Regulators of G Protein Signaling (RGS) Proteins Constitutively Activate Gβγ-gated Potassium Channels*

Here we report novel effects of regulators of G protein signaling (RGS) on G protein-regulated ion channels. RGS3 and RGS4 induced a substantial increase in currents through the Gβγ-regulated inwardly rectifying K+ channels,I K(ACh), in the absence of receptor activation. Concomitantly, the amount of current that could be activated by agonist was reduced. Pretreatment with pertussis toxin or a muscarinic receptor antagonist abolished agonist-induced currents but did not modify RGS effects. Cotransfection of cells with a Gβγ-binding protein significantly reduced the RGS4-induced basalI K(ACh) currents. The RGS proteins also modified the properties of another Gβγ effector, the N-type Ca2+ channels. These observations strongly suggest that RGS proteins increase the availability of Gβγ in addition to their previously described GTPase-activating function.

suggesting that RGS act by stabilizing the transition state of the GTPase (2). The first crystal structure of the core domain of RGS4 confirmed this hypothesis (3). Because RGS proteins can act as GAPs for heterotrimeric G proteins, they have been considered to be a missing link that might explain unresolved discrepancies in the deactivation kinetics of G protein signaling that have been observed in studies in vivo and in vitro (4 -8).
Ion channels that are directly gated by G proteins allow for the resolution of the kinetics of G protein signaling in intact cells. The prototype of such a channel is the cardiac atrial potassium channel that is activated by muscarinic acetylcholine receptors (I K(ACh) channel) (9). I K(ACh) channels are formed by heteromultimers of G protein-regulated inwardly rectifying K ϩ channels (GIRKs) 1 and 4 (Kir 3.1 and Kir 3.4) (10 -12). Upon activation of G i -coupled receptors, such as the M 2 muscarinic acetylcholine receptors (mAChRs) in membranes of atrial myocytes, G␤␥-subunits have been shown to directly bind to the channels and to increase their open probability (9,13,14).
When GIRK channels are studied in heterologous expression systems they exhibit a much slower turn off (deactivation) compared with the rapid deactivation that is observed in native atrial myocytes (6,7). However, several recent studies have demonstrated that RGS3, RGS4, and RGS8 can induce rapid deactivation of heterologously expressed I K(ACh) (6,7). These studies were the first to demonstrate that RGS proteins may function as GAPs in intact cells. However, the RGS proteins also caused an unexpected acceleration of the kinetics of activation of the channels (6,7). Furthermore, the RGS proteins did not cause a reduction in peak currents, an effect that would be predicted if the sole effect of RGS is to act as a GAP (8). These observations suggested that RGS may have additional roles.
In other studies, RGS proteins inhibited G␣-activated pathways in intact cells (15)(16)(17)(18)(19). These results could be explained in two ways: 1) RGS act as GAPs in vivo and decrease the concentration of active G␣ (GTP-G␣), or 2) RGS bind to G␣ and prevent binding of G␣ to its effectors. RGS proteins bind with high affinity to the transition state of GTP-G␣, but they can also bind with lower affinity to the GTP-and GDP-bound form of G␣ (1,20,21). The analysis of RGS effects on G␤␥ signaling should help to elucidate which of these two possibilities occur in vivo, because structural studies indicate that it is unlikely that G␤␥ and RGS can bind to the same G␣ at the same time (1,3,22). Therefore, if RGS inhibit signaling by binding to G␣, RGS will increase the concentration of free G␤␥ and activate G␤␥-signaling pathways. On the other hand, if RGS act solely as GAPs, there should be a decrease in free G␤␥ by promoting reassociation to heterotrimers (23). In this study we tested these hypotheses by examining the effect of RGS proteins on G␤␥-mediated signaling by measuring kinetics as well as amplitudes of basal and agonist-activated I K(ACh) .
Measurement of Membrane Currents-Membrane currents were recorded under voltage clamp using conventional whole cell patch-clamp techniques (28). Patch pipettes were fabricated from borosilicate glass capillaries (GF-150-10, Warner Instrument Corp.) using a horizontal puller (P-95 Fleming & Poulsen). The DC resistance of the filled pipettes ranged from 3 to 6 megaohms. Membrane currents were recorded using a patch-clamp amplifier (Axopatch 200, Axon Instruments). Signals were analog-filtered using a lowpass Bessel filter (1-3 kHz corner frequency). Data were digitally stored using an IBM compatible PC equipped with a hardware/software package (ISO2 by MFK, Frankfurt/ Main, Germany) for voltage control, data acquisition, and data evaluation. I K(ACh) was measured as an inward current using a holding potential of Ϫ90 mV as described (29). Voltage ramps (from Ϫ120 mV to ϩ60 mV in 500 ms, every 10 s) were used to determine current-voltage (I-V) relationships.
Barium currents through N-type calcium channels were measured as described (25). Briefly, Ba 2ϩ currents were measured every 10 s using two 25-ms test pulses of ϩ10 mV that were interrupted by a 70-ms period in which the potential was clamped either to the holding potential (without prepulses) or for a 65-ms long prepulse to ϩ80 mV followed by 5-ms period at the holding potential of Ϫ90 mV. All measurements were performed at room temperature. Summarized results are presented as mean values Ϯ S.E. Statistically significant differences were analyzed using Student's t test.

RGS Has Stimulatory Effects on I K(ACh) -
To measure agonist-induced I K(ACh) we routinely applied a saturating concentration of ACh (2 M for HEK cells and 1 M for CHO K1 cells) to fully activate I K(ACh) . To measure basal (nonagonist-dependent) currents, we measured the difference in currents before and after application of 0.2 mM Ba 2ϩ , which will completely block inwardly rectifying potassium currents. We first observed that RGS4 induced an unexpected increase in basal currents. In HEK cells transfected with GIRK1 and -4, only a small amount of basal current was observed as Ba 2ϩ reduced the current in the absence of ACh by 4.3 Ϯ 0.8 pA/pF, n ϭ 16 (Fig.  1, A and D). In marked contrast, in HEK293 cells transfected with GIRK1 and -4 plus RGS4, Ba 2ϩ decreased the basal currents by 14.3 Ϯ 2.7 pA/pF (n ϭ 14) (Fig. 1, B and D). In nontransfected cells, Ba 2ϩ did not affect background currents (five cells tested for each cell line). That the Ba 2ϩ blockable currents were attributable to expression of GIRK1 and -4 was confirmed by demonstrating that the I-V relationships for both the agonist-induced currents and basal currents were identical (Fig. 1C). Concomitant to the RGS-induced increase of basal currents, the ACh-induced currents were reduced in RGS4expressing cells (12.3 Ϯ 2.4 pA/pF in RGS-expressing cells versus 24.1 Ϯ 8.8 pA/pF in control cells), whereas the total (agonist ϩ basal) currents were not altered (Fig. 1, B and D). Thus expression of RGS4 induced activation of 53 Ϯ 8% of total I K(ACh) compared with basal activation of 19 Ϯ 4% of I K(ACh) in control cells. Similar results were obtained using RGS3 instead of RGS4 (basal activation, 43 Ϯ 5%, n ϭ 4).
Previously it was reported that RGS4 accelerated the kinetics of heterologously expressed I K(ACh) in Xenopus oocytes, as well as in CHO K1 cells (6); however, stimulation of basal currents was not apparent. To test whether or not the observed increase in basal I K(ACh) in HEK 293 cells was a cell typespecific effect, we tested the effect of RGS4 on I K(ACh) in transiently transfected CHO K1 cells. Upon cotransfection with FIG. 1. RGS4 induces basal I K(ACh) in HEK293 cells. GIRK1 and GIRK4 with or without RGS4 were transiently expressed in HEK293 cells stably expressing M 2 mAChRs. Inward I K(ACh) (holding potential of Ϫ90 mV, E K ϳϪ50 mV) was activated by superfusion of the cells with 2 mM ACh. To determine basal currents, Ba 2ϩ (0.2 mM) was used to specifically block inwardly rectifying K ϩ currents. A and B, representative current traces obtained from cells transfected without (A) or with (B) RGS4. C, by subtracting currents measured during voltage ramps (from Ϫ120 mV to ϩ60 mV within 500 ms) indicated by a, b, and c in B, I-V curves of agonist-activated currents (b-a) and basal Ba 2ϩ -sensitive currents (a-c) were constructed. D, summarized data of 12-14 experiments. Note the basal I K(ACh) is increased 3-fold in cells expressing RGS4, whereas the total current amplitude (agonist-induced current ϩ basal current) is not altered.
RGS4, most cells exhibited a large increase of ACh-independent, Ba 2ϩ -sensitive inwardly rectifying K ϩ currents (Fig. 2B, 27 Ϯ 20 pA/pF, n ϭ 5), whereas in the absence of expressed RGS4, the detected Ba 2ϩ -sensitive background currents were negligible ( Fig. 2A, 2 Ϯ 0.5 pA/pF, n ϭ 5). These results mirrored those from the HEK cells and demonstrated that RGS expression led to agonist-independent activation of I K(ACh) . We cannot explain why others did not detect an increase in basal current (6, 7); however, one obvious explanation is that there are differences in the heterologous expression systems used. In our experience the vast majority of the RGS4-transfected cells exhibited increased basal I K(ACh) .
Expression of RGS proteins also caused a small acceleration of the activation of I K(ACh) ( Table I); however, this effect was less significant than previously reported in studies of I K(ACh) measured in Xenopus oocytes or CHO K1 cells (6,7). A more prominent effect of RGS was to cause faster deactivation of ACh-induced currents after withdrawal of ACh in both HEK293 cells and CHO K1 cells (Table I, Fig. 2, A versus B), similar to previous observations (6,7). The acceleration of I K(ACh) kinetics by RGS proteins most likely reflects the GAP activity of RGS proteins (20, 30 -33). However, the induction of basal I K(ACh) by RGS3 and RGS4 cannot be easily explained by the function of RGS as a GAP.
To determine whether the various effects of RGS might be because of high levels of expression, we varied the amount of RGS cDNA used for transfections and tested for effects on basal currents and on acceleration of deactivation. When the amount of RGS cDNA was reduced as low as 0.1 g, the RGS effects on basal currents were still obvious, whereas the effects on deactivation were reduced (data not shown). Thus the newly appreciated effects of RGS to increase channel currents did not appear to be an artifact of overexpression.
Induction of Basal I K(ACh) by RGS4 Depends on G␤␥ but Not on Receptors-An RGS-induced promotion of the coupling of receptor, G protein, and channel was considered as a possible mechanism for the unexpected effects of RGS on basal currents and on activation kinetics (6). To test this, cells were treated with pertussis toxin. The uncoupling of receptors from G proteins by pertussis toxin pretreatment did not reduce the RGSinduced basal I K(ACh) whereas it completely abolished AChinduced I K(ACh) (data not shown). Therefore, the RGS-induced basal activation of I K(ACh) did not depend on the basal activity of G i -or G o -coupled receptors because it was not inhibited by pertussis toxin, which is known to prevent activation of G i and G o proteins by G protein-coupled receptors but not GTP hydrolysis and nucleotide exchange (34).
We determined that expression of G␤ 1 ␥ 2 resulted in a 3-4-fold increase of total I K(ACh) in HEK cells (data not shown) indicating that endogenous G i proteins are limiting for maximal activation of I K(ACh) in these cells. Next we tested whether RGS induced I K(ACh) by increasing free G␤␥ in the cell or by activating GIRK channels via a G␤␥-independent mechanism. To do this we coexpressed the channels and RGS with CD8-␤ARK-ct, a fusion protein of the ␣-subunit of the CD8 receptor, and the C terminus of the ␤-adrenergic receptor kinase 1 (␤ARK1), which has been demonstrated to act as a G␤␥ "sink" and block G␤␥-mediated signaling (35). The expression of CD8-␤ARK-ct reduced basal I K(ACh) in RGS4-expressing cells by 65% (Fig. 3, compare currents in cells expressing CD8 versus CD8-␤ARK-ct). The expression of the G␤␥-binding protein also reduced the ACh response to the threshold of detection, resulting in a total (basal ϩ agonist-activated) I K(ACh) of about 20% compared with the control (Fig. 3). Therefore, we concluded that RGS4 activated I K(ACh) via a G␤␥-dependent mechanism. Because GIRK1 and GIRK4 have been shown to bind G␤␥ directly (36) and therefore may act as sink for G␤␥ if overexpressed, it was possible that the observed RGS-induced increase in basal I K(ACh) was the result of an RGS-induced increase in GIRK expression. Immunoblotting experiments using antibodies against GIRK1, GIRK4, and hemagglutinin-tagged RGS4 demonstrated no increase of GIRK1 and GIRK4 expression upon cotransfection with hemagglutinin-tagged RGS4 (data not shown). Therefore, an increase of GIRK1 and GIRK4 expression as the mechanism by which RGS4 induces basal I K(ACh) can be ruled out. An increase in the expression level of the endogenous G␤␥ upon coexpression of RGS proteins could be another potential explanation for the observed increase in basal I K(ACh) . However, we have observed that the total I K(ACh) is limited by the expression level of endogenous G␤␥ proteins, as currents are greatly increased upon expression of G␤␥ (data not shown). Because neither RGS3 nor RGS4 increased total I K(ACh) (Fig. 1), an increase of G␤␥ expression induced by RGS proteins seems unlikely. The possibility that expression of RGS proteins caused down-regulation of the G␣ i proteins was not tested.
RGS4 Increases the Concentration of Free G␤␥-Because the increase of basal I K(ACh) by RGS was dependent on G␤␥, a reasonable hypothesis was that the underlying mechanism for the RGS-induced I K(ACh) was an increase in the concentration of free G␤␥ in the cell. To obtain further support of this hypothesis, we analyzed another G␤␥-regulated effector system that allowed for quantitative measurement of basal (nonagonist regulated) effects of G␤␥. N-type Ca 2ϩ channels are known to be inhibited by G␤␥, and this inhibition can be partially reversed by applying strong positive voltages prior to measuring channel currents with a procedure termed prepulse facilitation (25,(37)(38)(39)(40). Although prepulse facilitation is usually more apparent in the presence of an agonist that activates G i -coupled receptors, we attempted to reveal effects of RGS proteins on basal currents in the absence of an agonist. The effects of RGS on basal inhibition of N-type Ca 2ϩ currents were tested in HEK293 cells stably expressing ␣ 1B , ␤ 1 , and ␣ 2 ␦ calcium channel subunits (25) using Ba 2ϩ as the charge carrier. Ba 2ϩ currents were measured using a double test pulse proto-  col in which two successive measurements of current test pulses were separated by a 65-ms period during which the cells were or were not exposed to a strong depolarization (prepulse to ϩ80 mV) to test for relief of constitutive G␤␥-mediated inhibition. Prepulse facilitation was determined by the ratio of (I Ba measured after a prepulse to ϩ80 mV)/(I Ba measured at the second test pulse without a preceding positive prepulse). In the control condition, if there was no positive prepulse applied prior to the second test pulse, the current amplitude of Ba 2ϩ currents was 76 Ϯ 3%, n ϭ 11 compared with the current measured during the first test pulse (Fig. 4A, compare the first and second test pulses). In cells not transfected with RGS a positive prepulse did not increase I Ba (Fig. 4A). The test pulses Ϯ the prepulse were completely superimposable (Fig. 4, A and C, prepulse facilitation, 0.95 Ϯ 0.04, n ϭ 11). However, in cells transfected with RGS4, positive prepulses resulted in a small but reproducible and significant facilitation of I Ba (Fig. 4, B and C, prepulse facilitation, 1.13 Ϯ 0.05, n ϭ 15, p Յ 0.01). The effect of RGS4 to induce constitutive inhibition of the N-type channel was probably larger than it appeared in our measurements, because in independent experiments we determined that the prepulse led only to a ϳ60% release of the G␤␥-mediated inhibition of the channels. These results suggested that a G␤␥-dependent "basal" inhibition of channels was occurring in RGS4-transfected cells but not in control cells, as reflected by the increase of currents after reversal of the G␤␥mediated inhibition by positive prepulses. These results further suggested that RGS proteins cause constitutive G␤␥ signaling in HEK cells.
Both the N-type Ca 2ϩ channels as well as the I K(ACh) channels are well accepted to be direct effectors of G␤␥-subunits of pertussis toxin-sensitive G proteins (38,41,42). Because RGS4 in both cases induced a response that is usually mediated by G␤␥-subunits, we concluded that the expression of RGS4 increased the concentration of free G␤␥ in the cell and constitutively modulated the activation state of both channels. It should be mentioned that constitutive signaling through G␤␥ was not observed in previous studies where the effects of RGS proteins were assessed in systems involving G␤␥ signaling. Thus, an inhibition, rather than a constitutive activation, of agonist-induced mitogen-activated protein kinase activation was observed upon coexpression of RGS1 in HS-Sultan cell lines (16) or with RGS4 and SST2 in yeast (16,43). Although we have no definitive explanation for the differences between our observations and those made with the mitogen-activated protein kinase-linked systems, the different experimental conditions and the fact that the signaling cascades that lead to activation of mitogen-activated protein kinases are complex and involve considerable amplification make it difficult to compare results obtained in this system with ion channels that are directly G␤␥-modulated. It seems possible that the net effect of RGS proteins in different systems will reflect a summation of inhibitory (GAP activity) and stimulatory (increased availability of G␤␥) processes. This appears to be the case with the receptor-mediated activation of G␤␥-gated K ϩ channels. In all of the studies reported so far, although the kinetics of channel deactivation were increased, there was no depression of the peak currents upon expression of RGS proteins (6, 7) in contrast to what one would expect if the RGS proteins were solely acting as GAPs (see discussion in Ref. 8). Thus, in the mitogenactivated protein kinase systems studied previously (16,43), it is not possible to know if the inhibitions observed were also influenced by stimulatory actions of RGS proteins.
A Newly Appreciated Action of RGS-Several studies of intact cells have shown that RGS proteins inhibit G i , G o , or G q signaling (15-19, 21, 44, 45). These results could be explained by enhanced deactivation of the G proteins due to the GAP activity of RGS or by binding of RGS to G␣ and consequent block of effector activation. The efficiency of RGS proteins to act as GAPs in intact cells will strongly depend on the relative affinities of RGS to the different conformations of G␣ proteins. Our approach to measure basal and receptor-activated G␤␥ signaling in intact cells allowed for detection of GAP effects by RGS (i.e. the acceleration of I K(ACh) kinetics) as well as for indirect detection of RGS binding to G␣ proteins in the absence of agonists, by measuring basal G␤␥-induced currents. The concentration of free G␤␥ will most likely be increased if RGS
FIG. 4. RGS4 induces prepulse facilitation of N-type calcium channels. Ba 2ϩ currents in G1A1 cells transiently transfected with (B) or without (A) RGS4 were analyzed using a double pulse protocol with or without an intervening depolarization test prepulse to ϩ80 mV for 65 ms as indicated. The double pulse protocol was applied every 10 s, and the duration of the test pulses was 25 ms. The holding potential was Ϫ90 mV. A and B, representative current traces recorded with or without prepulse are overlaid. C, summarized data of prepulse facilitation for both conditions. Prepulse facilitation was calculated as the relative current amplitude of the second test pulse after the positive prepulse compared with the relative current amplitude of the second test pulse without positive prepulse. **, indicates that the two data groups are significantly different (p Ͻ 0.01). proteins bind to G␣ proteins, because G␤␥ will not be able to bind to RGS-occupied G␣-subunits (3). Assuming that the underlying mechanism for the turn off of G␤␥-mediated signaling is reassociation with the deactivated G␣-subunit, the GAP function of RGS should speed up deactivation of G␣ and consequently also turn off G␤␥-mediated signals. Considering this, we speculate that the observed and previously described acceleration of I K(ACh) deactivation by RGS is because of the GAP activity of these proteins (6,7). The increase of basal G␤␥ signaling upon coexpression of RGS3 or RGS4 may reflect binding of RGS to a substantial portion of the G␣ protein pool and a consequent decrease in the sequestering of G␤␥ by G␣subunits. Our studies report a novel effect of RGS on G␤␥ signaling and suggest that in addition to causing desensitization because of GAP activity, RGS proteins may also positively impact G␤␥ signaling.