The progressive loss of rod and cone photoreceptors in human subjects with retinitis pigmentosa causes a gradual decline in vision and can result in blindness. Current treatment strategies for the disease rely on the integrity of inner retinal neurons, such as amacrine cells, that are postsynaptic to photoreceptors. Previous work has demonstrated that a specialized subclass of retinal amacrine cell that synthesizes and releases the key neurotransmitter dopamine remains morphologically intact during the disease; however, the pathophysiological function of these neurons remains poorly understood. Here we examined spontaneous and light-evoked spike activity of genetically labeled dopamine neurons from the retinas of retinal degeneration 1 (rd1) mice. Our results indicated that rd1 dopamine neurons remained functionally intact with preserved spontaneous spiking activity and light-evoked responses. The light responses were mediated exclusively by melanopsin phototransduction, not by surviving cones. Our data also suggested that dopamine neurons were altered during photoreceptor loss, as evidenced by less spontaneous bursting activity and increased light-evoked responses with age. Further evidence showed that these alterations were attributed to enhanced GABA/melanopsin signaling to dopamine neurons during disease progression. Taken together, our studies provide valuable information regarding the preservation and functional modification of the retinal dopamine neuronal system in rd1; this information should be considered when designing treatment strategies for retinitis pigmentosa.
- amacrine cell
- retinitis pigmentosa
the vertebrate retina contains three unique classes of photosensitive cells: the outer retinal rod and cone photoreceptors, which are essential for image-forming vision, and the recently discovered inner retinal intrinsically photosensitive retinal ganglion cells (ipRGCs), which are obligatory for non-image-forming visual responses (e.g., circadian photoentrainment and pupillary light reflex) (Berson 2003). As a result, the loss of rods and cones in humans with retinitis pigmentosa leads to poor vision and can even cause blindness (Heckenlively 1988). Currently there is no effective treatment for the disease. Promising strategies for vision restoration include photoreceptor cell transplantation, stem cell and gene therapy, and electrical and cell type-specific optical stimulation (Acland et al. 2001; Bi et al. 2006; Pearson et al. 2012; Weiland et al. 2005). These approaches rely critically on the structural and functional integrity and stability of inner retinal neurons including horizontal cells, bipolar cells, amacrine cells, and ganglion cells. It is therefore imperative to understand how these neurons respond to rod and cone degeneration. Previous studies have attempted to evaluate the structure and function of these neurons during retinal degeneration, with a focus on horizontal cells, bipolar cells, and ganglion cells (Chen et al. 2012; Damiani et al. 2012; Lin et al. 2009; Mazzoni et al. 2008; Puthussery et al. 2009; Stasheff 2008; Strettoi and Pignatelli 2000; Strettoi et al. 2002; Varela et al. 2003; Yee et al. 2012), yet limited information is available for amacrine cells, the third-order retinal neurons that have diverse functions in visual information processing (Borowska et al. 2011; Strettoi et al. 2002; Trenholm et al. 2012).
A key subclass of retinal amacrine cells is the dopaminergic amacrine neuron. These cells are the sole source of retinal dopamine, which plays vital roles in a variety of retinal functions (Witkovsky 2004). This molecule can serve as a light mediator that reconfigures retinal neural circuits in light adaptation as well as acting as a clock regulator that controls retinal circadian rhythm. However, levels of retinal dopamine are decreased in dystrophic retinas; the underlying mechanisms are still unclear (Doyle et al. 2002; Hankins and Ikeda 1994; Nir et al. 2000; Nir and Iuvone 1994; Vugler et al. 2007). This decline could be due to a reduced number of retinal dopamine neurons, low levels of dopamine synthesis, or an impairment of dopamine release from the cells. Previous studies have demonstrated that the former two possibilities are unlikely, leaving the question of whether the latter is involved in the underlying mechanisms (Frucht et al. 1982; Hankins and Ikeda 1994; Kato et al. 1981; Strettoi et al. 2002). In addition, although rod and cone photoreceptors are degenerated in dystrophic retinas, light is still capable of increasing retinal dopamine release (Doyle et al. 2002; Vugler et al. 2007). Again, the phototransduction source and mechanisms are unclear. A direct strategy for addressing these crucial issues is to determine how individual dopamine neurons functionally respond to rod and cone loss.
Dopamine neurons are known to fire action potentials (spikes) that are thought to be the principal trigger for dopamine release (White 1996). The amount of dopamine released appears to be determined by the spike frequency and pattern (single spike, random, or bursting) (Floresco et al. 2003; Gonon 1988; Puopolo et al. 2001). Our recent studies show that retinal dopamine neurons in mice exhibit a mixture of spontaneous single spikes and bursts (Gustincich et al. 1997; Zhang et al. 2007). The spontaneous activity of the neurons is negatively regulated by GABAergic inhibitory synaptic inputs in darkness and elevated by glutamatergic excitatory synaptic inputs on light stimulation (Zhang et al. 2007, 2008). Further evidence demonstrates that the light-evoked activity is mediated by rods and cones through bipolar cells as well as by the novel photopigment melanopsin that is expressed in a small population of retinal ganglion cells (Zhang et al. 2007, 2008). This novel information obtained from wild-type (WT) mice has provided a basis for exploring the pathophysiology of dopamine neurons in diseased mice such as the retinal degeneration 1 (rd1) mouse—an animal model of retinitis pigmentosa. Our initial studies in rd1 mice have demonstrated that some dopamine neurons are still activated by light (Zhang et al. 2008, 2012). Here we expand our previous studies by demonstrating that 1) the intrinsic activity of dopamine neurons is preserved with a decrease in bursting in rd1 retinas, 2) burst reduction is mediated by enhanced GABA signaling, 3) the preserved light responses of dopamine neurons are mediated exclusively by melanopsin, and 4) the novel neural pathway from ipRGCs to dopamine neurons is enhanced in advanced retinal degeneration.
MATERIALS AND METHODS
C3H/HeJ mice homozygous for the Pde6brd1 mutation (rd1) were purchased from the Jackson Laboratory. This rd1 mouse model carries a mutation in the β-subunit of the rod photoreceptor cGMP phosphodiesterase-6 (PDE-6). The mutation is characterized by initial rod loss followed by secondary cone death. Rod loss occurs rapidly with onset at postnatal day (P)8 and is nearly complete by P21. By P90, virtually all outer photoreceptors have disappeared except for ∼3% of cone somata in the dorsal retina (Carter-Dawson et al. 1978).
We crossed the rd1 mice with our transgenic mice (on a C57BL/6J background) in which dopamine neurons are genetically marked by red fluorescent protein (RFP) under control of the tyrosine hydroxylase (TH) promoter (Zhang et al. 2004). Resulting F1 rd1 heterozygous (het) mice were genotyped for the presence of RFP, and the positive animals were further crossed with rd1 mice. Male and female offspring from the F2 generation were genotyped for RFP and the rd1 mutation. From this cross we obtained TH::RFP mice homozygous for the rd1 mutation (rd1 TH::RFP) to use for experiments. The rd1 het TH::RFP mice we obtained were used for control experiments. We examined the gross anatomy of the living retinas isolated from rd1 and rd1 het TH::RFP transgenic mice (1–13 mo old) and found that although these mice had a mixed C3H and C57BL/6J background, the rd1 TH::RFP transgenic mice had the same phenotype as the rd1 mice and the rd1 het TH::RFP transgenic mice had the same phenotype as the WT TH::RFP transgenic mice.
To genetically knock out the photopigment melanopsin in rd1 mice, rd1 TH::RFP transgenic mice were further crossed with melanopsin knockout (opn4−/−) mice on a 129SV background (kindly provided by Dr. Samer Hattar at Johns Hopkins University) (Hattar et al. 2003). RFP-positive F1 male and female offspring were crossed to gain an F2 generation. The resulting F2 offspring were genotyped, and the mice lacking opn4, homozygous for the rd1 mutation, and positive for RFP (opn4−/− rd1 TH::RFP transgenic mice) were used for experiments. Again, we found that the anatomical structures of the living retinas from this line of mice (1–4 mo old) had the same phenotype as the rd1 mice. All animals were maintained under 12:12-h light-dark conditions. All procedures conformed to National Institutes of Health guidelines for work with laboratory animals and were approved by the Institutional Animal Care and Use Committee at Oakland University.
To avoid a circadian effect, all experiments were performed during the day. Mice were dark adapted for 1–2 h prior to experiments and then euthanized by CO2 overdose and cervical dislocation. Their eyes were enucleated and hemisected at the ora serrata under infrared illumination. The cornea and lens were removed in a petri dish filled with oxygenated extracellular solution containing (in mM) 125 NaCl, 2.5 KCl, 1 MgSO4, 2 CaCl2, 1.25 NaH2PO3, 20 glucose, and 26 NaHCO4. The retina was separated from the sclera and then placed either photoreceptor side down (rd1 het retinas) or ganglion cell side down (rd1 retinas) in a recording chamber mounted on the stage of an upright conventional fluorescence microscope (Axio Examiner, Zeiss, Oberkochen, Germany) within a light-tight Faraday cage. Oxygenated extracellular medium (pH 7.4 with 95% O2-5% CO2) continuously perfused the recording chamber at a rate of 2–3 ml/min, and the superfusate was held at 32–34°C by a temperature control unit (TC-344A, Warner Instruments).
The retina was maintained in darkness for ∼1 h prior to recording. Cells and recording pipettes were viewed on a computer monitor coupled to a digital camera (AxioCam, Zeiss) mounted on the microscope. TH::RFP-expressing cells were first identified by fluorescence microscopy with a rhodamine filter set with a brief “snap-shot” of fluorescence excitation light (1–5 s). The identified cells and glass electrode were visualized with infrared differential interference contrast (IR-DIC) optics for patch-clamp recording. Experiments began 10–15 min after fluorescence was used to locate the cells, allowing the retinas to recover from photobleaching (caused by the brief fluorescence excitation light). The recovery may be incomplete during this short dark-adaptation period, so our experiments were likely performed in a partially dark-adapted state.
Patch-clamp recordings were made from the soma of RFP-labeled dopamine neurons with 4- to 7-MΩ electrodes, and signals were amplified with an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA). The pipette solution for whole cell current-clamp experiments contained (in mM) 125 K-gluconate, 10 KCl, 0.5 EGTA, and 10 HEPES adjusted to pH 7.3 with KOH. The pipette solution for loose-patch recordings contained 150 mM NaCl and 10 mM HEPES adjusted to pH 7.4 with NaOH. Data were acquired via a Digidata 1440A digitizer (Molecular Devices) and analyzed off-line with Clampex 10 software (Molecular Devices).
All drugs used for the experiments were purchased from Sigma Aldrich (St. Louis, MO) and prepared as concentrated stock solutions that were diluted to working concentrations in the extracellular medium.
Light stimuli were generated with light-emitting diode (LED) lamps with 375-, 470-, and 525-nm wavelengths (LED supply, Randolph, VT; L.C. Corp, Brooklyn, NY). An LED controller (Mightex, Pleasanton, CA) was used to drive the LEDs, and light intensity was adjusted by varying the driving current. Light intensity was measured at the surface of the retina with an optical power meter (units converted from μW/cm2 to photons·s−1·cm−2; model 840-C, Newport, Irvine, CA).
Data were analyzed with the Clampfit 10 and SigmaPlot 12 (SYSTAT Software) software packages. Firing rates and bursting activity were measured from 90-s recordings of individual cells with the Event Detection program of the Clampfit 10 software. The coefficient of variation (CV), defined as the standard deviation of the interspike intervals (ISIs) divided by the mean of the ISIs, was used to describe the variation of ISIs in spike trains. Compared with conventional measures of variation, such as sample variance and standard deviation, the CV is scale insensitive. Distributions of ISIs with a CV < 1 were considered to have low variance, whereas those with a CV ≥ 1 were considered to have high variance.
Bursts were detected by the “Poisson surprise” method (Legendy and Salcman 1985). It was assumed that spikes in a certain interval followed a Poisson distribution. This method calculated the probability (P) that a given spike train would be found. A group of spikes (n ≥ 3) was initially identified as a burst if the burst onset initiated by two consecutive spikes with an ISI < 80 ms and terminated with two spikes having an ISI > 160 ms (Grace and Bunney 1984). Bursts were assigned a Poisson surprise value as a quantitative measure of burst strength defined by −log10P (Legendy and Salcman 1985). After the initial calculation of the Poisson surprise value, this value was maximized by adding additional intervals after the initial interval or deleting first intervals from the initial interval. The spike train was judged to be a burst if the Poisson surprise value was >1.5, meaning that the particular group of spikes had less than a 1 in 32 chance of being a random event.
For peristimulus time histograms (PSTHs), light stimuli were presented once every 2 min and each PSTH was a summation of three to five stimulus presentations using 0.3-s bins. To calculate latency to peak spiking frequency, the distribution of spiking frequency during light responses was fit with a peak function; the peak time of the fitted function relative to stimulus onset was taken as the latency to peak spiking frequency. The distribution of spiking frequency of dopamine neurons from young mice (see Fig. 7A) had a symmetrical peak shape that was best fit with a Lorentzian function, whereas the distribution from old mice (see Fig. 7B) had a rapid onset, a sharp peak, and a slow decline that was best fit with a Weibull function.
We used one-way ANOVA to determine statistical significance of the firing properties across three independent groups of data. Comparison between two independent groups was made by a Student's t-test when the data followed a normal distribution. When the distribution was not normal, a nonparametric Mann-Whitney U-test was used. To assess the effects of drugs, we determined the significance of the firing properties before and during drug application, using a paired t-test when the data had a normal distribution or a Wilcoxon signed-rank test if the data were not normally distributed. In addition, proportion comparisons were accomplished with a two-tailed Fisher's exact test. Values of the normally distributed data are given as means ± SE; we have presented nonnormally distributed data as a scatterplot along the y-axis with the mean indicated. The levels of significance were set at P < 0.05, P < 0.01, and P < 0.001.
We first examined RFP expression in whole-mount living retinas isolated from rd1 TH::RFP and opn4−/− rd1 TH::RFP transgenic mice and found that in each retina 5–15 cells were clearly marked with RFP. Figure 1A displays an RFP-marked cell from an rd1 TH::RFP retina. Both the cell body and primary dendrites were strongly labeled by RFP, which allows the cells to be visualized and targeted for recording. We verified that these RFP-labeled cells were indeed dopamine neurons by performing immunostaining with a TH antibody. It was found that the RFP-labeled cells in both rd1 TH::RFP and opn4−/− rd1 TH::RFP retinas were positive for TH (data not shown), suggesting that they are dopamine neurons, not type 2 catecholaminergic amacrine cells (Zhang et al. 2004). In addition, we examined RFP expression in rd1 het TH::RFP retinas, in which photoreceptors remain intact; the number of RFP-expressing cells and the RFP intensity within the cells were indistinguishable from those we observed in WT TH::RFP retinas (Zhang et al. 2004). The exact reason for the disappearance of RFP in most rd1 dopamine cells is unclear, but the average of 10 RFP-expressing cells randomly distributed throughout each rd1 retina was sufficient for us to perform patch-clamp recording experiments.
Dopamine neurons in rd1 retinas fire dominantly single spontaneous spikes.
Dopamine neurons respond to intrinsic and extrinsic factors to modulate spike frequency and pattern, which controls the amount of dopamine release (White 1996). We therefore determined whether these critical firing properties were preserved or modified with the loss of rod and cone photoreceptors. RFP-expressing dopamine neurons were recorded in whole-mount retinas from rd1 and rd1 het mice with a loose-patch-clamp technique. We found that the neurons recorded from retinas of both mice were spontaneously active. Although they had nearly identical firing rates, the firing patterns were clearly different.
In a total of 23 rd1 dopamine cells recorded, 6 fired in a pacemaker-like single-spiking pattern; Fig. 1B, left, illustrates a typical example. The rest of the cells exhibited additional sparse clusters of spontaneous spikes (bursting) as shown in Fig. 1B, right. The value of the CV of ISIs, a commonly used parameter for measuring dispersion from a probability distribution, was calculated for each cell. The distribution of the CV values from 23 cells is illustrated in Fig. 1D, ranging from 0.08 to 1.82 with 96% of the values <1, indicating that most cells had relatively low ISI variance.
In contrast, the CV values for rd1 het dopamine neurons ranged from 0.46 to 4.1; 63% of them were >1, indicating that most cells had high variance of ISIs (Fig. 1D; n = 19). Figure 1C shows two representative traces recorded from rd1 het dopamine neurons: one that had a low CV value (Fig. 1C, left) and another that had a high CV value (Fig. 1C, right). Comparison of the CV values between rd1 het and rd1 dopamine neurons shows a significant difference (Fig. 1D; P < 0.001, Mann-Whitney U-test), suggesting that spontaneous bursting activity of dopamine neurons is significantly decreased in rd1 retinas. This observed reduction was conserved from young to old rd1 mice (1 to 13 mo); mean CV values of rd1 dopamine cells obtained from mice in age groups of 1–2, 6–7, and 12–13 mo show no significant difference between subjects (P > 0.05, 1-way ANOVA, data not shown). In addition, the spiking rate of dopamine neurons in rd1 het retinas ranged from 2.2 to 14 Hz with a median of 5.3 Hz (n = 19), whereas it ranged from 1.5 to 8.8 Hz with a median of 5.8 Hz (n = 23) in rd1 dopamine neurons; no significant difference was observed between these two groups (Fig. 1E; P > 0.05, Mann-Whitney U-test).
The above data demonstrated that ∼74% of rd1 dopamine neurons (17/23 cells) preserved their bursting activity to some degree; it was therefore necessary to further characterize the bursting properties of these cells and compare them to rd1 het dopamine neurons. Results showed contrasting bursting properties of rd1 (n = 17) and rd1 het (n = 19) dopamine cells, with rd1 cells having fewer bursts per second (bursting frequency; Fig. 2A; P < 0.01, Mann-Whitney U-test), longer intraburst ISIs (Fig. 2B; P < 0.05, Mann-Whitney U-test), and a drastically lower percentage of spikes within bursts (Fig. 2C; P < 0.001, Mann-Whitney U-test). In addition, burst duration tended to decrease in rd1 dopamine neurons; however, this decrease was not statistically significant (Fig. 2D; P > 0.05, Mann-Whitney U-test). Collectively, our data further indicate that dopamine cell bursting activity is substantially reduced in rd1 dopamine neurons compared with rd1 het dopamine neurons.
It is notable that the bursting properties of dopamine neurons in rd1 het retinas were quite similar to those observed in WT TH::RFP retinas (Zhang et al. 2007); this finding suggests that heterozygosity for rd1, without significant changes in the number of photoreceptors, does not significantly alter dopamine neuron bursting activity. Conversely, homozygosity for rd1, associated with photoreceptor loss, remarkably reduces dopamine neuron bursting activity, suggesting that this reduction is likely a result of photoreceptor loss. Since the bursting pattern of firing is predicted to result in more efficient dopamine release at target loci, less bursting in rd1 dopamine neurons may account for the low levels of basal dopamine previously reported in animal models of retinitis pigmentosa (Floresco et al. 2003; Hankins and Ikeda 1994; Nir and Iuvone 1994).
GABAA receptor blockade restores dopamine neuron bursting activity in rd1 retinas.
We next sought to determine the underlying mechanisms responsible for the reduction in rd1 dopamine cell bursts. We have previously demonstrated that GABAergic input to dopamine neurons inhibited their bursting activity in WT retinas (Zhang et al. 2007). If GABAergic signaling increases in rd1 retinas, then this increase could reduce dopamine neuron bursting activity. To test this possibility, GABAA receptor antagonists (GABAzine or bicuculline) were administered to the retinas. Figure 3A displays representative traces before and during bath application of 20 μM GABAzine with loose-patch extracellular recording. It is clearly shown that GABAzine drastically increased the bursting activity; this increase was reversed on washout (data not shown). Data analysis demonstrates that the CV value was increased from 0.86 ± 0.12 in control to 1.48 ± 0.19 in the presence of GABAzine (Fig. 3B; n = 15; P < 0.05, paired t-test), whereas the mean firing rate was increased from 5.01 ± 0.38 Hz to 8.26 ± 0.75 Hz (Fig. 3C; n = 15; P < 0.001, paired t-test). Similar results were obtained with bicuculline in whole cell current-clamp mode. As shown in Fig. 3D, left, the resting membrane potential of the cell was −55 mV and it was accompanied by dominant single action potentials. Bicuculline (100 μM) depolarized the cell and produced oscillatory potentials that were crowned by bursts of action potentials (Fig. 3D, right). Similar results were observed in two additional cells.
We further analyzed the bursting properties of rd1 dopamine neurons before and during application of GABAzine in detail. It was found that GABAzine significantly increased bursting frequency (Fig. 4A; 0.16 ± 0.03 Hz in control vs. 0.35 ± 0.04 Hz in GABAzine; n = 12; P < 0.001, paired t-test), the percentage of spikes within bursts (Fig. 4B; 19.8 ± 3.0% in control vs. 59.6 ± 7.3% during GABAzine; n = 12; P < 0.001, paired t-test), and burst duration (Fig. 4C; n = 12; P < 0.001, Wilcoxon signed-tank test). GABAzine also significantly decreased the intraburst ISIs (Fig. 4D; n = 12; P < 0.01, Wilcoxon signed-rank test). It is worth noting that these effects are quite distinct from what we previously observed in WT dopamine neurons (Zhang et al. 2007). In WT, GABAzine significantly increased only burst duration, indicating that GABAzine had a much stronger effect on dopamine neurons in rd1 retinas than in WT retinas. This profound difference suggests that inhibitory GABAergic signaling to dopamine neurons is substantially enhanced in the rd1 retina. This enhanced signaling appears to hyperpolarize the cells, which could result in a decrease of bursting activity.
Light-evoked responses remain intact in some rd1 dopamine neurons.
It was intriguing to observe that some dopamine neurons still responded to light although all rods and most cones are degenerated in rd1 retinas (Zhang et al. 2008, 2012); here we wanted to determine more detailed information regarding this observation. We tested light responses from dopamine neurons in rd1 retinas of mice ranging from 1 to 13 mo old. Each cell was subjected to various wavelengths of light (375 nm for ultraviolet cone opsin, 470 nm for melanopsin, and 525 nm for rod and middle-wavelength-sensitive cone opsins) as well as low and high light intensities (1.26 × 1012 and 2.25 × 1013 photons·s−1·cm−2, respectively). Figure 5A shows a loose-patch recording of one cell having a rapid increase in firing frequency to a 470-nm light pulse (3-s duration, 1.26 × 1012 photons·s−1·cm−2). We also conducted whole cell current-clamp recordings and found that the light-evoked spiking increases in dopamine neurons were associated with the membrane depolarization of the cells to light (Fig. 5B). This class of light response was observed from young to old mice, although response dynamics were quite distinct with age (see below). Of a total of 33 cells tested, 30 (∼91%) exhibited an increased spike frequency on light stimulation, which in most cases persisted during light stimulation. The other three cells did not show responses to 470-nm light (Fig. 5C) or the other wavelengths tested (data not shown).
We should point out that the percentage of light-responsive dopamine neurons (∼91%) obtained in rd1 retinas is not comparable to that reported previously in WT retinas (∼60%) (Zhang et al. 2007). This is because, as described above, only ∼2% of dopamine cells (∼10 dopamine cells/retina) were labeled by RFP in rd1 retinas and the recordings were made from this small proportion of the total cells. It is unclear whether the other 98% of dopamine neurons are light responsive. In WT retinas, 89% of dopamine neurons (∼400 dopamine cells/retina) were labeled by RFP and the recordings were randomly taken from this large proportion of the cells (Zhang et al. 2004); ∼60% of those cells (∼240 cells/retina) were light responsive, while the rest were resistant to light (Zhang et al. 2007). Although we previously suggested that dopamine neuron light responses in rd1 retinas are mediated by melanopsin in ipRGCs (Zhang et al. 2008, 2012), we had not yet ruled out the possibility that these light responses may be mediated by light activation of surviving cones or other unidentified photosensitive cells in the inner retina.
Melanopsin knockout eliminates light-evoked responses of dopamine neurons in rd1 retinas.
To test whether the remaining light responses of rd1 dopamine neurons were mediated by the photopigment melanopsin, we genetically knocked out this light-absorbing molecule in rd1 TH::RFP transgenic mice (opn4−/− rd1 TH::RFP mice) and reexamined dopamine neuron light responses in these subjects. We started with 1- to 2-mo-old animals, in which the cell bodies of some cone cells are still present in the retina (Lin et al. 2009). It was found that none of the 13 dopamine neurons recorded was sensitive to 375-, 470-, or 525-nm light (light intensities: 1.26 × 1012 or 2.25 × 1013 photons·s−1·cm−2). Figure 6, A–C, displays one such example using these three wavelengths at an intensity of 1.26 × 1012 photons·s−1·cm−2. The same results were observed in six cells recorded from the retinas of 3- to 4 mo-old mice (data not shown). We further compared the proportion of light-responsive cells in rd1 retinas with and without melanopsin (30/33 in rd1 retinas vs. 0/19 in opn4−/− rd1 retinas; Fig. 6D); the difference between them is statistically significant (P < 0.001, 2-tailed Fisher's exact test). These results clearly indicate that dopamine neuron light responses in rd1 retinas are mediated by melanopsin phototransduction and not by surviving cones.
Melanopsin-mediated light responses of dopamine neurons increase from young to old rd1 animals.
The structural and functional composition of retinal synaptic circuitry generally undergoes extensive remodeling during retinal degeneration, particularly during phase 3 (phases 1 and 2 are rod degeneration and cone degeneration, respectively) (Marc et al. 2003, 2007; Vugler 2010). We therefore hypothesized that the novel neural pathway from ipRGCs to dopamine neurons is functionally modified from phase 2 to phase 3. To test this hypothesis, we measured light-evoked responses of dopamine neurons from the retinas of rd1 mice at ages of 1–2 mo (phase 2) and 6–7 mo (phase 3) and then compared the age group characteristics. It appears that the responses from younger mice (Fig. 7A, left) peaked more slowly than those from older mice (Fig. 7B, left). The PSTH was successfully fit with a Lorentzian distribution for dopamine neurons recorded in younger mice (Fig. 7A, right) and a Weibull distribution for the recordings collected from older mice (Fig. 7B, right). From the fitted curves we measured the peak latency (time from light onset to peak) for each response and found that the peak latency significantly decreased from 1.95 ± 0.12 s (n = 5) in younger mice to 0.94 ± 0.14 s (n = 5) in older mice (P < 0.001, Student's t-test). In contrast, the peak firing frequency of the responses was much lower in younger mice than in older mice. The firing rate peak amplitudes (from curve: mean spontaneous firing rate subtracted from the peak firing rate) significantly increased from 9.7 Hz in younger mice (n = 5) to 85.8 Hz in older mice (n = 5; P < 0.05, Mann-Whitney U-test). These results suggest that the strength of melanopsin-driven signaling to dopamine neurons increases from phase 2 to phase 3 of retinal degeneration.
In this study, we have taken advantage of our transgenic mouse line in which dopamine neurons are marked by RFP under control of the TH promoter in combination with rd1 mutant mice in which rods rapidly degenerate soon after birth. Using this genetic strategy we have been able to visualize dopamine neurons and target them for recording in the diseased mice. The major findings of our study are that 1) dopamine neurons in rd1 retinas retain their spontaneous spiking activity with a decrease in bursting; 2) burst reduction is mediated by enhanced GABA signaling; 3) light responses of rd1 dopamine neurons are mediated exclusively by melanopsin photoreception and not by surviving cones; and 4) the retrograde signaling pathway from ipRGCs to dopamine neurons is enhanced in advanced retinal degeneration.
Physiological properties of dopamine neurons are partially preserved in absence of rods and cones.
Our results show that dopamine neurons in the rd1 retina, as in WT, fire spontaneously; this finding indicates that these diseased dopamine neurons have the ability to generate the action potentials required for dopamine release. In addition, dopamine neuron activity can also be regulated by external factors (e.g., light), suggesting that these cells continue to communicate with other retinal cells involved in regulating dopamine release. Furthermore, previous reports have indicated that dopamine neuron morphology is not significantly altered by the loss of rods and cones (Hankins and Ikeda 1994; Strettoi et al. 2002). Taken together, the overall results suggest that dopamine neuron structure and function are at least partially preserved in rd1 retinas. Additionally, retinal ganglion cells, another type of third-order retinal neuron, also show this preserved structure and function in rd1 retinas (Margolis et al. 2008; Mazzoni et al. 2008; Stasheff 2008; Ye and Goo 2007). Because some vision-restorative treatment strategies for retinal degeneration (e.g., photoreceptor cell transplantation) require donated cells to make connections with existing retinal circuitry and inner retinal neurons, the preservation of the third-order retinal neurons may be important for these therapies.
Reduction of dopamine neuron spontaneous bursting activity in rd1 retinas is mediated by GABA.
Although rd1 dopamine neurons retain a normal firing rate, our data show that their firing patterns are altered. The cells were observed to fire either in an irregular single-spiking pattern or with sparse bursts. This is in contrast to rd1 het dopamine neurons, which exhibit a relatively high bursting frequency, a high percentage of spikes within bursts, and short intraburst ISIs. Considering that the rd1 and the rd1 het mice were of the same genetic background, this distinction cannot be attributed to background difference.
Our results demonstrate that the loss of rods and cones affects the GABAergic system to alter dopamine neuron spontaneous firing pattern. This is shown through the observation that GABAA receptor antagonists depolarized rd1 dopamine neurons as well as restored their loss of bursts. The simplest interpretation of this effect is that GABA content increases during retinal degeneration (Murashima et al. 1990); however, we cannot rule out molecular alterations of GABA receptors on rd1 dopamine neurons, which could be causing the increased responsiveness to GABA (Yazulla et al. 1997). These possibilities include, but are not limited to, changes in receptor sensitivity, the number of receptors, receptor subunit composition, and endogenous ligands of auxiliary sites such as benzodiazepine (Wong et al. 2003).
Interestingly, the firing pattern changes seen in rd1 dopamine neurons are completely opposite of rd1 retinal ganglion cells, which display hyperactivity and rhythmic bursting (Margolis et al. 2008; Stasheff 2008; Ye and Goo 2007; Yee et al. 2012). Recent reports suggest that intrinsic oscillatory potentials were induced in bipolar cells (and AII amacrine cells) of rd1 retinas (Borowska et al. 2011; Menzler and Zeck 2011), thus providing the presynaptic mechanism responsible for retinal ganglion cell burst generation. If this were the case for dopamine neurons, these neurons would also have increased bursting activity; however, we observed the opposite effect that dopamine neuron bursting activity was decreased. It is evident that the loss of rods and cones alters dopamine neurons and retinal ganglion cells through distinctly different mechanisms; these mechanisms should be carefully considered when designing treatment strategies for retinitis pigmentosa.
Melanopsin mediates light-induced responses of dopamine neurons in rd1 retinas.
In our previous studies we have found that some dopamine neurons were responsive to light in the rd1 retina (Zhang et al. 2008). The present study provides evidence that these light responses are mediated exclusively by melanopsin phototransduction; this is evident because of the complete absence of a light-evoked response in opn4−/− rd1 retinas. This result is consistent with our recent report that normal retinas lacking melanopsin only exhibited rod- and cone-driven transient responses (Zhang et al. 2012). It is therefore conceivable that melanopsin mediates retrograde signaling from ipRGCs to dopamine neurons in normal retinas and that this signaling is well-preserved in hereditary retinal dystrophies.
Our study likely excludes the involvement of surviving cones in driving dopamine neuron activity in rd1 retinas. The youngest opn4−/− rd1 mouse used was 42 days old, an age at which a significant number of cones are still present in the retina; however, we were unable to obtain any light responses from these dopamine neurons with various light intensities and wavelengths. This finding is not surprising considering that previous studies have shown that cones surviving for extended periods of time in rd1 retinas appear incapable of sending light-driven information through the retina to mediate visual or nonvisual functions (Panda et al. 2003; Pearson et al. 2012; Provencio et al. 1994; Strettoi et al. 2002). Our results also suggest that other inner retinal photoreceptive sources, such as paradoxical opsin- or UV-sensitive Opn5 (Kojima et al. 2011; Semo et al. 2007), are unlikely to have any involvement in mediating light regulation of dopamine neuron activity.
Retrograde melanopsin signaling to dopamine neurons is enhanced in advanced retinal degeneration.
We compared light-evoked responses of dopamine neurons between young and old rd1 mice and found that signal transmission from ipRGCs to dopamine neurons is increased with age. The increase is unlikely to be caused by aging because such phenomena were not seen in our preliminary studies using the same ages of young and old WT mice (data not shown). Alternatively, this increase appears to be a result of retinal remodeling occurring in advanced retinal degeneration. At this point it is unclear which underlying mechanisms are involved in this process, but it is unlikely that this alteration is due to changes in ipRGCs, the presynaptic photosensitive neurons. This argument is strengthened by the observation that the number of ipRGCs does not increase and that melanopsin expression in individual cells remains unchanged in rd1 retinas (Vugler et al. 2008). We therefore speculate that the ipRGC-dopamine neuron neural circuit undergoes structural rewiring during progressive retinal degeneration that increases the number of contacts between dopamine neurons and ipRGCs. Other possibilities could be that during retinal degeneration glutamate released from ipRGCs increases and/or that there are an increased number of glutamate receptors on dopamine neurons, which could also result in enhanced melanopsin signaling to dopamine neurons. Future studies are needed to test these possibilities.
The retrograde melanopsin-signaling pathway to dopamine neurons is newly discovered, and its function remains fully established in normal and diseased retinas. Without rods and most cones in the rd1 retina, melanopsin is the only functional photopigment that is expected to drive dopamine release. Since there is a limited number of dopamine neurons driven by melanopsin signaling, the amount of retinal dopamine may be undetectable by high-performance liquid chromatography (a method that measures tissue dopamine content and does not detect the functional pool of dopamine) (Cameron et al. 2009); a future study will certainly need to utilize a highly sensitive carbon fiber electrode to detect the dopamine released from dopamine neurons in situ (Puopolo et al. 2001).
In summary, we have shown that dopamine neuron spontaneous activity is altered in the disease with less bursting, which could result in a decreased release of dopamine. This answers the long-standing question as to why dopamine release is decreased although the number of dopamine neurons and dopamine synthesis remain unchanged in retinal degeneration (Hankins and Ikeda 1994; Kato et al. 1981; Nir and Iuvone 1994). A similar mechanism may also apply to other retinal neurodegenerative diseases, such as diabetic retinopathy, in which the dopaminergic system is impaired (Nishimura and Kuriyama 1985). In addition, our results indicate that melanopsin is the photopigment mediating light regulation of dopamine neurons in rd1 mice. This provides a molecular mechanism by which light regulates the retinal circadian clock through the release of dopamine in the absence of rods and cones (Doyle et al. 2002; Ribelayga et al. 2008; Ruan et al. 2008; Vugler et al. 2007). Furthermore, dopamine can act via D1 family receptors on ipRGCs through a feedback mechanism that may shape the responses of ipRGCs and thus of non-image-forming visual functions including the pupillary light reflex and circadian photoentrainment (Panda et al. 2003; Van Hook et al. 2012).
The work was supported by an Oakland University Provost's Graduate Student Research Award (C. L. Atkinson), the Kwang-Hua Education Foundation (J. Feng), the Oakland University Research Excellence Fund (D.-Q. Zhang), and a Midwest Eye-Banks Research Award (D.-Q. Zhang).
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: C.L.A., J.F., and D.-Q.Z. conception and design of research; C.L.A., J.F., and D.-Q.Z. performed experiments; C.L.A. and D.-Q.Z. analyzed data; C.L.A. and D.-Q.Z. interpreted results of experiments; C.L.A. and D.-Q.Z. prepared figures; C.L.A. and D.-Q.Z. edited and revised manuscript; C.L.A., J.F., and D.-Q.Z. approved final version of manuscript; D.-Q.Z. drafted manuscript.
We thank Dr. Douglas McMahon for his support during the initial work of this project and Dr. Frank Giblin for his critical review of the manuscript.
- Copyright © 2013 the American Physiological Society