Three serotonin response types in vivo
Serotonin reuptake is a major focus of antidepressant agents. We previously reported stimulated [serotonin] vs. time in the SNr (Wood and Hashemi 2013) as a composite average of five experiments. Upon accumulation of more datasets, however, it became apparent that responses are heterogeneous, and averaging removes nuances that provide important information about serotonin neurochemistry. We found three distinct serotonin responses to a standard stimulation, primarily differentiated by the clearance slopes. Michael and colleagues found dopamine heterogeneity in the rising portion of extracellular concentration curves and proposed the terminology, slow, fast, and hybrid (Moquin and Michael 2009, 2011), which we adopted here. For serotonin, all three responses have a rapid rise. Fast responses are characterized by a rapid return to baseline, and slow responses are characterized by a more gradual return to baseline. Hybrid responses have both fast and slow attributes because they descend rapidly for a short time and then switch to slow decay.
Differences between electrode kinetics could account for erroneous assignment of our responses. We explore this in Fig. 4. Here, serotonin (1 μM) was injected in vitro onto eight electrodes; the responses are shown averaged with SEM. While there are differences in the response amplitude between electrodes, the difference in electrode kinetics is negligible (evidenced by the small error in the initial rising portion of the response shown between the two vertical green dashed lines). Therefore, it is likely that in vivo processes underlie our three response types.
Figure 4. In vitro fast scan cyclic voltammetry (FSCV) response to serotonin (1 μM) in a flow injection analysis (FIA) cell (n = 8 electrodes ± SEM). The green bar indicates the duration of serotonin injection. The two vertical green dashed lines indicate the rising response portion of the signal.
Download figure to PowerPoint
Two serotonin reuptake mechanisms
Visual inspection of our three serotonin response types shows two separate clearance slopes, suggesting involvement of two discrete reuptake mechanisms. Simple t1/2 analysis did not allow us to distinguish between hybrid and slow responses; therefore, we sought to employ kinetic models to determine any differences. However, we could not model our responses with models established for serotonin release in tissue slice preparations (Bunin et al. 1998). We found that incorporation of two separate reuptake mechanisms into our model, Vmax1;Km1 and Vmax2;Km2 allowed us to closely model our experimental data. Local stimulations in tissue slice preparations create a massive efflux of serotonin (Bunin and Wightman 1998; Bunin et al. 1998). In previous tissue slice experiments, serotonin clearance was apparently dominated by a reuptake mechanism that kinetically mirrors our values for Vmax2;Km2 while contributions from Vmax1;Km1 were reasonable to neglect. It is interesting that close inspection of [serotonin] versus time traces in these previous experiments (Bunin et al. 1998) shows that at low concentrations (low nM) the experimental data deviate from established models. At these low concentrations serotonin begins to decay slowly, in a similar way to our slow responses, described more aptly by Vmax1;Km1.
Snyder and colleagues suggested that serotonin clearance occurred via two reuptake mechanisms (Shaskan and Snyder 1970). They proposed Uptake 1 with high affinity and low efficiency and Uptake 2 with low affinity and high efficiency. Daws and colleagues verified pharmacologically that Uptake 1 is likely to occur primarily via the SERTs on serotonergic neurons and that Uptake 2 includes other transporters on other cells including the dopamine transporter, the norepinephrine transporter (NET), and the organic cation transporter (OCT) (Daws et al. 2013; Horton et al. 2013). Here, for the first time, we present endogenous in vivo data to support the concept of Uptake 1 and Uptake 2. Indeed our values for Vmax1:Km1 (17.5 nM s−1 and 5 nM) and Vmax2;Km2 (780 nM s−1 and 170 nM) agree remarkably well with high affinity, low efficiency uptake (Uptake 1), and with low affinity, high efficiency uptake (Uptake 2) respectively.
In vivo serotonin release is known to be highly regulated, and [serotonin]evoked is in the low nM range (Hashemi et al. 2009, 2011a, 2012; Wood and Hashemi 2013). Furthermore, it has been demonstrated that inhibiting serotonin reuptake and metabolism (with an SSRI and MAOI) leads to the potentially fatal serotonin syndrome (Hashemi et al. 2012). Therefore, it is not remarkable for multiple reuptake mechanisms to be charged with clearing serotonin from the synapse. It is probable that physiologically released serotonin is at low enough concentrations such that low efficiency, high affinity SERTs on serotonergic neurons, Uptake 1, can reuptake serotonin effectively. However, if serotonin release exceeds a certain limit, it may diffuse to other transport mechanisms, which are not as selective for serotonin and therefore have low affinity, but operate at high efficiency (Uptake 2).
To probe the effects of a commonly prescribed SSRI mechanistically, we administered ESCIT at a high dose. We chose to administer 100 mgkg−1 and compare data taken at 120 min after drug administration based on previous dose–response experiments that showed maximal and lingering effects with this dose and at this time (Wood and Hashemi 2013). Figure 2 shows the experimental data and corresponding models. SSRI administration substantially increased serotonin release and decreased its clearance, as previously seen (Wood and Hashemi 2013). This is not surprising since ESCIT is highly selective for the SERTs (Uptake 1). However, after considerable experimentation we found that our data were best fit with a model that included 95% inhibition of Uptake 1 and 40% inhibition of Uptake 2. This is not surprising given that there is evidence that SSRIs have affinity for Uptake 2 transporters. For example, ESCIT has been found to block NETs and have a significant effect upon OCT-sensitive serotonin uptake (Nguyen et al. 2013). Furthermore, many SSRIs inhibit the human plasma membrane monoamine transporter, also an Uptake 2 transporter (Haenisch and Bonisch 2010). Finally, Horton et al. (2013) found that, in the presence of fluvoxamine, blockage of Uptake 2 by Decynium-22 greatly raised both the extracellular 5-HT level and the clearance time.
It is important to note that the high dose of ESCIT in our experiment (100 mgkg−1) exceeds the minimal effective dose required for behavioral effects in mice (12 mgkg−1) (Sanchez et al. 2003). Although not likely to be encountered clinically, the high dose enables us to illustrate a central point in this work: that physiological deviations above normal extracellular serotonin concentration are cleared via SERTs, but larger deviations are cleared through combination of the SERTs and Uptake 2 transporters. Since different SSRIs have different chemical compositions, it is reasonable to expect that, at a given concentration, each blocks some percentage of Uptake 1 and a (presumably lower) percentage of Uptake 2. Thus, one would expect that both peak response and clearance time would vary among different SSRIs.
Serotonin autoreceptor regulation of serotonin transmission
FSCV does not determine the baseline or steady-state value of the extracellular serotonin concentration; this is an essential, previously unaccounted for, component of kinetic models for serotonin. The experimental curve for fast response (Fig. 1) descends 10–20 nM below baseline. While we cannot know what the absolute levels are, our data imply that the steady-state concentration of serotonin is between 10 and 20 nM; we therefore assumed [serotonin]baseline in our simulations as 20 nM. After completing simulations, we subtracted 20 nM from the model curves so that we could compare them directly to the experimental curves that are plotted with baseline as 0 nM. The value of 20 nM is not surprising because previous estimations of basal neurotransmitter concentrations (Justice 1993) are now thought to have underestimated the true concentrations (Wang et al. 2010; Owesson-White et al. 2012). In Fig. 1, and in most of our experimental data, [serotonin] dips below this baseline after stimulation has ceased. While a dip below baseline has previously been attributed to pH shifts for dopamine experiments (Venton et al. 2003), comparison of cyclic voltammograms suggests that this dip is, indeed, a substantial reduction in extracellular serotonin.
It was, again, not possible to utilize traditional models to account for this dip, likely because of the inherent differences between serotonin regulation in tissue slice preparations and in vivo. Autoreceptors are known to inhibit serotonin release (Barnes and Sharp 1999), in particular, prior FSCV studies in tissue slice preparations and chronoamperometry studies in synaptosomes and in vivo have uncovered important, discrete roles for different autoreceptor subtypes (Daws et al. 1999, 2000; Hopwood and Stamford 2001; Roberts and Price 2001; Threlfell et al. 2010; Hagan et al. 2012). In tissue slice experiments autoreceptors likely function differently than in in vivo because the cell body-terminal connections are severed. In vivo, in our circuitry, serotonin released from the DRN cell bodies stimulates 5-HT1A autoreceptors and serotonin released in SNr acts on 5-HT1B autoreceptors (Barnes and Sharp 1999). Therefore, we postulated that our dip below baseline could be autoreceptor mediated. Indeed, the gradually increasing autoreceptor effect in the model captures the experimental data very well. This is novel chemical data that implies ambient autoreceptor effects and the time scale on which they operate.
To experimentally test this autoreceptor hypothesis we employed methiothepin, a non-selective serotonin receptor antagonist with most affinity for the serotonin autoreceptors, to target the multiple autoreceptors that are involved in the DRN-SNr circuitry (Barnes and Sharp 1999). A prior FSCV study in DRN slices showed that combined 5-HT1A and 5-HT1B receptor antagonism produced greater serotonin efflux than targeting either receptor alone (Roberts and Price 2001). In our study, methiothepin greatly increased the t1/2 of clearance of our experimental data and our model could fit the experimental data by removing A(t). This simple, yet effective modeling strategy gives further evidence that autoreceptors may be acting within the timeframe of our collection window (30 s) to reduce serotonin transmission.
The advantage of our model is its simplicity; however, it carries limitations. The product term, R(t)(1-A(t)), cannot distinguish between lowering R(t) and raising A(t). For example, the autoreceptor effect may proceed earlier than 7 s after initiation of the stimulation. Here, we considered R(t) as release in the absence of autoreceptors and (1 – A(t)) as the modification of release when the autoreceptors are stimulated. We assumed that R(t) rapidly increased and decreased in correspondence to the stimulus and A(t) increased gradually thereafter (Fig. 1c). An additional limitation is that we cannot yet distinguish between the different serotonin autoreceptors. Finally, our data imply that basal serotonin levels are around 20 nM; this level needs to be verified independently with a method capable of reporting basal serotonin levels at carbon fiber microelectrodes. Addressing these three limitations requires method development, elaborate pharmacological experiments, a more sophisticated modeling approach (Reed et al. 2012) and is the focus of our future work.
We studied endogenous serotonin release and reuptake with FSCV. We took a novel mathematical approach by treating the data with Michaelis–Menten kinetics that incorporated two reuptake mechanisms, a baseline serotonin concentration, and autoreceptor functions. Experimentally, we discovered three serotonin chemical signatures which we termed fast, slow, and hybrid and mathematically we found that they could be explained with two reuptake mechanisms. We found a high affinity, low efficiency reuptake mechanism (Uptake 1), proposed to be via the SERTs and a low affinity, high efficiency reuptake system (Uptake 2) thought to represent the contribution of dopamine transporters, NETs, and OCTs. In addition, we outlined a timeframe for the inhibitory role of autoreceptors. Combining voltammetric and theoretical approaches gives us an ideal tool to study serotonin's dual-uptake mechanisms and autoreceptor control. This capability will be invaluable for characterizing the mechanisms of the pharmacological effects of existing antidepressant agents and to aide in the design of novel agents.