Phosphorylation-dependent Functional Coupling of hSlo Calcium-dependent Potassium Channel and Its hβ4 Subunit*

The auxiliary β4 subunit of the humanslowpoke calciumdependent potassium (slo) channel is expressed predominantly in the brain. Co-expression of β4 subunit with the slo channel α subunit in HEK293 and Chinese hamster ovary cells slows channel activation and deactivation and also shifts the voltage dependence of the channel to more depolarized potentials. We show here that the functional interaction between the hβ4 subunit and theslo channel is influenced by the phosphorylation state of hβ4. Treatment of cells with okadaic acid (OA) reduces the effect of hβ4 on slo channel activation kinetics and voltage dependence but not on slo channel deactivation kinetics. The effect of OA can be blocked by mutating three putative serine/threonine phosphorylation sites in hβ4 (Thr-11/Ser-17/Ser-210) to alanines, suggesting that OA potentiates phosphorylation of hβ4 and thereby suppresses its functional coupling to the slochannel. Mutation of Ser-17 alone to a negatively charged residue (S17E) can mimic the effect of OA. Mutating all three phosphorylation sites in hβ4 to negatively charged residues (T11D/S17E/S210E) not only suppresses the effect of hβ4 on slo channel activation kinetics and voltage dependence, it also suppresses its effect on slo channel deactivation kinetics. Co-immunoprecipitation/Western blot experiments indicate that all of these hβ4 mutants, as well as the wild-type hβ4, bind to theslo channel. Taken together, these data suggest that phosphorylation of the β4 subunit dynamically regulates the functional coupling between the β4 subunit and the pore-forming α subunit of the slo channel. In addition, phosphorylation of different residues in hβ4 differentially influences its effects onslo channel activation kinetics, deactivation kinetics, and voltage dependence.

The auxiliary ␤4 subunit of the human slowpoke calciumdependent potassium (slo) channel is expressed predominantly in the brain. Co-expression of ␤4 subunit with the slo channel ␣ subunit in HEK293 and Chinese hamster ovary cells slows channel activation and deactivation and also shifts the voltage dependence of the channel to more depolarized potentials. We show here that the functional interaction between the h␤4 subunit and the slo channel is influenced by the phosphorylation state of h␤4. Treatment of cells with okadaic acid (OA) reduces the effect of h␤4 on slo channel activation kinetics and voltage dependence but not on slo channel deactivation kinetics. The effect of OA can be blocked by mutating three putative serine/threonine phosphorylation sites in h␤4 (Thr-11/Ser-17/Ser-210) to alanines, suggesting that OA potentiates phosphorylation of h␤4 and thereby suppresses its functional coupling to the slo channel. Mutation of Ser-17 alone to a negatively charged residue (S17E) can mimic the effect of OA. Mutating all three phosphorylation sites in h␤4 to negatively charged residues (T11D/S17E/S210E) not only suppresses the effect of h␤4 on slo channel activation kinetics and voltage dependence, it also suppresses its effect on slo channel deactivation kinetics. Co-immunoprecipitation/Western blot experiments indicate that all of these h␤4 mutants, as well as the wild-type h␤4, bind to the slo channel. Taken together, these data suggest that phosphorylation of the ␤4 subunit dynamically regulates the functional coupling between the ␤4 subunit and the pore-forming ␣ subunit of the slo channel. In addition, phosphorylation of different residues in h␤4 differentially influences its effects on slo channel activation kinetics, deactivation kinetics, and voltage dependence.
Large conductance Ca 2ϩ -dependent potassium (K Ca or maxi K) channels are ubiquitously expressed in neurons and many other tissues. They contribute to action potential repolarization and influence neurotransmitter release (1)(2)(3)(4). Although the native K Ca channels exhibit diverse phenotypic properties, there is only a single gene encoding this potassium channel. The slowpoke gene that encodes the K Ca channel has been cloned in many species (5)(6)(7)(8). Alternative mRNA splicing of this gene can potentially generate a large number of distinct channel isoforms (9). Yet another mechanism of generating K Ca channel diversity is to associate the pore-forming ␣ subunit with a group of auxiliary ␤ subunits (10 -15). The ␤ subunits influence such diverse aspects of K Ca channel function as kinetic behavior, voltage dependence, and sensitivity to toxins and modulators (11, 13, 16 -19).
Like other types of potassium channels, as well as sodium and calcium channels, both native and recombinant K Ca channels are modulated by post-translation modification such as phosphorylation and oxidation/reduction (20 -25). Furthermore, K Ca channels from brain have been shown to be intimately associated with protein kinase and phosphoprotein phosphatase activity (4, 20, 26 -28). There is considerable variation in the sensitivity of different types of K Ca channel to protein kinase modulation (24,29). This variation could be due to different subunit composition of native K Ca channels, providing different substrates for post-translational modification (11,14,19,30,31). In the case of voltage-activated potassium (Kv) channels and L-type calcium channels, it is known that the ␤ subunits of these channels are substrates for protein kinases (32)(33)(34)(35). It has been shown previously that the human ␤4 subunit influences slo channel activation kinetics and voltage dependence (11,13,15). In the present study we investigated the effect of phosphorylation of the ␤4 subunit on its functional interaction with the slowpoke ␣ subunit.

EXPERIMENTAL PROCEDURES
Cloning and Transient Expression in CHO 1 Cells-h␤4 was cloned into the pIRES2-EGFP vector (CLONTECH), a bicistronic vector that allows co-expression of h␤4 and green fluorescent protein in the same cell. We used the Quick-change protocol (Stratagene) to make point mutations in the ␤4 subunit. Briefly, a PCR was performed, using the wild-type ␤4 as the template and a pair of complementary mutagenesis primers. The PCR mixture was then cut with the enzyme DpnI to digest the template wild-type ␤4. After DpnI digestion, the PCR product was used to transform competent bacterial cells, and the mutant plasmid of ␤4 was then amplified. All mutant constructs were verified by sequencing (University of Pennsylvania Sequencing Facility, Philadelphia).
CHO cells were maintained in Ham's F-12 nutrient mixture supplemented with 10% fetal bovine serum and penicillin/streptomycin. Cells were seeded on 35-mm culture dishes and transfected 2 days later with the appropriate slo and ␤4 DNA using the FuGENE 6 transfection reagent (Roche Molecular Biochemicals).
Electrophysiology-CHO cells were used for recording 1-3 days after transfection. The slo channel was cloned into the pcDNA3 vector. The ␤4 subunit was cloned into the pIRES2-EGFP vector (CLONTECH). Cells were transfected with both constructs, and transfected cells were identified by their green fluorescence; all such cells were found to express slo current. Recording electrodes were pulled to have resistances of 1.5-2 megohms when filled with regular pipette solution. slo current was recorded in the whole-cell recording mode. Solutions for the * This work was supported by grants from Millennium Pharmaceuticals and the National Institutes of Health (to I. B. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Recording was performed using an Axopatch 200A amplifier and pClamp8 software (Axon Instruments, Foster City, CA). Data were analyzed off line using pClamp8 software. For measuring the time course of slo channel activation and deactivation, the cells were held at Ϫ80 mV, and the membrane potential was stepped to ϩ40 mV for 200 ms and then back to Ϫ80 mV. For measuring the current-voltage relationship of the slo channel, the cells were held at Ϫ80 mV, and depolarizations in steps of 20 mV were then applied for 200 ms. The maximum instantaneous tail current after stepping the membrane potential back to Ϫ80 mV was measured. To quantify the activation kinetics, the activation of the slo current was fitted with a two-exponential function, giving rise to a fast and a slow time constant (). Similarly the deactivation current was fitted with a two-exponential function. The half-activation voltage (V1 ⁄2 ) was derived by fitting the current-voltage relationship curve with a Boltzmann function. All results were expressed as mean Ϯ S.E. Statistical significance was assessed using the Student's t test in SigmaPlot software.
Co-immunoprecipitation and Western Blot-These experiments were done as described previously (11,36). Epitope-tagged h␤4 in pCDNA3.1 was expressed in HEK293 cells, either alone or together with hSlo. 36 h after transfection, the cells were lysed and incubated with antibody raised against mSlo that also recognize hSlo. 2 The immune complexes were precipitated by incubation with protein A/G PLUS-agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA). Proteins in the lysate or immunoprecipitate were separated on polyacrylamide gels and transferred to a nitrocellulose membrane. After blocking with 5% nonfat milk in TBST (0.1% Tween 20 in Tris-buffered saline), the blots were probed with primary antibody directed either against mSlo or the V5 epitope tag. Horseradish peroxidase-coupled anti-V5 (Invitrogen) was used as secondary antibody for h␤4 blots. Horseradish peroxidase-coupled donkey anti-rabbit IgG (Amersham Biosciences) was used as the secondary antibody for hSlo blots. The membrane was washed with TBST, and proteins were visualized with an enhanced chemiluminescence detection system (Amersham Biosciences). 32 P Labeling of ␤4 -HEK293 cells were transfected with hSlo and h␤4 as described earlier. Eight hours after transfection, [ 32 P]orthophosphate was added into the culture medium to a final concentration of 0.1 mCi/ml. The cells were labeled for 12 h with 32 P and then lysed. hSlo was immunoprecipitated, and the immunoprecipitated proteins were separated on a polyacrylamide gel. 32 P-Labeled proteins, including h␤4, were detected by autoradiography.

Time-dependent Change of slo Channel Activation Kinetics in the Whole-cell Recording Configuration-
We showed previously that ␤4 slows slo channel activation in the inside-out recording configuration (11). Similar modulation of slo channel activation by ␤4 was observed in the whole-cell recording configuration ( Fig. 1, A and B). However, this effect of ␤4 on slo channel activation kinetics gradually decreased with time after going into whole-cell mode. In cells expressing both slo and ␤4, the time constant of slo channel activation decreased over the 3-min recording period, from an initial value of 3.4 ϩ 0.6 ms to a plateau level of about 0.8 ms (Fig. 1E). In contrast, in cells expressing slo alone, the time constant of slo channel activation stayed unchanged over the same recording period ( Fig. 1, C, D, and E). We have found a two-exponential function was necessary to fit accurately the time course of slo channel activation and deactivation. Because both the fast and the slow time constants were modulated similarly in the following experiments, as summarized in Table I, we have presented only the fast time constants in the figures.
Co-expression with ␤4 also slowed slo channel deactivation in the whole-cell recording mode. Unlike the time-dependent change of slo channel activation in the presence of ␤4, the time constant of slo channel deactivation did not change with time ( Fig. 1, A, B, and F). These data suggest that, in the presence of ␤4, there was a time-dependent change of slo channel acti-vation kinetics attributable to the whole-cell recording configuration. This change with time in the presence of ␤4 was unlikely to be due to the time required for Ca 2ϩ equilibration, which is much faster. Note that the time-dependent change in activation time constant was seen only in cells expressing both slo and ␤4 and not in cells expressing slo alone (Fig. 1E). In addition, the time constant of slo channel deactivation did not change with time in the absence or presence of ␤4 (Fig. 1F). All of these data suggest that the concentration of intracellular free Ca 2ϩ was equilibrated in less than 30 s. The same timedependent change in slo channel activation was also observed when the slo channel was co-expressed with mutant h␤4 with three serine/threonine residues mutated to alanines (T11A, S17A, S210A). The mechanism of this time-dependent change in slo channel activation has yet to be investigated, but it is unlikely that phosphorylation of h␤4 is responsible for this phenomenon. For the purpose of the present study, we focused on slo channel function only during the first 30 s after achieving the whole-cell recording mode.
Inhibition of Phosphoprotein Phosphatase with Okadaic Acid Suppresses the Effect of ␤4 on slo Channel Activation-Sequence analysis using ScanProsite revealed three potential serine/threonine phosphorylation residues in h␤4 ( 11 TEAE, 17 SIR, and RKFS 210 ). The threonine at position 11 is a consensus casein kinase II phosphorylation site; the serine at position 17 is a protein kinase C site; and the serine at position 210 is a PKA site. We were interested in whether phosphorylation of ␤4 could modulate its functional interaction with the slo channel. Preincubating cells expressing both slo and ␤4 in 0.5 M okadaic acid (OA) for 10 -30 min largely blocked the effect of ␤4 on channel activation ( Fig. 2A). This difference in channel activation kinetics is better revealed in the normalized current (normalized to the maximum current amplitude) as shown in Fig.  2B. Treatment of cells expressing both ␤4 and slo with OA decreased the time constant of slo channel activation, although the time constant of activation is still greater than that in cells expressing slo alone. These differences in slo channel activation are quantified in Fig. 2C. In cells expressing both slo and ␤4, OA treatment reduced the time constant of slo channel activation from 2.3 Ϯ 0.2 to 1.5 Ϯ 0.2 ms (p Ͻ 0.001) (Fig. 2C). The same OA treatment in cells expressing slo alone did not significantly change channel activation (p Ͼ 0.5) (Fig. 2C).
The effect of OA treatment on slo channel activation kinetics could be due to a change in the phosphorylation state of either the slo channel or the ␤4 subunit. We tried to distinguish these two possibilities by mutating all three putative phosphorylation residues in h␤4 to alanines (the ␤4 AAA mutation). The ␤4 AAA mutant increased the activation time constant to the same extent as the wild-type ␤4 (p Ͼ 0.05) (Fig. 2D). Interestingly, in cells expressing both slo and ␤4AAA, OA treatment did not alter the effect of ␤4 AAA on slo channel activation (Fig. 2D).
Mutation of the h␤4 Serine/Threonine Phosphorylation Sites to Negatively Charged Residues Suppresses the Effect of ␤4 on slo Channel Activation-The results with okadaic acid treatment and ␤4 AAA mutation suggest that phosphorylation of ␤4 could dynamically modulate the functional interaction between ␤4 and the slo channel. To test this hypothesis further, and to determine the critical phosphorylation site in ␤4, we mutated the three potential serine/threonine phosphorylation sites individually and in combination. Mutation to alanine removes any possibility of that residue being phosphorylated, and mutation to a negatively charged amino acid is assumed to mimic phosphorylation. Cells were transfected with slo and wild-type ␤4 or various mutant ␤4s, and the time constant of slo channel activation was measured. As shown in Fig. 3A, mutating all three phosphorylation sites to negatively charged residues (␤4 DEE) largely suppressed the effect of ␤4 on slo channel activation. The single mutation ␤4 S17E was sufficient to suppress the effect of ␤4 on slo channel activation. Neither the single mutation T11D nor the single mutation S210E had any effect on the functional interaction between ␤4 and the slo channel (data not shown). These data are quantified in Fig. 3B. The triple mutation ␤4 DEE decreased the effect of ␤4 on the time constant of slo channel activation; the time constant of activation was 2.3 Ϯ 0.2 ms for the wild-type ␤4 and 1.0 Ϯ 0.1 ms for the ␤4 DEE mutation (p Ͻ 0.001) (Fig. 3B). The time constant of activation for the triple mutation ␤4 DEE was still greater than that for slo alone (p Ͻ 0.002). The single mutation ␤4 S17E suppressed the effect of ␤4 on channel activation; in the presence of ␤4 S17E the activation time constant was 1.6 Ϯ 0.2 ms (p Ͻ 0.01) (Fig. 3B). This effect of the single mutation ␤4 S17E is similar to that of OA treatment (p Ͼ 0.7) (compare Figs. 2B and 3B). Mutating Ser-17 to alanine had no effect on the action of h␤4 on channel activation (Fig. 3B).
The Triple Mutation h␤4 DEE Suppresses the Effect of ␤4 on slo Channel Deactivation-Co-expressing ␤4 with the slo channel slowed slo channel deactivation (Fig. 1F). We tested whether the effect of ␤4 on the deactivation kinetics of the slo channel is also modulated by phosphorylation of the ␤4 subunit. Preincubating cells expressing both slo and ␤4 in 0.5 M OA for 10 -30 min did not alter the effect of ␤4 on the time constant of slo channel deactivation (p Ͼ 0.2) (Fig. 4, A and B). OA treatment also did not change the time constant of slo channel deactivation in cells expressing slo alone (Fig. 4B).  OA had no effect on channel activation in cells transfected with slo alone (p Ͼ 0.5) (n ϭ 8). * denotes significant difference from slo alone, and # denotes significant difference from slo ϩ ␤4. D, mutation of three serine/threonine residues in ␤4 to alanines (␤4 AAA) blocked the effect of OA. The effects of wild-type ␤4 and ␤4 AAA on slo channel activation time constant were not significantly different (p Ͼ 0.05) (n ϭ 9). Furthermore, the increase in activation time constant by ␤4 AAA coexpression was not affected by OA treatment (p Ͼ 0.5) (n ϭ 7). However, when we mutated all three putative serine/threonine phosphorylation sites in ␤4 to negatively charged residues (␤4 DEE mutation), the effect of ␤4 on the deactivation kinetics of the slo channel was eliminated (Fig. 5, A and B). In cells expressing both slo and the triple mutation ␤4 DEE, the time constant of slo channel deactivation was 1.3 Ϯ 0.3 ms, which was significantly different from that in cells expressing slo and wild-type ␤4 (p Ͻ 0.001) (Fig. 5B). The single mutation ␤4 S17E was not sufficient to suppress the effect of ␤4 on slo channel deactivation kinetics (p Ͼ 0.2) (Fig. 5B). ␤4 S17A was as effective as wild-type ␤4 in increasing the deactivation time constant (Fig. 5B).
Phosphorylation-dependent Modulation of slo Channel Voltage Dependence by ␤4 -We showed previously in the inside-out recording configuration that ␤4 modulates the voltage dependence of the slo channel (11). Under the conditions we used in the present study (i.e. the whole-cell recording configuration and overexpression in CHO cells), the current amplitude of the slo channel was so large (up to 20 nA) that an accurate measurement of the current-voltage relationship of the slo channel was impossible. In addition, we had limited our recording to the initial 30 s in order to see the modulation of the slo channel by ␤4, and this does not allow sufficient time to make the necessary input resistance compensation. Nevertheless, we made an estimate of the voltage dependence of the slo channel in the presence of either wild-type ␤4 or various mutant ␤4s. Because the current amplitude of the slo channel was similar under all these conditions, all the I-V measurements were subject to the same magnitude of error. Similar to what we showed with inside-out recordings, ␤4 shifted the current-voltage relationship curve to more depolarized membrane potentials (Fig. 6A). The V1 ⁄2 of slo alone was 14 Ϯ 3.2 mV, which was increased to 40 Ϯ 4.9 mV in the presence of wild-type ␤4 (p Ͻ 0.001) (Fig.  6B). The same okadaic acid treatment, which suppressed the effect of ␤4 on slo channel activation kinetics, also significantly suppressed the shift of the I-V curve caused by ␤4 (p Ͻ 0.01). The V1 ⁄2 of the slo channel in the presence of ␤4 and OA was 19 Ϯ 5.7 mV. As shown in Fig. 6, the single mutation ␤4 S17E also significantly suppressed the effect of the ␤4 subunit on the voltage dependence of the slo channel. The V1 ⁄2 of the slo channel was decreased from 40 Ϯ 4.9 mV in the presence of the wild-type ␤4 to 10 Ϯ 3.2 mV in the presence of ␤4 S17E (p Ͻ 0.001) (Fig. 6B). The triple mutation ␤4 DEE also suppressed the effect of ␤4 on slo channel voltage dependence (data not shown).
Phosphorylation-independent Binding of ␤4 to the slo Channel-The lack of effect of ␤4 DEE and ␤4 S17E on the functional properties of the slo channel could be due to two possibilities. The mutant ␤4 subunits may not bind to the slo channel; alternatively, they may bind to the slo channel but not interact with it functionally. To distinguish between these two possibilities, we tested the binding between the mutant ␤4 subunits and the slo channel by co-immunoprecipitation. HEK293 cells were transfected with hslo channel together with wild-type ␤4 subunit or mutant ␤4 subunits. The hslo channel was immunoprecipitated with a specific antibody recognizing the hslo channel. As shown in Fig. 7, we   FIG. 3. Mutation of three serine/threonine residues in ␤4 to negatively charged residues suppressed the effect of ␤4 on slo channel activation time constant. A, slo current was normalized to reveal that wild-type ␤4 slowed slo current activation. Mutating three putative phosphorylation sites to negatively charged residues in ␤4 (␤4 DEE) dramatically suppressed the effect of ␤4 on slo channel activation. Similarly, single mutation of the serine at position 17 (S17E) suppressed the effect of ␤4 on channel activation. B, quantification of the effect on ␤4 mutation on the functional interaction between ␤4 and the slo channel. ␤4 increased the time constant of slo channel activation. This effect was significantly suppressed by the triple mutation ␤4 DEE (p Ͻ 0.001) (n ϭ 11), although the time constant was still greater in cells expressing slo and ␤4 DEE than in cells expressing slo alone (p Ͻ 0.001). The single mutation ␤4 S17E also suppressed the effect of ␤4 on slo channel activation (p Ͻ 0.01) (n ϭ 19). Single mutation ␤4 S17A did not alter the effect of ␤4 (p Ͼ 0.9) (n ϭ 7).

FIG. 4. Okadaic acid treatment does not influence the effect of ␤4 on slo channel deactivation time constant.
A, ␤4 slowed slo current deactivation, and this effect of ␤4 on slo channel deactivation was not affected by OA treatment. B, quantification of the effect of OA. ␤4 significantly increased the time constant of slo channel deactivation (p Ͻ 0.001) (n ϭ 21). This effect of ␤4 on slo channel deactivation was not altered by OA treatment (p Ͼ 0.2) (n ϭ 13). OA treatment also had no effect on the time constant of slo channel deactivation in cells expressing slo alone (p Ͼ 0.1) (n ϭ 7). could detect both the wild-type ␤4 subunit and all the mutant forms of ␤4 subunit in the immunoprecipitates. As shown previously (11), there were two bands on the Western blot corresponding to the epitope-tagged ␤4 subunit and a higher molecular weight form of the ␤4 subunit. This higher molecular weight band was seen for both the wild-type ␤4 subunit and the ␤4 DEE and ␤4 S17E mutants, although it was largely reduced in the ␤4 AAA mutant (Fig. 7). The identity of this second band is unknown, but we currently are investigating it. In any event, it is clear that the ␤4 mutants we used in the present study still bind to the slo channel, and thus a lack of binding cannot account for the functional differences we observe.
h␤4 Is a Phosphoprotein-We also examined the phosphorylation of ␤4 directly using radioactive 32 P labeling of transfected cells. When wild-type ␤4 was co-expressed with hSlo in HEK293 cells and the hSlo protein was immunoprecipitated, we could detect a 32 P-labeled band in the immunoprecipitate corresponding to the molecular weight of ␤4 (Fig. 8). In contrast, there was no 32 P-labeled band of this molecular weight when hSlo was expressed alone in HEK293 cells (Fig. 8). This demonstrates that ␤4 is indeed a phosphoprotein. There are multiple potential intracellular phosphorylation sites in the ␤4 sequence, including the three serine/threonine residues on which we have focused and an additional threonine and two tyrosine residues. The triple alanine mutation (␤4 AAA) reduced the 32 P label in ␤4, confirming that one or more of these mutated residues is phosphorylated (Fig. 8). However there is still substantial 32 P label associated with ␤4 AAA (Fig. 8), suggesting that some or all of the other potential residues are also phosphorylated and contribute to the total 32 P signal we observe.

DISCUSSION
The function of an ion channel can be modulated by its associated auxiliary subunits. We show here that the functional interaction between the ␣ subunit of the slo channel and the ␤4 subunit is itself subject to modulation. Phosphorylation of the ␤4 subunit dynamically modulates its functional coupling to the ␣ subunit. This provides strong evidence that the protein complex associated with this ion channel is subject to dynamic modulation.
We showed previously that the slo channel ␣ subunit binds tightly to its ␤4 subunit (11). This physical interaction is stable, because it is maintained during co-immunoprecipitation in detergent solution. When co-expressed in HEK293 or CHO cells, ␤4 dramatically modulates the kinetic behavior and toxin sensitivity of the slo channel. There are three putative serine/ threonine phosphorylation sites in the ␤4 subunit, and we tested whether ␤4 could be phosphorylated and the ␣Ϫ␤ interaction could be modulated. For this purpose, we employed the whole-cell recording configuration in the present study. Similar to what we saw previously with recording from inside-out patches, the activation and deactivation kinetics of whole-cell slo current are modulated by ␤4. Interestingly, the modulation of slo channel activation kinetics is only observed within the 1st min of recording. This suggests that after going into the wholecell mode, either the slo channel ␣ subunit and/or the ␤4 subunit is modified, and the functional interaction between them disappears. This time-dependent phenomenon is currently being investigated.
Native K Ca channels in neurons, smooth muscles, and endocrine tissues have all been shown to be modulated by phosphorylation (4,20,24,25,37). In their recording of K Ca channels FIG. 5. The triple mutation ␤4 DEE suppresses the effect of ␤4 on slo channel deactivation. A, wild-type ␤4 slowed slo current deactivation. The triple mutation ␤4 DEE suppressed this effect of ␤4. In contrast, the single mutation ␤4S17E did not alter the effect of ␤4 on channel deactivation. B, the effect of ␤4 on slo channel deactivation time constant was suppressed by the triple mutation ␤4 DEE (p Ͻ 0.001) (n ϭ 11). In contrast, the single mutation ␤4 S17E or S17A did not alter the effect of ␤4 on slo channel deactivation (p Ͼ 0.2, n ϭ 22 and p Ͼ 0.5, n ϭ 6, respectively).
FIG. 6. OA treatment and single mutation of ␤4 S17E suppress the effect of ␤4 on slo channel voltage dependence. A, co-expression with wild-type ␤4 shifted the slo channel current-voltage relationship curve to the right (i.e. to more depolarized voltages). Either treating the cells with OA or mutating serine 17 to glutamate (␤4 S17E) suppressed this effect of ␤4. B, quantification of half-activation voltage (V1 ⁄2 ) of the slo channel under different conditions. Wild-type ␤4 increased slo channel V1 ⁄2 from 13.8 Ϯ 3.2 mV (n ϭ 31) to 40.6 Ϯ 4.9 mV (n ϭ 21) (p Ͻ 0.001). OA treatment significantly reduced the effect of ␤4 (p Ͻ 0.01), with the V1 ⁄2 of slo channel being 19.2 Ϯ 5.7 mV in the presence of OA (n ϭ 15). The single mutation ␤4 S17E also significantly reduced the effect of ␤4 on the slo channel current-voltage relationship (p Ͻ 0.001). The V1 ⁄2 of the slo channel co-expressed with ␤4 S17E was 9.6 Ϯ 3.2 mV (n ϭ 24), which was not significantly different from that of slo alone (p Ͼ 0.3). reconstituted in lipid bilayers, Reinhart et al. (23) classified the neuronal large conductance K Ca channels into two types according to channel kinetics. These two types of channel also respond differently to PKA treatment. The open probability of type 1 K Ca channels is increased by PKA, whereas the open probability of type 2 K Ca channels is decreased by PKA (24). The pore-forming ␣ subunits of K Ca channels have numerous phosphorylation sites. For example, serine 942 is a consensus site for PKA phosphorylation in Drosophila slo (27), although mutating this residue does not block the PKA modulation of slo function. 3 It is also conceivable that part of the diverse re-sponse of K Ca channels to protein kinase modulation is due to their association with different ␤ subunits. The data presented here strongly support this hypothesis.
When cells expressing slo channel ␣ subunit alone were treated with okadaic acid, an inhibitor of protein phosphatase 1 and 2A (38), there was no significant change in channel activity. On the other hand, when we treated cells expressing both slo ␣ subunit and ␤4 subunit with okadaic acid, a dramatic suppression of the effect of ␤4 on slo current activation kinetics and voltage dependence was observed. This suppression effect is likely to be due to changes in phosphorylation of the ␤4 subunit. First, mutating all three putative serine/threonine phosphorylation sites in ␤4 to alanines abolishes this effect of okadaic acid treatment. Second, mutating all three sites to negatively charged residues, or even the single mutation S17E, mimics the effect of okadaic acid treatment. The phosphorylation of ␤4 thus seems sufficient for uncoupling this functional interaction with the ␣ subunit. The loss of functional interaction between the ␤4 subunit and the slo channel ␣ subunit in the present study is not likely due to dissociation of the two proteins. This is supported by the co-immunoprecipitation experiments that show that the mutant ␤4 subunits bind to the slo ␣ subunit as well as does the wild-type ␤4. The results of these biochemical experiments also suggest that mutations of the putative phosphorylation residues did not grossly alter the structure of the ␤4 protein. The effect of these ␤4 mutations on the modulation of the slo channel thus must be due to alteration of functional interactions between these two channel subunits.
Our data suggest that the activation kinetics, deactivation kinetics, and voltage dependence of the slo channel are differentially modulated. For example, the time constant of slo channel activation decreases with time after going into whole-cell recording mode, whereas the time constant of slo channel deactivation does not change with time. It was also noted that the triple mutation ␤4 DEE influenced not only the effect of ␤4 on slo channel activation kinetics and voltage dependence but also the effect of ␤4 on slo channel deactivation (Figs. 3 and 5). In contrast, the single mutation ␤4 S17E or OA treatment altered the effect of ␤4 only on the activation kinetics and voltage dependence of the slo channel. These differential effects on activation kinetics, deactivation kinetics, and voltage dependence are summarized in Table II. One possible explanation is that uncoupling the effect of ␤4 subunit on slo channel deactivation kinetics may require phosphorylation of ␤4 on multiple residues, and this is not mimicked in the single mutation ␤4 S17E or achieved by OA treatment. These results suggest that different kinetic parameters of the slo channel can be modulated by fine-tuning of the phosphorylation state of the ␤4 subunit.
We found by 32 P labeling that h␤4 is indeed a phosphoprotein, and the ␤ subunits of some Kv channels and of L-type Ca 2ϩ channels have been shown to be modulated by phosphorylation (32,34). It was also shown that phosphorylation of Kv1.1 ␣ subunit at a specific residue is necessary for the channel to be sensitive to Kv␤1.1-induced fast inactivation (39). Similarly, the fast inactivation of Kv1.5 conferred by Kv␤1.3 is modulated by PKA activation, but in this case the effect of PKA is mediated by phosphorylation of the Kv␤1.3 subunit (34). PKA activation increases L-type Ca 2ϩ channel activity, and this modulation is at least partially mediated by the phosphorylation of the Ca 2ϩ channel ␤2 subunit, because mutating two serine residues in the ␤2 subunit blocked this up-regulation of the L-type Ca 2ϩ channel by PKA (32). These data, together with the present results implicating phosphorylation in the functional interaction between the slo channel and its ␤4 sub-3 Y. Zhou and I. B. Levitan, manuscript in preparation.  FIG. 8. ␤4 is a phosphoprotein. Shown here is an autoradiograph of 32 P-labeled proteins expressed in HEK293 cells. 1st lane, hSlo alone expressed in HEK293 cells; there was no 32 P-labeled band corresponding to the molecular weight of ␤4. 2nd lane, wild-type ␤4 co-expressed with hSlo; a 32 P-labeled band of the molecular weight of ␤4 was detected in the hSlo immunoprecipitate. 3rd lane, triple alanine mutation (␤4 AAA) co-expressed with hSlo; the 32 P label in ␤4 was reduced. unit, provide strong evidence that the interaction between pore-forming ␣ subunits and auxiliary subunits can be subject to dynamic regulation. Such dynamic regulation of subunit interactions thereby provides another mechanism for generating diversity of ion channel function. It will be interesting to determine how this type of modulation is utilized under physiological conditions.