Differential Modulation of N-Type Calcium Channels by μ-Opioid Receptors in Oxytocinergic Versus Vasopressinergic Neurohypophysial Terminals

Ethical approval

Rats were sacrificed as described below, using a method that causes minimum distress or discomfort, and is consistent with recommendations provided by the Panel on Euthanasia of the American Veterinary Medical Association. Rats were maintained and used in accordance with the Guide for the Care and Use of Laboratory Animals, created by the Institute of Animal Resources, National Research Council, DHHS, NIH and the protocol was approved by the Institutional Animal Care and Use Committee of the University of Massachusetts Medical School.

Isolation of nerve endings

Male Sprague–Dawley rats (Taconic Farms, Germantown, NY) weighing 200–250 g were sedated using CO₂ and immediately decapitated (Cazalis et al., 1987). Following removal of the anterior and intermediate lobes of the pituitary, the neural lobe was dissociated in a buffer at 37°C that contained (in mM): sucrose, 270; EGTA, 0.01; HEPES-Tris (N-2-hydroxyethylpiperazine-N′-2-ethanesulphonic acid-tris(hydroxymethyl)aminomethane), 10, buffered at pH 7.3; 295–300 mOsm/L. For electrophysiological, [Ca²⁺]_i measurements and immunolabeling experiments, the homogenate was placed directly into plastic Petri dishes and washed initially with a Low (3 μM free Ca²⁺) Locke's saline containing (in mM): NaCl 140; KCl 5; CaCl₂ 1.9; EGTA (ethyleneglycol-bis-(β-aminoethylether)-N,N,N′,N′-tetraacetic acid) 2; HEPES 10; Glucose 10; MgCl₂ 1.2; pH 7.3; 295–300 mOsmol/L, and slowly changed to a Normal (2.2mM CaCl₂; no EGTA) Locke's to remove any floating debris. For peptide release measurements, the dissociated neural lobe was centrifuged at 100g for 1 min, with the supernatant spun at 2,400g for 3 min. The resulting pellet, containing highly purified terminals, was resuspended in Normal Locke's saline and loaded onto filters (Acrodisc, LC13, 0.45 μm, Gelman Scientific, Ann Arbor, MI) where they were perfused for 1 h with Locke's buffer.

Peptide hormone release experiments

After 1 h of perfusion with Locke's saline, the terminals were perfused for another 10 min with 50mM K⁺ in the presence or absence of the drug being tested. The solutions containing elevated K⁺, to stimulate active release from the terminals, were prepared by replacing an equimolar amount of NaCl with 50mM KCl. High K⁺ solution contained in (mM): NaCl 90; KCl 55; MgCl₂ 1.2; CaCl₂ 2.2; glucose 10; HEPES-Tris 10; pH 7.2; at 37°C, and the osmolarity of the buffer was 295–300 mOsml/L. The solutions containing opioids were prepared just before they were needed. The eluate containing the released AVP and OT from the terminals was collected in 200 μl fractions and was assayed by radioimmunoassay using specific antibodies (kindly supplied by Dr. R.J. Bicknell, Babraham Institute, Cambridge, UK). AVP and OT antisera were raised in sheep (the AVP antiserum was raised by Dr. Bicknell; the OT antisera by Drs. Sheldric and Flint), and the final antibody dilution was 1:100,000 for AVP and 1:25,000 for OT. The cross-reactivity of OT antiserum with AVP was 0.015%, whereas that of the AVP antiserum with OT was 0.001%. The sensitivities of the assays were 0.5 pg (AVP) and 1pg (OT). The inter- and intra-assay coefficients of variation were 5–7% for AVP and 7–9% for OT.

Electrophysiological experiments

Using an inverted microscope the terminals were visually identified by their characteristic appearance, spherical shape, lack of nuclei, and relative small size (5–10μm in diameter). For the experiments in which Ca²⁺ currents were recorded, the terminals stayed in the 2.2mM Ca²⁺-Locke's solution described above. For the experiments where barium currents were recorded, a solution containing BaCl₂ was perfused into the recording chamber right before the beginning of the recording period. This bath solution contained (in mM): BaCl₂ 5; NaCl 145; KCl 5; HEPES 10; Glucose 15; MgCl₂ 1; pH 7.3. The terminals were patched in the perforated (Rae et al., 1991), whole-terminal configuration (Hamill et al., 1981) using freshly prepared amphotericin B (240μg/ml) dissolved in DMSO (dimethyl sulfoxide) and added to a pipette solution containing (in mM): Cs-glutamate 135; HEPES 10; Glucose 5; CaCl₂ 2; MgCl₂ 1; Tetraethylammonium (TEA) 20; pH 7.2. In some experiments Tetrodotoxin (TTX) ≥1μM was added to the bath. Pipettes were made of thin borosilicate glass (Drummond Scientific Co., Broomall, PA), using a vertical puller (David Kopf Instruments, Tujunga, CA) and fire-polished (Narashige, Tokyo, Japan) to display resistances in the range of 3–6MΩ. Series resistance was partially compensated using the patch clamp compensation circuit and was always below 15MΩ. Terminals were held at −80mV and depolarized to various test potentials every 30 sec using an Axopatch 200B patch interfaced via TL-1 DMA Interface (Axon Instruments, Foster City, CA) with a personal computer. Currents were low pass filtered at 2 kHz using an eight pole Bessel filter (Frequency Devises Inc., Haverhill, MA) and digitized at 10 kHz. pClamp software was used for data acquisition and analysis. Curve fitting was done with Originsoftware.

Current inactivation during voltage steps were obtained from a first order exponential fit (Chebyshev method) of the first 150 msec after peak currents were reached using the equation,

where A is the maximal amplitude of the component, t is the time t at which the current had decreased to 1/e of its initial amplitude, and C is the steady-state current level. Current–voltage curves were fitted using the equation

where G is the maximum conductance, E_rev is the apparent reversal potential, V is the test potential, V_0.5 is the potential at half activation, and k is the slope factor. Activation curves were fitted with a Boltzmann function of the type,

where k is the slope factor or the potential sensitivity for an e-fold change in G/G_max, V is the test potential, and V_0.5 is the voltage for half maximal values.

Identification of neuropeptide content on isolated individual terminals

At the end of each electrophysiological experiment, terminals were aspirated into the recording pipette. The pipette tip containing the terminal was broken against the bottom of a microcentrifuge tube and frozen until assayed for its neuropeptide content. Before assaying, 0.5% Triton-X 100 was added to each vial to maximize the amount of free peptide. The AVP or OT content of each terminal was determined by a specific enzyme-linked immunosorbant assay (ELISA; Assay Designs, Inc., Ann Arbor, MI). The sensitivity of the AVP and OT ELISA was previously determined (Custer et al., 2007) as 0.25 and 1 pg/well, respectively. When testing for only one peptide, the entire volume in each vial was assayed for AVP or OT. When both peptides were assayed from the same terminal, the content was divided so as to increase the probability to detect both peptides according to each assay's sensitivity.

Intracellular Ca^2R measurements

Dissociated nerve terminals were incubated with 2.5 μM Fura-2 AM in a low calcium Locke's saline for 45 min at 37°C. Terminals were thoroughly washed with normal Locke's saline and their resting global cytosolic [Ca²⁺]_i determined. This was performed according to the method of Grynkiewicz et al. (1985), with an assumed Ca²⁺-Fura K_d of 200 nM, as previously described by Becker et al. (1989). Resting values for global cytosolic [Ca²⁺] in the presence and absence of extracellular Ca²⁺ were 68.5±22.3 nM (n = 7) and 46.2±7.5 nM (n = 8), respectively, and these values were not significantly different (P = 0.4). High K⁺ solutions with and without toxins or DAMGO were delivered via a SF-77B microperfusion system (Warner Instruments, Hamden, CT) and gravity driven. Fluorescence images were viewed with a Nikon Diaphot TMD inverted microscope, using a quartz Plan-Neofluar 100×-oil immersion lens (Carl Zeiss Inc., Thornwood, NY) adapted with a 71000A Fura-2 filter cube (Chroma Technology, Corp., Rockingham, VT) and interfaced to a SenSys CCD camera (Photometrics Ltd., Tucson, AR). Terminals were excited using a Xenon arc lamp within a Lambda DG4 high-speed filter changer (Sutter Instruments Incorporated, Novato, CA) with the appropriate filters (340 and 380 nm wavelength) in place. Intraterminal emission of the Fura-2 indicator was gathered at 510 nm wavelength. Fluorescent images were acquired and processed with Axon Imaging Workbench 2.1 software (Molecular Devises, Foster City, CA).

Immunohistochemistry

Statistical analysis

Data are given as means±SEM. The effects of DAMGO on calcium currents, [Ca²⁺]_i, and peptide release were analyzed using a one-way ANOVA test followed by a Dunnetts test to determined significant differences between means (Sigma Stat; Jandel Scientific, San Rafael, CA). Statistical significance for all analyses was established as P<0.05.

Source of chemicals

All chemicals, including DAMGO, nicardipine and ω-conotoxin MVIIC were obtained from RBI/Sigma Chemical Co. (St. Louis, MO). The ω-conotoxins GVIA and MVIIA were obtained from Molecular Probes Inc. (Eugene, OR) and Alomone Labs Co. (Jerusalem, Israel). SNX 230 and SNX 482 were supplied by the Neurex Corporation (Menlo Park, CA). Nicardipine was first dissolved in DMSO at 1% of the final concentration and further diluted in Locke's buffer at 0.1–0.01% before used. DAMGO and all toxins stocks were prepared in distilled water, separated in aliquots and stored at −20°C. The final working dilutions were made in Locke's buffer and used immediately, with the exception of SNX 482 that was diluted 1 h before it was used, to allow for an optimal blocking conformation of the toxin.

N-type calcium currents are differentially modulated in OT versus AVP NH terminals

In order to test the effects of μ-opioid agonists on N-type channel activity, individual isolated terminals showing stable calcium currents were exposed to highly specific toxins able to block three out of the four possible current subtypes present in NH terminals (Fig. 1). The N-type current was isolated by inhibiting P/Q-type, L-type and R-type currents using 100–250 nM SNX230, 1–2 μM nicardipine, and 25 nM SNX482, respectively.

Figure 1A–C shows responses to DAMGO from a terminal that was positive for OT (OT+). The isolated N-type current comprised 38% of the total initial current. Peak currents were inhibited dose-dependently, with a 7% decrease by 100 nM DAMGO and a 29% by 300 nM DAMGO. The time-course of the current inhibited by DAMGO is displayed in Figure 1B. We used single exponential fits of the raw data to derive time constants of inactivation for the toxin's N-type isolated component as well as from the DAMGO-inhibited current component. Across all OT+ terminals, the opioid-inhibited current shows similar inactivation kinetics (38±2 msec) as the toxin-isolated N-type current (44±2 msec; P>0.05; n=5). Figure 1C shows the current–voltage relationship in control, isolated N-type, and 300 nM DAMGO conditions corresponding to the isolated OT+ terminal in Figure 1A,B. Inhibition by DAMGO was observed at all test voltages, with lower sensitivity to inhibition at negative potentials and without relief of inhibition at positive potentials. Steady-state activation curves obtained from all OT+ terminals showed a small shift of 5.4 mV in the voltage-dependence of activation toward more positive potentials upon exposure to DAMGO. V_0.5 values were estimated to be −7.8±0.9 mV, k=5.5±0.8 without DaMGO and −2.4±0.6 mV, k=7.6±1.8 in the presence of the opioid (Fig. 1D). Current inhibition by toxins and DAMGO was reversible, although required extensive washes (data not shown, but see, e.g., Fig. 3A). Mean inhibitory values for isolated N-type currents in all OT+ terminals were 6.8±7.3% at 10 nM, 14±6% at 100 nM and 42±5% at 300 nM DAMGO (Fig. 3C).

Vasopressinergic terminals exhibited a different N-type modulation profile by DAMGO. N-type channel responses to DAMGO from a positively identified AVP (AVP+) terminal are shown in Figure 2A. In this terminal, 42% of the total initial current remained after exposure to the same concentrations of the toxin mix described above. When the terminal was exposed to 100 nM DAMGO in the continuous presence of these toxins, the N-type current was increased by 43%. Although variable in magnitude, all AVP+ terminals exhibited current increments induced by DAMGO. Addition of 1.5 μM GVIA, a highly specific N-type toxin, completely blocked the potentiated, isolated current, corroborating its identity. The time-course and magnitude of the current increased by DAMGO is shown in Figure 2B. Time constants of inactivation obtained from the enhanced current component showed slower inactivation kinetics (65±6.4 msec) than the control N-type current (41±4.2 msec; P<0.05; n=7). The corresponding current–voltage relationship for this AVP+ terminal at all conditions tested is shown in Figure 2C. Current enhancements in 100 nM DAMGO were obtained at all test potentials, with higher increments at more positive potentials. The steady-state activation curve from the enhanced current shifted 6.6 mV per e-fold change toward more negative potentials when compared to control N-type currents. V_0.5 values were estimated to be −8.4±1.1 mV, k=5.3±0.7 without DAMGO and −14.4±0.6 mV, k=8.2±2.2 in the presence of the opioid (Fig. 2D). This shift in voltage was observed in all AVP+ terminals (n=7). As was observed with the inhibitory responses in OT+ terminals, current enhancements were reversible after washing with bath solution for 5–10 min (Fig. 3A). Mean values for N-type currents in all AVP+ terminals increased by 18±8% and 67±9% in the presence of 10 nM and 100 nM DAMGO, but decreased by 20 ±% at 300 nM DAMGO (Fig. 3C).

Approximately 5% of NH terminals express both types of neuropeptides (Custer et al., 2007; see also Kiyama and Emson, 1990; Mezey and Kiss, 1991). N-type currents from one of these colabeled terminals are shown in Figure 3B. Even though this kind of terminal expressed all four types of calcium conductances (not shown), 45% of the total initial current remained after exposure to the toxin mix isolating the N-type component. Exposure to low concentrations of DAMGO (15 nM) induced a robust 30% increase of the N-type current. However, increasing DAMGO's concentration to 115 nM caused a smaller 16% enhancement, thus following a similar response pattern as AVP+ terminals; that is, net current enhancements followed by inhibition at higher agonist concentrations. Similar responses were observed in two more colabeled terminals. Hence, current responses to μ-opioid agonist in colabeled terminals portray a dominant phenotype characteristic of terminals containing exclusively AVP+.

Modulation of intracellular calcium

Ca²⁺ influx measured with the calcium indicator Fura-2 was used as another independent method to test for differential behavior of Ca²⁺ influx through N-type channels in OT versus AVP terminals. If the N-type current is modulated differentially in each kind of peptidergic terminal, corresponding differences in [Ca²⁺]_i should also be detected. High K⁺ (50 mM) -induced changes in [Ca²⁺]_i were monitored in control conditions, in the presence of a toxin mix to isolate the N-type component, and in conjunction with increasing concentrations of DAMGO (10–500 nM). The toxin mixture was constant across experiments, with 25 nM SNX482, 100−150 nM SNX230 (or AgaIVA at equally effective concentrations), and 1−3 μM nicardipine to block R-, Q-, and L-type calcium channels, respectively. Three independent experiments were performed using this approach; two of them were followed with peptide immunolabeling to identify each terminal's content. The terminals were fixed in place after completing all [Ca²⁺]_i measurements and processed for immunolabeling as described in the methods section. Bright field images were obtained at the beginning of the experiment and cross-referenced with those taken during and at the end of the experiment to correct for any shift in the field at each experimental step. Figure 4 shows examples of one of those experiments. Out of nine individual nerve endings (Fig. 4A), three contained OT, four contained AVP, and two did not label for either peptide (assumed to be damaged during the isolation process). Only clearly labeled terminals were used in the analysis. High K⁺ induced changes in [Ca²⁺]_i for two strongly labeled terminals, #9 (OT) and #6 (AVP) are shown in Figure 4B,C. Consecutive images on the top part of Figure 4B, illustrate the response of terminal #6 to high K⁺ stimulation in the presence of the toxin mix and 10 nM DAMGO. The increased [Ca²⁺]_i is distinctively higher in this terminal than in terminal #9 (bottom part) following exposure to DAMGO. Quantification of the [Ca²⁺]_i responses (see Materials and Methods Section) for both types of terminals in the absence and presence of DAMGO at various concentrations provides a clear, distinctive pattern of enhancement versus inhibition of Ca²⁺ influx (Fig. 4C). Terminal #6 (AVP+), showed a 16% increase in the N-type dependent calcium influx when exposed to 10 nM DAMGO. This increment was decreased to only 3% above the toxin isolated level at 100 nM DAMGO and becomes inhibited by 14% with 500 nM DAMGO. To confirm the identity of the Ca²⁺ influx source, 800 nM MVIIA, a specific N-type channel blocker, was added to the dish. The final bar shows the remaining level of [Ca²⁺]_i after this application. Increments in [Ca²⁺]_i were absent in terminal #9 (OT+) and only inhibitory responses to DAMGO were obtained. A 4% decrease in [Ca²⁺]_i from the previous isolated level was obtained in 10 nM DAMGO followed by 22% and 33% decreases when exposed to 100 and 500 nM DAMGO, respectively. The final bar shows the remaining level of [Ca²⁺]_i after 800 nM MVIIA was added to the dish.

Data from all three independent experiments measuring [Ca²⁺]_i were pooled together in Figure 4D. In all, 23 Fura-2 loaded terminals were studied; 13 were positively identified as AVP or showed enhancements in [Ca²⁺]_i at low concentrations of DAMGO, 6 were OT positive or showed only inhibitory responses and 4 did not respond at all (and assumed to not be viable). The response patterns to DAMGO, although variable in magnitude, held true for both OT and AVP terminals. At low 10 nM DAMGO, AVP terminals showed a significant 21±3.7% increase in [Ca²⁺]_i compared to the previous isolated N-type dependent level. Enhancement was more pronounced (27 ± 6.4%) at 100 nM DAMGO but decreased beyond this point. At 300 nM DAMGO, [Ca²⁺]_i was 15 ± 4.6% higher than the toxin isolated level and at 500 nM DAMGO there were no longer significant differences (5.5 ± 6.1%) in [Ca²⁺]_i when compared to the initial toxin isolated level. However, close to half of AVP terminals showed moderate net inhibitions in [Ca²⁺]_i at high concentrations of DAMGO (see terminal #6 in Fig. 4C). In OT+ terminals, [Ca²⁺]_i levels were 1.5 ± 4.4% below the control N-type values at 10 nM DAMGO. At 100 and 300 nM DAMGO, intracellular Ca²⁺ values had diminished by 4.1 ± 3.9% and 7 ± 2.3%, respectively. At 500 nM DAMGO, [Ca²⁺]_i was signi ficantly inhibited (16 ± 2.1%) from control levels. Within all OT+ terminals, no significant enhancements in N-type dependent Ca²⁺ influx occur at either low or high concentrations of DAMGO.

Modulation of neuropeptide release

Since neuropeptide release is tightly controlled by Ca²⁺ influx, we expected a similar pattern of inhibition and enhancement of OT and AVP release by DAMGO. Figure 5A,B shows high K⁺-elicited, N-type dependent neuropeptide release in the absence and presence of DAMGO. Vasopressin release was significantly decreased by 51 ± 1.7% once terminals were exposed to the toxin mix able to block L-, P/Q-, and R-type calcium channels (Fig. 5A), thus, this amount of peptide release should solely depend on the activation of N-type calcium channels during high K⁺ depolarization. Exposure to 10 nM DAMGO in the continuous presence of the toxin mix induced a significant 27 ± 1% increase in AVP release. However, subsequent stimulation with a higher concentration (100 nM) of DAMGO induced a 45 ± 2.1% reduction in release. All effects by DAMGO and toxins were transient in nature as AVP release recovers to initial levels once all drugs were washed away. Figure 5B shows oxytocin levels, measured from the same samples used to quantify AVP. Oxytocin release was decreased by 65 ± 9% in the presence of the toxin mix. When this N-type-dependent component was tested with either low or high opioid agonist concentrations, all responses were inhibitory. Exposure to 10 and 100 nM DAMGO reduced OT release by 60 ± 7% and 64 ± 2%, respectively.

Thus, to summarize the work presented up to this point, the changes in N-type currents, [Ca²⁺]_i, and neuropeptide release displayed a consistent, distinct modulatory profile by DAMGO, characteristic of each terminal type.

N-type channel immunohistochemical distribution in NH terminals

We tried to determine whether mobilization and incorporation of N-type channels into the plasma membrane accounted for MOR current enhancement in AVP+ terminals. For these experiments, isolated terminals were equally distributed into two dishes; one exposed to a low concentration of DAMGO and one kept as a control (exposed only to Locke's saline) for a similar amount of time. Both dishes were immediately fixed and exposed to antibodies selective for the N-type α_1B (Ca_V2.2) subunit as well as antibodies against Neurophysin I (OT) and Neurophysin II (AVP) proteins. A total of 102 terminals were analyzed (see Materials and Methods Section and Fig. 6A) in four separate experiments. The distribution of α_1B immunoreactivity in OT+ and AVP+ terminals shows a distinctive, punctated pattern, both at the membrane as well as in the terminal's cytoplasm (see red labeling in Figs. 7 and 8). α_1B-dependent immunoreactivity is significantly higher (P < 0.05) in AVP+ terminals (2.76 × 10⁵ ± 2.38 × 10⁴ gfu/μm²) than in neighboring OT+ terminals (1.84 × 10⁵ ± 3.66 × 10⁴ gfu/μm²; Fig. 6B). However, when the immunoreactivity associated with the membrane was estimated (Fig. 6C; Control), no significant differences between terminal type was obtained. In AVP+ terminals, 29 ± 4% of the total α_1B fluorescence localized to the membrane area. Similar values (28 ± 5%) were obtained for OT+ terminals. Thus, N-type channels inside AVP+ terminals may mobilize to the membrane following exposure to DAMGO. To test this possibility, immunostaining of control terminals was compared to terminals exposed to 25 nM DAMGO prior to fixation. α_1B fluorescence associated with the membrane did not differ (P > 0.05) from control values (Fig. 6C). At this agonist concentration OT+ terminals exhibited 31 ± 1% of the total fluorescent signal to be associated to the membrane area versus 26 ± 2% in AVP+ terminals. Higher concentrations of DAMGO (100 nM) and longer incubation times—up to 2 h—did not significantly change α_1B fluorescence in the membrane area compared to control values. Thus, we could not detect any mobilization of N-type channels to the membrane in AVP+ terminals as a result of μ-opioid stimulation.

β₂ is present only in vasopressinergic NH terminals

As we described previously, the N-type current component increased by DAMGO showed slower inactivation kinetics than the current component lost during inhibition (Figs. 2B vs. 1B). Similar current enhancement such as the one induced by DAMGO in AVP+ terminals have been described in other systems. For example, activation of muscarinic receptors can lead to N-type current augmentation in sympathetic neurons and heterologous expression systems (Liu and Rittenhouse, 2003; Heneghan et al., 2009; Mitra-Ganguli et al., 2009). The crucial determinant for N-type current enhancement appears to be the presence of a palmitoylated β_2a subunit in the channel protein complex. However, expression levels of β_2a in brain tissue have been reported to be very low or undetectable (Scott et al., 1996; Ludwig et al., 1997; Vance et al., 1998). Nevertheless, using immunolabeling techniques, we tried to determine if β_2a was indeed present and, in particular, associated with the β_1B subunit in AVP terminals. Two examples of the expression pattern for both calcium channel subunits in each type of peptidergic terminal can be observed in Figure 7. Positive immunoreactivity against α_1B (red) and β_2a (green) was detected in 92% of AVP+ (blue) terminals (n = 42; Fig. 7A). The fluorescence pattern for β_2a was punctate and noticeable in the terminal's cytoplasm although, by far, the majority of the labeling was contained closer to the membrane area. Two dimensional profiles of the fluorescent signals taken at the white dotted line in each image shows that α_1B immunoreactivity seems to be closely associated –although not exclusively with β_2a immunoreactivity. However, colabeling analysis (Pearson's coefficient = 0.36 ± 0.09) could not provide clear evidence for a partial colocalization between both signals. Surprisingly, OT+ terminals did not express β_2a immunoreactivity (95%; n = 41; Fig. 7B). While a small number of OT+ terminals (5%) showed a slight amount of β_2a immunoreactivity, they were colabeled terminals and thus, dependent on the presence of AVP. Two distinct OT+ (blue) terminals and one not labeled (presumably AVP+) are shown in both images. All three terminals expressed α_1B, but only the OT-negative terminal showed β_2a immunoreactivity. These results could explain why only AVP+ terminals show N-type current enhancements when exposed to μ-opioid agonists.

To provide a more complete description of the subunit composition of calcium channels in NH terminals we determined which other β subunits, were present in OT+ and AVP+ terminals using immunofluorescence techniques (Fig. 8). Not surprisingly, both types of terminals show a substantial proportion of β₃ immunoreactivity: 86% of AVP+ terminals (n = 21) and 90% of OT+ terminals (n = 21) expressed β₃ immunoreactivity (third column in Fig. 8). Pearson's coefficients were estimated to be 0.62 ± 1.0 for AVP+ and 0.45 ± 0.13 for OT+ terminals, demonstrating a partial level of colocalization between both signals. The percentage of β₄ expression in AVP+ terminals was 91% (n = 23) and 73% OT+ terminals (n = 18; second column, in Fig. 8). OT+ terminals showed a low level of colocalization (PC = 0.41 ± 0.04), and in most instances, the β₄-dependent signal in AVP+ terminals was not associated with β_1B immunoreactivity (PC = 0.17 ± 0.02). Seventy-four percent of AVP+ terminals (n = 15) and only 23% (n = 17) of OT+ terminals showed β_1b immunoreactivity (first column, Fig. 8). Labeling of this subunit was very low and sparingly associated with α_1B immunoreactivity in AVP+ terminals (PC = 0.13 ± 0.06). In OT+ terminals the signal was too low and PC values were not determined. Thus, even though colocalization of α and β subunits was hard to asses in our system, in general we can still conclude that AVP+ terminals showed a higher variability by expressing all four β subunit types, while OT+ terminals expressed a smaller repertoire with only two main subunits. This difference could entail a higher functional variability in AVP+ terminals and a more limited one for those containing OT.