A P-loop Mutation in Gα Subunits Prevents Transition to the Active State: Implications for G-protein Signaling in Fungal Pathogenesis

Heterotrimeric G-proteins are molecular switches integral to a panoply of different physiological responses that many organisms make to environmental cues. The switch from inactive to active Gαβγ heterotrimer relies on nucleotide cycling by the Gα subunit: exchange of GTP for GDP activates Gα, whereas its intrinsic enzymatic activity catalyzes GTP hydrolysis to GDP and inorganic phosphate, thereby reverting Gα to its inactive state. In several genetic studies of filamentous fungi, such as the rice blast fungus Magnaporthe oryzae, a G42R mutation in the phosphate-binding loop of Gα subunits is assumed to be GTPase-deficient and thus constitutively active. Here, we demonstrate that Gα(G42R) mutants are not GTPase deficient, but rather incapable of achieving the activated conformation. Two crystal structure models suggest that Arg-42 prevents a typical switch region conformational change upon Gαi1(G42R) binding to GDP·AlF4 − or GTP, but rotameric flexibility at this locus allows for unperturbed GTP hydrolysis. Gα(G42R) mutants do not engage the active state-selective peptide KB-1753 nor RGS domains with high affinity, but instead favor interaction with Gβγ and GoLoco motifs in any nucleotide state. The corresponding Gαq(G48R) mutant is not constitutively active in cells and responds poorly to aluminum tetrafluoride activation. Comparative analyses of M. oryzae strains harboring either G42R or GTPase-deficient Q/L mutations in the Gα subunits MagA or MagB illustrate functional differences in environmental cue processing and intracellular signaling outcomes between these two Gα mutants, thus demonstrating the in vivo functional divergence of G42R and activating G-protein mutants.


Introduction
G protein-coupled receptors (GPCRs) convert extracellular signals to intracellular responses, primarily by stimulating guanine nucleotide exchange on heterotrimeric G-protein Ga subunits [1]. Upon receptor-stimulated exchange of GTP for GDP, Ga subunits undergo a conformational change, dominated by three mobile switch regions, resulting in separation of Ga from the obligate Gbc heterodimer [2]. Switches one and two directly contact the bound guanine nucleotide and include residues critical for catalyzing GTP hydrolysis, while switch three contacts switch two in the activated conformation [3]. The nucleotide-dependent conformational shift of Ga subunits can be monitored biochemically by differential resistance to proteolysis by trypsin or altered tryptophan fluorescence spectra [4,5]. The switch mechanism of activation is highly conserved among the mammalian Ga subunit family members, as well as in those found in fungi [6,7]. The activated Ga and free Gbc subunits propagate signals through numerous effectors, including adenylyl cyclases, phospholipases, ion channels, and phosphodiesterases [8]. Mammals express multiple Ga subunits which can be classified into subfamilies according to function. For example, members the Ga i/o subfamily have inhibitory effects on adenylyl cylase and stimulatory effects on cGMP-phosphodiesterase, while Ga q subfamily members stimulate phospholipase C isoforms, promoting hydrolysis of phosphatidylinositol bisphosphate to produce diacylglycerol and inositol triphosphate [9,10]. Ga signaling is terminated by intrinsic hydrolysis of bound GTP to GDP, a reaction accelerated by 'regulators of G-protein signaling' (RGS proteins), and reversion of the Ga switch conformation to the inactive, GDP-bound state [9,11]. Ga?GDP can then re-assemble a heterotrimer with Gbc or, in the case of the Ga i/o subfamily, engage GoLoco motif proteins that are also selective for the inactive Ga state [12]. In addition to naturally occurring conformationally selective binding partners, phage display peptides have also been engineered to discriminate between Ga?GDP and Ga?GTP. For example, the peptides KB-752 and KB-1753 selectively interact with the inactive GDP-bound and active GTP-bound states of Ga i1 , respectively [13].
Among the most stringently conserved motifs of Ga subunits is the phosphate-binding loop (P-loop) ( Figure S1). Very little variation in the P-loop sequence is seen across Ga subunits in distantly related species, including plants, fungi, and metazoans [27]. In fact, the P-loop is also conserved as a key phosphoryl group-interacting motif in ATP-binding kinases and members of the Ras GTPase superfamily [28].
A P-loop mutation to human Ras isoforms, Gly-12 to valine, is frequently found in human cancers. Ras G12V mutants are GTPase deficient, and thus constitutively active, leading to aberrant signaling and oncogenesis [29]. In fact, mutation of H-Ras Gly-12 to any residue other than proline results in constitutive activity [30]. Mutation of the corresponding P-loop residue in Ga i1 , Gly-42 to valine, also drastically reduces its GTPase activity [31]. Structural studies of Ga i1 (G42V) suggest that the introduced valine side chain sterically prevents appropriate positioning of Gln-204, a residue that coordinates a nucleophilic water molecule during GTP hydrolysis [31]. This glutamine is highly conserved and critical for GTPase activity; its mutation to leucine (''Q/L'') in Ras GTPases or Ga subunits also leads to constitutive activity [11,29].
Genetic studies of heterotrimeric G-protein function in fungal species have used GTPase deficient Ga Q204L mutants (referred to as Q/L mutants). Additionally, a Ga subunit P-loop mutation, G42R, has been utilized in a similar context. Given that Ga i1 (G42V) is GTPase-deficient and mutation of the corresponding glycine in Ras to any amino acid other than proline results in constitutive activation, it has been assumed that G42R mutants would be dominant and constitutively active [32]. Although the biochemical mechanism of the Ga G42R mutant has not previously been characterized, we and others have utilized it to probe the G-protein mediated biology of many fungal species (Table S1) [19,26,[32][33][34][35][36][37][38][39][40][41].
The phosphate-binding P-loop and switch mechanism of activation are both stringently conserved among Ga subunits from mammals to fungi [6,7] ( Figure S1). For example, human RGS2 recognizes the highly similar GTP hydrolysis transition state conformations of both human Ga q and a yeast Ga subunit (GPA1), such that RGS2 expression complements the deletion of an RGS protein gene in S. cerevisiae [42,43]. Furthermore, chimeras of GPA1 and human Ga subunits can function in the yeast pheromone signaling pathway [44]. The residue position corresponding to Gly-42 in Ga i1 is within potential contact distance of residues in the switch regions of the structurally conserved Ga subfamily members [3,10,[45][46][47]. The switch region sequences are highly conserved across mammalian Ga subfamilies, as well as in other species, including M. oryzae, A. nidulans, and S. cerevisiae ( Figure S1). Given the sequence and structural conservation of these regions in Ga subunits, as well as the demonstrated consistent behavior of other point mutations in these regions across multiple Ga subunits (e.g. the GTPasedeficient Ga i1 (Q204L) and the RGS-insensitive Ga i1 (G184S) [48]), the behavior of the G42R mutation is expected to be consistent in MagA, MagB, and the mammalian Ga subunits. Since we were unable to obtain properly folded recombinant MagA or MagB proteins and no direct cellular assays of MagA or MagB activity are currently available, we utilized three mammalian Ga subunits to investigate the behavior of G42R mutants.
Here, we determine through structural, biochemical, genetic, and cellular approaches that Ga subunit G42R mutants are neither GTPase deficient nor constitutively active. Rather, the mutant arginine side chain prevents transition to the activated state upon Ga binding to GTP. Direct phenotypic analyses of M. oryzae strains harboring either Ga G42R mutants or the GTPasedeficient Ga Q204L suggests that a re-evaluation of previous fungal genetic data generated with the G42R mutation is required.

Results
The G42R mutation minimally perturbs the inactive conformation of Ga To understand how the G42R P-loop substitution affects Ga subunit structure and function, we obtained a 3.0 Å resolution crystal structure model of Ga i1 (G42R) bound to GDP using the

Author Summary
Heterotrimeric G-proteins function as molecular switches to convey cellular signals. When a G-protein coupled receptor encounters its ligand at the cellular membrane, it catalyzes guanine nucleotide exchange on the Ga subunit, resulting in a shift from an inactive to an active conformation. G-protein signaling pathways are conserved from mammals to plants and fungi, including the rice blast fungus Magnaporthe oryzae. A mutation in the Ga subunit (G42R), previously thought to eliminate its GTPase activity, leading to constitutive activation, has been utilized to investigate roles of heterotrimeric G-protein signaling pathways in multiple species of filamentous fungi. Here, we demonstrate through structural, biochemical, and cellular approaches that G42R mutants are neither GTPase deficient nor constitutively active, but rather are unable to transition to the activated conformation. A direct comparison of M. oryzae fungal strains harboring either G42R or truly constitutively activating mutations in two Ga subunits, MagA and MagB, revealed markedly different phenotypes. Our results suggest that activation of MagB is critical for pathogenic development of M. oryzae in response to hydrophobic surfaces, such as plant leaves. Furthermore, the lack of constitutive activity by Ga(G42R) mutants prompts a re-evaluation of its use in previous genetic experiments in multiple fungal species.
inactive state-selective phage display peptide KB-752 as a crystallography tool [49]. The asymmetric unit contained three Ga i1 (G42R) subunits bound to GDP and Mg 2+ ; two of three monomers were bound to the KB-752 peptide, while the third (chain C) lacked electron density for the peptide and instead displayed switch region disorder characteristic of free, GDP-bound Ga subunits [31]. For data collection and refinement statistics, see Table S1. A comparison of our model with that of wild type Ga i1 ?GDP/KB-752 (PDB id 1Y3A) revealed minor perturbations to the inactive state upon introduction of Arg-42 ( Figure 1A). The side chain of Arg-42 projects away from the nucleotide-binding pocket, making no direct contacts with other Ga i1 (G42R) residues. Switch 1 and the adjacent b2 strand adopt slightly different conformations in the mutant Ga i1 (Ca atoms r.m.s.d. 1.3 Å ), likely The overall structure of Ga i1 (cyan) with switch regions in dark blue, bound to KB-752 (red) (current study; PDB 3QE0), is overlaid on the wild type Ga i1 ?GDP/KB-752 complex (wheat/red transparency) (PDB 1Y3A). GDP is represented by green sticks and magnesium by an orange sphere. (B) The Arg-42 side chain extends from the P-loop, making no polar contacts with other Ga i1 (G42R) residues, but preventing the wild type (transparent) switch conformation. Ga i1 (G42R) residues Arg-178 and Lys-180 are displaced relative to wild type due to steric and electrostatic repulsion by Arg-42. The G42R b2 strand and switch 2 also adopt slightly different conformations. For crystallographic data collection and refinement statistics, see Table S2. doi:10.1371/journal.ppat.1002553.g001 because the basic residues Arg-178 and Lys-180 are electrostatically and sterically repelled from their wild type orientations by the positively charged Arg-42 side chain ( Figure 1B). Arg-178 is known to stabilize the leaving phosphate group during GTP hydrolysis [11]; its perturbation in the Ga i1 (G42R) structure model is consistent with the previously assumed GTPase deficiency of G42R mutants.

Ga(G42R) is not GTPase deficient
Substitution of the corresponding Gly-12 in H-Ras for any amino acid other than proline yields GTPase deficiency and constitutive activity [30]. Thus it was previously reasoned that Ga(G42R) mutants were also incapable of GTP hydrolysis [26]. Binding of GTP by purified Ga subunits can be assessed with the non-hydrolyzable GTP analog, the radionucleotide GTPc[ 35 S]. Similarly, GTPase activity can be quantified by tracking release of radioactive inorganic phosphate from [c-32 P]GTP-loaded Ga subunits during a single round of hydrolysis [15]. GTPc[ 35 S] radionucleotide binding and [c-32 P]GTP single turnover hydrolysis assays indicated that the kinetics of GTP binding and hydrolysis by the equivalent G42R mutant Ga oA (G42R), in the most frequent splice variant of the mammalian adenylyl cyclase inhibitory Ga o1 , are not significantly different from wild type Ga oA (Figure 2A,B). Since the nucleotide binding and hydrolysis rate of this G42R mutant was unexpectedly not perturbed, we further examined the effect of the G42R mutation on Ga interactions with known protein binding partners.

The G42R mutation disrupts Ga interactions with RGS domains
RGS proteins accelerate the intrinsic GTPase activity of Ga subunits by stabilizing the transition state for GTP hydrolysis, a conformation mimicked by Ga binding to GDP, AlF 4 2 , and Mg 2+ [11]. Surface plasmon resonance (SPR) was utilized to detect optical changes upon injection of wild type or G42R mutant Ga oA over a surface coated with immobilized GST-RGS12 in the presence of either GDP, GTP, the non-hydrolyzable GTP analog GTPcS, or the hydrolysis transition state-mimetic GDP?AlF 4 2 [50]. The RGS domain of RGS12 bound selectively to wild type Ga oA in its GDP?AlF 4 2 -bound state (K D = 1.2760.06 mM), as measured by surface plasmon resonance (SPR) [50]. However, Ga oA (G42R) did not engage the RGS domain in any nucleotide state at concentrations up to 25 mM ( Figure 2C,D), suggesting that G42R mutants do not adopt a typical GTP hydrolysis transition state in the presence of AlF 4 2 and Mg 2+ (AMF), or alternatively that Arg-42 directly interferes with RGS domain binding. A superimposition of Ga i1 (G42R)/KB-752 and the Ga i1 /RGS4 complex (PDB 1AGR; not shown) indicated that the mutant arginine side chain likely directly perturbs the RGS-binding surface. To further characterize nucleotide state-dependent interactions of Ga(G42R), we measured binding affinity toward three additional state-selective Ga-binding partners: Gbc subunits, a GoLoco motif, and a phage display peptide, KB-1753 [13].
Ga(G42R) preferentially engages inactive conformationselective binding partners in any nucleotide state Ga subunits in their GDP-bound, inactive conformations form heterotrimers with Gbc subunits [6], and the interaction is disrupted by AlF 4 2 or GTP binding to the Ga subunit. As expected, wild type Ga i1 ?GDP bound Gb 1 c 1 as measured by SPR, but activation of the Ga subunit with GDP?AlF 4 2 prevented association with Gbc ( Figure 3A). However, Ga i1 (G42R) engaged Gb 1 c 1 in both nucleotide states. Interaction of Ga subunits with fluorophore-labeled peptides was assessed by detecting differences in fluorescence polarization between low molecular weight free peptide and the higher molecular weight Ga/peptide complex [51]. Similar to Gbc, the GoLoco motif of RGS14 was highly selective for binding the GDP-bound, inactive state of wild type Ga i1 (K D = 9.061.1 nM) over the activated GDP?AlF 4 2 -bound form, as determined by fluorescence polarization ( Figure 3B). Ga i1 (G42R) displayed a much reduced selectivity for RGS14 GoLoco motif binding between the GDP and AlF 4 2 nucleotide states, being only 3-fold selective for the GDP form, whereas wild type Ga i1 is .1000-fold selective. Finally, we tested two G42R mutant nucleotide states for interaction with the active conformation-selective phage display peptide KB-1753 using fluorescence polarization [13]. As expected, KB-1753 selectively interacted with wild type Ga i1 ?GDP?AlF 4 2 (K D = 470640 nM) relative to GDP-bound Ga i1 ( Figure 3C). In contrast, Ga i1 (G42R) displayed only weak affinity for KB-1753 in either nucleotide state, as measured by fluorescence polarization. Together these data indicate that Ga(G42R) mutants preferentially engage inactive conformation-selective binding partners regardless of the bound nucleotide. To assess the conformational shift of Ga(G42R) mutants upon activation with AlF 4 2 or a non-hydrolyzable GTP analog, we utilized intrinsic tryptophan fluorescence and limited trypsin proteolysis.

Ga(G42R) cannot assume the transition state-mimetic or activated conformations
Upon binding GDP?AlF 4 2 or GTP analogs, Ga subunits undergo conformational changes dominated by the three switch regions [52]. A tryptophan residue (Trp-211 in Ga i1 ) within switch 2 is shifted from a solvent-exposed to a buried orientation, resulting in a reduced efficiency of tryptophan fluorescence quenching that can be detected upon excitation of the Ga protein with light at 284 nm wavelength [5]. Wild-type Ga i1 displayed a large increase in tryptophan fluorescence upon exposure to AlF 4 2 , indicative of a shift to the activated conformation. In contrast, the shift in tryptophan fluorescence of Ga i1 (G42R) at the same concentration was blunted relative to wild type and occurred with faster kinetics (k obs = 0.1960.01 s 21 [95% C.I.], compared to k obs = 0.0560.01 s 21 for wild type Ga i1 ; Figure 4A).
The active and inactive states of Ga subunits are also differentially sensitive to proteolysis by trypsin; the more flexible loop conformations of Ga?GDP promote cleavage [4]. While the flexible N-terminus of wild type Ga i1 was cleaved in all three nucleotide states, the resulting ,38 kDa fragment was resistant to limited trypsin proteolysis in the GDP?AlF 4 2 or GTP-bound conformations relative to the inactive, GDP-bound form ( Figure 4B). Ga i1 (G42R), however, was readily proteolyzed in any nucleotide state. Addition of AlF 4 2 had no detectable effect on Ga i1 (G42R) resistance to trypsin proteolysis, while GTPcS provided only mild protection of the ,38 kDa species compared to that of wild type Ga i1 . These data further support the hypothesis that the switch regions of Ga(G42R) mutants do not assume appropriate transition state-mimetic or activated state conformations in the presence of AlF 4 2 and GTPcS, respectively.
The Arg-42 side chain prevents transition of the switch regions to an active conformation We next sought a structural explanation for the disrupted conformational switch of Ga(G42R) mutants. As previously mentioned, the Arg-42 side chain conformation, as modeled in the free GDP-bound Ga i1 (G42R), would not allow glutamine-204 to assume its critical position for orienting the nucleophilic water required for GTP hydrolysis ( Figure 1). However, unlike the G42V mutant of Ga subunits, the G42R mutant retains normal GTP hydrolysis kinetics ( Figure 2). Positioning of Gln-204 for hydrolysis may be possible if the Arg-42 side chain adopts an alternate rotamer. We also crystallized Ga i1 (G42R)?GDP in complex with the GoLoco motif from RGS14 and derived an independent structural model at 2.8 Å resolution (Table S2). In one of the two monomers of the asymmetric unit, Arg-42 adopts such an alternative rotamer that would allow Gln-204 to orient the nucleophilic water for hydrolysis ( Figures 4C and S2).
Since we are presently unable to crystallize Ga i1 (G42R) in either its GDP?AlF 4 2 or GTP analog-bound states, we superimposed our structural model of Ga i1 (G42R)?GDP (excluding the RGS14 GoLoco peptide) with the previously described, wild type Ga i1 ?GTPcS (PDB id 1GIA) ( Figure 4C,D). In the activated, GTPcS-bound state of wild type Ga i1 , switches 1 and 2 converge on the nucleotide c-phosphoryl group, while Glu-236 of switch 3 forms a new polar contact with the backbone of switch 2 [3]. The result is a convergence of the three switch regions near the P-loop to form a stable interface recognized by effector molecules. Superposition of Ga i1 (G42R)?GDP suggests that the bulky Arg-42 side chain would not be easily accommodated by the active switch conformations observed in wild type Ga i1 ?GTPcS ( Figure 4C,D). The arginine as modeled would sterically prevent the positioning of switch 3 residues Leu-234 and Glu-236 as seen in the wild type, activated state. Thus, the Arg side chain likely sterically prevents a normal activated conformation of the switch regions.
These data suggest that Arg-42 hinders attainment of the activated switch conformations seen in wild-type Ga subunits, but rotameric flexibility of the mutant side chain allows critical switch residues to effect GTP hydrolysis. Although the G42R mutants of Ga subunits have been shown to favor the inactive conformation despite retaining the ability to bind and hydrolyze GTP, we also sought to investigate their behavior in a cellular context. The G42R mutant is not constitutively active and displays a blunted response to stimulation by AlF 4 2 To investigate the effects of G42R mutants in a signaling pathway context, we introduced the corresponding P-loop mutation into the phospholipase C stimulating mammalian Ga subunit, Ga q (G48R). Wild-type Ga q ?GTP activates phospholipase Cb (PLCb), which in turn hydrolyzes phosphatidylinositol-4,5bisphosphate (PIP 2 ) to yield diacyl glycerol (DAG) and inositol triphosphate (IP 3 ) [10]. Phospholipase C activity can be quantified by measuring accumulation of radioactive IP 3 in cells pre-treated with tritiated inositol. Overexpression of wild type Ga q in COS-7 cells had little effect on inositol phosphate accumulation, while the GTPase-deficient and constitutively active Ga q (Q209L) markedly stimulated PLCb activity in a dose-dependent fashion ( Figure 5A,B). Ga q (G48R), however, had no significant effect on PLCb activity when overexpressed, confirming its lack of constitutive activity. Activation of PLCb by endogenous and overexpressed Ga q can be stimulated by exposure to AlF 4 2 , since Ga q ?GDP?AlF 4 2 has high affinity for PLCb [53]. As expected, endogenous Ga q was activated by AlF 4 2 , and the effect was enhanced by overexpression of wild type Ga q . However, overexpressed Ga q (G48R) did not respond to AlF 4 2 stimulation to the same extent as wild type Ga q , reflecting its inability to assume a fully-activated conformation ( Figure 5C,D). The Ga(G42R) mutant utilized in genetic studies of fungal species, such as Aspergillus nidulans and the rice blast fungus Magnaporthe oryzae, was assumed to be GTPase deficient and thus constitutively active [26,32], and has been used extensively to understand the biology of fungal G-protein signaling [19,26,[32][33][34][35][36][37][38][39][40][41]. Since the biochemical and structural characterization of such G42R mutants (Figures 1-4 above) indicate intact GTPase activity and, instead of constitutive activity, an inability to assume the activated conformation, we sought to clarify the behavior of G42R mutations in the Ga subunits of M. oryzae.
G42R and Q204L mutants of M. oryzae Ga subunits exhibit different effects on appressorium formation We directly compared strains of M. oryzae harboring mutations in the Ga subunits MagA or MagB. Since both Ga subunits are known to regulate appressorium formation in response to inductive, hydrophobic surfaces [24], we assessed appressorium formation by GTPase-deficient Q/L and non-activatable G42R mutant strains on both hydrophobic and hydrophilic surfaces. The magA(G45R) mutant formed slightly fewer appressoria on hydrophobic, inductive surfaces than wild-type M. oryzae, but maintained the differential response to surface hydrophobicity ( Figure 6A,B). In contrast, approximately 35% of magA(Q208L) conidia formed highly pigmented appressoria, albeit aberrant, after 16 hours, regardless of surface hydrophobicity. The magB(G42R) mutant strain resembled magA(Q208L), with ,30% appressorium formation independent of surface hydrophobicity ( Figure 6C,D). The magB(Q204L) strain, however, formed very few appressoria on either surface.
To further characterize differences between magA and magB G42R and Q/L mutant strains of M. oryzae, we compared colony 2 , the switch regions of Ga i1 undergo a conformational change, burying the switch 2 Trp-211 in a hydrophobic cleft [5]. As a result, the intrinsic tryptophan fluorescence of Ga i1 increases, and the activated switch conformation is protected from trypsin proteolysis, relative to the GDP-bound state. (A) The intrinsic tryptophan fluorescence of wild type Ga i1 increased upon injection of AlF 4 2 , while the response of Ga i1 (G42R) was blunted. (B) Ga i1 was relatively resistant to trypsin proteolysis upon loading with either GDP?AlF 4 2 or GTPcS. In contrast, Ga i1 (G42R) was efficiently proteolyzed in any nucleotide state. (C) The Ga i1 (G42R)?GDP/RGS14 GoLoco crystal structure model of this study (PDB 3QI2) is shown in cyan with the Arg-42 side chain in magenta sticks. GDP and magnesium are represented as green sticks and an orange sphere, respectively. The GoLoco motif peptide is excluded for clarity. For a complete model, see Figure S2. (D) The activated, GTPcS-bound form of wild type Ga i1 (PDB 1GIA) is shown in gray. Upon binding to the GTP analog, the switch regions (SI-III) of wild type Ga i1 converge on the phosphoryl groups of the nucleotide, resulting in a conformation recognized by effector molecules. However, the mutant Arg-42 side chain extending from the P-loop (superposed in magenta) is not sterically accommodated in a wild type-like activation state; switch 3 residues Leu-234 and Glu-236 would clash with the mutant residue. Thus, Arg-42 does not allow Ga i1 (G42R) to assume a typical active conformation, although the critical residues Glu-204 and Arg-178 apparently can be positioned for efficient GTP hydrolysis (see Fig. 2 and conidia morphology, as well as conidiation, to the wild type fungus. Both the magA and magB G42R mutants displayed different overall morphology from the corresponding Q/L mutants ( Figure S3). In the case of magA(G45R), morphology was indistinguishable from the wild type. Upon exposure to light, the magA(G45R) also produced slightly fewer conidia when compared to the wild-type M. oryzae, but magA(Q208L) formed very few heavily pigmented, aberrant conidia ( Figure 6A, inset and S4A). Both magB(G42R) and magB(Q204L) displayed enhanced conidiation relative to wild type, but those of magB(Q204L) were of a distinct morphology, with longer and thinner dimensions than either magB(G42R) or wild type ( Figure S4B, C).
These data indicate that fungal Ga G42R mutants exhibit markedly different phenotypes from truly GTPase-deficient Q/L mutants, consistent with aforementioned structural, biochemical, and cellular experiments that indicate an intact GTPase activity, but a marked inability to achieve an activated conformation.
M. oryzae expressing either G42R or Q204L mutant Ga subunits have differential effects on pathogenesis We next determined what effect the introduction of the nonactivatable G42R mutant Ga subunits has on fungal infection of barley leaves compared to constitutively active Q/L mutants. As expected, barley leaves inoculated with wild type M. oryzae showed the characteristic dose-dependent formation of disease lesions (Figure 7). The magA(G45R) strain showed similar pathogenicity as the wild type, consistent with intact surface-inducible appressorium formation ( Figure 6B). magB(G42R) displayed a reduced ability to cause disease, although small lesions were observed at the highest inoculations tested. Both magA(Q208L) and magB(Q204L) showed drastically reduced lesion formation relative to wild type and the corresponding G42R mutants. These data indicate that constitutive activity of either MagA or MagB can suppress the ability of M. oryzae to penetrate and infect the plant tissue. Additionally, we conclude that the ability of MagB to achieve its activated conformation is critical for Magnaporthe pathogenesis.

Discussion
Mutant Ga subunit strains have provided excellent tools for probing the functions of heterotrimeric G-proteins in many fungal species, including Aspergillus nidulans and Magnaporthe oryzae (Table S1) [19,26,[32][33][34][35][36][37][38][39][40][41]. Here, we have demonstrated that the P-loop mutant, G42R, is neither GTPase deficient nor constitutively active as assumed in previous studies. Rather, Ga(G42R) is unable to undergo a typical conformational change upon binding GTP, reflected by its inability to engage RGS domains or effector-like molecules. Consistent behavior of Ga(G42R) muta- Figure 5. Ga q G48R is not constitutively active in a cellular context. The analogous P-loop mutation in human Ga q , G48R, did not yield constitutive activity in contrast to the GTPase-deficient Ga q (Q209L) (A,B). Transfection of increasing amounts of Ga q (Q209L) markedly stimulated phospholipase C (PLC) activity in COS-7 cells, indicated by increased inositol phosphates (IP x ) accumulation. Like wild type Ga q , G48R overexpression did not alter PLC activity. (C,D) Endogenous and overexpressed KT3 epitope-tagged wild type Ga q stimulated PLC activity upon treatment with AlF 4 2 . The response of cells expressing Ga q (G42R) was blunted relative to wild type Ga q . doi:10.1371/journal.ppat.1002553.g005 tions was observed in three mammalian Ga subunit family members: Ga i1 , Ga oA , and Ga q . This finding, together with high sequence conservation surrounding the mutant residue ( Figure  S1) and distinct phenotypes of M. oryzae harboring either Ga(G42R) or truly GTPase-deficient Q/L mutants strongly support our hypothesis that MagA(G45R) and MagB(G42R) are structurally and biochemically similar to the corresponding mammalian Ga mutants. Our crystal structure models of Ga i1 (G42R) indicates that this perturbed conformational flexibility is likely due to steric hindrance and electrostatic repulsion between the mutant Arg-42 side chain and residues of the switch regions. The preserved GTPase activity of Ga(G42R) mutants implies that Gln-204 is still able to orient a nucleophilic water during GTP hydrolysis. The structural model of Ga i1 (G42R)?GDP bound to the GoLoco motif of RGS14 has provided a snapshot of an alternative Arg-42 rotamer that would indeed allow Gln-204 to access the orientation necessary for GTP hydrolysis. However, this rotamer still is expected to perturb the activated conformation of switch 3. We conclude that rotameric flexibility at Arg-42 allows the G42R mutant to retain GTPase activity while preventing appropriate active state switch conformations. Interestingly, previous work has identified another Ga i1 point mutation, K180P, that uncouples GTP hydrolysis from nucleotide-dependent conformational change [54]. Ga i1 (K180P) is capable of hydrolyzing GTP when not in a fully activated conformation, as also seen for Ga i1 (G42R).
Despite the retained ability of Ga(G42R) mutants to exchange and hydrolyze nucleotide, they favor an inactive state-like conformation, likely forming a less-dissociable heterotrimer with Gbc in a cellular context, thereby reducing Gbc/effector interactions. Since Ga(G42R) does not engage effectors with high affinity, it may be expected to behave as a dominant negative mutation; the Ga(G42R)/Gbc heterotrimer may serve as a substrate for receptor-stimulated exchange but fail to activate downstream signaling pathways. In Magnaporthe oryzae, it was previously unclear why strains with magB deleted or expressing the assumedly constitutively active magB G42R exhibited similar phenotypes regarding conidiation, sexual reproduction, and virulence on plant leaves [26]. The present study resolves this issue by demonstrating that the G42R mutant is not constitutively active, but likely exerts a dominant negative effect. The distinct behaviors of Ga(G42R) mutants are highlighted by a direct comparison to the truly GTPase-deficient and constitutively active Q/L mutants.
Although the magA G45R and magB G42R mutant strains do not reflect constitutive Ga subunit activity, as previously assumed [26,32], they do provide insight into fungal pathogenic development. A phenotypic deficiency upon expression of a Ga(G42R) mutant suggests that specific activation of the Ga of interest and subsequent engagement of its downstream effectors is necessary for a particular function of a cell or organism. For instance, both magB deletion [24] and magB G42R mutant strains display drastically reduced induction of appressoria by hydrophobic surfaces, while magA deletion [24] and magA G45R mutations each have minimal effects. Thus, it is likely that MagB transduces an external surface hydrophobicity signal, presumably through a GPCR. Use of the magB G42R mutant suggests that the conformational change accompanying MagB activation is necessary for the selective development of appressoria on hydrophobic surfaces ( Figure S6). It remains to be determined whether the Ga or Gbc subunits or both propagate signals required for appressorium formation and disease lesion formation in M. oryzae. Direct evidence of interactions between Magnaporthe heterotrimeric G-protein subunits and effector molecules is currently lacking. However, phenotypic similarities between the Ga subunit mutant and deletion strains [20,24,26], Gb subunit (MGB1) deletion [20], adenylyl cylase (Mac1) deletion [21], and cAMP phosphodiesterase (PdeH) deletion [22], suggest that MagA and MagB may modulate cellular cAMP level through mechanisms similar to those of mammalian Ga s and Ga i/o .
In conclusion, Ga(G42R) mutants are incapable of assuming a typical activated conformation, but their retained ability to hydrolyze GTP indicates an uncoupling of conformational change and enzymatic activity. Since G42R mutants are unable to separate from Gbc or to activate effectors, they provide tools for dissecting the functions of Ga subunits in cellular contexts. Utilizing both G42R and constitutively active Q/L mutants of two Ga subunits, we postulate a critical role for MagB activation in response to growth on hydrophobic surfaces, leading to appressorium formation in the rice blast fungus, M. oryzae.

Chemicals and other assay materials
Unless otherwise noted, all chemicals were the highest grade available from Sigma or Fisher Scientific. Peptides were synthesized by Fmoc (N-(9-fluorenyl)methoxycarbonyl) group protection, purified by HPLC, and confirmed using mass spectrometry by the Tufts University Core Facility (Medford, MA). Peptides used for crystallography and biophysical studies have been previously reported: FITC-RGS14 GoLoco [55], RGS14 GoLoco [56], FITC-KB-1753 [13], and KB-752 [49]. Barley leaf explants were spot inoculated in triplicate with the specified number of conidia (500, 100 and 2000 per inoculation site) from the magA G45R , magA Q208L , magB G42R , magB Q208L and wild type strains and the disease symptoms scored 7d post inoculation. The magA G45R caused typical disease lesions comparable to the wild type. The magA Q208L failed to cause typical blast lesions even at high spore counts. The magB G42R caused mild disease lesions on barley leaf explants inoculated with higher concentration of spores. Under comparable conditions, conidia from the magB Q208L were incapable of causing disease. doi:10.1371/journal.ppat.1002553.g007

Protein purification
Although we were unable to obtain properly folded, purified M. oryzae Ga subunits, the P-loop and surrounding switch regions are highly conserved from mammals to fungi (Figures S1). Thus, we utilized the readily available purified Ga i1 and Ga oA and corresponding G42R mutants. For biochemical experiments, full-length, hexahistidine-tagged Ga i1 and Ga oA , and G42R mutants thereof, were purified from E. coli by NTA affinity and gel filtration chromatography as previously described [57] (see Figure S5). A GST fusion of the RGS12 RGS domain (aa 664-885) was purified as described [58]. Biotinylated Gb 1 c 1 was purified as described [59]. For crystallization, an N-terminally truncated (DN30) Ga i1 (G42R) was expressed and purified by NTA affinity chromatography; the hexahistidine tag was cleaved by TEV protease, and the Ga subunit further purified by ion exchange (SourceQ, GE Healthcare) and gel filtration chromatography. Purified Ga i1 (G42R) was loaded with excess GppNHp or GDP for 3 hours at room temperature and concentrated to 15 mg/mL in GppNHp crystallization buffer (50 mM HEPES pH 8.0, 10 mM MgCl 2 , 10 mM GppNHp, 1 mM EDTA, 5 mM DTT) or GDP crystallization buffer (10 mM Tris pH 7.5, 1 mM MgCl 2 , 5% v/v glycerol, 5 mM DTT).

Crystallization and structure determination
The complex of Ga i1 (G42R) and synthetic KB-752 peptide was obtained by mixing a 1:1.5 molar ratio of protein to peptide in GppNHp crystallization buffer. Despite loading of Ga i1 (G42R) and crystallization in the presence of GppNHp, the crystal lattice contained Ga i1 (G42R) liganded with GDP and bound to KB-752. The selectivity of KB-752 for the GDP bound state [49] may account for the apparent absence of GppNHp. Crystals of Ga i1 (G42R)?GDP/KB-752 were obtained by vapor diffusion from hanging drops containing a 1:1 (v/v) ratio of protein/peptide solution to well solution (17% (w/v) PEG MME 5000, 200 mM MgCl 2 , 100 mM HEPES pH 7.0). Hexagonal rod crystals (,30061006100 mm) formed in 5 days at 18uC exhibited the symmetry of space group P6 1 22 (a = b = 106.6, c = 455.1, and a = b = 90u, c = 120u) and contained two Ga i1 (G42R)?GDP/KB-752 dimers and one Ga i1 (G42R)?GDP monomer in the asymmetric unit. For data collection at 100K, crystals were serially transferred into well solution supplemented with 30% saturated sucrose in 10% increments for ,30 s, followed by plunging into liquid nitrogen. A native data set was collected at the SER-CAT 22-ID beamline at the Advanced Photon Source (Argonne National Laboratory). Data were processed using the HKL-2000 program [60]. The crystal structure of the wild type Ga i1 /KB-752 heterodimer (PDB 1Y3A [49]), excluding the KB-752 peptide, nucleotide, and waters was used as a search model for molecular replacement using the Phaser program [61]. Refinement was carried out using phenix.refine [62], consisting of conjugate gradient minimization and refinement of individual atomic displacement and translation-libration-screw parameters, interspersed with manual revisions of the model using the Coot program [63]. For data collection and refinement statistics and a list of residues that could not be located in the electron density, see Table S2.
The complex of Ga i1 (G42R) and the RGS14 GoLoco motif peptide was generated by mixing a 1:1.5 molar ratio of protein to peptide in GDP crystallization buffer. Crystals of the complex were obtained by vapor diffusion from hanging drops containing a 1:1 ratio of protein/peptide solution to well solution (1.7 M ammonium sulfate, 100 mM sodium acetate pH 5.0, 200 mM MgCl 2 , 10% (w/v) glycerol). Crystals (,2006200650 mm) formed in 2-5 days at 18uC and exhibited the symmetry of space group C222 1 (a = 70.0, b = 131.0, c = 203.3, and a = b = c = 90u) and contained two Ga i1 (G42R)/GoLoco motif heterodimers in the asymmetric unit. Diffraction data were collected and processed, and the model refined as described for Ga i1 (G42R)/KB-752, above. The crystal structure of Ga i1 (Q147L)/RGS14 GoLoco motif (PDB 2OM2 [51]), excluding the peptide, nucleotide and waters was used as a molecular replacement search model. All structural images were made with PyMOL (Schrödinger LLC, Portland, OR).

Nucleotide binding and hydrolysis assays
The [ 35 S]GTPcS filter-binding assay used to measure rates of spontaneous GDP release from wild type and mutant Ga oA was conducted as described previously [64]. Intrinsic GTP hydrolysis rates of Ga oA and mutants were assessed by monitoring 32 Plabeled inorganic phosphate production during a single round of GTP hydrolysis as described previously [65].

Surface plasmon resonance assays
Optical detection of protein/protein interactions by surface plasmon resonance was performed using a Biacore 3000 (GE Healthcare). Carboxymethylated dextran (CM5) sensor chips (GE Healthcare) with covalently bound anti-GST antibody surfaces were created as described previously [50]. The GST-RGS12 RGS domain protein and GST alone (serving as a negative control) were separately immobilized on SPR chip surfaces. Biotinylated Gb 1 c 1 and mNOTCH peptide (serving as a negative control) were separately immobilized on a streptavidin (SA) sensor chip (GE Healthcare) as described previously [50].

Intrinsic tryptophan fluorescence measurements of Ga activation
Changes in tryptophan fluorescence of Ga i1 subunits were measured to assess activation by GDP?AlF 4 2 , as described previously [51]. Activation of Ga subunits results in translocation of a conserved switch 2 tryptophan into a hydrophobic pocket, increasing the quantum yield of tryptophan fluorescence [5]. Fluorescence intensity traces shown are representative of triplicate experiments.

Limited trypsin proteolysis
Ga subunits are relatively protected from trypsin-mediated proteolysis in the GDP?AlF 4 2 and GTP analog-bound, activated states [4]. Ten mg of wild type or mutant Ga i1 in 50 mM HEPES (pH 8.0), 1 mM EDTA, 5 mM DTT, 0.05% C 12 E 10 , and 10 mM MgCl 2 were incubated for three hours at room temperature with either 100 mM GDP, 100 mM GTPcS, or 100 mM GDP, 20 mM NaF, and 60 mM AlCl 3 . Five hundred ng of N-Tosyl-Lphenylalanine chloromethyl ketone (TPCK)-treated trypsin was added to each reaction, followed by a 10-minute incubation at room temperature. Proteolysis was stopped by addition of SDS-PAGE sample buffer and boiling. Samples were subjected to SDS-PAGE and stained with Coomassie Blue.
Quantitation of phospholipase C (PLC) activity COS-7 cells in 12-well culture dishes were transfected with KT3-tagged wild type or mutant Ga q , metabolically labeled with 1 mCi of [ 3 H]inositol/well and assayed for inositol phosphate accumulation using Dowex chromatography as described previously [66]. For AlF 4 2 stimulation experiments, final concentrations of 10 mM NaF and 30 mM AlCl 3 were added to cell media. To determine wild type and mutant Ga q expression levels, cells were lysed in SDS-PAGE sample buffer. Proteins separated by electrophoresis were immunoblotted with anti-KT3 antibody (Covance) or anti-actin antibody (Sigma).

Fungal strains, growth, and culture conditions
The M. oryzae wild-type strain B157 was obtained from the Directorate of Rice Research (Hyderabad, India). Magnaporthe strains carrying individual point mutations in the Ga subunits, namely: magA G45R , magA Q208L , magB G42R , magB Q208L have been described previously together with the rgs1D mutant [19]. Wild type and mutant strains cultures were maintained at 28uC in the dark on Prune Agar medium plates (PA; per L: 40 mL prune juice, 5 g lactose, 5 g Sucrose, 1 g yeast extract and 20 g agar, pH 6.5). Assessment of the radial growth, aerial hyphae and colony characteristics was carried out as previously described [22]. Conidiation was induced in the Magnaporthe colonies through exposure to continuous incandescent light at room temperature for 6 days.

Evaluation of conidiation status
Conidia were harvested by scraping the surface growth in water with an inoculation loop. The suspension was filtered through two layers of Miracloth (Calbiochem, San Diego, USA), collected in Falcon tubes (BD Biosciences, USA), vortexed for a minute to ensure complete detachment of conidia from the mycelia, and then pelleted by centrifugation at 3,000 rpm for 15 minutes. The conidia were washed twice and re-suspended in a fixed volume of sterile water. Prior to harvesting the spores, the radius of each colony was measured to calculate the surface area of the colony. Conidia produced by a given colony were quantified using a hemocytometer and reported as the total number of conidia present per unit area of the colony.

Appressoria formation assays
Droplets (20 ml containing 500 conidia) of conidial suspension were placed on plastic cover slips (hydrophobic surface) or hydrophilic side of GelBond membrane (Lonza Walkersville Inc., USA) and incubated in a humid chamber at room temperature. The total number of appressoria formed by each strain on either surface was quantified at 16 hpi (hours post inoculation).

Evaluation of pathogenicity in Magnaporthe strains
For pathogenicity assays, leaves from two week old barley seedlings were cut into smaller pieces (2-3 cm long) and washed in sterile water, following which the leaf bits were rinsed for 45 seconds in 40% ethanol. The leaf pieces were then washed twice with sterile antibiotic-containing distilled water. The washed leaves were placed on kinetin agar plates (2 mg/mL kinetin, 1% agar). Conidia were quantified and a dilution series of the conidial suspension was inoculated on detached barley leaves at the required concentrations. The samples were incubated in a humidified growth chamber with a 16 h light/8 h dark cycle at 22uC. Disease symptoms were assessed 5-7 days post inoculation.

Microscopic analysis
Samples were observed on a BX51 (Olympus, Japan) microscope equipped with UPlan FL N 60X/1.25 Oil objective with appropriate filter sets. Bright field images were captured using a Cool SNAP HQ camera (Photometrics, USA) and processed using Image J (National Institutes of Health, USA), MetaVue (Universal Imaging, USA) and Adobe Photoshop 7.0 (Adobe Inc, USA). Figure S1 The Ga subunit P-loop is highly conserved in fungi and mammals. The b1 strands, a1 helices, and intervening P-loops (gray), as well as the three switch regions of selected Ga subunits from humans and fungi are aligned. Nucleotide contacting residues are highlighted by black circles, and the mutated glycine by an arrowhead. (EPS) Figure S2 Arg-42 adopts an alternate rotamer in the crystal structure model of Ga i1 (G42R)?GDP/RGS14 GoLoco motif. Ga i1 (G42R) is shown in cyan with switch regions in dark blue and selected side chains in sticks. GDP is represented as green sticks, and a portion of the RGS14 GoLoco motif is orange. GoLoco motif residues 511 and 512 were disordered in the crystal structure; the cartoon shown is truncated at residue 510 (PDB 3QI2). The side chain of Arg-42 adopts a different rotamer than that seen in Ga i1 (G42R)?GDP/KB-752 (magenta sticks). Instead, the Arg side chain forms direct polar contacts with Glu-245 of Ga i1 (G42R) and the backbone carbonyl group of Val-507 from the RGS14 GoLoco motif. Arg-42 also coordinates a wellordered water molecule (yellow sphere) with Arg-242 and Gln-147 of Ga i1 (G42R). This Arg-42 rotamer would sterically prevent switch 3 from approaching the nucleotide upon binding to GTP. However, there is room for Arg-178 and Gln-204 to potentially assume their critical positions for GTP hydrolysis, providing a possible rationale for the normal GTPase activity of Ga i1 (G42R). (EPS) Figure S3 M. oryzae colony and growth characteristics. Morphology of the magA G45R , magA Q208L , magB G42R , magB Q208L , WT (wild-type) and rgs1D colonies. The indicated strains were grown in the dark on prune agar medium for a week and photographed (upper panels). The magB Q208L mutation lead to reduced rate radial growth. The radius of the magB Q208L colony was 2.2460.03 cm compared to 2.5260.03 cm in the magB G42R or the WT strain, when grown under identical conditions for a period of seven days at 28uC in the dark. Values represent the mean 6 SE (n = 5 colonies per strain; p,0.001). The lower panels represent cross sections at near-median planes. The magA Q208L showed dramatic reduction in aerial hyphal growth, compared to the magA G45R and WT. The magB G42R and magB Q208L mutants showed reduced aerial hyphal growth compared to the WT strain. (EPS) Figure S4 M. oryzae conidiation defects and conidial morphology. Comparative quantitative analysis of conidiation in the magA G45R , magA Q208L , magB G42R magB Q208L and wild type strains. The indicated strains were initially grown in the dark for a day and then exposed to constant illumination for 6 days. Data represents mean 6 SE based on three independent replicates. (A) Conidia per surface area unit were quantified for all five strains. Both magA G45R and magA Q208L produced fewer conidia than wild type fungi, although magA Q208L produced statistically significantly few conidia than magA G45R . The asterisk indicates the heavily pigmented aberrant structures and conidia with a single septum produced predominantly by the magA Q208L mutant. magB G42R and magB Q208L both displayed an increased number of conidia compared to wild type. (B) Conidia from magB Q208L displayed a thin, elongated morphology, while those of magB G42R were similar to wild type. (C) The dimensions (length and width) of conidia from the indicated strains were quantified. Values represent the mean 6 SE (n = 200 conidia per strain). (EPS) Figure S5 Purification of Ga i1 and Ga o G42R mutants. Wild type and G42R Ga i1 and Ga o were purified from E. coli by affinity chromatography, separated by SDS PAGE, and stained with Coomassie blue. (EPS) Figure S6 Activation of the Ga subunit MagB is required for selective appressorium formation on hydrophobic surfaces. Based on genetic data from the present and previous studies, a model of MagB-mediated regulation of appressorium formation in M. oryzae is hypothesized. Rgs1 was previously shown to modulate appressorium formation by negatively regulating MagA and MagB [19]. Experiments involving G42R and Q/L mutants of Ga subunits, from the present study, implicate MagB activation as a vital component of surface hydrophobicity sensing, putatively through a heptahelical GPCR.

(EPS)
Table S1 Previous studies utilizing G42R mutations in fungal Ga subunits. Investigations into Ga subunit function in multiple species have included G42R point mutations. In each case, the G42R mutant was assumed to be GTPase-deficient and constitutively active. (PDF)