Parvalbumin-immunoreactive (PV-IR) fast-spiking interneurons

One major category of cortical inhibitory interneuron can be identified by the expression of the calcium binding protein parvalbumin (PV). There are several subclasses of cortical inhibitory interneuron that express PV, which at least include a subset of basket cells and chandelier cells (for an excellent review, see KubotaReference Kubota²⁵). PV-IR interneurons primarily have fast-spiking behavior, and thus for the purposes of this article, the terms “PV-IR interneuron” and “fast-spiking interneuron” will be used interchangeably. Importantly, each PV-IR interneuron subtype is thought to have a selective subcellular axonal target. PV-IR basket cells, which comprise approximately 40–50% of cortical non-pyramidal neurons, arborize at the soma and dendrites of pyramidal neurons, with each basket cell innervating between 200 and 1000 pyramidal cells.Reference Kubota²⁵ PV-IR chandelier cells are a relatively small subset of PV-IR interneurons and synapse with pyramidal neurons almost exclusively at the axon hillock.Reference Kubota²⁵ Given the placement of these axonal arborizations, cortical PV-IR interneurons can be expected to have a strong inhibitory effect on the output of pyramidal neurons. Interestingly, PV-IR interneuron subclasses such as fast-spiking basket cells are thought to be connected to one another via electrical gap junctions,Reference Kubota²⁵ which may allow these cell types to easily form neural assemblies that could multiply the inhibitory signal they represent within the frontal cortex network.

Non-PV-IR interneurons

In addition to PV-IR interneurons, there are a number of other GABAergic interneuron classes in the frontal cortex. These subtypes can be identified by various neurochemical markers, including calbindin, calretinin, cholecystokinin, vasoactive intestinal peptide, somatostatin, and neuropeptide Y. Each interneuron subtype has a distinctive morphology, cellular target, and role in modulating the activity of frontal cortical networks.Reference Kubota²⁵ Furthermore, the electrical activity of these cells can have a regular-spiking, late-spiking, or bursting character. However, information on the placement of serotonergic targets on these cortical interneuron subtypes has generally not reached a high level of detail. Therefore, we have chosen to consider these various classes together as one group for the purposes of this article. The most important group of non-PV-IR interneurons to consider from the perspective of understanding serotonergic neurotransmission’s effects on PFC activity is a set of interneurons in the shallow cortical layers (ie, layers I–III) that inhibits cortical pyramidal neuron activity.

Pyramidal neurons

Pyramidal neurons are excitatory glutamatergic cells that are characterized by a triangular-shaped soma, apical dendrites emanating from the soma’s vertex, and basal dendrites as well as a single branching axon originating from the soma’s base. Within the cortex, pyramidal neurons are thought to fill multiple functional roles. The small pyramidal neurons prevalent in layers II and III integrate subcortical and cortico-cortical afferent inputs and send axonal projections horizontally to layers II and III of neighboring cortical columns.Reference Adesnik and Scanziani^26, Reference Gottlieb and Keller²⁷ Additionally, layer II/III pyramidal neurons project axons vertically to layer V pyramidal neurons.Reference Adesnik and Scanziani^26, Reference Gottlieb and Keller²⁷ This combination of local projections drives lateral inhibitory effects on layer II/III pyramidal neurons from neighboring cortical columns, presumably by activating inhibitory interneurons, while also activating layer V pyramidal neurons both within the originating column and in neighboring columns.Reference Adesnik and Scanziani²⁶ Thus, activation of a given layer II/III pyramidal neuron may suppress inputs from neighboring columns while at the same time magnifying the output of its own column.

The pyramidal neurons of the deep cortical layers (ie, layers V and VI) have a large soma in comparison to layer II/III pyramidal neurons, and layer V pyramidal neurons are responsible for sending excitatory glutamatergic projections to subcortical regions, such as the striatum. Thus, the pyramidal neuron is considered the principle cortical projection neuron.

In order to understand how 5-HT modulates PFC microcircuit activity, it is necessary to understand which cells within the network express each receptor and their effect on neural activity. Therefore, the following section will review information on 5-HT receptor expression within the PFC.

5-HT_1A receptors

5-HT_1A receptors are inhibitory G-protein coupled receptors. Autoradiographic and immunohistochemical studies have demonstrated strong 5-HT_1A receptor expression in some portions of the hippocampal formation (see the detailed review by Dale et al Reference Dale, Pehrson and Jeyarajah²⁸ elsewhere in this issue) and the brain stem raphe nuclei. 5-HT_1A receptor expression within the PFC is moderate in strength (Figure 1, panels A and B), and has been demonstrated in all cortical layers with the exception of layer I.Reference Aznar, Qian, Shah, Rahbek and Knudsen²⁹

Aznar et al Reference Aznar, Qian, Shah, Rahbek and Knudsen²⁹ investigated the cellular subtypes that express the 5-HT_1A receptor in the rodent cortex, and found that the majority of pyramidal neurons, PV-IR interneurons, and some non-PV-IR interneurons express 5-HT_1A receptors (approximately 85% or more in each case). However, it should be noted that other research groups have suggested lower estimates in PFC subregions, with approximately 60% of pyramidal neurons in the prelimbic, and 40% of those in the infralimbic subregion, expressing 5-HT_1A receptor messenger ribonucleic acid (mRNA).Reference Celada, Puig and Artigas³⁰ Approximately 20% of GABAergic interneurons in these PFC subregions express 5-HT_1A receptors.Reference Celada, Puig and Artigas³⁰

On a subcellular level, 5-HT_1A receptors are thought to be somatodendritically expressed, and in frontal cortex pyramidal neurons there is evidence supporting a placement on the axon hillock.Reference Cruz, Eggan, Azmitia and Lewis³¹

Based on this pattern of expression, 5-HT_1A receptors can be expected to have complex effects on the behavior of the frontal cortex. Activation of 5-HT_1A receptors expressed on PV-IR interneurons would inhibit these cells, and thereby allow pyramidal neuron firing to increase via disinhibition. However, 5-HT_1A receptor expression on pyramidal neurons themselves would reduce their activation state. Therefore, 5-HT_1A receptor activation should have mixed effects on the behavior of the PFC microcircuit. Moreover, electrophysiological evidence suggesting that selective activation of 5-HT_1A receptors leads to an increase in firing followed by a reduction for most pyramidal neuronsReference Llado-Pelfort, Santana, Ghisi, Artigas and Celada³² is consistent with this conceptual framework.

5-HT_1B receptors

Within the frontal cortex, 5-HT_1B receptors are expressed at low to moderate levelsReference Bruinvels, Palacios and Hoyer^33–Reference Sari, Miquel and Brisorgueil³⁵ (Figure 1, panels C and D) and may exist with a combination of presynaptic and postsynaptic expression patterns. The observation that 5-HT_1B receptor stimulation reduces extracellular 5-HT concentrations elicited by SSRIs in the frontal cortex suggests the presence of 5-HT_1B receptors on serotonergic terminals in this region.Reference de Groote, Klompmakers, Olivier and Westenberg³⁶ Additionally, Tanaka and NorthReference Tanaka and North³⁷ have suggested that 5-HT_1B receptors are present as postsynaptic terminal heteroreceptors on cortical glutamate terminals, as 5-HT_1B receptors modulate cortical glutamate release. However, this should be viewed with caution, since there is no histological confirmation. Egeland et al Reference Egeland, Warner-Schmidt, Greengard and Svenningsson³⁸ demonstrated 5-HT_1B receptor immunoreactivity within a subpopulation of PV-IR interneurons of the cingulate gyrus, and this included some weak immunoreactivity in the soma of these cells. It is not known whether this expression pattern reflects somatodendritic insertion of 5-HT_1B receptors on the neuronal membrane, as has been observed in the hippocampal formation,Reference Peddie, Davies, Colyer, Stewart and Rodriguez^39, Reference Peddie, Davies, Colyer, Stewart and Rodriguez⁴⁰ or simply 5-HT_1B receptors that are early in the process of being trafficked to axon terminals.

Thus, activation of 5-HT_1B receptors may negatively modulate the release of 5-HT and glutamate within the frontal cortex via their expression as terminal autoreceptors or heteroreceptors. Additionally, if 5-HT_1B receptors have a somatodendritic expression in cortical PV-IR interneurons, then activation of these receptors may also serve to reduce the activity of these neurons. However, we have not been able to identify any electrophysiological studies confirming the presence of a 5-HT_1B mediated current in cortical PV-IR interneurons.

Given that Bruinvels et al Reference Bruinvels, Palacios and Hoyer³³ found that the closely related 5-HT_1D receptor represents a negligible amount of binding in the frontal cortex, we will consider it to be absent in the PFC and will discuss it in more detail below, in the section dealing with the striatum.

5-HT_2A/2C receptors

5-HT₂ receptors are stimulatory G-protein coupled receptors that exist primarily as postsynaptic heteroreceptors within the CNS. 5-HT_2A receptors are strongly expressed in the cortex. Within the frontal cortex, 5-HT_2A receptor mRNA has been observed in pyramidal neurons and interneurons, with substantial variation in the proportion of these cellular populations expressing this receptor. In the prelimbic PFC, approximately one-half of pyramidal neurons and one-third of interneurons express 5-HT_2A receptor mRNA, while in the infralimbic region only about 12% of pyramidal neurons and 22% of interneurons expressed 5-HT_2A receptors.Reference Celada, Puig and Artigas³⁰ Moreover, the presence of 5-HT_2A receptor proteins has been confirmed in each of these cell populations.Reference Cornea-Hebert, Riad, Wu, Singh and Descarries⁴¹ Furthermore, the presence of 5-HT_2A receptors on GABAergic interneurons has been extended, showing these receptors are present in cortical PV-IR and non-PV-IR interneurons.Reference Puig, Watakabe, Ushimaru, Yamamori and Kawaguchi^42, Reference Weber and Andrade⁴³

The structurally similar 5-HT_2C receptor is present at moderate levels in the frontal cortex,Reference Abramowski, Rigo, Duc, Hoyer and Staufenbiel⁴⁴ but has not been found in fast spiking interneurons.Reference Puig, Watakabe, Ushimaru, Yamamori and Kawaguchi⁴²

Thus, within the frontal cortex, activation of 5-HT_2A and 5-HT_2C receptors is expected to have excitatory direct effects on pyramidal neurons, while 5-HT_2A receptors should also excite GABAergic interneurons. However, given the tight interconnection between pyramidal and GABAergic interneurons in the frontal cortical microcircuitry, the effects of 5-HT_2A/2C receptor stimulation are expected to be mixed. Moreover, the available electrophysiological evidence is consistent with this interpretation. In vivo recordings of the PFC after administration of the 5-HT_2A/2C receptor agonist 2,5-dimethoxy-4-iodoamphetamine (DOI) found that regular spiking cells had a mix of excitatory and inhibitory responses,Reference Puig, Celada, Diaz-Mataix and Artigas⁴⁵ although the largest group of DOI-responsive neurons exhibited an excitatory response.

5-HT₃ receptors

5-HT₃ receptors are unique among the known 5-HT receptor subtypes in that they are the only stimulatory ligand-gated ion channel. On a regional level, 5-HT₃ receptors are expressed at their strongest levels within the CNS in the nucleus of the solitary tract, and their level of expression in forebrain regions tends to be low to moderate by comparison.Reference Gehlert, Gackenheimer, Wong and Robertson⁴⁶ Within the frontal cortex, cells expressing 5-HT₃ receptor mRNA are segregated primarily into shallow cortical layers (ie, layers I–III), although some 5-HT₃ receptor mRNA-positive cells have been observed in deeper layers.Reference Puig, Santana, Celada, Mengod and Artigas⁴⁷ 5-HT₃ receptor protein expression in the frontal cortex tends to be present throughout the cortical layers, though again it is stronger in the shallow layersReference Gehlert, Gackenheimer, Wong and Robertson^46, Reference Morales, Battenberg and Bloom⁴⁸ (see Figure 1, panels E and F). On the level of cellular expression, 5-HT₃ receptors are nearly exclusively expressed within GABAergic interneurons, more specifically in those expressing cholecystokinin, calretinin, vasoactive intestinal peptide, or neuropeptide Y, but not in those expressing somatostatin or parvalbumin.Reference Puig and Gulledge^24, Reference Morales and Bloom⁴⁹

Thus, 5-HT₃ receptors represent a mechanism of 5-HT-mediated fast excitatory drive that is selective to non-PV-IR interneurons segregated into layers I–III. By extension, it can be expected that 5-HT₃ receptor-mediated 5-HT neurotransmission will have an overall inhibitory effect on the output of cortical pyramidal cells, especially in the PFC, where there are somewhat more 5-HT₃ receptors than in other cortical subregions. Electrophysiological data are consistent with this interpretation. Puig et al Reference Puig, Santana, Celada, Mengod and Artigas⁴⁷ demonstrated that raphe stimulation increased the activity of a subset of GABAergic interneuron that could be blocked by selective 5-HT₃ receptor antagonists, while Ashby et al Reference Ashby, Minabe, Edwards and Wang⁵⁰ found that 5-HT₃ receptor agonists reduced the activity of putative pyramidal neurons in the PFC. By extension, 5-HT₃ receptor antagonists can be expected to increase pyramidal neuron output via disinhibition.

5-HT₄ receptors

5-HT₄ receptors are stimulatory G-protein coupled receptors for which there is relatively little accumulated knowledge. Autoradiographic evidence suggests that 5-HT₄ receptors are expressed at very low levels in the frontal cortex.Reference Vilaro, Cortes and Mengod⁵¹ Furthermore, there is little and conflicting experimental data addressing their cellular expression in the frontal cortex.Reference Egeland, Warner-Schmidt, Greengard and Svenningsson^38, Reference Peñas-Cazorla and Vilaró⁵²

5-HT₅ receptors

5-HT₅ receptors are inhibitory G-protein coupled receptors that are broadly present within the rodent brain. Immunohistochemical evidence shows low to moderate 5-HT₅ receptor expression levels within several cortical regions, highest within layers II and IV, where expression tended to be in the soma,Reference Oliver, Kinsey, Wainwright and Sirinathsinghji⁵³ but the degree to which this information will translate to the PFC is questionable, since this region lacks a well-defined layer IV. Anatomical data on the cortical neuronal subtypes expressing 5-HT₅ receptors is largely absent; however, Goodfellow et al Reference Goodfellow, Bailey and Lambe⁵⁴ observed a 5-HT₅ receptor-mediated inhibitory current in cortical layer V pyramidal neurons, suggesting that they are at least present in cortical principle cells. Based on these data alone, activating 5-HT₅ receptors should be expected to reduce cortical output. However, given the relative lack of anatomical and electrophysiological data on this receptor subtype, 5-HT₅ receptors may turn out to have more complex actions on the PFC network.

5-HT₆ receptors

5-HT₆ receptors are stimulatory G-protein coupled receptors that are present at low to moderate levels within the rodent frontal cortex, with stronger areas of expression in rostral compared to caudal cortical regions.Reference Gerard, Martres and Lefevre⁵⁵ Similar results were found in human PFC, where 5-HT₆ receptor expression was present at relatively low levels.Reference Marazziti, Baroni and Pirone⁵⁶ The strongest 5-HT₆ receptor expression in the human cortex was observed in layer I, where 5-HT₆ receptors were primarily associated with glia. However, 5-HT₆ receptor expression was associated with pyramidal neurons in deeper cortical layers,Reference Marazziti, Baroni and Pirone⁵⁶ and expression in non-pyramidal neuronal types has not yet been reported. These data may suggest that 5-HT₆ receptor activation will activate pyramidal neuron firing; however, we are not aware of any electrophysiology data that have investigated the consequences of 5-HT₆ receptor stimulation in the PFC.

5-HT₇ receptors

5-HT₇ receptors are stimulatory G-protein coupled receptors that have varying levels of expression within the cortex, with strong expression in the shallow cortical layers and more moderate levels in deeper layersReference Neumaier, Sexton, Yracheta, Diaz and Brownfield⁵⁷ (Figure 1, panels G and H). More specifically, 5-HT₇ receptor expression has been observed in layer V pyramidal neurons, and in cells with relatively small oval-shaped soma that were not characterized further. Electrophysiological evidence indicates the presence of 5-HT₇ receptors on hippocampal pyramidal neuronsReference Tokarski, Zahorodna, Bobula and Hess⁵⁸ and GABAergic interneuronsReference Tokarski, Kusek and Hess⁵⁹; thus it is possible that they will also be present in cortical interneurons, but this remains to be experimentally confirmed. A study by Tokarski et al Reference Tokarski, Zelek-Molik and Duszynska⁶⁰ showed that acute intraperitoneal (i.p.) administration of the selective 5-HT₇ receptor antagonist SB269970 at 1.25 mg/kg had no effect on the firing of frontal cortex pyramidal neurons, while repeated administration reduced firing. But Lundbeck internal data suggest that a 1 mg/kg dose of this drug via the subcutaneous (s.c.) route, which is in many respects similar to the i.p. route, results in negligible occupancy at 5-HT₇ receptors. In fact, in our hands, a 30 mg/kg dose of SB269970 only engendered approximately 30% of 5-HT₇ receptors 1 hour after administration, owing to difficulty penetrating the blood–brain barrier. Given these receptor occupancy data, it remains unclear how 5-HT₇ receptor modulation will alter the activity of cells within the PFC.

PV-IR interneurons

In cortical PV-IR fast-spiking interneurons, there is evidence supporting the expression of inhibitory 5-HT_1A as well as excitatory 5-HT_2A receptors, although 5-HT_1A and 5-HT_2A receptors appear to be largely expressed by separate populations of PV-IR interneurons.Reference Puig, Watakabe, Ushimaru, Yamamori and Kawaguchi⁴² There is also some evidence suggesting that 5-HT_1B receptors are expressed on cortical PV-IR interneurons, but the subcellular localization of 5-HT_1B receptors on cortical PV-IR cells is not clear at this time. Based on this evidence, cortical PV-IR cells will likely respond to 5-HT with either an excitatory or inhibitory response on the single-cell level, while on the population scale there will likely be a mixed excitatory and inhibitory response. Whether this population level mix of excitatory and inhibitory responses skews to one side or the other may depend on a number of issues, including (1) how commonly 5-HT_1A and 5-HT_2A receptors are expressed within frontal cortical PV-IR cells, (2) the affinity of 5-HT for each of these targets (Table 1), and (3) the concentration of 5-HT applied to the neural system.

Table 1 5-HT receptor subtypes in the rat prefrontal cortex and striatum: expression patterns, affinities for 5-HT, and effects on neural activity.

Affinities for 5-HT were calculated based on pK_i/pK_D data from the IUPHAR database (www.iuphar-db.org).−: absent; +: low; ++: moderate; +++: strong; ?: unknown; ChAT-IR: choline acetyltransferase immunoreactive; I: inhibitory; IN: interneuron; MSN: medium spiny neuron; PFC: prefrontal cortex; PV-IR: parvalbumin immunoreactive; Pyr: pyramidal neuron; S: stimulatory.

Puig et al Reference Puig, Watakabe, Ushimaru, Yamamori and Kawaguchi⁴² have suggested that about half of PV-IR interneurons in layers II/III express 5-HT_1A receptor mRNA, while about 20% express 5-HT_2A receptor mRNA. However, in layer V, approximately one-third of PV-IR neurons express 5-HT_1A receptors, while a separate third of the population express 5-HT_2A receptors. Thus, these data may suggest that PV-IR interneurons in shallow cortical layers are more likely to be inhibited by 5-HT than those in deep cortical layers, where there is more likelihood for an even split in excitatory and inhibitory responses within this cell population. However, given that 5-HT’s affinity for the 5-HT_1A receptor is anywhere from 1.6- to 6.5-fold higher than for the 5-HT_2A receptor (Table 1), it is possible that the effect of 5-HT on cortical fast-spiking interneuron behavior will skew more heavily toward inhibition at lower levels of 5-HT concentrations, with excitation becoming more prominent as 5-HT concentrations increase.

Non PV-IR interneurons

In non-PV-IR cortical GABAergic interneurons, there is evidence to support 5-HT_1A receptor expression on calbindin-IR interneurons, and 5-HT₃ receptors have been found on CR-IR and CCK-IR interneurons located in shallow cortical layers. Additionally, electrophysiological evidence has suggested that 5-HT_2A receptors are also present in cortical non-PV-IR interneurons, although the specific subtype remains unclear at this time. When considering the overall effects of 5-HT in this subpopulation of cells, inhibitory 5-HT_1A receptors may again tend to predominate in low concentration ranges of 5-HT, given its higher affinity for the endogenous ligand by comparison to 5-HT_2A and especially 5-HT₃ receptors (Table 1). However, it is important to note that it is essentially unknown at this time whether these receptors are expressed in serotonergic synapses. Although SERT expression in the frontal cortex is relatively strong (see Figure 1, panels I and J), it is currently believed that only about one-quarter of serotonergic terminals in this brain region have synaptic specializations.Reference Seguela, Watkins and Descarries⁶¹ However, the serotonergic terminals that do form synapses are thought to be primarily on GABAergic interneurons residing in the shallow cortical layers,Reference Smiley and Goldman-Rakic⁶² which may include 5-HT₃ receptor-expressing interneurons. If true, this pattern may make it easier to achieve sufficiently high 5-HT concentrations to stimulate 5-HT₃ receptor activity than for those serotonergic receptor targets that are primarily modulated by extracellular serotonin concentrations. Additionally, 5-HT₃ receptors are a fast-desensitizing serotonergic target (for example, see Dale et al Reference Dale, Zhang and Leiser¹⁷); thus at high concentrations, it may be the case that a larger proportion of this receptor population is non-functional than at lower concentrations. Taken together, it is reasonable to suggest that serotonergic neurotransmission is more likely to activate non-PV-IR interneurons in the shallow cortical regions.

Pyramidal neurons

In cortical pyramidal neurons, there is evidence for the expression of a variety of serotonergic receptors. Inhibitory 5-HT_1A and 5-HT₅ receptors have been found in somatodendritic subcellular regions of pyramidal neurons, 5-HT_1B receptors may be expressed in the terminals of cortical pyramidal neurons. Additionally, several excitatory serotonergic receptors are present on these cells, including 5-HT_2A, 5-HT_2C, 5-HT₄, 5-HT₆, and 5-HT₇ receptors.

Given the wide range of excitatory and inhibitory serotonergic receptors that have been demonstrated in cortical pyramidal neurons, along with the differences in affinity these receptors have for 5-HT, it is not possible to accurately determine 5-HT’s overall effects on cortical pyramidal neurons. However, within the native frontal cortex microcircuit, pyramidal neurons are under powerful inhibitory control from a variety of inhibitory interneurons, and thus some of the effects 5-HT will have on pyramidal neurons will depend on its effects on these inhibitory interneurons.

In vivo electrophysiology

Studies that have investigated the effects of 5-HT on cortical fast-spiking (presumably PV-IR) interneuron behavior in vivo using anesthetized animals have generally found that 5-HT has inhibitory effects. For example, Puig et al Reference Puig, Watakabe, Ushimaru, Yamamori and Kawaguchi⁴² found that stimulating the midbrain raphe nuclei at low levels and recording in vivo from the medial PFC induced mixed effects in fast-spiking interneurons, with the majority of these cells exhibiting inhibition (61%), and small populations exhibiting excitation (approximately 10%) or excitation followed by inhibition (approximately 6%). Moreover, the inhibitory actions of raphe nucleus stimulation were antagonized by administration of the selective 5-HT_1A receptor antagonist WAY100635, whereas the few excitations that were observed were antagonized by the 5-HT₂ receptor antagonist ritanserin.

Superficially, these data appear to be at odds with data from the same lab demonstrating that separate but equal proportions of layer V PV-IR interneurons within the cortex express inhibitory 5-HT_1A receptors and excitatory 5-HT_2A receptorsReference Puig, Watakabe, Ushimaru, Yamamori and Kawaguchi⁴². A corollary to these data would seem to be that approximately equal proportions of layer V fast-spiking interneurons would exhibit an inhibitory or excitatory response to physiological 5-HT levels. However, during periods of light anesthesia that more closely resembled an “awake” cortical network, Puig et al.Reference Puig, Watakabe, Ushimaru, Yamamori and Kawaguchi⁴² also found a higher proportion of cells exhibiting an excitatory response to raphe stimulation (approximately 31% compared to 10%). These data fit with an interpretation that in vivo electrophysiology in anesthetized rats is biased towards 5-HT_1A receptor mediated inhibition.

Although inhibition of PV-IR interneurons would be expected to have a disinhibitory effect on pyramidal neurons, it seems that pyramidal neurons also have an overwhelmingly inhibitory response to raphe nucleus stimulation in anesthetized in vivo electrophysiological experiments. Puig et al Reference Puig, Artigas and Celada⁶³ demonstrated that low-level stimulation of the raphe nucleus inhibited activity in the majority of recorded putative pyramidal neurons (66%), while 13% exhibited an excitatory response and another 20% had mixed effects characterized by inhibition followed by excitation. 5-HT_1A receptor antagonism attenuated the inhibitory responses (but did not entirely block them), while 5-HT_2A receptor antagonism blocked the excitatory responses. Similar results were found by several other research groups after raphe nucleus stimulation in anesthetized in vivo electrophysiological preparations.Reference Amargos-Bosch, Bortolozzi and Puig^64–Reference Hajos, Gartside, Varga and Sharp⁶⁶

While the results of these in vivo anesthetized electrophysiology experiments consistently suggest that 5-HT has an overwhelmingly inhibitory effect on not only pyramidal neurons but also PV-IR interneurons, there are some important limitations to consider when evaluating these data. The presence of anesthesia may cause a number of deviations from normal cortical network states, for example a reduction in tonic or phasic 5-HT concentrations resulting from reduced raphe nucleus activity. This may bias the effects of 5-HT stimulation toward receptors with high affinity for 5-HT, such as the inhibitory 5-HT_1A receptor, which is present on pyramidal neurons, as well as multiple subtypes of non-pyramidal neurons in this region. It may be the case that in an awake and behaving animal, stimulation of the raphe nucleus would be more likely to induce excitatory responses mediated by lower-affinity receptors, such as the 5-HT_2A receptor (Table 1).

In addition to this distal effect of anesthesia, local cortical networks may also be abnormally regulated by these unusually high GABA_A receptor mediated currents. This may have the effect of partially divorcing the activity of pyramidal neurons and some types of interneurons from the activity of the inhibitory cells governing their behavior, therefore abrogating the downstream effects of serotonergic modulation. This, along with the apparent bias toward 5-HT_1A receptor mechanisms, may help to explain why pyramidal neuron firing is overwhelmingly reduced by raphe stimulation or SSRI treatments within anesthetized cortical networks in the face of reduced fast-spiking interneuron firing.

In vitro electrophysiology

In stark contrast to the in vivo electrophysiological techniques in anesthetized rodents, research groups investigating the effects of 5-HT on fast-spiking interneuron behavior in brain slices have generally found that 5-HT has an activating effect on fast-spiking interneurons. For example, Zhong and YanReference Zhong and Yan⁶⁷ found that 5-HT (20-40 µM) led to an increased firing rate of fast spiking interneurons, with relatively little effect of 5-HT at lower concentrations (1–2 µM). The idea that 5-HT activates fast-spiking interneuron behavior in slices is further supported by observations that 5-HT depolarized fast-spiking interneuronsReference Weber and Andrade⁴³ and increased spontaneous inhibitory postsynaptic currents (sIPSCs) recorded from pyramidal neurons.Reference Weber and Andrade^43, Reference Zhong and Yan^68, Reference Zhou and Hablitz⁶⁹

Additionally, in the concentration ranges that drove increases in fast-spiking interneuron activity, Zhong and YanReference Zhong and Yan⁶⁷ observed concomitant reductions in the firing rate of cortical pyramidal neurons. But the inhibitory effects of 5-HT on pyramidal neuron firing were not limited to downstream actions on GABAergic interneurons. Araneda and AndradeReference Araneda and Andrade⁷⁰ found that 5-HT hyperpolarized layer V cortical pyramidal neuronal membranes, which could be blocked by 5-HT_1A receptor antagonists but not a 5-HT₂ receptor antagonist.

As expected by the fact that both inhibitory 5-HT_1A and stimulatory 5-HT_2A receptors are expressed in cortical pyramidal neurons, 5-HT’s effects on pyramidal neuron firing in slice electrophysiological preparations are not universally inhibitory. Several groups have found that 5-HT application increases spontaneous excitatory postsynaptic currentsReference Zhou and Hablitz^69, Reference Beique, Chapin-Penick, Mladenovic and Andrade⁷¹ (sEPSCs), induces membrane depolarizations,Reference Araneda and Andrade⁷⁰ and modulates after hyperpolarization currents in a manner that supports increased activity.Reference Araneda and Andrade^70, Reference Villalobos, Beique, Gingrich and Andrade⁷²

Taken together, in contrast to in vivo recordings in anesthetized animals, these data from brain slices suggest that 5-HT has a primarily excitatory effect on cortical fast-spiking interneurons, as well as mixed excitatory and inhibitory effects on pyramidal neuron activation. Slice electrophysiology has a number of advantages, including a stronger ability to identify the cell type from which each recording is taken and the absence of anesthesia. However, there are also limitations to consider. First, although it can be argued that local neuronal circuits are largely intact, the process of sectioning tissue does attenuate some aspects of the neural network, most obviously from distal regions such as the raphe nucleus. Additionally, it should be noted that slice electrophysiological studies are often conducted in brain slices collected from adolescent rodents, and thus the degree to which the results translate to the adult cortical network are difficult to know. But perhaps the most important limitation of slice electrophysiology in the context of this article is that it is difficult to ascertain what the effects of physiological 5-HT concentrations would be. It is more or less not possible to determine whether the 5-HT concentrations used to elicit a primarily excitatory response via bath application are physiologically relevant, and in fact the concentrations commonly used to elicit these responses are in a very high range. Although it is common for slice electrophysiologists to use relatively high concentrations of a given ligand to ensure that it penetrates to the recorded cells (typically in the range of 100–1000 times Ki), concentrations of 5-HT of 20–40 µM seem disproportionately high from the perspective that the Ki for 5-HT at 5-HT_2A receptors is1.3 nM (Table 1).

Thus, it seems likely that in vivo investigations of 5-HT’s effects in anesthetized PFC networks primarily represent the actions of this neurotransmitter at low concentrations, while those from the slice represent 5-HT’s actions at high concentrations. In order to further advance our understanding of how 5-HT and serotonergic antidepressants modulate prefrontal cortical networks, it would be advantageous to conduct investigations in awake and behaving rodents.

5-HT and SSRIs

Based on the receptor profiles that are present in each of the major cell types we have considered in the frontal cortex microcircuitry, as well as electrophysiological evidence of 5-HT’s effects, the emerging theme seems to be that 5-HT has largely inhibitory actions at low concentrations dominated by 5-HT_1A receptor mediated actions at some PV-IR and non-PV-IR interneurons. Under normal circumstances, this would be expected to disinhibit pyramidal neuron activity, although pyramidal neurons themselves would also experience direct inhibitory effects of 5-HT_1A receptor activation. At higher levels of 5-HT tone, excitatory actions would start coming into play at inhibitory interneurons, primarily mediated by 5-HT_2A receptors at PV-IR fast-spiking interneurons and by 5-HT_2A and 5-HT₃ receptors in non-PV-IR interneurons. These actions on cortical interneurons would be expected to drive a GABA-mediated inhibitory influence on pyramidal neuron firing, particularly given the powerful inhibitory action PV-IR interneurons have on pyramidal neuron behavior. However, pyramidal neurons would also experience some direct excitatory actions of 5-HT mediated through a variety of serotonergic receptor targets. Therefore, at multiple levels of organization, 5-HT has opposite directed effects that overall do not appear to clearly influence the output of frontal cortical output cells toward excitation or inhibition. On a conceptual level, 5-HT may therefore play a role in fine-tuning the output of frontal cortical networks, acting in a sense as a buffer that will tend to normalize the activity of pyramidal output neurons. Given this concept and the idea that SSRI antidepressants increase synaptic and extrasynaptic serotonin concentrations without directly modulating serotonergic receptors, we hypothesize that, under acute conditions, these treatments will have mixed effects that will not clearly bias frontal cortex output in either direction. Moreover, this idea is supported by electrophysiological studies that have demonstrated that acute and short-term repeated SSRI administration have little effect on PFC firing in anesthetized rats.Reference Gronier and Rasmussen^73, Reference Ceci, Fodritto and Borsini⁷⁴ However, long-term SSRI administration seems to have an inhibitory effect on PFC firing,Reference Gronier and Rasmussen^73, Reference Ceci, Fodritto and Borsini⁷⁴ which may imply that desensitization of some serotonergic receptor systems are important for the long-term effects of SSRI administration on frontal cortical network behavior. It is not clear based on available anatomical and electrophysiological data precisely which serotonergic receptor mechanisms might be involved in these effects of chronic administration.

Vortioxetine

In contrast, vortioxetine should have markedly different effects on the output of frontal cortex networks that will be driven by its direct pharmacological actions on several serotonergic receptors. Vortioxetine’s potent antagonism of 5-HT₃ receptors should remove a source of 5-HT-mediated fast excitatory drive on a selective population of non-PV-IR cortical interneurons localized in shallow cortical layers. Based on the model presented here, this action should disinhibit cortical pyramidal neurons, although it is unclear whether this effect would be segregated purely in layer II/III pyramidal neurons or if it would generalize to layer V pyramidal output cells. 5-HT₃ receptor antagonism can also result in increased extracellular 5-HT concentrations when dosed along with an SSRI,Reference Bétry, Overstreet and Haddjeri⁷⁵ which may lead to more stimulation of serotonergic receptors that are not directly targeted by vortioxetine. Additionally, 5-HT_1A receptor agonism should induce a fast inhibitory effect mediated by inwardly rectifying potassium channels in both PV-IR and non-PV-IR interneurons, which may also disinhibit pyramidal neuron activity that would be balanced by inhibitory actions on the pyramidal neurons themselves. Additionally, vortioxetine’s partial agonist action at 5-HT_1B receptors, which acts to attenuate 5-HT_1B receptor actions in vivo,Reference El Mansari, Lecours and Blier⁷⁶ may increase serotonergic tone in the PFC by reducing inhibitory feedback at the site of the serotonergic terminal, which may again lead to greater stimulation of serotonergic targets not directly targeted by vortioxetine. Furthermore, this partial agonist effect at 5-HT_1B receptors may also attenuate 5-HT_1B receptor mediated inhibition of cortical glutamatergic terminals, leading to increased glutamate release from layer II/III pyramidal neurons at their local cellular targets. Finally, it is unclear at this time what effect vortioxetine’s 5-HT₇ receptor antagonist properties will have on PFC activity. But we expect that the net effect of vortioxetine in the PFC will be to reduce aspects of GABAergic neurotransmission and indirectly increase the activity of glutamatergic pyramidal cells.

In support of these ideas, recent in vivo electrophysiology experiments by Riga et al Reference Riga, Celada, Sanchez and Artigas¹⁸ in anesthetized rats demonstrated that acute administration of the SSRI escitalopram had no effect on the overall firing rate of putative frontal cortex pyramidal neurons, although a relatively equal minority of cells exhibited excitatory or inhibitory responses. However, in stark contrast, acute i.v. administration of vortioxetine led to an overwhelmingly excitatory response in cortical pyramidal neurons. This research group also found that an acute combination of escitalopram and the selective 5-HT₃ receptor antagonist ondansetron led to significant increases in pyramidal neuron firing, at least implicating a combination of 5-HT reuptake inhibition and 5-HT₃ receptor antagonism in vortioxetine’s excitatory effects.

In summary, SSRIs appear to have mixed effects on the function of the PFC circuit that do not clearly modulate activity either toward excitation or inhibition under acute conditions, while acute vortioxetine leads to an increase in PFC pyramidal neuron activity that is driven by a reduction in GABAergic inhibitory processes. After long-term administration, SSRI antidepressants lead to an overall reduction of PFC pyramidal neuron firing that is likely driven by the desensitization of some serotonergic receptors. At this time, empirical data on the long-term effects of vortioxetine on PFC pyramidal neuron activity are not available; however, we hypothesize that vortioxetine will still have an overall excitatory effect on these cells after chronic administration.

PV-IR fast-spiking interneurons

One of the most important classes of striatal GABAergic interneuron is identified immunohistochemically by the expression of the calcium binding protein parvalbumin (PV). As in the frontal cortex, PV-IR interneurons are characterized electrophysiologically by their fast firing rate and short-duration action potential. Therefore we will refer to these cell types interchangeably as striatal PV-IR interneurons or as striatal fast-spiking interneurons.

PV-IR interneurons are believed to make up between 3% and 5% of striatal neurons (for an excellent review, see Kawaguchi et al Reference Kawaguchi, Wilson, Augood and Emson⁷⁹). These interneurons are thought to receive a strong excitatory input from the neocortex, as well as an inhibitory input from other striatal PV-IR cells.Reference Kawaguchi, Wilson, Augood and Emson^79, Reference Wilson⁸⁰ In addition to these inputs, PV-IR cells are reported to form electrical networks with one another via gap junctions,Reference Kawaguchi, Wilson, Augood and Emson⁷⁹ which may allow for the formation of neural assemblies with other striatal fast-spiking interneurons.

The axons of striatal PV-IR cells have a basket-like arborization pattern,Reference Kawaguchi, Wilson, Augood and Emson⁷⁹ with synapses directed at both the dendritic tree and the soma of MSNs.Reference Wilson⁸⁰ Studies on the synaptic connections between this class of interneurons and MSNs have suggested that each MSN receives connections from multiple PV-IR cells.Reference Wilson⁸⁰ Thus, given the ideas that multiple PV-IR interneurons form inhibitory synapses on the soma of each MSN cell, and further that these cells may easily form inhibitory neural ensembles, it can be expected that PV-IR interneurons will have a powerful inhibitory effect on the output of striatal MSNs. Moreover, this idea is consistent with electrophysiological studies that have suggested that the majority of inhibitory processes in the striatum are mediated by PV-IR cells.Reference Jaeger, Kita and Wilson⁸¹

Although there are at least 2 other types of GABAergic interneurons that have been chemically or morphologically identified in the striatum,Reference Kawaguchi, Wilson, Augood and Emson⁷⁹ there is relatively little accumulated knowledge on the calretinin-IR and nitric oxide/neuropeptide Y/somatostatin-IR interneurons in the striatum, particularly in terms of their relationship to serotonergic neurotransmission. Therefore, this review will not consider these cell types further.

Cholinergic interneurons

Striatal non-GABAergic interneurons are differentiated from other striatal interneurons by a number of morphological, histochemical, and electrical properties, including a large soma with broadly branching dendritic and axonal arbors; the expression of the acetylcholine-producing enzyme choline acetyltransferase (ChAT); and a tonic, irregular firing pattern with low activation thresholds.Reference Kawaguchi, Wilson, Augood and Emson⁷⁹ These cells receive excitatory inputs from the cortex and thalamus, although the thalamic inputs are thought to predominate in this cellular population.Reference Kawaguchi, Wilson, Augood and Emson^79, Reference Goldberg and Reynolds⁸² Throughout this article, we will refer to this cellular population as ChAT-IR interneurons.

ChAT-IR interneurons modulate the behavior of other striatal neuronal populations in a complex manner, having a mix of excitatory and inhibitory effects on both MSN and PV-IR cell populations. These influences have been reviewed in detail in Goldberg and Reynolds,Reference Goldberg and Reynolds⁸² but for the purpose of understanding how serotonergic neurotransmission affects neuronal activity in the striatum, we will briefly discuss these effects here (summarized in Figure 4). ChAT-IR neurons have a fast excitatory influence on PV-IR cells that is mediated by nicotinic acetycholinergic receptors expressed on the soma.Reference Luo, Janssen, Partridge and Vicini⁸³ However, this excitatory effect is balanced by a slower, inhibitory influence on GABA release that is mediated by muscarinic M₂ receptors expressed on striatal GABA terminals.Reference Goldberg and Reynolds⁸² Additionally, Goldberg and ReynoldsReference Goldberg and Reynolds⁸² suggest that M₂ receptors can have a similar effect on the release of glutamate from cortical or thalamic afferent terminals and that striatal ChAT-IR neurons can have a direct stimulatory effect on MSN firing that is mediated via muscarinic M₁ receptors.

Medium spiny neurons

Whereas principle output cells in the cortex are excitatory glutamatergic pyramidal neurons, principle cells in the striatum are GABAergic in nature, and thus efferents originating from the striatum have an inhibitory effect on their projection regions. These striatal principle cells are MSNs, which are thought to represent between 80% and 98% of the neurons present in the neostriatum (for a review, see Tepper et al Reference Tepper, Koos and Wilson⁸⁴). From a morphological perspective, MSNs are characterized by their medium size and spiny dendritic arbors.Reference Kawaguchi, Wilson and Emson⁸⁵ MSNs can be further subdivided into at least 2 separate but equally represented populations based on neuropeptide and neurotransmitter receptor expression: those expressing enkephalin and dopamine (DA) D₂ receptors, and those expressing dynorphin, substance P, and DA D₁ receptors.Reference Augood and Emson^86–Reference Pollack, Harrison, Wooten and Fink⁸⁹ MSNs receive excitatory glutamatergic afferents from the neocortex and thalamus,Reference Wilson⁸⁰ as well as serotonergic afferents from the midbrain raphe nuclei. However, the MSN subpopulations noted above preferentially send efferent projections to separate portions of the basal ganglia. MSNs expressing enkephalin and DA D₂ receptors project preferentially to the globus pallidus, while those expressing dynorphin, substance P, and DA D₁ receptors send efferents to the substantia nigra.Reference Gerfen, Engber and Mahan^87, Reference Gerfen and Young⁸⁸

In addition to MSN axon efferents sent to basal ganglia targets, MSN neurons also send axon collaterals to other MSN cells in the striatum that synapse in the dendritic arbor.Reference Kawaguchi, Wilson, Augood and Emson⁷⁹ This phenomenon, in combination with the inhibitory nature of these cells, originally led to a hypothesis that striatal MSNs form a competitive lateral inhibitory network, where activation of one MSN leads to inhibition of surrounding MSNs. However, more recently this theory has been modified in light of electrophysiological evidence demonstrating that the inhibitory effect of a given MSN on the firing of surrounding MSNs is very weak, owing probably to the dendritic insertion of MSN-MSN axon collaterals.Reference Wilson^80, Reference Jaeger, Kita and Wilson⁸¹ A modified theory suggests that lateral inhibition may still be possible in MSN networks, based on the fact that each MSN sends axon collaterals to about 1 in 6 of its surrounding MSNs and receives inputs from approximately 450 other local MSNs. In this theory, MSNs that are not connected to one another via axon collaterals but receive similar excitatory inputs may form neural assemblies that can inhibit surrounding MSNs in the aggregate (reviewed in WilsonReference Wilson⁸⁰). Thus, it is possible that MSNs have a relatively weak lateral inhibitory influence on other MSNs in the striatum, but this is currently considered to represent a small minority of the total phasic inhibitory drive on striatal MSN activity.Reference Wilson⁸⁰

In the following section we will review evidence on the regional and cellular pattern of 5-HT receptor expression for each of the above-mentioned striatal cell types.

5-HT₁ receptors

5-HT_1A receptor binding in the caudate and putamen it is extremely lowReference Aznar, Qian, Shah, Rahbek and Knudsen^29, Reference Pompeiano, Palacios and Mengod⁹⁰ (Figure 2, panels A and B). Thus, although there may be some very weak expression of 5-HT_1A receptors within the striatum of the rodent brain,Reference Ghavami, Stark, Jareb, Ramboz, Segu and Hen⁹¹ we will consider it to be essentially absent in this region.

In contrast, studies of 5-HT_1B receptor expression in the striatum showed moderate to strong expression levels of mRNAReference Boschert, Amara, Segu and Hen⁹² as well as proteinReference Bruinvels, Palacios and Hoyer^33, Reference Sari, Miquel and Brisorgueil³⁵ (Figure 2, panels C and D), which may indicate that 5-HT_1B receptors are postsynaptically expressed in the striatum or are expressed on interneurons. A mouse study by Ghavami et al,Reference Ghavami, Stark, Jareb, Ramboz, Segu and Hen⁹¹ using a conditional 5-HT_1B receptor knockout approach, demonstrated that 5-HT_1B receptors made by striatal MSNs are primarily sent to the basal ganglia projection regions. These data might suggest that striatal 5-HT_1B receptors are localized primarily on terminals originating from the midbrain raphe.Reference Boschert, Amara, Segu and Hen^92, Reference Sari⁹³ However, dual-labeling immunohistochemistry has provided evidence demonstrating that 5-HT_1B receptors are also present on striatal PV-IR and ChAT-IR interneurons.Reference Egeland, Warner-Schmidt, Greengard and Svenningsson³⁸ It is not known whether there is a somatodendritic expression of 5-HT_1B receptors in these cell types, and it is possible that 5-HT_1B receptors act as terminal heteroreceptors in striatal PV-IR and ChAT-IR interneurons.

5-HT_1D receptors are closely related to 5-HT_1B receptors in the brain and often have a similar, although much more rarified, expression pattern. However, Bruinvels et al Reference Bruinvels, Palacios and Hoyer³³ demonstrated that, in the striatum, 5-HT_1D receptor specific binding constituted about 11% of the total binding of a radioligand that is nonselective for 5-HT_1B/1D receptors. However, given evidence that 5-HT_1D receptors form a heterodimer with 5-HT_1B receptors when they are co-expressed,Reference Xie, Lee, O’Dowd and George⁹⁴ it can be questioned whether these receptors should be considered separately.

Based on these data, striatal 5-HT_1B receptors may act primarily to modulate extracellular concentrations of 5-HT by acting as an inhibitory autoreceptor at serotonergic terminals. Moreover, this idea is supported by data from microdialysis experiments.Reference de Groote, Klompmakers, Olivier and Westenberg³⁶

If 5-HT_1B receptors are present in the somatodendritic region of PV-IR and ChAT-IR interneurons, then stimulation of these receptors should drive down the activation state of these interneurons. Alternatively, 5-HT_1B receptors expressed as terminal heteroreceptors in these cells may act to reduce release of GABA or acetylcholine, respectively, without markedly modulating cellular excitability at the soma. Which of these subcellular expression patterns is present in striatal interneurons is essentially unknown at this time.

5-HT_2A/2C receptors

5-HT_2A receptors are only moderately expressed in the striatum,Reference Cornea-Hebert, Riad, Wu, Singh and Descarries⁴¹ although it should be noted that at least one research group has noted a gradient in 5-HT_2A receptor mRNA expression, with more dense expression in the caudal portions of the striatum.Reference Ward and Dorsa⁹⁵ Cornea-Hebert et al Reference Cornea-Hebert, Riad, Wu, Singh and Descarries⁴¹ have confirmed 5-HT_2A receptor immunoreactivity in the majority of striatal MSNs. Furthermore, Ward and DorsaReference Ward and Dorsa⁹⁵ found that 5-HT_2A receptor mRNA is present in MSNs that project to both the globus pallidus and substantia nigra. Cornea-Hebert et al Reference Cornea-Hebert, Riad, Wu, Singh and Descarries⁴¹ also observed some 5-HT_2A receptor immunoreactivity in a population of neurons characterized by a large soma. These cells are presumably cholinergic interneurons based on the large soma.

5-HT_2C receptors are present in the striatum at moderate levelsReference Abramowski, Rigo, Duc, Hoyer and Staufenbiel⁴⁴ and are expressed in striatal MSN cells,Reference Ward and Dorsa^95, Reference Eberle-Wang, Mikeladze, Uryu and Chesselet⁹⁶ and once again do not appear to discriminate between MSN cells projecting to the globus pallidus vs the substantia nigra.Reference Ward and Dorsa⁹⁵ Although we have not been able to identify any systematic immunohistochemical studies that have examined whether 5-HT_2C receptors are present in striatal interneuron populations, electrophysiological experiments would suggest 5-HT_2C receptor expression on striatal fast spiking interneuronsReference Blomeley and Bracci⁹⁷ and ChAT-IR cells.Reference Bonsi, Cuomo and Ding⁹⁸

Within the frontal cortex, 5-HT_2A receptors have mixed effects on the behavior of the microcircuitry by activating pyramidal neurons and GABAergic interneurons. Within the striatum, a similar picture of the effects of 5-HT_2A/2C receptor stimulation emerges, with activation of these receptors likely to stimulate the striatal MSN projection cells as well as PV-IR and ChAT-IR interneurons.

5-HT₃ receptors

5-HT₃ receptor expression in the striatum is extremely lowReference Gehlert, Gackenheimer, Wong and Robertson⁴⁶ (Figure 2, panels E and F). There are reports of a few strongly 5-HT₃ receptor-IR cells present within the striatumReference Morales, Battenberg and Bloom⁴⁸; however, the identity of these cells is not known.

5-HT₄ receptors

5-HT₄ receptor expression is moderate to high in the striatum with an increasing rostro-caudal gradient.Reference Vilaro, Cortes and Mengod^51, Reference Vilaro, Cortes, Gerald, Branchek, Palacios and Mengod⁹⁹ Given that the regional profile of 5-HT₄ receptor mRNA is very similar to that seen using autoradiography within the striatum,Reference Vilaro, Cortes and Mengod^51, Reference Vilaro, Cortes, Gerald, Branchek, Palacios and Mengod⁹⁹ it can be hypothesized that 5-HT₄ receptors primarily have a somatodendritic expression pattern. Egeland et al Reference Egeland, Warner-Schmidt, Greengard and Svenningsson³⁸ found that striatal PV-IR and ChAT-IR interneurons do not express 5-HT₄ receptors. Given the prominence of 5-HT₄ receptors in the striatum, it is therefore likely that MSN cells express 5-HT₄ receptors; however, this awaits a definitive confirmation.

5-HT₅ receptors

Striatal 5-HT₅ receptor immunoreactivity is weak, and there is a paucity of evidence on the cell types expressing 5-HT₅ receptors. The little that is available suggests that these receptors are expressed by MSNs but not cholinergic cells.Reference Oliver, Kinsey, Wainwright and Sirinathsinghji⁵³ Thus, 5-HT₅ receptor activation might have an inhibitory effect on striatal MSN activity, and in fact this is one of the few serotonergic inhibitory receptors that have been found in the striatum. It remains to be seen whether 5-HT₅ receptors are present in any of the GABAergic interneurons of the striatum.

5-HT₆ receptors

5-HT₆ receptors are densely expressed in the striatum, where they are primarily associated with a postsynaptic, dendritic expression pattern.Reference Gerard, Martres and Lefevre⁵⁵ 5-HT₆ receptors have been observed in striatal MSN cellsReference Ward and Dorsa^95, Reference Helboe and de Jong¹⁰⁰ in a manner that does not differentiate between MSNs projecting to the globus pallidus or the substantia nigra,Reference Ward and Dorsa⁹⁵ continuing with the apparent theme that serotonergic receptors do not discriminate between these alternate projections to the basal ganglia. In human striatal tissue, Marazziti et al Reference Marazziti, Baroni and Pirone¹⁰¹ found that 5-HT₆ receptors were present in putatively PV-IR cells. However, in rodent tissue, another group found no evidence of 5-HT₆ receptor mRNA in PV-IR cells.Reference Helboe and de Jong¹⁰⁰ Electrophysiological data have suggested the presence of 5-HT₆ receptors in striatal cholinergic (ChAT-IR) interneurons.Reference Bonsi, Cuomo and Ding⁹⁸ Thus, stimulation of striatal 5-HT₆ receptors can be expected to activate striatal MSN and ChAT-IR interneurons, whereas it is unclear that these receptors are present in striatal PV-IR interneurons.

5-HT₇ receptors

Within the striatum, Neumaier et al Reference Neumaier, Sexton, Yracheta, Diaz and Brownfield⁵⁷ reported moderate levels of 5-HT₇ receptor-like immunoreactivity, and 5-HT₇ receptor-selective autoradiographic binding is present at relatively low levels in the rodent (see Figure 2, panels G and H) and human striatum.Reference Varnas, Thomas, Tupala, Tiihonen and Hall¹⁰² We have not been able to identify any anatomical studies that characterize the striatal cellular subtypes expressing 5-HT₇ receptors. However, Bonsi et al Reference Bonsi, Cuomo and Ding⁹⁸ found that cholinergic interneurons expressed 5-HT₇ receptor mRNA, and further that the activating effect of 5-HT on striatal cholinergic cells could be attenuated by application of the 5-HT₇ receptor selective antagonist SB269970. Thus, 5-HT₇ receptors are likely to be expressed at the least in striatal ChAT-IR interneurons, where they have a stimulatory effect.

PV-IR fast-spiking interneurons

In striatal PV-IR fast-spiking interneurons, there is evidence supporting the presence of excitatory 5-HT_2C receptors as well as inhibitory 5-HT_1B receptors. It is not known whether the 5-HT_1B receptor immunoreactivity is associated with a somatodendritic expression, as has been observed in the hippocampus,Reference Peddie, Davies, Colyer, Stewart and Rodriguez^39, Reference Peddie, Davies, Colyer, Stewart and Rodriguez⁴⁰ or if it is expressed primarily as a terminal heteroreceptor. Additionally, it is unknown whether 5-HT_1B and 5-HT_2C receptors constitute the entire complement of serotonergic receptors expressed by striatal PV-IR interneurons. Furthermore, it is somewhat difficult to gauge the relative level of affinity at these targets, as the range of observed affinities for 5-HT at these 2 receptors overlaps (Table 1). Based on these uncertainties, it is unclear how physiological levels of 5-HT will alter the behavior of striatal PV-IR interneurons, although one might expect to find mixed effects of 5-HT stimulation. However, it should be noted that electrophysiological investigations into the effects of 5-HT stimulation on PV-IR interneuron activity have not found any 5-HT_1B receptor mediated inhibitory currents in striatal PV-IR interneurons (see discussion below), which may either suggest this receptor is expressed as a terminal heteroreceptor, or may call its presence in these cells into question altogether. Taken together, these data may suggest that striatal PV-IR interneurons will be primarily excited by 5-HT, via a 5-HT_2C receptor mediated mechanism.

Cholinergic interneurons

There is evidence supporting the presence of excitatory 5-HT_2C, 5-HT₆, 5-HT₇, and inhibitory 5-HT_1B receptors in striatal ChAT-IR interneurons. This cellular profile may suggest that 5-HT will bias striatal ChAT-IR interneurons toward excitation.

Medium spiny neurons

In striatal MSNs, the available evidence primarily supports the presence of excitatory 5-HT receptors, including 5-HT_2A, 5-HT_2C, 5-HT₄, and 5-HT₆ receptors, although inhibitory 5-HT₅ receptors are also present. Given that 5-HT₅ receptors are expressed in a relatively weak manner in the striatum, and have a relatively low affinity for 5-HT by comparison to at least 5-HT_2A, 5-HT₄, and 5-HT₆ receptors (Table 1), it is likely that 5-HT will have primarily excitatory direct effects on MSN firing behavior.

In vivo electrophysiology

Although the available evidence suggests that primarily excitatory serotonergic receptor targets are expressed on the major cell types in the striatal circuitry, studies of the effects of 5-HT on neuronal firing have overwhelmingly shown an inhibitory effect. For example, Rebec and CurtisReference Rebec and Curtis¹⁰³ found in paralyzed (not anesthetized) rodents that 5-HT application by microinjection at a 10 µM concentration suppressed firing at neurons close to the microinjection cannula. Similar results were found in other studies that investigated this issue,Reference El Mansari, Radja, Ferron, Reader, Molina-Holgado and Descarries^104–Reference Olpe and Koella¹⁰⁶ although in some neurons a brief excitation was followed by inhibition.Reference El Mansari, Radja, Ferron, Reader, Molina-Holgado and Descarries¹⁰⁴ It should be noted that in at least one case, more units were excited than inhibited.Reference Contreras, Sanchez Estrada, Molina Hernandez and Marvan¹⁰⁷ The identities of the recorded units in these in vivo experiments were not further characterized, but, given the ascendant proportion of cells that MSNs represent within the striatum, it is most likely that the recorded neurons are primarily MSNs. Interestingly, although excitatory 5-HT_2A receptors are present on MSNs, and further that 5-HT_2C receptors are present on all 3 major striatal neuronal types discussed here, 5-HT_2A/2C receptor agonists mimicked the overwhelmingly inhibitory effects of 5-HT application. We were not able to identify any in vivo electrophysiology studies that examined the effects of 5-HT on identified fast-spiking interneurons or cholinergic interneurons within the striatum.

In vitro electrophysiology

In the one in vitro study we identified that attempted to examine the response of MSNs to 5-HT in acutely dissociated cells, the authors found that the majority of cells were excited by the bath application of 5-HT at 10–60 µM concentrations. This result stands in stark contrast to the behavior of putative MSNs recorded in vivo using anesthetized and paralyzed rats, but is in line with the primarily excitatory complement of 5-HT receptors that has been found on striatal MSNs. Given that the in vivo experiments reported above used similarly high 5-HT concentrations, the most probable explanation for this difference lies in the fact that the recorded MSNs were acutely dissociated from the striatal network, which divorces the subject cell from the inhibitory cells that would normally govern its behavior.

Blomeley and BracciReference Blomeley and Bracci⁹⁷ used a slice electrophysiology preparation to demonstrate that high concentrations of 5-HT (30 µM) strongly increased the excitability of striatal fast-spiking interneurons. These authors further demonstrated that this excitatory response was blocked by a 5-HT_2C receptor antagonist. Moreover, this research group did not report the presence of any 5-HT mediated inhibitory currents in these cells that could be attributed to 5-HT_1B receptors, which would seem to be at odds with the somatic 5-HT_1B receptor expression observed in striatal PV-IR interneurons by Egeland et al.Reference Egeland, Warner-Schmidt, Greengard and Svenningsson³⁸

In vitro electrophysiological evidence from at least 2 research labs also supports the idea that high concentrations of 5-HT have an excitatory effect on striatal cholinergic interneurons. Blomeley and BracciReference Blomeley and Bracci¹⁰⁸ found that 30 µM 5-HT excited striatal cholinergic cells and further that the 5-HT₂ receptor antagonist ketanserin antagonized this effect. Similarly, Bonsi et al Reference Bonsi, Cuomo and Ding⁹⁸ found that 50 µM 5-HT excited striatal cholinergic interneurons, that the 5-HT_2A/2C receptor agonist DOI mimicked this effect, and finally that a 5-HT_2C receptor antagonist reversed this excitation. This group also showed that the excitatory effect of 5-HT could be reduced by 5-HT₆ and 5-HT₇ receptor selective antagonists.

5-HT and SSRIs

The electrophysiological data on the effects of 5-HT on striatal PV-IR and ChAT-IR firing suggest that 5-HT has an excitatory effect on these cell types. Additionally, the profile of 5-HT receptor expression on striatal MSN cells also suggests that 5-HT should have primarily excitatory direct effects. This is supported by data showing that 5-HT activates MSN cell firing when these cells are dissociated from the striatal network, but it is firmly at odds with the electrophysiological effects of 5-HT on putative MSN cells recorded in vivo.

The most plausible explanation for this apparently paradoxical effect of 5-HT on striatal MSN activity is that the concomitant activation of striatal PV-IR interneurons has a downstream inhibitory action on MSNs that overwhelms 5-HT’s direct excitatory effects on these cells. This reasoning is based on the fact that striatal PV-IR interneurons have inhibitory axo-somatic insertions on numerous MSN cells, which will tend to powerfully and negatively modulate MSN activity. Furthermore, each MSN cell receives inhibitory inputs from multiple PV-IR interneurons. This phenomenon, along with the presence of electrical gap junctions between striatal PV-IR cells, will tend to multiply the inhibitory influence of PV-IR interneurons on MSN behavior. It is possible that some of 5-HT’s effects on MSN firing are also mediated via striatal ChAT-IR interneurons. However, ChAT-IR interneuron activation is expected to have a mixture of excitatory and inhibitory effects that seem unlikely to clearly bias MSN output in either direction. Therefore, the strong inhibitory effect of 5-HT on MSN firing is most likely mediated primarily through PV-IR interneuron activation.

Based on this line of reasoning, we hypothesize that SSRIs will have primarily inhibitory local effects on striatal MSN firing under acute conditions. Given that 5-HT_2C receptors are the only excitatory serotonergic targets identified on striatal PV-IR interneurons, the degree to which this local inhibitory effect of SSRIs is maintained after chronic administration might be dictated by the degree to which 5-HT_2C receptors desensitize or downregulate in response to long-term stimulation. Thus, it is possible that the local inhibitory effect of SSRIs on MSN firing will be decreased after chronic administration. Additionally, given that SSRIs tend to reduce cortical pyramidal neuron firing after chronic administration, striatal PV-IR cells may also receive an attenuated excitatory drive from the cortex after chronic SSRI treatment. Thus, it seems plausible to hypothesize that long-term SSRI treatment will attenuate the reduction in MSN firing. However, we were unable to identify in vivo or in vitro electrophysiological studies of the effects of acute or long-term SSRI administration on striatal neuron firing.

Vortioxetine

Although vortioxetine has markedly different effects by comparison to SSRIs within the frontal cortex, this may not be the case within the striatum. 5-HT₃ and 5-HT_1A receptors are largely absent in the striatum, and thus cannot be expected to modulate the effects of this network locally. 5-HT₇ receptors, where vortioxetine acts as an antagonist, have been confirmed in striatal ChAT-IR interneurons, and thus vortioxetine may remove some of the excitatory influence of 5-HT on this cell population. But it is seems unlikely that altered activation of striatal ChAT-IR interneurons will clearly bias MSN output in either direction. 5-HT_1B receptors, where vortioxetine functions as a partial agonist in vivo,Reference El Mansari, Lecours and Blier⁷⁶ are thought to primarily exist in the striatum as terminal autoreceptors. Thus vortioxetine’s primary 5-HT_1B receptor mediated effect in this region will probably be an enhancement of extracellular 5-HT by comparison to an SSRI. However, the question of whether vortioxetine drives greater extracellular 5-HT concentrations in the striatum than an SSRI is currently untested. Additionally, if 5-HT_1B receptors are expressed in a meaningful way on striatal PV-IR interneurons, then vortioxetine’s 5-HT_1B receptor partial agonism may have the effect of attenuating the only inhibitory influence 5-HT would be expected to have on PV-IR interneurons, thereby increasing their excitability and further driving down the activity of MSNs.

Therefore, when considering local effects of vortioxetine within the striatum only, the emerging theme is that vortioxetine will increase striatal PV-IR interneuron activation and reduce the output of striatal MSN projection neurons downstream. Moreover, these inhibitory effects on striatal MSN behavior will be qualitatively similar to those expected from SSRIs. However, the question of how vortioxetine may modulate striatal neuron activity is complicated by the fact that striatal MSN and PV-IR interneurons receive excitatory inputs from the cortex,Reference Wilson⁸⁰ where vortioxetine increases activity in projection neurons.Reference Riga, Celada, Sanchez and Artigas¹⁸ Thus, vortioxetine may have both local and non-local effects that will tend to activate striatal PV-IR interneurons, while in MSN cells, vortioxetine’s local GABA-mediated inhibitory effects and nonlocal excitatory effects may be at odds with one another. These contradictory actions may suggest that vortioxetine will produce a mix of excitatory and inhibitory effects on MSN activity, but we hypothesize that these effects will skew toward inhibition of MSN firing.