Extrinsic factors regulate partial agonist efficacy of strychnine-sensitive glycine receptors

Background Strychnine-sensitive glycine receptors in many adult forebrain regions consist of alpha2 + beta heteromeric channels. This subunit composition is distinct from the alpha1 + beta channels found throughout the adult spinal cord. Unfortunately, the pharmacology of forebrain alpha2beta receptors are poorly defined compared to 'neonatal' alpha2 homomeric channels or 'spinal' alpha1beta heteromers. In addition, the pharmacologic properties of native alpha2beta glycine receptors have been generally distinct from receptors produced by heterologous expression. To identify subtype-specific pharmacologic tools for the forebrain alpha2beta receptors, it is important to identify a heterologous expression system that closely resembles these native glycine-gated chloride channels. Results While exploring pharmacological properties of alpha2beta glycine receptors compared to alpha2-homomers, we found that distinct heterologous expression systems appeared to differentially influence partial agonist pharmacology. The β-amino acid taurine possessed 30–50% efficacy for alpha2-containing receptor isoforms when expressed in HEK 293 cells. However, taurine efficacy was dramatically reduced in L-cell fibroblasts. Similar results were obtained for β-alanine. The efficacy of these partial agonists was also strongly reduced by the beta subunit. There were no significant differences in apparent strychnine affinity values calculated from concentration-response data between expression systems or subunit combinations. Nor did relative levels of expression correlate with partial agonist efficacy when compared within or between several different expression systems. Finally, disruption of the tubulin cytoskeleton reduced the efficacy of partial agonists in a subunit-dependent, but system-independent, fashion. Conclusions Our results suggest that different heterologous expression systems can dramatically influence the agonist pharmacology of strychnine-sensitive glycine receptors. In the systems examine here, these effects are independent of both absolute expression level and any system-related alterations in the agonist binding site. We conclude that complex interactions between receptor composition and extrinsic factors may play a significant role in determining strychnine-sensitive glycine receptor partial agonist pharmacology.


Background
It has been well established that the amygdala is important in the acquisition and maintenance of fear/anxietyrelated behaviors [1]. Strychnine-sensitive glycine receptors have recently been found in the adult rat basolateral amygdala (BLA) using whole cell and intracellular electrophysiology [2,3]. Reverse transcription polymerase chain reaction on whole BLA tissue and single cells revealed a prominent expression of α 2 mRNA; and these receptors are likely to be α 2 β heteromers due to their low picrotoxin sensitivity [4]. This finding is consistent with prominent BLA 'general' immunoreactivity for α/β subunit protein but no apparent α 1 -specific protein expression [3]. A similar enrichment of α 2 /β heteromers is also evident in striatal cholinergic interneurons [5]. It is quite possible then that the α 2 β strychnine-sensitive glycine receptors present in the adult BLA and other forebrain areas represents a receptor population that could be functionally distinguished from those found in the spinal cord. Because the BLA regulates a number of anxiety-or fear-related behaviors [6], it is possible that this population of strychninesensitive glycine receptors may represent a novel therapeutic target for anxiety disorders. To insure that novel α 2 β compounds possess an appropriate therapeutic index, the pharmacology of these forebrain glycine receptors must be elucidated and extensively compared with the spinal isoform.
There have been conflicting reports regarding the details of glycine receptor pharmacology when expressed in heterologous systems. For example, taurine acts as a partial agonists (ca. 50% efficacy compared to glycine) for GlyRα 1 expressed in Xenopus ooctyes [7] whereas it shows nearly full agonist efficacy for GlyRα 1 expressed in HEK 293 cells [8]. Compared to GlyRα 1 , taurine efficacy is even weaker for GlyRα 2 (ca. 5-10% efficacy) when expressed in Xenopus oocytes [7]. However, native GlyRα 2 β receptors expressed by BLA neurons possess >50% efficacy for taurine and almost full efficacy for β-alanine [2]. While these results might initially be dismissed as expression systemdependent phenomena, brain region-specific effects are also evident in the literature. Taurine has markedly different efficacies at glycine receptors expressed by isolated adult lateral/basolateral amygdala neurons [2], adult hypothalamic magnocellular neurons [9], and juvenile spinal cord neurons [10]. It is therefore possible that the mechanisms regulating brain region-specific effects are related to those governing the divergence among heterologous expression systems. However, such mechanisms have not been systematically investigated, despite their potential usefulness in understanding region-to-region pharmacologic heterogeneity evident for some native receptors.
This study utilizes whole-cell patch clamp electrophysiology to examine the influence of distinct heterologous expression systems on the β-amino acid pharmacology of glycine receptors composed of distinct subunit combinations. We have focused on the α 2 and α 2 β receptors since these appear to be the predominate isoforms found in the embryonic and adult forebrain, respectively. Our results provide potentially important insight into the types of mechanisms that may govern brain region-to-brain region variation in glycine receptor pharmacology. Several aspects of this work have appeared in abstract form [11,12].

Subunit-and system-dependent effects on glycine pharmacology
Given the variation of glycine receptor partial agonist pharmacology in the literature, we specifically sought to identify any role that expression system may play in their pharmacological profiles. First, glycine concentrationresponse relationships were established for GlyRα 2 , and GluRα 2 /β in HEK-293 cells and in L-cell fibroblasts. Glycine-gated responses for each receptor isoform were elicited in a dose-dependent manner in both cell types (Fig.  1A). The apparent EC 50 of glycine HEK cells was 221 µM and 269 µM for α 2 (n = 4-6) and α 2 β (n = 7-8), respectively. GlyR subunits expressed in L-cells displayed a similar pharmacological profile. However, the apparent glycine EC 50 of both GlyRα 2 (446 µM, n = 5-7) and GlyRα 2 β (667 µM, n = 4-8) appeared lower than apparent affinities for the same subunits when expressed in L-cells. Two-way ANOVA on the Log (EC 50 ) values (Table 1) indicated a significant effect of system (F = 20.01, P < 0.001). However, the presence of the β-subunit did not significantly affect glycine apparent affinity in either system nor was there a significant interaction between system and subunit composition. These results indicate that glycine is less potent for receptors expressed in L cells compared to HEK cells.

Subunit-and system-dependent effects on β-Alanine pharmacology
With the glycine pharmacological profile established, we next examined the pharmacology of the partial agonists, β-alanine and taurine. The efficacy and potency of these βamino acids were compared to glycine by normalizing the current response at each concentration to a maximal glycine response in that same cell. In HEK cells (Fig. 1B), the average β-alanine EC 50 values calculated from individual cells were 770 µM (n = 6) and 570 µM (n = 7) for α 2 and α 2 /β receptors, respectively. In L-cells, β-alanine also elicited currents in a dose-dependent manner. And, like glycine, β-alanine appeared to be less potent in these cells compared to HEK cells. EC 50 values for α 2 and α 2 /β receptors were 2.0 mM (n = 8) and 2.9 mM (n = 7), respectively. Two-way ANOVA on LogEC 50 values from these studies indicate a significant effect of the expression system on βalanine potency (F = 43.52, P < 0.0001). There was a trend for the presence of the β-subunit to influence potency but this was not significant nor was there any significant interaction between expression system and subunit composition.
β-alanine efficacy was also examined in these same experiments by normalizing the maximal β-alanine response as a fraction of a maximal glycine response. In HEK cells, the α 2 and α 2 β isoforms had efficacies of 80 ± 6% and 55 ± 7% of the maximal glycine response, respectively. A similar trend was noted in L-cells with the α 2 and α 2 β isoform with β-alanine efficacies being 39 ± 8% and 25 ± 5% of the maximal glycine response. Two-way ANOVA analysis of these data indicate that both expression system and subunit composition had a significant influence on βalanine efficacy (F = 27.6, P < 0.0001 and F = 7.9, P < 0.01 respectively). There was no significant interaction between these variables. These data demonstrate that the presence of the β subunit reduced β-alanine efficacy of α 2containing receptors and that this efficacy was substantially smaller L-cells compared to HEK cells.

Subunit-and system-dependent effects on taurine pharmacology
Similar analysis of taurine pharmacology in HEK and Lcells revealed more dramatic effects of system and subunit on this partial agonist (Fig. 1C). In HEK cells, the apparent EC 50 for taurine was 501 µM for GlyRα 2 (n = 9) and 2 mM for GlyRα 2 β (n = 7). Because of its remarkably low efficacy in L-cells (see below), we can only provide estimates of taurine potency in this expression system. Regardless, apparent taurine affinity for both GlyRα 2 and GlyRα 2 β expressed in L cells were ~3 mM for both isoforms (n = 4 and 7, respectively). We did not compare L cell data with that obtained from HEK cells due to the uncertainty surrounding the fits. However, there was no significant difference in apparent taurine potency between the α 2 and α 2 β receptors expressed in HEK cells (P >> 0.05, t-test).
Taurine efficacy was obviously quite different between the two expression systems. In HEK cells, taurine efficacy was 48 ± 12% of glycine for GlyRα 2 and 32 ± 4% of glycine for the GlyRα 2 β isoform. Efficacy for these same receptors was reduced to approximately 6 ± 1% and 5 ± 0.7% of glycine when they were expressed in L-cells in these particular studies. The system difference was significant with twoway ANOVA (F = 17.4, P < 0.001) with no substantial effects of subunit composition or interactions between these variables.

Expression level and system-dependent pharmacology
The preceding results suggest that there may be a complex interaction between subunit composition and the expression system in which the receptor is produced. Specifically, the system-dependent agonist pharmacology could be related to differences in the relative expression levels between various systems. Expression level has clearly been demonstrated to influence agonist pharmacology for G protein-coupled receptors (e.g. [13]), where the levels of G-protein bound to receptor and thus the relative levels of high affinity receptor can vary from system to system. However, the influence of expression level on ligandgated channel function has not been extensively explored (see Discussion). Unfortunately, it is problematic to compare expression levels between HEK and L-cells since the relative efficiency of transfection varied widely between these systems. Indeed, liposome-mediated transfection is remarkably efficient in HEK 293 cells (70-90% of cells based on GFP fluorescence) but only marginally effective in L-cells (10-20% of cells, not shown). To get around these differences in transfection efficiency, we examined the relative expression level of GlyRα 2 protein using western analysis of total lysate derived from the same number of GFP + HEK 293 or L cells from transfected cultures ( Fig.  2A). For this experiment, cells were harvested under native conditions, GFP + cells were counted, and volumes of lysate corresponding to equivalent numbers of GFP + cells was loaded onto the gel. Western blots from two separate experiments demonstrate that transfected HEK 293 cells expressed 4-to 5-fold more GlyRα 2 protein than transfected L-cells. The mean optical density from the two experiments was 83 ± 2 units for HEK cells and 17 ± 2 units for L-cells.
Maximal conductance is an independent measure of functional expression and was also larger for both α 2 and α 2 β receptors expressed in HEK cells compared to receptors expressed in L cells (Fig. 2B). Across all experiments where maximal glycine concentrations were assayed, the conductance of α 2 receptors expressed in L-cells was 65 ± 11 nS and was 114 ± 21 nS in HEK cells. Similarly, L-cells expressed α 2 β receptors at 42 ± 9 nS while HEK cells expressed this isoform at 104 ± 17 nS. Two-way ANOVA using subunit and system as variables revealed a significant effect of system (F = 15.4, P < 0.001) but not subunit, nor was there a significant interaction between variables. Results from both westerns and functional experiments therefore indicate that relative expression levels of glycine receptor were different between HEK and L-cells.
To further explore the interaction between expression level and partial agonist efficacy, both current density and taurine efficacy were compared for α 2 and α 2 β glycine receptors in a number of different heterologous systems, as well as for native receptors expressed in rat lateral/basolateral amygdala. In addition to HEK and L-cells, the heterologous systems included mouse 3T3 fibroblasts and MDCK kidney cells. α 2 β receptors expressed in mouse 3T3 fibroblasts had twice the current density (121 ± 34 pA/pF) of the mouse L-cells (59 ± 19 pA/pF) but had a similar taurine efficacy (13 ± 8% of glycine in 3T3 cells versus 8 ± 1% in L-cells). Similarly, α 2 β receptors expressed in HEK293 cells had a current density similar to GlyRs expressed in 3T3 fibroblasts (115 ± 11 pA/pF) but had a taurine efficacy compared to glycine of 48 ± 3%. This efficacy was similar to glycine receptors expressed by acutely isolated adult rat basolateral amygdala neurons (46 ± 5% of glycine) although the current density in this native system was only 57 ± 14 pA/pF. Note that the channels expressed by these neurons are composed primarily of α 2 +β subunits [4]. Canine kidney MDCK cells expressed the lowest α 2 β current density (15 ± 5 pA/pF); yet the channels expressed by this system had the highest taurine efficacy of any cell tested (101+6%). For α 2 GlyRs, the rank order of glycine receptor density was 3T3 (111 ± 21 pA/pF)> HEK cell (90 ± 11 pA/pF)> L-cell (50 ± 9 pA/pF); while the a -efficacy relative to maximal glycine response b -p < 0.0001 for system but not subunit using two-way ANOVA c -estimate due to low efficacy Relative expression levels does not influence taurine efficacy C rank order of taurine efficacy for these same receptors was HEK (74 ± 9%)> 3T3 (23+7%)> L-cells (11 ± 2%). Across all subunit combinations and systems, there was no significant correlation (R 2 = 0.14, P >> 0.05) between taurine efficacy and glycine current density (Fig. 5C). Indeed, no correlation between expression level and taurine efficacy was evident within any given population of cells whether the receptors were expressed in native or heterologous systems. For example, the correlation coefficients for α 2 β receptors between taurine efficacy and glycine current density in individual systems were 0.11, 0.13, 0.19, 0.05, and 0.31 for HEK, L-cells, 3T3 cells, MDCK cells, and amygdala neurons, respectively (P >> 0.05). Thus, while there is clearly a difference in expression level between both the systems as well as between individual cells in a given system, this particular characteristic cannot account for the apparent taurine efficacy.

The agonist-binding site is not affected by expression system
There are a variety of possible mechanisms to account for the disparities in partial agonist pharmacology between two expression systems. One way to address this is to examine competitive antagonist binding properties in HEK and L-cells. We therefore examined the potency of the glycine receptor competitive antagonist strychnine in both systems. Following a 30 second pretreatment with the antagonist [2], we co-applied strychnine and an EC 50 concentration of glycine. The strychnine K B was estimated for HEK and L-cells expressing either the GlyRα 2 or GlyRα 2 +β subunits using the Cheng-Prusoff relationship (see Methods). This relationship takes into account the divergent Hill-slope and potencies for glycine found in these two expression systems. Receptors composed of the GlyRα 2 subunit (Fig. 4A) had very similar K B values when expressed in either HEK (K B = 49 ± 8 nM, n = 11) or L-cells (K B = 38 ± 7 nM, n = 16; Fig. 4C). The same was true for cells expressing the GlyRα 2 β subunits where strychnine apparent affinity was 32 ± 7 nM (n = 10) in HEK cells and 38 ± 8 nM (n = 10) in L-cells. Two-way ANOVA did not reveal any significant effect of either system or subunit composition. Since strychnine is a competitive antagonist and site-directed mutation studies suggests that strychnine and glycine interact with overlapping regions of the receptor [14][15][16], our results strongly suggest that functional strychnine affinity, and hence the general structure of the agonist binding pocket, was not substantially influence by expression system.

Cytoskeletal components influence glycine receptor pharmacology
A third possible mechanism for reduced efficacy in L-cells compared to HEK cells or neurons could be related to intracellular factors that influence channel gating [17]. This hypothesis was examined by disrupting the cytoskeletal protein tubulin, which has been shown to be important for glycine receptor localization [18]. Direct application of 100 µM colchicine did not elicit any membrane currents. Furthermore, acute application of 100 µM colchicine and an EC 50 concentration of glycine (300 µM) did not significantly affect glycine-gated currents themselves. Glycine currents were 17.3 ± 2.3 pA/pF while gly-cine+colchicine currents were 16.7 ± 2.2 pA/pF (p > 0.5, paired two-tail t-test, n = 7).

Discussion
We have expressed several the 'embryonic' (α 2 homomeric) and 'forebrain' (α 2 β heteromeric) isoforms in two distinct expression systems to understand the influence of endogenous and exogenous factors on receptor partial agonist pharmacology. Although the pharmacology of the 'embryonic' GlyRα 2 isoform and the 'adult spinal' isoform (GlyRα 1 β) have been explored more frequently in the literature, the pharmacology of GlyRα 2 β receptors has remained largely unexplored. Despite this, there is strong evidence that the adult 'forebrain' isoforms, specifically in the rat basolateral amygdala, is indeed α 2 β [4]. The current study indicates a general trend for decreased apparent affinity and reduced relative efficacy of agonists when receptors consist of the α 2 β subunits compared to their homomeric α 2 counter parts.
We were particularly surprised to find that receptors expressed in different expression systems possessed markedly different partial agonist efficacies. The remainder of our study focused on identifying extrinsic factors that influence difference in ligand-gated receptor pharmacology in distinct expression systems. While differences in efficiency of cDNA expression/transfection between systems could explain such differences, the expression levels of GlyRα 2 β receptors measured by current density was not correlated with taurine efficacy across several different cell types or within any given system. Importantly, the efficacy of β-alanine and taurine in HEK cells agree with previous findings where cells expressing GlyRα 2 show almost full efficacy for taurine and β-alanine [21]. Similarly, distinct ligand binding characteristics of receptors expressed in different expression systems seemed to be another possible mechanism governing agonist efficacy or potency. For the glycine receptor, the binding site for the competitive antagonist strychnine is believed to be adjacent to the agonist-binding site, sterically hindering agonist binding. A gross alteration in the agonist binding pocket, particularly one that hindered agonist binding, would most likely affect strychnine binding as well. To examine this, strychnine K B was calculated for GlyRα 2 and GlyRα 2 β isoforms expressed in both HEK and L-cells. In order to decrease the error in estimating K B , a derivation of the Cheng-Prusoff equation was used that takes in account variations in the slopes of the inhibition curves [22]. Differences in K B were negligible between expression systems, indicating the strychnine binding site, and presumably the agonist binding site, was altogether similar in these different systems. Differences in pharmacology between systems therefore cannot be explained by substantial alterations in the agonist/competitive antagonist binding pocket.
Receptor gating is another mechanism by which receptor function may be altered. Cytoskeletal elements have been shown to play a crucial role in neurotransmitter receptor clustering [17] and may have a role in receptor function as well. For example, cytoskeletal stabilization has been shown to reduce Ca ++ -dependent inactivation of Ca ++ channels in snail ganglia [23]; and, actin has been shown to modulate several different types of membrane ion channel [24][25][26]. Cytoskeletal depolymerization has also been found to inhibit the function of GABA A receptors, which share significant sequence homology and functional characteristics with strychnine-sensitive glycine receptors [27]. And the tubulin-gephyrin-glycine receptor interaction is critical for establishing functional glycinergic synapses [28]. The current study suggests that cytoskeletal elements may play a functional role in α 2 β glycine receptor pharmacology as well. β-containing glycine receptors are intimately associated with the tubulin-associated protein gephyrin [29] via the gephyrin binding site that lies within the intracellular domain of this subunit [30]. In our studies, the efficacy of both taurine and βalanine were reduced in cells expressing GlyRα 2 β subunits. This despite the finding that gephyrin-like immunoreactivity in L-cells was apparently distinct from that in HEK cells, suggesting that distinct cytoskeletal components in these systems may have profound influence over GlyR α 2 β pharmacology. This is further supported by suggestions that gephyrin may exist in multiple, tissue-specific isoforms with potentially distinct functional roles [31][32][33]. It should be noted however that colchicine treatment of α 2 β-expressing HEK cells did not suppress partial agonist efficacy to a level that approached that found in Lcells or 3T3-fibroblasts. Given that colchicine had no perceptible effect of α 2 -homomeric channels expressed in HEK cells, our results suggest that additional systemdependent factors may have a more pronounced influence on the partial agonist pharmacology of strychninesensitive glycine receptors.

Conclusions
It is of particular interest that the β subunit appears to play a functional role in the pharmacology of α 2 -containing receptors regardless of expression system. For example, the beta subunit decreased the apparent efficacy of the partial agonists. Since the β-subunit itself does not appreciably interact with the competitive antagonist strychnine [34], these results are at least consistent with some allosteric interaction between the β-subunit and the agonist binding site present the α subunit. This may indicate that α 2 β glycine receptors in the forebrain may be distinguishable from other receptor isoforms given the appropriate pharmacologic agent. Although cytoskeletal components potentially play some role for these 'forebrain' receptors, there appear to be other 'extrinsic' factors governing expression system-dependent effects on agonist pharmacology. Since it is conceivable that such factors may be differentially distributed between different forebrain regions, the large apparent differences between glycine receptor pharmacology reported by various studies may not necessarily depend upon differential expression of glycine receptor subunits per se. At the very least, our findings suggest that great care should be taken when utilizing different expression systems to develop screens for novel pharmacophores acting on this receptor.  [4]. Briefly, a total of 2-3 µg of plasmid constructs was added to 100 µL serum-free media along with 10 µL Superfect reagent, vortexed for several seconds, and incubated for 10 minutes to allow DNA/liposome formation. Standard media (600 µL) was added and this mixture applied the cells and incubated for 2-3 hours at 37°C. The cells were subsequently washed twice with phosphate-buffered saline, fresh media was applied and cells were then incubated for 24-48 hours prior to recording. Cells were co-transfected with green fluorescent protein (pEGFP-C1, Clontech, Pal Alto, CA, USA) to identify transfected cells before recording. Mass ratios of 1:1:5 were used in the GFP:α 2 :β transfections. An equal mass of the cloning vector pCI (Promega Corp) replaced the β subunit when examining α 2 homomeric channels.

Electrophysiology
Whole cell recordings were performed at room temperature using standard patch-clamp techniques and the axopatch-1D amplifier (Axon Instruments, Inc., Foster City CA, USA) in the voltage clamp mode. Gigaohm seals were formed using patch pipettes made from borosilicate glass (World Precision Instruments, Sarasota FL, USA). For most experiments, the internal solution contained (in mM): CsCl 100, EGTA 11, HEPES 10, CaCl 2 1, Mg-ATP 4, pH 7.2 with methane sulfonic acid; adjusted to 290-295 mmol kg -1 with sucrose. Whole cell capacitance and series resistance was manually compensated after opening the cell. Cells were continuously bath perfused with a HEPESbuffered saline (in mM): NaCl 150, Glucose 10, HEPES 10, KCl 2.5, CaCl 2 2.5 MgCl 2 1.0, pH 7.4, 305-320 mmol kg -1 Data will be analyzed off-line using pClamp software (Axon). Numerical analysis was performed using commercially available software. Independent student's t-test and two-way ANOVA were used for comparisons where appropriate; and statistical significance was based on p < 0.05. Concentration-response curves were generated from fits of data to a standard logistic equation as previously described (McCool & Botting, 2000). To derive K B from functional IC 50 and EC 50 data, the Cheng-Prusoff equation was used: where S = hillslope of agonist curve, A = concentration of agonist, EC 50 = half-maximal agonist concentration, and IC 50 = half-maximal antagonist concentration.
Agonists and antagonists were applied for 4-10 sec from an array of eight HPLC-grade capillary tubes (150 µm i.d.; Hewlett Packard Analytical Direct) placed within 100 µm of the cell of interest.