HDAC inhibition in the cpfl1 mouse protects degenerating cone photoreceptors in vivo

Abstract

Cone photoreceptor cell death as it occurs in certain hereditary retinal diseases is devastating, with the affected patients suffering from a loss of accurate and colour vision. Regrettably, these hereditary cone diseases are still untreatable to date. Thus, the identification of substances able to block or restrain cone cell death is of primary importance. We studied the neuroprotective effects of a histone deacetylase inhibitor, Trichostatin A (TSA), in a mouse model of inherited, primary cone degeneration (cpfl1). We show that HDAC inhibition protects cpfl1 cones in vitro, in retinal explant cultures. More importantly, in vivo, a single intravitreal TSA injection significantly increased cone survival for up to 16 days post-injection. In addition, the abnormal, incomplete cone migration pattern in the cpfl1 retina was significantly improved by HDAC inhibition. These findings suggest a crucial role for HDAC activity in primary cone degeneration and highlight a new avenue for future therapy developments for cone dystrophies and retinal diseases associated with impaired cone migration.

Introduction

Cone photoreceptors in the human retina are responsible for sharp high-resolution vision and colour discrimination. Hereditary cone degenerations, such as in Stargardt ´ s and Best ´ s disease, achromatopsia, and cone dystrophies are caused by mutations in single genes, and lead to severe visual impairment, reduced visual acuity and loss of colour vision. Cone dystrophies are characterized by a high genetic heterogeneity, with disease causing mutations in at least 27 genes (RetNet: https://sph.uth.edu/retnet, date last accessed May 2016). In addition, a genetic contribution has also been proposed for complex diseases affecting cones, such as in age-related-macular degeneration (AMD) or diabetic retinopathy (DR). Irrespective of the genetic cause, a common outcome of retinal diseases is neuronal cell death, giving a strong rationale for targeted neuroprotective approaches that may prevent or delay cell death execution (1).

The cone photoreceptor function loss-1 (cpfl1) mutant mouse is an animal model for autosomal recessive achromatopsia or progressive cone dystrophy (2). This model is unique in the sense that it is characterized by an early onset of cone loss at post-natal day 14 (PN14) and a fast progression with the peak of cell death at PN24 (3). We have previously shown that photoreceptor cell death in the cpfl1 mouse, as well as in nine other animal models for inherited retinal degeneration, follows a non-apoptotic mechanism, characterized by accumulation of cGMP, increased activities of cGMP-dependent protein kinase (PKG), histone deacetylase (HDAC), poly-ADP-ribose-polymerase (PARP), and calpain proteases, as well as accumulation of poly-ADP-ribose (PAR) (4,5). In addition, the cpfl1 cone degeneration is also associated with prominent defects in cone migration (3).

A common non-apoptotic cell death mechanism in different retinal degeneration models provides a rationale for the identification of therapeutic targets that would prove beneficial for a broader population of patients suffering from a variety of different genetic causes (4,6). While, numerous attempts to restrain the execution of rod photoreceptor cell death were described (6–11) to date, an effective clinical treatment for primary cone photoreceptor degeneration has not been reported. Since humans use mostly cones for their vision, a development of treatment protocols to preserve cones is of highest priority in clinical ophthalmology.

The interplay between histone acetylation and deacetylation, performed by histone acetyltransferases (HATs) and histone deacetylases (HDACs) respectively, determines the transcription state of genes in general (12), and this is also true for photoreceptor-specific genes (13,14). The aberrant activity of histone deacetylases is associated with a number of diseases with very different aetiology, ranging from cancer to neurodegenerative diseases (15). Consequently, in current medicine, HDAC inhibition is discussed as one of the most promising novel therapeutic approaches for various diseases (16).

In the present study, we tested the hypothesis that hereditary cone degeneration can be prevented or delayed by HDAC pharmacological inhibition. We assessed cone photoreceptor survival after HDAC inhibition with Trichostatin A (TSA) in vitro, on retinal explant cultures obtained from cpfl1 animals, as well as in vivo, after intravitreal injection. We show that TSA protects cpfl1 cones both in vitro and in vivo. Importantly, a single TSA injection in vivo achieved a significant protection of cone photoreceptors, up to sixteen days post-injection. In addition, we observed a significant improvement of impaired cone migration, present in the degenerating cpfl1 retina. Our study highlights the possibility to use pharmacological HDAC inhibition for substantial cone protection in an inherited cone dystrophy, and may forward the future development of therapies aimed at cone photoreceptor preservation.

Results

HDAC activity is increased in cpfl1 photoreceptors

We have previously shown that the cpfl1 cone photoreceptor degeneration follows a non-apoptotic cell death mechanism characterized by increased HDAC activity at the peak of degeneration at PN24 (4). In this study, we evaluated HDAC activity also at the onset of cone degeneration (PN14) and found significantly more HDAC activity positive photoreceptors in mutant retinas compared to wild-type (wt) (Supplementary Material, Fig. S1A–F). Increased HDAC activity present also at PN24 was significantly decreased after the TSA treatment (P = 0.004; Supplementary Material, Fig. S1G).

HDAC inhibition with TSA rescues degenerating cones in vitro

To test whether HDAC inhibition could promote cone photoreceptor survival, we used cpfl1 long-term retinal explants cultured under entirely defined conditions, in serum-free medium. Both cpfl1 and wt retinas were explanted at the onset of cpfl1 degeneration (PN14) and cultured for 2 days in regular R16 retinal culture medium, without any treatment. Retinas were then exposed to 10 nM TSA every second day, until PN24. In the untreated cpfl1 explants almost half of the cones were lost at PN24, as assessed by the percentage of cones in the outer nuclear layer (ONL), identified by labeling with an antibody for glycogen phosphorylase (Glyphos) (17). When compared to untreated wt explants (n = 4), only ∼53% of cones are still present in cpfl1 retina as shown in Figure 1 (52.91% ± 7.18 SEM, n = 7). Treatment with 10 nM TSA increased the percentage of surviving cones in cpfl1 explants to wt levels (109.76% ± 10.34, n = 6, P= 0.0023), indicating near complete cone protection. On the other hand, TSA had no significant effect on cone photoreceptors in wt cultures (110.27% ± 23.18 SEM, n = 3).

Figure 1.

HDAC inhibition protects cone photoreceptors in vitro. Long term retinal explant cultures from cpfl1 and wt animals were subjected to 10 nM TSA or control condition from PN14 to PN24. Glyphos staining (green), specific for cone photoreceptors, suggested that few cones survived at PN24 in untreated cpfl1 explant cultures (A), while TSA treated retinas displayed a clear increase in the numbers of cones (B). TSA treatment showed no adverse effect on wt treated retinas (C, D). The protective effect of TSA on cones is summarized in the bar graph (E). Error bars represent SEM; Scale bars are 20 µm. ONL-outer nuclear layer containing photoreceptors.

Intravitreal TSA injection protects degenerating cones in vivo

These promising in vitro results prompted us to further evaluate the protective effects of HDAC inhibition in vivo. To guide and optimize the in vivo treatment scheme, we first tried to predict the potential half-life of the drug inside the eye. Model calculations suggested that intravitreal clearance of TSA in the rabbit eye was 0.478 ml/h, and the half-life was in the range of 1.7–3.3 h. In the mouse, the calculated intravitreal half-life of TSA was an order of magnitude shorter (about 17 min), and more than 90% of the dose was expected to be eliminated in 1 h after injection (Supplementary Material, Fig. S2 and Note 1).

Since our in vitro experiments suggested that 10 nM TSA was efficiently protecting cones, we wanted to test if the same concentration would be effective also in vivo. As controls, we included both non-treated and sham treated eyes in the analysis. At PN14, nine animals from three different litters received an injection of 0.5 µl of a 100 nM TSA solution in one eye, with an estimated final TSA concentration of 10 nM. The contralateral eyes were sham injected with vehicle to control for injection-specific effects. Mice were sacrificed and analysed 10 days later, at PN24. Consistent with our in vitro results, TSA significantly increased the percentage of surviving cones in treated vs. sham treated eyes (cpfl1 sham: 84.92% ± 3.1 SEM, n = 9; cpfl1 treated: 99.88% ± 4.83, n = 9, P = 0.026; Fig. 2). The percentage of cones in sham injected eyes was comparable to the percentage of cones in non-treated cpfl1 mice (cpfl1: 78.24% ± 2.53 SEM, n = 8; cpfl1 sham: 84.92% ± 3.1, n = 9), suggesting that the injection itself had no effect on cone survival. The number of cones in sham- or 10 nM TSA- treated wt retinas were similar to non-treated wt retinas confirming the in vitro observation that TSA in the studied dosage had no deleterious effect on healthy cones (wt sham: 102.37% ± 4.72, n = 8; wt treated: 106.11% ± 7.02, n = 8, Fig. 2). The in vivo results suggested a remarkable protection of degenerating cones by a single intravitreal injection of 10 nM TSA with the number of cones in treated retinas reaching 99.8% of wt levels. To then test if TSA could efficiently protect degenerating cones at even lower concentrations, we performed a single intravitreal injection of TSA with a final estimated concentration of 1 nM, at PN14, and evaluated the protective effects at PN24. Remarkably, 1 nM TSA single treatment yielded a protective effect comparable to the 10 nM dose (Fig. 2, cpfl1 sham: 83.24% ± 2.92 SEM, n = 10; cpfl1 treated: 98.3% ± 5.25, n = 11, P = 0.03).

Figure 2.

HDAC inhibition protects degenerating cones in vivo. A single intravitreal injection of 1 nM or 10 nM TSA at PN14 was sufficient to significantly increase the number of cpfl1 mutant cones in treated compared to sham injected retinas at PN24 (A, B). In addition, cones from treated retinas were positioned mainly in the upper part of the ONL and had visibly longer inner segments (arrows), compared to sham treatment. In vivo treatment of wt animals showed no effects on the cone number, cone positioning, or structure (C, D). Quantification of the cone percentage of non-treated, sham treated, and TSA treated (1 nM and 10 nM) cpfl1 and wt animals is shown in the bar graph (E). Remarkably, the percentage of cones present was similar to wt for both TSA concentrations. Scale bar in a-d is 20 µm. ONL-outer nuclear layer. INL-inner nuclear layer containing interneurons. GCL-ganglion cell layer.

To assess the protective effect of HDAC inhibition at longer time points, we treated the cpfl1 animals with a single intravitreal injection of 10 nM TSA and assessed cone survival at PN30, i.e. after the peak of degeneration. When compared to untreated PN30 wt retina, cone loss in sham-treated cpfl1 had progressed further, with 71.6% ± 3.77 cones remaining (Fig. 3). In contrast, in 10 nM TSA treated eyes, even 16 days after a single intravitreal injection, the percentage of cones was significantly higher than in sham-treated, with 88.74% ± 2.79 cones still present (n = 5, P = 0.007). However, the protective effect at PN30 was less pronounced than at PN24, with the number of surviving cones lower than in wt retina (P = 0.03).

Figure 3.

Cpfl1 cones are protected also at PN30. Cone survival was analysed in animals which had received a single intravitreal injection at PN14. At PN30 (i.e. 16 days post-injection) sham-treated cpfl1 retina displayed a strong loss of cone-arrestin-labeled cones (green; A). By comparison, the 10nM TSA treatment preserved almost 90% of cones (B), similar to what was observed in age-matched wt (C). The protective effect of TSA on cones is summarized in the bar graph (D). Error bars represent SEM; Scale bars are 20 µm. ONL-outer nuclear layer.

To understand whether the observed increase in the cone numbers after TSA treatment was due to an indirect effect of a reduced clearance by microglia, we assessed the status of microglia using a staining against ionized-calcium-binding-adapter molecule 1 (IBA1) (18). Although activated microglia appeared to be present to some extent in the different layers of the retina, we neither detected a microglial invasion of the ONL, nor differences in IBA1 positive cells, with or without TSA treatment (Supplementary Material, Fig. S3). Alternatively, the protective effect of TSA might have been related to a delay of cone maturation in cpfl1 animals. To address this possibility, we also characterized the status of cone differentiation by looking at the expression and localization of M- and S-opsins, as well as cone-specific transducin (GNAT2), previously reported to be abnormally distributed in degenerating cones (19). Although M-opsin was seen in untreated cone outer segments, cpfl1 cones frequently showed opsin mislocalization to the cell body (Fig. 4G, arrows) and synaptic terminals (Fig. 4G, arrowheads). S-opsin staining revealed short outer segments as well as opsin mislocalization (Fig. 4H, arrows) in untreated cpfl1 cones and a similar but more pronounced mislocalization was observed for GNAT2 (Fig. 4I, arrows). In contrast, double staining for M- and S-opsin with Glyphos indicated that cones protected by TSA had normal morphology, with clearly developed inner and outer segments (Fig. 4J–L and P–R). In addition, the correctly positioned cones in the upper part of the ONL show both M- and S-opsin properly localized in their outer segments (Fig. 4J and K). Similarly, in surviving cones GNAT2 was properly localized in outer segments Fig. 4L. The correct localization of cone opsins and cone transducin suggested that TSA-mediated cone preservation had no obvious adverse effects on cone development and maturation.

Figure 4.

Cone specific opsins and transducin are expressed and correctly positioned in protected cones. In untreated PN24 cpfl1 retinas, cones were stained with Glyphos (green, A–C, D–F), and colocalized with M-opsin (magenta, G; merged, M), which was localized appropriately in only some of the cone outer segments, mainly in those positioned in the upper parts of the ONL. In the remaining cone cells, M-opsin was also mislocalized in what seemed to be cytoplasm of misplaced cones close to the border with the INL (arrows) and in cone end feet (arrowheads). A similar mislocalization was observed also with antibodies directed against S-opsin (H, N) and transducin (GNAT2; arrows; I, O). After TSA treatment, not only were there more cones, but also M-opsin (J, P), S-opsin (K, Q), as well as GNAT2 (L, R) were correctly localized almost exclusively in outer segments (P-R). Scale bar is 20 µm. ONL-outer nuclear layer.

HDAC inhibition significantly improves impaired cone migration

We previously reported that cpfl1 retina is characterized by impaired migration of developing cones (3). Glyphos staining of cones in degenerating cpfl1 retina confirmed such aberrant cone positioning, both in vitro and in vivo (Fig. 5). Importantly, we observed an improved migration of cones after both in vitro and in vivo TSA treatments of retinas (Fig. 5). To evaluate the extent of the cone migration improvement, we measured the distance of the centre of individual cone nuclei from the outer plexiform layer (OPL, Fig. 5, dashed red lines), which demarcates the lower boundary of the photoreceptor area. The percentage of cone migration distance from the OPL is presented in relative terms to the thickness of the ONL. In the in vitro situation, at PN24, wt cones migrated approximately to 70% of the ONL thickness (wt untreated: 69.25% ± 2.3 SEM, n = 3; wt treated: 72.66% ± 5.1, n = 3), while in non-treated cpfl1 retinas they reached only 55%. After treatment with 10 nM TSA of cpfl1 retinas in explant cultures, cone nuclei reached a position very similar to that of wt cones (cpfl1 untreated: 55.31% ± 3.1 SEM, n = 6; cpfl1 treated: 65.95% ± 2.8, P = 0.029). Similarly, cone migration was also improved in cpfl1 retinas in vivo, after treatment with 1 nM TSA. Here, cones were localized significantly more distant from the outer plexiform layer (cpfl1 untreated: 71.68% ± 1.09 SEM, n = 10, cpfl1 treated: 77.18% ± 0.84, n = 11, P = 0.02). Comparable results were obtained after treatment with 10 nM TSA (cpfl1 untreated: 67.47% ± 1.92 SEM, n = 9, cpfl1 treated: 76.38% ± 0.73, n = 9, P = 0.0005). We did not observe changes in cone localization in wt retinas after TSA treatment (wt untreated: 86.89% ± 0.26 SEM, n = 7; wt treated: 85.49% ± 0.23, n = 7). Nevertheless, even though TSA treatment significantly improved cpfl1 cone migration, cone positioning was still significantly different compared to wt cones (cpfl1 1 nM TSA: 77.18% ± 0.84 SEM vs. wt untreated: 86.89% ± 0.26, P = 0.0002; cpfl1 10nM: 76.38% ± 0.73 vs. wt untreated: 86.89% ± 0.26, P = 0.0004).

Figure 5.

HDAC inhibition improves cone migration. Cpfl1 retina displayed cone mislocalization, both in vitro and in vivo, when compared to wt (A-C). At PN24, cone nuclei were scattered throughout the ONL in cultured (A) and in vivo cpfl1 mutant retinas (B), while in wild-type retinas cones were positioned exclusively in the upper part of the ONL (C). TSA treatment significantly improved cone migration, both in the in vitro (D) and in vivo (E) treatment paradigms, while no effect was detectable in wt treated retinas (F). For quantification, the cone migration distance was assessed by measuring the distance between the center of cone nuclei from the outer plexiform layer (dashed line), relative to ONL thickness, measured for each individual section of analysed retinas. The quantifications show a significant improvement of cone migration in vitro (G) and in vivo (H), with TSA treatment. Scale bars are 20 µm. ONL-outer nuclear layer.

We previously found high cGMP accumulation and increased activity of cGMP-dependent protein kinase G (PKG) to be associated with cpfl1 cone phenotypes (3). To investigate whether HDAC inhibition had an effect on cGMP and PKG activity after ten days of treatment with TSA, we looked for cGMP accumulation and the phosphorylation status of a well-known PKG substrate, vasodilator-stimulated-protein (pVASP) (20), with and without TSA treatment. Untreated cpfl1 retinas showed cGMP accumulation in cone segments as well as in cone cell bodies (as evidenced by Glyphos co-labeling). There were no obvious differences in cone cGMP accumulation after TSA treatment. (Supplementary Material, Fig. S4A). Likewise, some cones, visualized by Glyphos staining, were also positive for pVASP in both TSA treated and untreated cpfl1 retinas, with no evident differences between the two (Fig. S4B). Taken together, these results indicated that HDAC activity (and TSA inhibition) operated down-stream of the mutation-induced cGMP accumulation.

Discussion

Hereditary diseases of cone photoreceptors are currently untreatable and lead to major visual impairment and blindness. Our study in the cpfl1 mouse model demonstrates prominent cone protection after HDAC inhibition, in vitro and in vivo, and thus confirms that HDAC activity is causally involved in hereditary cone photoreceptor degeneration. This provides a proof-of-principle for the further development of HDAC inhibition as a novel therapeutic approach for cone dystrophies.

HDAC inhibition was found to induce cell death in tumour cells derived from various cancers (21), while in neurodegenerative diseases, the same treatment may prolong cell survival (22). In the retina, HDAC inhibitors have been discussed as a potential therapeutic strategy for ischemic retinal injury, with different HDAC inhibitors affording structural and functional neuroprotection after experimental ischemia (23). Furthermore, HDAC inhibition was shown to slow rod photoreceptor degeneration in an animal model for Retinitis Pigmentosa (24). Importantly, current clinical trials are suggesting beneficial effects of treatments with an HDAC inhibitor in patients suffering from Retinitis Pigmentosa (25,26). We show here for the first time that HDAC inhibition significantly protects cones in a mouse model for hereditary cone dystrophy. Besides its implication for future treatment of rare forms of retinal degeneration, the results presented here may also extend to common diseases of the retina, including DR and AMD, where cone degeneration is the cause of legal blindness.

An important result from our study is that a single intraocular application significantly protected cpfl1 cones for at least two weeks post-injection. This is unexpected, as our calculations on TSA clearance suggested that the drug was below threshold concentration, after less than 2 h and may point to an imprinting mechanism, beyond the transient presence of TSA. This is in line with previous studies which found that TSA leads to robust protein acetylation after 3 h (13) or to transcriptional changes after only 5 min (27), supporting the possibility of epigenetic mechanisms. The molecular mechanism underlying treatments with HDAC inhibitors are still not well characterized. In cpfl1 retina, a previous study found an upregulation of STAT3 signalling which was suggested as an endogenous neuroprotective response (28). Interestingly, STAT3 activation may be controlled by HDAC activity (29).

Importantly, the TSA clearance estimation suggested that cone degeneration was halted already after a short-time intraocular exposure. This observation could be especially relevant for patients with retinal dystrophies where monthly intravitreal injections are usually well tolerated (30). Since in mice the protective effect was present for at least 10 days post-injection, a rough extrapolation from mouse to man, would suggest that in humans an intravitreal dosing once per three months might be sufficient (31). Alternatively, formulations for local application to the eye would minimize the potential side-effects observed in some cases of long-term, systemic HDAC inhibition (26,32,33).

Cone photoreceptor development is characterized by a migration within the ONL, from the outer limiting membrane inwards to the OPL, and then outwards, until they reach their final position just below the outer limiting membrane, from PN12 onwards (34). Cone degeneration in mouse models is characterized not only by cone loss, but also by improper developmental cone migration (3,35). In fact, delayed cone migration was also reported in mouse models with rod degeneration as an indirect consequence of rod cell death (34). Migration of cone photoreceptors also occurs in the human retinal development, where in the fovea cones are migrating laterally towards the foveal pit (36). Cone misplacement has been associated with human retinal diseases of different aetiology as improper cone migration can lead to foveal hypoplasia (37). This is observed, for instance, in patients suffering from albinism (38), idiopathic congenital nystagmus (37), and also from achromatopsia (39). In addition, cone misplacement was reported in aged patients suffering from AMD (40). We found that TSA treatment not only prevented cone degeneration but also significantly improved aberrant cone migration in degenerating cpfl1 retina. There are numerous reports on the effects of HDAC inhibition of cell migration. The effects of HDAC inhibition on improved cell migration could take place via inhibition of the cytoplasmic histone deacetylase (HDAC6) resulting in higher acetylation of α-tubulin necessary for stability of microtubules in migrating cells (41). Alternatively, HDAC inhibition could lead to epigenetic changes governing the migration of photoreceptors, a process still not fully understood. TSA treatment significantly improved the migration of misplaced cpfl1 cones both in vitro and in vivo. Nevertheless, HDAC inhibition alone was not sufficient to fully repair the cone migration defect in cpfl1 retinas. A plausible explanation could relate to increased PKG activity in cpfl1 retina, as we previously reported (3). PKG-specific phosphorylation of the serine 239 residue of VASP was shown to negatively regulate neuronal migration (42). Proteins of the Ena/VASP family are regulators of actin assembly and cell motility as they are localized in focal adhesions (43). cGMP-dependent PKG overactivation leads to VASP phosphorylation, which will result in removal of VASP from focal adhesions, contributing to altered cell migration (44). Our data suggest that TSA inhibition of HDACs did not reduce the VASP phosphorylation of serine 239 as assessed by immunostaining. The possible dual regulation of cone photoreceptor migration, one via PKG overactivation and the other involving aberrant HDAC activity, will require further specific studies. In addition, in the cpfl1 mouse model, the mutation-induced loss of cone photoreceptors coincides in part with the postnatal retinal development (3,45). As the TSA treatment during the third postnatal week overlaps in time with the establishment of retinal connectivity (46), further studies (i.e. rod ERG analysis) are also necessary to define how restoration of the normal developmental cone migration pattern under the influence of HDAC inhibitors may alter rod connectivity and function.

Degenerating cones in cpfl1 retina are characterized by cGMP accumulation as a consequence of the non-functional cone photoreceptor PDE6 enzyme (3,4). We did not observe obvious changes in the numbers of cells showing cGMP accumulation after the treatment, indicating that TSA-driven protection of cones probably operates down-stream in the degenerative pathway. Increasing lines of evidence suggest PKG as one of the main initiators of a photoreceptor cell death cascade (4,6,47). Previously, we found that increased PKG and HDAC activities were temporally connected in 10 different animal models for inherited retinal degeneration (4). This corresponds to observations in C. elegans where PKG activity was found to regulate HDAC activation (48). Since HDAC inhibition did not seem to affect VASP phosphorylation in cones, our study indicates that during cell death HDAC activity may occur independently of- or downstream of PKG. The underlying mechanisms through which HDAC inhibition offers cone protection remains to be determined.

HDAC activity is also crucial for mouse photoreceptor development as HDAC inhibition at early post-natal retinal development (PN2) leads to a complete loss of developing rod photoreceptors (49). In the mouse retina cone photoreceptors are born prenatally (45), therefore we do not expect an effect of TSA on cone birth, but the full maturation and commitment towards blue cones (S cones) or red/green cones (M cones) takes place during the second postnatal week (50). Since we performed HDAC inhibition just after this time-period, we assessed the expression and localization of S- and M-opsin, to address whether the observed increase in cone survival following HDAC inhibition was due to cone development and maturation delays. In addition, we checked for the cone specific transducin (GNAT2), a member of the cone phototransduction cascade (51). The observed correct localization of cone opsins and transducin suggests that late postnatal HDAC inhibition did not affect cone cell development and maturation. However, we cannot exclude the possibility that the rapid degeneration of cpfl1 cones was in part caused by developmental defects. In this case TSA treatment, in a critical time-frame, may have supported normal development as the observed effects lasted for more than two weeks. Future studies, for instance on more slowly progressing cone degeneration models, may reveal whether TSA and HDAC inhibition mediated cone protection can be applied to cone degeneration in general, irrespective of developmental stage and disease pathogenesis.

In summary, this study is the first to show that HDAC inhibition can significantly delay cone loss in a mouse model for a human cone dystrophy. While the effect of TSA on cpfl1 retina in vivo lasts until at least one month post-natal, further studies will be needed to determine the duration of the protection and to establish long-term treatment paradigms. Remarkably, HDAC inhibition appeared to have a dual effect on both cone survival and cone migration without obvious effects on cone differentiation and development. Our study thus provides a proof-of-principle for the potential of HDAC inhibitors as therapeutic agents for cone photoreceptor protection. Importantly, this may be applicable not only to hereditary cone dystrophies, but possibly also to common retinal diseases associated with cone degeneration such as diabetic retinopathy (52,53), as well to diseases where cone degeneration is accompanied by improper cone migration, such as age-related macular degeneration (40).

Materials and Methods

Animals

Cpfl1 and congenic C57BL/6 wild-type (wt) animals were housed under standard white cyclic lighting, had free access to food and water, and were used irrespective of gender. All procedures were approved by the Tübingen University committee on animal protection (Einrichtung für Tierschutz, Tierärztlichen Dienst und Labortierkunde) and performed in accordance with the ARVO statement for the use of animals in ophthalmic and visual research. Procedures performed at the polistab of the University of Modena and Reggio Emilia (in vivo treatment on cpfl1 and wt animals) were reviewed and approved by the local ethical committee (Prot. N. 50 12/03/2010). All efforts were made to minimize the number of animals used and their suffering.

Immunofluorescence

Retinae were fixed in 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (pH 7.4) for 45 min at 4 °C. Sagittal 12 µm sections were collected, air-dried and stored at -20 °C. Sections were incubated overnight at 4 °C with primary antibodies. The primary antibodies used were specific for Glycogen phosphorylase (Glyphos, 1:1000 dilution, kindly provided by Prof. Pfeiffer-Guglielmi (17)), cone arrestin (1:1000, AB15282, Millipore, Darmstadt, Germany), IBA1 (1:200, 019-19741, Wako, Richmond, USA), M-opsin (1:200, AB5405, Millipore), S-opsin (1:200, AB5407, Millipore), GNAT2 (1:200, sc-390, Santa Cruz, Dallas, USA), cGMP (1:500, kindly provided by Prof. Steinbusch, University of Maastricht, The Netherlands (54)) and phospho-VASP (Ser239) (1:200 dilution, 0047-100/VASP-16C2, Nanotools, Teningen, Germany (3,44)). Alexa Fluor 488- or 566-conjugated were used as secondary antibodies (Molecular Probes, Inc. Eugene, USA). Negative controls were carried out by omitting the primary antibody. Specificity of cone labeling with Glyphos was shown by co-labelling with peanut-agglutinin (Supplementary Material, Fig. S5), a well-established marker for cones (55).

HDAC in situ activity assay

HDAC activity assays were performed on cryosections of 4% PFA- fixed eyes. The assay is based on an adaptation of the Fluor de Lys Fluorescent Assay System (Biomol, Hamburg, Germany). Retina sections were exposed to 200 μM Fluor de Lys-SIRT2 deacetylase substrate (Biomol) with 500 μM NAD+ (Biomol) in assay buffer (50 mM Tris/HCl, 137 mM NaCl; 2.7 mM KCl; 1 mM MgCl2; pH 8.0) for 3 h at room temperature (8). Sections were then washed in PBS and fixed in methanol at -20 °C for 20 min. Slides were mounted with developer (Biomol; KI105, diluted 1:20 in assay buffer) containing 2 μM TSA (Sigma, Steinheim, Germany) and 2 mM NAM (Sigma). Negative controls consisted of omitting the substrate (Supplementary Material, Fig. S1E).

Retinal explant cultures

Organotypic retinal cultures from cpfl1 (n = 7) and wt (n = 4) animals that included the retinal pigment epithelium (RPE) were prepared under sterile conditions. Briefly, PN14 animals were sacrificed, the eyes enucleated and pretreated with 0.12% proteinase K (ICN Biomedicals Inc., OH, USA) for 15 min at 37 °C in R16 serum-free culture medium (Order No.: 074-90743A; Gibco, Paisley, UK). Proteinase K activity was blocked by the addition of 10% fetal bovine serum, followed by rinsing in serum-free medium. Following this, the cornea, lens, sclera and choroid were removed carefully, with only the RPE remaining attached to the retina. The explant was then cut into four wedges, to give a clover-leaf like structure, which was transferred to a culture membrane insert (Corning Life Sciences, Lowell, USA) with the RPE facing the membrane. The membrane inserts were placed into six well culture plates with R16 medium (Gibco, Paisley, UK) and incubated at 37 °C in a humidified 5% CO₂ incubator. The culture medium was changed every 2 days during the 10 culturing days.

Retinal explants were left without treatment for 2 days (until PN16), followed by 10 nM Trichostatin A (TSA) treatment (≥98% (HPLC), from Streptomyces sp., Sigma-Aldrich, St-Louis, USA). TSA was dissolved in 0.2% dimethyl sulfoxide (DMSO; Sigma) and diluted in R16 culture medium. For controls, the same amount of DMSO was diluted in culture medium. Culturing was stopped at PN24 by 2 h fixation in 4% PFA, cryoprotected with graded sucrose solutions containing 10, 20, and 30% sucrose and then embedded in the tissue freezing medium (Leica Microsystems Nussloch GmbH, Nussloch, Germany).

In vivo injections

Animals were anaesthetized with an intraperitoneal injection of Avertin (250 mg/kg) (1.25% (w/v) 2,2,2-tribromoethanol and 2.5% (v/v) 2-methyl-2-butanol; Sigma, Milan, IT) and kept warm during injections. Single intravitreal injections were performed at PN14 on one eye while the other eye was sham injected with 0.0001% DMSO (in 0.9% NaCl solution) as contralateral control. TSA (1 and 10 nM) was diluted in 0.9% NaCl solution. Injections were performed with 0.5 µl of 10 nM and 100 nM TSA in order to have a final concentration of 1 nM and 10 nM, respectively, assuming the free intraocular volume of mouse eye to be 5µl (http://prometheus.med.utah.edu/∼marclab/protocols.html; date last accessed May 2016). Eleven cpfl1 animals and eight wt from three different litters were used for intravitreal injections and were sacrificed 10 days after treatment (PN24). For the treatment of cpfl1 animals until PN30 injections were performed at PN14 in five different animals with one eye being injected with 10 nM TSA, while the other eye was sham injected. Eyes were immediately enucleated, fixed for 2 h in 4% PFA and prepared for cryosectioning (see ‘Retinal explant cultures’). Sections were labelled in a non-conclusive fashion to allow for an investigator-blind analysis.

Microscopy, cell counting, and statistical analysis

Fluorescence microscopy was performed on an Axio Imager Z1 ApoTome Microscope, equipped with a Zeiss Axiocam MRm digital camera. Images were captured using Zeiss Axiovision 4.7 software. Adobe Photoshop CS3 (Adobe Systems Incorporated, San Jose, CA, USA) was used for primary image processing. Quantifications were performed on pictures captured on at least nine different random positions of at least three sagittal sections for at least four different animals for each genotype and treatment using Z-stacks mode of Axiovision 4.7 at 20x magnification. The average area occupied by a photoreceptor cell (i.e. cell size) for each individual eye was determined by counting DAPI-stained nuclei in nine different areas (50 x 50 µm) of the retina. The total number of photoreceptor cells was estimated by dividing the outer nuclear layer (ONL) area by this average cell size. The quantification of cones was performed by manually counting the number of positively labelled cones in the ONL. Values obtained are given as a fraction of total cell number in ONL (i.e. as percentage) and normalized to the wt situation. Values are expressed as average ± standard error of the mean (SEM). For statistical comparisons the two-tailed, unpaired Student t-test as implemented in Prism 6 for Windows (GraphPad Software, La Jolla, CA, USA) was employed.

To assess differences in cone migration, the distance in µm between the outer plexiform layer (OPL) and the centre of Glyphos positive cell bodies was measured using Axiovision software (Zeiss). Distance values for 150–250 cones in the entire retinas were averaged on Glyphos immunostained sections from at least 5 different animals for each genotype. The migration profile of cones was presented as the relative migration distance to ONL thickness measured using Axiovision software. Statistical differences between experimental groups were calculated using Student’s two-tailed, unpaired t-test and Microsoft Excel software (Microsoft, Seattle, WA, USA). Error bars in the figures indicate SEM, levels of significance were: * = P < 0.05, ** =  P < 0.01, *** =  P < 0.001.

Assessment of intravitreal clearance of TSA

The intravitreal clearance (CL_ivt) of TSA was calculated in silico using the Quantitative structure-property relationships (QSPR) model published recently (56) based on comprehensive rabbit data from intravitreal injection experiments: LogCL_ivt= −0.25269–0.53747 (LogHD) + 0.05189 (LogD_7.4), where HD is the number of hydrogen bond donor atoms and LogD_7.4 is the calculated n-octanol/water distribution coefficient at pH 7.4 of the compound. The model is built based on small molecular weight compounds using the linear multivariate analysis tools: principal component analysis (PCA) and linear partial least square (PLS) (Simca Plus, version 10.5, Umetrics AB, Umea, Sweden). Firstly, the chemical structure of the TSA was retrieved from ACD/Dictionary from ACDlabs software (version 12, Advanced Chemistry Development, Inc., Toronto, Canada) and was used as input in ACDlabs software to generate 30 molecular descriptors: pK_a for the most acidic molecular form, pK_a for the most basic form, LogD at pH 5.5 and 7.4, LogP, MW, PSA (polar surface area), FRB (freely rotatable bonds), HD (hydrogen bond donors), HA (hydrogen bond acceptors), Htot (HD + HA), rule of 5, molar refractivity, molar volume, parachor, index of refraction, surface tension, density, polarizability, C ratio, N ratio, NO ratio, hetero ratio, halogen ratio, number of rings and number of aromatic, 3-, 4-, 5- and 6-membered rings. The applicability domain of the intravitreal clearance model for TSA was inspected generating the PCA score plot of the training set of the model together with TSA (Supplementary Material, Fig. S2). Compounds that lie inside the ellipse depicted in the plot belong to the same chemical space of the model and are predictable by the model. Once the applicability domain was confirmed, the intravitreal clearance value of TSA was calculated with the above equation using the descriptor values (LogD at pH 7.4 and HD). Half-life was calculated using equation t_1/2 = ln2 V_d/CL, where V_d is the volume of distribution and CL is the intravitreal clearance. For a brief discussion on the transferability between rabbit and mouse data, see supplementary note 1.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We are grateful to Prof. E. Zrenner for fruitful discussions and support, we also thank K. Masarini and N. Rieger for skilful technical assistance, and M. Power for critically reading of the manuscript.

Conflict of Interest Statement. None declared.

Funding

This work was supported by the Kerstan Foundation, Deutsche Forschungsgemeinschaft [DFG TR 1238/4-1, DFG PA1751/4-1], Alcon Research Institute, European Commission [DRUGSFORD: HEALTH-F2-2012-304963], German Ministry of Education and Research [BMBF HOPE2 FKZ 01GM1108A] and European Union s Horizon 2020 Research and Innovation Programme under Grant Agreement No 634479 (EyeRisk).

Author notes

†

The authors wish it to be known that, in their opinion, the first 2 authors should be regarded as joint First Authors.