Stem cell-mediated cell replacement therapy for retina has advanced rapidly in the past several years and has the possibility of becoming a treatment method for some retinal degeneration (RD) conditions [35
]. Apart from reproducibility of the data from different hESC lines, many issues require further evaluation; these include OLM barrier [23
], immunorejection of graft by a host [27
] (excellent discussion in [60
]), and the formation of glial scar containing extracellular matrix and Müller glia endfeet, preventing further cell integration [58
]. In addition, the host retinal niche and preservation of retinal architecture of the recipient seem to contribute to the complexity of any graft’s survival and functional integration [24
We considered it important to investigate two separate recurrent questions frequently reported in retinal cell transplantation papers: the survival of the retinal grafts in a non-immunocompatible recipient and the population of retinal layers with grafted hESC-RPCs. We approached this by first selecting normal (non-RD) young adult mouse eyes as recipients of hESC-RPC grafts to avoid the influence of a degenerating and rapidly changing neural niche on the survival of the graft [24
]. Such reports, although debated, suggest that an injured or degenerative neural environment might adversely affect the survival of human stem cell-derived grafts. We also chose not to apply immunosuppression, as retina is considered an immunoprivileged site due to the blood-retinal barrier (BRB). In addition, survival of xenogenic human grafts in retina has been reported [50
]. To account for the expected differences in graft survival, we correlated the survival of transplanted cells with the overall integrity of the RPE/choroid tissue, which comprises the BRB [65
]. Lastly, we compared the dynamics of cell integration into the host’s retina from the subretinal and epiretinal space to circumvent the OLM barrier. The advantage of such an approach is that in any given transplantation case the grafting niche remains the only difference, which may be informative for data interpretation. Overall, we find that both hESC lines UC06 and TE03 (cultured for 50+ passages) can differentiate to mature retinal phenotypes using the noggin/Dkk-1/IGF-1/bFGF/FGF9 protocol. After 3 months in a subretinal environment, transplanted cells demonstrated the ability to acquire mature PR-specific immunophenotypes (e.g., recoverin and rhodopsin staining) and no tumorigenicity was detected in all examined grafts. Importantly, we observed that the survival of xenogenic grafts with no immunosuppression correlates with the integrity of the RPE/choroid structure (BRB) but not the NR. Whenever the histology showed no damage to the RPE/choroid, the graft survived and thrived for up to 12 weeks with no immunosuppression and no signs of deterioration. The damage to the host’s NR alone and/or strong activation of GFAP by reactive Müller glia of the host ( and ) did not affect graft survival. In cases when the RPE/choroid showed signs of substantial damage by a blunt needle guided by the nano-injector, xenogenic hESC-RPC grafts did not survive, displayed lysed human cells, were filled with host’s Iba-1 [+] microglia, and were GFAP [+]. Therefore, we conclude that the xenogenic grafts may survive and thrive in the subretinal space when the BRB is intact. Consequently, systemic immunosuppression may not be necessary for graft survival when nonautologous PR progenitors are transplanted into retina.
Our results showed limited integration of subretinally grafted hESC-RPCs into the host’s retina and only in cases when the ONL had some structural damage. However, no HNu [+] cells (except one case) were found in INL or RGC layers, likely due to intact OLM present in the wild-type retina, consistent with other reports [19
]. In contrast, integration of hESC-RPCs into the host INL and especially the RGC layers was efficient from the epiretinal grafts, irrespective of whether the retina had any structural damage. Some HNu [+] cells were co-localized with host RGCs and also expressed RGC marker Tuj1 [66
]. qRT–PCR analysis of cells at the time of grafting showed that hESC-RPCs upregulated RGC markers (such as MATH5
) and the horizontal neuronal marker (e.g., PROX1
). Thus, hESCs could potentially generate RGCs and horizontal cells.
A limited number of human synaptophysin [+] boutons en passant could be detected in the INL and RGC layer, indicating initiation of synaptogenesis. Due to the lack of a barrier for cell penetration from the epiretinal side, such grafts may be used for long-term trophic support of degenerating retina [67
], including the trans-synaptic transport of neurotrophins [68
], as well as for potential RGC and INL cell-replacement strategies. Although the migration of cells into the ONL from subretinal grafts was clearly impeded, we suggest that in RD conditions this migration could be helped by a porous OLM [69
] as well as guided by tropism of grafted progenitor cells to the sites affected by degeneration [5
]. It is also possible that the maturation state of hESC-RPCs affects integration as some studies have reported integration of postmitotic progenitors and even mature PRs into normal retina [35
]. Although immunosuppression may not be crucial for xenogenic graft survival, it may be beneficial for retinal integration in a clinical setting when nonautologous (i.e., stem cell-bank-derived) hESC-RPCs are transplanted subretinally. For example, removal of glial barrier in GFAP−/−
and vimentin −/−
mice provided a permissive environment for retinal integration of transplanted neurons [31
]. Such a glial barrier, induced by the host, may be partially alleviated by immunosuppression and chondroitinase ABC [61
We also noted that the subretinal but not the epiretinal niche can provide further cues for hESC-RPC maturation to PRs, resulting in a sharp gain of mature PR marker recoverin, a neuronal calcium-binding protein found almost exclusively in PRs [16
]. However, the epiretinal grafts demonstrated no cell maturation and retained the original, mostly nestin [+] immunophenotype. This is consistent with a previous observation [49
] indicating that paracrine morphogens in the host retina and/or RPE can promote further maturation of hESC-RPCs.
As the cell population at the time of grafting showed almost 100% neuralization with noggin and over 67% of cells in grafts were positive for PR marker recoverin by 3 months, the overall efficiency of PR-fate specification from both hESC lines appears to be comparable to that reported [16
]. Only a small number of cells neuralized by noggin may be expected to remain non-neural after 4 weeks in culture [53
]. Since only neural rosettes were collected for further induction with Dkk-1 and IGF-1, the number of non-neural cells in such cultures should be minimal by day 50 (grafting), thus reducing tumorigenicity. There are several important distinctions resulting in faster derivation of recoverin/rhodopsin immunophenotypes in cultures reported earlier [16
]. These differences potentially originate from somewhat longer exposure to Dkk-1 and IGF-1, culturing on Matrigel rather than defined gelatin/laminin coating, and likely different culturing densities, which may profoundly influence the dynamics of neuronal cell fate acquisition and maturation [72
]. We also chose to maintain both bFGF and FGF9 in neural cultures, which earlier received anteriorizing Dkk-1 and IGF-1 induction, as both bFGF [75
] and FGF9 [76
] reportedly bias early retinal cells to an NR rather than an RPE cell fate.
The influence of FGF9 on the NR versus RPE cell fate is especially interesting as it is unexplored in retinal differentiation protocols. Fgf9
is expressed in the distal part of the developing optic vesicle in the mouse that is destined to become a NR and was reported to induce activation of Ras
by receptor tyrosine kinase in early optic neuroepithelium [76
]. Ectopic expression of Fgf9
in the proximal region of the optic vesicle destined to become RPE promotes conversion of the RPE cell fate to an NR cell fate in early retinal development by suppressing the expression of RPE marker Mitf
and induction of NR-specific markers Rx
]. As a result of such ectopic expression, a duplicated NR has been produced. Notably, the original NR and duplicated NR differentiated and laminated symmetrically but with a mirror-image polarity. The same study delineated the likely downstream target of FGF9 signaling, promoting the acquisition of the NR cell fate: the RAS-mediated RAF-MEK-mitogen-activated protein kinase pathway. Specifically, the transient expression of a constitutively active human Ras
oncogene by tyrosinase-related protein2 (TRP2) promoter in mouse transgenic embryos also converted the developing RPE to a second NR. Because the retinal development in both types of transgenic mice was overall normal, it was concluded by Zhao et al. [76
] that FGF9 signaling was needed to define the boundary between the retina and the RPE. Collectively, transient FGF9 signaling, likely through RAS signaling, was sufficient to promote NR cell fate at the expense of RPE, which was one of the goals of our differentiation protocol. Other factors, such as ectopic Pax6
expression or null mutation of Chx10,
are known to shift the cell fate in the developing retina from RPE to NR and vise versa, respectively. However, such signaling requires genetic manipulations in hESCs compared to easy delivery of FGF9 (and bFGF) morphogens during the differentiation protocol.
FGF9 belongs to a different subfamily of FGF factors compared to bFGF (FGF2) and can inhibit the canonical Wnt
pathway via upregulation of Dkk-1
, a canonical Wnt
antagonist, and regulate the transcription of Hedgehog targets patched homolog 1 (Ptch1
) and glioma-associated zinc finger 1 (Gli1
) independently of the Hedgehog ligand [77
]. Both effects may promote NR differentiation [16
]. Additional investigations are necessary to clearly delineate the role of FGF9 in NR differentiation.
In summary, we show that (i) xenogenic human hESC-RPC grafts from both hESC lines survive in the subretinal space without immunosuppression when little structural damage occurs to the RPE/choroid; (ii) gradual maturation of hESC-RPCs in subretinal but not epiretinal grafts occurs over a period of 3 months, indicating that the subretinal but not the epiretinal (vitreous) niche provides further differentiation cues for retinal cell fate maturation; (iii) substantial migration and integration of hESC-RPCs into the RGC and INL layers from epiretinal grafts occurs, even when the host retina lacked signs of damage. Our data provide new insights into differentiation and integration of grafted cells and may advance the protocols for cell therapies of retinal degenerative diseases.