Delineating the Site of Interaction of the 5-HT_3A Receptor with the Chaperone Protein RIC-3

Abstract

The serotonin type 3A (5-HT_3A) receptor is a homopentameric cation-selective member of the pentameric ligand-gated ion channel (pLGIC) superfamily. Members of this superfamily assemble from five subunits, each of which consists of three domains: extracellular (ECD), transmembrane (TMD), and intracellular domain (ICD). Previously, we have demonstrated that the 5-HT_3A-ICD is required for the interaction between 5-HT_3A and the chaperone protein resistance to inhibitors of choline esterase (RIC-3). Additionally, we have shown that 5-HT_3A-ICD fused to maltose-binding protein (MBP) directly interacts with RIC-3, without the involvement of other protein(s). To elucidate the molecular determinants of this interaction, we developed different MBP-fused 5-HT_3A-ICD constructs by deleting large segments of its amino acid sequence. We expressed seven engineered ICDs in Escherichia coli and purified them to homogeneity. Using a RIC-3 affinity pull-down assay, the interaction between MBP-5HT_3A-ICD constructs and RIC-3 was investigated. In summary, we identify a 24-amino-acid-long segment of the 5-HT_3A-ICD as a molecular determinant for the interaction between the 5-HT_3A-ICD and RIC-3.

Significance

The chaperone protein RIC-3 is known to modulate the functional surface expression of cation-conducting pentameric ligand-gated ion channels. Previously, we have demonstrated that the intracellular domain of serotonin channels mediates this effect. Here, we provide experimental evidence for a 24-amino-acid-long segment within the 115-amino-acid-long intracellular domain as a determinant for RIC-3 interaction. Recently, it was found experimentally that the identified segment contains an α-helix that has been observed or predicted to be present in other cation-conducting channels. This work provides, to our knowledge, novel insights into protein-protein interactions that are likely also relevant for other cation-conducting members of this large ion channel family that includes nACh and 5-HT₃ receptors.

Introduction

Serotonin type 3 (5-HT₃) receptors, like the nicotinic acetylcholine (nACh), ɣ-aminobutyric acid (GABA) type A, and glycine receptors, are members of the pentameric ligand-gated ion channel (pLGIC) superfamily. Ion channels of this cysteine-loop superfamily are homo- or heteropentamers. 5-HT₃ receptors are composed of five identified subunits (A–E) and are involved in excitatory neurotransmission (1). These receptors are abundant in the central nervous system (CNS) and peripheral nervous system (PNS) and are found pre- and postsynaptically (2). Among all subunits of 5-HT₃ receptors, the 5-HT_3A subunit is widely distributed in the CNS and PNS and is also found in nonneural cells and tissues across the human body (3). Whereas the 5-HT_3B subunit is less abundant, the messenger RNA (mRNA) of this subunit has been detected in several regions of the human brain, intestines, and kidney (4). The mRNAs of the remaining 5-HT₃ subunits (C–E) have been found in the PNS and CNS, organs, and extraneuronal cells (5,6). Currently, these receptors are mainly targeted for the treatment of vomiting in patients undergoing chemotherapy or in patients suffering from irritable bowel syndrome (5,7). Recent studies have shown potential correlations between 5-HT₃ receptor activity and several neurological disorders such as anxiety, psychosis, nociception, bipolar disorder, schizophrenia, and cognitive function (8, 9, 10, 11, 12).

Each subunit of a pLGIC pentamer consists of three domains, namely, the extracellular domain (ECD), transmembrane domain (TMD) with four membrane-spanning helices (M1–M4), and intracellular domain (ICD), which is present only in eukaryotic pLGICs (13). The assembly, functional maturation, and membrane trafficking of pLGIC subunits are modulated by different chaperone proteins depending on their subtype. The protein resistance to inhibitors of cholinesterase (RIC-3) is known as a modulator of functional surface expression of nAChRs as well as 5-HT_3ARs (14,15). RIC-3 was originally identified in Caenorhabditis elegans, where it was shown to be necessary for assembly and trafficking of DEG-3 acetylcholine receptors but not for GABA receptors (16). The same study observed increased acetylcholine receptor activity after C. elegans RIC-3 coexpression with C. elegans DEG-3 and rat nAChR α7 in Xenopus laevis oocytes, whereas a RIC-3 effect was not observed for glutamate channels (GluR₃). The study noted the remarkable protein-specific effect of RIC-3 and postulated that RIC-3 may be involved in a mechanism of regulating the level of specifically one class of receptors while not affecting others or an involvement in stoichiometric assembly. The coexpression of human RIC-3 (hRIC-3) with heteromeric human nAChR α3β4 and α4β2 as well as mouse 5-HT_3AR in X. laevis oocytes inhibits the functional maturation of these receptors, whereas human RIC-3 enhances human nAChR α7 expression (17). Later, it was shown that additional host cell factors may contribute to the activity of RIC-3 as a chaperone based on the discrepancy in potentiating or inhibiting effects of RIC-3 on pLGICs from different species in various host cells (18, 19, 20, 21). A further complication arises from time-dependent as well as transfection-ratio-dependent effects (20). In summary, although RIC-3 coexpression may have potentiating or inhibiting effects on nAChR and 5-HT₃ channels (the most prominent cation-conducting pLGICs), RIC-3 has never been shown to modulate anion-conducting pLGICs (16,22).

Previously, we have demonstrated that the ICD of 5-HT_3A receptors (5-HT_3A-ICD) is required for receptor modulation by hRIC-3 because the removal of the ICD eliminated the inhibitory effects of hRIC-3 coexpression on 5-HT_3A functional surface expression in X. laevis oocytes (23). Further, we designed chimeras in which we added the 5-HT_3A-ICD of 115 amino acids to the prokaryotic homolog of 5-HT_3A, the Gloeobacter violaceus ligand-gated ion channel (GLIC), to substitute the GLIC heptapeptide linker between M3 and M4 transmembrane helices. This addition led to the observation of modulatory effects on expression of GLIC-5-HT_3A-ICD chimeras when coexpressed with hRIC-3, whereas wild-type GLIC was insensitive to coexpression (22,24). Additionally, we demonstrated that the interaction between GLIC-5-HT_3A-ICD and hRIC-3 is direct and does not require the presence of additional proteins. Both proteins were overexpressed in and purified from Escherichia coli and were subjected to a pull-down assay through which we observed a direct interaction between the two proteins (25). Subsequently, in a separate study, we examined if the 5-HT_3A-ICD alone is sufficient to retain an interaction with hRIC-3. We used a chimera obtained by fusing the 5-HT_3A-ICD to the C-terminus of maltose-binding protein (MBP). The addition of MBP to the 5-HT_3A-ICD facilitated the expression, purification, and stability of this domain. We determined that the purified 5-HT_3A-ICD maintained an interaction with hRIC-3 (26). In summary, previous studies have demonstrated that the 5-HT_3A-ICD is required and sufficient for interaction of the 5-HT_3AR and hRIC-3. Furthermore, this direct interaction between the ICD and hRIC-3 does not require the presence of other host cell proteins (22, 23, 24, 25, 26).

In the current study, we aimed to identify the segment of the ICD responsible for this interaction with the hypothesis that the interaction site lies within an independent segment of the ICD. The complete ECD and TMD of mouse 5-HT_3AR were resolved by x-ray crystallography, whereas part of the ICD (62 amino acids) was missing because of trypsin treatment of the receptor before crystallization. This unresolved segment of the ICD, here named L2, is located between the MX helix that follows a short post-M3 loop (L1) and the MA helix that continues into M4 (27). Recently, structures of full-length 5-HT_3ARs obtained by single-particle cryoelectron microscopy have been published. However, almost the entirety of the L2 segment remained unresolved (28,29). For the purpose of this study, we used the x-ray structure as a guide to design our constructs. We used a process of elimination by dividing the ICD into its known structural elements (L1-MX, L2, and MA) (Fig. 1) and determined if hRIC-3 interacted with the different segments. Here, we demonstrate that the L1-MX region within the 5-HT_3A-ICD interacts with hRIC-3. Our results identify a stretch of 24 amino acids within the 115-amino-acid-long 5-HT_3A-ICD as the interaction site with hRIC-3.

Figure 1.

Structural representation and alignments of constructs.

For a Figure360 author presentation of this figure, see https://doi.org/10.1016/j.bpj.2019.11.3380.

(A) A cartoon representation of the pentameric m5-HT_3AR x-ray structure viewed parallel to the plane of the membrane is shown. The ECD is in green, TMD is in purple, MA helix is in violet, and the MX helix is in cyan. (B) A single subunit of the 5-HT_3A receptor viewed parallel to the membrane is shown. (C) A cartoon representation of MBP-5HT_3A-ICD-WT is shown. The cytoplasmic domain of 5-HT_3AR is fused to the C-terminus of MBP (black structure) and serves as the template for respective constructs. The relative orientation of MBP with regard to the ICD is entirely hypothetical and only meant for overall illustration purposes. The 62 residues removed from L2 (charcoal) are shown as a dashed line. (D) The MBP-MA has the N-terminus of the MA helix attached to MBP. (E) The MBP-ΔMA has the entire MA-helix removed. (F) MBP-Δ34, Δ44, and Δ55 constructs were obtained by deleting 34 (red), 44 (dark green), and 55 (dark blue) amino acids from L2, respectively. (G) The MBP-L2 contains only the unresolved segment of the ICD attached to the C-terminus of MBP. (H) The MBP-L1-MX contains only the short loop and the helical structure post the M3 transmembrane segment. (I) Shown are the multiple sequence alignment of 5-HT_3AR and the ICD constructs highlighting deletions (dashed lines).

Materials and Methods

Materials

The following materials were used: BL21-CodonPlus-(DE3)-RIPL cells (Agilent Technologies, Santa Clara, CA); ampicillin (Thermo Fisher Scientific, Fair Lawn, NJ); chloramphenicol (Thermo Fisher Scientific); IPTG (Fisher Scientific, Fair Lawn, NJ); leupeptin (AdipoGen Life Sciences, San Diego, CA); pepstatin (AdipoGen Life Sciences); PMSF (Research Products International, Mt. Prospect, IL); TCEP-HCl (Oakwood Chemical, Estill, SC); lysozyme (MP Biomedicals, Solon, OH); protease inhibitor cocktail III (Research Products International); DNase I (Alfa Aesar, Ward Hill, MA); DDM (Anatrace, Maumee, OH).

Molecular biology

The cytoplasmic domain of the mouse 5-HT_3A receptor (GenBank: AAT37716.1, QDLQ… RVGY, 115 amino acids) fused to the C-terminus of MBP in pMALX vector (New England Biolabs, Ipswich, MA) (MBP-5-HT_3A-ICD-pMALX) was used as the template to design constructs for this study (26,30). The fusion constructs were generated by deletion of amino acids from the 5-HT_3A-ICD using appropriate partially overlapping primers with the QuikChange II Site-Directed Mutagenesis kit (Agilent Technologies) and were confirmed by DNA sequencing (GENEWIZ, South Plainfield, NJ) (31).

Human RIC-3 (GenBank: NM_001206671.2) in the pGEMH19 (hRIC-3-pGEMH19) expression vector was generously provided by Dr. Millet Treinin (Hebrew University, Israel). hRIC-3 was cloned into the prokaryotic expression vector pMAL-c2x fused to N-terminal MBP and C-terminal His₆ tag (25,32,33). The MBP-hRIC-3-His₆-pMAL-c2x and hRIC-3-pHEMH19 constructs were used in pull-down and X. laevis oocyte interaction assays, respectively.

For X. laevis oocyte interaction assays, the 5-HT_3A subunit containing the V5 epitope tag (GKPIPNPLLGLDSTQ) close to the N-terminus and the 5-HT_3B subunit were engineered into the expression vector pGEMHE and were used for full receptor expression in oocytes (23). These complementary DNAs along with complementary DNA of hRIC-3-pHEMH19 were linearized with the restriction enzyme NheI (New England Biolabs). Subsequently, complementary RNA (cRNA) of each construct was prepared using T7 RNA polymerase (mMESSAGE mMACHINE T7 Kit; Applied Biosystems/Ambion, Austin, TX) in an in vitro transcription process. The cRNAs were then purified and precipitated with MEGAclear kit (Applied Biosystems/Ambion) and dissolved in nuclease-free water.

Oocytes

Oocytes for interaction assays were harvested from X. laevis frogs and defolliculated in house. Maintenance and surgery procedures were approved by the Texas Tech University Health Sciences Center animal welfare committee. Before injection with an automatic oocyte injector (Nanoject II; Drummond Scientific, Broomall, PA), the oocytes were washed using Ringer’s buffer (OR2: 82.5 mM NaCl, 2 mM KCl, 1 mM MgCl₂, 5 mM HEPES (pH 7.5)). Injected oocytes were maintained in standard oocyte saline medium (SOS: 100 mM NaCl, 2 mM KCl, 1 mM MgCl₂, 1.8 mM CaCl₂, 5 mM HEPES (pH 7.5)) supplemented with a 1% antibiotic-antimycotic (100×, 10,000 units/mL of penicillin, 10,000 μg/mL of streptomycin, and 25 μg/mL of Gibco amphotericin B (Gaithersburg, MD)) and 5% horse serum (Sigma-Aldrich, St. Louis, MO) at 15°C.

SDS-PAGE

For sodium dodecyl sulfate (SDS) electrophoresis, 4–15% Mini-PROTEAN TGX stain-free gels (Bio-Rad Laboratories, Hercules, CA) were used and visualized by stain-free-enabled imager (Gel Doc EZ Imager; Bio-Rad Laboratories). The stain-free gels contain trihalomethane compounds that upon ultraviolet irradiation covalently modify tryptophan residues that can be subsequently visualized by their fluorescence.

Protein expression and purification

MBP-5-HT_3A-ICD-pMALX fusion proteins

The MBP-5-HT_3A-ICD-pMALX soluble fusion proteins were overexpressed in E. coli BL21-CodonPlus (DE3)-RIPL cells. The cells were harvested by centrifugation (5000 × g for 15 min at 4°C) and lysed in a two-step process using freshly prepared buffer (buffer A: 20 mM Tris, 150 mM NaCl, 1 mM EDTA, 1 mM tris(2-carboxyethyl)phosphine (TCEP) (pH 7.4)). First, the cell pellets were resuspended by stirring for 2 h at 4°C using buffer A containing added protease inhibitors (1 mM phenylmethylsulfonyl fluoride (PMSF); 10 μg/mL leupeptin (Research Products International); 10 μg/mL pepstatin (Sigma-Aldrich); Thermo Fisher Scientific) and 1 mg/mL lysozyme (Sigma-Aldrich). The cell lysate solution was frozen at −80°C overnight. Next, 50 μg/mL of DNase I (Sigma-Aldrich) and 1 mM PMSF were added to the thawed cell lysate solution, and the mixture was stirred for 2 h at 4°C. The cell suspension was passed through a 15 G needle several times to create a homogeneous solution. A clear soluble fraction of the lysate was obtained by centrifugation at 30,000 × g for 1 h at 4°C. The supernatant was passed through a 0.2 μm pore-size bottle-top filter and then loaded onto a gravity-packed amylose-resin column. The unbound proteins were washed extensively using buffer A (3 × 10 bed volume), and the bound proteins were eluted by buffer A (pH 7.4) containing 20 mM maltose.

MBP-hRIC-3-His₆

The MBP-hRIC-3-His₆-pMAL-c2x membrane protein was overexpressed in E. coli BL21-CodonPlus (DE3)-RIPL cells. The cells were harvested by centrifugation (5000 × g for 15 min at 4°C) and lysed in a two-step process similar to above using freshly prepared buffer (10 mL/g of cell pellet buffer A′: 20 mM Tris, 100 mM NaCl, 2 mM EDTA, 10 mM DTT, 10% glycerol (v/v) (pH 7.4)). After the lysis, the debris and unlysed cells were separated from the cell lysate solution by centrifugation (10,000 × g for 20 min at 4°C). Next, cell membranes were collected by centrifugation of supernatant at a higher speed (100,000 × g for 1 h at 4°C). The membranes were resuspended in solubilization buffer (5 mL/g of initial cell pellet buffer B: 20 mM Tris, 100 mM NaCl, 10% glycerol (v/v) (pH 7.4)). An equal amount of buffer B containing 1% (w/v) n-dodecyl β-D-maltoside (DDM) was added dropwise at the rate of 1 mL/min to the resuspended membrane solution while slowly stirring at 4°C. The solubilized cell membranes were further cleared by centrifugation (100,000 × g for 1 h at 4°C) and the supernatant was loaded onto a gravity-packed amylose-resin column. The resin was washed extensively with buffer B containing 0.05% DDM (3 × 10 bed volumes), and bound proteins were eluted with a solution of 20 mM maltose and 0.05% DDM in buffer B. Eluted proteins were concentrated to 2 mL using 100 kDa Amicon Ultra-15 centrifugal filter units (MilliporeSigma, Burlington, MA) and subjected to a batch of Talon Superflow (MilliporeSigma) resin incubation overnight at 4°C. The resin beads were preequilibrated before adding the protein solution using cold buffer (buffer C: 20 mM Tris, 500 mM NaCl, 10% glycerol (v/v), 0.05% DDM (w/v), 0.5% Triton X-100 (v/v) (pH 7.5)) containing 10 mM imidazole. After overnight incubation, the protein-resin mixture was transferred to a gravity column and buffer removed by gravity flow at a rate of 2 mL/min. The column was washed with buffer C containing 10 mM imidazole (3 × 5 bed volumes). Next, the protein-bound resin bed was washed with a Triton X-100 free buffer (buffer D: 20 mM tris, 500 mM NaCl, 1 mM TCEP, 10% glycerol (v/v), 0.05% DDM (w/v) (pH 7.5)) containing 15 mM imidazole (3 × 5 bed volumes). The bound protein was then eluted with 10 mL of 300 mM imidazole in buffer D. The buffer was exchanged immediately after elution to reduce the imidazole concentration using a 100 kDa Amicon centrifugal filter unit.

SEC

The amylose/Talon-column-purified proteins were subjected to size-exclusion chromatography (SEC) for final purification on a Superdex 200 10/300 GL column (Bio-Rad Laboratories). The column was equilibrated with SEC buffer (buffer S: 20 mM Tris, 150 mM, NaCl, 1 mM TCEP, 5 mM maltose, 0.01% NaN₃ (w/v) (pH 7.4) for MBP fusion constructs; buffer S′: 20 mM tris, 500 mM NaCl, 1 mM TCEP, 10% glycerol (v/v), 0.05% DDM (w/v) (pH 7.5) for MBP-hRIC-3-His₆). Ultraviolet absorbance at 280 nm (A₂₈₀) was utilized to monitor the elution of the proteins. Fractions were analyzed after separation on 4–15% Mini-PROTEAN TGX stain-free gels. Standard proteins (thyroglobulin 669 kDa, ferritin 440 kDa, aldolase 158 kDa, conalbumin 75 kDa, and ovalbumin 44 kDa) were used according to the instruction manual (Millipore) to determine a standard curve, and the void volume (V_o) was determined with Blue Dextran 2000. The gel-phase distribution coefficient (K_av) is graphed versus log molecular weight (log M_r). The gel-phase distribution coefficient is calculated by the equation K_av = (V_e − V_o)/(V_c − V_o), where V_e is elution volume, V_c is geometric column volume, and V_o = column void volume.

Pull-down assay

In this assay, 0.05% DDM and 1× protease inhibitor cocktail (Research Products International) were added to all buffers just before use in each step. First, 2.5 μg of freshly SEC-purified MBP-hRIC-3-His₆ (bait) was added to 15 μL of HisPur cobalt resin (Thermo Fisher Scientific) equilibrated with and resuspended in 200 μL of binding buffer (buffer A: 20 mM Tris, 150 mM NaCl, 1 mM MgCl2, 0.05% Triton X-100, 10 mM imidazole (pH 7.4)) inside a Pierce cellulose acetate filter spin cup (Thermo Fisher Scientific). After 30 min of incubation at 4°C, unbound MBP-hRIC-3-His₆ was washed away with 300 μL of wash buffer (buffer B: 20 mM Tris, 150 mM NaCl, 1 mM MgCl2, 0.05% Triton X-100, 15 mM imidazole (pH 7.4)) and centrifugation at 1000 × g for 30 s. This step was repeated five times, and the resin was resuspended with 200 μL of buffer B. Next, the prey proteins (MBP-5-HT_3A-ICD-pMALX constructs) were added to each individual spin cup $\left((\text{pMole of MBP}-\text{RIC}-3-\text{His}6)/\left(\text{pMole of prey protein}\right)=1/14\right)$, and mixtures were incubated at 4°C for 2 h. Additionally, similar conditions were prepared using only the prey proteins with 15 μL of cobalt resin to serve as controls. Unbound proteins were washed 12 times with 300 μL of buffer B and then eluted with 25 μL of elution buffer (buffer C: 20 mM Tris, 150 mM NaCl, 1 mM MgCl₂, 0.05% Triton X-100, 200 mM imidazole (pH 7.4)) and centrifugation at 1200 × g for 2 min. The eluted samples were immediately analyzed by SDS polyacrylamide gel electrophoresis (SDS-PAGE) using 4–15% Mini-PROTEAN TGX stain-free gels.

X. laevis oocyte interaction assay

The X. laevis oocyte interaction assay was designed to probe the modulation of functional surface expression of 5-HT_3A and serotonin type 3AB (5-HT_3AB) receptors by hRIC-3. 1.5 ng of hRIC-3 in the pGEMH19 cRNA was coinjected with 6 ng of 5-HT_3A-wt cRNA and 6 ng of 5-HT_3AB-cRNA (3:1 ratio of A/B) into X. laevis oocytes, respectively. Oocytes injected with 5-HT_3A-wt cRNA or 5-HT_3AB cRNA alone were used as controls. Current amplitudes induced by application of 10 μM of 5-HT in OR2 were measured 48 h after injection using two-electrode voltage-clamp (TEVC) recordings. Currents were recorded from individual oocytes perfused with OR2 at a holding potential of −40 mV. They were amplified using a TEV-200A amplifier (Dagan, Minneapolis, MN), digitized using a Digidata 1440A analog-to-digital converter (Molecular Devices, Sunnyvale, CA), and analyzed with pCLAMP (Clampex/Clampfit) software (Molecular Devices).

Statistics

Statistical significance was determined with one-way analysis of variance and Tukey’s multiple comparisons test (^∗∗∗∗ denotes p < 0.001) of oocytes injected with 5-HT_3A or 5-HT_3AB and hRIC-3 vs. 5-HT_3A or 5-HT_3AB alone (Prism 6; GraphPad Software, San Diego, CA).

Results

Construct design

To further investigate the interaction between the 5-HT_3AR and hRIC-3 and determine the interaction site within the 5-HT_3A-ICD, we designed a series of complementary deletion constructs of the 5-HT_3A-ICD in a stepwise manner (Fig. 1). Initially, we engineered two deletion constructs based on structural elements observed within the 5-HT_3A-ICD (Fig. 1 C) in the x-ray structure of the mouse 5-HT_3A receptor (27). We hypothesized that the interaction with hRIC-3 involves specific and likely independent segments of the 5-HT_3A-ICD. The x-ray structure of the 5-HT_3A contains a partially resolved ICD with three distinct segments (Fig. 1 A): the MX helix following a short loop (L1) post the M3 segment of the TMD; the unresolved loop in between the two MX and MA helices (L2), and the MA helix, the 31-amino-acid-long helical structure preceding the M4 segment of the TMD. The first construct was obtained by deletion of L1-MX-L2 leaving only the MA helix, MA (GLLQ…RVGY) (Fig. 1 D). To obtain the second construct, we deleted the MA helix from the 5-HT_3A-ICD, yielding ΔMA (QDLQ…LAVR) (Fig. 1 E). Three additional deletion constructs were designed comparable to a previous study investigating the functional impact of deletions of the ICD (34) (Fig. 1 F): Δ34 was generated by deletion of 34 amino acids from the C-terminus of L2 (ΔA358–L392); Δ44 was generated by deletion of 44 amino acids from the C-terminus of L2 (ΔF348–L392); Δ55 was generated by deletion of 55 amino acids from the C-terminus of L2 (ΔE337–L392); L2 was generated by deletion of L1-MX and MA (LGEQ…LAVR) (Fig. 1 G); and L1-MX was generated by deletion of L2 and MA (QDLQ…WILC) (Fig. 1 H). All constructs were fused to MBP at their N-terminus, forming soluble constructs.

Protein expression and purification

Next, we overexpressed the constructs in E. coli and purified the cell lysates utilizing an amylose-resin affinity column for the first step of purification. The eluted fractions alongside with loading material, flow-through, and washes were then analyzed using SDS-PAGE (Fig. 2 A). Whereas many impurities were observed in lysate material, affinity purification yielded significant enrichment of ICD constructs with more than 90% purity observed upon protein visualization. Eluate fractions from affinity chromatography with the most protein were then subjected to SEC for further purification, and purified protein fractions were visualized by SDS-PAGE (Fig. 2 B). SEC chromatograms are shown in Fig. S1. For some constructs, lower molecular weight bands are observed for some fractions that mainly represent MBP. For subsequent experiments, we therefore used only fractions without significant impurities. Some of the constructs have been previously used in a study that investigated the oligomerization motif that leads to pentameric assembly of full-length 5-HT_3A-ICD and identified a structural role for three conductance-limiting arginines within the MA helix and an aspartate within L1 (35). We have observed that the lack of the MX helix in both the MA and L2 constructs resulted in lower yield and higher degradation, which may suggest that MX plays a role in the stability and expression of the ICD. It is important to mention that the deletions did not cause any significant changes in pI of the proteins; therefore, the same expression and purification conditions were used for all MBP-5-HT_3A-ICD constructs.

Figure 2.

Two-step purification of the ICD constructs. (A) Proteins were purified by amylose-resin affinity column chromatography. The soluble protein fraction was loaded onto columns and was passed through via gravity flow, followed by washes (W) and eluted in fractions (Elution). Stain-free SDS-TGX gels indicate the purity of eluted protein fractions, identify peak fractions, and determine the weight of denatured monomeric ICD constructs. Note that different numbers of elution fractions were run for individual gels. From top: ICD (53 kDa), MA (45 kDa), ΔMA (49 kDa), Δ34 (50 kDa), Δ44 (48 kDa), Δ55 (46 kDa), L2 (45 kDa), and L1-MX (42 kDa) are shown. A standard protein ladder is shown on the left for each gel. (B) The second purification step was size-exclusion chromatography (SEC). Peak fractions were separated on stain-free SDS-TGX gels to assess purity.

We additionally overexpressed MBP-hRIC-3-His₆ in E. coli and significantly optimized our previous procedure (25) for the lysis and purification of this membrane protein to obtain a homogenous and stable protein with appropriate yield for our direct protein-protein interaction studies. hRIC-3 was purified using two sequential affinity purification steps with amylose and Talon Superflow resins and additionally SEC. The solubilized protein is observed at 100 kDa after SDS-PAGE (theoretical MW: 84.9 kDa) (Fig. 3).

Figure 3.

hRIC-3 pull-down assay. To assess protein-protein interaction, each 5-HT_3A-ICD construct (prey) was incubated with purified MBP-hRIC-3-His₆ (bait) bound to cobalt resin. Complexes retained on beads were eluted and analyzed using stain-free SDS-TGX gel electrophoresis. MBP-hRIC-3-His₆ is observed at 100 kDa (upper band), and the presence of an interaction is indicated by the appearance of a second band corresponding to the size of the protein of interest. Lack of a lower band indicates no interaction with hRIC-3. Experiments were repeated 2–10 times per protein and we observed the same results. (A) SDS-PAGE analysis of eluted complexes is as follows: lane 1: hRIC-3 only to ensure binding to the resin; lanes 2–8: eluates from pull-down assays with hRIC-3 and constructs as indicated. (B) SDS-PAGE analysis of eluted complexes is as follows: lanes 1–3: eluates from pull-down assays with hRIC-3 and constructs as indicated; lane 1: hRIC-3 only to ensure binding to the resin. The last two lanes are control proteins to indicate the localization on SDS-PAGE without either resin or hRIC-3.

Effect of deletions on interaction with RIC-3

We investigated the interaction between the ICD constructs and hRIC-3 by conducting pull-down assays directed against the His-tag of the RIC-3 construct. The resin-bound hRIC-3 protein was incubated with equimolar amounts of ICD, ΔMA, and MA constructs, respectively, and the eluates were analyzed by SDS-PAGE under reducing conditions to visualize both the bait and prey proteins (Fig. 3). Under these conditions, observing only one single band on the gel corresponding to the size of hRIC-3 (bait) migrating as a major band with a relative molecular weight of 100 kDa indicates that there was no interaction between hRIC-3 and the added ICD construct. On the other hand, observing two bands, one corresponding to the size of hRIC-3 and one to the size of the respective chimera, illustrates the presence of both bait and prey protein and therefore an interaction. We observed a second band in the SDS-PAGE gel in conditions in which hRIC-3 was incubated with the ICD (∼53 kDa) or ΔMA (∼50 kDa), indicating interactions of the involved proteins. On the contrary, no band was observed for the MA construct (∼41 kDa) (Fig. 3 A). The direct interaction of the full-length ICD and hRIC-3 had been shown previously in a similar experiment, and the ICD construct here was used as a control (26). MBP alone or omitting the hRIC-3 in the pull-down assay did not reveal the presence of prey proteins.

Eliminating the MA helix from potential interaction sites of the ICD with hRIC-3 shifted our focus to ΔMA, which contains L1, MX, and the structurally unresolved L2 segments. We designed three constructs to understand the role of L2 (the disordered flexible region between MA and MX helices) in the observed interaction between hRIC-3 and ΔMA: Δ34, Δ44, and Δ55. Using the pull-down assay, we observed strong interactions between all three deletion constructs and hRIC-3 using SDS-PAGE analysis (Fig. 3 A). The second band for each condition was observed at relative molecular weights of 50 kDa for Δ34, 48 kDa for Δ44, and 46 kDa for Δ55, consistent with the calculated weights of the prey proteins (Table 1). The L2 region of the ICD is formed by 61 amino acids, and by deleting 55 amino acids in the Δ55 construct, we eliminated all but the three most N- and C-terminal amino acids from the ICD. Therefore, we hypothesized that the interaction site must lie within the remaining segment of ΔMA: the L1-MX region.

Table 1.

Constructs and Respective Molecular Weight and Number of Amino Acids

Construct	Theoretical Molecular Weight of the Monomer Fused to MBP (MW, 103 g/mole)	Number of Amino Acids
ICD	53.4	115
MA	44.3	31
ΔMA	49.8	84
Δ34	50.0	81
Δ44	48.9	71
Δ55	47.7	60
L2	46.8	60
L1-MX	43.4	24

Identifying the short peptide responsible for hRIC-3 interaction

We continued our studies following our initial strategy of stepwise complimentary deletion of amino acids. Our final constructs were designed by using ΔMA as the template. We deleted the L2 region from ΔMA to create the L1-MX construct and, to complement this construct, we deleted the L1-MX segment to create the L2 construct. These constructs were designed to confirm that the L2 region does not confer the interaction of 5-HT_3A and hRIC-3 and that L1-MX is independently responsible for mediating this interaction. Using the hRIC-3 pull-down assay, we observed a second band corresponding to the size of the L1-MX construct (42 kDa) and no band in case of the L2 construct (Fig. 3 B).

X. laevis oocyte assay comparing interaction of homomeric and heteromeric 5-HT₃ receptors with hRIC-3

A sequence alignment between 5-HT_3A and 5-HT_3B subunits revealed that 5-HT_3B potentially lacks the helical structure post-M3, namely, the MX helix (Fig. 4 A). Because we had just demonstrated that the L1-MX segment mediates the hRIC-3 modulation of 5-HT_3A subunits, we wanted to determine if hRIC-3 coexpression impacts functional heteromeric 5-HT_3AB expression similar to homomeric 5-HT_3A expression. 5-HT_3B subunits alone are unable to form functional channels. Coexpression of 5-HT_3A with 5-HT_3B is required for functional expression (36). We used the cRNA of 5-HT_3A and 5-HT_3B subunits to express homomeric 5-HT_3A and heteromeric 5-HT_3AB in X. laevis oocytes. Both 5-HT_3A and 5-HT_3AB receptors were also coexpressed with hRIC-3 (hRIC-3) in oocytes. The current amplitudes of the receptors with and without hRIC-3 coexpression were measured in response to 10 μM 5-HT by TEVC technique 48 h postinjection (n ≥ 3) (Fig. 4, B and C). There was no significant difference (p = 0.93) in current amplitudes measured from oocytes injected with 5-HT_3A versus 5-HT_3AB. We observed a significant reduction in current amplitudes in oocytes coinjected with 5-HT_3A and hRIC-3 (p ≤ 0.001) as well as in oocytes coinjected with 5-HT_3AB and hRIC-3 (p ≤ 0.001) compared to the respective 5-HT₃ injected alone. The difference in inhibition in the latter two conditions was not statistically significant.

Figure 4.

Comparison of homomeric and heteromeric 5-HT₃Rs in hRIC-3 interaction. (A) Secondary structure prediction of 5-HT_3A and 5-HT_3B subunits obtained with PSIPRED is shown. Predicted α-helical segments of the ICDs are highlighted in gray within the sequence alignment. Consistent with structural data, 5-HT_3A contains a larger helical segment post-M3 (15 amino acids), whereas 5-HT_3B only has a four-amino-acid-long helix predicted post-M3. (B) 10 μM 5-HT-induced current amplitudes (nA) were measured 48 h after injection of cRNA into X. laevis oocytes by TEVC experiments. Current amplitudes recorded in oocytes injected with 5-HT_3A and 5-HT_3AB (gray bars) and after coexpression together with hRIC-3 (white bars) are shown. Error bars represent standard errors. Statistical significance was determined with one-way analysis of variance and Tukey’s multiple comparisons test (^∗∗∗ denotes p < 0.001). n > 3 oocytes were used for each condition. Currents were almost completely abolished after coexpression with hRIC-3 for both 5-HT_3A and 5-HT_3AB; therefore, bars are very small. (C) Sample traces of currents for each condition are representative of all data used for (B). To see this figure in color, go online.

Discussion

With more than 40 different subunits of the pLGIC superfamily found in humans, there is a great abundance of these ion channels in neuronal tissues and some expression in nonneuronal cells and tissues (13). The role of pLGICs is extremely important in signaling mechanisms and functioning of the nervous system, and disruptions to their function lead to a wide variety of diseases and disorders (9, 10, 11,13). According to existing structures of several pLGICs, such as 5-HT₃, nAChRs, GABA type A β3, and Glyα1 receptors, there is a high conservation in composition and structure of the ECD and TMD of these receptors (27, 28, 29,37, 38, 39, 40). The commonality of most of these structures is that ICDs have been removed, shortened, or in the case of full-length receptor structures, the ICD remained partially resolved (27,28). Additionally, all current drugs targeting these receptors interact with either the ECDs or TMDs (5,7, 8, 9,41). Considering the significant sequence similarities of these domains among the superfamily, drugs interacting with these two conserved domains have been shown to cause undesired effects by interacting with related subunits. Interestingly, ICDs of these receptors are diverse in length (73–262 amino acids in humans) and amino acid composition. This indicates that targeting the ICD may yield subtype-specific drugs. Therefore, the ICDs may represent safer and more effective potential drug targets for many diseases and disorders (8,42). hRIC-3 is a chaperone protein known to modulate maturation and membrane trafficking of some pLGIC members. Previously, we have shown that the ICD mediates this interaction (22, 23, 24) and that the ICD alone, in the absence of both ECD and TMD, interacts with hRIC-3 (26).

In this study, we aimed to identify the segment of the ICD involved in the interaction with hRIC-3 using a series of constructs that were designed with different deletions of segments of the ICD. After the α-helical M3 transmembrane segment, the ICD begins with a short loop (L1) that leads into an α-helical segment of ∼18 amino acids in length (MX helix), which is followed by a loop (L2) of 60 amino acids that has not been resolved in the structures. At the C-terminus of the ICD, an α-helical segment of 31 amino acids (MA helix) is continuous with the last transmembrane segment, M4. We had initially hypothesized that the MA helix that is only present in cation-conducting pLGICs mediates the interaction with hRIC-3. Using a direct protein-protein interaction assay, we observed that hRIC-3 directly interacts with ΔMA and not with the MA-helix segment of the ICD. These results indicated that the area of interaction lies within the L1-MX-L2-segment. To dissect a potential interaction site within the long L2 region, we designed three constructs that lacked 34, 44, and 55 amino acids, respectively, of L2 within the parent L1-MX-L2-MA construct. These three ICD constructs maintained interactions with hRIC-3. Furthermore, a direct correlation between intensity of the interaction (observed by intensity of the corresponding band on the gel) and the number of amino acids deleted from the L2 region was observed. We suspected that the deletion of amino acids from the L2 region caused changes in conformation or molecular crowding that facilitated interaction between ICD and hRIC-3. Consequently, we inferred that neither the MA helix nor L2 was the primary contributor to interaction sites of the ICD with hRIC-3. At this point, we hypothesize that the L1-MX region contains the interacting segment. From the x-ray structure of the 5-HT_3A, it is possible to infer that the MX helix would be responsible for mediating the interaction between the 5-HT_3A receptor and a membrane protein such as hRIC-3. The conformation of the MX helix in the structure and positional proximity of this helix to the membrane-spanning helices creates a favorable position for protein-protein interaction. One caveat was mentioned in the article: “However, whether the conformation of the post-M3 loop and of the MX helix in the detergent-solubilized, trypsin-treated, crystallized receptors do accurately represent a physiological conformation of intact receptors at the plasma membrane remains to be investigated” (27). To assess our hypothesis, we designed two constructs: L1-MX and L2. The observed interaction between the L1-MX, but not the L2 construct, and hRIC-3 indicated our hypothesis to be true. Both the strong interaction between Δ55, in which only the three most N- as well as C-terminal amino acids of L2 are present, and the lack of interaction of the L2 construct indicate that L2 is not directly mediating hRIC-3 interaction. In conclusion, we have identified the 24-amino-acid-long L1-MX segment of the ICD to be the interaction site of the 5-HT_3A receptor and hRIC-3.

RIC-3 has been shown to be a modulator of functional surface expression of 5-HT_3ARs, the only homomeric receptors in 5-HT₃R families. However, because the 5-HT_3A subunit is required for assembly with other 5-HT(B-E) subunits to form heteromeric receptors, RIC-3 may have an effect on heteromer formation (3,13,36,43). In 2007, it was discovered that a specific isoform of RIC-3 enhances the surface expression of homomeric 5-HT_3ARs while inhibiting the expression of heteromeric 5-HT_3ABRs in COS-7 cells (44). Shortly afterward, a separate study revealed that RIC-3 strictly enhances the surface expression of 5-HT_3ARs in HEK cells, whereas its mRNA was found in tissues also expressing C, D, and E 5-HT₃ subunits. RIC-3 and 5-HT_3C, 5-HT_3D, or 5-HT_3E subunits were found to be colocalized in the endoplasmic reticulum suggesting possible interactions of RIC-3 with subunits other than 5-HT_3A (45). These studies suggest that RIC-3 plays a role in the composition of 5-HT₃Rs and is involved in folding, assembly, or transport of homomeric 5-HT₃ARs at the expense of heteromeric receptor formation (20,45). However, 5-HT_3ABRs expressed in HEK cells are the only identified heteromeric receptors that resemble characteristics of native 5-HT₃Rs (36). In this study, we have identified the L1-MX segment of the 5-HT_3A subunit to mediate the interaction between 5-HT_3A subunits and hRIC-3. Interestingly, according to a sequence alignment between 5-HT_3A and 5-HT_3B subunits and secondary structure predictions using PSIPRED (46), the 5-HT_3B subunit lacks the MX helix (Fig. 4 A). Here, we investigated the effect of hRIC-3 on heteromeric 5-HT_3ABRs by coexpressing both 5-HT_3A and 5-HT_3AB with hRIC-3 in X. laevis oocytes and recording the currents in response to 5-HT application. Coexpression of hRIC-3 attenuated 5-HT-induced currents in both 5-HT_3A and 5-HT_3AB expressing oocytes, with no significant difference between 5-HT_3ARs and 5-HT_3ABRs (Fig. 4, B and C). It has been suggested that heteromerization of 5-HT_3B subunits with 5-HT_3A subunits covers ER-retention signals found in 5-HT_3B subunits (47). Our results, although not conclusive, suggest different possibilities leading to hRIC-3 attenuating surface expression of 5-HT_3AB: hRIC-3 may bind to unassembled 5-HT_3A subunits and inhibit their assembly with 5-HT_3A or 5-HT_3B subunits and thus inhibit both homomer and heteromer formation, or the interaction may block trafficking to the membrane. Alternatively, hRIC-3 may additionally bind to 5-HT_3B subunits and inhibit their assembly and/or trafficking. Future studies are necessary to directly investigate whether an interaction occurs between RIC-3 and 5-HT_3B. For this purpose, direct protein-protein interaction studies between hRIC-3 and a soluble 5-HT_3B-ICD construct, similar to the pull-down experiments in this study, are desirable.

Conclusion

In this study, we used a series of deletions and a direct protein-protein interaction assay to more closely define the interaction site of RIC-3 within the 5-HT_3A-ICD. We identified a 24-amino-acid-long peptide of the 5-HT_3A-ICD, L1-MX, to bear the interaction site. Of note, a short post-M3 loop and α-helical MX segment have been observed in structures of eukaryotic cation- (27, 28, 29,48, 49, 50, 51) but not anion-conducting pLGICs (52,53). This is in essential agreement with RIC-3 modulation of functional surface expression of cation- but not anion-conducting pLGICs. Further studies are needed to identify the face of the α-helix or the exact amino acids responsible for this interaction and additionally to delineate the binding site on RIC-3. Our data provide a potential surface area for future drug design to specifically target 5-HT₃ receptors and, by extension, also other cation-conducting pLGICs.

Author Contributions

M.J. designed research. E.P., A.G.S., and M.J. conceived the wild-type and engineered constructs. E.P. and A.G.S. performed experiments and analyzed data. A.P. performed the initial optimization of the hRIC-3 purification. E.P., A.G.S., A.P., and M.J. wrote the article.

Editor: Andrew Plested.