The β1-Subunit of the MaxiK Channel Associates with the Thromboxane A2 Receptor and Reduces Thromboxane A2 Functional Effects*

Background: The vasoconstricting thromboxane A2 prostanoid receptor (TP) is physically coupled to the MaxiK channel α-subunit, down-regulating its activity, but the role of the MaxiK β1-subunit is unknown. Results: The MaxiK β1-subunit can interact with TP independently of the MaxiK α-subunit and counteracts activated TP-induced current inhibition, and its ablation increases TP vasoconstricting potency. Conclusion: The β1-subunit modifies TP actions. Significance: This is the first demonstration that the MaxiK β1-subunit associates with a vasoconstricting receptor. The large conductance voltage- and Ca2+-activated K+ channel (MaxiK, BKCa, BK) is composed of four pore-forming α-subunits and can be associated with regulatory β-subunits. One of the functional roles of MaxiK is to regulate vascular tone. We recently found that the MaxiK channel from coronary smooth muscle is trans-inhibited by activation of the vasoconstricting thromboxane A2 prostanoid receptor (TP), a mechanism supported by MaxiK α-subunit (MaxiKα)-TP physical interaction. Here, we examined the role of the MaxiK β1-subunit in TP-MaxiK association. We found that the β1-subunit can by itself interact with TP and that this association can occur independently of MaxiKα. Subcellular localization analysis revealed that β1 and TP are closely associated at the cell periphery. The molecular mechanism of β1-TP interaction involves predominantly the β1 extracellular loop. As reported previously, TP activation by the thromboxane A2 analog U46619 caused inhibition of MaxiKα macroscopic conductance or fractional open probability (FPo) as a function of voltage. However, the positive shift of the FPo versus voltage curve by U46619 relative to the control was less prominent when β1 was coexpressed with TP and MaxiKα proteins (20 ± 6 mV, n = 7) than in cells expressing TP and MaxiKα alone (51 ± 7 mV, n = 7). Finally, β1 gene ablation reduced the EC50 of the U46619 agonist in mediating aortic contraction from 18 ± 1 nm (n = 12) to 9 ± 1 nm (n = 12). The results indicate that the β1-subunit can form a tripartite complex with TP and MaxiKα, has the ability to associate with each protein independently, and diminishes U46619-induced MaxiK channel trans-inhibition as well as vasoconstriction.

The regulation of vascular tone is an integrative physiological process involving a broad range of vasoconstricting and vasorelaxing factors. The large conductance voltage-and Ca 2ϩ -activated K ϩ channel (MaxiK, BK Ca , BK) 6 is one of the key regulators of vascular tone depending on the particular level of activity achieved in response to vasoactive substances.
We have recently shown that activation of TP trans-inhibits MaxiK channel activity in native coronary arterial muscle; this G-protein independent coupling is supported by the ability of TP to physically interact with the MaxiK ␣-subunit (MaxiK␣) (12). Because in coronary arterial muscle the majority of currents resemble channels accompanied by the ␤1-subunit (13), the question of whether (and how) ␤1 relates to vasoconstricting G protein-coupled receptors (in this case) to TP actions becomes relevant.
The results from this study reveal the following. (i) The MaxiK regulatory ␤1-subunit can by itself interact with vasoconstricting TP. (ii) ␤1 expression decreases activated TP-induced positive shift in MaxiK␣ voltage dependence of the fractional open probability. (iii) ␤1 expression is linked to a reduced potency of the thromboxane A2 analog U46619 in producing vasoconstriction.

Animals
Three-month-old wild-type (C57BL/6NCrL, Charles River Laboratories) and ␤1 knock-out (␤1 Ϫ/Ϫ ) male mice were used. The ␤1 Ϫ/Ϫ line was prepared using conventional homologous recombination by deleting exons II and III. Gene knock-out was confirmed by genotyping and the absence of ␤1 mRNA in the aorta, uterus, and bladder. All animal protocols received institutional approval.

HEK293T Cell Culture and DNA Transfection
HEK293T cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 2 units/ml penicillin, and 2 mg/ml streptomycin at 37°C in a humidity-controlled incubator supplemented with 5% CO 2 . HEK293T cells at 80% confluence were transiently transfected using Lipofectamine 2000 (Invitrogen). Plasmid concentrations used for transfection are given in the figure legends.
Lysates were centrifuged at 13,000 ϫ g for 10 min at 4°C, and the supernatants were precleared with 10 l of protein A/G resin (Pierce)/mg of protein for 1 h at 4°C with shaking and centrifuged at 2000 ϫ g for 2 min. The precleared lysates (1 mg of protein) were incubated overnight at 4°C with 10 l of antibody-saturated protein A/G resin (2 g of antibody/10 l of resin were incubated for 2 h at 4°C with shaking) in a final volume of 500 l. Samples were centrifuged at 2000 ϫ g for 2 min and washed five times with lysis buffer. The immunoprecipitated proteins were eluted from the beads with 30 l of 3ϫLaemmli sample buffer at 37°C for 1 h and centrifuged at 13,000 ϫ g for 3 min at 4°C. Immunoprecipitated proteins and input lysates were analyzed by SDS-PAGE and immunoblotting. Molecular weight markers were from LI-COR Biosciences (catalogue no. 928-40000) except those used for TP, which were low-range SDS-PAGE standards from Bio-Rad (catalogue no. 161-0305). The Odyssey infrared imaging system (LI-COR Biosciences) was used to analyze single-or double-labeled immunoblots.

Immunolabeling
HEK293T cells were plated onto poly-D-lysine-coated coverslips 24 h after transfection and incubated at 37°C for another 24 h. Live cells were incubated with 5 g/ml anti-c-Myc polyclonal antibody for 1 h on ice in a 37°C incubator supplemented with 95% air and 5% CO 2 atmosphere; washed once with PBS (2.67 mM KCl, 138 mM NaCl, 1.47 mM KH 2 PO 4 , and 8.1 mM Na 2 HPO 4 (pH 7.4)), and then fixed with 4% paraformaldehyde in phosphate buffer (0.1 M Na 2 HPO 4 , and 0.022 M NaH 2 PO 4 (pH 7.4)). Cells were permeabilized and blocked with PBS containing 0.2% Triton X-100 and 10% normal goat serum for 30 min at room temperature, followed by incubation with 5 g/ml anti-FLAG monoclonal antibody in PBS containing 0.2% Triton X-100 and 1% normal goat serum overnight at 4°C. Cells were washed three times with PBS containing 0.2% Triton X-100 and labeled with 2 g/ml secondary antibodies (Alexa Fluor 568 anti-mouse and Alexa Fluor 488 anti-rabbit) for 1 h at room temperature. Cells were rinsed twice with PBS containing 0.2% Triton X-100 and once with PBS alone. Cells were then mounted on slides using ProLong Gold (Invitrogen). Images were taken with an Olympus confocal microscope. All conditions, including optical sectioning and acquisition parameters, were identical for a given experiment.

Protein Proximity Index Analysis
To calculate the protein proximity index (PPI) (12,15), pairs of digital images were acquired at 0.0288 m/pixel and subjected to the following analysis using a custom-made program.
Median Filter-First, the nonspecific background value at each pixel was estimated by calculating the median value of a 32 ϫ 32 pixel square centered at the target pixel. This value was then subtracted from the intensity of the target pixel.
Autocorrelation and Cross-correlation Analysis-Second, three-dimensional autocorrelation (of each TP and ␤1 image) and cross-correlation (of TP and ␤1 images) plots as a function of pixel shift in the x,y axis were constructed. Corresponding contour plots were generated, and line scans were obtained to plot the correlation intensity values as a function of pixel shift. FEBRUARY 1, 2013 • VOLUME 288 • NUMBER 5

JOURNAL OF BIOLOGICAL CHEMISTRY 3669
Fitting-Third, line scan plots were then fitted to the sum of two Gaussian functions: (A 1 exp(Ϫ(M 1 Ϫ x) 2 /2B 1 2 )) ϩ (A 2 exp(Ϫ(M 2 Ϫ x) 2 /2B 2 2 )) Ϫ Base, where A 1 and A 2 are the amplitudes, B 1 and B 2 are the widths, M 1 and M 2 are the means of the position of the peaks in the x axis, x is the pixel shift, and Base is a constant value. The fits displayed sharp and shallow components. The sharp components (A 1 ) correspond to specific labeling, and the shallow components (A 2 ) correspond to antibody background and random colocalization.
PPI Calculation-Finally, the PPI values were obtained by dividing A 1 from the cross-correlation analysis by A 1 from each of the autocorrelation analyses.

HEK293T Cell Patch Recording
TP and c-Myc-MaxiK␣ were subcloned into the pIRES vector with TP under the control of the CMV promoter and with c-Myc-MaxiK␣ under the control of the internal ribosome entry site (TP-pIRES-c-Myc-MaxiK␣). This was necessary to ensure coexpression of both proteins in the same cell. GFP was coexpressed to help monitor the transfected HEK293T cells. Twenty-four hours after transfection with TP-pIRES-c-Myc-MaxiK␣ with or without the ␤1-subunit, macroscopic currents were recorded in the inside-out patch configuration. The halfactivation potential (V 1 ⁄2) was measured to monitor the coexpression of ␤1. A negative shift of ϳ90 mV by the ␤1-subunit indicated full coupling (2). The pipette and bath solution contained 105 mM potassium methanesulfonate, 5 mM KCl, 5 mM HEDTA, 3.9 mM CaCl 2 , and 10 mM HEPES (pH 7.4). The free calcium concentration ([Ca 2ϩ ] i ) of the bath solution measured with a Ca 2ϩ electrode was 6.7 M (facing the intracellular side of the channel). U46619 (500 nM) was dissolved in bath solution and perfused. Pipette resistances were 2.5-3.5 megohms. Currents were filtered at 1 kHz and digitized at 10 kHz. Instantaneous tail currents (I) measured at the beginning of a constant repolarizing pulse to Ϫ70 mV were used to fit a Boltzmann distribution of the following form. FP o ϭ G/G max ϭ 1/(1 ϩ exp((V 1 ⁄2 Ϫ V)z␦F/RT)), where FP o is the fractional open probability; G is the macroscopic conductance; G max is the maximum macroscopic conductance; V is the voltage of the test pulse (preceding the repolarizing pulse); V 1 ⁄2 is the half-activation potential; z␦ is the effective valence; and F, R, and T have their usual thermodynamic meanings. G ϭ I/(V Ϫ E K ), where I is the instantaneous tail current, V is the voltage of the constant repolarizing pulse (in this case, Ϫ70 mV), and E K is the reversal potential for K ϩ (in this case, 0 mV).
where I max is the maximum instantaneous tail current and the other parameters have been defined. Note that the voltage dependence of the macroscopic conductance (G) reflects the voltage dependence of channel P o (16), and thus, G/G max ϭ FP o .

Isometric Contraction
Aortic rings (endothelium denuded with a cotton thread) were mounted in an organ bath kept at 37°C, filled with modified Krebs solution (119 mM NaCl, 4.7 mM KCl, 1.6 mM CaCl 2 , 1.2 mM MgSO 4 , 1.2 mM KH 2 PO 4 , 22 mM NaHCO 3 , 8 mM HEPES, 5 mM creatine, 20 mM taurine, 5 mM pyruvate, and 5 mM glucose (pH 7.4); gassed with 95% O 2 and 5% CO 2 ), and connected to an isometric force transducer (World Precision Instruments). Tension was recorded with WinDaq (DATAQ Instruments). After equilibration for 60 min at the optimum resting tension (ϳ7.8 millinewtons), rings were contracted with 80 mM KCl for ϳ15 min and then washed out with Krebs solution. After equilibration for another 30 min, cumulative concentrations of U46619 (0.01 nM to 3 M) were applied. Contraction was normalized to KCl-induced contraction and expressed as a percentage. The calculation was as follows: % contraction ϭ 100 ϫ (U46619-induced aortic tension Ϫ basal tension)/(80 mM KCl-induced aortic tension Ϫ basal tension). Half-maximum effective concentration values (EC 50 ) were calculated using a Hill function: % contraction ϭ E max /(1 ϩ EC 50 / [U46619]) H , where E max is the maximum contraction and H is the Hill coefficient. To compare KCl contraction in wild-type (␤1 ϩ/ϩ ) versus knock-out (␤1 Ϫ/Ϫ ) aortic rings, maximum contraction was normalized to the dry weight of each ring.

Statistical Analysis
Data are presented as means Ϯ S.E. Statistical comparisons between groups were made using Student's t test. A p value Ͻ0.05 was considered statistically significant.  (Fig. 1E, white and black bars); co-immunoprecipitation efficiency was calculated by normalizing the band intensities of co-immunoprecipitated MaxiK␣ (Fig. 1B, lanes 2 and 4) to those of immunoprecipitated TP (Fig. 1F, lanes 2 and 4) and to MaxiK␣ expression in the corresponding input lysates (Fig. 1H, lanes 2  and 4). Overall, the experiments demonstrate that ␤1 is unable to disrupt MaxiK␣-TP association but, at the same time, can interact with MaxiK␣.

␤1 Does Not Prevent MaxiK␣-TP Interaction and Is Able to
Unexpectedly, we also found that ␤1 could readily immunoprecipitate TP without MaxiK␣ being present (Fig. 1I, lane 3 versus lane 4). This novel interaction was verified by reverse co-immunoprecipitation, where TP pulled down ␤1 in the absence of MaxiK␣ (Fig. 1J, lane 3). Moreover, the presence of MaxiK␣ did not modify the degree of interaction between TP and ␤1, as co-immunoprecipitation efficiency was the same in the absence (control) and presence of MaxiK␣ (Fig. 1E, dotted and checkered bars). Efficiency was obtained by normalizing

␤1-Subunit Regulates TP-MaxiK Function
co-immunoprecipitated ␤1 (Fig. 1J, lanes 3 and 4) to immunoprecipitated TP (Fig. 1F, lanes 3 and 4) and to ␤1 expression in the corresponding cell lysates (Fig. 1K, lanes 3 and 4). The proper expression of ␤1, TP, and MaxiK␣ is shown in immunoblots of input lysates in Fig. 1 (D, H, and K). In summary, ␤1 is able to associate with both MaxiK␣ and TP, forming a tripartite complex, but its association with TP is independent of its interaction with MaxiK␣.
␤1 and TP Colocalize at the Plasma Membrane-We next examined whether ␤1 and TP can colocalize at the plasma membrane without the assistance of MaxiK␣. HEK293T cells cotransfected with ␤1-FLAG (tagged at its intracellular C terminus) and c-Myc-TP (tagged at its extracellular N terminus) were first labeled live to assess the plasma membrane localization of TP. This was followed by fixation, permeabilization, and labeling of ␤1. Confocal images of immunolabeled cells show a high degree of correlation between TP signals at the plasma membrane ( Fig. 2A) and ␤1 labeling (Fig. 2B), which is emphasized in the overlay (Fig. 2C). PPI (12,15) was calculated to quantify ␤1 and TP "colocalization" as described under "Experimental Procedures." Panels D and E in Fig. 2 show the autocorrelation three-dimensional plots as a function of pixel shift of TP (panel A) and ␤1 (panel B) images, respectively. The crosscorrelation three-dimensional plot of both TP and ␤1 images is shown in Fig. 2F. At zero pixel shift, the surface plots have a peak that decays abruptly by shifting the image a few pixels, indicating specific TP and ␤1 signals (Fig. 2, D and E) and specific colocalization (Fig. 2F).   (Fig. 2K) indicating that at least 58% of labeled TP is in close proximity to ␤1 and vice versa. These results map ␤1-TP interaction to the plasma membrane, although other trafficking locations are not excluded, and confirm that ␤1-TP association does not require MaxiK␣ expression.
␤1 Extracellular Loop Contributes to ␤1-TP Interaction-Which domain of ␤1 is required for association with TP? Fig. 3A illustrates the topology of wild-type ␤1 (␤1(1-191)) and deletion constructs (␤1(1-102) and ␤1(1-71)) designed to circumscribe the region relevant for TP association. Black circles represent residues encoded by exon II, gray circles represent those encoded by exon III, and open circles represent those encoded by exon IV. All three constructs were tagged with a FLAG epitope at the C terminus. The first deletion construct tested, ␤1(1-102), lacked all residues encoded by exon IV (residues  103-191, open circles). The idea behind this was that sequences in exons may have been evolutionarily selected as functional cassettes and thus could be a good starting point to look for residues that interact with TP. Fig. 3B shows that similar to wild-type ␤1 (␤1(1-191)), ␤1(1-102) readily interacted with
Used as a positive control, TP was effectively immunoprecipitated in all instances (Fig. 3C); both monomeric and dimeric TP species were detected at ϳ40 and ϳ80 kDa (arrow and double dots, respectively). Proper expression of ␤1 constructs (Fig.  3D) and TP (Fig. 3E) was tested by immunoblotting cell lysates used for immunoprecipitation. Wild-type ␤1 and ␤1(1-102) deletion constructs showed clear bands at higher molecular masses, which would correspond to dimers of the protein (Fig.  3D, lanes 1 and 2, dashed boxes). To compare the efficiency of TP in pulling down the ␤1 constructs, the density of co-immunoprecipitated ␤1 bands (Fig. 3B) was quantified in each case and normalized to immunoprecipitated TP signals (Fig. 3C) and to the signal of ␤1 bands in cell lysates (Fig. 3D). The bands corresponding to the protein monomer and dimer were taken into account when analyzing the data. Mean values of normalized density are shown in Fig. 3F. The efficiency of TP in coimmunoprecipitating ␤1(1-102) was 94 Ϯ 14% (n ϭ 3), which was not significantly different from that of TP in pulling down wild-type ␤1(1-191) (set to 100%). These data demonstrate no dramatic change in ␤1-TP interaction after deletion of residues encoded by exon IV. On the other hand, further deletion of residues 72-102 of the ␤1 extracellular loop (␤1(1-71)) yielded a protein that was no longer able to interact well with TP. The co-immunoprecipitation efficiency dropped nearly 90% to an efficiency of 13 Ϯ 6% (n ϭ 3), indicative of the role of the ␤1 extracellular loop, likely of residues 72-102, in ␤1-TP association.
␤1 Decreases Activated TP Reduction in MaxiK␣ Fractional Open Probability-The results so far indicate that ␤1 can independently interact with either TP or MaxiK␣. Therefore, we next asked the question whether ␤1 could modify TP and MaxiK␣ functional coupling, whereby activated TP trans-inhibits MaxiK␣ (12). To address this question, we compared the effect of a stable analog of thromboxane A 2 (U46619) on MaxiK␣ macroscopic currents coexpressed with TP in the presence and absence of ␤1 in HEK293T cells (Fig. 4). In all experiments, the same patch expressing TP and MaxiK␣ alone (TP-pIRES-c-Myc-MaxiK␣) or in conjunction with the ␤1-subunit was used to measure the current before (control) and after U46619 treatment. The holding potential was 0 mV, [K ϩ ] was the same in the bath and patch pipette, and [Ca 2ϩ ] facing the intracellular side of the membrane was 6.7 M. Under these conditions, at 0 mV, there was no net current flux (Fig. 4, A-D), as the K ϩ concentration was equal at both sides of the membrane, yet under control conditions, channels had a significant FP o of Ͼ0.6 at this voltage (Fig. 4, E and F, black circles). As a consequence of the latter, when patches were stimulated by a series of test pulses, control currents displayed very fast activation kinetics (unresolved by the acquisition system, yielding   FEBRUARY 1, 2013 • VOLUME 288 • NUMBER 5

␤1-Subunit Regulates TP-MaxiK Function
traces that were almost square) (Fig. 4, A, C, and D), and at very negative potentials, it was possible to observe "tail" currents (arrowheads) that deactivated to reach the steady state for the particular voltage. Confirming our previous findings (12), in patches expressing TP and MaxiK␣ alone, application of 500 nM U46619 to the same cell patch dramatically slowed down MaxiK␣ current activation kinetics at positive potentials (Fig. 4, B versus A) and reduced tail currents at the beginning of negative test pulses (arrowheads), both indicative of a TP-mediated decrease in channel P o or channel inhibition. This inference is based on the well established concepts that the time course of channel average P o corresponds to the macroscopic current (16) and that the instantaneous tail current is proportional to channel P o at the end of the preceding pulse (17).
The macroscopic current (I) at the end of test pulses was measured to construct current-voltage relationships (I-V curves). The I-V curves show that TP activation by U46619 produced a decrease in macroscopic current amplitude at several potentials (Fig. 4, B versus A, insets). Because I ϭ NiP o , where N is the number of channels, i is the unitary conductance, and P o is the channel open probability, a reduction in I could be due to an inhibition of any of the three parameters. However, we showed previously that activated TP inhibits channel P o but does not decrease i or N in inside-out patches (12). To confirm an inhibition of MaxiK P o by activated TP, we evaluated the voltage dependence of the macroscopic conductance or FP o before (control) and after U46619 treatment (Fig. 4E) by measuring instantaneous tail currents at the beginning of the repolarizing pulse (Fig. 4, A and B, arrows) (see "Experimental Procedures"). It is clear from these measurements that U46619 induced a 51 Ϯ 7 mV shift of the voltage dependence of FP o from a half-activation potential (V 1 ⁄2) of Ϫ15 Ϯ 5 mV (control; n ϭ 7) to a V 1 ⁄2 of 37 Ϯ 7 mV (n ϭ 7). This positive shift by U46619 can be interpreted as an inhibition of MaxiK␣ channel P o by activated TP.
Parallel experiments were performed in cells coexpressing TP and MaxiK␣ (in the pIRES vector) together with the ␤1-subunit (Fig. 4, C and D). To saturate the effect of ␤1, the molar ratio of transfected ␤1 to TP ϩ MaxiK␣ was 3:1. Under these conditions and consistent with previous studies (2), in the presence of ␤1, the V 1 ⁄2 of MaxiK␣ was Ϫ99 Ϯ 2 mV at [Ca 2ϩ ] i ϭ 6.7 M (Fig. 4F, Control).
The presence of ␤1 modified the U46619 effect on MaxiK␣ macroscopic currents. The kinetics of current activation at positive test potentials and tail currents at the beginning of negative test pulses induced by U46619 remained practically the same as under control conditions (Fig. 4, D versus C), whereas I-V curves measured at the end of test pulses showed a modest decrease (Fig. 4, D versus C, insets). The voltage dependence of FP o was also calculated from tail currents at the beginning of the repolarizing pulse (Fig. 4, C and D, arrows). In this case, U46619 caused a smaller but significant shift of 20 Ϯ 6 mV in the voltage dependence of FP o from a V 1 ⁄2 of Ϫ99 Ϯ 2 mV (n ϭ 7) in the control to Ϫ79 Ϯ 8 mV (n ϭ 7) after treatment (Fig. 4F).
To rule out the remote possibility that the decreased effect of U46619 in cells coexpressing TP ϩ MaxiK␣ ϩ ␤1 was due to a selective inhibition of TP expression compared with MaxiK␣ when coexpressed with ␤1, we measured protein expression levels in these cells and compared them with the levels of expression in cells transfected with TP and MaxiK␣ alone. Immunoblot analysis showed that when the ␤1-subunit was expressed, both TP and MaxiK␣ relative expression levels were proportionally decreased, TP by 65 Ϯ 1% (n ϭ 3) and MaxiK␣ by 59 Ϯ 7% (n ϭ 3) (p ϭ 0.37), ruling out selective decreased TP expression by ␤1 cotransfection. Together, the results support the view that ␤1 modifies the way activated TP modulates MaxiK␣ activity.
␤1 and U46619-induced Aortic Vasoconstriction-In addition to causing MaxiK␣ channel inhibition, activation of TP by the thromboxane A 2 analog U46619 induces vasoconstriction

DISCUSSION
In this study, we present the first demonstration that the MaxiK regulatory ␤1-subunit can by itself interact with TP and at the same time can assemble in a tripartite complex with MaxiK␣ and TP. Furthermore, our study provides evidence that ␤1 is able to reduce TP agonist thromboxane A 2 -induced functional effects, i.e. thromboxane A 2 -induced MaxiK␣ inhibition as well as vasoconstriction.
TP, ␤1, and MaxiK␣ in a Macromolecular Complex-MaxiK␣ and ␤1 are known to associate with each other without the need of an intermediary protein, e.g. TP. Expressed ␤1 is known to be in close contact with MaxiK␣, enabling covalent cross-linking of engineered cysteines in both proteins (18). Our experiments also confirmed this physical association by co-immunoprecipitation without coexpression of TP (Fig. 1C). On the other hand, our previous studies have shown a close proximity of MaxiK␣ with TP, enabling Förster resonance energy transfer without the need of ␤1 (12). Based on these premises, it was possible that ␤1 and TP might associate via MaxiK␣ (Fig.  6A) or that ␤1 could disrupt MaxiK␣-TP association (Fig. 6B). Unexpectedly, we found that ␤1 could co-immunoprecipitate TP without the assistance of MaxiK␣ (Fig. 1I) and vice versa (Fig. 1J) and that ␤1 did not disrupt MaxiK␣-TP association (Fig. 1, B and E, white and black bars). Moreover, co-immunoprecipitation efficiency between ␤1 and TP was not modified by the additional coexpression of MaxiK␣ (Fig. 1, E (dotted and  checkered bars), I, and J). Together, these data demonstrate that ␤1-TP association can occur independently of MaxiK␣ (Fig.  1B), reminiscent of MaxiK␣-TP interaction that occurs independently of ␤1 (12). In this scenario, one can picture that in the tripartite complex, ␤1 interacts with both MaxiK␣ and TP via different regions (Fig. 6C). At present, however, we cannot discard the possibility that ␤1-TP interaction involves an intermediary protein.

A C B
MaxiKα β1 TP FIGURE 6. Model of ␤1, TP, and MaxiK␣ tripartite complex. A, model in which MaxiK␣ mediates ␤1-TP association. B, model in which ␤1 disrupts the previously demonstrated MaxiK␣-TP physical interaction (12). These possibilities are not supported by the experiments in Fig. 1. C, model in which ␤1 interacts with MaxiK␣ (18), MaxiK␣ interacts with TP (12), and ␤1 associates with both via independent mechanisms. This model is supported by the experiments in Fig. 1. Whether ␤1 interacts directly with TP or via an intermediary protein is unknown. FEBRUARY 1, 2013 • VOLUME 288 • NUMBER 5 cellular N and C termini are engaged in functional regulation of MaxiK␣ (19), whereas extracellular loop residues proximal to both TM1 and TM2 (positions 40 -45 and 152-155) are close to MaxiK␣, as interprotein disulfide bridges can be formed after appropriate cysteine engineering (18). In conjunction, these studies point to the view that the interacting sites between ␤1 and TP are distinct from those between ␤1 and MaxiK␣ and are consistent with the observation that the association efficiency of TP and ␤1 was not modified by coexpression of MaxiK␣. In the scenario that TP interacts with ␤1 through its extracellular loop via residues within positions 72-102, as suggested by our studies, MaxiK␣ could still bind/interact with ␤1 through residues proximal to TM1 and TM2 without interrupting ␤1-TP interaction.

␤1-Subunit Regulates TP-MaxiK Function
␤1 and TP-mediated Function-␤1 is an important regulatory subunit of MaxiK channels in the vasculature (7). In human coronary smooth muscle cells, single-channel voltage activation curves could be sorted in five groups, suggesting different degrees of MaxiK channel association with the ␤1-subunit, with ϳ70% of the channels likely in complex with 3-4 ␤1-subunits, ϳ25% of the channels in complex with 1-2 ␤1-subunits, and ϳ5% of the channels not in apparent association with the ␤1-subunit (13). Moreover, reduced expression of ␤1 is related to vascular dysfunction in hypertension (20,21), supporting a key role of this MaxiK channel subunit in vasoregulation.
Our present results indicate that the ␤1-subunit regulates TP-mediated MaxiK channel trans-inhibition in a way that the activated TP effect would be exerted most potently on channels formed by the ␣-subunit alone than on channels formed by ␣/␤1-subunits (Fig. 4). Still, TP activation induces a significant inhibition of channels expressing saturating levels of the ␤1-subunit (confirmed by the very negative V 1 ⁄2 value of approximately Ϫ100 mV at 6.7 M Ca 2ϩ i under control conditions) as evident from a 20 mV rightward shift of the voltage dependence of the FP o curve. By comparison, in human coronary cells, TPmediated MaxiK inhibition yielded an ϳ35 mV rightward shift of the channel voltage activation curve (12), which is consistent with the majority of MaxiK channels formed by ␣and ␤1-subunits in this smooth muscle cell type.
TP-mediated channel inhibition of cells expressing the ␣-subunit alone showed an ϳ50 mV rightward shift (Fig. 4E) of the FP o versus voltage curve, which is larger than the ϳ30 mV shift reported in our previous studies under similar experimental conditions (12). At present, we do not have a clear explanation for this difference, but we have noted that distinct batches of the commercially available TP agonist U46619 may display different potencies. Nevertheless, the present studies performed in parallel clearly show that channels formed by the ␣-subunit alone respond more effectively to U46619-induced inhibition than those formed by ␣/␤1-subunits.
We also analyzed the potential effect of ␤1-subunit expression on the end result of TP activation, vasoconstriction. Utilizing aortas from ␤1 knock-out and wild-type mice, we found a significant EC 50 reduction in U46619-induced aortic contraction (without a significant change in maximum contraction) in the absence of ␤1 expression. This result reveals a protective effect of ␤1 against thromboxane A 2 -induced vasoconstriction and also provides evidence for a physiological coupling between ␤1 and TP. A lack of change in the maximum response in ␤1 Ϫ/Ϫ versus ␤1 ϩ/ϩ mice indicates no change in the number of activated TP receptors after knocking out ␤1 expression. However, the change in EC 50 after normalization to KCl contraction reflects an intrinsic change in the TP-triggered contractile system in ␤1 Ϫ/Ϫ mice, which could comprise (i) a change in cellular signaling pathways, including decreased MaxiK activity due to loss of ␤1, and/or (ii) a change in TP agonist affinity due to loss of ␤1-TP interaction. Interestingly, norepinephrine-induced aortic contraction normalized to KCl contractile response has been shown to be identical in ␤1 Ϫ/Ϫ and ␤1 ϩ/ϩ animals (8). It would be interesting to determine whether ␤1 associates or not with ␣-adrenergic receptors.
In summary, ␤1 not only modifies MaxiK channel voltage/ calcium sensitivity but also regulates TP-mediated channel inhibition and vasoconstriction. ␤1-TP association might be a force to dampen the inhibitory signal from TP to MaxiK␣ and to modify TP vasoconstricting potency.