Abstract
Purpose.:
Gap junctional coupling between rod and cone photoreceptor cells is regulated by light and the circadian clock, and contributes to retinal light adaptation. Phosphorylation of connexin 36 (Cx36) has been proposed as the mechanism involved. We investigated whether retinal Cx36 is also regulated at the level of transcript and protein expression.
Methods.:
At specific time points in a diurnal or circadian cycle, Cx36 protein was assessed by Western blotting and immunohistochemistry, and Cx36 transcript by quantitative real time PCR in a melatonin-deficient (C57BL6/FVB) and two melatonin-proficient (C3H+/+ and C3Hrd/rd) mouse strains.
Results.:
In C57BL6/FVB mice during a diurnal cycle, Cx36 protein expression was rhythmic, peaking at approximately zeitgeber time (ZT) 20. However, this rhythm was not maintained in the circadian cycle. In C3H+/+ mice levels of Cx36 protein were higher at night and subjective night relative to day and subjective day, respectively. These patterns of Cx36 expression were localized primarily to the outer plexiform layer in both strains. Cx36 transcript expression was higher at night and subjective night relative to day and subjective day in C57BL6/FVB and C3H+/+ mice. Rhythmic expression of Cx36 transcript was lost in retinally degenerate C3Hrd/rd mice.
Conclusions.:
The results suggested the circadian control of Cx36 protein expression is dependent on melatonin, whereas the circadian regulation of Cx36 transcript expression may be controlled directly by the circadian clock. In addition to post-translational modification, regulation of Cx36 transcript and protein expression may be important during retinal light adaptation.
The retina is a dynamic tissue that adapts to light stimuli, which can vary 10
9-fold within a normal light-dark cycle. One mechanism the retina uses to adapt to changes in ambient illumination is the regulation of gap junctional coupling. Gap junctions are present between nearly every neuronal cell type in the retina.
1 They are composed of 2 opposing connexons, which in turn are composed of 6 connexin (Cx) subunits. At low light levels increased cell-to-cell coupling occurs at the rod–cone,
2 horizontal cell (HC)–HC,
3,4 and AII amacrine cell (AC)–AII AC junctions.
5 This can enhance signal-to-noise ratio and, at the rod–cone junction, permit mixing of signals from different pathways. OFF alpha ganglion cells demonstrate the opposite pattern with increased coupling in the light adapted state, possibly to correlate activity during light stimulation.
6
The circadian clock also contributes to light adaptation and permits retinal pathways to anticipate changes in background illumination. Ribelayga et al. demonstrated that the circadian clock controls the day–night difference in rod–cone coupling.
2 The mechanism involved has been proposed to involve phosphorylation of Cx36
2,7–9 and studies have implicated dopamine as a regulator of this process.
2,10 Dopamine release in the retina is regulated by light
11–13 and also by the circadian clock via rhythmic melatonin release.
14,15 In the day/subjective day, activation of dopamine D2/D4 receptors on cone photoreceptors inhibits adenylate cyclase activity, preventing cAMP-dependent protein kinase A (PKA)–mediated Cx36 phosphorylation.
2 In the night/subjective night, the inhibition of adenylate cyclase is relieved and phosphorylation of Cx36 results in an increase in gap junctional conductance.
In addition to regulation of connexins by post-translational modification, complex regulation at the transcriptional and translational levels also have been described in various tissues.
16,17 In the retina, transcriptional control of the Cx36 gene (
Gjd2 [gap junction protein, delta 2]) has been observed following prolonged dark adaptation, during development and in conditions of retinal degeneration.
18–20 However, it is unknown whether Cx36 expression is regulated at the transcriptional and/or translational level during the diurnal or circadian cycle.
We quantified Cx36 protein levels, assessed Cx36 protein localization, and quantified Cx36 transcript levels during the diurnal and circadian cycle in melatonin-proficient (C57BL6/FVB) and melatonin-deficient (C3H/HeJ) mouse strains.
C57BL6/FVB, C3H/HeJ-
Pde6b+ (C3H
+/+), and C3H/HeJ-
Pde6brd1 (C3H
rd/rd) mice of either sex were maintained under a controlled 12:12 light–dark (LD) cycle (zeitgeber time [ZT] 0 lights on, ZT12 lights off). Food and water were available ad libitum. Eyes were collected at ZT 0, 4, 8, 12, 16, and 20 for C57BL6/FVB mice, and at ZT8 and 20 for C3H/HeJ strains. For circadian experiments, animals were released into constant darkness (DD) for one day and eyes collected at the equivalent circadian time (CT) points on the following day. The time in DD was kept relatively short (<48 hours) to enable accurate approximation of the CT.
21,22 Retinas were dissected and flash frozen for RNA or protein extraction. Alternatively eyes were processed for immunohistochemistry. All experiments were performed in accordance with local and UK Home Office guidelines, and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.
Retinal tissue samples were homogenized in 2% (wt/vol) SDS buffer in PBS with mini complete protease inhibitors (Roche, Basel, Switzerland) and centrifuged at 23,000g for 30 minutes. The protein concentration was quantified using the Micro BCA assay (Thermo Scientific, Waltham, MA). The remaining supernatant was mixed with 0.25 volumes of 4 × SDS sample buffer and sonicated for protein denaturation. The samples (20 μg) were run on 10% SDS-PAGE polyacrylamide gels and transferred onto polyvinylidine fluoride (PVDF) membranes (Bio-Rad, Hercules, CA). The PVDF membranes were stained with Fast Green to assess the quality of the transfer and blocked in 3% (wt/vol) BSA in Tris buffered saline, 1% (vol/vol) Tween 20 (TBST) for 1 hour. The membranes were incubated with a mouse monoclonal antibody against mouse Cx36 (37-4600; Invitrogen, Carlsbad, CA) diluted 1:600 in 3% (wt/vol) BSA in TBST overnight at 4°C. Membranes were washed in TBST and incubated with HRP-linked secondary antibody (Abcam, Cambridge, MA) for 1 hour at room temperature. The secondary antibody was detected using an Enhanced Chemiluminescence (ECL) system (Thermo Scientific). The membranes were exposed to X-ray film (Santa Cruz Biotechnology, Santa Cruz, CA) that was developed with standard procedures (Xograph Imaging Systems; Xograph Healthcare, Gloustershire, United Kingdom). After ECL development the membranes were stripped with 50 mM glycine, 1% (vol/vol) SDS, pH2 solution, washed, and blocked as described earlier. To assess gel loading, the stripped blots were incubated with a rabbit polyclonal antibody against β-actin (Abcam), which was detected using a donkey anti-rabbit secondary antibody (Abcam).
Quantity One v.4.5.1 software (Bio-Rad) was used to quantify the Cx36 bands on Western blots. The mean intensity of the Cx36 band was expressed relative to the respective actin mean band intensity. Where a full diurnal or circadian profile of Cx36 expression was assessed, values were normalized to the lowest value in the data set.
Eyes were enucleated, and the cornea and lens removed before immersing in 4% paraformaldehyde in PBS for 5 minutes. After fixation, the eyecups were washed in PBS, cryoprotected in 30% sucrose for 2 to 3 days, and embedded and frozen in Tissue-Tek OCT (Sakura, Torrance, CA). A Leica CM1850 cryostat (Leica Microsystems, Wetzlar, Germany) was used to cut 18 μm frozen sections, which were collected on polylysine-coated slides and stored at −20°C until further processing. Sections were stained using standard immunohistochemistry procedures. Briefly, sections were washed with PBS, permeabilized for 30 minutes in 0.5% Triton X-100 in PBS, blocked in 10% normal goat serum, 0.1% Triton X-100 in PBS, and incubated with the primary antibody overnight at 4°C. Cx36 protein was immunostained using the same monoclonal antibody used for Westerns at a dilution of 1:40,000 (at lower dilutions, saturation of the signal intensity made analysis difficult). Following overnight incubation, sections were washed and incubated for 1 hour with an Alexa 555 goat anti-mouse IgG1 secondary antibody. The sections were washed and mounted using ProLong Gold antifade medium (Invitrogen). Images were acquired in the inner (IPL) and outer (OPL) plexiform layers separately due to the difference in signal intensity between layers, using an inverted Zeiss LSM710 laser scanning confocal microscope (Carl Zeiss, Oberkochen, Germany). To minimize bleaching, only one image was collected from each retinal section. Three retinal sections from each eye were imaged per synaptic layer. The same acquisition settings were used for each image set (i.e., IPL or OPL). To quantify connexin staining, the mean pixel intensity was quantified in three regions of interest (ROI) using Image J (available in the public domain at
http://rsbweb.nih.gov/ij/). The ROI was 100 × 20 μm within the OPL and 100 × 40 μm within the IPL. The averaged data for each retinal layer were compared between ZT8 and ZT20 or CT8 and CT20. Staining in the IPL was analyzed further by placing ROI (100 × 20 μm) in either the inner or outer regions of this layer.
RNA from each retina was extracted using RNeasy Mini columns (Qiagen) following the manufacturer's protocol. To avoid genomic DNA contamination RNA was treated on column with RNase-free DNase (Qiagen, Germantown, MD). The resulting RNA was quantified and 500 μg of each sample were reverse transcribed into cDNA with random hexamers using the SuperScriptIII Reverse-Transcriptase system (Invitrogen) following the manufacturer's instructions. Quantitative real-time PCR was performed with the QuantiFast Probe PCR kit (Qiagen) or Brilliant III Ultra Fast SYBR Green (Agilent, Santa Clara, CA) using the StepOnePlus Real Time PCR System (Applied Biosystems, Foster City, CA). The primers used are described in the
Table. Expression levels of each transcript were normalized to the geometric mean of three housekeeping genes: acidic ribosomal phosphoprotein (ARP), beta-2 microglobulin (B2M), and proteasome subunit beta type-2 (PSMB2).
ANOVA analysis was used to test for significant differences between diurnal and circadian data sets. Periodicity was analyzed by fitting a sinusoidal wave through the data and its significance tested against fitting the data with a constant value (CircWave v1.4 software, available in the public domain at
http://www.euclock.org/).
23 When the data set only included two time points during a diurnal or circadian cycle, unpaired Student's
t-tests were used to test for significant differences.
Cx36 Protein Expression Is Rhythmic in the Diurnal but Not Circadian Cycle in C57BL6/FVB Mice
Cx36 protein expression was assessed by Western blot in C57BL6/FVB retinae. The Cx36 antibody (Invitrogen) used recognized two bands in Western blots of retinal homogenates (
Fig. 1). The 36 kDa band represents Cx36 protein. It has been reported previously that the upper ∼38 kDa band recognizes a nonspecific product.
24 We quantified Cx36 protein expression (36 kDa band) over the diurnal and circadian cycle (
Figs. 2A,
2B). Cx36 protein expression was relatively low at the beginning of the light phase of the diurnal cycle and increased steadily from lights OFF peaking at ZT20 (
Figs. 2A,
2C). This pattern of expression was significantly rhythmic (
F = 4.72,
P < 0.05). In the circadian cycle Cx36 expression fluctuated with a pattern similar to the diurnal cycle; however, this rhythm was less pronounced and not statistically significant (
F = 0.94,
P = 0.49;
Figs. 2B,
2D).
The localization of the protein rhythm was examined using immunohistochemistry in retinal sections. Cx36 protein is found in the synaptic (plexiform) layers of the retina.
25–27 We quantified the Cx36 fluorescence signal in the OPL and IPL at ZT20 when Cx36 protein expression was high, and at ZT8 when Cx36 protein levels were low. In the OPL, Cx36 fluorescence intensity was significantly higher at ZT20 relative to ZT8 (
P < 0.001;
Figs. 3A,
3B). This pattern of expression was not maintained in conditions of constant darkness. There was no significant difference in Cx36 fluorescence intensity at CT20 relative to CT8 in the OPL (
P = 0.27;
Figs. 3C,
3D). In the IPL we did not observe any differences in Cx36 fluorescence intensity between the two time points in the diurnal (
P = 0.68) or circadian (
P = 0.32) cycle (
Figs. 3E–H). Separate analysis of Cx36 fluorescence intensity in the outer (corresponding to sublamina a) and inner (corresponding to sublamina b) regions of the IPL also did not reveal significant differences between the two time points (outer IPL ZT8 vs. ZT20,
P = 0.45 and CT8 vs. CT 20,
P = 0.35; inner IPL ZT8 vs. ZT20,
P = 0.21 and CT8 vs. CT 20,
P = 0.92; data not shown).
Cx36 Transcript Expression Is Rhythmic in the Diurnal and Circadian Cycle in C57BL6/FVB Mice
To determine if the protein rhythm was reflected at the transcriptional level, Cx36 (
Gjd2) transcript expression was assessed in the C57BL6/FVB mouse retina. In the diurnal cycle, Cx36 transcript expression was low in the light phase and increased steadily during the dark period, peaking in the late night phase (
Fig. 4A). A highly significant rhythm in the pattern of expression was detected (
F = 7.83,
P < 0.01). In the circadian cycle, low transcript levels were observed in the subjective day and expression increased subsequently from CT16 to the end of the subjective night phase. This pattern of expression also was significantly rhythmic (
F = 4.90,
P < 0.05,
Fig. 4B), but with a lower amplitude than in LD. Thus, at the transcriptional level, a diurnal and circadian rhythm in Cx36 expression was detected.
Cx36 Protein Expression Is Increased during the Night in the Diurnal Cycle and Subjective Night in the Circadian Cycle in C3H+/+ Mice
To investigate whether the lack of a circadian rhythm in Cx36 protein expression was due to the strain used, we assessed Cx36 protein levels in C3H
+/+ mice. Unlike C57BL6/FVB mice, the C3H strain is melatonin proficient.
28–30 Cx36 protein was examined at one day/subjective day (ZT8/CT8) and one night/subjective night (ZT20/CT20) time point, corresponding to the approximate peak and trough of Cx36 protein expression in C57BL6/FVB mice. In the diurnal cycle, C3H
+/+ mouse retinae showed a significantly higher level of Cx36 protein at ZT20 in comparison to ZT8 (
P < 0.05;
Figs. 5A,
5B). In conditions of constant darkness, this increase was maintained with significantly higher levels of Cx36 at CT20 relative to CT8 (
P < 0.05;
Figs. 5D,
5E).
Using immunohistochemistry, we investigated the localization of the diurnal and circadian changes in Cx36 protein expression in C3H
+/+ mice. Cx36 fluorescence intensity in the OPL was significantly higher at night or in the subjective night relative to the day (ZT8 vs. ZT20
P < 0.05) or subjective day (CT8 vs. CT20
P < 0.05), respectively (
Figs. 5G–I). We did not observe any significant change in fluorescence intensity in the IPL in the diurnal (ZT8 vs. ZT20
P = 0.16) or circadian (CT8 vs. CT20
P = 0.24) cycle (
Figs. 5J–L). Separate analysis of Cx36 fluorescence intensity in the outer and inner regions of the IPL also revealed no significant difference between time points in LD or DD (outer IPL ZT8 vs. ZT20,
P = 0.19 and CT8 vs. CT 20,
P = 0.62; inner IPL ZT8 vs. ZT20,
P = 0.09 and CT8 vs. CT 20,
P = 0.44; data not shown).
Cx36 Transcript Expression Is Increased during the Night in the Diurnal Cycle and Subjective Night in the Circadian Cycle in C3H+/+ Mice but Not in C3Hrd/rd Mice
Cx36 transcript expression in C3H
+/+ mice was higher at night and in the subjective night relative to day (ZT8 vs. ZT20
P < 0.05) and the subjective day (CT8 vs. CT20
P < 0.05), respectively (
Figs. 6A,
6B). To examine whether the transcript rhythms were dependent on the presence of a functional outer retina, we assessed Cx36 transcript expression in C3H
rd/rd retinae. C3H
rd/rd mice have a mutation in the
Pde6b gene resulting in the degeneration of the outer retina. In these mice, Cx36 transcript expression was not significantly different between ZT8 and ZT20 (
P = 0.66) or between CT8 and CT20 (
P = 0.36;
Figs. 6C,
6D). To confirm that C3H
rd/rd were not expressing a rhythm in Cx36 transcript expression that was out of phase to the C3H
+/+ rhythm, we also assessed Cx36 transcript at other time points. Sampling every 4 hours in C3H
rd/rd mouse retinae did not show a significant Cx36 transcript rhythm (data not shown).
Rod–cone coupling in the retina is under circadian control and the regulation of Cx36 phosphorylation has been suggested to be the mechanism involved.
2 Here, we showed that rhythmic regulation of Cx36 expression at the transcript and protein level also may contribute to the circadian regulation of gap junctional coupling. We propose that an interaction between the circadian clock, light, melatonin, and dopamine regulate Cx36 transcription and translation in the diurnal and circadian cycles.
Ribelayga et al. first described the circadian control of rod–cone coupling in the retina and proposed that this was regulated by dopamine-controlled PKA-mediated phosphorylation of Cx36
2. In our study, we found that Cx36 protein expression is rhythmic over a diurnal cycle in C57BL6/FVB mice. Low protein levels were observed in the day with higher levels at the night. We did not find a significant rhythm in Cx36 protein expression over a circadian cycle. C57BL6/FVB mice do not produce melatonin
28–30 and, therefore, do not have a strong dopamine rhythm in constant darkness.
14 Therefore, we examined Cx36 protein expression in a melatonin-proficient mouse strain, C3H
+/+ 28,29 . These mice do demonstrate a retinal dopamine rhythm in constant darkness.
14 Cx36 protein expression was higher at night and in the subjective night in the diurnal and circadian cycle, respectively, in C3H
+/+ mice. This would suggest that a melatonin/dopamine rhythm is necessary to maintain the increase in Cx36 expression in the subjective night of the circadian cycle. Doyle et al. have shown that cyclic light can drive a dopamine rhythm in the absence of melatonin.
14 Thus, in melatonin-deficient C57BL6/FVB mice, light-induced retinal dopamine release could be sufficient to drive the Cx36 protein expression rhythm in the diurnal cycle. These data represent to our knowledge the first evidence for nonpost-translational regulation of Cx36 in the diurnal and circadian cycles.
We observed an increase in Cx36 immunoreactivity in the OPL at ZT20 relative to ZT8 in C57BL6/FVB mice, and at ZT20/CT20 relative to ZT8/CT8 in C3H
+/+ mice. No significant change in Cx36 fluorescence intensity was detected in the IPL at the same time points. These results imply that the Cx36 protein rhythm was localized primarily to the OPL. Cx36 is expressed at cone pedicles at the outer region of the OPL and at OFF-cone bipolar cell dendrites in the inner region of the OPL.
26 Within the OPL, we observed strongest Cx36 labeling in the outer OPL. However, due to the narrow dimensions of the OPL, quantification was difficult. We propose that the observed rhythm in Cx36 expression originates in the photoreceptor cell layer.
There could be several possible reasons why we could not detect day/night or subjective day/subjective night differences in Cx36 immunoreactivity in the IPL, where the majority of Cx36 is expressed. Perhaps the simplest explanation is that no rhythm exists in the IPL. It also is possible that a rhythm is present, but out of phase with the OPL rhythm. However, if this was the case it would not support the protein expression data as assessed by Western blot, which peaked at ZT20. Furthermore, a rhythm in-phase with the OPL rhythm may be present in the IPL, but of low amplitude and, therefore, nonsignificant. The presence of a low amplitude in-phase rhythm within the IPL could explain how a 2-fold change in Cx36 protein expression is observed by Western blot when only ∼10% of Cx36 expression is present in the OPL.
26
We investigated whether the Cx36 protein rhythm was reflected at the level of Cx36 transcript expression. Although Cx36 transcript has been shown to be regulated during various remodeling processes in the retina, such as degeneration, dark adaptation, and in development,
18–20 we provide the first evidence that Cx36 transcript expression is regulated over a diurnal and circadian cycle in the retina. Cx36 transcript levels in C57BL6/FVB and C3H
+/+ mice were low in the day and subjective day, and increased significantly at night and in the subjective night. In mice lacking an outer retina (C3H
rd/rd), all rhythmicity was lost. This would suggest the Cx36 transcript rhythms either originated in or were dependent on the outer retina (supporting the Cx36 protein localization data). It is interesting that the transcript rhythm peaked later than the protein rhythm. It is unclear why this should occur, and could suggest a long lag between transcription and translation, or possibly due to repression of translation/protein degradation in the late night phase.
In a previous study, Urschel et al. observed no day/night changes in Cx36 transcript expression.
31 The discrepancy between these findings and ours could be due to many reasons. Urschel et al. used Northern blots to quantify Cx36 transcript, whereas we used real-time quantitative PCR. Therefore, sensitivity of the techniques may be an issue. Furthermore, in the previous study Cx36 expression was normalized to GAPDH.
31 Data from various species, including mice, show that GAPDH expression is rhythmic over a diurnal and circadian cycle,
32–34 and, therefore, any difference in Cx36 transcript expression may have been masked by a rhythm in the normalization gene.
The fact that the Cx36 transcript rhythm was not dependent on melatonin proficiency suggests that Cx36 transcription may be controlled directly by a circadian oscillator. The inner and outer retina contain self-sustaining oscillatory mechanisms that are independent of the central circadian pacemaker, the suprachiasmatic nucleus.
32,35–38 Indeed, it has been proposed that expression of other transcripts in the outer retina are controlled by the circadian clock, including adenylyl cyclase 1 (
Adcy1) and arylalkylamine N-acetyltransferase.
35,39 Interestingly, rhythmic expression of
Adcy1 can be entrained to the external day/night cycle by dopamine.
40 We observed a higher amplitude rhythm of Cx36 transcript expression in the diurnal cycle relative to the circadian cycle, and it is possible that dopamine may be responsible for this difference.
The results suggested Cx36 transcription is controlled directly by the circadian clock, whereas Cx36 protein expression involves melatonin/dopaminergic modulation. This discrepancy between transcription and translation of Cx36 could be explained by a requirement of a melatonin/dopamine rhythm in post-transcriptional regulation. One example of post-transcriptional regulation of connexins is by microRNAs, which can reduce translation without affecting mRNA expression.
41,42 The pathways that regulate microRNAs in the retina are largely unknown, but micro RNAs have been shown to be regulated by dopamine in the brain.
43 Therefore, it is possible that in the absence of a melatonin/dopamine rhythm in C57BL6/FVB mice, microRNAs may suppress Cx36 translation in the circadian cycle.
We have demonstrated that light and/or the circadian clock can control transcription and translation of Cx36, which may be important in the regulation of rod–cone gap junctional coupling during light/dark adaptation.
CircWave analysis software was provided by Roelof Hut.
Supported by the Biotechnology and Biological Sciences Research Council (BB/G003602/1 [S.S]).