Efficient Expression of Functional (α6β2)2β3 AChRs in Xenopus Oocytes from Free Subunits Using Slightly Modified α6 Subunits

Human (α6β2)(α4β2)β3 nicotinic acetylcholine receptors (AChRs) are essential for addiction to nicotine and a target for drug development for smoking cessation. Expressing this complex AChR is difficult, but has been achieved using subunit concatamers. In order to determine what limits expression of α6* AChRs and to efficiently express α6* AChRs using free subunits, we investigated expression of the simpler (α6β2)2β3 AChR. The concatameric form of this AChR assembles well, but is transported to the cell surface inefficiently. Various chimeras of α6 with the closely related α3 subunit increased expression efficiency with free subunits and produced pharmacologically equivalent functional AChRs. A chimera in which the large cytoplasmic domain of α6 was replaced with that of α3 increased assembly with β2 subunits and transport of AChRs to the oocyte surface. Another chimera replacing the unique methionine 211 of α6 with leucine found at this position in transmembrane domain 1 of α3 and other α subunits increased assembly of mature subunits containing β3 subunits within oocytes. Combining both α3 sequences in an α6 chimera increased expression of functional (α6β2)2β3 AChRs to 12-fold more than with concatamers. This is pragmatically useful, and provides insights on features of α6 subunit structure that limit its expression in transfected cells.

Chimeras reveal a6 sequences that limit expression of transfected a6* AChRs. Chimeras with the extracellular domain of a6 and the remainder of a3 or a4 expressed in combination with b2 are efficiently transported to the surface [12]. The large cytoplasmic domain of some AChR subunits promotes assembly and transport of AChRs [17]. This suggests that the large cytoplasmic domain of a3 or a4 might provide efficient transport to the surface that a6 does not. The cytoplasmic domain of a3 is smaller than that of a4 and more closely resembles the sequence of a6.
Analysis of chimeras of a6 and a3 identified a3 sequences that permitted expression of a6* AChRs and sequences of a6 that inhibited expression of a3 AChRs [16]. A region of a6 in the first half of transmembrane domain 1 inhibited expression of a3 AChRs [16]. This region contains a unique methionine at position 211 which is occupied by leucine in a3 and other a subunits. M211 is in a sequence that in a1 subunits governs stability, assembly, and transport to the cell surface [18]. This sequence is on the side of the a1 subunit that assembles with the accessory subunit (e.g. b1 in a1* AChRs or b3 in a6* AChRS) [19]. This suggests that the a6 sequence containing M211 might be important for associating with b3 accessory subunits. a6b2b3* AChRs are usually expressed in aminergic neurons at presynaptic locations [1,20,21]. These neurons may express a chaperone for assembly of b3 subunits with a6 that is missing in other cell types such as Xenopus oocytes or human embryonic kidney (HEK) cell lines. These cells efficiently assemble b3 with other a subunits to form AChRs such as (a4b2) 2 b3 [22].
Here we report efficient expression of (a6b2) 2 b3 AChRs in Xenopus oocytes using free subunits with only small changes in a6 subunits, while not altering AChR pharmacology or channel structure. We explored the effects of incorporating M211L and a3 cytoplasmic domain alone, and together, into chimeras with a6. M211L increased assembly with b3, and a3 cytoplasmic domain increased assembly with b2 and transport to the surface. Together, these two modifications synergistically permitted expression of high levels of (a6b2) 2 b3 AChRs from free subunits. These AChRs exhibited the same pharmacological properties as concatameric AChRs, but were expressed on the oocyte surface in much greater amounts.

Construction of a6/a3 Chimeras
We prepared chimeras from a3 and a6 subunits [12]. To replace the cytoplasmic domain of a6 with the corresponding part of a3, the ApaLI restriction site was introduced at position Ile297H is 298 of a6 cDNA using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). This restriction site is present in native a3 cDNA. Using this restriction site and the NcoI site that is common between a6 and a3 in the beginning of the M4 domain, we replaced the cytoplasmic domain of a6 with the a3 domain, leaving intact the M4 transmembrane domain and C-tail of a6. To form the a6 a3cyt-C construct, the cytoplasmic domain, M4 transmembrane domain and C-tail of a6 were replaced with a3 using an introduced ApaLI site and an EcoRI site from the psp64 (polyA) plasmid.
To make M211L chimeras, we made the M211L change in a6 sequence using a GAAGATTGCCGCTGTTTTACACG oligo and a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) for native a6 or a6 with an a3 cytoplasmic domain.
b3 and a6 joined with a Q 4 A 3 PAQ 3 AQA 3 PA 2 Q 5 (QAP) linker to form the b32a6 concatamer was used here. Using previously prepared b3 with a BspEI site at the end of the coding domain and a6 with a FspI site at the end of signal peptide [16], we inserted QAP linker with these oligos: CCGGGAACAGCAACAGCAAG-CAGCGGCTCCGGCCCAACAGC.
AAGCACAGGCGGCTGCCCCCGCAGCGCAACAACAG-CAACAGTGC and GCACTGTTGCTGTTGTTGCGCTGC-GGGGGCAGCCGCCTGTGCTTGCTGTTGGGCCGGAG-CCGCTGCTTGCTGTTGCTGTTC. This QAP linker is modeled on one used to express concatameric GABA A receptors in HEK 293 cells [23]. To prepare the b32a6 concatamer with the M211L chimera and a3 cytoplasmic domain, we used the unique site of EcoRV in the a6 sequence and the unique site of PvuI in the psp64 (polyA) plasmid.

DNA Preparation
Two microliters of DNA ligations were transformed into XL-10 Gold Ultracompetent cells (Stratagene, La Jolla, CA) using the protocol in the kit. Colonies were selected using the QIAquick spin miniprep kit. Miniprep DNA was tested for correct sequence by restriction enzyme digestion and subsequent agarose gel electrophoresis for correct size of fragments. DNAs were purified using Qiagen plasmid midiprep kit (QIAGEN) and concentrations were calculated using spectrophotometry.

Oocyte Injection
Xenopus laevis oocyte harvest from Xenopus laevis frogs was performed in accordance with our approved IACUC protocol. The cRNA encoding desired subunits was synthesized from 1 mg of linearized cDNA templates in the pSP64 vectors using SP6 RNA polymerase from the mMessage mMachine kit (Ambion Inc, Austin, TX). Subunit cRNAs were mixed at a 1:1:1 ratio of a6:b2:b3 for constructs of free subunits. Dimeric concatamer was mixed with a6 chimera and b2 subunits at a 1:1:1 ratio. Xenopus laevis oocytes were injected with 100 ng of cRNA mix per oocyte, then incubated in 50% L-15 (Invitrogen, San Diego, CA), 10 mM HEPES, pH 7.5, 10 units/ml penicillin, 10 mg/ml streptomycin, and 50 mg/ml gentamycin [24]. This medium was refreshed daily, and replaced with gentamycin-free medium the day before recording.

Surface Expression of AChR
Surface expression was determined by binding of 125 I monoclonal antibody (mAb) 295 performed on the same day as electrophysiological recording of responses to 30 mM ACh. Groups of 8 oocytes were placed in an Eppendorf tube with 525 ml of L-15 media containing 10% horse serum and 5 nM 125 I mAb 295 [25] at room temperature for 3 hours. Unbound 125 I mAb was removed by 3 washes with 1 ml of L-15, then 125 I bound to individual oocytes was determined in a c-counter. Nonspecific binding was determined using non-injected oocytes.

H Epibatidine Binding
Groups of 8 oocytes were homogenized in 1 ml of buffer A (50 mM NaCl, 50 mM sodium phosphate buffer, pH 7.5, 5 mM EDTA, 5 mM EGTA, 5 mM benzamide, 5 mM iodoacetamide, 2 mM phenylmethylsulfonyl fluoride). Then a crude membrane fraction was pelleted by centrifugation for 15 minutes at 13,400 rpm. Membrane proteins were resuspended by pipetting and solubilized in 150 ml of buffer A containing 2% Triton X-100 for 1 hour at room temperature. Debris was removed by centrifugation at 13,400 g for 15 min. Next, mAb 295 coated wells were loaded with aliquots of detergent extracts with 2 nM 3 H epibatidine (PerkinElmer Life Sciences, Emeryville, CA) in a total volume of 100 mL in phosphate-buffered saline (PBS) buffer containing 0.5% Triton X-100 and 10 mM NaN 3 [12]. These plates were left overnight on a shaker at 4uC. Wells were then washed three times with 0.5% Triton X-100 in PBS. 3 H epibatidine bound to AChRs bound to the wells through mAb 295 or mAb 210 was eluted with 30 ml of 0.1 M NaOH and quantitated by liquid scintillation counting [22]. Nonspecific binding was determined using non-injected oocytes.

Sucrose Gradient Sedimentation
Triton X-100-solubilized AChRs from oocytes were prepared as described above from groups of 30-50 oocytes [16]. Aliquots (150 ml) of the extracts, mixed with 2 ml of 1 mg/ml of Torpedo californica electric organ AChR, were loaded onto 11.3 ml sucrose gradients [linear 5-20% sucrose (w/v) in 10 mM sodium phosphate buffer, pH 7.5, that contained 100 mM NaCl, 1 mM NaN 3 , 5 mM EDTA, 5 mM EGTA, and 0.5% Triton X-100] [16,26]. Gradients were centrifuged for 16 hours at 40,000 rpm in a SW-41 rotor (Beckman Coulter, Fullerton, CA) at 4uC. Fractions were collected at 15 drops per well from the bottom of the tubes. Fifty microliters of each fraction were transferred to mAb 295coated wells to isolate b2-containing AChRs for measurement of 3 H epibatidine binding, and 20 ml of each fraction were transferred to mAb 210-coated wells to isolate Torpedo californica AChRs used as internal molecular weight standards [26]. Fractions in mAb 295-coated wells were incubated with 2 nM 3 H epibatidine at 4uC overnight. Fractions in mAb 210 coated wells was incubated with 1 nM 125 I a bungarotoxin at 4uC overnight. Wells were then washed three times with PBS and 0.5% Triton X-100. Bound 3 H epibatidine was determined by liquid scintillation counting.

Electrophysiology
Electrophysiological recording was performed with a twomicroelectrode voltage clamp amplifier (Oocyte Clamp OC-725; Warner Instruments, Hamden, CT) on Xenopus oocytes with a constant flow of ND-96 solution (96 mM NaCl, 1.8 mM CaCl 2 , 1 mM mgCl 2 , and 5 mM HEPES, pH 7.5) containing 0.5 mM atropine [16,27]. Whole-cell membrane currents were recorded in response to application of agonists 5-7 days after RNA injection at a clamp potential of 270 mV. Currents were measured in response to application of various concentrations of agonists for 4 seconds. Between agonist applications, the recording chamber was washed with ND-96 buffer containing 0.5 mM atropine for 3 minutes. Responses were normalized to the maximum response induced by acetylcholine (ACh)(30 mM). The mean value of at least five oocytes was used for graphing concentration/response curves. Values are expressed 6 standard error.
Data were analyzed with KaleidaGraph version 4.1 software for common statistical determinants (Synergy Software, Reading, PA). EC 50 , efficacy and Hill co-efficiency were obtained from the Hill equation as described [24].

Design of (a6b2) 2 b3 AChR Constructs
Subunit concatamers and chimeras of a6 and a3 subunits were incorporated into a series of constructs for expressing (a6b2) 2 b3 AChRs as shown in Figure 1. Concatameric pentamer (construct 1) linked through (AGS) n linkers ( Figure 1B and 1D) enabled successful expression of functional (a6b2) 2 b3 AChRs in oocytes [16]. Free native a6, b2, and b3 subunits (construct 2) did not permit assembly of functional AChRs [12,16]. To investigate the effect of exchanging methionine 211 and/or cytoplasmic domain of a6 subunit to that of a3, three chimeras were made to build constructs 3, 4 and 5 ( Figure 1C and 1D). Chimera a6 a3cyt-C and construct 6 were made to investigate the effect of the short Cterminus following the last transmembrane domain of the a6 subunit. To increase b3 incorporation, a QAP linker was used to link b3 with a6. Combining b3-QAP-a6 concatamer with various chimeras shown in Figure 1C, constructs 7, 8 and 9 were made with one or two chimeric a6 subunits. Numbering and nomenclature of the constructs described above is used in the subsequent data figures.

Expression Efficiency in Oocytes
To evaluate efficiency of expression of the nine constructs in Figure 1, Xenopus oocytes were injected with mRNAs and six days later tested for total binding site assembly, surface protein level and responses to ACh. ACh binding sites were measured by binding of 3 H epibatidine to immunoisolated detergent solubilized components containing b2 subunits. Expression in the surface membrane was assayed by binding to oocytes of 125 I mAb 295 to b2 subunits. Biophysical properties were determined by examining the currents induced by 30 mM ACh. Results are shown in Figure 2.
Various constructs produced very different amounts of ACh binding sites (Figure 2A). The amounts of AChRs on the oocyte surface were not proportionate to the total amount of ACh binding sites. For example, constructs 2, 4 and 5 all yielded more than 90 fmols of 3 H epibatidine bound per oocyte but expressed very different amounts of AChRs on the surface, as low as 0.16860.070 fmol for construct 2 or as high as 8.3164.97 fmol for construct 5. This indicates that some constructs result in incompletely assembled AChRs or properly assembled AChRs that were not transported to the cell surface, as shown previously [12,16].
AChR function and AChR expression on the oocyte surface were closely correlated ( Figure 2B), indicating that AChRs on the surface produced by most constructs had similar functional properties. The pentameric concatamer 1 (b32a62b22a62b2) exhibited the expected pharmacological properties and is considered the positive control [16]. Mature concatameric AChRs were assembled, but were inefficiently transported to the oocyte surface. Free native construct 2 (a6+b2+b3) subunits are the negative control. Free native subunits assembled large numbers of ACh binding sites. These a6b2* complexes assemble into amorphous aggregates, but virtually no mature AChRs [12]. Construct 3 (a6 211L +b2+b3) did not increase surface expression or function, and decreased assembly of ACh binding sites. Construct 4 (a6 a3cyt +b2+b3) greatly increased assembly with b2, surface expression, and function compared to construct 3. Construct 5 (a6 211L,a3cyt +b2+b3) combines the a3 components of constructs 3 and 4 in one a6 chimera. Construct 5 increased expression on the surface and function 40-80 fold compared to free subunits and 10-13 fold compared to concatamer 1.
Other constructs did not approach the efficiency of expression obtained with construct 5 (a6 211L,a3cyt +b2+b3). Construct 6 (a6 a3cyt-C +b2+b3) reduced assembly of ACh binding sites, surface expression, and function compared to construct 4 (a6 a3cyt +b2+b3). Thus, including a3 transmembrane domain 4 and C-terminal domain impaired assembly and transport. Subsequent experiments will show that pharmacological function was also altered. Construct 7 (b32a6+a6 a3cyt +b2) was intended to test whether the b32a6 concatamer equaled or exceeded the effect of the a6 211L chimera in promoting assembly with b3 and whether a single a3 cytoplasmic domain was sufficient for enhanced assembly and transport. Construct 7 assembled fewer ACh binding sites than construct 4 with two a3 cytoplasmic domains, and was equally expressed on the surface. This is consistent with the ideas that assembly of b2 with a6 in the concatamers was reduced due to the absence of an a3 cytoplasmic domain in this a6 and that one a3 cytoplasmic domain per AChR is sufficient for transport to the surface. The b32a6 concatamer provided no benefit. Construct 8 (b32a6+a6 211L,a3cyt +b2) had no better surface expression or function than 7 and lower assembly of ACh binding sites. Thus, the b32a6 concatamer with a single or double a6 chimera provided no benefit. Construct 9 (b32a6 211L,a3cyt +a6 211L,a3cyt + b2) expressed functional AChRs better than 8. However, the b32 a6 211L,a3cyt concatamer substantially impaired assembly relative to that achieved with the a6 double chimeras as free subunits. This is consistent with the idea that this concatamer did not promote assembly of b3 with a6 and prevented M211 from promoting this assembly which does occur with free subunits. The b32a6 concatamer used a QAP linker rather than the (AGS) n linkers used in construct 1. The change in linker may account for the ineffectiveness of the b32a6 concatamer, but this linker has proven effective in several other constructs (unpublished).
Most constructs exhibited a similar ratio of current per surface AChR in response to activation by 30 mM ACh (Figure 3). This suggests that they have similar channel opening and/or channel conductance. Construct 6 (a6 a3cyt-C +b2+b3) had larger current per surface AChR. This suggests that a3 transmembrane domain 4 and/or the extracellular C-terminal domain altered channel Figure 1. Illustration of (a6b2) 2 b3 AChR constructs. A) Diagrammatic representation of an AChR subunit. B) Diagrammatic representation of two AChR subunits joined by a linker. Direction of the linker is indicated by arrows. C) Representation of a6 and a3 sequences used in the a6/a3 chimeras studied. D) Representation of (a6b2) 2 b3 AChRs assembled from the various constructs. Agonist binding sites are shown as solid triangles between two subunits. The number and nomenclature for each construct depicted here are used in the following data figures. doi:10.1371/journal.pone.0103244.g001 properties. Construct 9 also exhibited larger current per surface AChR for reasons which are not evident.
Incorporation of b3 Accessory Subunit b3 functions only as an accessory subunit, thus does not form ACh binding sites with a subunits. As expected, incorporation of b3 is complete in concatamers of construct 1 (b32a62b22a62 b2, Figure 4). Incorporation of b3 is least in the large number of partially assembled and aggregated a6b2 ACh binding site complexes formed from the free native subunits of construct 2 (a6+b2+b3, Figure 4). Changing only the a6 methionine 211 to leucine in construct 3 (a6 211L +b2+b3) results in complete incorporation of b3 ( Figure 4) and formation of functional AChRs ( Figure 2B). Thus, this part of a6 transmembrane domain 1 on the side of a6 where the accessory subunit is expected to assemble [19], contributes to assembly of b3. Changing only the large cytoplasmic domain of a6 for that of a3 in construct 4 (a6 a3cyt + b2+b3) increases assembly of b3, but not as effectively as changing the single amino acid M211 (Figure 4). However, the a3 cytoplasmic domain is efficient at increasing both assembly with b2 and transport to the cell surface (Figure 2A and Figure 4). Combination of the two chimeras in construct 5 exhibited complete incorporation of b3 ( Figure 4) and efficient assembly of mature AChRs contributed by M211L ( Figure 2B) with the efficient assembly with b2 and transport to the surface contributed by the cytoplasmic domain of a3 (Figure 2A) to produce large amounts of functional AChRs on the cell surface.

Efficiency of Mature (a6b2) 2 b3 AChR Assembly
Efficiency of assembly of mature AChR was analyzed by sedimentation velocity analysis on sucrose gradients ( Figure 5). Confirming previous observations [16], the pentameric concatamer construct 1 exhibited a high proportion of mature AChRs of the expected size, intermediate between monomers and dimers of Torpedo AChRs. Construct 2 (a6+b2+b3) resulted in high proportions of partially assembled AChRs and large aggregates, but virtually no mature AChRs. This confirms previous observations that free wild type subunits do not assemble into mature AChRs in Xenopus oocytes [12]. Construct 3 (a6 211L +b2+b3) resulted in a substantial proportion of mature AChRs. Thus, replacing the methionine unique to a6 in transmembrane domain 1 with the leucine found in a3 and most other a subunits greatly promoted assembly of mature AChRs. Construct 4 (a6 a3cyt +b2+ b3) resulted in mature AChRs, but also some partially assembled AChRs and a substantial proportion of large aggregates. Figure 2 showed that the absolute amounts of ACh binding sites assembled, AChRs transported to the surface, and functional AChRs were greater with the a6 chimera incorporating only the large cytoplasmic domain of a3 (construct 4) than with the chimera incorporating only leucine 211 of a3 (construct 3). Combining the two chimeras in an a6 subunit in construct 5 (a6 211L,a3cyt +b2+b3) gave a synergistic effect. The proportion of mature AChRs was large, and partially assembled AChRs and the largest aggregates were eliminated ( Figure 5E). In addition, the absolute amount of AChRs assembled and transported to the surface membrane where their function was assayed increased 13-80 fold compared to when the chimeras were expressed individually ( Figure 2). Construct 6 (a6 a3cyt-C +b2+b3) produced a high proportion of mature AChRs ( Figure 5F). Thus, the a3 transmembrane domain 4 and/or the a3 C-terminal extracellular domain contributed to assembly of an increased proportion of mature AChRs and a decreased proportion of large aggregates and assembly intermediates compared to the a3 cytoplasmic domain alone ( Figure 5D). However, the absolute amount of functional AChRs on the cell surface was lower than with only the a3 cytoplasmic domain in construct 4 ( Figure 2B). Constructs containing b32a6 concatamers in constructs 7, 8, and 9 exhibited substantial proportions of mature AChR ( Figure S1), although the absolute amounts of AChRs assembled or functional AChRs on the cell surface were much smaller than achieved with construct 5 containing the a6 double chimera, b2, and b3 as free subunits.
Most constructs exhibited similar high sensitivities to activation by agonists ( Figure 6, Table 1). Constructs 2 and 3 showed very little activation by acetylcholine, with maximum responses less than 20 nA. Thus their pharmacology was not studied in detail. EC 50 of acetylcholine for activating all constructs, except 9, is around 1 mM. Kinetics of activation by ACh are displayed in Figure 7 and Figure S2. Constructs exhibited similar kinetics of activation and desensitization by various concentrations of ACh. For example, construct 5 (a6 211L,a3cyt +b2+b3) responded to 30 mM ACh with 10 fold greater amplitude than construct 1 (b32a62b22a62b2) but had similar response kinetics. There were noticeable concentration-dependent shifts in peak times in Figure 3. Effect of various constructs on channel properties. Evoked current was divided by the number of AChRs on the oocyte surface (i.e., mAb 295 surface binding) for each construct presented in Figure 2B. Values shown are a measure of effects on probability of channel opening and/or channel conductance. doi:10.1371/journal.pone.0103244.g003 constructs 5, 6 and 7 (Figure 7 and Figure S2). The response persists after the ACh application period due to design of our electrophysiology testing apparatus, permitting washout to be slower than wash in. This effect might also result from calciumactivated chloride channels that remain open longer than the AChRs channels. Theoretically this effect could be avoided by omitting Ca ++ from outside solution or clamping at a lower voltage. However, Ca ++ may potentiate activation of AChRs by agonists and a6* AChR evoked peak current is strongly influenced by extracellular Ca ++ [28,29]. Considering the overall low functional responses of the nine constructs, experiments were all executed at a holding potential of 270 mV and in a buffer with normal concentration of Ca ++ . Finally, altered kinetics of agonist responses were only exhibited by constructs that were not especially effective, thus characterizing them in detail is tangential to our goal of effectively expressing (a6b2) 2 b3 AChRs.
Besides difference in activation, construct 6 (a6 a3cyt-C +b2+b3) also exhibits more rapid desensitization than the other constructs ( Table 2 and Figure S3). The desensitization rate of construct 6 increased 48% compared to that of construct 1. Construct 5, which has both M211L and the a3 large cytoplasmic domain, desensitized at a rate close to construct 1 (Table 2). Thus, a3 transmembrane domain 4 and/or the a3 C-terminal extracellular domain increase the rate of desensitization.

Discussion
It is hard to study a single AChR subtype in vivo because various subtypes often co-express together in brain. Establishing in vitro model systems expressing a6-containing AChRs is challenging. By changing a single unique amino acid in transmembrane domain 1 of a6 and the cytoplasmic domain of a6 to that of a3, we succeeded in efficiently expressing in Xenopus oocytes large amounts of human (a6b2) 2 b3 AChRs. This will be very useful for characterizing functional properties of these AChRs and drugs directed at them.
These AChRs have the pharmacological properties of (a6b2) 2 b3 AChRs formed from wild type subunits linked in a concatamer. As discussed in our previous study, concatameric (a6b2) 2 b3 AChRs showed the pharmacology expected from measuring dopamine release in the brain tissue [16]. EC 50 values for nicotine obtained in vivo ranged from 0.1 mM to 1 mM for activation of a6* AChRs not containing a4 [5,21,30]. Such variation may be due to technical variations. The pentameric concatamer and AChRs expressed from the free subunits of our highest expressing free subunit construct 5 have very similar pharmacological properties, (e.g. EC 50 for nicotine = 0.239 or 0.203 mM) but most of the constructs exhibit similar properties.
Expression from these modified free subunits is much more efficient than expression from (a6b2) 2 b3 concatamers [16]. Concatamers will still be required to ensure efficient expression and subunit order of the more complex (a6b2)(a4b2)b3 AChR subtype [16]. Dopaminergic neurons which uniquely express a6b2b3* AChRs may have chaperones for efficiently assembling b3 accessory subunits with a6 and ensuring efficient assembly of (a6b2) ACh binding sites in AChRs with (a4b2) ACh binding sites [12,13]. In cells expressing free a4, a6, and b2 subunits, a6 and b2 do not effectively assemble to form (a6b2)(a4b2)* AChRs, instead a4b2 AChRs predominate (Kuryatov unpublished).  . Assembly of (a6b2) 2 b3 AChRs constructs evaluated by sucrose sedimentation velocity gradient analysis. After centrifugation, gradient fractions were immunoisolated on microwells coated with mAb 295 to b2 to isolate a6b2b3 AChRs prior to labeling with 3 H epibatidine. Properly assembled mature (a6b2) 2 b3 AChRs sediment between the two internal standards, 9S monomer and 13S dimer of Torpedo californica AChRs. Peaks on the left of dimers in the gradient indicate multimers or aggregates of AChRs, while peaks on the right of monomers represent partially assembled AChRs. A) Expression of pentameric concatamer construct 1 (b32a62b22a62b2) resulted in a high proportion of mature AChRs, as expected (16). B) Expression of construct 2 (a6+b2+b3) resulted in a high proportion of aggregates and partially assembled AChRs and a very low proportion of mature AChRs, as expected (12). C) Expression of construct 3 (a6 211L +b2+b3) resulted in a large proportion of mature AChRs. D) Expression of construct 4 (a6 a3cyt +b2+b3) resulted in mature AChRs but also partially assembled AChRs and both large and very large aggregates. E) Expression of construct 5 (a6 211L,a3cyt +b2+b3) showed mature AChRs and some aggregates. F) Expresion of construct 6 (a6 a3cyt-C +b2+b3) showed a high proportion of mature AChRs and few aggregates. doi:10.1371/journal.pone.0103244.g005 Figure 6. Concentration/response curves for constructs that resulted in significant amounts of functional AChRs. Full agonist (ACh) and partial agonists (cytisine, nicotine, and varenicline) were used on (a6b2) 2 b3 AChRs. Each point is the average response of at least 5 oocytes. Arrows indicate construct 5, that behaves like construct 1, and constructs 6 and 9 that are divergent. doi:10.1371/journal.pone.0103244.g006 Chimeras of a6 subunits with a3 subunits have revealed sequences of a6 that inhibit expression of a3b2 AChRs and sequences of a3 that permit expression of a6b2 AChRs [12,16]. Studies reported here build on this foundation to achieve efficient expression of (a6b2) 2 b3 AChRs and further explain structural limits to a6* AChR expression.
The unique methionine 211 in the first transmembrane domain of a6, when exchanged for the leucine present in most other a subunits, promotes efficient assembly with b3 and formation of a  high proportion of mature AChRs. This suggests that a unique chaperone in dopaminergic neurons might bind to this region of a6 to promote assembly with b3. No chaperone is needed to get efficient assembly of b3 with a4 subunits that have a leucine at this position [22]. Replacing leucine 211 of a3 with methionine inhibits assembly of a3b2 AChRs [16]. The concatameric construct 1 (b32a62b22a62b2) efficiently assembles into mature AChRs, but these are not efficiently transported to the cell surface, as are the concatamers b32a62 b22a42b2 or b32a42b22a62b2 [16]. This suggests that the large cytoplasmic domain of a4 may contribute to transport to the cell surface. The cytoplasmic domain of a4 is uniquely large. Here we show the normal sized cytoplasmic domain of a3, a subunit closely related to a6 in sequence, when substituted for that of a6 can promote both transport to the cell surface and increase assembly with b2 subunits. It is known that large cytoplasmic domain chimeras can increase the assembly and transport of AChRs [17]. Incorporation of shorter sequences of a3 cytoplasmic domain will be required to determine the minimum sequences required to promote assembly with b2 or surface transport. Use of shorter a3 cytoplasmic domain sequences might avoid incorporation of the a3 amphipathic a helix which could alter cation selectivity [17,31]. Identification of these shorter sequences may further suggest their mechanisms of action in transport and assembly and help to explain their collaboration with the single amino acid change M211L to promote expression of human (a6b2) 2 b3 AChRs from free subunits in Xenopus oocytes.
The synergism of these two chimeras appears to result from 211L promoting a high proportion (but low amount) of assembly of b3 combined with the cytoplasmic domain of a3 promoting extensive assembly with b2, and to a lesser extent b3, and facilitating transport of assembled AChRs to the oocyte surface. Linking b3 to a6 in a concatamer can force their assembly [16]. However, the dimeric concatamer used here did not exceed the effect of combining 211L and a3 cytoplasmic domain in generating functional (a6b2) 2 b3 AChRs.
The a6 chimeras with 211L and the a3 large cytoplasmic domain (construct 5) effectively expressed functional (a6b2) 2 b3 AChRs without altering their pharmacological properties. Transfection of HEK 293 cells with a 2:1:10 ratio of a chimera with the extracellular domain of a6 and the rest of a3, b2, and b3 with and a V9'S gain of function mutation resulted in functional (a6b2) 2 b3 AChRs [32]. Despite the extensive modifications necessary in those studies to assemble b3 and get a6 to the surface of the cell line, some of the pharmacological properties of its AChRs were similar to constructs 1 and 5, while others were divergent. On the cell line, EC 50 values for ACh (0.23 mM), nicotine (0.15 mM), cytisine (0.072 mM), and varenicline (0.022 mM) were lower than we observed by 1.4 to 5.9 fold, as might be expected as a result of their using a hyperactive mutant b3 subunit. Similarly, the efficacy for nicotine they observed was two fold greater than we observed for constructs 1 and 5.
(TIF) Figure S4 Stability of concatamer in construct 8 (b32 a6+a6 211L , a3cyt +b2) was confirmed by western blot. AChR solubilized from 100 oocytes solubilized by Triton X-100 was purified and concentrated by immunoaffinity chromatography using mAb 295 linked to resin before being resolved to subunits by SDS-polyacrylamide gel electrophoresis. After overnight incuba-tion at 4uC, the column was spun at 5,000 rpm for 15 minutes to remove unbound material and washed with PBS, 0.5% Triton solution. 40 ml of LDS sample buffer (Invitrogen) was placed in the column and heated for 30 minutes at 37uC. The eluent was resolved by SDS-polyacrylamide gel electrophoresis and then transferred using a semidry electroblotting method [16]. Blots were then quenched with 5% Carnation dried nonfat milk for in PBS, 0.5% Triton X-100, 10 mM NaN 3 for one hour. Blots were probed with rat antiserum to a6 (1:500) [13], then incubated with 2 nM 125 I-labeled goat anti-rat IgG for 3 h at room temperature. After washing in 0.5% Triton with NaN 3 , blots were visualized by autoradiography. Proteins of the sizes expected of QAP linker