Expression of functional inhibitory neurotransmitter transporters GlyT1, GAT-1, and GAT-3 by astrocytes of inferior colliculus and hippocampus

Neuronal inhibition is mediated by glycine and/or GABA. Inferior colliculus (IC) neurons receive glycinergic and GABAergic inputs, whereas inhibition in hippocampus (HC) predominantly relies on GABA. Astrocytes heterogeneously express neurotransmitter transporters and are expected to adapt to the local requirements regarding neurotransmitter homeostasis. Here we analyzed the expression of inhibitory neurotransmitter transporters in IC and HC astrocytes using whole-cell patch-clamp and single-cell reverse transcription-PCR. We show that most astrocytes in both regions expressed functional glycine transporters (GlyTs). Activation of these transporters resulted in an inward current (IGly) that was sensitive to the competitive GlyT1 agonist sarcosine. Astrocytes exhibited transcripts for GlyT1 but not for GlyT2. Glycine did not alter the membrane resistance (RM) arguing for the absence of functional glycine receptors (GlyRs). Thus, IGly was mainly mediated by GlyT1. Similarly, we found expression of functional GABA transporters (GATs) in all IC astrocytes and about half of the HC astrocytes. These transporters mediated an inward current (IGABA) that was sensitive to the competitive GAT-1 and GAT-3 antagonists NO711 and SNAP5114, respectively. Accordingly, transcripts for GAT-1 and GAT-3 were found but not for GAT-2 and BGT-1. Only in hippocampal astrocytes, GABA transiently reduced RM demonstrating the presence of GABAA receptors (GABAARs). However, IGABA was mainly not contaminated by GABAAR-mediated currents as RM changes vanished shortly after GABA application. In both regions, IGABA was stronger than IGly. Furthermore, in HC the IGABA/IGly ratio was larger compared to IC. Taken together, our results demonstrate that astrocytes are heterogeneous across and within distinct brain areas. Furthermore, we could show that the capacity for glycine and GABA uptake varies between both brain regions. Electronic supplementary material The online version of this article (10.1186/s13041-018-0346-y) contains supplementary material, which is available to authorized users.


Introduction
Glycine and GABA mediate synaptic inhibition in matured circuits. Glycinergic and GABAergic inhibition are attributed predominantly to caudal and rostral brain regions, respectively. Nonetheless, both neurotransmitters coexist throughout the whole brain.
In a previous study, we analyzed the expression of functional GlyTs and GATs in astrocytes in the lateral superior olive (LSO)a conspicuous auditory brainstem center whose main inhibitory input is glycinergic after early postnatal development [2,3]. Astrocytes in this nucleus express functional GlyT1, GAT-1, and GAT-3 [19]. To study the region-dependent heterogeneity of GlyT and GAT expression in astrocytes, we chose two systems that contrast the LSO with respect to the utilization of glycine and GABA for inhibitory synaptic transmission: 1) The inferior colliculus (IC) residing in the midbrain belongs to the rostral part of the auditory brainstem and serves as a major hub for processing auditory cues [4,27]. Afferents from all auditory brainstem centers converge in the lateral lemniscal tract (LL) and project to the IC (Fig. 1a) [4,8,28]. The inhibitory part of the tract consists of glycinergic and GABAergic projections [8,[29][30][31]. Accordingly, IC astrocytes can be proposed to express GlyTs and GATs to account for neurotransmitter uptake. GlyT1 expression was found in the IC and attributed to glial cells [11,13,32]. Likewise, GAT-1 and GAT-3 are present in the IC [33,34]. However, GlyTs and GATs in IC astrocytes have not yet been electrophysiologically characterized. 2) The HC is the second system of interest. Whereas its main circuitry is glutamatergic [35,36], inhibitory synaptic transmission arises from GABAergic interneurons [9,10]. In line with this, astrocytes in the stratum radiatum express GAT-3, whereas GAT-1 has been attributed to interneurons [21,37]. Glycine is co-released from glutamatergic terminals and modulates NMDAR-mediated signaling [11,12]. For uptake of released glycine, GlyT1 is expressed in astrocytes and presynaptic terminals [11,[38][39][40]. However, functionality of GlyT1 in HC astrocytes has not been demonstrated prior to this study.
Here we analyzed the heterogeneity of expression and function of inhibitory neurotransmitter transporters in astrocytes from IC and HC. Using wholecell patch-clamp recordings from sulforhodamine 101 (SR101)-labeled astrocytes [19,41,42] and concomitant application of glycine or GABA to provoke transporter activation, together with single-cell reverse transcription (RT)-PCR, our results demonstrate that all IC astrocytes and about half of the HC astrocytes expressed functional GlyT1, GAT-1, and GAT-3. In contrast, GlyT2, GAT-2, and BGT-1 were never found. From our experiments, we can exclude that transporter currents were contaminated by respective ionotropic receptor-mediated currents. As expected, GAT activity was much stronger in HC astrocytes compared to IC astrocytes. Concurrently, our results show that IC and HC astrocytes exhibit heterogeneous properties, which reflect region-specific adaptation to local circuitry.
Glycine and GABA, both 1 mM in ACSF, were applied in two ways: 1) In experiments analyzing the maximal neurotransmitter-induced current and depolarization (Figs. 2 and 4) both transmitters were administered using a peristaltic pump (Reglo, Ismatec) at a rate of 1-2 ml/min. Data were sampled at 100 Hz. We monitored putative changes of membrane resistance (R M ) and series resistance (R S ) every 30 s (≙ 0.033 Hz) using hyperpolarizing test pulses (ΔU = 5 mV) [44]. The resulting currents were sampled at 20 kHz. 2) When pharmacologically isolating transporters mediating membrane currents (Figs. 3a and 5a) neurotransmitters were applied via focal pressure injection (PDES-2 T, NPI; 12 psi). Therefore, a pipette with a resistance of 3-7 MΩ was filled with glycine or GABA and positioned approximately 20 μm apart from the recorded cell [19]. Membrane currents were sampled at 1 kHz. In order to detect additionally shortlasting receptor-mediated changes in membrane conductance during focal application of neurotransmitters (Figs. 3c and 5c, Additional file 2: Figure S2B-E), hyperpolarizing test pulses were applied at 1 Hz (Additional file 2: Figure S2A) and R M of astrocytes and neurons was calculated [44]. All recordings were low-pass filtered at 2.9 kHz. Data were processed and analyzed using "IGOR Pro 6.2" software (Wavemetrics). Measurements were rejected if R S exceeded 15 MΩ.

Single-cell RT-PCR
The patch pipette was filled with 3 μl of intracellular solution. Next, astrocytes were patch-clamped as described in the preceding paragraph. After determination of the I-V relationship, the cytoplasm was sucked into the patch pipette, which was then retracted from the slice. The remaining cell parts were sucked into the patch pipette and the intracellular solution containing the cytoplasm was put into a 50 μl PCR reaction tube containing 3 μl of diethyl pyrocarbonate (0.1%)-treated water (Ther-moFisher Scientific). To avoid degradation by RNAse activity the sample was immediately frozen in liquid nitrogen and stored at −80°C. Samples were rejected, if the patch was unstable during cell extraction or fragments from neighboring cells stuck at the pipette.
For transcription of mRNA into cDNA reverse transcriptase (SuperScript III, 100 U; ThermoFisher Scientific), RNAse inhibitor (RNAseOUT, 40 U; Ther-moFisher Scientific), random hexamers (50 μM, Ther-moFisher Scientific), first-strand buffer (ThermoFisher Scientific), and dithiothreitol (DTT; 10 mM; Thermo-Fisher Scientific) were added to the frozen sample (total volume: 13 μl). Next, RT was performed for 1 h at 37°C. Subsequently, a multiplex PCR was performed to identify transcripts of inhibitory neurotransmitter transporters. MPprimer software [45] was used to create primer sequences (Table 1). Primers were chosen to be located on different exons. Thus, amplification of DNA, which contains exons and introns, would result in larger product length compared to the amplicon of spliced mRNA that could be distinguished after gel electrophoresis. The PCR reaction mix contained: 5× PCR buffer including dNTPs (50 μM; Bioline), Taq Polymerase (4 U, Bioline), 200 nM primers (Eurofins Scientific), 10 μl of the RT reaction product, H 2 O (ad 50 μl, Ampuwa, Fresenius Kabi). Fifty PCR cycles were performed: denaturation for 25 s at 94°C, annealing for 2 min (first 5 cycles) and 45 s (subsequent 45 cycles) at 51°C, and elongation for 25 s at 72°C. Afterwards, a second PCR with nested primers and 40 cycles was conducted: denaturation for 25 s at 94°C, annealing for 2 min (first 5 cycles) and 45 s (subsequent 35 cycles) at 54°C, and elongation for 25 s at 72°C. The second PCR reaction mix contained Platinum Taq Polymerase (1 U, Thermo-Fisher Scientific), 10× PCR buffer (MgCl 2 -free; Thermo-Fisher Scientific), 2.5 mM MgCl 2 (ThermoFisher Scientific), 50 μM dNTPs (Bioline), nested primers (200 nM, Eurofins Scientific), and 2 μl of the first PCR reaction product. Positive controls were performed with mRNA extracted from mouse brainstem by using an mRNA extraction kit (Dynabeads mRNA Purification Kit, Invitrogen; Additional file 3: Figure S3). For negative controls, the patch pipette was positioned close to the tissue in the recording chamber and ACSF was sucked into the pipette. Subsequently, the probe was frozen in liquid nitrogen and used for RT-PCR (Additional file 3: Figure S3). All amplified PCR products were loaded on an agarose gel (1.5%), labeled with 1% ethidium bromide (Carl Roth), and analyzed using a transilluminator (Biometra TI 1). To determine the PCR product length we used a standard DNA ladder (HyperLadder 50 bp, Bioline).
Initial experiments showed that some HC astrocytes were devoid of any target RNA (GlyTs or GATs). To prove successful RNA extraction from HC astrocytes transcripts for the inwardly rectifying K + (Kir) channel 4.1 were detected, which are present in all HC astrocytes [46].

Confocal microscopy
The labeling with SR101 -used for a priori identification of IC and HC astrocytes -and AF568 was documented as described before [41] using a confocal microscope (Leica TCS SP5 LSM: HC PL FLUOTAR 10 × 0.30 DRY; HCX PL APO Lambda blue 63 × 1.4 OIL UV) and LAS AF software. Fluorophores were detected as follows (excitation wavelength/filtered emission wavelength): SR101 (SP5: 561 nm/580-620 nm) and AF568 (561 nm/580-620 nm). To improve the quality of confocal micrographs and reduce background fluorescence, we used a Kalman filter (averaging of four identical image sections). Images were processed using Fiji software [47].

Statistics
Results were statistically analyzed using WinSTAT (R. Fitch Software). Data were tested for normal distribution with Kolmogorov-Smirnov test. In case of normal distribution, results were assessed by one-tailed, paired or non-paired Student's t-tests. In the absence of a normal distribution, results were assessed by Wilcoxon test for paired or U-test (Mann-Whitney) for non-paired data. P represents the error probability, *P < 0.05, **P < 0.01, ***P < 0.001; n represents the number of experiments or cells/slices/animals. In case of multiple comparisons data were statistically analyzed by the tests described above under post hoc Šidák correction of critical values [48]: two comparisons: Fig. 2a Table 2; *P < 0.025, **P < 0.005, ***P < 0.0005; three comparisons: Fig. 2c 3 , Fig. 4c 3 , Additional file 2: Figure  S2B 3 -E 3 , Table 3; *P < 0.017, **P < 0.0033, ***P < 0.0003. Data are provided as mean ± SEM.

Results
Identification of IC and HC astrocytes SR101 labeling is used in many different brain regions to identify astrocytes in acute tissue slices [19,41,42]. We mentioned before that incubation of acute slices with SR101 results in labeling of putative astrocytes in the IC, but we did not electrophysiologically confirm the identity of these SR101 + cells [19]. In IC and HC (CA1, stratum radiatum), SR101-labeled cells comprised a small sized soma with several branching processes. In HC, these cells exhibited strong branching, whereas in IC they appeared to be less complex (Fig. 1b, Fig. 1c, f ), which is typical for that developmental stage.
Interestingly, we never observed a glycine-induced outward current or changes in R M (tested every 30 s ≙ 0.033 Hz) upon the activation of putatively expressed GlyRs (not shown). However, glycine-induced outward currents and R M changes upon short term activation of GlyRs during the first seconds of glycine wash-in might be overlooked due to the relatively slow exchange of ACSF in the recording chamber and concomitant slow rise of the neurotransmitter concentration in combination with receptor desensitization [50]. Thus, bath application of glycine is not a suitable approach to prove the presence of functional GlyRs. Therefore, we designed a new protocol for fast and focal pressure injection of neurotransmitters in combination with a voltage-clamp protocol including a higher frequency of test pulses assessing R M changes now at 1 Hz (Additional file 2: Figure S2A).
Astrocytes, for example in the HC, express GABA A Rs [21,22]. Accordingly, I GABA (max) might be underestimated if GABA A R-mediated Cl − influx causes an outward current that counteracts the GAT-mediated inward current. To verify this, we performed fast and focal pressure injection of GABA and assessed R M changes.
Subsequently, we did focal application and analyzed putative GABA A R-mediated R M changes in IC and HC astrocytes (Fig. 5c). At any time, GABA induced an inward but no outward current (Fig. 5c 1-2 ). In IC astrocytes, the offset current induced by the test pulses did not change (Fig.  5c 1 ). Accordingly, at t 1 R M was not reduced (+1.1 ± 4.4%, n = 15/5/5, P = 0.403; Fig. 5c 3 ). Thus, R M was GABAindependent arguing against an activation of GABA A Rs. In HC astrocytes however, GABA increased the offset current in response to the test pulses (Fig. 5c 2 ). In turn, R M was reduced (t 1 : −7.6 ± 1.1%, n = 8/5/2, P < 0.001; Fig. 4c 3 ) demonstrating activation of GABA A Rs in HC astrocytes. In the prolonged presence of GABA, R M recovered completely, indicating desensitization of GABA A Rs (t 10 : −2.4 ± 1.3% of resting R M , P = 0.055 compared to t 0 ; P < 0.001 compared to t 1 ; Fig. 5c 3 ). Thus, I GABA (max) was not contaminated by GABA A R activation as it was not determined within the first 10 s of GABA wash-in. Taken together, IC and HC astrocytes co-expressed functional GAT-1 and GAT-3, whereas GABA A Rs were only found in HC astrocytes. Data are summarized in Tables 2 and 3.
In summary, our results show that SR101-labeled cells in the IC and HC exhibited properties of classical astrocytes. In all IC and about half of the HC astrocytes, GlyT1, GAT-1, and GAT-3 were present, whereas GlyT2, GAT-2, and BGT-1 were not found. In both regions, astrocytes exhibited a stronger GAT than GlyT activity. However, in HC astrocytes the I GABA (max) /I Gly (max) ratio was remarkably higher. In comparison to IC astrocytes HC astrocytes showed faster kinetics for the transport of glycine and slower kinetics for the transport of GABA. Finally, GlyRs could not be detected in astrocytes of IC and HC. However, expression of GABA A Rs was heterogeneousit was found in HC but not in IC astrocytes.

Discussion
In the present study, we investigated the expression and function of GlyTs and GATs in astrocytes from IC and HC. In both regions, astrocytes generally expressed the three inhibitory neurotransmitter transporters GlyT1, GAT-1 and GAT-3, whereas GlyT2, GAT-2, and BGT-1 were not detected. Remarkably, IC astrocytes exhibited larger I Gly (max) and smaller I GABA (max) compared to HC astrocytes. In turn, this resulted in a higher I GABA (max) / I Gly (max) ratio in HC astrocytes.

Basic properties of IC and HC astrocytes
Astrocytes were labeled with SR101, by which classical astrocytes in acute tissue slicescontaining the superior olivary complex (SOC) or the HCcan be identified [19,41,42,55]. We mentioned before that in the IC SR101 labels small sized and highly branched cells [19]. However, their identity was not verified yet by electrophysiological recordings. Here we show that SR101-labeled IC cells exhibit a highly negative E M and a low R M . They are not NG2 glia as these exhibit completely different electrophysiological properties, i.e. a more positive E M , a tremendously higher R M , and currents through voltage-activated sodium channels [42,[56][57][58]. Furthermore, they are unlikely to be oligodendrocytes as they are if at all just weakly labeled by SR101 [41]. In contrast, SR101-labeled IC cells exhibited a non-linear or linear current-voltage relationship corresponding to non-passive and passive astrocytes, respectively, which are found throughout the auditory brainstem ( Fig. 1) [19,41,55,59,60]. Furthermore, these from now on IC astrocytes-termed cells were distributed homogeneously within the nucleus (Fig. 1) like astrocytes in SOC nuclei [19,41]. HC astrocytes exhibited properties like reported in previous studies, e.g. [42,56].
IC and HC astrocytes express functional GlyT1, GAT-1, and GAT-3 Glycine and GABA activate respective transporters that mediate an inward current and a concomitant depolarization due to their stoichiometry: 1 glycine/1 GABA: 2 Na + : 1 Cl − [15,17]. Both inward current and depolarization sometimes partially recovered in the prolonged presence of the agonist (Figs. 2 and 4). This was observed before in LSO astrocytes and may be due to a reduced driving force [19]. Both IC and HC astrocytes showed sarcosine-sensitive I Gly (max) , demonstrating the presence of functional GlyT1 (Fig. 3). Sarcosine is a competitive agonist and therefore inhibited just about 60-70% of I Gly (max) [61]. Thus, the co-expression of the neuron-typical GlyT2 could not be excluded per se. GlyT2 was reported to be present occasionally in astrocytes [40,49]. However, here we never found transcripts for GlyT2 in IC and HC astrocytes indicating the absence of GlyT2. GlyT1 mRNA was present in all IC astrocytes sufficiently explaining I Gly . However, GlyT1 transcripts were found only in about half of the HC astrocytes (Fig. 3). There are several possible explanations: 1) Although the scRT-PCR reliably detected GlyT1 transcripts in the positive controls, it was eventually not sensitive enough to detect single transcripts in all HC astrocytes. 2) There is effectively a mosaic expression of GlyT1. However, all recorded HC astrocytes exhibited an I Gly (Fig. 2). Thus, HC astrocytes  (19) 11.17 ± 0.54 (19) The data are given as mean ± SEM. The number of experiments is given in parenthesis. IC Inferior colliculus, HC Hippocampus, P Probability value, n.s. Not significant **: P < 0.01, ***: P < 0.001 putatively express further transporters that are capable to transport glycine. The neutral amino acid transporter ASCT2 (Slc1A5) as well as sodium-coupled neutral amino acid transporters (system N) SNAT3 (Slc38A3) and SNAT5 (Slc38A5) are expressed by astrocytes and transport glycine, but are electroneutral and accordingly do not generate currents [62][63][64][65]. 3) HC astrocytes are extensively coupled [66][67][68][69] and allow direct electrical communication between neighboring astrocytes [70][71][72]. Here, about half of the HC astrocytes lacked GlyT1 expression, but can be expected to be surrounded by and coupled to GlyT1 expressing HC astrocytes. Therefore, GlyT1 negative astrocytes might indirectly experience I Gly .
Likewise, IC and HC astrocytes exhibited NO711-and SNAP5114-sensitive I GABA (max) showing the co-expression of functional GAT-1 and GAT-3 in both regions (Fig. 5). Hitherto, in the HC GAT-1 and GAT-3 were attributed to interneurons and astrocytes, respectively [21,37]. To our surprise, we found prominent expression of functional GAT-1 in HC astrocytes. NO711 and SNAP5114 inhibited I GABA (max) by about 20 to 40% (Fig. 5), which is similar to our former study on LSO astrocytes [19]. However, simultaneous administration of NO711 and SNAP5114 did not completely abolish I GABA (max) (Fig. 5). Both antagonists dose-dependently inhibit respective GATs [21]. As we here used a low drug concentration to retain the specificity of GAT inhibitors [16] it was not expected to achieve a complete blockage. However, to that point our data did not exclude the possibility of co-expression of further GATs, such as GAT-2 or BGT-1. The latter are predominantly found at the meninges and neuronal somata, respectively [16]. In accordance, we found only transcripts for GAT-1 and GAT-3 but not for GAT-2 and BGT-1 in IC and HC astrocytes. These results indicate that I GABA was solely mediated by GAT-1 and GAT-3 (Fig. 5). Surprisingly, GAT-1 and GAT-3 mRNA exhibited a mosaic pattern in HC astrocytes. In 3/7 cases HC astrocytes did not exhibit transcripts for any GAT. There are two possible explanations: 1) Although the scRT-PCR detected transcripts in the positive controls, it was not sensitive enough to detect single transcripts on the single cell level. 2) There is effectively a mosaic expression pattern. However, the second explanation contrasts with the finding that all HC astrocytes exhibited I GABA that was always sensitive to the GAT-1 and GAT-3 inhibitor NO711 and SNAP5114, respectively (Figs. 4 and 5). Again, the extensive coupling of and direct electrical communication between HC astrocytes [66][67][68][69][70][71][72] could explain why I GABA was recorded in all cells independent from GAT expression.
The co-expression of GlyTs and GATs in the same astrocyte raises the question of transporter interference. Such interference of different transporters was seen before [6,73,74]. In a previous study on LSO astrocytes, we could show that GlyT and GAT activity influence each other [19]. The reciprocal reduction of activity likely refers to changes in their commonly used gradients for Na + and Cl − . Those gradients become weakened upon transporter activation thereby reducing the driving force for the transport. Especially in the IC, where neurons simultaneously receive glycinergic and GABAergic synaptic inputs [29,30], transporter interference might occur during synchronous activation of astrocytic GlyTs and GATs [15]. However, it remains to be elucidated to which extend this interplay takes place and how altered neurotransmitter clearance putatively modulates neuronal signaling [15,17,18].
Taken together, all IC and about half of the HC astrocytes expressed functional GlyT1, GAT-1, and GAT-3. In this respect, these astrocytes can express the same combination of inhibitory neurotransmitter transporters like astrocytes located in LSO, thalamus, cortex or Bergmann glia in the cerebellum or Müller cells in the retina [15,19,[75][76][77][78][79][80]. The potentially heterogeneous expression in HC astrocytes could be indicative of functional domains in which glycinergic transmission arising from excitatory projections and GABAergic transmission from interneurons are segregated from each other.

Transporter currents are not contaminated by ionotropic receptor activation
Both glycine and GABA act on respective transporters and ionotropic receptors. While activation of GlyTs and GATs by exogenous accessible neurotransmitters necessarily causes an inward current, activation of GlyRs and GABA A Rs can result in either an inward current or an outward current. The underlying Cl − efflux or influx depends on [Cl − ] i and subsequently on E Cl . Under physiological conditions astrocytic [Cl − ] i amounts to about 30 mM [81] causing an inward current and concomitant depolarization upon receptor activation. However, our pipette solution contained 2 mM Cl − and receptor activation would have caused an outward current. In our recordings, we never observed glycine-or GABA-induced outward currents in IC and HC astrocytes (e.g. Figs. 2 and  4), which was surprising as at least HC astrocytes express functional GABA A Rs [21,22]. Two possible scenarios could explain this discrepancy: 1) The GABA A R-mediated outward current was too small and consecutively masked by the large GAT-mediated inward current. This in turn would suggest that the amount of GAT-mediated inward current would be underestimated. 2) GABA A Rs rapidly desensitize [53,54,82]. In combination with slow wash-in of GABA in our experiments, this early desensitization might hamper the accurate detection of GABA A R activation. To answer the question of masked activation and/or desensitization of ionotropic receptors we measured R M changes that could result from increased membrane permeability (see Methods). Proof of principle experiments on GlyR and GABA A R expressing IC and HC neurons validated the method (Additional file 2: Figure S2). Our results convincingly demonstrated the capability to detect R M changes upon GlyR and GABA A R activation with the utilized test pulse protocol.
With this tool at hand, we were able to detect GABA A R activation in HC astrocytes (Fig. 5). GABA A R activation was detected by temporary R M reduction that vanished within 10 s indicating receptor desensitization. However, we never observed any outward current that had to arise from Cl − influx due to the low [Cl − ] i of the intracellular solution. We reason that any small Cl − influx-mediated outward current is instantly masked by strong electrogenic transporter current. Nonetheless, I GABA (max) , which was measured earliest after 10 s, was not contaminated by GABA A R-mediated currents. The R M reduction in HC astrocytes was rather small (~8%) compared to HC neurons (~74%). Astrocytes express various K + channels that are constitutively open at resting conditions (inwardly rectifying K + channels, twopore-domain K + channels) [46,83]. In turn, these channels cause the very high K + conductance observed in astrocytes [44]. Accordingly, it is not surprising that the R M reduction was relatively small. At the same time, IC astrocytes exhibited no R M reduction upon GABA application (Fig. 5). Thus, either GABA A Rs are absent or their amount is essentially not high enough to be relevant. Interestingly, using this method on LSO astrocytes we detected a small R M reduction indicating the presence of GABA A Rs (Vanessa Augustin and Simon Wadle, unpublished). We previously reported that I GABA in LSO astrocytes mainly constitutes of GAT-mediated current [19]. Similar to HC astrocytes, the GABA-induced R M reduction in LSO astrocytes vanished within 10 s after the beginning of GABA application. Thus, our previously reported I GABA (max) in LSO astrocytes was not contaminated by GABA A R activation.
Similarly, we used the same method to examine a possible influence of GlyR activation onto our recorded I Gly (max) . We could show that neither IC nor HC astrocytes exhibited glycine-induced R M changes or outward currents (Figs. 2 and 3). Likewise, LSO astrocytes lack glycine-induced R M changes (Vanessa Augustin and Simon Wadle, unpublished). Accordingly, functional GlyRs appear to be absent in those astrocytes. This is consistent with the observation that GlyRs were described only in astrocytes located in most caudal brain regions, i.e. spinal cord and caudal brainstem (ventral respiratory group) [20,24,25]. However, this contrasts with the wide distribution of GABA A Rs throughout the brain [22]. In summary, I Gly (max) and I GABA (max) were not affected by GlyRs and GABA A Rs, respectively, and the transporter currents were accordingly not underestimated.

Activity and kinetics of GlyTs and GATs
IC and HC astrocytes are differently capable to take up glycine and GABA (Table 2). While there is no statistical difference for glycine transport among the two brain regions, the GABA transport is stronger in HC astrocytes. In the LSO, which is located more caudal compared to IC and HC, astrocytes exhibit a similar capability to take up glycine. However, their ability for GABA clearance is much lower [19]. Thus, astrocytic I GABA (max) increases from caudal to rostral brain regions (LSO < IC < HC). Consequently, the ratio of I GABA (max) /I Gly (max) is elevated in more rostral brain regions (HC (3.6) > IC (1.9) > LSO (1.6; data from [19])). This was expected, as the need to take up GABA rather than glycine is higher in rostral brain regions, which arises from the glycine-to-GABA shift as the predominant inhibitory neurotransmitter [2,3,9,10,29,30]. Noticeably, GlyT-mediated I Gly (max) substantially persists in HC astrocytes. This allows the clearance of glycine that is co-released from excitatory presynaptic terminals [11,12]. Taken together, I Gly (max) is similar in the three brain regions, whereas I GABA (max) as well as the I GABA (max) /I Gly (max) ratio are regiondependent and increases with the prevalence of GABA as inhibitory neurotransmitter.
Beside inter-region differences of amplitudes, we additionally found region-dependent alterations of the kinetics of transporter-mediated currents (Table 4). Whereas IC astrocytes exhibit similar kinetics for the transport of glycine and GABA, HC astrocytes are marked by faster glycine and slower GABA transport. However, LSO astrocytes generally outperform IC and HC astrocytes regarding kinetics of GlyTs (rise time: 1.05 ± 0.18 s; decay time: 4.88 ± 1.11 s; n = 6/6/6) and GATs (rise time: 0.61 ± 0.13 s; decay time: 4.52 ± 0.52 s; n = 12/12/11; data from [19]). GlyTs and GATs can be modulated by several mechanisms: e.g., enhancement of transporter activity can be achieved by transporter glycosylation and [Ca 2+ ] i elevation [84][85][86], whereas decrease of transporter activity can be caused by activation of protein kinase C and de-glycosylation [84,85,[87][88][89]. If one or more of those mechanisms are relevant in astrocytes of the three brain regions, is yet unexplored. However, the different transport kinetics correlate with different precision of signal processing in those three brain regions. The auditory system in general requires temporal precise coding to compute correctly e.g. interaural time and level differences in the medial superior olive and the LSO, respectively, and synapses show relatively weak depression allowing high rates of synaptic transmission [4,[90][91][92]. Furthermore, the synaptic signaling in the LSO is considerably faster and more precise compared to the hippocampus [93]. Like the LSO, the IC belongs to the auditory brainstem. However, it is not used for sound source localization, but serves as an information hub.