Synaptic Plasticity at Inhibitory Synapses in the Ventral Tegmental Area Depends upon Stimulation Site

Abstract Drug exposure induces cell and synaptic plasticity within the brain reward pathway that could be a catalyst for progression to addiction. Several cellular adaptations have been described in the ventral tegmental area (VTA), a central component of the reward pathway that is the major source of dopamine release. For example, administration of morphine induces long-term potentiation (LTP) of excitatory synapses on VTA dopamine cells and blocks LTP at inhibitory synapses. Drug-induced synaptic changes have a common endpoint of increasing dopamine cell firing and dopamine release. However, gaining a complete picture of synaptic plasticity in the VTA is hindered by its complex circuitry of efferents and afferents. Most studies of synaptic plasticity in the VTA activated a mixed population of afferents, potentially yielding an incomplete and perhaps misleading view of how drugs of abuse modify VTA synapses. Here, we use midbrain slices from mice and find that electrical stimulation in two different regions induces different forms of plasticity, including two new forms of LTP at inhibitory synapses. High-frequency stimulation (HFS) induces LTP independently of NMDA receptor (NMDAR) activation, and surprisingly, some inhibitory inputs to the VTA also undergo NMDAR-independent LTP after a low-frequency stimulation (LFS) pairing protocol.


Introduction
The ventral tegmental area (VTA) contains dopaminergic cells that receive inhibitory innervation from GABAer-gic cell bodies originating within the VTA and from many other brain regions (Watabe-Uchida et al., 2012;Beier et al., 2015). Despite a wealth of anatomical and behavioral studies investigating the diversity of VTA afferents, plasticity at inhibitory synapses was historically described without identification of the presynaptic partner (Melis et al., 2002;Liu et al., 2005;Nugent et al., 2007Nugent et al., , 2009Niehaus et al., 2010;Dacher and Nugent, 2011;Padgett et al., 2012;Graziane et al., 2013;Kodangattil et al., 2013;Polter et al., 2014). For example, nitric-oxide-dependent long-term potentiation (LTP GABA ) can be triggered using electrical stimulation within the VTA (Nugent et al., 2007); however, when specific afferents were isolated using optogenetics, induction of LTP GABA was found to depend on the presynaptic partner (Simmons et al., 2017;Polter et al., 2018). Specifically, LTP GABA is expressed at nucleus accumbens and VTA GABA A synapses, but not rostromedial tegmental nucleus (RMTg)-originating GABA A synapses. These observations demonstrate that all GABAergic synapses cannot be assumed to share a common plasticity mechanism. The idea that plasticity is segregated to specific populations is not a new one, and in fact many reports segregate experiments by postsynaptic cell identity. For example, long-term depression (LTD) induced by low-frequency afferent stimulation is only expressed in putative dopamine cells in the VTA that express large H currents (I h ; Dacher and Nugent, 2011). With local electrical stimulation in acute slices, it is possible to isolate synapses of one neurotransmitter type pharmacologically, but the identity of the presynaptic source is not always as easy to manipulate or determine. Although the location of the postsynaptic VTA cell (e.g., medial vs lateral VTA) can sometimes predict output site, inputs from over 20 brain regions contact dopamine cells in all VTA subregions (Beier et al., 2015). It is possible that other plasticity mechanisms have yet to be uncovered because their expression is limited to a subset of inputs, and therefore not apparent with global activation of all inputs. Here, we use different electrical stimulation sites, and report two ways of inducing LTP at inhibitory synapses in the VTA with a mechanism(s) that is non-overlapping with that of LTP GABA or other known forms of LTP at inhibitory synapses in the VTA.

Preparation of brain slices
Horizontal brain slices (220 m) were prepared from deeply anesthetized mice. Briefly, anesthetized mice were perfused with ice-cold oxygenated artificial CSF (ACSF): 126 mM NaCl, 21.4 mM NaHCO 3 , 2.5 mM KCl, 1.2 mM NaH 2 PO 4 , 2.4 mM CaCl 2 , 1.0 mM MgSO 4 , 11.1 mM glucose, and 5 mM sodium ascorbate. Following perfusion, the brain was rapidly dissected and horizontal slices (220 m) were prepared using a vibratome. Slices recovered for 1 h at 34°C in oxygenated HEPES holding solution: 86 mM NaCl, 2.5 mM KCl, 1.2 mM NaH 2 PO 4 , 35 mM NaHCO 3 , 20 mM HEPES, 25 mM glucose, 5 mM sodium ascorbate, 2 mM thiourea, 3 mM sodium pyruvate, 1 mM MgSO 4 , and 2 mM CaCl 2 (Ting et al., 2014), and then were New Research held in the same HEPES solution at room temperature until use. Slices were then transferred to a recording chamber where they were submerged in ACSF without sodium ascorbate.

Electrophysiology
Electrophysiological experiments were performed in horizontal midbrain slices containing the VTA that were continuously perfused with ACSF containing 10 M 6,7dinitroquinoxaline-2,3-dione (DNQX) and 1 M strychnine, AMPA and glycine receptor antagonists respectively. Except where noted, recordings also included the NMDA receptor (NMDAR) antagonist D-APV (50 or 100 M). Whole-cell recordings were performed from neurons in the lateral VTA with KCl pipette solution and voltageclamped at -70 mV. Patch pipettes were filled with the following: 125 mM KCl, 2.8 mM NaCl, 2 mM MgCl 2 , 2 mM ATP-Na ϩ , 0.3 mM GTP-Na ϩ , 0.6 mM EGTA, and 10 mM HEPES. In some experiments, EGTA was increased to 15 mM, or GDP-␤-S (1 mM) was included in the pipette solution as noted. The presence of a large hyperpolarization-activated inward current (I h ) was used to select postsynaptic cells for recording, although we are aware that this metric can allow inclusion of a subset of nondopamine neurons. If the steady-state I h was Ͼ25 pA during a step from -50 to -100 mV, the cell was included in analyses. In a subset of recordings, dopamine cells were also identified via fluorescence imaging using a DAT::IRES-cre x TdTomato reporter line. All experiments were performed at 30°C, maintained by an automatic temperature controller. The series resistance was monitored continuously during the experiment and cells were discarded for deviations Ͼ15%.

Stimulation protocols
For electrical stimulation, a bipolar stainless-steel stimulating electrode was placed caudal to the VTA ϳ500 m from the recorded cell; for rostral placement the stimulating electrode was placed within the VTA at 200 -500 m from the recorded cell (Fig. 1A). IPSCs were evoked at 0.1 Hz using 100-s current pulses. We used input-output curves to identify the stimulation intensity used for plasticity experiments for both rostral and caudal afferents, and the baseline amplitude was at 50% of this generated curve. No correlation was observed between stimulation intensity and LTP magnitude. This stimulation protocol did not produce action potentials escaping voltageclamp, but in the rare cases that cells began spiking later in the recording, they were excluded from analysis. Channelrhodopsin-induced synaptic currents were evoked at 0.033 Hz using 0.1-to 5-ms light pulses from a white LED (Mightex) controlled by driver (ThorLabs) and reflected through a 40ϫ water immersion lens. When feasible, IPSCs were shown to be GABA A receptor mediated by bath application of 10 M bicuculline at the end of recordings. For all stimulus frequencies, intensity remained constant throughout the experiment.

Statistical analysis
Results are expressed as mean Ϯ SEM. Significance was determined using a two-tailed paired Student's t test or one-way ANOVA with significance level of p Ͻ 0.05 (Table 1). LTP values are reported as averaged IPSC amplitudes for 10 min just before LTP induction compared with averaged IPSC amplitudes during the 10-min period from 10 -20 min after manipulation. Paired-pulse ratios (50-ms interstimulus interval) and coefficient of variation were measured over 10-min epochs of 10 -30 IPSCs each. The paired pulse ratio was calculated using the average value for all IPSC2 amplitudes divided by the average value for the corresponding IPSC1 amplitudes and reported as the mean paired pulse ratio for that epoch. 1/CV 2 values were determined by dividing the mean amplitude of IPSCs squared recorded over 10-min epochs by the mean variance of these IPSCs.

Location of electrical stimulation determines expression of synaptic plasticity
Most reports examining synaptic plasticity in the VTA have used a stimulating electrode placed within the VTA 200 -500 M rostral to the recorded cell in a horizontal slice (Fig. 1A). This approach has been assumed to randomly sample the synaptic inputs onto cells within the VTA. We hypothesized that stimulating caudal to and outside of the VTA, ϳ500 M from the recorded cell, might bias the inputs differently than with a rostral placement (Fig. 1A). We refer to this as "caudal" stimulation. We recorded IPSCs in putative dopamine cells identified by a large I h and compared synaptic properties using either caudal or rostral electrical stimulation. We did not detect any differences between IPSCs evoked by rostral or caudal stimulation in the onset delay, rise slope, or time of peak amplitude of IPSCs ( Fig. 1B-D; paired t test of rostral versus caudal onset delay: p ϭ 0.82 a ; rise slope: p ϭ 0.74 b ; time of peak: p ϭ 0.98 c ; rostral: n ϭ 15 cells, caudal: n ϭ 18 cells). Opioids depress GABAergic inhibition in the VTA; therefore, we compared the opioid-sensitivity of caudal and rostral-stimulated IPSCs. IPSCs from both stimulating locations were depressed by 1 M DAMGO to the same degree (Fig.  1E,F; rostral ϭ -55 Ϯ 8%, n ϭ 6 cells; caudal ϭ -58 Ϯ 9%, n ϭ 13 cells; rostral versus caudal: p ϭ 0.86 d ). Thus, synaptic properties were similar when stimulating either the rostral or caudal site.
We next used a stimulation protocol known to induce the nitric oxide-dependent LTP GABA : high-frequency stimulation (HFS) consisting of two 100-Hz tetani separated by 10 s (Nugent et al., 2007). LTP GABA is dependent on NMDAR activation that leads to the release of nitric oxide and activation of a signaling cascade that increases presynaptic GABA release (Nugent et al., 2007(Nugent et al., , 2009). Instead, HFS of the caudally stimulated site resulted in LTP, even with the NMDAR antagonist, APV (100 M), in the bath solution ( Fig. 2A,B,E; 157 Ϯ 23% of baseline value; paired t test: p ϭ 0.018 e , n ϭ 16 cells). Conversely and consistent with prior results, the same tetanus of a rostrally-placed electrode did not potentiate IPSCs in APV (Nugent et al., 2007;Fig. 2C-E; 90 Ϯ 11% of baseline value; paired t test: p ϭ 0.73 f , n ϭ 6 cells). Potentiation after HFS of the caudally-stimulated electrode was correlated with a decrease in paired pulse ratio for cells that potentiated at least 10% ( Fig. 2F; baseline: 1.1 Ϯ 0.2, after HFS: 0.9 Ϯ 0.1), without a change in 1/CV 2 values ( Fig.  2G; baseline: 7.5 Ϯ 2.1, 10 -20 min after HFS: 8.1 Ϯ 1.4; paired t test: p ϭ 0.048 g and p ϭ 0.70 h , respectively, n ϭ 12 cells). These findings supported our hypothesis that electrode placement may activate different subsets of afferents and lead to a different outcome when performing protocols to induce synaptic plasticity.

Low-frequency stimulation (LFS) potentiates caudalevoked inhibitory inputs
Inhibitory synapses can be regulated bidirectionally by different afferent stimulation patterns. In an earlier study, LFS of afferents by a stimulating electrode placed rostral to the VTA cell being recorded was used to elicit LTD. LTD is induced by LFS, 6 min of 1-Hz stimulation while voltage clamping the postsynaptic cell at -40 mV (LFS-LTD; Dacher and Nugent, 2011). LFS-LTD occurs independently of NMDAR activation and is partially blocked by a dopamine D2 receptor antagonist (Dacher and Nugent, 2011). Given the surprising result with HFS of a caudally-  3F; baseline: 5.7 Ϯ 1.1, 10 -20 min after LFS: 7.8 Ϯ 1.8; p ϭ 0.09 k , n ϭ 19 cells). This surprising finding led us to conclude that as with HFS-induced LTP, previous observations of LTD following LFS were likely dependent on activation of a subset of VTA afferents.

LFS of optically-evoked inhibitory inputs in the VTA does not induce plasticity
There are many sources of GABAergic inhibition in the VTA and given that LFS can induce either LTD or LTP, depending on stimulation site, we wondered which form of plasticity was predominant when activating GABAergic synapses more globally. We hypothesized that just a subset of VTA synapses express LFS-LTP GABA , so that when using optical stimulation of VGAT ϩ inputs in a VGAT-ChR2 transgenic mouse line, both forms of plasticity might occur at different synapses on the same dopamine cell. We used a BAC transgenic mouse line, VGAT-ChR2(H134R)-EYFP, to activate multiple GABAergic inputs in the VTA (Zhao et al., 2011), and used whole-field LED illumination of the slice to activate inhibitory inputs. After generating a stable 10-min baseline of light-evoked IPSCs, we delivered optical LFS while depolarizing the postsynaptic cell to -40 mV. In contrast to what we observed with caudal electrical stimulation, the mean light-evoked IPSC amplitude was unchanged after optical LFS (Fig. 4A-C; 102 Ϯ 10% of baseline value; p ϭ 0.83 l , n ϭ 7 cells). PPR was not significantly altered after LFS (data not shown; baseline: 0.72 Ϯ 0.07, 10 -20 min after LFS: 0.72 Ϯ 0.05, p ϭ 0.97 m , n ϭ 7 cells) and neither were 1/CV 2 values (data not shown; baseline: 20.0 Ϯ 2.4, 10 -20 min after LFS: 20.2 Ϯ 3.6, p ϭ 0.95 n , n ϭ 7 cells).

LFS-induced LTP does not require postsynaptic G-protein-coupled receptor (GPCR) signaling and is not prevented by postsynaptic EGTA
Potentiation after LFS was not associated with a significant change in the paired pulse ratio or coefficient of variation (Fig. 3E,F), suggesting that the mechanism may reflect an increased GABAergic sensitivity of the postsynaptic cells. Most forms of LTP are triggered by increases in calcium concentration in the postsynaptic cell, and so we tested whether LFS-LTP requires a rise in postsynaptic calcium. When the concentration of the calcium chelator, EGTA, was raised to 15 mM in the patch pipette, LFS still resulted in potentiation of caudally-evoked IPSCs (Fig. 6A,C,D; 155 Ϯ 24% of baseline value; n ϭ 8 cells). An alternative postsynaptic mechanism might require activation of receptors on the postsynaptic cell other than GABA A receptors. For example, the report of LFS-LTD found that depression was partially dependent on dopamine D2 receptors (Dacher and Nugent, 2011), which are coupled to the G i subtype of GPCRs. When we included an inhibitor of GPCR activity (1 mM GDP-␤S) in the patch pipette to block all postsynaptic GPCR signaling, LFS still potentiated caudally-evoked IPSCs (Fig. 6B,C,D; 135 Ϯ 8% of baseline value; n ϭ 8 cells). The magnitude of LTP after LFS was not significantly different for high EGTA or GDP-␤-S conditions than experiments with normal KCl internal solution (F (2,27) ϭ 0.77, p ϭ 0.48 r , n ϭ 8 cells high EGTA, n ϭ 8 cells GDP-␤-S, n ϭ 14 cells normal KCl). Together our results suggest a mechanism that does not require postsynaptic GPCRs or Ca 2ϩ influx. Future experiments will be needed to understand this novel form of LTP.

Discussion
Most reports describing synaptic plasticity in the VTA used electrical stimulation, which can miss the possibility of circuit specificity that can now be probed using optogenetic tools. For example, LTP GABA (Nugent et al., 2007) was found to vary depending on presynaptic source (Simmons et al., 2017;Polter et al., 2018). Here, we report that inhibitory synapses in the VTA have different requirements for inducing LTP depending on the placement of the stimulating electrode.

Synaptic plasticity induction
Using the same stimulation protocol but with the electrode at a site that deviated from the usual placement, we serendipitously discovered that we could induce NMDARindependent LTP using either HFS or LFS. Pairing postsynaptic cell depolarization with afferent stimulation, to substitute for a strong tetanus, is a classic approach used to induce LTP in the hippocampus (Nicoll, 2017). However, this method for inducing LTP is generally due to NMDAR activation. Instead, robust caudal LFS-induced LTP in the VTA was elicited in the presence of an NMDAR antagonist. Previously described LFS-LTD in the VTA is also NMDAR independent (Dacher and Nugent, 2011). How might the same pattern of afferent stimulation result in opposite synaptic plasticity outcomes? One likely explanation is that different electrode locations preferentially activate different subsets of afferents in the VTA slice. The VTA dopamine cells are innervated both by local GABA neurons and by GABA projections originating in regions throughout the brain that may differ in protein expression leading to different forms of synaptic plasticity. Another possibility is that the timing of inputs differs when using the two stimulating electrode locations; however, we did not observe a significant difference in onset delay, rise slope, or time of peak amplitude of IPSCs from caudal versus rostral. We speculate that LFS with mild depolarization may lead to release of a signaling molecule from the postsynaptic or presynaptic cell. If different synapses express different receptor subtypes for that signaling molecule, then release via LFS could result in distinct synaptic strength changes. Future experiments will be needed to determine whether HFS-LTP and LFS-LTP result, e.g., from activation by metabotropic glutamate, endocannabinoid, or dopamine receptors.

LFS and synaptic plasticity
Numerous studies have shown that LFS induces LTD, often when paired with modest postsynaptic depolariza-  (Bear and Malenka, 1994;Gutlerner et al., 2002). There are fewer instances where LFS induces LTP, and these generally required pairing with strong depolarization to activate NMDARs (Bonci and Malenka, 1999;Lanté et al., 2006;Ikeda et al., 2007;Dringenberg et al., 2014). Here, we found that LFS potentiates VTA GABA synapses with mild depolarization and LFS that did not require NMDAR activation, an apparently rare mechanism at CNS synapses. One other example is at the excitatory synapse from lateral perforant path to dentate gyrus cells, where LFS also potentiates synapses independently of NMDAR activation (Gonzalez et al., 2014). However, to our knowledge, ours is the first report of LTP elicited by LFS at GABAergic synapses.

GABAergic afferents innervating the VTA
Exposure to drugs of abuse causes LTP at excitatory synapses on VTA DA cells (Ungless et al., 2001;Saal et al., 2003). Many drugs of abuse also block LTP GABA in the VTA (Nugent et al., 2007;Guan and Ye, 2010;Niehaus et al., 2010;Graziane et al., 2013). The net result of these drug-induced changes in synaptic strength is thought to be increased dopamine cell firing via enhanced excitatory drive and disinhibition. However, if more types of synaptic plasticity exist than previously suspected, differential effects of drugs of abuse on VTA afferents may produce a more nuanced effect on dopamine cell firing. LTP GABA is expressed at VTA GABA ¡VTA but not at RMTg GABA ¡VTA synapses; by analogy, it is likely that the HFS-induced and LFS-induced LTP we report here are expressed only at a subset of inputs. It is difficult to be certain precisely which afferents are sufficiently close to the caudal stimulation site to be activated by our stimulus protocol. Regions other than the RMTg that are located caudal to the VTA with reported GABAergic innervation include: the dorsal raphe, periaqueductal gray, pedunculopontine nucleus, and laterodorsal tegmentum (Beier et al., 2015;Omelchenko and Sesack, 2010;Faget et al., 2016;Ntamati et al., 2018). Given the placement of the caudal electrode in our experiments, it is possible that the presynaptic source of those inputs is from one of these caudal brain regions, although it is alternatively possible that regions located elsewhere in the brain send projections that pass through the caudal stimulation location.
In conclusion, depending on stimulation site, HFS and LFS can induce LTP at GABAergic synapses in the VTA via a mechanism that does not require NMDAR activation. These results support the recent findings that some forms of plasticity, like LTP GABA (Simmons et al., 2017;Polter et al., 2018), are selectively expressed at some synapses but not others. Together, this points toward a specificity of synaptic plasticity based on presynaptic partner and postsynaptic cell identity. Furthermore, this study highlights the fact that there may be plasticity mechanisms in the VTA still to be identified.