Go to Recent Publications
Our laboratory continues its interest in the molecules of membrane excitability--ion channels, 7-helix receptors, and neurotransmitter transporters. We are interested in these molecules individually, as part of interacting complexes, and as part of cells.
Over the past year, our structure-function studies on nicotinic acetylcholine receptors have taken on a new a new dimension with the development of methods to incorporate unnatural amino-acid residues in ion channels expressed in oocytes. This work is conducted in collaboration with the groups of D. Dougherty in Chemistry at Caltech, J. Abelson in Biology at Caltech, and P. Schultz at Berkeley. We foresee a number of applications, including (1) high-resolution mapping of ligand sites, (2) further information about ion interactions in the conducting pore, (3) exploration of membrane topology, and (4) design of new proteins with novel properties. We also found that conserved leucine residues govern nicotinic receptor gating in a surprisingly symmetric and independent fashion.
Neurotransmitter transporters have fascinating functional properties amenable to study by electrophysiology; they can be modulated by membrane trafficking in heterologous expression systems; and structure-function studies are possible using site-directed mutagenesis. The GABA transporter GAT1 has been characterized further with mutants and with concentration-jump experiments. Our experiments on the serotonin transporter have revealed a surprisingly complex set of permeation properties; we now find that actual single-channel event underlie some of these properties and in turn reveal possible conformational changes that produce coupled transport.
Our work on G protein-gated K+ channels, the GIRK/Kir3 family, is progressing with new information about heteromultimerization and details about ion permeation. We believe that GIRKs exemplify a class of proteins whose function can be understand in intact neurons only by acutely expressing and/or suppressing these proteins, followed by functional assays. These experiments are under way with adenovirus vectors as the major expression tools.
Our collaborative work with the Zinn laboratory on olfaction has resulted in the characterization of a new ion channel in olfactory epithelia. We are performing structure-function studies on this channel. We also continue to study the possibility of simplified assays for olfactory transduction.
Conserved leucine residues in the M2 domain of nicotinic receptors govern gating independently and symmetrically
Haiyun Zhang, Purnima Deshpande, Cesar Labarca
We are studying the gating mechanism of the nicotinic acetylcholine receptor (AChR) by site-directed mutagenesis of the M2 transmembrane region. Structural data for the nAChR suggest that a leucine residue at the approximate midpoint of the M2 transmembrane domain (the 9' position) occupies a "kink" in each of the 5 M2 helices (a, a, b, g, d) and points into the closed channel. In the open channel configuration, the M2 helices rotate so that Leu9' side chains no longer occlude the conduction pathway. We expressed mouse nAChR with varying numbers of Leu9' Ser mutated subunits in Xenopus oocytes. The dose-response relation shifted to the left by about tenfold for each mutation, so that a nAChR with four Leu9' mutations is 104-fold more sensitive than wild type. The results suggest that each of the 5 Leu9' residues participates independently and symmetrically in a key step in the structural transition between the closed and open states.
In single-channel studies with the Leu9' mutated receptors (a2bdg*, a2b*gd, a*2bdg, a*2bd*g receptors were tested), ACh evokes bursts of openings, within which longer channel open times and shorter shut times are observed. For the a*2bdg receptors, for instance, there are many brief (time constant ~0.15 ms) closings within a burst, so that the longest component of open time was <3-fold longer for the mutant receptor than for the wild type receptor. In a simple structural explanation of these data, mutant channels with 2 bound ACh molecules would remain open roughly as long as normal channels, then close, then rapidly reopen because the axial cluster of side chains that closes the channel is less stable with Ser than Leu. Single-channel current amplitudes (~6 pA at -100 mV in symmetrical 100 mM KCl solutions) differed by <10% among the receptors tested.
Mutations at many M2 position affect gating. The suggestion that Leu9' plays a unique role in receptor function led us to examine the consequences of mutating Leu16', which is also well conserved among AChR subunits and would lie on the same face of an a-helix as the 9' position. (aLeu16'Ser)2bgd receptors had EC50 values of 320 nM, near that for the corresponding Leu9'Ser mutation. However, Leu16'Ser mutations in b and g had little effect by themselves on dose-response relations, and none of the non-a Leu16'Ser subunits produced further shifts in the dose-response relation when combined with the aLeu16'Ser subunit. Thus the symmetrical interactions involving the M2 domains do not extend to the 16' position.
We are currently studying the effect of additional mutations and performing more detailed electrophysiological analysis in order to understand the gating mechanism of this receptor.
Domains of the muscle and neuronal acetylcholine receptor involved in ligand binding and channel gating
Mark W. Nowak, Patrick C. Kearney, Purnima Deshpande, Cesar Labarca, Margaret E. Saks, Jeffrey R. Sampson, John N. Abelson, Wenge Zhong, Scott Silverman1, Dennis A. Dougherty
We have continued research on the relationship between agonist binding and gating for the acetylcholine receptor (AChR) by introducing unnatural amino acids into AChR using synthetic "nonsense suppressor" tRNAs (1). A general outline of the methodology is as follows. We first introduce a TAG codon at the desired amino acid position of the AChR a subunit by site-directed mutagenesis. A "nonsense suppressor" tRNA with CUA at the anticodon loop and charged with the unnatural amino acid is then synthesized. Finally, we coinject Xenopus oocytes with the wild-type b, g, and d subunit mRNA transcripts along with the mRNA of the mutated a subunit and synthetic tRNA charged with the unnatural amino acid. Electrophysiological characteristics of the AChRs containing unnatural amino acids are compared to those of wild-type AChR.
Previous studies have suggested that tyrosine residues in the a subunit at positions 93, 190 and 198 contribute to the agonist binding pocket and may interact with acetylcholine (2). To further explore the role of these tyrosine residues in ACh binding to and gating of AChR we introduced a variety of unnatural amino acids at positions 93, 190 and 198 (see table). Receptors containing the unnatural amino acids displayed characteristics similar to wild-type receptors, including activation by micromolar concentrations of ACh, Hill coefficients near 2, inhibition by the cholinergic competitive antagonist d-tubocurarine and receptor desensitization.
Introduction of unnatural amino acids at position 93 established a prominent role for the 4'-OH group of the wild-type tyrosine. Replacement of the 4'-OH group with either H (Phe) or MeO (4'-MeO-Phe) caused a tenfold increase in the EC50 for ACh activation of AChR. To examine the role of the OH group in detail, we inserted a series of fluorinated tyrosine residues with a wide range of pKa values (10, 9.2, 8.7 and 5 for Tyr, 3-F-Tyr, 2-F-Tyr and F4-Tyr, respectively). Despite the range of pKa values, receptors containing the fluorinated tyrosine residues had EC50 values similar to those for the wild-type receptor (EC50 = 50 µM). The most straightforward interpretation of these findings was that the Tyr and the fluorinated Tyr residues are in the same protonation state and that this state is OH, not O-. Finally, insertion of 4-COOH-Phe, which can contribute an OH group but perhaps with suboptimal positioning, has an EC50 value only threefold that of wild type. Taken together, these findings indicated that an aromatic OH group at position 93 functions as a hydrogen bond donor.
The rank order of EC50 values for receptors containing unnatural amino acids at position 198 differs markedly from that observed at position 93. For example, replacement of Tyr with 4-MeO-Phe does not significantly alter the EC50 for ACh, suggesting that the OH group of Tyr 198 does not function as a hydrogen bond donor or acceptor. In addition, neither hydrophobicity scales nor any of the well-known parameters that model aromatic substituent effects rationalized the observed data. This suggested that the aromatic ring, not the 4'-substituent, was the primary contributor to ACh recognition at position 198. It can be further concluded that the residues at positions 93 and 198 interact differently with acetylcholine. Of course, some of the variations measured at these sites could have resulted from indirect effects of the side chain structure.
We next examined competitive inhibition by d-tubocurarine of receptors containing unnatural amino acids at position 198. In general, it was observed that replacement of the 4'-OH group of Tyr with less polar groups increases the affinity for d-tubocurarine. Consistent with previous suggestions (3) these data indicate that the interaction between the residue at position 198 and d-tubocurarine is predominantly hydrophobic.
At position 190, only Phe, 4-MeO-Phe and 4-NH2-Phe produced detectable responses; and the EC50 values were more than tenfold greater than that for wild-type AChR. Position 190, therefore, was the most sensitive to structural modifications of the three positions.
We have demonstrated that subtle chemical changes in the structure of an individual side chain resulted in readily detectable changes in the function of AChR, providing decisive tests for previous hypotheses concerning ligand-receptor interactions.
References:
1. M.W. Nowak et al., Science 268, 439 (1994).
2. M.J. Dennis et al., Biochemistry 27, 2346 (1988); S.N. Abramson et al., J. Biol. Chem. 264, 12666 (1989); J.-L. Galzi et al., ibid. 265, 10430 (1990); J.B. Cohen et al., ibid. 266, 23354 (1991).
3. M.E. O' Leary et al., Am. J. Physiol. 266, C648 (1994); S.M. Sine et al., J. Biol. Chem, 269, 8808 (1994).
Unnatural amino acid incorporation in the binding and pore regions of the nicotinic acetylcholine receptor
Patrick C. Kearney, Mark W. Nowak, Scott Silverman1, Wenge Zhong1, Margaret E. Saks, Jeffrey R. Sampson, John N. Abelson, Cesar Labarca, Dennis A. Dougherty
Using the ability to express in Xenopus oocytes functional receptors incorporating unnatural amino acids within their polypeptide sequences, we continue to explore the roles played by residues thought to be central to the binding and gating events of the nicotinic acetylcholine-gated receptor (AChR).
We previously identified a key role for the hydroxyl group of the tyrosyl side chain at position aY93, and for the aromatic group at position aY198. Presently, we have broadened our study of these effects by examining the dose-response relations of acetylcholine and tetramethylammonium (TMA) with AChRs containing bL262S and gL260S mutations (M29' position; see abstract by Haiyun Zhang et al) and unnatural amino acids at either the aY93 or aY198 positions. The bL262S, gL260S mutations are located in the pore region greatly decrease the EC50s of AChRs compared to the wild-type receptor. These modifications are used here, in order to lower the dose response relationship of the weak agonist TMA. To date, we have determined that the bL262S and gL260S mutations do not effect the rank ordering of EC50s of acetylcholine with receptors containing various aY93 mutations. TMA displays an identical ordering of EC50s with the mutant receptors, strongly suggesting that the acetyl moiety of acetylcholine does not interact with residues at the aY93 position. Studies of mutations at the aY198 position are in progress.
In the pore region, we have begun to replace the five highly conserved leucine residues at positions aL251, bL262, gL260 or dL265 with unnatural amino acids. It has been suggested that these residues form a hydrophobic plug, sealing the channel in the closed state, a view bolstered by recent structural and electrophysiological evidence (Unwin, 1995; see abstract by Haiyun Zhang et al). Using unnatural amino acids, we hope to obtain information regarding the arrangement and packing of these residues can be obtained. Also, it is hoped that these sites will prove suitable for testing chemistry capable of selectively trapping the receptor in the closed and open states. Presently, we are determining which of these positions are suitable for unnatural amino acid incorporation. Simple expression experiments carried out to date suggest the feasibility of substitutions on b but not a subunits.
References:
Unwin, N. (1995) Nature 373, pp. 37-43.
Ligand binding site studies on nicotinic acetylcholine receptor
Wenge Zhong, Mark W. Nowak, Patrick C. Kearney, Dennis A. Dougherty
Based on experiments with introduction of unnatural tyrosine derivatives acids at position 93, we recently suggested that all the tyrosine derivatives are in the protonated state as OH, not as O-. To test this idea, we sought to study pH effects on the 4-COOH-Phe mutant receptor, since there are many similar studies on pH effects on enzyme activities. We chose to use EC50 as the functional probe to evaluate the pH effects on the mutant in the pH range from 5.2 to 8.5. Preliminary efforts suggest such studies are feasible, providing a sensitive probe of the microenvironment of the agonist binding site.
We are also introducing tethered agonists into several positions in the N-terminal region (residues 180-200) in the a subunit as well as in the b, g, and d subunits. We have synthesized a tethered carbamylcholine based on Lys, and a tethered trimethylalkyl quaternary ammonium agonist based on Tyr. We are also planning to introduce photocleavable, photoswitchable tethered agonists, and tethered open channel blocker QX-222. It is conceivable that introduction of tethered agonists and channel blockers with variable tether lengths will provide distance and geometry information on the agonist binding site and pore region of the channel receptor.
Reference:
Nowak, M. W., Kearney, P. C., Sampson, J. R., Saks, M. E. and Labarca, C. G. (1994) Science 268, 439-442.
Expression of the extracellular domain of a nicotinic acetylcholine receptor
Anthony West, Dennis Dougherty
The a subunits of nicotinic acetylcholine receptors have an approximately 210 amino acid extracellular N-terminal domain that contains the agonist-binding site. The secondary and tertiary structure of this domain, as well as of the whole receptor, are unknown, although Unwin has obtained a 9 Å resolution structure of the receptor using electron microscopy. In hopes of obtaining high-resolution structural information, we have been exploring various strategies for obtaining the extracellular domain in a pure, soluble, and correctly folded state. We have been focusing on the mouse muscle a and chicken neuronal a7 subunits.
By expressing these extracellular domains fused to a bacterial secretion signal peptide, we were able to produce large amounts of these proteins in an insoluble state in the periplasm of E. coli. Attempts to refold this material in vitro led to soluble, but apparently misfolded protein, based on an a-Bungarotoxin binding assay.
To avoid protein-folding or refolding difficulties, we next used Xenopus oocytes to screen various constructs that contained truncated a subunits. One construct that we have examined appears to work well: the mouse muscle a extracellular domain. Related approaches have been used by the Bjorkman group and others to express extracellular domains of various membrane proteins which could not be expressed in a simple truncated form. We are now scaling up production of this protein in Chinese hamster ovary cells.
Voltage-jump relaxation kinetics for wild type and chimeric b subunits of neuronal nicotinic receptors
Antonio Figl, Cesar Labarca, Bruce N. Cohen
We have studied voltage-jump relaxations for a series of neuronal nicotinic acetylcholine receptors constructed by coexpression of wild type and chimeric b4/b2 subunits with a3 subunits in Xenopus oocytes. With acetylcholine or nicotine as agonists, wild type a3b4 receptors displayed five- to eight-fold slower and larger voltage-jump relaxations than did wild-type a3b2 receptors. In both cases, the relaxations could best be described by a two exponential-component fit, each component contributing approximately equal amplitudes over a wide range of [ACh]. Relaxation rate constants increased with [ACh] and saturated at 20- to 30-fold lower concentrations for the a3b2 receptor than for the a3b4 receptor, as observed previously for steady-state conductances. Furthermore, the chimeric b4/b2 subunits showed a transition in the concentration dependence of the rate constants in the region between residues 94 and 109, analogous to our previous observation with steady-state conductances. However, the series of b-subunit chimeras did not reveal localized residues that govern the absolute value of the kinetics or of the voltage sensitivity. Hill coefficients for the relaxations also differed from those previously measured for steady state responses. Relaxation kinetics were virtually unchanged when the a3 subunit was replaced by the a4 subunit. The data reinforce previous conclusions that the region between residues 94 and 109 on the b subunit plays a role in binding agonist but also show that other regions of the receptor control gating kinetics subsequent to the binding step.
Structure-function relations of cyclic nucleotide-gated channels
Jun Li, Jonathan Bradley, Kai Zinn
We are interested in the structure/function relations of cyclic nucleotide-binding domains of cyclic nucleotide-gated (CNG) channels. Most of the previous work involved site-directed mutagenesis, or probing the binding site with various cyclic nucleotide analogues. None of these experiments yielded conclusive results. Even combining the two approaches, one can still not rationalize, for example, differences in binding selectivity for cAMP and cGMP among channels.
Occasionally these experiments were aided by structural models, developed in light of the rather limited sequence similarities between CNG channels and the bacterial catabolite gene activator protein (CAP), the only cyclic nucleotide-binding protein with known atomic structure. But the computer predictions are ambiguous in pointing out key residues involved in binding.
We think a high resolution structure should be directly sought. We plan to express this domain as a cytoplasmic protein fragment, in the absence of the rest of the channel, then to assay for the ligand-binding capability of the expressed fragment. If the binding is measurable, we will proceed 1) to purify the protein, and attempt to crystallize it, and 2) to make mutations, and test the binding of each mutant. In contrast to mutagenesis on a complete channel, work on the binding alone allows the measurement of binding strength without the influence of channel gating.
Another planned experiment addresses the limitation of conventional mutagenesis: the finite repertoire of twenty naturally-occurring amino acids. It is now possible to incorporate unnatural amino acids into specific sites in a protein expressed in Xenopus oocytes. We can now put in residues with small structural variations in order to fine-tune the interaction between the ligand and its binding site.
Modulation of a GABA transporter expressed in Xenopus oocytes requires components of the secretory vesicle docking and fusion apparatus
Michael W. Quick, Norman Davidson, Henry A. Lester, Janis L. Corey1
Recent evidence indicates that a cloned GABA transporter, GAT1, is targeted to a regulated secretory pathway in Xenopus oocytes, and that protein kinase C (PKC) affects GAT1 transport activity by redistribution of transporters between cytoplasmic vesicles and the plasma membrane. We are presently elucidating components of this process. CaM kinase and calmodulin inhibitors decreased GABA transport; however, while CaM kinase inhibition reduced basal GABA uptake and eliminated the PKC-induced increase in transport, it did not alter the subcellular distribution of GAT1 or the PKC-induced translocation of GAT1 to the plasma membrane. Both the PKC and CaM kinase results suggested an interaction between GAT1 and components of the vesicular docking/fusion apparatus, so we used botulinum toxins to inactivate individual proteins known to be involved in the docking and fusion of vesicles during exocytosis. Inactivation of SNAP-25 and syntaxin-related proteins reduced basal GAT1 surface expression and prevented PKC-induced translocation. Inactivation of vesicular proteins synaptobrevin and cellubrevin prevented PKC-induced translocation but did not change basal GAT1 expression. Thus, expression and modulation of GAT1 in oocytes occurs via a pathway that uses components common to vesicle secretion. This system may help delineate individual components in vesicular docking/fusion and determine how transporters interact with this process.
1SIBIA, 505 Coast Boulevard South, La Jolla CA 92037-4641.
Ion binding and permeation at the GABA transporter GAT1 site
Sela Mager, Gilmor I. Keshet1, Nurit Kleinberger-Doron1, Darlene Gabeau, Baruch I. Kanner1
Transporters for small molecular weight neurotransmitter transporters are expressed on neuronal cell membranes and play an important role in synaptic transmission. The energy for the transport process is provided by the electrochemical potential of Na+; other ions such as Cl- and K+ may also contribute to the transport process. Nevertheless, the mechanism by which the ion flux is coupled to transport of the organic substrate is unclear. We are studying the ion binding and permeation at the GABA transporter site. The Na+ driving force was jumped rapidly by either 1) increasing the Na+ concentration (concentration jumps) or 2) applying more negative membrane potential (voltage jumps); in response, there was transient inward current. This current reflects the Na+ binding process. Replacement of Na+ by other monovalent cations such as Li+ and Cs+ transformed the transporter into an ionic channel and allowed ion permeation. The direct measurements of ion binding and permeation give new insight into the mechanism by which ions interact with the transporter and are coupled to the transport of the organic substrate. Presently we are combining this approach with site-directed mutagenesis to identify amino acid residues and domains involved in the ion binding and permeation. The ultimate goal of this work is to obtain a structure-function model of the transporter.
1Department of Biochemistry, Hadassah Medical School, The Hebrew University, Jerusalem, Israel.
Single-channel studies of the serotonin transporter: (A) Different conducting states and (B) an amino acid in the permeation pathway.
Fan Lin, Sela Mager
The mammalian 5-HT transporter is an important target for drugs of therapy and abuse. We have combined single-channel recording techniques and site-directed mutagenesis to study permeation properties of the cloned rat 5-HT transporter expressed in oocytes. For the first time, it is demonstrated that the permeation pathway of the transporter behaves like an ionic channel. There are at least two distinct transport-related single channel conducting states. 1) In the absence of 5-HT, the transporter displays an ohmic conductance of ~6 pS. When Na+ is replaced by Li+, the open probability of this state increases. 2) In the presence of 5-HT, the transporter appears to open a different permeation pathway. This transmitter-dependent ionic current requires the presence of Na+ and is blocked if Na is replaced by Li+.
We mutated asparagine177 in the third transmembrane domain to glycine. This mutation more than doubles the single-channel conductance of both states, without significantly changing the other kinetic parameters. Therefore, we conclude that this residue is located in the permeation pathway.
Are ABC transporters ion-coupled transporters?
Yongwei Cao
We are expressing several ABC transporters in Xenopus oocytes and mammalian cells. We have subcloned the human MDR1 cDNA and the human multidrug resistance-associated protein (MRP, another member of the ABC transporter family) into a modified expression vector that is designed for high levels of expression in Xenopus oocytes. Our preliminary results showed that oocytes injected with MDR1 mRNA displayed a 2-10 fold decrease in vinblastine accumulation compared with uninjected oocytes. Oocytes injected with MRP mRNA did not show significant increase in leukotriene C4 efflux. This is probably due to a high background of endogenous leukotriene C4 export in uninjected oocytes. However, MDR1-injected oocytes consistently showed a faster deterioration rate than that of MRP-injected or water-injected oocytes, indicating that the MDR1 protein is functioning even in the absence of externally applied substrates. Two- electrode voltage-clamp recordings of MDR1-injected oocytes occasionally revealed a small outward current (~10 nA) that was induced by vinblastine. Attempts are being made to increase this response and to characterize this electrical signal. In addition, we have subcloned the MRP cDNA into a mammalian expression vector and are generating a stably transfected HEK293 cell line that expresses MRP protein. With this cell line and other MDR-expressing cell lines, we will test the detailed physiology of transport in either intact cells or membrane vesicles.
Inhibition of function in Xenopus oocytes of the G protein-activated atrial K channel (GIRK1) by over-expression of a membrane attached form of the C-terminal tail
Nathan Dascal1, Craig A. Doupnik, Tatiana Ivanina1, Suzanne Bausch3, Weizhen Wang, Catherine Lin, Justine Garvey, Charles Chavkin2,3
Last year we reported the cloning and initial characterization of a G-protein activated K channel from rat atrial RNA. This gene, now known as GIRK1, is a member of the IRK or inward rectifier K channel family. The monomer subunits of these channels all have a short cytosolic N-terminal domain, then a conserved segment consisting of two transmembrane domains flanking the highly conserved pore region common to all K channels (which is believed to consist of a loop inserted into the membrane), all followed by a longer C-terminal cytosolic domain. The functional channels are believed to be either homo- or heterotetramers. The C-terminal cytosolic region, consisting of amino acids 183-501, is longer for GIRK1 than for most members of the IRK family. Work in several laboratories has indicated that it is Gbg subunits that gate the opening of GIRKs. Several lines of evidence indicate that the C-terminal segment of GIRK1 plays a role in G-protein binding and also interacts with the conducting pore.
In order to explore these interactions further, we have coexpressed, in Xenopus oocytes, GIRK1 and the 183-501 segment that has been engineered to be a peripheral membrane protein. Membrane attachment is provided by an N-terminal myristoylation signal that is provided by the first fifteen amino acids of the p60src protein fused in frame to amino acid 183.
The important result is that coexpression of an excess of the src+ (183-501) with a lesser amount of GIRK1 gives a marked decrease in hormone evoked GIRK1 currents in oocytes. Certain other members of the IRK family show a weaker degree of inhibition when this segment is coexpressed. Immunological analyses confirm that there is a very high concentration of this segment at the plasma membrane of the oocyte along with lesser amounts of the full length GIRK1 protein. Our results are consistent with the hypothesis that the 183-501 segment functions as a blocking particle, analogous for example to the N-terminal peptide of certain rapidly inactivating Shaker type K channels; in addition, it may inhibit gating of the channel by competitive binding of Gbg subunits.
These results suggest a general method for creating dominant negative truncation mutants of GIRKs.
1Department of Physiology and Pharmacology, Sackler School of Medicine, Tel Aviv University, Ramat Aviv 69978, ISRAEL.
2Computation and Neural Systems, California Institute of Technology, Pasadena, CA 91125.
3Department of Pharmacology, University of Washington, Seattle, WA 98195.
Intrinsic gating properties of a cloned G protein-activated inward rectifier K+ channel
Craig A. Doupnik, Nancy F. Lim, Paulo Kofuji
The voltage, time, and K+-dependent properties of a G protein-activated inwardly rectifying K+ channel (GIRK1/KGA/Kir3.1) cloned from rat atrium were studied in Xenopus oocytes under two-electrode voltage clamp. During maintained G protein activation and in the presence of high external K+ (VK = 0 mV), voltage jumps from the VK to negative membrane potentials activated inward GIRK1 K+ currents with three distinct time-resolved current components. GIRK1 current activation consisted of an instantaneous component that was followed by two components with time constants tf ~50 ms and ts ~400 ms. These activation time constants were weakly voltage-dependent, increasing ~2-fold with maximal hyperpolarization from the VK. Voltage-dependent GIRK1 availability, revealed by tail currents at 80 mV following long prepulses, was greatest at potentials negative to VK and declined to a plateau of approximately half the maximal level at positive voltages. Voltage-dependent GIRK1 availability shifted with VK and was half-maximal at VK - 20 mV; the equivalent gating charge was ~1.6 e-. The voltage-dependent gating parameters of GIRK1 did not significantly differ for G protein activation by three heterologously expressed signalling pathways: m2 muscarinic receptors, serotonin 5HT1A receptors, or G protein b1g2 subunits. Voltage dependence was also unaffected by agonist concentration. These results indicate that the voltage-dependent gating properties of GIRK1 are not due to extrinsic factors such as agonist-receptor interactions or G protein-channel coupling, but instead are analogous to the "intrinsic" gating behaviors of other inward rectifier K+ channels.
A unique P-region residue is required for slow voltage-dependent gating of a G protein-activated inward rectifier K+ channel
Paulo Kofuji, Craig A. Doupnik
G-protein activated inwardly rectifying potassium channels (GIRKs) exert an important role on cardiac and neuronal excitability. We have continued to investigate the structural determinants for the voltage- and time-dependent gating a G protein-activated inwardly rectifying potassium channel, GIRK1/KGA (Kir3.1), by heterologous expression of chimeric constructs and point mutants in Xenopus oocytes. As reported previously we have found a residue in the pore-lining domain of GIRK1 (P-region) critical for the manifestation of the time- and voltage-dependent channel gating. This residue, phenylalanine 137, when mutated to serine leads to expression of currents with nearly time-independent activation instead of the characteristic slow time dependent activation on step changes of membrane potential to more negative values. While this work was in progress others have reported that in another inwardly rectifying potassium channel, IRK1, an acid residue (aspartate) in the putative second membrane spanning domain partially controls the time and voltage-dependent gating in these channels by affecting the binding of intracellular polyamines to the internal side of the channel pore. To investigate whether a similar gating mechanism is present in GIRK1, we mutated the equivalent aspartate 173 to glutamine in GIRK1. Currents evoked from this mutant channel still displayed the characteristic time-dependent activation, although with a decreased degree of inward rectification. These results reveal an important role, not previously described, of the P-region in controlling the gating of an inwardly rectifying potassium channel and suggest a close relation between permeation and gating in this family of K+ channels. We are currently investigating the role of intracellular factors (e.g., magnesium, polyamines, ATP) in GIRK1 gating by recording of macroscopic currents in excised macropatches from oocytes expressing wild-type GIRK1 and F137S GIRK1 channels.
Interaction between G protein bg dimers and a family of G protein-gated inward rectifier K+ channels
Craig A. Doupnik, Vlad Slepak, Melvin I. Simon, Carmen Dessauer1, Alfred Gilman1
Receptor activation of G protein-gated inward rectifier channels is rapid (t ~300 ms), does not involve diffusible cytoplasmic second messengers, and can be mimicked by cytoplasmic application of purified or recombinant Gbg subunits, but not Ga subunits. These findings are consistent with a direct interaction between Gbg dimers and the Kir channel protein. Recent cloning of several Kir channels has revealed a subfamily (Kir3.0) of Kir channel subunits (Kir3.1-3.4) that, like the native channels, are gated by Gbg dimers released from heterologously expressed 7-helix receptors or by coexpressed Gb1g2 proteins. Amino acid sequence comparisons of Kir3.0 channel subunits with other signal transduction molecules known to interact with Gbg dimers fail to yield domains of significant similarity. Comparisons with other non-G protein-activated (constitutively active) Kir channels show sequence divergence primary at the amino- and carboxy-terminal regions. Hence, channel domains involved in Gbg interactions are initially suspected to reside in the C- and N-termini. To investigate the interaction between Kir3.0 channel proteins and Gbg heterodimers directly, we have engineered and expressed glutathione-S-transferase (GST)-Kir3.1 fusion proteins that contain either the N-terminal portion of Kir3.1 [a.a. 1-84], the Kir3.1 C-terminal tail [a.a. 184-501], or the fused N- and C-termini together [a.a. (1-84)-(184-501)]. The Kir3.1 fusion proteins each contain a (His)6 tag designed to immobilize the protein to a biosensor chip via biotinylated Ni2+-nitriloacetic acid bound to streptavidin-linked dextran on the surface of the chip. Using the BIAcore facility, interactions between perfused recombinant Gb1g2 proteins and the immobilized Kir3.1 proteins are quantitatively measured by surface plasmon resonance. Similar fusion proteins of Kir3.2, Kir3.3, and Kir3.4, as well as constructs of a non-G protein-gated Kir channel (i.e., Kir2.1) are being synthesized. Using site-directed mutagenesis, we plan to investigate 1) channel domains necessary for Gbg coupling, and 2) modulatory sites (consensus phosphorylation sites) that may affect channel-Gbg interactions. The functional consequences of positive mutants will then be studied electrophysiologically on heterologously expressed full-length channel proteins.
1Department of Pharmacology, University of Texas Southwestern Medical School, Dallas, TX 75235.
A putative Kir3.1 channel domain involved in G protein bg coupling
Craig A. Doupnik, Catherine Lin, Weizhen Wang1
To identify putative regions of Kir3.1 involved in Gbg subunit interactions, we constructed and expressed in Xenopus oocytes a series of deletion/substitution mutant Kir3.1 channels that consist of incrementally mutating 10 a.a. of wild-type sequence to a hydrophilic "tether" sequence (AKGAQGAGDG) over a 170 a.a. segment in the carboxy terminus. Starting from the C-terminal end, nine 10 a.a. deletion/substitution mutants covering a.a. 432 to 363 produced G protein-gated K+ currents similar in amplitude to wild type as determined by activation of co-expressed muscarinic m2 receptors. Thus residues 432-363 are not expected to directly participate in Gbg interactions. Beginning with mutant *(353-362), there was a significant reduction (~60%) in the amplitude of the ACh-evoked current compared to wild-type responses. Mutants *(343-352), *(333-342), *(323-332), *(313-322) all had responses <5% of the wild-type response and were essentially nonfunctional. Mutant *(284-293), however, displayed functional currents ~40% the amplitude of wild-type responses. Together, these results indicate a domain from a.a. 313 to 352 as essential for functional Kir3.1 channels. Since the nonfunctional mutant channels could result from nonspecific structural perturbations such as unassembled channel subunits, the involvement of this region in Gbg coupling is only suggestive. Corresponding mutations in GST-Kir3.1 fusion proteins and the effects on interaction with recombinant Gb1g2 proteins (see previous Abstract by Doupnik et al) will be tested to examine directly the role of this channel domain in Gbg coupling.
1Present address: University of California, Los Angeles, CA 90024.
Evidence that neuronal G protein-gated inwardly rectifying K+ channels are activated by Gbg subunits and function as heteromultimers
Recently, complementary DNAs encoding three isoforms of a G protein-coupled inwardly rectifying K+ (GIRK) channel have been cloned from cardiac (GIRK1/Kir 3.1) and brain cDNA libraries (GIRK2/Kir 3.2 and GIRK3/Kir 3.3). We are currently investigating the mechanisms of activation and the relationship among these various inwardly rectifying K+ channels. Injection of synthetic RNAs encoding GIRK2 channel and m2 type muscarinic receptor in Xenopus oocytes showed agonist evoked currents characteristic of GIRK channels. Furthermore, overexpression of G protein subunits Gb1 and Gg2 with GIRK2 but not GIRK3 elicited K+-sensitive inwardly rectifying currents independent of receptor activation suggesting the involvement of G proteins Gbg subunits in channel activation.
However, when either GIRK3 or GIRK2 was coexpressed with GIRK1 and activated either by muscarinic receptors or by Gbg subunits several effects were noted. First, G protein-mediated inward currents were dramatically increased in amplitude (5 to 40 fold) with new kinetics upon step changes of membrane potential. Second, the unitary channel properties on GIRK1 and GIRK2 coexpression were distinct from those observed for GIRK1 alone or for GIRK2 alone. These studies suggest that formation of heteromultimers involving the several GIRKs is an important mechanism for generating diversity in expression level and function of neurotransmitter-coupled inward rectifier K+ channels.
The future goals are: 1) To investigate the dose-response relationship for exogenously applied Gbg in inside-out macropatches from oocytes expressing the various combinations of GIRK channels; 2) to determine the subunit composition for the GIRK channels in various cell types by using polymerase chain reaction methods, and 3) to determine the viability of using the GIRKs as a reporter for expression cloning of new seven-helix receptors (e.g., GABAB)
Transient expression of inwardly rectifying K+ channels in HEK 293 cells
Lisa DiMagno, Jennifer Stary
We are attempting to optimize the transient expression of inward rectifying K+ channels in mammalian cells using adenovirus as an expression vector. Current work involves transfecting HEK 293 cells not only with inwardly rectifying K+ channels but also with the T-cell antigenic determinant CD8. Our approach utilizes plasmids that contain ORF sequences of both an inward rectifier K+ channel and the a subunit of the human CD8 gene. Two different constructs are being used: 1) a plasmid containing both genes each under the control of their own promoter and polyadenylation sequence and 2) a plasmid utilizing the CITE (cap independent translation enhancer) sequence to express both genes each under the control of only one promoter and polyadenylation sequence. Therefore, when we select transfected HEK 293 cells which fluoresce when labeled by either an FITC conjugated or R-PE conjugated CD8 antibody, we also know that the cells are expressing the inward rectifier K+ channel as well. These fluorescing HEK 293 cells can then be characterized electrophysiologically. To study the expression of these inward rectifier K+ channels in the rat brain, stereotactic injections of these constructs, sense and antisense, will also be performed.
Antisense cDNA inhibition of G protein-gated K+ channels expressed in Xenopus oocytes
Markus U. Ehrengruber, Lou Byerly, Youfeng Xu, Catherine Lin, Jennifer Stary
We plan to use antisense cDNA plasmids to construct recombinant adenovirus for in vivo suppression of G protein-gated inwardly rectifying K+ channels (GIRKs), in order to identify their roles in neuronal function. In preparation for these experiments, we are cloning sense and antisense cDNAs for GIRKs (types 1, 2, and 3) into expression plasmids that are then used for construction of adenoviral recombinants. We will first test these plasmids by nuclear injection into Xenopus oocytes, where GIRK expression can easily be measured by two-electrode voltage clamp.
To understand the principles of expression by nuclear injection of eukaryotic expression plasmids into oocytes, our test system is the Shaker K+ channel gene. A volume of 10 nl/oocyte is injected in all studies. Good expression of Shaker currents (several µA) has been achieved by injecting 0.1 ng plasmid DNA/oocyte; surprisingly, higher amounts (>1 ng/oocyte) give poorer expression. We also find that coinjection of other expression constructs, especially those encoding for membrane proteins, greatly suppress the Shaker current. The problem therefore is to recognize a specific antisense effect against this non-specific background.
Coinjection of GIRK1 or GIRK2 (0.1 ng/oocyte) with muscarinic ACh receptor type 2 (equal amount) causes expression of ACh-activated K+ currents of up to 1 µA. As with Shaker, coinjection of additional plasmids suppresses GIRK expression. However, coinjection of GIRK1 and GIRK2 gives an ACh-sensitive inwardly rectifying K+ current which is 13-fold larger than the current given by injection of an equal amount of either plasmid alone. This result agrees with the conclusion of Kofuji et al. (see abstract by Kofuji) that GIRK1 and GIRK2 form heteromultimers. In our first test of antisense suppression of GIRK currents, antisense GIRK2 cDNA suppressed the expression of GIRK2 three times more than it suppressed Shaker expression. We are now testing various antisense GIRK cDNAs for specificity and for the most effective ratios of antisense to sense. These studies will indicate which of the antisense cDNAs can be used to construct recombinant adenovirus for suppression of GIRK currents in neurons.
Incorporation of unnatural amino acids into inward-rectifier potassium channels
Scott K. Silverman, Dennis A. Dougherty
We are extending the nonsense codon suppression method for incorporation of unnatural amino acids (see Abstract by Nowak et al.) to potassium channels. As our first targets we have focused on inward-rectifier potassium channels. We initially investigated rescue of wild-type ROMK1 channels expressed heterologously in Xenopus oocytes. Separate replacement of two wild-type tyrosine codons with TAG (stop) and suppression with a designed tRNA synthetically acylated with tyrosine established rescue of wild-type function. Current work is on GIRK1, a G-protein-coupled inward-rectifier potassium channel. Initial studies are in progress for wild-type recovery by nonsense codon suppression at a critical tyrosine residue in the K+ channel pore region signature sequence (TXXTXGYG). The ability to substitute a variety of aromatic and nonaromatic amino acids at this position will provide information on the origin of potassium selectivity in this channel. Future work is planned for residues contributing to the rectification and G-protein-subunit binding of GIRK1.