A). The transcription factors MITF and OTX2 are both required for RPE differentiation, whereas PMEL17 is a matrix protein present in the melanosome precursors of pigmenting cells. Only one of the compounds tested, CTM, was found to consistently up-regulate all three RPE markers: 40 nM CTM increased the expression of the markers 9- to 15-fold compared with DMSO-treated controls (Fig. 2B), whereas higher concentrations of CTM caused marked cell death. CTM is a dimeric epidithiodiketopiperazine (ETP) metabolite of the fungus Chaetomium cocliodes and Chaetomium seminudum species (Fig. 2C).

Table S1.

Quantitative real-time PCR sequence list

Gene		Primer sequence (5′ to 3′)	Annealing temperature, °C	Source
SRP72	F	TCTGCCTCTACAAGTAACATCAT	58	Synnergren (28)
	R	CTCATCACCAGCCACCTT
CREBBP	F	GAGAGCAAGCAAACGGAGAG	60	Synnergren (28)
	R	AAGGGAGGCAAACAGGACA
FBXL12	F	GCCTTGGTCATATCATCAG	60	Synnergren (28)
	R	TTCTTCATCCGTCCTGTT
OTX2	F	ACCTTGAACTCCACCTCT	56	This study
	R	GCTTCTCTTCTCTGACTCTCTTTG
PAX6	F	AGTGAATCAGCTCGGTGG	60	Meyer et al. (29)
	R	TGCAGAATTCGGGAAATGTCGCAC
MITF	F	TTCACGAGCGTCCTGTATGCAGAT	60	Meyer et al. (29)
	R	TTGCAAAGCAGGATCCATCAAGCC
TYROSINASE	F	ATTGGGACTGGCGGGATG	56	This study
	R	GCATAAAGACTGATGGCTGTTG
GATA6	F	AAAGACTTGCTCTGGTAATAGC	60	This study
	R	GGCTGTAGGTTGTGTTGTGG
T(brach)	F	TAGCGAGAAATATGCCGAGGAG	60	This study
	R	GAGCTGCGTGATCCGATGG
EOMES	F	AAGGCTTCAGAGACAACTATG	60	This study
	R	CGACCTCCAGGGACAATC
HAND1	F	TCAAGGCTGAACTCAAGAAGG	58	This study
	R	CGGCTCACTGGTTTAACTCC
NODAL	F	AGAAGATGTGGCAGTGGATGG	58	This study
	R	GCTCTGTGTCGGGCTTTGG

Fig. 2.

Identification and validation of CTM as a promoter of RPE differentiation. (A) Schematic of the HT qPCR screening strategy. (B) RPE marker expression in wells treated with CTM (n = 1 for each concentration). Expression levels were normalized by mean of DMSO-treated samples using the ΔΔCt method. (C) Chemical structure of CTM. (D) RPE marker expression at day 15 following CTM treatment in four different hPSC lines, as measured by qPCR. Data were normalized by the expression levels in control conditions. (E) Flow-cytometric analysis of the expression of Tyr-GFP at day 35 in control and CTM-treated samples (red). 3D1 hiPSC in the pluripotent state were used as a negative control (black). (F) Expression of Tyr-GFP at day 35 following CTM treatment, as measured by flow cytometry. (G) Flow-cytometric analysis of the expression of PMEL17 at day 35 in control and CTM-treated samples (red). The isotype control is displayed in black. (H) Expression of PMEL17 at day 35 following CTM treatment, as measured by flow cytometry in four different hPSC lines. For each hPSC line, differing letters indicate statistically significant mean differences (P < 0.05 by ANOVA). In all cases, error bars represent SD of three biological replicates.

Next, we validated the ability of CTM to induce RPE marker expression in four different hPSC lines. Following the protocol of the primary screen, cells were cultured in the presence of CTM from day 0 to day 10 at concentrations ranging from 1.25 to 80 nM in twofold increments. About a week after initiation of CTM treatment, a high level of cell death was observed above 20 nM CTM in the hESC lines H7 and H9, whereas the hiPSC lines IMR904 and 3D1 tolerated up to 40 and 80 nM CTM, respectively. Although the dose–response curves differed, all three markers were up-regulated by CTM in a dose-dependent pattern in all four hPSC lines (Fig. 2D). These results confirm the HT qPCR data and indicate that CTM’s effect on RPE marker expression is not limited to the specific cell line used in the screen.

To further validate CTM’s effect on RPE differentiation in an orthogonal assay, 3D1 hiPSC were treated with 10–80 nM CTM for 10 d, and flow-cytometric analysis of Tyr-GFP expression was performed at day 35. The percentage of Tyr-GFP+ cells followed a dose-dependent pattern, with a plateau above 40% positive cells for 40–80 nM CTM compared with about 15% in control (P < 10⁻⁴ by ANOVA) (Fig. 2 E and F). We repeated the dose–response experiments and assessed the extent of RPE differentiation by immunostaining for PMEL17 followed by flow cytometry (Fig. 2G). Control conditions yielded 10–15% PMEL17+ cells after 35 d of differentiation, whereas treatment with 20 nM CTM increased the percentage to 30% for H7 and H9 cells, and treatment with 40 nM CTM increased efficiency to ∼40% for IMR904 and 3D1 cells (P < 0.05 by ANOVA for all four hPSC lines) (Fig. 2H), demonstrating that the differentiation efficiency is increased threefold to fourfold by CTM treatment.

Despite the presence of large areas of PMEL17+ cells at day 15, and Tyr-GFP+ cells at day 30 of differentiation (Fig. S1 A and B), we surprisingly did not see pigmented clusters in the CTM-treated cultures, although they were readily apparent in the control wells (Fig. S1C). To assess whether the Tyr-GFP+ cells obtained after CTM treatment were indeed bona fide RPE cells, the whole culture was passaged onto a new dish and grown in RPE medium (RPEM) (Materials and Methods). During the next month, cells grew into a confluent monolayer made of pigmented and polygonal cells (Fig. S1D), a morphology characteristic of RPE cells. They also homogeneously expressed the RPE markers MITF and the tight junction protein ZO-1 (Fig. S1D), and ∼95% were Tyr-GFP+ (Fig. S1 D and E). On the contrary, when control cultures were passaged into RPEM, they failed to grow into a homogenous pigmented monolayer, and instead formed pigmented clusters surrounded by fibroblast looking cells, as previously described (16). These results demonstrate that CTM is a potent and reproducible inducer of RPE differentiation.

Fig. S1.

Morphology and expression of RPE markers during hiPSC differentiation following CTM treatment, and in hiPSC-RPE after whole-dish passage. (A) Microscope images showing expression of PMEL17 by immunostaining in hiPSC IMR904, after 15 d of differentiation. (Scale bar: 100 μm.) (B) Microscope image showing Tyr-GFP expression in differentiating 3D1 hiPSC at day 35 of the protocol. (Scale bar: 40 μm.) (C) Morphology of 3D1 hiPSC after 35 d of differentiation. Note the absence of pigmented clusters in the CTM-treated well. (D) Microscope images displaying the expression of RPE markers and Tyr-GFP in 3D1 hiPSC-RPE, 1 mo after whole-dish passage. (Scale bar: 40 μm.) (E) Flow-cytometric analysis of the expression of Tyr-GFP (in red) in 3D1 hiPSC-RPE, 1 mo after whole-dish passage. 3D1 hiPSC in the pluripotent state were used as a negative control (in black).

Disruption of the CH1 Domain of p300 Is Not the Main or Only Mechanism Involved in CTM-Induced RPE Differentiation.

CTM has known antimicrobial activity (17) and is a disruptor of the CH1 domain of the p300 coactivator (18). To test whether CTM’s RPE-inducing activity was related to its effect on p300, we assessed the ability of ETP2, another dimeric ETP compound (Fig. S2A) that binds the CH1 domain with high affinity, to induce RPE differentiation in hPSC. Dose–response treatment of 3D1 hiPSC with ETP2 for 10 d was followed by qPCR at day 15. Massive cell death was observed with ETP2 at 50 nM and above. MITF and OTX2 were both significantly up-regulated upon ETP2 treatment, although their respective peak increase was reached at different concentrations (Fig. S2B). On the other hand, PMEL17, a more mature RPE marker, did not show any significant change in expression with ETP2 treatment.

Fig. S2.

Probing the effect of CH1 domain disruption on hPSC differentiation using ETP2. (A) Chemical structure of ETP2, a member of the dimeric epidithiodiketopiperazine class. (B) Expression of key RPE markers at day 15 following dose–response treatment with ETP2, as measured by qPCR. Data were normalized by the expression levels of control samples. Error bars represent SD of three biological replicates. (C) Expression of Tyr-GFP by flow cytometry at day 35 following dose–response treatment with ETP2. As a positive control, 3D1 hiPSC were treated in parallel with 40 nM CTM (same samples as in Fig. 2F). Error bars represent SD of three biological replicates. (D) Expression of key lineage markers during embryoid body differentiation in the presence of CTM. Error bars represent SD of three biological replicates.

We further assessed the extent of RPE differentiation following ETP2 treatment by analyzing Tyr-GFP expression at day 35 by flow cytometry (Fig. S2C). We did not observe any statistically significant change between control and ETP2-treated samples (P > 0.27 by ANOVA). Interestingly, pigmented colonies were not observed above 6 nM ETP2. These results suggest that CTM’s differentiation-promoting activity, such as RPE lineage induction or initiation of early differentiation, may be due in part to CH1 domain disruption, whereas the ability to promote more mature RPE differentiation requires an activity in addition to CH1 domain disruption.

To gain a better understanding of CTM action at a broader level, hPSC-derived embryoid bodies were grown for 15 d in the presence or absence of 50 nM CTM, followed by qPCR for key markers of the primordial lineages (Fig. S2D). Although neuroectodermal markers (PAX6 and OTX2) were up-regulated, as might be expected based on our earlier results, CTM also strongly inhibited expression of markers of the mesodermal and endodermal fates (P < 0.05 by multiple T test). These results indicate that CTM’s mechanism of action might be quite complex and multifactorial, as it induces neuroectodermal differentiation while actively repressing alternative cell fates.

NIC Prevents Excessive Cell Death During CTM-Induced RPE Differentiation.

We next tested different lengths of time for CTM treatment and found that expression of key RPE markers was optimal when CTM was added during the first 2 wk of differentiation.

Importantly, while performing these experiments, we observed significant cell death of H7 cells with CTM treatment length of 2 wk and above. CTM cellular toxicity has been described previously (18). In sensitive hPSC lines, such as H7, we observed patches of differentiating cells after 2-wk CTM treatment at 25 nM, whereas there was only minimal cell survival at 50 nM. On the contrary, in less sensitive lines, such as 3D1, cells formed a confluent monolayer regardless of the culture conditions. These observations were quantified by counting the remaining live cells by flow cytometry, after 2-wk CTM treatment (Fig. S3). Apoptotic and dead cells were excluded from the analysis based on SytoxRed staining. As NIC has been reported to protect hPSC from cell death during neuroectoderm differentiation through PARP1 inhibition (11), we tested whether it could reduce CTM-induced cell death. Cotreatment with 10 mM NIC prevented cell loss in H7 at 50 nM CTM, and cotreatment with NIC more than doubled the number of live 3D1 cells when 50 nM CTM was used (Fig. S3). Overall, CTM cellular toxicity can be prevented by combination with NIC, so confluent cell monolayers can be maintained throughout hPSC differentiation.

Fig. S3.

Optimization of CTM treatment. Live-cell counting by flow cytometry after 2-wk treatment with small molecules. Data were normalized by the number of live cells in control conditions.

Combined Small-Molecule Treatment Followed by Culture in RPEM Leads to High-Efficiency Generation of RPE Cells with Characteristic Morphology.

Although CTM, or CTM plus NIC, treatment led to RPE induction as measured by marker expression and Tyr-GFP reporter activation, surprisingly pigmented colonies were not readily observed (Fig. S1B and Fig. 3A). We previously showed that, after passage in RPEM, a medium originally designed for fetal RPE culture, CTM-induced RPE cells were able to acquire characteristic RPE morphology. Therefore, we decided to test whether growing differentiating hPSC in RPEM would promote RPE pigmentation. After 2-wk induction with CTM and NIC in DM, hPSC were switched to RPEM (Tables S2 and S3). Within 1–2 wk of switching to RPEM, pigmented clusters started to become visible, and by the third week they covered the majority of the well surface (Fig. 3A). For cultures switched to RPEM, robust Tyr-GFP expression was observed in colonies with polygonal and pigmented cells, whereas for cells maintained in DM, there was no perceptible change in morphology despite Tyr-GFP activation (Fig. 3B). Flow cytometry was used to quantify the effect of RPEM on Tyr-GFP expression. Cells cultured in RPEM showed higher levels of Tyr-GFP expression (Fig. 3C), with a 24% increase in median fluorescent index (MFI) for the whole population compared with cells maintained in DM (P = 0.02, t test) (Fig. 3D). The percentage of Tyr-GFP+ cells was moderately increased, from 55.6% (±5.5%) for hiPSC differentiated in DM to 63.9% (±5.4%) for their counterparts grown in RPEM (P = 0.04, t test). Altogether, these results indicate that switching to RPEM following cotreatment with CTM and NIC both promotes RPE morphology and increases differentiation efficiency.

Fig. 3.

Differentiation of hPSC into RPE following treatment with combination of CTM and NIC. (A and B) Morphology of differentiating hPSC at day 35 in DM vs. RPEM, with A showing macroscopic and B showing microscopic images. (Scale bar: 40 μm.) (C) Flow-cytometric analysis of Tyr-GFP expression at day 35 after culture in DM or RPEM. 3D1 hiPSC in the pluripotent state were used as a negative control (black). (D) Histogram displaying the percentage of Tyr-GFP+ cells and the median fluorescent intensity (MFI) of the whole cell population in 3D1 hiPSC. (E) Expression of RPE markers at day 35, as measured by flow cytometry in 3D1 hiPSC. (F) Flow-cytometric analysis of PMEL17 expression at day 35 in hPSCs cultured in RPEM (red). The isotype control is displayed in black. (G) Expression of PMEL17 at day 35 in four different hPSC lines, as assessed by flow cytometry. The difference between NIC and NIC plus CTM is statistically significant by T test (P < 0.001 for all four hPSC lines). (H) Schematic of hPSC differentiation into RPE using combination of NIC and CTM, followed by culture in RPEM. RPE purification is achieved at day 35 by whole-dish passage (P1). Error bars represent SD of three biological replicates.

Table S2.

Composition of Differentiation Medium

Medium and supplements	Concentration	500 mL
DMEM/F12 (Invitrogen; 11330)		425 mL
KO SR (Invitrogen; 10828)	15% (vol/vol)	75 mL
Glutamine 100× (Invitrogen; 25030)	2 mM	5 mL
NEAA 100× (Invitrogen; 11140)	0.1 mM	5 mL
Anti-Anti 100× (Invitrogen; 15240)	1%	5 mL
β-Mercaptoethanol (Sigma; M3148)	0.1 mM	3.5 µL

Table S3.

Composition of RPE medium (30)

Media and supplements	Concentration, % (vol/vol)	500 mL
DMEM (Invitrogen; 11965)	70	350 mL
F12 (Invitrogen; 11765)	30	150 mL
B27, 50× (Invitrogen; 17504)	2	10 mL
Anti-Anti, 100× (Invitrogen; 15240)	1	5 mL

In a parallel set of experiments, we observed that switching to RPEM after 2 wk of CTM treatment also led to the emergence of pigmented colonies at 35-d differentiation (data not shown). However, the percentages of both Tyr-GFP+ and PMEL17+ cells were significantly lower (P < 0.0054 for Tyr-GFP, P < 0.0026 for PMEL17) compared with cotreatment with CTM and NIC (Fig. 3E). Thus, the addition of NIC to CTM also helps further improve RPE generation efficiency, in agreement with recent studies on its role during RPE differentiation (6, 10).

Compared with control and NIC alone, cultures treated with CTM or with a combination of CTM and NIC showed a strong induction of the eye field transcription factor paired box gene 6 (PAX6) and MITF, whereas OTX2 expression was maintained at higher levels throughout the first month of differentiation (Fig. S4). As a presumed consequence of the strong up-regulation of RPE-inducing transcription factors (14), PMEL17 expression was dramatically increased after the second week. Compared with NIC alone, tyrosinase (TYR), which encodes the enzyme responsible for the conversion of tyrosine to pigment melanin, also displayed sixfold up-regulation by the fourth week of differentiation in cultures treated with CTM alone, or cotreated with CTM and NIC. This is consistent with the extended area of pigmented cells observed in cultures treated with CTM and NIC.

Fig. S4.

Kinetics of marker expression of differentiating hPSC as measured by qPCR. Error bars represent SD of three biological replicates. Data for CTM alone were obtained from an independent experiment and normalized to an internal control common with the three other conditions tested.

Finally, we validated our optimized approach for RPE differentiation (Fig. 3H) on four additional hPSC lines. All four lines responded to the protocol by yielding RPE cells with efficiencies ranging from 45% to 60% of PMEL17+ cells at day 35 (Fig. 3 F and G). Altogether, these data show that during CTM-plus-NIC–directed differentiation, RPE markers are strongly up-regulated. As a consequence, high yields of RPE cells are obtained from a wide range of hPSC lines.

Pure, Polarized, and Functional RPE Monolayers Are Obtained After a Single Whole-Dish Passage.

After 2-wk cotreatment with CTM and NIC in DM, differentiating hPSC lines 3D1, IMR904, and H7 were switched to RPEM. At day 35, whole cultures were dissociated to single cells, passaged at high density (P1), and cultured in RPEM (Fig. 3H). A month later, the 3D1 culture had grown into a monolayer of pigmented and polygonal cells, and displayed a majority of Tyr-GFP–expressing cells (Fig. 4A and Fig. S5A). Domes had also formed (Fig. S5B), suggesting that the cells were functional and actively transporting fluid, as is known to occur in mature, tight junction-bearing, RPE cells (1). Key RPE markers such as MITF, PMEL17, and the tight junction protein ZO-1 were strongly expressed in all three tested hPSC lines (Fig. 4B). When hPSC-RPE cells were left to mature for another month, they also expressed the chloride channel-related protein Bestrophin 1 (BEST1), as well as retinaldehyde-binding protein 1 (RLBP1) and retinal pigment epithelium-specific protein of 65 kDa (RPE65), both proteins involved in the visual cycle (Fig. 5B). Additionally, the cells show the polarity expected of RPE cells, with ZO-1 and BEST1 showing apical (Fig. 4C) and basolateral (Fig. 4D) localization (19, 20). Quantification of marker expression by flow cytometry indicated that the hPSC-RPE cultures were highly pure: 95.1–98.6% MITF+, 92.6–98.3% PMEL17+, and >95% Tyr-GFP+ (Fig. 4 E–G and Table S4).

Fig. 4.

Characterization of hPSC-RPE obtained after NIC-plus-CTM induction. (A) Microscope images of 3D1 hiPSC-RPE. (Scale bar: 40 μm.) (B) Microscope images of IMR90-4–derived hPSC-RPE immunostained for key RPE markers. Insets display higher magnification of the images with Hoechst (blue) counterstaining. (Scale bar: 40 μm.) Immunostaining of the top three panels was performed on hPSC-RPE cultured for 60 d after P1, whereas the bottom three panels were obtained following 30 d culture after P1. (C and D) Z-stack confocal micrographs showing typical polarized expression of RPE proteins, with ZO-1 (red) demonstrating apical localization (C) and BEST1 (red) demonstrating basolateral localization (D). The nuclei were counterstained with Hoechst (cyan). (Scale bars: 20 μm.) (E–G) Flow-cytometric analysis of the expression of RPE markers in hPSC-RPE 1 mo after purification by whole-dish passage. The profile of cells stained with RPE marker antibodies is shown in red, and the isotype control is displayed in black. (H) Flow-cytometric analysis of hPSC-RPE cells phagocytosis. The profile of hPSC-RPE incubated with photoreceptor outer segment (POS) is shown in red. The no-POS control is displayed in black, whereas POS internalization was specifically blocked with an anti-integrin α_Vβ₅ antibody in the control shown in blue. (I) Confocal microphotograph showing phagocytosis of POS (green) by IMR90-4–derived hPSC RPE. The apical sides of hPSC-RPE cells are stained with ZO-1 (red), whereas nuclei are counterstained with Hoechst (cyan). (Scale bar: 20 μm.) Inset shows a higher magnification with yellow arrows pointing at phagocytosed POS. (J) Polarized secretion of VEGF-A from the apical and basal sides of hPSC-RPE cells grown on Transwells. (K) Polarized secretion of PEDF from the apical and basal sides of hPSC-RPE cells grown on Transwells. For both J and K, the difference between apical and basal secretion is statistically significant by t test (P < 0.01 for all three hPSC lines). In all cases, error bars represent SD of three biological replicates.

Fig. S5.

Morphology of hPSC-RPE cells following whole-dish passage. (A) Several wells of a six-well plate with hPSC-RPE after 1 mo of maturation. (B) Microscope images showing the formation of a fluid-filled dome by the hPSC-RPE monolayer (arrow) in two different focal planes. (Scale bar: 40 μm.)

Fig. 5.

Transplantation of 3D1 hiPSC-RPE into the subretinal space of NOD-SCID albino mice. Confocal micrograph showing the presence of rhodopsin-positive fragments (yellow arrowheads) 2 wk posttransplantation within the RFP-positive cytoplasmic compartment of injected 3D1 hiPSC-RPE.

Table S4.

Purity and functionality of hPSC-RPE cells, as assessed by flow cytometry analysis of key RPE marker expression, and phagocytosis of POS

	Percentage of positive cells
hiPSC line	MITF, %	PMEL17, %	Tyr-GFP, %	POS, %
3D1	95.1 (±0.6)	92.6 (±1.7)	95.5 (±0.5)	NA
IMR904	98.6 (±1.8)	98.3 (±2.7)	NA	71.4 (±3.5)
H7	96.3 (±0.9)	93.8 (±1.1)	NA	66.7 (±4.9)

NA, not applicable.

Next, we tested the phagocytosis of photoreceptor outer segments (POS), which is an essential function assumed by RPE in vivo (1). hPSC-RPE cells were incubated in the presence of FITC-labeled POS for 5 h. Free POS were washed away and external fluorescence was quenched with trypan blue to visualize internalized POS, whereas ZO-1 staining marked the apical side. Apically localized POS were extensively observed within hPSC-RPE layers (Fig. 4I). Flow cytometry analysis (21) indicated that 66.7–71.4% of hPSC-RPE cells had phagocytosed POS (Fig. 4H and Table S4). RPE cells in vivo are also known to secrete preferentially vascular endothelial growth factor (VEGF) and pigment epithelium-derived factor (PEDF) to the basal and apical side, respectively (1). We therefore grew hPSC-RPE cells on Transwells and collected medium from the upper and lower reservoirs. VEGF was preferentially found in the lower reservoir, reflecting basal secretion (Fig. 4J), whereas PEDF secretion was predominant in the upper reservoir, reflecting apical secretion (Fig. 4K). This polarized pattern of growth factor secretion is in accordance with previous studies using native RPE and hPSC-RPE (19, 22). Taken together, these results demonstrate that, following small-molecule–directed differentiation of hPSC, a polarized and functional monolayer of hPSC-RPE showing a high degree of cell purity is obtained from a single whole-dish passage.

Directed RPE Differentiation Is Clinically Relevant.

To increase the clinical relevance of our small-molecule–only approach for RPE differentiation, we adapted it to a completely defined and xeno-free process: 3D1 hiPSCs were differentiated with NIC and CTM on synthetic vitronectine peptide-acrylate surfaces using xeno-free versions of B27 and KnockOut serum (Supporting Information). This process yielded over 60% (64.1 ± 3.8%) Tyr-GFP+ cells after 1 mo of differentiation (Fig. S6A), whereas high-purity RPE monolayers (95.6 ± 0.4% PMEL17+, 95.9 ± 0.1% Tyr-GFP+) expressing key markers were obtained following a single passage (Fig. S6 B–D). Additionally, after verifying that they had a normal karyotype (Fig. S7), 3D1 hPSC-RPE cells at P1 were injected into the subretinal space of albino, NOD-SCID mice. Fundus imaging taken 2 wk later displayed the presence of numerous pigmented clusters in the injection area, suggesting that the injected cells had survived the transplantation process and maintained their RPE phenotype in vivo (Fig. S8A). Retinal section confirmed this observation showing RFP- and GFP-positive cells, integrated within the host RPE layer (Fig. S8B). The presence of rhodopsin-positive fragments within the RFP-positive cytoplasmic compartment of hiPSC-RPE cells suggests that these cells were functional and phagocytosed POS from the endogenous photoreceptor cells (Fig. 5, arrowheads). Significantly, we did not detect any cellular overgrowth that would have been suggestive of tumor formation in any of the injected animals (n = 8 injected eyes).

Fig. S6.

Generation of 3D1 hiPSC-RPE in defined and xeno-free conditions. (A) Flow-cytometric analysis of the expression of Tyr-GFP (in red) in 3D1 hiPSC-RPE, after 35 d of differentiation. 3D1 hiPSC in the pluripotent state were used as a negative control (in black). (B) Bright-field microscope image showing hiPSC-RPE 1 mo after whole-dish passage. (Scale bar: 100 μm.) (C) Microscope images displaying the expression of Tyr-GFP and RFP from (B) at higher magnification. (Scale bar: 100 μm.) (D) Microscope images displaying the expression of RPE markers by immunostaining in 3D1 hiPSC-RPE, 1 mo after whole-dish passage. (Scale bar: 100 μm.) (E) Flow-cytometric analysis of the expression of PMEL17 and Tyr-GFP (in red) from 1 mo 3D1 hiPSC-RPE. An Isotype control antibody and 1 mo GFP negative hiPS RPE cells were used as negative control (black) for analyzing PMEL17 and Tyr-GFP expression, respectively.

Fig. S7.

G-banded karyotype analysis of 3D1 hPSC-RPE at P1.

Fig. S8.

Transplantation of 3D1 hiPSC-RPE into the subretinal space of NOD-SCID albino mice. (A) Fundus photograph of an injected eye, 2 wk posttransplantation. (B) Confocal micrograph showing the presence of RFP- and GFP (evidenced by anti-GFP antibody immunostaining)-positive cells within the host RPE layer. Abbreviation: INL, inner nuclear layer; IS, inner segment; ONL, outer nuclear layer; OPL, outer plexiform layer; OS, outer segment; RPE, retinal pigment epithelium.

Discussion

In most studies reported to date, directed differentiation of hPSC into RPE is induced, at least in part, by the use of growth factor and related molecules that recapitulate in vivo developmental cues. In this report, we used a discovery-based HTS approach to identify small molecules that promote RPE generation and differentiation. Fluorescence-based HTS have been successfully developed in previous studies to identify compounds directing hPSC differentiation toward a number of lineages (23). However, fluorescence-based HTS are prone to false-positive hits. Sources of apparent activity in such assays include fluorescent compounds and impurities that may have been present in the chemical sample (24) as well as autofluorescence of apoptotic/dead cells. Perhaps reflecting these issues, when we performed preliminary screens using high-throughput flow cytometry to assess the expression of Tyr-GFP in 3D1 hiPSCs treated with small-molecule libraries, we were unable to validate any of the nine primary hits obtained out of almost 1,000 compounds tested. Compared with a fluorescence-based reporter system, qPCR directly measures native mRNA expression levels, which limits artifacts that can cause false discovery. The Tyr-GFP 3D1 hiPSC line then proved useful as an orthogonal assay to validate the primary hits from the qPCR screen. HT qPCR screening has been used successfully in a previous study with mouse embryonic stem cells (25); however, to the best of our knowledge, this is the first report of its application in the context of hPSC differentiation.

Following the above-described approach, we performed a pilot screen on a focused library of about 300 compounds. Despite the small size of this library, we identified CTM as a novel and potent promoter of RPE differentiation. Protocol optimization, with the addition of NIC to CTM, led to high yields of RPE cells after 1 mo of differentiation. A pure and functional RPE monolayer was obtained from a single whole-dish passage. This is an improvement over our previous approach based on low-efficiency spontaneous differentiation, for which two rounds of whole-dish passages were necessary to obtain a pure RPE population (16). Although high-efficiency RPE differentiation protocols have been reported, with yields as high as 60–97%, these high-efficiency approaches all involve the addition of growth factors that are suboptimal for clinical use (4, 6), or were tested on only a limited number of hESC lines (5). A small-molecule–only differentiation protocol has been reported, but yield and efficacy were limited, with 26% of colonies showing pigmented cells after 2 mo of differentiation (9). For a common hiPSC line, IMR904, our small-molecule–only approach is comparable to the best growth factor-based protocol published to date (6). To test the reproducibility of our protocol, the hiPSC line BC1 (26), which had never been cultured in our hands, was amplified by clonal propagation for two passages before beginning our RPE differentiation protocol. The fact that it generated RPE cells as efficiently as the other hPSC lines suggests that our protocol should be readily implementable in other laboratories and with a variety of ESCs and patient-derived iPSCs.

A striking observation made during CTM-directed, or CTM- and NIC-directed, differentiation into RPE was the absence of pigmented colonies when cultures were maintained in DM, whereas pigmented cells were readily observed in control and NIC-only conditions. In the light of these results, it is advisable to remain cautious with the use of pigmented colonies to determine RPE differentiation, as we have demonstrated that RPE-committed cells can exist in a culture without showing pigmentation or RPE-specific morphology. Interestingly, a simple medium change was sufficient to induce the acquisition of hallmarks of pigmentation and RPE morphology in cultures treated with small molecules, suggesting that CTM-committed RPE cells are not fully mature in DM and require RPEM to acquire their characteristic morphology.

Finally, we observed that during optic vesicle formation from hPSC (27), CTM treatment seems to increase the number of neuronal folds, suggesting a possibly wider role for this compound in retinal differentiation.

Materials and Methods

Differentiation and Culture of RPE from hPSC.

A detailed version of the optimized differentiation protocol and other methods can be found in Supporting Information. Briefly, hPSCs were seeded at 20,000 cells per cm² on Matrigel and grown in mTeSR1 in a 10% CO₂/5% O₂ incubator for 5 d. Afterward, they were transferred to a 5% CO₂/20% O₂ incubator and cultured for 5 additional days. At this point, culture medium was switched to DM, and in some experiments supplemented with 10 mM NIC and/or CTM (Sigma) at concentrations ranging from 1.25 to 80 nM. After 10–15 d, cells were maintained in DM only for 3 more weeks, or switched to RPEM. Differentiated cells were dissociated in Accumax (Sigma), plated at 250–300,000 cells per cm² and grown in RPEM. All surgical and animal care procedures were performed in compliance with the Animal Care and Use Committee of the Johns Hopkins University School of Medicine and the Association for Research in Vision and Ophthalmology statement on the Use of Animals.

SI Materials and Methods

Generation of Tyr-GFP 3D1 hiPSC.

A construct containing the tyrosinase enhancer and an hsp70 minimal promoter was described previously (13, 14). TFEC cDNA in the original construct was replaced with a GFP cDNA using an XhoI/SalI restriction sites. Transgenic mice (C57BL/6N) were generated using this construct. Two different transgenic mouse lines were obtained, both expressing GFP in the RPE. The tyr-enhancer GFP construct was cloned into a PacI site using primers (forward, 5′-ccagtttaattaaaggtgccagaacatttctctatc-3′, and reverse, 5′-ataacttaattaacgcctttgagtgagctgatacc-3′) in a FUGW lentiviral vector backbone that contains an RFP expressed by a constitutive Ubiquitin-c gene promoter (kind gift from Dr. Sally Temple, Neural Stem Cell Institute, Rensselaer, NY). VSV-G pseudotyped lentiviral particles prepared using NIH 293T cells were obtained from SBI Bioscience. The iPS cell line B42 iPS3 (15) was transduced at a concentration of 100 viral particles per cell. Positive colonies were selected and expanded after 72–96 h of transduction based on red fluorescence. Four different iPS cell colonies were selected and two showed reproducible green fluorescence when differentiated into RPE. GFP-positive and -negative cells were sorted using FACSAria sorter (BD Bioscience) and replated in Primaria plates (Corning).

Human Pluripotent Stem Cell Culture.

Human pluripotent stem cell (hPSC) lines were cultured as previously described (16). Briefly, the hiPSC lines IMR904 (WiCell), 3D1, and BC1 (27), and the hESC lines H1, H7, and H9 (WiCell) were maintained on growth factor-reduced Matrigel (BD Biosciences), in mTeSR1 medium (Stem Cell Technologies), in a 10% CO₂/5% O₂ incubator, and amplified by clonal propagation using ROCK pathway inhibitor blebbistatin (Sigma).

Embryoid bodies were generated following 48-h culture in 96-well ultra-low attachment plates (5,000 cells per well) and transferred onto 100-mm ultralow attachment dishes (both from Corning) in differentiation medium (DM). In some instances, 50 nM chetomin (CTM) was added to DM during the first 15 d of differentiation.

High-Throughput Quantitative PCR Screen.

Cell culture and small-molecule library treatment was done as follows: 3D1 hiPSCs were plated at 5,000 cells per well in Matrigel-coated 384-well cell culture plates (Falcon). They were maintained in mTeSR1, with addition of blebbistatin for the first 48 h. Daily medium exchange was performed using an automated washer/dispenser (EL406; Biotek). After 10 d, mTeSR1 was switched to DM consisting in DMEM/F12 (Invitrogen), 15% (vol/vol) KnockOut serum (Invitrogen), 2 mM glutamine (Invitrogen), 1× nonessential amino acids (Invitrogen), 0.1 mM β-mercaptoethanol (Sigma), and 1× antibiotic–antimycotic (Invitrogen). During the next 10 d, each quadrant of the 384-well plate received compounds from the StemSelect small-molecule library (Millipore) at four different concentrations ranging from 10 μM to 80 nM (or 1 μM to 8 nM according to the initial compound concentration) following a fivefold dilution scheme, using a syringe-based automated liquid handler (Tango; Matrix Technologies). Each concentration was assayed once. Due to the StemSelect library design, the first and last two columns of the 384-well plate received DMSO only. Afterward, cells were maintained for 5 additional days in DM only using the automated washer/dispenser for media changes.

Cells were lysed using the RealTime ready Cell Lysis kit, while RNA reverse-transcription (RT) was performed with the Transcriptor kit following the manufacturer’s instructions (both kits from Roche). For the last two columns of the 384-well plate, reverse-transcriptase was omitted (RT minus). qPCR reactions were performed with the Sso advanced qPCR mix (Bio-Rad) using an automated liquid handler for reagent and sample dispensing and run on a CFX 384 (Bio-Rad). Expression levels were normalized using the reference gene CREBBP. If the threshold cycle (Ct) of the reference gene for a given well was more than 2 SDs below the average Ct of the whole plate (without RT minus wells), it was omitted from the subsequent analysis to avoid biasing the normalization. Hit identification was based on the ΔΔCt method: compounds were selected for further analysis if the ΔΔCt for each of the three target genes, MITF, OTX2, and PMEL17, was greater than 3 SDs of the DMSO-treated controls. Gene-specific primers (Table S4) were obtained from published sequences or designed using Beacon Designer (PREMIER Biosoft).

Quantitative Real-Time PCR.

Total RNAs were extracted with Isolate II RNA minikit (Bioline) and reverse-transcribed (High Capacity cDNA kit; Applied Biosystem), before qPCRs were performed with EVA green qPCR mastermix (Abm). Quantitative PCR samples were run in biological triplicates and expression levels normalized using the geometric mean of three reference genes: FBXL12, SRP72, and CREBBP. Gene-specific primers can either be found in Table S1, or were part of a custom primer array (reference CAPH10651A; SABiosciences).

Flow Cytometry.

Immunostaining was performed using the IntraPrep Permeabilization kit (Beckman Coulter) according to the manufacturer’s instructions. Primary antibody concentration was 1 µg per 1 million cells for mouse anti-MITF (C5) (Thermo Fisher) and 0.046 µg per 1 million cells for mouse anti-PMEL17 (Abcam). Goat anti-mouse conjugated to Alexa 647 (Invitrogen) was used as a secondary antibody. A nonspecific, species-appropriate isotype control was included in all flow cytometry experiments and stained cells were analyzed using a C6 flow cytometer (Accuri). In cell-counting experiments, apoptotic and dead cells were excluded using Sytox red staining (Invitrogen) according to the manufacturer’s instruction. Analyses were performed on three biological repeats.

Immunostaining.

After fixation with 4% paraformaldehyde in PBS, cells were blocked and permeabilized for 30 min in 5% goat serum, 0.25% Triton X-100 in PBS, and then incubated for 1 h at room temperature with one of the following primary antibodies: mouse anti-MITF (C5) (1/100) (Thermo Fisher), mouse anti–ZO-1 (1/500) (Invitrogen), mouse anti-BEST1 (1/150) (Novus Biologicals), mouse anti-RPE65 (1/100) (Abcam), mouse anti-RLBP1 (1/100) (Abcam), or mouse anti-PMEL17 (1/500) (Abcam). Cells were then incubated for 1 h at room temperature with the corresponding secondary antibody conjugated to Alexa 488 or Alexa 647 (Invitrogen), and counterstained with Hoechst 33342 (Invitrogen). Antibodies conjugated to Alexa 647 were used in all secondary stainings with Tyr-GFP 3D1.

Mouse eyecups were fixed in 2% paraformaldehyde in 0.1 M phosphate buffer with 5% sucrose. They were then cryoprotected in 20% sucrose in 0.1 M phosphate buffer and rapidly frozen in OCT (Tissue-Tek). Because it was not possible to detect any GFP signal in otherwise RFP-positive cells after fixation and freezing, an anti-GFP antibody was used to evidence the presence of Tyr-GFP in eye sections. Eight micron sections were therefore stained with 1/400 rabbit anti-GFP antibody conjugated to Alexa 488 (Molecular Probes) in 2% goat serum, 0.1% Triton X-100 in PBS, or with 1/400 mouse anti-Rhodopsin (Sigma) followed by incubation with secondary antibodies conjugated to Alexa 647.

Pictures were taken with an epifluorescent microscope (Nikon Eclipse TE) or a confocal microscope (Zeiss LSM 710).

Phagocytosis of Photoreceptor Outer Segment.

Photoreceptor outer segments (POS) were a kind gift from Dr. Emeline Nandrot (Institut de la Vision, Paris, France). hPSC-RPE were challenged for 5 h with 3 × 10⁶ POS per cm² in RPEM. To assess the specificity of the POS phagocytosis, control wells were also incubated in the presence of 50 μg/mL anti-integrin α_Vβ₅ antibody (Abcam). The extent of POS phagocytosis was then determined by flow cytometry, as previously described (21). Briefly, hPSC-RPE cells were dissociated with TrypLE (Invitrogen), and live cells were stained with Vybrant DyeCycle Ruby Stain (Invitrogen) and immediately analyzed on a C6 flow cytometer. Cells with internalized POS were DyeCycle Ruby, GFP double positive.

For imaging of internalized POS, hPSC-RPE were vigorously washed with PBS, and then an equal volume of 0.4% trypan blue was added for 10 min to quench extracellular fluorescence, before proceeding with immunostaining.

VEGF and PEDF ELISA.

RPE cells derived from hPSC were plated at 250,000 cells per cm² and grown on Transwell membrane (Corning) coated with Matrigel. Cell culture supernatants from the hPSC-RPE apical and basal sides, that is, upper and lower compartments of the Transwell, respectively, were collected after 72 h of cell culture, and sent on dry ice to Elisatech. Measurements were done in biological triplicates. Standard ELISA was performed using antibody-coated plates, biotin-labeled antibodies, and streptavidin-HRP for the detection. Dilutions were 1/20 for VEGF and 1/2,000 for PEDF (both antibodies from R&D Systems).

Subretinal Transplantation of RPE Cells.

hiPSC 3D1-RPE cells were cultured for 20 d after their first passage (P1), until they formed a lightly pigmented monolayer suitable for transplantation. NOD-SCID mice (NOD.CB17-Prkdescid/J; 8–10 wk old; The Jackson Laboratory) were anesthetized by i.p. injection of a mixture of ketamine and xylazine. After making a hole at the pars plana with a 30-gauge sterile needle (BD), a micropipette glass needle (tip internal diameter, 25–30 μm) mounted on a Pico-Injector holder (PLI-100; Harvard Apparatus) was inserted through the hole into the vitreous, stopping at the surface of the retina. One microliter of hiPSC-RPE cell suspension (5 × 10⁴/μL) was delivered into the subretinal space of each eye. Fundus photographs of the retina were taken with a Micron III imager (Phoenix Research Lab, Inc.).

Statistical Analysis.

Analysis of variance (ANOVA), or T test, was performed accordingly with Prism 6.01 (GraphPad Software) on n = 3 biological repeats. We considered that statistical significance was reached for P < 0.05.

SI Detailed Protocol

Small-Molecule–Directed Differentiation of RPE from hPSC.

• i)Plate hPSC colonies at high density (20,000 cells per cm²) on Matrigel (BD). Culture cells in TesR1 medium in 10% CO₂, 5% O₂. At day 5 after seeding, transfer cells to 5% CO₂, 20% O₂, use 3 mL of medium per well (6wp), and wait until the cell layer becomes a monolayer (usually around day 7 or 8). After cells have formed a monolayer, keep in TesR1 for 3 additional days.
• ii)Switch culture medium to DM (Table S2). Use 3 mL of DM per well (6wp).
• iii)The next day, start feeding cells with DM plus 10 mM NIC plus 25–100 nM CTM (DM NC) for 2 wk. Change medium every day (3 mL per well of 6wp), adding fresh NIC and CTM. Note: we suggest to start with a 25–100 nM range for CTM, so that the most efficient concentration for a given hPSC line can be identified to direct RPE differentiation.
• iv)After 2 wk in DM NC, switch medium to RPE medium (Table S3). Use 2 mL per well (6wp). Feed cultures daily until day 30–35.
• v)Around day 35, remove RPE medium and add 1 mL of Accumax (Sigma). Incubate for 10–20 min at 37 °C, 5% CO₂ until most cells look rounded.
• vi)Vigorously dissociate cells in Accumax until most or all clumps have disappeared. Add 2 mL of RPE medium and transfer to 15-mL tube.
• vii)Centrifuge, 130 × g, for 5 min. Remove supernatant and resuspend cell solution in RPE medium.
• viii)Filter through a 40-μm nylon mesh (BD Falcon) and count the cells.
• ix)On dishes precoated with Matrigel, plate RPE cells at a density of 250–300,000 cells per cm².
• x)For routine passage of RPE layer, use Accumax at day 15–20 after P1, and seed on Matrigel at 100,000 cells per cm².

Material.

• Tissue culture plates (Falcon);

• Matrigel: growth factor-reduced Matrigel (BD; 356231);

• Accumax (Sigma; A7089);

• TeSR1 (Stem Cell Technologies; 05850).

For xeno-free RPE differentiation.

• Vitronectine peptide-acrylate surface (Corning; Synthemax II-SC 3535);

• KnockOut Serum Replacement CTS (Invitrogen; 12618);

• B27 CTS (Invitrogen; A14867).

Small molecules.

NIC (Sigma; N3376), 1 M stock (100×): to 50 mL of cell culture water, add 6.1 g of NIC, filter sterilize. Keep at 4 °C for up to 1 mo.

CTM (Sigma; C9623), 1 mM stock: dilute 1 mg of CTM into 1.4 mL of cell culture-tested DMSO (Sigma; D2650). Aliquot and store at −80 °C.

Note: After thawing, keep CTM aliquots at room temperature in dessicator for up to 1 wk. It is also possible to freeze/thaw CTM at least two or three times without affecting its efficiency in driving RPE differentiation.