Following photoreceptor injury, miR-18a is expressed in both the inner and outer nuclear layers, including in Müller glia and Müller glia-derived progenitors
To initially determine if miR-18a might play a role in photoreceptor regeneration, RT-qPCR and in situ hybridization were performed to determine the temporal and spatial expression of miR-18a following photoreceptor injury. Taqman qPCR showed that, compared with control uninjured retinas, miR-18a expression is significantly higher at 3 days, 5 days and 7 days post-retinal injury (dpi) (Fig. 1a). In situ hybridization using an LNA ribroprobe for mature miR-18a, combined with immunolabeling for green fluorescent protein (GFP) in Tg(gfap:egfp)mi2002 fish, showed that at in control uninjured retinas, there is only very slight expression of miR-18a (Fig. 1b) but by 24 hpi, around the time that Müller glia divide once to produce a neuronal progenitor [see 41], miR-18a is expressed in many cells throughout the inner nuclear layer (INL), including Müller glia (Fig. 1c). Then at 3 dpi, during the peak of progenitor proliferation, miR-18a remains strongly expressed in the INL and ONL, including in Müller glia and MG-derived progenitors (Fig. 1d). These progenitors are clearly visible as cells in the ONL that express GFP in their cytoplasm [described in 2], shown in Fig. 1d with black outlined arrows. Finally, by 7 dpi, when MG-derived progenitors are normally exiting the cell cycle, miR-18a expression is again similar to the expression pattern seen between 0 and 24 hpi (Fig. 1e, f). At all time points, all GFP-labeled cells appear to express miR-18a, the intensity of miR-18a labeling increases in these GFP-labeled cells at 1, 3 and 5 dpi, and miR-18a labeling is stronger in GFP-labeled cells than in surrounding cells without GFP labeling. Together, these results show that miR-18a expression is upregulated in the retina during the time periods when Müller glia and then MG-derived progenitors divide.
miR-18a regulates the timing, but not the extent, of photoreceptor regeneration
In the embryonic zebrafish retina, the absence of miR-18a results in accelerated photoreceptor differentiation but, by 6 days post-fertilization, does not affect the number of photoreceptors produced . Previous work did not compare photoreceptor numbers between WT and miR-18ami5012 fish in the adult retina. To determine this, in situ hybridization was used to label mature cones (arr3a) and rods (rho) in both WT and miR-18ami5012 retinas, and this showed that there are no differences in photoreceptor numbers (Fig. 2a, b, i, j).
To determine if miR-18a regulates the timing and/or extent of photoreceptor regeneration in adults, in situ hybridization was used to label mature cones (arr3a) and rods (rho) in injured retinas. First, mature photoreceptors were quantified in WT retinas at 3 dpi, to establish a baseline for comparison and determine the numbers of photoreceptors destroyed by the photolytic lesioning method. This determined that at 3 dpi, in the central retina just dorsal to the optic nerve, 96.4% of cones (Figure S1a, c, e) and 61.6% of rods (Figure S1b, d, f) had been destroyed. Next, quantitative comparisons were made between WT and miR-18ami5012 retinas at 7 dpi, when large numbers of newly differentiated photoreceptors can first be detected, 10 dpi, when most new photoreceptors have been normally regenerated, and 14 dpi, when photoreceptor regeneration is largely complete. At 7 dpi, miR-18ami5012 retinas have fewer mature cones than WT, but the number of mature rods does not differ (Fig. 2c, d, i, j). At 10 dpi, miR-18a retinas have fewer mature cones and rods than WT (Fig. 2e, f, i, j) but at 14 dpi, the number of mature photoreceptors does not differ (Fig. 2g-j). Together, these data show that in miR-18ami5012 retinas compared with WT, the same overall numbers of photoreceptors are present prior to injury and are regenerated by 14 dpi, but maturation of cone and rod photoreceptors is delayed.
During photoreceptor regeneration, miR-18a regulates proliferation among Müller glia-derived progenitors
The delay in photoreceptor maturation in the injured miR-18ami5012 retina indicates that photoreceptor regeneration is delayed, and this could be linked to the timing of proliferation among photoreceptor progenitors. To determine if miR-18a regulates cell proliferation among MG-derived progenitors, BrdU labeling and immunostaining were used to quantitatively compare the number of proliferating cells in WT and miR-18ami5012 retinas. In the injured WT retina, Müller glia divide between 1 and 2 dpi to produce neural progenitors, MG-derived progenitors then divide multiple times with proliferation peaking around 3 dpi, many MG-derived progenitors normally stop dividing between 4 and 5 dpi, and the first regenerated photoreceptors can be detected between 5 and 6 dpi [see 42]. By 7 and 10 dpi, proliferation is markedly reduced, and very few progenitors normally continue to proliferate. Compared with WT retinas, the total number of BrdU-labeled cells [i.e. cells that were actively dividing during the 24-hour period of BrdU exposure] in miR-18ami5012 retinas did not differ at 3 dpi (Fig. 3a, b), and there were no differences in numbers of BrdU-labeled cells in either the INL (WT 50.0 ± 7.2 SD, miR-18ami5012 40 ± 10.8 SD cells/0.3 mm, p = 0.25) or ONL (WT 17.7 ± 3.8 SD, miR-18ami5012 14.2 ± 3.2 SD cells/0.3 mm, p = 0.29). This indicates that the initial proliferative response is unaltered in miR-18ami5012 retinas. However, at 7 dpi, there were significantly more BrdU-labeled cells in the miR-18ami5012 retinas than in WT (Fig. 3c, d), and this difference was most pronounced in the ONL (WT 21.8 ± 5.9 SD, miR-18ami5012 51.4 ± 6.7 SD cells/0.3 mm, p = 0.0003). At 10 dpi, there were also significantly more BrdU-labeled cells in the miR-18ami5012 retinas (Fig. 3e, f), and significantly more cells in both the INL (WT 1.7 ± 0.8 SD, miR-18ami5012 14.3 ± 2.5 SD cells/0.3 mm, p = 0.001) and ONL (WT 11.0 ± 1.8 SD, miR-18ami5012 30.2 ± 4.9 SD cells/0.3 mm, p = 0.003).
To confirm the excess proliferation observed for miR-18ami5012 retinas, a second, independent approach was used to deplete miR-18a during photoreceptor regeneration. A miR-18a morpholino oligonucleotide, previously confirmed to effectively knock down miR-18a [38, 30], was used to knock down miR-18a in WT retinas during photoreceptor regeneration. Because miR-18a expression peaks at 3 dpi, for the present experiment, morpholinos were injected and electroporated at 2 dpi and eyes collected at 7 dpi. Compared with a standard control morpholino (SC MO), at 7 dpi, miR-18a knockdown resulted in more PCNA-immunolabeled progenitor cells [i.e proliferating cells that are in G1 to S-phase of the cell cycle] in the retina (Fig. 3g, h). The MO injection and electroporation technique is invasive and could possibly increase the overall proliferative response; PCNA might also label more cells than the 24-hour BrdU exposure; together, these differences likely explain the higher overall numbers of PCNA-labeled cells observed in MO injected compared with BrdU-labeled cells in 7 dpi uninjected retinas (compare with Fig. 3c, d). Together, along with miR-18ami5012 data, these results show that following photoreceptor injury, in the absence of miR-18a, the timing of the initial proliferative response is unchanged, but that progenitors continue to proliferate longer than in WT retinas.
Following photoreceptor injury in miR-18ami5012 retinas, excess MG-derived progenitors migrate to all retinal layers
In miR-18ami5012 retinas, MG-derived progenitors proliferate for a longer period of time, suggesting that excess MG-derived progenitors might be produced. However, since extra photoreceptors are not generated, the excess progenitors might either die or migrate to other retinal layers, possibly differentiating into other cell types. To investigate this, TUNEL was used to label apoptotic cells at 10 dpi, when the largest differences in numbers of photoreceptors and MG-derived progenitors were identified between WT and miR-18ami5012 retinas. This experiment showed that, at 10 dpi, there were no TUNEL-positive cells in either WT or miR-18ami5012 retinas (Fig. 4a), although some TUNEL-positive cells were observed in extraocular tissues in both WT and miR-18ami5012, thereby, demonstrating the sensitivity of the assay. The absence of TUNEL-positive cells in the retinas of WT and mutants indicates that excess progenitors in miR-18ami5012 retinas are not eliminated by cell death. To determine the potential fates of the excess MG-derived progenitors in miR-18ami5012 retinas, fish were exposed to 5 mM BrdU from 3 to 4 dpi during the peak of cell proliferation and sacrificed at 14 dpi when photoreceptor regeneration is complete. Immunolabeling for BrdU at 14 dpi shows that compared with WT, miR-18ami5012 retinas have more BrdU-labeled cells in both the INL and in/around the ganglion cell layer (GCL) (Fig. 4b-d), indicating that the excess progenitors produced at 3–4 dpi either remain in the INL or migrate to the inner retina, suggesting that they may adopt other cell fates, although this was not yet been evaluated.
During photoreceptor regeneration, miR-18a regulates the extent and duration of the inflammatory response
The effect of miR-18a on the cell cycle during photoreceptor regeneration is strikingly different from embryonic development, in which miR-18a regulates photoreceptor differentiation but not cell proliferation . This indicates that in the injured retina, miR-18a regulates pathways that are specific to the post-injury response. Silva et al.  showed that in injured mmp9 mutant retinas, there was increased inflammation resulting in excess proliferation among MG-derived progenitors. This finding helped lead to the rationale that increased or prolonged inflammation might cause the excess proliferation observed in miR-18ami5012 retinas. Indeed, the miRNA target database TargetscanFish (http://www.targetscan.org/fish_62) predicts that miR-18a interacts with mRNA for many molecules involved in inflammatory pathways. To determine if miR-18a regulates inflammation, RT-qPCR was used to compare the mRNA expression levels of key inflammatory molecules at 1, 3, 5 and 7 dpi in WT and miR-18ami5012 retinas. These data show that at 1 dpi, around the time that Müller glia normally begin to divide, expression of tnfα is higher in miR-18ami5012 retinas compared with WT (WT 2.31 ± 0.99 SD, miR-18ami5012 4.51 ± 1.14 SD, p = 0.033) (Fig. 5a). Then at 3 dpi, around the normal peak of MG-derived progenitor proliferation, il1β expression is higher in miR-18ami5012 retinas compared with WT (WT 0.97 ± 0.06 SD, miR-18ami5012 1.61 ± 0.21 SD, p = 0.007) (Fig. 5b). At 5 dpi, when many photoreceptor progenitors normally stop proliferating and begin to differentiate, expression of il6, il1β, and the cytokine regulator Nuclear Factor Kappa B 1 (nfkb1) are higher in miR-18ami5012 retinas compared with WT (il6 WT 0.85 ± 0.25 SD, miR-18ami5012 1.32 ± 0.26 SD, p = 0.042; il1b WT 0.97 ± 0.08 SD, miR-18ami5012 1.71 ± 0.47 SD, p = 0.028; nfkb1 WT 1.19 ± 0.17 SD, miR-18ami5012 1.68 ± 0.23 SD, p = 0.021) (Fig. 5c). Finally, at 7 dpi, when many photoreceptor progenitors have normally stopped proliferating and have fully differentiated, nfkb1 expression is still higher in miR-18ami5012 retinas (WT 1.03 ± 0.07 SD, miR-18ami5012 1.33 ± 0.22 SD, p = 0.044) (Fig. 5d). Taken together, these results indicate that following photoreceptor injury, in miR-18ami5012 retinas compared with WT, there is both a higher level of inflammatory pathway activity and a prolonged inflammatory response, indicating that miR-18a regulates the extent and duration of the inflammatory response following photoreceptor death.
miR-18a regulates the timing of the microglia response to photoreceptor injury
Microglia respond to retinal injury by undergoing a morphological change from ramified to amoeboid, migrating to the injury site and phagocytosing apoptotic cells, and releasing pro- and anti-inflammatory cytokines and other signaling molecules [reviewed in 24]. The function of miR-18a to regulate inflammation during photoreceptor regeneration, and the central role that microglia play in the injury-induced inflammatory response, suggest that miR-18a might regulate the microglia response to injury. To determine if miR-18a is expressed in microglia following photoreceptor injury, in situ hybridization for miR-18a was used in combination with immunolabeling with the 4c4 antibody that labels macrophages (e.g. microglia) in zebrafish. At 3 dpi, when microglia are active and miR-18a expression is the highest, mir-18a is expressed rarely in microglia (arrowhead, Fig. 6a), suggesting that miR-18a does not directly regulate microglia. Next, to determine if miR-18a regulates the response of microglia to photoreceptor injury, microglia were immunolabeled and counted in WT and miR-18ami5012 retinas at 3, 5, 7 and 10 dpi. The results showed no differences in the number of microglia at 3, 5 or 7 dpi (Fig. 6b-g). In contrast, at 10 dpi, miR-18ami5012 retinas had more microglia (Fig. 6h, i) than controls. These data indicate that miR-18a does not regulate the overall number of microglia responding to photoreceptor injury, but that it regulates the duration of the microglia response.
Supressing inflammation in miR-18a mutants rescues both the excess proliferation and delayed photoreceptor regeneration
The increased and prolonged expression of inflammatory genes and prolonged microglia response in the injured miR-18ami5012 retina led to the hypothesis that, in the absence of miR-18a, prolonged and excessive inflammation is causally related to the excess proliferation and delayed photoreceptor maturation observed in miR-18a mutants. To test this hypothesis, dexamethasone was used to suppress inflammation in WT and miR-18ami5012 fish from 2 to 6 dpi, the interval during which miR-18a expression is upregulated in injured retinas (see Fig. 1a, b) and when in miR-18ami5012 retinas inflammatory genes are expressed higher levels than in WT retinas (see Fig. 5). Fish were then exposed to 5 mM BrdU from 6 to 7 dpi, to label proliferating cells, and in situ hybridization was used to label mature rod or cone photoreceptors. Compared with controls, the dexamethasone treatment fully rescued the excess proliferation in miR-18ami5012 retinas, reducing the number of BrdU-labeled cells to that observed in controls (Fig. 7a, b). Dexamethasone treatment also reduced the number of BrdU-labeled cells in WT retinas by a smaller amount (Fig. 7a, b). Further, dexamethasone treatment rescued the delay in cone maturation and regeneration at 7 dpi, matching the number of arr3a-expressing mature cones with WT control levels (Fig. 7c, d), and also matching the number of Hoechst-labeled cone nuclei (WT control 76.7 ± 9.2 cells/0.3 mm, miR-18ami5012 control 57.2 ± 11.4 cells/0.3 mm, p = 0.021; WT control 76.7 ± 9.2 cells/0.3 mm, dexamethasone treated miR-18ami5012 76.3 ± 4.04 cells/0.3 mm, p = 0.451). Dexamethasone treatment had no effect on the number of arr3a-expressing cone cells in WT retinas (Fig. 7c, d). Together, these data indicate that in the injured retina, miR-18a regulates the proliferation of MG-derived progenitors and photoreceptor regeneration by regulating inflammation.