The plasma membrane of the California poppy is known to harbour a PLA2 (phospholipase A2) that is associated with the Gα protein which facilitates its activation by a yeast glycoprotein, thereby eliciting the biosynthesis of phytoalexins. To understand the functional architecture of the protein complex, we titrated purified plasma membranes with the Gα protein (native or recombinant) and found that critical amounts of this subunit keep PLA2 in a low-activity state from which it is released either by elicitor plus GTP or by raising the Gα concentration, which probably causes oligomerization of Gα, as supported by FRET (fluorescence resonance energy transfer)-orientated fluorescence imaging and a semiquantitative split-ubiquitin assay. All effects of Gα were blocked by specific antibodies. A low-Gα mutant showed elevated PLA2 activity and lacked the GTP-dependent stimulation by elicitor, but regained this capability after pre-incubation with Gα. The inhibition by Gα and the GTP-dependent stimulation of PLA2 were diminished by inhibitors of peptidylprolyl cis–trans isomerases. A cyclophilin was identified by sequence in the plasma membrane and in immunoprecipitates with anti-Gα antibodies. We conclude that soluble and target-associated Gα interact at the plasma membrane to build complexes of varying architecture and signal amplification. Protein-folding activity is probably required to convey conformational transitions from Gα to its target PLA2.
- California poppy (Eschscholzia californica)
- phospholipase A2
- plant G-proteins
- plant plasma membrane
- plant signalling
- protein–protein interaction
Heterotrimeric G-proteins are conserved modules of signal transfer in all eukaryotic organisms. Although their peptide sequences and three-dimensional structures display some basic similarities between evolutionarily recent animal and plant cells , the mode of interaction with their targets obviously differs between the kingdoms. In animal cells, signal perception causes one or more out of several task-specific heterotrimers to release their subunits for interaction with target proteins, which is best documented for the large subunit Gα and its various isoforms [2–4]. A plant cell, which typically contains only one Gα, one Gβ and two Gγ subunits  (a third atypical Gγ was found in Arabidopsis and rice [6,7]), obeys different strategies to allow a similar diversity of G-protein functions in the signal transfer [8–11]. A recent example that underlines the variability of plant G-protein signalling is the loss of the Gα-activating protein RGS in the evolution of eudicots and the acquisition of receptor-based activation mechanisms . One important clue arises from the fact that a significant part of Gα or the complete heterotrimer is embedded in large protein complexes at the plasma membrane [11,13,14], and some of these aggregates are known to contain target proteins [11,15]. This promoted the idea that task-specific combinations of G-proteins and their target enzymes might form a framework for the multitude of G-protein-triggered signal pathways in plants .
The inventory and architecture of such complexes are far from being understood, as is the mode of intrinsic interaction between Gα, other subunits and target enzymes. As an example, conformational changes triggered by GTP or GTPγS (guanosine 5′-[γ-thio]triphosphate; GTP[S]) are reported to cause the total dissociation of Gα-containing protein complexes in rice . In contrast, in Arabidopsis and Eschscholzia, no or very limited liberation of Gα was found upon activation [11,14,16], implying that GTP-triggered conformational transitions are conveyed to the target without dissociation of the protein complex.
Actually, an increasing number of potential Gα-binding plant proteins has been deduced from the interaction of the recombinant proteins expressed in yeast or plant protoplasts, or from co-immunoprecipitation experiments [17,18]. Confronting and complementing such studies with biochemical analyses of G-protein-containing complexes might allow new insights into the functional specificity of the signal process. Among the problems to be addressed is the stoichiometry of interaction, e.g. the number of G-protein subunits per target molecule required for regulatory impacts, and the structural and functional plasticity of such complexes, e.g. the ability of membrane-associated Gα to exchange or interact with the soluble protein components faced in its cytosolic environment .
In the last few years, the plasma membrane of Eschscholzia californica, the California poppy, has been established as an in vitro model system to study the control exerted by Gα over the target enzyme PLA2 (phospholipase A2; EC 188.8.131.52). The highest and most reproducible levels of PLA2 activity and its stimulation by GTP were obtained after solubilization of the plasma membrane with cholate. Among the solubilized proteins, co-immunoprecipitation and non-denaturing electrophoresis detected large detergent-resistant complexes that contain PLA2 together with Gα. The in vitro activity of the enzyme increased upon contact with yeast elicitor plus GTP, thus reflecting its Gα-dependent control . In intact cells, the same elicitor activates PLA2 at the plasma membrane [20,21], and a product of this enzyme, lysophosphatidylcholine, acts as a second messenger for the induction of alkaloid biosynthesis [22,23]. Thus Gα-controlled PLA2 initiates a signal chain that induces the overproduction of phytoalexins . This antimicrobial response is triggered very selectively, i.e. it is independent of the ubiquitous jasmonate-dependent hypersensitive cell death. The latter can be triggered in addition by exposing our strains to high elicitor concentrations [21,24]. Although the hypersensitive response might involve other PLA enzymes , there are no indications of their G-protein-dependent control [11,25]. The Gα-dependent elicitor-responsive PLA2 of Eschscholzia has now been cloned, sequenced and deposited in GenBank® (accession number JQ886492).
In the present study, we use recombinant Gα of Eschscholzia to mimic the impact of soluble Gα at the target enzyme PLA2 of the isolated plasma membrane.
Cell suspensions of Eschscholzia californica Cham. were originally derived from hypocotyl discs via callus cultures . For experiments, cultures were maintained in a 9–10 day growth cycle in Linsmaier–Skoog medium supported with the hormones 2,4-D (2,4-dichlorophenoxyacetic acid) and α-naphthalene acetic acid (1 μM each). Cultivation was carried out on a gyrotary shaker (100 rev./min) at 24°C under continuous light (~7 μmol·m−2·s−1). Cells were used for experiments after 6 or 7 days of growth as described previously .
Cell fractionation and preparation of plasma membrane
Eschscholzia or Arabidopsis cell suspensions (80 g) were filtered, frozen in liquid nitrogen and fractionated as described previously . The microsomal pellet was obtained after centrifugation at 100000 g for 1 h (4°C; Beckman LE-80K, TI-70 angle rotor, 37000 rev./min) and used to purify the plasma membrane by liquid two-phase partioning . Soluble and membrane fractions were also analysed by SDS/PAGE and subsequent Western blotting with IgG fractions purified from the polyclonal rabbit anti-Gα antiserum at a dilution of 1:500 in TBS [Tris-buffered saline; 25 mM Tris/HCl (pH 7.5), 150 mM NaCl and 3 mM KCl].
The plasma membrane fraction was finally solubilized by mild shaking (2 h, 8°C) in TENC buffer (20 mM Tris/HCl, pH 7.5, 1 mM EDTA, 50 mM NaCl and 0.5% sodium cholate). This detergent proved to be superior to a number of other compounds tested (including lubrol and n-dodecyl-β-D-maltoside) with respect to the yield of solubilized PLA2 activity and its stimulation by GTPγS. The supernatant obtained by centrifugation (100000 g, 40 min, 4°C; Beckman LE-80K, TI-70 angle rotor, 37000 rev./min) was adjusted to a protein content of 500 μg/ml and used in all experiments.
Assay of PLA2 activity
PLA2 activity was quantified by monitoring the fluorescence emission from the artificial substrate BPC (bis-BODIPY®-FL-C11 phosphatidylcholine) (Molecular Probes) as described in [11,29]. A typical assay contained per well of a 96-well microtitre plate (Microfluor II; Greiner) 170 μl of substrate solution (0.4 μM BPC) followed by 30 μl of effectors (see below). After 10 min of equilibration (mild shaking at room temperature, 22°C), 20 μl of solubilized plasma membrane preparation (containing 10 μg of protein) were added and the fluorescence was recorded at λex 485 nm/λem 528 nm over 10 or 60 s (dependent on the rate of fluorescence development) in a FLX800 microplate fluorescence reader (BioTek). To eliminate non-enzymatic generation of fluorescence, a control experiment was performed with the enzyme preparation replaced by TENC buffer. Initial rates of hydrolysis were converted into enzyme activities by using calibration assays with bee venom PLA2 and a series of substrate concentrations.
The final concentrations of effectors in the 220 μl reaction mixture were as follows: Gα, 45–1350 nM as indicated; cyclosporin A, 1 μM; GTP, 5 μM; yeast elicitor, 1 μg/ml (see below); and antibodies (IgG or scFv), 68 nM. These concentrations were optimized for each compound or antibody in a test series and yielded either saturating or optimum effects. In short-time incubations as used in the present study, GTP and GTPγS displayed similar effects on PLA2. Yeast elicitor is a glycoprotein fraction prepared from baker’s yeast according to  and further purified by ultrafiltration (30 kDa molecular mass cut-off), FPLC (anion exchange and size exclusion) and SDS/PAGE. The active fraction comprises glycoproteins of 30–42 kDa that contain approximately 40% mannose . Dosage refers to the dry weight of the crude preparation.
Preparation of recombinant Gα and Gγ proteins
The ORF (open reading frame) of EcGPA1 (GenBank® accession number HQ830348) encoding the Gα protein was amplified from a cDNA library of Eschscholzia  and expressed in the pET-23(+) vector. Similiarly, the Gγ1 coding gene from Arabidopsis thaliana (AT3G63420.1) was cloned into the same vector. Expression was performed using standard procedures. Briefly, competent cells of Escherichia coli BL21(DE3) were transformed with these plasmids by the heat-shock procedure (42°C) and grown on ampicillin selection agar. Colonies were transformed to liquid LB (Luria–Bertani) medium, grown for 24 h at 35°C and 1 mM IPTG (isopropyl β-D-thiogalactopyranoside) was added to stimulate protein production. After 3 h the bacterial pellet was harvested by centrifugation (20 min, 4°C; Sorvall GSA rotor, 4000 rev./min) and lysed using lysozyme and ultrasonic treatment. In the 30000 g supernatant, the C-terminally His6-tagged protein was trapped on 50% Ni-NTA (Ni2+-nitrilotriacetate)–agarose (Ni Sepharose™ High Performance, GE Healthcare), purified by FPLC (ÄKTA, GE Healthcare) and finally eluted with buffer containing 250 mM imidazole.
The recombinant Gα protein contained <1% impurities, most probably from the bacteria used for cloning, as determined by SDS/PAGE. To confirm that the bacterial impurities did not mimic or change the observed effects of Gα, test experiments were performed in the absence of Gα but with the contaminating proteins, obtained by the same expression and purification procedure using an empty pET-23(+) vector. Addition of this test extract together with Gα did not significantly change its effects on PLA2.
Preparation of native Gα by immunoabsorption
Gα was isolated from the soluble Eschscholzia proteins (100000 g supernatant, obtained according to ). An immunomatrix was prepared by binding and crosslinking the IgGs from the polyclonal anti-Gα antiserum (see above) to Protein A–Sepharose 4 Fast Flow (GE Healthcare). After incubation with the Gα-containing protein fraction, the bound Gα was eluted at alkaline pH. The whole procedure followed essentially a protocol published by I.J. Delgado (http://www.ivaan.com/protocols/126.html).
The Gα was further purified by SDS/PAGE, the 42 kDa band was excised from the gel and the protein was renatured in the presence of 5 mM CaCl2, 5 M cysteine and 50 mM ammonium acetate, pH 7.0, according to . Gel extraction was facilitated by a freeze–thaw cycle, followed by dialysis and ultrafiltration, as detailed in .
To raise antisera against recombinant Gα, two rabbits were each immunized with 1 mg of Gα and Freund's complete adjuvant and boostered with 500 μg of Gα and Freund's incomplete adjuvant after 4 weeks, 5 weeks and 6 weeks respectively. The anti-Gα sera were affinity-purified via Gα–Sepharose columns made from CNBr-activated Sepharose by loading Gα following the manufacturer's protocol (GE Healthcare). The resulting IgG fraction was used in PLA2 assays.
Isolation of specifc scFv against Gα was performed as described previously . Briefly, the phage display libraries Tomlinson A and B  were screened against Gα produced in pET-23a . After four rounds of selection one monoclonal scFv (anti-Gα scFv2) was isolated and characterized by sequencing and indirect ELISA as described previously . Briefly, antigens were adsorbed to Maxisorb plates at neutral pH overnight, the wells blocked and specific scFv, anti-c-Myc antibody and anti-mouse alkaline phosphatase conjugate were added sequentially after washing between steps. Alkaline phosphatase activity was measured using p-nitrophenylphosphate. A selectivity test of anti-Gα scFv2 is exemplified in Supplementary Figure S1 at http://www.biochemj.org/bj/450/bj4500497add.htm by showing an indirect ELISA experiment with Gα, Gγ1 and Gγ2 as antigens. Anti-Gα scFv2 binds to Gα with a dissociation constant of 1.3×10−7 M as determined by competitive ELISA (results not shown).
Anti-cyclophilin antibodies, raised against human cyclophilin A, were purchased from NovusBio, NB300-553. The recombinant human cyclophilin A used to raise this IgG fraction shows 71% sequence identity with the cyclophilin isolated from Eschscholzia (see Figure 10B).
All animal work was performed according to national guidelines described in the German Animal Welfare Act AWA (Deutsches Tierschutzgesetz). Immunization studies were carried out by an authorized employee under licence according to Section 8b from the district veterinary office Aschersleben-Stassfurt.
Co-immunoprecipitation of plasma membrane proteins
IgG fractions from antisera raised in rabbits against Gα (see above), or against human cyclophilin A were immobilized on a Protein A–Sepharose immunomatrix (Fast Flow 4; GE Healthcare) and incubated with the solubilized plasma membrane as described in .
The mating-based split-ubiquitin assay
The assay was performed in principle as described previously . Briefly, the gene encoding Gα and the gene of a potential interactor protein were fused to the C-terminal or the N-terminal half respectively of the ubiquitin gene and expressed in yeast strains of different mating type. During mating, interactions between the membrane-bound fusion proteins are expected to create a ubiquitin-like structure by interacting the N-terminal part of ubiquitin (Nub) with the C-terminal part of ubiquitin (Cub) that harbours the PLV transcription factor (Cub-PLV). This structure activates ubiquitin-specific proteases, thus causing the release of PLV, which can be quantified by the expression of the yeast reporter genes lacZ, HIS3 and ADE2.
In our experiments, the ORF of Gα from E. californica was fused by in vivo recombination cloning to the C-terminal part of ubiquitin (Gα-Cub) followed by PLV. In the same way, this ORF was also fused to the N-terminal part of the ubiquitin-13 mutant (NubG).
The haploid yeast strains THY.AP4 (harbouring the Cub-fused Gα proteins) and THY.AP5 (harbouring the the Nub-fused test protein) were mated and allowed to generate diploid yeast cells on medium lacking histidine and adenine. Growth of diploids in SC medium and in addition the expression of LacZ and β-galactosidase [assayed with X-Gal (5-bromo-4-chloroindol-3-yl β-D-galactopyranoside)], was taken as an indicator of interaction between the Nub- and Cub-fused proteins . In order to detect false-positive effects of test proteins, which may not arise from an interaction with the bait, e.g. direct influences at the yeast reporter genes and growth (if any), negative control experiments were performed by mating the THY.AP5 strain (with the Nub-fused test protein) to a THY.AP4 strain that did not contain the Gα gene in the Cub-vector. No colony growth was detectable in such experiments.
The expression of Gα encoded in the Cub-PLV vector can be controlled via a methionine-sensitive promoter and thus the Gα available for binding can be reduced to definable levels. In our hands, methionine concentrations between 0.15 mM and 5 mM caused the Gα content in the diploid clones to decrease from approximately 90% to 15% . This feature was used to favour the formation and growth of clones with higher affinity interactors and thus to rank the interactors by their binding strength. To measure growth rates at different methionine concentrations, diploid clones obtained after mating on solid medium were picked and each subcultured in liquid medium with no methionine. After another 2 days of growth, all cell suspensions were adjusted to an attenuance of 0.1 and 3 μl aliquots were spotted on to solid medium with the same nutrient composition but different methionine concentration. After 7 days of incubation at 30°C digital images of the colonies were recorded. The increase in colony size was compared between cultures at the specified methionine concentration and without methionine, and normalized to the increase displayed by the same clone on a methionine-free agar plate.
Purification, MS-based sequencing and PCR-based cloning of cyclophilin
The immunoprecipitate obtained from the solubilized plasma membrane with the anti-Gα antibody (see above) was separated by SDS/PAGE. The cyclophilin band (approximately 19 kDa) was identified by comparison with a Western blot detected with the anti-cyclophilin antibody. The band was excised from the gel, digested with trypsin and the fragments identified by MS/MS (tandem MS)-based sequencing, as detailed with Gα in . On the basis of two peptide sequences, a forward primer (5′-AAAGTTTTCTTCGATATGACTGTTGGTGG-3′) and an antisense reverse primer (5′-ATGTTTACCATCAGGCCATGAAGTTTT-3′) were synthesized based on Eschscholzia codon usage (http://www.kazusa.or.jp), which amplified a 390 bp fragment from a cDNA template of Eschscholzia californica. PCR conditions were: 5 min at 94°C followed by 35 cycles with 20 s at 94°C, 1 min at 50°C and 1 min at 72°C, finalized by one cycle at 72°C for 10 min. The full-length ORF was completed from the above fragment by genome walking (Genome Walker™ Universal Kit; Clontech) according to the manufacturer's protocol. Genomic DNA as required for genome walking was isolated from the Eschscholzia cell culture, digested with EcoRV and adaptor-ligated. The gene-specific primers were as follows: forward, 5′-GTCAAGGTGGAGATTTCACAGCCGGAAACG3′; and antisense reverse, 5′-AGATCCATTTGTGCCTGGACCGCGGTTAGC-3′. Two overlapping fragments were obtained, from which the full-length cDNA was obtained by PCR (see above) with primers from non-coding regions, 5′ (upstream) 5′-CCCCGATATAACCCCTATCCAGAAA-3′ and 3′ (downstream) 5′-GAGGGAGAGATAGGGTGAATTGGGGATGT-3′. The ORF of the cyclophilin gene was cloned into the pDRIVE vector (Qiagen) and expressed in E. coli.
Construction of vectors for plant transformation
The 3′-end of the ORF of Gα of Eschscholzia (GenBank® accession number HQ830348) was fused to the 5′-end of the ORFs of eGFP [enhanced GFP (green fluorescent protein)], CFP (cyan fluorescent protein) or YFP (yellow fluorescent protein). A spacer of three alanine triplets (5′-GCTGCCGCG-3′) was placed between the two genes. ORFs encoding eGFP, CFP or YFP were used as references.
The plasmid DNAs were amplified by PCR with gene-specific primers, which introduced flanking BsaI restriction sites and were finally cloned into the binary vector pICH56022 (Icon Genetics) using similar restriction sites. The desired vectors were produced in a one-step ligation/digestion reaction [35,36]. They were used for the heat-shock transformation of competent DH10B E. coli cells. Transformed colonies were selected on LB-Amp50 plates containing X-Gal and IPTG as selection markers for intact β-galactosidase. Successfully transformed bacteria were selected by colony PCR using the same Gα- and fluorophor-specific primers (Table 1) and the presence of the amplified vectors was confirmed by DNA sequencing. For each transformation vector, three individual DNA clones were isolated by alkaline-lysis-based minipreparation (Gene Jet Kit; Thermo Fisher Scientific).
Biolistic gene transfer
Transient gene expression in Eschscholzia cell suspensions was achieved by bombardment with DNA-coated gold particles. Plasmid DNA of each vector was precipitated on to gold particles by adding 60 μl of water containing 5 μg of DNA to 2.5 mg of gold particles (1.0 μm diameter, Bio-Rad Laboratories), prepared according to the manufacturer's protocol. After addition of 0.1 M spermidine (20 μl) and 2.5 M CaCl2 (50 μl) and intensive vortex-mixing for 3 min, the suspension was kept on ice for 3 min and precipitated by centrifugation for 1 min at 11000 g. The gold pellet was resuspended in 200 μl of 96% (v/v) ethanol, centrifuged again (1 min at 11000 g) and finally resuspended in 85 μl of 96% (v/v) ethanol and used for biolistic gene transfer.
Eschscholzia cell suspensions (6 days old, approximately 50 mg of wet weight) were filtered on to sterile paper discs and placed on to Petri dishes containing 1.5% agar with M21 medium. The biolistic bombardment with the cell-suspension-containing filter pieces was done with a PDS1000/He Biolistic Particle Delivery System (Bio-Rad Laboratories) following the manufacturer's instructions. After bombardment, the cells on the filter were transferred to a new Petri dish with the same medium containing 50 μM paramomycin, 5 μM benomyl and 5 μM nystatin. After 2 h, the cells were transfered to 1.5 ml of M21 liquid medium with supplements as mentioned above. These samples were incubated on a horizontal shaker at 120 rev./min at 25°C.
The biolistic bombardment resulted in a mixture of paromomycin-resistant (i.e. stably transformed) and -sensitive cells. The former initiated regenerative growth 2–4 days after bombardment and often generated multicellular aggregates. The latter included cells that clearly suffered under the influence of the antibiotic treatment as indicated by the expression of fluorescent CFP and/or YFP but irregular cell shapes. The microscopic analysis was done with newly formed cells that were clearly distinguishable from the detrimental cells.
Verification of transformation by RT (reverse transcription)–PCR
At 2 days after biolistic treatment, 100 mg [fwt (fresh weight)] samples of the transformed cultures were harvested by vacuum filtration and stored at −80°C. RNA was isolated by the NucleoSpin RNA Plant isolation kit (Macherey-Nagel). cDNA synthesis was performed with the Qiagen RevertAid™ H Minus First Strand cDNA Synthesis kit using an oligo(dT)18 primer, according to the manufacturer's instructions. An aliquot (2 μl) of the resulting cDNA mixture was used as a template for PCR, which was performed in an Eppendorf Mastercycler gradient PCR machine: 2 min at 94°C; 1 min at 55°C, 1 min at 72°C, 30 s at 94°C (35 repeats); and 10 min at 72°C. The primers used are listed in Table 1. The resulting PCR products were separated (Supplementary Figure S2 at http://www.biochemj.org/bj/450/bj4500497add.htm) and identified by DNA sequencing.
Confocal laser-scanning microscopy
Confocal microscopy was done with the LEICA TCS SP laser-scanning unit mounted on a DCM RE fluorescence microscope equipped with a 63-fold apochromate objective. Fluorescence was excited at 488 nm (Argon-Ion laser) and detected in the range between 496 and 530 nm. The detection pinhole was set at 1.0. In parallel with the laser-excited fluorescence emission, a (non-confocal) transmission image of each specimen was scanned. All images are averaged from six subsequently scanned frames of 1025 pixels×1025 pixels or 512 pixels×512 pixels. Intensities were visualized by a standard lookup table consisting of different shades of green (dark at low and bright at high intensity).
The red fluorescence of benzophenanthridine alkaloids, which is partially emitted in the transformed cells, was detected in a parallel precautionary scan at 570–670 nm. Cells with detectable fluorescence in this wavelength area were discarded from analysis.
Fluorescence microscopy and image-processing for FRET (fluorescence resonance energy transfer) analysis
Transformed cell suspensions were examined with a Nikon Optiphot fluorescence microscope equipped with a Sony 3-chip CCD (charge-coupled-device) camera. Images were obtained successively by using three fluorescence filters: For YFP FRET, λex=380–425 nm, λem=520 nm BP (bandpass filter); for CFP, λex=380–425 nm, λem=50 nm BP; and for the YFP reference, λex=482–500 nm, λem=520–560 nm, followed by a transmitted light image. The digitalized images were converted into greyscale and ratioed with the Optimas 6.2 software: (i) the acceptor image (obtained via the YFP-FRET filter) was divided by the donor image (obtained via the CFP filter) and amplified by 100; (ii) the resulting quotient image was again divided by the donor image and amplified by 100; and (iii) the new image was divided by the YFP reference image (obtained via the YFP reference filter) and amplified by 30. The final ratio image was processed with Adobe Photoshop CS5, firstly by color coding with a spectral lookup table and secondly aligned with the transmitted light image. The procedure is exemplified in Figure 5. Apart from the linear amplifications, the final ratio image contains the original image informations converted as: donor/acceptor2 per YFP (reference).
When fluorescence intensities were to be compared between different channels (e.g. in Figure 4), the emission in the FRET-YFP channel was corrected for the ‘spillover’ of CFP-derived emission. For this purpose, a correction factor was established by dividing the fluorescence intensities in the FRET-YFP channel (λex=380–425 nm, λem=520 nm BP) of cells transformed with Gα–CFP alone with cells that expressed both fusion proteins. A total of 12 areas of cytoplasmic regions (31 pixels×31 pixels) with maximum intensity were averaged in each cell strain. When measured under identical conditions, the greyscale images in Gα–CFP cells displayed a mean intensity of 36% compared with Gα–CFP+Gα–YFP cells. This value was used as an average for the CFP-derived ‘spillover’.
Intracellular distribution of Gα
The isolated plasma membrane preparation of Eschscholzia used in our previous and current studies contains approximately 12 μg (286 pmol) of Gα per mg of protein. This amount represents only approximately 0.5% of the total cellular Gα (cell homogenate excluding mitochondria), of which >50% is not membrane-bound, i.e. detectable in the 100000 g supernatant. This distribution contrasts with analogous data obtained from Arabidopsis cell cultures: as seen in Figure 1, the Gα of Arabidopsis is located almost completely in the microsomal pellet and in the isolated plasma membrane, whereas only a small percentage is detectable in the soluble fraction. This finding is in line with earlier cell fractionation studies in meristematic cells of Arabidopsis and cauliflower, where Weiss et al.  found most, if not all, Gα associated with the plasma membrane and ER (endoplasmic reticulum) membranes. Binding of Gα to the plasma membrane and the ER has now been confirmed by the visualization of Gα–GFP hybrids in protoplasts (e.g. ) or by electron microscopy studies in immunogold-labelled tissues . In order to substantiate the apparent species difference, a localization study of Gα was performed in Eschscholzia cells by confocal microscopy. As seen in Figure 2, a Gα–eGFP hybrid protein spreads over all detectable cytoplasmic areas and does not report a substantial accumulation at the plasma membrane.
This result is in line with the data of the cell fractionation study described above. It appears therefore that substantial amounts of soluble Gα are present together with the membrane-bound species, at least in some plants. For instance, in growing wheat tissue cultures the larger part of Gα was recovered in a 120000 g supernatant and is thus considered a soluble protein , similar to our experience with Eschscholzia. More comparative experiments are justified to reveal whether the localization of Gα changed during species evolution (actually, the monocot wheat and the early eudicot Eschscholzia most probably share a pattern that differs from the late eudicots Arabidopsis and Brassica) or whether the reported differences represent expression profiles of different cell types and growth stages.
The present study is based on the assumption that a small fraction of Gα tightly bound to the plasma membrane faces an excess of soluble or weakly bound Gα at the cytoplasmic surface. We address the question whether and how the interplay of both fractions can influence the impact of this subunit at its target protein(s).
Interaction with Gα is inhibitory to PLA2
In the isolated and solubilized plasma membrane of Eschscholzia the activity of PLA2 was assayed as the rate of hydrolysis of a fluorigenic substrate, BPC, according to an optimized protocol [11,20,23,29]. Recombinant Gα, encoded by Eschscholzia cDNA and produced in E. coli, inhibits PLA2 activity if added in concentrations between 20 and 100 nM (Figure 3, columns 2–4). Typically, 45 nM Gα caused a reduction in activity of approximately 66%. This inhibition disappeared if the concentration of recombinant Gα was raised, and above a saturating level (approximately 900 nM), the enzyme activity was even stimulated (140–145%, Figure 3, columns 6–8). Neither the inhibition nor the activation caused by recombinant Gα was due to its His6-tag or potential bacterial contaminants, as in test experiments native Gα, isolated from the soluble proteins of Eschscholzia cell cultures (see the Experimental section), displayed the same effect as the recombinant protein (see columns 2/3 and 6/7 in Figure 3). Furthermore, the effects of native Gα on PLA2 activity did not change if it was added together with extracts prepared in a similar manner from untransformed bacteria or from those transformed with the empty vector (results not shown).
The concentration-dependence in Figure 3 suggests that the mode of interaction between PLA2 and external Gα changes above a critical concentration of this protein. Gα molecules might then either access additional activating binding site(s) at the enzyme or lose their affinity by self-interaction, thus competing with the target. As the latter idea appears more plausible, we tested the ability of Gα to dimerize or oligomerize by two alternative procedures: (i) by fluorescence microscopy to search for FRET-like interactions between different Gα–GFP hybrids in the Eschscholzia cell; or (ii) by the mating-based split-ubiquitin assay with the Gα of Eschscholzia expressed in yeast cells.
FRET caused by interaction of fluorescently labelled Gα molecules in Eschscholzia cells
In this approach, cultured cells were biolistically transformed with two DNA vectors, each encoding a fusion protein of Gα with either CFP or YFP. Control cell lines were established similiarly by transferring the CFP or the YFP gene alone, or a mixture of both. After 2–4 days of regeneration, the transformed cultures expressed the desired mRNAs (Supplementary Figure S1) and most cells emitted fluorescence typical of YFP and CFP.
In cells expressing both Gα–CFP and Gα–YFP, the excitation of CFP with a wavelength of ~400 nm caused a substantial increase in fluorescence in the YFP channel, i.e. at >520 nm. This effect persisted after correcting for the average ‘spillover’ of CFP (donor) fluorescence into the YFP (acceptor) channel, and was not found in cells that expressed Gα–YFP alone (Figure 4). This finding was taken as a first hint for energy transfer between Gα–CFP and Gα–YFP, but could not easily be quantified in individual cells as the proportions between CFP- and YFP-fusion proteins appeared to differ significantly from cell to cell, which would require extensive individual correction measurements.
We thus established an alternative procedure to visualize potential interactions between the Gα-moieties by comparing fluorescence properties between cells that expressed both hybrid proteins (Gα–CFP plus Gα–YFP) and others that contained the pair of unsubstituted CFP plus YFP. Three images from each cell were obtained and processed by ratioing acceptor emission (FYFP-FRET) to donor emission (FCFP). The resulting map (FYFP-FRET/FCFP) was divided by an image of the specific YFP emission in order to correct for different concentrations of the acceptor in individual cells (Figure 5 and see the Experimental section). The ratio maps obtained by this procedure are comparable with those produced by the FRET algorithm of Gordon et al.  and a commercial software (Zeiss AxioVision SE64), but in our hands yielded more intracellular details (Figure 5).
The results collected in Figure 6 suggest that the simultaneous presence of both fusion proteins (Gα–CFP plus Gα–YFP) causes higher acceptor/donor ratios than the unsubstituted fluorescent proteins (CFP plus YFP) in a substantial majority of cells. The ratio maps differ mainly in the peripheral cytoplasmic areas, whereas central regions including the nucleus display lower ratios and more similarities between Gα-containing and Gα-free fluorescent cell lines.
Taken together, the fluorescence imaging data support a FRET-like energy transfer between the Gα-coupled fusion proteins that is probably caused by interaction of Gα-moieties.
Gα-interaction studies with the split-ubiquitin system
The mbSUS (mating-based split-ubiquitin system) was established to detect plant membrane protein interactions . This and similar two-hybrid assays proved successful in the characterization of protein complexes with a variety of membrane-integral and membrane-associated proteins [41–44]. For the present study, we expressed Gα of Eschscholzia as a part of hybrid proteins with ubiquitin moieties in yeast cells and studied its interaction with itself and with other G subunits. The results presented in Figure 7(A) indicate that two Gα molecules can form dimers with high affinity. As the mbSUS allows the ranking of interactions by their binding strength (via a methionine-controlled promoter ), it could be shown that the self-interaction between two Gα molecules is in the same order as the well-known interaction between Gα and Gβ [1,14], whereas the binding in the Gα–Gγ couple was much weaker. Figure 7(B) shows the influence of increased methionine concentrations that cause a decrease in available Gα. The assay does not discriminate between dimers and higher oligomers, and therefore it cannot be excluded that higher oligomers might be formed as well.
Thus the split-ubiquitin studies support the findings of a tight interaction between two Gα molecules obtained from the above-mentioned FRET experiments. This is especially interesting as it suggests that the dimerization of Gα molecules occurs not only in the cells of their origin but also in the heterologous yeast cell system. Currently, the self-interaction of Gα appears the most likely reason why its inhibitory power is lost at increasing concentrations. The stimulation of PLA2 by saturating amounts of Gα might reflect the involvement in the dimerization process of the inhibitory Gα present in the original plasma membrane preparation.
The functional specificity of the interaction and stimulation between Gα and PLA2
The observed impacts on PLA2 activity involved immunologically targetable sites at the Gα protein, as shown by the effects of antibodies raised against the Gα protein of Eschscholzia. As seen in Figure 3 (columns 10, 11, 13 and 14), both the inhibition by low and the stimulation by high Gα concentrations are relieved by polyclonal antisera or monoclonal single-chain (scFv) antibodies. Both types of antibodies did not affect the activity of PLA2 if added in the absence of external Gα. Further data support their selective binding to Gα: (i) the polyclonal antiserum detected only the expected 43 kDa band among the plasma membrane proteins separated by SDS/PAGE; and (ii) the anti-Gα scFv antibody preparation was pre-selected by a four-step process of repeated panning and screening for Gα-binding molecules in a relevant phage-display library (see the Experimental section) and did not detect the His6-tag of the recombinant protein (Supplementary Figure S1). Thus it appears that binding of antibody to the Gα protein impedes both its interaction with PLA2 and its self-interaction.
It is known that the association of Gα and PLA2 allows activation of the enzyme by yeast glycoprotein elicitor in the presence of GTP/GTPγS, but not GDP/GDPβS (guanosine 5′-[β-thio]diphosphate) . In the present study, this stimulation attained the same level irrespective of whether inhibitory concentrations of external Gα were present or not (Figure 8, columns 1–3 and 8–10). Although GTP and GTPγS act similarly, GDβS had no stimulatory effects under any of the conditions tested (Figure 8, columns 1, 5 and 6). It is thus likely that elicitor plus GTP act at the Gα molecules bound (or are accessible) to PLA2, thus releasing the enzyme from an inhibited state. Two further experiments support this conclusion.
First, the increase caused by elicitor plus GTP is significantly smaller if Gα is present at high stimulating concentrations (Figure 8, compare columns 1–3 with 11–13, the high conversion rates proved not to be limited by substrate supply). This is consistent with the tendency of Gα to undergo dimerization, as this would shift the binding equilibrium away from the PLA2-associated form of the Gα protein, which confers the stimulation by GTP plus elicitor.
Second, in the antisense-Gα strain TG11, whose plasma membrane contains only approximately 15% of the wild-type Gα content , the specific activity of PLA2 at the outset was significantly higher than in the wild-type (120%, Figure 9). In the same plasma membrane, titration with Gα causes the same effects as in the wild-type (Figure 9, columns 2 and 4–7). Importantly, in the mutant plasma membrane, PLA2 was not stimulated by yeast elicitor plus GTP, but regained such stimulation after addition of 45 nM Gα, which alone caused strong inhibition (Figure 9, columns 2, 3 and 8–10). These findings indicate that the PLA2-inhibiting fraction of Gα is identical with the GTP-stimulated fraction and thus the addition of Gα restores the regulatory features of the TG11 mutant to those of the wild-type.
A cyclophilin is involved in the elicitor-triggered activation of PLA2
The plasma membrane preparation contains a PPIase (peptidylprolyl cis–trans isomerase) of the cyclophilin type (E.C. 184.108.40.206). Its 18 kDa polypeptide was found as a common constituent of immunoprecipitates obtained either with an anti-cyclophilin IgG or with the anti-Gα antibody raised against the recombinant Gα. It was partially sequenced by MS/MS. On the basis of three tryptic peptides, it was possible to clone the complete ORF (Figure 10A). The deduced complete peptide sequence indicates a protein of high similarity with other plant cyclophilins, notably an ABH-type cyclophilin of Digitalis lanata  (Figure 10B). A significant degree of sequence identity (71%) even with human cyclophilins (Figure 10B, last line) prompted us to test a commercially available antibody raised against the human homologue. This antibody caused a significant inhibition of PLA2 activity (Figure 11, column 9), which further supported an association of cyclophilin and Gα in the plasma membrane. Pharmacological data obtained with cylosporin A, a specific inhibitor of PPIases of the cyclophilin type , suggest that the protein-folding activity of cyclophilin is required for the interaction of Gα and PLA2: first, cyclosporin A depressed the PLA2 activity in the non-treated plasma membrane by 40–45% (Figure 11, column 10) but exerted no significant inhibition in the low-Gα mutant (Figure 11, columns 7 and 8). Secondly, the stimulation exerted by elicitor plus GTP was completely blocked (Figure 8, column 7). Thirdly, the inhibitory effects of low Gα concentrations were less severe in the presence of cyclosporin A or of the anti-cyclophilin antibody (Figure 11, columns 1, 3 and 4). Finally, the stimulation by high Gα concentrations was reduced to an extent comparable with the non-treated plasma membrane (Figure 11, columns 5 and 6).
These results are the first indications that a cyclophilin is required for Gα to exert its inhibitory effect on PLA2. Although low and high Gα concentrations shift the enzyme activity into different directions, both effects were attenuated by cyclosporin A.
The relevance of our experimental data for the understanding of the PLA2–Gα complex rests on the assumption that the added Gα protein is functionally intact and equivalent to the native protein associated with the plasma membrane of Eschscholzia cells. This is supported by the following facts: first, the recombinant Gα and the native Gα are products of the same gene, but produced by entirely different procedures (a bacterial His6-tagged product, purified by Ni2+-affinity chromatography, or a soluble protein of the plant cell homogenate, purified by immune absorption) and are thus unlikely to contain similar contaminants. Nonetheless, either protein caused the same inhibitory or stimulatory effects on PLA2 if added to the plasma membrane. Secondly, either protein reconstitutes the low-Gα mutant TG11 with respect to the GTP-dependent activation of PLA2. Thirdly, potential bacterial contaminants of the recombinant Gα preparation had no influence on PLA2 activity and the impact of Gα, as tested with extracts from non-Gα-expressing bacteria.
The most plausible interpretation of our results leads to the following five conclusions. They are followed by the main results supporting them. Figure 12 summarizes our interpretation of the flexible architecture and PLA2 activities of protein complexes in the form of a hypothetical model.
Soluble Gα interacts with the plasma membrane complex of PLA2 and Gα at the cytoplasmic side, thereby changing its structure and functionality
Gα is a soluble protein of the cytoplasm in Eschscholzia as shown by cell fractionation and confocal imaging of a Gα–GFP hybrid protein (Figures 1 and 2). Addition of increasing amounts of Gα, recombinant or native, changed the PLA2 activity of the solubilized plasma membrane in an unique manner (Figure 3). The Gα-deprived mutant TG11 reacted to the addition of Gα much like the wild-type, indicating that the recombinant protein compensates for the lack of native Gα (Figure 9).
The impact of Gα on PLA2 is inhibitory in nature
Low Gα concentrations inhibited the enzyme activity under a variety of conditions (Figures 3 and 8). In the absence of extra Gα, the low-Gα mutant TG11 displayed a higher specific activity of PLA2 than the wild-type (Figure 9). The loss of Gα (85%) is stronger than the gain of enzyme activity (20%), which is consistent with earlier findings that not all Gα molecules of the plasma membrane are part of the complex  and could also reflect a different architecture of the mutant plasma membrane complex.
The activation of PLA2 by elicitor in the presence of GTP  probably reflects the relief of the enzyme from a pre-existing inhibited state
GTP (or GTPγS) plus elicitor evoked the same increase in enzyme activity in the absence of Gα as in the presence of inhibitory concentrations (Figure 8). Replacement of GTP by GDP or GDPβS prevented any stimulation. The Gα-deprived mutant TG11 lacked the stimulation by elicitor plus GTP but regained this property after addition of Gα (Figure 9). This also supports the ability of added Gα to functionally replace the originally bound Gα.
With increasing Gα concentrations, this protein undergoes self-interaction, thereby competing with its binding to PLA2
Both the FRET-based analysis and the yeast split-ubiquitin assay indicated a high affinity of Gα for di- or oligo-merization (Figures 4, 6 and 7). Although the actual outputs of both assays are only qualitative or semiquantitative in nature, it is remarkable that products of the same Gα gene display the same tendency in the Eschscholzia cell and in yeast. The di- or oligo-merization at increasing concentrations most likely explains why the effect of added Gα changes from inhibition to stimulation of PLA2 (Figure 3). The stimulation of PLA2 by elicitor plus GTP was higher at 45 nM Gα than at 900 nM Gα, which is consistent with the loss of PLA2-bound Gα at high concentrations. The latter conditions allow the highest specific activity of PLA2 to be measured in the solubilized plasma membrane (approximately 190%, Figure 8). Under such conditions, most of the inhibitory Gα molecules are expected to be inaccessible to the enzyme due to di- or oligo-merization, with the remainder maximally stimulated by elicitor plus GTP.
The interaction between PLA2 and Gα involves the catalytic activity of a cyclophilin
A plant-type PPIase co-precipitated with Gα and the cognate gene could be cloned (Figure 10). Also, an anti-cyclophilin antibody caused a significant inhibition of PLA2 activity (Figure 11). The basal level of PLA2 activity, its increase caused by elicitor plus GTP, the inhibitory effect of low Gα and the stimulation caused by high Gα concentrations were all attenuated by cyclosporin A (Figures 8 and 11).
The effect of cyclosporin A is related to the endogeneous Gα content: the low-Gα mutant is less strongly inhibited than the untreated wild-type (Figure 11, TG11). Taken together, our data show that PLA2 is regulated by Gα via controlled inhibition. In the absence of elicitor, the Gα contained in the plasma membrane complex keeps the enzyme in a low activity state. This inhibition can be intensified by increasing the native Gα content (approximately 20 nM under assay conditions) up to 3-fold, i.e. through addition of approximately 45 nM of the native or recombinant protein.
The tight binding to PLA2 of low (native) Gα concentrations is well documented, e.g. by various immunoprecipitations and native protein electrophoresis that revealed detergent-resistant complexes of approximately 130 and 170 kDa which did not dissociate upon addition of GTP plus elicitor . The elicitor signal causes a release from inhibition, most probably via the known conformational activation of Gα that requires GTP. Although not the focus of the present study, we assume that this elicitor effect is mediated via a GPCR (G-protein-coupled receptor), as a seven-transmembrane protein (subfamily MLO1) was found among the interactors of Gα in the split-ubiquitin assay (Figure 7B and C. Massalski and W. Roos, unpublished work). Classically, plant G-proteins are thought to be coupled to seven-transmembrane receptors  that share limited homology with animal GPCRs  and homologues of this type are present in an actual interactome of Gα of Arabidopsis .
A novel mechanism that tends to shift Gα away from the PLA2-inhibiting state is the self-interaction of this protein at high local concentrations. The competition of oligomerization with the binding to PLA2 probably gives rise to complexes of different Gα/PLA2 stoichiometry. Such complexes would also display different influences of GTP at the enzyme activity level. As a consequence, a broad range of PLA2 activities can be expected, each depending on the actual Gα content and GTP level. Not only would their basal PLA2 activity be different, but also the final amplification of the elicitor signal would differ significantly (Figure 12).
From the biological point of view, the results of the present study support the idea that soluble Gα in the neighbouring cytoplasm can modulate the activity of the target PLA2 at the plasma membrane via exchange and equilibration with the bound G-protein. Although, in the cell culture used in the present study, the cellular content of Gα did not undergo dramatic changes (M. Heinze and W. Roos, unpublished work), differences in the Gα levels associated with growth and cellular development [49,50] are worthwhile for investigation in future studies. In either case, local imbalances in soluble Gα probably influence the architecture and effectivity of the PLA2-containing complex(es).
Cyclophilins, which have not yet been reported as components of the Gα-containing complexes, may now be considered as intrinsic catalysts or mediators of the interaction between Gα and its neighbouring targets. The involvement of these chaperone-like proteins, as suggested from the results of the present study, could motivate new experiments that aim at understanding the conformation-based signal transfer within a Gα-target complex.
It also remains an aim of further experiments to determine whether or not the control of PLA2 at the plasma membrane involves the Gβγ complex or its components. In vitro studies compatible with that shown in the present study are hampered by the lack of supply of Gβ, as the recombinant overproduction of this plant plasma membrane-associated protein has not yet proved successful (U. Conrad, unpublished work). The Gγ subunit of Arabidopsis is available via recombinant expression in bacteria, and preliminary data show a significant influence of this protein on the interaction between Gα and PLA2 (Supplementary Figure S3 at http://www.biochemj.org/bj/450/bj4500497add.htm). This might serve as a first indication that Gγ, whose interaction with Gα is weak but still detectable (Figure 7B), can modulate the interaction between Gα and its target. However, to obtain biologically relevant data, experiments that include the simultaneous presence of the Gβ and Gγ subunit of Eschscholzia are indispensable. Thus, although the regulatory effect of the Gα subunit clearly dominates the PLA2-dependent signal path in Eschscholzia (the present study and ), modulating influences of the other subunits cannot be excluded. In Arabidopsis, the use of mutants either lacking or with reduced amounts of a distinct G-subunit, has yielded evidence that distinct signal paths only require Gα, whereas others require the Gβγ complex or the heterotrimer [8,10,51,52]. To date, the interaction of Gα and PLA2 as documented by the present study is not reflected in genome-wide interactome studies in Arabidopsis: one of the published sequences in  shows detectable, but very low, similarity to the PLA2 of Eschscholzia. Such discrepancies might be resolved in the near future by completing the coverage of this and related genomes and performing a more thorough comparison with other species. Although in Arabidopsis Gα is almost completely located at the plasma membrane and ER (see above), the co-existence of monomeric and complex-bound Gα has been reported .
Thus the present data may open new research avenues to understand the diversity of available and new Gα interactors. The interplay of bound and soluble Gα can create membrane-associated complexes of varying architecture and efficiency. They constitute a hitherto underestimated level of tuning in G-protein-controlled signalling which contributes much to its flexibility and adaptability.
Michael Heinze did the assays of PLA2 activity, plasma membrane preparations, and established and characterized all of the transgenic Eschscholzia strains. Madeleine Herre, supervised by Michael Heinze and Werner Roos, contributed with PLA2 assays in the G-protein-treated plasma membrane. Werner Roos and Michael Heinze did the FRET analysis of Gα-interaction. Carolin Massalski did all the mbSUS approach experiments and supplied plasma membrane samples of Arabidopsis. Udo Conrad and Isabella Herrmann produced and characterized recombinant Gα and Gγ, and anti-Gα scFv antibodies and antisera. Werner Roos co-ordinated the study and wrote the paper.
This work was supported by the German Research Council (DFG) [grant number RO 889/12 (to W.R.)], the Graduate College “Plant protein complexes” (to C.M.) and the Martin-Luckner Foundation, Halle (to M.H.).
We thank Dr Angelika Schierhorn (Max Planck – Forschungsstelle für Enzymologie der Proteinfaltung, Halle, Germany) for MS/MS protein sequencing, Gabriele Danders and Beate Schöne for engaged technical assistance, and Professor Gary Sawers for expert linguistic help. The yeast strains THY.AP4 and THY.AP5 as well as the pSUgate vectors pMetYCgate, pNubWtXgate, pNXgate32-3HA, pXNgate21-3HA and pNX32-DEST were provided by Wolf Frommer (Department of Plant Biology, Carnegie Institution for Science, Washington DC, U.S.A.) and Klaus Harter (Plant Science Center, University of Tübingen, Tübingen, Germany). The cloning vectors vectors pICH 56022 and pICH49100 (with eGFP insert) were supplied by Icon Genetics; we thank Dr Sylvestre Marillonnet and Dr Carola Engler for expert help with the cloning of Gα–eGFP vectors. The microscopic imaging software AxioVision SE64 (release 220.127.116.11) was made available by Carl Zeiss Microscopy.
The nucleotide sequence data of the Eschscholzia cyclophilin reported will appear in GenBank®, EMBL, DDBJ and GSDB Nucleotide Sequence Databases under the accession number JQ886493.
Abbreviations: BP, bandpass filter; BPC, bis-BODIPY®-FL-C11 phosphatidylcholine; CFP, cyan fluorescent protein; Cub, C-terminal part of ubiquitin; eGFP, enhanced green fluorescent protein; ER, endoplasmic reticulum; FRET, fluorescence resonance energy transfer; fwt, fresh weight; GDPβS, guanosine 5′-[β-thio]diphosphate; GFP, green fluorescent protein; GPCR, G-protein-coupled receptor; GTPγS, guanosine 5′-[γ-thio]triphosphate; IPTG, isopropyl β-D-thiogalactopyranoside; LB, Luria–Bertani; mbSUS, mating-based split-ubiquitin system; MS/MS, tandem MS; Nub, N-terminal part of ubiquitin; ORF, open reading frame; PLA, phospholipase A; PPIase, peptidylprolyl cis–trans isomerase; RT, reverse transcription; YFP, yellow fluorescent protein; X-Gal, 5-bromo-4-chloroindol-3-yl β-D-galactopyranoside
- © The Authors Journal compilation © 2013 Biochemical Society