Human Wharton’s jelly mesenchymal stem cells protect axotomized rat retinal ganglion cells via secretion of anti-inflammatory and neurotrophic factors

Mesenchymal stem cell (MSC) transplantation is emerging as an ideal tool to restore the wounded central nervous system (CNS). MSCs isolated from extra-embryonic tissues have some advantages compared to MSCs derived from adult ones, such as an improved proliferative capacity, life span, differentiation potential and immunomodulatory properties. In addition, they are more immunoprivileged, reducing the probability of being rejected by the recipient. Umbilical cords (UCs) are a good source of MSCs because they are abundant, safe, non-invasively harvested after birth and, importantly, they are not encumbered with ethical problems. Here we show that the intravitreal transplant of Wharton´s jelly mesenchymal stem cells isolated from three different human UCs (hWJMSCs) delays axotomy-induced retinal ganglion cell (RGC) loss. In vivo, hWJMSCs secrete anti-inflammatory molecules and trophic factors, the latter alone may account for the elicited neuroprotection. Interestingly, this expression profile differs between naive and injured retinas, suggesting that the environment in which the hWJMSCs are modulates their secretome. Finally, even though the transplant itself is not toxic for RGCs, it is not innocuous as it triggers a transient but massive infiltration of Iba1+cells from the choroid to the retina that alters the retinal structure.

Immunological properties of hWJMSCs. First, we studied the capacity of hWJMSCs to suppress the proliferation of T cells stimulated by co-culture with allogeneic myeloid dendritic cells (mDCs), using mixed lymphocyte cultures. In Fig. 1A it is shown that hWJMSCs mediate a dose-dependent inhibitory effect on T cell proliferation compared to T cell proliferation in the absence of hWJMSCs. At the lowest ratio hWJMSCs: effector T cells (1:100), T cell proliferation was significantly reduced compared to the control co-culture in the absence of hWJMSCs (11% reduction, p < 0.05). Furthermore, the addition of a higher number of hWJMSCs to the co-culture led to a more marked inhibition of T cell proliferation, from 44% (ratio 1:25) to 96% (ratio 1:1) (p < 0.001).
Next, we analyzed the effect of hWJMSCs on the production of inflammatory cytokines by stimulated T cells. The co-culture of stimulator mDCs and responder T cells in the presence of hWJMSCs significantly decreased the secretion of IFN-γ (Fig. 1B) and inversely, increased the levels of the anti-inflammatory factors TGFβ, PGE 2 , and IDO ( Fig. 1C-E) in a dose-dependent manner which was significant even at the lowest hWJMSCs: effector T cells ratio tested here (1:100) for IFN-γ and TGFβ, and at 1:25 ratio for PGE 2 and IDO. For all of them the maximum modulation was reached at 1:1 ratio.
Anti-inflammatory PGE 2 , TGFβ, and IDO play a major role mediating the immunosuppressive effects of hWJMSCs on T cell proliferation. Next, we measured T cell proliferation after mDC stimulation using a 1:1 ratio of hWJMSCs:T cells and different specific inhibitors of the biosynthesis or signaling of TGFβ, PGE 2 and IDO (SB-431542, indomethacin (IDM) or 1-methyl-tryptophan (1-MT), respectively, Fig. 1F). The addition of SB-431542, IDM, and 1-MT in the absence of hWJMSCs did not affect the maximum proliferative capacity of effector T cells, nor the basal proliferation of stimulator mDCs, responder T cells, or hWJMSCs alone. Again, the proliferation of mDCs-stimulated T cells significantly decreased when hWJMSCs were present in the  Responder T cell proliferation was significantly inhibited in a dose-dependent manner when hWJMSC were added to the mixed lymphocyte cultures compared to T cell+mDCs without hWJMSC (ANOVA, *p < 0.05 or ***p < 0.001). (B-E) Column graphs showing the normalized production of IFN-γ, TGFβ, PGE 2 , and IDO vs. control (T cell+mDCs without hWJMSC) in supernatants from the mix lymphocyte cultures in the absence or presence of hWJMSC as before. Compared to control, the production of IFN-γ decreased significantly when hWJMSC were added, while the secretion of TGFβ, PGE 2 , and IDO increased significantly (*p < 0.05, **p < 0.01 or ***p < 0.001). In all cases, this modulation was dependent on the hWJMSC dose. (F) Column graph showing T cell proliferation after stimulation with allogeneic mDCs without or with hWJMSC (ratio 1:1), and the absence or presence of the inhibitors SB-431542, IDM or 1-MT. None of the inhibitors alone or in combination influenced the proliferation of T cells in the presence of mDCs. However, the inhibition of T cell proliferation by hWJMSC co-culture was overcome by each inhibitor individually and by their combination absence of any inhibitor (94% reduction). However, the addition of any of the inhibitors significantly recovered the proliferative response of T cells (p < 0.001), although this proliferation was still significantly lower than in the absence of hWJMSCs. Remarkably, the simultaneous treatment with the three inhibitors recovered almost entirely the proliferation rate of the T cells, suggesting that PGE 2 , TGFβ, and IDO play a major role mediating the immunosuppressive effects of hWJMSCs on T lymphocytes proliferation after allogeneic stimulation.

Intravitreally administered hWJMSCs are not toxic for retinal ganglion cells (RGCs).
To assess the possible toxicity of different doses of hWJMSCs injected intravitreally on RGCs, we injected either vehicle or 10,000, 20,000 or 40,000 cells into the vitreous chamber of otherwise intact retinas, and 7 days later we quantified the total number of RGCs (n = 4/dose). As shown in Fig. 2A, and compared to vehicle-injected retinas none of the hWJMSCs doses decreased significantly the population of RGCs.
hWJMSCs are neuroprotective for RGCs. Next, we wondered whether hWJMSCs exerted a neuroprotective effect on axotomized RGCs and if so, whether it was dose-dependent. The same doses as before were injected right after ONC (n = 4/dose), and retinas analyzed 7 days later when without treatment more than half of the RGCs have died 18,21,39,40 . In Fig. 2B, is observed that in all hWJMSCs-treated retinas there were significantly more surviving RGCs than in the vehicle-treated ones (p < 0.001). Furthermore, all doses seemed to be similarly neuroprotective. Thus, for subsequent experiments we chose the medium dose, 20,000 cells/injection. hWJMSC from different umbilical cords delay axotomy-induced RGC death. To investigate whether hWJMSCs isolated from different umbilical cords produced the same neuroprotection or if it was cordon-specific, as well as to know for how long this neuroprotection lasted, hWJMSCs isolated from 3 different umbilical cords were administered intravitreally right after ONC (n = 4-7 retinas/time point and umbilical cord) and the retinas analyzed 7, 14 or 30 days later. Quantitative data in Fig. 2C show that the number of RGCs in the transplanted retinas was similar among umbilical cords at all time points analyzed, and significantly higher than in vehicle-treated retinas at 7 and 14 days after the lesion (p < 0.001). Averaging the data of all the hWJMSCs-treated retinas the mean number of RGCs ± standard deviation at 7, 14 and 30 days was 61,773 ± 9371; 34,338 ± 4254 and 7291 ± 1379, respectively, while in vehicle-treated retinas was 34,549 ± 9168; 11,772 ± 3535 and 7810 ± 2840 at the same time points. In percent, this increased survival amounts to 179 ± 21 and 291 ± 36 at 7 and 14 days post-lesion, respectively (p < 0.001) and to 96 ± 19 at 30 days (Fig. 2D). Topographically (Fig. 3A), RGC neuroprotection by hWJMSCs occurrs across the retina, with no sectors of survival, whether the neuroprotection is mediated by cell-contact and/or by a paracrine mechanism.
Intravitreally transplanted hWJMSCs integrate in the ganglion cell layer and last up to 30 days. Seeing the pattern of RGC survival, it was important to know for how long hWJMSCs survived and/or whether they were able to integrate in the ganglion cell layer. Seven days after the injection, hWJMSCs were found in the central retina, spreading along the inner retinal arteries (Fig. 3B,C). By day 14, they had extended all across the retina, and by day 30 there were fewer although still abundant (Fig. 3B). It is worth noting that RGCs and hWJMSCs were photographed without changing the focus, indicating that hWJMSCs integrate in the ganglion cell layer (GCL, see below). hWJMSCs transiently over-express anti-inflammatory cytokines and trophic factors after intravitreal administration, preferentially when transplanted into injured retinas. First, we analyzed whether once transplanted, hWJMSCs expressed PGE 2 , TGFβ, and IDO as they did in the mixed cultures shown above. ELISA data in Fig. 4A show that the levels of both, PGE 2 and TGFβ, were highly increased in extracts from injured and transplanted retinas compared to intact, intact+hWJMSC and ONC+vehicle retinal extracts. Importantly, both PGE 2 and TGFβ levels were highest at 7 days post-transplantation (p < 0.001), decreasing gradually thereafter (p < 0.01). Of note, although both ELISAs are human specific, PGE 2 has been reported to work on rat's extracts 41 . Nevertheless the concentration of PGE 2 found in retinal extracts without hWJMSC was ~100 times lower than in transplanted ones. Finally, contrary to the in vitro results, human IDO was not detected in the transplanted retinas (not shown).
Then, we measured the levels of several trophic factors known to be neuroprotective for axotomized RGCs. NGF and BDNF levels were quantified by ELISA, and CNTF and VEGF by western blotting. In all cases we used human-specific antibodies. As shown in Fig. 4B, NGF and BDNF levels increase significantly in injured and transplanted retinas compared to the control extracts (p < 0.01). Again, their levels peak at 7 days, declining thereafter (p < 0.01, p < 0.05). Regarding CNTF and VEGF, their levels increase in hWJMSCs-transplanted retinas, both intact and injured, compared to non-transplanted ones, where they are almost undetectable. In intact+hWJMSC extracts, CNTF and VEGF levels were highest at 7 days decreasing thereafter, while in ONC+hWJMSC samples, they were highest at 14 days declining at 30 days.
( ΔΔΔ p < 0.001). Full recovery of T cell proliferation was only achieved when the three inhibitors were added at the same time (*p < 0.05, **p < 0.01, or ***p < 0.001, respectively compared to the proliferation of control (T cell+mDCs without hWJMSC). Results are shown as mean ± SD of three independent experiments performed in triplicate according to one-way ANOVA. Abbreviations: Stim mDCs: stimulator mature myeloid dendritic cells, Resp T cells: responder T cells, hWJMSC: human Wharton's jelly mesenchymal stem cells, IFN-γ: interferon gamma, TGFβ: transforming growth factor beta, PGE 2 : prostaglandin E2, IDO: indoleamine 2,3-dioxygenase, SB-431542: TGFβ inhibitor, IDM: indomethacin (PGE 2 inhibitor), 1-MT: 1-methyltryptophan (IDO inhibitor). n = 3 independent experiments/assay.   Bottom row, western blotting of CNTF and VEGF in the same extracts as above (hWJMSC extracts were not used in the western blots). The expression levels of these proteins were higher in injured retinas treated with hWJMSC compared to intact, intact+hWJMSC or ONC+vehicle. Note that all these assays were done with human-specific antibodies, although species cross-reactivity exists, mostly for PGE 2 41 . Extracts are pools from n = 4 retinas/time point and group. *p < 0.05; **p < 0.01; ***p < 0.001, ANOVA Tukey's post-hoc test. is observed that in ONC+hWJMSC retinas, but not in ONC+vehicle ones, the outer retina was folded and separated from the choroid (retinal detachment, arrows) breaking the retinal layered structure. Furthermore, these folds were observed in all the transplanted retinas, included intact ones (not shown), and were more frequent at 14 than at 30 days. In these sections, we immunodetected macrophages/microglial cells (Iba1) and hWJMSCs (h-mitochondria) together with a marker of cell proliferation (PCNA). In ONC+vehicle retinas, microglial cells were observed in the ganglion cell layer and both plexiform layers and some of them were PCNA + (Fig. 5B). In intact retinas, no dividing microglial cells were observed (not shown). By contrast, in the transplanted retinas (Fig. 5C) Iba1 + cells, displaying the typical reactive hypertrophic phenotype, were observed entering the retina from the choroid plexus and causing the retinal grooves. Also, these invading macrophages underwent division. In some cases, hWJMSCs were seen in the outer retina (Fig. 5C,b') but they were mostly located in the inner retina, as abovementioned (Fig. 5C,a') and around blood vessels, where microglial cells were also concentrated.

Discussion
In this work we have demonstrated that hWJMSCs isolated from three different human umbilical cords and transplanted into the rat vitreous body are not toxic for RGCs, and neuroprotect axotomized RGCs during the quick phase of death, up to 14 days. Thereafter RGC loss equals that observed in vehicle-treated retinas. Interestingly, the decline of neuroprotection coincides with fewer hWJMSCs present in the retina and a decrease of secreted human trophic factors whose levels, unexpectedly, differ when the transplanted retina is intact or injured. Finally, we have observed that hWJMSCs induce a massive migration of microglial/macrophages from the choroid to the inner retina that disrupts the retinal architecture hWJMSCs are non-immunogenic since a single injection of MHC-mismatched unactivated umbilical cord tissue-derived cells does not induce a detectable immune response. However, when hWJMSCs are injected into an inflamed region, repeatedly injected in the same area or stimulated with IFN-γ before injection, they are immunogenic 42,43 . The finding that MSC induce inhibition of T cells has clinical implications since allogeneic stem cell transplantation could modulate the immune response during stem cell-based regenerative therapies.
Several studies have shown that extra-embryonic MSC significantly express a higher level of immunomodulatory factors (e.g. TGFβ1, IDO, PGE 2 ) in comparison to adult bone marrow-derived MSCs 44 . This fact matches well with what we found after performing the in vitro characterization of the immunological properties of the hWJM-SCs. Here we show that hWJMSCs: i/do not induce proliferation of allogeneic T cells; ii/suppress the proliferation of T cells induced by allogeneic mDCs cells; iii/secrete soluble factors that mimic the immunosuppressive effects associated with the co-culture of the MSCs with the T cells (i.e. TGFβ, IDO, and PGE 2 ), and iv/inhibit the production of pro-inflammatory cytokines (e.g. IFN-γ) of T cells stimulated by an allogeneic stimuli. Importantly, our data are consistent with previously reported results that showed that hWJMSCs exhibit more potent immunomodulatory properties than adult bone marrow MSCs 7 .
In vivo, hWJMSCs integrate into the ganglion cell layer and survive up to 30 days. They first locate along the retinal arteries vessels, a behavior that has been also observed after intracerebral grafting 45 . By day 14 they cover the retina and it is at this time point when they reach their peak. During this period it is possible that they divide, though we did not observe tumors in agreement with previous works in the retina 46 or the brain 45,47 . Thirty days after the transplant their number has greatly decreased. Although the survival found here is in accordance with other works in brain 45,48 , it is important to have in mind that because this is a xenograft, hWJMSCs decrease at 30 days may be caused by tissue rejection and clearance 49,50 . In fact, the massive macrophage/microglia activation observed in the transplanted retinas advocates for this possibility.
In intact retinas, hWJMSC transplant does not cause RGC death. This is, to our knowledge, the first work assessing the toxicity of UC-derived cells on RGCs. This is an important assay firstly, because safety experiments should be done in animal models if these, or other cells, are going to be translated into the clinic. Secondly, because the neuroprotection elicited by UC-derived MSCs in models of RGC axonal damage is often very small (hWJMSCs, 22% higher than no treatment after ocular hypertension 35 ; UC-MSC, 28% after ONT 51 ) and this could be because the transplanted cells exert a direct or indirect (through glial cells) toxic effect that interferes with the neuroprotection.
In line with this, we show that while the graft does not kill RGCs, it is not completely harmless, as the retinal architecture is damaged by the infiltration and activation of Iba1 + cells. This response is not surprising since this is a xenotransplant albeit of cells with limited immunogenicity 3 and into an immuno-privileged environment. Nevertheless, this activation has been observed as well in allograft transplants into the mouse retina 52 . It is important to have in mind that the infiltration of Iba1 + cells was transient, decreasing at 30 days post-transplant when the number of hWJMSCs in the retina had also diminished. Importantly retinal architecture at 30 days was better than at earlier time points, indicating that once the hWJMSCs have disappeared, the system can be restored.
Iba1 + cells might be either microglial cells, the resident macrophages of the CNS, or invading macrophages which upon entering the CNS parenchyma are not distinguishable from microglial cells. Here, those Iba1 + cells located in the inner retina are most probably resident microglial cells, while those in the outer retina would be a mixture of both populations. Nonetheless, these cells have the morphological attributes of activation. To what extent and whether the excess of activated Iba1 + cells impairs retinal function or affects the observed neuroprotection in injured retinas we do not know. The role of microglial cells in neurodegeneration is controversial 53 , and it has been reported that MSC modulate them towards a restorative phenotype 5 . Although Giunti et al. 5 used allografts and the experiments here are xenografts, the same modulation may happen and thus, the excess of Iba1 + cells may not be harmful to neurons, being their role a cleansing one. Furthermore, part of the neuroprotection observed here may be due to them being in a restorative state. Although the infiltration of these Iba1 + cells might be caused by tissue rejection as above-mentioned, we cannot forget that MSC express macrophage attracting-chemokines 54 . Finally, we did not investigate astrocytes and Müller cells, but these glial cells are activated by topical treatment 41 , intravitreal injections 55 , RGC death 56 , and MSC grafts 52 .
We show that hWJMSCs isolated from three different umbilical cords elicit the same RGC neuroprotection. This is a promising result for translational medicine because due to the high genetic variability among human donors, each cord may have had different properties.
We observed RGC neuroprotection at 7 and 14 days (179 and 291% higher compared to untreated retinas, respectively). This is, to our knowledge, the higher neuroprotection reported in in vivo models of RGC axonal damage treated with MSC derived from the bone-marrow (160% higher than no treatment at 14 days after ONT 27 ), UC-blood (28% after ONT 51 ), or WJ (22% after ocular hypertension 35 ). However, these percentages may not be fully comparable because of the different axonal injuries, cellular doses used in each work, and RGC quantification methods (sampling vs. whole population). Nevertheless, there are two common denominators among these works and ours: RGC survival is transitory, and the transplanted cells secrete neuroprotective trophic factors. In fact, the higher RGC survival by hWJMSC transplant, may be explained alone by the higher levels of trophic factors found in the transplanted retinas [20][21][22][24][25][26] . Interestingly, our data show that the higher levels of BDNF, NGF, CNTF and VEGF are not better neuroprotecting RGCs than a single treatment with BDNF or CNTF alone [20][21][22] , suggesting that the effect of these factors is not additive, at least at the levels expressed in the transplanted retinas.
The loss of neuroprotection coincides not only with fewer hWJMSCs in the retina, and consequently with a lower level of released PGE 2 , TGFβ and neurotrophic factors, but also with the second, slow phase of axotomy-induced RGC death [16][17][18]40 . With the exception of knock-out strains lacking key pro-apoptotic genes 57 , most neuroprotective treatments after optic nerve axotomy last no longer than 14-21 days [20][21][22][24][25][26] , even when prolonged delivery of BDNF, the best neuroprotectant, is achieved 58 . It has been postulated that this happens because RGCs are no longer receptive to the treatment. We propose here that it could be also related to their mode of death. During the first two weeks most of the RGCs die by apoptosis 16,59 , and during this phase, microglial cells do not have a role in their death 40,60 . During the second slow phase, the mechanism of RGC death has not been described, but hypothetically it may relate to an altered microglial activation. PGE 2 increases in axotomized retinas with or without transplant as shown before 41 and here, and TGFβ increases in axotomized and transplanted retinas. Both soluble factors are immunomodulators with a described anti-inflammatory function [61][62][63] and PGE 2 may also be neuroprotectant [64][65][66] . When these two mediators decrease, the retinal inflammatory environment may change to neurotoxic, by altering the microglial phenotype and causing the second wave of RGC loss.
It has been suggested that depending on the pathophysiological microenvironment MSC release different cytokines and trophic factors 47,[67][68][69] . We show here that the secretion of TGFβ, PGE 2 , NGF, BDNF, VEGF and CNTF in intact and transplanted retinas is much lower than in axotomized and transplanted ones. Because we used human-specific ELISAs or antibodies (although some cross-reactivity may occur), these factors are mostly secreted by the hWJMSC. Thus, it becomes clear that injured neurons emit signals that directly or indirectly (through glial cells) stimulate hWJMSC to a given secretome. It is, therefore, tempting to speculate that hWJMSC produce bespoken secretomes to accommodate the needs of the injured tissue.
In summary, hWJMSCs transplanted into the axotomized rat retina are neuroprotective most possibly due, but not limited to, their release of anti-inflammatory and neurotrophic factors. We also demonstrate that this xenotransplant is not innocuous, as it attracts Iba + cells that invade the retina disrupting its architecture. Whether this infiltration has a role in the elicited neuroprotection is not known but it would be of interest to investigate the rescue of RGCs and the inflammatory response in immunosuppressed animals or in syngenic transplants, since allografts seem to cause the same response 52 . Finally, the secretome of hWJMSCs changes depending on the retinal state, widening the potential therapeutic use of these cells for a variety of neurodegenerative diseases. Animals undergoing surgery were anesthetized by intraperitoneal injection of a mixture of ketamine (60 mg/ kg; Ketolar, Pfizer, Alcobendas, Madrid, Spain) and xylazine (10 mg/kg; Rompun, Bayer, Kiel, Germany). Analgesia was provided by simultaneous administration of buprenorphine (0.1 mg/kg; Buprex, Buprenorphine 0.3 mg/mL; Schering-Plough, Madrid, Spain). During and after surgery, the eyes were covered with an ointment (Tobrex; Alcon S. A., Barcelona, Spain) to prevent corneal desiccation. Euthanasia was carried out by an intraperitoneal injection of an overdose of sodium pentobarbital (Dolethal, Vetoquinol; Especialidades Veterinarias, S.A., Alcobendas, Madrid, Spain).

Isolation of Wharton's jelly mesenchymal stem cells from human umbilical cords.
Three different human umbilical cords were collected from patients undergoing full-term pregnancy elective for caesarean section. Tissue collection was approved by the biobank of the Instituto Murciano de Investigación Biosanitaria-Virgen de la Arrixaca, (IMIB-Arrixaca Murcia, Spain). Donors provided an informed written consent which was approved by and followed the guidelines of the Ethics Committee the Hospital Clínico Universitario Virgen de la Arrixaca (Murcia, Spain).
Human Wharton's jelly mesenchymal stem cells (hWJMSC) were isolated by the explant method according to previously described protocols 70,71 . Briefly, each umbilical cord was sectioned into 3-5-cm long pieces, amnion was cut along the horizontal axis, and blood vessels with clots inside were removed. Then, cord pieces were placed with the inside facing to the bottom of a sterile 10-cm 2 petri dish. Explants were left to attach to the plate and complete culture medium (DMEM medium supplemented with 15% (v/v) fetal bovine serum (FBS), 1% (v/v) L-glutamine and 1% (v/v) penicillin/streptomycin, all from Life Technologies, Carlsbad, CA, USA) was added. The medium was replaced every three days and the hWJMSC attached to the plate were grown up to 80-90% confluence before doing serial passages.

Characterization of hWJMSC.
Human WJMSC were analyzed for the positive expression of the mesenchymal stem cell markers CD73, CD90 and CD105, and negative expression of the hematopoietic markers CD14, CD20, CD34 and CD45 by flow cytometry (all antibodies for Miltenyi Biotec, Bergisch Gladbach, Germany), following the guidelines of the International Society for Cellular Therapy to confirm mesenchymal phenotype 72,73 .
In addition, the expression of the co-stimulatory molecules CD80 and CD86 (Biolegend, San Diego, CA, USA) or major histocompatibility complex class II (HLA-DR) (Biolegend) was analyzed. Flow cytometry experiments were performed with a BD FACS Canto II flow cytometer (BD Biosciences, San Diego, CA, USA) and further analyzed with Kaluza analysis software (Beckman Coulter, Inc., Brea, CA, USA). Human WJMSC were tested as well for their capacity to differentiate to adipocytes, chondroblasts, and osteoblasts as previously described 74 Optic nerve crush (ONC) and intravitreal injection. The left optic nerve was crushed at 3 mm from the optic disc, following previously described methods 18,20 . In intact animals or immediately after the axotomy, cells or vehicle were injected into the vitreous chamber of the left eye using standard procedures in our group 20-22 . Tissue preparation and sectioning. Unless otherwise stated, all reagents were from Sigma-Aldrich Quimica S.A. Madrid, Spain.
For anatomical analyses, all animals were perfused transcardially with 0.9% saline solution followed by 4% paraformaldehyde in 0.1 M phosphate buffer.
For RGC quantification and identification of hWJMSC, retinas (n = 4-7/group and time point) were dissected as flattened whole-mounts as previously described 39 . Some eye cups (n = 3/group and time point) were cryoprotected in a series of sucrose gradients, embedded in optimal cutting temperature medium (OCT, Tissue-Tek, Sakura-Finetek, VWR, Barcelona, Spain) maintaining their orientation (the dorsal rectum muscle was kept as mark), frozen, and sectioned in a cryostat at 14 µm thickness.
Image acquisition and analysis. All images were acquired using an epifluorescence microscope (Axioscop 2 Plus; Zeiss Mikroskopie, Jena, Germany) equipped with a computer-driven motorized stage (ProScan H128 Series; Prior Scientific Instruments, Cambridge, UK) controlled by image analysis software (Image-Pro Plus, IPP 5.1 for Windows; Media Cybernetics, Silver Spring, MD, USA). Retinal photomontages (flat mounts or cross-sections) were reconstructed from 168 (12 × 14) individual 10x images by zig-zag tiling, as reported 39 . Individual images were acquired with a 20x objective.
The whole population of Brn3a + RGCs was quantified automatically and their distribution assessed by isodensity maps using previously reported methods 39 . Maps were plotted using SigmaPlot (SigmaPlot 9.0 for Windows; Systat Software, Inc., Richmond, CA, USA).
Immunomodulatory properties of hWJMSC. Mixed lymphocyte culture (MLC) experiments were used to evaluate hWJMSC ability to suppress T cell proliferation after stimulation with allogeneic mature dendritic cells. Firstly, monocytes were isolated from peripheral blood samples from healthy volunteers using the Human Monocyte Enrichment Cocktail (StemCell Technologies, Grenoble, France). Then, for immature myeloid dendritic cells (iDCs) differentiation, 50 ng/ml granulocyte-macrophage colony-stimulating factor (GM-CSF) (Sigma-Aldrich, St. Louis, MO, USA) and 25 ng/ml interleukin-4 (IL-4) (Invitrogen, Waltham, MA, USA) were added to the culture medium every three days for seven days. By day seven, more than 95% of cells were CD14 − CD11c + CD1a + BDCA-1 + (not shown). Finally, mature myeloid dendritic cells (mDCs) were obtained from iDCs by stimulation with 200 ng/ml lipopolysaccharide (LPS; Sigma-Aldrich) for additional 24 hours as reported 77 . On the other hand, human T cells were purified from peripheral blood mononuclear cells using Pan T cell isolation kit (Miltenyi Biotec). Then, the MLC was