Isolation and characterization of rat mesenchymal stem cells
To investigate the presence of characteristic markers of MSCs harvested from rat BM, MSCs were characterized by FCM analysis. The data showed that MSCs highly expressed positive MSC markers CD29, CD73, and CD105 from passage 1 until the last passages for use in culture. In contrast, negative MSC markers CD11b/c, CD45, CD68, and the MHC Class II Ia marker decreased with successive passages, until reaching 0% expression before retinal explant co-culture in accordance with the International Society for Cellular Therapy .
RGC degeneration in a retinal explant culture model
The best therapeutic window to assess neuroprotection was determined by evaluation of RGC degeneration following optic nerve axotomy in the retinal explant culture model, which can vary between laboratories, handlers, or culture conditions. Brn3a and RBPMS RGC markers were used for RGCs counting from ‘Day ex vivo’ (DEV) 0 to 7 (Fig.1A, B, C, and D) . Loss of RGCs occurred with both markers after 24 hours of culture but without significance. The number of RGCs was statistically different at DEV 3 and later compared to DEV 0 for Brn3a quantification (44% RGC death at DEV 3 from DEV 0, P<0.01) or DEV 5 for RBPMS quantification (36% RGC death compared DEV 0, P<0.01) (fig.1). From DEV 5 to DEV 7, a plateau of the RGC loss curves for Brn3a and RBPMS quantification was obtained. From these results, we determined an optimal therapeutic window from DEV 5 to 7 of culture for testing neuroprotective agents or MSCs.
Retinal explant responses to neuroprotective or excitotoxic stimuli
To evaluate the ability of the retinal explant model to respond to neuroprotective or excitotoxic agents, we exposed explants for 5 days to BDNF or NMDA [31,32]. At DEV 5 with Brn3a and RBPMS markers, quantification of RGCs on whole-mount retinas confirmed the neuroprotective effect of BDNF, with a significant increase in the survival of Brn3a+ (110±5.1 vs 186±24.8/306µm2, P<0.01) and RBPMS+ (130±5 vs 190±23/306µm2, P<0.05) RGC populations (fig.2, A and B). NMDA exposure at 50 µM caused significant RGC loss compared to controls at DEV 5 with both markers Brn3a (109±13 vs 1.8±1.3/306 µm2, P<0.0001) and RBPMS (135±10.2 vs 49±10.4/306 µm2, P<0.01) (fig.2, C and D).
Coculture of MSCs with retinal explants confers RGC neuroprotection
In accordance with the publications of Johnson et al., we investigated whether the presence of MSCs could limit or prevent RGC loss in retinal explants after an even longer duration, namely at DEV 7 [28,33]. At DEV 7, the number of RGCs stained with Brn3a and NeuN was significantly higher in cryosections of retinal explants cocultured with 1.104 MSCs compared to the control group (18.7±6.7/mm vs 8.5±1.8/mm, P<0.05 and respectively). RBPMS quantification did not show statistically significant differences at DEV 7 between control and MSC co-culture groups. No significant difference in RGC numbers was found between DEV 0 and DEV 7 with 104 MSCs for RBPMS, Brn3a, or NeuN staining (fig 3).
MSCs decrease gliosis in retinal explants
To determine the inflammatory response following MSC implantation, we analyzed GFAP immunostaining and GFAP mRNA expression in retinal explants at DEV 0 and DEV 7. GFAP immunoreactivity in astrocytes and Müller cells was upregulated in all retinal layers at DEV 7 in the control group compared to DEV 0, where GFAP was limited to the nerve fiber layer (NFL) and outer plexiform layer (OPL). In contrast, at DEV 7 in the MSC coculture group, GFAP immunostaining was limited to the NFL and to a lesser extent to the OPL (fig 4). GFAP mRNA was significantly higher at DEV 7 in the control and MSC coculture groups (38.5-fold, P<0.0001 and 11.4-fold, P<0.0001 respectively) compared to DEV 0. However, GFAP mRNA was significantly lower at DEV 7 in the MSC coculture group compared to the control group (P<0.0001) (fig 5B). These data demonstrate that MSC coculture conferred limited glial activation in retinal explants at DEV 7 compared to the control group. However, GFAP activation was limited but still robust in the NFL and OPL layer in the MSC coculture group, proving a significant, localized reactive gliosis following MSC implantation.
MSCs reduce microglial activation
In order to determine the microglial cell response following coculture with MSCs, we analyzed Iba1 and CD68 immunostaining and ITGAM and CD68 mRNA expression in retinal explants at DEV 0 and DEV 7. At DEV 0, Iba1+ microglial cells were located in the inner plexiform layer (IPL) and ganglion cell layer (GCL), and no CD68+ cells were found in retinal layers. At DEV 7 in the control group, Iba1+ and CD68+ cells were found in all retinal layers, contrary to the MSC group, where Iba1+ and CD68+ cells were limited to the inner retinal layers (GCL, IPL and inner nuclear layer (INL)). Moreover, at DEV 7 in the MSC group, microglial cells were present mainly at the interface between the GCL and MSCs (fig 5 A, B). ITGAM and CD68 mRNA fold inductions were significantly lower in the MSC group compared to the control group (respectively 17.3 vs 33.1 and 83.4 vs 169, p<0.0001) (fig 5C, D). These data demonstrate that microglial cells were distributed differently throughout the retina at DEV 7 in the control and MSC groups, with a limited but strong distribution of microglial cells to the internal retinal layers in the MSC group. In the control group at DEV 7, Iba1+ microglial cells migrated and proliferated toward the outers layers of the retina. Furthermore, microglial activation and proliferation were higher at DEV 7 in the control group compared to the MSC coculture group. However in the MSC coculture group this microglial activation was concentrated at the RGC/MSC interface.
MSCs have an immunomodulatory effect on retinal explants
To investigate whether the presence of MSCs in the retinal explant co-culture model could exert immunomodulatory properties and influence on microglial phenotypes, we analyzed microglial polarization markers through the mRNA expression of type M1 classically activated, namely TNFα, IL1β and IL6, and M2 alternatively activated, namely Arginase 1, IL10, CD163 and TNFAIP6 [15,20]. Our data showed that mRNA expression of M1 phenotype markers TNFα, and IL1β levels were significantly lower at DEV 7 in the MSC group compared to the control group (p<0.05 and p<0.001 respectively) (fig 6 A). There was a significant increase in the level of IL-6 mRNA expression at DEV 7 but no significant difference between the control group and the MSC co-culture group. The mRNA expression levels of the markers of M2-polarized microglia, Arginase 1, IL10, and TNFAIP6 were significantly lower in the MSC coculture group at DEV 7 compared to the control group (p<0.0001, p<0.01, and p<0.0001, respectively) (fig 6 B). Only mRNA expression of the M2 marker CD163 was significantly lower in the control group compared to DEV 0, with no significant difference between the MSC coculture group and the control group at DEV 7. However, there were no significant differences between the DEV 7 MSC group and DEV 0 for IL10, CD163 or TNFAIP6 mRNA expression levels.
MSCs affect retinal architecture and induce an ERM-like phenotype
Before the co-culture, DiO labeling of MSCs allowed tracking of the MSCs to determine the ability of MSCs to graft into the retinal explant. At DEV 7, a few MSCs were found in the retinal explants at the GCL level. Most MSCs did not penetrate the retina and remained at the surface of the explant, probably where the 2 µl of MSC suspension was deposited (fig 3B).
Explant “swelling” was observed in all explants in the MSC co-cultured group, with increased explant thickness and retinal folding, compared to DEV 0 and to DEV 7 controls. MSCs also induced the appearance of an epiretinal membrane at the surface of the retinal explants. In order to investigate whether this distortion was associated with an epiretinal membrane phenotype, we used an anti-fibronectin antibody, since this protein is known to be upregulated in idiopathic epiretinal membranes . Figure 7 shows the increase in fibronectin labelling at the surface of the explant co-cultured with MSCs. This fibronectin expression was higher on the apical side of the explant, at the contact area between the MSCs and GCL. Likewise, H&E staining was performed to assess retinal micro-architectural organization and to quantify explant swelling in presence of MSCs. The laminar structure of the retinal explants at DEV 7 in the control group showed a folded appearance to the outer segments (OS) and outer nuclear layer (ONL) compared to DEV 0 (figure 8A and 8B). Explants in the co-culture group exhibited a thicker internal retinal layer. Figure 8C shows a significant increase in retinal explant thickness, which was found in all the explants in the MSC co-cultured group compared to the DEV 7 control and DEV 0 groups (405±30 µm vs 218±20 µm and 220±28 µm respectively, P<0.0001).