Identification of Amino Acids in the Glutamate Receptor, GluR3, Important for Antibody-binding and Receptor-specific Activation*

(Received for publication, November 19, 1996, and in revised form, January 27, 1997)

Noel G. Carlson Dagger §, Lorise C. Gahring Dagger par **, Roy E. Twyman **Dagger Dagger and Scott W. Rogers Dagger §**§§

From the Dagger  Salt Lake City Veteran's Administration Medical Center and Geriatrics Research, Education and Clinical Center, Salt Lake City, Utah 84148, the § Department of Neurobiology and Anatomy and the  Program in Human Molecular Biology and Genetics, Eccles Institute of Genetics, University of Utah, Salt Lake City, UT 84112, and the par  Division of Geriatrics, Department of Medicine, the ** Huntsman Cancer Institute, and the Dagger Dagger  Departments of Neurology and Pharmacology, University of Utah School of Medicine, Salt Lake City, Utah 84112

ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS AND DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES


ABSTRACT

We reported (Twyman, R. E., Gahring, L. C., Speiss, J., and Rogers, S. W. (1995) Neuron 14, 755-762) that antibodies to a subregion of the glutamate receptor (GluR) subunit GluR3 termed GluR3B (amino acids 372-395), act as highly specific GluR agonists. In this study we produced additional rabbit anti-GluR3B-specific antibodies, ranked them according to their ability to function as GluR agonists and characterized the immunoreactivity using deletion and alanine substitution mutagenesis. These anti-GluR3B antibodies bound to a subset of the residues in GluR3B (amino acids 372-386), of which glutamate 375, valine 378, proline 379, and phenylalanine (Phe) 380 were preferred. The level of GluR activation correlated with the binding of antibody to Phe-380, which suggests that immunoreactivity directed toward Phe-380 is an index for the anti-GluR agonist potential. Since the identity of this residue varies between respective GluR subunits, this suggested that this residue may be important for imparting antibody subunit specificity. To test this possibility, the alanine in GluR1 was converted to a phenylalanine, which extended the subunit specificity from GluR3 to the modified GluR1. We conclude that antibody contacts with key residues in the GluR3B region define a novel GluR subunit-specific agonist binding site and impart subunit-specific immunoreactivity.


INTRODUCTION

Ionotropic glutamate receptors (1) are activated following the binding of an agonist, such as glutamic acid, through highly specific interactions with amino acids within the receptor protein structure. We have reported (2-4) that autoantibodies specific to certain GluRs from patients with either Rassmusen's encephalitis (anti-GluR3) or antibodies prepared in rabbits injected with a bacterially derived fusion protein containing a portion of the GluR3 subunit can also function as direct agonists of a subset of non-NMDA1 GluRs expressed by cultured cortical neuronal cells. The specificity of the agonist-like antibody-receptor interaction contained within this polyclonal serum background was demonstrated when a synthetic peptide containing the GluR3 subunit sequence of amino acids 372-395 (termed GluR3B) blocked the anti-GluR3 antibody-activating property. A different peptide with the amino acid sequence 245-274 (termed GluR3A) failed to exhibit antagonism of this response. Two peptides from a subportion of the GluR3B epitope, GluR3B1 (amino acids 372-384) or GluR3B2 (amino acids 381-395), both failed to block the antibody agonist-like property, suggesting that amino acids important to the antibody-receptor interaction was interrupted by this division of the peptide.

Additional support for the "GluRB" peptide region serving as a subunit-specific epitope for binding by receptor-activating antibodies was discovered in a patient with advanced olivopontocerebellar atrophy (5). Unlike the anti-GluR3B IgG found in Rassmusen's encephalitis, this patient harbors autoantibodies to GluR2 that are of the multimeric IgM immunoglobulin class. However, similar to rabbit anti-GluR3 antibodies or autoantibodies to GluR3B in Rassmusen's encephalitis patients, this anti-GluR2 autoantibody functions as an agonist toward a subclass of CNQX-blockable GluRs expressed by cultured cortical neuronal cells. Furthermore, this activity is antagonized by preincubation with a synthetic peptide containing the GluR2 amino acid sequence, amino acids 369-393 (termed GluR2B), which is the region in GluR2 homologous to GluR3B that shares little sequence identity. Notably, GluR3B synthetic peptides failed to block the anti-GluR2 autoantibody agonist effect. Therefore, these studies support the hypothesis that antibodies prepared to the B-epitope region of a GluR subunit may function as highly specific receptor agonists. Determining which amino acids in GluRs that are most commonly bound by agonist-like antibodies and determining if these can be used to predict relative agonist efficacy and subunit specificity is central to understanding the molecular nature of an antibody-mediated agonist effect, the further development of GluR-subunit-specific agonists, and possibly the development of treatment strategies that have clinical relevance.

In this study, we have produced additional anti-GluR3B-specific antibodies in rabbits that display agonist-like activation of a subset of non-NMDA GluRs expressed by cortical neuronal cells in culture. In combination with deletion mapping and alanine substitution mutagenesis, we have resolved the minimum region of the receptor protein sufficient to produce anti-GluR antibodies with agonist-like properties and defined those residues most commonly important to antibody recognition and binding. Antibody-mediated receptor activation correlated with immunoreactivity to the phenylalanine at position 380, and this amino acid is demonstrated to be important for determining anti-GluR subunit immunospecificity. We suggest that anti-GluR3 antibodies to the GluR3B region define a novel glutamate receptor subunit-specific agonist site.


MATERIALS AND METHODS

Choice of Animals

New Zealand White female 4-6-pound rabbits were used for the present study. They were screened for pathogens and maintained in the University of Utah sanctioned animal research center. Food and water were provided ad labium and animals were maintained at constant temperature and light-dark cycle.

Antibody Production in Rabbits

Two sources of immunogen were used in the present study. As in earlier studies (2, 3), a portion of GluR3 (amino acids 245-457) was expressed in bacteria as a trpE-fusion protein extension. A second immunogen used the GluR3B peptide (amino acids 372-395) that was expressed in bacteria as a ubiquitin-peptide fusion protein as described in detail elsewhere (6, 7), but summarized as follows. Overlapping DNA oligonucleotides encoding the desired GluR3B amino acid sequence were synthesized, annealed, and subcloned into the AflII-HindIII site on the ubiquitin (Ub) expression plasmid pDS78/RBSii-Ub as a 3'-extension of the ubiquitin protein. In the initial construction of the GluR3B or GluR2B peptide extensions of Ub, an additional restriction site (NarI) was added to facilitate cloning of various versions of GluR3B (pUb3B). Generation of the NarI site at the 3'-end of ubiquitin adds a codon for alanine at the COOH terminus of ubiquitin. Bacterial strain AR13/pDMI was then transformed, and the fusion protein under control of the Lac promoter (LacP) was induced by addition of isopropyl-1-thio-beta -D-galactopyranoside (1 mM). The overexpressed fusion protein was purified to greater than 95% homogeneity as described (6, 7) and subsequently used in protein assays or as immunogen. For sizing chromatography a Pharmacia Superdex HiLoad 16/60 was employed that resolved the final ubiquitin-peptide fusion protein as a single peak. On average greater than 25 mg/liter of soluble ubiquitin-peptide fusion protein per respective construct was obtained. The integrity of the fusion peptide protein was confirmed by mass spectroscopy conducted at the University of Utah Mass Spectroscopy Facility (not shown).

Immunization

Ubiquitin- or trpE-GluR-fusion proteins of choice (described above) were emulsified in complete Freund's adjuvant and injected subcutaneously into the rabbit flank as before (2, 8, 9). For ubiquitin-peptide immunogen, each rabbit was injected with 100 µg of ubiquitin-GluR3B fusion protein of which approximately 25 µg is composed of the GluR3B peptide. Four weeks after the primary immunization, animals were boosted by subcutaneous injection of fusion protein emulsified in incomplete Freund's adjuvant. Animals were bled 12 days after boosting, and the serum was tested for immunoreactivity to the trpE-GluR3 and Ub-GluR proteins by Western blot (8, 9) ELISA or by electrophysiological analysis as described below. In this study, all serum was from rabbits that were boosted three times.

Antibody Screening and Analysis by ELISA

Soluble Ub-peptide proteins were quantitatively adsorbed to Immulon-4 ELISA plates (Dynatech Laboratories) using a concentration of 40 µg of immunogen/well. Plates were washed repeatedly with phosphate-buffered saline, blocked for 2 h at room temperature with 3% bovine serum albumin (Boehringer Mannheim) in phosphate-buffered saline, washed, incubated with immune serum diluted serially (see below) in 3% bovine serum albumin in phosphate-buffered saline for 2 h at room temperature. The plate was again washed and reacted with a goat-anti-rabbit secondary coupled to peroxide (1:7500; Jackson ImmunoResearch) for 1 h at room temperature, developed for peroxidase activity, and the plate read in a Molecular Dynamics ELISA dish reader as before (3). All samples were tested in duplicate wells and on multiple duplicate ELISA plates. Sample dilutions exhibiting specific anti-GluR3B immunoreactivity in the linear range (1:400 to 1:3200 except for rabbit antiserum 2785 which was 1:200 to 1:1600) were used in subsequent calculations. Background immunoreactivity to either Ub alone or to Ub-R2B was measured in parallel and served as the negative control (specific cross-reactivity to GluR2B was not observed in the antisera tested). Immunoreactivity to wild-type Ub-GluR3B or Ub-GluR3BDelta S389 was also measured in parallel as the positive control. Immunoreactivity toward each peptide was assessed after subtracting the background (generally less than 10% of the total anti-Ub-GluR3B immunoreactivity) from the immunoreactivity of the mutant and wild-type GluR3B peptides. For each antisera the ratio of the immunoreactivity toward each mutant peptide compared with the wild-type immunoreactivity was calculated (mutant immunoreactivity divided by the wild-type immunoreactivity). An index of the net decrease in immunoreactivity toward each GluR3B mutant compared with wild-type (Table I and Fig. 2) was calculated by subtracting 1 from this ratio. Thereby, a negative number indicates the net decrease in immunoreactivity while a positive number shows the net increase in immunoreactivity. Standard Student's t test statistical analyses were applied to samples as noted in the text.

Table I.

The effect of alanine substitutions on immunoreactivity toward GluR3B as measured by ELISA

The data reflect the net change in immunoreactivity in serum prepared to the trpE-GluR3 fusion protein toward the peptide Ub-GluR3BDelta S or this peptide in which alanine residues were substituted at the position indicated (see "Materials and Methods"). The rabbit antiserum number is found in the first column with the relative GluR3-agonist value in parentheses (see Fig. 1). The mean net change in immunoreactivity (bold values) and S.E. of the mean (light print) for each mutation collected from at least four tests conducted in duplicate is shown.


 


Fig. 2. ELISA measurement of immunoreactivity following alanine substitution mutagenesis. A, the immunoreactivity profile of antiserum 2784 to alanine-substituted Ub-GluR3BDelta S389 peptides. Values reflect the net decrease of immunoreactivity toward the mutant sequence compared with that for the wild-type sequence (see "Materials and Methods"). B, the mean decrease in immunoreactivity upon alanine substitution for each trpE-GluR3 antiserum (n = 6; increases were assigned a 0 value) is averaged to generate a "typical" immunoreactivity profile (see text). Also shown are the ELISA results from three additional mutants where valine 378 was converted to either a glycine (G), glutamine (Q), or arginine (R). The remaining single letter codes are as in Fig. 1. The results of a single-tailed Student's t test analyses that determine the significance of the observed decrease from 0 are shown by double asterisks for p <= 0.01 or by a single asterisk for p <=  0.05. Error bars reflect ± S.E. of the individual means.
[View Larger Version of this Image (33K GIF file)]


Electrophysiology

Anti-GluR immune serum was tested for agonist-like reactivity toward GluRs expressed on 3-5-week-old mouse E14-E16 cortical neuronal cell cultures as described previously (3). All sera were exchanged into electrophysiology buffer by repetitive washing of the serum with buffer using an Amicon 30-kDa cut-off filter to remove all small molecules. The final filter flow-through is used as the negative control and all responses are compared with 100 µM kainic acid. Specificity of receptor activation was confirmed by inclusion of the receptor antagonist, CNQX (40 µM), in the washed serum. Some Amicon filter units introduce a nonspecific, non-CNQX-blockable, component into the external buffer that produces a slow activating component (not shown). Samples exhibiting this activity were discarded.

Alanine Substitution Mutagenesis

All alanine-substituted mutant versions of GluR3B were constructed as fusion proteins linked to the 3'-end of the ubiquitin gene. Each mutant was generated using a PCR-based mutagenesis scheme with a pair of primers that anneal to the 5'- and 3'-coding regions of GluR3B. The upstream primer contained sequences encoding a NarI site 5' to the GluR3B NH2 terminus, while the downstream primer contained a HindIII site which followed the COOH-terminal sequences of the GluR3B coding region. The PCR product generated with this primer pair using pUb3B as a template was subsequently purified over a NucTrap® column (Stratagene), restricted with NarI and HindIII, and ligated into the pUb3B plasmid, which was previously digested with NarI and HindIII. For each mutant an (either the upstream or downstream primer depending on the location of the mutation) oligonucleotide was synthesized containing the desired nucleotide changes followed by 8-12 nucleotides of the wild-type GluR3B coding region. This mutant-containing oligonucleotide was then paired with the wild-type counterpart oligonucleotide in a PCR to generate the desired DNA fragment for cloning as described above. All mutant constructs were verified by DNA sequence analysis (DNA sequencing facility, University of Utah). Each mutant protein was subsequently expressed in Escherichia coli and purified as described above.


RESULTS AND DISCUSSION

Production of Agonist-like Anti-GluR3 Antibodies

In a previous study (3), we found that two rabbits (5209 and 5210) injected with trpE-GluR3 fusion protein produced antibodies with agonist-like properties when applied to GluRs expressed by cultured cortical neurons. Furthermore, the agonist-like properties of these polyclonal antibodies were blocked specifically by a synthetic peptide prepared to the GluR3B sequence (amino acids 372-395) (3). To increase our sample size for the present study, we injected six additional rabbits with trpE-GluR3 fusion protein as described in our original studies (2, 8) (see Fig. 1). The antisera used in this study were collected 12 days subsequent to the third boost, which was approximately 4 months following the initial injection. Immunoreactivity against GluR3 was measured toward the trpE-GluR3 fusion protein construct, where antibodies specific to trpE were removed (8) or to the Ub-GluR3B peptide fusion protein (see "Materials and Methods"). All six rabbits produced a robust antibody response to the GluR3 region of trpE-GluR3 fusion protein (not shown); however only five of six exhibited immunoreactivity to Ub-GluR3B (Fig. 1). The exception was rabbit 2786 that exhibited a robust and highly specific anti-GluR3 reaction that was consistent with immunoreactivity toward GluR3A (3) and other regions of the GluR fusion protein (not shown). Anti-GluR3B reactivity from rabbit 2786 was present, but was small (1:50 dilutions in Ub-GluR3B ELISA were required for detection, not shown) and could not be measured with sufficient reproducibly or confidence to be included in the present study. Immunoreactivity to GluR3B in the five responding anti-GluR3 rabbit antisera exhibited serum-comparable titers and immunoreactivity could be measured to dilutions of at least 1:5000 (not shown, see below). Antisera from rabbits that were prepared to Ub-GluR3B, or subregions of this peptide, are discussed below.


Fig. 1. Regions of GluR3 protein used as immunogens for the rabbit immunological response. On the left is a schematic representation of GluR3 (I-III designates proposed transmembrane domains; RL is the re-entry loop; and LS is the leader sequence) (see Ref. 12). The regions cloned into the respective bacterial overexpression systems (i.e. trpE-GluR3, residues 245-457; or Ub-GluR3B, residues 372-395 and Ub-GluR3BDelta S 372-388) are shown. Immunized rabbits are identified by number. All rabbits exhibited anti-GluR3 immunoreactivity, and the presence of antibodies to the GluR3B peptide region as measured by ELISA is indicated by a + sign. Rabbit 2786, identified by a - sign contained strong immunoreactivity to GluR3 (not shown), but immunoreactivity to the GluR3B region was poor. The relative efficacy of each antiserum for activating GluRs expressed on 3-week-old murine cortical neurons in culture was normalized to antiserum 5209, and this value is shown in the far right column. Rabbit 2786 exhibited low or no receptor agonist-like properties, and its relative efficacy was not numerically ranked (Not Ranked). The one letter codes for the amino acids shown are: A = alanine; D = aspartic acid; E = glutamic acid; F = phenylalanine; I = isoleucine; N = asparagine; Q = glutamine; R = arginine; S = serine; Y = tyrosine.
[View Larger Version of this Image (23K GIF file)]


Rabbit anti-trpE-GluR3 antisera were next tested for their ability to act as a GluR agonist. Serum from each rabbit was prepared for electrophysiological analysis by repeated washing with electrophysiology external solution (3) (see "Materials and Methods"). These sera were then applied at a dilution of 1:8 to 3-6-week-old neurons in culture prepared from E14-E16 CF1 mice. The evoked current peak amplitude of responding neurons was averaged from 4 to 24 neurons (median 13) per sample. This value was normalized to the response from rabbit antiserum 5209 (905 ± 79 pA (mean ± S.E.), n = 24), and they were ranked according to their relative ability to activate GluRs as shown in Fig. 1. Rabbit 5210 (2, 3) was not included in the present study due to the lack of sufficient quantities of serum for testing. As noted above, five of six rabbits produced antibodies that directly activated GluRs expressed by cultured cortical neuronal cells, and these antibodies also exhibited anti-GluR agonist-like properties. The GluR currents evoked by antiserum from rabbit 2786, which also exhibited poor immunoreactivity toward Ub-GluR3B, were too small or inconsistent to be confidently ranked, and this antiserum was not studied further.

The relative antibody anti-GluR agonist-like activity varied considerably between these rabbits, and it ranged from almost 150% to less than 25% relative to rabbit antiserum 5209. The notable diversity of agonist-like activity in these sera would predict that considerable variability might also result in the seizure phenotype reported previously for rabbit 5209 (2). However, to assure adequate antiserum supplies for the present study, animals were sacrificed and the serum collected prior to the onset of severe clinical manifestations as reported for rabbits 5209 and 5210 (2). This result also addresses the potential diversity to be anticipated in the anti-GluR3 autoantibodies from serum of patients with Rasmussen's encephalitis (2, 3). Consequently, further animal studies and the direct comparison of these results with those from ongoing clinical studies of Rasmussen encephalitis patients will require the strict implementation of quantitative behavioral analyses of injected animals and careful analysis of both antiserum efficacy and total immunoreactivity including the proportion of immunoreactivity to relevant epitopes.

A Subregion of GluR3B Is Sufficient as Immunogen for Anti-GluR Activating Antibodies

The possibility that a region of the GluR3 fusion protein other than GluR3B contains antigenic determinants important for antibody agonist-like activation was determined first. Three rabbits (designated 2787, 2788, and 2789, respectively) were injected with Ub-GluR3B protein as immunogen, boosted with immunogen three times, and the serum was collected. The Ub-GluR3B immunogen produced antibodies with immunoreactivity toward the native antigen and each exhibited agonist-like activity when applied to non-NMDA GluRs (Fig. 1). Antibodies from these rabbit sera were also immunoreactive toward the trpE-GluR3 fusion protein, but they failed to exhibit immunoreactivity toward trpE-GluR1, trpE-GluR2, trpE-GluR5, or UB-GluR2B protein (not shown). This result confirms that the peptide GluR3B is alone sufficient as immunogen to produce antibodies with GluR agonist-like properties. Similar to rabbits injected with trpE-GluR3 immunogen, the antibodies from rabbits injected with just the Ub-GluR3B constructs also exhibited variable agonist efficacy when compared relative to rabbit serum 5209 (Fig. 1).

We next determined the minimum GluR3B peptide size that retained full anti-GluR3B immunoreactivity. The GluR3B peptide consists of 24 amino acids (NEYERFVPFSDQQISNDSSSSENR) and antagonizes antibodies that activate GluRs expressed in culture (3). Subdivision of this region into two regions (NEYERFVPFSDQQ or SDQQISNDSSSSENR, respectively) failed to produce peptides with the antagonist properties of full-length GluR3B (3). This result suggested that either the entire GluR3B peptide sequence was required to antagonize the antibody-mediated activation of GluRs or that the specific epitope for antibody binding within the GluR3B peptide had been disrupted. To distinguish between these possibilities, we produced a panel of carboxyl-terminal deletion mutants of the GluR3B peptide (expressed as a Ub-extension peptide) and used these for immunogen. Rabbits injected with the GluR3B subsequence, NEYERFVPFSDQQISND (amino acids 372-388; termed Ub-GluR3BDelta S389; rabbits U301 and U302, respectively), were alone sufficient to produce anti-GluR immunoreactivity that had agonist-like properties (Fig. 1, see below). Since peptides lacking the carboxyl-isoleucine 385 and serine 386 failed to inhibit antibody-mediated GluR activation (3), and deletion of Asn-387 or Asp-388 did not effect anti-GluR3 immunoreactivity (see below), further deletions were not tested for their ability to induce agonist-like immunoreactivity.

Alanine Substitution Mutagenesis Defines GluR3 Amino Acids Important for Agonist-like Immunoreactivity

To define the amino acids in GluR3B required for antibody recognition and binding, we used alanine substitution mutagenesis (10, 11). Usually antigen-antibody interactions are best characterized using monoclonal antibodies with the desired properties, and the exact epitope usage of an agonist-like anti-GluR antibody will await its creation. Nevertheless, a first approximation of the residues important to binding in this region can be derived from the polyclonal background through applying a strategy that, like peptide blocking of specific immunoreactivity or affinity purification of antibodies to a desired region of the protein, can be made by assessing residue-specific binding of the dominant anti-GluR3B reactivity contained within the serum. To do this we measured immunoreactivity to just the GluR3B peptide through the use of the ubiquitin-fusion peptide overexpression system (see "Materials and Methods"). This system enabled us to generate large quantities of the mutant versions of the GluR3B peptide and to rapidly quantitate the relative immunoreactivity to only this protein region.

Using alanine substitution mutagenesis, we surveyed the amino acid sequence to determine those residues important for anti-GluR3B immunoreactivity. A series of mutant peptides were constructed in which each amino acid in the sequence (amino acids 372-386) was changed to an alanine using PCR-mediated mutagenesis (see "Materials and Methods"). Each fusion-peptide protein was then tested for retention of immunoreactivity to our panel of rabbit antisera using ELISA analyses. To reduce any possibility that the results of alanine substitution mutagenesis would be compromised by context effects of deletions at the carboxyl terminus of the peptide, we performed this mutagenesis in the peptide background of Ub-GluR3BDelta S389 (NEYERFVPFSDQQISND; Ub-GluR3BDelta S). Immunoreactivity to each mutant variant was then measured as depicted for rabbit serum 2784 in Fig. 2A and the quantitative measurement summarized for each trpE-fusion protein immunized rabbit antiserum as the change in immunoreactivity relative to wild-type Ub-GluR3BDelta S as described under "Materials and Methods" (Table I).

Each rabbit antiserum produced a unique, but overlapping, pattern of amino acid recognition. Consequently a "composite" immunoreactivity interaction profile for all of the anti-GluR3B antisera could be prepared by summing the average relative decrease of immunoreactivity following alanine substitution as shown in Fig. 2B. This plot reveals that immunoreactivity is most often directed toward residues E375A, P379A, and F380A. A concern regarding the possibility that a valine to alanine substitution was insufficient to produce a change in immunoreactivity was examined further by generating three additional amino acid substitutions at this site that included the nonconservative changes of V378G, V378Q, and V378R. In all cases, these substitutions either diminished or completely abolished immunoreactivity in the ELISA assays (Fig. 2B, Table I), suggesting that substitution of valine with amino acids that are less conservative in their relative properties (size, hydrophilicity, or charge) disrupts immunoreactivity. Of interest is that one of these substitutions (V378Q, antiserum 2785) produced the only increase in immunoreactivity toward a tested peptide that was observed in the present study. Antiserum 2785 exhibited a 160% increase in immunoreactivity specifically toward only the V378Q construct over wild-type (Table I).

An interesting result was the small but significant (p < 0.05) decrease (average decrease of 19%) in immunoreactivity against the I385A construct. Notably, the synthetic peptide that lacked this residue (amino acids 372-384) failed to antagonize the agonist-like antibody effect on GluRs (3), which, in combination with the present result, suggests that isoleucine 385 is important through a subtle effect on antibody binding or perhaps maintaining the structure of the GluR3B region to which antibody binds. Immunoreactivity toward the F380A construct is notable because the effect on immunoreactivity is greatest in the antisera that also exhibit the greatest GluR efficacy (e.g. 2784), whereas there is essentially no effect on the immunoreactivity of antisera that exhibit poor efficacy (e.g. 0290). This result and the potential relationship between the interaction of antibody with residue F380 and anti-GluR3 agonist-like efficacy is discussed in greater detail below.

Similar to the antisera prepared to the trpE-GluR3 protein, immunoreactivity in antisera to the Ub-GluR3BDelta S fusion peptide was reduced by the substitution of alanine for Glu-375, Val-378, or Pro-379, respectively (not shown). Also, similar to trpE-GluR3 immunized rabbits, these respective antisera exhibited varied use of amino acids for binding across the peptide sequence. In particular, the V378A mutation had little effect on immunoreactivity (<= 25% reduction), whereas substitutions with glycine, glutamine, or arginine essentially abolished essentially all reactivity. A notable difference between these antisera and those prepared to trpE-GluR3 immunogen was that immunoreactivity was diminished or essentially abolished by the alanine modification of Asn-372 and Gln-384. Consequently, this result supports the conclusion that this region alone is sufficient to impart the agonist-like activity of this antibody although different antibody-amino acid interactions may be favored when the immunogen fusion protein partner of GluR3B is changed.

These findings also hold implications for the structure of the GluR3B region. First, there are two possible N-linked glycosylation sites in the GluR3B peptide region (asparagines 387 and 394, respectively) (12), but neither seems important for antibody recognition, since their deletion does not effect immunoreactivity and antibodies bind to both synthetic peptides and bacterially produced fusion proteins that are not glycosylated. However, the possibility that glycosylation of one or both of these residues could actually inhibit or modify antibody interaction with the receptor has not been investigated. Second, the fact that residues important to immunoreactivity as measured by ELISA span as many as 14 residues (e.g. 5209) suggests that these amino acids may reside more proximally relative to each other in three-dimensional space. This would also be consistent with major histocompatibility class II systems for antigen presentation that often present discontinuous or relatively large epitopes from intracellular proteins that often span between 13 and 25 amino acids (13). Third, prolines are common in beta -turns at the protein surface, and they frequently participate in the transition between various secondary structural motifs or between protein structural domains (see Ref. 14). Consequently, the binding by anti-GluR3B antibodies to the proline 379 in GluR3B may play a significant role in defining the antibody-receptor interaction important to receptor activation. Notably, only antiserum 2785 failed to show any diminished immunoreactivity toward the P379A mutation. However, this immunoreactivity was sensitive to alanine substitutions in the adjacent residue, F380A, and it was the only antiserum that exhibited significant loss of immunoreactivity to the S381A construct. Given that prolines most commonly reside in the second position of the beta -turn (14), this would suggest that residues VPFS (amino acids 378-381) would compose this structure. Notably, the anti-GluR3B antibodies with agonist-like properties bind to at least two and up to three of these residues (Fig. 3). Finally, structural models of GluRs proposed by O'Hara et al. (15) and Stern-Bach et al. (16) place the GluR3B sequence at a hinge between the two prominent domains of the NH2-terminal portion of the receptor preceding the first membrane spanning segment that is adjacent, but not overlapping, with the highly conserved sequences that are thought to participate in ligand binding. Furthermore, we have found that deletion of 9 amino acids from the carboxyl-terminal segment of the original GluR3B sequence failed to alter the immunoreactivity of any antisera prepared to this region. This places the antibody binding region to GluR3B at least 14 amino acids toward the amino terminus away from the first region of amino acids that have been proposed to participate in ligand binding region as suggested by Stern-Bach et al. (16). This is also consistent with recent reports that expression of portions of the GluR extracellular domains that lack the GluR3B region retain their ability to bind [3H]AMPA (17). Collectively, this argues that the GluR3B region represents a novel non-NMDA receptor agonist-binding site. Activation of ligand-gated ion channels independent of the nature of the native agonist binding also occurs in GABA-A receptors when pentobarbital can directly activate chloride channels without interacting at the GABA binding sites (18). It remains to be determined if the antibody binding of the receptor at this region will fully mimic the conformational changes produced during glutamate, AMPA, or kainic acid receptor activation.


Fig. 3. Plot of the correspondence between relative anti-GluR antibody receptor activation and immunoreactivity to representative amino acids. The relative anti-GluR agonist-like activity of the antisera prepared in this study (see Fig. 1) was plotted against the decrease in immunoreactivity for four different alanine substituted mutants relative to wild-type (see Table I). The examples shown are for Tyr-374 (an amino acid that does not significantly participate in antiserum binding; Glu-375, Pro-379, and Phe-380 (amino acids that participate in antiserum binding to this region; see Fig. 2). Diamonds reflect values for antisera prepared to the trpE-GluR3 antigen. Squares represent antiserum prepared to the Ub-GluR3B antigen, and circles represent Ub-GluR3BDelta N prepared antisera that were measured in ELISA assays as for anti-trpE-GluR3 antisera (see text). Error bars reflect ± S.E. of the mean. The solid line is the calculated best fit linear regression line, which has a correlation coefficient of R2 = 0.813. The correlation predicted by this relationship is highly significant (p < 0.005).
[View Larger Version of this Image (34K GIF file)]


The increased binding of one antiserum, 2785, to the mutant V378Q GluR3BDelta S sequence (Table I) may suggest how molecular mimicry might occur (see Ref. 19) toward a protein that contains an epitope with a similar, but likely divergent, amino acid sequence. Interestingly, anti-GluR3 agonist-like autoantibodies from patients with Rasmussen's encephalitis (3) exhibited relatively rapid activation and relaxation kinetics on the receptor, suggesting relatively low affinity by antibodies specific to the GluR3B region. Consequently, the enhanced immunoreactivity toward mutant peptides such as V378Q may provide support to the notion that autoantibodies from patients with Rasmussen's encephalitis to the GluR3B region (3, 4) could actually be of high affinity toward an unrelated antigen that nevertheless closely resembles the GluR3B region. The collection of sufficient quantities of serum from enough patients to conduct the above analyses will address this possibility in future studies.

Immunoreactivity toward Amino Acid GluR3 Phenylalanine 380 Correlates with Relative Agonist-like Efficacy and Can Determine Antibody Subunit Specificity

The variability in amino acid usage toward the peptide region GluR3BDelta S and the distribution in relative antibody-GluR agonist efficacy offers the opportunity to determine if a simple correlation exists between these two variables. To test this, we examined if the absolute decrease in immunoreactivity toward any alanine substitution in Ub-GluR3B correlated with the relative agonist efficacy of these antisera (Table I). Only one of these tests, between agonist efficacy and the change in immunoreactivity upon alanine substitution of F380A, produced a convincing relationship between these parameters relative to other amino acids that were most commonly found to be important to anti-GluR3B immunoreactivity (Fig. 3). This observation suggests that efficacy may be predicted, or perhaps determined, by immunoreactivity toward Phe-380 but not other amino acids that are often involved in antibody binding to the GluR3B region. Nevertheless, the binding of this residue by antibody may not be absolutely required to activate receptors, since rabbit antisera 0290 and 2789 immunoreactivity was not altered by the F380A mutation, yet both antisera still activated GluRs albeit at relatively low efficacy.

The importance of Phe-380 to anti-GluR antiserum subunit recognition and specificity was also investigated. The alignment of sequences of the B-epitope region of GluR3 with the homologous region of GluR1 (amino acids 365-389) (12) shows that notable sequence identity and differences exist between amino acids in this region of these otherwise highly conserved subunits (Fig. 4A). Notably, antisera prepared to the GluR3B region showed no appreciable immunoreactivity toward the GluR1B region (Fig. 4B). This result was not unexpected since GluR1B differs in amino acid sequence identity with GluR3B at residues that alanine substitution mutagenesis revealed to be important for antibody recognition and binding. One such difference between the sequences of GluR3B and GluR1B is seen at amino acids 377-380 in GluR3B which is "FVPF" compared with the homologous region of GluR1B (amino acids 370-373) which is "FVPA" (Fig. 4A). Rabbit antiserum 2784 immunoreactivity was defined by alanine substitution to use amino acids that are in common with GluR1 except for Phe-380, which is an alanine in GluR1, and E375, which is an aspartic acid in GluR1 (see Figs. 2 and 4A). Although an exchange of a glutamic acid for an aspartic acid is a highly conservative substitution (14), the difference of an alanine for the phenylalanine at GluR3 residue position 380 is more severe. Therefore we tested the possibility that changing this alanine in GluR1 to phenylalanine would alter subunit specificity of this antiserum. The results of this experiment support the conclusion that GluR1B acquires anti-GluR3B immunoreactivity by the exchange of this single residue (Fig. 4). In contrast, the antiserum 2785, which requires Gln-383 in addition to Phe-380 for full immunoreactivity, failed to react with either GluR1B or the phenylalanine-substituted GluR1B (not shown). This result also predicts that these antisera would exhibit poor efficacy toward a receptor containing just GluR1. Unfortunately, consistent with the findings of Seeburg and colleagues (20), the relative poor expression of functional GluR3 homomeric receptors in transfected cell systems has thus far limited our ability to determine if the agonist-like properties of these antibodies is conferred by these simple amino acid substitutions in either GluR3 or GluR1 homomeric receptors. Nevertheless, the close correlation between immunoreactivity toward Phe-380 and agonist-like activity as well as subunit specificity suggests that the determinants of these properties may, like other agonist-binding sites, be determined by a small number of key amino acids.


Fig. 4. The specificity of anti-GluR3B immunoreactivity in antiserum 2784 is changed by a single amino acid substitution. A, the GluR3BDelta S389 protein sequence (amino acids 372-388) is aligned with the homologous sequence from GluR1 designated "R1B" (amino acids 365-384) (see Ref. 12). The highlighted amino acids are important to anti-GluR3B immunoreactivity for antiserum 2784. In the homologous position of Phe-380 in GluR3, an alanine is present in GluR1 (arrow). B, conversion of this phenylalanine to alanine in GluR3 (F380A) essentially abolished 2784 immunoreactivity as described earlier (see text); however, the reciprocal mutation in GluR1 (A373F) acquired immunoreactivity to the level measured for wild-type GluR3. Immunoreactivity to GluR1 wild-type sequence or GluR2B was equivalent to background in these assays.
[View Larger Version of this Image (49K GIF file)]


Collectively, these studies further define the physical interaction between antibodies that function as GluR agonists and the amino acids of the receptor required for imparting this effect. The critical amino acids involved in receptor binding and activation of the ion channel do not appear to comprise the proposed site of binding by traditional ligands (16, 17), and consequently these antibodies define a novel agonist binding site. The ability to alter subunit specificity of these antisera between GluR3 and the homologous region of GluR1 through conversion of a single amino acid that correlates well with antibody efficacy provides further evidence of the potential for these antibodies to function as highly specific GluR agonists and provide important information into how such reagents may be designed. Furthermore, it remains an important note that antibodies to other regions of the GluR3 protein were present in these rabbits (e.g. GluR3A (see above and Ref. 3)). This is also true of patients with Rasmussen's encephalitis (2, 3), and it argues for the position that a full understanding of how immune-mediated process toward GluRs progress in the nervous system will require that multiple variables be simultaneously assessed in addition to agonist-like activity in both animal models and the patient with disease. The well documented effects of antibody-mediated receptor down-regulation or complement-mediated lysis on receptors in diseases such as myasthenia gravis (see Ref. 21) could, in addition to, or separately from receptor activation, participate in disease pathology, etiology, and severity.


FOOTNOTES

*   This work was supported by National Institutes of Health Grants AA11418 (to L. C. G.) and NS35181 (to R. E. T. and S. W. R.), by a VA-Merit Award (to L. C. G.), and by the PEW Charitable Trust and the Esther and Joseph Klingenstein Foundation (to S. W. R.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§§   To whom correspondence should be addressed. Tel.: 801-585-6338; Fax: 801-585-3884; E-mail: srogers{at}genetics.utah.edu.
1   The abbreviations used are: NMDA, N-methyl-D-aspartic acid; CNQX, 6-cyano-7-nitroquinoxaline-2,3-dione; Ub, ubiquitin; ELISA, enzyme-linked immunosorbent assay; PCR, polymerase chain reaction; AMPA, alpha -amino-3-hydroxy-5-methyl-4-isoxazole propionate.

ACKNOWLEDGEMENTS

We thank Drew Wagner for his invaluable technical assistance. The additional expertise of Katrina Anderson and David Grimes is also appreciated. The ubiquitin overexpression system was generously provided by Dr. Martin Rechsteiner and Greg Pratt.


REFERENCES

  1. Hollmann, M., and Heinemann, S. (1994) Annu. Rev. Biochem. 17, 31-108
  2. Rogers, S. W., Andrews, P. I., Gahring, L. C., Whisenand, T., Cauley, K., Crain, B., Hughes, T. E., Heinemann, S. F., and McNamara, J. O. (1994) Science 265, 648-652 [Medline] [Order article via Infotrieve]
  3. Twyman, R. E., Gahring, L. C., Speiss, J., and Rogers, S. W. (1995) Neuron 14, 755-762 [Medline] [Order article via Infotrieve]
  4. Rogers, S. W., Twyman, R. E., and Gahring, L. C. (1996) Mol. Med. Today 2, 76-81 [CrossRef][Medline] [Order article via Infotrieve]
  5. Gahring, L. C., Rogers, S. W., and Twyman, R. E. (1997) Neurology 48, 494-500 [Abstract]
  6. Yoo, Y., Rote, K., and Rechsteiner, M. C. (1989) J. Biol. Chem. 264, 17078-17083 [Abstract/Free Full Text]
  7. Yoo, Y., and Rechsteiner, MC. (1990) Anal. Biochem. 191, 35-40 [Medline] [Order article via Infotrieve]
  8. Rogers, S. W., Hughes, T., Hollmann, M., Gasic, G., Deneris, E., and Heinemann, S. (1991) J. Neurosci. 11, 2713-2724 [Abstract]
  9. Rogers, S. W., Mandelzys, A., Deneris, E., Cooper, E., and Heinemann, S. (1992) J. Neurosci. 12, 4611-4623 [Abstract]
  10. Cunnigham, B. C., and Wells, J. A. (1989) Science 244, 1081-1085 [Medline] [Order article via Infotrieve]
  11. Jin, L., Fendley, B. M., and Wells, J. A. (1992) J. Mol. Biol. 226, 851-865 [Medline] [Order article via Infotrieve]
  12. Hollmann, M., Maron, C., and Heinemann, S. (1994) Neuron 13, 1331-1343 [Medline] [Order article via Infotrieve]
  13. Sant, A. J. (1994) Immunol. Res. 13, 253-267 [Medline] [Order article via Infotrieve]
  14. Creighton, T. E. (1993) Immunol. Res.Proteins, W. H. Freeman and Co., New York
  15. O'Hara, P., Sheppard, P. O., Thogersen, H., Venezia, D., Haldeman, B. A., McGrane, V., Houamed, K. H., Thomsen, C., Gilbert, T. L., and Mulvihill, E. R. (1993) Neuron 11, 41-52 [Medline] [Order article via Infotrieve]
  16. Stern-Bach, Y., Bettler, B., Hartely, M., Sheppard, P. O., O'Hara, P. J., and Heinemann, S. F. (1994) Neuron 13, 1345-1357 [Medline] [Order article via Infotrieve]
  17. Arvola, M., and Keinänen, K. (1996) J. Biol. Chem. 271, 15527-15532 [Abstract/Free Full Text]
  18. Amin, J., and Weiss, D. S. (1993) Nature 366, 565-569 [CrossRef][Medline] [Order article via Infotrieve]
  19. Ehrenstein, M., and Isenberg, D. (1991) Curr. Opin. Immunol. 3, 930-932 [Medline] [Order article via Infotrieve]
  20. Keinänen, K., Wisden, W., Sommer, B., Werner, P., Herb, A., Verdoorn, T. A., Sakmann, B., and Seeburg, P. H. (1990) Science 249, 556-560 [Medline] [Order article via Infotrieve]
  21. DeBates, M. H., and Vincent, A. (1993) in The Molecular Pathology of Autoimmune Diseases (Bona, C. A., Siminovitch, K. A., Zanetti, M., and Theofilopoulos, A. N., eds), pp. 1-14, Harwood Academic Publishers GmbH, Amsterdam

©1997 by The American Society for Biochemistry and Molecular Biology, Inc.