(Received for publication, November 19, 1996, and in revised form, January 27, 1997)
From the 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.
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.
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.
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 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.
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-GluR3B
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-GluR3B
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.
All alanine-substituted
mutant versions of GluR3B were constructed as fusion proteins linked to
the 3 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.
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.
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-GluR3B 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-GluR3B 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-GluR3B 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
The increased binding of one antiserum, 2785, to the mutant V378Q
GluR3B The variability in amino acid usage
toward the peptide region GluR3B 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.
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.
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.
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
Division of Geriatrics,
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
Choice of Animals
-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-
-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).
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.
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-GluR3BS389 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)]
-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.
Production of Agonist-like Anti-GluR3 Antibodies
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-GluR3BS 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)]
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.
S389
(NEYERFVPFSDQQISND; Ub-GluR3B
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-GluR3B
S as described
under "Materials and Methods" (Table I).
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.
-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
-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-GluR3BN 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)]
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.
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.
Fig. 4.
The specificity of anti-GluR3B
immunoreactivity in antiserum 2784 is changed by a single amino acid
substitution. A, the GluR3BS389 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)]
*
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, -amino-3-hydroxy-5-methyl-4-isoxazole propionate.
©1997 by The American Society for Biochemistry and Molecular Biology, Inc.