Loss of HCN1 enhances disease progression in mouse models of CNG channel-linked retinitis pigmentosa and achromatopsia

Abstract

Most inherited blinding diseases are characterized by compromised retinal function and progressive degeneration of photoreceptors. However, the factors that affect the life span of photoreceptors in such degenerative retinal diseases are rather poorly understood. Here, we explore the role of hyperpolarization-activated cyclic nucleotide-gated channel 1 (HCN1) in this context. HCN1 is known to adjust retinal function under mesopic conditions, and although it is expressed at high levels in rod and cone photoreceptor inner segments, no association with any retinal disorder has yet been found. We investigated the effects of an additional genetic deletion of HCN1 on the function and survival of photoreceptors in a mouse model of CNGB1-linked retinitis pigmentosa (RP). We found that the absence of HCN1 in Cngb1 knockout (KO) mice exacerbated photoreceptor degeneration. The deleterious effect was reduced by expression of HCN1 using a viral vector. Moreover, pharmacological inhibition of HCN1 also enhanced rod degeneration in Cngb1 KO mice. Patch-clamp recordings revealed that the membrane potentials of Cngb1 KO and Cngb1/Hcn1 double-KO rods were both significantly depolarized. We also found evidence for altered calcium homeostasis and increased activation of the protease calpain in Cngb1/Hcn1 double-KO mice. Finally, the deletion of HCN1 also exacerbated degeneration of cone photoreceptors in a mouse model of CNGA3-linked achromatopsia. Our results identify HCN1 as a major modifier of photoreceptor degeneration and suggest that pharmacological inhibition of HCN channels may enhance disease progression in RP and achromatopsia patients.

Introduction

Retinal neurodegeneration is a key feature of many inherited blinding eye diseases with high clinical and socioeconomic impact. The degenerative process often concerns the photoreceptors and can be of cell autonomous or non-cell autonomous nature. A large number of disease causing mutations have been identified (https://sph.uth.edu/retnet) and some mechanistic insights on disease-related functional defects [e.g. lack of photoresponse in the absence of cyclic nucleotide-gated (CNG) channels (1)] have been gained (2–4).

Photoreceptors are retinal neurons specialized on the detection of light and the translation of light-encoded information into electrical activity in a process called phototransduction. The photoreceptor membrane potential is controlled by CNG channels found in the plasma membrane of photoreceptor outer segments (5). Moreover, a number of proteins mediate or modulate additional ion conductance and help in adjusting the electrical properties of photoreceptors (6–8). One of these proteins is the hyperpolarization-activated cyclic nucleotide-gated (HCN) channel found in photoreceptor inner segments. Four HCN channel genes (Hcn1–4) exist (9) and all of them are expressed in the mammalian retina with distinct patterns of localization (10). Among the Hcn channel genes, Hcn1 shows the highest expression levels in photoreceptors and, overall, is one of the most highly expressed retinal genes (11). The HCN1 channel is strongly enriched in rod and cone photoreceptor inner segments where it contributes to shaping the photoresponse (12–14). In line with this, genetic deletion of Hcn1 in mice results in prolongation of the scotopic and photopic electroretinogram (ERG) responses (14). Moreover, in the absence of HCN1 sustained rod responses after bright light illumination saturate the retinal network and impair downstream cone signaling (15). Systemic pharmacological inhibition of HCN channels in animal models also results in characteristic changes in the ERG responses (16,17). Ivabradine, the first clinically approved hyperpolarization activated current (I_h) inhibitor acting on HCN channels is used for the treatment of stable angina pectoris (18). In agreement with the functional role of HCN1 in shaping the retinal photoresponse, ∼15% of the patients report the occurrence of light-stimulus-independent visual sensations, so called phosphenes, under ivabradine medication (19,20). Therefore, it is suggested that caution should be exercised when prescribing ivabradine to patients with chronic retinal diseases like retinitis pigmentosa (RP). RP is a hereditary ocular disorder characterized by progressive degeneration of rod photoreceptors. Secondary to rods, cone photoreceptors also degenerate by an unknown non-cell-autonomous mechanism (21). The disease progression can vary significantly between patients, and the severity mainly depends on the nature of the disease causing gene mutation, but can also be influenced by additional confounding factors.

We hypothesized that proteins controlling the electrical properties of photoreceptors might also influence the disease progression. Given the high levels of expression and its important physiological role in retinal photoreceptors HCN1 was our major candidate. To analyze the importance of HCN1 for photoreceptor cell viability in the context of degenerative retinal diseases we cross-bred retinal degeneration mouse models with mice lacking the Hcn1 gene and followed the disease progression using in vivo and in vitro techniques. In addition, we explored the effect of pharmacological inhibition of HCN channels in retinal degeneration.

We show that genetic deletion or pharmacological inhibition of HCN1 in degenerating photoreceptors dramatically enhances the disease progression suggesting that HCN1 is an important factor that counteracts degeneration of photoreceptors.

Results

Loss of Hcn1 enhances photoreceptor degeneration in Cngb1 knockout mice

To test whether HCN1 influences the function and structure of degenerating rod photoreceptors, we cross-bred Hcn1 KO mice with the Cngb1 KO mouse model of RP (22) to obtain Cngb1 × Hcn1 double-KO (Cngb1/Hcn1 DKO) mice. We first performed ERG measurements to assess the effect of Hcn1 deletion on the photoresponse in Cngb1 KO mice. We recorded both scotopic and photopic ERGs in 4-month-old Cngb1/Hcn1 DKO mice and found an almost complete loss of rod- and cone-driven responses (Fig. 1A). This result was quite surprising since the deletion of Cngb1 alone strongly impairs the rod-driven ERG responses whereas cone-driven ERG responses are still detectable at this age (Fig. 1A). To further elucidate this inconsistency, we recorded ERGs in 4-week-old Cngb1/Hcn1 DKO and age-matched Cngb1 KO control mice. As shown in Figure 1B, both rod- and cone-driven responses were present in 4-week-old Cngb1/Hcn1 DKO mice, but amplitudes were smaller than those in 4-week-old Cngb1 KO control mice (Fig. 1B–C). These findings suggest a faster or exacerbated degeneration of both rod and cone photoreceptors in the absence of HCN1. This phenomenon was investigated further at the level of retinal morphology. First, we examined the morphology of rod and cone photoreceptors in the retinas of 90-day-old Cngb1/Hcn1 DKO mice. We labeled retinal cross sections with a peripherin-2 (Prph2)-specific antibody to reveal rod morphology and peanut agglutinin (PNA) to mark cones. Both markers had a similar appearance in wild-type and Hcn1 KO mice (Fig. 2A) confirming that HCN1 channel deletion alone has no effect on photoreceptor morphology. In 90-day-old Cngb1 KO mice both morphological markers were preserved (Fig. 2A). However, some signs of a moderate degeneration were already evident, including a reduction in both rod outer segment length and photoreceptor nuclear layer thickness [Fig. 2A and (22)]. In striking contrast, the photoreceptor layer in Cngb1/Hcn1 DKO mice of the same age was already reduced to only 1–2 rows, and rod and cone cell markers were barely preserved (Fig. 2A). In comparison, the retina of a 28-day-old Cngb1/Hcn1 DKO showed great similarity with that of a 90-day-old Cngb1 KO mouse (Fig. 2A).

Figure 1.

KO of Hcn1 leads to premature vision loss in Cngb1-deficient mice at 4 months. (A) Superimposed representative dark-adapted (scotopic, left) and light-adapted (photopic, right) single-flash ERG series of 4-month-old Cngb1 KO (blue traces) and Cngb1/Hcn1 DKO mice (red traces). (B) Superimposed representative dark-adapted (scotopic, left) and light-adapted (photopic, right) single-flash ERG series of 4-week-old Cngb1/Hcn1 DKO mice (green traces) and Cngb1 KO mice (black traces). The vertical line indicates the timing of the light stimulus in each panel. (C) Box-and-Whisker plots of the b-wave amplitudes plotted as a function of the logarithm of the flash luminance for entire groups: boxes indicate the 25 and 75% range, the whiskers the 5 and 95% quantiles, asterisks mark the median of the data.

Figure 2.

KO of Hcn1 enhances photoreceptor degeneration in Cngb1-deficient mice. (A) Retinal cryo-sections of wild-type, Hcn1 KO, Cngb1 KO and Cngb1/Hcn1 DKO mice stained with markers specific for rods (peripherin 2, red) and cones (peanut agglutinin, green). (B) Upper part: in vivo OCT scan of a wild-type retina illustrating the quantification of photoreceptor layer thickness. Lower part: representative OCT images revealing the progressive photoreceptor layer thinning in Cngb1 KO and Cngb1/Hcn1 DKO mice. (C) Quantification of photoreceptor layer thickness of the different genotypes from 2 to 22 weeks of age. Each data point represents one retina (n = 4–8 per genotype and time point). (D) Infrared images of the mouse fundus revealing a severe RPE atrophy in Cngb1/Hcn1 DKO at 5 months of age. onh, optic nerve head; onl, outer nuclear layer; os, outer segments.

To characterize the time course of degeneration, we applied optical coherence tomography (OCT), a powerful imaging technique that generates virtual cross sections through tissues and enables the visualization and quantification of retinal layer thickness (Fig. 2B) in vivo (23). We started imaging wild-type, Cngb1 KO and Cngb1/Hcn1 DKO mice right after eye opening (Day 14) and followed them up to 5 months of age. At postnatal day 14, the gross retinal layer morphology was similar in all three genotypes (Fig. 2B–C). In wild-type mice, the photoreceptor layer thickness did not substantially change over the observation period (Fig. 2C). At 5 months of age, the photoreceptor layer thickness in Cngb1 KO mice had decreased by almost one half to 49.50 ± 1.31 µm (n = 6) (Fig. 2C). In Cngb1/Hcn1 DKO mice, the disease progression was significantly faster (Fig. 2C). Three months after birth, the photoreceptor layer in the double knockout (DKO) was so much reduced that it could no longer be resolved by OCT (Fig. 2B). In contrast, the deletion of Hcn1 alone had no negative effect on retinal morphology and photoreceptor layer thickness (Supplementary Material, Fig. S1). Accordingly, the thickness of the photoreceptor layer in aged Hcn1 KO retina (100.2 ± 2.0 µm, n = 6) was similar to wild-type (101.8 ± 1.6, n = 6; Supplementary Material, Fig. S1). The examination of the ocular fundus using confocal laser scanning ophthalmoscopy (cSLO) confirmed the observed changes in 5-month-old Cngb1/Hcn1 DKO mice and revealed a marked atrophy of the retinal pigment epithelium (RPE), which was not observed in wild-type or Cngb1 KO mice (Fig. 2D).

Taken together, these findings point to a context-specific deleterious effect of a genetic deletion of Hcn1 on retinal morphology in the Cngb1 KO mouse model of RP. This effect manifests as an enhanced degeneration of rod photoreceptors, associated with an accelerated secondary degeneration of cone photoreceptors and RPE cells.

Pharmacological inhibition of I_h enhances photoreceptor degeneration in Cngb1 KO mice

To test if the disease-amplifying effect resulted specifically from the lack of HCN channel function we treated Cngb1 KO mice with the I_h current blocker zatebradine and analyzed the progression of photoreceptor degeneration using OCT. Wild-type and Cngb1 KO mice received daily i.p. injections of zatebradine (10 µg/g body weight/day) or vehicle for 10 consecutive days starting on postnatal day 11. The photoreceptor layer thickness was then assessed at Days 21, 60 and 90 (Fig. 3A). Confirming the morphological data from Hcn1 KO mice (Figure 2A and Supplementary Material, Fig. S1), there was no sign of retinal thinning in zatebradine-treated wild-type mice (Fig. 3B–C). Moreover, zatebradine had no short-term effects on the Cngb1 KO retina since the photoreceptor layer thickness was similar in treated and untreated KO mice at the end of the 10-day treatment period (Fig. 3C). However, at 40 and 70 days after zatebradine treatment we observed a significant reduction of the photoreceptor layer thickness in Cngb1 KO mice [Fig. 3B–C; two-way-analysis of variance (ANOVA), P < 0.001], supporting the view that the loss of HCN1 channel function in rod photoreceptors enhances the disease progression in the Cngb1 KO mouse model of RP.

Figure 3.

Pharmacological HCN inhibition accelerates the retinal degeneration of Cngb1 KO mice. (A) Cngb1 KO or wild-type mice were treated with the HCN inhibitor zatebradine (P11–20, daily 10 µg/g i.p.) or vehicle. Repeated OCT measurements were performed at Days 21, 60 and 90. (B) Retinal morphology of treated and non-treated wild-type and Cngb1 KO mice at Days 21 and 90. (C) Quantification of photoreceptor layer thickness of treatment groups at Days 21, 60 and 90. Each data point represents the mean of photoreceptor layer thickness (n = 4–8) ± SEM; ***P < 0.001. (D) Overview on a Cngb1/Hcn1 DKO retina (Week 6) injected with pAAV2.1-Rho-YFP-HCN1-WPRE at Day 14. The AAV-mediated HCN1 expression (indicated by YFP, green) substantially rescued photoreceptor degeneration in the treated part of the retina. The vertical bars mark the corresponding ONL thickness. gcl, ganglion cell layer; inl, inner nuclear layer; onl, outer nuclear layer. Scale bar in (D) marks 100 µm.

Adeno-associated viral vectors-mediated expression of HCN1 in Cngb1/Hcn1 DKO mice rescues photoreceptors from enhanced degeneration

To confirm that HCN1 function in rods is crucial for survival of Cngb1-deficient rod photoreceptors, we generated recombinant adeno-associated virus (AAV) vectors expressing the mouse HCN1 channel as an N-terminal yellow fluorescent protein (YFP)-fusion protein under control of the rod photoreceptor-specific rhodopsin promoter and delivered them into the subretinal space of 14-day-old Cngb1/Hcn1 DKO mice. YFP fluorescence, indicative for the YFP-HCN1 channel expression, became detectable 7-day post-injection in the treated part of the retina (data not shown). Importantly, 4 weeks later the treated (fluorescent) part of the Cngb1/Hcn1 DKO retina was substantially better preserved compared with the untreated (non-fluorescent) part (Fig. 3D). In particular, we found that AAV-mediated HCN1 expression resulted in preservation of approximately 2-fold more photoreceptor rows in the treated compared with the non-treated region of these retinas. This effect could be reproduced in three animals (Fig. 3D).

Effect on Hcn1 deletion on the photovoltage of degenerating rods

In healthy photoreceptors, HCN1 channels open in response to light-evoked hyperpolarization and contribute to shaping the photovoltage (12–14). Our results so far suggested that the disease-amplifying effect in degenerating Cngb1/Hcn1 DKO rod photoreceptors can be attributed to the missing function of the HCN1 channels. This prompted us to determine the range of membrane potentials in which mutant rods operate. To this end, we used a previously described perforated patch recording technique to current- or voltage-clamp rod photoreceptors in mouse retinal slices (24). As an important initial experiment, we compared wild-type and Hcn1 KO rod photoreceptors (Supplementary Material, Fig. S2). The mean dark membrane potential (V_dark) in adult wild-type rods was −35.4 ± 1.6 mV (mean ± SEM, n = 19). Hcn1 KO rods had similar V_dark values at −34.1 ± 2.0 mV (n = 4), confirming that HCN1 does not significantly contribute to the dark membrane potential of rod photoreceptors. Moreover, saturating light stimuli resulted in similar levels of hyperpolarization in wild-type −55.4 ± 2.2 mV (n = 19) and Hcn1 KO rods −54.2 ± 2.6 mV (n = 4).

To test whether HCN1 influences the electrical properties of degenerating rod photoreceptors, we next compared wild-type, Cngb1 KO and Cngb1/Hcn1 DKO rod photoreceptors. To exclude measuring end-stage degenerating photoreceptors, we performed the experiments on retinal slices from 15- to 20-day-old mice, which is prior to the peak of rod degeneration in Cngb1 KO mice (22,25). As expected I_h was present in wild-type and Cngb1 KO rods, but absent in DKO rods (Fig. 4A) confirming that HCN1 is the major HCN channel isoform in rod photoreceptors. Rod photoreceptors in young wild-type mice had a mean V_dark of −37.9 ± 1.1 mV (n = 5) (Fig. 4B). Although the V_dark of Cngb1 KO rods (−34.9 ± 0.9 mV, n = 14) was slightly more depolarized than in wild-type rods, there was no statistically significant difference between the groups (one-way ANOVA) (Fig. 4B). The additional deletion of HCN1 in Cngb1 KO mice did not change V_dark significantly (−36.4 ± 1.0 mV, n = 8) (Fig. 4B), arguing against a major role of HCN1 in setting the dark membrane potential in degenerating rods. Short light (flash) stimuli cause a fast hyperpolarization of the mouse rod photovoltage, which returns to the V_dark after a few seconds (24). The response of young wild-type rods to flash stimuli of varying luminance is shown in Figure 4C. In wild-type rods the saturating flash stimulus caused a maximal hyperpolarization of about −15 mV in amplitude to −53.4 ± 2.5 mV (n = 5) (Fig. 4B). In line with the previously reported suction-pipette recordings from rod outer segments (22), the light responses in Cngb1 KO rods were strongly compromised: saturating flash stimuli only weakly hyperpolarized Cngb1 KO rods to −37.7 ± 1.5 mV (n = 14) (Fig. 4B). Similarly, in Cngb1/Hcn1 DKO rods the bright flash resulted only in a minor peak hyperpolarization to −37.7 ± 1.4 mV (n = 8) (Fig. 4B). The rare Cngb1 KO and Cngb1/Hcn1 DKO rods that showed a flash response displayed a greatly reduced amplitude, slower onset and recovery kinetics and a less pronounced ‘nose’ (13) compared with the wild-type rods (Fig. 4C).

Figure 4.

KO of Hcn1 has no major effect on the membrane potential of depolarized rods in the Cngb1 KO. (A) Representative families of current traces during hyperpolarizing voltage clamp steps from a holding potential of −53 mV, to −60/−67/−74/−81/−88/−95/−102/−109 mV, and depolarization to −65 mV (wild-type, Cngb1 KO and Cngb1/Hcn1 DKO rods). (B) Box plot showing V_dark (shaded box) and V_dark–max flash response (empty box) of juvenile wild-type, Cngb1 KO and Cngb1/Hcn1 DKO rods. (C) Photovoltage responses to flashes of increasing strength measured in wild-type, Cngb1 KO and Cngb1/Hcn1 DKO rods, respectively (flash strengths in the range of 0.5–398 photons/µm²). Data represented in (B) are the median value (thick horizontal line), interquartile range (the box), min and max values (narrow error bars). ***P < 0.001.

In the absence of Cngb1, only small amounts of homotetrameric CNGA1 channels are present in rod outer segments (22). Given that the dark current is mediated by CNG channels (26), the smaller number of functional CNG channels should have led, other factors being equal, to a more hyperpolarized dark membrane potential. The finding that the actual measured V_dark in Cngb1 KO and Cngb1/Hcn1 DKO rods was not significantly different from wild-type suggests that unidentified conductances result in a constitutive depolarization of Cngb1-deficient rods. To test this hypothesis, we analyzed the peak membrane potential reached in response to a saturating flash (V_dark max flash response), since saturating light eliminates the contribution of the dark current. In line with this idea, both Cngb1 KO and Cngb1/Hcn1 DKO rods were significantly depolarized relative to wild-type rods (one-way ANOVA, P < 0.0001), while no significant difference was detected between the two mutants (Fig. 4B).

Taken together, these data indicate that the protective role of HCN1 channels during the early stages of photoreceptor degeneration cannot be attributed to a straightforward effect on cellular membrane potential. In fact, we found that juvenile Cngb1 KO rods are in a constitutively depolarized state.

Involvement of Ca²⁺ and calpain in photoreceptor degeneration in Cngb1/Hcn1 DKO mice

To investigate if impaired Ca²⁺ signaling is involved in the accelerated disease progression of Cngb1/Hcn1 DKO mice, we analyzed the levels of calpain activation using an in situ activity assay. Calpain is a Ca²⁺-dependent protease involved in neuronal and photoreceptor cell death (27,28). Calpain activity was found to be elevated in a subset of photoreceptors in Cngb1 KO mice (25). Here, we found that in Cngb1/Hcn1 DKO mice significantly more photoreceptors showed elevated calpain activity compared with Cngb1 single KO mice (Fig. 5A–B).

Figure 5.

Involvement of calpain and Cav1.4 in the photoreceptor degeneration of Cngb1/Hcn1 DKO mice. (A) Calpain activity assay performed on unfixed retinal cryo sections of Cngb1 and Cngb1/Hcn1 DKO mice. (B) Quantification of photoreceptors exhibiting calpain activity. Bars represent the mean percentage of photoreceptors with calpain activity (n = 3) ± SEM; ***P < 0.001. (C) Representative OCT images of Cngb1/Hcn1 DKO and Cngb1/Hcn1/Cacna1f TKO mice at day 59 revealing the less pronounced thinning of the photoreceptor layer in the absence of Cav1.4 channels. (D) Photoreceptor layer thickness of Cngb1/Hcn1 DKO and Cngb1/Hcn1/Cacna1f TKO mice at Days 32 and 59. The quantification is based on repeated in vivo OCT measurements. Each data point represents the mean photoreceptor layer thickness (n = 5–6) ± SEM; **P < 0.01. gcl, ganglion cell layer; inl, inner nuclear layer; onl, outer nuclear layer. Scale bar in (A) marks 25 µm.

The main gates for calcium entry into photoreceptors are the outer segment plasma membrane CNG channel and the synaptic voltage-gated calcium channel (VGCC). The CNG channel function is strongly impaired by the deletion of Cngb1 (22). To test for the contribution of the VGCC, we cross-bred the Cngb1/Hcn1 DKO mice with Cacna1f KO mice-lacking expression of the Ca_v1.4 α1 subunit of the VGCC channel (29,30) and compared the development of photoreceptor layer thinning in these triple KO (TKO) mice to that in the Cngb1/Hcn1 DKO mouse. Indeed, in Cngb1/Hcn1/Cacna1f TKO mice the thinning of the photoreceptor layer progressed significantly slower compared with Cngb1/Hcn1 DKO mice (Fig. 5C–D) (two-way ANOVA, P < 0.01). However, the disease progression was not halted and still progressed faster than in Cngb1 KO mice suggesting that additional Ca²⁺ sources contribute to the effect. Given that the deletion of Cacna1f alone does not result in rod photoreceptor degeneration itself (30), the residual CNG channel and/or unknown calcium sources might contribute to the activation of calpain.

Effect of Hcn1 deletion on cone photoreceptor degeneration

HCN1 is expressed in both types of photoreceptors, rods and cones (14). Thus, the neuroprotective mechanism of HCN1 in degenerating rods might also be relevant for degenerative cone photoreceptor diseases. To test if deletion of Hcn1 exerts a similar disease-amplifying effect in cone degeneration, we cross-bred Hcn1 KO mice with the Cnga3 KO mouse model of achromatopsia (1). Cone photoreceptor degeneration in Cnga3 KO mice progresses slowly in the superior (dorsal) part of the retina and rather fast in the inferior (ventral) retina (31). Accordingly, at 3 months of age the cone density is significantly lower in the ventral compared with the dorsal part of the Cnga3 KO retina, resulting in areas with distinct levels of disease progression. We compared the cone photoreceptor density in 3-month-old Cnga3 KO with age-matched Cnga3/Hcn1 DKO mice, and found that the loss of Hcn1 significantly reduced the number of surviving cone photoreceptors by 1.5-fold in the dorsal and by 10-fold in the ventral part of the Cnga3-deficient retina (Fig. 6A–B, Supplementary Material, Fig. S3). Thus, the neuroprotective effect of HCN1 is not only restricted to rod photoreceptors, but also applies to diseases with cell-autonomous cone photoreceptor degeneration.

Figure 6.

KO of Hcn1 enhances photoreceptor degeneration in Cnga3-deficient mice. (A) Retinal cryo-sections of Cnga3 KO and Cnga3/Hcn1 DKO mice stained with two specific markers for cones, glycogen phosphorylase (red) and peanut agglutinin (green). Representative confocal images captured from the dorsal or the ventral part of the retina on cryo-sections through the level of the optic nerve. (B) Graph showing the quantification of cone photoreceptor density (cones/mm², n = 3) in the dorsal and ventral part as mean ± SEM; **P < 0.01, ***P < 0.001. onl, outer nuclear layer; os, outer segments. Scale bar in (A) marks 25 µm.

Discussion

HCN1 is a highly abundant photoreceptor inner segment plasma membrane ion channel (10,14) involved in shaping the rod and cone photoresponses (12–14). In particular, studies in Hcn1 KO mice revealed that the lack of HCN1 prolongs the rod photoresponse (14), which in turn inhibits cone signaling under mesopic conditions (15). Pharmacological inhibition of HCN channels also affects retinal function. Patients treated with the I_h blocker ivabradine report the occurrence of phosphenes (19,20) and ERG measurements in animal models demonstrated a modulation of the photoresponse after pharmacological HCN inhibition (32,33). However, no reports exist linking pharmacological inhibition of I_h or a genetic deficiency of HCN1 with photoreceptor degeneration.

In this study, we analyzed the effect of HCN1 on the viability of degenerating rod or cone photoreceptors. We show that the genetic ablation of Hcn1 or systemic administration of the I_h blocker zatebradine both exacerbate degeneration and loss of rod photoreceptors in the Cngb1 KO mouse model of RP. The Hcn1 KO also exacerbated non-cell autonomous cone degeneration secondary to rod cell death—a phenomenon known as bystander effect from RP patients and Cngb1 KO mice (22). In addition, a deleterious cell autonomous effect on cone viability was observed after genetic deletion of Hcn1 in the Cnga3 KO mouse model of achromatopsia. Genetic deletion of Hcn1 alone, however, had no effect on the retinal morphology and photoreceptor survival. Based on these findings, we propose that HCN1 function serves a neuroprotective role in inherited retinal diseases with a primary defect in rod or cone photoreceptors like RP or achromatopsia.

A previous study by Della Santina et al. (17) addressed the effects of the I_h blocker ivabradine on the progression of retinal degeneration in the rd10 mouse model of RP. The study was designed to analyze the acute and short-term (up to 3 weeks) effects of systemic ivabradine administration on retinal function and morphology. The authors found no effect of ivabradine on photoreceptor degeneration in the rd10 mice during a 3-week observation period. In line with Della Santina et al. (17), we also did not observe any short-term effects of I_h inhibition on retinal morphology in our study. However, when examining longer term effects at 7 or approximately 11 weeks after treatment, we found that inhibition of I_h resulted in enhanced thinning of the photoreceptor layer in the Cngb1 KO mouse model of RP. In addition to the different observation periods, the two studies used distinct I_h blockers (ivabradine versus zatebradine), but with similar potency on HCN channels (34). We cannot exclude that factors other than the observation period and the type of I_h blocker might have contributed to the different outcome. One additional factor may be the use of differing mouse models of RP. In particular, the lack of PDE6B function in the rd10 mouse (35,36) analyzed in Della Santina et al. (17) results in a much more rapid photoreceptor degeneration than the lack of CNGB1 in the Cngb1 KO mouse model (22) utilized in our study. Given the relatively fast disease progression in the rd10 mouse, it might be more difficult to detect an enhancement of photoreceptor degeneration by inhibition of I_h than in the slow degenerating Cngb1 KO retina.

Mechanistically, the loss of HCN1 function leads to increased activation of the Ca²⁺-dependent protease calpain, which is known to be involved in cell death in many mouse models of photoreceptor degeneration (25). It appears that the Ca²⁺/calpain effect depends at least partially on the presence of the synaptic voltage-gated calcium channels Cav1.4 (Cacna1f). The role of other Ca²⁺ sources in this regard is still unclear. However, a major contribution of CNG channel-mediated Ca²⁺ influx is less likely in both mouse models analyzed in this study (Cngb1 KO and Cnga3 KO).

In summary, we show here a protective effect of HCN1 channels on photoreceptor degeneration on the basis of both pharmacological data and genetic inactivation studies in murine models of RP and achromatopsia. Together, these data strongly support the view that HCN1 is a major factor for outer retinal viability in primary photoreceptor diseases.

The HCN inhibitor ivabradine is clinically used for the treatment of stable angina pectoris or heart failure. In the light of these findings, it may be recommended to reevaluate whether I_h blockers might be harmful to the eyesight in patients suffering from retinal degenerative disorders.

Materials and Methods

Animals

Hcn1 KO mice (B6;129-Hcn1^tm2Knd1/J, stock number #005034, The Jackson Laboratory) (37) were cross-bred with mice lacking either Cngb1 (22) or Cnga3 (1) to generate Cngb1/Hcn1 and Cnga3/Hcn1 DKO mice, respectively. To obtain Cngb1/Hcn1/Cacna1f TKO mice, Cngb1/Hcn1 DKO mice were crossed with a Cacna1f-deficient line obtained from Dr Marion Maw, University of Otago, Dunedin, New Zealand (29,30). Control experiments were conducted on mice with the same genetic background. All procedures concerning animals were performed with permission of local authorities (Regierung von Oberbayern, Regierungspräsidium Tübingen and Ethical Committee of the University of Pisa) and in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research.

Ophthalmological examinations

For ophthalmological examinations, adult mice received intraperitoneal injections of ketamine (0.1 mg/g) and xylazine (0.02 mg/g). For mice younger than 3 weeks a lower dose of ketamine (0.05 mg/g) and xylazine (0.01 mg/g) was used. Before the scanning procedure, Tropicamide eye drops were applied to the mice eyes for pupil dilation (Mydriadicum Stulln, Pharma Stulln GmbH, Stulln, Germany). Subsequently, hydroxylpropyl methylcellulose (Methocel 2%; OmniVision, Puchheim, Germany) was applied to keep the eyes moist. The examination was performed with an adapted Spectralis HRA + OCT system from Heidelberg Engineering (Dossenheim, Germany) in combination with optic lenses described previously (38). The system allowed for imaging of the eye fundus by cSLO and examination of the retinal morphology by OCT.

OCT scans were conducted with a 12° circular scan mode centered at the optic nerve head. This procedure enabled measurements of the photoreceptor layer thickness at a comparable distance from the optic nerve head and allowed for comparison of values in longitudinal examinations of the same eye and between individuals. In detail, photoreceptor layer thickness was measured between the clearly visible outer plexiform layer and the border of neuronal retina and the RPE. The photoreceptor layer thickness is equivalent to the term ‘photoreceptor plus’ occasionally used in other studies for quantification of OCT data. For statistical analysis, the mean photoreceptor layer thickness was calculated from single values measured in the dorsal, temporal, nasal and ventral region around the optic nerve. cSLO images of the eye fundus were obtained using the infrared laser (820 nm) and the scanner set to a 30° field of view at high resolution mode.

Pharmacological treatment

In general, pharmacological experiments were conducted on littermates. For the pharmacological inhibition of HCN channels, mice received a daily injection of 10 mg/kg Zatebradine (Z0127, Sigma-Aldrich) dissolved in 0.9% NaCl. Littermate controls received a daily injection of 0.9% NaCl.

Production and in vivo application of AAV vectors

Cloning was performed by standard techniques. All sequence manipulations were confirmed by sequencing. The YFP sequence was fused with a glycine-serine-glycine linker to the N-terminus of mouse Hcn1 using overlap PCR (39) and ligated with the human rhodopsin promoter sequence (40) into pAAV2.1-MCS (41) to generate pAAV2.1-RHO-YFP-HCN1-WPRE. Single-strand AAV vectors were produced by triple calcium phosphate transfection of HEK 293T cells with pAdDeltaF6 (42), pAAV2/8 Y733F (43) and pAAV2.1-RHO-YFP-HCN1-WPRE plasmids followed by iodixanol-gradient purification of cell lysates after 48 h (44). The 40–60% iodixanol interface was further purified and concentrated by ion-exchange chromatography on a 5 ml HiTrap Q Sepharose column using an ÄKTA Basic FPLC system (GE Healthcare, Munich, Germany) followed by further concentration using Amicon Ultra-4 Centrifugal Filter Units (Millipore, Schwalbach, Germany). Physical titers (in vector genome copies/µl) were determined by quantitative polymerase chain reaction with primers specific for the woodchuck stomatitis posttranslation regulatory element (WPRE) sequence (41).

Subretinal injections were performed as described previously (45). In brief, anesthetized mice received one microliter containing 7 × 10⁸ AAV genomic particles injected into the subretinal space. Special care was taken to avoid damage of the lens. The success of the procedure was monitored immediately following the injections using cSLO and OCT.

Immunohistochemistry

Vertical cryosections (10 µm) of the mouse retina were prepared for immunohistochemical staining as described previously (31). A mouse anti-Prph2 monoclonal antibody (2B7) (46) was applied (at 1:1000 dilution) for labeling of rod photoreceptor outer segments. Cone photoreceptors were stained with a guinea pig anti-glycogen phosphorylase (Glypho) polyclonal antibody [(47); 1:1000] and with fluorescein isothiocyanate conjugated PNA (1:100, Sigma-Aldrich). Confocal images were collected on a laser scanning microscope (LSM) 510 (Carl Zeiss, Oberkochen, Germany) or a true confocal scanner SP8 (Leica, Wetzlar, Germany) microscope.

Calpain activity assay

For detection of calpain activity, we performed an in situ enzymatic assay on unfixed retinal cryo-sections (48). In detail, sections were covered for 15 min with calpain reaction buffer (CRB: 25 mm 2-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid, 65 mm KCl, 2 mm MgCl₂, 1.5 mm CaCl₂, 2 mm dithiothreitol, pH 7.2]. Subsequently, 20 µm of the fluorescent calpain substrate 7-amino-4-chloromethylcoumarin, t-butyloxycarbonyl-Leucyl-L-methionine amide (Life Technologies, Darmstadt, Germany) dissolved in 1% dimethyl sulfoxide and 99% CRB were added. After an incubation of 1 h 30 min at 37°C slices were washed 3 × 10 min with CRB. Cells with increased calpain activity could be clearly identified due to the fluorescence of the cleaved substrate. Confocal images were obtained using a Zeiss LSM510 at 364 nm excitation (detection LP 385).

Electroretinography

ERG analysis was performed according to procedures described previously (49,50). In short, single-flash ERG responses were obtained under dark-adapted (scotopic; no background illumination: 0 cd/m²) and light-adapted (photopic; 30 cd/m², starting 10 min before recording) conditions. Single white-flash stimuli ranged from −4 to 1.5 log cd × s/m² under dark-adapted conditions, and from −2 to 1.5 log cd × s/m² under light-adapted conditions. Ten responses were averaged with interstimulus intervals of 5 s (for −4 to −0.5 log cd × s/m²) or 17 s (for 0 to 1.5 log cd × s/m²).

Patch clamp recordings

Current- and voltage-clamp perforated patch clamp recordings of rod photoreceptors in mouse retinal slices were performed as previously described (24,51). All procedures were approved by the Ethical Committee of the University of Pisa (prot. no. 2891/12) and were conducted in accordance with Italian (legislative degree 116/92) and EU regulations (Council Directive 86/609/EEC). Briefly, juvenile (P15–20) or adult (>P30) mice (see above for strain details) were anaesthetized and euthanized. Their retinas were extracted in the ice-cold AMES' medium (A1420; Sigma-Aldrich, St Louis, MO, USA), made to adhere on filter paper and sliced on a manual tissue chopper at a thickness of 250 µm. After being transferred in the recording chamber they were superfused at a temperature of ∼24°C with bicarbonate-buffered AMES' medium and visualized with a differential interference contrast infrared microscope. Seals were obtained on rod somata with perforated patch pipettes filled with a solution containing (in mm): 90 potassium aspartate, 20 K₂SO₄, 15 KCl, 10 NaCl, 5 K₂Pipes and 0.4 mg/ml Amphotericin-B. Final pH was set at 7.2 with HCl/KOH. Based on the expected liquid junction and Donnan potentials in our recording conditions (24), we report uncorrected values of the membrane potential. Full-field flashes were delivered with a green light emitting diode (OD520; Optodiode Corp., Newbury Park, CA, USA) having its emission peak at 520 nm.

Statistical analysis

Statistical analysis was performed using the Graph Pad Prism 5 software. To compare two or more groups at one time point, unpaired Student's t-test or one-way ANOVA tests were applied, respectively. For comparing groups in longitudinal examinations, two-way ANOVA with Bonferroni post-tests were performed. Unless otherwise stated, all values are given as mean ± SE.

Funding

This work was supported by the Deutsche Forschungsgemeinschaft (DFG) and the Tistou and Charlotte Kerstan Foundation (RD-CURE).

Acknowledgements

We thank Gudrun Utz, Pia Lacroix, Jennifer Schmidt and Elisabeth Schulze for excellent technical help, Dr Marion Maw (University of Otago, Dunedin, New Zealand) for providing the Cacna1f-deficient mouse line, Drs James M. Wilson (University of Pennsylvania) and Alberto Auricchio (TIGEM) for the gift of AAV plasmids and Drs Muna Naash (Oklahoma State University) and Brigitte Pfeiffer-Guglielmi (University of Tübingen) for the gift of antibodies.

Conflict of Interest statement. None declared.