Histone Deacetylase 1 Is Essential for Rod Photoreceptor Differentiation by Regulating Acetylation at Histone H3 Lysine 9 and Histone H4 Lysine 12 in the Mouse Retina*

Histone acetylation has a regulatory role in gene expression and is necessary for proper tissue development. To investigate the specific roles of histone deacetylases (HDACs) in rod differentiation in neonatal mouse retinas, we used a pharmacological approach that showed that inhibition of class I but not class IIa HDACs caused the same phenotypic changes seen with broad spectrum HDAC inhibitors, most notably a block in the differentiation of rod photoreceptors. Inhibition of HDAC1 resulted in increase of acetylation of lysine 9 of histone 3 (H3K9) and lysine 12 of histone 4 (H4K12) but not lysine 27 of histone 3 (H3K27) and led to maintained expression of progenitor-specific genes such as Vsx2 and Hes1 with concomitant block of expression of rod-specific genes. ChiP experiments confirmed these changes in the promoters of a group of progenitor genes. Based on our results, we suggest that HDAC1-specific inhibition prevents progenitor cells of the retina from exiting the cell cycle and differentiating. HDAC1 may be an essential epigenetic regulator of the transition from progenitor cells to terminally differentiated photoreceptors.

Histone modifications regulate both chromatin structure and gene expression, and changing patterns of such modifications are an integral part of normal tissue development (1). In particular, a series of acetylated derivatives of histone 3 (H3) 4 and histone 4 (H4) are thought to be sensitive indicators, and possibly predictors, of gene expression, but the complexity of this histone code is far from understood (2,3). The positively charged lysines in the N-terminal tails of histones H3 and H4 bind to negatively charged phosphates of DNA and restrict DNA accessibility, promoting chromatin condensation and repressing gene transcription. Acetylation of lysines 9, 14, 18, and 27 in H3 and lysines 8, 12, and 16 in H4 histones inhibits DNA-nucleosome interaction, opening chromatin for transcription and activating gene expression (1, 4 -8).
Acetylation of histones is regulated by two antagonist enzyme groups: histone acetyltransferases and histone deacetylases (HDACs) (9). The family of mammalian HDACs is composed of four classes based on their homologies to yeast proteins and enzymatic function (10,11). HDAC1, HDAC2, HDAC3, and HDAC8 belong to class I (12,13). Because of their higher levels of expression, nuclear localization, and high enzymatic activity toward histone substrates, class I HDACs are thought to play a central role in cell differentiation and tissue development, especially in neurogenesis (11,14). Class I HDACs exert their functions as subunits in chromatin complexes such as Sin3, nucleosome remodeling and histone deacetylation (NuRD), corepressor for element-1-silencing transcription factor (CoREST) and nuclear receptor co-repressor/silencing mediator for retinoid or thyroid hormone receptors (NCoR/SMRT). Class II HDACs have several additional protein domains, demonstrate low enzymatic activity in vitro, and are found in cytoplasm as well as in nuclei. They are subdivided into class IIa (HDAC4, HDAC5, HDAC7, and HDAC9) and class IIb (HDAC6 and HDAC10). Class II HDACs are expressed at lower levels, show more restricted patterns of expression, and shuttle between cytoplasm and nuclei. They are thought to function by recruiting class I HDACs or other transcriptional repressors to chromatin complexes (10,15).
In the central nervous system (CNS), HDACs regulate differentiation of both neuronal and glial lineages (16,17). Several studies of histone acetylation in neural development using pan-HDAC inhibitors or knockdown or knock-out models have given somewhat contradictory results (for a review, see Ref. 18), suggesting that spatial-temporal differences in expression of different HDACs are crucial for their function. So far, there have been few studies of the actions of individual HDACs.
Sodium butyrate was first discovered by Riggs et al. (20) to cause acetylated histones to accumulate in HeLa and Friend erythroleukemia cells, and further studies by Candido et al. (21) showed that the effect was due to the inhibition of histone deacetylases in a reversible manner. In a deacetylase activity assay with recombinant human HDACs (rhHDACs), entinostat exhibits EC 50 in the 100 nM range for rhHDAC1. No inhibition was found for rhHDACs 4, 6, 7 and 8 and an EC 50 in the low M range was described for rhHDACs 2, 3, and 9 (22). The same assay showed that apicidin inhibits rhHDACs 2 and 3 in the nanomolar range (120 and 8 nM, respectively) with no inhibitory effect on rhHDAC1 and class II rhHDACs (22). CAS 193551-00-7 is a benzamide derived from entinostat with selective inhibitory activity for HDAC1 in the range of 100 nM (23). An in vitro enzyme inhibition assay has shown the selectivity for only HDAC1 with an effect on cell proliferation in murine erythroleukemia cells (23). In primary cell cultures, 1 M CAS 193551-00-7 had the same effect as specific HDAC1 siRNA (24). At a concentration of 1 M, CAS 193551-00-7 does not inhibit HDAC3 (25).
Romidepsin is produced by Chromobacterium violaceum and in an assay using HDACs prepared from 293T cells showed that the IC 50 for HDAC1 and HDAC2 is in the 40 nM range, whereas inhibition for class II HDAC4 is 500 nM and for HDAC6 is 14,000 nM (26 -28). The class IIa inhibitor MC-1568 was shown to be selective for this class of HDACs with an IC 50 of 220 nM. No inhibitory activity was found for HDAC1, whereas 55% of HDAC4 activity was inhibited in these enzymes immunoprecipitated from human breast cancer ZR-75.1 cell lysates (29).
We have used the mouse retina as a model to investigate the role of HDACs in terminal differentiation of a specific neuronal cell type, the rod photoreceptor. All retinal cell types originate from a multipotent retinal progenitor cell in a time-dependent manner with horizontal, cone, amacrine, and ganglion cells born embryonically and the great majority of rod, bipolar, and Müller cells being born postnatally (for a review, see Ref. 30). After postnatal day 1 (PN1), rod photoreceptors, which represent the majority of cells in the retina, begin differentiation and start expressing a number of cell type-specific molecules, including the visual pigment protein rhodopsin. Previous studies have linked non-selective inhibition of HDACs by TSA or FIGURE 1. HDAC expression in the mouse retina during development from E16 to PN21measured by RT-PCR. RNA was isolated from mouse retina at specific developmental times, and samples from brain and body mixed tissue (adult) were used as controls. ␤-Actin RT-PCR was used as a loading control. expression of rod-specific genes. HDAC1 appears to be key enzyme controlling the passage of cells from a progenitor to a terminally differentiated state.

Inhibition of Class I HDAC Selectively Increases H4K12ac and H3K9ac Acetylation and Decreases Rhodopsin
Expression-To analyze HDAC expression in the retina, total RNA was isolated from retinas at embryonic day 16 (E16), PN1, PN7, and PN21 and compared with expression in the body and in brain. Most HDACs are expressed in retina, with the exception of HDAC7 (HDAC8 was not tested), and show varying levels throughout development (Fig. 1). General inhibition of HDACs by nonselective inhibitor TSA or sodium butyrate blocks rod photoreceptor development in mouse retina (31). To test whether there was selectivity in the inhibition of rod differentiation by HDACs, we compared the effects of inhibitors of two classes. In retina explants cultured from PN1-PN4, we tested a panel of inhibitors for class I and class IIa HDACs (Table 1) and followed rhodopsin expression as a marker of rod photoreceptor differentiation. No difference in rhodopsin staining was observed in retinas treated for 96 h with class IIa HDAC inhibitor MC-1568 when compared with control ( Fig. 2, A, B, and E). In contrast, inhibition of class I HDACs by entinostat for 96 h showed a 60% decrease in rhodopsin labeling, a decrease comparable with rhodopsin down-regulation by sodium butyrate (Fig. 2, C, D, and E). Co-staining with antibodies recognizing one histone acetylation site, H3K9ac, showed that the level of acetylation was dramatically increased under entinostat and sodium butyrate treatment but not under MC-1568 treatment (Fig. 2, C, D, and F).
Entinostat inhibits all class I HDACs. Because selective inhibitors for HDAC1 and HDAC3 are available, we wanted to determine whether inhibition of one or both of these enzymes was responsible for the results seen with entinostat. We incubated retinas with either CAS 193551-00-7, an HDAC1 selective inhibitor, or apicidin, an HDAC3 selective inhibitor. Inhibition of HDAC1 using CAS 193551-00-7 resulted in a dose-dependent decrease in rhodopsin (Fig. 3, A-G and I) and an increase in labeling by antibodies recognizing H4K12ac (Fig.  3, A-D and J) and H3K9ac (Fig. 3, F and K) but no change in labeling with anti-H3K27ac (Fig. 3, G, H, and L). At the highest concentration of CAS 193551-00-7, there was a clear loss of retinal structure. To confirm that HDAC1 inhibition was not selectively impacting rhodopsin expression but had a more general effect on photoreceptors generation, we examined another photoreceptor specific protein, recoverin, and found that levels were decreased under specific HDAC1 inhibition similarly to rhodopsin (Fig. 4).
Inhibition of HDAC3 by apicidin (Fig. 5, A-C) resulted in a modest decrease (25%) in rhodopsin staining (Fig. 5D) and no visible change in anti-H4K12ac (Fig. 5E) or H3K9ac labeling (Fig. 5F). There is no HDAC2-specific inhibitor, and we were unable to assess HDAC2-specific regulation of retina development, but we could not rule out a possible role of HDAC2 in this process. From these results, we conclude that class I HDACs are important for rod differentiation as defined by the appearance of rhodopsin-or recoverin-expressing cells in developing retina and that within this class HDAC1 is the most important for this process.
To confirm the effects of the different HDAC inhibitors (HDACis) on the global pattern of histone acetylation in the retina shown by immunofluorescence staining, we performed Western blotting using antibodies against H3K9ac, H3K27ac, and H4K12ac marks. PN1 explants were incubated in the presence of six different HDACis for 96 h. As shown in Fig. 6, inhibition of all HDACs with sodium butyrate increased the acetylation level of all three histone modifications. Selective inhibition of HDACs 1 and 2 by romidepsin or class I HDACs by entinostat also increased the levels of the three acetylated histones. Selective inhibition of HDAC1 by CAS 193551-00-7 showed an increased acetylation for H4K12ac and H3K9ac but not for H3K27ac, matching the immunofluorescence results. In contrast, inhibition of HDAC3 by apicidin resulted in little to no change in the acetylation levels when compared with the control, confirming the lack of change in immunocytochemical labeling of H4K12ac and H3K9ac described above. Similarly, inhibition of class IIa HDACs by MC-1568 had no effect on the acetylation levels. From these results, we conclude that HDAC1 has a more widespread effect on histone acetylation in PN1-PN4 retina explants than either HDAC3 or class IIa HDACs.
HDAC1 Inhibition Results in Less Cell Proliferation and More Cell Death-Pan-HDAC inhibition in retina explants by TSA resulted in increased cell death and reduction in proliferation (31). We tested whether an HDAC1-specific inhibitor had the same effect by assaying cell death by TUNEL and cell proliferation by BrdU incorporation and proliferating cell nuclear antigen (PCNA) staining. Retina explants treated for 48 h with CAS 193551-00-7 (HDAC1i) showed an increase in the number of apoptotic cells by TUNEL assay (Fig. 7, A-C) and reduction in BrdU incorporation (Fig. 7, D-F). For comparison, explants treated with apicidin (HDAC3i) had no difference in proliferation as judged by BrdU incorporation (data not shown). Retinal explants treated for 96 h with HDAC1i showed a similar number of cells positive for PCNA when compared with the control (Fig. 7, G-I).
HDAC1 Inhibition Effect on Retina Is Reversible and Stagespecific-To determine whether HDAC1 inhibition merely paused development or permanently altered the retinal cells, we studied whether CAS 193551-00-7 treatment is reversible. Retina explants were incubated for 96 h in the presence of CAS 193551-00-7 (Fig. 8, A-C) or entinostat for comparison (Fig. 8, D-F), then inhibitors were washed out, and retinas were cultured for another 96 h without the inhibitors. After HDAC1i washout, rod photoreceptor development resumed with an increase in rhodopsin levels compared with explants maintained with inhibitor ( Fig. 8G). H4K12 acetylation levels following washout were reduced (Fig. 8H). Both results suggest that the effects of HDAC1 inhibition on rod development are reversible. Although washout of entinostat partially decreased H4K12ac level, it failed to promote rhodopsin expression. We next investigated whether the effect of selective HDAC1 inhibition is stage-specific and treated explant cultures from PN7 retinas for 48 h with CAS 193551-00-7 (Fig. 9A). We did not observe any changes in retina structure or rhodopsin level compared with control ( Fig. 9, A, left panels, and B) despite prominent up-regulation of acetylation of H3K9 (Fig. 9, A and C). This suggests that HDAC1 acts at a distinct developmental window during the first postnatal week of mouse retina maturation when precursors differentiate into rod photoreceptors.

Inhibition of HDAC1 Prevents the Increased Expression of a Specific Subset of Photoreceptor Genes while Maintaining the Levels of Expression of Genes Associated with Retinal
Progenitors-To determine whether the effect of HDAC1 inhibition was on rod photoreceptor development in general or selectively on expression of rod-specific genes such as rhodopsin or recoverin, we tested RNA levels of a panel of 31 genes that represent different groups of genes known to be down-regulated or up-regulated during retina maturation. RNA was col-  FEBRUARY 10, 2017 • VOLUME 292 • NUMBER 6 lected from retinal explants treated for 96 h with 0.5 or 1.0 M CAS 193551-00-7, and genes were considered up-or downregulated if the expression was significantly different from controls with a p value Ͻ0.05.

HDAC1 Is Essential for Rod Photoreceptor Development
These data revealed a number of important results (Fig. 10A). First, all nine rod-specific genes tested were expressed at lower levels in both treated groups with Rho, Pde6b, Samd11, Sag, Rom1, and Nrl showing more than 10-fold lower expression, demonstrating that HDAC1 inhibition blocks the program of rod photoreceptor differentiation rather than affecting any individual gene.
Second, seven genes of 13 that were up-regulated in both treated samples were progenitor genes, Otx2, Dll1, Notch1, Vsx2, Foxn4, and Hes1, or a cell cycle regulator, Ccnd1, with 2-5 times higher expression in treated compared with untreated explants. This suggests that a second major effect of HDAC1 inhibition is the maintenance of expression of the group of key progenitor genes that are normally turned off as the retina develops and could be achieved by keeping histone acetylation at promoters of these progenitor genes.
Third, several genes specific for non-rod retinal cell types were also up-regulated under HDAC1 inhibition. Expression levels were increased for a marker of bipolar cells (42), Pkc␣ (2.6-fold); a marker of Müller glia cells (43), Hes5 (2.1-fold); a marker of retinal ganglion cells (44), Isl (2.2-fold); and four markers of cone photoreceptors (45), Pde6c, Opn1sw, Opn1mw, and Gnat2 (Ͼ1.5 times) but not Kcnv2 gene that is expressed in both cone and rod photoreceptors. Bipolar and Müller cells differentiate postnatally, and our results are in accordance to previous studies and might indicate that bipolar and Müller cell fate may be driven by a loss of HDAC activity (31). We conclude from this that HDAC1 is essential and specific for rod photoreceptor differentiation. Fourth, in agreement with results of a decrease in cell proliferation and an increase in cell death under HDAC1 inhibition (Fig. 7), expression of apoptotic genes (46) Bcl2, Apaf1, Casp3, and Casp9 was up-regulated.
For comparison, we tested the expression pattern of a panel of nine genes selected from the earlier panel to represent genes that showed significant changes in expression in retinal explants treated with HDAC3 inhibitor apicidin as a decrease in rhodopsin staining was also observed in retinas treated with this inhibitor. Transcription factors Crx and Nrl and rhodopsin expression were modestly down-regulated (Ͻ2-fold) in the treated retinas as compared with control (Fig. 10B). Conversely, Pax6 and Hes1 expression was up-regulated slightly (Ͻ 1.5fold) as compared with control.
We then evaluated the time course of changes in expression level under HDAC1i for several key genes. At 3-h incubation of retina explant culture, there were no changes in Rho level with HDAC1i, but after 3 h its expression increased in control samples but markedly declined in HDAC1i-treated explants (Fig.  11A). Crx expression had a different time course compared with rhodopsin; at 3-h incubation with HDAC1i, its level was not significantly different from that in control explants, but then its level fell and remained significantly below control levels even at 96 h of incubation (Fig. 11B). We used caspase3 expression as a proxy to assess the process of apoptosis under HDAC1i. The level of Casp3 was already up-regulated at 3 h under HDAC1i, but it returned to the level in the control explant after 48 h (Fig. 11C). Levels of Hes1 decreased in control samples up to 48 h and then plateaued, whereas under HDAC1i expression the level of Hes1 was higher compared with control at all time points, and it began to increase after 48 h (Fig. 11D). We conclude from these experiments that there is a burst of apoptosis in retina explants under HDAC1i that starts earlier than down-regulation of Rho and that low RHO levels are due to combination of apoptosis of progenitors and maintained expression of the suppressor HES1.
HDAC1 Inhibition Leads to Maintenance of High Histone Acetylation Levels on Promoters of Progenitor Genes-As described in the previous section, HDAC1 inhibition led to maintenance of expression of a group of key regulatory progenitor genes. To test whether this was achieved by keeping histones acetylated at the promoters of these genes, we examined the promoters of progenitor gene Vsx2 (Fig. 12A) and rod-specific gene Rho (Fig. 12B) by chromatin immunoprecipitation (ChIP) and RT-PCR. We first analyzed existing retina developmental genome-wide data sets to understand how chromatin organization changed during development for these two genes. We mapped ChIP-sequencing data for inhibitory mark trimethylation of lysine 27 of histone 3 (H3K27me3) (E17, PN1, PN7, and PN15); active mark dimethylation of lysine 4 of histone 3 (H3K4me2) (E17, PN1, PN7, and PN15) (47); binding sites for photoreceptor-specific transcription factors CRX (PN56) (48) and NRL (PN28) (49); polymerase (Pol) II-binding sites (PN2 and PN25) (50); and DNase-hypersensitive sites (DHS) (51), which correspond to open chromatin regions (PN1, PN7, and PN56) (Fig. 12, A and B). The Vsx2 gene showed chromatin features characteristic for developmental down-regulation (Fig.  12A) as H3K4me2 marks and DHS decreased during retina maturation, but H3K27me3 accumulated around the promoter. There are no Pol II-or CRX-binding sites around the promoter of Vsx2. The Rho promoter showed the opposite patterns of chromatin features (Fig. 12B) with essentially no H2K27me3, increasing amounts of the active mark H3K4me2, and increased DHS as well as Pol II-, NRL-, and CRX-binding sites in mature retina.  FEBRUARY 10, 2017 • VOLUME 292 • NUMBER 6

JOURNAL OF BIOLOGICAL CHEMISTRY 2429
We next assessed levels of acetylation marks H3K27ac and H4K12ac and compared these with H3K4me2 on the Vsx2 and Rho promoters and transcription start site (TSS) by ChIP-RT-PCR at three stages of retina development, PN1, PN7, and PN15, using three pairs of genomic primers. Levels of H4K12ac (Fig. 12D) at the Vsx2 gene locus were decreased during retina development, similar to those of H3K4me2 (Fig. 12C), whereas H3K27ac levels did not change (Fig. 12E). Addition of the pan-HDAC inhibitor sodium butyrate during the ChIP procedure substantially increased the level of all three active histone modification marks at Vsx2 locus, especially H4K12ac at all developmental stages. In comparison, Rho locus was characterized by increased accumulation of all three active histone modifications during development (Fig. 12, F-H), but addition of sodium butyrate had no effect, or only a small effect, on this accumulation.
Because we saw an effect of histone modification marks at the promoter of Vsx2 using the pan-HDAC inhibitor sodium butyrate, we decided to verify the effects of HDAC1 inhibitor CAS 193551-00-7 and HDAC3 inhibitor apicidin on the histone acetylation mark H4K12ac around the promoter and TSS of this and several other progenitor genes that were up-regulated under HDAC inhibition as well as several rod-specific genes that were down-regulated under HDAC inhibition. PCRs used primers specific for promoter regions in 11 genes that showed statistical difference in the expression levels when treated with both inhibitors (Fig. 12I). Treatment with HDAC1i increased the H4K12ac level in most of the genes analyzed, especially Ccnd1, Crx, Hes1, Notch1, and Vsx2 (Ͼ2-fold change). Only Nrl, Rho, and Gapdh showed no difference in H4K12ac between treatment and control. In contrast, we observed no increase in the H4K12ac mark in retinas treated with HDAC3i on any of the 11 genes (Fig. 12I). Interestingly, Hbb, a gene not expressed in the retina, showed substantial increases in H4K12ac after treatment with either HDAC1 or HDAC3 inhibitor.

Discussion
Histone acetylation is generally thought to be associated with the opening of chromatin and the activation of transcription.  FEBRUARY 10, 2017 • VOLUME 292 • NUMBER 6

JOURNAL OF BIOLOGICAL CHEMISTRY 2431
The phenotypic effect of modulating histone acetylation is, however, more complex. Inhibiting HDACs can induce differentiation in both embryonic and adult CNS progenitors (16,52,53), but following conditional deletion of HDAC1 and HDAC2 progenitors were unable to undergo differentiation into mature neurons and underwent cell death (17).
Some of the differences in reported roles of HDACs may well be due to the different temporal and spatial distribution of the various isoforms. We found members of most classes of HDACs in the retina, each with its own developmental profile, corroborating data from others (31, 54 -56). Previous studies have shown that non-selective inhibition of HDACs in the developing retina blocked formation of rod photoreceptors (31,32), raising the question of which class of HDACs might be involved in regulating rod development. Using a variety of HDAC inhibitors, we found that HDAC1 was key in allowing the differentiation of rod photoreceptors. Although this was originally defined using expression of opsin and recoverin as markers of rods, a group of other rod-specific genes showed similar blocked expression, indicating that the whole program of rod development was inhibited. Interestingly, genes for other retinal cell types showed increased expression. This was true for

the other late developing retinal cells (bipolar cells and Müller glial cells) as well as early developing retinal cells (ganglion cells and cone photoreceptors)
. It is possible that when the progenitor pool is prevented from moving into a rod differentiation pathway the cells take on other fates. HDAC1-null mice are lethal, and the animals die before E10.5 (57). Lethal defects in neuronal elements are seen also in zebrafish deleted for HDAC1 (58 -60). In mice lacking HDAC1 in the central nervous system, neuronal progenitors do not differentiate and undergo cell death (17). Similarly, we found an increase in cell death and a decrease in cell proliferation when retinal explants were incubated with an HDAC1 inhibitor. A decrease in BrdU incorporation has been observed in the retinas of zebrafish HDAC1 mutants (60). One of the genes whose expression is increased when HDAC1 is inhibited is Ccnd1. The up-regulation of Ccnd1 in the treated retinas was unexpected because a decrease in BrdU incorporation was observed. At this point in development, photoreceptor progenitor cells become postmitotic and begin to differentiate. It is likely that because cells fail to leave the cell cycle cyclin D1 accumulates and induces apoptosis (61). This is with agreement with our observed up-regulation of the apoptotic genes Apaf1, Bcl2, and Casp3.
Although BrdU incorporation decreased in the presence of HDAC1 inhibitors, PCNA labeling remained the same. This is probably due to the different distribution of these two proliferation markers during the cell cycle. PCNA starts to increase through G 1 , peaking at the G 1 /S phase, whereas BrdU is only incorporated in the S phase. This would suggest that inhibiting HDAC1 prevents cells from moving into S phase, leading to an increase in cell death and movement into different differentiation pathways. Not all the rod progenitors die because washout of the HDAC1 inhibitor led to resumption of rod generation, indicating survival of a pool of rod progenitors.
Although we have not tested all HDAC isoforms, the only other enzyme regulating rod formation was HDAC3. The effect of HDAC3 was less prominent as determined both by reduction of rhodopsin expression and by changes in histone acetylation assayed by both immunocytochemistry and Western blotting. More modest changes in expression of particular genes were also observed. In particular, levels of RNA for the transcription factor Crx were reduced almost 2-fold, and this may have been sufficient to account for the reduced rhodopsin expression.
Acetylation of non-histone proteins can affect a number of cellular functions because it influences mRNA localization and stability as well as protein signaling, transcription, and degradation (for a review, see Ref. 62). Because HDAC1 lacks a nuclear export signal, this enzyme is primary localized in the nucleus; therefore, the effect of its inhibition is probably due to its action on histone proteins and not on non-histone targets. In contrast, HDAC3 has a nuclear export signal. It has been shown that HDAC3 can influence STAT3 phosphorylation at serine 727 through the interaction of phosphatase 2A protein (63). The persistent activation of STAT3 was shown to suppress rod cell fate by decreasing Crx and rhodopsin expression and increasing the precursor markers Hes1 and OTX2 (64 -66). This effect on non-histone HDAC substrates could explain our results for HDAC3 inhibition because the mRNA levels for Crx and rhodopsin were down-regulated and Hes1 was up-regulated in retinas treated with apicidin as compared with control.
In previous experiments, we showed that inhibiting histone demethylation by lysine demethylase 1 (LSD1) also blocked differentiation and maintains expression of progenitor genes that are normally down-regulated as the tissue matures (67). Our conclusion in that study was that among the genes whose expression was maintained were transcriptional repressors such as Hes1 that are known to block rod differentiation. Similarly, inhibition of HDAC1 caused maintained expression of Hes1. From these data, we suggest that the changes of gene expression seen with HDAC1 inhibition may be due to its direct action on the chromatin of progenitor genes. Higher levels of histone acetylation lead to an open chromatin conformation around the promoter and TSS and prevent the normal developmental switching off of the gene with subsequent effects on cell proliferation and further cell differentiation. It has been shown that when HDAC1 and LSD1 deacetylate histones and demethylate H3K4me2/1, respectively, as parts of complexes CoREST and NuRD this causes repression of both neuronal and non-neuronal genes (68 -72). In the course of retina development, the two histone-modifying enzymes HDAC1 and LSD1 may work in a common complex at the point of transition from late progenitor to differentiated rod photoreceptor (schematized in Fig. 13). Because neither HDAC1 nor LSD1 have DNA binding properties, the temporal and spatial control of the actions of these enzymes during development must reside in other components of the complexes. In retinoblastoma cells, HDAC1 is found together with LSD1 in CoREST complex-repressed genes where the complex is specifically bound to target gene promoters through the nuclear receptor NR2E1 (TLX) (73).
In summary, we have shown that inhibition of HDAC1, but not HDAC3, can prevent the developmental decrease in expression of progenitor genes and the developmental increase in expression of genes characteristic of terminally differentiated rod photoreceptors. The actions of HDAC1 affect both H3K9 and H4K12 but not H3K27, indicating targeting of this enzyme to specific chromatin sites. The similarities in phenotype observed by inhibiting HDAC1 and LSD1 and their co-expression in chromatin-modifying complexes suggest a coordinated interaction between acetylation and methylation in the regulation of the transition from progenitor to differentiated photoreceptor in the mammalian retina and possibly other tissues. Thus, our results demonstrate that, in addition to important roles of transcriptional factors during development and cell type specification, epigenetic regulation is essential for proper retina maturation. An interesting application for epigenetic modifiers such as inhibitors of HDACs or LSD1 could be for reprogramming cell phenotype during development as our studies show that, in parallel with a decrease in expression of rod photoreceptor-specific genes, genes specific for other retina cell types are up-regulated. This reprogramming could be a useful tool to fight a number of degenerative diseases of the CNS.

Experimental Procedures
Mice-C57Bl/6j mice were purchased from The Jackson Laboratory (Bar Harbor, ME). All animal experiments were approved by the Animal Care and Use Committee of the Pennsylvania State University School of Medicine (Protocol number 2009-061) and were conducted in accordance with National Institutes of Health guidelines.

Retina Isolation and Explant
Culture-Retinas from pups at PN1 were isolated by removing the sclera and most of the retinal pigmented epithelium and cultured in 1 ml of Ultra-Culture TM serum-free medium (Cambrex Bio Science, Rockland, ME) supplemented with gentamycin (10 g/ml) as described previously (76,77). Retinas were treated with different concentrations of the specific HDACi for 96 h unless otherwise mentioned. Medium was changed every other day by replacing 0.5 ml with fresh medium and adding the proper concentration of the inhibitor. For the "washout" experiments, retinas were incubated in the presence of the inhibitor for 96 h, washed with medium for five times, and incubated another 96 h with fresh medium with 1% DMSO.   FEBRUARY 10, 2017 • VOLUME 292 • NUMBER 6 Immunofluorescence Staining-Retina explants were fixed in 4% paraformaldehyde overnight at 4°C; washed three times in PBS; incubated in 5% sucrose, PBS for 30 min; and then dehydrated in 20% sucrose, PBS overnight at 4°C. Retinas were embedded in a 2:1 mixture of 20% sucrose and OCT (Optimal Cutting Temperature Compound) (Sakura Finetek, Torrance, CA) and stored at Ϫ80°C. Blocks with tissue samples were sectioned to 8 m on a Cryostat Microtome HM550 (Thermo Fisher Scientific) and stored at Ϫ20°C. Antigen retrieval was performed by incubating the slides in 10 mM sodium citrate, pH 6, for 30 min at 80°C. Double labeling immunohistochemistry was performed as described previously (64, 65) using fluorescent dye-conjugated secondary antibodies diluted 1:500 (BD Biosciences). Primary antibodies were diluted as follow: anti-H3K9ac, ant-H3K27ac, and anti-H4K12ac, 1:500; anti-rhodopsin, 1:100; anti-BrdU, 1:20; and anti-PCNA, 1:1000. Slides were counterstained with Hoechst 33258 (1 g/ml) and visualized using an Olympus Fluoview FV1000 confocal microscope (Olympus, Center Valley, PA). The acquisition parameters were maintained constant for each set of experiments. Fluorescence intensity was assessed and quantitated using the ImageJ (1.48v/Java1.6.0 -20) software package.

HDAC1 Is Essential for Rod Photoreceptor Development
Western Blotting-PN1 retina explants were cultured in the presence of 54 nM romidepsin, 1 M CAS 193551-00-7, 5 M entinostat, 36 M MC-1568, 16 nM apicidin, or 5 M sodium butyrate. Proteins were extracted with lysis buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA, 1% SDS, 1 mM PMSF, 1ϫ Halt proteinase inhibitor mixture (Thermo Fisher Scientific)). Sam-ples were sonicated twice, and protein concentration was measure using Bio-Rad DC TM protein assay according to the manufacturer's instructions. Samples were diluted to 1 mg/ml, resolved on a Criterion TGX precast gel Any kD (Bio-Rad), and transferred to nitrocellulose membranes followed by immunoblotting with antibodies anti-H3K9ac, anti-H3K27ac, and anti-H4K12ac diluted 1:5000. Secondary goat anti-rabbit HRP (Jackson ImmunoResearch Laboratory, West Grove, PA; 111-035-144; lot number 87644) was diluted 1:5000. An ECL Western blot detection system (Thermo Fisher Scientific) was used for visualization according to the manufacturer's instructions. The membrane was exposed to Hyperfilm TM film (GE Healthcare) for 20 s for H4K12ac, 5 s for H3K9ac, and 30 s for H3K27ac. For quantitative analysis of relative protein levels in samples, the Coomassie-stained gels and autoradiographs after ECL detection were scanned and digitized, and the intensity of protein bands was quantitated using the ImageJ (ImageJ 1.48v/Java1.6.0 -20) software package.
Isolation of nuclei from mouse retinas was carried out according to Popova et al. (67). For Western blots presented in Fig. 14, isolated nuclei from retina samples were dissolved in SDS-containing loading buffer, and electrophoresis was carried out in 15% polyacrylamide-SDS gels. Proteins were transferred to Immobilon-P polyvinylidene difluoride membranes (Millipore) and detected with primary and secondary HRP-conjugated antibodies.
TUNEL Assay-PN1 retina explants were treated with 0.5 or 1.0 M CAS 193551-00-7 or 1% DMSO for 48 h. The TUNEL assay was carried out using the In Situ Cell Death Detection kit (fluorescein) from Roche Applied Science according to the manufacturer's instructions. Sections were counterstained with Hoechst 33258 (1 g/ml) and analyzed by confocal microscopy.
cDNA Synthesis and RT-PCR-Total RNA was isolated using QIAshredder and an RNeasy mini kit (Qiagen, Germantown, MD) according to the manufacturer's instructions. RNA concentration and purity were determined using a GeneSpectIII (Hitachi, Tokyo, Japan) spectrophotometer. cDNA was synthesized using the SuperscriptIII First-strand Synthesis System (Invitrogen) according to the manufacturer's instructions.   FEBRUARY 10, 2017 • VOLUME 292 • NUMBER 6

HDAC1 Is Essential for Rod Photoreceptor Development
Quantitative real time PCRs were made in triplicates using the 2ϫ iQ-SYBR Green PCR Supermix (Bio-Rad) on a CFX96 Touch Real-Time PCR Detection System (Bio-Rad). Lists of primers used are in Tables 2 and 3. The relative expression level for each gene was calculated by the 2 Ϫ⌬⌬Ct method (78,79) where ⌬Ct ϭ Ct(target gene) ϪCt(Gapdh) and ⌬⌬Ct ϭ ⌬Ct(target sample) Ϫ ⌬Ct(reference sample), and the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (Gapdh) was used as a reference gene. During retina development in PN1-PN4 explant cultures, the Gapdh expression level is stable (76). Genes were considered up-or down-regulated if the p value was Ͻ0.05.
Chromatin Immunoprecipitation-PN1 retina explants were treated for 96 h with either 1.0 M CAS 193551-00-7, 16 nM apicidin, or DMSO. ChIPs were done in triplicates for the HDAC1i and in duplicate for the HDAC3i. ChIP was done as previously described (37) and subjected to quantitative PCR. Absolute quantification was done by the standard curve method using primers listed in Table 2 (80,81). For quantitative real time PCR, we used 2ϫ iQ-SYBR Green PCR Supermix. Samples in triplicate were run on the iQ5 Multicolor Real Time PCR Detection System (Bio-Rad). Each run included suitable standard curve samples using a wide range of concentrations for genomic DNA prepared similarly to inputs.
Statistical Analysis-Graphs were generated using GraphPad Prism 5 software. Results are presented as means Ϯ S.E. Unpaired, one-tailed Student's t test was used to evaluate statistical significance between groups. If variances between samples were significantly different, Welch's correction was applied. A p value Ͻ0.05 was considered significant. To compare more than two groups, a one-way analysis of variance with Bonferroni correction was used.
Author Contributions-R. C. F. and E. Y. P. designed, performed, analyzed the experiments shown in Figs. 3-11 and wrote the paper. S. S. Z. designed, performed, and analyzed the experiment shown in Fig. 1. R. C. F., E. Y. P., and J. J. designed, performed, and analyzed the experiment shown in Fig. 2. M. R. S. B. conceive the study of HDAC3 and wrote the paper. C. J. B. and S. S. Z. conceived and coordinated the study and wrote the paper. All authors reviewed the results and approved the final version of the manuscript.