Targeted mutagenesis of the farnesylation site of Drosophila Ggammae disrupts membrane association of the G protein betagamma complex and affects the light sensitivity of the visual system.

Activation of phototransduction in the compound eye of Drosophila is mediated by a heterotrimeric G protein that couples to the effector enzyme phospholipase Cβ. The γ subunit of this G protein (Gγe) as well as γ subunits of vertebrate transducins contain a carboxyl-terminal CAAX motif (C, cysteine; A, aliphatic amino acid; X, any amino acid) with a consensus sequence for protein farnesylation. To examine the function of Gγe farnesylation, we mutated the farnesylation site and overexpressed the mutated Gγe in Drosophila. Mass spectrometry of overexpressed Gγe subunits revealed that nonmutated Gγe is modified by farnesylation, whereas the mutated Gγe is not farnesylated. In the transgenic flies, mutated Gγe forms a dimeric complex with Gβe, with the consequence that the fraction of non-membrane-bound Gβγ is increased. Thus, farnesylation of Gγe facilitates the membrane attachment of the Gβγ complex. We also expressed human Gγrod in Drosophila photoreceptors. Despite similarities in the primary structure between the transducin γ subunit and Drosophila Gγe, we observed no interaction of human Gγrod with Drosophila Gβe. This finding indicates that human Gγrod and Drosophila Gγe provide different interfaces for the interaction with Gβ subunits. Electroretinogram recordings revealed a significant loss of light sensitivity in eyes of transgenic flies that express mutated Gγe. This loss in light sensitivity reveals that post-translational farnesylation is a critical step for the formation of membrane-associated Gαβγ required for transmitting light activation from rhodopsin to phospholipase Cβ.

Phototransduction in Drosophila and related flies represents the fastest G protein-coupled signaling system known to date (for reviews, see Refs. [1][2][3][4][5]. In this signaling pathway a rhodopsin molecule is activated by light absorption and transmits the signal to a heterotrimeric Gq protein consisting of G␣q, 1 G␤e, and G␥e. As a result of G protein activation G␣q couples to the norpA-encoded effector enzyme phospholipase C␤, which is assembled together with an eye-specific protein kinase C and the cation channel transient receptor potential protein into a signaling complex by the scaffolding protein INAD (inactivation no afterpotential D protein) (6 -9). Activated phospholipase C␤ hydrolyzes the membrane phospholipid phosphatidylinositol bisphosphate to form the second messengers diacylglycerol and 1,4,5-inositol trisphosphate. The phototransduction cascade terminates in the opening of cation channels composed of ion channel subunits of the transient receptor potential protein family (10) and thus results a depolarization of the photoreceptor cell membrane.
Most of the proteins mediating phototransduction in Drosophila have been identified, and their function has been determined with the aid of Drosophila mutants. These mutants exhibit distinct defects in the electrical response to light stimuli which can be recorded by measuring electroretinograms, transepithelial potentials reflecting light-dependent photoreceptor cell depolarization. Thus, the ␣ and ␤ subunits of the visual G protein, referred to as DGq (G␣q) and G␤e, respectively, have been cloned and functionally characterized (11)(12)(13)(14). The visual ␥ subunit (G␥e) has also been cloned (15). Characterization of G␥e revealed that it exhibits a carboxylterminal CAAX box (C, cysteine; A, aliphatic amino acid; X, any amino acid, Fig. 1A), which is a common motif of G protein ␥ subunits. ␥ Subunits are prenylated at the conserved cysteine residue of this motif (for review, see Refs. 16 and 17). Following prenylation, the carboxyl-terminal tripeptide (AAX) is cleaved off, and the newly exposed cysteine residue is usually methylated at its carboxyl terminus (Fig. 1A). With regard to the type of prenylation, G protein ␥ subunits can be divided in two groups that are either farnesylated (when the last amino acid of the CAAX box is serine, methionine, or glutamine) or geranylgeranylated (when the last amino acid is leucine or valine) (18 -21). Accordingly, Drosophila G␥e as well as the visual G␥ subunits of vertebrate transducins are likely to become farnesylated because they have a methionine or serine in the carboxyl-terminal position. In contrast, nonvisual G␥ subunits are typically geranylgeranylated. A second G␥ subunit isolated from Drosophila (G␥1, Ref. 22), which is expressed preferentially in the brain, has a consensus CAAX box for geranylgeranylation. The conservation of the farnesylation site of visual G␥ subunits of vertebrates and flies indicates that the farnesyl residue could contribute to a specific role of visual G␥ subunits in phototransduction which is common to both vertebrates and invertebrates, for example binding of the visual G protein to activated rhodopsin. For vertebrate transducin it has been shown that synthetic ␥ 1 peptides modified with a farnesyl residue interact more effectively with activated rhodopsin than peptides with a geranyl or geranylgeranyl residue (23,24). Evidence for a role of G␥ farnesylation in rhodopsin interaction is further provided by studies using recombinantly expressed ␤␥ complexes (25). In addition to the type of prenylation, the carboxyl-terminal sequence of the ␥ subunit also confers receptor selectivity (25).
Previous studies analyzing the function of post-translational isoprenylation of G␥ proteins were primarily carried out with synthetic peptides or heterologously expressed ␥ subunits which were then studied by in vitro assays. Although these studies provide important quantitative data, e.g. on the interaction of the ␤␥ complex with the receptor, they cannot account for effects caused by the complex structural and functional organization provided by the light-transducing compartment of a photoreceptor cell. In the present study we assessed the functional importance of the conserved farnesyl modification of visual G␥ subunits in vivo by generating transgenic flies that express a mutated G␥e without a farnesylation site. Transgenic flies expressing nonfarnesylated G␥e were investigated for alterations in membrane anchoring of the different G protein subunits, for ␤␥ complex formation, and for abnormalities in the light response. We show that overexpression of mutated G␥e interferes with membrane attachment of the ␤␥ complex and that it renders the photoreceptor cells less sensitive to light.

Generation of DNA Constructs and Transgenic Drosophila-In vitro
oligonucleotide-mediated mutagenesis was carried out by amplifying the coding region of Drosophila G␥e using the sequence-specific primers 5Ј-ATTACCCGGGATGGAACAAAAACTTATTTCTGAAGAAGATCTG-ATGGATCCCAGTGCTCTACA-3Ј (SmaI restriction site underlined and myc tag italicized) and 5Ј-GGCGCAAGCTTTTACATAATAACGCTTT-GCCC-3Ј (HindIII restriction site underlined; bold type indicates the miss-matching nucleotide) to induce the amino acid mutation C69G. A nonmutated, myc-tagged G␥e amplificate was generated using the same sequence-specific forward primer and the nonmutated reverse primer 5Ј-GGCGCAAGCTTTTACATAATAACGCATTTGCCC-3Ј (with HindIII restriction site underlined). The coding sequence of the human G␥rod subunit was amplified from an expressed sequence tag clone (accession number AA015841) and combined with a myc tag by using the forward and reverse primers 5Ј-ATTACCCGGGATGGAACAAAAA-CTTATTTCTGAAGAAGATCTGATGCCAGTAATCAATATTGAGGAC-C-3Ј (SmaI restriction site underlined and myc tag italicized) and 5Ј-G-GCGCAAGCTTTTATGAAATCACACAGCCTCC-3Ј (HindIII restriction site underlined), respectively. The obtained PCR products G␥eC69G, G␥e, and hG␥rod ( Fig. 1B) were cloned into the SmaI/HindIII restriction site of the pBluescript II SK vector (Stratagene), 3Ј of a Drosophila Rh1-promoter fragment (bp Ϫ833 to ϩ67; Ref. 26). The constructs were sequenced to confirm the correct identity of the expressed sequence tag clone and successful mutagenesis. Rh1 promoter and G␥ coding sequences were then cloned into the NotI/XhoI restriction site of the P-element transformation vector YC4 (a gift from S. Britt, University of Colorado; the YC4 vector is derived from the vector Y.E.S., Ref. 27). P-element-mediated transformation of host strain Drosophila yellow white was carried out as described (28). The transformants were made homozygous for the P-element inserts. Flies were raised on a standard corn meal diet under a 12-h light/12-h dark cycle and were used for the experiments at an age of 2-3 days posteclosion.
Protein Extraction-For extracting the G protein subunits expressed in the transgenic lines, dissected compound eyes were collected in 1ϫ SDS-buffer (5% SDS, 65 mM Tris-HCl, pH 6.8) and homogenized with a plastic pestle. The proteins were extracted for 20 min at room temperature. Insoluble material was sedimented by centrifugation at 100,000 ϫ g for 10 min at 4°C. To obtain soluble proteins, compound eyes or Drosophila heads were collected in 10 mM sodium phosphate buffer, pH 7.0, 1 mM phenylmethylsulfonyl fluoride and homogenized with a plastic pestle, and residual chitinous material was removed (centrifugation at 3,900 ϫ g for 1 min at 4°C). The homogenate was then incubated for 20 min at 4°C. Eye membranes were sedimented by centrifugation at 386,000 ϫ g 4°C for 30 min to separate the soluble and membrane fraction. The supernatant contained the soluble proteins. To isolate the membrane-bound proteins, the pellet was resuspended in 1ϫ SDS-buffer, incubated for 20 min at room temperature, and the extract was centrifuged for 10 min at 116,000 ϫ g and 4°C. The supernatant of this centrifugation contained the membrane-bound proteins.
SDS-PAGE, Western Blot Analysis, and Protein Quantification-Protein extracts were subjected to SDS-PAGE according to Laemmli (29) on 8 -20% gradient gels (Amersham Biosciences, Midget System). Immunoblotting was performed as described earlier (15). Membranes were incubated with either polyclonal antibodies directed against Bound antibodies were detected with protein A conjugated with alkaline phosphatase, with 125 I-labeled secondary antibodies (1:1000, Amersham Biosciences) or with anti-mouse IgG conjugated with alkaline phosphatase (1:25,000, Sigma) and visualized through a chromogenic reaction with 5-bromo-4-chloro-3-indolyl phosphate/4-nitro blue tetrazolium chloride or by autoradiography, respectively.
To quantify the relative amount of visual G protein subunits present in the soluble and membrane-bound protein fraction, extracts from dissected compound eyes of wild type and P[G␥eC69G] flies were subjected to Western blot analysis with anti-G␣q, anti-G␤e, and anti-G␥e antibodies. For detection of bound antibodies the blot membrane was incubated with 125 I-labeled secondary antibodies. The radioactive sig- A, scheme of the protein prenylation mechanism at the cysteine (Cys) of the carboxyl-terminal CAAX motif of G␥ subunits. The cysteine is modified with an isoprenyl residue by a CAAX prenyltransferase. The amino acid in position X determines the type of isoprenylation. After the post-translational modification, the last three amino acids AAX are removed, and the newly exposed carboxyl group of the cysteine residue is methylated. B, schematic representation of the G␥ constructs used to generate transgenic Drosophila. The exchange of the invariant Cys-69 of the CAAX motif for glycine (C69G) in G␥e was performed by site-directed mutagenesis using appropriate primer pairs. A myc tag was added at the 5Ј-end of the constructs. The rectangle represents the coding region of Drosophila G␥e and human G␥rod. Numbers indicate amino acid positions. nals were quantified using a bioimaging analyzer (Bas 1000 MacBAS, Fujix, program Quant Mode, software MacBAS V1.0). Each of four independently prepared extracts was analyzed twice. For determining the relative amounts of G␣q, G␤e, and G␥e the percentage of radioactivity measured in the soluble and in membrane fraction was determined, and mean values and S.D. were calculated.
Immunocytochemistry-Labeling of myc-tagged G␥ subunits on sections through Drosophila heads was carried out as described previously for native G protein subunits (15). The primary anti-c-myc antibody (New England Biolabs, dilution 1:20) was detected with a Cy5-conjugated anti-mouse IgG (Sigma, dilution 1:100).
Immunoprecipitation-For immunoprecipitation studies of overexpressed Drosophila G␥ subunits and human G␥rod, proteins from Drosophila heads were extracted for 20 min at 4°C with Triton X-100 buffer (1% Triton X-100, 150 mM NaCl, 50 mM Tris-HCl, pH 8.0). Extracted proteins were separated from insoluble material by a centrifugation at 116,000 ϫ g, 4°C, and 10 min. The extracts were added to protein A/G-Sepharose (Pierce) which had been preincubated with monoclonal anti-c-myc antibodies (New England Biolabs) for 1-1.5 h. Immunoprecipitation was performed for 1-2 h at 4°C and was followed by five washes with 500 l of Triton X-100 washing buffer (0.1% Triton X-100, 100 mM NaCl, 50 mM Tris-HCl, pH 8.0). Precipitated proteins were eluted from protein A/G-Sepharose beads with 1ϫ SDS-buffer for 10 min at 80°C and were subjected to SDS-PAGE and Western blot analysis. For detection of immunoprecipitated proteins by Western blot analysis or Coomassie Blue staining of SDS gels, proteins extracted from 90 or 1000 Drosophila heads were used, respectively.
Mass Spectrometry-For molecular mass determination of proteins by matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry, immunoprecipitated G␥ subunits of 1,000 Drosophila heads were eluted from the Sepharose beads with 20 l of 3% trifluoroacetic acid. The eluted proteins were purified and concentrated using ZipTip pipette tips with C 18 reversed phase medium (Millipore) and obtained in 4 l of 0.1% trifluoroacetic acid and 50% acetonitrile. A 0.5-l aliquot of protein solution was mixed with 0.5 l of matrix solution (␣-cyano-4-hydroxycinnamic acid in 0.1% trifluoroacetic acid and acetonitrile, 7:3 (v/v)) on the mass spectrometer sample target and left to crystallize at ambient temperature. Mass analysis was performed in linear mode on a Bruker BIFLEX IV time-of-flight mass spectrometer (Bruker Daltonics). Apomyoglobin (M r ϭ 16952.27), cytochrome c (M r ϭ 12361.96), and bovine insulin (M r ϭ 5734.51) (Sigma) were used as calibration standards.
Electroretinogram Recordings-Light responses were recorded from white-eyed Drosophila raised for 3-5 days in a 12-h dark/12-h light regime. Flies were immobilized with a mixture of Kroenig's wax and thermal conductive paste, and their temperature was maintained at 24°C with a Peltier element. Test stimuli (500-ms duration, 20 nm band width at 520 nm, maximally 2.5 ϫ 10 19 photons m Ϫ2 s Ϫ1 ) were generated by a 150-W xenon arc lamp (Osram, Germany), a monochromator (Oriel Instruments, Stanford, CA), and reflective neutral density filters (Melles Griot, Zeevenar, The Netherlands). To maintain the rhodopsin-metarhodopsin concentration constant, each test stimulus was followed by a bright orange light pulse generated with an additional xenon arc lamp and an OG 580 orange filter (Schott, Germany). Glass electrodes were filled with insect saline (0.67% NaCl, 0.015% KCl, 0.012% CaCl 2 , 0.015% NaHCO 3 , pH 7.2) and inserted just below the cornea. An Ag/AgCl wire in the thorax served as a reference electrode. Signals were detected by an AI 401 SmartProbe Differential Amplifier connected to a CyberAmp 320 signal conditioner (both from Axon Instruments, Inc., Foster City, CA), low pass filtered at 1 kHz, sampled at 7.5 kHz by a Lab-PC-1200 A/D converter (National Instruments), and analyzed by a Whole Cell Analysis program (Strathclyde Electrophysiology software, Whole Cell Analysis Program version 3.0.8, John Dempster, University of Strathclyde, Glasgow, UK).

P-element-mediated Expression of G␥ Subunits of Visual G
Proteins in the Compound Eye of Drosophila-By PCR-mediated site-directed mutagenesis the cysteine (Cys-69) of the carboxyl-terminal CAAX box of Drosophila G␥e was exchanged for a glycine. A myc tag, which allowed us to identify the mutated G␥e on Western blots, was attached at the N terminus of the protein. The mutated G␥e (G␥eC69G, Fig. 1B) was overexpressed in Drosophila wild type flies under the control of the Drosophila Rh1 promoter, which normally directs the expression of the major Drosophila rhodopsin Rh1 in R1-6 cells of the compound eye. We also generated transgenic Drosophila expressing nonmutated, myc-tagged ␥ subunits of the Drosophila visual G protein (G␥e) or of human transducin (hG␥rod, also referred to as human G␥ 1 , Ref. 30) under the control of the Rh1 promoter. We obtained four independent transgenic lines for G␥eC69G, two lines for G␥e, and one line expressing the human G␥ subunit of transducin.
The flies were tested for the expression of the individual G protein subunits present in the transgenic lines by Western blot analysis (Fig. 2). Anti-G␥e recognizes the native G␥e, expressed in the photoreceptor cells of wild type and transgenic flies ( Fig. 2A, transparent arrowheads), as well as recombinantly expressed G␥eC69G and nonmutated G␥e ( Fig. 2A, black arrowheads). By using anti-c-myc antibodies which specifically detect the myc epitope, it was possible to distinguish the recombinantly expressed, myc-tagged G␥eC69G from the native G␥e (Fig. 2B). Because of the myc tag the apparent molecular mass of myc-tagged G␥e subunits is ϳ1300 kDa higher than that of native G␥e, which has a molecular mass of ϳ8400 kDa. Except for one line (P[G␥eC69G]-4) all transgenic lines showed an overexpression of recombinantly expressed myc-tagged G␥e compared with native G␥e. The myc-tagged human G␥rod was specifically detected by anti-bovine G␥rod antibodies which reacted with hG␥rod but not with Drosophila G␥e (Fig. 2C). We failed to detect hG␥rod with anti-myc antibodies on Western blots, suggesting that the human G␥ subunit is expressed at a lower level than the myc-tagged Drosophila G␥ subunit. To obtain information on possible effects of G␥ overexpression on the formation of the trimeric G protein complex, the expression level of the visual G␣ and G␤ subunits was investigated ( Fig. 2A). Despite the presence of overexpressed forms of G␥, all transgenic lines expressed an amount of G␣q and G␤e protein similar to that of wild type flies, showing that neither G␣q nor G␤e was up-or down-regulated in the transgenic lines. Lines P[G␥eC69G]-1, P[G␥e]-1, and P[hG␥rod] were chosen for all following experiments.
To determine whether the localization of the recombinantly expressed G␥ subunits is restricted to photoreceptor cells R1-6 of the compound eye, we carried out immunohistochemical experiments (Fig. 3). Longitudinal sections through heads of transgenic Drosophila were probed with anti-c-myc antibodies. Labeling of the myc-tagged G␥ subunits expressed in P[G␥eC69G], P[G␥e], and P[hG␥rod] was restricted to the retina and the lamina, i.e. the first optic ganglion in which axons of R1-6 photoreceptors terminate. This labeling pattern is almost identical to the pattern described previously for the native ␣, ␤, and ␥ subunits of the Drosophila visual G protein (15), except that the native subunits can additionally be detected in the medulla, i.e. the second optic ganglion to which R7 and R8 photoreceptor cells project. Thus, in line with the expected expression pattern for genes controlled by the Rh1 rhodopsin promoter, the spatial distribution of the myc-tagged G␥ subunits in head sections reveals a highly specific expression of these proteins in R1-6 photoreceptor cells.
The Point Mutation C69G Disrupts in Vivo Farnesylation of G␥e-Although Drosophila G␥e harbors a bona fide farnesylation site, it has not yet been demonstrated that this G␥ subunit is indeed prenylated. To show that G␥e is modified by posttranslational farnesylation, and, equally important, that the mutated G␥eC69G does not carry a farnesyl residue, we analyzed the recombinantly expressed G␥e subunits by MALDI-TOF mass spectrometry (Fig. 4, A and B). The molecular mass determined for the c-myc-tagged mutated G␥e (9709.55) was about 42 higher than the theoretical mass calculated for the unmodified amino acid sequence (M r ϭ 9668.06). This finding suggests a post-translational modification of the protein by acetylation. Indeed the c-myc tag that is attached at the amino terminus of both mutated and nonmutated G␥e has a consen-

FIG. 4. Mass spectrometry of myc-tagged G␥eC69G and G␥e.
Myc-tagged G␥eC69G, G␥e, and hG␥rod were purified from Triton X-100 extracts prepared from 1,000 Drosophila heads by immunoprecipitation with anti-myc antibodies and analyzed by MALDI-TOF mass spectrometry in linear mode in a detection range of 5000 -20,000. Oneeighth of the immunoprecipitated material, i.e. the equivalents of 125 fly heads, was analyzed in each measurement. The insets show the same G␥ peaks at higher resolution. A, the determined mass of G␥eC69G (9709.55) corresponded to the calculated mass of myc-tagged, acetylated G␥eC69G (calculated M r ϭ 9710.10). B, the two peaks observed for G␥e (9617.88 and 9631.97) fitted the theoretical masses of myc-tagged, acetylated and farnesylated G␥e, which lacks the carboxylterminal tripeptide and represents either the carboxyl methylated (calculated M r ϭ 9631.09) or a nonmethylated state (calculated M r ϭ 9617.06) state. C, mass spectrum of hG␥rod. The most prominent peak of the spectrum at 9773.40 corresponded to acetylated, fully modified hG␥rod (calculated M r ϭ 9773.51). The shoulder observed at the left side of this peak may represent a fraction of hG␥rod which has not been carboxyl methylated. sus sequence (methionine followed by a strongly hydrophilic amino acid, Ref. 31) for amino-terminal acetylation. The native G␥e also shows such a consensus sequence at its amino terminus. Further support for post-translational acetylation of the myc-tagged G␥e was obtained by mass spectrometric analysis of peptides generated by digestion of the mutated or nonmutated Drosophila G␥ subunits with endoproteinase Asp-N (data not shown). The molecular mass spectrum obtained for the nonmutated form of G␥e revealed two peaks differing in molecular masses by 14.09 (9617.88 and 9631.97, respectively). The higher molecular mass corresponds to the theoretical mass of an acetylated G␥e subunit that is fully modified by farnesylation, removal of the last three carboxyl-terminal amino acids, and carboxyl methylation. The more prominent peak at 9617.88 most likely represents a farnesylated state of G␥e which lacks carboxyl methylation (14.14). Taken together, the mass spectrometric analysis of the mutated and nonmutated Drosophila G␥e subunits showed that both proteins were modified by amino-terminal acetylation, but only the nonmutated G␥e subunit was farnesylated. This farnesylation was followed by the removal of the three carboxyl-terminal amino acids. A significant fraction of the farnesylated G␥e did not seem to be carboxyl methylated. The myc-tagged human G␥rod expressed in Drosophila photoreceptor cells was also analyzed by mass spectrometry (Fig. 4C). In this mass spectrum a prominent peak at 9773.40 was detected, indicating that the human transducin ␥-subunit, like nonmutated Drosophila G␥e, becomes acetylated and fully modified by farnesylation, removal of the carboxyl-terminal tripeptide, and carboxyl methylation in Drosophila photoreceptor cells (calculated M r of acetylated, fully modified hG␥rod is 9773.51). Additional peaks observed in this mass spectrum may result from contaminations in the immunoprecipitated probe because their molecular masses do not correspond to nonfarnesylated or to partially modified hG␥rod.
Interaction with G␤e and Membrane Attachment of ␥ Subunits-To determine the relevance of the farnesylation site of G␥e for dimerization of G␤e and G␥e, we performed coimmunoprecipitation studies with anti-c-myc antibodies of protein extracts obtained from Drosophila heads (Fig. 5). SDS-PAGE of the precipitated proteins followed by staining with Coomassie Blue revealed two protein bands with molecular masses of ϳ42,000 and 10,000, which were present in immunoprecipitated proteins from transgenic flies expressing myc-tagged G␥e or G␥eC69G (Fig. 5A) but were absent in controls. These two proteins were identified as G␤e and G␥e by Western blot analysis (Fig. 5B). Consequently, G␤e coimmunoprecipitated with both farnesylated G␥e present in P[G␥e] flies as well as with nonfarnesylated G␥e expressed in P[G␥eC69G] flies. G␣q was detected in neither of the immunoprecipitates, indicating that the interaction between the ␤␥ complex and G␣q is not maintained during the immunoprecipitation procedure applied here. Because G␤e is coprecipitated with nonfarnesylated G␥e, it is concluded that the farnesyl modification of the visual G␥e subunit of Drosophila is not required for G␤␥ complex formation. We also determined whether or not the human G␥rod subunit forms a complex with Drosophila G␤e when it is expressed in Drosophila photoreceptors R1-6. Although our mass spectrometric measurements (see Fig. 4C) showed that human G␥rod becomes farnesylated in Drosophila photoreceptor cells, which suggests functional expression of hG␥rod in this system, immunoprecipitation experiments with protein extracts from the heads of P[hG␥rod] transgenic flies did not reveal coprecipitation of G␤e with human G␥rod (Fig. 5). This finding argues against the formation of ␤␥ dimers composed of the human and Drosophila G protein subunits, and it suggests that human G␥rod cannot functionally replace Drosophila G␥e, de-spite the conservation of domains that characterize Drosophila G␥e and human G␥rod as visual G protein subunits (15).
The farnesyl anchor of the vertebrate transducin ␥ subunit has been shown to facilitate its membrane attachment (32). Therefore, it is likely that the absence of a farnesylation site in G␥C69G interferes with membrane attachment of this Drosophila G␥ subunit. Furthermore, the absence of such a membrane anchor in G␥e could also influence the membrane association of G␤e and G␣q. To examine the effect of the C69G mutation on the subcellular location of the G protein subunits, Western blot analyses of soluble and membrane-bound eye proteins of P[G␥eC69G], P[G␥e], and wild type flies were carried out (Fig. 6A). In wild type flies G␥e is distributed equally between the soluble and the membrane fraction (Fig. 6A, third  panel). This finding is in agreement with results showing that farnesylated visual G␤␥ dimers can be partially solubilized without detergent, whereas solubilization of geranylgeranylated G␤␥ dimers, which are more hydrophobic, requires the presence of a detergent (33). For example, human G␥11 carrying a farnesyl residue also showed an equal distribution between the soluble and membrane fraction, whereas a geranylgeranyl modification of G␥11 resulted in a complete membrane-association of the G␥ subunit (34). As in wild type flies, in P[G␥e] flies that overexpress nonmutated G␥e, about half of the total G␥e was detected in the soluble and in the  1, 5, and 9) and immunoprecipitates (lanes 2, 6, 10, and 13) were subjected to SDS-PAGE and stained with Coomassie (A) or to Western blot analysis with the indicated antibodies (B). Black arrowheads in A point at coimmunoprecipitated G␤e. Lanes 3,4,7,8,11, and 12 show control experiments; protein A/G beads without antibody were used to rule out nonspecific binding of proteins to the beads (lanes 3, 7, and 11). No probe was added to the antibody-conjugated protein A/G beads to determine which of the protein bands resulted from the presence of antibodies (lanes 4, 8, and 12). In the immunoprecipitation experiment with hG␥rod (right panels) the immunoprecipitation from P[G␥e] was repeated as a control (lane 13). In A the input lanes contain the equivalent of 3 fly heads, and the immunoprecipitates contain the equivalent of 1,000 fly heads. For the Western blot analysis (B) the equivalents of 3 or of 90 fly heads were loaded on the input lane or on the lanes showing the immunoprecipitates, respectively. membrane fraction, respectively (Fig. 6A, second panel). Mass spectrometric analysis of myc-tagged G␥e immunoprecipitated from either the soluble fraction or the membrane fraction revealed that both fractions contained fully modified farnesylated G␥e as well as farnesylated G␥e lacking carboxyl methylation (data not shown). In contrast to the equal distribution in the soluble and membrane fraction of G␥e, the mutated G␥eC69G was localized predominantly in the soluble fraction of the photoreceptor cells (Fig. 6, A, first panel, and B). Only about 20% of G␥e was retained in the membrane fraction in P[G␥eC69G] transgenic flies (Fig. 6B). To determine whether the altered membrane association of nonfarnesylated G␥eC69G influences the membrane association of G␣q and G␤e in P[G␥eC69G] flies, the relative amount of G␤e and G␣q in the soluble and membrane fraction was quantified (Fig. 6, C and  D). For G␤e, the distribution between the soluble and membrane fraction was similar to that observed for G␥e (Fig. 6, B and C). In P[G␥eC69G] flies 70% of G␤e was localized in the soluble fraction compared with 50% in wild type flies. In con-trast, there was no change in the membrane association of G␣q compared with the wild type situation (Fig. 6D). In both wild type and P[G␥eC69G] flies 70% of G␣q was soluble, and 30% was localized in the membrane fraction. These results show that the membrane association of G␥e and G␤e is disrupted by the C69G mutation. They strongly suggest that the membrane attachment of the G␤␥ complex in the eye of Drosophila is regulated by farnesylation of G␥e.
Overexpression of Mutated G␥eC69G Reduces the Sensitivity of the Visual System-The analysis of a Drosophila G␤e mutant revealed that lowering the level of G␤e to 5% of the wild type level results in a dramatic loss of light sensitivity (14). Therefore, it is to be expected that alterations in the membrane localization of the G␤␥ complex have similar physiological consequences. By recording electroretinograms, we compared the light sensitivity of wild type photoreceptors with those that express the mutated G␥eC69G, nonmutated G␥e, or hG␥rod. Fig. 7 shows intensity-response functions (V-log I curves) which relate the normalized response amplitude to the relative stimulus intensities. A shift of the V-log I curve to the right, i.e. to higher light intensities, corresponds to lower sensitivity of the examined visual system. Such a shift of the V-log I curve is observed in flies expressing the mutated G␥eC69G, whereas the V-log I curves of flies expressing G␥e or hG␥rod are superimposed with the one obtained from wild type flies. A quanti- tative evaluation of the V-log I curves reveals that the light intensity required to obtain half-maximal responses is shifted by 0.4 log unit to a higher light intensity in flies expressing G␥eC69G, which corresponds to a decrease in sensitivity to 40 Ϯ 15% (Fig. 7B) of the wild type sensitivity. Because light sensitivity of fly photoreceptors is directly proportional to the rhodopsin content (35), by analogy this decrease in sensitivity would approximately correspond to a halving of the photoreceptor quantum yield. On the other hand, expression of the nonmutated G␥e and of the human G␥rod had no significant effect on the sensitivity of the visual system.

DISCUSSION
In vitro studies of isoprenylation of G␥ subunits have assigned a functional relevance of this post-translational modification for activation of the G protein by its receptor, for interaction of the ␤␥ complex with the ␣ subunit, possibly via an S-prenyl binding site in the ␣ subunit (36) and for the membrane attachment of the ␤␥ complex (for review, Refs. 16 and 17). To gain insight into the functional role of the putative farnesyl modification of G␥e in the intact visual system of Drosophila photoreceptors, we generated transgenic Drosophila expressing a mutated G␥e without a farnesylation site. The study of these flies revealed that farnesylation of G␥e facilitates membrane association of the G␤␥ complex but is not required for the association of G␥e with G␤e. These findings are in agreement with earlier studies on mammalian visual as well as nonvisual G␥ subunits showing that ␤␥ dimerization occurs independently of G␥ isoprenylation (37,38).
Our mass spectrometric analysis of immunoprecipitated G␥e subunits revealed that the myc-tagged G␥e analyzed here is acetylated at the amino terminus and, most importantly, that the nonmutated G␥ subunit is indeed modified by farnesylation and cleavage of the last three carboxyl-terminal amino acids. As has been described for the G␥ subunit of bovine transducin (39,40), a major fraction of G␥e expressed in Drosophila photoreceptors does not seem to be carboxyl methylated. Methylation of the farnesylated carboxyl-terminal cysteine of transducin G␥ was shown to be a reversible process (39). Together with farnesylation the methylation has been implicated in facilitating the membrane association of G␥ and the interaction of G␣ with G␤␥ (40) as well as the coupling of transducin to metarhodopsin II (41). On the other hand, enzymatic demethylation of transducin G␥ did not affect the ability of G␤␥ to interact functionally with G␣ in detergent and had only a small (2-fold) effect in the presence of rod outer segments, which may be attributed to enhanced membrane association of the carboxyl methylated G␥ (42). Likewise, only minor effects on signal transduction processes were observed after enzymatic demethylation of G␥2 (43). Our finding indicating that a significant fraction of the G␥e subunits isolated from Drosophila photoreceptors are farnesylated but not carboxyl methylated lends support to the hypothesis that carboxyl methylation is a reversible process in vivo with a possible regulatory function.
Besides the altered membrane association of G␤␥, we recorded a decrease in the sensitivity of the visual responses in flies which express the mutated G␥ subunit. There are a number of possible explanations for this result. First, the change in the membrane binding properties of the mutated ␤␥ subunit may result in a decreased number of available ␤␥ dimers in the photoreceptive membrane for forming functional heterotrimeric G proteins. Although native G␥ is still present in the eyes of these flies, the majority of ␤ subunits will interact with the overexpressed nonfarnesylated ␥ subunits. In addition to a functional role of the farnesylation site of G␥e in membrane attachment of the G␤␥ complex it is possible that farnesylation of G␥e is required for the binding of G␤␥ to G␣q. From studies using, for example, recombinantly expressed human ␥ 1 and ␥ 2 subunits, there is clear evidence that only those ␤␥ complexes containing a prenylated G␥ are able to interact with G␣ (24,37,38). Consequently, less heterotrimeric G proteins, which can be activated by the receptor, will be present in the photoreceptive membrane of transgenic flies which overexpress the mutated G␥. Thus, overexpression of the mutated G␥ may partially phenocopy a hypomorphic mutation in the visual G␤ subunit which leads to a severe defect in receptor G protein coupling and hence to a reduced light sensitivity (14). Second, the affinity between activated receptor and G protein in vertebrate photoreceptors is in part determined by the presence of a farnesyl residue at the G␥ subunit (23,24,41,44,45). Mutations in the farnesylation site of Drosophila G␥e could therefore lead to a reduced affinity of Gq for rhodopsin and consequently to a defect in G protein activation, even if heterotrimeric G proteins containing the mutated G␥ were formed. A third possibility is that the mutation of the farnesylation site of G␥ interferes with the interaction of downstream effectors of the visual G protein.
Indeed, there is evidence that the prenyl moiety of vertebrate ␥ subunits has a function in the activation of phospholipase C␤ (46 -48). However, cumulative evidence for G protein function in the phototransduction cascade of Drosophila suggests that G␣q alone and not the ␤␥ subunit activates the downstream effector phospholipase C␤ (see, e.g. Ref. 5). A direct interaction with phospholipase C␤ has only been shown for G␣q but not for G␤␥ (49). This renders it unlikely that the decreased light sensitivity of flies expressing mutated G␥ results from less effective activation of phospholipase C␤ by the G␤␥ subunit. Overexpression of mutated G␥e does not abolish the visual response but only leads to a moderate decrease in light sensitivity. This is explained by the presence of nonmutated, native G␥e in the eyes of the transgenic flies, which allows the formation of fully intact heterotrimeric visual G proteins. By using mutants that express different amounts of G␣q it has been shown that a reduction in the amount of G protein to less than 50% is required to record a reduction in light sensitivity and that flies that express only 1% G␣q still show residual light responses (13). Thus, the overexpression of a G␥e subunit lacking the farnesylation site in Drosophila photoreceptor cells must result in a decrease of functional heterotrimeric Gq protein to less than 50% of the wild type level.