In brain slices, changes in serotonin concentration can be evoked using local electrical stimulation in brain regions containing serotonergic neurons or their axonal projections. The dorsal raphe nucleus (DRN), a tiny hub in the core of the medulla, contains the majority of serotonin-producing neurons that send ascending projections into the brain. Voltammetric measurements detect serotonin efflux from both axonal and somatodendritic sites in this region because a subset of serotonergic neurons synapse locally. Although axonal serotonin release is prevalent throughout the central nervous system, experiments employing FSCV are typically constrained to brain regions dense with serotonergic terminals and limited interference from other neurotransmitters and metabolites. These studies predominantly take place in the substantia nigra, a midbrain region composed of the pars compacta, packed with dopamine-synthesizing neurons, and the pars reticulata (SNr), a networked relay region that includes the densest serotonergic projections from the DRN to any forebrain region (9 × 10⁶ sites per mm³) (Moukhles et al., 1997). In the SNr, serotonin is the predominant electroactive neurotransmitter evoked by electrical stimulations, frequently observed in the absence of somatodendritic dopamine release (Cragg et al., 1997). However, Moukhles et al. (1997) reported that serotonergic processes form synaptic junctions at a high rate in the SNr than any other brain region. It should be considered, therefore, that serotonin dynamics described in this region may be dissimilar to the dynamics in other brain regions, including the cerebral cortex, neostriatum, and hippocampus, where a majority of serotonin terminals form non-junctional synapses (Descarries et al., 1990). Serotonin efflux has also been described using FSCV in brain slices containing the suprachiasmatic nucleus (SCN) and ventral lateral geniculate nucleus (vLGN), hypothalamic and thalamic areas with similarly robust serotonergic innervation.

Serotonin measurements in in vivo FSCV experiments have taken place exclusively in the SNr. Thick vasculature and meninges above the DRN make targeting this region in the intact brain with a fragile carbon-fiber microelectrode difficult. Other serotonergic regions of interest, such as the hippocampus and prefrontal cortex, have not been explored due to significant chemical interference from other monoamines. However, recent advancements in neuronal stimulation technology may help circumvent this problem, and these potential future directions will be discussed in more detail in the conclusion of this article.

Electrochemical methods, including FSCV, lack absolute chemical specificity. Some chemical species, particularly those with similar structure, can interfere with detection of the desired substance by oxidizing at similar or identical potentials. Therefore, voltammetric measurements rely on five criteria for identification of endogenously released substances: First, cyclic voltammograms obtained under experimental conditions must have high correlation with cyclic voltammograms of the authentic compound. Second, presence of the neurotransmitter must be validated by independent chemical identification, such as microdialysis, tissue content analysis, or radioligand binding in the targeted brain region. The third criterion requires precise anatomical positioning of the sensor into the brain region of interest. Fourth, observed release should follow known physiological properties for the neurotransmitter and target brain region. Finally, identification of the released substance is dependent on pharmacological validation.

O'Connor and Kruk (1991a) systematically addressed the criteria for electrochemical validation in the first published report of endogenous serotonin measured in rat brain slices using FSCV (O'Connor and Kruk, 1991a). The cyclic voltammogram obtained from electrically-evoked serotonin is highly correlated to the one obtained after adding known concentrations of serotonin to the bath solution. Stimulation trains (500 ms in duration) elicited transient flux in serotonin levels in the DRN, where serotonin-synthesizing neurons are located, and SCN, a region dense in serotonin projections from the DRN (Fuxe, 1965). The evoked concentrations measured in both brain regions was completely and reversibly abolished by removal of calcium from the buffer solution or addition of sodium-channel blocker tetrodotoxin, complying with known physiological properties of exocytotic release. RO 4-1284, an irreversible vesicular monoamine transporter 2 (VMAT2) inhibitor, attenuated release, confirming that observed release was vesicular in nature. Inhibition of monoamine oxidase (MAO) had no effect on stimulated efflux, ruling out interference from serotonin's metabolite, 5-HIAA. Finally, the clearance rate of serotonin in the DRN and SCN could be decreased after application of citalopram, a selective serotonin uptake inhibitor, but not by benztropine, a norepinephrine uptake inhibitor, to bath solution. Similar procedures validate the identity of serotonin detected in subsequent experiments by this and other groups.

Bunin and Wightman (1998) later investigated an aspect of serotonin's physiological release properties that had not been addressed by initial voltammetric characterizations. The dimensions of carbon-fiber microelectrodes are considerably larger than the synaptic cleft into which neurotransmitters are released (Figure 2). Consequently, FSCV detects extracellular, not intrasynaptic, changes in concentration, and its measurements are limited to the neurotransmitter concentration that diffuses into the extrasynaptic space following release. A number of neuromodulators diffuse beyond the synaptic space to reach their receptors and transporters in a process called volume transmission (Fuxe et al., 2010), and prior evidence from non-voltammetric techniques implicated serotonin as a volume neurotransmitter. Ultrastructural studies of serotonergic terminals throughout the brain suggest that they form predominantly non-junctional synapses (Chazal and Ralston, 1987). This terminal architecture, together with reports that expression of serotonin transporters and receptors occurs primarily on extrasynaptic regions of neuronal processes (Kia et al., 1996; Zhou et al., 1998), is indicative of volume transmission. In light of this information, Bunin and Wightman (1998) hypothesized that electrically-evoked serotonin should reach the extracellular space via diffusion, without buffering from uptake and receptor binding sites. This was found to be the case for both somatodendritic and terminal release, where the concentration of serotonin evoked per stimulation pulse during 20-pulse trains was equivalent to the concentration evoked by a single pulse (Bunin and Wightman, 1998). Therefore, the authors concluded that serotonin concentrations measured by voltammetry reflect physiological volume transmission from the synapse to its extrasynaptic targets.

Since O'Connor and Kruk's first report of voltammetric detection of serotonin, several modifications have been implemented to adapt and improve the use of FSCV for novel applications. The voltage potential waveform (−1 to +1.4 to −1) used by the Stamford and Kruk labs, as well as others, in serotonin studies cited throughout this review was adjusted by Jackson et al. (1995) to improve temporal resolution. This modification to an N-shaped waveform, which scans from +0.2 to +1.0 to −0.1 back to +0.2 (Figure 1A), was designed to reduce serotonin adsorption to the electrode surface, as this slows electrode response times. It also avoids fouling reactions of serotonin's oxidative and reductive byproducts, improving electrode sensitivity and stability over time (Jackson et al., 1995). The modified waveform's improvement enhanced electrode response times and enabled more accurate measurements of release and uptake rates, facilitating closer examination of the kinetic parameters of serotonin release.

Improvement of carbon-fiber microelectrode sensors has been another ongoing adaptation to voltammetric measurements of serotonin. Brazell et al. (1987) first reported that dip-coating a carbon-fiber microelectrode in Nafion, a cation-selective polymer, improves serotonin and dopamine detection (Brazell et al., 1987; Jackson et al., 1995). Nafion enhances serotonin detection in two ways: first, by directly increasing the electrode's sensitivity to (positively-charged) serotonin, and second, by reducing its sensitivity to interfering anionic species such as uric acid and serotonin's metabolites. Years later, the success of the first in vivo voltammetric measurements of endogenous serotonin concentrations in a rat owe their success to the enhanced sensitivity and temporal resolution facilitated by Nafion-coated sensors and the modified voltage potential waveform (Hashemi et al., 2009).

Many labs continued their investigations of serotonin release without adopting either modification. Because each study reports on electrically-evoked changes in serotonin concentration, which are derived from in vitro calibrations, comparing findings between labs is not considered an issue within this review. It is important to note that these calibrations do not take into account the deleterious effects of electrode fouling that may be appreciably different over the course an experiment depending on the waveform used. Regardless of waveform choice, however, demonstration of a linear relationship between concentration applied and the current evoked establish the suitability of a waveform for stable detection of serotonin.

Three subtypes of serotonin receptors, all 5-HT1-type, are expressed on serotonergic axons, soma, and dendrites and function as autoreceptors that provide inhibitory feedback. 5-HT1-type receptors are found throughout the brain as autoreceptors, expressed on pre-synaptic serotonin terminals, and also as heteroreceptors, expressed on post-synaptic targets. The most well-studied autoreceptors, 5-HT1A, 1B, and 1D are seven transmembrane, G-protein coupled receptors (GPCRs). 5-HT1B and 1D autoreceptors negatively couple to adenylyl cyclase (Yocca and Maayani, 1990). 5-HT1A heteroreceptors throughout the brain also inhibit adenylyl cyclase activity, but autoreceptors in the DRN apparently function through a different G_i-coupled mechanism (Clarke et al., 1996). In vivo studies of these autoreceptors are challenging because even highly-selective drugs inadvertently target pharmacologically-identical heterosynaptic receptors, which are often expressed at high levels in the same brain region as the autoreceptor. Intact circuitry thus makes it difficult to extricate direct effects of autoreceptor activity from indirect regulation by heteroreceptors.

Voltammetric measurements in brain slices avoid some of the problems associated with 5-HT1-type receptor pharmacology. In slices, the absence of spontaneous activity in serotonergic cells, due to either separation from cell bodies in a terminal slice or elimination of noradrenergic inputs in a DRN slice, results in loss of endogenous serotonin tone (Judge and Gartside, 2006). Therefore, these experiments avoid tonic activation of autoreceptors and can also avoid transient autoreceptor activity, when appropriate, using POP stimulations. This provides an opportunity to study the timing and function of these receptors in relative isolation. O'Connor and Kruk (1991b) showed that the non-selective autoreceptor antagonist methiothepin did not affect the concentration of serotonin evoked by POP stimulations, but increased serotonin elicited by longer stimulations. Further exploration with stimulations of varying frequency and duration determined that activation of autoreceptors requires a stimulation period of at least 400 ms (O'Connor and Kruk, 1991b). This time frame is comparable to the activation window for dopamine autoreceptors in striatal and limbic regions. Phillips et al. (2002) found that the activation delay observed for dopamine autoreceptors reflects timing of intracellular cascades added to the rate of neurotransmitter diffusion in a given brain area (Phillips et al., 2002).

5-HT1A, 1B, and 1D receptors are expressed at high levels in the DRN, where they negatively influence neuronal firing rate and extracellular levels of serotonin (Sprouse and Aghajanian, 1987; Pineyro et al., 1995; Moret and Briley, 1997; Adell et al., 2001). Voltammetric studies corroborate the inhibitory functions of all three receptors in this region by demonstrating that their selective agonists can reduce the amplitude of electrically-evoked serotonin release (Davidson and Stamford, 1995b; Hopwood and Stamford, 2001). Although its heteroreceptor analogues are prominently expressed in limbic regions, 5-HT1A autoreceptors are only expressed in the DRN and median raphe nucleus (Verge et al., 1985). Serotonin levels in forebrain terminal regions are affected by 5-HT1A-mediated changes in DRN unit activity (Kreiss and Lucki, 1994; Casanovas et al., 1997), but only 5-HT1B and 1D autoreceptors are expressed locally to functionally inhibit release in these regions. Voltammetric measurements in the SCN and vLGN confirm absence of 5-HT1A autoreceptor function in these terminal regions. 5-HT1B and 1D receptors, and not 5-HT1A receptors, negatively influence serotonin efflux in brain slices containing terminal regions (O'Connor and Kruk, 1992; Davidson and Stamford, 1996).

The 5-HT1A receptor may be the trump card in this family of autoreceptors: 5-HT1A receptor mRNA is expressed in nearly 100% of serotonergic cells and up to 15% of GABAergic interneurons in the DRN (Day et al., 2004). This receptor robustly regulates both neuronal firing rates and extracellular serotonin levels in the DRN (Sprouse and Aghajanian, 1987; Hjorth and Sharp, 1991). Voltammetric measurements find that antagonists for 5-HT1A and 1B have a supra-additive effect when administered together: increase in serotonin efflux is greater when both receptors are blocked than would be expected given the effect of each antagonist alone (Roberts and Price, 2001). In addition, the effects of 5-HT1B receptor antagonists on serotonin efflux are overpowered by 5-HT1A receptors unless they are also blocked, suggesting that these receptors compensate for reductions of 5-HT1B activity. Given these results, it is suggested that 5-HT1A and 1B receptors exhibit a functional interaction that is facilitated by proximal expression sites on serotonin neurons. Interest in 5-HT1A receptors has increased in the last decade, as they may play a role in depression and anxiety-related disorders (Ohno, 2010). Use of FSCV in future studies could meaningfully contribute to our understanding of how 5-HT1A receptor-mediated modulation of serotonin release plays a role in the etiology and treatment of these disorders.

Much speculation has occurred regarding the explanation for seemingly parallel functions of 5-HT1B and 1D autoreceptors. Both receptors are expressed in most serotonergic brain regions and have superficially redundant effects. One theory posits that these autoreceptors differ in their affinity for serotonin: one high affinity and the other low affinity. However, it has since been demonstrated that their affinities are nearly identical (Boess and Martin, 1994). More likely, the two receptors are expressed in different anatomical locations, and thus provide site-specific regulation of serotonin release, e.g., dendritic vs. axonal localizations in the DRN. Stamford et al. (2000) have reviewed the evidence supporting this hypothesis (Stamford et al., 2000).

The SNr expresses the highest concentration of 5-HT1B autoreceptors and heteroreceptors in the murine brain (Pazos and Palacios, 1985). 5,7-HT-induced lesions of serotonin neurons reduced 5-HT1B expression level by 37%, presumably due to degradation of serotonin terminals (Verge et al., 1986); this suggests that over 1/3 of 5-HT1B receptors expressed in the SNr could function as autoreceptors. Heterosynaptic function of 5-HT1B receptors on presynaptic sites in the SNr has been well-documented and may yield important therapeutic findings (Sari, 2004), but its functionality as an autoreceptor in the SNr remains controversial. Iravani and Kruk (1997) found no effects of 5-HT1B receptor antagonists on electrically-evoked serotonin concentrations in SNr slice preparations (Iravani and Kruk, 1997). However, Threlfell et al. (2010) report that these autoreceptors influence short-term depression of serotonin efflux. In paired stimulation trains, the concentration of serotonin evoked by the second stimulation (S2) reached 30% of that evoked by the first stimulation (S1) when there was a 1 second delay between S1 and S2. Antagonists of 5-HT1B receptors relieved this depression by up to 20% (Threlfell et al., 2010). 5-HT1B autoreceptors are thus apparently functional in the SNr, although their modulatory effects may be less robust than in other brain regions.

It is possible that the role of autoreceptors could be better elucidated by in vivo voltammetric studies, where endogenous serotonin tone is undisturbed and autoreceptor function is closer to normal physiological levels. However, a limited number of studies currently address the effects of serotonin's autoreceptors in vivo. In practice, it is difficult to selectively target 5-HT1-type receptors on serotonin terminals when pharmacologically indistinct 5-HT1-type heteroreceptors are expressed throughout the brain. As with in vivo microdialysis, the direct roles of the autoreceptor would be difficult to extricate from indirect modulation by in situ circuitry. Recent technological advancements in iontophoresis enable spatially-resolved, quantitative drug delivery at the site of voltammetric measurements (Herr and Wightman, 2013). Future studies using FSCV combined with this drug-delivery method have great potential to answer important questions about serotonin's autoreceptors.

In addition to their inhibitory influence on release, serotonin's autoreceptors appear to modulate response to SERT inhibition. A number of studies report that autoreceptor antagonists can potentiate the rise in extracellular serotonin levels elicited by SSRIs (Hjorth, 1993; Artigas et al., 1994), and 5-HT1A autoreceptors also mediate reduction of firing rate by SSRIs in the DRN (Gartside et al., 1995). The concentration change evoked by POP stimulations, deliberately rapid enough to avoid creating an endogenous tone, typically does not activate autoreceptors and is thus not affected by their antagonists. However, in brain slices of the DRN, paroxetine-induced increases in serotonin efflux were potentiated by 5-HT1A and 1B/D receptor antagonists (Davidson and Stamford, 1995a). Therefore, it is hypothesized that SERT inhibition causes an increase in extracellular serotonin levels sufficient to activate autoreceptors, even in brain slices. This produces an inhibitory tone, such that autoreceptor antagonists can further unmask SSRI-induced increases in release. 5-HT1B and 1D autoreceptors appear to similarly potentiate the effects of SSRIs in distal brain regions, as the Stamford lab also reports increases in paroxetine's effects in the vLGN when co-administered with 5-HT1B and 1D receptor antagonists (Davidson and Stamford, 1997). The interaction between regulation of release and uptake functions may also be an important detail in understanding how chronic uptake inhibition functions in treating depressive disorders.

The gap between onset of acute physiological effects and the therapeutic efficacy achieved in a chronic treatment period implies that SSRI-induced increases in serotonin levels are not directly producing antidepressant effects. Instead, elevated serotonin levels may influence long-term changes in serotonin signaling and its downstream targets to relieve symptoms of depression (Blier et al., 1987). The effects of long-term SERT inhibition are conflicting: some find increases in extracellular serotonin levels, and some find no changes. Associated with these outcomes are variable reports of autoreceptor desensitization or hypersensitization of 5-HT1A and 1B autoreceptors (Chaput et al., 1986; Invernizzi et al., 1992, 1995; Bosker et al., 1995a,b; Moret and Briley, 1996). Studies examining the effects of chronic SSRI treatment using FSCV have produced more consistent findings.

FSCV measurements of serotonin signaling after 21 days of SSRI exposure reveal that rate of clearance, measured by t_1/2, is unchanged by this treatment. This lack of change is intriguing because radioligand binding studies report brain-wide reductions in SERT density after chronic inhibition (Kovacevic et al., 2010). It may reflect compensation by other clearance mechanisms, such as low affinity serotonin transporters. High and low affinity transport systems have been described for other monoamine neurotransmitters (Iversen, 1974; Stamford et al., 1984, 1990; Hagan et al., 2011). Although studies suggest that these transporters may play an important role in serotonin signaling (Daws, 2009), there are presently no FSCV studies describing their role in modulating release and uptake. The role of non-selective uptake transporters in modulating serotonin signaling, particularly following chronic SSRI treatment, would be interesting to investigate using FSCV.

Long-term SSRI treatment increases stimulation-evoked serotonin concentrations in the DRN and other brain regions by 20–100%, depending on the experiment and brain region studied (O'Connor and Kruk, 1994; Davidson and Stamford, 1998, 2000). These findings concur with the results of Schoups et al. (1986), who found that electrically-evoked release of tritiated serotonin (³[H]5-HT) in the hypothalamus increased after 21 days of SSRI treatment (Schoups et al., 1986). Although increases in serotonin efflux are observed after acute SERT inhibition, these can be explained by changes in rate of uptake. However, t_1/2 was not altered in any voltammetric investigation of long-term SSRI treatment. Therefore, increases in evoked concentrations induced by chronic treatment must rely on another mechanism. Changes in other aspects of release may contribute to this effect, for example: the quantity or composition of serotonin stored in vesicles, regulation of intracellular calcium, or excitability of the synaptic membrane. In-depth exploration of these mechanisms has not yet been explored using voltammetric methods.

Alterations in 5-HT1A autoreceptors contribute to the effects of chronic SSRIs on serotonin signaling. Under normal conditions, activated 5-HT1A receptors inhibit serotonin release and neuronal firing rates, and chronic SSRI treatment may modify this activity. Selective suppression of 5-HT1A autoreceptors can produce antidepressant behavioral effects in the absence of SSRIs (Bortolozzi et al., 2012). Many investigations have described functional desensitization of 5-HT1A receptors after chronic SERT inhibition, but to varying degrees across brain regions (Kreiss and Lucki, 1994, 1997; Cremers et al., 2000; Bosker et al., 2001; Rossi et al., 2008). Davidson and Stamford (1998) compared serotonin release and uptake and neuronal firing rates in the DRN of rats treated with water or paroxetine for 21 days. Paroxetine-treated rats had significantly higher serotonin release rates but exhibited no differences in firing rate. Interestingly, application of a 5-HT1A receptor agonist revealed that firing rate was less sensitive, and release amplitude more sensitive, to this manipulation. Contradictory findings of 5-HT1A receptor sensitivity was not a total surprise: prior studies found similar desensitization of 5-HT1A receptors in the control of firing rate after chronic paroxetine treatment (Chaput et al., 1986; Blier et al., 1988, 1990), and O'Connor and Kruk (1994) had previously reported sensitization of 5-HT1A receptors controlling release amplitude. The dichotomous effect of chronic SSRIs on 5-HT1A receptor sensitization indicates a functional distinction between the receptors mediating neuronal firing and those controlling release. Given the complex effects of chronic SSRIs on 5-HT1A autoreceptors in the DRN, it would be interesting to see how these changes translate to serotonin release in an intact brain. Currently, however, no studies employing FSCV have examined the effects chronic SERT inhibition in vivo.

5-HT1B and 1D receptors also desensitize after chronic SERT inhibition, although the extent to which this occurs appears to vary between brain regions. O'Connor and Kruk (1994) reported desensitization of 5-HT1B receptors in SCN after chronic treatment with fluoxetine. In contrast, the Stamford lab found no changes in the sensitivity of 5-HT1B receptors in the vLGN, instead finding desensitization of 5-HT1D receptors after chronic paroxetine. This inconsistency may reflect differences in autoreceptor expression in the SCN and vLGN, or result from difficulty in selectively targeting the 5-HT1B receptor pharmacologically (O'Connor and Kruk do not address the effects of 5-HT1D receptors in their study). Additionally, while O'Connor and Kruk (1994) found no desensitization of 5-HT1B receptors in the DRN, Davidson and Stamford (2000) later demonstrated that 5-HT1B receptor desensitization was apparent only when the 5-HT1A autoreceptor is antagonized (Davidson and Stamford, 2000). This adds further weight to the conjecture that 5-HT1A and 1B receptors functionally interact in the DRN.

Metabolic degradation of serotonin by the enzyme MAO also contributes to serotonin clearance, especially in the developing brain (Cases et al., 1995, 1998). However, MAO inhibition in brain slices has no reported effect on release amplitudes or uptake (O'Connor and Kruk, 1991a), a finding used to confirm absence of serotonin's metabolites from the voltammetric signal. Owesson et al. (2002) showed a greater role of MAO in regulating serotonin efflux using transgenic mice lacking MAO-A expression. MAO-A is the isoenzyme that preferentially degrades norepinephrine, epinephrine, dopamine, and serotonin, and mice lacking this enzyme have decreased neuronal firing rates in the DRN and increased extracellular serotonin levels (Evrard et al., 2002). In brain slices of the DRN, MAO-A-deficient mice displayed significantly greater serotonin efflux and reduced clearance rates compared to wild-type controls. Additionally, the effects of citalopram were smaller and radioligand binding showed significantly lower expression of SERT in these mice (Owesson et al., 2002). This suggests that serotonin signaling is subject to regulation by MAO under the right experimental conditions. In vivo work supports this idea, as a recent study has shown that MAO inhibitors dramatically increase serotonin efflux in the SNr (Hashemi et al., 2012). MAO inhibition also has a much greater effect on serotonin than dopamine efflux when compared in vivo, suggesting a unique role for metabolic degradation in the regulation of serotonin transmission compared to other monoaminergic systems.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.