We recently described new methods to efficiently generate human iPSCs with only one or two of the conventionally used four transcription factors and treatment with small molecules [13
]. These naïve iPSCs expressed key pluripotency markers and formed teratomas in vivo. Whether these iPSCs possess full differentiation capabilities that match conventionally derived iPSCs has not been evaluated. In this study, we assessed the differentiation potential of one- and two-factor-derived iPSCs in comparison with conventional four-factor-derived iPSCs.
We subjected one conventionally reprogrammed iPSC clone (4F-iPSCs), four different clonal lines of iPSCs generated by using two reprogramming factors (2F-iPSCs), and four clonal lines of one-factor-derived iPSCs (1F-iPSCs) to either published spontaneous or directed differentiation protocols [20
]. (The reagents used for reprogramming are listed in .) Directed differentiation resulted in more rapid and efficient generation of pigmented cells (6 weeks; technique depicted in supplemental online Fig. 1). Pigmentation in embryonic RPE cells is observed in a similar time frame in vivo [34
]. Although differentiating iPSC cultures demonstrated a decrease in expression of the pluripotency marker Nanog, expression of the melanogenesis marker tyrosinase and the RPE terminal differentiation markers bestrophin, CRALBP, and RPE65 increased (A; supplemental online Fig. 2). These results are similar to those observed when RPE cells are differentiated from conventional iPSCs [22
]. Immunohistochemical analysis revealed a close colocalization of terminal differentiation marker expression and cellular pigment formation (B; supplemental online Fig. 3). When these pigmented cell clusters were isolated and replated, they expanded as homogeneous monolayers of cells with characteristic RPE polygonal morphology, dense pigmentation, and ZO-1-positive intercellular tight junctions (C).
Figure 1. (A): mRNA expression was analyzed in 4F-, 2F-, and 1F-iPSCs at the beginning and end of 8 weeks of directed differentiation. Expression of the pluripotency marker Nanog decreased with differentiation but remained detectable in 4F-iPSCs that were transfected (more ...)
To use iPSC-derived RPE as autologous grafts to replace aged and dysfunctional RPE cells in diseases such as AMD, these cells need to perform essential RPE functions such as apical-to-basolateral fluid transport, barrier formation, and phagocytosis of POSs. The RPE monolayer in vivo forms a tight barrier between the retina on its apical side and the choroidal circulation on its basolateral side (outer blood-retinal barrier); RPE cells constantly pump fluid from the apical (retinal) to the basolateral (choroidal) side. For functional RPE cells in vitro, these two characteristics have been shown to result in the accumulation of fluid under the cell monolayer, leading to the formation of fluid-filled cellular domes [35
]. We consistently observed this phenomenon in all cultures of iPSC-RPE cells, demonstrating cellular barrier formation and directional apical-to-basolateral fluid transport (D). Another RPE-specific function in vivo is the phagocytosis and degradation of shed POSs [36
]. When we challenged iPSC-RPE cells with isolated POSs, we indeed detected pronounced phagocytosis activity (D).
To more closely determine how iPS-RPE and hfRPE functionality correlates, we compared the metabolomes of these cell types. The metabolomes of keratinocytes and fibroblasts, naïve stem cells, and differentiated RPE were analyzed with a novel and high-resolution approach using LC coupled to ESI Q-TOF MS. Our MS-based platform enabled us to detect thousands of metabolite features, defined as molecular entities with a unique mass/charge and retention time value. When directly compared, the percentage of dysregulated features between populations can be used to correlatively determine how similarly they function biochemically. The largest disparities were observed between naïve 4F- and 1F-iPSCs and the fetal fibroblasts and neonatal keratinocytes lines from which they were reprogrammed. Remarkably, few differences were observed when comparing the naïve 4F- and 1F-iPSCs with hESCs (and between themselves), even though they originated from different cell types at different developmental stages. Importantly, few metabolic differences (0.5%) were also observed between 1F-iPS-RPE and hfRPE (A). 4F-iPS-RPE and hfRPE also differed only slightly (5.0%), but this difference was 10-fold higher than that observed between 1F-iPS-RPE and hfRPE. Furthermore, the differences between 1F- and 4F-iPS-RPE were low (7.9%), but higher than when either sample was compared with hfRPE.
Figure 2. (A): The percentage of dysregulated metabolic features (fold change ≥2 with high statistical significance, p ≤ 0.001) was compared between hfRPE, 1F-iPS-RPE (1F-iPS-2), and 4F-iPS-RPE. Nominal disparities were observed between hfRPE and (more ...)
Furthermore, antibody-based in-cell Western proteomic analyses revealed that 1F-iPS-RPE and hfRPE share far more similarities than 1F-iPS-RPE and aged primary hRPE cells do. Protein expression in hfRPE, 1F-iPS-RPE, primary adult hRPE cells (from 62- and 88-year-old donors), and the ARPE-19 cultures was examined using 28 known protein markers of RPE development and terminal differentiation. One-way analysis of variance (ANOVA) tests revealed, with few exceptions, that expression levels of the markers chosen were not statistically different. Exceptions included markers of advanced differentiation RPE65 (p = .02), tyrosinase (p = .04), and connexin 43 (Cx43; p = .002). Closer analyses revealed that the expression levels of the markers examined in 1F-iPS-RPE consistently correlated best with hfRPE (B). Interestingly, when 1F-iPS-RPE and hfRPE were directly compared, significant differences were observed in the expression of terminal differentiation markers, including bestrophin (p = .02), RPE65 (p = .05), CRALBP (p = .01), peropsin (p = .003), and Cx43 (p = .02) (C).
ELISAs were used to quantify the amount of secreted PEDF and VEGF from 1F-iPS-RPE and hfRPE because these factors have been implicated as modulators of choroidal neovascularization in AMD [37
]. Although no significant difference was observed in the amount of VEGF secreted by 1F-iPS-RPE and hfRPE (p
= .68), the amount of PEDF secreted by 1F-iPS-RPE was substantially higher (5.6-fold; p
= .02) than that secreted by hfRPE (D). This result is significant since increased PEDF may reflect an advanced differentiation status [38
]. Additionally, the expression of bestrophin in 1F-iPS-RPE was localized to the basolateral membranes but diffusely expressed in hfRPE, and the 1F-iPS-RPE synthesized much more melanin pigment than hfRPE did (E).
We next examined whether 1F-iPS-RPE can prevent photoreceptor death in vivo using a rat model of RPE-mediated retinal degeneration, the Royal College of Surgeons (RCS) rat. Retinal degeneration in this strain is complete and rapid since the RPE cells cannot phagocytose POSs. We injected 1F-iPS-RPE into the subretinal space of albino RCS rat retinas using a novel technique after transiently labeling the cells with fluorescent CellTracker and immediately imaging the retina in vivo to determine the efficacy of the injection. To facilitate in vivo monitoring we used cSLO, fundus autofluorescence, and optical coherence tomography imaging. Transient CellTracker labeling avoids the risk of altering gene expression or cellular functions, problems sometimes observed after introducing transgenic markers, such as green fluorescent protein (supplemental online Fig. 5A). To demonstrate that 1F-iPS-RPE cells are functional in vivo after transplantation in RCS rats, we monitored autofluorescence as a biomarker of active RPE phagocytosis and visual cycling. Normally, the fundus of the retina exhibits uniform fluorescence, but in human cases of sensory retina and RPE atrophy, profoundly reduced fluorescence is observed [40
]. At the margins of the degenerating regions, however, elevated autofluorescence is often detected [40
]; these signals may correlate with early-stage RPE dysfunction (impaired phagocytosis) and buildup of degenerating photoreceptor outer segments in a debris layer similar to that observed in the RCS rat [42
]. In fact, a thick autofluorescent debris layer is observed in human cases of retinal degeneration linked to a mutation in the MerTK gene [43
], the same gene mutated as in RCS rats.
We determined after careful analyses of cryosectioned RCS rat retinas that detectable autofluorescent patterns differed in regions where 1F-iPS-RPE cells were transplanted. In uninjected animals, the autofluorescent layer was more thick and diffuse, occupying the entire subretinal space and debris layer (A, arrows). In regions implanted with 1F-iPS-RPE, however, the autofluorescence was detected in a restricted and punctate pattern in the RPE cell layer (B, arrows; D), demonstrating very efficient delivery and incorporation of the 1F-iPS-RPE cells. Remarkably, the implanted 1F-iPS-RPE cells could also be visualized 1.4 years after implantation (C). To examine the ultrastructural features of the implanted cells and their spatial orientation in relation to Bruch's membrane and the outer retina, we used electron microscopy. Examinations performed 7 months after transplantation revealed that the 1F-iPS-RPE (easily identifiable as pigmented cells in albino animals) were perfectly polarized in normal proximity to Bruch's membrane and correctly extended apical processes toward the photoreceptor cells (D). Similar findings were observed 1.4 years after implantation (E), and we discovered that at this age the implanted cells were also easily identifiable (and functional) since they had phagocytosed large amounts of debris from the subretinal space (F, arrows). On the basis of our analyses, the majority of the transplanted cells, despite being injected as a bolus, incorporated as a monolayer into the RPE cell layer. We did observe occasional instances where the RPE cells existed in multilayered aggregates (supplemental online Fig. 8). However, it is very encouraging that the 1F-iPS-RPE appeared, overall, to integrate very efficiently into the correct cell layer after transplantation after such a simple experimental delivery.
Figure 3. (A): Autofluorescence in uninjected Royal College of Surgeons (RCS) rat eyes was detected in a broad diffuse region correlating with the subretinal space. (B): In contralateral eyes injected with 1F-iPS-RPE (1F-iPS-2), the autofluorescence pattern was (more ...)
Figure 4. 1F-iPS-RPE (clone 1F-iPS-2) phagocytosed photoreceptor outer segment debris and reduced the levels of A2E accumulation in Royal College of Surgeons (RCS) rat eyes. (A): A2E accumulation in RCS rat as detected by liquid chromatography-mass spectrometry. (more ...)
We speculated that the large amount of debris phagocytosed by the implanted cells contributed to the bright autofluorescence we observed previously. The dominant fundus fluorophore is RPE lipofuscin [44
], and the major RPE lipofuscin fluorophore is A2E [45
], a nondegradable pyridinium bisretinoid probably generated in photoreceptor outer segments that accumulates in the RPE after phagocytosis [46
]. Bisretinoid biosynthesis is significantly enhanced in the RCS rat because of improper clearance of outer segments [42
]. We confirmed rapid accumulation of A2E in RCS rat retinas using mass spectrometry over several developmental time points concurrent with photoreceptors degradation (A). The mass spectrometry signal intensity of A2E increased 3.9-fold between 3 and 10 weeks (289,115–1,122,397) but was significantly reduced 8 (66.9%) and 10 (71.2%) weeks after 1F-iPS-RPE injection (11 and 13 weeks old, respectively) (B; mass spectrometry data shown in supplemental online Fig. 6). These data suggest that transplanted 1F-iPS-RPE cells phagocytose much of the debris before A2E can be formed. Autofluorescent lipofuscin can be directly visualized using fluorescent microscopy on frozen sections of RCS rat retinas. In uninjected eyes, nominal autofluorescence and recoverin immunoreactivity was observed in RPE cells (C, bracket). Conversely, in 1F-iPS-RPE injected eyes, intense autofluorescence was observed in the RPE cell layer (D, bracket). These data strongly suggest that the transplanted 1F-iPS-RPE cells are functional and are phagocytosing outer segments from the debris layer. In fact, using electron microscopy, we detected large amounts of debris highly resembling photoreceptor outer segments only in implanted 1F-iPS-RPE cells 7 months after transplantation (E, blue). Other subcellular materials were detected only in implanted 1F-iPS-RPE, including lipofuscin (F, green) and extracellular deposits (G, red), known byproducts of photoreceptor outer segment phagocytosis 15 months after transplantation. At this stage, the host RPE cells (which are unable to phagocytose photoreceptor outer segments) were much smaller and did not contain debris (H), whereas the implanted 1F-iPS-RPE (I) were much larger. Additionally, since retinal macrophages are known to contribute to clearance of the debris layer in RCS rats [47
], we looked for cells with macrophage features and morphologies in the sections. These were observed very sporadically in the subretinal space of the prepared sections (J), suggesting that the implanted 1F-iPS-RPE cells were primarily responsible for clearance of the debris layer.
The accumulation of lipofuscin in 1F-iPS-RPE cells may also indicate that the visual cycle is functional in transplanted RPE and host photoreceptor units. In mice with defective visual cycles (RPE65
mutants), lipofuscin levels are reduced >90% [50
] since the retinals from which A2E are generated are not manufactured [51
]. Although retinal levels dropped rapidly in RCS rat eyes at time points consistent with photoreceptor degradation, in 1F-iPS-RPE transplanted eyes, retinal levels were significantly elevated 8 and 10 weeks postinjection (1.74× and 1.69×, respectively; supplemental online Fig. 7). Consequently, detection of retinal and lipofuscin in transplanted RPE cells likely indicates that the transplanted RPE cells integrated with the host photoreceptors and proceeded to phagocytose outer segments and recycle retinal in the visual cycle, demonstrating functional rescue.
We performed histological examinations 6 weeks after 1F-iPS-RPE injection to monitor the extent of anatomical rescue. These examinations revealed that a significantly higher number of rows of photoreceptor cell nuclei were observed in the ONL in RCS rat retinas injected with 1F-iPS-RPE as compared with untreated control animals (8.5-fold; p
< .001; n
= 4; A), demonstrating that 1F-iPS-RPE protected photoreceptors from degeneration in the RCS line. High-resolution novel in vivo imaging techniques were used for longitudinal tracking of the status of the retinal degeneration in transplanted animals. Retinal degeneration was quantified by measuring the thickness of the neurosensory retina. Reduced thickness in degenerating retinas corresponds to the loss of rows of photoreceptors in the ONL [52
]. Significantly increased retinal thicknesses of 1F-iPS-RPE injected RCS rat retinas were observed at 4, 5, 6, 12, and 18 weeks after injection compared with age-matched untreated or sham-injected controls (B; supplemental online Fig. 9). By 6 months after injection, differences in retinal thickness in animals injected with iPS-RPE compared with controls were no longer statistically significant. These findings are comparable to those reported by other groups using hES- or four-factor iPSC-derived RPE [20
]. We also used optical coherence tomography imaging techniques to obtain spatial maps of retinal thickness around the injection area after 1F-iPS-RPE injections. These novel analyses revealed that the thickest regions of the retina correlated with the regions where transplanted pigmented RPE cells were present (C). Since the thickness is measured between the inner surface of the retina and photoreceptor outer segments, the increases in thickness values were not due to increased layers of transplanted RPE cells. Importantly, tumor formation was not observed in any rat eye successfully injected with 1F-iPS-RPE at any time point up to 1.4 years after injection (n
Figure 5. (A): Immunohistochemistry analyses were performed 6 weeks after transplantation of 1F-iPS-RPE (clone 1F-iPSC-2) in the subretinal space of RCS rats. Recoverin staining labeled photoreceptors (marked with a red line) and some bipolar cells. Note that at (more ...)
We performed novel image-guided fERG to determine whether the photoreceptors rescued by iPS-RPE injection in the RCS rats are functionally active. Conventional attempts to directly measure electrical activity of the retina using whole-field ERG have been largely unsuccessful since RPE incorporate into such a small region of the retina. Using fERG, we could focus on a discrete region of the retina (0.7%) and adjust the diameter and intensity of the light beam to deliver a stimulus directly onto the area of interest, and a corneal electrode on the front of the lens measured the elicited electrical response. Fundus images of the regions from which the recordings were measured could also be captured (D). When light was injected into regions into which 1F-iPS-RPE had incorporated, we detected significantly more electrical activity than when it was injected into regions in which no cells have been transplanted (D, middle and right panels), from eyes injected with only PBS (D), and from the contralateral uninjected eyes (data not shown). Collectively, therefore, we demonstrated significant anatomical and functional rescue of photoreceptors in RCS rat retinas in which 1F-RPE were implanted.