Four-turn α-Helical Segment Prevents Surface Expression of the Auxiliary hβ2 Subunit of BK-type Channel*

Large conductance, voltage- and Ca2+-activated K+ (BK) channels encoded by the mslo α and β2 subunits exist abundantly in rat chromaffin cells, pancreatic β cells, and DRG neurons. The extracellular loop of hβ2 acting as the channel regulator influences the rectification and toxin sensitivity of BK channels, and the inactivation domain at its N terminus induces rapid inactivation. However, the regulatory mechanism, especially the trafficking mechanism of hβ2, is still unknown. With the help of immunofluorescence and patch clamp techniques, we determine that the hβ2 subunit alone resides in the endoplasmic reticulum, suggesting that trafficking mechanism of hβ2 differs from that of hβ1 opposite to what we predicted previously. We further demonstrate that a four-turn α helical segment at the N terminus of hβ2 prevents the surface expression of hβ2, that is, the helical segment itself is a retention signal. Using the c-Myc epitope-tagged extracellular loop of hβ2, we reveal that the most accessible site by antibody is located at the middle of the extracellular loop, which might provide clues to understand how the auxiliary β subunits regulates the toxin sensitivity and the rectification of BK-type channels.

Large conductance voltage-and Ca 2ϩ -activated K ϩ (BK) 4 channels associated with its auxiliary ␤ subunits show functional varieties in many native cells (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13). The auxiliary ␤1-␤4 subunits of BK channels are composed of two transmembrane segments and a large extracellular segment (5, 14 -16). For h␤2 subunits, three hydrophobic residues FIW at its N terminus produces a rapid inactivation of BK channels (15), whereas it does not follow a typical N-type inactivation mechanism (17). The NMR experiment reveals a four-turn helix structure located at the middle of its N terminus (24). Recently, Zhang et al. (18) reported that trypsin can easily target the Arg 8 and Arg 9 of nine basic residues before the helix. In addition, its extracellular loop prevents the scorpion toxin charybdotoxin from approaching the channel pore (9) and induces the rectification of BK channels (19). It has been reported that the ␤1 and ␤2 subunits contain endocytic sorting signals at their C termini, which can down-regulate the association MaxiK channel surface expression levels (14,20). The ␤1 subunit can reach the cell membrane alone, whereas the ␤2 subunit only appears underneath the plasma membrane (14,20,21). Obviously, the trafficking mechanism of h␤2 subunits is still unknown so far.
There are many mechanisms for governing the surface expression of ion channels and thus the electrical activity of a cell. A channel protein in a misfolded or unfolded state can be trapped in the ER as demonstrated by measurements of the surface labeling, luminomitry. Some retrieval/retention and anterograde signals determine the ER exit of proteins (22). The retention sequence RKR appearing in potassium inward rectifying channel (Kir) and its auxiliary SUR1 subunit is due to improperly folded or assembled channels (22,23). Both the Kir and SUR1 subunits have the cytosolic RKR sequences, which must be masked by the assembly of octamer before channels can be transported to the cell surface (23). Zarei et al. (24) also report that a nonbasic hydrophobic retention/retrieval motif CVLF prevents the surface expression of mslo ␣ and ␤1-subunits.
With the help of immunofluorescence and patch clamp techniques, we find that the auxiliary h␤2 subunit cannot traffic to the cell surface without associating with the mslo ␣ subunit. Fortunately, the h␤2 subunit can solely traffic to the cell surface only after demolishing or destroying a four-turn ␣ helical segment of its N terminus (25), which may implicate a novel trafficking mechanism. We also find that the most accessible region by antibody is located at the middle position of the h␤2 outer loop near the channel filter, suggesting that it may be the important site in regulating the toxin sensitivity and rectification of inactivating BK currents.

MATERIALS AND METHODS
Constructions and Mutagenesis-The full-length cDNA for each of the mslo ␣ or h␤2 were subcloned into pcDNA3.1/Zeo(ϩ) (Invitrogen). For the construct h␤2-Tdimer2, h␤2 cDNA, and Tdimer2 cDNA were subcloned into pcDNA3.1/Zeo(ϩ). The BK pore-forming ␣ subunit (mslo) was tagged with the c-Myc epitope (EQKLISEEDL) at the N terminus. To get a green fluorescence protein targeted ER, pEGFP-ER, the enhanced green fluorescence protein (EGFP) was used to substitute the DsRed2 in pDsRed2-ER (Clontech). The human ␤ subunits (h␤1/h␤2) were tagged with epitope (c-Myc, EQKLISEEDL; hemagglutinin, YPYDVPDYA) at different positions of the putative extracellular loop by sequential overlap extension PCR. Series of constructs for deletions of the h␤2-137myc N terminus were made by deleting appropriate amino acids from h␤2-137myc and introducing a new start codon with PCR primers. The mutations on h␤2-137myc were created with the QuikChange site-directed mutagenesis kit (Stratagene). All of the constructs and mutations were verified by direct DNA sequence analysis.
Immunofluorescence Imaging in HEK293 Cells-HEK293 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 IU/ml penicillin, and 100 IU/ml streptomycin. The cells were transiently transfected using Lipofectamine 2000 (Invitrogen). After transfection, the live cells were transferred to a poly-D-lysine-coated chamber. The next day, the cells were fixed with 2% paraformaldehyde in phosphate-buffered saline for 5 min. For cell permeabilization, 0.2% Triton X-100 was added for 5 min. After blocked with 5% goat serum for 1 h, the cells were incubated with a monoclonal anti-human c-Myc antibody (1:200) for 3 h, washed, and incubated with fluorescein isothiocyanate-conjugated anti-mouse IgG (HϩL) (1:150) for 1 h. All of the experiments were performed at room temperature (22-25°C). Mouse anti-human c-Myc and fluorescein isothiocyanate rabbit anti-mouse IgG (HϩL) conjugate were purchased from Zymed Laboratories Inc.
We took advantage of the high numeric aperture objective (APO ϫ100 OHR, NA ϭ 1.65, Olympus) to take high resolution fluorescence images of transfected cells. Excitation light from a fiber optical coupled monochromator (Polychrome IV; TILL Photonics GmbH, Germany) was passed through a shutter that opened only during camera exposure. The wavelength selection and switch were controlled by the image acquiring software (TILL vision 4.0; Till Photonics GmbH). The images were acquired with a cooled CCD (PCO SensiCam; Germany) with pixel size of 0.067 m at the specimen plane.
Image Analysis and Statistics-Images were viewed, processed, and analyzed in TILL Vision (TILL.Photonics, Germany) and Adobe Photoshop (Adobe Systems) and IMAGE J (National Institutes of Health, Public Domain). Electrophysiology-Patch pipettes were pulled from borosilicate glass capillaries with a resistance of 2-4 megohms when filled with pipette solution. Inside-out patches were used in experiments. During recordings, 0 and 10 M Ca 2ϩ solutions were applied onto the cell via a perfusion pipette containing seven solution channels. All of these experiments were performed at room temperature (22-25°C). The experiments were performed using a PC2C patch clamp amplifier with its software (InBio). The currents were typically digitized at 100 kHz. Macroscopic records were filtered at 5 kHz. Patch clamp recording data were analyzed with IGOR (Wavemetrics, Lake Oswego, OR), Clampfit (Axon Instruments, Inc.), and Sigmaplot (SPSS, Inc.) software.

RESULTS
Auxiliary h␤2 Subunits Alone Reside in the ER-The h␤ subunits of a four-member ␤ family (h␤1-h␤4) have the similar structure, i.e. with two transmembrane segments, an extra large loop, and a few consensus glycosylation sites (26) (Fig. 1A). The h␤2 subunits like other ␤ subunits have many ways to regulate the mslo BK channels. For example, its extracellular loop alters the conductance and toxin sensitivity of BK channels, but its N terminus introduces a rapid inactivation of BK channels (9,17,19). The h␤1 subunits have been reported and can traffic to the cell surface alone (14). However, it is unclear whether the auxiliary h␤2 subunits also traffic to the plasma membrane as h␤1 subunits do. To investigate the trafficking mechanism of auxiliary h␤2 subunits, a c-Myc epitope was tagged to the center of the loop for measuring the surface expression of h␤2 subunits (Fig. 1A). Considering that recognition signals (export and retention/retrieval) of proteins in the ER are mostly found at the N or C termini, we constructed a series of mutations at the N terminus to investigate the trafficking mechanism of the h␤2 subunits (Fig. 1B). A three-dimensional structure of the N-terminal helix from 18 to 31 is plotted in the Fig. 1C (25).
Because coexpression of mslo ␣ and h␤2 subunits can make the inactivating functional BK channels, we expected that images for coexpression of Myc-mslo (green) and h␤2-Tdimer2 (red) should show their coexistence in HEK293 cells. In fact, images for Myc-mslo, h␤2-Tdimer2, and Myc-msloϩh␤2-Tdimer2 show green (left), red (middle), and yellow (right) brims, respectively, which indicates coexistence of two proteins on the cell surface ( Fig. 2A). In contrast, there was no red contour distinctly appearing on the cell brim, whereas the h␤2-Tdimer2 was transfected alone in HEK293 cells. The image of coexpression of pEGFP-ER with h␤2-Tdimer2 demonstrates that two proteins coexist in the ER (Fig. 2B). Furthermore, there was no detectable colocalization with coexpression of the h␤2-Tdimer2 with an EGFP-tagged Golgi marker (data not shown) that meant the h␤2 subunit should be retained in ER. Because the Tdimer2 is a protein with more than 450 amino acids, it may influence the traffic of h␤2. To precisely recognize and quantify surface expression of h␤2 subunits, a c-Myc epitope was tagged at the loop. Similarly, the image for coexpression of mslo with h␤2-137myc revealed that the h␤2-137myc subunits were transported to the membrane surface (Fig. 2C, right panel). It is interesting that the surface signals (green) of the h␤2-137myc subunits alone could be detected only under permeabilized conditions (Fig. 2C, left and middle panels). In other words, the green signals were undetectable under nonpermeabilized conditions, whereas the h␤2-137myc subunits were transfected alone in HEK293 cells (Fig. 2C, left panel). In conclusion, a surprising finding is that the h␤2 subunit cannot traffic to the plasma membrane alone except that it expresses with the mslo ␣ subunits together.
The h␤2 Subunit Is Restricted in the ER by Four-turn ␣-Helical Segment-To look for the retention signals of h␤2 subunits, we deleted the C terminus (⌬C) first and then the initial three hydrophobic residues FIW (⌬FIW). In Fig. 3A, mutations ⌬C and ⌬FIW failed to relieve h␤2 from being retained in ER, but they did traffic to the cell surface after coexpressing with the mslo subunits (data not shown). After deleting the first 10, 20, and 30 amino acids of N terminus, we found that the construct ⌬30A was the only one appearing on the cell surface, which illuminates that the retention signals of the h␤2 subunits must be concealed among the residues ranging from 20 to 29. Examining their fractional intensities plotted in Fig. 3D, we further confirmed that only ⌬30A could traffic to the cell surface.
The constructs ⌬23A, ⌬26A, and ⌬30A were designed to precisely determine the retrieval signal or domain. After gradually decreasing the intensity of retention signals, we determined that the retrieval signal is a motif of 23-29 amino acids (Fig. 3B). To further determine the exact sequence of retention signals, the mutations ⌬KIR and ⌬DHDL were designed to relieve the retention signal of ER. In Fig. 3B, the construct ⌬KIR is clearly a better mutation than the construct ⌬DHDL for that purpose. It indicated that the motif KIR played a major role in trapping h␤2 in the ER, which is also consistent with the statistical data of the fractional intensities shown in Fig. 3E.
Because alanine-scanning mutations from Q23A to L30A did not show any remarkable surface expression (data not shown), we thus inferred that the motif KIR might be only a core of an expanded sequence. Considering that the motif KIR is located within a four-turn ␣ helical segment ranging from residues 18 to 31 (25), we tested whether the whole helix served as a retention signal. A series of mutations, i.e. 23A-26A, 24A-26A, 24A-27A, and 23A-27A, were made to flatten the helix (Fig.  3C). Furthermore, mutant Q23P,D27P was constructed to destroy the helix by two prolines, because proline residue is usually thought to promote distortions of transmembrane helices (27,28) (Fig. 3C). In Fig. 3 (E and F), the expressions of 23A-27A is about 30% more than that of 24A-26A or ⌬KIR, which indicates that the motif KIR plays a role in trafficking through its secondary structure. A, topology of h␤2 subunits exhibits a putative structure, which consists of a N terminus, two transmembrane domains (TM1 and TM2), a long extracellular loop, and a short C terminus. The N terminus contains three components: an inactivation "ball" domain, am ␣-helical segment (dark gray), and two flexible linkers. Four conserved cysteines (mediumdarkgray)intheextracellularlooparelocatedtopositions84,113,142,and174,respectively.Inmostexperiments,ac-Myc epitope(lightgray)wasinsertedatposition137oftheextracellularloop.B,aseriesofN-terminalmutationsandconstructs of h␤2 subunits is shown in the list. The residues 18-31 of the ␣ helical segment are labeled in light gray, and the substituting residues are in dark gray. C, a three-dimensional structure of the four-turn helix from 18 to 31 of h␤2 subunits gives more detailed orientation of residues. The hydrophobic residue is labeled in light gray, the negatively charged one is in dark gray, and the positively charged one is in medium-dark gray.  FEBRUARY 1, 2008 • VOLUME 283 • NUMBER 5

JOURNAL OF BIOLOGICAL CHEMISTRY 2711
To further demonstrate whether the retention signal is a fixed sequence or a secondary structure, we inserted the QKIRD sequence into the N terminus of h␤1 subunits i.e. QKIRD-h␤1 (Fig. 3, C and G). The mutant QKIRD-h␤1 of h␤1 did not show any clear impact on the surface expression of h␤1 subunits. Therefore, the retention signal of h␤2 subunits is the four-turn helix itself rather than the primary sequence QKIRD. Furthermore, the motif KIR also appears in the ␤E of the C terminus of mslo subunits (29), which is obviously not a retention signal of BK channels.
To test whether the retention signal at the N terminus was able to present at other locations of h␤2, we translocated the potential retention signal, i.e. the first 33 amino acids of N terminus, to the C terminus of h␤2 after Arg 235 (C33). The C33 alone shows little expression, whereas the mSloϩC33 only shows a little reduction compared with that of the wild type h␤2 (Fig. 4). It means that the same retention signal presents at the new location. We thus believe that the four-turn helix structure at the N terminus is a retention signal of h␤2.

Accessibility of the Extracellular Segment of h␤2 for Antibody Varies with Loci of c-Myc Epitope-We
first define the c-Myc antibody accessibility at different positions in the outer segment/loop of h␤2. The differential fluorescence intensity of c-Myc antibody at positions in the h␤2 outer loop may arise from either the cysteine-rich and highly glycosylated structure of the loop or restrictions on access of the big antibody molecule to small volumes. The stronger fluorescence intensity means the c-Myc antibody molecule is easier to reach and bind with its antigen through a more spacious pathway. We thus examined the position dependence of antibody accessibility by inserting c-Myc antigen at different sites in the loop, which was used to provide the detailed structural conformation of the extracellular segment of h␤2.
In addition to the site Lys 137 in the loop, c-Myc epitope was also introduced either at the C terminus or at the sites 72, 76, 114, 126, 158, and 185 in the loop. However, Lys 137 is the only site with the clear c-Myc signals in either h␤2 alone or h␤2 with mslo together (Table 1). It is abnormal that we cannot see the c-Myc signals at positions 76 and 158 with coexpression of mslo with h␤2, in the presence or absence of Triton X-100 (Table 1). This may suggest that the mslo ␣ subunit can affect the c-Myc accessibility of h␤2.
To remove the interference of mslo, the construct ⌬33A was selected for this purpose. In this construct, the c-Myc epitope was introduced at the positions 76, 100, 114, 126, 137, 158, and 185 in the loop, respectively. In this work, we found that the positions 126 and 137 are two easiest sites for access of antibody based on their fluorescence intensities, whereas ␤2 expressed alone ( Fig. 5 and Table 1). It indicates that the fluorescence signal at position 126 disappears in the coexpression case. Thus, the c-Myc signal of the h␤2 loop can be covered by the ␣ sub-unit, specially, by the P-loop of S5-S6 while coexpressed with ␣ subunits.
Because the higher fractional intensity means the higher accessibility of the loop, the most accessible site was plotted at the highest level of the extracellular loop, the second was plotted at the second highest level, and so on. In this way, the "structural" profile of the extracellualr loop can be depicted in a letter "W," of which the apex was located around the middle of the loop (Fig. 5C). On the basis of its hydrophobicity plot (9), the hypothetical topology of h␤2 has two transmembrane domains with a large extracellular loop, of which positions 76 and 185 are just located at the N and C termini of the extracellular loop, respectively. Therefore, they should be the easier sites to be approached by antibody. Moreover, the central part of the loop is usually considered to be difficult to approach because it lays over the channel pore and deep into the cell surface (30,31). According to the above rule of accessibility, the structural profile of the extracellualr loop can be depicted in a letter "M," of which the valley is around the middle too. To our great surprise, A, the images for HEK293 cells transfected with h␤2-137myc or msloϩh␤2-137myc. B, the images for HEK293 cells expressed with C33 or msloϩC33 (The first 33 amino acids of h␤2 N terminus were translocated to its C terminus after Arg 235 .) C, the percentage intensity of h␤2, msloϩh␤2, C33, and msloϩC33 are 12.8 Ϯ 0.9 (n ϭ 6), 100.0 Ϯ 0.0 (n ϭ 95), 13.0 Ϯ 5.6 (n ϭ 8), and 72.4 Ϯ 13.9 (n ϭ 63), respectively. The scale bar indicates 5 m.  To make it simpler, the representative traces are only coming from some of the mutations (Fig. 6). Typically, mutations with the initial three hydrophobic residues FIW at their N termini always show a rapid inactivation with a time constant of about 25 ms at 100 mV in 10 M Ca 2ϩ . The rest exhibits the larger inward currents at negative voltages, which indicates that there is a huge negative shift (usually more than 50 mV) in conductance-voltage (G-V) curves relative to the G-V curve of mslo ␣ channels. The above results indicate that the h␤2 including its mutations that we constructed in this study can traffic onto the membrane surface of cells after being functionally associated with mslo ␣ subunits.

DISCUSSION
Among ␤-subunit trafficking, ␤1 subunit can traffic to the cell surface alone (14,21), whereas ␤2 subunit can reach the plasma membrane with the ␣ subunit together. Through the biotinylation experiments, Jin et al. (32) reported that the ␤4 subunit could target to the cell membrane based on the existence of double glycosylation sites in ␤4 subunits. There is no information about trafficking about ␤3 subunit so far. Furthermore, Zarei et al. (14,20) observed a putative endocytic signal in the C terminus of the h␤2, which reduced the surface expression of the coexpressed hslo subunits. With polyclonal antibody against ␤2 subunit using the permeabilized labeling protocol, they found that the h␤2 alone appeared to reach the plasma membrane. We infer that the h␤2 alone stays on the ER membrane just underneath the plasma membrane.
In this study, we were attempting to determine a retention motif in h␤2 like the RXR in potassium inward rectifying channel (Kir) and its auxiliary SUR1 (22,23) or the CVLF in BK splice variant (24). There are many reasons against KIR being a retention signal. For instance, the surface expression of ⌬29A is more than that of ⌬26A; QKIRD-h␤1 or Q23PD27P can traffic to membrane surface. Therefore, the second structure of the h␤2 N-terminus is our exclusive choice.
Using NMR spectroscope, Bentrop et al. (25) investigated the solution structure of the h␤2 N terminus (amino acids 1-45, BK␤2N). The BK␤2N structure comprises two domains connected by a flexible linker: the ball domain (formed by residues 1-17) and the chain domain (between residues 20 -45) linking it to the membrane segment of h␤2. The chain domain consists of a four-turn helix including a 3 10 -helix (between residues 18 -31) with an unfolded linker at its C terminus. The BK␤2N with a four-turn helix is a properly folded peptide as it can functionally inactivate BK channels when applied to the cytoplasmic side (25). Furthermore, our results suggest that the retention signal may be derived from protein-protein or protein-lipid mechanism. Because there is no report about the specific anchoring protein of h␤2, we would rather consider a protein-lipid mechanism. From the three-dimensional structure of the four-turn helix shown in Fig. 1C (25), we found that the positively charged residues (Lys 18 , Arg 19 , and Arg26) were clustering on one side, and the hydrophobic residues (Ile 21 /Tyr 22 and Leu 30 /Leu 31 ) are lining on each side. Considering that the head group of phospholipid is negatively charged and the tail domain is hydrophobic, we infer that the resultant force coming from electric field and hydrophobic interaction may distort the helix itself to lead the h␤2 subunit coming off the membrane of the ER completely. Furthermore, the positively charged residues Lys and Arg seem to play a crucial role, because the reduction in the number of positively charged residues would somehow weaken the protein-lipid interaction.
It is well known that trypsin can attack and cleave any exposed basic residue in the h␤2 N terminus. Zhang et al. (18) reported that the first four basic residues, i.e. Arg 8 , Arg 14 , Lys 18 , and Arg 19 , had the highest trypsin accessibility of basic residue. During h␤2 assembling with mslo in the ER, the C terminus of mslo ␣ subunit may somehow wrap the N-terminal helix of h␤2 to avoid trypsin attack (18). Correspondingly, ␤ subunits reciprocally modulate many properties of BK ␣ channels by the looploop interaction of ␣ and ␤ subunits. For instance, the BK ␣ϩh␤4 channel is becoming very resistant to CTX or IbTX because of the glycosylation in the loop domain of h␤4 subunits, which prevents the access of toxins to the channel vestibule (32). The outward rectification of BK ␣ϩh␤2 channels can be demolished completely by the c-Myc insert h␤2-137myc (data not shown). This study may provide us a new way to understand the interaction of the BK ␣ subunit and ␤ subunit.
The extracellular loop of h␤2 plays an important role in preventing the toxin from approaching the pore of BK channels (9) and producing the rectification of BK channels (19). However, the lack of the structural information about the extracellular FIGURE 6. All of the BK channels encoded with mslo and mutations of h␤2 subunits make functional currents. All of the currents encoded by mslo and h␤2 mutation subunits were recorded in inside-out patches in presence of intracellular 10 M Ca 2ϩ . The traces of msloϩh␤2-mutations were evoked by a voltage step from a holding potential of Ϫ180 mV to voltages ranged from Ϫ150 to 150 mV in an increment 20 mV. The scale bars represent 30 ms (time) and 2 nA (current), respectively. The voltage protocol is plotted on the upper left.
loop prevents us from precisely determining the interacting or binding sites between the loop of h␤2 and the pore of BK channels. Our results reveal that three locations at 137, 100, and 185 represent the peak and valleys of accessibility of the extracellular loop, respectively, which may provide us with important clues to explore the interacting or binding sites in the future. In addition, four conserved cysteines influence the rectifying and pharmacological characteristics of BK channels via forming disulfide bridges to constitute a specific structure in the extracellular loop, that can be disrupted by 20 mM extracellular dithiothreitol (19). Two cysteine mutations (Cys 2 and Cys 3 ) were constructed for this purpose, whereas we did not find any changes in accessibility. Hopefully, a complete removal of four conserved cysteines may provide more clues for re-evaluating the accessibility of c-Myc antibody to the extracellular loop in the future.