(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



GABAergic regulation of striatal spiny projection neurons depends upon their activity state [1]

['Michelle Day', 'Department Of Neuroscience', 'Feinberg School Of Medicine', 'Northwestern University', 'Chicago', 'Illinois', 'United States Of America', 'Marziyeh Belal', 'William C. Surmeier', 'Alexandria Melendez']

Date: 2024-02

(A) The RiboTag construct AAV5-DOI-hSyn-RpI22I1-3Xflag-2A-eGFP was injected into the striatum of Adora2a-cre mice at P18 or at 6 months of age. (B) The coronal slice images demonstrate both the coverage and restriction to the striatum of the stereotaxically injected AAV carrying the RiboTag and eGFP genes (stereotaxic injection coordinates: ML = −1.85, AP = +0.74, DV = −3.50). Scale bars = 1 mm and 20 μm. Ten days later, the infected tissue (green fluorescence) was dissected out with the aid of fluorescence microscopy and qPCR was performed. (C) mRNA abundance (ΔCT) levels for the chloride cotransporters NKCC1 (SLC12A1) and KCC2 (SLC12A5) were determined by qPCR in striata from Adora2a-cre mice 4 weeks and 6 months of age. The data underlying the graphs shown in the figure can be found in dx.doi.org/10.5281/zenodo.10386854 . AAV, adeno-associated virus; qPCR, quantitative polymerase chain reaction; SPN, spiny projection neuron.

It is commonly thought that the reversal potential of GABA A Rs [ 14 , 15 ] is governed in large part by the balance between the plasma membrane cation/Cl - co-transporters–NKCC1 and KCC2 [ 8 , 16 ]. To determine whether SPNs expressed NKCC1 and KCC2, the striata of 1- and 6-month-old Adora2-Cre mice were stereotaxically injected with an adeno-associated virus (AAV) carrying a DIO-RiboTag expression construct [ 17 ] ( Fig 1A ) . Four weeks later, mice were killed; total striatal mRNA and RiboTag-associated mRNA were harvested for quantitative polymerase chain reaction (qPCR) and RNASeq analyses ( Fig 1B ) . These experiments revealed that iSPNs robustly expressed mRNA coding for KCC2 (Slc12a5), but not NKCC1 (Slc12a2) ( Fig 1C ) . The relative expression of these transcripts did not change within the time window examined ( Fig 1C ) . The expression of RiboTag harvested, iSPN-specific transcripts was like the mRNA harvested from the entire striatum ( Fig 1C ).

Why is the reversal potential of the GABA A Rs relatively depolarized? The striatal circuitry is largely quiescent in the ex vivo brain slice, making it highly unlikely that ongoing GABAergic signaling was loading neurons with Cl - and pushing the reversal potential in a depolarized direction. Despite the absence of detectable levels of its mRNA, the functional contribution of NKCC1 to the reversal potential of GABA A R was tested by bath application of the NKCC1-selective antagonist bumetanide (10 μM); bumetanide did not change the GABA A R reversal potential ( Fig 2G ). Since GABA A Rs exhibit a significant permeability to HCO 3 - , the other determinant of the GABA A R reversal potential is the HCO 3 - equilibrium potential [ 8 ]. To assess the role of intracellular HCO 3 - in determining the GABA A R reversal potential, perforated patch recordings were obtained from SPNs in ex vivo brain slices (as described above) and then the reversal potential of GABA A Rs determined before and after inhibition of carbonic anhydrase (CA) with acetazolamide (10 μM). In vivo, CA catalyzes the conversion of cytosolic CO 2 to H + and HCO 3 - ( Fig 2H ) [ 8 ]. RiboTag/RNASeq analysis revealed that iSPNs expressed 2 cytosolic CA subtypes with intracellular catalytic domains (Car2>Car7; RNASeq read ratio = 4.3)—in agreement with previous work [ 19 ]. Nonspecific inhibition of these CAs with acetazolamide led to a significant negative shift in the reversal potential of GABA A Rs ( Fig 2I and 2J ), consistent with the inference that the relatively depolarized GABA A R reversal potential in SPNs was attributable to HCO 3 - flux.

To determine if there was any shift in the GABA A R reversal potential after weaning, gramicidin perforated patch recordings were made from SPNs in ex vivo brain slices taken from mice at 3 ages: young (approximately 1 month old), young adult (6 to 7 months old), and adult (approximately 9 months old) mice. SPNs recorded in this mode displayed the characteristic inward rectification, delayed time to the first spike at rheobase, and sustained repetitive spiking with suprathreshold current injection ( Fig 2D ). The membrane potential changes evoked in SPNs by RuBi-GABA uncaging on the peri-somatic membrane reversed near −60 mV at all ages ( Fig 2E and 2F ).

(A) Clomeleon-expressing iSPNs allowed visual identification of dendrites in gramicidin perforated-patch recording conditions where cells cannot be loaded with dyes via internal-pipette solution (920 nm laser maximum projection image, scale bar = 40 μm). When low (<100 MΩ) access-resistance was achieved in voltage-clamp mode, RuBi-GABA (10 μM) was uncaged with a 473 nm laser spot (approximately 1 μm diameter, 1 ms) in the presence of the synaptic blockers: TTX (1 μM), AP5 (50 μM), NBQX (5 μM), CGP-55845 (1 μM). The laser was targeted to the somatic region or to distal dendrites (blue spots, projection image). (B) Representative voltage traces showing GABA responses, recorded in serial, from the soma (top traces, scale bars = 20 pA/2 s) or the dendrite (lower traces, scale bars = 10 pA/2 s) as the membrane was manually stepped from −70 mV to −50 mV. (C) Plot of the current/voltage relationship between somatic and dendritic activation. The data, represented by medians with interquartile ranges, did not differ significantly between the soma and dendritic compartments (n = 5 each; soma, dendrite; slope = 0.96, 0.71; x-intercept = −55.9, −59.6 mV; R 2 = 0.77, 0.85, respectively). Current measurements were rounded to the nearest 0.5 pA. (D) Current-clamp experiments in gramicidin perforated-patch mode were performed to examine age-dependent shifts in reversal potential. Here, Adora2a-eGFP positive iSPNs could be visually identified and patched. When low (<100 MΩ) access-resistance was achieved in current-clamp mode, the resting membrane potential along with series of hyperpolarizing and depolarizing steps were used to examine cell health (traces, scale bars = 20 mV/200 ms). (E) RuBi-GABA (10 μM) was uncaged over the full-field (3 ms duration, 60× lens) with a 473 nm LED in the presence of the synaptic blockers: AP5 (50 μM), NBQX (5 μM), CGP-55845 (1 μM). Representative current traces showing GABA responses as the membrane was manually stepped from −80 mV to −50 mV, scale bars = 5 mV/200 ms. (F) Plot of the change in PSP amplitude at P30, P90, and P270. The data, represented by medians with interquartile ranges, did not differ significantly between the 3 ages tested (n = 5 each P30, P90, P270; slope = 0.75, 0.70, 0.75; x-intercept = −61.6, −61.5, −61.5 mV; R 2 = 0.93, 0.91, 0.94, respectively). Values were calculated to the nearest 0.5 mV for the ΔV/V measurements. The data shows that the reversal for GABA-induced current is a full 20 mV+ above the resting membrane potential for SPNs, typically, −80 to −85 mV. (G) The addition of bumetanide (NKCC1 blocker, 10 μM) did not change the GABA A R reversal potential significantly (n = 5, p = 0.4076). (H) Perforated patch recordings were obtained from Adora2a-eGFP iSPNs and then the reversal potential of GABA A Rs determined before and after inhibition of CA with acetazolamide (aceta, 10 μM). (I) Representative traces recorded from a visually identified iSPN from an Adora2a-eGFP mouse in gramicidin perforated patch in current-clamp mode in the synaptic blockers: AP5 (50 μM), NBQX (5 μM), CGP-55845 (1 μM), MPEP (1 μM), and CPCCOEt (50 μM). RuBi-GABA (15 μM) was uncaged using a single LED pulse (470 nm, 25 ms). The pulse was applied at an interval of 30 s while manually stepping the cell to different potentials from −80 to −50 mV, scale bars = 10 mV/100 ms. (J) Summary data shows that application of acetazolamide shifted the reversal of the GABA-induced current to more negative potentials (p = 0.03125, n = 6). Membrane potentials were adjusted to correct for the estimated liquid junction potential and then binned into 5 mV increments (-70, -65, -60, -55 and -50 mV). The data underlying the graphs shown in the figure can be found in dx.doi.org/10.5281/zenodo.10386854 . CA, carbonic anhydrase; PSP, postsynaptic potential; SPN, spiny projection neuron.

To determine the reversal potential of GABA A Rs in SPNs, ex vivo brain slices were prepared from young adult (6 to 7 months old) mice and then gramicidin perforated patch recordings were made from identified SPNs. Gramicidin is selectively permeable to monovalent cations, leaving the intracellular Cl - concentration ([Cl - ] i ) unperturbed. To visualize dendrites, SPNs were sparsely labeled using an AAV carrying a SuperClomeleon expression plasmid [ 18 ] ( Fig 2A ) . To activate GABA A Rs, RuBi-GABA was uncaged on the soma and dendrites using a blue laser spot ( Fig 2A ) . The somatic membrane potential was clamped at membrane potentials between −50 and −70 mV prior to uncaging GABA and the resulting currents monitored ( Fig 2B ). The amplitude and polarity of uncaging evoked currents were then plotted as a function of somatic membrane potential. The estimated reversal potential for somatic GABA A Rs was near −55 mV ( Fig 2C ) . Dendritic uncaging of GABA evoked currents which also reversed in polarity near −60 mV ( Fig 2C ) .

To simplify the afferent circuitry engaged in these experiments, mice expressing Cre recombinase under the control of the NPY promoter (NPY-Cre) were injected with the same AAV vector used in the ChAT-Cre mice ( Fig 3D ). NPY is expressed by NGFis and low-threshold spike GABAergic interneurons (LTSIs) [ 4 ]—both of which make GABAergic synapses primarily on SPN dendrites [ 22 ]. Optogenetic activation of NPY-expressing interneurons alone produced depolarizing PSPs that were kinetically similar to those evoked by optogenetic stimulation of ChIs ( Fig 3E and 3F ) .

(A) AAV9-hSyn-chronos-flex-eGFP was stereotaxically injected into the striatum of two-month-old ChAT-Cre X D1tdTomato mice (Stereotaxic coordinate injection: ML = −1.2, AP = −0.7, DV = −3.4). The coronal confocal slice image shows the expression of Chronos (green cells) in a ChAT-cre neuron (cholinergic interneurons) along with dSPNs expressing tdTomato (red cells, scale bar = 40 μm). The tissue was dissected and recorded from 21 days postinjection. (B) The mean (± SEM) of ChI-evoked EPSP responses recorded from visually identified SPNs in gramicidin perforated patch in current-clamp mode in the presence of synaptic blockers: NBQX (5 μM), AP5 (50 μM), CGP-55845 (1 μM), MPEP (1 μM), and CPCCOEt (50 μM). The LED pulse (470 nm, 5 ms) was applied at an interval of 60 s. The traces recorded before and after the addition of gabazine (10 μM). Scale bars = 1 mV/100 ms. (C) Box plots of data from dSPNs (n = 8) and iSPNs (n = 6). (D) NPY-Cre X D1tdTomato mice were injected as described in (A). Confocal image showing NPY-Cre neurons expressing Chronos (green) and dSPNs expressing tdTomato (red, scale bar = 40 μm). (E) Mean (+ SEM) of NPY-Cre-evoked EPSP responses recorded from visually identified dSPNs in gramicidin perforated patch in current-clamp mode in the presence of blockers as described in (B) before and after the addition of Gabazine (10 μM). Traces from dSPN recorded in NPY (n = 4). Scale bars = 1 mV/100 ms. (F) Summary data for dSPNs (n = 4) and for iSPNs (n = 4). The data underlying the graphs shown in the figure can be found in dx.doi.org/10.5281/zenodo.10386854 . ChAT, choline acetyltransferase; ChI, cholinergic interneuron; PSP, postsynaptic potential; SPN, spiny projection neuron.

To study the role of synaptic GABA release, a mixed population of striatal GABAergic interneurons were activated by optogenetic stimulation of cholinergic interneurons (ChIs) [ 20 ]. Working through nicotinic acetylcholine receptors (nAChRs), ChIs can activate both neurogliaform interneurons (NGFIs) and tyrosine hydroxylase interneurons (THIs), giving rise to GABA A R-mediated currents in SPNs [ 21 ]. To monitor evoked responses in SPNs, perforated patch recordings were made from identified iSPNs or dSPNs using the approach described above. To optogenetically activate ChIs, an AAV carrying a Cre recombinase-dependent expression construct for Chronos was injected into the striatum of transgenic mice expressing Cre recombinase under the control of the choline acetyltransferase (ChAT) promoter ( Fig 3A ). In the ex vivo brain slice, SPNs are quiescent and reside in the down-state near −80 mV [ 10 ]. As predicted from the GABA uncaging studies above, optical stimulation of ChIs in the presence of iGluR antagonists evoked depolarizing, PSPs in SPNs that were blocked by the GABA A R antagonist gabazine ( Fig 3B and 3C ) .

(A) Maximum projection image of a visually identified dSPN from a D1R-tdTomato x ChAT-cre mouse with a high magnification image of a distal dendrite where 720 nm 2PLSM spot uncaging of DNI-Glu (2PLU, 5 mM) was conducted (red dots). Tomato+ dSPNs were patched in whole-cell mode and the cells were loaded with Alexa 568 for clear identification of dendrites and spines. Scale bars = 40 μm cell, 5 μm dendrite. (B) Scheme for interrogating endogenous GABA release from NGFIs onto SPNs via optogenetic stimulation of ChAT-cre mice expressing Chronos. (C, D) Throughout the dendrites, glutamate uPSPs in dSPNs and iSPNs can be evoked by uncaging DNI-Glu (5 mM, 1 × 15 spines, 1 ms pulses at 500 Hz, red traces, 720 nm laser) while stimulating GABA release from NGFIs with the blue laser (1 × 3 ms pulse, blue traces, 473 nm, within approximately 20 μm of the dendrite). From the quiescent down-state, GABA A R activation is depolarizing and pushes SPNs toward enhanced dendritic integration in both dSPN and iSPN dendrites (Glu-2PLU + GABA A opto = black trace, scale bars = 5 mV/200 ms). (E) Summary data showing the enhancement in amplitude and duration of the plateaus at ½ the maximum amplitude (1/2max) in iSPNs and dSPNs combined (n = 11 total: 3 iSPNs + 8 dSPNS; p < 0.001 for both amplitude and 1/2max duration, respectively). (F) Scatter plot of duration at ½ maximum amplitude vs. amplitude for clustered glutamate alone (red) and following GABA A R activation (black). Median effects (open circles) and the median absolute difference as capped lines are also illustrated. All experiments are conducted in the appropriate cocktail of synaptic blockers: CGP-55845 (1 μM), MPEP (1 μM), and CPCCOEt (50 μM). The data underlying the graphs shown in the figure can be found in dx.doi.org/10.5281/zenodo.10386854 . 2PLSM, two-photon laser scanning microscopy; ChAT, choline acetyltransferase; ChI, cholinergic interneuron; SPN, spiny projection neuron.

But, what if the GABA A R activation precedes the glutamatergic input to dendrites? A priori, one might predict that the dendritic depolarization produced by GABA A R opening would enhance the response to trailing glutamatergic input, much like the situation described at the soma. To test this hypothesis, 2 sets of experiments were performed. Identified iSPNs or dSPNs were recorded from in whole-cell mode to allow them to be filled with a dye (Alexa 568) and imaged using two-photon laser scanning microscopy (2PLSM) [ 25 ]. The [Cl - ] in the pipette was adjusted to yield a GABA A R reversal potential near −60 mV. Next, a region of parfocal dendrite was identified to allow two-photon uncaging of DNI-glutamate at visualized spine heads [ 11 , 26 , 27 ]. In the first set of experiments, dendritic GABA A Rs were activated by optogenetic stimulation of ChIs as described above. Because of their large axonal field and those of the NGFIs/THIs they activate [ 4 , 28 ], optogenetic stimulation of ChIs should produce a diffuse GABAergic input to the dendrites of the recorded SPN ( Fig 4A and 4B ) . As shown above, optogenetic stimulation of ChIs alone evoked a consistent but modest somatic depolarization ( Fig 4C and 4D ) . Dendritic uncaging of glutamate alone also evoked a somatic depolarization. The number of axospinous sites stimulated was adjusted to be subthreshold for dendritic spike generation (assessed by the decay of membrane potential after termination of uncaging) ( Fig 4C and 4D ) . When this uncaging event was preceded by ChI-evoked GABA A R depolarization, the resulting magnitude and duration of the somatic depolarization was significantly increased in both types of SPN ( Fig 4E ) . A scatter plot of the algebraic sum of the amplitudes of the GABAergic and glutamatergic PSPs in isolation against the amplitude of the response to the combined stimulation revealed that the 2 inputs almost invariably summed linearly or supra-linearly ( S1 Fig ). A scatter plot of the amplitude and duration of iGluR-mediated responses demonstrated that prior engagement of GABAergic interneurons (by ChI stimulation) enhanced the iGluR-mediated responses ( Fig 4F ). Thus, transiently opening dendritic GABA A Rs produced a dendritic membrane potential change that enhanced the ability of subsequent dendritic glutamatergic input to push SPNs toward the local spike threshold.

How might the interaction between GABA A Rs and iGluRs play out in dendrites? A key feature of SPN dendrites beyond about the first major branch point (approximately 80 μm from the soma) is the ability to generate dendritic spikes or plateau potentials that can last for 50 to 200 ms [ 11 – 13 , 24 ]. These dendritic spikes require the temporal convergence of 10 to 15 glutamatergic inputs over a relatively short stretch (approximately 20 μm) of dendrite, which produces enough of a local depolarization to engage NMDARs and voltage-dependent Ca 2+ channels. Previous experimental and modeling work has shown that opening GABA A Rs near the site of glutamatergic stimulation after spike initiation can truncate them, much like somatic situation described above. Indeed, as modeling suggests that the dendritic membrane potential during these spikes rises close to 0 mV, GABA A R opening should hyperpolarize the dendrites [ 13 ].

As shown previously, in both SPNs and pyramidal neurons [ 5 , 6 , 23 ], a depolarizing GABA A R input can boost the response to a trailing intrasomatic current injection and enhance the probability of spiking. However, GABA A R activation also can suppress spike generation by membrane shunting and pushing the membrane potential below spike threshold, which is typically between −45 and −50 mV [ 23 ].

Computational modeling of dendritic integration in SPNs

Although intriguing, the experimental results presented are limited by the inability to control the timing and location of GABAergic input to dendrites in a rapid precise manner. Understanding how the timing and dendritic location of GABA A R activation modulates the response to clustered excitatory input could provide insight into the role of GABAergic interneurons in striatal computation. To help achieve a better grasp of the mechanisms underlying this interaction, a modified NEURON model of a dSPN [13,29–31] was used to assess the impact of timing and location of GABAergic input on the response to clustered glutamatergic synaptic input to a stretch of distal dendrite. As observed experimentally, clustered glutamatergic input was able to generate NMDAR-dependent, dendritic spikes or plateau potentials when delivered to distal dendrites of a quiescent neuron (Fig 5A–5C).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Computational modeling of dSPN dendritic activity. (A) Morphology of reconstructed dSPN with cartoon to illustrate the direction of stimulation along a dendrite. (B) Synaptic potentials recorded at the soma in response to clustered spine activation (18 neighboring glutamatergic synapses stimulated sequentially at 1 ms intervals) in 2 separate dendrites. A dendritic spike is generated to the distal (orange trace) but not proximal dendritic input (blue trace). The quarterdrop duration is defined as the time interval between the last stimulation and the time for the membrane voltage to drop by one quarter of its peak value. (C) Quarterdrop interval (ms) plotted as a function of path distance from the center of that dendrite to the soma (μm) for every dendrite with spines (50) of the reconstructed dSPN. Dendritic spikes were only reliably observed in distal dendrites (>100 μm from cell soma). (D) Onsite phasic GABAergic activation and glutamatergic activation delivered to the same distal dendrite. Synaptic potentials recorded at the dendrite (E) and soma (F). GABA synaptic activation comprised 3 simultaneous stimulations delivered 5 times at an interval of 1 ms to the midpoint of the dendrite. The timing of this phasic input was varied relative to a fixed clustered supra-threshold glutamatergic input (delivered to 18 spines; 1 ms interval as before) in intervals of 10 ms from −10 (blue) to 80 ms (orange). For comparison, the effect of glutamatergic activation alone is illustrated by a thick gray line. Onsite GABAergic activation causes a dramatic cessation of dendritic and somatic potentials in a manner consistent with the relative timing of the 2 inputs. (G) Offsite phasic GABAergic activation delivered to 4 distal dendritic locations. A clustered glutamatergic input (15 spines; 1 ms interval) delivered to the same dendrite as before resulted in a subthreshold synaptic potential (red) at the dendritic site of delivery (H) and soma (I). Similarly, the effect of only activating GABAergic synapses at 1 ms intervals at each of the 4 offsite dendritic locations simultaneously (3 per dendrite; 12 in total) resulted in a moderate postsynaptic potential (blue trace). When delivered sequentially, with GABAergic activation preceding glutamatergic by 10 ms, the previously subthreshold glutamatergic input (red traces) resulted in the generation of a spike (black trace; H and I). The data underlying the graphs shown in the figure can be found in dx.doi.org/10.5281/zenodo.10386854. SPN, spiny projection neuron. https://doi.org/10.1371/journal.pbio.3002483.g005

When GABAergic synapses were activated near glutamatergic synapses, the model behaved as previously described by Du and colleagues. That is, GABAergic input at almost any point during the dendritic spike (when the local membrane potential was near −30 mV) led to inhibition of both the dendritic and somatic membrane potential (Fig 5D–5F). However, the impact of GABAergic input was very different when the site of stimulation was at some distance from that of glutamatergic stimulation. For example, if the GABAergic input was distributed at distal locations across the dendritic tree, as predicted to happen following ChI or NGFI/THI activation, the effect was consistently excitatory. To illustrate this point, the distributed GABAergic input was followed by a subthreshold dendritic glutamatergic input. In this scenario, the combination of GABAergic and glutamatergic input led to an NMDAR-dependent dendritic spike (Figs 5G–5I and S4A–S4C)—just as seen experimentally using optogenetic stimulation of GABAergic interneurons and 2P uncaging of glutamate.

The default value for cytoplasmic resistivity (R a ) in these computational simulations was 200 Ω cm. However, R a is a difficult parameter to measure experimentally and there is little overall consensus as to its true value and estimates vary at least 5-fold (70 to 350 Ω cm) [32]. For this reason, simulations were repeated with R a set to 100 Ω cm. The results were qualitatively similar to those described above (S2 Fig), suggesting that R a is not a major factor in our simulations when varied within the proposed physiological range.

In many of the previous studies examining the impact of GABA A Rs on dendritic integration of glutamatergic input, the focus has been on the role of timing and location dependent GABA A R -mediated shunting of iGluR-evoked EPSPs on the same dendrite [33–35]. As shown above, our results are largely consistent with this literature. Of particular interest is the timing dependence of the interaction. To explore this relationship in SPNs, NEURON simulations were run with on-site glutamatergic and GABAergic input to the same distal dendrite (Fig 6A). As expected, there was a strong timing dependence on the interaction between synaptic events. As experimentally shown by others [6,23], when the glutamatergic EPSP preceded a neighboring GABAergic input, the effect of opening GABA A Rs on the input impedance (i.e., shunting) was clearly evident (Fig 6B and 6C). However, when the glutamatergic input arrived later, there was synaptic summation, albeit sublinear at both dendritic (Fig 6B) and somatic locations (Fig 6C). To better illustrate the quantitative interaction between the 2 inputs at the dendritic site of stimulation, 2 plots were generated. In one, relative amplitude of glutamatergic EPSP (P 2 ) with a concomitant GABAergic input was divided by the amplitude of the glutamatergic EPSP alone (P 1 ) and then plotted as a function of the relative timing of the 2 inputs (Fig 6D). This ratio (P 2 /P 1 ) fell when the glutamatergic input preceded the GABAergic input and then rose when it trailed the GABAergic input. Similarly, if the ratio of the peak amplitude of the aggregate potential (P 3 ) was divided by the peak amplitude of the isolated glutamatergic EPSP (P 1 ) and plotted as a function of the relative timing of the 2 inputs, the ratio fell when the glutamatergic input preceded the GABAergic input, but then rose above 1 when the glutamatergic trailed the GABAergic input (Fig 6E).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 6. The interaction between glutamatergic and GABAergic synaptic activity. (A) On-site phasic GABAergic activity was delivered to the same dendrite as glutamatergic synaptic input. Synaptic potentials are illustrated at (B) dendritic site of glutamatergic activity and (C) cell soma. The black trace represents the effect of glutamate-mediated excitation alone (15 neighboring spines activated at 1 ms intervals along the chosen dendrite). The light gray trace illustrates the effect of phasic GABAergic stimulation (3 simultaneous synaptic stimulations delivered 4 times at an interval of 1 ms to the same dendritic site giving a total of 12 GABA synapses activated). Color traces represent the effect of varying glutamatergic stimulation at temporal intervals relative to the fixed GABAergic input described (blue to orange illustrate 5 traces with Δt = t GLUT −t GABA in the range of −10 to 30 ms, respectively, at 10 ms intervals). Inset illustrates the measurement of P 1 , P 2 , and P 3 . The absolute amplitude of the synaptic potential in the dendrite was measured in the absence of GABAergic activity (P 1 ) or either relative to the amplitude of the underlying depolarizing phasic GABAergic potential at the peak of the postsynaptic response (P 2 ) or relative to the underlying baseline (P 3 ). (D) and (E) show the sublinear effect of varying glutamatergic spine activation relative to a fixed onsite GABA synaptic input on P 2 and P 3 normalized to P 1 , respectively. (F) Off-site phasic GABAergic activity was delivered to 4 distal dendrites distinct from the dendrite receiving clustered spine excitation. As before, (G) and (H) show simulated synaptic potentials recorded at dendrite and soma. The black trace represents the same glutamatergic input as above (i.e., 15 spines activated at 1 ms intervals). The light gray trace represents the effect of phasic GABAergic activity delivered to the 4 distal dendrites at the same time (as 3 synaptic simulations at 1 ms intervals per dendrite giving a total of 12 GABA synapses activated). Again, as before, color traces show the effect of altering the timing of clustered spine activation relative to GABA activity. In contrast to on-site activity, Δt = 0, 10, and 20 ms results in the generation of a dendritic spike. Note that the peak of the EPSP GLUT is still clearly visible before the trailing spike manifests. (I) and (J) illustrate the supralinear effect of varying glutamatergic spine activation relative to a fixed off-site GABA synaptic input on P 2 and P 3 normalized to P 1 , respectively. The data underlying the graphs shown in the figure can be found in dx.doi.org/10.5281/zenodo.10386854. https://doi.org/10.1371/journal.pbio.3002483.g006

Of greater interest, given the architecture of striatal circuits, was how a diffuse GABAergic input (mimicking conditions produced by ChI activation) would affect the interaction between synaptic events. To explore this interaction, the temporal relationship between a focal glutamatergic input to a distal dendrite and a GABAergic input to 4 neighboring dendrites was examined (Fig 6F). Not surprisingly, in this situation there was no shunting and the 2 inputs summed at both the dendritic (Fig 6G) and somatic (Fig 6H) locations—regardless of relative timing. In fact, GABAergic input enhanced the ability of glutamatergic synapses to trigger a dendritic spike (Fig 6G and 6H). To probe the dendritic interaction, the relative amplitude of the mixed PSP (measured relative to the underlying GABAergic PSP—P 2 ) was divided by the amplitude of the isolated glutamatergic EPSP (P 1 ) and then plotted as a function of the relative timing of the 2 inputs. At all intervals, the ratio was greater than or equal to 1 (Fig 6I). A qualitatively similar plot was obtained by computing the ratio of the peak amplitude of the mixed PSP (P 3 ) divided by the glutamatergic EPSP amplitude (P 1 ) as a function of relative timing of the 2 inputs (Fig 6J).

In these simulations, 3 GABAergic synapses were activated in sequence (1 ms interval) at 4 distal, off-path dendrites (a total of 12 GABAergic synapses activated). Increasing the number of dendrites stimulated to 12, each receiving 1 GABAergic synapse (i.e., to maintain a total of 12 synapses activated) produced qualitatively similar results (S4D–S4H Fig). The widening of the temporal window for supralinear summation presumably resulted from an increase in amplitude of the GABAergic PSP at the site of glutamatergic input. In this simulation, 9 (or more) GABAergic synapses were needed for supralinear summation (S4I and S4H Fig). When the location of GABAergic synapses was shifted to a more proximal dendritic location, shunting decreased (S3A–S3C Fig) and the temporal window for supralinear summation broadened (S3D and S3E Fig). Thus, only GABAergic synapses near the site of glutamatergic input prevented supralinear summation. However, if a dendritic spike was generated, proximal GABAergic input had little effect on its propagation to the soma.

SPNs exhibit tonic GABA A R-mediated currents [36]. To assess its effect on dendritic integration, tonic GABA A R current was modeled as a uniformly distributed conductance (Fig 7A). As expected, increasing tonic GABA A R conductance density progressively depolarized the membrane potential, approaching the reversal potential for E GABA (−60 mV) (Fig 7B and 7C). A modest elevation in the tonic GABA A R current led to a dendritic spike in response to a previously subthreshold, clustered excitatory input. These spikes, like those described previously (S4A–S4C Fig), were dependent upon NMDARs (Fig 7E). The excitatory effect of tonic GABAergic currents on dendritic excitability was evident over a broad range of conductance values, with shunting becoming significant only at large values (Fig 7D). Thus, the “dose-response” relationship between tonic GABA A R current and “boosting” of the dendritic response to glutamatergic input had an inverted “U” shape.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. The interaction between tonic GABA A R signaling and glutamatergic activity. (A) Tonic GABAergic signaling was modeled as a conductance that was evenly distributed throughout the soma and dendritic tree. Glutamatergic synaptic input was delivered to the dendrite illustrated as clustered spine excitation as before. Synaptic potentials at the dendritic site of glutamatergic input (B) and at soma (C) are illustrated. The black trace represents the effect of synaptically released glutamate alone (15 neighboring synapses excited at 1 ms intervals). Color traces (blue to red) illustrate the effect of increasing tonic GABA conductance (from 10−6 to 3 × 10−2 S/cm2) on the glutamate response and resting membrane potential. Inset of (D) illustrates the measurement of P 3 relative to P 1 with absolute amplitudes measured relative to its underlying resting membrane potential. (D) Increasing tonic GABA conductance density caused an incremental depolarization (ΔV in red) that approached the reversal potential for E GABA (−60 mV) at >10−2 S/cm2. Increasing depolarization (2–3 mV to 10−5 S/cm2 density) was accompanied by a dendritic spike in response to a glutamatergic input that was subthreshold in the absence of tonic GABA activation. This supralinear summation persisted for greater than an order of magnitude increase in tonic GABA activation (3 × 10−4 S/cm2); larger values led to local shunting. (E) Illustrates the requirement for glutamate-mediated synaptic activation of NMDAR for tonic GABA-mediated spike generation. The simulation was identical to that illustrated in (B–D) except that the NMDAR conductance at glutamatergic synapses was zero. In the absence of NMDARs, a combination of onsite shunt and reduced AMPAR driving force arising from depolarization presumably underlies the GABA conductance density-dependent decrease in response. The data underlying the graphs shown in the figure can be found in dx.doi.org/10.5281/zenodo.10386854. NMDAR, N-methyl-D-aspartate receptor. https://doi.org/10.1371/journal.pbio.3002483.g007

To better illustrate the role of location in dictating the shunting effect of a GABAergic synapse, the dendritic voltage and input impedance at the distal dendritic site was computed for a range of GABAergic synapses/dendrite (0–24). Near the site of GABAergic input, the evoked dendritic depolarization progressively increased with the number of GABAergic synapses, but the input impedance fell in parallel (Fig 8A–8C). In contrast, on neighboring dendrites, the depolarization grew with the number of synapses, but there was little local change in input impedance (Fig 8D–8F).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 8. On vs. off-site GABAergic activation and local input impedance. (A) On-site phasic GABAergic activity was delivered and recorded in the same dendrite. (B) Color traces (blue to orange; above) represent simulated GABAergic PSPs and local dendritic impedance (to 10 Hz; below) to increasing synaptic activation (1, 3, 6, 12, and 24 GABA synapses per dendrite). (C) Increasing PSP amplitude (blue) to GABAergic synaptic stimulation reduced local impedance (red). (D) Off-site phasic GABAergic activity was delivered to 4 distal dendrites and recorded in the same dendrite as in (A). (E) and (F) as above. Note the lack of effect of increasing off-site GABAergic activity on local impedance. The dotted lines on panels (C) and (F) illustrate the effect of activating 12 GABAergic synapses in total, which represents the common scenario setting for these 2 conditions throughout the study. The data underlying the graphs shown in the figure can be found in dx.doi.org/10.5281/zenodo.10386854. PSP, postsynaptic potential. https://doi.org/10.1371/journal.pbio.3002483.g008

The inference to be drawn from these simulations is that the interaction between dendritic glutamatergic and GABAergic synapses depends upon both location and timing. When co-localized, the timing of the 2 inputs dictates their interaction, as previously described [6,23]. However, when the GABAergic input is more diffuse, as predicted to occur with activation of ChIs or NGFIs, the relative timing of GABAergic and glutamatergic inputs becomes less important, and the 2 depolarizing inputs sum. This additivity allows GABAergic and glutamatergic synapses to work together to trigger dendritic spikes—as observed experimentally.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002483

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.