Generation of immature retinal neurons from proliferating cells in the pars plana after retinal histogenesis in mice with retinal degeneration

Animals

All experimental procedures were performed in accordance with the guidelines of the Institute for Laboratory Animal Research (Guide for the Care and Use of Laboratory Animals) and Nagoya University School of Medicine for the use of animals. C57BL/6J mice were used as wild-type controls and for the N-methyl-N-nitrosourea (MNU; Clea, Tokyo, Japan) injection study. C3H/HeJ mice (rd1 mice; Clea, Tokyo, Japan) were used as a murine model of inherited retinal degeneration [20]. All mice were fed on a basal diet (CE-2; Clea, Tokyo, Japan) and water. They were kept on a 12-h light-dark cycle in a standard cage placed in a temperature- and humidity-controlled environment. The number of mice used in each quantitative experiment is indicated in Table 1. A total estimate of at least 200 mice was used throughout the study (128 mice were used for quantification purpose).

BrdU labeling

To label cells in the S-phase of the cell cycle, we injected mice intraperitoneally with bromodeoxyuridine (BrdU; Sigma, St. Louis, MO). A dose of BrdU used for each experiment is indicated in the figure legend. These mice were euthanized either by inhalation of 100% CO₂ or by cervical dislocation after 2 h to 24 days to collect the eyes.

Immunohistochemistry

Immunohistochemical analyses were performed as described previously [16,21]. Sections and flat-mount specimens were stained with primary antibodies for BrdU (1:1,000; Oxford Biotechnology, Oxford, UK), Pax6 (1:200; marker for retinal progenitors; Developmental Studies Hybridoma Bank, Iowa City, IA), rhodopsin (1:1,000; marker for rods; Chemicon, Temecula, CA), recoverin (1:1,000; marker for rods and cones and subset of cone bipolar cells; Chemicon), synaptophysin (1:1,000; marker for synaptic vesicles; Sigma) and K_i-67 (1:1,000; cell proliferation marker [22]; BD PharMingen; San Diego, CA) followed by combinations of 1:1,000 Alexa 405-, 1:1,000 488-, and 568-conjugated secondary antibodies (1:1,000; Molecular Probes, Eugene, OR), PNA (1:100; marker for the inner/outer segments and pedicles of cones [23,24]; Molecular Probes), and diamidino-2-phenylindole (1:1,000; DAPI; Molecular Probes). After the staining procedure, mild overnight pressure was applied to the flat-mount specimens to remove the undesired retinal/choroidal/scleral folds. As a result, the thickness of the retina scanned for histological analyses would not necessary correspond to that in living eyes.

Identification of cells in the pars plana within flat-mounts

With regard to the images of flat-mount specimens acquired by confocal microscopy (Eclipse C-1/ Nikon, Tokyo, Japan) cells in the pars plana were presented in two different ways: thin scan and thick scan (Figure 1). The former is a scan with its focus on the ciliary epithelium and the retinal surface (Figure 1B). When evaluating photoreceptor markers, we visualized only cells in the pars plana in a circumferential and sparse distribution. The latter was presented as an image of a single scan or as a merged image of 2 to 4 scans (the detail is indicated in the corresponding legend) that included the cells of ciliary epithelium as well as those of the photoreceptor layer (Figure 1C). Only some images were presented with thin scan (indicated in the legend) while all the other flat-mount data were presented with thick scan. Using this method, we found cells in the pars plana showed a distribution similar to that of the thin scan. Meanwhile, cells in the retina typically showed much denser staining demarcating the cilioretinal border. In addition, the following features distinguished the ciliary body from the retina. First, the inner surface of the pars plana was flat, showing a contour similar to the retina, while in the deeper layers it often had discernable folds that appeared to be the continuum of the pars plicata. Second, an intervening space or gap was frequently detected between immunopositive cells at the retinal margin and pars plana.

Almost all figure panels that were obtained with thick scans are merged images of multiple scans as described in the legend. The interval of each optical section comprising the merged image was set at approximately 75% of the thickness of the scan itself. This is to merely show the retina as a “plane” with dense photoreceptors. A single thick scan was applied to all images obtained for statistical purpose (this is enough to just visualize the cilioretinal border and the cells in the pars plana). A thick scan was adopted for statistical purposes because a thin scan had a higher chance of underestimating the cells in the pars plana, which might result in larger inter-sample variations. In addition, it was more convenient to visualize the cilioretinal border by thick scan in the eyes with a thin degenerated retina. In wild-type mice or developing rd1 mice that have thick retinas, we often found it easier to isolate cells in the pars plana with a thin scan.

MNU injection

C57BL/6J mice were injected intraperitoneally with 60 mg/kg MNU (Sigma). Seven days after MNU administration, severe photoreceptor degeneration was observed in MNU-injected animals [25].

Quantification and statistical analysis

The numbers of cells positive for recoverin, rhodopsin, PNA, BrdU, and K_i-67 per 320 µm width of the pars plana were quantified from at least 3 independent images obtained per animal. The number of animal used for each experiment is outlined in Table 1. For quantification of cells in the pars plana, images were obtained with a single scan with its focus on the ciliary epithelium. Cells that encompassed both the retina and pars plana were excluded from the analyses. Student’s unpaired t-test was used for the statistical analysis of the differences. p<0.05 was considered significant.

Pars plana of rd1 mice during ocular development

To avoid ambiguous interpretation, we defined retinal progenitors as proliferating cells with the potential to generate retinal neurons. Post-mitotic cells that had some morphological features of neurons in the ciliary body were defined as immature retinal neurons/retinal precursors (termed retinal precursors in the previous report) [16] in this study. First, we examined the hypothesis that retinal progenitors or immature retinal neurons/retinal precursors are also present in the developing ciliary epithelium of rd1 mice. In rd1 mice, photoreceptors begin to degenerate rapidly at postnatal day 12 (P12), shortly after histogenesis of the retina is complete [20]. Mitotic cells were labeled by intraperitoneal injection of BrdU into P6 rd1 mice 2 h before enucleation. Histological sections showed that cells positive for both BrdU and Pax6 were most abundant in the peripheral retina and ciliary epithelium, many of which formed the neuroblast layer (Figure 2A) similar to the observation in wild-type mice [16]. In flat-mounts, the BrdU-positive cells comprising neuroblast layer in the retina extended into areas of the pars plana where recoverin-positive or rhodopsin-positive cells were also found (Figure 2B-D). Further analyses of the pars plana revealed that most of the recoverin-positive cells were of cone photoreceptor lineage positive for PNA (Figure 2E,F). Such cells had various degrees of morphological features of immature cone photoreceptors or its precursors in the developing retina. These include a large budding at the end of one of its processes called the pedicles, which are sometimes as large as the cell body, and a conical process at the end of the other process corresponding to the inner and outer segments. Unfortunately, calculating the proportion of rod versus cone lineage cells in the pars plana was difficult because of the relatively weak signal and the high background of PNA staining, particularly during retinal development [23,26]. Nonetheless, these finding were consistent with the results of wild-type mice that showed the development of the pars plana and peripheral retina were closely related to each other until the latest stage of postnatal ocular development [16].

Pars plana of adult rd1 mice

In wild-type mice, the existence of immature retinal neurons/retinal precursors generated in the ciliary epithelium of the pars plana is mostly restricted to a short period during retinal development and is rare at P12 [16] when gross retinal histogenesis is complete [27-29]. To determine whether immature photoreceptors/photoreceptor precursors were present after retinal histogenesis in the eyes of rd1 mice, we examined the ciliary epithelium and peripheral retina of adult rd1 mice. Although BrdU-positive cells were generally increased in rd1 mice compared to the wild-type controls, such cells mostly disappeared from the retina by P12 and continued to decline in number also in the ciliary epithelium (data not shown). However, many cells of rod and cone photoreceptor lineage, positive for rhodopsin and PNA respectively, were found in the pars plana of rd1 mice at P30 (Figures 3A,B). The former comprised approximately 28% and the latter approximately 61% of recoverin-positive cells, while the identity of an estimated 11% remained unclear (Figure 3C). We found no recoverin-positive cells that were positive for both PNA and rhodopsin. Some of these immature retinal neurons/retinal precursors bridged both the ciliary epithelium and retina (Figure 3D). Rhodopsin-positive cells persisted in the pars plana up to P90 (data not shown) and recoverin-positive cells up to at least P120 (Figure 3E). Consequently, in comparison to the number of recoverin-positive cells in the pars plana of wild-type mice at P60, those of rd1 mice at P60 and P120 were 31.2 and 7.6 fold higher, respectively (Figure 3F,G). Many of these cells had detectable synaptophysin-positive vesicles in their synaptic terminals (Figures 3H-I), suggesting that they may be capable of generating and releasing neurotransmitters. Interestingly, we also found a few recoverin-positive cells in the pars plicata of adult rd1 mice (data not shown). Similar cells were found in the developing but not in the adult pars plicata of wild-type mice (data not shown).

In the pars plana and areas of the peripheral retina of P60 rd1 mice, the recoverin-positive cells showed common morphological characteristics that were distinct from the posterior parts of the retina (Figure 4). These cells had short as well as randomly oriented processes in the pars plana and peripheral retina, while those in the remaining posterior part of the retina had longer processes and were aligned radially in an organized manner. However, the location of the border of such a transition in cellular alignment and morphology varied considerably between mice and even within the same retina. Similar regional differences in cell morphology were also observed in rhodopsin- and PNA-positive cells at P30 (Figure 5).

Generation of recoverin-positive cells after retinal histogenesis

We identified many immature neurons/neural precursors mostly of photoreceptor lineage in the pars plana of adult as well as young rd1 mice. However, whether these cells in the adult pars plana were produced as a result of neurogenesis, i.e., differentiation from their proliferating progenitors, after retinal histogenesis or were merely persistence of or differentiation from post-mitotic cells that were produced during ocular development remains unknown. To clarify this issue, we performed pulse-chase assays to define the temporal origins of these cells. BrdU used to label dividing cells may have negative influence on the neural differentiation [30]. To avoid such an undesired effect, a minimal dosage of BrdU injection that allowed a reliable detection of the agent was sought. We found that a BrdU dosage of 50 mg/kg was sufficient for this study after trying various injected doses of BrdU (12.5, 25, 50, 100, and 150 mg/kg) and histologically evaluating its incorporation (data not shown). Using this dosage, dividing cells were labeled with BrdU at P6, P12, P18, and P24, and the fates of these cells were examined later at P30. Anti-recoverin antibody that mostly stained cells of photoreceptor lineage provided the best staining quality with least background was used for the neural marker. Numerous cells were labeled with BrdU injection at P6, i.e., during retinal histogenesis. When these mice were later examined at P30, some recoverin-positive cells with morphological features of retinal neurons that incorporated BrdU were found in the ciliary epithelium (Figure 6A). These cells were evaluated thoroughly by continuous confocal scanning microscopy to confirm that the BrdU-positive nucleus resided within the recoverin-positive cells. Among the recoverin-positive cells in the pars plana, approximately 7.5% were BrdU-positive and were generated after P6 (Figure 6E). By P12, histogenesis of the retina was complete [20] as evidenced by examination of histological sections (Figure 6C) and the presence of rare BrdU-positive cells only in the peripheral retina (Figure 6B). This is also consistent with our previous quantitative evaluation of BrdU positive cells in retinal sections from developing mice [16]. Analysis of mice injected with BrdU at P12 indicated that an estimated 2.9% of recoverin-positive cells in the pars plana were generated after histogenesis, i.e., P12 (Figures 6B,E). The percentage declined further in the eyes with BrdU labeling at P18 (Figures 6D,E). The evidence of neurogenesis was extremely rare when mice were injected with BrdU at P24. Similar experiments using wild-type mice indicated that the evidence of neurogenesis in the P30 pars plana observed when BrdU was injected at P6 became undetectable with injection at P12 (Figure 6E).

Next, we injected BrdU at P6 and analyzed the number of recoverin-positive cells with and without BrdU incorporation in the pars plana of rd1 mice at P9, P12, P20, and P30. The evidence of neurogenesis was detectable by P9, 3 days after BrdU injection. The rate of cells positive for BrdU among recoverin-positive cells in the pars plana became greatest at P20, 14 days after BrdU injection, but showed a modest decrease at P30 (Figure 6F). The number of newly generated immature neurons/neural precursors after histogenesis may actually be greater than detected in the present study, as the presence of unlabeled proliferating cells that escaped the single injection of BrdU was confirmed (Figure 7); these unlabeled cells might also give rise to immature retinal neurons/retinal precursors. Interestingly, such cells negative for BrdU and positive for K_i-67 were mostly observed in the pars plana and occasionally formed a cluster that resembled the circumferential distribution of retinal precursors (Figure 7B).

Neurogenesis in mice with acquired retinal degeneration

In rd1 mice with inherited retinal degeneration, many recoverin-positive cells were found in the adult pars plana, a proportion of which were generated after retinal histogenesis. No evidence of neurogenesis was observed after P30 in rd1 mice (data not shown) or P12 in wild-type mice. Therefore, the important question of whether the pars plana of uninjured adult wild-type mice has the potential to newly produce immature retinal neurons/retinal precursors remains unsolved. Previously, we reported a marked increase in number of recoverin-positive cells in the pars plana of adult mice when acquired retinal degeneration was induced by MNU [16]. However, whether any of these cells were generated from proliferating retinal progenitors remains unknown. To resolve this issue, we injected MNU into wild-type mice at P30 and labeled proliferating cells with BrdU at P32. Mice eyes were subsequently collected for histological analyses. When the eyes were analyzed at P37, the number of BrdU-positive cells in the pars plana was increased by an estimated 4.3-fold in the MNU-treated animals as compared to untreated controls (Figures 8A,B,F). The number of recoverin-positive cells was also increased by approximately 17.1 fold in the MNU-treated mice (Figure 8G). Similar to the case of rd1 mice, most of recoverin-positive cells in the pars plana were cells of cone photoreceptor lineage positive for PNA (roughly 77%) and a minor proportion were cells of rod photoreceptor lineage positive for rhodopsin (approximately 15%; Figure 8D,H). This was in contrast with the peripheral retina, which had a much larger proportion of rod photoreceptors. Therefore, the cilioretinal border could often be recognized as an abrupt alteration of cone-rod ratio (Figure 8D). However, most importantly, a small number of recoverin-positive cells in the pars plana was labeled with BrdU (approximately 1.52%); these cells were newly generated at or after P32. Such cells were observed infrequently also in the peripheral retina (Figure 8C). Some appeared to be cells of cone photoreceptor lineage (data not shown). No evidence of neurogenesis was detected in control mice (Figure 8I) at or after P32.