Human wild type (WT) forms of TREK-2 (NCBI reference sequence NM_138318)-expressing vector (pGH19-TREK-2) was constructed using Bgl II and Hind III sites. Point mutations and deletions were engineered using the MutanBEST kit (TaKaRa, Dalian, China)-guided high-fidelity PCR. The concatenated dimeric constructs were built to contain a different combination of WT and/or mutant (or deletant) TREK-2 cDNAs in defined position. The two consecutive subunits were connected by a flexible linker encoding the AAAGSGGSGGSTGGSSGSSGS sequence (Bagriantsev et al., 2012) with a little modification. A Sal I site was designed in the middle of the linker to facilitate cloning. One mutation (G312A) and two deletions (residues from W326 to A374 or from I323 to N543 were deleted in ΔpCt and ΔCt, respectively) were used in this study. G312 is located in the middle of M4 domain, and pCt is covalently linked with M4 and resides on the cytoplasmic side. For all the monomer (WT, G312A, ΔpCt and ΔCt) used to make dimers, Bgl II/Sal I was used to clone the former subunit, and Sal I/Hind III to the latter subunit. The name of concatenated dimers was defined as the relative position of monomeric channels. For example, WT-G312A represented that the first subunit was WT channel, and the second one was G312A channel. To investigate whether the flexible linker between monomers was degraded in mammalian cells during processing of the protein, either prior to or after membrane insertion, the concatenated WT dimeric construct (WT-WT) was also inserted into the pEGFPN1 vector (Clontech Laboratories, Mountain view, CA, USA) by using Bgl II and Sal I. All the constructs were confirmed by DNA sequencing. All the pGH19-based Plasmids were linearized by Xho I before in vitro transcription. cRNA was synthesized using the RiboMAX^TM Large Scale RNA Production Systems (Promega, Madison, WI, USA) kit.

Procedures used for harvesting oocytes from Xenopus laevis were approved by the Animal Care and Use Committee at Beijing Institute of Pharmacology and Toxicology. After excised from Xenopus laevis, pieces of ovarian lobes were subjected to collagenase (Sigma Aldrich, St Louis, MO, USA) digestion. Stage V or VI oocytes were selected and cRNA (0.1~10 ng/46 nl for each oocyte) was microinjected. Injected cells were incubated at 18°C in ND96 medium (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, 5 mM pyruvate, 100 mg/ml gentamycin, pH 7.2).

Whole-cell currents were measured 1–3 days after injection, and amplified using an Axoclamp2B amplifier (Axon Instruments, Union City, CA, USA) in two-electrode voltage clamp (TEVC) mode. Microelectrodes were pulled with a tip resistance of 0.1–1 MΩ when filled with 3 M KCl. Data were sampled at 2 kHz and filtered at 0.5 kHz with Clampex 10.0 software (Axon Instruments). Recordings were performed under standard, physiological extracellular solution (5 mM KCl, 93 mM NaCl, 1 mM MgCl₂, 1.8 mM CaCl₂, 5 mM HEPES, pH 7.4, adjusted with NaOH), and constant perfusion at room temperature. TREK-2 currents were elicited by continuous voltage-ramps from −120 to +60 mV from a holding potential of −80 mV, with 2 s in duration (current-voltage relationship, I-V curve). The currents of TREK-2 undergo obvious “run-up” at the beginning of recording, thus, each measurement was performed after the stabilization was reached (about 20 min) before applying stimulus. 2-APB (Promega, Madison, WI, USA) was diluted with the standard solution freshly.

HEK 293 cells were cultured in DMEM supplemented with 10% fetal bovine serum and 2 mM L-glutamine and held at 37°C in humidified air with 5% CO₂. Constructs encoding GFP (vector), TREK-2-GFP (monomer) and WT-WT-GFP (tandem-linked dimer) were transiently transfected into HEK 293 cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). Two days after transfection, the cells were harvested. Protein extracts were prepared by solubilizing in buffer (pH 7.4, 50 mM Tris, 270 mM NaCl, 1% Triton X-100) for 1 h and clarified by centrifugation at 12,000 × g for 30 min. Oocytes were injected with 10 ng cRNA, and then they were lyzed after 48 h in a cold buffer of 150 mM NaCl, 1.06 mM KH₂PO4, 2.07 mM Na₂HPO₄, 1% Triton X-100, pH 7.4 supplemented with antiproteases and clarified by centrifugation at 20,000 × g for 15 min. The above lysates were subjected to SDS-PAGE on 10% gel and wet-transferred onto a PVDF membrane. The lysates were hybridized with anti-TREK-2 antibody (Abgent, San Diego, 1:200) or anti-GFP (Cell Signaling Technology, Danvers, MA, USA, 1:1000) for oocytes and HEK293 in Western blotting, respectively. Actin (anti-actin antibody was purchased from ZSGB-BIO, China, 1:5000) was used as loading control. Reaction bands were visualized by incubating the blots with appropriate fluorescence-tagged secondary antibodies (Lincoln, NE, USA, 1:15,000).

Voltage clamp data were analyzed with origin 8.0 (OriginLab Corporation, Northampton, MA, USA) and GraphPad prism version 5.0 (GraphPad Software Inc., La Jolla, CA, USA). The currents recorded at 0 mV were used to calculate all the ratios, which were presented as mean ± SEM, and reflected n = 5 or more from at least two batches of oocytes. Activation ratio (AR) was used to observe the stimulatory effects of externally applied 2-APB on TREK-2. AR was calculated from I_2-APB/I_o, where I_2-APB represented the currents in the presence of 2-APB, and I_o was the baseline, stabilized currents in the absence of 2-APB, and was plotted as function of the concentration of 2-APB ([2-APB]_o). To estimate the inhibitory effects of extracellular alkalization, inhibition ratio (IR) was calculated from 1-I/I_pH6.5, where I represented the currents in a specific pH, and I_pH6.5 was the stabilized currents recorded at pH 6.5, and was plotted as a function of different pH (pH_o). The resulted ratios were fitted to the Hill equation. Statistical differences were tested by using unpaired Student’s t test where appropriate. Differences were considered significant if p < 0.05.

Mature TREK-2 channel forms as homodimers, so mutations or deletions of the channel can involve two subunits simultaneously. To perform a quantitative study of intersubunit or intrasubunit interaction, a strategy that employs concatenated dimer was adopted. TREK-2 tandem constructs, in which the two subunits are connected by a 21-amino-acid flexible linker (Bagriantsev et al., 2012; WT-WT, Figure 1A) were produced and expressed in Xenopus oocytes. To investigate whether the tandem channels are properly expressed, we firstly detected the expressed proteins by monomeric (TREK-2) and concatenated dimeric constructs (WT-WT) in Xenopus oocytes using western blotting. As shown in Figure 1B (left panel), the monomeric construct produces one specific band corresponding to the approximate size of mature monomer (red arrow). Notably, two nonspecific bands were also detected by the anti-TREK-2 primary antibody in oocytes (asterisks). The WT-WT construct produces a specific band (green arrow, overlapped with a nonspecific band) corresponding with dimeric TREK-2. No monomeric signal was detected by In WT-WT construct-injected oocytes. To rule out the nonspecific expression, we further assessed their expressions in HEK293 cells by transfecting GFP-tagged monomeric and tandem-linked vectors. As shown in Figure 1B (right panel), distinct monomeric or dimeric bands were observed at the predicted molecular mass in TREK-2- or WT-WT-transfected group, respectively, but no monomeric signals were found in WT-WT group. Notably, no immunoreactivity was detected in the pEGFPN1-transfected cells (vector). These data indicate that the concatamers remain intact and the linker is not susceptible to degradation, and the WT-WT constructs can be used for further investigations.

To investigate the gating properties of the tandem-linked dimer, 2-APB response of the channel was investigated and compared with that of monomeric TREK-2. As the current-voltage relationships (I-V curves) shown in Figure 1C, the currents recorded from WT-WT channels was radically stimulated by 100 μM 2-APB. The activation ratio (AR, measured using the currents at 0 mV) values were plotted against their 2-APB concentrations, and results show that the activating effects of 2-APB on WT-WT channels was similar with that of monomeric TREK-2 (Figure 1D). The maximal AR (AR_max, produced by 333 μM 2-APB, unless otherwise specified) produced by TREK-2 monomer and WT-WT tandem-linked dimer was 19.46 ± 1.94 and 18.13 ± 1.30, respectively (Table 1).

To further evaluate the gating properties of the concatenated dimer, we also measured its response to extracellular alkalization. When the extracellular pH (pH_o) was switched from 6.5 to 8.0 or 9.3, the currents of WT-WT channels was drastically inhibited (Figure 1E). Importantly, the dose response curves (Figure 1F) demonstrating that the IRs against their pH_o clearly revealed that the tandemly linked dimer was inhibited by extracellular alkalization in a similar degree with monomer. The maximal IR (IR_max, produced by pH 9.3) for monomeric TREK-2 and WT-WT channels was 0.78 ± 0.02 and 0.73 ± 0.02, respectively (Table 1). Taken together, these data suggest that the tandem-linked strategy does not influence the gating properties of TREK-2 channels. Thus, this method was further used to investigate how the intersubunit and intrasubunit interplay.

Our previous study have identified that the cytoplasmic pCt (from W326 to A374, the relative position was illustrated in Figure 2A) plays an essential role in gating TREK-2 (Zhuo et al., 2016), so we further used this negative gating mutation to investigate how these two subunits may interact with each other. ΔpCt was constructed as either the first subunit (ΔpCt-WT) or the second subunit (WT-ΔpCt). The response of these heterodimers to 2-APB were investigated and compared with WT-WT and ΔpCt-ΔpCt homodimers.

When the gating of one subunit in TREK-2 dimer does not alter the gating kinetics of another one, the heterodimers should respond to stimuli in a linear combination for their two homodimers (such as WT-WT and ΔpCt-ΔpCt in our study), and such interactive pattern belongs to independent or sequential manner. If the response of heterodimers to stimuli is similar with those of homodimers, the gating between subunits is compatible with concerted cooperative fashion (Zandany et al., 2008; Tombola et al., 2010; Meisel et al., 2012; Thomson et al., 2014). As shown in Figure 2B, 100 μM 2-APB strongly stimulated the currents of WT-ΔpCt, while the activity of ΔpCt-ΔpCt was only slightly stimulated. Comparison of the concentration-dependent curves of the four channels further indicated that the 2-APB responses of both WT-ΔpCt and ΔpCt-WT channels were more similar to that of WT-WT than to that of ΔpCt-ΔpCt (Figure 2C). The AR_max for WT-ΔpCt and ΔpCt-WT was 15.50 ± 0.74 and 17.37 ± 1.56, respectively. However, the ratio of ΔpCt-ΔpCt was only 4.34 ± 0.10 even when the concentration of 2-APB was enhanced to 800 μM (Table 1).

Our previous study has also revealed that the full Ct regulates the gating of TREK-2 in a similar manner with pCt (Zhuo et al., 2016), so here we measured 2-APB responses of ΔCt deletant incorporated heterodimers (WT-ΔCt and ΔCt-WT). Consistent with the results from ΔpCt heterodimer, both WT-ΔCt and ΔCt-WT exhibited similar sensitivity to 2-APB as that of WT-WT channels (Figure 2D and Table 1). Taken together, the WT subunit exerts strong, positive effects on the ΔCt or ΔpCt subunit in response to 2-APB stimulation, suggesting that the concerted intersubunit cooperativity gates TREK-2 channels at 2-APB pathway.

To confirm above conclusion, we also evaluated the ΔCt- or ΔpC-contained heterodimers to pH_o changes (pH_o pathway). Similar with the situation of 2-APB response, WT-ΔpCt channels were strongly inhibited by extracellular alkalization, whereas the currents from ΔpCt-ΔpCt channels were slightly blocked (Figure 3A). Furthermore, the responses of WT-ΔpCt (IR_max = 0.79 ± 0.01) and ΔpCt-WT (IR_max = 0.77 ± 0.02), WT-ΔCt (IR_max = 0.76 ± 0.03) and ΔCt-WT (IR_max = 0.81 ± 0.02) to pH_o changes were more resembled with that of WT-WT channels than that of ΔpCt-ΔpCt channels (Figures 3B,C and Table 1). Taken together, these results reveal that the heterodimers including pCt or Ct deletant have similar sensitivity in response to pH_o changes as that of WT-WT channels, suggesting that the pCt (or Ct) domain controls the channel in a concerted intersubunit cooperative manner in the pH_o pathway.

The glycine hinge of M4, G312 (the relative position was illustrated in Figure 2A), is involved in the regulations of both 2-APB and extracellular alkalization (Zhuo et al., 2016). By introducing G312A mutation to either the first or second subunit of WT-WT channels and evaluating their responses to 2-APB and ΔpH_o, we investigated whether G312 gates TREK-2 in a cooperative or seqential manner between subunits. WT-G312A was activated dramatically by 100 μM 2-APB, while the activated degree to G312A-G312A was radically decreased (Figure 4A). The dose response curves of WT-G312A (AR_max = 17.16 ± 1.00) and G312A-WT (AR_max = 16.84 ± 2.00) were more similar to that of WT-WT channels than to the channel comprising two G312As (AR_max = 7.80 ± 0.50; Figure 4B and Table 1). In addition, dimerizing G312A with WT subunit exhibited more resembled sensitivity with WT-WT channels to ΔpH_o than that of G312A-G312A (Figures 4C,D and Table 1). These results indicate that G312 of the WT subunit exerts positive effects on the mutated one (G312A) during gating processes induced by 2-APB and pH_o, suggesting that the glycine hinge of M4 also gates TREK-2 in a concerted cooperative manner between subunits.

Since both Gly312 and pCt control TREK-2 gating cooperatively between subunits in modulating the bidirectional communication between SF gate and pCt domain (Figures 2–4), it is critical to understand how these elements interact with each other during this communication: a cis-type mechanism in which the interaction only occurs within one subunit, or a trans-type mechanism that involves both subunits. To discriminate these two possible mechanisms, we performed mutant cycle analysis. For this purpose, a concatenated dimer with one nonfunctional subunit was required. G312A-WT dimer was chosen because both 2-APB and ΔpH_o induced gating pathways were interrupted by G312A mutation in the former subunit. G312A transition and ΔpCt deletion were individually or simultaneously introduced into the WT subunit to obtain G312A-G312A (as also described in Figure 4), G312A-ΔpCt and G312A-G312A/ΔpCt. Then their responses to 2-APB and pH_o were investigated. If the interaction between G312 and pCt is trans-type, the pCt of the G312A subunit and the G312 of the ΔpCt subunit will rescue the channel function due to their cooperative natures. Accordingly, the effects of 2-APB and pH_o changes on G312A-ΔpCt will be similar with those of G312A-WT. Meanwhile, such effects on G312A-G312A/ΔpCt will be analogous with those of G312A-G312A. Nevertheless, the ability of 2-APB to activate G312A-ΔpCt was decreased radically compared with WT-ΔpCt, and such ability was also decreased in G312A/ΔpCt when compared with G312A-G312A (Figure 5A). According to the concentration-dependent curves represented in Figure 5B and Table 1, the AR_max of G312A-WT was 16.84 ± 2.00, while the ratio of G312A-ΔpCt was decreased to 5.32 ± 0.46. Likewise, incorporation of ΔpCt into G312A-G312A led to the decrement of AR_max from 7.80 ± 0.50 in G312A-G312A to 3.36 ± 0.37 in G312A-G312A/ΔpCt. The pH_o response of G312A-ΔpCt also exhibited blunted gating kinetics (Figures 5C,D) when compared with WT-ΔpCt. The IR_max of WT-ΔpCt was 0.79 ± 0.01, whereas the ratio of G312A-ΔpCt was decreased to 0.43 ± 0.03 (Table 1). Meanwhile, very similar with the situation in 2-APB, introduction of ΔpCt into G312A-G312A also resulted in blunted pH_o response of G312A-G312A/ΔpCt (IR_max = 0.12 ± 0.05) when compared with G312A-G312A (IR_max = 0.23 ± 0.02; Figures 5C,D). Taken together, further deleting the function of pCt of the WT subunit in G312A-WT or G312A-G312A channel attenuates its sensitivity to 2-APB and pH_o changes, suggesting that the allosteric regulations triggered by these stimuli occur along one subunit, rather than across both subunits. Namely, G312 and pCt gate TREK-2 channels via a cis-type mechanism.