Understanding the regulation of L-type voltage-gated Ca2+ current is an important component of elucidating the signaling capabilities of retinal amacrine cells. Here we ask how the cytosolic Ca2+ environment and the balance of Ca2+-dependent effectors shape native L-type Ca2+ channel function in these cells. To achieve this, whole cell voltage clamp recordings were made from cultured amacrine cells under conditions that address the contribution of mitochondrial Ca2+ uptake (MCU), Ca2+/calmodulin (CaM)-dependent channel inactivation (CDI), protein kinase A (PKA), and Ca2+-induced Ca2+ release (CICR). Under control conditions, repeated activation of the L-type channels produces a progressive enhancement of the current. Inhibition of MCU causes a reduction in the Ca2+ current amplitude that is dependent on Ca2+ influx as well as cytosolic Ca2+ buffering, consistent with CDI. Including the Ca2+ buffer bis-(o-aminophenoxy)-N,N,N',N'-tetraacetic acid (BAPTA) internally can shift the balance between enhancement and inhibition such that inhibition of MCU can enhance the current. Inhibition of PKA can remove the enhancing effect of BAPTA suggesting that cyclic AMP-dependent phosphorylation is involved. Inhibition of CaM suppresses CDI but spares the enhancement, consistent with the substantially higher sensitivity of the Ca2+-sensitive adenylate cyclase 1 (AC1) to Ca2+/CaM. Inhibition of the ryanodine receptor reduces the current amplitude, suggesting that CICR might normally amplify the activation of AC1 and stimulation of PKA activity. These experiments reveal that the amplitude of L-type Ca2+ currents in retinal amacrine cells are both positively and negatively regulated by Ca2+-dependent mechanisms.
Amacrine cells are retinal interneurons that signal extensively in the inner plexiform layer of the retina. The functions of amacrine cells are diverse but include shaping the response properties of ganglion cells, the output cells of the retina (Baccus 2007; Demb 2007; Zhou and Lee 2008). These key players in retinal signal processing often participate in serial (Dowling and Boycott 1966; Dubin 1970; Guiloff et al. 1988; Pollard and Eldred 1990; Zhang et al. 1997) and reciprocal synapses (Hartveit 1999; Shields and Lukasiewicz 2003; Vigh and von Gersdorff 2005), implying that local synaptic environments might be regulated independently of one another. It has been established for several classes of amacrine cells that L-type Ca2+ channels are employed at their synapses to control neurotransmitter release (Bieda and Copenhagen 2004; Gleason et al. 1994; Habermann et al. 2003; Vigh and Lasater 2004). Thus the regulation of these channels can play a central role in visual signal processing.
The pore-forming region of the L-type channel is encoded by one of four genes: CaV1.1–1.4. CaV1.1 is expressed in skeletal muscle, CaV1.2 and -1.3 are the dominant L-type channels in the brain (Hell et al. 1993), and CaV1.4 is expressed predominately at ribbon synapses in the retina (Strom et al. 1998). L-type Ca2+ channels are distinctive in that they can support fairly sustained levels of Ca2+ influx. This Ca2+ influx can have a variety of effects including regulation of the channel itself via Ca2+/calmodulin (CaM)-dependent inactivation (CDI) that occurs for most CaV1 and -2 (non-L-type) Ca2+ channels (for review see, Halling et al. 2005).
The molecular players and details of this inactivation have been described by an elegant set of experiments on CaV1/2 channels (Dick et al. 2008; Tadross et al. 2008). The efficiency of the inactivation process is optimized by the preassociation of CaM to the channel (Erickson et al. 2001; Pitt et al. 2001). Ca2+ entering through the channel binds CaM and inactivation is initiated. In CaV1.2/1.3 channels, the CaM sensors detect both local and global Ca2+ (Dick et al. 2008). The local concentration of Ca2+ eliciting this response is on the order of 100 μM, which only exists within hundreds of angstroms of the channel pore (Neher 1998; Sherman et al. 1990).
Another known regulator of L-type Ca2+ channels is protein kinase A (PKA). Phosphorylation of L-type channels by PKA enhances the whole cell current amplitude by increasing the open time of the channels (Bean et al. 1984; Yue et al. 1990b). The level of PKA activity can be regulated by cell surface receptors linked to G proteins that either stimulate (Gs) or inhibit (Gi) adenylate cyclase (AC). There are nine membrane bound isoforms of AC, all of which can be stimulated by activated Gs (for review, see Willoughby and Cooper 2007). Alternatively, AC1 and AC8 can be directly activated by the Ca2+/CaM complex with AC1 being about five times more sensitive to Ca2+/CaM (Kds ∼100 nM, AC1; ∼500 nM, AC8) (Fagan et al. 1996; Wu et al. 1993). We have previously reported that metabotropic glutamate receptor 5- and phospholipase C-dependent activation of PKA enhances the amplitude of L-type Ca2+ currents in retinal amacrine cells, possibly via an AC1-dependent mechanism (Sosa and Gleason 2004).
If these two Ca2+-dependent Ca2+ channel regulators (CDI and AC1/8) coexist in amacrine cells, then we would predict that mechanisms regulating cytosolic Ca2+ will influence the outcome of L-type Ca2+ channel regulation. It has been previously shown that synaptic transmission between retinal amacrine cells is affected by mitochondrial Ca2+ uptake (MCU) (Medler and Gleason 2002). This work led us to hypothesize that at least part of the impact of MCU on synaptic transmission was in maintaining the L-type Ca2+ channels in a relatively noninactivated state. Here we test this hypothesis by examining the effects of disrupting MCU on L-type Ca2+ channel function.
Entry of Ca2+ through L-type channels is known to elicit Ca2+-induced Ca2+ release (CICR) in amacrine cells (Mitra and Slaughter 2002; Warrier et al. 2005). Because this amplification of the Ca2+ signal has the potential to affect other Ca2+-dependent processes, we also test the role of ryanodine receptor (RyR) activity in the Ca2+-dependent regulation of these channels.
Using a primary cell culture system consisting of identified GABAergic amacrine cells (Gleason et al. 1993), we have begun to clarify the physiological relationships between L-type Ca2+ channel inactivation, MCU, PKA activity, and CICR. Although the molecular details of CDI and PKA-dependent current enhancement have been worked out, most of this work has been done in expression systems or in cardiac myocytes. It remains to be determined how these factors interact in the native environment of retinal amacrine cells; an interneuron critical to shaping the output of the retina. Given the dependence of synaptic transmission on L-type Ca2+ channel function in these cells (Gleason et al. 1994), our aim is to investigate the balance of Ca2+-dependent mechanisms regulating L-type Ca2+ channel functions in retinal amacrine cells.
Primary cell cultures of chick retinal amacrine cells were used in our experiments. The chicken embryos (Gallus gallus, Animal Science Department, Louisiana State University, Baton Rouge, LA) were dissected on embryonic day 8, and retinal cells were dissociated and cultured as previously described (Hoffpauir and Gleason 2002). Cell cultures were maintained at 37°C under 5% CO2 atmosphere until they were ready for experiments, 8–14 days after plating. For electrophysiology experiments, amacrine cells were identified based on their morphology. Cells with large somas (10–15 μm) with two to five primary processes have been previously identified as amacrine cells based on immunocytochemical and physiological criteria (Gleason et al. 1993; Huba and Hofmann 1990, 1991; Huba et al. 1992).
Unless otherwise indicated, all reagents were purchased from Sigma-Aldrich (St. Louis, MO). External solutions consist of the following (in mM): 116.7 NaCl, 20.0 TEACl, 3.0 CaCl2, 0.4 MgCl2, 5.6 glucose, and 10.0 HEPES. Voltage clamp experiments performed in the perforated-patch configuration employed the following internal solution (in mM): 135.00 CsAc, 10.0 CsCl, 1.0 NaCl, 2.0 MgCl2, 0.1 CaCl2, 1.1 EGTA, 10.0 HEPES, and 200 μg/ml amphotericin B. Voltage clamp experiments performed in the ruptured-patch configuration used the following internal solution (in mM): 100.00 CsAc, 10.0 CsCl, 2.0 MgCl2, 0.1 CaCl2, 10.0 HEPES, 3.0 ATP (dipotassium), 1.0 ATP (disodium), 20.0 phosphocreatine, 2.0 GTP, and 50 U/ml creatine phosphokinase. Solutions were adjusted to pH 7.4 with NaOH for external solutions and with CsOH for internal solutions. Two different Ca2+ buffers were also included in internal solutions in ruptured-patch recordings: Ethylene glycol-bis (2-aminoethyl-ether)-N,N, N′,N′-tetraacetic acid (EGTA, 1.1 or 14 mM) and 1,2-bis-(o-aminophenoxy) ethane-N,N, N′,N′-tetraacetic acid (BAPTA, 10 mM, Enzo Life Sciences, Plymouth Meeting, PA).
A pressurized gravity flow perfusion system (1.5–2 ml/min) was used to deliver the external solutions (AutoMate Scientific, Berkeley, CA). Unless otherwise indicated, the following reagents were purchased from Enzo Life Sciences. Reagents added via the bath included the protonophore carbonylcyanide-4-(trifluoromethoxy)-phenylhydrazone (FCCP, 1 μM), a PKA inhibitor N-[2-(p-bromocinnamylamino) ethyl]-5-isoquinolinesulfonamide · 2HCl (H89, 1 μM,), an adenylate cyclase (AC) inhibitor 9-(tetrahydro-2-furanyl)-9H-purin-6-amine, (SQ 22,536, 200 μM) a general phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (IBMX, 100 μM), a calcineurin inhibitor cyclosporine A (CsA, 1 μM), and a selective inhibitor of Ca2+/CaM-dependent phosphodiesterase (PDE 1) 8-methoxymethyl-1-methyl-3-(2-methylpropyl) xanthine (8-M-IBMX, 100 μM), inhibitors of the RyR, ryanodine (14 μM) and dantrolene (20 μM, Sigma). CaM inhibitor calmidazolium chloride (CMZ, 10 μM) was added to the pipette solution. In all electrophysiology experiments, (−)-bicuculline methobromide (10 μM, Tocris Bioscience, Ellisville, MO) was included in external solutions to block GABAA receptor-mediated autaptic currents (Gleason et al. 1993). Tetrodotoxin (TTX, 300 nM, Alomone Labs, Jerusalem, Israel) was included in external solutions to block voltage gated Na+ currents.
Cell culture dishes were mounted on the stage of an Olympus IX70 inverted microscope. A reference Ag/AgCl pellet served to ground the bath. Patch electrodes were pulled from thick-walled borosilicate glass with a filament (1.5 mm OD, 0.86 mm ID; Sutter Instrument, Novato, CA) using a Flaming-brown Micropipette Puller (Sutter Instruments). For electrophysiology experiments, either ruptured- or perforated-patch whole cell recording was performed. For perforated-patch recordings, only cells with stable resistances (changes of <5 MΩ) were used in the experiments. Recordings were made using Axopatch 1D-patch clamp amplifier (Molecular Devices, Sunnyvale, CA). Data were recorded using Clampfit 9.2 and 10.0 software (Molecular Devices). Electrode resistance values were monitored and ranged from 3 to 8 MΩ. Junction potential corrections (−8 mV) were made for the data in Fig. 2, D and E.
Because we were concerned that some experimental manipulations might produce long term changes in the cells, experimental and control recordings were often done in separate groups of cells. We also used separate groups of cells for experiments with different internal solutions. When this sort of protocol was used, the different treatments (or internal solutions) were alternated from cell to cell. As such, comparisons were made between cells from the same culture and often from the same culture dish.
Chicken and mouse brains were homogenized in nondenaturing lysis buffer containing a cocktail of protease inhibitors [PMSF (1 mM), leupeptin, (5 μg/ml) aprotinin (2.5 μg/ml), 1,10 ortho-phenantrolin (0.2 μg/ml), and pepstatin (0.7 μg/ml)]. Samples were spun at 4,000 rpm for 20 min at 4°C. Protein content was determined using the BCA protein assay kit from Pierce (Rockford, IL). Proteins (300 μg) were separated on a 7.5% SDS gel along with 10 μl Pageruler molecular weight markers (Fermata, Glenburnie, MD). Proteins were transferred to nitrocellulose membranes. Membranes were blocked overnight at 4°C in 4% milk, 0.1% Tween 20 in Tris buffered saline. The polyclonal antibody raised against human AC1 (Abcam, Cambridge, MA) was diluted 1:500 in PBS with 1% BSA. Goat anti-rabbit secondary antibody conjugated on horseradish peroxidase (Pierce) was diluted to 1:1,000 in PBS with 1% milk. Membranes were incubated in primary and secondary antibodies for 1.5 h each, at room temperature. Proteins were visualized using the Supersignal Western Pico Reagent (Pierce).
Adult White Leghorn chickens were killed by intraperitoneal injection of sodium pentobarbital (500 mg/kg, Sigma-Adrich) followed by decapitation. These methods were approved by the Institutional Animal Care and Use Committee, Louisiana State University. The eyes were enucleated and hemisected. After removing the vitreous, 4% paraformalydehyde was added to the eye cups and kept at 4°C for 1 h. Following fixation, eye cups were washed in PBS +1% glycine. Retinae were then dissected from eye cups and incubated in 15% sucrose for 30 min, 20% sucrose for 1 h, and 30% sucrose solution overnight at 4°C. Retinae were embedded in OCT compound (Sakura Finetek, Torrence, CA) by freezing in dry ice and isopentane. Sections (12–16 μm) were cut on a Leica CM1850 cryostat (Wetzlar, Germany) and mounted on presubbed slides (Southern Biotech, Birmingham, AL).
Cells grown on glass coverslips were fixed in 2% paraformaldehyde for 30 min at 4°C after 8 days in culture. Fixed cells and retinal sections were preincubated for 1 h at room temperature in a blocking solutions consisting of dilution solution (see following text) with 10% normal goat serum. Primary polyclonal antibodies raised against human AC1 were diluted at 1:100 in dilution solution (PBS, 1% bovine serum albumin, 0.5% saponin) and applied to either retinal sections or cells for 1 h at room temperature then washed in PBS. Secondary goat-anti-rabbit antibodies conjugated to Cy3 were obtained from Millipore (Temecula, CA) and were diluted 1:1,000 in dilution solution. Cells were incubated in secondary antibodies for 1 h at room temperature. After washing, coverslips and slides were mounted in a medium containing 70% glycerol, 28% PBS and 2% n-propy gallate. Cells and retinal sections were viewed on an Olympus 1X70 microscope equipped with epifluorescence and images were captured using Slidebook software and hardware (Intelligent Imaging Innovations, Denver, CO).
The Origin 7.5 and 8.0 (OriginLab, Northampton, MA) software package was used to analyze and plot the data. Images in Fig. 8 were adjusted for brightness and contrast in Adobe Photoshop (San Jose, CA). Equivalent adjustments were made for experimental and control images. Figures were assembled in Adobe Illustrator. Statistical analyses were done using the t-test and data are presented as mean ± SE. Maximum P value for significance was 0.05.
Ca2+ current amplitude increases with repeated depolarizations
Whole cell recordings were made from single isolated amacrine cells in the perforated-patch configuration. It has been previously established that these cells express L-type but not N or P-type Ca2+ channels (Gleason et al. 1993). To examine the normal variability in current amplitude over time, Ca2+ currents were elicited by depolarizing amacrine cells from −70 to 0 mV either for 1,000 ms, every 60 s, or for 100 ms every 30 s (Fig. 1, A and C, respectively). It is evident, especially for the longer voltage steps (Fig. 1A), that inactivation of the current occurs over the duration of the step. It was also observed that the time course and degree of inactivation could vary from cell to cell. The physiological basis for these differences are not known but could feasibly be due to different expression levels of calmodulin or other effectors (see discussion). The tail currents after the voltage step (observable in Fig. 1C as well as in subsequent figures) are primarily due to the activity of the plasma membrane Na/Ca exchanger transporting Ca2+ back out of the cell. Our ability to identify and to measure electrogenic Na/Ca exchange activity has been firmly established for these cells (Gleason et al. 1994, 1995; Hurtado et al. 2002; Medler and Gleason 2002).
Although inactivation occurred during the voltage step, the peak current amplitudes tended to increase over time, implying that multiple levels of regulation are occurring. Figure 1, B and D, shows the peak current amplitude for steps (1 s and 100 ms) delivered every 60 and 30 s (respectively) from separate populations of amacrine cells. A progressive increase in the current amplitude was typically observed with both protocols. This increase is not due to changes in series resistance because only cells with stable series resistances (see methods) were included in the analysis. The rate of the increase in current amplitude was inherently variable among cells as indicated by the substantial error bars that tended to increase over the duration of the recording. Because the only “treatment” that the cells received in these recordings was the history of voltage steps and the resulting Ca2+ influx, we predicted that the primary sources of enhancement were Ca2+ dependent. Perhaps then, variability in current amplitude enhancement was related to the Ca2+ current density in each amacrine cell. To determine whether the rate of increase in current amplitude was linked to Ca2+ current density, these quantities were plotted for a population of cells that had been recorded under the same conditions (step to 0 mV for 100 ms, every 30 s, Fig. 1E). Regression analysis revealed an R2 value of 0.02 (Fig. 1E) and does not indicate a dependence on current density. This implied that the regulation under these stimulus protocols is not a simple transform based on the amount of Ca2+ influx. Instead it suggested that the regulation of these channels has multiple Ca2+-dependent components that vary intrinsically among cells. This suggestion is borne out by much of the data presented in subsequent sections.
Disruption of MCU inhibits the Ca2+ current
To test the hypothesis that L-type Ca2+ channel inactivation in amacrine cells is regulated by MCU (Medler and Gleason 2002), we used the protonophore FCCP to temporarily collapse the proton gradient across the inner mitochondrial membrane and disrupt MCU via the Ca2+ uniporter (Herrington et al. 1996; Werth and Thayer 1994; White and Reynolds 1997). Whole cell current recordings were made in the perforated-patch configuration. Amacrine cells were depolarized from −70 to 0 mV for 1 s, every 60 s. Disrupting MCU had two effects: an increase in the inward current amplitude recorded at −70 mV(before and after the voltage step to 0 mV) and a decrease in the Ca2+ current amplitude recorded during the step to 0 mV. We have previously shown that the relatively time-invariant increase in inward current at −70 mV (Fig. 2A, arrow) is due to a persistent FCCP-dependent increase in cytosolic Ca2+ that activates the electrogenic plasma membrane Na/Ca exchanger (Medler and Gleason 2002). Furthermore, it has been established that the FCCP-dependent Ca2+ increase driving this exchanger activity is due to the RyR-dependent leakage of Ca2+ from stores that is normally sequestered by mitochondria via the uniporter (Sen et al. 2007). Importantly, we have also previously demonstrated that Na/Ca exchange activity is negligible at 0 mV and thus does not contribute significantly to the current recorded during the voltage step to 0 mV (Gleason et al. 1995). To simplify the appearance of the data, the FCCP-dependent Na/Ca exchange current at −70 mV has been subtracted from subsequent data (as in Fig. 2B). However, we show this current in insets (Figs. 5 and 6) to confirm that this FCCP-dependent Ca2+ elevation and exchanger activity persists under some key experimental conditions.
More directly relating to our hypothesis, disrupting MCU also caused a reversible decrease in the Ca2+ current amplitude (Fig. 2, A and B, gray trace). Under these conditions, disrupting MCU significantly reduced the Ca2+ current amplitude by 35 ± 7% (n = 7, P = 0.03, Fig. 2C). A series of control experiments previously established that the effects of FCCP on depolarization-induced Ca2+ elevations in these cells are not due to ATP depletion (Medler and Gleason 2002) (also see discussion) or changes in pH (Sen et al. 2007). Another possibility was that the decrease in current amplitude was due to an FCCP-dependent shift in the activation range of the channels. To address this, currents were elicited by steps over a range of voltages from −80 to +10 mV in 5 mV increments to reveal the voltage of activation. These experiments were done in ruptured-patch because the voltage of activation can be better resolved with the relatively low series resistance recordings achieved in this configuration. These recordings were made with two different internal Ca2+ buffering conditions (1.1 mM EGTA, 10 mM BAPTA). Although we observed a difference in the effects of FCCP on Ca2+ current amplitude in 10 mM BAPTA (see following text), no FCCP-dependent shift in activation range was observed under either Ca2+ buffering condition (Fig. 2, D and E, respectively). These data are consistent with a role for MCU in limiting the degree of Ca2+/CaM-dependent inactivation for these channels.
Effects of disrupting MCU are dependent on the duration and frequency of the voltage step
To examine whether increases in Ca2+ influx enhanced the effects of disrupting MCU, we first altered the duration of the voltage steps. Ca2+ currents were recorded in response to voltage steps lasting either 50 ms or 3 s (Fig. 3, A and B, respectively). Currents from 50 ms steps were reduced by 35 ± 4% (n = 7), whereas currents from 3 s steps were reduced by 60 ± 5% in the presence of FCCP (n = 6, C). The significantly (P = 0.002) larger effect of FCCP for longer voltage steps supports the possibility that Ca2+-dependent inactivation was being altered by disrupting MCU.
If the level of FCCP-dependent inhibition of L-type channel current increased with prolonged voltage steps, and this was due to the greater Ca2+ influx during longer steps, we would also predict that increasing the frequency of depolarization would intensify the effects of inhibiting MCU. To test this, single amacrine cells were depolarized from −70 to 0 mV, for 100 ms, and voltage steps were delivered either every 60 s (Fig. 3D) or every 5 s (E). Disrupting MCU caused a reduction in the Ca2+ current amplitude under both recording condition; however, the FCCP-dependent reduction in current amplitude was significantly larger for higher frequency depolarizations [61 ± 4% reduction for every 5 s (n = 6), and 35 ± 4% reduction for every 60 s (n = 7), P = 0.001, Fig. 3F]. These results further support the possibility that inhibition of the Ca2+ current amplitude is dependent on cytosolic Ca2+ levels.
Effects of MCU are not dependent on calcineurin
The established role of PKA-dependent channel phosphorylation in L-type current enhancement raises the possibility that the effect of inhibiting MCU is to reduce channel phosphorylation. The most likely candidate for such an activity under these conditions would be the Ca2+/CaM-dependent phosphatase, calcineurin. To test for the involvement of calcineurin, we asked whether the potent (IC50 = 7 nM) (Fruman et al. 1992) calcineurin inhibitor CsA (for review, see Kunz and Hall 1993) would block the effects of FCCP. Recordings were made using 100 ms steps from −70 to 0 mV, every 30 s. Cells were preincubated for 10–20 min in CsA (1 μM) before FCCP was applied. The inhibitory effect of FCCP persisted in the presence of CsA (Fig. 4, A and B), and CsA alone had no consistent effect on the current (C). Negative results require cautious interpretation but it is important to note that with similar exposure times (10–20 min) and concentrations (0.1–1 μM), CsA has been demonstrated to be effective in inhibiting calcineurin in intact neurons (Xu and Krukoff 2007) and smooth muscle cells (Schuhmann et al. 1997). These data therefore suggest that calcineurin activity is not mediating the FCCP-dependent inhibition of the current and may not be a major regulator of L-type Ca2+ channel activity in these amacrine cells.
Effects of MCU are Ca2+ influx-dependent
Were the effects of disrupting MCU on the current amplitude due to an excess of Ca2+ originating from channel entry or were they due to the FCCP-dependent elevation in cytosolic Ca2+ as revealed by the increase in the activity of the Na/Ca exchanger (Fig. 2A)? To distinguish between these two possibilities, we replaced external Ca2+ with Ba2+. Ba2+ is known to carry the current through L-type channels but is a poor substitute for Ca2+ with respect to inactivation (Brehm and Eckert 1978; Chad and Eckert 1986; Tillotson 1979; Yue et al. 1990a; Zuhlke et al. 1999). Perforated-patch recordings were made in amacrine cells bathed with either normal external (3 mM Ca2+, Fig. 5A) or external solution with equimolar Ba2+ replacing Ca2+ (B). Cells were stepped from −70 to 0 mV for 1 s. The currents recorded in Ba2+ inactivated relatively little over the course of the 1 s depolarization consistent with minimal Ca2+-dependent inactivation. With Ba2+ as the charge carrier, the effect of disrupting MCU was significantly reduced (18 ± 3% reduction with Ba2+, n = 7; 44 ± 7% reduction with Ca2+, n = 7; P = 0.007) indicating that the primary source of the inhibition was from Ca2+ crossing the plasma membrane (Fig. 5C). The relatively small inhibitory effect of FCCP on Ba2+ current amplitude that was observed may be due to either the low level of CaM activation known to occur with Ba2+ (Dick et al. 2008) and/or the FCCP-dependent Ca2+ elevation (Medler and Gleason 2002; Set et al. 2007). The persistence of the FCCP-dependent Ca2+ elevation in external Ba2+ is demonstrated by the presence of the FCCP-dependent Na/Ca exchange current shown in Fig. 5B, inset. The relatively small effect of inhibiting MCU on Ba2+ current amplitude is consistent with the hypothesis that MCU normally sequesters Ca2+ entering through L-type Ca2+ channels.
Increasing cytosolic Ca2+ buffering alters the effects of inhibiting MCU
Different levels of internal Ca2+ buffering are known to alter CDI (Brehm and Eckert 1978; Dick et al. 2008; Kalman et al. 1988; Kohr and Mody 1991; Tadross et al. 2008). To further explore the role of CDI in the effects of blocking MCU, experiments were repeated under different Ca2+ buffering conditions. Either EGTA (1.1 or 14 mM) or the faster buffer BAPTA (10 mM) (Adler et al. 1991) was included in recording pipette. For these experiments, Ca2+ currents were recorded in the ruptured-patch configuration. Currents were recorded before (Figs. 6, A–C, black traces) and during application of FCCP (gray traces). First, it should be noted that the different buffering environments differentially affect the two components of the Na/Ca exchange current. The FCCP-dependent component of the exchange current occur (∼35 pA in each cell) under all three Ca2+ buffering conditions (Fig. 6, A–C, insets), indicating that Ca2+ elevations can persist under these buffering conditions. The Ca2+ current-dependent component of the Na/Ca exchange current (visible as tail currents after the voltage step) is more sensitive to Ca2+ buffering conditions with the current nearly eliminated in 10 mM BAPTA. With 1.1 mM EGTA, the effect of FCCP was not significantly different from perforated-patch recordings (perforated 45 ± 4% reduction, n = 6; 1.1 mM EGTA 36 ± 6% reduction, n = 6, P = 0.23, Fig. 6,A, D, and E), suggesting that this level of artificial Ca2+ buffering approximates that found in intact amacrine cells. However, with 14 mM EGTA, the effects of disrupting MCU were significantly suppressed [13 ± 6% reduction (vs. 36 ± 6%), n = 6, P = 0.02] and delayed (no Ca2+ current reduction observed until a minute of FCCP exposure) when compared with 1.1 mM EGTA (Fig. 6, B, D, and E). Hence more EGTA minimized but did not eliminate the effect of MCU on the Ca2+ current amplitude, consistent with the idea that stronger buffering reduces CDI. Overall, blocking MCU in the presence of 10 mM BAPTA produced a 12 ± 5% (n = 22) enhancement in the Ca2+ current that was significantly different from the results in 1.1 mM EGTA (P = 0.0002, Fig. 6E). The sign of the effect varied from cell to cell with current enhancement in 12/22 cells (31 ± 6% enhancement) or small reductions in the current in 10 /22 cells (0.7 ± 2%, Fig. 6, C, D, and E). These data indicated that strong Ca2+ buffering reduced the impact of MCU on channel inactivation. We hypothesize that in BAPTA, CDI is reduced. In some cells, this reveals another Ca2+-dependent process that enhances current amplitude.
Inhibition of PKA and AC decreases the Ca2+ current amplitude
It is well known that PKA-dependent phosphorylation can enhance L-type Ca2+ currents by increasing the open time of the channels (Bean et al. 1984; Yue et al. 1990b). In amacrine cells, we have previously demonstrated that metabotropic glutamate receptor 5 activation leads to a PKA-dependent enhancement of the L-type current that does not result from changes in voltage sensitivity (Sosa and Gleason 2004). To test for the involvement of PKA in the FCCP-dependent enhancement of the Ca2+ current in BAPTA, we examined the effects of the PKA inhibitor, H89 (1 μM) with BAPTA (10 mM) internal. H89 has been shown to be specific for PKA when used at concentrations <10 μM (Chijiwa et al. 1990). For these experiments, different cells were used for the different treatments to avoid complications due to previous drug exposures (see methods). On average, BAPTA-loaded cells tested with FCCP showed enhancement of the Ca2+ current (12 ± 5%, n = 22, Fig. 6E). Inhibition of PKA with H89 reduced the Ca2+ current amplitude in all cells tested indicating basal PKA activity (41 ± 5%, n = 4, Fig. 7, A and C). When used in combination (H89 + FCCP), the effect on the Ca2+ current was generally larger but the difference was not statistically significant (58 ± 8% reduction, n = 6, P = 0.17, Fig. 7, B and C) than with either reagent alone indicating that with reduced PKA activity, inhibition of MCU still contributes to inactivation. Our interpretation of the target of H89 (PKA) is consistent with our previous observation that 8-bromo cAMP enhances the Ca2+ current in these cells (Sosa and Gleason 2004).
PKA is activated by cAMP which is generated by the enzymatic activity of AC. To confirm the involvement of this classical pathway, we asked whether inhibition of AC would have the same effects on the Ca2+ current as inhibition of PKA. The general AC inhibitor SQ 22,536 (200 μM) was used to inhibit AC (Fabbri et al. 1991). In all cells tested (84 ± 4% reduction, n = 7), SQ 22,536 exposure reduced the amplitude of the Ca2+ current (Fig. 7, D and E), consistent with basal AC activity producing cAMP and driving basal PKA activity.
Expression of AC1
Because the PKA activity appeared to be independent of cell surface receptor activation in these experiments, we postulated that the enzyme is stimulated by cAMP that has been generated through the activity of the Ca2+/CaM-dependent adenylate cyclase AC1. The Kd of AC1 for Ca2+ 100 nM (Fagan et al. 1996; Wu et al. 1993) is near resting cytosolic Ca2+ levels in these cells (50–100 nM, Hurtado et al. 2002) making this enzyme a good candidate for mediating both basal PKA activity as well as enhanced activity due to Ca2+ influx via L-type V-gated Ca2+ channels. We have previously demonstrated AC1-like immunoreactivity in cultured amacrine cells (Sosa and Gleason 2004). Here we further examine the expression of AC1 using a different, and more fully characterized, polyclonal antibody raised against human AC1. The specificity of this antibody was confirmed in Western blots using homogenates of both chicken and mouse brain (Fig. 8A). Single bands near the predicted molecular weight of AC1 (130 kDa) were detected for both chicken and mouse brain homogenate, indicating that the antibody recognizes the avian form of AC1. On sections of chicken retina, the anti-AC1 antibody labeled photoreceptors most strongly (Fig. 8C), but labeling was also strong in cell bodies in the ganglion cell layer. Cells at the inner border of the inner nuclear layer (most likely amacrine cells) were also labeled in a distinctly punctate pattern. Processes could be observed extending from these cells down into the inner plexiform layer where amacrine cell synapses form (Fig. 8D). In culture, cone photoreceptors were usually the most strongly labeled cells. The intensity of anti-AC1 labeling was variable among amacrine cells but all amacrine cells showed some level of AC1 expression (Fig. 8F). Punctate AC1 expression was detected both in cell bodies and processes of amacrine cells in culture (Fig. 8G). It is important to note that the AC inhibitor SQ 22,536 (Fig. 7) has been demonstrated to be an effective inhibitor of AC1 in neurons at the concentration used in our experiments (Liauw et al. 2005). These results support the hypothesis that AC1 can be involved in L-type Ca2+ channel regulation in retinal amacrine cells.
Inhibition of CaM primarily affects inactivation
From the results presented thus far, a scenario emerges where at least two Ca2+/CaM-dependent mechanisms might collaborate to regulate the function of L-type Ca2+ channels in retinal amacrine cells: first, the Ca2+/CaM-dependent inactivation that is sensitive to Ca2+ from 5 to 100 μM (Tadross et al. 2008) and second, the Ca2+/CaM-dependent activity of AC1 that is more Ca2+-sensitive (Kd∼100 nM). If these suggestions are valid, then we would predict that inhibition of CaM activity would be least effective in blocking the effects of AC1 activity. To test this, we looked at the effects of the calmodulin inhibitor calmidazolium (CMZ, 10 μM) (Weiss et al. 1982) on the FCCP-dependent alterations in current amplitude. Ruptured-patch recordings were made with either EGTA (1.1 mM) and CMZ in the pipette (Fig. 9A) or BAPTA (10 mM) and CMZ in the pipette (Fig. 9B) and then tested the effects of blocking MCU with FCCP. With inhibition of CaM, suppression of MCU produced an enhancement of the Ca2+ current under either buffering condition (EGTA + CMZ 36 ± 12% enhancement, n = 6, P = 0.0003; BAPTA + CMZ 19 ± 9% enhancement, n = 6, P = 0.5; Fig. 9, C and D). Recall that in the absence of CMZ, inhibition of MCU caused a decrease in current amplitude (presumably due to increased CDI) with 1.1 mM EGTA internally (EGTA 36 ± 6% reduction, n = 6, Fig. 6A). The switch in the sign of the response to FCCP when calmodulin is partially inhibited is consistent with a shift in the balance toward AC1 activation. The enhancement tended to be larger in EGTA than BAPTA however this difference was not statistically significant.
Phosphodiesterases play a role in regulating Ca2+ current amplitude
If the enhancement is due to the generation of cAMP via AC1 and subsequent activation of PKA, then phosphodiesterase (PDE) activity could affect the current amplitude. To test the involvement of PDEs, we employed the general PDE inhibitor IBMX (100 μM) (Beavo et al. 1970). Ca2+ current recordings were made in the perforated-patch configuration. At the outset, we reasoned that if we boosted cAMP levels by inhibiting its degradative enzyme, then the current amplitude should be enhanced. Our results, however, did not conform to expectations in that the effect of IBMX on the current amplitude varied from cell to cell. Twenty six percent (Fig. 10, A and B) of amacrine cells tested (n = 23) responded with the expected increase in current amplitude (75 ± 29% enhancement P = 0.04). However, IBMX produced a decrease in the current amplitude in 26% of cells (Fig. 10, C and D, 13 ± 2% reduction, P = 0.00002) and no change (less than ±5%) in 48% of cells (E and F). Some of the variability in responses could be due to the diversity of PDEs that can be inhibited by IBMX. To address this possibility we used 8-M-IBMX (100 μM), an inhibitor that is specific for PDE1 a Ca2+/CaM-dependent phosphodiesterase (Fig. 10G) (Wells and Miller 1988). All cells (n = 12) tested responded to 8-M-IBMX with small, consistent decrease in the current amplitude (4 ± 1% reduction, P = 0.004, Fig. 10H). These results imply that relatively high levels of cAMP generated in an environment with reduced PDE activity can have an inhibitory effect on the L-type channels (Ishikawa et al. 1993).
Ca2+- induced Ca2+ release functions to enhance the L-type calcium channel current
Thus far our data suggest that influx of Ca2+ can regulate L-type channels in at least two ways: by activating Ca2+/CaM-dependent inactivation and by activating AC1 and ultimately PKA. Ca2+ released from internal stores might also have regulatory effects on L-type channels. It is established that in amacrine cells, activation of L-type Ca2+ channels leads to CICR (Mitra and Slaughter 2002; Warrier et al. 2005). To examine the role of CICR, we looked at the effects of inhibiting the RyRs using a blocking concentration of ryanodine (14 μM) (Meissner 1986) and the RyR inhibitor dantrolene (20 μM) (Nelson et al. 1996). If CICR normally contributes to channel inactivation, then blocking CICR should enhance the current. If CICR normally contributes to activation of AC1 and ultimately stimulation of PKA, then inhibition of RyRs would cause a decrease in current amplitude. We found that inhibition of the RyRs with either ryanodine (Fig. 11A) or dantrolene (B) consistently produced suppression of the Ca2+ current (ryanodine 25 ± 4%, n = 3, P = 0.002, Fig. 11C; dantrolene 11 ± 3%, n = 7, P = 0.0002, D). These results were consistent with the hypothesis that CICR normally functions to enhance the L-type Ca2+ current, possibly by increasing the activation of AC1. Interestingly, 8-M-IBMX tended to block the effects of these RYRs inhibitors although this effect was not statistically significant for dantrolene (ryanodine +8-M-IBMX P = 0.002; dantrolene +8-M-IBMX P = 0.2). This observation implied that in the absence of PDE activity, increased cAMP levels can compensate for the lower level of AC1 activation during CICR suppression.
We find that in retinal amacrine cells, L-type Ca2+ channels are regulated by multiple Ca2+-dependent processes (Fig. 12). Under control conditions, Ca2+ currents tend to increase in current amplitude over time under our recording conditions. This increase in amplitude may represent the balance of PKA-dependent enhancement via Ca2+/CaM-dependent AC1 activity and CDI, which is largely mediated by Ca2+ entering through the Ca2+ channels. Interestingly, we observe an enhanced rate of Ba2+ current amplitude increase before FCCP application for cells recorded in external Ba2+ (Fig. 5C, 1st 3 data points). This enhanced rate of increase might represent the smaller contribution from CDI relative to basal AC1 activity. Disruption of MCU reduces the Ca2+ current amplitude in a Ca2+-dependent manner, suggesting that under normal conditions, mitochondria function to limit CDI and thus maximize the availability of L-type Ca2+ channels for signaling. We provide evidence that the reduction of the inhibitory effects of blocking MCU in BAPTA reveals the enhancing effect of PKA activity that is most likely due to the Ca2+/CaM-dependent activation of AC1. Inhibition of RyRs reduces the Ca2+ current amplitude suggesting that internal Ca2+ stores normally contribute to Ca2+ current enhancement. We propose that the link between the PKA-dependent enhancement of the Ca2+ current and the enhancing effects of CICR might also be the activity of AC1. Together, these studies indicate that L-type Ca2+ channels in amacrine cells are regulated by Ca2+ via complex and interacting mechanisms.
Variability in cultured GABAergic amacrine cells
Throughout this work there is variability observable in our data. One example appears in Fig. 1, B and D. Under control conditions the current amplitude increases over time, but as indicated by the large error bars, different levels of this effect in different cells becomes apparent over time. The origin of this variability is not known, but if the enhancement is due to AC1 activity (as we have suggested), it is perhaps relevant that our AC1 antibody labeling intensity varies among amacrine cells in culture implying differing AC1 expression levels among these cells (Fig. 8F). Another example of variability would be in the time course of inactivation during the voltage steps. We have not yet investigated the source of this variability except to confirm that cells that are different in this regard to not have distinctive response properties in our experiments. These examples of variability (and others) raise the question of whether some of the variability is the representation of the multiple amacrine cell types known to exist in the vertebrate retina. It is important to consider, however, that our culture conditions have apparently narrowed the range of possible amacrine cells fates. In the retina, different classes of amacrine cells can release GABA, glycine, acetylcholine, or dopamine at their synapses. In our cultures, however, we have only observed GABAergic synaptic transmission from these cells. We have viewed this as a benefit of the system that allows us to study a specific subset of amacrine cells that represent a substantial fraction of amacrine cells in the retina. Amacrine cells in the vertebrate retina can be further categorized by their morphology including their lamination pattern in the inner plexiform layer (MacNeil and Masland 1998). In culture, however, this information is lost. There appear to be a few morphological groups of amacrine cells identifiable in the two dimensions of the culture dish, but we have not made a systematic attempt to relate these morphological groups to those found in the intact retina. We considered the possibility that different levels of AC1 expression might correspond to different morphological types of amacrine cell in culture. On examination, this seemed unlikely because it was clear that cells with similar morphologies had different levels of AC1 antibody labeling.
L-type Ca2+ channels expressed by amacrine cells
The molecular identity of the L-type Ca2+ channels expressed by amacrine cells is not fully determined. However, it has been established that CaV1.3 mRNA is expressed by AII amacrine cells in the mouse retina (Habermann et al. 2003). In the chicken retina, an immunohistochemistry study demonstrated amacrine cell expression of CaV1.3 and possibly CaV1.4 but not CaV1.2 (Firth et al. 2001). In cultures of chick retinal neurons enriched for photoreceptors, other neuronal cells (presumably including amacrine cells) are reported to also express CaV1.3 (Ko et al. 2007). Interestingly the characteristics of CDI differ between CaV1.3 and CaV1.4 in that CaV1.3 contains sequence bestowing local as well as global Ca2+ sensing, whereas CaV1.4 should only sense global Ca2+ levels (Dick et al. 2008). This combination would be consistent with our data showing that 10 mM BAPTA removes a variable fraction of CDI but not all of it.
Effects of FCCP in amacrine cells
Our interpretation of the results in FCCP are based on the assumption that the primary effect of FCCP in amacrine cells is to reduce the proton gradient across the inner mitochondrial membrane and thus inhibit Ca2+ uptake via the uniporter. Several pieces of evidence from our previous work indicate that this is a fair assumption. It is well known that in the absence of a proton gradient, the ATP synthase can consume ATP. We have previously established, however, that at the same FCCP concentration (1 μM) and time frame of exposure used here (1 min), the interruption of ATP synthesis does not alter the amplitude or time course of depolarization-dependent Ca2+ elevations (Medler and Gleason 2002). This is consistent with reports from other neuronal cell types indicating that the glycolytic pathway can maintain ATP levels for ten's of minutes (Kauppinen and Nicholls 1986; Peng 1998; Werth and Thayer 1994; White and Reynolds 1995). Another potential complication would be if FCCP (as a protonophore) altered cytosolic pH. However, using SNARF-1 pH imaging we have established that under similar conditions, cytosolic pH is unperturbed by FCCP in amacrine cells (Sen et al. 2007). Furthermore, we have demonstrated that the effects of FCCP on cytosolic Ca2+ in amacrine cells are localized to within ∼10 μm of a mitochondrion indicating that the effects of FCCP are mitochondria-associated (Sen et al. 2007).
Mitochondrial Ca2+ uptake regulates Ca2+-dependent inactivation
We know that the inhibition of uniporter activity causes an increase in basal (un-stimulated) cytosolic Ca2+ levels. This Ca2+ elevation is independent of extracellular Ca2+ (Medler and Gleason 2002) but is dependent on internal Ca2+ stores (Sen et al. 2007). As such, we cannot rule out the possibility that some fraction of the effects we see in FCCP are due to the increase in basal Ca2+ levels. Nonetheless we propose that when the uniporter is inhibited, the primary effect is that normally sequestered Ca2+ now initiates additional CDI. What is our evidence that CDI is being enhanced when the uniporter is inhibited? The activity dependence of the effect of FCCP is consistent with a Ca2+-dependent inhibition of the current such as CDI in that both increasing the duration of the voltage steps and the frequency of the voltage steps intensified the effects of FCCP. Furthermore, replacement of extracellular Ca2+ with Ba2+ reduces the effects of FCCP and Ba2+ is known to substitute poorly for Ca2+ in calmodulin binding and activation (Dick et al. 2008).
Interestingly, the sign of the effect of FCCP on the Ca2+ current amplitude is often inverted when BAPTA is present internally. Because BAPTA is estimated to bind Ca2+ ∼400 times faster than EGTA (Adler et al. 1991), the differential effects of the two Ca2+ buffers can provide information about the spatial relationships between sources of Ca2+ and Ca2+ targets. The greater effectiveness of BAPTA over EGTA in reducing MCU-sensitive CDI suggests that mitochondria are in close proximity to the L-type Ca2+ channels, on the order of tens of nanometers (Burrone et al. 2002).
L-type Ca2+ channel CDI has also been shown to be regulated by MCU in chromaffin cells (Hernandez-Guijo et al. 2001). In these cells, however, 14 mM EGTA internally eliminated the inhibitory effect of disrupting MCU on CDI. In amacrine cells, FCCP-dependent effects on CDI were clearly observed in 14 mM EGTA suggesting a more intimate association between mitochondria and L-type Ca2+ channels in amacrine cells than in chromaffin cells. Ca2+-dependent inactivation has also been demonstrated for store-operated CRAC channels (Parekh 1998; Zweifach and Lewis 1995), and this inactivation is also minimized by MCU in T lymphocytes and basophilic leukemia cells (Gilabert and Parekh 2000; Hoth et al. 2000; for review, see Gilabert and Parekh 2000). These related observations in a neuron, a secretory cell, and cells of the immune system suggests that a widely expressed function of mitochondria is their ability to maintain the availability of Ca2+ influx pathways.
PKA-dependent regulation of L-type Ca2+ channels
PKA is known to phosphorylate L-type Ca2+ channels and to increase their open time (Bean et al. 1984; Yue et al. 1990b). PKA and the Ca2+/CaM-activated phosphatase calcineurin are known to be localized to CaV1.2 channels by the anchoring protein AKAP79/150 in neurons (Oliveria et al. 2007; for review, see Dai et al. 2009). Our evidence that calcineurin is not a major regulator of L-type Ca2+ current in amacrine cells is consistent with the report that amacrine cells do not express CaV1.2 (Firth et al. 2001). Comparatively little is known about the regulation of CaV1.3, but consistent with our observations, it has been demonstrated that PKA-dependent phosphorylation of CaV1.3 leads to current enhancement (Liang and Tavalin 2007; Qu et al. 2005). Current enhancement for CaV1.4 is apparently voltage dependent rather than phosphorylation dependent (Kourennyi and Barnes 2000).
AC1 in the retina
Intriguingly, expression of AC1 mRNA is enriched in the retina in comparison to brain and spinal cord (Xia et al. 1993). In the developing mouse retina, an in situ hybridization study demonstrated AC1 mRNA in photoreceptors and ganglion cells (Nicol et al. 2006). An immunocytochemistry study in the mouse retina found AC1 expression in the inner nuclear layer. In the chicken retina, we find AC1 protein strongly expressed in photoreceptors, in nearly all cells in the ganglion cell layer, and in amacrine cells. This labeling pattern is consistent with what we find in culture where AC1 expression was detected in amacrine cells and quite strongly in cone photoreceptors. AC1 expression was not determined for cultured ganglion cells because they do not persist under our culture conditions (Hyndman and Adler 1982).
Ca2+-dependent AC activation
It has been established that AC1 and AC8 can be activated by Ca2+ entering the cell across the plasma membrane. The strongest evidence for influx-dependent AC1/8 activation is for capacitative Ca2+ entry (CCE), but it is clear that entry via voltage-gated Ca2+ channels is effective as well (for review, see Willoughby and Cooper 2007). In rat cerebellar granule cells, depolarization-dependent Ca2+ influx was shown to promote cAMP accumulation (Cooper et al. 1998). An expression study with AC8 in a pituitary cell line has demonstrated that both CCE and voltage-gated Ca2+ channel activity were effective in stimulating the enzyme (Fagan et al. 2000). Interestingly, although release of Ca2+ from stores was substantial in this preparation, it was not effective in stimulating AC8 activity. In immortalized gonadotropin–releasing hormone neurons derived from the rat hypothalamus, it was also demonstrated that AC1 was activated by voltage-dependent Ca2+ influx but not by release of Ca2+ from stores (Krsmanovic et al. 2001). Here we provide evidence that in amacrine cells, both Ca2+ influx and release from stores enhances the L-type Ca2+ current amplitude, and we propose that the effect of store release could be due to further stimulation of AC1. It might be that the structure and organization of amacrine cells differ from the cells discussed in the preceding text such that release of Ca2+ from stores has an impact on AC1 activity. The spatial relationship between RyRs, L-type Ca2+ channels and AC1 is not known. The resistance of AC1 activity to 10 mM BAPTA could imply that AC1 is intimately associated with both channel types. The importance of localization is also made relevant in light of our previous work showing that the metabotropic glutamate receptor 5 (coupled to the IP3 pathway)-dependent enhancement of this same Ca2+ current can be suppressed by 5 mM BAPTA internally. This receptor-mediated enhancement of the current is also thought to involve AC1, but the source of the activating Ca2+ here is presumably IP3 receptors rather than RyRs (Sosa and Gleason 2004). An additional consideration is the high affinity of AC1 for Ca2+ that is about five times that of BAPTA [100 nM AC1 (Fagan et al. 1996; Wu et al. 1993) vs. 500 nM BAPTA (Adler et al. 1991].
It has been shown in amacrine cells that RyR-mediated CICR can activate Ca2+-sensitive K+ currents (Mitra and Slaughter 2002). A previous study on cultured amacrine cells demonstrated a role of CICR in enhancing synaptic GABA release, but IP3 receptors rather than RyRs were involved (Warrier et al. 2005). Although RyRs and IP3 receptors are expressed throughout the cell bodies and dendrites of cultured amacrine cells (Warrier et al. 2005; Sen et al. 2007), they might be differentially regulated in the two regions. If this is case, then nonsynaptic L-type Ca2+ channels might be the primary target of the regulatory mechanism involving CICR and RyR demonstrated here.
In the retina, bipolar cells can support sustained release of glutamate (for review, see Heidelberger et al. 2005), and amacrine cells are one of their postsynaptic targets. Although some amacrine cells are known to feedback onto bipolar cells, limiting the duration of excitation (Chavez et al. 2006; Dong and Hare 2003; Hartveit 1999; Singer and Diamond 2003), others may be subjected to prolonged depolarization and potentially significant elevations in cytosolic Ca2+. On the whole, the mechanisms described here (MCU, PKA activity, and CICR-dependent enhancement) tend to promote and preserve the activity of the L-type channels in amacrine cells. This may be important in maintaining signaling capabilities of amacrine cells at their cell bodies or at their synapses or, in some cases, both.
This work was supported by National Eye Institute Grant R01EY-012204 to E. Gleason.
There are no disclosures to be made.
We thank Drs. Jackie Stephens and Ursula White for advice, assistance, and expertise and Dr. Emily McMains for carefully reading of the manuscript. We also thank S. Prasad for technical assistance.
- Copyright © 2010 the American Physiological Society