Dicer Plays Essential Roles for Retinal Development by Regulation of Survival and Differentiation

Abstract

Purpose.: Much attention has been paid to the roles of microRNA in developmental and biological processes. Dicer plays essential roles in cell survival and proliferation in various organs. We examined the role of Dicer in retinal development using retina-specific conditional knockout of Dicer in mice.

Methods.: Dkk3-Cre expressed the Cre gene in retinal progenitor cells from an early embryonic stage. The authors analyzed Dkk-Cre/Dicer-flox (Dicer-CKO) mice for their survival, proliferation, and differentiation. To analyze the role of Dicer in later stages of retinal development, a Cre expression plasmid was introduced into the neonatal retina by electroporation, and retinal differentiation was examined.

Results.: Dicer-CKO mice were born at the numbers we expected, based on Mendelian genetics, but their eyes never opened. Massive death of retinal progenitor cells occurred during embryogenesis, resulting in microphthalmia, and most retinal cells had disappeared by postnatal day 14. In vitro reaggregation culture of Dicer-CKO retinal cells showed that cell death and the suppression of proliferation by Dicer inactivation occurred in a cell-autonomous manner. Cell differentiation markers were expressed in the Dicer-CKO retina; however, these cells localized abnormally, and the inner plexiform layer was absent, suggesting that cell migration and morphologic differentiation, especially process extension, were perturbed. Forced neonatal expression of Cre induced apoptosis and affected the expression of differentiation markers.

Conclusions.: Taken together, these results show that Dicer is essential during early retinal development.

The vertebrate neural retina is organized into a laminar structure comprising six types of neuron and glial cell, including Müller glia and microglia. During retinogenesis, these various cell types are derived from a common population of multipotent retinal progenitor cells in a relatively fixed chronological sequence. ¹ Intrinsic cues and extrinsic signals play critical roles in defining the types of cells generated from common retinal progenitor cells, ^2,3 and various molecules are involved in this process. The expression of these genes in retinal development is regulated at various levels; microRNA (miRNA) is one such regulator. 

MicroRNAs are small, noncoding RNAs that are encoded in the genomes of all metazoans. They are essential in the proliferation and differentiation of various tissues, including stem cells. ^{4 –6} The roles of miRNAs in retina have been reported, ^7,8 and a recent study showed the presence of light-regulated retinal miRNAs. ⁹ In addition to suppressing the function of certain miRNAs, we can remove all miRNAs by deleting enzymes that are essential in their biosynthesis. DGCR8 is required for the production of all canonical miRNAs, and Dicer is an enzyme that cleaves double-stranded RNA into miRNA. ¹⁰ The removal of DGCR8 or Dicer results in a defective cell cycle and silencing of the self-renewal program of embryonic stem cells. ^11,12 Because the complete loss of Dicer in mice results in early embryonic death, ¹³ mice with a conditional allele of the Dicer gene have been produced, ¹⁴ enabling the study of the roles of miRNA in organogenesis. Subsequently, essential roles of Dicer in organogenesis have been revealed by studying mice with various tissue-specific expression of Cre. ^14,15  

In neurons, the deletion of Dicer by α-calmodulin kinase II Cre results in an array of phenotypes, including microcephaly and reduced dendritic branch elaboration, suggesting that the loss of Dicer disrupts cellular and tissue morphogenesis in the cortex and hippocampus. ¹⁶ The first study examining the roles of Dicer in the retina using Chx10-Cre transgenes, expected to express Cre in retinal progenitor cells, showed that although Chx10 was expressed in the embryonic retina, morphologic defects were observed at postnatal day (P) 16 with the formation of photoreceptor rosettes, accompanied by abnormal electroretinogram responses. ¹⁷ However, the relatively mild phenotype of the mice is surmised to be caused by mosaic expression of Cre in the Chx10-Cre transgenic retina ¹⁷ because subsequent work by George and Reh using αPax6-Cre-retina specific Dicer conditional knockout showed that Dicer is required in retinal development. ¹⁸ In this work, we evaluated the effects of deleting Dicer using Dkk3-Cre mice, which expressed the Cre gene in retinal progenitor cells from an early embryonic stage. ¹⁹  

Materials and Methods

Mice and Reagents

EGFP transgenic mice, which express the EGFP gene ubiquitously through the CAG promoter, were kindly provided by Masaru Okabe (Osaka University). ^20,21 Dicer^flox mice ¹⁴ were kindly provided by Michael McManus (University of California, San Francisco), and Dkk3-Cre BAC transgenic mice were as previously described. ¹⁹ Dicer^flox/flox (Dicer-fl/fl) or Dicer^flox/wild were used as controls for experiments shown in Figures 1 and 2. ICR mice were obtained from Japan SLC Co. All animal experiments were approved by the Animal Care Committee of the Institute of Medical Science, University of Tokyo, and were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Figure 1.

 

Dicer^lox/lox:Dkk3^+/cre (Dicer-CKO) mice were born but had microphthalmia. (A–C) Images of Dicer-CKO (A) and littermate (B) mice at 2 weeks of age. Images of embryos of Dicer-CKO and littermate at E16.5 (C). (D) Immunostaining of Cre expression of Dicer-CKO retina at E17.5 was performed using frozen sections. (E–N) Structure of the eye of Dicer-CKO and littermate at E16.5 (E, F), E18.5 (G, H), P1 (I, J), P5 (K, L), and P14 (M, N) stages. Head (E–J) or whole eyes (K–N) were frozen sectioned, and nuclei were visualized by staining of DAPI. L, lens. Scale bar, 200 μm unless indicated. L, lens; NBL, neuroblastic layer. (O) Thickness of retina of Dicer-CKO or fl/fl control mice. Measurements were made under a microscope, and the thickness of retina at central region was examined at indicated stages. Average of three independent retinas with SD is shown. **P <0.01, Student's t-test.

Figure 2.

 

Detailed examination of retinal development by immunostaining of retinas of Dicer-CKO mice revealed perturbation of retinal development. (A–D) Retinas from Dicer-CKO or littermate control mice at E16 (A–C) or indicated stages (D) were frozen sectioned. Immunostaining using indicated antibodies was performed, and nuclei were visualized by staining of DAPI. (D) Lower panels are enlarged images of the white squared regions in the upper panels. L, lens. Scale bars: 100 μm (A, D); 200 μm (B, C).

DNA Construction

pxCANCre containing CAG promoter followed by Cre genes was the gift of Izumu Saito (University of Tokyo). CAG-Cre-IRES-EGFP was constructed by the ligation of fragments of CAG-Cre (SalI-BglII), IRES-EGFP (BglII-NotI), and the vector portion from pEGFP2 (SalI-NotI). 

Immunostaining

Immunostaining of sectioned or dissociated retina was performed as described previously. ²² OCT compound (Tissue-Tek)–embedded samples were sectioned with 10-μm thickness by a cryostat (CM3050S; Leica, Wetzlar, Germany). Primary antibodies used were the following: mouse monoclonal antibodies against βIII tubulin (Covance, Princeton, NJ), photoreceptor-specific nuclear receptor ([PNR], ppmx), rhodopsin (Rho4D2, kindly donated by Robert S. Molday, University of British Columbia), glutamine synthetase (GS; Chemicon, Temecula, CA), HuC/D (Molecular Probes, Eugene, OR), Ki67 (BD Biosciences, Franklin Lakes, NJ), Cre (Millipore, Billerica, MA), rabbit polyclonal antibody against GFP (Clontech, Palo Alto, CA), Pax6 (Covance), calbindin (Millipore), active-caspase3 (Promega, Madison, WI), and goat polyclonal antibody anti-Brn3b (Santa Cruz Biotechnology, Santa Cruz, CA). All antibodies against retinal subtypes have been used by us and confirmed to recognize mouse retina. ^{23 –25} The first antibodies were visualized by using appropriate Alexa 488 or Alexa 546-conjugated secondary antibodies (Molecular Probes). Samples were mounted in reagent (VectaShield; Vector Laboratories, Burlingame, CA) and analyzed under a microscope (Axioplan; Zeiss, Oberkochen, Germany). 

Retinal Cultures and Electroporation

Reaggregation cultures were set up as described earlier. ²³ Briefly, retinal cells of Dicer-CKO:GFP or GFP mice at embryonic day (E) 16 were dissociated and mixed with far larger numbers of host retinal cells isolated from normal ICR mice at E16. The ratio of donor to host cells was 5:95. Electroporation was performed using an electroporator (CUY21; Nepa Gene, Chiba, Japan) and electrode (CYU520P5; Nepa Gene), as described. ²⁶ Briefly, retinas were transferred to a micro-electroporation chamber filled with plasmid solution (1 mg/mL in Hanks' balanced salt solution), and four square pulses (25 V) of 50-μs duration with 950-μs intervals were applied using a pulse generator (CUY21; Nepa Gene). 

Results

Inactivation of Dicer in Retinal Progenitor Cells Results in Severe Retinal Malformation

To inactivate Dicer in retinal progenitor cells, we used Dkk3-Cre mice, which express Cre recombinase beginning on at least E10.5 in a retina-specific manner. ¹⁹ The expected numbers of Dicer^flox/flox/Dkk3^cre/− (Dicer-CKO) mice, based on Mendelian genetics, were born, but they died approximately 4 to 6 weeks after birth for unknown reasons. Their eyes never opened (Fig. 1A), and they had small earlobes (Fig. 1A, red arrows), probably because of undetectable expression of Dkk3 in this region. The Dicer-CKO embryos were indistinguishable from their littermates in their appearance, except for their small eyes (Fig. 1C). During development, we examined eye structure in more detail using frozen sections. We first confirmed that Cre was expressed in nearly the whole area of the retina at E17.5 (Fig. 1D), as expected from the expression pattern of Dkk3-Cre original mice. ¹⁹ At E16.5, the retinas of the Dicer-CKO mice were already smaller than those of control mice (Figs. 1E, 1F). At this stage, the ganglion cell layer (GCL) was visible in control (Fig. 1E, arrows) but not in the Dicer-CKO (Fig. 1E) mice. At E18.5, the difference between retinal sphere diameters in the Dicer-CKO and littermates became clearer (Figs. 1G, 1H). At P1, the GCL and the inner plexiform layer (IPL) were clearly formed in the control retinas (Fig. 1J) but had not formed in the Dicer-CKO retinas. Retinal diameters were even smaller than those at E16.5 in Dicer-CKO mice, and the cells were not tightly linked (Fig. 1I). At P5, the retina was very thin, and no layered structure was observed in the Dicer-CKO mice (Fig. 1K'). At P14, the retina had no visible structure in Dicer-CKO mice, and there only a few cell aggregates remained in the central region in Dicer-CKO retinas (Fig. 1M). 

We measured retinal thickness. Retinas of Dicer-CKO were thinner than those of control mice at E17 and constantly became thinner as development proceeded (Fig. 1O). 

Differentiation Markers of Retinal Subtypes Were Once Expressed, Then Disappeared, as Retinal Development Proceeded

To examine the differentiation of retinal subtypes in the Dicer-CKO retina, we immunostained various markers of retinal subtypes using frozen sections. At first, control staining to examine nonspecific signal was performed using frozen, sectioned retinas at E17 and P6. Control immunoglobulin was used as the first antibody, and appropriate second antibodies conjugated with either Alexa 488 or Alexa 594 were stained. With E17 samples, mouse IgG showed nonspecific staining in regions around the GCL and inner nuclear layer (INL), which was thought to be the blood vessel, and rat primary antibody gave no significant nonspecific staining (Supplementary Fig. S1). At P6, signals around GCL in mouse IgG antibodies, but not in the rat IgG antibody, were observed (Supplementary Fig. S1). At E16, although the control retina had no layer structure except for the GCL, the inner half of the cells had became postmitotic whereas the outer half was still composed of undifferentiated progenitor cells, as shown by the restricted expression of the early neural marker βIII-tubulin in the inner half (Fig. 2A). In Dicer-CKO mice, although the βIII-tubulin signal was observed in the inner side of the Dicer-CKO retina, there were also signals in the outer half of the retina; consequently, there was no clear boundary between βIII-tubulin–positive and –negative fields, as seen in the controls (Fig. 2A). Then we examined the expression of differentiation markers. GS, a marker of Müller glia cells, was expressed in the inner half of the retina in controls (Fig. 2A). Again, like the βIII-tubulin staining pattern, GS expression was scattered throughout the Dicer-CKO retina, and no boundary between GS-positive and -negative regions was observed. HuC/D, a marker for amacrine and ganglion cells, and Brn3b, which is expressed in ganglion cells, were expressed in the innermost part of the control retina, forming a layer-like structure (Fig. 2B). In the Dicer-CKO retina, HuC/D was weakly expressed with relatively stronger intensity in the inner half of the retina (Fig. 2B). The Brn3b pattern was also expressed in the inner side, and strong signals were also observed in the outer region (Fig. 2C). Although Brn3-positive cells were seen in the innermost side of the retina, the IPL was not observed, suggesting that process extension is inhibited by the depletion of Dicer. These results indicate that, in Dicer-CKO mice, early differentiation of retinal progenitor cells was under way. We next examined the time course of the expression of several markers. Expression of calbindin, an amacrine and horizontal cell marker, was clearly observed as making lines in the outer region and GCL at E19 in controls (Fig. 2D, blue arrows). In the Dicer-CKO retina, expression of calbindin was observed, but positive cells did not make lines and were scattered in the whole area of the retina (Fig. 2D). At P1, controls showed an expression pattern similar to that at E19, but in Dicer-CKO retina, the expression of calbindin was diminished (Fig. 2D). 

Brn3 was expressed in GCL at all examined stages in control retinas. At E17 and E18, Brn3 signals were observed at the innermost side of the retina, and it was also scattered at all areas of the retina. At later stages, the number of positive cells decreased, but signals were observed in all areas of the retina (Fig. 3A). We counted the Brn3-positive cells semiquantitatively, and the cell numbers of Brn3 were slightly fewer in control than in Dicer-CKO (Fig. 3B) mice. Because the Dicer-CKO retina was thinner than the control retina (data not shown), the percentage of Brn3-positive cells in total retinal cells was much larger in Dicer-CKO than in control. At the P1 stage, Pax6 and PKC were weakly expressed in the whole retinal area (Fig. 4A). GS was not expressed in either control or Dicer-CKO retina, and some PNR-positive cells were observed in the Dicer-CKO retina although no signal was observed in the control retina (Figs. 4A, 4B). At P5, Hu, calbindin, and Pax6 signals were making lines near IPL in the control retina (Fig. 4B). However, in Dicer-CKO, no layer structure was observed even at this stage, but Hu, calbindin, Pax6, and Brn3B expression was observed in the whole retinal area (Fig. 4B). 

Figure 3.

 

Enhanced expression of Brn3 in developing retinas of Dicer-CKO mice. (A) Retinas from Dicer-CKO or littermate control mice at indicated stages were frozen sectioned. Immunostaining using anti-Brn3 antibody was performed, and nuclei were visualized by DAPI staining. Scale bars, 100 μm. Lower panels are enlarged images of the white squared regions in the upper panels. (B) The number of Brn3-positive cells in the retina at indicated stages was examined. Brn3-positive cells at the central region of the retina were counted under a microscope in an area 100-μm wide. The average of three independent retinas with SD is shown. **P <0.01 and *P <0.05, were calculated by Student's t-test.

Figure 4.

 

Disturbed expression of retinal marker proteins of Dicer-CKO mice after birth. (A, B) Retinas from dicer-CKO or littermate control mice at P1 (A) or P5 (B) were frozen sectioned. Immunostaining using indicated antibodies was performed, and nuclei were visualized by staining of DAPI. (A) Lower panels are enlarged images of the white squared regions in the upper panels. (B) DAPI-stained images. Scale bars are as indicated.

Next, we analyzed cell proliferation by examining anti–phospho-Histone H3, which is a marker of cells at the M phase of the cell cycle. As expected, signals were observed in the most apical edge in both control and Dicer-CKO retinas at E17 (Figs. 5A, 5B). In control samples at E18 and E19, patterns of staining were similar to those at E17 (Fig. 5A). In Dicer-CKO retina, positive signals were observed at the apical-most side, but in some regions, signals were also observed throughout the retinal region (Fig. 5B), suggesting that the layer structure was perturbed in the Dicer-CKO retina. In the P1 control retina, signals had disappeared from some portions of the central region but remained at the periphery (Fig. 5A, P1, inset). In the Dicer-CKO retina, strong signals were still observed in the central region. At P5, signals had completely disappeared from the central region (Fig. 5A, P5, upper panel) but remained in the peripheral region (Fig. 5A, P5, lower panel). In contrast, signals were still observed in both central and peripheral regions in the Dicer-CKO retina (Fig. 5B, P5). Semiquantitative counting of positive cells showed that until P19, the total number of phospho-Histone H3–positive cells was comparable between control and Dicer-CKO retinas (Fig. 5E). At P5, although no signal was observed in the central region of the control retina (Fig. 5A), the total number of phospho-Histone H3 in the control retina was larger than in the Dicer-CKO retina (Fig. 5E), probably because of the small size of the Dicer-CKO retina; this is supported by the finding that the population (%) of phospho-Histone H3–positive cells was bigger in Dicer-CKO than in controls in all examined stages (Fig. 5F). Then we examined cell apoptosis using anti–active caspase3 antibody, which was rarely expressed in any of the examined developmental stages in controls (Fig. 5C, E17∼P5) but was strongly expressed in Dicer-CKO retinas at all stages (Fig. 5D, E17∼P5), suggesting that abnormal apoptosis was induced in the Dicer-CKO retinas. 

Figure 5.

 

Proliferation was slightly suppressed, but massive cell death occurred in Dicer-CKO retinas. Retinas of Dicer-CKO or control Dicer-fl/fl mice at indicated developmental stages were frozen sectioned, and immunostaining was performed using anti–phospho-Histone H3 (A, B) or anti–active Caspase3 (C, D) antibodies. Nuclei were visualized by staining with DAPI. (A–D) E17 to P1. Lower panels are enlarged images of the white squared regions in the upper panels. Lower right panels are without DAPI signals. (A, B) P5. Upper panels show peripheral retinas, and lower panels show central retinas. (C, D) P5. Images of the central region of retinas are shown. Right panels are without DAPI. (E, F) Phospho-Histone 3–positive cells in the central region of retina (100-μm wide) were counted at each stage, and the cell number (E) and positive cell population in percentages (F) are shown. The average of three independent retinas with SD is shown. Scale bars are as indicated.

Apoptosis Induced by the Deletion of Dicer Is an Autonomous Cell Phenomenon

We examined whether the apoptosis observed in the Dicer-CKO retina was cell autonomous using reaggregation cultures, which are a good model for evaluating the intrinsic characteristics of proliferation and differentiation of donor cells in a defined environment. ^23,27 To prepare reaggregation cultures, dissociated retinal cells from Dicer^flox/flox:GFP ^+/GFP:Dkk3^+/cre (Dicer-CKO/GFP) or GFP mice at E16.5 were mixed with an excess number of dissociated host retinal cells from wild-type mice at E16.5. After 5 or 8 days of culture, samples were harvested, frozen-sectioned, and immunostained with anti-GFP antibody. We found that although we used the same number of GFP-positive control or Dicer-CKO/GFP-derived cells in the aggregate cultures, the Dicer-CKO/GFP cells, but not the control GFP cells, decreased quickly during culture. After 8 days of culture, there were fewer than 50 Dicer-CKO cells but a large number of GFP-positive control cells (Fig. 6A). To quantify the results after culturing, we dissociated reaggregations, immunostained the dissociated cells, and counted immunopositive cells semiquantitatively. It was revealed that approximately 25% of Dicer-CKO/GFP cells were active caspase3-positive after 5 and 8 days of culture, whereas fewer than 2% (5 days) or 0% (8 days) of control GFP-positive or -negative cells were positive (Fig. 6B). We also examined the expression of rhodopsin and Pax6; no significant difference in expressing cell populations was observed (Figs. 6C, 6D). 

Figure 6.

 

Cell death of retinal cells by deletion of Dicer occurred cell autonomously. Re-aggregation culture of retinal cells from Dicer-CKO/GFP or control was performed. Retinal cells at E16.5 were dissociated, mixed with excessively large numbers of host normal cells, and cultured for 12 days. Then re-aggregation cultures were harvested, frozen sectioned, and examined for proliferation and differentiation by immunostaining. (A) Sections were immunostained with anti-GFP antibody, and nuclei were visualized by staining of DAPI. Scale bar, 100 μm. (B–D) Apoptotic cells (B), rhodopsin (C)-, and Pax6 (D)-positive cells were examined by immunostaining using anti active-caspase3, rhodopsin, and Pax6 antibody, respectively. Double staining with GFP antibody was performed, and marker and EGFP double-positive populations (%) in total EGFP-positive cells are shown. (B, right) Caspase 3–positive cells in the EGFP-negative population. The same set of experiments was conducted three times, and essentially the same results were obtained.

Forced Expression of Cre around Birth Induced Apoptosis and Affected the Expression of Differentiation Markers

To examine the effect of deleting Dicer at a later stage of retinal development, the Cre gene was introduced into retinas isolated from Dicer-fl/fl or control mice at P1 using in vitro electroporation. First, we examined the expression of Cre protein by immunostaining frozen sections after 12 days of culture. Strong anti-Cre signals were observed, and most of the signals overlapped GFP signals (Fig. 7A). Apoptosis was also induced by Cre expression in the Dicer-fl/fl retina, but only a very small number of apoptotic cells appeared in the control retina (Fig. 7B). However, we cannot rule out the possibility that the electroporation procedure has a stronger apoptotic effect on Dicer-CKO retinas than on controls. Then we examined the expression of PNR and GS by immunostaining because rod photoreceptors and Müller glia differentiate at a relatively later stage of retinal development. Nearly 90% of the GFP-positive cells were PNR-positive in both control and Dicer-fl/fl retinas (Figs. 7C-E). There were slightly fewer GS-positive cells in control than in Dicer-fl/fl retinas (Figs. 7F, 7I). These results suggest that the differentiation of retinal cells into rod photoreceptors and Müller glia may not be perturbed by the deletion of Dicer. However, when we examined PKC (bipolar) and Islet1 (ganglion and amacrine) markers, we were not able to observe any PKC/EGFP double-positive cells in the Dicer-fl/fl retina (Fig. 7G, 7J). In addition, the number of islet1/EGFP-positive cells in the Dicer/fl/fl retina was significantly lower than in the control retina (Fig. 7H, 7K). 

Figure 7.

 

Expression of Cre at P1 Dicer-CKO retina resulted in enhanced apoptosis and perturbation of differentiation. (A) pCAG-Cre-IRES-EGFP was introduced into the retina at P1 of wild-type mice by electroporation. After 12 days of culture, expression of Cre and EGFP was examined by anti-Cre and -EGFP antibodies, respectively, by immunostaining of frozen sections. Similar results were obtained when we used retinas from Dicer-fl/fl mice. Scale bar, 50 μm. (B–G) pCAG-Cre-IRES-EGFP was introduced into retinas at P1 of control (wild-type) or Dicer-CKO and was cultured for 12 days. Apoptosis was examined by anti–active Caspase 3 antibody, and positive cells in the central retinal region (100 μm wide) were counted semiquantitatively in EGFP-positive cells (B). Differentiation of cells into photoreceptor (C–E), Müller glia (F, I), bipolar (G, J), or ganglion/amacrine (H, K) was examined by immunostaining with anti-PNR, GS, PKC, or Islet1 antibodies, respectively. Populations of PNR (E)-, GS (I)-, PKC (J)-, or Islet1 (K)-positive cells in total EGFP-positive cells were calculated semiquantitatively in the central retinal region (100 μm wide). The average of three independent retinas with SD is shown. **P <0.01 and *P <0.05 were calculated by Student's t-test. Nuclei were visualized by staining of DAPI. Scale bars: 100 μm (B), 50 μm (C), 100 μm (E), and 50 (F) μm.

Discussion

We found that the deletion of Dicer in retinal progenitor cells during early development resulted in severe malformation of the retina; before P14, the Dicer-deleted retina had totally degenerated. In the Dicer-CKO retina, caspase was activated at all the examined developmental stages, suggesting that apoptosis was induced by the expression of Cre. Our finding is consistent with a previous study of αPax6-enhancer–dependent Dicer-CKO retinas, ¹⁸ reporting increased apoptosis by the deletion of Dicer in retinal progenitor cells. In addition, when we expressed Cre at a later stage (P1), the active caspase signal was observed to be at a significantly higher level than control, suggesting that Dicer is also essential for the survival of retinal cells after birth. This notion is supported by the finding that although some cells expressed differentiation markers and then survived after birth, all the cells ultimately disappeared, and the retina had completely degenerated before P14. 

In contrast, retinal differentiation was less affected by the deletion of Dicer. Although localization was perturbed, retinal subtype marker–positive cells were present in the Dicer-CKO retina. In addition, in terms of marker expression, the forced expression of Cre in the later phase of development by electroporation supported the concept of the nonessential role of Dicer in differentiation. However, the detailed examination of marker expression patterns revealed that the effects of deletion of Dicer for each marker are different. The most striking is the upregulation of Brn3. Georgi and Reh ¹⁸ also observed a similar phenomenon: the enhanced expression of Brn3 in αPax6-Cre/Dicer-fl/fl mice. They report observing both the upregulation of early neuronal types (such as horizontal cells) and the downregulation of late progenitor cell markers. ¹⁸ We examined horizontal cell differentiation by the expression of calbindin, which appeared to be expressed around E19 in the control retina. We did not observe either ectopic expression before E19 or enhanced expression of calbindin in Dicer-CKO retina. In our mice, there was a possibility that the delayed onset of induction of Cre expression in some retinal progenitor cells might have resulted in their survival, allowing them to differentiate. In the experiments involving Cre expression at the P1 stage, we found a lack of expression of PKC, which is a marker for bipolar cells, and a lower level of Islet1, which is a marker of ganglion and amacrine cells. Taking all our results together, the effects of the deletion of Dicer in retinal progenitor cells might not have been simply a shift of competency of retinal progenitor cells to retinal cells born early. The explanation may be more complex and may depend on the stage of retinal cells. However, given that the lamination of retinal cells had not been observed in any of the differentiation markers, we cannot use information about the subretinal localization of cells to determine whether the expression marker represented fully differentiated retinal cells. 

In fact, we observed that the Dicer-CKO retina failed to form laminated retinal structures, and it is difficult to identify which marker-positive cells are equivalent to those in control retinas. Georgi and Reh ¹⁸ reported that the formation of GCL and INL was absent. However, Sox2 and Pax6—both retinal progenitor markers—showed relatively normal lamination in the embryonic retina of Dicer-CKO. We observed—at least from E16—no segregation of postmitotic cells or proliferating cells in the Dicer-CKO retina. Among all examinations of immunostaining, only normal positioning of cells was observed in those cells at the M phase that were marked by anti–phospho-Histone H3 antibody. In addition, we observed no layer structure in Dicer mice. In Dicer-CKO mice, the IPL was not clearly observed, suggesting that miRNA regulates the formation process of retinal cells. This suggests that Dicer is less critical for determining the fate of the retina but is critical for the migration and maturation of retinal cells. Taken together, these results show that Dicer is essential to retinal progenitor cell proliferation and survival in the retina during its early development, as in other organs. In addition, even after differentiation, Dicer is essential to cell survival and the final differentiation of retinal cells. 

The initial report of retina-specific inactivation of Dicer by Chx10-Cre showed that morphologic defects at P16 progressed to more general cellular disorganization and widespread degeneration of retinal cell types as the animals aged. ¹⁷ In this study, the authors stated that the crucial role of Dicer is long-term regulation or retinal cell lamination, survival, and function, with no visible impact on early postnatal retinal structure or function. In our Dicer-CKO mice, at P16, we could not detect any retina-like structure; although some cells remained around the lens, these cells were active caspase3-positive. In contrast, Georgi and Reh ¹⁸ and we observed massive cell death at an early stage of retinal development. Because the same Dicer-flox mice ¹⁴ were used in the studies, this might have been due to differences in the Cre mice. Damiani et al. ²⁸ used Chx10-Cre mice, made by using a Chx10-BAC construct. Chx10-BAC reporter analysis showed that the Chx10 enhancer drives downstream genes beginning from at least E11.5. However, mosaic expression of target genes in the retina was observed. ²⁸ Mosaic expression of Cre was also observed in Chx10-Cre/Dicer-flox mice. ¹⁷ Based on the lack of a severe phenotype, it was surmised that either miRNAs in the retina are extremely stable or that an additional protein can compensate for Dicer function during early postnatal life. ¹⁷ Georgi and Reh ¹⁸ discussed the possibility of non–-cell-autonomous rescue of the phenotype of the Chx10-Cre/Dicer-deficient retina and the difference of onset of Cre in these mice. However, our observation of reaggregation culture suggested that the effects of deletion of Dicer are cell autonomous. The electroporation of Cre-expressing plasmid suggested that the severe effects of Cre deletion may be unrelated to the timing of expression, at least until the neonatal stage. Therefore, we postulated that mosaic expression of Cre is a less likely explanation of the phenotype. Furthermore, it seems unlikely that stable miRNA and other Dicer-like proteins are present in Chx10-Cre/Dicer-deficient retina. One possible reason is that the numbers of retinal progenitor cells expressing Cre may be too small in Chx10-Cre/Dicer-flox mice during early developmental stages and that the elimination of these cells was negligible in comparison with healthy cells or did not affect the gross morphology of the retina during development. Consequently, Chx10-Cre may turn on after birth in bipolar cells and cause the later phenotype. 

The ubiquitous expression of Cre in retinal progenitor cells afforded by use of the Dkk3-promoter has enabled clarification of the essential role played by Dicer in retinal development. 

Supplementary Materials

Figure sf01, TIF - Figure sf01, TIF 

Footnotes

 Supported by MEXT, Japan.

Footnotes

 Disclosure: A. Iida, None; T. Shinoe, None; Y. Baba, None; H. Mano, None; S. Watanabe, None

The authors thank Takahisa Furukawa for providing Dkk3-Cre mice; Robert Whittier, Itsuki Ajioka, Shuji Takada, and Yoko Tabata for discussions and technical advice; Yumiko Ishii and the FACS core laboratory for technical support with sorting; and Dovie Wylie for excellent language assistance.