Kainate-binding Proteins Are Rendered Functional Ion Channels upon Transplantation of Two Short Pore-flanking Domains from a Kainate Receptor*

Kainate-binding proteins belong to an elusive class of putative ionotropic glutamate receptors that to date have not been shown to form functional ion channels in heterologous expression systems, despite binding glutamatergic agonists with high affinity. To test the hypothesis that inefficient or interrupted signal transduction from the ligand-binding site via linker domains to the ion pore (gating) might be responsible for this apparent lack of function, we transplanted the short homologous linker sequences from the fully functional rat kainate receptor GluR6 into frog kainate-binding protein. We were able to generate chimeric receptors that are functional in the Xenopus oocyte expression system and in human embryonic kidney 293 cells. The linker domains A and B in particular appear to be crucial for gating, because a functional kainate-binding protein was observed when at least parts of both linkers were derived from GluR6. We speculate that to enable signal transduction from the ligand-binding site to the ion pore of the frog kainate-binding protein, the linker structure of the protein has to undergo an essential conformational alteration, possibly mediated by an as yet unknown subunit or modulatory protein.

The KBPs have attracted much attention because they were shown to bind agonists such as glutamate, KA, and domoate with high affinity (5,7,9,10), in contrast to the orphan receptors for which no ligand binding has yet been established (3). To date five different KBP genes have been identified from five different species: one subunit each from the frog Rana pipiens (5), chicken (4), duck (12), Xenopus laevis (7), and two different subunits from the goldfish Carassius auratus (6). Chicken and duck KBP are 92.8% identical at the amino acid level, indicating that they represent the same gene. The other subunits share between 49.8 and 67.9% sequence identity and thus are derived from different genes. When compared with other GluRs the KBPs exhibit one compelling structural difference that is characteristic for all KBPs: a short N-terminal domain of only 128 -148 amino acids, as opposed to ϳ520 amino acids for other GluR subunits.
The transmembrane topology of KBPs, however, was shown to be identical to that of other GluRs receptors (6). The pore domains of all five known KBPs have been investigated for intrinsic ion channel function by domain transplantation and were proven to be capable of conducting cations (11).
Because KBPs bind agonists with high affinity and possess functional ion channels, we speculated that inefficient "gating," i.e. inefficient or interrupted signal transduction between the ligand-binding site and the ion pore, might be responsible for the observed lack of ion channel function of KBPs. The sequences linking the ligand-binding domains to the three transmembrane domains (linkers A-C; see Fig. 1A) were considered prime candidates for such gating domains. Because the kainate receptor GluR6 is fully functional and 41% identical at the amino acid level to frog KBP (see Fig. 1B), we chose GluR6 as the donor for gating domains of proven functionality and transplanted putative GluR6 gating domains into frog KBP. This approach for the first time succeeded in generating a mutant kainate-binding protein with demonstrable ligand-gated ion channel function.

EXPERIMENTAL PROCEDURES
Mutagenesis-Rat GluR6(Q) and frog KBP wild type cDNAs were used for construction of chimeras between KBP and GluR6. In the present study, we define the connecting stretches of sequence between transmembrane domain (TMD) A and the S1 domain, between TMD B and the S2 domain, and between the S2 domain and TMD C as linkers A, B, and C, respectively (see Fig. 1A). The amino acid sequences of these three linker domains are GTNPGVFSFLNPLSPD (linker A, Glu-R6), AAESSYMFGFLNPFSKE (linker A, KBP), NLAAFLTVERMES (linker B, GluR6), SFASYINSNTNQT (linker B, KBP), EESKEASALG-VQN (linker C, GluR6), and GWVPVQPHT (linker C, KBP) (see Fig.  1B).
The chimeras KBP-(linkerA)GluR6 (frog KBP with the linker A derived from GluR6; see Fig. 2), KBP-(linkerB)GluR6 (frog KBP with the linker B derived from GluR6; see Fig. 2), and KBP-(linkerC)GluR6 (frog KBP subunit with the linker C derived from GluR6; see Fig. 2) were generated by PCR. They do not contain any amino acid alterations except for the respective inserted native GluR6 linker sequences (see Fig. 1B). We furthermore engineered KBP constructs containing combinations of two or all three linker domains of GluR6, resulting in the constructs KBP-(linkerAϩB)GluR6, KBP-(linkerAϩC)GluR6, KBP-(linkerBϩC)GluR6, and KBP-(linkerAϩBϩC)GluR6. For a graphic representation of the various linker transplantation constructs see Fig. 2. All of the mutated clones were sequenced across the PCR-amplified regions.
In preliminary experiments we had generated KBP-(linkerA)GluR6 and KBP-(linkerB)GluR6 constructs using a different approach that made use of intermediary constructs we had available from previous pore domain transplantation studies (11,13). In that approach, we used four introduced restriction sites at homologous positions in KBP and GluR6 to allow easy construction of chimeras between these two receptors (see Fig. 1A): KpnI and NdeI sites N-terminal of TMD A flanking linker A in KBP at amino acids 130 -132 (AAE) and 146 -148 (ELW), respectively, and in GluR6 at amino acids 516 -518 (GTN) and 531-533 (DIW), respectively; furthermore, in KBP NruI and EcoRV sites Cterminal of TMD B flanking linker B at amino acids 235-237 (YIN) and 241-243 (NQT), respectively, and in GluR6 at amino acids 627-629 (FLT) and 633-635 (MES), respectively. Numbering in each case starts with the first codon of the mature protein.
The mutant GluR6, which contains all four newly introduced restriction sites KpnI, NdeI, NruI, and EcoRV was named GluR6-PCL-PXL, and the equivalent KBP mutant was named KBP-PCL-PXL. These names are in keeping with our previously established nomenclature for mutants designed for ion pore exchange (13). They indicate that the constructs allow exchange of a "large" pore domain (a PCL, between sites NdeI and NruI) as well as an "extra large" domain (a PXL, between sites KpnI and EcoRV). These constructs, in addition to facilitating the exchange of ion pores, allow for easy exchange of linkers A and/or B, albeit at the expense of a few additional amino acid changes caused by the introduction of the four restriction sites (see Fig. 1B). Two additional KBP mutants were generated from KBP-PCL-PXL, which contain only the N-terminal linker A exchange sites KpnI and NdeI (KBP-PCL-PXL/N) or the C-terminal linker B exchange sites NruI and EcoRV (KBP-PCL-PXL/C).
To enable the exchange of PXL, KpnI and EcoRV restriction sites (see above) were introduced into KBP and GluR6, resulting in the constructs KBP-PXL and GluR6-PXL. These constructs were then used as parent clones to create KBP-PXLGluR6 (frog KBP with the extra large pore cassette derived from GluR6) and GluR6-PXLKBP (rat GluR6 with the extra large pore cassette derived from frog KBP), respectively (see Fig.  1B).
cRNA Synthesis-cRNA synthesis was performed as described earlier (14). Briefly, template DNA was linearized with a suitable restriction enzyme. cRNA was synthesized from 1 g of linearized DNA using an in vitro transcription kit (Stratagene) with a modified protocol that employs 800 M m7 GpppG (Pharmacia Corp.) for capping and an extended reaction time of 3 h with T7 polymerase. Trace labeling was performed with [ 32 P]UTP to allow calculation of yields and evaluation of transcript quality by agarose gel electrophoresis.
Electrophysiological Measurements in Xenopus Oocytes-Oocytes of stages V and VI were surgically removed from the ovaries of X. laevis as described elsewhere (15). The oocytes were injected with 10 ng of cRNA using a 10-l Drummond (Broomall, PA) microdispenser. Two-electrode voltage clamp recordings were performed 4 -8 days after cRNA injection with a TurboTec 10CD amplifier (npi, Tamm, Germany) by superfusion of the oocyte with glutamatergic agonists (300 M) prepared in normal frog Ringer's solution (115 mM NaCl, 1.5 mM CaCl 2 , 2.5 mM KCl, and 10 mM HEPES-NaOH, pH 7.2). Current electrodes were filled with 3 M CsCl and had resistances of Ϸ0.5-1.5 M⍀. Voltage electrodes were filled with 3 M KCl and had resistances of Ϸ 4 M⍀. The oocytes were held at Ϫ70 mV, and agonists (kainate and glutamate) were applied for 10 s at a flow rate of 10 -14 ml/min. To minimize receptor desensitization, bath pretreatment of oocytes with concanavalin A (ConA, 10 M for 8 min) preceded agonist application (15).
Labeling of Cell Surface Proteins Using Biotinylated ConA-To identify the fraction of receptor protein inserted in the plasma membrane, surface proteins were tagged with biotinylated ConA and isolated by streptavidin/Sepharose-mediated precipitation of the biotinyl-ConA-protein complex. Briefly, intact oocytes were incubated in 10 M biotinyl-ConA (Sigma) for 30 min at room temperature. After five 10-min washes in normal frog Ringer, intact oocytes were homogenized with a Teflon pestle in H buffer (20 l/oocyte; 100 mM NaCl, 20 mM Tris-HCl, pH 7.4, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride plus a mixture of proteinase inhibitors (Complete TM tablets; Roche Molecular Biochemicals)) and were kept at 4°C for 1 h on a rotating rod. After centrifugation for 60 s at 16 000 ϫ g, the supernatants were supplemented with 20 l of washed streptavidin-Sepharose beads (Sigma) and incubated at 4°C for 3 h on the rotating rod. The streptavidin-Sepharose beads were then pelleted by a 120-s spin at 1600 ϫ g and washed three times in H buffer. The final pellets were boiled in 40 l of SDS-PAGE loading buffer (0.8 M ␤-mercaptoethanol, 6% SDS, 20% glycerol, 25 mM Tris-HCl, pH 6.8, 0.1% bromphenol blue).
Gel Electrophoresis and Western Blotting-Proteins were separated on 20-cm SDS-PAGE gels (16). The gels were blotted (17) onto Hybond ECL nitrocellulose membranes (Amersham Biosciences). For development of the blots, a previously described protocol was followed (11). Because antibodies directed against KBPs are not commercially available, we used a polyclonal anti-KBP antiserum directed against the C terminus of KBP, which had been generated in the laboratory of Robert J. Wenthold. That antiserum showed stronger reactivity to KBP on Western blots than the monoclonal antibody that had been produced previously in our laboratory and that had been successfully used in the cloning of the frog KBP cDNA (18).
HEK 293 Cell Transfection-HEK cells 293 (ATCC number CRL 1573) were transfected with recombinant vector DNA using a modified Ca 3 (PO 4 ) 2 precipitation technique (19). Exponentially growing cells in 3-cm dishes with four coverslips (polylysine-coated) were transfected with 1 g of DNA. For precipitation, a mixture of 1 l of glutamate receptor DNA (1 g/l), 1 l of green fluorescent protein DNA (1 g/l), 5 l of CaCl 2 (2.5 mM), 50 l of 2ϫ BBS (50 mM BES, 280 mM NaCl, 1.5 mM Na 2 HPO 4 ϫ 2ϫ H 2 O, pH 6.95), and 43 l of H 2 O were incubated for 10 min at room temperature. Droplets of calcium phosphate-DNA were added to the growing HEK cells while gently shaking. The dishes were then incubated for 15-20 h at 35°C with 3% CO 2 . The cells were washed twice with 5 ml of medium (500 ml of minimum essential medium with Earle's salts from Invitrogen, plus 10% (v/v) fetal calf serum, 1.4 mM L-glutamine, 100 units of penicillin, and 0.1 mg of streptomycin) at 37°C. After transfection, HEK 293 cells were allowed to express receptor for 24 h at 37°C with 5% CO 2 .
Electrophysiological Recordings in HEK 293 cells-Whole cell recordings were performed using a HEKA EPC9 amplifier (HEKA Electronics) controlled by Pulse software (HEKA). Recording pipettes were pulled from borosilicate glass (WPI) using a Sutter P-97 horizontal puller. Ligand was applied using a U-tube that bathed the suspended cell in a laminar flow of solution, giving a time resolution for equilibration of 10 -30 ms (20,21). The external buffer consisted of 137 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl 2 , 1.0 mM MgCl 2 , 5.0 mM Hepes, pH adjusted to 7.2 with NaOH; the internal buffer was 120 mM CsCl, 20 mM N(Et) 4 Cl, 1.0 mM CaCl 2 , 2.0 mM MgCl 2 , 11 mM EGTA, 10 mM HEPES, pH adjusted to 7.2 with CsOH. The current responses were measured at room temperature (21-23°C) at a holding potential of Ϫ60 mV. Desensitization time constants were estimated from monoexponential fits of individual traces using the Microcal Origin program.
[ 3 H]Kainate Binding to HEK 293 Cell Membranes-HEK 293 cells transfected with KBP or KBP mutants were harvested into ice-cold 0.5 mM EDTA, 100 M 4-(2-aminoethyl) benzenesulfonyl fluoride, and phosphate-buffered saline. After centrifugation at 4000 ϫ g, the pelleted cells (from 10 -30 plates 10 cm in diameter) were homogenized with a Teflon glass homogenizer in ice-cold 50 mM Tris acetate buffer containing 10 mM EDTA, 100 M 4-(2-aminoethyl) benzenesulfonyl fluoride, 10 g/ml leupeptin, 10 g/ml aprotinin, and 1 g/ml pepstatin A and centrifuged at 8000 ϫ g. The supernatant was collected and centrifuged for 60 min at 10,000 ϫ g. The latter step was repeated twice more in NaCl-free 50 mM Tris acetate buffer at pH 7.3. After suspension and homogenization of the pellets in NaCl-free 50 mM Tris acetate buffer at pH 7.3, the membranes were frozen and kept in liquid nitrogen until use for the [ 3 H]kainate binding assay. The displacement curves were constructed by incubating on ice membranes (50 -150 g of protein) with 80 nM [ 3 H]kainate (58 Ci/mmol) in a total volume of 250 l of 50 mM Tris acetate buffer at pH 7.3 in the presence of increasing concentrations of unlabeled kainate (10 nM, 30 nM, 100 nM, 300 nM, 1 M, 3 M, and 10 M). After 60 min, the membranes were centrifuged at 12,000 ϫ g for 1 h, and the pellet was resuspended in 100 l of 0.5 M NaOH. After a 1-h incubation, 75 l of 12% acetic acid were applied for neutralization. The samples were counted with scintillation fluid (Lumax-xylene). The specific binding of [ 3 H]kainate was defined as the total binding minus the binding obtained in the presence of 1 mM kainate. All of the experiments were performed in triplicate. The binding data were analyzed using the PRISM program (GraphPad Inc., San Diego, CA).
All of the chimeras were expressed in both Xenopus oocytes and HEK 293 cells for electrophysiological analysis. Because it FIG. 1. A, schematic representation of the structure of a typical ionotropic glutamate receptor. The N terminus is extracellular, the C terminus is intracellular, and there are three TMDs, A, B, and C. The pore-forming region consists of a hairpin loop plus two small intracellular loops (L1 and L2). The hairpin loop inserts into the membrane from the cytosolic side and is thought to line the ion permeation pathway (the pore), whereas loops L1 and L2 connect the pore to TMDs A and B, respectively. A large, extracellular domain resides between TMDs B and C. S1 and S2 are two extracellular domains homologous to bacterial amino acid-binding proteins that interact to form the ligand-binding site. Linkers A, B, and C are here defined as the connecting sequences between the S1 and S2 domains and TMDs A, B, and C, respectively (see also Fig. 1B). KpnI, NdeI, NruI, and EcoRV are introduced restriction sites used for constructing some of the chimeras between GluR6 and frog KBP. B, amino acid sequences of linkers A, B, and C of wild type rat GluR6 and frog KBP and linker transplantation mutants used in this study. The overall sequence identity between GluR6 and frog KBP subunits is 41.2%. Listed are the amino acid sequences of linkers A (16 amino acids in low affinity kainate receptors/17 amino acids in frog KBP), B (13/13 amino acids), and C (13/9 amino acids) of wild type GluR6, frog KBP, three linker exchange constructs, and the PCL-PXL and PXL mutants of KBP and GluR6. Amino acids in bold type indicate differences in the PCL-PXL and PXL mutants compared with the respective wild type proteins. Amino acids that frame the linkers are underlined. The arrows indicate positions of introduced restriction sites (KpnI, NdeI, NruI, and EcoRV) used for constructing some of the chimeras between GluR6 and frog KBP. The restriction sites, when present, are indicated by dashed lines below the respective sequence.
Surprisingly, the construct KBP-PCL-PXL, which carries four introduced restriction sites and which was used in preliminary experiments as an intermediary construct to enable the quick exchange of linkers A and B between KBP and GluR6, also gave small currents, of 2.8 -5.2 nA (n ϭ 18), in Xenopus oocytes after application of glutamate or kainate (Fig. 3 and Table I). This result was verified by expression of KBP-PCL-PXL in HEK 293 cells where the mutant also expressed functional ion channels. In these cells the construct showed current amplitudes of 107 Ϯ 42 pA (n ϭ 5 responding cells, of 39 tested). Steady state current amplitudes could be recorded without application of inhibitors of desensitization such as cyclothiazide or ConA. Treatment with ConA did not lead to any potentiation of current amplitudes, unlike what is observed for kainate receptors. The desensitization time constant for the construct KBP-PCL-PXL was 827 Ϯ 332 ms (n ϭ 8 traces averaged from three cells, S.D. given) and thus was significantly larger than in GluR6(Q) wild type receptors (19 Ϯ 13 ms, n ϭ 12 traces from four cells, S.D. given).
KBP-PCL-PXL contains a total of eight amino acid point mutations in linker domains A (three amino acids mutated) and B (five amino acids mutated), caused by the introduction of the four restriction sites in the cDNA to prepare this cDNA for domain transplantations (see "Experimental Procedures"). The currents observed with KBP-PCL-PXL support the finding with KBP-(linkerAϩB)GluR6 that a combination of amino acids in the linker domains A and B is indeed critical for gating. We found that the separation of the mutated amino acids in linker A of KBP-PCL-PXL from the mutated amino acids in linker B leads to loss of ion channel function; chimeras KBP-PCL-PXL/N and KBP-PCL-PXL/C, which contain only the three amino acid mutations in linker A and the five amino acid mutations in linker B, respectively, could not be activated by kainate or glutamate (data not shown). This parallels the finding that in KBP-(linkerAϩB)GluR6 both linkers are required for function and supports our conclusion that linkers A and B interact directly or indirectly with each other. It also demonstrates that the linker domains are sensitive regions that indeed influence the gating mechanism.
The mutants KBP-(linkerAϩB)GluR6 and KBP-PCL-PXL showed clear responses in only two of five batches of oocytes tested. However, nonfunctional chimeras never gave any measurable current. A similarly inconsistent expression of functional receptors was observed in HEK 293 cells, where only 25 of 55 cells for wild type GluR6 and 5 of 39 cells for KBP-PCL-PXL gave currents in response to stimulation by kainate or glutamate.
Western blot data demonstrated that all mutants and chimeras, functional as well as nonfunctional ones, were expressed and incorporated into the oocyte plasma membrane. Expression levels of chimeras were generally comparable with those of the wild type KBP receptor subunit (Fig. 4). To rule out the possibility that the limited reproducibility of ionic currents depended on varying expression levels of an endogenous glutamate receptor subunit, Xenopus KBP, also called XenU1, that is known to be expressed in Xenopus oocytes (7, 24), we coexpressed each of the functional chimeras KBP-(linkerAϩB)-GluR6 and KBP-PCL-PXL with recombinant Xenopus KBP (11) in Xenopus oocytes. However, no currents could be measured under these conditions. This control experiment was repeated with the same result in a different batch of oocytes.
For KBP-(linkerAϩB)GluR6 and KBP-PCL-PXL, the observed gain of function might potentially be explained by a simple change in agonist affinities. with wild type KBP. The binding data for KBP-(linkerB)GluR6 and KBP-(linkerC)GluR6 are interesting results in themselves, because they show that regions other than the well established ligand-binding domains (25) can influence agonist affinities. The K d values of the functional chimeras KBP-(linkerAϩB)-GluR6 and of KBP-PCL-PXL are not significantly different from wild type KBP, thus excluding the possibility that an increase in agonist affinity could be responsible for the observed gain of function. By contrast, KBP-(linkerC)GluR6 has a significantly decreased agonist affinity (K d ϭ 104 Ϯ 12) compared with wild type KBP, which potentially could be the reason for the observed lack of function of KBP-(linkerAϩBϩC)GluR6).
By constructing functional GluR6 mutants containing the pore domains of all five known KBPs, we previously showed that all KBPs possess ion channel domains capable of supporting ion flux (11). In those constructs, the pore-forming regions of KBP subunits starting C-terminal of TMD A and ending N-terminal of TMD B were transplanted into GluR6 recipient subunits. These chimeras, named GluR6-PKBP, were functional ion channels. It was of interest to test whether an extension of the transplanted pore domain such as to include the adjacent linker domains A and B would still produce a functional chimera. This extended pore domain of the KBP, which we named PXL and which comprises the pore-forming region plus TMDs A and B as well as linkers A and B of the KBP from R. pipiens, was transplanted into GluR6, resulting in the chimera GluR6-PXLKBP (Fig. 1B). Like the other chimeras, this subunit was expressed and functionally analyzed in oocytes. However, neither after application of kainate nor after application of glutamate were any currents observed (data not shown). Thus, although the GluR6-derived linkers in the construct GluR6-PKBP evidently support gating of the transplanted KBP pore, the KBP-derived linkers in GluR6-PXLKBP do not support gating of their "own" pore, the cotransplanted KBP pore. This result, together with the results of the linker transplantation chimeras, suggests that the two pore-flanking linker domains indeed play a crucial role in the gating mechanism of KBPs. The reverse construct, KBP-PXLGluR6, a KBP subunit containing the extra large pore domain of GluR6 (Fig.  1B), unexpectedly also lacked ion channel function (data not shown), despite the fact that it contains the native GluR6 linkers A and B in addition to the GluR6 ion pore. We interpret this to indicate that it is the interaction between the ligandbinding domain, gating domains (linker domains), and ion pore domain, rather than any particular single domain, that enables successful gating.

DISCUSSION
The kainate-binding proteins are of interest because of their considerable sequence homology with mammalian glutamate receptor subunits and because high expression levels have been detected in nonmammalian vertebrate central nervous systems. Although researchers have put considerable emphasis on studying the molecular make-up of KBPs in the past, the physiological roles of these proteins are still unknown. Because these proteins are integral membrane proteins it appears unlikely that they serve as simple binding proteins for glutamatergic agonists, analogous to the recently discovered soluble glia-derived acetylcholine-binding protein that modulates synaptic transmission (26). Rather, it seems likely that they indeed serve some function in signal transduction across the plasma membrane.
Possible reasons for the lack of observable ion channel function of KBPs in heterologous expression systems in principle could be: 1) a lack of protein expression; 2) a lack of agonist binding; 3) an absence of a functional ion pore; or 4) a defect in translating the information of agonist binding into pore opening (gating). Protein expression of the recombinant KBP subunits has been demonstrated repeatedly (4,8,18,22,23). For the KBP chimeras in the present study, protein expression comparable with that of wild type was verified (Fig. 4). Similarly, high affinity agonist binding has been clearly shown for all KBPs (8,9,18,22,23). Furthermore, it has been demonstrated conclusively that domain swapping between different glutamate receptor subtypes is possible without loss of function (11,13,14,27). With this domain swapping technique, the ion pores of all KBPs have been assayed for function and, without exception, were proven to be capable of conducting currents (11).
Obviously, homomerically expressed KBP subunits fail to translate ligand binding into channel opening. It is assumed that ligand binding causes a conformational change in the extracellular domain of the receptor (25, 28 -30). This conformational change has been dubbed "venus fly trap" or "clamshell mechanism," alluding to the closure that, upon agonist binding, occurs between two extracellular domains. Although details of that mechanism remain controversial (25,28), it presumably represents the gating step required to open the ion channel. In recent publications by Armstrong et al. (25,30), data on the crystal structure of the ligand-binding domain of GluR2 complexed with kainate or other agonists strongly support the assumption that conformational changes take place after ago-  nist binding. For heterologically expressed KBP subunits, it appears that homomeric subunit assemblies either fail to generate the appropriate conformational change after binding of agonist or fail to communicate such change to the pore. In any case, this constitutes a disturbance in the gating mechanism.
Based on our previous studies showing that the pore domains of KBPs are inherently functional domains capable of fluxing ions (11), we used the frog KBP to investigate the function of the linker domains A, B, and C, which connect the TMDs and the ion pore domain to the ligand-binding domain. Our two functional KBP mutants, KBP-(linkerAϩB)GluR6 and KBP-PCL-PXL, for the first time demonstrate directly that the linker domains are involved in the gating mechanism of KBPs. It is not clear whether exclusively the linker domains A and B are important for gating or whether linker C plays an additional role. The latter is suggested by the chimera KBP-(linkerAϩBϩC)GluR6, which is nonfunctional but shows a K d value for kainate binding comparable with that of wild type KBP despite the facts that KBP-(linkerC)GluR6 has decreased kainate affinity and that KBP-(linkerAϩB)GluR6 agonist affinity is unchanged. It cannot be excluded off-hand that other regions of the receptor, previously undiscussed, play a role in gating. However, at least linkers A and B are clearly involved in and interact during the gating process, and, in wild type KBP, seem to be either disabled or extremely inefficient in triggering pore opening.
It is conceivable that, in the three-dimensional structure, the linker domains are in close proximity. The data on the linker chimeras KBP-(linkerAϩB)GluR6, KBP-(linkerA)GluR6, and KBP-(linkerB)GluR6 as well as on KBP-PCL-PXL, KBP-PCL-PXL/N, and KBP-PCL-PXL/C, indicate that amino acid sequences in the linker domains A and B influence each other. Thus, linkers A and B cannot be exchanged independently of each other and appear to be functionally coupled. Interestingly, if linker C of GluR6 is added to linkers A and B to produce the chimeric construct KBP-(linkerAϩBϩC)GluR6, the function gained by transplanting linkers A and B of GluR6 is lost again, suggesting that amino acids in linker C contribute to the gating domain structure. This latter conclusion is supported by additional data. In a recent publication, we presented evidence for the importance of the linker C domain of GluR7 for the gating mechanism (14). Single amino acid point mutations at three different positions within the linker domain C each converted the kainate receptor subunit GluR7, which, at least in oocytes, is nonfunctional under physiological conditions, into a clearly functional subunit. In a publication by Taverna  and in a sense may reconcile those apparently contradictory findings. We suggest that an appropriate linker B has to inter-act with an appropriate linker A to gate KBPs. It is certainly possible that KBPs and NMDA receptors differ in the mechanism of the linker structure required for successful gating. However, the contradictory data of Krupp et al. and Jones et al. together with our data may be taken to indicate that also for NMDA receptors there is not just one single determinant that mediates signal transduction between the ligand-binding site and the ion pore, as proposed by Krupp et al. (34) for linker A and by Jones et al. (33) for TMD B/linker B.
In another recent publication (35) transposon-mediated insertion of green fluorescent protein into GluR1 was used as a means to randomly fluorescence label the receptor protein.
Interestingly, many of the insertion mutants were functional ion channels despite the large protein domain integrated into the sequence. However, green fluorescent protein insertion into linkers A and C were as incompatible with function as insertion into the pore domain or the ligand-binding domain, underscoring the importance of functional integrity of the linker domains.
Furthermore, Sun et al. (36) in their comparative analysis of the crystal structures of the ligand-binding domains of wild type GluR2 and a nondesensitizing mutant GluR2 recently suggested a crucial role for the linker regions in channel activation. They proposed that the ligand-binding cores are organized as dimeric units in a tetrameric assembly where interactions of the dimer interfaces determine the level of desensitization of the receptor. The mechanistic scheme the authors present entails an agonist-triggered increase in the separation of the linker regions resulting in the activation of the pore if a stable dimer interface is present. Our experimental data on KBP linker regions presented here, which suggest the participation of both linkers A and B in the gating process, are perfectly in line with the hypothesis of Sun et al. Our data as well as those of Sun et al. stress the need for structural models incorporating not only the ligand-binding domain of the receptor but also the linker domains and, ideally, the ion channel domain.
An unexpected observation in the present study was the unreliable current expression of KBP/GluR6 chimeras. Only in two of five experiments (two of five batches of Xenopus oocytes) did KBP-(linkerAϩB)GluR6 and KBP-PCL-PXL show currents. In the two experiments in which these two constructs were functional, other chimeras were not. The question arises as to the reason for this phenomenon. Obvious explanations such as different concentrations of injected RNA or different protein expression levels can be excluded because these parameters were controlled for and monitored, respectively. We can also exclude a varying contribution by the endogenous Xenopus kainate-binding protein XenU1 (7, 24) being responsible for FIG. 4. Western blot demonstrating protein expression of the mutant frog KBP receptor subunits analyzed. Plasma membrane protein was labeled with biotinyl-ConA, solubilized, and then streptavidin-precipitated (11 oocytes/lane). Samples including controls from uninjected oocytes were separated on a SDS gel, Western blotted, and probed with a polyclonal antiserum against a C-terminal peptide of KBP (18). Arrows point to the position of the ϳ49-kDa band of wild type KBP and chimeras between KBP and GluR6. The asterisk denotes a known unspecifically cross-reacting protein band that is not found in control oocytes (11). this effect. Coexpression of the functional chimeras KBP-(linkerAϩB)GluR6 and KBP-PCL-PXL with recombinant Xenopus KBP (7) did not increase the probability of current expression. Because the phenomenon of limited reproducibility has never been observed before for any wild type or mutant glutamate receptor subunit investigated in our laboratory, we conclude that lack of reproducibility might be a characteristic feature of our chimeric constructs or even of KBPs in general. Recently, Ayalon and Stern-Bach (37) have shown that the assembly of glutamate receptor subunits is determined by the proximal extracellular N-terminal domain of the subunits. Exactly this part of the receptor is missing in all KBPs, which is a hallmark structural feature of these receptors. Lack of the N-terminal domain potentially could result in hampered assembly of subunits, which might explain varying expression levels and low current amplitudes.
Several genes encoding glutamate receptor subunits were identified in the genomes of Drosophila melanogaster and Caenorhabditis elegans by homology search with amino acid sequences that are present in the highly conserved ion pore domain (38 -40). Interestingly, in contrast to the other vertebrate glutamate receptor subtypes, the kainate-binding proteins have no sequence homologs in Drosophila and C. elegans (39). This might indicate that kainate-binding proteins are restricted to particular groups of species.
With regard to the in vivo function of KBPs, we speculate that an endogenous modulator or an endogenous receptor subunit needs to bind to KBPs, thereby causing a conformational change within the linker domains that in turn trigger pore opening. Such a mechanism of tweaking the linker domain into gating mode might represent the long-sought on switch for the ion channel function of KBPs and might be a special feature that sets KBPs apart from other glutamate receptor subunits. The functional mutants described here could thus prove useful in the search for this endogenous trigger mechanism and eventually for the elucidation of the physiological role of KBPs.