Abstract
The main immunogenic region (MIR) is a conformation-dependent region at the extracellular apex of α1 subunits of muscle nicotinic acetylcholine receptor (AChR) that is the target of half or more of the autoantibodies to muscle AChRs in human myasthenia gravis and rat experimental autoimmune myasthenia gravis. By making chimeras of human α1 subunits with α7 subunits, both MIR epitopes recognized by rat mAbs and by the patient-derived human mAb 637 to the MIR were determined to consist of two discontiguous sequences, which are adjacent only in the native conformation. The MIR, including loop α1 67–76 in combination with the N-terminal α helix α1 1–14, conferred high-affinity binding for most rat mAbs to the MIR. However, an additional sequence corresponding to α1 15–32 was required for high-affinity binding of human mAb 637. A water soluble chimera of Aplysia acetylcholine binding protein with the same α1 MIR sequences substituted was recognized by a majority of human, feline, and canine myasthenia gravis sera. The presence of the α1 MIR sequences in α1/α7 chimeras greatly promoted AChR expression and significantly altered the sensitivity to activation. This reveals a structural and functional, as well as antigenic, significance of the MIR.
Introduction
Myasthenia gravis (MG) and experimental autoimmune myasthenia gravis (EAMG), are caused by antibody-mediated autoimmune responses to nicotinic acetylcholine receptors (AChRs) which impair neuromuscular transmission (Lindstrom, 2000). At least half of the autoantibodies in both MG and EAMG are directed at the main immunogenic region (MIR) on AChR α1 subunits (Tzartos et al., 1998). The MIR is defined by the ability of a single rat mAb to inhibit binding of many autoantibodies from MG patients or rats with EAMG (Tzartos and Lindstrom, 1980; Tzartos et al., 1982, 1983). The α1 sequence 66–76, the MIR loop, is crucial to the MIR (Gullick and Lindstrom, 1983; Das and Lindstrom, 1989; Saedi et al., 1990; Tzartos et al., 1990; Tzartos et al., 1998). The antigenicity and myasthenogenicity of the MIR depend greatly on the native conformation of the AChR (Lindstrom et al., 1978; Lindstrom and Einarson, 1979; Lennon et al., 1991; Im et al., 2000).
mAbs to the MIR can passively transfer EAMG into experimental animals (Tzartos et al., 1987; van der Neut Kolfschoten et al., 2007). These mAbs exhibit the primary pathological activities of serum antibodies: complement-dependent focal lysis of the postsynaptic membrane (which destroys AChRs and disrupts synaptic morphology) and antigenic modulation (which reduces the number of AChRs through crosslinking AChRs and thereby increasing their internalization). The orientation of the MIR at the outer perimeter and away from the central axis of the AChR explains why mAbs to the MIR are very effective at crosslinking adjacent AChRs and triggering antigenic modulation (Conti-Tronconi et al., 1981; Beroukhim and Unwin, 1995).
Antibody competition experiments reveal that the antibody repertoire in MG patients is similar to that in EAMG rats immunized with purified AChRs (Tzartos et al., 1998). Like human MG, canine MG also has a high proportion of autoantibodies to the MIR (Shelton et al., 1988). Some rat mAbs to the MIR also bind in a conformation-dependent manner to human neuronal α3, α5, and β3 subunits (Kuryatov et al., 1997; Wang et al., 1998), but autoantibodies from MG patient sera do not react with human neuronal AChRs (Vernino and Lennon, 2004). This implies that these rat mAbs and human MG autoantibodies recognize different epitopes within the MIR.
Here, we precisely mapped MIR epitopes recognized in MG or EAMG by making chimeras in which sequences of human muscle α1 subunits replaced parts of the human neuronal α7 AChR or Aplysia ACh binding protein (AChBP). Two sequences, which were adjacent only in the native α1 conformation, formed the MIR, thereby explaining conformation dependence of the MIR. The great influence of the α1 MIR sequences on AChR expression and sensitivity to activation implies important roles of the MIR in conformation changes associated with subunit conformational maturation and assembly as well as AChR activation.
Materials and Methods
α7 chimeras.
Human α7 cDNA was subcloned into the BglII site of the PMXT vector as previously described (Peng et al., 1994). The sequences of α7 2–14 and 66–76 were substituted with homologous human α1 sequences by multiple-step PCR using appropriate pairs of forward and reverse synthetic oligonucleotide primers (Invitrogen). The extra α1 sequences 60–65 and 77–81 substituted for corresponding sequences of the above chimera using a similar approach to express an extended MIR loop.
To incorporate the α1 sequence 1–32, we engineered a BamH I site and a BstE II site at each end of the target sequence in α7 cDNA, or the chimera with the extended MIR loop. Using a similar approach, we introduced a BamH I site between the sequences coding for signal peptide and N terminus of α1 subunit in α1 cDNA by PCR. The mutated α1 and α7 cDNA clones were digested with BamH I and BstE II, and a purified 212-bp fragment from α1 cDNA was ligated together with the remaining fragments of α7 into the PMXT vector.
Chimera cDNAs were checked for accuracy by DNA sequencing before cRNA preparation. cRNAs were synthesized in vitro using the SP6 mMessage mMachine kit (Ambion). Schematic diagrams of the wild-type subunits and chimeras are shown in Figure 1B.
AChBP chimeras.
A KpnI site was engineered at position Phe 35 of a cDNA encoding the Aplysia acetylcholine binding protein (A-AChBP) in the FLAG-CMV-3 expression vector (Hansen et al., 2002). A 123-bp PCR product coding for the N-terminal sequence α1 1–30 (Val 31 and Thr 32 are conserved between α1 subunit and A-AChBP) was produced using the full-length α1 cDNA clone in TE1.1 as a template. The upstream and downstream sequences were constructed to contain HindIII and KpnI restriction sites respectively. Both the purified PCR product and A-AChBP cDNA were cut with the restriction enzymes HindIII and KpnI. A fragment coding for α1 1–30 was ligated together with the remaining fragments of A-AChBP. Then the KpnI site was removed by PCR.
To graft the MIR loop α1 60–81 into the above chimera, another KpnI site was introduced at position Lys 61 by two-step PCR. An 87-bp PCR product encoding α1 60–81, which was engineered to have a KpnI site at 5′-end and a EcoR V site at 3′-end, was produced from the full-length α1 cDNA. The purified PCR product was cut with the restriction enzymes KpnI and EcoR V for ligation with the remaining fragments of chimeric α1(1–30)/AChBP, which was cut with the same pair of enzymes. Then the second KpnI site was removed by PCR.
mAbs.
Monoclonal anti-FLAG M2 antibody was purchased from Sigma-Aldrich. All subunit-specific mAbs used here have been characterized and described previously (Lindstrom, 1996). Properties of mAbs to the MIR used are summarized in Table 1. mAb 306 and mAb 319 are directed at the cytoplasmic domain of α7 (McLane et al., 1992).
Properties of mAbs to the MIR used in this study
MG patient sera.
Serum from human MG patients was provided by archived samples from the Lindstrom lab and the de Baets lab. Sera from canine and feline MG patients were provided from the Shelton lab and, chosen to be ≥0.6 nm titer for canines and ≥0.3 nm titer for felines.
Xenopus oocyte expression.
Oocytes were prepared for microinjection as described by Colman (1984), and injected with 50 ng of cRNA of each of wild-type or chimeric α7 subunits unless otherwise specified. They were incubated for 3–4 d after injection in a modified L-15 medium containing 50% Leibovitz L-15 (Invitrogen), 10 mm HEPES, pH 7.5, 10 unit/ml penicillin, and 10 mg/ml streptomycin at 18°C. Surface expression was determined by incubating oocytes in L-15 medium containing 1% BSA and 4 nm 125I αBgt for 1 h at room temperature followed by washing steps with L-15 medium. Nonspecific binding was determined by incubating noninjected oocytes under the same conditions.
Radioimmunoassay.
On day 5 after injection, groups of oocytes were homogenized in buffer A (50 mm Na2HPO4-NaH2PO4, 50 mm NaCl, 5 mm EDTA, 5 mm EGTA, 5 mm benzamidine, 15 mm iodoacetamide, 2 mm phenylmethylsulfonyl fluoride, pH7.5). Cell membranes were pelleted by centrifugation at 15,000 × g for 20 min, followed by solubilizing in buffer A containing 2% Triton X-100 for 1 h at room temperature. After cellular debris was removed, aliquots of oocyte extract were incubated with appropriate amounts of a 5 mg/ml stock of mAbs with 5 μl of normal serum and 10 nm 125I αBgt overnight at 4°C. AChR-antibody complexes were precipitated using 40 μl of a standardized stock of goat anti-human IgG (for human IgG) or 100 μl of a standardized stock of sheep anti-rat IgG (for rat mAbs) for 2 h at room temperature. This precipitate was pelleted by centrifugation and washed two times with 1 ml of 0.5% Triton X-100 in PBS (10 mm sodium phosphate buffer, 100 mm NaCl, pH 7.5). The pellets were assayed in a γ counter.
Electrophysiological recordings.
Currents in oocytes were measured using a standard two-microelectrode voltage-clamp amplifier setup (oocyte clamp OC-725, Warner Instrument) as previously described (Gerzanich et al., 1998). Dose–response curves were derived by determining the maximum ACh response obtainable on each oocyte, then normalizing these responses as a fraction of the maximum response. Normalized responses from several oocytes were analyzed using the Hill equation (R = 1/(1 + 10(LogEC50− Log[ACh]) × Hill slope)) and the curve-fitting program Kaleidagraph (Synergy) to determine the EC50 values reported.
Western blots.
The groups of oocytes were homogenized in buffer A containing 10 μg/ml DNase followed by three freezing/melting cycles. Cell membranes were pelleted by centrifugation and resuspended in 1 ml of buffer A containing 10 μg/ml DNase, followed by an 1 h incubation at room temperature. Membrane fractions were collected by centrifugation and solubilized in buffer A containing 2% Triton X-100 for 1 h at room temperature. After cellular debris was removed, aliquots of oocyte extract (corresponding to one oocyte) were separated on pre-cast NuPAGE 10% Bis-Tris Gels (Invitrogen). The transfers were conducted in a SemiPhor semi-dry electroblotting chamber (Hoefer) to Trans-Blot Medium PVDF membrane (Bio-Rad). The blots were quenched using 5% dried milk after transfer for 80 min. The subunits were detected by a 1:1 mixture of mAbs 306 and 319.
Expression and purification of chimeric AChBP.
HEK293S cells lacking the N-acetylglucosaminyltransferase I (GnTI−) gene (Reeves et al., 2002) were maintained in DMEM (high glucose) (Invitrogen) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific), 100 units/ml penicillin, 100 μg/ml streptomycin, and 2 mm l-glutamine (all from Invitrogen) in a CO2 (10%) incubator. Wild-type AChBP or chimeric α1(1–30, 60–81)/AChBP with a N-terminal FLAG tag were transfected into HEK293S(GnTI−) cells using the FuGene6 transfection agent (Roche Diagnostics), followed by selection with G418 to yield stable cell lines secreting AChBP. Both wild-type AChBP and chimera were expressed as a soluble exported protein. Culture media containing AChBP or chimera were collected and applied onto a FLAG antibody column. Elution with the 3×FLAG peptide yielded purified proteins (Hansen et al., 2004).
Immobilization of chimeric AChBP on Activated CH Sepharose 4B.
α1(1–30, 60–81)/AChBP chimera was immobilized on Activated CH Sepharose 4B at a protein/resin ratio of 2 mg/ml according to the protocol recommended by the manufacturer (GE Healthcare). The prepared resin was stored in PBS, pH 7.4, containing 0.05% sodium azide. In parallel, an equal amount of bovine serum albumin (BSA) was immobilized on Activated CH Sepharose 4B under the same conditions as a control.
Immunoadsorption of anti-MIR autoantibodies.
Twenty-five microliters of the α1(1–30, 60–81)/AChBP chimera beads or BSA beads were incubated overnight at 4°C with 70 μl of diluted MG sera in PBS, 0.2% BSA containing 35 fmol of anti-AChR antibodies in a compact reaction column (USB), followed by centrifugation and a wash with 70 μl of PBS, 0.2% BSA. Combined supernatants were assayed for unbound anti-AChR antibodies using a radioimmunoassay described as above with a substitution of TE671 AChR (transfected with ε subunit) for chimeric AChR. The percentage of immunoadsorption was calculated as follows: % immunoadsorption = [1 − (anti-AChR titers after adsorbed with chimera beads)/(anti-AChR titers after adsorbed with BSA beads) × 100.
mAb 35 competition assay.
Anti-AChR titers of MG sera were determined using a radioimmunoassay as above in the presence or absence of 50 nm of mAb 35. The immune complex of 125I-αBgt labeled AChR and MG autoantibody was precipitated using goat anti-human IgG depleted with normal rat serum. The percentage of inhibition was calculated as follows: % inhibition = [1 − (anti-AChR titers in the presence of mAb 35)/(anti-AChR titers in the absence of mAb 35) × 100.
Statistics.
Student's two tailed t test was used to determine the significance of differences between group means. All data represent the mean ± SD.
Results
Mapping the antigenic structure of the MIR
Sequences of the extracellular domain of human AChR α7 subunits were replaced by homologous human α1 sequences to determine which α1 sequences would confer antigenicity of an α1 MIR to an α1/α7 chimera (Fig. 1). Properties of the rat and human mAbs to the MIR used in his study are summarized in Table 1. Rat mAb 210 to the MIR binds both native and denatured α1 (Lindstrom, 2000). Incorporating only the putative MIR loop α1 66–76 into α7 allowed binding of mAb 210 (Table 2), but at only 1/500th of the affinity with which it bound native AChR. mAb 198 had similarly low affinity for this chimera compared to native AChR. The α1(66–76)/α7 chimera was not bound by the absolutely conformation-dependent mAb 35. Increasing the length of the α1 insert to 60–81 increased the binding affinities of mAb 210 and mAb 198 by 227- and 79-fold, respectively. The α1(60–81)/α7 chimera was bound by mAb 35, but with only 1/1000th the affinity for native AChR (Table 2).
Structures of MIR chimeras. A, Putative MIR components are highlighted on the crystal structure of an Aplysia AChBP subunit (Hansen et al., 2005). A top-view space-filling model is shown at the top. A front-view ribbon diagram, rotated to permit viewing of the MIR from its underside as a prominent appendage, is shown at the bottom. Red highlights the α helical ribbon corresponding to α1 1–14. The thin segment proceeding the α helix represents a FLAG tag on the AChBP constructs. Yellow highlights sequence corresponding to α1 15–32. Green highlights sequence corresponding to α1 60–81, including the MIR loop α1 66–76. Dark blue designates the remainder of the α1 subunit. A circle with a radius of 15.5Å and center at the MIR indicates the area typically buried upon antibody binding to a protein antigen (Mariuzza et al., 1987; Konstantakaki et al., 2007). The prominent loop on the right is the C loop which closes over the ACh binding site when agonists are bound. B, The crystal structure of a Fab of mAb 192 (Kontou et al., 2000) is accompanied by the mouse α1 extracellular domain (Dellisanti et al., 2007). This allows comparison between the similar structures of the α1 extracellular domain and AChBP subunit and it allows comparison between the size of an Fab and the MIR. This makes it evident that while very small differences in the sequence and conformation of epitopes can profoundly influence the affinity with which antibodies are bound, the large size of bivalent IgG molecules can result in competitive binding between different closely spaced epitopes within the MIR. Parts of the MIR are highlighted in the same colors as in the AChBP subunit. The light chain of the Fab is gray and the heavy chain is black. Six hypervariable loops, which form the antigen binding sites, are highlighted in cyan. This mAb to the MIR does not bind to the MIR loop but competes for binding with mAbs which do. The Fab is angled to suggest this, but not actually docked on the subunit. Future studies in which the crystal structures of MIR chimeras with Fabs bound are determined should reveal the contact amino acids involved in binding of mAbs to the MIR. C, A schematic diagram of the α1/α7 chimeras used to map MIR epitopes is shown. The extracellular domain of each chimeric subunit is displayed with the α1 sequences indicated by black boxes. D, Alignment of extracellular domain sequences of human and rat muscle AChR α1 subunits. Rodent α1 differs significantly at the sequence α1 23–30 from human α1 (in red), making this sequence a likely candidate for contributing to MIR epitopes which differ between rats and humans, such as the epitope for mAb 192.
Dissociation constants (KD) of mAbs to the MIR for α1/α7 and α1/AChBP chimeras
Examination of the structure of Aplysia AChBP revealed that the N-terminal α helical sequence 1–14 paralleled the putative MIR loop 66–76 (Fig. 1A). α1/α7 chimeras containing both the α1 N-terminal α helix and the MIR loop were efficiently bound by mAbs with sub-nanomolar affinities (Table 2). The contribution to the MIR of the two segments of the extracellular domain of the α1 subunit, which are adjacent only in the native conformation, explains the conformation dependence of the binding of these mAbs to the MIR. The presence of α1 2–14 along with α1 60–81 increased the affinity with which mAb 210 was bound 2600-fold. Then the binding affinity of mAb 210 for the chimera exceeded that for muscle AChR by fourfold. This may be because the chimera is a homopentamer, which may permit both mAb binding sites to bind within a single pentamer, whereas in muscle type AChRs the two α1 MIRs are oriented so that they cannot be crosslinked by a single mAb (Conti-Tronconi et al., 1981; Beroukhim and Unwin, 1995). The affinity of mAb 198 for this chimera was also greater than for native AChR. The presence of α1 2–14 had the greatest effect on the conformation-dependent mAb35, increasing the affinity with which it was bound 91,000-fold, also to greater than for native AChR. However, the above chimeras bound to neither rat mAb 192 nor human mAb 637. This suggests that some human autoantibodies to the MIR bind to epitopes distinct from those which are recognized by most mAbs to the MIR from rats with EAMG, but close enough to compete for binding to human muscle AChRs (Tzartos and Lindstrom, 1980). The ring in Figure 1A showing the area on a protein typically obscured by a bound antibody illustrates how several adjacent epitopes could be obscured by a single bound mAb. Comparison of the size of an Fab fragment of an IgG molecule to the size of the α1 extracellular domain further emphasizes this point (Fig. 1B).
mAb 192 depends absolutely for its binding on the native conformation of α1, and binds to rat AChRs with 10,000-fold lower affinity than to human AChRs (Tzartos et al., 1998). It seems likely that its epitope includes sequences within α1 23–30, because these are the only nonconservative sequence differences between human and rat anywhere near the MIR (Fig. 1C). An α7 chimera containing both the α1 1–32 and the longer 60–81 version of the MIR loop was bound by the human mAb 637 with sub-nanomolar affinity, although it was not recognized by mAb 192 (Table 2). These results emphasize the importance of both AChR sequence and conformation to binding affinity of antibodies.
α1 MIR increases expression of α1/α7 chimeras
Rather than reducing expression of α1/α7 chimeras, as one might expect from disruption due to a combination of muscle and neuronal AChR sequences, the chimeras exhibited >10-fold, on the average, increased expression on the oocyte surface compared to wild-type α7 (Fig. 2A). The α1(2–14, 60–81)/α7 chimera showed a 53-fold increase over wild-type α7 on the oocyte surface. This suggests that interactions between the MIR loop and the N-terminal α helix, which produce epitopes recognized with highest affinity by mAbs to the MIR, may also provoke conformational maturation of the α1/α7 subunits. In contrast, incorporating only the N-terminal sequence α1 2–14 or 1–32 into α7 completely prevented expression of mature AChR on the oocyte surface (Fig. 2A) or within the oocyte (data not shown). Thus, interactions between the N-terminal sequence and the MIR loop must occur when both are from α1, which make the α1 N terminal sequences tolerable within the chimera. However, incompatible interactions between N-terminal α1 sequences and the α7 MIR loop may prevent conformational maturation, thereby preventing assembly and the formation of binding sites for 125I αBgt. Sequences within α1 15–32 must be poorly compatible with parts of α7 and lead to reduced expression (Fig. 2A).
α1/α7 chimeras were more efficiently expressed than wild-type α7 AChRs. A, Expression on the oocyte surface detected by binding of 125I αBgt to intact oocytes is shown. Presence of the α1 MIR increased assembly of mature α1/α7 AChRs as measured by the total amount of chimeric AChRs on the cell surface. Only mature pentameric AChRs are expected to be transported to the cell surface. All pairs of values shown are significantly different from each other (p < 0.05) except α1(66–76)/α7 and α1(1–32, 60–81)/α7. B, Similar amounts of wild-type α7 and α1/α7 chimera proteins were synthesized as measured by a Western blot (top). However, the α1/α7 chimera which forms the most antigenic MIR structure was assembled much more efficiently into mature AChRs expressed on the oocyte surface (bottom, p < 0.05 for all pairs). A chimera containing only α1 1–32 was not expressed on the cell surface at all. C, The fractions of wild-type α7 and α1(2–14, 60–81)/α7 which could be adsorbed by αBgt-coupled beads were compared on a Western blot using mAb 319 followed by 125I labeled goat anti-rat IgG. Radioactive bands were cut out and quantitated using a γ counter. Only mature pentameric AChRs are capable of binding αBgt. With wild-type α7, 6.6% of subunits were assembled into mature AChRs that were adsorbed to αBgt-coupled beads. Of α1(2–14, 60–81)/α7 subunits, 71% formed mature AChRs. The α1(2–14, 60–81)/α7 chimera always had more protein on Western blots than wild-type α7. One possible explanation is that, as with α1 (Merlie and Lindstrom, 1983), unassembled subunits turn over much more rapidly than mature AChRs. Thus the remarkable assembly efficiency of α1(2–14, 60–81)/α7 (Fig. 2A) results in accumulation of much more total α7 subunits.
Similar amounts of wild-type and chimeric subunits were detected on a Western blot when equal amounts of cRNAs were injected in Xenopus oocytes (Fig. 2B, top). Both mAbs 306 and 319 used for the blot are directed at the large cytoplasmic domain of α7 subunit (McLane et al., 1992), thus their binding to α7 should not be affected by mutations of the extracellular domain. It is striking that the same large amount of denatured α7 protein is found in oocytes expressing wild-type α7, α1(1–32)/α7 chimeras which assemble no 125I αBgt binding sites of mature AChRs, and α1(1–32, 60–81)/α7 chimeras which assemble 8.3-fold more mature AChRs than wild type. This shows that the great increase in mature AChRs resulting from expression of α1/α7 chimeras is caused by increased assembly rather than by increased synthesis of subunits. Even with wild-type α7, assembly is very inefficient (6.6%). Adsorption with αBgt beads reveals that 71% of α1(2–14, 60–81)/α7 chimeras were assembled into pentameric AChRs (Fig. 2C). Conformational maturation of subunits before assembly is probably what limits formation of mature α7 AChRs, as is the case for muscle AChRs (Merlie and Lindstrom, 1983). The MIR, and its interactions with the N-terminal sequences which produces high affinity MIR epitopes, may also be responsible for nucleating conformational maturation of α7 subunits that permits greatly increased assembly of mature AChRs.
Functional chimeric α7 AChRs expressed in oocytes
α1/α7 chimeras retain their ability to function as AChRs, as shown in Figure 3. Presence of the MIR sequences greatly influences the sensitivity to activation of chimeric AChRs by acetylcholine (ACh). The α1(1–32, 60–81)/α7 chimera was tenfold more sensitive than wild-type α7 to activation by ACh. On the other hand, the α1(2–14, 60–81)/α7 chimera was 13-fold less sensitive than wild-type α7. These marked changes in sensitivity to activation by ACh were not accompanied by significant change in agonist-induced current per AChR (Fig. 3B). This suggests that the differences between chimeras resulted from differences in probability of channel opening, rather than duration of channel opening due to altered rates of desensitization or altered channel conductance.
α1/α7 chimeras greatly influence the sensitivity to activation of chimeric AChRs by ACh. A, Dose/response curves for ACh show that these chimeras exhibited significantly different sensitivities to activation by ACh. Normalized currents are plotted against agonist concentrations. Averaged data (each point from three to five experiments) were fitted by the Hill equation. The single-fit errors were shown. B, The α1 sequences in the chimeras did not change the maximum current per AChR at saturating ACh concentrations of chimeric AChRs. Chimeras were expressed in Xenopus oocytes and function was studied with whole-cell recordings using a standard two-microelectrode voltage clamp. The total amount of chimeric AChRs on the cell surface was detected by binding of 125I αBgt to intact oocytes. C, Time course of currents induced by EC50 concentrations of ACh. Currents were normalized to the peak currents for each chimera.
Response kinetics are similarly rapid for α7 and the much more sensitive α1(1–32, 60–81)/α7 chimera (Fig. 3C). This shows that the greater sensitivity does not result from greatly decreased desensitization but instead, most likely, from a greater probability of being opened when liganded. The kinetics of both activation and desensitization are slower for the insensitive α1(2–14, 60–81)/α7 chimera, even though its EC50 is 13-fold greater than that for α7 and 132-fold greater than that for the sensitive α1(1–32, 60–81)/α7 chimera (Fig. 3A,C). Binding affinity for αBgt to the α1(2–14, 60–81)/α7 chimera (2.79 ± 0.34 nm) was not significantly different from that to wild-type α7 (2.18 ± 0.52 nm). These observations are consistent with the hypothesis that the low sensitivity of α1(2–14, 60–81)/α7 results from a low probability of opening when bound by agonist at one or two sites, and an increasing probability of opening when liganded at 3 sites. The rapid desensitization kinetics of α7 AChRs results in underestimation of their sensitivity to activation (Papke and Thinschmidt, 1998), and inhibition of their desensitization moves their dose/response curves to the left (Hurst et al., 2005). However, that is not what is seen here. Activation of α7 AChRs (or others) by binding of one agonist is very inefficient, binding of two agonists is most efficient at activation, and binding at three or more sites enhances the rate of desensitization thereby inhibiting the response (Papke and Thinschmidt, 1998).
mAb 210 did not block muscle AChR responses (data not shown). In contrast, binding of mAb 210 to the α1(2–14, 60–81)/α7 chimera reduced the response to an EC50 concentration of ACh by 70%. In this experiment, oocytes were incubated with 250 μg/ml of mAb for 1 h before retesting. As a control, similar exposure to mAb 306, which is directed at the cytoplasmic domain of α7, produced no effect. In muscle AChR, the two α1 subunits are not adjacent and the orientation of the MIRs are away from the central axis of the AChR (Beroukhim and Unwin, 1995). This explains why a single mAb to the MIR cannot crosslink the two MIR epitopes in a single AChR (Conti-Tronconi et al., 1981). However, the presence of five MIR epitopes in the chimera rather than two in a muscle AChR might allow the mAb to the MIR to crosslink adjacent subunits, perhaps thereby impairing function. The higher affinity with which mAbs can bind most chimeras compared to native α1 AChRs (Table 2) might also be explained by the ability of both mAb binding sites to bind to homopentameric chimeras, rather than only 1 site to bind muscle type AChRs (Conti-Tronconi et al., 1981).
Expression and characterization of AChBP chimeras
α1(1–30, 60–81)/AChBP, like wild-type AChBP, is water soluble, sediments on sucrose gradients at 4.9 S, and binds αBgt (data not shown). 125I labeled chimera was used in immune precipitation studies.
As expected, wild-type AChBP was not bound by any rat mAbs to the MIR (data not shown). The α1(1–30, 60–81)/AChBP chimera showed high affinity binding to all rat MIR-reactive mAbs tested as well as the human mAb 637, as shown in Table 2. In contrast to the α1(1–32, 60–81)/α7 chimera, which contained the same α1 sequences, this AChBP chimera was bound by mAb 192 with subnanomolar affinity (Table 2). This suggests that the rest of AChBP caused conformational changes in the mAb 192 epitope that allowed the binding of mAb 192. This might involve different interactions between the α1 15–30 sequence and adjacent AChBP sequences. The small differences between affinities of rat mAbs to α1/α7 and α1/AChBP chimeras may also result from subtle conformation differences between α1 sequences in α1/α7 chimeras and α1/AChBP chimeras.
The only other region which is near the classic MIR loop on the structure of AChBP is the loop between segments β5 and β6, which corresponds to α1 107–115 (Fig. 1A). Including the sequence α1 107–115 did not improve the binding of mAbs to the MIR (Table 2). Thus the β5-β6 loop does not appear to contribute to the antigenic structure of the MIR.
Binding of either nicotine (1 mm) or αBgt (2.5 μm) to the α1/AChBP chimeras decreased binding affinity of mAb 637 slightly (KD = 0.075 nm with no cholinergic ligand, 0.18 nm with nicotine and 0.22 nm with αBgt) (Fig. 4). Binding affinity of mAb 192 was also reduced slightly (KD = 0.14 nm with no cholinergic ligand, 0.23 nm with nicotine and 0.29 nm with αBgt). Thus, the differences between resting, activated and desensitized conformations do not appear to greatly alter the epitopes within the MIR.
Binding affinity of mAb 637 to the α1(1–30, 60–81)/AChBP chimera is slightly decreased in the presence of saturating nicotine or αBgt. The binding of various concentrations of mAbs to 125I labeled α1/AChBP chimera was determined by immunoprecipitation in the absence (circles) and presence of either 1 mm nicotine (square) or 2.5 μm αBgt (triangle). The averages of duplicate determinations were fitted by the Hill equation to determine KD values (see Table 2).
Expression of α7(66–76)/α1 AChR chimeras
Since neither α1/α7 chimeras nor α1/AChBP chimeras incorporating only the N-terminal sequence α1 2–14 or 1–32 were able to form mature pentamers (Fig. 2A), it was still unknown whether the N-terminal sequence alone contains the epitopes recognized by mAb 637 or 192 or whether parts of α1 60–81 were also required to form the epitopes. To answer this question, a chimera in which the MIR loop formed by α1 66–76 was replaced in α1 by corresponding parts of the human neuronal α7 AChR was expressed in oocytes. The α7(66–76)/α1 chimeric subunit in combination with β1, δ, and γ or ε subunits was able to form functional AChRs on the oocyte surface in either adult (ε) or fetal (γ) forms (Fig. 5A). The EC50 for ACh of the wild type (71 ± 7 μm) and chimeric type (60 ± 6 μm) were similar in the adult form. In the fetal form, wild type (27 ± 3 μm) had somewhat greater sensitivity to ACh than the chimeric type (65 ± 8 μm). The fetal AChR with mutant α1 desensitized more rapidly than fetal AChR with native α1, suggesting that α1 MIR, interacting with γ, influenced conformation changes associated with desensitization. The α7(66–76)/α1 AChRs solubilized in Triton X-100 could only be immunoprecipitated by mAb 192, but not by any other classic rat MIR mAbs (like mAbs 198 and 210) or human mAb 637 (Fig. 5B). This confirmed that most mAbs to the MIR absolutely require the α1 66–76 MIR loop for binding (Saedi et al., 1990), but mAb 192 does not.
Chimeric α7(66–76)/α1 AChRs lacking the α1 MIR loop were not immunoprecipitable by most mAbs to the MIR or by many MG patient autoantibodies. A, Chimeric α7(66–76)/α1 subunits formed functional AChRs when coexpressed with β1, δ, and either γ or ε subunits. Chimeric α7(66–76)/α1 subunits expressed with γ subunits to produce a fetal type AChR desensitized more rapidly than wild-type fetal AChR. B, 125I αBgt labeled (α7(66–76)/α1)2β1δγ AChRs were immunoprecipitated from Triton X-100 extracts by mAbs 61 and 192, but not by mAbs 198, 210 or 637 (*p < 0.05). mAb 61 is directed at the large cytoplasmic domain of the α1 subunit (Ratnam et al., 1986). C, The chimeric (α7(66–76)/α1)2β1δγ AChR prevented binding of an average of 57% of tested MG sera when compared to wild-type (α1)2β1δγ. It is consistent with inhibition of binding of these sera to AChR by mAb 35 (average 55%). The sera are numbered as in Figure 7B.
The MIR loop was required for binding of 57% of MG autoantibodies, which corresponds closely to the 55% of serum antibodies inhibited from binding by mAb 35 in the MG sera tested (Fig. 5C). Replacing the α1 MIR did not prevent binding of mAb 192, although it competes with mAb 35 (Tzartos et al., 1998). The sequence 1–32 may contain part or all of the epitope for mAb 192 because this sequence contains rat-specific sequences which may account for the low affinity of mAb 192 for rat α1 AChRs. mAb 192 is an absolutely conformation-dependent antibody, which bound α1(1–30, 60–81)/AChBP, but not α1(1–32, 60–81)/α7. Thus, the fact that the α7(66–76)/α1 AChRs bound mAb 192 suggests that the α7 MIR loop did not significantly alter the conformation of other parts of α1.
Changes in MIR conformation during assembly of α1 AChRs
Assembly of human α1 subunits into mature AChRs causes conformation changes which increase the binding affinity of mAbs to the MIR (Merlie and Lindstrom, 1983; Conroy et al., 1990). The human rhabdomyosarcoma cell line TE671 expresses fetal muscle (α1γ)(α1δ)β1 AChRs and a nearly equal amount of monomeric α1 subunits (Conroy et al., 1990). Both mature AChRs and α1 bind 125I αBgt, although α1 has fivefold lower affinity. Only mature AChRs bind small cholinergic ligands, because the ACh binding sites are formed at the interfaces between α1 and δ, γ or ε subunits. Thus, in the presence of a high concentration of nicotine, 125I αBgt is not bound to mature AChRs in TE671 extracts, but only to free α1. Using this approach to distinguish unassembled α1 and mature AChRs, it was shown that the conformation-dependent rat mAb 35 reacted 20-fold better with mature AChRs than with unassembled α1, while the less conformation-dependent mAb 210 bound mature AChRs only fivefold better (Conroy et al., 1990). A selection of 45 MG patient sera reacted on average 14-fold better with mature AChRs than unassembled α1.
Here we took a similar approach to assaying the binding of the highly conformation-dependent human mAb 637, and found that it had 1 × 104 greater reaction with TE671 AChRs than with unassembled α1 (Fig. 6). Thus, the epitope of the MG mAb within the MIR is remarkably more dependent on the native conformation than are either of the rat mAbs 35 or 210 to the MIR. High affinity binding of mAb 637 to α1/α7 and α1/AChBP chimeras is strong evidence for how close the conformations of the α1 MIR in these chimeras approaches that of human muscle AChRs.
mAb 637 bound to the MIR on unassembled α1 subunits with low affinity compared with native AChR. The binding of different concentrations of mAb to 125I αBgt labeled Triton X-100 extracts of TE671 cells was determined by immunoprecipitation assay in the presence (open circles) and absence (closed circles) of 10 mm nicotine. The binding curve in the absence of nicotine showed a high and a low affinity component. The low affinity phase did not reach a plateau within the tested concentration range. The monotonic calculated high affinity binding curve, determined by subtracting the low affinity binding in the presence of nicotine, is shown as a dotted line (× symbols). KD1 for high affinity to the MIR in mature AChRs was determined from the monotonic calculated binding curve by fitting with the Hill equation as described in Table 2. KD2 for low affinity binding to unassembled α1 was determined from the binding curve in the presence of nicotine by fitting with a two-component Hill equation (r = (1 − A)/(1 + 10(LogKD1 − Log[ACh]) × Hill slope1) + A/(1 + 10(LogKD2 − Log[ACh]) × Hill slope2), where A is the proportion of the low affinity component). The binding curve in the absence of nicotine was also fit with this two-component Hill equation. Nicotine inhibits binding of 125I αBgt to mature AChRs but not to the unassembled α1 subunits which are present in these extracts. Thus, in the presence of nicotine binding to only unassembled α1 is assayed.
The α1 MIR expressed in chimeric AChBP was recognized by MG patient sera
Sixty-one randomly chosen MG patient sera archived for several decades were tested for their ability to bind 125I α1(1–30, 60–81)/AChBP. Crossreaction with parts of the AChBP sequences other than the MIR was determined using 125I labeled wild-type AChBP and was subtracted from the total binding to the chimeras. Of these, 59% bound with titers ranging from 0.012 to 19 nm, averaging 1.7 nm.
To examine whether the β5-β6 loop sequence α1 107–115 contributes to the epitope recognized by MG autoantibodies, the sixty-one MG patient sera were tested for their ability to bind 125I labeled α1/AChBP chimeras. α1(1–30, 60–81, 107–115)/AChBP was not bound by MG sera significantly better than was α1(1–30, 60–81)/AChBP (Fig. 7A). Thus, the β5-β6 loop does not contribute to MIR epitopes recognized by MG patient sera, just as it dose not contribute to binding of mAbs to the MIR (Table 2).
The MIR expressed in chimeric AChBP was recognized by MG patient sera. A, Titers of sera from 61 MG patients were determined by immunoprecipitation of 125I α1(1–30, 60–81)/AChBP or 125I α1(1–30, 60–81, 107–115)/AChBP. A linear regression line was fitted to the data with the following parameters: r = 0.90 and slope = 0.49. These results indicated that both chimeras were recognized equally, implying that the sequence α1 107–115 did not contribute significantly to autoantibody binding. B, The proportion of MG autoantibodies specific to the MIR was determined in sera from 13 other MG patients by two methods. The fraction of autoantibodies to human muscle AChR that could bind to the α1(1–30, 60–81)/AChBP chimera was determined by pre-adsorbing the antisera with the α1(1–30, 60–81)/AChBP chimera coupled to agarose. The fraction of autoantibodies that could be inhibited from binding to AChR by the presence of excess mAb 35 was also determined. Binding of mAb 35 inhibited the binding of a greater proportion of the autoantibodies than could bind the α1 epitopes expressed in the chimera.
A collection of recently collected MG sera was used to assay both the fraction of autoantibodies which could bind to the MIR in the α1(1–30, 60–81)/AChBP chimera and the fraction which could be inhibited from binding to human α1 AChR by mAb 35 to the MIR (Fig. 7B). The chimera adsorbed an average of 15% of the autoantibodies. mAb 35 inhibited the binding of an average of 63% of the autoantibodies. These results are consistent with the concepts that: (1) the chimera contains several closely spaced conformation-dependent epitopes which are bound with high affinity by some mAbs and autoantibodies to the MIR, (2) muscle AChR contains more MIR epitopes, the majority of which include the MIR loop, and (3) binding of a mAb to the MIR can inhibit the binding of autoantibodies to many overlapping and closely spaced epitopes within the MIR.
The MIR in the α1/AChBP chimera was recognized by a majority of feline and canine MG sera
Canine and feline MG are good naturally occurring models for human MG (Shelton and Lindstrom, 2001). mAb 35 inhibited binding of an average of 68% of the autoantibodies in canine MG (Shelton et al., 1988). Since mAbs to the MIR from rats with EAMG recognize overlapping but distinct epitopes other than those that are recognized by MG patients, we asked whether canine or feline MG autoantibodies bind to the same epitopes recognized by MG patients. To answer this question, 21 canine and 39 feline MG sera were randomly chosen and tested for crossreactions with 125I labeled α1(1–30, 60–81)/AChBP. The results are shown in Figure 8. The chimera was recognized by 28% of canine MG autoantibodies and 24% of feline MG autoantibodies. Fifty-seven percent of canine MG sera, and 80% of feline MG sera, had >5% crossreactivity with α1(1–30, 60–81)/AChBP chimera.
The MIR in the α1 1–30, 60–81/AChBP chimera was recognized by a majority of feline and canine MG sera. A, C, The apparent titers of 21 canine and 39 feline MG sera against 125I labeled α1(1–30, 60–81)/AChBP were determined by immunoprecipitation assay and reported as nanomoles of 125I labeled α1(1–30, 60–81)/AChBP bound per liter of serum. B, D, Proportion of anti-MIR autoantibodies was expressed as the percentage of each serum titer against 125I labeled α1(1–30, 60–81)/AChBP relative to the corresponding titer against muscle AChR from that species. From left to right, MG sera were arranged in ascending order of titers against muscle AChR.
Discussion
The MIR is important because half or more of the autoantibodies to AChRs in EAMG and MG in several species are directed at this region, and because these antibodies contribute to pathology at the neuromuscular junction (Tzartos et al., 1982, 1983; Shelton et al., 1988). Our data showed that two α1 sequences (the MIR loop and the N-terminal α helical region) must interact to form MIR epitopes recognized by mAbs from rats with EAMG or humans with MG. These results are consistent with studies of the crystal structure of mouse AChR α1 subunit extracellular domain, in which the MIR loop is in close association with the N-terminal α helix and the β5-β6 turn loop (Dellisanti et al., 2007). Our results showed that the N-terminal α helix was critical to the antigenic structure of the MIR, but the β5-β6 turn loop (i.e., α1 107–115) was not. Our results further showed that the conformation of the MIR changes when α1 subunits are assembled into mature human AChRs, conferring higher affinities for binding of mAbs to the MIR. Thus, the conformation of crystallized unassembled α1 subunits (Dellisanti et al., 2007) must differ somewhat from α1 in mature AChR pentamers. However, details of the structure of unassembled α1 subunits are consistent with our observations. Amino acids 68 and 71, which are critical to the antigenic structure of the MIR loop (Saedi et al., 1990), pack closely with amino acids of the N-terminal α helix (Dellisanti et al., 2007).
The α1 MIR loop is critical for binding of most antibodies to the MIR. Replacing the MIR loop in α1 subunits assembled in α1 AChRs with the corresponding α7 loop inhibited the binding of MG autoantibodies in the same large proportion that they are inhibited by binding of mAb 35. Although the α1(1–30, 60–81)/AChBP chimera was recognized by 59% of human MG sera, the proportion of autoantibodies bound to this chimera was still lower than the proportion of autoantibody binding blocked by mAb 35.
MG sera and mAbs bind to unassembled α1 with much lower affinity than to α1 assembled in mature AChRs (Conroy et al., 1990). This could reflect a conformation change in α1 on assembly. Alternatively, or in addition, contributions to the MIR from adjacent subunits could explain why α1/ACHBP MIR chimeras adsorb a much smaller fraction of MG patient autoantibodies than are prevented from binding by mAb 35. Construction of more complex AChBP chimeras containing both α1 and other subunit sequences will be required to test this possibility.
α1 subunits differ greatly from both α7 and AChBP in the sequences 1–14 and 15–32 (Fig. 1C). Thus, it is not surprising that α1 1–14 or 15–32 are incompatible with forming mature AChRs alone in α7 chimeras. The interactions between α1 1–14 and 66–76 permit assembly of this domain with α7, while conferring a particular conformation on the domain with high affinity for conformation-dependent mAbs. The stability of this domain helps to overcome the incompatibility of 15–32 with α7, but leaves the AChR in a conformation which is more easily activated by agonists. The 15–32 sequence apparently assumes a different conformation in association with adjacent AChBP sequences than α7 sequences, because the α1(1–30, 60–81)/AChBP permits binding of mAb 192 and α1(1–32, 60–81)/α7 does not.
MG in dogs and cats provide a good animal models for human MG (Shelton et al., 1988, 2000). A majority of feline (80%), and canine (57%) MG sera bind to a chimera forming MIR epitopes recognized by human MG patients.
The rat mAb 210 to the MIR recognizes human α3 (Wang et al., 1998) but not rodent α3 (Feng et al., 1998) because of two amino acid differences (amino acids 68 and 73) in the MIR loop between human α3 and rodent α3. Changing these two amino acids in rat α3 to their human α3 equivalents allows mAb 210 to bind rodent α3 (Ficklin et al., 2005). Rat mAbs to the MIR may recognize epitopes slightly different from those recognized by humans, because B cells which produce antibodies that can react with the human α3 epitope are not eliminated.
mAb 192 competes for binding with mAb 35 (Tzartos et al., 1998). Its binding did not require the MIR loop upon which binding of mAb 35 is dependent. This can be explained by the observation that mAbs can obscure large areas (750 Å) on proteins to which they are bound (Mariuzza et al., 1987; Konstantakaki et al., 2007).
Fostieri et al. (2005) reported another example of the heterogeneity of MIR epitopes. A Fab derived from a MG patient, sh6.4, inhibited binding of mAb 35 and 198, but not mAb 192. sh6.4 did not bind free α1 subunits, but it did bind human αγ and αε dimers. Some MIR epitopes may contain some parts of neighboring subunits. This could explain the higher affinities of mAbs for AChRs as compared to monomeric α1 subunits (Conroy et al., 1990). The absence of parts of the MIR formed by adjacent subunits in the α1/AChBP chimeras may reduce binding of some MG autoantibodies.
α1/α7 chimeras expressed on the oocyte surface retain their ability to function as AChRs. The sensitivity to activation of chimeric AChRs by ACh was greatly influenced by the α1 sequences within them. Addition of the sequence 15–32 to the α1(2–14, 60–81)/α7 chimera resulted not only in binding of the MG patient mAb 637 but also in a 130-fold increase in sensitivity to activation by ACh. This suggests that interaction between parts of the sequences that form the conformation-dependent MIR may influence conformational changes in this region associated with activation (Henchman et al., 2003). Although closing of the C loop over the ACh binding site when agonists bind is a critical conformational change associated with activation (Hansen et al., 2005), and is propagated through the AChR to regulate opening of the distant channel gate, AChR activation must be associated with conformation changes throughout the AChR. For example, changing accessory subunits (like β1 of muscle AChRs), which do not participate in forming ACh binding sites, has large effects on the sensitivity of AChRs to activation (Kuryatov et al., 2008). The decreased binding affinity of mAbs in the presence of either nicotine or αBgt suggests small effects on the conformation of the MIR with a desensitized state or a resting state. Also, the α7(66–76)/α1 chimera desensitized more rapidly than wild-type α1. Binding of mAb 210 to the α1(2–14, 60–81)/α7 chimera reduced activation by 70%, consistent with there being some conformational feedback between the MIR and the ACh binding sites.
Disrupting the structure of the MIR can disrupt conformational maturation of α1 subunits and expression of α1 AChRs. Incorporation of the 25 amino acid P3A exon between amino acids 58 and 59 of human AChR α1 subunits prevents expression of the AChR (Newland et al., 1995).
The importance of the MIR in conformational maturation is demonstrated by our observation that incorporation of the α1 MIR into α7 greatly increases chimeric subunit conformational maturation and expression of mature AChRs. Efficient assembly of α7 AChRs has been difficult to achieve from cloned subunits (Gee et al., 2007). In α1 subunits, conformational maturation of the MIR and the ACh binding site (detected by the ability to bind mAb 35 and αBgt, respectively) occur simultaneously before assembly (Merlie and Lindstrom, 1983). Conformational maturation of chimeric α1/α7 subunits may be nucleated by the association of the N-terminal and MIR loop components of the MIR, and this may drive efficient assembly of these chimeric AChRs. One of the normal functional roles of the MIR may be to nucleate conformational maturation of unassembled subunits.
Recently Castillo et al. (2009) found that deleting or disrupting the N-terminal α helix of α7, α3, α4, β2 or β4 AChR subunits or 5-HT3A subunits prevented assembly of mature AChRs. In congenital myasthenic syndromes, some mutations in or near the N-terminal α helix and MIR loop in the AChR subunits reduce expression of AChR (Engel and Sine, 2005). These observations suggest that the importance of the interaction between the N-terminal α helix and MIR loop, for conformational maturation and assembly, which we have demonstrated in α1/α7 chimeras, is a general phenomenon applying to all subunits of AChRs and related receptors. Transmembrane domain interactions and cytoplasmic regions adjacent to the transmembrane domains can also influence expression and function of α7 AChRs (Castelán et al., 2007; Gee et al., 2007).
In the future, x-ray crystallography of mAbs bound to the MIR in α1/AChBP chimeras should allow us to determine the structure of the MIR at atomic resolution. These chimeras should also permit testing the ability of a native α1 MIR structure to induce or inhibit EAMG.
Footnotes
-
This research was supported by National Institutes of Health Grants NS11323 and NS052463 to J.L. and UO1 DA019372 and GM 18360 to P.T. This work was also supported by grants from the Prinses Beatrix Fonds, L'Association Française contre les Myopathies and the MYASTAID project grant of the Sixth Framework Programme of the European Community to M.d.B. We thank Barbara Campling for comments on this manuscript.
- Correspondence should be addressed to Dr. Jon M. Lindstrom, Department of Neuroscience, University of Pennsylvania Medical School, 217 Stemmler Hall, 36th and Hamilton Walk, Philadelphia, PA 19104-6074. jslkk{at}mail.med.upenn.edu
