ELIC-α7 Nicotinic Acetylcholine Receptor (α7nAChR) Chimeras Reveal a Prominent Role of the Extracellular-Transmembrane Domain Interface in Allosteric Modulation*

Background: Allosteric modulators bound to the transmembrane domain (TMD) of the α7 nicotinic acetylcholine receptor (α7nAChR) can potentiate channel function. Results: ELIC-α7nAChR showed potentiation only when the extracellular-transmembrane domain (ECD-TMD) interface matched that of α7nAChR. Conclusion: PAM modulation through the TMD requires a more specific ECD-TMD interface than agonist activation. Significance: The study provides insight into the basis for positive allosteric modulation of α7nAChR. The native α7 nicotinic acetylcholine receptor (α7nAChR) is a homopentameric ligand-gated ion channel mediating fast synaptic transmission and is of pharmaceutical interest for treatment of numerous disorders. The transmembrane domain (TMD) of α7nAChR has been identified as a target for positive allosteric modulators (PAMs), but it is unclear whether modulation occurs through changes entirely within the TMD or changes involving both the TMD and the extracellular domain (ECD)-TMD interface. In this study, we constructed multiple chimeras using the TMD of human α7nAChR and the ECD of a prokaryotic homolog, ELIC, which is not sensitive to these modulators, and for which a high resolution structure has been solved. Functional ELIC-α7nAChR (EA) chimeras were obtained when their ECD-TMD interfaces were modified to resemble either the ELIC interface (EAELIC) or α7nAChR interface (EAα7). Both EAα7 and EAELIC show similar activation response and desensitization characteristics, but only EAα7 retained the unique pharmacology of α7nAChR evoked by PAMs, including potentiation by ivermectin, PNU-120596, and TQS, as well as activation by 4BP-TQS. This study suggests that PAM modulation through the TMD has a more stringent requirement at the ECD-TMD interface than agonist activation.


The native ␣7 nicotinic acetylcholine receptor (␣7nAChR) is a homopentameric ligand-gated ion channel mediating fast synaptic transmission and is of pharmaceutical interest for treatment of numerous disorders. The transmembrane domain (TMD) of ␣7nAChR has been identified as a target for positive allosteric modulators (PAMs), but it is unclear whether modulation occurs through changes entirely within the TMD or changes involving both the TMD and the extracellular domain (ECD)-TMD interface.
In this study, we constructed multiple chimeras using the TMD of human ␣7nAChR and the ECD of a prokaryotic homolog, ELIC, which is not sensitive to these modulators, and for which a high resolution structure has been solved. Functional ELIC-␣7nAChR (EA) chimeras were obtained when their ECD-TMD interfaces were modified to resemble either the ELIC interface (EA ELIC ) or ␣7nAChR interface (EA ␣7 ). Both EA ␣7 and EA ELIC show similar activation response and desensitization characteristics, but only EA ␣7 retained the unique pharmacology of ␣7nAChR evoked by PAMs, including potentiation by ivermectin, PNU-120596, and TQS, as well as activation by 4BP-TQS. This study suggests that PAM modulation through the TMD has a more stringent requirement at the ECD-TMD interface than agonist activation.
The ␣7 nicotinic acetylcholine receptor (␣7nAChR) 2 is a homopentameric acetylcholine-gated cation channel mediating fast synaptic transmission in neuronal cells (1) and calcium signaling in nonneuronal cells (2). The diverse biological functions of ␣7nAChR have made it a promising therapeutic target for numerous medical conditions, including pain (3,4), inflammation (5), cardiovascular disease (6), and a variety of psychiatric and neurological disorders (7)(8)(9). ␣7nAChR opens the channel gate in the transmembrane domain (TMD) upon agonist binding to the extracellular domain (ECD), but channel function can also be modulated by ligand binding to allosteric sites distinct from the orthosteric agonist binding-sites in the ECD. Positive allosteric modulators (PAMs), especially those specific for ␣7nAChR, such as PNU-120596 (10) and TQS (11), are particularly of therapeutic potential (8). Based upon mutational analysis, certain PAMs are thought to act at the ECD or ECD-TMD interface (12,13), whereas others, such as PNU-120596 and TQS, are thought to act through binding to intrasubunit sites near the middle of the ␣7nAChR TMD (13)(14)(15)(16)(17). The molecular mechanisms of allosteric modulation are not understood.
␣7nAChR is a member of the pentameric ligand-gated ion channel (pLGIC) superfamily, including different subtypes of nAChRs, 5-HT 3 A receptors (5-HT 3 ARs), glycine receptors (GlyRs), and GABA A receptors (GABA A Rs). ␣7nAChR is also homologous to GluCl, and the prokaryotic pLGICs ELIC and GLIC, whose crystal structures have been determined in the absence and presence of ligands (18 -28). A wealth of data from decades of biochemical characterization and more recent structural characterization of pLGICs provide a broad understanding of the events relating to agonist binding and channel activation (28 -35). Agonist binding to orthosteric sites in the ECD leads to conformational transitions that propagate through the ECD-TMD interface to the TMD and open the channel gate. Upon prolonged exposure to agonist, the channel desensitizes and is no longer sensitive to agonist binding until it returns to the resting state. Among pLGICs, ␣7nAChR demonstrates particularly rapid desensitization. PAMs such as ivermectin, PNU-120596, and TQS can significantly slow down the desensitization of ␣7nAChR (10,11,36,37). Several studies have demonstrated involvement of the ECD-TMD interface in desensitization (38 -41), but the role of the ECD-TMD interface in allosteric modulation through the TMD is less known.
Chimeras joining the ECD from one pLGIC with the TMD from another provide valuable opportunities to understand how the interplay among different domains/regions underlies pLGIC function. Previous studies of GLIC-␣1GlyR (42), ␣7nAChR-␣1GlyR (43), ␣4␤2nAChR-5HT 3 R (44), and ␣7nAChR-5HT 3 R (45) found that functional chimeras could be obtained by splicing the ECD prior to the first transmembrane helix of the TMD, although kinetic parameters could be improved by optimization of the ECD-TMD interface. In contrast, it has proven difficult to express a functional chimera of an ␣7nAChR TMD (13). A 5HT 3 R-␣7nAChR chimera expressed a small current in only a few oocytes, which was sufficient to show that potentiation by the PAM PNU-120596 could be conferred by the ␣7nAChR TMD when activated by a suboptimal agonist (13). However, given the high homology between loop 7 and the TM2-3 linker of ␣7nAChR and 5HT 3 R (approximately 78% similarity), it is unclear what role the ECD-TMD interface might have played in the observed potentiation.
In this study, we constructed multiple chimeras combining the TMD of human ␣7nAChR and the ECD of the prokaryotic homolog ELIC (Fig. 1). ELIC is a cation channel activated by a group of primary amines, including cysteamine (46). In contrast to ␣7nAChR, ELIC is inhibited by the general anesthetic propofol but unaffected by PAMs like ivermectin (21,47). We show here that ELIC is also unaffected by the ␣7nAChR-specific PAMs PNU-120596 and TQS. We examined the effect of the ECD-TMD interface on allosteric modulation through the TMD by engineering two classes of ELIC-␣7nAChR (EA) chimeras: EA ELIC , with an ECD-TMD interface resembling ELIC, and EA ␣7 , with an ␣7nAChR ECD-TMD interface. Both classes of EA chimeras resulted in functional channels with similar activation and desensitization kinetics. However, only EA ␣7 retained sensitivity to the PAMs ivermectin, PNU, and TQS as well as the ability to be activated by 4BP-TQS. Thus, allosteric modulation through the TMD requires a specific ECD-TMD interface similar to that of ␣7nAChR, a more stringent requirement than that needed to activate the channel by agonists.

EXPERIMENTAL PROCEDURES
Construction of EA Chimeras-EA chimeras were constructed by fusing the ECD of ELIC ending after pre-TM1 (ELIC-S201) with the beginning of the human ␣7nAChR TMD (␣7-Y210) using overlapping PCR (48). The resulting construct was subcloned into a Xenopus expression vector, pOTV (49) for expression of RNA. Mutations to match the sequence of interface elements between ELIC and ␣7nAChR were introduced using QuikChange Lightning Site-directed Mutagenesis kits (Agilent). All constructs were confirmed by sequencing in full.
Modeling of Chimeras-Homology models for EA chimeras were generated using Modeler 9v8 (51, 52) based on the crystal structure of ELIC (Protein Data Bank (PDB) ID code 3RQW) and the NMR structure of the ␣7nAChR TMD (PDB ID code 2MAW). Five independent models were generated, and the model with the lowest discrete optimized potential energy was selected for presentation.

RESULTS
Engineering Functional EA Chimeras-Multiple EA chimeras were constructed to assess the requirements for efficient coupling at the ECD-TMD interface. The function of each chimera was tested in Xenopus oocytes by two-electrode voltage clamp electrophysiology. The data for agonist response are summarized in Table 1.
The original ECD-TMD interfaces for ELIC and ␣7nAChR are not compatible. A chimera (EA ELIC 1 ) simply connecting the ELIC ECD with the ␣7nAChR TMD at the beginning of TM1 did not show any response to the ELIC agonist cysteamine when the chimera was expressed in Xenopus oocytes (Table 1). This was unexpected. Chimeras between most members of the pLGIC family were found to have at least some degree of function when joined in this manner (13,(42)(43)(44)(45), with the exception of the AChBP-5HT 3 A chimeras (53). To rule out interference by the extended C terminus of ␣7nAChR, we replaced the last 18 residues of EA ELIC 1 with the last 4 residues of ELIC. This construct EA ELIC 2 also failed to exhibit current in response to cysteamine ( Table 1).
The sequence, length, and position of the TM2-3 linker vary among the known pLGIC structures. The TM2-3 linker comprises 8 residues beginning at 22Ј in ELIC (PDB ID code 3RQU) (20) and 10 residues beginning at 22Ј in the NMR structure for the TMD of ␣7nAChR (PDB ID code 2MAW) (54). The TM2-3 linker in the cryo-EM structure of Torpedo nAChR has only 7 residues in the flexible loop beginning at 29Ј. If one counts the TM2-3 linker of Torpedo nAChR starting from 22Ј, it has a total of 14 residues (PDB ID code 2BG9) (55). A recent disulfide trapping experiment with the ␣ subunit of mouse muscle nAChR showed a shift in register between TM2 and TM3, which could not be reconciled with 2BG9 (56), but was consistent with the ␣7 NMR structure (54).
To improve coupling of the ␣7nAChR TM2-3 linker to the ELIC ECD without disrupting TMD interactions, we substituted varied lengths of the TM2-3 linker in ␣7nAChR with their counterparts in ELIC. Substitution of only 4 residues in the middle of the TM2-3 linker (EA ELIC 3 ) or more extensive substitution from 19Ј to 28Ј (EA ELIC 4 ) was not functional ( is not the length of the TM2-3 linker, but its position at the ECD-TMD interface that is critical for efficient coupling. To engineer a chimera with a more native ␣7nAChR ECD-TMD interface, we kept the ␣7nAChR TM2-3 linker intact but mutated the ELIC ECD to match the sequence of ␣7nAChR at loops 2, 7, 9, and the pre-TM1 region. This construct, EA ␣7 , is functional with an EC 50 of 1.3 mM for cysteamine (Table 1 and Fig. 2). EA chimeras with mismatches at loop 2 (EA ␣7 1 ) or loop 9 (EA ␣7 2 ) still expressed as functional channels, but with a rightward shift in EC 50 for cysteamine, suggesting a lower coupling efficiency (Table 1 and Fig. 2c). Mismatches at both loop 2 and loop 9 (EA ␣7 3 ), or at loop 7 alone (EA ␣7 4 ), resulted in chimeras that did not exhibit current in response to cysteamine (Table 1). Taken together, these results suggest that an ensemble of coupling among loop 2, loop 7, loop 9, the pre-TM1 region and the TM2-3 linker is essential for obtaining functional channels. Similar modifications were also needed for the functional AChBP-5HT 3 chimera (53).
In contrast to EA ␣7 , EA ELIC 5 and EA ELIC were insensitive to ivermectin, TQS, or PNU-120596 (Fig. 6). They were also insensitive to propofol. The lack of sensitivity to propofol for the EA chimeras can be explained by the loss of the binding site within the ELIC TMD. However, EA ELIC 5 , EA ELIC , and EA ␣7 all have the ␣7nAChR TMD, which contains the binding sites for the tested PAMs. These chimeras differ only in the ECD-TMD interface, suggesting a functional role of the ECD-TMD interface in allosteric potentiation through the TMD.

DISCUSSION
Unlike other Cys-loop receptor chimeras (42)(43)(44)(45), the engineering of functional chimeras between ELIC and ␣7nAChR requires more extensive optimization of an ensemble of interactions at the ECD-TMD interface. It does not matter how the optimization is achieved: chimeras with either an ELIC interface (EA ELIC 5 , EA ELIC ) or an ␣7nAChR interface (EA ␣7 ) show similar agonist response and desensitization characteristics. They desensitize much slower than native ␣7nAChR. PAM modulation through the TMD only occurs with the ␣7nAChR ECD-TMD interface, suggesting that PAM modulation through the TMD requires a more specific ECD-TMD interface than agonist activation.
One of the primary effects noted for PAMs like PNU-120596 and TQS is a marked decrease in the ␣7nAChR desensitization rate (10,11,36,37). Our study shows that EA ␣7 does not reproduce the fast desensitization characteristic of ␣7nAChR, yet can still be modulated by these PAMs, suggesting that fast desensitization may not be required for potentiation by PNU-120596 or TQS . Because EA ␣7 and EA ELIC show similar desensitization rates, desensitization cannot explain their distinct responses to the tested PAMs.
What may have contributed to the different responses of EA ␣7 and EA ELIC to PAMs? The PAMs tested here bind to the TMD (13)(14)(15)58). The differences between EA ␣7 and EA ELIC / EA ELIC 5 are localized only to the ECD-TMD interface. One may suspect that lack of modulation in EA ELIC 5 results from interference with PAMs binding to a structurally distorted TMD due to a shortened TM2-3 linker. However, such a possibility is dismissed by EA ELIC , which has a full-length TM2-3 linker that is unlikely to distort the TMD.
Because no high resolution structural data are available, the precise binding sites for ivermectin, PNU-120596, or TQS in ␣7nAChR are unknown. In the Caenorhabditis elegans GluCl crystal structure, ivermectin contacts both the TM2-3 linker and the residues toward the middle of the TMD (18). Mutational analysis on ␣7nAChR also suggests that ivermectin interacts with residues near the TM2-3 linker and residues near the middle of the TMD (16,59). If direct contacts to specific residues in the TM2-3 linker are critical for ivermectin modulation, the ELIC TM2-3 linker may not provide proper contacts in EA ELIC and EA ELIC 5 , thereby preventing ivermectin modulation.
In the case of PNU-120596 and TQS, they are much smaller than ivermectin and cannot simultaneously contact residues at both the ECD-TMD interface and near the middle of the TMD. If they directly contact the ECD-TMD interface, the role of the ECD-TMD interface in PAM modulation will be the same as discussed above for ivermectin. However, previous mutation studies suggest an intrasubunit binding site near the middle of the TMD for PNU-120596 and TQS (13)(14)(15)58). In this case, the binding site is remote from the ECD-TMD interface. How can the interface still be important for modulation? It is possible that the PAM binding to the proposed site near the middle of TMD allosterically induces changes to the ECD-TMD interface that are required for potentiation. These changes can be accommodated by the ECD-TMD interface in ␣7nAChR, but not in ELIC. This hypothesis is supported by a substituted cysteine accessibility study (60), which demonstrates that PNU-120596 binding can independently induce conformational changes to the ECD and ECD-TMD interface. The conformational changes induced by PNU-120596 are similar but not identical to those induced by the agonist acetylcholine (60).
Structural characterization of the precise binding sites for these PAMs is essential for understanding the mechanism of PAM modulation. Future high resolution structural studies of the EA chimeras in the absence and presence of modulators will provide insights into binding of PAMs to the TMD and the role of the ECD-TMD interface in allosteric modulation. Because of the prokaryotic origin of its ECD, the EA chimeras are good candidates for production in quantities suitable for high resolution structural studies (19,20).