Non-linear relationship between 5-HT transporter gene expression and frequency sensitivity of 5-HT signals


Address correspondence and reprint requests to Dr Katie Jennings, University Department of Physiology, Anatomy and Genetics, Sherrington Building, Parks Road, Oxford OX1 3PT, UK. E-mail:


J. Neurochem. (2010) 115, 965–973.


Much evidence suggests that variation in expression of the 5-hydroxytryptamine (5-HT) transporter (5-HTT) is linked to risk of psychiatric illness, but the neurobiological basis of this association is uncertain. In this study, we investigated the impact of variation in 5-HTT expression on subsecond fluctuations in extracellular 5-HT concentrations ([5-HT]o). Stimulus-evoked [5-HT]o was detected using fast-scan cyclic voltammetry at carbon-fibre microelectrodes in the substantia nigra in brain slices from 5-HTT knockout (KO) and 5-HTT over-expressing (OE) mice. Compared with wild-type (WT) controls, evoked [5-HT]o was greater in KO and less in OE mice. In WT controls, evoked [5-HT]o was frequency-sensitive; however, in both KO and OE mice, evoked [5-HT]o showed a striking loss of frequency sensitivity. The latter was observed in WT mice after application of a 5-HTT blocker. These data show that while variation in 5-HTT expression modified the peak magnitude of [5-HT]o evoked by any given stimulus in a gene dose dependent manner, there was a non-linear relationship between 5-HTT expression and frequency sensitivity. Overall, the findings suggest that variation in 5-HTT expression has a marked effect on frequency sensitivity which is a fundamental property of normal 5-HT transmission.

Abbreviations used:



5-HT transporter


extracellular 5-HT concentration


carbon fibre microelectrode




DA transporter


fast scan cyclic voltammetry

−/+ KO

heterozygote knockout

−/− KO

homozygote knockout




substantia nigra pars reticulata



The human 5-hydroxytryptamine (5-HT; serotonin) transporter (5-HTT, SERT, SLC6A4) gene plays a critical role in the regulation of extracellular 5-HT, and is one of the most intensively investigated genetic risk factors for psychiatric disorders (e.g. Murphy and Lesch 2008). Evidence suggests that 5-HTT levels vary naturally by up to sevenfold between individuals (Lundberg et al. 2006), due in part to the existence of several 5-HTT gene variants (Lesch et al. 1996; Ogilvie et al. 1996; Zoroglu et al. 2002; Ozaki et al. 2003; Hu et al. 2006). Studies of the 5-HTT gene-linked polymorphic region of the 5-HTT gene promoter demonstrated that low 5-HTT-expressing alleles resulted in increased neuroticism scores (Lesch et al. 1996), increased anxiety-related responses to fearful stimuli (Pezawas et al. 2005), and increased risk of depression when accompanied by stressful life events (Caspi et al. 2003, 2010), whilst the high-expressing allele protected from these phenotypes.

However, subsequent studies have reported that gain-of-function 5-HTT gene variants are also associated with disabling psychiatric phenotypes, for example, obsessive–compulsive disorder, comorbid autism-related psychopathology and attention deficit disorder (Zoroglu et al. 2002; Ozaki et al. 2003; Hu et al. 2006). Together, these different findings suggest that there is a non-linear relationship between 5-HTT expression and detrimental behavioural phenotypes.

How both low and high 5-HTT expression are associated with undesirable behavioural disturbances remains poorly understood. Studies of the impact of 5-HTT expression on 5-HT transmission suggest a linear gene-dose dependent effect on brain levels of [5-HT]o. 5-HTT knockout (KO) mice show decreased clearance of extracellular 5-HT (Montanez et al. 2003), and correspondingly increased basal and potassium-evoked extracellular 5-HT ([5-HT]o) in in vivo microdialysis studies (Shen et al. 2004), and there are similar findings in 5-HTT KO rats (Olivier et al. 2008). Heterozygote 5-HTT KO mice had 5-HT clearance and [5-HT]o levels intermediate between homozygotes and wild-type (WT) controls (Montanez et al. 2003; Mathews et al. 2004). In contrast to 5-HTT KO mice, transgenic 5-HTT over-expressing (OE) mice have decreased basal and potassium-evoked levels of [5-HT]oin vivo (Jennings et al. 2006).

However, the insights provided by these studies to date are limited. These studies have identified effects on 5-HT transmission with only low temporal resolution, that is, on a timescale that is not commensurate with the dynamic changes in firing rates of 5-HT neurons which range from low frequency tonic firing of 0.5–10 Hz to brief, high frequency burst firing of up to 100–200 Hz (Hajos and Sharp 1996; Hajos et al. 2007). If we are to understand the impact of 5-HTT variation on 5-HT function, it is important to appreciate how 5-HT release is influenced dynamically during different physiological firing frequencies. Here, we used fast-scan cyclic voltammetry (FCV) to explore the subsecond control of evoked [5-HT]o under conditions of altered 5-HTT expression in heterozygote (−/+) and homozygote (−/−) 5-HTT KO mice, as well as 5-HTT OE mice. Our data demonstrate that variation in 5-HTT expression impacts not only on the absolute magnitude of [5-HT]o evoked by any given stimulus, but also on frequency sensitivity of 5-HT transmission.

Materials and methods


5-HTT KO and OE mice were generated as previously described (Bengel et al. 1998; Jennings et al. 2006). Male genetically modified mice and WT littermates (3–9 months, 3–6 mice per genotype per data set) were used in each experiment. Animals were kept on a 12 h light : dark cycle (lights on at 7 am) and had free access to lab chow, water and nesting materials. All animals were kept and killed in accordance with local rules and Home Office guidelines.

Slice preparation

After cervical dislocation, mouse brains were dissected into ice-cold HEPES ringer containing (in mM) 120 NaCl, 5 KCl, 20 NaHCO3, 6.7 HEPES acid, 3.3 HEPES salt, 2 CaCl2, 2 MgSO4, 1.2 KH2PO4 and 10 glucose saturated with 95% O2/5% CO2. Slices (300 μm) containing the substantia nigra were cut in ice-cold HEPES ringer and maintained in HEPES ringer for ≥ 1 h at 21°C. Approximately 30 min before the experiment slices were transferred to a bath and superfused at a flow rate of ∼1.5 mL/min with a bicarbonate-buffered artificial CSF maintained at 30–32°C ((in mM) 124 NaCl, 3.7 KCl, 26 NaHCO3, 2.4 CaCl2, 1.3 MgSO4, 1.3 KH2PO4, and 10 glucose saturated with 95% O2/5% CO2).

Fast scan cyclic voltammetry

Electrically evoked [5-HT]o was monitored using FCV in the substantia nigra pars reticulata (SNr). Recordings of [5-HT]o were made using a carbon fibre microelectrode (tip diameter ∼7 μm, length 20–50 μm, 100 μm tissue depth) as described previously (Cragg et al. 1997; Threlfell et al. 2010). In brief, voltammograms were obtained using a triangular scan waveform (−0.7 to +1.3 V to −0.7 V versus Ag/AgCl) at a scan rate of 800 V/s, switching out of circuit (0 V) between scans (8 Hz scan frequency). This scan waveform is an alternative to the ‘N’ waveform also used to detect 5-HT (see Threlfell and Cragg 2007 for full discussion) and allows detection of both of the 5-HT reduction currents, minimises adsorption of 5-HT to the electrode by switching to non-negative holding potentials between scans (0 V) and gives rise to fewer artefacts resulting from switching in polarity compared to ‘N’ waveforms (see Threlfell and Cragg 2007). The sensitivity to 5-HT of each electrode was assessed by post-experiment calibration in 0.5 μM 5-HT in artificial CSF at 32°C. Electrode sensitivity to applied [5-HT] is linear over a range including 10–1000 nM (data not illustrated). Electrodes are placed in tissue for a minimum of 30 min prior to the start of recording each day, to allow any effects on electrode sensitivity of exposure to the tissue macromolecular environment to occur prior to data acquisition.

Release of 5-HT was evoked electrically using a concentric bipolar electrode positioned on the tissue surface ∼50–100 μm from the recording electrode. Stimulus pulses (0.2 ms duration; 0.67 mA) were applied out-of-phase with FCV scans in trains of 20 pulses at 10, 20, 50 and 100 Hz at intervals of 5 min.

Data analysis and statistics

Data were acquired and analysed using Strathclyde Whole Cell Program (University of Strathclyde, Glasgow, Scotland, UK) and Microsoft Excel. 5-HT oxidation currents were measured from background-subtracted voltammograms and converted into [5-HT]o concentration using the calibration factor obtained for each electrode. [5-HT]o was plotted against time to give a [5-HT]o profile for each stimulation. [5-HT]o evoked in response to 20-pulse 50 Hz stimulation was measured in several sites in each slice prior to assessing frequency sensitivity to confirm that effects of 5-HTT expression on magnitude of [5-HT]o was not caused by site bias (data not shown).

Sample size (n) refers to number of experiments. No more than two experiments were performed on tissue from each animal, and number of animals used ranged from a minimum of three, up to nine, for each experiment. As release can vary between sites and slices to a similar extent as between animals, all experiments from different slices were treated independently for the purposes of statistical analysis.

Each stimulation type was performed in triplicate within each experiment and averaged to produce a [5-HT]o profile per stimulation from each experiment. Profiles from all experiments were then averaged to give an overall mean [5-HT]o profile for each stimulation type. Mean peak [5-HT]o± SEM values (n = no. of experiments, not number of stimulations) were determined from this mean [5-HT]o profile and taken into statistical analysis. For ease of illustration of frequency sensitivity, [5-HT]o profiles were expressed as a percentage of mean peak [5-HT]o evoked by 10 Hz. Statistical analyses were performed using SPSS v16 (SPSS Inc. an IBM company, Chicago, Illinois, USA). Unless otherwise stated, statistics reported in text refer to analyses of raw data. Analyses comprised two-way anova of genotype, treatment and frequency for each mouse cohort (i.e. KO, HET, WT and OE, WT); frequency (OE data) or treatment and frequency (KO data) were treated as repeated measures. Post hoc testing was only carried out for significant interactions and was performed using simple effect analysis followed by pairwise comparisons (Fisher’s LSD) as necessary and appropriate.


Effect of loss of 5-HTT expression on electrically evoked [5-HT]o transients

In the SNr of WT mice, electrically evoked [5-HT]o was stable and reproducible for > 3 h and abolished by local application of 1 μM tetrodotoxin, indicating dependence on action potentials (Figure S1). Evoked 5-HT currents were identifiable as 5-HT because of peak oxidation current at + 500–600 mV and dual reduction currents at ∼0 and −600 mV, as seen for 5-HT applied in calibration solutions (Fig. 1). Recordings made in the very dorsal portion of the SNr regularly yielded pure 5-HT signals (Fig. 1), in agreement with another study of mouse SN (John et al. 2006) but any sites exhibiting a reduction current peak potential of a value between those of the dual reduction peak potentials of 5-HT were not included for analysis because this is suggestive of co-detection of a catecholamine, for example, dopamine (DA). Electrochemical signals were confirmed further as 5-HT because they were enhanced by 5-HTT inhibitors (in WT, 5-HTT −/+ and OE mice, see later).

Figure 1.

 Evoked voltammograms can be identified as 5-HT. Representative background-subtracted voltammograms at peak response recorded during calibration with 500 nM 5-HT and in slices containing SNr from homozygote (KO) and heterozygote (HET) knockout mice and wild-type littermates (WT1), and 5-HTT over-expressing (OE) and their wild-type littermates (WT2). Dashed lines illustrate the two reduction peaks characteristic of 5-HT and the solid line indicates the oxidation peak.

In WT mice, electrically evoked [5-HT]o transients were strongly frequency-sensitive over the range of 10–50 Hz (Figs 2 and 3; frequency F(3,42) = 100, p <0.001 and F(3,72) = 33 p <0.001 respectively, p <0.05 for all Fisher’s LSD comparisons of frequency in WT except 50 vs. 100 Hz). No difference between WT strains was observed in control conditions (strain p >> 0.05 and strain*frequency p >> 0.05). Furthermore, although a slight difference between the two WT strains was noted after citalopram, this was only significant at the lowest stimulus frequency (strain p >> 0.05; strain*frequency F(3,27) = 8.5 p <0.05, WT1 vs. WT2 p <0.05 at 10 Hz p >> 0.05 at all other stimulus frequencies).

Figure 2.

 Decreased 5-HTT expression reduces frequency sensitivity and increases magnitude of evoked [5-HT]o. (a) Averaged profiles ± SEM of stimulus-evoked [5-HT]o versus time in wild-type (WT) and heterozygote (HET) and homozygote 5-HTT knockout (KO) mice. Stimuli (arrows) indicate start of 20-pulse trains (= 5–6). (b) Mean peak [5-HT]o ± SEM taken from evoked [5-HT]o profiles. (c) Mean peak [5-HT]o ± SEM (expressed as percentage of [5-HT]o evoked by 10 Hz) versus stimulus frequency in each genotype. In (b), no genotype*frequency*treatment interaction; genotype*treatment and genotype*frequency interactions (< 0.05) in analysed data set (control and citalopram data), * indicates results of post hoc testing of genotype*treatment interaction, < 0.05 versus WT in control (Fisher’s LSD). In (c), * and • indicate genotype*frequency*treatment interaction (< 0.05) and < 0.05 for KO versus WT (*) or KO versus HET (•) in control (Fisher’s LSD test).

Figure 3.

 Increased 5-HTT expression reduces frequency sensitivity and magnitude of evoked [5-HT]o. (a) Averaged profiles ± SEM of stimulus-evoked [5-HT]o versus time in wild-type (WT) and 5-HTT over-expressing (OE) mice. Stimuli (arrows) indicate start of 20-pulse trains (= 5–6). (b) Mean peak [5-HT]o ± SEM taken from evoked [5-HT]o profiles. (c) Mean peak [5-HT]o ± SEM (expressed as percentage of [5-HT]o evoked by 10 Hz) versus frequency in each genotype. For both (b) and (c), * indicates genotype*frequency*treatment interactions (< 0.05) and < 0.05 versus WT in citalopram (Fisher’s LSD test). Insert illustrates the correlation between 5-HTT expression (as a % of wild-type) and the maximum mean peak [5-HT]o ± SEM seen during experiments for each genotype (this was in response to 20 pulses at 50 Hz in all genotypes).

The effect of loss of 5-HTT levels on electrically evoked [5-HT]o was examined in the SNr of −/+ and −/− 5-HTT KO mice. In −/− 5-HTT KO mice, evoked [5-HT]o was of greater magnitude and duration than in −/+ 5-HTT KO mice, which was greater than WT mice (Fig. 2a and b; genotype F(2,13) = 25, p <0.001 genotype*treatment F(2,13) = 10 p <0.01, in control p <0.05 for WT vs. HET, WT vs. KO and HET vs. KO; genotype*treatment*frequency p >> 0.05). Strikingly, evoked [5-HT]o demonstrated a profound loss of normal frequency sensitivity in −/− 5-HTT KO mice (Fig. 2c; genotype*frequency F(6,39) = 110, p <0.001, simple effect of frequency in all genotypes p <0.05). Although this loss was not significant in raw data, probably as a result of the divergent [5-HT]o between genotypes, it was clearly apparent when data were expressed as a percentage of 10 Hz (genotype*frequency F(6,39) = 110, p <0.001 and genotype*frequency*treatment F(6,39) = 8.0 p <0.001 simple effect of frequency in WT and HET p <0.05 and in KO p >> 0.05) In comparison, the frequency sensitivity of −/+ 5-HTT KO mice was not different to WT controls (Fig. 2c).

Effect of gain of 5-HTT expression on electrically evoked [5-HT]o transients

The effect of gain of 5-HTT levels on electrically evoked [5-HT]o was examined in the SNr of 5-HTT OE mice. The magnitude of electrically evoked [5-HT]o was markedly less in 5-HTT OE compared with WT mice (Fig. 3a and b; genotype F(1,24) = 35, p <0.001, genotype*frequency F(3,72) = 7.6 p <0.001, genotype*frequency*treatment F(3,72) = 3.0 p < 0.05 WT vs. OE in control p <0.05 for all frequencies except 10 Hz). Importantly, as observed in the 5-HTT KO mice, evoked [5-HT]o in 5-HTT OE mice demonstrated a clear-cut loss of frequency sensitivity (Fig. 3c; frequency F(3,72) = 33 p <0.001 frequency*genotype*treatment F(3,72) = 3.0, p <0.05; simple effect of frequency in OE control p >> 0.05 and in WT control p <0.001), as observed in −/− 5-HTT KO mice.

Effect of 5-HTT blockade on electrically evoked [5-HT]o transients

To investigate whether the loss of frequency sensitivity of evoked [5-HT]o in the −/− 5-HTT KO and OE mice reflects simply the outcome of a reduced or enhanced rate of 5-HT removal from the extracellular space on the extracellular summation of [5-HT]o, the effect of pharmacological inhibition of the 5-HTT by citalopram (100 nM) was tested. In WT and −/+ 5-HTT KO mice, citalopram evidently prolonged the extracellular lifetime of 5-HT, and significantly increased the magnitude of evoked [5-HT]o to levels of [5-HT]o equivalent to those seen in −/− 5-HTT KO mice (Fig. 4a, compare Fig. 2a; treatment F(1,14) = 72, p <0.001, genotype*treatment F(2,13) = 10, p <0.01, simple effect of genotype in control p <0.05 and in citalopram p >> 0.05; WT vs. HET, WT vs. KO and HET vs. KO in citalopram p >> 0.05). Furthermore, although citalopram evidently reduced the frequency sensitivity of the evoked [5-HT]o in WT and HET mice (Fig. 4C compare Fig. 2c, treatment*frequency F(3,39) = 9.7, p <0.001), a significant effect of frequency remained in both the simple effect analysis of either raw or normalized data (p <0.05 for all frequency comparisons). Citalopram had no effect on evoked [5-HT]o in −/− 5-HTT KO mice (Fig. 4a–c, compare Fig. 2a–c genotype*treatment F(2,13) = 10, p <0.01, Simple effect of treatment in WT p <0.05, HET p <0.05 but KO p >> 0.05).

Figure 4.

 5-HTT inhibition modifies frequency sensitivity of [5-HT]o. (a, d) Averaged profiles ± SEM of stimulus-evoked [5-HT]o versus time (a) in wild-type (WT) and heterozygote (HET) and homozygote 5-HTT knockout (KO) mice and (d) wild-type (WT) and 5-HTT over-expressing (OE) mice in the presence of the 5-HTT blocker, citalopram (100 nM). Stimuli (arrows) indicate start of 20-pulse trains (= 5–6). (b, c, e, f) Mean peak [5-HT]o ± SEM taken from evoked [5-HT]o profiles expressed as raw data (b, e) and, as a percentage of [5-HT]o evoked by 10 Hz (c, f), from WT, HET and KO mice (top panel) and WT and OE mice (bottom panel). In (c) and (f), wild-type control data from Figs 2 and 3 (grey line) is illustrated for comparison purposes. In (b) no genotype*frequency*treatment interaction; genotype*treatment interaction (< 0.05) but >> 0.05 for HET versus WT, KO versus WT and HET versus KO after citalopram. (c) * indicates genotype*frequency*treatment interaction (< 0.05) and < 0.05 versus both HET and WT (Fisher’s LSD test). In (e), * indicates genotype*treatment*frequency (p < 0.05) and < 0.05 versus WT (Fisher’s LSD test). In (f), no significance difference between genotypes was observed was observed.

In 5-HTT OE mice, citalopram increased the magnitude of evoked [5-HT]o (Fig. 4d compare Fig. 3a; treatment F(1,24) = 89, p <0.001), although not to the level of WT mice (genotype*treatment F(1,24) = 6.5, p <0.05; simple effect of genotype in control p <0.05, in citalopram p <0.001; WT vs. OE p <0.05 at all frequencies after citalopram). Citalopram reduced the effect of frequency on [5-HT]o (treatment*frequency F(3,72) = 4.2, p <0.01), although a significant effect of frequency remained after citalopram treatment (simple effect of frequency after citalopram p <0.001). Although when expressed as a percentage of 10 Hz, the effect of citalopram on reducing frequency sensitivity of [5-HT]o in WT mice was significant (Fig. 4f, compare Fig. 3c; frequency*genotype*treatment F(3,72) = 3.9, p <0.05, simple effect of frequency in WT before citalopram p <0.001 and after citalopram p >0.05). Moreover, the drug did not reverse the loss of frequency sensitivity of [5-HT]o in the 5-HTT OE mice (Fig. 4d) (effect of frequency after citalopram in OE p >> 0.05).


Much evidence suggests that both loss- and gain-of-function alleles of the human 5-HTT gene confer vulnerability to psychiatric illness and disabling behavioural traits. However, the impact of this variation in 5-HTT expression on neurochemical signalling at the level of 5-HT release sites is little explored. The present study utilised mice with genetically altered 5-HTT expression levels in combination with FCV detection of [5-HT]o, to investigate the effect of variation in 5-HTT expression on 5-HT signalling in real-time. Our data show that loss of 5-HTT levels caused an increase in the levels of [5-HT]o evoked by brief electrical stimulation whereas a gain of 5-HTT levels caused a decrease. Strikingly, both a complete loss and large magnitude gain of 5-HTT levels resulted in insensitivity to stimulation frequency. The latter finding is important in the context of evidence that 5-HT neurons spontaneously fire action potentials over a wide range (0.5–200 Hz) of frequencies (Hajos and Sharp 1996; Allers and Sharp 2003; Kocsis et al. 2006; Hajos et al. 2007).

In the current study, electrically evoked [5-HT]o transients were measured by FCV in the SNr in brain slices. The SNr provides an excellent model of 5-HT terminals in that this region receives amongst the highest density of 5-HT terminals seen in the CNS (Moukhles et al. 1997) and correspondingly, expresses the highest levels of 5-HTT protein in the mouse brain (Perez et al. 2006). Earlier FCV studies of [5-HT]o transients in rats and guinea pigs have validated the SNr as a neurochemical model of 5-HT synapses, and demonstrated that there is minimal contamination of electrochemical signals by the related monoamines DA and noradrenaline, which can complicate FCV measurement of 5-HT in other brain regions (Cragg et al. 1997; Bunin et al. 1998; Bunin and Wightman 1998; Threlfell et al. 2004; John et al. 2006; Threlfell et al. 2008). The applicability of the FCV method to the mouse SNr is evident by our findings that electrically evoked [5-HT]o transients were reduced and increased by inhibitors of voltage-gated sodium channels (tetrodotoxin) and a 5-HTT inhibitor (citalopram), which confirm the results of another recent FCV study of [5-HT]o in the mouse SNr (John et al. 2006).

Our data show that the magnitude of electrically evoked [5-HT]o at any given stimulus was greater in 5-HTT KO mice compared to those detected in WT controls, with heterozygote 5-HTT KO mice having [5-HT]o levels intermediate between homozygotes and WT controls. These findings are in accordance with previous in vivo microdialysis studies (Mathews et al. 2004; Shen et al. 2004). Moreover, a recent FCV study reported an increase in the duration of electrically evoked [5-HT]o transients in SNr slices from 5-HTT KO mice (John et al. 2006). In contrast to 5-HTT KO mice, in 5-HTT OE mice the magnitude of electrically evoked [5-HT]o was much smaller than observed in WT mice. This finding is also in accordance with earlier findings from in vivo microdialysis studies showing that regional brain levels of basal and depolarization-evoked [5-HT]o were lower in these mice compared with WT controls (Jennings et al. 2006). All-in-all, these findings suggest that there is a gene-dose dependent effect of 5-HTT expression on the magnitude of 5-HT signals in the brain.

The latter conclusion, however, contrasts with our striking observation that both loss and gain of 5-HTT levels were associated with insensitivity of [5-HT]o to stimulation frequency. Specifically, whereas electrically evoked [5-HT]o was sensitive to stimulation frequency in WT mice, it was insensitive in both 5-HTT OE and −/− 5-HTT KO mice. Interestingly, this frequency sensitivity was maintained in −/+ 5-HTT KO mice. As previous data showed that −/+ 5-HTT KO mice have 50% of WT 5-HTT levels, whereas −/− KO and OE mice have 0% and 200–300%, respectively (Bengel et al. 1998; Jennings et al. 2006), the present data suggest that normal frequency sensitivity can be maintained within a limited range of 5-HTT levels. Our data suggest that a range in 5-HTT levels of more than twofold, attenuates frequency sensitivity at 5-HT release sites. In humans, neuroimaging data show that 5-HTT levels vary naturally by a range as large as sevenfold between individuals (Lundberg et al. 2006).

The frequency sensitivity of stimulus-evoked [5-HT]o observed here in WT mice has been previously reported for 5-HT (Bunin et al. 1998; Bunin and Wightman 1998; John et al. 2006) and also for another monoamine DA for which extracellular levels are proposed to be a function of a dynamic equilibrium between uptake and release (Wightman et al. 1988; Chergui et al. 1994). According to this proposal, the longer the interval between stimulus pulses (i.e. the lower the frequency), the better uptake can reduce extracellular monoamine concentration and thus limit the extracellular summation of monoamine after release by the next stimulus (e.g. Wightman et al. 1988). Applying this well-defined model for DA to 5-HT, the lack of frequency sensitivity observed for both 5-HTT KO and OE mice is in keeping with a simple and predictable shift in the equilibrium between release and reuptake driven by altered 5-HTT expression. Thus, abolishing 5-HT re-uptake through 5-HTT KO would be expected to permit great extracellular summation of [5-HT]o irrespective of interpulse interval. Conversely, greatly increasing 5-HT uptake by 5-HTT over-expression would be expected to limit the [5-HT]o remaining between stimuli even at short interpulse intervals (high frequencies), leading to minimal [5-HT]o summation, irrespective of interpulse interval.

Note that this straightforward interpretation of the effects of loss and gain of 5-HTT expression on 5-HT transmission is in keeping with our observation that acute pharmacological blockade of the 5-HTT by citalopram also reduced frequency sensitivity in WT mice, thereby approaching the phenotype seen in 5-HTT KO. However, our data do not exclude the possibility that adaptation in the 5-HT release process caused by life-long changes in 5-HTT expression contributes to the loss of frequency sensitivity observed in the 5-HTT KO and 5-HTT OE mice. Indeed, our observations that citalopram in WT and 5-HTT OE mice did not fully replicate the KO phenotype, and that the drug failed to restore the magnitude of evoked [5-HT]o to WT levels in OE mice, suggests that underlying adaptations in 5-HT release are present. This idea is further supported by evidence that both 5-HTT KO and OE mice have depleted tissue 5-HT levels (Bengel et al. 1998; Shen et al. 2004; Jennings et al. 2006).

Note that it is unlikely that altered nerve terminal 5-HT(1B) autoreceptor function contributes to the loss of frequency sensitivity observed in 5-HTT KO and OE mice, even though reductions in 5-HT1B autoreceptor expression and function in the SN of 5-HTT KO mice have been reported (Fabre 2000; Shanahan 2009). Previous studies have shown that under the present experimental conditions (i.e. ≤ 5 s stimulus trains) the autoinhibitory influence of 5-HT1B receptor activation by endogenous 5-HT is absent (Iravani and Kruk 1997). In this laboratory, we have found that 5-HT1B receptor-mediated autoinhibition by endogenous 5-HT release can only be observed using paired-pulse paradigms. Moreover, even when this control is unmasked using specific stimulation protocols it remains extremely mild (Threlfell et al. 2010).

It is of interest that over-expression of the DA transporter (DAT), yields similar effects on [DA]o as observed here for the 5-HTT on [5-HT]o. Thus, mice that show a ∼40% increase in plasmamembrane expression of DAT exhibit a 40% decrease in [DA]o measured by microdialysis, and a decrease in evoked [DA]o detected with FCV (Salahpour et al. 2008). By comparison, DAT KO mice exhibit a fivefold increase in basal [DA]o although [DA]o evoked by discrete stimuli is reduced to approximately 25% of WT (Jones et al. 1998, 1999). These findings contrast with those of the current study where 5-HTT KO mice exhibit increased basal and evoked [5-HT]o. This disparity between the effects of DAT versus 5-HTT KO on underlying transmitter availability cannot be explained by different changes in pre-synaptic DA or 5-HT pools as DAT and 5-HTT KO mice both have low tissue levels of DA or 5-HT respectively (Bengel et al. 1998; Gainetdinov et al. 1998; Jones et al. 1998). Although frequency responsiveness has not been assessed in detail, one study demonstrated, in accordance with the findings of the present study that DAT KO results in a loss of sensitivity to high frequency stimuli (Benoit-Marand et al. 2000).

Notably, alterations in the function of post-synaptic 5-HT receptors have been reported in both 5-HTT OE and KO mice (Jennings et al. 2008; Fox et al. 2010), and these changes might compensate for, or offset, changes in pre-synaptic 5-HT function such as net levels of released 5-HT. However, although such alterations might offset the effects of a change in the magnitude of tonic 5-HT levels, they are unlikely to compensate for the loss of ability of 5-HT release to reflect subsecond phasic changes in neuronal activity. In vivo, 5-HT neurons fire at frequencies and patterns ranging from slow (0.5–5 Hz), regular firing to short stereotyped high frequency (100–200 Hz) bursts (Hajos and Sharp 1997; Hajos et al. 2007). The loss of frequency sensitivity of [5-HT]o observed here would translate into an inability of 5-HT release sites to generate an appropriate range of concentration of 5-HT signals in response to these dynamic firing patterns. In turn, this might severely impair information relay and, consequently, behaviour. For example, loss of high frequency bursting in other neurotransmitter systems has been associated with impaired behavioural function (Jeans et al. 2007; Zweifel et al. 2009). Both the 5-HTT KO and OE mice have potentially disabling, behavioural phenotypes which could be explained by aberrant information processing and maladaptation to environmental stimuli; heightened anxiety, loss of fear extinction and increased helplessness in the case of 5-HTT KO mice, and reduced anxiety, decreased fear conditioning and increased impulsivity in the case of 5-HTT OE mice (Jennings et al. 2006; Holmes 2008; Murphy and Lesch 2008; S. J. Line, unpublished data). The findings herein may therefore represent a neural correlate of the dysfunctional information processing and maladaptation displayed by both genotypes. It is important to note however, that whilst loss of frequency sensitivity occurs in both 5-HTT KO and OE mice the neurochemical consequences are distinct. Thus, 5-HTT KO mice exhibit high [5-HT]o regardless of stimulation frequency whereas 5-HTT OE mice exhibit very low [5-HT]o. In conclusion, the current data suggest that dynamic, sub-second signalling by 5-HT is critically dependent on the level of 5-HTT expression. Although variation in 5-HTT expression modified the magnitude of evoked [5-HT]o transients in a gene dose-dependent manner, there was a non-linear relationship between 5-HTT gene dose and frequency sensitivity. Thus, these data predict that both large magnitude loss and gain of 5-HTT expression will impair phasic 5-HT transmission and, in turn, the ability to appropriately relay 5-HT-mediated information through neural networks. Such abnormalities in 5-HT transmission may contribute to an explanation of the association between certain functional variants of the 5-HTT gene and psychiatric phenotypes.


Supported by the Deutsche Forschungsgemeinschaft (Le 629/4-2, SFB 581 and SFB TRR 58/A5 to K.P.L.), a European Community Integrated Network (NEWMOOD; LSHM-CT-2004-503474 to T.S.) and a Royal Society Equipment Grant (to S.J.C). The authors report no conflicts of interest.