All of the animal procedures were approved by the Animal Welfare Committee of the University of Texas Health Science Center at Houston. Prior to use, adult male and female mice, strain C57BL/6J (Jackson Laboratories, Bar Harbor, ME, United States) were maintained on a standard 12 h light/dark cycle. Sacrifice was performed by cervical dislocation followed by decapitation. To minimize potential effects of circadian rhythm on syntaxin 3 (STX3) phosphorylation, animals for each set of experiments were sacrificed at the same time of day, typically between 12 and 1 pm.

To measure exocytosis from RBCs, retinal slices (≈200 μm thick) were prepared under dim red light illumination using a handmade, semiautomatic chopper. The external bath solution (bubbled with a mixture of 95% O₂ and 5%CO₂) contained (in mM): 125 NaCl; 1.2 CaCl₂; 0.5 MgCl₂; 1.5 KCl; 1.3 NaH₂PO₄; 10 D-Glucose; 23 NaHCO_3, with pH 7.3, 294–296 mOsm. Retinal slices were allowed to rest in the dark for ≥1 h under continuous perfusion prior to use. Slices were transferred under dim light to the recording chamber. The external bath solution used for recordings was identical to that above except that it additionally contained (in μM): 100 picrotoxin, 50 μM 1,2,5,6-Tetrahydropyridin-4-yl methylphosphinic acid (TPMPA), 10 strychnine to block GABA_A, GABA_C, and glycine receptors, respectively. To minimize artifacts in the capacitance record associated with perfusion, bath perfusion was halted during capacitance measurements.

Membrane capacitance was measured in whole-cell recording mode with an 8–12 MΩ borosilicate pipette placed on the soma of an intact RBC. The control internal pipette solution contained (in mM): 100 Cs-Methanesulfonate; 20 TEA-Cl; 0.2 MgCl₂; 10 phosphocreatine-Na₂; 0.5 Na-GTP; 1 mM EGTA, 20 HEPES, pH = 7.2, 286–290 mOsm. To test the effects of CaMKII inhibition on RBC exocytosis, the selective CaMKII inhibitor AIP was added to the internal recording solution to achieve a final concentration of 25 μM (Sakaba and Neher, 2001; Tatone et al., 2002; Gao et al., 2006; Mockett et al., 2011). Measurement of baseline parameters began 10–15 min after achieving the whole-cell recording configuration to allow time for reagents to reach stable levels within the synaptic terminal. To confirm the entry of Cs⁺ and TEA⁺ into the synaptic compartment, a linear voltage ramp (−65 mV to +20 mV, 24.2 mV/s) was applied approximately 1 min prior to the depolarizing pulse train stimulus and the current–voltage relationship evaluated for efficacy of K⁺ channel blockade and isolation of the presynaptic, L-type Ca²⁺ current (Heidelberger and Matthews, 1992; Heidelberger et al., 2005; Wan and Heidelberger, 2011). Exocytosis was driven by a train of ten depolarizing voltage steps (−65 to −20 mV) given at 4 Hz (Wan et al., 2010).

Membrane capacitance measurements were made with the use of an EPC-10 patch-clamp amplifier and Patchmaster software (v3.92 Heka Electronics, Lambrecht, Germany). In brief, a 500-Hz, 20-mV peak-to-peak sinusoidal voltage stimulus was superimposed on a holding potential of −65 mV (Zhou et al., 2006). The resultant signal was processed using the Lindau-Neher technique to yield estimates of membrane capacitance, C_m, and membrane conductances G_m and G_s. For each 100 ms sine wave segment, the mean values for C_m, G_m, and G_s were recorded. Measurement of C_m was temporally halted during a depolarizing voltage step and for the first 25 ms thereafter to minimize the confounding contributions of current flow and channel gating to C_m (Horrigan and Bookman, 1994; Wan et al., 2008, 2010). RBCs were included in the final data set if they had an access resistance of less than ≈75 MΩ and a well-isolated, measurable Ca²⁺ current. Control and AIP-exposed cells in the final data set were similar with respect to the mean resting access conductance, G_s, (control: 2.06e-8 ± 2.04e-9S; AIP: 1.86e-8 ± 1.71e-9S; p = 0.466, n = 6,5) and the mean resting membrane conductance, G_m, (control: 1.46e-10 ± 5.2e-11S; AIP: 5.0e-11 ± 1.6e-11S; p = 0.138; n = 6,5 cells). In addition, there was no statistical difference in the resting C_m (control: 6.48 ± 0.811 pF; AIP: 4.61 ± 0.385 pF; p = 0.083, n = 6,5) or in the standard deviation of the resting C_m signal (baseline noise) (control 11.4 ± 2.37 fF; AIP: 9.31 ± 1.74 fF; p = 0.494. n = 6,5). Data were exported to IGOR Pro (v6.3.7.2, WaveMetrics) for analysis of membrane current and C_m, G_m, and G_s. Two-tailed independent sample t-tests with or without Welch’s correction were used, as appropriate. Statistical analyses were performed in Excel (Microsoft, Redmond, WA, United States) and Prizm 7 (GraphPad Software, Inc., San Diego, CA, United States).

Staurosporine (ThermoFisher, Waltham, MA, United States) was made as a 2.15 mM stock solution in DMSO and used at a final concentration of 100 nM. KN93 (Selleckchem, Houston, TX, United States) was made up as a 3.35 mM stock solution in MQ water and used at a final concentration of 1 μM. KN92 (Biovision, Milpitas, CA, United States) was made up as a 3.6 mM stock in DMSO and used at a final concentration of 1 μM. Myristoylated-AIP (mAIP; Tocris, Bristal, United Kingdom), a membrane permeant analog of the CaMKII inhibitor AIP (Autocamtide-2-related inhibitory peptide), was made as a 293 μM stock solution in MQ water and used at a final bath concentration of 100 nM. In the isolated cell experiments, drugs were added to the external bath solution. For drugs made up in DMSO, addition of DMSO to the bath solution at the same concentration as in the test condition had no effect on pSTX3B or pCaMKII levels (not shown). For electrophysiological measurement of exocytosis, a 1 mM stock solution of AIP (Tocris, Bristal, United Kingdom) was made in MQ water and added to the internal pipette solution at a final concentration of 25 μM (Sakaba and Neher, 2001; Tatone et al., 2002; Gao et al., 2006; Mockett et al., 2011). All other chemicals were purchased from Sigma.

We have shown previously that in the mammalian retina, STX3 can be phosphorylated at T14 under physiological conditions (Liu et al., 2014). Furthermore, in vitro biochemical studies have shown that T14 of STX3 is phosphorylated by Ca²⁺/CaMKII (Risinger and Bennett, 1999; Liu et al., 2014). This raises the possibility that phosphorylation at this site in vivo may be regulated by synaptic activity. In the first part of the present study, we ask whether phosphorylation of STX3 at T14 in the mammalian retina is regulated in vivo by light.

Levels of STX3 phosphorylated at T14 (pSTX3) and total STX3 levels in the synaptic terminals of rod and cone photoreceptors and RBCs were measured using quantitative immunofluorescence. To this end, retinal sections were double-labeled with a phospho-specific antibody that recognizes pSTX3, which we previously developed and extensively characterized (Liu et al., 2014) and a STX3 antibody (Zulliger et al., 2015) that we further characterized as part of this study. As shown in Supplementary Figure 1, the latter recognizes both STX3 and a STX3 T14 phosphomimetic mutant and is not occluded by the binding of an antibody raised against an epitope in the N-terminus of STX3 that encompasses T14. This pair of antibodies was used to provide a read-out of pSTX3 and STX3, respectively, which in retinal ribbon-style synapses, is indicative of pSTX3B and STX3B (Curtis et al., 2008). The retinae from dark-adapted mice and dark-adapted mice that were subsequently exposed to 15 min of a light stimulus were compared. Data were quantified by determining the ratio of pSTX3 to total STX3 of individual synaptic terminals of rod and cone photoreceptors and rod bipolar cells for each condition. For each animal, results were compiled across several retinal sections for each type of synaptic terminal and a single composite data point per type of synaptic ending point per animal reported (see section “Materials and Methods”).

Figure 1A depicts a set of representative confocal images of the OPL of vertical retinal sections from a dark-adapted mouse (top) and from a dark-adapted mouse exposed to 15 min of a light stimulus (bottom) that are triple-labeled for cone arrestin (blue), STX3 (red), and pSTX3 (green). A strong pSTX3 signal is evident in the section from the dark-adapted mouse (Figure 1A, top) relative to the light-exposed animal (Figure 1A, bottom). Quantitation of the results in rod terminals from multiple histological sections and animals demonstrated that total STX3 levels, expressed as mean pixel intensity, were virtually identical in the dark-adapted and light-exposed mice (Figure 1B: dark: 87 ± 5 vs. light: 79 ± 5; p = 0.2673, t = 1.182, and n = 6,5). By contrast, pSTX3 levels were approximately two times higher in the dark-adapted mice than in light-exposed mice (Figure 1B; dark: 46 ± 4 vs. light: 23 ± 2; p = 0.0006, t = 5.207). These data suggest a selective, light-driven decrease in the phosphorylation of STX3 at T14. To quantify the change in STX3 phosphorylation, we calculated the pSTX3/STX3 ratio for each condition. Results show that the pSTX3 to STX3 ratio in rod terminals was nearly two times higher in the dark-adapted mouse, when rods are depolarized and releasing neurotransmitter, as compared to the light-exposed mouse, when rods are expected to be quiescent (Figure 1C; dark: 0.497 ± 0.020 vs. light: 0.276 ± 0.016; p < 0.0001, t = 8.372, and n = 6,5). Cone terminals, on the other hand, showed little evidence of a light-regulated change in the pSTX3 to STX3 ratio (Figures 1D,E). This may reflect the broader operational range of cone photoreceptors and methodological limitations (see section “Discussion”).

Figure 2 shows a representative set of confocal images of vertical retinal sections taken of the IPL double-labeled for STX3 (red) and pSTX3 (green). In the dark-adapted state, a modest level of pSTX3B labeling can be observed at the bottom of the proximal inner plexiform layer, where the synaptic terminals of RBCs reside (Figure 2A, top). The level of pSTX3 labeling in this region is enhanced following light exposure (Figure 2A, bottom). Quantification of multiple experiments revealed that pSTX3B to STX3B ratio is significantly increased in RBC synaptic terminals of light-exposed mice relative to those of dark-adapted mice (Figure 2C; light: 0.672 ± 0.028 vs. dark: 0.385 ± 0.013; p < 0.0001, t = 9.258; n = 4,4 mice). Furthermore, the light-dependent increase in pSTX3 labeling does not arise from a light-regulated change in total synaptic STX3 (Figure 2B; dark: 64 ± 4 vs. light: 69 ± 13; p = 0.6747, t = 0.4409) but rather a specific light-dependent increase in pSTX3B (Figure 2B: dark: 26 ± 2 vs. light: 46 ± 7; p = 0.0415, t = 2.584) Given that RBCs are depolarized by light, these findings suggest that pSTX3 levels are higher in depolarized terminals relative to quiescent ones. Taken together, results obtained rod photoreceptors and rod bipolar cells indicate a correlation between periods of neuronal activity and the phosphorylation of STX3 at T14.

We next asked whether external Ca²⁺, which enters nerve terminals during synaptic activity, was required for phosphorylation of STX3 at T14. To this end, eyecups were isolated from dark-adapted mice and bathed in our standard, Ca²⁺-containing bath solution or in a nominally Ca²⁺-free bath solution. Isolated eyecups were then subjected to either an additional 15 min of darkness or a light stimulus for 15 min.

As shown in Figure 3, in rod terminals under normal Ca²⁺ conditions, we saw a significantly higher pSTX3 to STX3 ratio in the dark than in the light (Figure 3B; dark: 0.563 ± 0.040 vs. light: 0.196 ± 0.035; n = 4,4; p < 0.0001), recapitulating the results observed in the live mouse experiments. This dark-driven phosphorylation of STX3B at T14 in rod terminals (Figure 3A, top) was prevented when the eyecup was bathed Ca²⁺-free external solution (Figure 3A, bottom). Quantification of multiple experiments revealed that the pSTX3/STX3 ratio of rod terminals in dark-adapted eyecups was approximately 2.5 times larger in the presence of external Ca²⁺ than in its absence (Figure 3B; 0.563 ± 0.040 vs. 0.140 ± 0.023; n = 4,4; p < 0.0001). In the light-exposed eyecup, removal of external Ca²⁺ did not further reduce the pSTX3 to STX3 ratio (Figure 3B 0.140 ± 0.023 vs. 0.092 ± 0.023, n = 4,4, p = 0.1374).

In cone terminals, the ability of the light stimulus to reduce the pSTX3 to STX3 ratio attained in the dark-adapted state under normal Ca²⁺ conditions approached significance (Figure 3C; dark: 0.796 ± 0.074 vs. light: 0.445 ± 0.084; n = 4,4; p = 0.0551), possibly reflecting the greater ease with which isolated eyecups could be oriented to face the light stimulus relative to live mice. Importantly, removal of external Ca²⁺ significantly reduced the dark-adapted levels of STX3 phosphorylation at T14 (Figure 3C; 0.796 ± 0.074 vs. 0.222 ± 0.055; n = 4,4; p = 0.0023), indicating that cone terminals also exhibit Ca²⁺-dependent STX3 phosphorylation at T14. There was no further decrease in the pSTX3 to STX3 ratio in photoreceptor terminals in light-exposed eyecups in the absence of external Ca²⁺ (Figures 3B,C; rods: 0.140 ± 0.023 vs. 0.092 ± 0.023, n = 4,4, p = 0.1374; cones: 0.222 ± 0.055 vs. 0.243 ± 0.116, n = 4,4 p = 0.3802).

Rod bipolar cells are depolarized by light. Therefore, if phosphorylation of STX3B at T14 requires activity-dependent Ca²⁺ entry, then in RBC terminals, phosphorylation levels should be highest in light-exposed eyecups when external Ca²⁺ is present and remain at or near baseline levels when in the dark or when bathed in a Ca²⁺-free solution in the light. The results depicted in Figure 4 supports this prediction. In the presence of external Ca²⁺, RBC terminals in light-stimulated eyecups exhibited an approximately two-fold higher pSTX3B to STX3B ratio relative to those of dark-adapted eyecups (Figure 4B; 0.673 ± 0.027 vs. 0.295 ± 0.150, p = 0.0014, n = 2,2). That external Ca²⁺ was required for this light-driven doubling of the pSTX3 to STX3 ratio was evidenced by the virtual abolition of the light-driven increase in Ca²⁺-free external solution (0.673 ± 0.027 vs. 0.318 ± 0.032, p < 0.0019, n = 2,2) and maintenance of the pSTX3 to STX3 ratio at a level indistinguishable from that observed in the dark-adapted eyecup in the absence of Ca²⁺ (0.318 ± 0.032 vs. 0.238 ± 0.223, n = 2,2 p = 0.238). Taken together, the results of this set of experiments demonstrate that the light-regulated phosphorylation of STX3 on T14 in synaptic terminals of retinal photoreceptor and rod bipolar cells is dependent upon external Ca²⁺, which presumably enters the synaptic terminals via voltage-gated Ca²⁺ channels during synaptic activity (Heidelberger et al., 2005; Wan and Heidelberger, 2011).

To probe the signaling pathway downstream of Ca²⁺ entry, we turned to the acutely isolated rod bipolar cell preparation (Zhou et al., 2006; Wan et al., 2008, 2010). This preparation allows us to manipulate intraterminal Ca²⁺ and examine Ca²⁺ signaling pathways in the absence of local circuit interactions (Wan et al., 2010, 2012). To elevate the level of free Ca²⁺ in the rod bipolar cell synaptic terminal, we used a low Na⁺ external solution to reverse the Na⁺/Ca²⁺ exchanger. This manipulation produces a modest, sustained elevation of the spatially-averaged intraterminal Ca²⁺ concentration (Wan et al., 2012). As shown in the representative PKC-positive terminal cluster of synaptic boutons of an individual RBC shown in Figure 5A (upper panel, “high Ca”), elevation of intraterminal Ca²⁺ via the reverse mode of the Na⁺/Ca²⁺ exchanger method produced a robust pSTX3 labeling that contrasted strongly with the minimal labeling observed in nominally Ca²⁺-free external solution (Figure 5A, lower panel, “low Ca”). Compiled data from seven experiments, with each experiment representing the average result obtained from numerous RBC terminals, showed that there was a significant increase in the pSTX3 immunolabeling in response to elevated intraterminal Ca²⁺ with no change in total STX3 levels (Figure 5B), suggestive of a Ca²⁺-dependent rise in phosphorylation of STX3 at T14. The mean pSTX3 to STX3 ratio in RBC synaptic terminals under conditions of elevated intraterminal Ca²⁺ was more than two-fold higher than those of cells bathed in a nominally Ca²⁺-free solution (Figure 5C; 0.944 ± 0.110 vs. 0.260 ± 0.052, p < 0.0001, n = 7,7), confirming this interpretation.

Given that STX3 is phosphorylated at T14 by CaMKII in vitro (Risinger and Bennett, 1999; Liu et al., 2014), we additionally asked whether CaMKII is activated via our stimulation paradigm. To this end, we performed parallel experiments in which immunolabeling for CaMKII at phosphorylated at T286/287 (pCaMKII) was used a read-out of constitutively-activated CaMKII (Colbran et al., 1989; Hudmon and Schulman, 2002). As shown in Figure 5D, elevation of intraterminal Ca²⁺ (“High”) produced an increase in pCaMKII labeling in RBC synaptic terminals relative to RBC terminals bathed in nominally Ca²⁺-free solution (“Low”) (High: 78.73 ± 7; Low: 54.99 ± 5.68; t = 2.562, p = 0.0283, n = 6,6).

We next took advantage of our ability to manipulate Ca²⁺ and pSTX3 levels in the synaptic terminals of acutely isolated RBCs to test for a role for CaMKII. To validate this approach, we first ascertained whether inhibition of phosphorylation, in general, would prevent the Ca²⁺-mediated increase in pSTX3. To this end, acutely isolated RBCs were treated with the non-specific kinase inhibitor staurosporine (Figure 6). As shown in the images of representative RBC terminals (Figure 6A), treatment with staurosporine (100 nM) dramatically reduced both the Ca²⁺-triggered elevation of pSTX3 and the expected Ca²⁺- dependent rise in the pSTX3 to STX3 ratio. On average, the Ca²⁺-triggered increase in the pSTX3 to STX3 ratio was reduced by more than 75% with staurosporine treatment (Figure 6B; 0.167 ± 0.027 vs. 0.756 ± 0.046, p < 0.0001, n = 4,4). Furthermore, staurosporine lowered the residual pSTX3 to STX3 ratio of RBC synaptic terminals bathed in Ca²⁺-free solution (Figure 6B; 0.090 ± 0.004 vs. 0.278 ± 0.051, p = 0.0366, n = 4,3). Staurosporine treatment was also found to suppress the Ca²⁺-dependent increase in pCaMKII immunolabeling (Figure 6C).

To probe the ability of CaMKII to phosphorylate native STX3B at T14 in RBCs, we first turned to KN-93, a well-characterized, membrane-permeant CaMKII inhibitor that prevents the activation of CaMKII (Hudmon and Schulman, 2002). Treatment with KN-93 (1 μM, Supplementary Figure 2A) prevented the Ca²⁺-triggered elevation of the pSTX3 to STX3 ratio (high Ca: 0.944 ± 0.11; high Ca +KN-93: 0.375 ± 0.72; p = 0.0003, n = 7,7), holding the ratio at a level indistinguishable from that of unstimulated RBC terminals (0.260 ± 0.52, p = 0.996, n = 7,7). It also suppressed the Ca²⁺-induced increase in pCaMKII levels (Supplementary Figure 2B). However, treatment with 1 μM of KN-92 (Supplementary Figure 2A), an inactive analog of KN-93 that shares some but not all off-target effects with KN-93 (Pellicena and Schulman, 2014), also reduced the Ca²⁺-dependent rise in the pSTX3B to STX3B ratio and blocked the increase in CaMKII autophosphorylation (Supplementary Figure 2). These results raised the possibility that one or both compounds acted on targets other than or in addition to CaMKII, rendering it impossible to draw a conclusion about a possible in vivo role for CaMKII in STX3 phosphorylation using KN93.

Therefore, to probe the role of CaMKII in STX3 phosphorylation at T14, we turned to the potent and highly selective peptide inhibitor of CaMKII called autocamtide-2-related inhibitory peptide (AIP; Ishida et al., 1995). AIP, derived from the autoinhibitory domain of CaMKII, inhibits all CaMKII activity regardless of activation mode (Ishida and Fujisawa, 1995; Ishida et al., 1995; Erickson et al., 2011; Pellicena and Schulman, 2014) and does not directly affect voltage-gated ion channels (Gao et al., 2006). Figure 7A shows representative examples of acutely dissociated PKC-positive RBC terminals under low and high Ca²⁺ conditions and RBCs treated with mAIP, a membrane-permeant analog of AIP, under high Ca²⁺ conditions. Incubation in external solution containing mAIP (100 nM) blocked the Ca²⁺-evoked increase in pSTX3. Pooled data from multiple experiments show that the Ca²⁺-triggered elevation of the pSTX3 to STX3 ratio was reduced in the presence of mAIP (Figure 7B; High: 0.852 ± 0.058; High + AIP: 0.573 ± 0.076, p = 0.0171, n = 7,7) to a ratio that was not dramatically different from that observed when external Ca²⁺ was nominally zero (low: 0.343 ± 0.058, p = 0.0616, n = 7,6). Interestingly, mAIP did not suppress the Ca²⁺-evoked elevation of pCaMKII in these experiments (Figure 7C). This suggests that the inhibitory effect of mAIP on STX3 phosphorylation observed here is most likely attributable to the ability of AIP to block constitutively-activated CaMKII (Ishida and Fujisawa, 1995; Erickson et al., 2011; Murakoshi et al., 2017).

The above data are consistent with a model in which STX3 is phosphorylated at T14 in a Ca²⁺-dependent manner by CaMKII. To address the effect of this activity-dependent modification on synaptic function, we next examined the effect of CaMKII inhibition on RBC exocytosis. Membrane capacitance measurements were used to quantify exocytosis from an individual RBC in the retinal slice configuration held under voltage-clamp control. Exocytosis was triggered via a brief train of depolarizing voltage-pulses that depletes the ribbon-associated releasable vesicle pool (Wan et al., 2008, 2010; Wan and Heidelberger, 2011). To avoid modulating effects of AIP on targets outside of the rod bipolar cell under investigation, AIP [25 μM; (Sakaba and Neher, 2001; Tatone et al., 2002; Gao et al., 2006; Mockett et al., 2011)] was delivered directly to an individual rod bipolar cell via the whole-cell patch pipette. At this concentration, AIP is expected to selectively block both the autophosphorylation of CaMKII and activated CaMKII (Ishida et al., 1995). To allow time for both K⁺ channel blockers and AIP to reach the synaptic terminal from the somatically-positioned recording electrode, the stimulus was given approximately 15 min after achieving the whole-cell recording configuration. In addition, effectiveness of the delivery of blockers to the synaptic terminal was confirmed prior to stimulation by visual inspection of the current–voltage relationship and isolation of the presynaptic L-type Ca²⁺ current (Heidelberger and Matthews, 1992; Heidelberger et al., 2005; Wan and Heidelberger, 2011). Evaluation of the Ca²⁺ channel current–voltage relationship did not reveal a significant effect of AIP on the voltage-gated Ca²⁺ current either with respect to the mean activation voltage (Control: −42 ± 2 mV; AIP: −44 ± 1 mV; p = 0.4887, t = 0.7219; n = 6,5) or the mean Ca²⁺ current amplitude measured at −20 mV [Control: −23 ± 6 pA (median: −18 pA); AIP: −14 ± 2 pA (median: −16 pA); p = 0.2279, t = 1.343; and n = 6,5].

As shown in Figure 8, control RBCs exhibited a mean depolarization-evoked capacitance increase at the end of the pulse train of ≈50 fF, a value slightly larger than the amplitude of the total releasable vesicle pool reported for dissociated mouse rod bipolar cells (Zhou et al., 2006; Wan et al., 2008). In the presence of AIP, however, the depolarization-evoked rise in membrane capacitance was reduced by approximately an order of magnitude (Figure 8A; Control: 50 ± 11 fF; AIP: 3.4 ± 2.7 fF; p = 0.0059, t = 4.309, and n = 6,5). These results indicate a major role for CaMKII in the regulation of exocytosis in RBC synaptic terminals. Because changes in exocytosis could also result from changes in Ca²⁺ entry, we additionally evaluated the effects of AIP on the train-evoked Ca²⁺ current. To correlate the total train evoked exocytosis with total train-evoked Ca²⁺ entry, we calculated the mean total Ca²⁺ current for the pulse train by taking the sum of each Ca²⁺ current evoked by an individual depolarizing voltage step within a stimulus train. As shown in Figure 8B, comparison of the total Ca²⁺ current for the pulse train failed to reveal a significant difference between control and AIP-treated rod bipolar cells [Control: −268 ± 85 pA (median value: −206 pA); AIP: −166 ± 24 pA (median value: −180 pA); p = 0.2948, t = 1.151; and n = 6,5]. This stimulation protocol and the levels of total Ca²⁺ current produced are sufficient to trigger fusion of the rapidly-releasable and releasable vesicle pools (Wan et al., 2008, 2012; Heidelberger, unpublished observations). While we cannot rule out the possibility that there could be differences in the intraterminal Ca²⁺ levels that might influence synaptic vesicle dynamics, the data suggest that AIP acts at a site other than Ca²⁺ entry to reduce exocytosis. A leading possibility, consistent with our previously published model (Liu et al., 2014), is that AIP inhibits the activity-dependent phosphorylation of STX3 at T14 by CaMKII, thereby modulating the assembly of SNARE complexes required for membrane fusion and the release of neurotransmitter.