Heterotrimeric G-protein Signaling Is Critical to Pathogenic Processes in Entamoeba histolytica

Heterotrimeric G-protein signaling pathways are vital components of physiology, and many are amenable to pharmacologic manipulation. Here, we identify functional heterotrimeric G-protein subunits in Entamoeba histolytica, the causative agent of amoebic colitis. The E. histolytica Gα subunit EhGα1 exhibits conventional nucleotide cycling properties and is seen to interact with EhGβγ dimers and a candidate effector, EhRGS-RhoGEF, in typical, nucleotide-state-selective fashions. In contrast, a crystal structure of EhGα1 highlights unique features and classification outside of conventional mammalian Gα subfamilies. E. histolytica trophozoites overexpressing wildtype EhGα1 in an inducible manner exhibit an enhanced ability to kill host cells that may be wholly or partially due to enhanced host cell attachment. EhGα1-overexpressing trophozoites also display enhanced transmigration across a Matrigel barrier, an effect that may result from altered baseline migration. Inducible expression of a dominant negative EhGα1 variant engenders the converse phenotypes. Transcriptomic studies reveal that modulation of pathogenesis-related trophozoite behaviors by perturbed heterotrimeric G-protein expression includes transcriptional regulation of virulence factors and altered trafficking of cysteine proteases. Collectively, our studies suggest that E. histolytica possesses a divergent heterotrimeric G-protein signaling axis that modulates key aspects of cellular processes related to the pathogenesis of this infectious organism.


Introduction
GTP-binding proteins (G-proteins) are important transducers of cellular signaling [1]. Heterotrimeric G-proteins are composed of three distinct subunits (Ga, Gb, and Gc) and typically coupled to seven-transmembrane domain, G-protein coupled receptors (GPCRs). Ga binds guanine nucleotide while Gb and Gc form an obligate heterodimer [1]. Conventionally, Ga forms a highaffinity binding site for Gbc when Ga is in its inactive GDPbound state. Activated receptor acts as a guanine nucleotide exchange factor (GEF) for Ga, releasing GDP and allowing subsequent GTP binding. The binding of GTP causes a conformational change in three flexible ''switch'' regions within Ga, resulting in Gbc dissociation. Ga?GTP and freed Gbc independently activate downstream effectors, such as adenylyl cyclases, phospholipase C isoforms, and Rho-family guanine nucleotide exchange factors (RhoGEFs) to modulate levels of intracellular second messengers [1,2]. 'Regulator of G-protein signaling' (RGS) proteins generally serve as inhibitors of Gamediated signaling [3]; however, one class of RGS protein, the RGS-RhoGEFs, serve as positive ''effectors'' for activated Ga signal transduction [2,4].
Heterotrimeric G-protein signaling has provided a wealth of targets amenable to pharmacologic manipulation, most prevalent being the GPCR itself [5]. Heterotrimeric G-proteins in mammals regulate processes as diverse as vision, neurotransmission, and vascular contractility [1,5]. Heterotrimeric G-proteins in nonmammalian organisms also exhibit a wide range of functions; for example, pheromone and nutrient sensing in yeast [6], hydrophobic surface recognition in the rice blast fungus [7], and cellular proliferation and chemical gradient sensing in the slime mold Dictyostelium discoideum [8,9].
Entamoeba histolytica causes an estimated 50 million infections and 100,000 deaths per year worldwide [10]. E. histolytica infection is endemic in countries with poor barriers between drinking water and sewage; however, outbreaks also occur among travelers and susceptible subpopulations in developed countries [11]. Upon cyst ingestion, the amoeba may colonize the human colon. Although the majority of infections are asymptomatic (e.g. ref [12]), a fraction results in symptomatic amoebic colitis. Migratory E. histolytica trophozoites attach to intestinal epithelial cells through a Gal/Gal-NAc lectin [13]. Amoebae subsequently kill host cells through a number of mechanisms, including secretion of cell-perforating amoebapores [14,15] and release of cytotoxic cysteine proteases [16].
E. histolytica has been studied for more than 50 years, and some of the signaling pathways important for pathogenesis have been identified. Several transmembrane kinases have been implicated cellular proliferation, phagocytosis, and the establishment of intestinal infection [17,18,19]. Calcium signaling is also involved in phagocytosis; for instance, calcium binding protein 1 (Eh-CaBP1) modulates the actin cytoskeleton at phagocytic cups and, together with the EhC2PK kinase, is involved in phagosome maturation [20,21,22,23]. Rho family GTPases and their activating exchange factors are also involved in a variety of pathogenic processes, including migration, phagocytosis, and surface receptor capping [24,25,26,27]. The related Rab family small GTPases control trafficking and maturation of cellular vesicles, and are implicated in processes such as phagocytosis and cysteine protease secretion [28,29,30,31].
However, many E. histolytica signaling components, and thus potential targets for therapeutic intervention, remain understudied. For example, recent sequencing of the E. histolytica genome identified multiple potential cell signal transduction components; e.g., 307 putative protein kinases representing all seven eukaryotic kinase families have been identified, including receptor tyrosine kinases [17,32]. In this paper, we describe genetic, structural, and biochemical data establishing the identity of E. histolytica heterotrimeric G-protein signal transduction components as well as their regulatory roles in pathogenesisrelated behaviors of E. histolytica.

Identification of E. histolytica heterotrimeric G-protein subunits
By a BLAST sequence similarity search with human Ga i1 (Evalue cutoff of 10 230 ), we identified a single gene in E. histolytica encoding a putative Ga subunit (EhGa1; AmoebaDB EHI_140350) also present in the related E. dispar, E. invadens, E. moshkovskii, and E. terrapinae. One Gb subunit was also identified (AmoebaDB EHI_000240) by sequence similarity to human Gb1 (E-value cutoff of 10 230 ), termed EhGb1 (Fig. S1B). As Gb subunits form obligate heterodimers with short Gc polypeptides or Gc-like (GGL) domains [3], we also searched for putative Gcencoding genes. Based on sequence similarity with S. cerevisiae Ste18 and D. discoideum gpgA, together with alignment of candidate protein sequences and identification of key functional residues, we identified two putative Gc-encoding genes named EhGc1 and EhGc2; these two open-reading frames (in the NCBI E. histolytica genomic contigs AAFB02000029.1 and AAFB02000157.1, respectively) each possess a C-terminal CAAX-box that specifies isoprenylation in conventional Gc subunits [33].

Functional assessments of E. histolytica G-protein subunits
To determine whether the identified EhGa1, EhGb1, EhGc1, and EhGc2 subunits form conventional heterodimeric (Gbc) and heterotrimeric (Ga?GDP/Gbc) complexes, bimolecular fluorescence complementation and co-immunoprecipitation assays were performed (Fig. 1A, B). The N-terminal half of yellow fluorescent protein (YFP N ) was fused to EhGc1 and EhGc2 open reading frames while the C-terminus of YFP (YFP C ) was fused to EhGb1. Only when YFP C and YFP N are fused to interacting proteins will the fluorescent protein fold and function correctly [34], as shown with the human G-protein subunits Gb 1 and Gc 2 (Fig. 1A). Significant cellular fluorescence was observed only when EhGa1 was co-transfected with YFP C -EhGb1 and either YFP N -EhGc1 or YFP N -EhGc2 (Fig. 1B). Example epifluorescence micrographs are show in Figure S3. As expected, co-expression of YFP C alone with any of the YFP N -fusions did not yield measurable cellular fluorescence. EhGb1/c1 and EhGb1/c2 dimers were found to interact with EhGa1 only in the presence of GDP (and not GTPcS) (Fig. 1C), consistent with canonical Ga?GDP/Gbc interaction selectivity.
To determine if E. histolytica Ga binds and hydrolyzes GTP, EhGa1 was purified from E. coli. Spontaneous nucleotide exchange (as measured by [ 35 S]GTPcS binding) was determined to be 0.27 min 21 and 0.064 min 21 at 30uC for EhGa1 and human Ga i1 respectively ( Fig. 2A). The observed EhGa1 exchange rate is comparable to that of Ga o [35], one of the faster spontaneous exchangers among mammalian Ga subunits. EhGa1 exhibited an intrinsic GTP hydrolysis rate of 0.21 min 21 at 20uC (Fig. 2B), comparable to rates previously observed for human Ga i1 and Ga i3 under the same conditions (e.g., ref. [36]).
Trp-196 in switch 2 of EhGa1 is universally conserved among Ga subunits (e.g., Fig. S1A) and translocates to a hydrophobic pocket upon Ga activation -an event which is easily measured as a dramatic change of intrinsic tryptophan fluorescence in select Ga subunits that lack multiple additional tryptophan residues (e.g.,

Author Summary
Entamoeba histolytica causes an estimated 50 million intestinal infections and 100,000 deaths per year worldwide. Here, we identify functional heterotrimeric G-protein subunits in Entamoeba histolytica, constituting a signaling pathway which, when perturbed, is seen to regulate multiple cellular processes required for pathogenesis. Like mammalian counterparts, EhGa1 forms a heterotrimer with EhGbc that is dependent on guanine nucleotide exchange and hydrolysis. Despite engaging a classical Gprotein effector, EhRGS-RhoGEF, EhGa1 diverges from mammalian Ga subunits and cannot be classified within mammalian Ga subfamilies, as highlighted by distinct structural features in our crystal structure of EhGa1 in the inactive conformation. To identify roles of G-protein signaling in pathogenesis-related cellular processes, we engineered trophozoites for inducible expression of EhGa1 or a dominant negative mutant, finding that G-protein signaling perturbation affects host cell attachment and the related process of contact-dependent killing, as well as trophozoite migration and Matrigel transmigration. A transcriptomic comparison of our engineered strains revealed differential expression of known virulence-associated genes, including amoebapores and cytotoxic cysteine proteases. The expression data suggested, and biochemical experiments confirmed, that cysteine protease secretion is altered upon G-protein overexpression, identifying a mechanism by which pathogenesis-related trophozoite behaviors are perturbed. In summary, E. histolytica encodes a vital heterotrimeric G-protein signaling pathway that is likely amenable to pharmacologic manipulation.
ref. [37]). Exposure to the activating reagent AlF 4 2 and magnesium (AMF) increases tryptophan fluorescence (Fig. 2C), and thus EhGa1 appears to assume a similar, activated switch conformation as conventional Ga subunits. Since the measured rates of EhGa1 nucleotide exchange (0.27 min 21 at 30uC) and hydrolysis (0.21 min 21 at 20uC) were on the same order of magnitude, we tested whether hydrolysis was rate-limiting, as seen for the A. thaliana Ga protein, AtGPA1 [38]. While EhGa1 assumes an activated conformation upon exposure to the nonhydrolyzable GTP analog, GppNHp, as indicated by intrinsic tryptophan fluorescence, addition of hydrolyzable GTP was insufficient to activate EhGa1 (Fig. 2D). Thus, nucleotide Figure 1. E. histolytica G-protein subunits form a heterotrimer in a nucleotide-dependent manner. Interactions between Gb and Gc subunits were detected with split-YFP protein complementation in COS-7 cells. (A) Human Gb1 heterodimerized with human Gb2, but not with E. histolytica Gc subunits. (B) EhGb1 interacts with EhGc1 or EhGc2 when co-expressed with EhGa1. (C) G-protein heterotrimer formation in the presence of excess GDP (''D'') or the non-hydrolyzable GTP analog, GTPcS (''T''), was examined with co-immunoprecipitation. EhGb1 and EhGc1 or EhGc2 interacted selectively with EhGa1 in its GDP-bound, inactive state. Error bars represent standard error of the mean for three experiments. * represents statistically significant difference from zero, as determined by 95% confidence intervals excluding zero. doi:10.1371/journal.ppat.1003040.g001 exchange is the rate-limiting step in the steady-state nucleotide cycling of EhGa1, as for mammalian Ga subunits, indicating that activation likely relies on GEF-stimulated exchange.

EhGa1 functional mutants
To further characterize EhGa1 activation properties and provide tools for probing G protein function in E. histolytica trophozoites, we mutated presumed key residues of the nucleotide-cycling function of EhGa1. Gln-189 in switch 2 ( Fig. S1A) is predicted to coordinate the critical nucleophilic water responsible for c-phosphoryl group hydrolysis [39]. Mutation of this residue to leucine in mammalian Ga subunits results in inability to hydrolyze GTP even in the presence of GTPase-accelerating proteins [35]. The corresponding EhGa1(Q189L) mutation abolished the ability of EhGa1 to hydrolyze GTP (Fig. 2E), suggesting a conserved role for the switch 2 Gln-189 residue in orienting the nucleophilic  60.02, as determined by single turnover hydrolysis assays. No difference was observed for selenomethionine, lysine-methylated EhGa1 used for crystallization. (C) EhGa1 changes conformation upon binding the transition state mimetic aluminum tetrafluoride. Intrinsic EhGa1 fluorescence following excitation at 285 nm increases upon activation, reflecting burial of a conserved tryptophan on switch 2 (Trp-196). (D) EhGa1 adopts an active switch conformation upon addition of the nonhydrolyzable GTP analog GppNHp, as reflected by increased intrinsic tryptophan fluorescence. The kinetics of GppNHp-mediated activation are consistent with the kinetics of radiolabeled GTP analog binding (Fig. 1A). In contrast, addition of hydrolyzable GTP does not result in EhGa1 activation, indicating that nucleotide exchange, rather than GTP hydrolysis, is the rate-limiting step in the nucleotide cycle of EhGa1. (E, F) Two EhGa1 point mutants were profiled for effects on nucleotide cycle. The dominant negative S37C possessed negligible GTP binding. The constitutively active Q189L bound but did not hydrolyze GTP. Error bars in all panels represent standard error of the mean. doi:10.1371/journal.ppat.1003040.g002 water. The Q189L mutant also exhibited a slower rate of 0.026 min 21 (95% C.I., 0.021-0.031 min 21 ) for GTPcS binding compared to wildtype (Fig. 2F), likely due to the slow rate of GTP dissociation in the absence of hydrolysis. Co-immunoprecipitation experiments demonstrated that EhGa1(Q189L) did not interact with EhGb1/c2 dimers when cell lysates were incubated with either GDP or GTPcS (Fig. S4), consistent with a state of constitutive activation.
In a mutagenesis screen [40], the mammalian Ga residue corresponding to Ser-37 of EhGa1, when mutated to cysteine, was identified as constitutively binding Gbc irrespective of whether presented with GDP or GTP analogs. We hypothesized that we could create an EhGa1 variant that constitutively binds GDP by mutating Ser-37 to cysteine. The EhGa1(S37C) mutant showed no appreciable GTPcS binding (Fig. 2F), consistent with dominant negative behavior due to disrupted GTP/Mg 2+ binding. Given that the EhGa1(S37C) mutant did not bind GTP, single turnover assays were not possible with this mutant. However, EhGa1(S37C) was observed to form a heterotrimer with EhGb1/c2 in the presence of either GDP or GTPcS (Fig. S4), consistent with dominant negative character.

Evolutionary analysis of EhGa1 and identification of a putative effector
In an attempt to identify the Ga subunit family that EhGa1 most closely resembles, we generated a phylogenetic tree comparing Ga subunits from multiple species (Fig. 3A) using MEGA5 [41]. EhGa1 is only distantly related to the metazoan Ga subunits, including the Ga 12/13 subfamily that couples to RGS-RhoGEFs. EhGa1 is most similar to D. discoideum Ga9, a Ga subunit involved in cellular proliferation [8], although low bootstrap values in the phylogram region surrounding EhGa1 indicate uncertain topology. EhGa1 also has similarity to A. thaliana GPA1 and the yeast Ga subunits, GPA1 and GPA2, the latter with roles in pheromone response and nutrient sensing, respectively [6]. The A. thaliana GPA1 regulates diverse processes, such as transpiration and cellular proliferation in response to glucose [42,43]. We also calculated sequence similarity between EhGa1 and an array of human Ga subunits based upon multiple sequence alignments. In calibrating this method, the five known Ga subunits of Drosophila melanogaster showed sequence similarity patterns allowing facile classification into each of the Ga subfamilies (Ga s , Ga i/o , Ga q , Ga 12/13 ) (Fig. S5A); however, both EhGa1 and GPA1 from Saccharomyces cerevisiae exhibited low sequence similarities to each of the human Ga subfamilies (Fig.  S5B). EhGa1 exhibits the lowest similarity to each mammalian Ga tested, implying a likely early evolutionary departure from an ancestral Ga.
The E. histolytica genome was found to encode an RGS domaincontaining RhoGEF (AmoebaDB EHI_010670; named EhRGS-RhoGEF) with distant homology to the RGS-RhoGEF effectors of mammalian Ga 12/13 subunits; no other canonical Ga effector proteins, such as adenylyl cyclases or phospholipase Cb isoforms, were identified. The transcript encoding EhRGS-RhoGEF was detected within trophozoite mRNA using quantitative RT-PCR (Fig. S2). Recombinant EhRGS-RhoGEF was therefore expressed and purified from E. coli; as measured by surface plasmon resonance, immobilized EhRGS-RhoGEF protein was found to bind EhGa1 selectively in its GDP?AlF 4 2 (AMF) nucleotide state (Fig. 3B). This selective binding is consistent with a putative EhGa1 effector function for EhRGS-RhoGEF, yet occurs in the absence of significant homology of EhGa1 to the mammalian Ga 12/13 subunits that interact with mammalian RGS-RhoGEFs [2,4].

A crystal structure of EhGa1
To gain better insight into the distant homology of EhGa1 versus other Ga subunits, we determined a crystal structure of EhGa1 bound to GDP by single-wavelength anomalous dispersion (SAD) using data to 2.6 Å resolution (Table S1; Fig. S6). To obtain high-quality diffracting crystals, we modified EhGa1 by removing its extended N-terminal helix (a.a.  and subjecting it to reductive lysine methylation. Neither alteration perturbed the nucleotide cycle or activation kinetics of EhGa1 (Fig. 2B,C). EhGa1 features the highly conserved Ras-like and all-helical domain structure and nucleotide-binding pocket characteristic of Ga subunits (Fig. 4A). The three switch regions are ordered in one of the two monomers in the asymmetric unit, likely due to crystal contacts (Fig. S6B). EhGa1 exhibits a highly conserved mode of nucleotide interaction, including the dispositions of residues Ser-37 and Gln-189 (Fig. 4B). The guanine ring is embraced by the conserved NKxD motif (residues 254-257; Fig. S7), with the hydrophobic portion of Lys-255 packing against the planar guanine ring. The phosphate-binding loop (P-loop) forms numerous polar contacts with the aand b-phosphoryl groups of GDP [39].
Unique to EhGa1 is the absence of an aB helix in the all-helical domain (Fig. 4A). Although the segment between aA and aC (aA-aC loop) could be affected by crystal packing, five prolines scattered throughout this loop (positions 84, 89, 99, 103, and 106; Fig. S1A) suggest this region likely also lacks helical structure in solution. GoLoco motif-containing proteins are one of the few molecules that interact with the aB helix (e.g., ref. [37]); not surprisingly, given the lack of a structurally-conserved binding site on EhGa1, the E. histolytica genome does not seem to encode any GoLoco motifs. EhGa1 also harbors a unique 16-residue insert in the Ras-like domain following the a4 helix (Figs. 4A, S1A). A portion of this insert forms a short b-strand (here termed b7) that extends the six-stranded b-sheet common to all heterotrimeric and Ras-family GTPases [44,45], followed by a 15-residue loop that is disordered in our crystal structure. This region of Ga is critical for interaction with GPCRs as seen, e.g., in the crystal structure of the b2 adrenergic receptor/Gs complex [46]. Because this region is important for receptor coupling and/or specificity, the existence of this insert in EhGa1 suggests a potentially unique GPCR-coupling mechanism in E. histolytica, but no receptor has yet been identified (see Discussion).

G-protein signaling perturbation modulates trophozoite migration, Matrigel transmigration, and host cell attachment and killing
To determine roles of heterotrimeric G-protein signaling in pathogenesis-related behaviors of E. histolytica, HM-1:IMSS trophozoites were stably transfected with tetracycline-inducible expression plasmids [47] encoding either wildtype EhGa1 or the dominant negative EhGa1 S37C (Fig. 5A). A strain expressing the constitutively active EhGa1 Q189L could not be established, potentially due to cellular toxicity; however, overexpression of wildtype EhGa1 is expected to result in a moderately higher basal level of signaling to downstream components. Overexpression of signaling components is subject to limitations, including the possibility that supra-physiological expression levels and/or protein mislocalization result in toxicity or other cellular effects not typically mediated by endogenous signaling. However, this approach is useful to suggest cellular processes that may be regulated by heterotrimeric G-protein signaling and to mimic the gross perturbation that may be achieved with pharmacological agents acting on this pathway. Immunofluorescence of overex-pressed EhGa1 revealed a diffuse, cytoplasmic cellular distribution that did not differ significantly between the wild type and S37C mutant strains (Fig. S8A). Endogenous EhGa1 was not assessed due to a current lack of specific antibodies. To assess potential effects of Ga subunit overexpression on trophzoite growth and viability, growth curves were assessed for the parent HM-1:IMSS, EhGa1 wt , and EhGa1 S37C strains in the presence and absence of tetracycline. No significant differences in growth or viability (.90% at all time points) were observed over three days, although trophozoites expressing EhGa1 S37C displayed a trend toward slower growth at day 3 (Fig. S8B). All subsequent cellular experiments were conducted following growth with or without tetracycline for 24 hours.
Trophozoite motility is related to the pathogenesis of amoebic colitis, likely contributing to tissue invasion [48,49]. Tetracyclineinduced EhGa1 wt overexpression increased migration in the absence of a serum stimulus while EhGa1 S37C expression reduced migration in the presence or absence of serum in Transwell migration assays (Fig. 5B), suggesting that perturbation of heterotrimeric G-protein signaling may regulate motility at baseline and potentially in response to serum factor stimuli. However, the reduced migration of the EhGa1 S37C strain in the presence of serum may be due to the lower baseline trophozoite motility, as observed in the absence of a serum stimulus, rather than due to specific heterotrimeric G-protein involvement in a signaling response to serum factors. Tetracycline treatment had no measurable effect on the migration of the HM-1:IMSS parent strain or trophozoites transfected with an empty expression vector (Fig. S9A).
E. histolytica invades the intestinal mucosa, giving rise to ulcers and, in rare cases, systemic amoebiasis [50,51]. To assess migration across a barrier, transfected trophozoite strains were profiled by a Transwell assay, with upper and lower chambers separated by Matrigel. Induced expression of EhGa1 wt enhanced, but EhGa1 S37C reduced, Matrigel transmigration relative to uninduced controls (Fig. 5C), revealing a potential regulatory role for heterotrimeric G-protein signaling. Tetracycline treatment had no effect on the transmigration of HM-1:IMSS or empty vectortransfected trophozoites (Fig. S9B). The effects of EhGa1 wt and EhGa1 S37C overexpression on Matrigel transmigration displayed the same trends seen for migration in the absence of serum (Fig. 5B). Thus, differential baseline migration rates may account for part or all of the observed differences in Matrigel transmigration.
E. histolytica trophozoites also attach to and kill host cells, including intestinal epithelium and responding immune cells. Host cell attachment, achieved primarily through a galactose-inhibitable lectin [13,52], is required for subsequent cell killing. Trophozoites expressing EhGa1 wt displayed greater attachment to CHO cell monolayers than uninduced controls, and the opposite effect was seen in the EhGa1 S37C strain (Fig. 6A, S10). EhGa1 wt overexpression enhanced Jurkat cell killing, as assessed with a membrane integrity assay, while trophozoites expressing the dominant negative EhGa1 S37C were less cytotoxic (Fig. 6B). Tetracycline treatment had no effect on host cell attachment or killing by HM-1:IMSS or empty vector-transfected trophozoites ( Fig. S9C, D). Thus, perturbation of heterotrimeric G-protein signaling also regulates host cell killing by E. histolytica. Similar patterns were observed in host cell attachment and cell killing assays; different degrees of attachment upon expression of EhGa1 wt or EhGa1 S37C may be partially or wholly responsible for the observed changes in contact-dependent cell killing.

Regulation of transcription by perturbed heterotrimeric G-protein signaling
To gain insight into potential mechanisms by which perturbation of EhGa1 expression controls pathogenesis-related behaviors in E. histolytica, RNA-seq was performed on mRNA isolated from trophozoites expressing EhGa1 wt , EhGa1 S37C , and uninduced controls. To emphasize highly transcribed genes and eliminate potential transcriptional effects of tetracycline treatment, transcripts with a Fragments Per Kilobase of exon per Million fragments mapped (FPKM) value less than 10 and transcripts that were up-or downregulated (in the same direction) in both EhGa1 wt and EhGa1 S37C samples (24 hour tetracycline treatment at 5 mg/mL) relative to uninduced (tetracycline-free) trophozoites were excluded. Twentyone genes were differentially transcribed in opposite directions upon expression of either EhGa1 wt or EhGa1 S37C (Fig. 7A). Transcriptional changes of multiple genes were verified over a 24 hour time course by RT-PCR (Fig. S11). For instance, EhGb1 was found to be more highly expressed in trophozoites expressing EhGa1 S37C . Analysis of putative functions for the differentially transcribed genes revealed a diversity of responses to altered heterotrimeric G-protein signaling (Fig. 7B). Stress responserelated transcripts, such as those encoding heat shock proteins, were exclusively down-regulated upon EhGa1 wt expression and up-regulated in the dominant negative EhGa1 S37C strain; conversely, numerous metabolic enzymes were selectively upregulated following expression of EhGa1 S37C , suggesting that heterotrimeric G-protein signaling may be involved in sensing and responding to vital extracellular nutrients.
Genes with known effects on E. histolytica pathogenesis were also differentially transcribed, as measured by RNA-seq. (Table S2). For example, the host cell lytic factor amoebapore C was upregulated upon EhGa1 wt expression, while the amoebapore A precursor was down-regulated by EhGa1 S37C (Table S2), consistent with the higher or lower cell killing efficiencies, respectively, of each strain (Fig. 6B) [14,15,48]. Down-regulation of amoebapore A upon expression of EhGa1 S37C was confirmed by RT-PCR at the transcriptional level, and by western blot at the protein level ( Fig. S11; anti-amoebapore A was a gift from Dr. M. Leippe, U. of Kiel, Germany). A number of cysteine proteases, known factors in both host cell killing and Matrigel transmigration [53], were differentially transcribed following expression of EhGa1 S37C (Table S2). The down-regulation of one cysteine protease (EHI_006920) was confirmed by RT-PCR (Fig. S11). Ten Rab family GTPases, known to regulate vesicular trafficking and cysteine protease secretion [28], as well as other putative secretion/trafficking proteins, were also differentially transcribed. Specifically, four cysteine protease binding factors (CBPFs), recently shown to modulate cysteine protease secretion [54], were down-regulated in trophozoites expressing EhGa1 S37C (Table S2). These transcriptional effects suggested that altered cysteine protease activity and/or secretion may be a mechanism by which perturbation of heterotrimeric G-protein signaling modulates Matrigel transmigration and host cell killing (Figs. 5C & 6B). To test this hypothesis, intracellular and secreted cysteine protease activities were each measured in the EhGa1 wt and EhGa1 S37C strains. EhGa1 wt expression increased extracellular and decreased intracellular cysteine protease activity, likely reflecting more efficient vesicular trafficking and secretion (Fig. 7C). In contrast, EhGa1 S37C expression resulted in a trend toward more intracellular protease activity, although not statistically significant (p = 0.07), and significantly less extracellular protease activity relative to uninduced control trophozoites, correlating with Figure 4. Structure of EhGa1 reveals a conserved fold with unique features. The crystal structure of EhGa1 was determined by single anomalous dispersion (SAD) using 2.6 Å resolution data (Table S1). (A) The EhGa1 Ca backbone is shown in green, bound to GDP in purple sticks. Conserved switch regions (SW 1-3) are dark blue. Trp-196 is solvent-exposed in the inactive state and buried when switch 2 adopts its activated conformation (e.g., Fig. 2C). Unique among Ga subunits, EhGa1 lacks an aB helix in the all-helical domain (red; labeled 'aA-aC loop') but possesses a unique short b-strand insert (b7) and a loop (orange) between the conserved a4 helix and b6 strand. Disordered regions in switch 3 (residues 222 and 223) and the b7-b6 loop (residues 302-310) are indicated by dashed lines. (B) Ser-37, conserved among Ga subunits, is an important ligand for Mg 2+ , a cofactor for GTP binding and hydrolysis. Mutation of Ser-37 to Cys is predicted to produce a dominant negative EhGa1 [40]. Gln-189 is required for orienting the nucleophilic water during GTP hydrolysis; its mutation to Leu is predicted to cripple GTPase activity, yielding a constitutively active EhGa1. doi:10.1371/journal.ppat.1003040.g004 Trophozoites were stably transfected to express wildtype EhGa1 or dominant negative EhGa1 S37C under tetracycline control. (B) EhGa1 wt -expressing trophozoites showed greater migration across a porous membrane in the absence of stimuli (serum-free) while amoebae expressing EhGa1 S37C showed lower migration toward both serum-free and serum-containing nutritive media. Migration of HM-1:IMSS trophozoites was not significantly different from the non-induced EhGa1 wt and EhGa1 S37C strains. Tetracycline treatment was 5 mg/mL over 24 hours. (C) Trophozoites expressing EhGa1 wt were better able to migrate through a Matrigel layer than uninduced controls. Conversely, EhGa1 S37C expression greatly reduced Matrigel transmigration. Parent strain HM-1:IMSS trophozoites were unaffected by tetracycline treatment and were indistinguishable from non-induced EhGa1 wt and EhGa1 S37C . Error bars represent standard error of the mean. * represents statistical significance by an unpaired, two-tailed Student's ttest (p,0.05) for four independent experiments. doi:10.1371/journal.ppat.1003040.g005 Trophozoites attach to CHO cell monolayers, primarily through a galactose-inhibitable lectin. Overexpression of EhGa1 wt enhanced monolayer attachment, while expression of EhGa1 S37C reduced attachment. Parent strain HM-1:IMSS trophozoites were unaffected by tetracycline treatment and were indistinguishable from non-induced EhGa1 wt and EhGa1 S37C . Attached trophozoites quantities were obtained by multiplying detached cell concentrations by a dilution factor. * indicates a statistically significant difference (p,0.05) between quadruplicate experiments. Error bars represent standard error of the mean. * indicates statistical significance by an unpaired, two-tailed Student's t-test (p,0.05) for four independent experiments. (B) Amoebae overexpressing EhGa1 wt or EhGa1 S37C displayed enhanced or reduced abilities to kill Jurkat (human T-lymphocyte) cells, respectively, as measured by LDH release in a membrane integrity assay. Cell killing by HM-1:IMSS trophozoites was not altered by tetracycline treatment. 0.5% Triton X-100 was added to Jurkat cells to define 100% host cell lysis. Tetracycline treatment was 5 mg/mL over 24 hours. Error bars represent standard error of the mean. * indicates statistical significance by an unpaired, two-tailed Student's t-test (p,0.05) for three independent experiments, with four technical replicates each. doi:10.1371/journal.ppat.1003040.g006

Discussion
Here we demonstrate that functional heterotrimeric G-protein subunits are encoded by the pathogen Entamoeba histolytica, including single Ga and Gb subunits, and two Gc subunits. Like their mammalian counterparts, EhGa1, EhGb1, and EhGc1/2 form a nucleotide state-dependent heterotrimer. EhGa1 binds and hydrolyzes GTP and its switch regions undergo a conserved conformational change. When in an activated state, EhGa1 is seen to engage a putative effector protein, namely an RGS domaincontaining RhoGEF (EhRGS-RhoGEF). EhRGS-RhoGEF likely represents a functional signaling link between heterotrimeric Gproteins and Rho family GTPases in E. histolytica. Indeed, Rho GTPases and other Dbl family RhoGEFs in E. histolytica have been implicated in multiple processes important for pathogenesisrelated processes such as actin reorganization during chemotaxis, surface receptor capping, cell killing, phagocytosis, and tissue destruction [24,25,26,27,55].
The sequence of EhGa1 diverges from each of the mammalian Ga subunit subfamilies, including the Ga 12/13 subfamily that couples to RGS-RhoGEFs. Thus EhGa1 likely represents an early evolutionary departure from the metazoan Ga/RGS-RhoGEF signaling axis, or possibly a signaling pathway of similar function with an independent evolutionary origin. A search of publicly available genome sequences using SMART [56] identified the RGS and DH-PH domain combinations exclusively in metazoan species, with the only exception being the amoebazoans. This lack of RGS-RhoGEF related proteins in non-metazoan species suggests an independent origin of the E. histolytica Ga/RGS-RhoGEF interaction; however, we cannot rule out the possibility that a Ga/RGS-RhoGEF interaction arose early in evolutionary history, such as an ancestral Unikonta supergroup member (e.g. [57]), and was later lost in fungal species, but retained in metazoans and amoebae. Among the species compared in this study, EhGa1 was found to be most similar in sequence to the D. discoideum Ga9, followed more distantly by S. cerevisiae GPA1 and GPA2, as well as A. thaliana GPA1. This set of Ga subunits is only loosely related by function, with D. discoideum Ga9 regulating cellular proliferation [8], while yeast GPA1 and GPA2 transduce signals in response to pheromones and nutrients, respectively [6]. A variety of downstream signaling machinery is utilized as well, with S. cerevisiae pheromone signaling occurring predominantly through Gbc subunit effectors, while S.c. GPA2 engages an adenylyl cylase effector [6]. The current study clearly differentiates EhGa1 from these relatively similar Ga subunits on the sequence level, demonstrating interaction with an RGS-RhoGEF effector and no significant effect on cellular proliferation, but apparent roles in multiple pathogenesis-related processes of E. histolytica.
Perturbation of heterotrimeric G-protein signaling in E. histolytica trophozoites was observed to modulate migration, Matrigel transmigration, and host cell attachment and killing. Notably, trophozoite Matrigel transmigration is dependent on general migration to some degree, and host cell killing is dependent on attachment. Thus, the effects of heterotrimeric Gprotein perturbation on Matrigel transmigration and host cell killing may be partially or wholly due to the alterations in migration and attachment, respectively. Induced expression of the dominant negative EhGa1 S37C impaired these pathogenic processes, suggesting that antagonizing G-protein signaling may reduce E. histolytica virulence. The complete mechanisms by which heterotrimeric G-proteins are linked to specific trophozoite behaviors remain to be elucidated. For instance, it is presently unclear which signaling cascades are utilized to effect transcriptional changes in response to perturbed EhGa1 expression. EhGa1 likely engages its RGS-RhoGEF effector, leading to activation of specific Rho GTPases, some of which are known to regulate cytoskeletal dynamics required for such processes as migration and Matrigel transmigration [24,27,55,58]. EhGbc may also engage as yet unidentified effectors, like its homologs in other species, leading to changes in pathogenic processes [1].
It is presently unclear how heterotrimeric G-protein signaling is activated in E. histolytica. Since nucleotide exchange is the ratelimiting step in the nucleotide cycle of EhGa1, an exchange factor, such as a GPCR, is likely required for high levels of EhGa1 activation. At this time, the only putative GPCR described is the Rab GTPase-binding protein EhGPCR-1 [59]. While it would be compelling to demonstrate receptor-mediated nucleotide exchange on EhGa1, our own bioinformatic analysis revealed that EhGPCR-1, while containing seven-transmembrane spanning regions, is more likely a conserved Wnt-binding factor required for Wnt secretion (as seen in C. elegans) [60]. Identification of a bona fide GPCR/ligand pair or other heterotrimeric G-protein activation mechanism in E. histolytica will provide powerful tools for further probing of the roles of heterotrimeric G-protein signaling in trophozoites.

Cloning of E. histolytica G-protein subunits
The open reading frame (ORF) of EhGa1 was amplified from E. histolytica genomic DNA (Dr. M. Vargas, Center of Investigation and Advanced Studies, Mexico City) by polymerase chain-reaction (PCR) using Phusion polymerase (New England BioLabs) and Invitrogen primers. Amplicons were subcloned using ligationindependent cloning [61] into a Novagen pET vector-based prokaryotic expression construct (''pET-His-LIC-C'') to form Nterminal tobacco etch virus (TEV) protease-cleavable, hexahistidine-tagged fusions. Mutations were made using QuikChange sitedirected mutagenesis (Stratagene). ORFs of EhGa1, EhGb1, EhGc1, and EhGc2, codon-optimized for mammalian cells, were obtained from Geneart (Regensburg, Germany); EhGa1 with an internal FLAG epitope, DYKDDDK inserted after His-83, was also obtained for co-immunoprecipitations. Sequences for EhGc1 and EhGc2, identified in genomic shotgun sequences were MSQQQLTRLLQEKERLMKNFERSKNLMKVSEACSDLV-NFTKSKVDPFSPEFKDSNPWDKNNEGGCCALV and MSQ-QQLIRLLQEKERLMKNFERSKNLMKVSEACSELVNFTK-NKIDPFSPEFKDTNPWDKSSNAGCCSLM, respectively. genes were categorized by putative function based on prior studies, homology to genes of known function, or predicted protein domains of known function. ''Virulence/encystation'' category includes genes known to modulate E. histolytica pathogenesis, such as cysteine proteases [48]. (C) Both intracellular and secreted cysteine protease activities were assessed with an azo-collagen assay. EhGa1 wt overexpression enhanced cysteine protease secretion, while EhGa1 S37C expression resulted in less extracellular (E), despite higher intracellular (I), cysteine protease activity, suggesting that transcriptional responses downstream of heterotrimeric G-protein signaling modulate E. histolytica pathogenic processes in part by regulating cysteine protease secretion. Tetracycline treatment in all experiments was 5 mg/mL over 24 hours. * = statistical significance by an unpaired, twotailed Student's t-test (p,0.05) for four independent experiments. doi:10.1371/journal.ppat.1003040.g007 Protein purification, crystallization, and structure determination See the Supplementary Methods for details.

Fluorescence complementation and coimmunoprecipitation
Yellow fluorescent protein (YFP) bimolecular fluorescence complementation was performed as described [62] with modifications below. Codon-optimized ORFs of EhGc isoforms were subcloned as HA-tagged fusions to the N-terminal 159 amino acids of YFP-venus (pcDNA3.1-YFP N ; Dr. Nevin Lambert, MCG). The EhGb1 ORF was subcloned as an HA-tagged fusion with a C-terminal fragment (residues 159-239) of YFP-venus (pcDNA3.1-YFP C ; also obtained from Dr. Lambert, along with control YFP N -human Gc 2 and YFP C -human Gb 1 fusions). 200,000 COS-7 cells per well in 6-well dishes were transfected with 1 mg DNA using FuGENE-6 as per manufacturer's directions. Empty pcDNA3.1 DNA was used to maintain a constant amount of total DNA per well. Forty-eight hours post-transfection, epifluorescence was observed using an Olympus IX70 microscope with Hamamatsu monochrome CCD camera. Digital images were imported into MATLAB 2007a and quantified as previously described [62]. Pixels with greater than 40 units of intensity were considered to be fluorescent, and the percentage of positive pixels was quantified. All experiments were repeated three times. Coimmunoprecipitation was performed using the YFP-fusion proteins as previously described [62].

Nucleotide binding, hydrolysis, and EhGa1 activation
Spontaneous GDP release, measured by [ 35 S]GTPcS incorporation, and [c-32 P]GTP hydrolysis by single turnover assays were both quantified as previously described [37]. For GTPase acceleration assays, increasing concentrations of purified EhRGS-RhoGEF were added along with the hydrolysis-initiating magnesium. Real-time monitoring of EhGa1 tryptophan fluorescence (excitation 280 nm; emission 350 nm) was conducted as described for Ga i1 [37].

Evolutionary analysis
The protein sequences of Ga subunits from humans, S. cerevisiae, A. thaliana, D. melanogaster, and D. discoideum were aligned and an unrooted phylogram derived using T-coffee [63]. Percent amino acid sequence similarities of EhGa1 and S. cerevisiae GPA1 were calculated relative to each human Ga subunit, using a multiple sequence alignment, as described previously [64]. The Ga family of Drosophila melanogaster served as a positive control for subfamily classification.

Surface plasmon resonance
Optical detection of protein binding was conducted as described previously [65]. Briefly, His 6 -tagged EhRGS-RhoGEF was immobilized on an NTA chip surface and increasing concentrations of wildtype EhGa1 and mutants were flowed over at 10 mL/s in various nucleotide states.

Trophozoite migration and Matrigel transmigration
Trophozoite migration assays were performed essentially as described previously [67]. Briefly, amoebae were grown in the presence or absence of 5 mg/mL tetracycline for 24 hours, harvested in log growth phase, suspended in serum free TYI growth medium, and 50,000 cells loaded in the upper chamber of a Transwell migration chamber (Costar, 8 mm pore size). The lower chamber contained growth medium with or without 15% adult bovine serum. Transwell plates were incubated at 37uC for 2 hr under anaerobic conditions (GasPak EZ, BD Biosciences). Matrigel transmigration assays were performed in similar fashion, except that Matrigel was first diluted to 5 mg/mL in serum free TYI growth medium, layered on the Transwell porous filter, and allowed to gel for 6 hr prior to assay initiation. Incubation time was also extended to 16 hr to allow penetration. Migrated trophozoites attached to the lower chamber wall were detached on ice, fixed, and counted. Each experiment was performed in triplicate and statistical significance among four independent experiments was determined by an unpaired, two-tailed Student's t-test.

Host cell attachment
Attachment of E. histolytica trophozoites to epithelial monolayers was assessed as previously described [68]. Chinese hamster ovary (CHO) cells were grown to confluency in 24-well plates, washed, and fixed in 4% paraformaldehyde for 30 minutes. Trophozoites (3610 5 ) grown in the presence or absence of 5 mg/mL tetracycline for 24 hours were added to the fixed monolayers in medium 199 supplemented with 5.7 mM cysteine, 1 mM ascorbic acid, and 25 mM HEPES (pH 6.9). After incubation at 37uC for 30 minutes, each well was washed gently two times with warm PBS to remove unattached trophozoites. Monolayer-attached trophozoites were detached on ice and quantified by counting with an inverted microscope. In similar experiments, trophozoites were labeled with carboxyfluorescein diacetate succinimidyl ester (CFDA-SE). Attached fluorescent trophozoites were counted in three microscopic fields at 106 magnification. Each experiment was performed in quadruplicate and statistical significance determined by an unpaired two-tailed Student's t-test.

Cell killing
Killing of mammalian cells (Jurkat) was assessed using the CytoTox-ONE membrane integrity assay (Promega). In 96-well plates, 5610 5 Jurkat cells and/or 2.5610 4 trophozoites, grown with or without 5 mg/mL tetracycline for 24 hours, were incubated at 37uC in 200 mL of medium 199 (Sigma) supplemented with 5.7 mM cysteine, 0.5% BSA, and 25 mM HEPES pH 6.8. After 2.5 hr, 50 mL of medium from each well was incubated with Cytotox reagent and a colorimetric measure of extracellular lactate dehydrogenase activity was obtained after 10 min. 0.5% Triton X-100 was used to define 100% host cell death. Each experiment was performed with five replicates and statistical significance among three independent experiments was determined by an unpaired two-tailed Student's t-test.

Whole transcriptome shotgun sequencing
Total RNA from 10 6 trophozoites each of the tetracyclineinduced (5 mg/mL tetracycline for 24 hours) EhGa1 wt and EhGa1 S37C strains, as well as a tetracycline-free control, was isolated using an RNeasy Mini Kit (Qiagen) per manufacturer's instructions. Duplicate RNA purifications and sequencing were obtained for each condition.
Quality of total RNA from each sample was estimated by automated electrophoresis (Bioanalyzer, Agilent). Libraries were constructed using TruSeq RNA library preparation kits (Illumina) according to manufacturer's recommendations; molarity was estimated by analysis of DNA concentration from fluorometer detection and DNA fragment size. Prepared libraries with equal molarity were pooled and used for multiplex sequencing reactions. Libraries were sequenced using 57 cycles in a single end Illumina flowcell v.3 on a HiSeq2000 instrument (Illumina) at the UNC High Throughput Sequencing Facility. Primary data analysis and demultiplexing was performed using a standard Illumina pipeline 1.8.2.
Resulting mRNA sequence reads were mapped to the annotated Entamoeba histolytica genome (AmoebaDB.org) using Bowtie v0.12.7 [69]. Between 12610 6 and 32610 6 reads were aligned for each sample. Aligned reads were further analyzed with Cufflinks v1.3.0 [70] and visualized using the Integrative Genomics Viewer (www.broadinstitute.org/igv/). Cuffdiff was used to determine differential expression by comparing relative transcript abundances between pairs of duplicate experiments: EhGa1 wt expression vs tetracycline-free control, EhGa1 S37C expression vs tetracycline-free control, and EhGa1 wt vs Eh-Ga1 S37C expression. Genes exhibiting statistically significant differential transcription were compiled and corresponding annotations retrieved using software from Dr. Chung-Chau Hon (Institut Pasteur) [71]. Transcripts that were either up-or downregulated in both the induced EhGa1 wt and EhGa1 S37C strains were excluded from further analysis, because of potential transcriptional modulation due to tetracycline treatment. Functions of the associated proteins were inferred from prior E. histolytica studies, by similarity to mammalian protein families, or from conserved domains of known function. All encoded proteins without annotated conservation and those with domains of unknown function were classified as ''unknown''.

Cysteine protease activity
Intracellular cysteine protease activity in amoebic lysates was assayed essentially as described previously [72]. Crude extracts of 10 6 trophozoites, grown with or without 5 mg/mL tetracycline for 24 hours, were obtained by lysing with 5 cycles of freeze-thaw. Total protein concentration was quantified by Bradford's method. 2 mg of azo dye-impregnated collagen (Sigma) with 100 mg of crude extract in 500 mL of protease activation buffer (100 mM Tris pH 7.0 and 10 mM CaCl 2 ) were incubated at 37uC for 18 hr, then terminated with 500 mL of 10% TCA. Samples were centrifuged to exclude intact collagen fibers, and supernatants collected for absorbance reading at 540 nm. In parallel experiments, the inhibitor p-hydroxy-mercuribenzoic acid (PHMB) was included at 1 mM to assess the fraction of specific cysteine protease activity. Residual protease activity (after PHMB treatment) was subtracted to determine total cysteine protease activity.
Extracellular cysteine protease activity was also assayed with azo-collagen as described above. However, 10 6 trophozoites were incubated at 37uC for 3 hr in 500 mL PBS supplemented with 20 mM cysteine, 0.15 mM CaCl 2 , and 0.5 mM MgCl 2 , conditions known to sustain E. histolytica growth and allow cysteine protease secretion [53]. Following centrifugation, the cell-free conditioned medium was assayed for cysteine protease activity as above. Statistical significance was determined by an unpaired, two-tailed Student's t-test.
Accession numbers for proteins used in this study EhGa1, AmoebaDB EHI_140350; EhGb1, AmoebaDB EHI_000240; glyceraldehyde-3-phosphate dehydrogenase, Amoe-baDB EHI_167320; EhRGS-RhoGEF, AmoebaDB EHI_010670. EhGc1, identified within the NCBI genomic contig AAFB02000029.1; EhGc2, identified within the NCBI genomic contig AAFB02000157.1; amoebapore A, AmoebaDB EHI_159480; cysteine protease, AmoebaDB EHI_006920. Figure S1 The genome of Entamoeba histolytica encodes heterotrimeric G-protein subunits. (A) A multiple sequence alignment of EhGa1 with selected Ga subunits from other species (Dd = Dictyostelium discoideum, Sc = Saccharomyces cerevisiae, Hs=Homo sapiens). The secondary structure information above the aligned sequences reflects the crystal structure of EhGa1 (this study), with naming adapted from human transducin (PDB 1TND). Residues mutated in this study are marked with black arrowheads, and gray bars indicate relative sequence identity. A 110-residue insert within Sc GPA1 (gray box) was omitted for clarity. Although a number of E. histolytica proteins are reportedly ADP-ribosylated by pertussis toxin [73], EhGa1 is not likely to be a substrate as it lacks the C-terminal cysteine ADP-ribosylation site shared among conventional Ga i/o subunits (e.g., Cys-351 in human Ga o ). Based on the sequence of the amino terminus of EhGa1, it is likely that this protein is myristoylated on its second residue (glycine) and palmitoylated on its third residue (cysteine) [33]. (B) EhGb1 is aligned with selected Gb subunits in a fashion identical to panel A with secondary structure elements as found in transducin Gbc (PDB 1TBG). (EPS) Figure S2 Heterotrimeric G-protein signaling components are expressed in E. histolytica. qRT-PCR amplification of RNA isolated from HM1 E. histolytica trophozoites (a kind gift of Dr. William Petri, Jr.) confirmed transcription of EhGa1, EhGb1, EhGc1, and EhRGS-RhoGEF genes. The basally expressed housekeeping gene GAPDH was included as a control. DC t reflects the difference in threshold cycle relative to reactions lacking reverse transcriptase, used as a control for DNA contamination. Error bars represent standard error of the mean. (EPS) Figure S3 Example bimolecular fluorescence complementation micrographs. YFP fluorescence was detected microscopically in COS-7 cells expressing heterotrimeric G-protein subunits. YFP complementation was observed when EhGa1 was coexpressed with EhGb1 and EhGc1 (A, B) or EhGc2 (C, D). The human subunits Gb1 and Gc 2 exhibited complementation, while the expressed N-and C-terminal fragments of YFP did not (E, F). For a quantification of fluorescence, see Figure 1. (EPS) Figure S4 The inactive EhGa1(S37C) constitutively binds to EhGb1c2, while the constitutively active EhGa1(Q189L) mutant does not. Co-immunoprecipitations of EhGa1 and mutants with EhGb1 and EhGc2 were conducted as in Figure 1. As predicted, the dominant negative S37C mutant remains bound to EhGb1c2, even in excess GTPcS. The constitutively active, GTPase-deficient Q189L mutant does not bind EhGb1c2 in either nucleotide state. (EPS) Figure S5 Mammalian Ga subfamily homology analyses. Sequence similarity to human Ga subunits was plotted for the Ga subunits from Drosophila melanogaster (A), Saccharomyces cerevisiae GPA1, and EhGa1 (B). In contrast with D. melanogaster subunits, EhGa1 cannot be classified as a member of any particular Ga subfamily. (EPS) Figure S6 Structural comparison of EhGa1 with Hs transducin and switch 2 crystal contacts. (A) The two EhGa1 molecules in the asymmetric unit are highly similar, although switch 2 of chain B (wheat) is partially disordered. (B) Crystal contacts between the ordered switch 2 of chain A (blue) and a neighboring molecule (orange) likely account for the structural differences between the two molecules in the asymmetric unit. The non-polar  interface with a hydrophobic patch on a neighboring molecule. Switch 2 may be drawn away from the nucleotide pocket, accounting for the absence of bound AlF 4 2 (see discussion below). (C, D) The model of EhGa1 is superposed with human transducin in two nucleotide states (slate blue, AMF, PDB 1TAD; teal, GDP, PDB 1TAG). EhGa1 lacks an aB helix seen in transducin and all other Ga subunits and contains a unique a4-b6 insert (orange). Switch 2 of EhGa1 (chain A) adopts a distinct conformation from both the active and inactive forms of transducin, likely due to crystal contacts with a neighboring molecule. (EPS) Figure S7 Electron density map of guanine nucleotide binding pocket of EhGa1. A region of the 2F o -F c electron density map is shown in stereo view from the structure of EhGa1 (yellow sticks) bound to GDP (purple sticks). The nucleotide binding pocket is highly similar to mammalian Ga subunits, featuring a conserved phosphate binding loop (P-loop; Glu-33 shown) and an NKxD motif (residues 254-257). Switch one also directly contacts the nucleotide, and Arg-163 forms polar contacts with the P-loop Glu-33. (EPS) Figure S8 Expression of EhGa1 wt or EhGa1 S37C does not significantly alter trophozoite proliferation. (A) The cellular distribution of overexpressed FLAG-EhGa1 wt and FLAG-EhGa1 S37C were assessed by immunofluorescence with a Cy3 anti-FLAG conjugate. Both wild type and mutant EhGa1 exhibited similar diffusely cytoplasmic localizations following induced expression by treatment with 5 mg/mL tetracycline for 24 hr. Nuclei were stained with DAPI. (B) Trophozoites of the parent HM-1:IMSS, EhGa1 wt , and EhGa1 S37C strains were seeded in TYI medium with or without 5 mg/mL tetracycline and cell numbers assessed over 3 days. Cell viability was .90% at each measurement, as determined by trypan blue dye exclusion. No significant differences in growth were identified among the strains, although trophozoites induced to express EhGa1 S37C trended toward slower growth at day 3. Error bars represent standard error of the mean for three independent experiments. (EPS) Figure S9 E. histolytica transfected with empty vector is not affected by tetracycline treatment. HM-1:IMSS trophozoites were stably transfected with empty tetracyclineinducible expression vector. (A) Transwell migration and (B) Matrigel transmigration of parent strain and vector-transfected trophozoites did not differ significantly upon tetracycline treatment of 24 hours prior to the assay. Similarly, transfection with empty vector and tetracycline treatment had no significant effect on host cell attachment (C) or host cell killing (D). Error bars represent standard error of the mean for four independent experiments in panels A-C and three independent experiments in panel D. Statistical significance was tested using an unpaired, two-tailed Student's t-test. (EPS) Figure S10 Microscopic analysis of perturbed E. histolytica attachment to host cells upon overexpression of EhGa1 wt or EhGa1 S37C . (A) Trophozoites grown in the presence or absence of 5 mg/mL tetracycline were fluorescently labeled with CFDA and allowed to attach to fixed, confluent layers of CHO cells. Phase contrast (upper panels) and epifluorescence (lower panels) images were obtained of attached trophozoites. (B) Attachment was quantified by counting trophozoites in three microscopic fields (106). Overexpression of EhGa1 wt enhanced monolayer attachment, while expression of EhGa1 S37C reduced attachment. Parent strain HM-1:IMSS trophozoites were unaffected by tetracycline treatment and were indistinguishable from non-induced EhGa1 wt and EhGa1 S37C . Error bars represent standard error of the mean. * represents statistical significance by an unpaired, two-tailed Student's t-test (p,0.05) for three independent experiments. (EPS) Figure S11 RT-PCR analysis of differentially transcribed genes and altered expression of amoebapore A protein. (A) qRT-PCR amplification of RNA isolated from HM1 E. histolytica trophozoites confirmed differential transcription of EhGa1, EhGb1, amoebapore A, and a cysteine protease (EHI_006920) upon tetracycline treatment of the parent HM-1:IMSS, EhGa1 wt , or EhGa1 S37C strains over 24 hours. * indicates statistically significant difference from time zero (no tetracycline exposure), using an unpaired, two-tailed Student's t-test for two technical duplicates of two independent experiments. EhGa1 expression was significantly up-regulated in the EhGa1 wt and EhGa1 S37C strains, while EhGb1 was up-regulated and amoebpore A and cysteine protease (EHI_006920) were down-regulated upon expression of EhGa1 S37C . (B) Trophozoite lysates were subjected to western blotting with anti-amoebapore A (kind gift of M. Leippe, U. of Kiel, Germany), with actin serving as a loading control. Amoebapore A protein expression is reduced in parallel with its transcriptional downregulation upon overexpression of EhGa1 S37C . (EPS)

Supporting Information
Table S1 Data collection and refinement statistics for lysine-methylated selenomethionine EhGa1. (PDF)