G Protein β2 Subunit-derived Peptides for Inhibition and Induction of G Protein Pathways

Voltage-gated Ca2+ channels of the N-, P/Q-, and R-type and G protein inwardly rectifying K+ channels (GIRK) are modulated via direct binding of G proteins. The modulation is mediated by G protein βγ subunits. By using electrophysiological recordings and fluorescence resonance energy transfer, we characterized the modulatory domains of the G protein β subunit on the recombinant P/Q-type channel and GIRK channel expressed in HEK293 cells and on native non-L-type Ca2+ currents of cultured hippocampal neurons. We found that Gβ2 subunit-derived deletion constructs and synthesized peptides can either induce or inhibit G protein modulation of the examined ion channels. In particular, the 25-amino acid peptide derived from the Gβ2 N terminus inhibits G protein modulation, whereas a 35-amino acid peptide derived from the Gβ2 C terminus induced modulation of voltage-gated Ca2+ channels and GIRK channels. Fluorescence resonance energy transfer (FRET) analysis of the live action of these peptides revealed that the 25-amino acid peptide diminished the FRET signal between G protein β2γ3 subunits, indicating a reorientation between G protein β2γ3 subunits in the presence of the peptide. In contrast, the 35-amino acid peptide increased the FRET signal between GIRK1,2 channel subunits, similarly to the Gβγ-mediated FRET increase observed for this GIRK subunit combination. Circular dichroism spectra of the synthesized peptides suggest that the 25-amino acid peptide is structured. These results indicate that individual G protein β subunit domains can act as independent, separate modulatory domains to either induce or inhibit G protein modulation for several effector proteins.

Transmitter-mediated activation of G proteins causes inhibition of voltage-gated Ca 2ϩ channels of the N-, P/Q-, and R-type. The inhibition is mediated by G protein ␤␥ subunits (1,2) and involves the direct binding of the G protein to the channel. Ca 2ϩ channels consist of a pore-forming ␣ 1 subunit and several attached ancillary subunits (for review see Refs. [3][4][5]. During binding to the channel complex, the G protein presumably stabilizes a conformational state of the channel, which causes the reluctant opening of the channel (6). This biophysical model is supported by the fact that the G protein alters the FRET 1 signal between the FRET pair, the Ca 2ϩ channel ␣ 1 , and the Ca 2ϩ channel ␤ subunit (7,8). This mechanism was proposed in earlier biophysical studies on G protein modulation of the Ca 2ϩ channels (9 -14). The binding sites of the G protein on the channel have been carefully characterized. G protein ␤␥ subunits bind to the intracellular loop I-II and to the C terminus of the pore-forming ␣ 1 subunit (15)(16)(17)(18)(19)(20)(21). G protein modulation also involves the N terminus of the Ca 2ϩ channel (22). Most surprisingly, little is known about the interaction of the G protein with the channel and how the G protein domains are involved in modulation. Earlier studies characterized single point mutations on the G protein ␤ subunit, suggesting that the entire G protein surface rather than a single amino acid may determine specific effector functions (23,24). Recently, Doering et al. (25) identified several structural components of the G protein ␤ subunit as important for the specificity of G protein modulation of N-type channels.
In this study, we addressed whether we could identify specific G protein domains that either inhibited or induced G protein modulation of different channel (effector) proteins. By applying electrophysiological studies and FRET measurements, we identified two structural components of the G protein important for ion channel modulation. Peptides derived from the N terminus of G␤ 2 , including the G protein ␤␥ interaction site, inhibit G protein modulation of GIRK and voltagedependent Ca 2ϩ channels. The inhibitory effect is correlated with a change in the interaction between G protein ␤ and ␥ subunits. In contrast, protein domains derived from the C terminus of G␤ 2 induce G protein modulation of GIRK and voltage-dependent Ca 2ϩ channels. This effect is correlated with an increase in FRET between GIRK channel subunits mediated by either the G␤␥ complex or the modulatory peptide. Our data suggest that peptides derived from G protein ␤ subunits can act as independent domains for G protein ␤␥ effector protein such as ion channels. These peptides will thus provide useful tools for characterizing and manipulating G protein pathways in living cells.

MATERIALS AND METHODS
Construction of pYFP-nuc/HA-C1 for Co-expression Studies-All G protein constructs were cloned into a new expression vector (pYFP-nuc/ HA-C1). In this vector, YFP is under the control of a second and separate cytomegalovirus promoter and tagged with a nuclear localization signal, which allows the constitutive expression and transport of YFP to the nucleus as well as the expression of the nonfused protein or peptide of interest. Thus, cells with a nuclear targeted YFP contain the cDNA encoding for the peptide of interest. For constructing the pYFPnuc/HA-C1 vector used for co-expression of the G␤ 2 -derived deletion constructs, two SV40 nuclear localization signals were added 3Ј inframe into the YFP coding sequence in pEYFP-C1 (Clontech) (pYFPnuc). Then the cytomegalovirus promoter, polylinker region, and bovine growth hormone polyadenylation signals were PCR-amplified from pcDNA3.1 and subcloned into the AflII site of the pYFP-nuc-C1. An HA tag polylinker was directly cloned into the Nhe-PmeI site of the amplified polylinker via blunt end cloning.
Cell Culture-Human embryonic kidney (HEK) 293 cells (tsA201 cells) were transfected with 2 g of the indicated DNAs in each set of experiments with Quantum Prep R Cytofectene TM Transfection Reagent (Bio-Rad) and incubated for 24 - 1-25, 26 -50, 201-235, 236 -270, 271-305, and 306 -340) were PCR-amplified and subcloned in-frame into pYFP-nuc/HA-C1 for electrophysiological recordings and into pEYFP-C1 for localization within the cell. The cellular localization of the YFP-tagged G␤ constructs is shown in the supplemental Figs. 1-3. Other constructs, i.e. G␥ 3 and GIRK1 and GIRK2, were either PCR-amplified or if restriction sites were suitable cloned into either pECFP-C1, pECFP-C2, pECFP-N1, pEYFP-C1, and pEYFP-N1 and as non-tagged versions into pcDNA3.1. All PCR-amplified products were verified by sequencing, and protein expression was verified by Western blot analysis after expression of the constructs in COS-7 cells. Western blots of transfected COS-7 cells (with G␤ constructs) were performed with a polyclonal anti-green fluorescent protein antibody (Molecular Probes, Inc., Eugene, OR) according to standard procedures as described by Mark et al. (26). Muscarinic AChR M2 (human) was cloned into pcDNA3.1(ϩ) and purchased from UMR cDNA resource center (Rolla, MO). Reverse transcription and PCR amplification were used to clone ␥ 2 (stargazin) from rat brain RNA using the following oligonucleotide primers: sense, 5Јatggggctgtttgatcgaaggtgt-tcaaatg3Ј; antisense, 5Јtcatacgggcgtggtccggcggttggctgt3Ј. The nucleotide sequence was verified to the rat ␥ 2 (GenBank TM accession number NM_053351) by cDNA sequencing.
FRET Measurements-FRET measurements were performed by using a modified version of the 3-cube FRET method (27) as described in Hü mmer et al. (7). Briefly, the average FRET signal (FRET ratio) was calculated using the following equation: FR ϭ (FRET fluorescence Ϫ RD1 ϫ CFP fluorescence )/(RA1 ϫ (YFP fluorescence )), where the constants RD1 and RA1 were derived from the CFP and YFP constructs expressed alone in HEK293 cells. The average FRET signal was confirmed for each FRET pair by acceptor photobleaching. Here FRET samples were exposed 30 min to fluorescent light through the YFP filter cube (excitation, HQ 500/20; band pass filter 535/30; beam splitter JP4 PC 535/30), and the FRET efficiency (Eff) was calculated according to the increase in CFP emission: Eff ϭ (1 Ϫ CFP fluorescence before/CFP fluorescence after) ϫ 100%. In our experimental setting, we calculated that the YFP signal was reduced by 70% after 30 min of light exposure. CFP, FRET, and YFP pictures were analyzed with Volocity software (Improvision).
For live cell FRET measurements, cells were incubated in extracellular solution containing (in mM) 172 NaCl, 2.4 KCl, 10 HEPES, 10 glucose, 4 CaCl 2 , 4 MgCl 2 (pH 7.3). CFP, FRET, and YFP pictures were taken before and 30 min after peptide application. For calculation of the FRET ratio over the whole image, images were saved as TIFF files and processed with Igor software (WaveMetrics, Inc.). The FRET equation (FR ϭ (FRET fluorescence Ϫ RD1 ϫ CFP fluorescence )/(RA1 ϫ (YFP fluorescence )) was applied to every pixel of the image for all three images. FR within these images was encoded by 256 gray values. These values were converted into color values, where the intensity of the color represents the FR value. For FR calculation, the gray scale value outside of the cell of interest was determined (typically under Ͻ33 gray scale values). This value was subtracted from the whole image. The FR image of Figs. 6 and 7 without background subtraction are shown in the supplemental Fig. 4.
Fluorescence was detected with a conventional fluorescence microscope (Leica DMLFSA) equipped with one filter wheel on the emission site controlled by a Sutter Lambda10-2 and a Sutter DG5 light source on the excitation site. For CFP and YFP detection, the following filters were used: CFP, excitation, D436/10; emission, D470/30; YFP, excitation, HQ 500/20; band pass filter 535/30; beam splitter JP4 PC. All filters were obtained from Chroma. The following objectives were used: a 63ϫ water corrected 1.2NA objective for fixed FRET samples and a 63ϫ water corrected long distance 0.9 NA objective for live cell recordings.
Peptide Delivery-The synthesized peptides (pep1-25, pep271-305, and pep1-25scrambled) were dissolved in H 2 O and kept as 10 mM stock solution at Ϫ20°C. From this stock solution further dilutions (100 M) were prepared in the intracellular solutions used for GIRK, P/Q-type, and neuronal non-L-type channel recordings for the intracellular application of the peptides. The peptides have the following sequences: pep1-25, MSELEQLRQEAEQLRNQIRDARKAC; pep1-25scrambled, MRLQAEQRNCSLEIRLEQDKEQARA; and pep271-305, CGITS-VAFSRSGRLLLAGYDDFNCNIWDAMKGDRA.
For peptide transfection, the Chariot transfection reagent (Active Motif, Carlsbad, CA) was used. According to the instructions manual of the manufacturer, 500 ng of peptide was mixed in the transfection reagent, incubated for 30 min at room temperature, and then added directly to the imaged cells in the recording chamber. Unless otherwise indicated, electrophysiological recordings were performed 20 min after chariot/peptide application, and FRET measurements were performed 30 min after chariot/peptide application.
Circular Dichroism Spectra-Peptides were dissolved in H 2 O. Peptide concentration were 29 M for pep1-25 and 13 M for pep271-305. Far-UV circular dichroism spectra were measured on an Aviv instrument in a thermostatted 0.3-cm quartz cell (Helma). Measurements were carried out at 25 and 5°C. A single scan was acquired at 0.5 nm intervals between 250 and 185 nm with an averaging time of 2 s. The signal of buffer without peptide was subtracted from each measurement. Molar ellipticity was calculated as [⌰] ϭ CD signal in millidegree (peptide concentration M ϫ cell length cm ϫ number of residues). The helical content was estimated from the ellipticity at 222 nm using the procedure of Rohl and Baldwin (28).
Electrophysiology-For P/Q-type channel recordings in HEK293 cells, Ca 2ϩ channel subunits (␣ 1 2.1, ␤ 1b , and ␣ 2 ␦) and the G protein subunit (G␤ 2 ) or G protein ␤ fragments were co-expressed in tsA201 cells, and Ca 2ϩ channel-mediated Ba 2ϩ currents were measured and analyzed as described previously (1,29,30). The following bath pipette solutions were used: for external solution (mM), 100 Tris, 4 MgCl 2 , 10 BaCl 2 with pH adjusted to 7.3 with methanesulfonic acid; for internal solution (mM), 120 aspartic acid, 5 CaCl 2 , 2 MgCl 2 , 10 HEPES, 10 EGTA and 2 Mg-ATP with pH adjusted to 7.3 with CsOH. Endogenous G proteins were activated by adding 0.6 mM GTP␥S to the internal solution. Tail currents (supplemental Table) were fitted with a single exponential using Igor software.
PKC inhibitor peptide (Sigma) was dissolved in the intracellular recording solution at a concentration of 100 M. 4␣-Phorbol 12-myristate 13-acetate (PMA) (Sigma) was dissolved in Me 2 SO as a 1 mM stock solution. 2 l of this solution was added to 2 ml of the extracellular bathing solution 5 min prior to the experiment (final concentration 1 M).
All whole cell patch clamp recordings (32) were performed with an EPC9 amplifier (HEKA). Currents were digitized at 10 kHz and filtered with the internal 10-kHz three-pole Bessel filter (filter 1) in series with a 2.9-kHz 4-pole Bessel filter (filter 2) of the EPC9 amplifier. Series resistances were partially compensated between 70 and 90%. Leak and capacitative currents were subtracted by using hyperpolarizing pulses from Ϫ60 to Ϫ70 mV with the p/4 method.
Statistical significance throughout the experiments was tested with ANOVA using Igor Software. Standard errors are mean Ϯ S.E.

RESULTS
G protein modulation of N-, P/Q-, and R-type channels is mediated by G protein ␤␥ subunits (1, 2) and most likely involves the direct binding of the G protein to the channel (16,18,19). Co-expression of G␤ 2 together with the P/Q-or N-type channel ␣ 1 , ␤ 1b , and ␣ 2 ␦ subunits results in channel modulation, i.e. a shift in the voltage dependence of activation, a more shallow activation curve, and prepulse facilitation (1,29,33). In this study we addressed whether deletion fragments of the G FIG. 1. Induction of P/Q-type channel G protein modulation by G protein ␤ 2 deletion constructs. A, examples of Ba 2ϩ current traces in the presence and absence of G protein ␤ peptides with ATP in the intracellular recording solution. Ba 2ϩ currents were elicited from a holding potential of Ϫ60 mV by a 5-ms test pulse to ϩ20 mV. After 1 s a 10-ms prepulse to ϩ150 mV was applied, and a second 5-ms test pulse to ϩ20 mV was elicited after stepping back for 2 ms to Ϫ60 mV. The facilitation ratios throughout the experiments were determined by dividing the peak current of test pulse 2 by the peak current of test pulse 1. The current traces indicate that in the presence of G␤ 2 full-length (middle trace) and G␤-(271-305) (lower trace) facilitation is increased over facilitation in the absence of activated G proteins (upper trace). B, facilitation ratios were determined as described in A for whole cell patch clamp recordings of P/Q-type channels expressed in HEK293 cells transfected without or with G␤ 2 or the indicated G␤ 2 peptide fragments. Numbers on the left indicate the length of the G␤ 2 deletion constructs in amino acids, i.e. 70 or 35 amino acids. The number in parentheses indicates the number of experiments. **, p Ͻ 0.01, ANOVA. C, voltage dependence of activation for P/Q-type channels expressed in HEK293 cells co-expressed with or without G protein ␤ peptides. Voltage pulses were elicited from a holding potential of Ϫ60 mV to various step potentials in 5-mV steps. Tail currents were related to the largest tail current and plotted in %. The voltage dependence of activation was determined by a Boltzmann fit according to I/I max ϭ 1/(1 ϩ exp((Vh Ϫ V)/K))), where I/I max is the normalized tail current; Vh is the half-activation voltage, and K determines the steepness of the curve. The current traces and plots indicate that in the presence of G␤-(201-340) and for the control where GTP␥S was used in the intracellular solution, the voltage-dependent activation curve of the expressed P/Q-type channels is shifted to more positive potentials in comparison to currents recorded with an intracellular solution containing only ATP. The inset shows the tail currents of the traces above at ϩ20 and ϩ65 mV at a larger time scale to illustrate that tail currents were fast throughout the experiments (see also the supplemental table). The activation curves represent single examples of cells for the indicated condition. D, voltage dependence of activation (half-activation voltage (Vh)) for P/Q-type channels transfected without or with G␤ 2 or the indicated G␤ 2 peptide fragments. Numbers on the left indicate the length of the G␤ 2 deletion constructs in amino acids, i.e. 70-or 35-amino acids. The number in parentheses indicates the number of experiments. *, p Ͻ 0.05; **, p Ͻ 0.01, ANOVA.
protein ␤ subunit were able to either induce or block G protein modulation of the P/Q-type channels when co-expressed together in a heterologous expression system. We divided the human G protein ␤ 2 subunit consisting of 340 amino acids into three domains encoding amino acids 1-100, 101-200, and 201-340. We then further subdivided these protein domains into smaller fragments to identify small protein regions that can be synthesized and directly applied to ion channels (e.g. GIRK channels and neuronal Ca 2ϩ channels) and to investigate the potential of these peptides in manipulating G protein-modulated effector proteins. The distribution of the YFP-tagged G␤ constructs in HEK293 cells is shown in the supplemental Figs. 1-3.
C-terminally Derived G Protein ␤ 2 Deletion Constructs Induce P/Q-type Channel Modulation in HEK293 Cells-We first investigated if G protein ␤ 2 -derived constructs induce modulation of P/Q-type channels comparable with levels observed with the G protein ␤ 2 subunit alone ( Fig. 1, A, middle trace, and B). G protein modulation was measured as prepulse facilitation, a characteristic of G protein-mediated voltage-dependent inhibition of the Ca v 2 channel family (9). To investigate prepulse facilitation, we compared the peak current amplitude of currents elicited by the two test pulses throughout the experiments. The test pulses were separated by a depolarizing pulse to relief G protein modulation from the Ca 2ϩ channel.
As indicated in Fig. 1, A and B, prepulse facilitation of P/Q-type channels was observed in the presence of the G␤-(201-340) but not in the presence of G␤-(1-100) and G␤-(101-200). Thus, the C terminus of the G protein ␤ 2 subunit induces P/Q-type channel modulation.
We next investigated whether the modulatory G␤ 2 -derived constructs induce a shift in the voltage dependence of activation, as would be predicted for the voltage-dependent G protein modulation of P/Q-type channels. In order to do so we analyzed the tail currents after the voltage pulse for the applied voltage pulse protocols. Indeed, as shown in Fig. 1, C and D, an ϳ8-mV shift in the voltage dependence of activation to more depolarized potentials was observed for P/Q-type channels expressed with G␤-(201-340), G␤-(201-270), G␤-(271-340), G␤-(271-305), and G␤-(305-340) in comparison to channels expressed without the G␤ fragments. In contrast, G␤-(1-100), G␤-(101-200), and G␤-(201-235), which did not induce facilitation, also did not shift the voltage dependence of activation to more positive potentials. G␤-(236 -270) shifted P/Q-type channel activation slightly to more positive potentials. The data suggest that co-expression of the protein fragments derived from the C-terminal end of the G protein ␤ 2 subunits induce P/Q-type channel modulation comparable with the modulatory effects described for the full-length G␤ 2 subunit.
N-terminally Derived G Protein ␤ 2 Deletion Constructs Reduce P/Q-type Channel Modulation in HEK293 Cells-We next investigated whether the modulation of P/Q-type channels is reduced in the presence of G protein ␤ fragments when endogenous G proteins are activated by GTP␥S. As shown in Fig. 2B, we observed that the first 100 amino acids (G␤-1-100) of the G protein ␤ subunit decreased significantly G protein modulation when endogenous G proteins were activated by GTP␥S, whereas P/Q-type channel modulation was not altered in the presence of the G␤-(101-200) and G␤-(201-340) deletion constructs.
We next analyzed smaller deletion constructs of the first 100 amino acids of G␤ 2 . Whereas the 1-50 construct reduced facilitation significantly, the 51-100 peptide did not. Furthermore, the two 25 amino acids long constructs, G␤-(1-25) and G␤-(26 -50) (Fig. 2B) were sufficient for a reduction in modulation. Thus, our data suggest that the N-terminal region of the G protein ␤ subunit can diminish P/Q-type channel modulation in HEK293 cells.
Application of Synthesized G Protein ␤ 2 -derived Peptides, Pep1-25 and Pep271-305, Inhibit or Induce G Protein Modulation of P/Q-type, Neuronal Non-L-type, and GIRK Channels-In our overexpression study, we found that the N-terminal region of the G protein ␤ 2 subunit inhibits G protein modulation of P/Q-type channels, whereas the C-terminal part induces modulation. We next investigated if direct application of synthesized peptides reproduced the effects observed with the overexpression of the G␤ constructs. We therefore synthesized one 25-amino acid-long peptide derived from the N terminus of G␤ 2 containing amino acids 1-25 (pep1-25) and one 35-amino acid long peptide derived from the C terminus containing amino acids 271-305. The reason for choosing these regions was that we observed strong effects on inhibition and induction of P/Q-type channel modulation for the corresponding cDNA constructs (see Figs. 1 and 2). As a control peptide, we synthesized a peptide that contained the amino acids of pep1-25 in a random order (pep1-25scrambled). In addition to testing these peptides on the P/Q-type channel modulation in HEK293 cells, we also wanted to investigate if these peptides are specific for P/Q-type channels or if other G protein ␤␥-modulated channels, e.g. neuronal non-L-type channels and GIRKs, are also affected.
We first tested the effects of the synthesized peptides on Ca 2ϩ channel modulation, and we applied the peptides to the intracellular side of the P/Q-type channels expressed in HEK293 cells. Intracellular application of pep1-25 reduced the GTP␥S-mediated P/Q-type channel facilitation significantly, whereas neither the pep271-305 nor the pep1-25scrambled had any effect (Fig. 3, A and B). To rule out the possibility that pep1-25 acts through activation of PKC, which would also result in a reduction in P/Q-type channel modulation (14,34), we performed the pep1-25 experiments in the presence of the PKC inhibitor peptide. In these experiments P/Q-type facilitation was still reduced when pep1-25 and PKC inhibitor peptide were co-applied to the channel, suggesting that pep1-25 does not reduce channel modulation via activation of PKC. To demonstrate that the PKC inhibitor peptide can antagonize the action of PKC on the P/Q-type channel, we tested the PKC inhibitor peptide in the presence of the phorbol ester PMA. Extracellular application of 1 M PMA to P/Q-type channels 5 min prior to the experiments reduced GTP␥S-mediated modulation to control levels, whereas in the presence of 100 M PKC inhibitor peptide in the intracellular solution and 1 M PMA in the extracellular solution the GTP␥S-mediated facilitation was not reduced (Fig. 3B). We next tested the effects of pep271-305. As shown in Fig. 3, A and B, pep271-305 increased P/Q-type facilitation comparable with the facilitation ratio measured in the presence of the full-length G␤ 2 subunits, whereas pep1-25 or pep1-25scrambled had no effect. The time course of the peptide-mediated channel modulation occurred within 1 min after seal formation and was significantly faster than the inhibitory effect of pep1-25 (Fig. 3C). We next investigated if the modulatory effect of pep271-305 and G␤ 2 is occluded by the pep1-25 peptide. Application of pep1-25 neither significantly reduced the pep271-305 nor the G␤ 2 full-length (fl) induced prepulse facilitation of the P/Q-type channels, suggesting that pep271-305 acted independently from pep1-25. The results indicate that the pep1-25 inhibits and the pep271-305 induces P/Q-type channel modulation.
The pep271-305-mediated P/Q-type channel modulation re-sembles channel modulation induced by G␤ 2 subunits. To investigate if the effects on channel modulation are analogous to the action of G␤ 2 fl, we characterized more extensively the properties of the pep271-305 peptide on Ca 2ϩ channel modulation according to studies performed previously (see Refs. 35 and 36). We first compared the shift in the voltage dependence of activation between GTP␥S-, G␤ 2 fl-, and pep271-305-modulated channels, and we found that in all cases the voltage dependence of activation is shifted to more positive potentials (Fig. 4, A and B). Application of high positive prepulses to ϩ150 mV shifted the voltage dependence of activation curve for GTP␥S-, G␤ 2 fl-, and pep271-305-modulated channels by 7-9 mV to more negative potentials, and the curves became steeper (Fig. 4, C and D). This result indicates that the amount of inhibition relief is comparable between G␤ 2 fl and pep271-305. We next compared the voltage-dependent relief from channel inhibition by increasing the voltage of the prepulse. In this experiment, we increased the test pulse duration to 20 ms to reveal possible changes on the steady state currents by pep271-305 in comparison to GTP␥S and G␤ 2 fl (Fig. 4E). 50% of G protein relief is caused by prepulse voltages around 60 mV for GTP␥S-, G␤ 2 fl-, and pep271-305-modulated P/Q-type channels (Fig. 4F). No differences in the facilitation ratios of GTP␥S-, G␤ 2 fl-, and pep271-305-modulated channels were observed when ratios where taken after 5 or 20 ms during the test pulse after the prepulse (Fig. 4G). According to Arnot et al. (36), the G␤ 2 subunits induce a kinetic slowing of P/Q-type channel currents by a factor of around 1.4 when the activation time of the test pulse after the prepulse was compared with the activation time of the test pulse before the prepulse. Comparing 5-ms test pulses to ϩ20 mV before and after application of a prepulse to ϩ150 mV between channels modulated by G␤ 2 or FIG. 3. G protein ␤ 2 -derived peptides (pep1-25 and pep271-305) reduce or induce P/Q-type channel modulation in HEK293 cells. A, pep1-25 reduces and pep271-305 induces P/Q-type channel modulation in HEK293 cells. Ba 2ϩ current traces are shown in the presence or absence of G protein ␤ peptides without (upper trace) or with GTP␥S (lower trace) in the intracellular recording solution. Ba 2ϩ currents and facilitation ratios were elicited and determined as in Fig. 1A. The current traces indicate that in the presence of pep271-305 P/Q-type channel facilitation is induced, whereas pep1-25 reduces the GTP␥S-mediated P/Q-type channel facilitation. Peptides were applied through the patch pipette in a concentration of 100 M. B, facilitation ratios were determined as described in Fig. 1A for P/Q-type channels expressed in HEK293 cells in the absence or presence of GTP␥S and of the indicated G␤ 2 -derived peptides (pep1-25 and pep271-305). Peptides were applied through the patch pipette in a concentra- FIG. 4. G protein ␤ 2 -derived peptide pep271-305 and full-length G␤ 2 induce analogous effects on P/Q-type channel modulation. A, voltage dependence of activation for P/Q-type channels expressed in HEK293 cells in the presence or absence (control) of GTP␥S, G␤ 2 fl, or pep271-305. Voltage pulses were elicited from a holding potential of Ϫ60 mV to various step potentials in 5-mV steps. Tail currents were related to the largest tail current and plotted in %. The voltage dependence of activation was determined by a Boltzmann fit according to I/I max ϭ 1/(1 ϩ exp((Vh Ϫ V)/K))), where I/I max is the normalized tail current; Vh is the half-activation voltage, and K determines the steepness of the curve. The plots indicate that in the presence of G␤ 2 fl or pep271-305 and where GTP␥S was used in the intracellular solution the voltage-dependent activation curve of the expressed P/Q-type channels is shifted to more positive potentials in comparison to currents recorded with an intracellular solution containing only ATP (see also supplemental data for tail current kinetics). The activation curves represent single examples of cells for the indicated condition. B, voltage dependence of activation (half-activation voltage (Vh)) for P/Q-type channels expressed in HEK293 cells in the presence or absence (control) of GTP␥S, G␤ 2 fl, or pep271-305. The number in parentheses indicates the number of experiments. **, p Ͻ 0.01, ANOVA. C, voltage dependence of prepulse facilitation for P/Q-type channels expressed in HEK293 cells in the presence of G␤ 2 fl or pep271-305. A 5-ms voltage pulse (test pulse 1) was elicited from a holding potential of Ϫ60 mV. After 1 s, a 10-ms prepulse to ϩ150 mV was applied. After stepping back for 2 ms to Ϫ60 mV, a second 5-ms voltage pulse was elicited. The potential of the test pulse was increased in 5-mV steps from Ϫ10 to ϩ65 mV. Tail currents were related to the largest tail current and plotted in %. As described in A, the voltage dependence of activation was determined by a Boltzmann fit for the activation curve before and after the prepulse. The plots indicate that the prepulse shifts the voltage dependence of activation to more negative potentials for P/Q-type channels co-expressed with G␤ 2 fl or for channels where the pep271-305 was applied through the patch pep271-305, the activation kinetics were slowed by a factor of 1.2 (supplemental Table). We next compared the inhibition kinetics between the modulated channels. The recovery of P/Qtype channel modulation was analyzed by increasing the prepulse duration. For GTP␥S-, G␤ 2 fl-, and pep271-305-modulated channels, the recovery time constants were 7-8 ms (Fig.  4, G and H) These values are in agreement with results published previously for G␤ 2 fl-mediated inhibition of P/Q-type channels (36). The time course of G protein re-inhibition was analyzed by increasing the time between prepulse and the following test pulse. Re-inhibition was faster for channels modulated by pep271-305 in comparison to G␤ 2 fl and GTP␥S. The re-inhibition of the P/Q-type channel is predicted to depend on the amount of applied or available G proteins (37). It is therefore likely that the peptide concentration delivered through the patch pipette on the channel, at least in this set of experiments, is higher than the G␤ 2 fl subunit concentration in the cell. Thus our data revealed no significant differences on the voltage dependence of activation, the voltage-dependent relief of the G protein modulation, the effect on steady state current, and the recovery kinetics of the modulation between G␤ 2 fl and pep271-305, suggesting that pep271-305 mimics the action of the G␤ 2 fl subunit.
We next investigated if these peptides also act on neuronal voltage-gated Ca 2ϩ channels. 10 -14-Day-old cultured hippocampal neurons were analyzed for their expression of non-L-type Ca 2ϩ channels. Non-L-type Ca 2ϩ currents were isolated by blocking Na ϩ and K ϩ channels with TTX and TEA and voltage-gated L-type channels with nimodipine (Fig. 5A). Peptides were delivered by the chariot transfection reagent 20 min before the recordings. As observed for the heterologously expressed recombinant P/Q-type channels and shown in Fig. 5, pep1-25 decreased the facilitation ratio of neuronal non-L-type Ca 2ϩ currents in the presence of GTP␥S. Neither the pep271-305 nor the chariot transfection reagent alone reduced significantly GTP␥S-mediated non-L-type channel modulation. In contrast, application of the pep271-305 in the absence of GTP␥S increased Ca 2ϩ channel facilitation, whereas neither the chariot transfection reagent alone nor the pep1-25 had an effect on channel modulation. Because cultured hippocampal neurons develop extensive dendritic and axonal processes, proper voltage control of the somatic neuronal Ca 2ϩ channels is difficult, and the results have to be interpreted carefully. However, the effects of the peptides on neuronal Ca 2ϩ channels resemble the effects observed on P/Q-type channels expressed in HEK293 cells.
We next analyzed the action of these peptides on the G protein modulation of GIRK channels. We co-expressed GIRK1 and GIRK2 subunits together with the muscarinic AChR-M2, which couples to pertussis toxin-sensitive G proteins and activates GIRK channels (38). We first showed that application of 10 M ACh increased inward basal GIRK currents by 54% measured at Ϫ80 mV (Fig. 6A, top trace). Direct application of the pep1-25 peptide through the patch pipette decreased the basal GIRK current by 15% (Fig. 6A, middle trace). In the presence of pep1-25, application of ACh only increased the GIRK current by 4%, indicating that the pep1-25 peptide blocks G protein-coupled receptor-mediated activation of GIRK channels. This effect was peptide-specific, because application of pep1-25scrambled reduced basal GIRK current by 2%, but extracellular application of ACh still increased the GIRK current by 31%. In contrast to the inhibitory action of pep1-25 on GIRK channel activity, the C-terminally derived peptide pep271-305 increased the basal GIRK current by 26%. Application of ACh 10 min after peptide application further increased the GIRK current (Fig. 6A, bottom trace) by 60%. The time constants for the peptide delivery through the patch pipette was significantly faster for pep271-305 in comparison to the pep1-25, whereas peptide delivery via chariot transfection reagent had a time constant of ϳ200 s for both peptides (pep1-25 and pep271-305) (Fig. 6D). These results indicate that the pep271-305 increases basal GIRK currents.
In summary, the G␤ 2 -derived N-terminal 25-amino acid-long peptide reduces GIRK and Ca 2ϩ channel modulation, whereas the synthesized G␤ 2 -derived C-terminal 35-amino acid-long peptide induces GIRK activation and Ca 2ϩ channel modulation. The effects seem to be peptide-specific because the inhibition of Ca 2ϩ channel modulation is only achieved by pep1-25 but not by pep271-305 or pep1-25scrambled, whereas the induction of Ca 2ϩ channel modulation is only mediated by pep271-305 but not by the pep1-25 or pep1-25scrambled.
Peptide Pep1-25 Decreases the FRET Signal between G Protein ␤ 2 and ␥ 3 Subunits in Living HEK293 Cells-In order to investigate a possible mechanism for the reduction in GIRK pipette (see also supplemental data for tail current kinetics). The activation curves represent single examples of cells for the indicated condition. D, shift in the voltage dependence of activation between activation curves before and after the prepulse for non-modulated P/Q-type channels or channels modulated by GTP␥S, G␤ 2 fl, or pep271-305. For calculating ⌬Vh (mV), the half-activation voltage of the activation curve after the prepulse was subtracted from the half-activation voltage of the activation curve before the prepulse (⌬Vh ϭ Vh2 Ϫ Vh1). The number in parentheses indicates the number of experiments. **, p Ͻ 0.01, ANOVA. E, voltage dependence of relief from G protein inhibition for P/Q-type channels expressed in HEK293 cells in the presence of G␤ 2 fl or pep271-305. From a holding potential of Ϫ60 mV, a prepulse was elicited for 10 ms to various step potentials (step size 5 mV). After 10 ms, a 20-ms test pulse to ϩ20 mV was elicited. Test current traces are shown for prepulses to Ϫ10, ϩ20, ϩ120, and ϩ150 mV for P/Q-type channels modulated by G␤ 2 fl and pep271-305. F, voltage dependence of relief from G protein inhibition for P/Q-type channels expressed in HEK293 cells in the presence of GTP␥S, G␤ 2 fl, or pep271-305. The amplitude of test currents was measured after 15 ms and normalized to the current amplitude produced by the most positive prepulse (ϩ150 mV). The curves were fitted with a sigmoidal curve, and the voltages to elicit half-maximal current increase were determined (see supplemental Table). No difference between the voltage-dependent relief from G protein inhibition was observed. G, comparison between facilitation ratios measured after 5 or 20 ms during the test pulse for P/Q-type channels expressed in HEK293 cells in the presence or absence (control) of GTP␥S, G␤ 2 fl, or pep271-305. From a holding potential of Ϫ60 mV, 10 ms after a 10-ms prepulse to Ϫ10 or ϩ150 mV, a test pulse was applied to 20 mV for 20 ms. Facilitation ratio was determined at different time points of the test pulse (5 and 20 ms) by dividing the test pulse current after a prepulse to Ϫ10 mV by the test pulse current after a prepulse to ϩ150 mV (see E). The number in parentheses indicates the number of experiments. H, comparison between the re-inhibition and recovery from G protein inhibition of P/Q-type channels modulated by G␤ 2 fl or pep271-305. To investigate the re-inhibition of P/Q-type channels, Ba 2ϩ currents were elicited from a holding potential of Ϫ60 mV by a test pulse to ϩ20 mV. After 1 s a 10-ms prepulse to ϩ150 mV was applied, and a second 5-ms test pulse to ϩ20 mV was elicited after stepping back to Ϫ60 mV. The time between prepulse and second test pulse was increased between 1 and 195 ms by 1 ϩ 1.5^x, where x is the number of sweeps. The facilitation ratio was plotted versus the time interval between prepulse and test pulse. A single exponential fit of this curve was used to determine the time constant of re-inhibition. To investigate the recovery from inhibition, the same voltage protocol as described above was used with two alterations. The time of the prepulse but not the time interval between prepulse and test pulse was increased between 1 and 20 ms by 1 ϩ 1.5^x, where x is the number of sweeps. The time interval between prepulse and test pulse was 2 ms. The facilitation ratio was plotted versus the time interval between prepulse and test pulse. A single exponential fit of this curve was used to determine the time constant of recovery of P/Q-type channel inhibition. The re-inhibition and recovery curves represent single examples of cells for the indicated condition. I, time constants for the re-inhibition and recovery from inhibition for P/Q-type channels modulated by GTP␥S, G␤ 2 fl, or pep271-305. Time constants were determined as described in H. The number in parentheses indicates the number of experiments. *, p Ͻ 0.05, ANOVA. and Ca 2ϩ channel modulation in the presence of the peptide pep1-25, we investigated the interaction between G protein ␤ and ␥ subunits. Considering the x-ray structure of the G protein, the N terminus (first 48 amino acids) of the G protein ␤ subunit is expected to interact with the G protein ␥ subunit ( Fig. 9) (39, 40). Thus, it is possible that the pep1-25 peptide may act by perturbing the interaction between G protein ␤␥ subunits. To demonstrate the interaction and/or proximity between the G protein ␤ 2 and G protein ␥ 3 subunits, we performed fluorescent resonance energy transfer measurements based on the method of Erickson et al. (27 and see Ref. 7). The G protein ␤ subunit was used as the acceptor (tagged to YFP), whereas the G protein ␥ subunit (tagged to CFP) was used as a fluorescent donor in the FRET assay. Both fluorophores were tagged to the N terminus of the G protein subunit. According to the crystal structure, both fluorophores should be in close proximity because the N termini of the G protein ␤ and ␥ lie next to each other. We first verified that both constructs localized to the cell membrane when expressed in HEK293 cells (Fig. 7A). We next analyzed if these two proteins were in close proximity and thus able to induce FRET. The most accepted method to demonstrate FRET is the dequenching of the CFP emission by acceptor (YFP) photobleaching. As shown in Fig. 7B, 30 min of photobleaching increased the CFP emission of G␤ 2 -YFP-and G␥ 3 -CFP-transfected cells by 15%. As a positive control for the acceptor bleaching experiments, we used a chameleon Ca 2ϩ sensor. Within this protein CFP is fused to YFP via calmodulin and the calmodulin-binding peptide of the myosin light chain kinase (41). After 30 min of acceptor bleaching, the CFP fluorescence increased by 21%. In contrast the non-interacting protein pairs (G␤ 2 -YFP ϩ CFP, G␥ 3 -CFP ϩ YFP, CFP alone, and CFP ϩ YFP) expressed in HEK293 cells revealed only a slight increase in CFP emission. As an additional negative control, we analyzed the FRET signal between a membranelocalized protein and the G protein subunits (i.e. ␥ 2 (stargazin)-CFP ϩ G␤ 2 -YFP and G␥ 3 -CFP ϩ ␥ 2 (stargazin)-YFP). The constructs did not induce FRET. These results indicate that G protein ␤␥ subunits induced a FRET signal.
We next analyzed if the FRET signal between G protein ␤␥ subunits in HEK293 cells is altered when the peptides (pep1-25 and pep271-305) were delivered into the cells via the chariot transfection reagent. We calculated first the average FRET signal between G␤␥ subunits for transfected HEK293 cells and then applied the peptide dissolved in the chariot transfection reagent to the cells (Fig. 7C). After 30 min of incubation time we again calculated the FRET signal of the same cells analyzed before. We found that 30 min after application of the pep1-25, the average FRET signal for G␤ 2 ␥ 3 was reduced significantly in the presence of the pep1-25. This is also demonstrated in Fig. 7D, where a strong FRET signal between G␤ 2 ␥ 3 was detected at the outer rim of the cell, probably representing the cell membrane. This signal decreased significantly after 30 min (Fig. 7D). In contrast neither the pep271-305 nor the chariot transfection reagent alone altered the average FRET signal between G protein ␤␥ subunits. These data indicate that the peptide pep1-25 decreases the FRET FIG. 5. G protein ␤ 2 -derived peptides (pep1-25 and pep271-305) reduce or induce neuronal non-L-type channel modulation. A, example current trace of a non-L-type current elicited by a 500-ms-long voltage ramp protocol from Ϫ60 to ϩ80 mV. The current was measured in the presence of TTX, nimodipine, and TEA to block Na ϩ , K ϩ , and L-type Ca 2ϩ channels. B, pep1-25 reduces and pep271-305 induces non-L-type channel modulation in cultured hippocampal neurons. Ca 2ϩ current traces in the presence or absence of G protein ␤ peptides with (upper trace) or without GTP␥S (lower trace) in the intracellular recording solution. Ca 2ϩ currents were elicited from a holding potential of Ϫ60 mV by a 10-ms test pulse to ϩ5 mV. After 2 s a 10-ms prepulse to ϩ100 mV was applied, and a second 10-ms test pulse to ϩ5 mV was elicited after stepping back for 10 ms to Ϫ60 mV. Facilitation ratios were determined as described in Fig. 1A. The current traces indicate that in the presence of pep271-305 P/Q-type channel facilitation is induced, whereas pep1-25 reduces the GTP␥S-mediated P/Q-type channel facilitation. 20 min before the recordings, peptides were delivered via the chariot transfection reagent in a concentration of 10 M. signal between G protein ␤ 2 and ␥ 3 subunits probably by changing the orientation or distance between the attached CFP and YFP.
Peptide Pep271-305 Increases the FRET Signal between GIRK1/2 Subunits in Living HEK293 Cells-We next investigated the effects of pep271-305 on GIRK channel subunit interaction. Recently, it has been shown that activation of the muscarinic ACh receptor M2 or co-expression of G protein ␤␥ altered the FRET signal between CFP and YFP-tagged GIRK subunits (42). We therefore asked whether we could change the FRET signal between YFP-tagged GIRK1 and CFP-tagged GIRK2 subunits when we co-expressed the G protein ␤␥ subunits and whether the G protein-mediated FRET change could be elicited by pep271-305. Again we first verified the FRET signals for the given protein interactions with acceptor bleaching experiments (Fig. 8B), and we compared the results to the measured average FRET signal for the given interaction (Fig.  8C). These experiments were performed on transfected fixed HEK293 cells. We found that co-expressed GIRK2-CFP and GIRK1-YFP induced a robust average FRET signal, and that after YFP bleaching the CFP emission increased for this FRET pair by 9%. Both constructs localized to the same areas in HEK293 cells as indicated in Fig. 8A. Co-expression of G protein ␤␥ increased the FRET signal between GIRK1/2 significantly, which correlated well with a significant increase in the CFP emission to 15%. In contrast GIRK1-YFP co-expressed with cytosolic CFP or membrane-targeted stargazin-CFP and GIRK2-CFP co-expressed with cytosolic YFP or membranetargeted stargazin-YFP did not induce a FRET signal.
We next investigated in live HEK293 cells if the G protein ␤␥-mediated FRET increase between GIRK1/2 subunits could be mediated by the pep271-305 peptide. We therefore calculated the FR for GIRK1/2 before and after application of the pep271-305 peptide dissolved in the chariot transfection reagent, and we found that the pep271-305 increased FR significantly. In contrast the application of the pep1-25 or the chariot transfection reagent alone did not induce a change in FR. We further verified our results by analyzing the change in FR over FIG. 6. G protein ␤ 2 -derived peptides (pep1-25 and pep271-305) reduce or induce GIRK channel modulation in HEK293 cells. A, K ϩ current traces of GIRK channels expressed in HEK293 cells in the absence and presence of the G protein ␤ peptides pep1-25 and pep271-305 before and after GIRK channel activation via muscarinic ACh receptors. GIRK channel subunits 1 and 2 were co-expressed with muscarinic AChR-M2 in HEK293 cells. K ϩ currents through GIRK channels were elicited by a voltage ramp from Ϫ100 to ϩ50 mV with a duration of 50 ms (control; thick line). 10 M ACh was applied after 30 s to activate mAChR-M2. In the absence of G␤ 2 -derived peptides, activation of mAChR-M2 increased the GIRK current amplitude (ϩACh; thin line; upper trace). Peptides were applied through the patch pipette in a concentration of 100 M, and GIRK currents were recorded immediately after whole cell configuration was achieved (control; thick line). Peptides were allowed to diffuse into the cell for 5-10 min until the GIRK current amplitude was stable (ϩpeptide; dotted line). After stabilization of the GIRK current amplitude in the presence of the peptide, ACh was applied (ϩAChϩpeptide; thin line). In the presence of the pep1-25 peptide the control current was reduced, and ACh application only caused a small increase in GIRK currents (middle trace). In contrast, in the presence of the pep271-305 peptide GIRK current amplitude was increased, and ACh-mediated GIRK currents were further increased (lower trace). B, current change (increase or decrease) in GIRK current amplitude in the presence and absence of G protein ␤ 2 -derived peptides (pep1-25 and pep271-305) before and after activation of mAChR-M2. Positive current indicates an increase, and negative current indicates a decrease in GIRK current amplitude measured at Ϫ80 mV. ϩpep indicates the GIRK current change in the presence of peptide. ϩACh indicates the GIRK current change after application of 10 M ACh in the absence or presence of pep1-25, pep271-305, or pep1-25scrambled. The number in parentheses indicates the number of experiments. C, example traces of the time course of decrease and increase of GIRK currents in the presence of pep1-25 during ACh application (upper trace) and pep271-305 before ACh application (lower trace). Currents were elicited as described in A. Each point represents the current measured at Ϫ80 mV within the ramp protocol. D, time constants of the increase and decrease of GIRK currents in the presence of pep1-25 during ACh application and pep271-305 before ACh application applied through the patch pipette or via the chariot transfection reagent. The number in parentheses indicates the number of experiments. **, p Ͻ 0.01, ANOVA. time. HEK293 cells were transfected with GIRK1/2 subunits, and cells were monitored before and after application of the pep271-305 peptide (Fig. 8, D and E). As indicated in Fig. 8D an increase in the FR was measured after application of the pep271-305 but not pep1-25. This is also demonstrated in Fig.  8E. Here a larger FR was detected in particular at the outer edge of the cell (probably membrane region) after application of the peptide. These results indicate that the pep271-305 peptide changes the FRET signal between GIRK1/2 subunits probably by slightly reorienting the GIRK1/2 subunits.

Circular Dichroism Spectra Predicts ␣-Helical Structures for
Pep1-25-In order to investigate whether the synthesized peptides used in our experiments are structured, we analyzed the CD spectra of the peptides. Far-UV CD spectra of pep1-25 showed features that are characteristic for polypeptides that are significantly helical with a maximum near 190 and minima at 208 and 222 nm (Fig. 9B). However, the data demonstrate also that this peptide is not fully helical, because the minimum at 205 is slightly shifted and the ratio of ellipticity at 222 and 208 nm is still not close or is greater than 1. Estimates for the FIG. 7. G protein ␤ 2 subunit-derived peptide, pep1-25, reduces the FRET signal between YFP-tagged G protein ␤ 2 and CFP-tagged G protein ␥ 3 subunit. A, distribution of YFP-tagged G protein ␤ 2 and CFP-tagged G protein ␥ 3 subunits in HEK293 cells. Deconvolved images of the co-distribution of YFP-G␤ 2 (left) and CFP-G␥ 3 (right) in HEK293 cells 24 h after transfection. Note fluorescence for both constructs is brighter at the edges, suggesting membrane localization of G␤ 2 and G␥ 3 . B, acceptor photobleaching for FRET pairs co-transfected in HEK293 cells: CFP-G␥ 3 ϩ YFP-G␤ 2 ; CFP ϩ YFP-G␤ 2 ; CFP-G␥ 3 ϩ YFP; CFP; CFP ϩ YFP; the chameleon Ca 2ϩ sensor protein; CFP-G␥ 3 ϩ YFP-(stargazin)␥ 2 and CFP-(stargazin)␥ 2 ϩ YFP-G␤ 2 . CFP fluorescence for each FRET pair was compared before and after 30 min of YFP photobleaching. The efficiency reflects the increase in CFP fluorescence. The number in parentheses indicates the number of experiments. *, p Ͻ 0.05; **, p Ͻ 0.01, ANOVA. C, FRET ratio for the FRET pair CFP-G␥ 3 ϩ YFP-G␤ 2 before and after application of G protein ␤ 2 -derived peptides pep1-25 and pep271-305. CFP, FRET, and YFP pictures were taken before application of the peptide, and the average FR was calculated (see "Materials and Methods"). Peptides dissolved in the chariot transfection reagent were applied to the live cells in a 10 M concentration. Cells exposed to the peptides were incubated for 30 min at room. After 30 min, CFP, FRET, and YFP pictures were taken again, and the average FR was calculated. The control FRET pairs CFP ϩ YFP-G␤ 2 , CFP-G␥ 3 ϩ YFP, CFP-G␥ 3 ϩ YFP-(stargazin)␥ 2 , and CFP-(stargazin)␥ 2 ϩ YFP-G␤ 2 were measured in fixed cells. (RD1 and RA1 constants used: G␤␥ ϩ pep1-25 (0.554;0.03); G␤␥ ϩ pep271-305 (0.551;0.028); G␤␥ ϩ chariot (0.558;0.036); G␤ ϩ CFP and G␥ ϩ YFP (0.568;0.054); CFP-G␥ 3 ϩ YFP-(stargazin)␥ 2 (0.597; 0.12); CFP-(stargazin)␥ 2 ϩ YFP-G␤ 2 (0.59; 0.118).) **, p Ͻ 0.01, ANOVA. D, example of a change in the YFP-G␤ 2 /CFP-G␥ 3 FR over time in the presence of the G protein ␤ 2 -derived peptide pep1-25. Upper, CFP, FRET, and YFP pictures taken from HEK293 cells transfected with YFP-G␤ 2 /CFP-G␥ 3 before peptide delivery. From these pictures an image was calculated (lower), where the FR per pixel is encoded in color value. The arrow indicates that a high FR is detected on the outer edge of the cell and that the FR is larger before application of the pep1-25 peptide. In contrast, the * indicates a cell with YFP-G␤ 2 expression but no or very little CFP-G␥ 3 expression and therefore no change in the FR. The bar on the right relates the calculated FR to color scale values. helical content are 22% at 25°C and 30% at 5°C. The results are in reasonable agreement with the 29% estimated from a sequence-based prediction for the pep1-25 peptide (43). Both a temperature increase and addition of 200 mM NaCl reduce helical content slightly, as may be expected from a weakening of intrapeptide hydrogen bonds and salt bridges, respectively. In contrast pep271-305 showed no appreciable ellipticity at 222 nm (Fig. 9B). Both estimates and predictions (also for pep1-25scrambled) gave helix content of less than 3%. Addition of 50% trifluoroethanol, a co-solvent that stabilizes helical structure, had only a marginal effect on pep271-305, confirming that they have little or no intrinsic propensity to adopt helical structure (44). The results are in agreement with the structure of the corresponding protein region in the G protein ␤␥ complex (Fig. 9A). DISCUSSION We demonstrate in this study that the G␤ 2 N-terminally derived G protein ␤␥ interacting domain reduces G protein modulation of GIRK and Ca 2ϩ channels and shows that the first 25 amino acids of this domain alters the interaction between the G protein ␤ and ␥ subunits. In addition we demonstrate that the G␤ 2 C-terminally derived peptide pep271-305 induces G protein modulation of GIRK and Ca 2ϩ channels, and we further show that this peptide increases the FRET signal between interacting GIRK channel subunits. Thus, both peptides act on multiple G protein-modulated target proteins. Possible mechanisms for these effects are discussed below.
Circular Dichroism Suggest That Pep1-25 Is Partly ␣-Helical, whereas Pep271-305 Is Unstructured in Solution-By contrast to the recent advent of designer peptides, which have been optimized for very high helical propensity, peptide fragments from proteins are generally known to contain only poorly stabilized secondary structures. Thus it is surprising that pep1-25 has a relatively strong helical propensity, which may aid folding or recognition, both in a kinetic and thermodynamic sense. A helix that is populated to 50% relative to unstructured state has a thermodynamic stability of ⌬G ϳ0 kcal/mol as ⌬G ϭ ϪRT ln (fraction helical). At first this value may appear inconsequential, but it is significant compared with peptides, such as pep271-305, that have no tendency to form helical structure, even in the helix-stabilizing co-solvent trifluoroethanol. In these cases additional interactions of considerably magnitude may be needed to stabilize the secondary structures. By comparison, pep1-25 is kinetically and thermodynamically primed to form a helical structure, which may be of considerable importance for the interaction of this region with the remainder of G␤, G␥, or the channels.

N-terminally Derived G Protein ␤ Peptides Reduce Ion Channel Modulation and Change the Interaction between G Protein
␤ and ␥ Subunits-G proteins consist of three subunits (␣, ␤, and ␥), which work together to confer extracellular transmitter signals to intracellular effectors (45). The crystal structure has recently been resolved and confirms the biochemical data that G protein ␤ and ␥ subunits tightly interact (39,40). Both N termini lie next to each other (Fig. 9A). Therefore, one would predict that fusion of suitable fluorophore combinations allowing FRET (e.g. CFP and YFP) to the N terminus of both G protein ␤ and ␥ subunits would demonstrate that the fluorophores come into close contact resulting in energy transfer. In fact, FRET between N-terminally YFP-tagged G␤ 1 and N-terminally CFP-tagged G␥ 2 was recently demonstrated by Ruiz-Velasco and Ikeda (46) and was also observed for the G protein subunits used in our study (G␤ 2 ␥ 3 ). In addition, biomolecular fluorescence complementation was achieved by fusing the separated N and C termini of YFP to the N termini of G␤ and G␥ (47). By applying the G␤ 2 N-terminally derived pep1-25 pep-tide, FRET between G protein ␤ and ␥ subunits was diminished, indicating that the orientation and/or distance between these two subunits had been altered in the presence of the peptide. Because the pep1-25 peptide comprises part of the binding site between G protein ␤␥ and shows a high degree of ␣-helical structure, it is most likely that this peptide acts like a wedge to separate the two proteins. In the presence of the peptide, we showed that modulation of GIRK and Ca 2ϩ channels was reduced. This may have important implications for the function of the G␤␥ subunit interaction site for the correct orientation and placement of the G protein ␤␥ complex in the membrane to allow protein target modulation.
However, other possibilities have to be considered, which may account for the loss of channel modulation in the presence of pep1-25. For example the peptide may interfere with GTP␥S binding and/or nucleotide-dependent stimulation of the endogenous G proteins, the peptide may terminate or accelerate the G protein signal, or pep1-25 may interfere with the G protein binding to the channel itself. We tried to address and rule out at least some of these issues in our studies. For example, we increased the GTP␥S concentration in the pipette 2-fold to 1.2 mM, but we could still observe the blocking effect on channel modulation of pep1-25 (data not shown), suggesting that at least at these GTP␥S concentrations the pep1-25 most likely does not interfere with the activation of endogenous G protein.
We also tried to exclude the possibility that pep1-25 activates PKC, which has been described as a mechanism to counteract N-and P/Q-type channel modulation (14,34). We therefore performed the experiments in the presence of the PKC inhibitor peptide but could still block GTP␥S-mediated Ca 2ϩ channel modulation with pep1-25, suggesting that pep1-25 does not act via PKC activation or competes with the PKC inhibitor peptide on PKC. It has to be mentioned at this point that other possibilities terminating G protein signaling such as up-regulation of RGS proteins or stimulation or stabilization of the reassociation of the heterotrimeric G protein (G␣␤␥) may still be able to occur, and we cannot exclude that these processes are involved in the reduction of channel modulation. We also tested if the pep1-25 may interfere with G protein ␤␥ binding to the channel. If this would be the case, then pep1-25 would interfere with the modulatory action of G␤ 2 or pep271-305 if these two proteins/peptides modulate P/Q-type channels via direct binding to the channel. We tested this possibility by applying pep1-25 together with pep271-305 and in the presence of exogenously expressed G␤ 2 subunits. In both cases modulation was still observed arguing against the interference of pep1-25 for binding of G␤ subunits or domains to the channel but rather point to an important interaction between G protein ␤ and ␥ subunits for G protein modulation. The results concerning the action of G␤ 2 or pep271-305 independent of G␥ and the change in G␤␥ interaction/orientation with loss in G protein signaling are puzzling. One interpretation might be that in isolation the modulatory domains within G␤ 2 can act independently on the channel, but in the G protein complex G␤␥ has to be oriented precisely on the effector protein.
The importance of the N-terminal G␤␥ interaction site is also demonstrated by the fact that Asn-35 and/or Asn-36 in the G␤ 1 subunit are necessary for sensing the PKC-mediated phosphorylated state of the N-type channel leading to stronger Ca 2ϩ channel inhibition (25). This suggests that this part of the G␤ subunit comes in close proximity to the binding site on the Ca 2ϩ channel intracellular loop I-II.
Given our observation that pep1-25 decreases the FRET signal between G␤␥ and reduces G protein modulation, the G␤␥ interaction site is most likely important for precise orientation of the G␤␥ complex on the target protein. It is interesting to note that the N-terminal part of the intracellular loop III-IV of the P/Q-type Ca 2ϩ channel ␣ 1 subunit reveals structural similarities to the G protein ␥ 2 subunit (50).
C-terminally Derived G Protein ␤ Peptides Induce Ion Channel Modulation and Change the Interaction between the Ion Channel Subunits-Several groups have tried to identify the amino acids within the G␤ subunit responsible for specificity and/or target protein function. One mutagenesis study analyzed G protein ␤-alanine mutations on various effector functions (23,51). These mutations could either increase, decrease, or abolish G␤␥-dependent interactions and were distributed over the entire G protein with clusters of amino acid residues in positions 51-100. Here, in particular the point mutations L55A and I80A largely increased the prepulse facilitation ratio of N-type Ca 2ϩ channels (23). Most interestingly, the L55A G␤ mutant attenuated specifically the modulation of P/Q-but not N-type channels (51). The position of this mutation is adjacent to the interaction site between G protein ␤␥, and mutations at this position may therefore alter the orientation of the G␤ protein relative to the G␥ subunits. Most of the identified amino acids are located on the large surface area of G␤, suggesting that this site of the G␤ subunit anchors the G protein to the channel. The recent study by Doering et al. (25) analyzed the G␤ subunit specificity for N-type channel modulation by creating chimeras between the G␤ 1 and G␤ 5 subunits. The authors found two 20-amino acid-long regions (positions 140 -168 and 186 -204) and an Asn-Tyr-Val motif that contributed to the specific action of the G protein for N-type channel modulation. Most interestingly, exchange of the N terminus (amino acid 1-47) and the C terminus (amino acid 280-C terminus) between the two subunits did not change the subunit specificity in Ca 2ϩ channel modulation. These are the two regions that were identified in our deletion approach. In particular, we found that the C-terminal region of the G protein ␤ 2 subunit was able to induce Ca 2ϩ channel modulation as well as GIRK channel activation. The three-dimensional structure reveals that the C terminus contains the WD domains WD6 and WD7. Most interestingly, this C-terminal site is distinct from the FIG. 9. Circular dichroism spectra and location of modulatory peptides within the G protein ␤␥ protein complex. A, structure of the G protein ␤␥ subunit complex (upper and lower surface and side view of the G protein shown as 180°and 90°rotations) and location of the modulatory peptides within the G protein ␤ subunit: G protein ␤ subunit (light gray) with amino acids 1-25 (red) and 271-305 (orange); G protein ␥ subunit (black). Structure is from Refs. 39 and 40. B, circular dichroism spectra of pep1-25 and pep271-305 at 25°C. from HEK293 cells transfected with YFP-GIRK1/CFP-GIRK2 before peptide delivery. From these pictures an image was calculated (lower) where the FR per pixel is encoded in the color values (see "Materials and Methods"). The arrow indicates that a high FR is detected on the outer edge of the cell and that the FR is further increased after application of the pep271-305 peptide. The bar on the right relates the calculated FRET ratio to color scale values. contact sites between G␤ and G␣ subunits in heterotrimeric G proteins (39,52), and it overlaps with the location of the binding site of phosducin, which inhibits G protein action (53). In addition, mutations in C-terminal peptide regions of G␤, which undergo a conformational change during phosducin binding and come into contact with the farnesyl group of G␥, led to decreases in activity on phospholipase C-␤ and adenylyl cyclase type II (54). Thus, the surface area including the WD repeat 6 and 7 (C-terminal end) may determine the effector function for several proteins, i.e. phosducin, phospholipase C-␤, adenylyl cyclase type II, GIRK, and voltage-gated Ca 2ϩ channels. It seems likely that during activation of the G protein, a conformational change within the G protein (55) induces a conformational change of the target protein, such as ion channels. A G␤␥-mediated change of a FRET signal has been demonstrated for P/Q-type channel ␣ 1 and ␤ subunits (7,8) and for the GIRK channel subunits (42). We used the GIRK1/2 interaction to demonstrate that the pep271-305 can induce a change in the FRET signal between the channel subunits. The physiological action of the peptide, which mimics the action of the G protein itself, and the relatively fast time course of its action on channel modulation in comparison to the pep1-25 suggest a direct interaction between the peptide and the channel proteins (i.e. GIRK and Ca 2ϩ channel). However, we do not know at this point if the peptide directly interacts with the channel to induce modulation, because we were not able to co-immunoprecipitate the HA-tagged 271-305 deletion construct with the P/Q-type Ca 2ϩ channel co-expressed in HEK293 cells (data not shown). Another critical point for induction of modulation by G protein-derived peptides is that for the overexpressed peptides it might be considered that differences in the targeting of the construct, aggregation within the cell, or different expression levels may account for the differential effects seen for modulation of the channels. Therefore, we looked at the localization and possible aggregation of the constructs when tagged to YFP. The constructs were either localized to the membrane or within the cytoplasm, and no aggregation of the constructs was observed within the analyzed time period, i.e. 24 h of expression. This excludes that G protein aggregation within the endoplasmic reticulum or Golgi might be the cause of the modulatory effects seen on the channels (supplemental Figs. 1-3). In addition, application of brefeldin A, which leads to 90% Golgi disassembly within 20 min (56), did not change the pep1-25 or pep271-305 action on GIRK currents, suggesting that these peptides do not block or activate G protein transport from Golgi structures to the plasma membrane (data not shown). Because the C-terminal region of the G␤ subunit also comes into contact with the G␥ subunit (see Fig. 9) and the G␣ subunit, we cannot rule out that the C-terminally derived peptides change the orientation between G protein ␤ and ␥ subunits relative to the target protein or release G protein ␤␥ subunits from the G␣␤␥ complex, which could explain the modulatory effect of the pep271-305. However, in our experiments (Fig. 7) we did not detect a change in the FRET signal between G␤␥ in the presence of pep271-305.
In summary, we have identified the structural domains of the G protein ␤ 2 subunit, which can either suppress (i.e. the G␤ 2 N-terminal G␥ interacting domain) or induce ion channel modulation (i.e. the G␤ 2 C-terminal domain). By using the synthesized peptides of these domains, we revealed that the peptides most likely change the orientation between the G protein ␤␥ subunits (pep1-25) or the GIRK channel subunits (pep271-305). Besides their importance for understanding ion channel modulation, the peptides will provide useful tools for manipulating the amount of G protein activation within a cell and could therefore be used as magic shotguns for treatment of complex nervous system disorders (57). For example, the most effective drug in the treatment of schizophrenia (clozapine) interacts with at least four classes of different G protein-coupled receptors (i.e. serotonin, ACh, adrenergic, and other biogenic amine receptors). We predict that the G protein ␤ subunit-derived peptides will modulate these intracellular signaling pathways.