Mechanisms and Functions of GABA Co-Release

PROGRESS

into individual synaptic vesicles or from the coincident membrane fusion of Mechanisms and functions of GABA different vesicles filled with distinct neuro transmitters (FIG. 1a–d). Because small co‑release molecule transmitters and neuropeptides are often distributed to separate vesicular compartments (clear synaptic vesicles and Nicolas X. Tritsch, Adam J. Granger and Bernardo L. Sabatini large dense core vesicles) and differentially Abstract | The ‘one neuron, one neurotransmitter’ doctrine states that synaptic mobilized by intracellular Ca2+ (see REF. 2), communication between two neurons occurs through the release of a single the term co‑release typically applies to chemical transmitter. However, recent findings suggest that neurons that exocytosis of two classical neurotransmitters. The term co‑transmission also refers to communicate using more than one classical neurotransmitter are prevalent synaptic release of multiple substances from a throughout the adult mammalian CNS. In particular, several populations of single cell, but it does not specify mechanism neurons previously thought to release only glutamate, acetylcholine, dopamine and, therefore, also extends to gaseous or histamine also release the major inhibitory neurotransmitter GABA. Here, we transmitters. It does, however, emphasize review these findings and discuss the implications of GABA co‑release for synaptic function, as it implies that both transmitters contribute to synaptic transmission, either transmission and plasticity. by evoking postsynaptic electrical responses or by modulating the excitability of pre- and Neurons in the vertebrate and invertebrate capable of modulatory functions, because postsynaptic membranes1. Hence, co‑release nervous systems contain and release several glutamate, GABA, ACh and purines can might not result in co‑transmission if the small-molecule and peptide transmitters, activate GPCRs in addition to ligand-gated target cell does not express receptors for both each capable of signalling through a variety ion channels. Second, the synaptic transmitters (FIG. 1e). We note that the relative of receptors. A distinction is often made actions of individual neurons vary with ease with which fast synaptic currents can between classical small-molecule neuro- developmental age, membrane potential, be detected experimentally compared with transmitters, including glutamate, GABA, cellular identity and the biochemical state of the detection of slow modulatory influences glycine, acetylcholine (ACh), purines and the postsynaptic cell. Third, in the nervous has biased the study of co‑transmission monoamines, which are the primary means system of invertebrates and in the peripheral to classical neurotransmitters acting on of transferring electrical signals between nervous system and developing CNS of ionotropic receptors. synaptic partners, and neuropeptides such vertebrates, many neuronal populations as somatostatin, neuropeptide Y, substance P have been shown to release more than one Examples of GABA co‑release and enkephalin (among many others), which classical neurotransmitter1–4. However, At a minimum, the co‑release of two neuro slowly alter the properties of target neurons the prevalence and physiological role of transmitters requires that presynaptic by activating G protein-coupled receptors such co‑release in the adult mammalian terminals express the molecular machinery (GPCRs). It has been established for several CNS are less well established. In this necessary to acquire both transmitters and to decades that exocytosis of one or more Progress article, we discuss recent reports package them into synaptic vesicles. GABA neuropeptides accompanies the release of of neuronal populations in the rodent is typically synthesized from glutamate by classical small-molecule neurotransmitters adult CNS that co‑release GABA, examine glutamate decarboxylase 65 kDa isoform in most neurons1,2. the molecular mechanisms involved and (GAD65; encoded by Gad2) and/or glutamate Nevertheless, our understanding consider the functional significance of decarboxylase 67 kDa isoform (GAD67; of neural function and rapid synaptic GABAergic co‑transmission for neuronal encoded by Gad1), and it is transported into communication has been dominated by communication and circuit computations. synaptic vesicles using the vesicular GABA the idea that a neuron releases a single transporter (VGAT; also known as VIAAT, classical neurotransmitter at all of its Defining co‑release which is encoded by Slc32a1)5. The degree synapses — a notion commonly known as The term neurotransmitter co‑release of plasma membrane uptake of GABA and Dale’s principle (BOX 1). This principle has describes the process by which two (or glutamate has also been shown to influence helped reduce the complexity of the nervous more) neurotransmitters are released by the amount of GABA that is exocytosed, system by assigning each neuron to one a single neuron in response to an action particularly during sustained synaptic of three functional categories: excitatory, potential. This description emphasizes stimulation6–8. GABA signals through − inhibitory, or modulatory. However, several the mechanism involved and is restricted Cl -permeable type A GABA (GABAA) important facts complicate the simple char- to the release of neurotransmitters by and GABAC receptors, as well as 4 acterization of neurons in this way. First, vesicular exocytosis . Co‑release results Gαi/o‑coupled GABAB receptors. In this excitatory and inhibitory neurons are also from packaging multiple transmitters section, we summarize the classical

PROGRESS

Box 1 | In defence of Sir Henry Dale from starburst amacrine cells (SACs) onto direction-sensitive retinal ganglion In the early 1900s, Otto Loewi (1873–1961) observed that stimulation of frog parasympathetic and cells (DSGCs)15. Interestingly, this study sympathetic nerves evoked the release of substances that inhibited or accelerated heart rate, revealed that vesicular release of GABA respectively. Henry Dale (1875–1968) subsequently helped identify the released transmitters — and ACh is differentially regulated by acetylcholine and noradrenaline — and coined the terms ‘cholinergic’ and ‘adrenergic’ to describe 2+ nervous systems that liberated these molecules. These findings, which earned Loewi and Dale the intracellular Ca and is spatially uncoupled; 1936 Nobel Prize in Physiology or Medicine, established the chemical basis of neurotransmission ACh transmission is reliably evoked in all and fostered the concept that neurons have distinct chemical identities that specify their function. SAC–DSGC pairs whereas GABA release Dale is often cited for proposing that each neuron releases a single classical transmitter — a rule occurs preferentially in response to visual that is increasingly proving not to hold. However, in his defence, Dale recognized the possibility stimuli moving in one particular direction that nerves release more than one molecule and never explicitly formulated or embraced the and is significantly biased to one half of principle that has become synonymous with his name70. In fact, the phrase ‘Dale’s principle’ was the radial dendritic tree of a DSGC15. The introduced by John Eccles (1903–1997) in 1954 in reference to Dale’s hypothesis that neurons spatial bias emerges from the asymmetric release the same transmitter (or set of transmitters) across all of their synapses. Interestingly, development of SAC GABAergic synapses recent work10,38,71 suggests that this may not be universally true. onto DSGCs16,17. These data strongly suggest that SACs package GABA and ACh into distinct populations of synaptic neurotransmitter systems in which the with the antidepressant citalopram11. vesicles that distribute to non-overlapping co‑release of GABA has been observed in These findings add to a growing literature or partially-overlapping presynaptic the adult mammalian CNS and discuss the implicating LHb dysfunction in major terminals (FIG. 1d). molecular and cellular mechanisms involved depressive disorders13 and suggest that More recently, functional evidence of in each example. changes in glutamate–GABA co‑release co‑release of GABA and ACh has been within the LHb might be a therapeutic target found in the cortex of mice, which receives Glycine–GABA. Functional co‑release of for these illnesses. most of its cholinergic input from basal GABA and glycine in the adult CNS was first Most axons from EP and VTA that forebrain projection neurons. Optogenetic reported in the late 1990s and has since been terminate in LHb co‑express vesicular activation of cholinergic fibres in general18, extensively documented in several neuronal transporters for GABA and glutamate11,12, or of cholinergic fibres originating in globus populations in the spinal cord, brainstem and indicating that individual synapses are pallidus specifically19, evoked postsynaptic cerebellum (reviewed in REF. 4). Glycine and capable of packaging and releasing both responses in inhibitory cortical interneurons

GABA are released from the same synaptic transmitters. Quantal synaptic events with mediated by both GABAA and nicotinic vesicles, as both neurotransmitters utilize biphasic waveforms have occasionally ACh receptors. The short onset latency and VGAT for packaging into synaptic vesicles5 been observed11, providing compelling pharmacological properties of GABAergic (FIG. 1a). The relative contribution of each functional evidence that GABA and currents suggest that cholinergic fibres of these transmitters to neurotransmission glutamate are co‑packaged at some release directly release GABA. As in the retina, at individual synapses varies with the local sites in LHb (FIG. 1b). However, it is difficult cortical ACh–GABA co‑release seems to presynaptic concentration of GABA and to experimentally exclude the possibility occur through separate populations of glycine and with the density and composition that there are also intermingled synaptic synaptic vesicles because individual cortical of postsynaptic receptors8–10. vesicles preferentially enriched with one or interneurons often exhibited responses the other transmitter (FIG. 1c). Interestingly, to only one transmitter, and labelling Glutamate–GABA. Classically, it is thought individual mesohabenular boutons establish of cholinergic presynaptic terminals that glutamate and GABA are released from symmetric and asymmetric postsynaptic from the globus pallidus showed spatial distinct sets of excitatory and inhibitory contacts enriched in GABAA and glutamate segregation of vesicular transporters for neurons in the adult CNS. However, despite receptors, respectively, which is suggestive GABA and ACh18,19. However, further the opposing functions of these neuro- of a postsynaptic division of labour12. research is required to determine the precise transmitters, two recent reports describe The finding that multiple synaptic inputs presynaptic mechanism and postsynaptic the co‑release of glutamate and GABA into this small brain area share the peculiar targets of ACh–GABA co‑release in cortex. from individual CNS axons. Neurons in property of releasing both GABA and the ventral tegmental area (VTA) and glutamate suggests that co‑release may Dopamine–GABA. Dopamine is synthesized entopeduncular nucleus (EP) were recently confer a specific function on this region. from tyrosine by the enzymes tyrosine shown to release both glutamate and hydroxylase (TH) and aromatic-l‑amino- GABA onto neurons in the lateral habenula ACh–GABA. Cholinergic neurons in acid decarboxylase (AADC) and is loaded (LHb)11,12. Activation of fibres from VTA the CNS have long been suspected into synaptic vesicles via the vesicular most frequently resulted in short latency of co‑releasing GABA, on the basis of monoamine transporter 2 (VMAT2; suppression of firing in LHb neurons. extensive co‑expression of GABAergic and encoded by Slc18a2). The identification of By contrast, stimulation of EP inputs reliably cholinergic markers in many vertebrate neurons capable of releasing dopamine has evoked spiking in LHb neurons, reflecting species14. Unambiguous functional historically been carried out by labelling TH a predominantly excitatory mode of action. evidence of co‑release was found in an in neurons and, more recently, by using mice Interestingly, the ratio of GABAergic elegant study of the mouse retina that expressing transgenes under the control to glutamatergic signalling from EP is demonstrated, using paired whole-cell of the TH promoter. These approaches decreased in animal models of depression recordings, that monosynaptic GABAergic revealed that a surprisingly high number and is increased in response to treatment and cholinergic transmission occurs of TH‑expressing cells also express GADs

PROGRESS

and might therefore co‑release GABA20. a Presynaptic b c neuron However, careful analyses revealed that low levels of TH mRNA are transcribed in many neurons during development and in adulthood without generating Synaptic detectable amounts of TH protein or vesicle dopamine, and that several populations of TH‑expressing neurons in the cortex, striatum, hypothalamus and midbrain Postsynaptic neuron lack expression of AADC and VMAT2 and the ability to release dopamine20–23. d e Therefore, care should be taken when determining the neurochemical identity of a cell and identifying neuronal candidates for co‑release in the adult brain. This is important in all optogenetic studies, as nonspecific opsin expression in even a small number of off-target cells could yield results that may easily be misconstrued as reflecting 24 Postsynaptic Postsynaptic co‑release from one population of neurons . neuron A neuron B Nevertheless, functional evidence of co‑release of dopamine and GABA has been Transmitter 1 Receptor for Vesicular transporter Common vesicular transporter found in several brain areas. The olfactory transmitter 1 for transmitter 1 for transmitter 1 and 2 bulb and retina contain populations of Transmitter 2 Receptor for Vesicular transporter cells that express GADs, TH, VGAT and transmitter 2 for transmitter 2 VMAT2. Depolarization of these neurons Figure 1 | Distinct cellular and molecular mechanisms of co‑transmission.Nature Reviews There are| Neuroscience several pos‑ results in the release of both transmitters sible mechanisms by which co-transmission of two neurotransmitters might occur, each offering dis‑ from small clear vesicles in a Ca2+-dependent tinct functionality and plasticity. Unfortunately, these mechanisms can be difficult to distinguish fashion25–28. Interestingly, quantal release experimentally using electrophysiological and anatomical approaches. a | The transmitters may be of dopamine and GABA is mostly loaded into individual synaptic vesicles using a common vesicular transporter. The abundance of the asynchronous — the release of dopamine transmitters in the cytosol and the affinity of the transporter for those transmitters will determine the relative content of synaptic vesicles. Such co‑packaging ensures that both transmitters are outlasts GABA exocytosis by several 26,27 released at the same time and location, and that they are subject to similar presynaptic short-term seconds — suggesting that packaging into plasticity. b | The transmitters may be packaged into the same vesicles using distinct vesicular trans‑ largely non-overlapping synaptic vesicles porters. The relative abundance of each co‑transmitter in vesicles varies with cytosolic transmitter occurs (FIG. 1c). availability and with the expression levels of the vesicular transporters. Unless all vesicles express both We recently reported that dopa- transporters, this type of co‑packaging probably occurs together with the co‑packaging described in mine-containing neurons in the VTA part c. c | The transmitters may be packaged into separate vesicles that are found within individual and substantia nigra pars compacta presynaptic boutons. Although both transmitters would be exocytosed from the same presynaptic (SNc) monosynaptically inhibit spiny bouton, their release could be modulated independently over short and long timescales, depending projection neurons (SPNs) in the striatum on transmitter abundances, synaptic vesicle release probability and the rate of synaptic vesicle recy‑ through Ca2+-dependent release of a cling and loading for each transmitter. d | The transmitters may be packaged into separate vesicles 29,30 that distribute to distinct presynaptic release sites. Physical separation would allow the presynaptic GABAA receptor agonist . Dopamine − release of both transmitters to be modulated separately and the targeting of each transmitter to dif‑ does not gate Cl conductances in striatal ferent postsynaptic compartments. For example, glutamate release may target dendritic spines while 29,31 neurons , and several lines of evidence GABA is exocytosed onto dendritic shafts. e | An example of transmitter co‑release that does not result instead suggest that GABA is the released in synaptic co‑transmission is depicted, in which postsynaptic membranes (neuron A and neuron B) transmitter30. However, whereas GABA express receptors for only one of the two transmitters. This mechanism enables presynaptic neurons synthetic enzymes have been observed in a to broadcast a signal that is differentially interpreted by distinct neuronal elements. small fraction of SNc and VTA neurons in the rat32, striatum-projecting dopaminergic neurons in mice do not contain detectable chemicals from cells, and it is the most studies detected GABA associated with levels of GAD65, GAD67 or VGAT mRNA promiscuous of the vesicular neurotrans- VMAT2‑containing synaptic vesicles within or protein23,29,30,33, which raises the question mitter transporters, with over a dozen dopaminergic terminals35. Additional of how SNc and VTA neurons acquire and reported substrates, including methyl- biochemical studies are needed to package GABA for exocytosis. Our studies enedioxymethamphetamine (MDMA), determine the relative affinity of GABA for revealed that VMAT2 is necessary for ethidium and rhodamine34. VMAT2 — or a VMAT2 and whether other proteins also GABAergic transmission from nigrostriatal molecular complex incorporating VMAT2 enable VMAT2‑dependent transport of afferents and restores vesicular release of — may therefore transport GABA into GABA. The absence of such a cofactor in GABA when re‑expressed in GABAergic synaptic vesicles, raising the possibility that neurons containing dopamine and GABA neurons from which VGAT had been GABA and dopamine are co‑packaged in in the retina and olfactory bulb could knocked out29. VMAT2 shows homology SNc and VTA axons (FIG. 1a). In agreement explain why co‑packaging is not observed to a large family of transporters that expel with this possibility, ultrastructural more frequently in these cells26,27.

PROGRESS

Midbrain dopaminergic neurons in several species indicating that there is are widely expressed, and might contribute express several genes typically found in substantial overlap in the expression of to the accumulation of extracellular GABA neurons that release GABA36, including dopaminergic and GABAergic markers in by stimulating the release of GABA from plasma membrane GABA transporters, the developing and mature nervous systems, cortical interneurons or by preventing which are involved in terminating the under both physiological and pathological its reuptake. Second, the experiments synaptic actions of GABA30. In mice, conditions42–45, which collectively suggest that established the dependence of TMN these transporters are also required to that GABA is a ubiquitous co‑transmitter stimulation-induced cortical and striatal sustain GABAergic transmission from SNc in dopaminergic neurons. responses on the expression of VGAT neurons30, suggesting that reuptake is an involved comparisons between two important means of supplying GABA for Histamine–GABA. Neurons of the separate strains of mice, viral vectors exocytosis. This mechanism may allow tuberomammilary nucleus (TMN), which and specificities of opsin expression. It is the degree of GABAergic transmission are the only source of histamine in the therefore difficult to assess whether TMN to vary along dopaminergic axons, vertebrate CNS, exhibit a GABAergic neurons were recruited to a similar extent depending on extracellular GABA levels or phenotype characterized by expression of under both conditions. In addition, the the membrane distribution of the GABA GABA, GAD67 and VGAT46,47. However, small magnitude of the induced GABAergic transporter37. The latter may explain why functional evidence of GABA release currents complicates the assessment of the striatal cholinergic neurons do not receive from these cells has remained elusive. contribution by VMAT2 to these currents GABAergic co‑transmission from SNc Electrical stimulation of TMN neurons as, presumably, VMAT2 does not transport 38 axons , despite expressing GABAA receptors elicits inhibitory postsynaptic potentials GABA as efficiently as VGAT. and receiving dopaminergic innervation39. in supraoptic nucleus neurons that 40 Functions of GABA co‑release A recent study by Kim and colleagues are insensitive to the GABAA receptor supports the finding that mouse midbrain antagonist bicuculline48, and optogenetic GABA is a remarkably versatile neurotrans- dopaminergic neurons do not express activation of TMN neurons fails to evoke mitter: it can exert excitatory, inhibitory GADs23,29,30,33, and it reveals a non-canonical phasic GABAergic responses locally within or trophic influences, it may act either pathway for the synthesis of GABA by TMN, in preoptic nucleus target neurons tonically or phasically, and can bind to aldehyde dehydrogenase 1A1 (ALDH1A1). or in cortical pyramidal neurons, despite either ionotropic or metabotropic receptors Pharmacological inhibition, global genetic evidence of histamine release in those that may be localized pre- and postsyn- deletion, and midbrain knockdown of areas47,49. This may reflect the fact that TMN aptically. This flexibility, combined with ALDH1A1 in mice all reduced GABA axonal varicosities are not typically found the diversity of GABA co‑transmitters, release from SNc axons by half, and the closely apposed to postsynaptic structures46, neural circuit architectures and co‑release administration of GABA reuptake blockers preventing the efficient detection of mechanisms, suggest it is unlikely that abolished the remaining GABA release, co‑released GABA. Alternatively, GABA GABA co‑transmission serves a single, indicating that ALDH1A1 and membrane and histamine may be packaged into universal function. Here, we propose some GABA transporters both contribute to distinct vesicles in TMN neurons50 and interesting possible functions of GABA the cytosolic accumulation of GABA for released onto different targets that remain co‑release in synaptic transmission that we exocytosis40. Interestingly, dopamine release to be identified. hope will stimulate further investigation. is also dependent on both de novo synthesis However, it was recently shown that We note that in addition to the possible roles and membrane reuptake41. In addition, GABA co‑release might participate in discussed below, GABA co‑release might the study by Kim et al.40 reported that the control of arousal and wakefulness by have a role in regulating the neurotransmitter prolonged incubation of acute brain slices neurons of the TMN by ablating VGAT cycle in the developing and diseased CNS3, in ethanol depresses GABA co‑release from these cells in mice47. Recordings from the specification of neurotransmitters44,45, from SNc axons, and that this effect cortical and striatal neurons in slices and the functional establishment and requires ALDH1a1. Furthermore, during from wild-type mice revealed a slow and reorganization of synapses (see REFS 4,51–53). a continuous two-bottle-choice test in sustained increase in holding current on their home cage, mice lacking ALDH1A1 prolonged optical stimulation of TMN Inhibition. A major function of GABAergic consumed more alcohol than wild-type axons (900 light pulses over 3 min) that was signalling is the inhibition of activity in controls, suggesting a functional role for mediated by GABAA receptors. This current target neurons by decreasing exocytosis, GABA co‑release in regulating behavioural was dependent on transmitter release hyperpolarizing membranes and shunting responses to alcohol. from TMN axons, was partially blocked depolarization54–56 (FIG. 2a,b). When Co‑release of GABA may also occur by the inclusion of H1 and H2 histamine paired with excitatory or modulatory in other populations of dopaminergic receptor antagonists and was not observed co‑transmitters, GABA co‑release may neurons in the mammalian CNS, as in slices from mice with genetic deletion provide a faster and more targeted form dopaminergic nuclei A11–A14 all express of VGAT in the TMN, providing the first of inhibition than that resulting from GADs20. Although the presence of VMAT2 functional evidence that GABA may be typical disynaptic mechanisms. For alone should render these cells capable co‑released by TMN neurons. However, instance, in the LHb, a region lacking local of releasing both transmitters, they may two caveats should be noted that prevent interneurons, GABA co‑release may balance also express VGAT. Co‑expression of both more definitive conclusions being drawn glutamatergic excitation from EP11 and vesicular transporters might allow for from this study. First, the authors did not VTA12, and possibly provide a form of gain the presence of separate pools of vesicles exclude a contribution of H3 histamine control to the LHb that in other brain areas enriched with each neurotransmitter. These receptors to the postsynaptic response to arises from feedforward recruitment of observations support a large body of work TMN stimulation; H3 histamine receptors fast-spiking interneurons.

PROGRESS

a b c Presynaptic terminal * * *

VGKC K+ 2 mV

20 ms ∆V m Rin V > E m Cl Vm > ECl Ca2+ 10 mV Metabotropic Ionotropic GABA receptor GABA receptor 20 ms GABA NMDAR Cl–

VGCC 2 mV 5 mV

Rin 20 ms 20 ms ∆Vm

Postsynaptic neuron Vm = ECl Vm < ECl

GABA only Glutamate only GABA and glutamate Stimulation of presynaptic inputs

d e Baseline

Excitatory GABA Hist axon Single stimulus Inhibitory interneuron Basal TMN forebrain axon axon ACh GABA Pyramidal GABA Co-released modulator neuron Burst stimulus

Stimulation of the excitatory axon GABA Co-released modulator 0.5 s Figure 2 | Functions of GABA co‑release. a | Once released in the synaptic depicts fluctuations in membrane potential upon stimulation of a presynap‑ Nature Reviews | Neuroscience cleft, GABA can signal pre- and postsynaptically, homo- and heterosynapti‑ tic GABAergic synapse (asterisk denotes truncated action potential).

cally and via ionotropic type A GABA (GABAA) receptors and GABAC recep‑ d | Co‑transmission can be directed in space to achieve synergistic effects

tors and metabotropic GABAB receptors. Gating of GABAA and GABAC in complex neuronal circuits. Three cortical microcircuits are shown under

receptors shunts synaptic conductances by decreasing input resistance (Rin). different conditions, each composed of a pyramidal neuron, a local inhibi‑ − Net flow of Cl through these receptors alters the membrane potential (Vm), tory interneuron and an excitatory afferent. Under baseline conditions (left), which affects Ca2+ influx through voltage-gated Ca2+ channels (VGCCs) and excitatory afferent stimulation evokes three spikes in the pyramidal cell.

NMDA receptors (NMDARs). GABAB receptors act to inhibit presynaptic Target-specific co‑release of GABA at some synapses, but not others, ena‑ exocytosis and postsynaptic depolarization through modulation of voltage bles basal forebrain neurons to enhance (middle) and enables tuberomam‑ gated potassium channels (VGKCs), VGCCs and the presynaptic release milary nucleus (TMN) neurons to dampen (right) the output of pyramidal − machinery. b | Depending on the Cl reversal potential (ECl) relative to the cells. e | GABA co‑release provides a temporally precise signal. Schematic

resting Vm of a neuron, GABA co‑release may dampen cellular excitability by representation of the spontaneous action potential discharge of a target cell physically hyperpolarizing postsynaptic membranes (top) or by shunting under baseline conditions (top) or upon stimulation of afferents that co‑ synaptic depolarizations (bottom). To illustrate this, the postsynaptic poten‑ release GABA along with a neuromodulator that increases membrane excit‑ tials evoked by presynaptic release of glutamate only, GABA only and co‑ ability by activating G protein-coupled receptors (middle and bottom). release of glutamate and GABA are depicted. c | GABA release can exert a Phasic inhibition of action potential firing by GABA co‑release (the phasic depolarizing influence if the postsynaptic cell expresses hyperpolarization- inhibitory influence is shown in red) allows the postsynaptic cell to readily activated channels that depolarize membranes (top) or if intracellular Cl− is distinguish between single and burst stimulation of presynaptic afferents.

elevated, such that ECl lies above the resting Vm (bottom). Each example ACh, acetylcholine; Hist, histamine.

What function might the co‑release of two accelerated by GABA57. Co‑transmission been identified in mammals, co‑release of inhibitory transmitters, such as GABA and of two inhibitory neurotransmitters may GABA also confers on glycinergic synapses glycine (which both gate Cl− conductances) therefore allow fine-tuning of the magnitude the ability to modulate transmission in a serve? Glycinergic currents are considerably and duration of synaptic inhibition. Because homosynaptic and heterosynaptic manner, 58 shorter than GABAergic currents and are metabotropic receptors for glycine have not via metabotropic GABAB receptors .

PROGRESS

Providing a depolarizing influence. GABA spines that receive glutamatergic inputs60; Conclusions can evoke action potentials in target because dopamine signals in a diffuse It has long been appreciated that individual neurons if intracellular Cl− is elevated or manner, GABA co‑release may specifically neurons have the capacity to express and if the target neurons express ion channels shunt excitatory potentials within release multiple small-molecule, peptide that promote rebound excitation (FIG. 2c). individual spines. and gaseous transmitters1,2. However, the In mature neurons, the Cl− equilibrium notion that the primary synaptic actions potential (ECl) is typically below the Providing a temporally precise signal. of many neurons in the adult CNS are not spike threshold, precluding GABA from Monoamines signal almost exclusively limited to the release of a single classical directly evoking firing. However, it is through GPCRs. Their synaptic effects transmitter has only recently been widely 3,4 not uncommon for ECl to lie above the are therefore achieved over hundreds recognized . Rather than representing resting membrane potential of a neuron. of milliseconds to minutes, notably an evolutionary oversight, it is likely that

In such cases, activation of GABAA slower than those of transmitters acting co‑release of multiple classical transmitters receptors would cause a subthreshold on ionotropic receptors. Therefore in has emerged because of its ability to provide membrane depolarization, with important monoaminergic neurons, GABA co‑release additional means of modulating and consequences for somatic excitability may provide the means to rapidly fine-tuning synaptic transmission between and the induction of synaptic plasticity53. convey signals time-locked to sensory individual synaptic partners over many Co‑release of GABA along with excitatory or motor events (FIG. 2e). SNc and VTA different timescales. As discussed in this or modulatory transmitters may therefore neurons fire tonically and modulate their article, there is now compelling functional confer flexibility on synaptic partners to discharge in response to salient stimuli by evidence that GABAergic transmission can fine-tune the membrane potential of target adding or omitting a few spikes. GABA occur from neurons that also release glycine, cells so as to optimize the ionotropic or co‑release might have an important role glutamate, ACh and monoamines in the metabotropic actions of co‑transmitters. in communicating these phasic changes in adult CNS. Many of the functions of GABA firing to striatal neurons. Interestingly, co‑release remain to be determined, but Providing a spatially targeted signal. GABA worms and insects have evolved Cl− the diversity and flexibility of GABAergic co‑transmission, where it does occur, is not channels gated by biogenic amines in signalling are likely to factor greatly in the necessarily uniform across all synapses, addition to metabotropic receptors61–63, prevalence of GABA as a co‑transmitter. either because presynaptic neurons restrict suggesting an important role for coincident Nicolas X. Tritsch was previously at the Howard GABA release to particular sites, or fast inhibitory and slow modulatory Hughes Medical Institute, Department of because some postsynaptic targets do not signalling by aminergic cells. Neurobiology, Harvard Medical School, Boston, express GABA receptors. Thus, rather than Massachusetts 02115, USA. broadcasting the same information to all Regulation of synaptic plasticity. Present address: New York University Neuroscience of the postsynaptic targets of a presynaptic By directly and indirectly influencing Institute, Department of Neuroscience and Physiology, New York University Langone Medical Center, 2+ (FIG. 2a) neuron, GABA co‑transmission may serve axonal and dendritic Ca influx , New York, New York 10016, USA. several synergistic functions in complex GABAergic signalling plays an important circuits (FIG. 2d). For example, the fact role in the induction of short- and Adam J. Granger and Bernardo L. Sabatini are at the Howard Hughes Medical Institute, Department of that some terminals of individual SACs long-term synaptic plasticity at GABAergic Neurobiology, Harvard Medical School, Boston, 54,55 predominantly release ACh, whereas others and non-GABAergic synapses . Like Massachusetts 02115, USA. within the same cell release both ACh any mild depolarization, subthreshold Correspondence to B.L.S. and GABA, enables downstream DSGCs GABAA-mediated depolarization might [email protected] 2+ to encode both motion sensitivity and favour Ca influx and synaptic plasticity doi:10.1038/nrn.2015.21 15 direction selectivity . As another example, by reducing the efficacy of NMDA-type Published online 11 Feb 2016 2+ in the cortex, coincident ACh-mediated glutamate receptor Mg blockade and 1. Burnstock, G. Cotransmission. Curr. Opin. Pharmacol. excitation of layer I disinhibitory increasing the opening probability of 4, 47–52 (2004). 2+ 64–66 2. Nusbaum, M. P., Blitz, D. M., Swensen, A. M., interneurons and GABA-mediated voltage-gated Ca channels (VGCCs) , Wood, D. & Marder, E. The roles of co‑transmission in inhibition of deeper layer interneurons while simultaneously dampening excitatory neural network modulation. Trends Neurosci. 24, 146–154 (2001). may account for the role of basal potentials via shunting inhibition. 3. Hnasko, T. S. & Edwards, R. H. Neurotransmitter forebrain neurons in arousal59. Similarly, These mechanisms would be predicted corelease: mechanism and physiological role. Annu. Rev. Physiol. 74, 225–243 (2011). TMN neurons may increase cortical to facilitate the induction of long-term 4. Vaaga, C. E., Borisovska, M. & Westbrook, G. L. inhibitory synaptic transmission via the potentiation, although such an effect Dual‑transmitter neurons: functional implications of co‑release and co‑transmission. Curr. Opin. Neurobiol. release of histamine, while simultaneously not yet been demonstrated for GABA. 29, 25–32 (2014). 5. Wojcik, S. M. et al. A shared vesicular carrier allows providing slow and sustained GABAergic Conversely, activation of GABAB receptors, 47 synaptic corelease of GABA and glycine. Neuron 50, inhibition to pyramidal neurons . often localized at excitatory postsynaptic 575–587 (2006). GABAergic co‑transmission might be terminals, can reduce Ca2+ influx through 6. Mathews, G. C. & Diamond, J. S. Neuronal glutamate uptake contributes to GABA synthesis and inhibitory further restricted to distinct subcellular NMDA receptors and VGCCs by a variety synaptic strength. J. Neurosci. 23, 2040–2048 regions of a target neuron, enabling fine of mechanisms54,67–69, which would limit (2003). 7. Wang, L., Tu, P., Bonet, L., Aubrey, K. R. & spatial control of presynaptic release the induction of synaptic plasticity. Supplisson, S. Cytosolic transmitter concentration and postsynaptic activity. This may be Thus, GABA co‑release could be a regulates vesicle cycling at hippocampal GABAergic terminals. Neuron 80, 143–158 particularly important for co‑transmitter complementary signal to the metabotropic (2013). systems that signal by volume transmission. actions of transmitters such as ACh, 8. Apostolides, P. F. & Trussell, L. O. Rapid, activity- independent turnover of vesicular transmitter content Midbrain dopaminergic neurons, for dopamine and histamine in modulating the at a mixed glycine/GABA synapse. J. Neurosci. 33, example, form synapses on dendritic plasticity of synapses and neuronal circuits. 4768–4781 (2013).

PROGRESS

9. Awatramani, G. B., Turecek, R. & Trussell, L. O. 30. Tritsch, N. X., Oh, W. J., Gu, C. & Sabatini, B. L. 51. Owens, D. F. & Kriegstein, A. R. Is there more to Staggered development of GABAergic and glycinergic Midbrain dopamine neurons sustain inhibitory GABA than synaptic inhibition? Nat. Rev. Neurosci. 3, transmission in the MNTB. J. Neurophysiol. 93, transmission using plasma membrane uptake of 715–727 (2002). 819–828 (2005). GABA, not synthesis. eLife 3, e01936 (2014). 52. Le Magueresse, C. & Monyer, H. GABAergic 10. Dugue, G. P., Dumoulin, A., Triller, A. & Dieudonne, S. 31. Kress, G. J. et al. Fast phasic release properties of interneurons shape the functional maturation of the Target-dependent use of co‑released inhibitory dopamine studied with a channel biosensor. cortex. Neuron 77, 388–405 (2013). transmitters at central synapses. J. Neurosci. 25, J. Neurosci. 34, 11792–11802 (2014). 53. Kesner, P., Gobert, D. & Ruthazer, E. S. Formula for 6490–6498 (2005). 32. Hedou, G. et al. Immunohistochemical studies of the unsilencing plasticity: spike with GABA. Neuron 87, 11. Shabel, S. J., Proulx, C. D., Piriz, J. & Malinow, R. localization of neurons containing the enzyme that 915–917 (2015).

Mood regulation. GABA/glutamate co‑release controls synthesizes dopamine, GABA, or γ-hydroxybutyrate in 54. Chalifoux, J. R. & Carter, A. G. GABAB receptor habenula output and is modified by antidepressant the rat substantia nigra and striatum. J. Comp. modulation of synaptic function. Curr. Opin. Neurobiol. treatment. Science 345, 1494–1498 (2014). Neurol. 426, 549–560 (2000). 21, 339–344 (2011). 12. Root, D. H. et al. Single rodent mesohabenular axons 33. Poulin, J. F. et al. Defining midbrain dopaminergic 55. Higley, M. J. Localized GABAergic inhibition of release glutamate and GABA. Nat. Neurosci. 17, neuron diversity by single-cell gene expression dendritic Ca2+ signalling. Nat. Rev. Neurosci. 15, 1543–1551 (2014). profiling. Cell Rep. 9, 930–943 (2014). 567–572 (2014). 13. Proulx, C. D., Hikosaka, O. & Malinow, R. Reward 34. Yelin, R. & Schuldiner, S. The pharmacological 56. Isaacson, J. S. & Scanziani, M. How inhibition shapes processing by the lateral habenula in normal and profile of the vesicular monoamine transporter cortical activity. Neuron 72, 231–243 (2011). depressive behaviors. Nat. Neurosci. 17, 1146–1152 resembles that of multidrug transporters. FEBS Lett. 57. Lu, T., Rubio, M. E. & Trussell, L. O. Glycinergic (2014). 377, 201–207 (1995). transmission shaped by the corelease of GABA in a 14. Granger, A. J., Mulder, N., Saunders, A. & 35. Stensrud, M. J., Puchades, M. & Gundersen, V. GABA is mammalian auditory synapse. Neuron 57, 524–535 Sabatini, B. L. Cotransmission of acetylcholine and localized in dopaminergic synaptic vesicles in the rodent (2008). GABA. Neuropharmacology 100, 40–46 (2016). striatum. Brain Struct. Funct. 219, 1901–1912 (2013). 58. Magnusson, A. K., Park, T. J., Pecka, M., Grothe, B. & 15. Lee, S., Kim, K. & Zhou, Z. J. Role of ACh–GABA 36. Karayannis, T. et al. Cntnap4 differentially contributes Koch, U. Retrograde GABA signaling adjusts sound cotransmission in detecting image motion and motion to GABAergic and dopaminergic synaptic transmission. localization by balancing excitation and inhibition in direction. Neuron 68, 1159–1172 (2010). Nature 511, 236–240 (2014). the brainstem. Neuron 59, 125–137 (2008). 16. Wei, W., Hamby, A. M., Zhou, K. & Feller, M. B. 37. Scimemi, A. Structure, function, and plasticity of 59. Fu, Y. et al. A cortical circuit for gain control by Development of asymmetric inhibition underlying GABA transporters. Front. Cell. Neurosci. 8, 161 behavioral state. Cell 156, 1139–1152 (2014). direction selectivity in the retina. Nature 469, (2014). 60. Moss, J. & Bolam, J. P. in Dopamine Handbook 402–406 (2011). 38. Straub, C., Tritsch, N. X., Hagan, N. A., Gu, C. & (ed. Iversen, L. L.) 49–60 (Oxford University Press, 17. Morrie, R. D. & Feller, M. B. An asymmetric increase in Sabatini, B. L. Multiphasic modulation of cholinergic 2010). inhibitory synapse number underlies the development interneurons by nigrostriatal afferents. J. Neurosci. 61. Zheng, Y. et al. Identification of two novel Drosophila of a direction selective circuit in the retina. J. Neurosci. 34, 8557–8569 (2014). melanogaster histamine-gated chloride channel 35, 9281–9286 (2015). 39. Guo, Q. et al. Whole-brain mapping of inputs to subunits expressed in the eye. J. Biol. Chem. 277, 18. Saunders, A., Granger, A. J. & Sabatini, B. L. projection neurons and cholinergic interneurons in the 2000–2005 (2002). Corelease of acetylcholine and GABA from cholinergic dorsal striatum. PLoS ONE 10, e0123381 (2015). 62. Gengs, C. et al. The target of Drosophila forebrain neurons. eLife 4, e06412 (2015). 40. Kim, J. I. et al. Aldehyde dehydrogenase 1a1 mediates photoreceptor synaptic transmission is a histamine- 19. Saunders, A. et al. A direct GABAergic output from the a GABA synthesis pathway in midbrain dopaminergic gated chloride channel encoded by ort (hclA). J. Biol. basal ganglia to frontal cortex. Nature 521, 85–89 neurons. Science 350, 102–106 (2015). Chem. 277, 42113–42120 (2002). (2015). 41. Gainetdinov, R. R. & Caron, M. G. Monoamine 63. Ringstad, N., Abe, N. & Horvitz, H. R. Ligand-gated 20. Bjorklund, A. & Dunnett, S. B. Dopamine neuron transporters: from genes to behavior. Annu. Rev. chloride channels are receptors for biogenic amines in systems in the brain: an update. Trends Neurosci. 30, Pharmacol. Toxicol. 43, 261–284 (2003). C. elegans. Science 325, 96–100 (2009). 194–202 (2007). 42. Barreiro-Iglesias, A., Villar-Cervino, V., Anadon, R. & 64. Jahr, C. E. & Stevens, C. F. Voltage dependence of 21. Li, X., Qi, J., Yamaguchi, T., Wang, H. L. & Morales, M. Rodicio, M. C. Dopamine and γ-aminobutyric acid are NMDA-activated macroscopic conductances predicted Heterogeneous composition of dopamine neurons colocalized in restricted groups of neurons in the sea by single-channel kinetics. J. Neurosci. 10, 3178–3182 of the rat A10 region: molecular evidence for lamprey brain: insights into the early evolution of (1990). diverse signaling properties. Brain Struct. Funct. 218, neurotransmitter colocalization in vertebrates. J. Anat. 65. Magee, J. C. & Johnston, D. Synaptic activation 1159–1176 (2013). 215, 601–610 (2009). of voltage-gated channels in the dendrites of 22. Xenias, H. S., Ibanez-Sandoval, O., Koos, T. & 43. Svensson, E., Proekt, A., Jing, J. & Weiss, K. R. hippocampal pyramidal neurons. Science 268, Tepper, J. M. Are striatal tyrosine hydroxylase PKC-mediated GABAergic enhancement of 301–304 (1995). interneurons dopaminergic? J. Neurosci. 35, dopaminergic responses: implication for short-term 66. Carter, A. G. & Sabatini, B. L. State-dependent 6584–6599 (2015). potentiation at a dual-transmitter synapse. calcium signaling in dendritic spines of striatal 23. Lammel, S. et al. Diversity of transgenic mouse models J. Neurophysiol. 112, 22–29 (2014). medium spiny neurons. Neuron 44, 483–493 for selective targeting of midbrain dopamine neurons. 44. Velazquez-Ulloa, N. A., Spitzer, N. C. & Dulcis, D. (2004). Neuron 85, 429–438 (2015). Contexts for dopamine specification by calcium spike 67. Sabatini, B. L. & Svoboda, K. Analysis of calcium 24. Pupe, S. & Wallen-Mackenzie, A. Cre-driven activity in the CNS. J. Neurosci. 31, 78–88 (2011). channels in single spines using optical fluctuation optogenetics in the heterogeneous genetic panorama 45. Dulcis, D., Jamshidi, P., Leutgeb, S. & Spitzer, N. C. analysis. Nature 408, 589–593 (2000). of the VTA. Trends Neurosci. 38, 375–386 (2015). Neurotransmitter switching in the adult brain 68. Skeberdis, V. A. et al. Protein kinase A regulates 25. Maher, B. J. & Westbrook, G. L. Co‑transmission regulates behavior. Science 340, 449–453 (2013). calcium permeability of NMDA receptors. of dopamine and GABA in periglomerular cells. 46. Haas, H. L., Sergeeva, O. A. & Selbach, O. Histamine Nat. Neurosci. 9, 501–510 (2006). J. Neurophysiol. 99, 1559–1564 (2008). in the nervous system. Physiol. Rev. 88, 1183–1241 69. Lur, G. & Higley, M. J. Glutamate receptor modulation 26. Borisovska, M., Bensen, A. L., Chong, G. (2008). is restricted to synaptic microdomains. Cell Rep. 12, & Westbrook, G. L. Distinct modes of dopamine 47. Yu, X. et al. Wakefulness is governed by GABA and 326–334 (2015). and GABA release in a dual transmitter neuron. histamine cotransmission. Neuron 87, 164–178 70. Strata, P. & Harvey, R. Dale’s principle. Brain Res. Bull. J. Neurosci. 33, 1790–1796 (2013). (2015). 50, 349–350 (1999). 27. Hirasawa, H., Betensky, R. A. & Raviola, E. Corelease 48. Hatton, G. I. & Yang, Q. Z. Ionotropic histamine 71. Zhang, S. et al. Dopaminergic and glutamatergic of dopamine and GABA by a retinal dopaminergic receptors and H2 receptors modulate supraoptic microdomains in a subset of rodent mesoaccumbens neuron. J. Neurosci. 32, 13281–13291 (2012). oxytocin neuronal excitability and dye coupling. axons. Nat. Neurosci. 18, 386–392 (2015). 28. Liu, S., Plachez, C., Shao, Z., Puche, A. & Shipley, M. T. J. Neurosci. 21, 2974–2982 (2001). Olfactory bulb short axon cell release of GABA and 49. Williams, R. H. et al. Optogenetic-mediated release Acknowledgements dopamine produces a temporally biphasic inhibition- of histamine reveals distal and autoregulatory The authors would like to acknowledge members of the excitation response in external tufted cells. mechanisms for controlling arousal. J. Neurosci. 34, Sabatini laboratory for insightful discussions on this and other J. Neurosci. 33, 2916–2926 (2013). 6023–6029 (2014). topics and for fostering a learned and collegial research 29. Tritsch, N. X., Ding, J. B. & Sabatini, B. L. 50. Kukko-Lukjanov, T. K. & Panula, P. Subcellular environment, which greatly helped to shape this manuscript. Dopaminergic neurons inhibit striatal output through distribution of histamine, GABA and galanin in non-canonical release of GABA. Nature 490, 262–266 tuberomamillary neurons in vitro. J. Chem. Neuroanat. Competing interests statement (2012). 25, 279–292 (2003). The authors declare no competing interests.