The present address of Judith A. Heads is the Howard Florey Institute, University of Melbourne, Parkville, VIC 3052, Australia.
Structure-activity analysis of ginkgolide binding in the glycine receptor pore
Version of Record online: 21 JAN 2008
© 2008 The Authors. Journal Compilation © 2008 International Society for Neurochemistry
Journal of Neurochemistry
Volume 105, Issue 4, pages 1418–1427, May 2008
How to Cite
Heads, J. A., Hawthorne, R. L., Lynagh, T. and Lynch, J. W. (2008), Structure-activity analysis of ginkgolide binding in the glycine receptor pore. Journal of Neurochemistry, 105: 1418–1427. doi: 10.1111/j.1471-4159.2008.05244.x
- Issue online: 21 JAN 2008
- Version of Record online: 21 JAN 2008
- Received October 28, 2007; revised manuscript received January 10, 2008; accepted January 11, 2008.
- binding site;
- channel block;
- Cys-loop receptor;
- Ginkgo biloba;
- ligand-gated ion channel;
- site-directed mutagenesis
Ginkgolides, active constituents of Ginkgo biloba extracts, potently block the glycine receptor chloride channel (GlyR). Ginkgolides A, B, C and J are structurally similar, varying only by the presence or absence of oxygens at their R1 and R2 positions. The aim of this study was to understand how variable ginkgolide groups bind to pore-lining 2′ and 6′ residues in the α1 GlyR. Ginkgolide potency was not affected by G2′A or G2′S mutations, suggesting 2′ residues are not important for ginkgolide coordination. Analysis of the α1T6′S GlyR suggests that ginkgolides bind to this receptor via hydrogen bonds between T6′S and ginkgolide R1 hydroxyls. The abolition of block by the T6′A and T6′V mutations but not by the T6′S mutation implies the existence a second transmembrane domain α-helical kink formed by hydrogen bonding between 6′ threonine and serine sidechains and backbone carbonyl oxygens. We also found that ginkgolide A binds in different orientations in the closed and open states of a mutant GlyR, possibly reflecting its enhanced flexibility relative to other ginkgolides. Together these results indicate that small variations in ginkgolide structure or pore structure can lead to drastic potency variations. This property may be exploited to create improved pharmacological probes for discriminating among anionic Cys-loop receptor isoforms with 6′ structural variations.
half-maximal concentration for activation
GABA type-A receptor chloride channel
glycine receptor chloride channel
half maximal concentration for inhibition
second transmembrane domain
The glycine receptor (GlyR) chloride channel mediates inhibitory neurotransmission in spinal motor reflex circuits and nociceptive pathways (Lynch 2004; Zeilhofer 2005; Betz and Laube 2006). It belongs to the Cys-loop ligand gated ion channel family that also includes the GABA type A receptor (GABAAR), the nicotinic acetylcholine receptor, and the serotonin type 3 receptor. Cys-loop receptor subunits comprise a large N-terminal extracellular ligand binding domain, followed by four α-helical transmembrane domains (M1–M4), with the second transmembrane domain (M2) domain lining the central water-filled pore (Unwin 2005).
Five GlyR subunits have been identified (α1–α4, β). Embryonic receptors generally comprise α2 homomers or α2β heteromers, whereas the α1β heteromer is the dominant adult subtype (Lynch 2004; Betz and Laube 2006). Although the β subunit is broadly distributed throughout the CNS, it is poorly expressed in immature neurons. The α1, α2 and α3 subunits all exhibit different neuronal distribution patterns that are particularly evident in the superficial dorsal horn of the spinal cord (Harvey et al. 2004) and retina (Haverkamp et al. 2003, 2004; Heinze et al. 2007). The functional consequences of the differential distribution patterns are difficult to establish as there are few pharmacological probes that can effectively discriminate among GlyR isoforms (Webb and Lynch 2007).
Ginkgolides, which are among the main active constituents of herbal extracts from the Ginkgo biloba tree, are terpene trilactone molecules with a cage-like structure consisting of six adjacent rings each containing five bonded atoms (van Beek 2005; Nakanishi 2005). Ginkgolides A, B, C, and J (GA, GB, GC, and GJ) are structurally similar and vary only by the presence or absence of oxygen atoms at the R1 and R2 positions (Fig. 1a). The highly conserved ginkgolide structure means that structure-activity analyses performed on GlyRs incorporating mutations at candidate binding sites should identify specific molecular interactions between the ginkgolide variable groups and the receptor.
Several lines of evidence suggested that ginkgolides bind in the 2′–6′ region of the pore Early studies on native rat neuronal GlyRs suggested that GB acted as a classical pore blocker (Kondratskaya et al. 2002; Ivic et al. 2003). A molecular docking model supported the feasibility of a binding site in this region (Hawthorne et al. 2006) and mutagenesis studies provided limited experimental evidence for such a model. For example, the α1 subunit G2′A mutation (the α1 to α2 subunit substitution) successfully replicated the modest ginkgolide sensitivity difference between the α1 and α2 GlyR homomers (Kondratskaya et al. 2005; Hawthorne et al. 2006). However, apart from the non-conservative T6′F mutation (the α1 to β subunit substitution) which abolished ginkgolide binding (Hawthorne et al. 2006), the role of 6′ threonines in coordinating ginkgolides has not been investigated.
An understanding of how ginkgolides bind in the GlyR pore could place constraints on pore structure and could lead to the synthesis of more specific subunit selective antagonists, which could be valuable tools to pharmacologically separate native GlyR subtypes. Accordingly, the aim of this study was to investigate the effects of conservative and non-conservative 2′ and 6′ mutations on receptor sensitivity to GA, GB, GC, and GJ in an attempt to understand how they bind in the α1 GlyR pore.
Mutagenesis and expression of GlyR cDNAs
The human GlyR α1 subunit cDNA was subcloned into the pCIS2 plasmid vector (Clontech, Palo Alto, CA, USA), and the β subunit was subcloned into the pIRES2-EGFP plasmid vector (Clontech). Mutations were incorporated using the QuickChange mutagenesis kit (Stratagene, La Jolla, CA, USA). The presence of the correct mutation was confirmed by automated DNA sequencing of plasmid DNA isolated using the BioRad Aurum miniprep kit (Hercules, CA, USA).
Recombinant GlyRs were functionally expressed in HEK293 cells. Cells were cultured in Dulbecco’s Modified Eagle medium supplemented with 10% fetal calf serum and penicillin 30 U/mL (GibcoBRL, Grand Island, NY, USA). The day before transfection, cells were split onto glass coverslips in 3 cm diameter culture dishes. Transfection was then performed via a calcium phosphate transfection protocol. After exposure to the transfection solution overnight, the cells were washed twice with phosphate buffered saline and the culture medium was replaced. Electrophysiological recordings were then performed in the 2 days following washing. To visually identify cells that had taken up exogenous cDNA, empty pEGFP vector (Clontech) was co-transfected with the α1 GlyR cDNA in a 1 : 10 ratio. This was not used when transfecting α1β heteromers as green fluorescent protein fluorescence was used to identify β subunit expression.
Currents were measured by whole-cell patch clamp recording of cells visually identified through a fluorescent microscope. Each coverslip was used for up to 1 h of recording after placement in control solution containing (mM): 140 NaCl, 5 KCl, 2 CaCl2, 1 MgCl2, 10 HEPES, 10 glucose, adjusted to pH 7.4 with NaOH. Patch-clamp pipettes were made from borosilicate glass hematocrit tubing (Vitrex, Modulohm, Denmark) pulled with a horizontal puller (P97, Sutter Instruments, Novato, CA, USA). The resistance of the tips varied between 1 and 3 MΩ when filled with CsCl pipette solution containing (in mM): 145 CsCl, 2 CaCl2, 2 MgCl2, 10 HEPES, 10 EGTA, adjusted to pH 7.4 with NaOH. Cells were voltage-clamped at −40 mV. Solutions were applied to cells via gravity-induced perfusion through parallel micro-tubules. The perfusion system was under the control of a manual micromanipulator and solution exchange between adjacent tubes was routinely complete within 100 ms. Current responses were recorded with an Axopatch 1D amplifier and pCLAMP9 software (Axon Instruments, Union City, CA, USA), digitally filtered at 500 Hz and digitized at 1 kHz. Experiments were performed at room temperature (21–24°C).
Glycine and the ginkgolides were dissolved in water to make 100 mM and 10 mM stocks, respectively, and stored frozen at −20°C. On the day of recording, frozen stocks were thawed and diluted with control solution to the desired concentration. The ginkgolides were obtained as follows: GA from Sigma (St Louis, MO, USA) or Tauto Biotech Co. Ltd. (Shanghai, China), GB from Biomol (Plymouth Meeting, PA, USA), GC from MP Biomedicals (Eschwege, Germany), and GJ from Tauto Biotech Co. Ltd.
All results are expressed as mean ± SEM of three or more independent experiments. The Hill equation was used to calculate the Hill coefficients (nH) and the half-maximal concentrations for activation and inhibition (EC50 and IC50, respectively). To achieve this, individual dose-response curves were fitted using a non-linear least squares algorithm (Sigmaplot 9.0, Jandel Scientific, San Rafael, CA, USA). The effect of each ginkgolide concentration on current magnitude was normalized and presented as I/Imax, where I is the amplitude of the inhibited current and Imax is the amplitude of the current activated by glycine only. Statistical significance was determined by paired or unpaired Student’s t-test, as stated in the text, with p < 0.05 representing significance.
Functional characterization of GJ inhibition of recombinant GlyRs
We first investigated the subunit-sensitivity and functional properties of GJ inhibition as these have not previously been performed using electrophysiology. The inhibitory potency of GJ was compared at α1 homomeric and α1β heteromeric GlyRs. The GJ dose-responses were quantitated in the presence of an EC30 (20 μM) glycine concentration. Figure 1(b) and (c) shows example current traces in the presence of indicated GJ concentrations together with the averaged inhibitory dose response curves for each receptor. GJ inhibited the α1 GlyR with a mean IC50 of 5.3 ± 1.3 μM and an nH of 0.9 ± 0.1 (both n = 5). It was significantly more potent (p < 0.05, Student’s unpaired t-test) at the α1β GlyR, with a mean IC50 of 1.6 ± 0.5 μM and an nH of 0.7 ± 0.1 (both n = 5). Previous electrophysiological studies on recombinantly expressed α1 GlyRs (Jensen et al. 2007) and native neuronal GlyRs (Chatterjee et al. 2003) also reported IC50 values for GJ in the low micromolar range.
We next investigated the use-dependence and glycine concentration-dependence of GJ inhibition at the α1 GlyR. Use-dependence was investigated as shown in Fig. 1(d). The rate of recovery from inhibition by 3 μM ginkgolide was quantitated by fitting a single exponential to the recovery phase (as indicated by 2 in Fig. 1d). Although these single time constant fits oversimplify the complex activation process, we employ them here only to compare differences in current activation rates for the purpose of detecting bound ginkgolide molecules. Accordingly, we also fitted single exponentials to the glycine-only activation phase both without GJ pre-incubation (as indicated by 1 in Fig. 1d) and following incubation with 3 μM GJ in the closed state (as indicated by 3 in Fig. 1d). The three time constants (as indicated by 1, 2, and 3 in Fig. 1d) were all quantitated in each of six cells with the averages and statistical comparisons presented in Fig. 1(e). These results indicate that GJ can access its binding site only when channels are bound by glycine. Thus, GJ shares a use-dependent blocking mechanism with the other three ginkgolides (Kondratskaya et al. 2002; Hawthorne et al. 2006).
The glycine concentration-dependence of GJ inhibition was tested by comparing the percentage block caused by 3 μM GJ on currents activated by EC30 (20 μM) and saturating (2 mM) glycine concentrations. Averaged over five cells, a similar proportion of block at both glycine concentrations showed that GJ inhibition is non-competitive (Fig. 1f and g). GA, GB, and GC also inhibit homomeric GlyRs in a non-competitive manner (Kondratskaya et al. 2002; Hawthorne et al. 2006).
The R19′ residue lies at the external end of the M2 domain. We recently discovered that the GlyR α1 R19′C mutation modifies the pore structure so that substances that bind in the pore are unable to dissociate from their binding sites while the channel is closed but can dissociate efficiently when the channel is opened (Hawthorne and Lynch 2005). Pore-binding compounds are thus ‘trapped’ in the closed state. So far we have shown that picrotoxin (PTX), bilobalide and GA exhibit trap behavior (Hawthorne and Lynch 2005; Hawthorne et al. 2006), with evidence to date suggesting that all three compounds bind in the 2′–6′ region of the GlyR pore (Hawthorne and Lynch 2005; Hawthorne et al. 2006; Yang et al. 2007) We thus infer that any substance that exhibits trap behavior also binds in the pore. We would also infer that any substance that exhibits a different recovery rate after having been trapped in the pore is likely to adopt a different binding confirmation in the closed channel state. In the present study we used this approach to test whether ginkgolides adopt different binding conformations in the closed and open channel states. Figure 2 shows the outcome of these experiments. In the case of GA, we used a saturating (20 mM) glycine concentration and a 10 μM GA concentration. GA was first applied and removed from α1R19′C GlyRs that were maintained in the open conformation by the constant application of glycine (Fig. 2a, left). In a second experiment, glycine and GA were removed simultaneously after GA-induced block (Fig. 2a, middle). In this experiment, recovery from block was initiated only after the re-introduction of glycine, indicating that GA was effectively trapped in the pore. The recovery rates from GA block in the continued open state and in the open state following a 60 s trap were compared as shown in Fig. 2(a) (right). The recovery rate was significantly slower after trap, suggesting that GA is forced into a different binding conformation by channel closure. Similar experiments were performed with the other ginkgolides with the only difference being that 5 mM (EC50) glycine was used to reduce the recovery rate (Hawthorne and Lynch 2005) in order to detect small recovery rate differences. Ginkgolides B, C, and J were all applied at 10 μM. As shown in Fig. 2(b), GJ was also trapped in the closed state of the channel, although there was no significant difference in its recovery rate before and after trap. Thus, GJ seems to maintain its initial binding position even after being trapped in the closed state. Channels blocked by GB and GC showed little recovery even after several minutes, indicating that these ginkgolides were not able to dissociate efficiently from their binding sites in the open state (Fig. 2c and d). Similar results were observed in four cells exposed to GB and GC. Thus, recovery rates could not be quantitated for these ginkgolides.
Effects of 2′ and 6′ mutations on ginkgolide sensitivity
As our previous study employed relatively non-conservative, subunit-specific pore mutations (e.g., G2′P, P2′G, T6′F) (Hawthorne et al. 2006), it provided little detail about the chemical nature of binding interactions. In the present study we used a mutagenesis strategy designed to probe the binding mechanisms of the four ginkgolides. Four 6′ mutants were chosen for analysis: α1T6′S, α1T6′A, α1T6′G, and α1T6′V. The small β-branched sidechain of threonine exposes a polar hydroxyl group and a non-polar methyl group to the pore, giving it the ability to participate in both hydrogen bonds and non-specific hydrophobic interactions, respectively. Serine, generally considered a conservative substitution for threonine, maintains the reactive hydroxyl group attached to the β-carbon, although the hydrophobic methyl group is lost. The absence of a methyl group increases the accessibility of the hydroxyl group, possibly offering more opportunities for hydrogen bond formation. Valine maintains the β-branched sidechain structure but replaces the threonine hydroxyl with a second hydrophobic methyl. Both the hydroxyl and methyl groups are absent in alanine, which has a short, very non-reactive side chain. Glycine was employed to test the effect of completely removing the threonine side chain. While the single proton side chain of glycine is non-reactive, its small size can expose reactive backbone groups. At the 2′ position, we tested the ginkgolide sensitivity of the α1G2′A and α1G2′S GlyRs.
The glycine EC50 and nH values for most of these mutant GlyRs have recently been published (Yang et al. 2007). The only exception is the α1T6′G GlyR which exhibited mean glycine EC50 and nH values of 516 ± 119 μM and 1.81 ± 0.21 (both n = 4 cells), respectively. Ginkgolide inhibitory dose-responses were recorded in the presence of EC50 glycine concentrations for each mutant GlyR. These concentrations were as follows (in μM): α1T6′S, 1; α1T6′A, 6; α1T6′G, 500; α1T6′V, 400; α1G2′A, 30; and α1G2′S, 40.
We found that the T6′A and T6′V mutations completely eliminated the ability of all ginkgolides to block the pore. Shown in Fig. 3(a) and (b) are sample currents recorded from cells expressing α1T6′A and α1T6′V GlyRs. The α1T6′A GlyRs showed a transient potentiation upon initial application of a 30 μM concentration of any of the four ginkgolides. This potentiation decayed to near the control current level, thus providing no evidence for block (Fig. 3a). A transient inhibition was also observed when ginkgolides were removed. The α1T6′A GlyRs also consistently showed a transient potentiation upon removal of glycine before channel closure (e.g., Fig. 3a). Similar results were observed for all four ginkgolides in each of nine cells expressing α1T6′A GlyRs. Note that the off-transients generally exhibited significant rundown with repeated glycine applications and this may be the reason they are absent in the α1T6′A GlyR traces displayed previously (Yang et al. 2007).
The α1T6′V GlyR displayed a rapid and almost complete desensitization upon glycine application (Fig. 3b). Surprisingly, we observed that 30 μM concentrations of each of the four ginkgolides produced strong potentiation (e.g., Fig. 3b). Similar effects were observed for all four ginkgolides (data not shown). Averaged over four cells, 30 μM concentrations of GA, GB, GC, and GJ potentiated the EC50 glycine current by 70 ± 15, 65 ± 10, 77 ± 6 and 53 ± 18%, respectively. Using a paired t-test, none of these values are statistically different from each other.
Averaged inhibitory dose-response curves for each ginkgolide at the α1T6′S, α1T6′G, α1G2′A, and α1G2′S GlyRs are presented in Fig. 3(c–f) with mean parameters of best fit summarized in Table 1. Both the T6′S and T6′G mutations caused a statistically significant reduction in sensitivity to all ginkgolides relative to the wild type (WT) control (Fig. 3c and d, Table 1). Although the four ginkgolides were equipotent inhibitors of the α1WT GlyR, large variations in potency were present at both 6′ mutant receptors. Of particular interest, 100 μM GA produced no significant inhibition of the α1T6′G GlyR and 100 μM GJ produced no significant inhibition of the α1T6′S GlyR. Pooled results from both mutant GlyRs, which are discussed in detail below, do suggest any obvious pattern of interaction between the 6′ sidechains and the variable ginkgolide groups. The nH for GB inhibition of the α1T6′S GlyR was increased significantly relative to the α1WT GlyR, implying the existence of multiple GB inhibitory sites or an allosteric mechanism of inhibition.
|IC50 (μM)||nH||n||IC50 (μM)||nH||n||IC50 (μM)||nH||n||IC50 (μM)||nH||n||IC50 (μM)||nH||n|
|GA||5.9 ± 1.7||0.9 ± 0.1||5||3.0 ± 0.6||0.9 ± 0.1||3||4.0 ± 0.8||1.0 ± 0.2||5||121 ± 16***||1.2 ± 0.33||3||>200||5|
|GB||7.9 ± 1.8||0.9 ± 0.1||5||6.7 ± 2.7||1.1 ± 0.1||5||8.3 ± 1.8||1.2 ± 0.1||5||45.7 ± 5.3**,##||1.6 ± 0.2*||6||18.1 ± 3.4*||0.7 ± 0.1||8|
|GC||10.5 ± 0.8#||1.0 ± 0.1||5||4.5 ± 0.6*||0.9 ± 0.1||5||4.2 ± 1.4*||1.0 ± 0.1||5||32.6 ± 3.8**,##||0.8 ± 0.1||3||123 ± 26***,###||0.6 ± 0.1**||5|
|GJ||5.3 ± 1.3||0.9 ± 0.1||5||2.7 ± 0.6||1.0 ± 0.1||8||5.6 ± 0.8||1.3 ± 0.2||5||>200||5||45.8 ± 4.5***,##||1.3 ± 0.1*||5|
In contrast to 6′ mutations, the 2′ mutations had little influence on ginkgolide sensitivity (Fig. 3e and f). The only exception was a modest but significant increase in GC sensitivity at both the α1G2′Aand α1G2′S GlyRs (Table 1). As alanine and serine contribute to different bond types and GC incorporates hydroxyl groups at the R1 and R2 positions, this does not point to a specific interaction. These data suggest that the 2′ residue has little involvement in directly coordinating ginkgolides within the GlyR pore.
Do ginkgolides bind in the pore?
Several lines of evidence strongly suggest that ginkgolides in the 6′ pore-lining region. First, molecular docking simulations demonstrated the feasibility of GA binding in this region (Hawthorne et al. 2006). Second, ginkgolides display classic pore blocker attributes including use-dependence and non-competitive inhibition (Kondratskaya et al. 2002; Ivic et al. 2003; Hawthorne et al. 2006). Third, ginkgolides and PTX are both trapped in the α1R19′C GlyR in the closed state (Hawthorne and Lynch 2005; Hawthorne et al. 2006) (present study). Because PTX is known to bind in the 6′ pore-lining region (Yang et al. 2007), it is reasonable to conclude by functional association that ginkgolides bind in a similar location. Finally, as discussed in detail below, the variable effect of 6′ mutations on the sensitivity of different ginkgolides (Fig. 3) provides evidence for a direct interaction.
It is unlikely that the ginkgolide potentiation seen in the α1T6′V GlyR is mediated by a 6′ binding site as the ginkgolide diameter is equal to the pore diameter at this point (Hawthorne et al. 2006). On the basis of the anomalously low ginkgolide sensitivity of the fluorescence of a rhodamine label attached to R19′C, we recently postulated the existence of an additional low affinity ginkgolide site outside the pore (Pless et al. 2007). One possibility is that the non-conservative T6′V mutation alters M2 conformation in a manner that decrypts an external potentiating site.
Variable ginkgolide groups do not interact strongly with 2′ residues
The effects of the G2′S and G2′A mutations on ginkgolide potency were virtually insignificant, indicating that reactive ginkgolide groups are unlikely to interact strongly with 2′ residues. Our previous conclusion for a direct interaction between the R2 ginkgolide group and the 2′ residue involved analysis of glycine to proline mutations (and proline to glycine mutations) in the α1β heteromeric GlyR (Hawthorne et al. 2006). The highly non-conservative nature of this substitution coupled with the poorly conserved M2 domain sequences of the α1 and β subunits mean that the results of that study are unlike to pertain to the homomeric α1 GlyR investigated here. Together, these ginkgolide results fit well with a recent report characterizing PTX interactions in the homomeric α1 GlyR pore (Yang et al. 2007). That study found firstly that PTX was ligated by 6′ threonines but had little energetic interaction with 2′ pore-lining residues, and secondly that introduction of the β subunit altered the way in which PTX sensitivity was influenced by 2′ mutations.
Interactions between ginkgolides and the 6′ threonine
Ginkgolides possess two invariant OH groups plus two variable OH groups (at R1 and R2), any or all of which may be important for hydrogen bonding with T6′. Elimination of the T6′ hydrophobic bonding capability by the T6′S substitution caused a significant but highly variable reduction in the potency of all four ginkgolides. GA and GJ, which exhibited the most dramatic loss of potency, both have a non-reactive hydrogen atom at R1. One possibility is that the flexible serine hydroxyl may move sufficiently to coordinate GB and GC via their R1 hydroxyl, which may partially compensate for the loss of a crucial hydrophobic interaction between the T6′ methyl group and a moiety common to all ginkgolides. Alternatively, the R1 hydroxyl of GB and GC could allow an intermolecular bond with the shared R10 hydroxyl in the α1WT GlyR (Ivic et al. 2003). In the α1T6′S GlyR, it might be more energetically favorable for the R1 hydroxyl to form a hydrogen bond with the pore rather than with R10. Experiments aimed at testing the role of 6′ hydrophobic interactions (via the T6′V and T6′A mutations) were not informative as they rendered the ginkgolide inhibitory site dysfunctional.
Because GB and PTX exhibit strong similarities in the locations of key functional groups (Ivic et al. 2003), knowledge of the PTX binding mechanism (Yang et al. 2007) could may be useful in predicting the ginkgolide binding mechanism. Overlaying the molecular structures of PTX and GB in the α1 GlyR pore predicts that the R1 ginkgolide group lies in the vicinity of the 6′ residue and that the R2 group is close to the 2′ residue. This fits with the functional data from the α1T6′S GlyR described above. Thus, the four ginkgolides may bind in very similar orientations in the α1WT and α1T6′S GlyRs.
The pattern of IC50 values at the α1T6′G GlyR does not assist in clarifying the role of T6′ in ginkgolide coordination. GA and GC exhibited the most dramatic losses of potency although they differ in structure at both the R1 and R2 positions. GJ and GB, which were affected to a lesser extent, also differ in structure at both positions. These results strongly suggest that different ginkgolides bind in different orientations in the α1T6′G GlyR pore. It is difficult to predict how each might bind as 6′ glycine substitutions could cause complex changes to the binding pocket structure and chemistry by enhancing backbone flexibility and exposing reactive backbone groups.
It is relevant to consider why ginkgolides do not inhibit α1T6′A or α1T6′V GlyRs despite T6′A being a generally more conservative mutation than T6′G. Theoretical studies have postulated the existence of a hydrogen bond between the Oγ atom of serine and threonine and the i-3 or i-4 backbone carbonyl oxygen that induces a stereotypical kink in the α-helix backbone (Gray and Matthews 1984; Ballesteros et al. 2000). Indeed, such a bond involving 6′ serines or threonines is thought to be important for forming the central kink in the nicotinic acetylcholine receptor M2 domain (Sankararamakrishnan and Samsom 1994). An experimental analysis of a G protein-coupled receptor transmembrane helix provided evidence not only that serines and threonines induce a local helical kink but that valines and alanines do not (Govaerts et al. 2001). Because glycine essentially behaves as a swivel, the glycine substitution may achieve a similar kink. The observed correlation between 6′ mutation and the magnitude of its effect on ginkgolide sensitivity suggests that a backbone kink induced by 6′ threonines or serines may be essential for the formation of a high affinity ginkgolide inhibitory site.
Ginkgolide A may bind in multiple orientations
Figure 2(a) showed that GA dissociated at a slower rate after being trapped in the closed conformation of the α1R19′C GlyR. This is strong evidence that the GA binding orientation is different in the closed state. On the other hand, GJ dissociates at the same rate from both the open and closed channel conformations implying either that interactions initially formed by GJ in the pore remained fixed when the channel closed, or that its orientation in the closed state was energetically equivalent. A comparison of ginkgolide crystal structures revealed two of the six five-membered rings in the GA molecule deviate significantly from those of the other ginkgolides (Zhao et al. 2002). One possibility suggested by (Ivic et al. 2003) is that GA has the least number of hydroxyl groups, and hence its molecular conformation is not held as stiffly. This flexibility may be the reason why GA can take advantage of a lower energy state within the closed pore, whereas the greater rigidity of other ginkgolides may prevent this adaptation.
It is evident in Fig. 2(a) and (b) that the recovery from ginkgolide block in the α1R19′C GlyR exhibits both fast and slow components. This could result from different rates of unbinding from high and low affinity sites or from different rates of unbinding in liganded closed and open states. Furthermore, as noted above, GJ exhibits a finite rate of unblock in the closed unliganded state as noted above. As these mechanisms appear to be specific to the α1R19′C GlyR, we have not investigated them.
Ginkgolides are sensitive reporters of their binding site environment
Two studies have screened synthetic ginkgolide derivatives with the aim of identifying the structural requirements required for high potency at the GlyR (Jaracz et al. 2004; Jensen et al. 2007). Both found that minor alterations to structure were enough to eliminate potency, and that rigidity of the skeleton was important for activity. This fits well with the findings of the present study, with the exception that we have made minor modifications to pore structure rather than to ginkgolide structure.
The results of all three analyses suggest that variations in the number and position of attached hydroxyls, originally thought to be a useful tool for deducing ginkgolide binding orientation, may greatly impact on the ability of ginkgolides to bind. This greatly complicates the task of finding precise molecular binding determinants. If slight changes in bond angles because of subtle structural differences imposed by single non-conserved groups is the reason for significant variations in potencies, then it may be possible to synthesize molecules with enhanced subunit specificity and greater differences in specificity for a given channel isoform. This could lead to an optimized set of pharmacological tools for probing the roles of different GlyR isoforms.
As ginkgolides also block GABAARs (Huang et al. 2004), it is feasible that ginkgolide derivatives may also be developed as subunit-specific GABAAR antagonists. The GABAAR may be of particular interest because although most GABAAR subunits also contain threonines at the 6′ position, some subunits do not. For example, the mouse and rat ε subunits contain 6′ serines (Sinkkonen et al. 2000) and the rat ρ2 subunit contains a 6′ methionine (Zhang et al. 1995). In addition, mouse and human τ subunits contain 7′ serines (Sinkkonen et al. 2000). Because naturally occurring ginkgolides are highly sensitive to the identity of the α1 GlyR 6′ residue, it is possible they may be useful for pharmacologically identifying GABAARs incorporating ε, τ or ρ2 subunits.
We conclude that GA binds in different orientations in the closed and open states of the α1R19′C GlyR, which may reflect its enhanced flexibility relative to the other ginkgolides. Ginkgolide sensitivity was not strongly affected by conservative mutations to the 2′ pore-lining glycine. In contrast, the T6′S mutation affects the potency of ginkgolides in a pattern that suggests that a hydrogen bond between the variable R1 hydroxyl group and the 6′ serine is important for ginkgolide potency. However, there is no obvious pattern in the effect of the T6′G mutation on the ginkgolide potency sequence. The retention of high potency block in the T6′S and T6′G mutants and the abolition of block in the T6′A and T6′V mutants may be because of the elimination of a local kink in the M2 helix formed by hydrogen bonding between the sidechains of threonines and serines and backbone carbonyl oxygens. Together these results are consistent with previous studies showing that small variations in ginkgolide structure or pore structure can lead to large differences in potency. We speculate that this property may be exploited to create improved pharmacological probes for discriminating among GlyR and GABAAR isoforms with structural variations at the 6′ level.
We thank Dr Tim Webb for insightful comments on the manuscript. We also thank Ms Agnieszka Ney for expert assistance with site-directed mutagenesis and Mr Xuebin Chen for help with plasmid preparation and HEK293 cell transfection. This study was supported by research grants from the Australian Research Council and the National Health and Medical Research Council of Australia. JWL is supported by National Health and Medical Research Council Research Fellowship.
- 2000) Serine and threonine residues bend alpha-helices in the chi(1) = g(-) conformation. Biophys. J. 79, 2754–2760. , , , and (
- 2005) Ginkgolides and bilobalide: their physical, chromatographic and spectroscopic properties. Bioorg. Med. Chem. 13, 5001–5012. (
- 2006) Glycine receptors: recent insights into their structural organization and functional diversity. J. Neurochem. 97, 1600–1610. and (
- 2003) Structure-activity studies with Ginkgo biloba extract constituents as receptor-gated chloride channel blockers and modulators. Pharmacopsychiatry 36(Suppl 1), S68–S77. , and (
- 2001) The TXP motif in the second transmembrane helix of CCR5. A structural determinant of chemokine-induced activation. J. Biol. Chem. 276, 13217–13225. , , , , , , , and (
- 1984) Intrahelical hydrogen bonding of serine, threonine and cysteine residues within alpha-helices and its relevance to membrane-bound proteins. J. Mol. Biol. 175, 75–81. and (
- 2004) GlyR alpha3: an essential target for spinal PGE2-mediated inflammatory pain sensitization. Science 304, 884–887. , , et al. (
- 2003) Diversity of glycine receptors in the mouse retina: localization of the alpha3 subunit. J. Comp. Neurol. 465, 524–539. , , , , and (
- 2004) Diversity of glycine receptors in the mouse retina: localization of the alpha2 subunit. J. Comp. Neurol. 477, 399–411. , , , and (
- 2005) A picrotoxin-specific conformational change in the glycine receptor M2-M3 loop. J. Biol. Chem. 280, 35836–35843. and (
- 2006) Molecular determinants of ginkgolide binding in the glycine receptor pore. J. Neurochem. 98, 395–407. , , , and (
- 2007) Diversity of glycine receptors in the mouse retina: localization of the alpha4 subunit. J. Comp. Neurol. 500, 693–707. , , and (
- 2004) Ginkgolides, diterpene trilactones of Ginkgo biloba, as antagonists at recombinant alpha1beta2gamma2L GABAA receptors. Eur. J. Pharmacol. 494, 131–138. , , , , and (
- 2003) Terpene trilactones from Ginkgo biloba are antagonists of cortical glycine and GABAA receptors. J. Biol. Chem. 278, 49279–49285. , , , , and (
- 2004) Ginkgolides and glycine receptors: a structure-activity relationship study. Chemistry 10, 1507–1518. , , and (
- 2007) Probing the pharmacophore of ginkgolides as glycine receptor antagonists. J. Med. Chem. 50, 1610–1617. , , , , , , , , and (
- 2002) BN52021, a platelet activating factor antagonist, is a selective blocker of glycine-gated chloride channel. Neurochem. Int. 40, 647–653. , , and (
- 2005) The beta subunit increases the ginkgolide B sensitivity of inhibitory glycine receptors. Neuropharmacology 49, 945–951. , , and (
- 2004) Molecular structure and function of the glycine receptor chloride channel. Physiol. Rev. 84, 1051–1095. (
- 2005) Terpene trilactones from Ginkgo biloba: from ancient times to the 21st century. Bioorg. Med. Chem. 13, 4987–5000. (
- 2007) Conformational variability of the glycine receptor M2 domain in response to activation by different agonists. J. Biol. Chem. 282, 36057–36067. , , and (
- 1994) Kinked structures of isolated nicotinic receptor M2 helices: a molecular dynamics study. Biopolymers 34, 1647–1657. and (
- 2000) GABA(A) receptor epsilon and theta subunits display unusual structural variation between species and are enriched in the rat locus ceruleus. J. Neurosci. 20, 3588–3595. , , and (
- 2005) Refined structure of the nicotinic acetylcholine receptor at 4A resolution. J. Mol. Biol. 346, 967–989. (
- 2007) Molecular pharmacology of the glycine receptor chloride channel. Curr. Pharm. Des. 13, 2350–2367. and (
- 2007) A proposed structural basis for picrotoxinin and picrotin binding in the glycine receptor pore. J. Neurochem. 103, 580–589. , , , and (
- 2005) The glycinergic control of spinal pain processing. Cell. Mol. Life Sci. 62, 2027–2035. (
- 1995) Cloning of a gamma-aminobutyric acid type C receptor subunit in rat retina with a methionine residue critical for picrotoxinin channel block. Proc. Natl Acad. Sci. USA 92, 11756–11760. , , , and (
- 2002) Three ginkgolide hydrates from Ginkgo biloba L.: ginkgolide A monohydrate, ginkgolide C sesquihydrate and ginkgolide J dihydrate, all determined at 120 K. Acta Crystallogr. C 58, o195–o198. , , , , and (