Toxoplasma GRA7 effector increases turnover ofimmunity-related GTPases and contributes toacute virulence in the mouseAditi Alaganan, Sarah J. Fentress, Keliang Tang, Qiuling Wang, and L. David Sibley1

Department of Molecular Microbiology, Washington University School of Medicine, St. Louis, MO 63110

Edited by Ralph R. Isberg, Howard Hughes Medical Institute, Tufts University School of Medicine, Boston, MA, and approved December 9, 2013 (received forreview July 20, 2013)

The intracellular parasite Toxoplasma gondii enjoys a wide hostrange and is adept at surviving in both naive and activated macro-phages. Previous studies have emphasized the importance of theactive serine-threonine protein kinase rhoptry protein 18 (ROP18),which targets immunity-related GTPases (IRGs), in mediating mac-rophage survival and acute virulence of T. gondii in mice. Here, wedemonstrate that ROP18 exists in a complex with the pseudoki-nases rhoptry proteins 8 and 2 (ROP8/2) and dense granule protein7 (GRA7). Individual deletion mutant Δgra7 or Δrop18 was par-tially attenuated for virulence in mice, whereas the combinedΔgra7Δrop18 mutant was avirulent, suggesting these proteinsact together in the same pathway. The virulence defect of thedouble mutant was mirrored by increased recruitment of IRGsand clearance of the parasite in IFN-γ–activated macrophages invitro. GRA7 was shown to recognize a conserved feature of IRGs,binding directly to the active dimer of immunity-related GTPase a6in a GTP-dependent manner. Binding of GRA7 to immunity-relatedGTPase a6 led to enhanced polymerization, rapid turnover, andeventual disassembly. Collectively, these studies suggest thatROP18 and GRA7 act in a complex to target IRGs by distinct mech-anisms that are synergistic.

pathogenesis | innate immunity | cooperative polymerization

The apicomplexan parasite Toxoplasma gondii has a re-markable host range and is capable of infecting most

warm-blooded animals (1). The cellular life cycle of this op-portunistic pathogen involves active invasion of nucleatedhost cells and establishment of an intracellular niche withinthe parasitophorous vacuole. This compartment is demarcatedby the parasitophorous vacuole membrane (PVM), which avoidsfusion with the host endomembrane system, although beingstudded with many parasite proteins (2). PVM-localized proteinsderive from two major organelles: the rhoptries (ROP proteins)and the dense granules (GRA proteins), which are sequentiallysecreted upon invasion (3). The strategic location of GRAs andROPs on the PVM positions them to play important roles ininteracting with the host.ROP proteins are secreted directly into the host cell cytosol at

the time of invasion, after which they target to the PVM or otherlocations within the cell (4). Although many ROP proteinscontain a kinase fold, nearly half of these are predicted to bepseudokinases because they lack the critical catalytic residuesthat are normally required for phosphate transfer (5). Often,these pseudokinases (i.e., ROP5, ROP8/2, ROP4/7) exist astandem gene duplications and show evidence of positive selec-tion (5). Following secretion, ROP2 family members are targetedto the cytoplasmic face of the PVM via a series of amphipathicα-helical regions in their N termini (6, 7).ROP proteins generated renewed interest when it became

evident that some ROPs confer critical strain-specific virulencein mice (8–12). In particular, the active kinase ROP18 and thepseudokinase ROP5 defend the parasite vacuole by blocking thefunction of mouse immunity-related GTPases (IRGs) (4). Fol-lowing up-regulation by IFN-γ, IRGs target susceptible parasites

for destruction by relocating to the parasite vacuole and vesic-ulating the PVM (13). Under normal conditions, immunity-related GTPase family M (IRGM) proteins that are found on hostendomembranes bind IRGs and maintain them in an inactivestate (14). Upon infection, IRGs are preferentially recruited topathogen-containing vacuoles due to an absence of IRGMproteins (14). However, virulent strains of T. gondii that expressa type I allele of ROP18, an active serine/threonine kinase, areable to phosphorylate IRGs, and thus to decrease their re-cruitment to the PVM (15, 16). The activity of ROP18 is reg-ulated by a related pseudokinase ROP5 (17), which also bindsto monomeric Irga6 and blocks its assembly in vitro (18, 19).Recent evidence implicates ROP5 and ROP18 in avoidance ofguanylate binding proteins, which also contribute to restrictionof T. gondii in IFN-γ–activated macrophages (20).Kinases often interact with pseudokinases and scaffolding

proteins for regulation or substrate recruitment (21, 22). Con-sequently, we were interested in whether ROP18 forms a com-plex with other parasite proteins that might influence its activityor function. Our studies establish that ROP18 is part of a com-plex with the related pseudokinases ROP8/2 and the densegranule protein GRA7, the latter of which acts synergisticallywith ROP18 in controlling acute virulence in the mouse.

ResultsCharacterization of ROP18-Containing Complexes. To identify ROP18interacting partners, we immunoprecipitated ROP18 from cellsinfected with a transgenic type III strain parasite expressing activeROP18 (i.e., CTG + ROP18-Ty) vs. CTG, which lacks ROP18expression (9). TandemMS (MS/MS) analysis revealed that ROP18

Significance

Toxoplasma gondii is a widespread parasite of animals thatfrequently infects humans. Infection in mice, a natural host fortransmission, is strongly dependent on IFN-γ, which up-regu-lates innate defenses. The parasite secretory kinase rhoptryprotein 18 (ROP18) phosphorylates immunity-related GTPases(IRGs) and prevents recruitment to the vacuole surroundingvirulent parasites. Here, we demonstrate that ROP18 is foundin a complex with other pseudokinases (i.e. ROP8/2) and thedense granule protein GRA7. Our studies reveal that GRA7binds to oligomers of Irga6, resulting in increased polymeri-zation initially but then leading to rapid disassembly. Theactivity of GRA7 may prevent proper IRG loading onto theparasite-containing vacuole and may make substrates availablefor ROP18, thus explaining its contribution to virulence.

Author contributions: A.A., S.J.F., and L.D.S. designed research; A.A., S.J.F., and Q.W.performed research; K.T. contributed new reagents/analytic tools; A.A. and L.D.S. ana-lyzed data; and A.A. and L.D.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: sibley@wustl.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1313501111/-/DCSupplemental.

1126–1131 | PNAS | January 21, 2014 | vol. 111 | no. 3 www.pnas.org/cgi/doi/10.1073/pnas.1313501111

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

021

(detected by 26 unique peptides) was specifically associated with therhoptry pseudokinase ROP2 (detected by 15 unique peptides) andGRA7 (detected by 11 unique peptides) (Table S1). A majority ofpeptides matched ROP2, whereas four peptides also matchedROP8, a closely related protein that is located immediately up-stream in the same locus, referred to here as ROP8/2 (Fig. S1). Twopeptides of ROP4/7 were also identified in the ROP18-Ty immu-noprecipitation (IP), which may reflect the previously describedinteraction between GRA7 and ROP4/7 (23). To validate theROP18 binding partners identified by MS/MS, GRA7 was immu-noprecipitated from WT RH strain vs. Δgra7 mutant parasites un-der similar conditions and analyzed by MS/MS. This reciprocal IPrevealed an almost identical set of proteins comprising the baitprotein GRA7 (14 peptides), ROP2 (14 peptides), ROP4 (8 pep-tides), ROP7 (5 peptides), ROP18 (4 peptides), and ROP5 (19peptides) (Table S2). These proteins were enriched in the IPsamples from WT parasites but not in those from Δgra7 mutantparasites (Table S2). Although ROP5 was not seen in the ROP18-Ty IP, its presence in the GRA7 complex, together with ROP18, isconsistent with the observation that ROP5 regulates the kinaseactivity of ROP18 (17). To confirm the interaction shown by MS, weimmunoprecipitated ROP18 from CTG + ROP18-Ty parasites vs.CTG parasites and performed Western blotting to detect interact-ing partners. ROP2 and GRA7 were specifically coprecipitated withROP18 (Fig. 1A), whereas GRA5, another secretory protein thatalso localizes to the PVM, did not associate with ROP18 (Fig. 1A),demonstrating the specificity of this complex.

ROP18 and GRA7 Localize to the Cytosolic Face of the PVM. Theidentification of GRA7 as part of a complex with ROP18 sug-gested that it might localize to the external surface of the PVMafter secretion, similar to ROP18 (6). To determine whetherGRA7 also localizes to this interface, digitonin was used topermeabilize infected host cells selectively. Under these semi-permeabilized conditions, antibodies to both ROP18 and GRA7stained the cytoplasmic surface of the PVM at both 30 min and12 h, whereas staining of the parasite surface antigen (SAG1)remained negative (Fig. 1B, Upper). In contrast, when Triton

X-100 was used to permeabilize all membranes fully, positivestaining for SAG1 was observed, as well as staining for GRA7 indense granules and ROP18-Ty in rhoptries (Fig. 1B, Lower; bothproteins are extracted from the PVM under these conditions).Combined with the MS/MS data above, these findings supportthe existence of a complex consisting of ROP18, GRA7, andROP8/2 on the surface of the PVM in infected cells.

Analysis of ROP18 Complex Deletion Mutants. To explore thefunctions of GRA7 and ROP8/2, we made single deletions ofeach gene by homologous recombination (Fig. S1 and TableS3). Furthermore, ROP8/2 and ROP18 were each separatelydeleted in the Δgra7 parasite line to create two double-deletionmutants (Fig. S2). Drug-resistant parasite pools were clonedby limiting dilution and screened by PCR (Figs. S1 and S2 andTable S3) and Western blotting (Fig. 1C) for their respectivegene deletions. The Δgra7Δrop18 parasites were then com-plemented with either GRA7 or ROP18-Ty, and single-cellclones were confirmed for proper protein expression and lo-calization (Fig. S3).GRA7 was previously implicated in nutrient acquisition (24),

so we were interested in whether the Δgra7 parasites would ex-hibit any growth defect in vitro. Plaque assays were performed onthe WT parasites and various mutant parasites to test whetherdeletion caused a growth defect. At 7 d postinfection, the numberof plaques was comparable for WT and all mutant lines (Fig. 1Dand Fig. S4A). Relative plaque size measurements and replicationassays further confirmed the absence of an in vitro growth defectof the Δgra7 mutant (Fig. S4 B and C). We also analyzed the ul-trastructure of WT and Δgra7Δrop18 parasite vacuoles by EM andobserved no differences in either the presence of the tubulove-sicular network (TVN) or the general architecture of the network(Fig. S5). In addition, we observed normal localization of bothROP18 and ROP5 on Δgra7 vacuoles (Fig. S6 A and B). Hence,the defects described below for the Δgra7Δrop18 parasites arenot related to adverse effects on growth or differences invacuole morphology or composition.

Fig. 1. Interaction of ROP18 and GRA7 and generation of KO and complemented lines. (A) Co-IP of proteins with ROP18. Lysates from CTG or CTG + ROP18-Ty parasites were immunoprecipitated with mAb BB2 to Ty (Mo αTy). Western blotting (WB) of the pellet (Pell, 100% loaded) and supernatant (Sup, 25%loaded) fractions was probed for rabbit anti-ROP18 (Rb αROP18), rabbit anti-ROP2 (Rb αROP2), rabbit anti-GRA7 (Rb αGRA7), or mAb to GRA5 (Mo αGRA5). (B)Immunofluorescence localization of ROP18 and GRA7 in selectively permeabilized (digitonin) vs. fully permeabilized (Triton X-100) cells. Permeabilized cellswere incubated with rabbit anti-GRA7 (red), mAb BB2 to Ty tag to detect ROP18-Ty (blue), and mAb DG52 to SAG1 (green). (Scale bars: 5 μm.) (C) Lysates fromWT or mutant parasites were resolved by SDS/PAGE and probed with Rb αROP18 and rabbit anti-T. gondii actin (Rb αTgACT) as a loading control (Upper) orwith Rb αROP2 and Rb αGRA7 (Lower). (D) Plaque assay of in vitro growth of the WT and mutant parasites.

Alaganan et al. PNAS | January 21, 2014 | vol. 111 | no. 3 | 1127

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

021

Combined Deletion of ROP18 and GRA7 Leads to Severe Attenuationof Acute Virulence. Because RO18 plays an important role in thevirulence of T. gondii in mice (15–17), we were interested inwhether the ROP8/2 and GRA7 proteins might also contributeto this phenotype. When challenged with Δrop8/2 parasites,outbred CD-1 mice died with almost identical kinetics to WTparasites (Fig. 2A). In comparison, CD-1 mice infected withΔgra7 and Δgra7Δrop8/2 parasites experienced a slight delay intime to death that was not as severe as that seen with Δrop18parasites (Fig. 2A). In contrast, all CD-1 mice challenged withthe Δgra7Δrop18 parasites survived (Fig. 2A). To establish the ex-tent of this attenuation, we examined the dose-dependent mortalityof mice following challenge with Δgra7Δrop18 parasites. Althougha dose of 103 parasites led to survival of CD-1 mice, higher dosesresulted in a dose-dependentmortality (Fig. 2B). Complementationof theΔgra7Δrop18 parasites with eitherGRA7 (Δrop18/GRA7) orROP18-Ty (Δgra7/ROP18-Ty) restored virulence to the approxi-mate levels of the single mutants (Fig. 2C).To determine the importance of immunity in the attenuation

of the double mutant, we injected the Δgra7Δrop18 parasites intoIFN-γ receptor (IFN-γR)−/− mice, which are highly susceptible toΔrop18 and Δrop5 mutants of T. gondii (17). Mutant Δgra7Δrop18parasites killed IFN-γR−/− mice within 10 d postinfection butremained avirulent in control C57BL/6 mice (Fig. 2D). Thisfinding indicates that the attenuation of Δgra7Δrop18 parasites isa consequence of immunological control and not due to intrinsicdifferences in growth.

Δgra7Δrop18 Parasites Show Increased IRG Recruitment and AreCleared by Activated Macrophages. The observation that the at-tenuation of Δgra7Δrop18 parasites relies on IFN-γ signalingsuggested that the double mutant may be more susceptible to theIRG pathway, which is a known target of ROP18 (15, 16).Consistent with this, we observed a significant increase in theextent of Irga6 (Fig. 3 A and B) and Irgb6 (Fig. 3 C and D) lo-calization to Δgra7Δrop18 parasite vacuoles in IFN-γ–activatedmacrophages. When complemented with GRA7 or ROP18,levels of Irga6 and Irgb6 recruitment to the vacuole were re-stored to WT levels (Fig. 3 A–D).Based on the increased levels of IRG recruitment, we hy-

pothesized that the double-mutant parasites might also be cleared

more effectively in IFN-γ–activated macrophages. AlthoughΔrop18 parasites were more efficiently cleared than WT par-asites at 20 h postinfection, the survival of Δgra7 mutants wasunchanged (Fig. 3E). Consistent with the higher level of IRGrecruitment to vacuoles surrounding the double mutant, sig-nificantly fewer Δgra7Δrop18 parasites survived compared withthe single Δrop18 mutant (Fig. 3E). Complementation with eitherGRA7 or ROP18 restored the survival defect seen in the doublemutant (Fig. 3E).The enhanced recruitment and clearance of the double mutant

suggested that ROP18 and GRA7 act together to block the IRGpathway, albeit with ROP18 having a more dominant role. Totest this prediction, we examined the survival of Δgra7 andΔgra7Δrop18 parasites in Irgm3−/− mice, which are defective inloading of IRGs onto the PVM (25). Although we observeda slight delay in time to death of outbred CD-1 mice challengedwith Δgra7 mutants (Fig. 2A), this effect was eliminated in themore susceptible C57BL/6 strain and in Irgm3−/− mice (Fig. 3F).Consistent with the idea that GRA7 and ROP18 act synergisti-cally to avoid IRG-mediated killing, the normally attenuatedΔgra7Δrop18 mutant killed Irgm3−/− mice within 10 d (Fig. 3F).

GRA7 Binds to and Stimulates Rapid Turnover of the GTP-ActivatedIrga6 Oligomer in Vitro. Increased recruitment of IRGs to thevacuole surrounding Δgra7Δrop18 parasites, as well as thecomplete reversal of the Δgra7Δrop18 avirulence in Irgm3−/−mice, led us to test whether GRA7 directly affects IRG function.Irga6 oligomerization was monitored by 90° light scattering, asdescribed previously (26). Addition of GTP to Irga6 alone causedan increase in light scattering consistent with polymerization (Fig.4A). Addition of substochiometric amounts of GRA7 resulted inan increase in light scattering (Fig. 4A), suggesting it promotesoligomerization and/or cross-links and bundles Irga6. Sedimenta-tion assays revealed more Irga6 in the pellet fraction at the peak oflight scattering when GRA7 was present compared with Irga6alone (Fig. S7A). The subsequent decrease in light scattering,which corresponds to depolymerization (26), occurred much fasterand dropped to a much lower level in the presence of GRA7 (Fig.4A and Fig. S7A). Altered kinetics of polymerization and turnoverof Irga6 showed a dose-dependent response with addition ofGRA7 (Fig. 4B).To determine the mechanism of this disruption, we examined

the interaction of His6-tagged GRA7 with recombinant Irga6.GRA7 coprecipitated recombinant Irga6 in the presence of GTPbut not GDP (Fig. 4C), suggesting that it recognizes GTP-bounddimers and/or oligomers but not GDP-bound monomers of Irga6(26, 27). Precipitation was not due to formation of large olig-omers or aggregates of Irga6 and GRA7, as shown by the failureof these proteins to pellet in the absence of Ni beads (Fig. S7B).The interaction between GRA7 and Irga6 was also not the resultof nonspecific binding to Ni beads, because an untagged versionof GRA7 did not bring down Irga6 in the presence of beads;however, it was still able to alter the polymerization of Irga6(Fig. S7B). Moreover, His6-tagged aldolase, an unrelated gly-colytic enzyme, did not precipitate Irga6 (Fig. 4C). In addition,a GRA7 mutant (GRA7 MUT) that was modified to eliminatethe hydrophobic transmembrane domain (Fig. S7C) was alsoable to precipitate Irga6 efficiently (Fig. 4C), suggesting thatthe interaction does not depend on the putative transmem-brane domain of GRA7. To test the specificity of the GRA7–Irga6 interaction further, we took advantage of previous studiesthat have defined the importance of GTP hydrolysis on theformation of Irga6 oligomers. Irga6 binds to 2′dGTP and 3′dGTP but only oligomerizes with 2′dGTP (27). GRA7 bound toIrga6 in the presence of 2′dGTP but not 3′dGTP (Fig. 4D),indicating that GRA7 binds to oligomers of Irga6. In addition,two Irga6 point mutants, T102A and T108A, which are in-capable of hydrolyzing GTP or oligomerizing (16), were alsounable to bind to GRA7 (Fig. 4D). In accordance with the in-crease in light scattering of Irga6 in the presence of GRA7,addition of GRA7 stimulated higher phosphate release from theGTPase activity by Irga6 compared with Irga6 alone (Fig. 4E).

Fig. 2. Virulence of WT and mutant parasites in laboratory mice. (A) WTand deletion mutants were injected into CD-1 mice (100 parasites, s.c. in-oculation), and survival was followed for 30 d (n = 10 mice for all strainsexcept Δrop8/2 and Δgra7Δrop8/2, where n = 5 mice). ****P < 0.0001,Gehan–Breslow Wilcoxon test. (B) Dose-dependent mortality of Δgra7Δrop18parasites. CD-1 mice (n = 5) were injected i.p with varying doses of parasites,and survival was followed for 30 d. (C) Δgra7Δrop18 parasites and theirsingly complemented strains (i.e., Δrop18/GRA7, Δgra7/ROP18-Ty) wereinjected into C57BL/6 mice (100 parasites, s.c. inoculation), and survival wasmonitored for 30 d (n = 10 mice for all strains). ****P < 0.0001, Gehan–Breslow Wilcoxon test. (D) Survival of WT C57BL/6 mice vs. IFN-γR−/− micechallenged with Δgra7Δrop18 parasites. Mice (n = 10) were injected i.p. with100 Δgra7Δrop18 parasites, and survival was followed for 30 d.

1128 | www.pnas.org/cgi/doi/10.1073/pnas.1313501111 Alaganan et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

021

DiscussionOur findings reveal that ROP18 exists in a protein complex onthe PVM with the parasite pseudokinases ROP8/2 and GRA7.Previous co-IPs also identified an interaction between thepseudokinases ROP4/7, ROP8/2, and GRA7 and demonstratedthat these proteins become phosphorylated within the host cell(23). Deletion of either GRA7 or ROP8/2 alone did not decreasevirulence, unlike the loss of ROP18, suggesting that GRA7 andROP8/2 are not involved in regulating ROP18 function. How-ever, the combined loss of ROP18 and GRA7 led to a muchgreater attenuation than the loss of either gene alone. Double-Δgra7Δrop18 mutants showed increased recruitment of IRGs andclearance in IFN-γ–activated macrophages. The avirulencephenotype of this double mutant was reverted in Irgm3−/− mice,implicating the complex in avoidance of the IRG pathway.GRA7 bound to Irga6-GTP in vitro, enhancing the initial rate ofpolymerization but eventually leading to a rapid disruption of theoligomeric complex. These findings suggest a unique role forGRA7 in combating cell-autonomous defenses by altering IRGpolymerization kinetics.Despite a previously suggested role of GRA7 in nutrient ac-

quisition (24), we did not observe any defect in growth of theΔgra7 mutant under conditions examined here. Moreover, theΔgra7 mutant showed a normal PVM and TVN structure andnormal distribution of the virulence factors ROP18 and ROP5on the PVM. Loss of GRA7 did not directly affect IRG loadingor clearance in IFN-γ–activated macrophages, unlike deletion ofGRA2, which leads to alterations in the TVN and increased IRGrecruitment (18). Deletion of GRA7 did lead to a slight delay in

time to death in CD-1 outbred mice, although not in more sus-ceptible C57BL/6 mice, suggesting there is no intrinsic growthdefect of the Δgra7 mutant. Although loss of GRA7 alone hada minimal effect on survival, when combined with deletion ofROP18, it led to enhanced IRG recruitment and clearance inIFN-γ–activated macrophages in vitro and a profound loss ofvirulence in both CD-1 and C57BL/6 mice. The dramatic changein virulence is unlikely to be due to defects in growth in vivo,because the double-Δgra7Δrop18 mutant showed normal growthin vitro. Importantly, the decreased virulence of the Δgra7Δrop18mutant was reversed in IFN-γR−/− mice, leading to a similar timeto death as that observed previously with WT parasites (17). Thedecreased virulence of the Δgra7Δrop18mutant was also reversedin Irgm3−/− mice, implicating IRGs in this control mechanism.The IRG proteins form the major IFN-γ–dependent resistancemechanism to T. gondii in the mouse (13). Collectively, thesefindings indicate that GRA7 and ROP18 act synergistically toblock the IRGs and that in the absence of this host defensepathway, they are no longer required.GRA7 dramatically increased polymerization of Irga6 in vitro,

as revealed by 90° light scattering and sedimentation. The accel-erated rate of Irga6 assembly suggests that GRA7 may act asa nucleation or stabilizing factor for polymer formation. Alterna-tively, GRA7 may bind to small oligomers of Irga6 and cross-linkor bundle them, leading to greater light scattering and sedi-mentation. Intriguingly, GRA7 only bound to forms of Irga6 thatwere competent for oligomerization. Structural and modelingstudies have defined a dimer interface in Irga6 that is necessaryfor homo-oligomerization and mediates binding to heterodimer

Fig. 3. IRG recruitment and clearance in activated macrophages. (A) Immunofluorescence staining of Irga6 on parasite-containing vacuoles in activated RAWmacrophages at 30 min postinfection stained with mAb 10D7 for Irga6 (green), rabbit polyclonal antibody to ROP5 (red), and DAPI (blue). (Scale bar: 5 μm.) (B)Quantification of Irga6 localized to the parasite vacuole in IFN-γ–activated RAW cells. Mean ± SD of the population (SDP) (n = 3 experiments). *P < 0.05 and**P < 0.001, Student t test. (C) Immunofluorescence staining of Irgb6 on parasite-containing vacuoles in IFN-γ–activated RAW macrophages at 30 minpostinfection stained with rabbit polyclonal α-Irgb6 (green), mAb Tg17-113 to GRA5 (red), and DAPI (blue). (Scale bar: 5 μm.) (D) Quantification of Irgb6localized to the parasite vacuole in activated RAW cells. Mean ± SDP (n = 3 experiments). *P < 0.05, Student t test. (E) Quantification of parasite survival inIFN-γ–activated RAW macrophages. Mean ± SD values for two experiments with three replicates each are given (n = 6). *P < 0.05 and **P < 0.001, Mann–Whitney test. (F) Survival of Irgm3−/− vs. WT C57BL/6 mice injected s.c. with 100 parasites (n = 7 mice for Δrop18 and Δgra7Δrop18, n = 5 mice for Δgra7).****P < 0.0001, Gehan–Breslow Wilcoxon test.

Alaganan et al. PNAS | January 21, 2014 | vol. 111 | no. 3 | 1129

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

021

formation with IRGM proteins that negatively regulate polymerassembly (27). GRA7 was unable to bind to mutants (T102 andT108) that fail to hydrolyze GTP, and hence do not undergooligomerization (16). Similarly, GRA7 was only able to bind toIrga6 in the presence of 2′dGTP, which induces polymerizationmore slowly than GTP, whereas it did not bind to Irga6 in thepresence of 3′dGTP, which fails to drive oligomerization (27).

These results suggest that GRA7 may recognize the active dimerthat leads to IRG assembly, thereby taking advantage of a naturalregulatory feature to alter polymerization kinetics.The function of GRA7 contrasts with that of known mediators

of IRG function or GTPase effectors. IRGs exist as monomerswhen bound to GDP, and they oligomerize in the presence ofGTP to undergo hydrolysis by a cooperative mechanism (26).IRGs have intrinsic GTPase activating protein (GAP) activityand exchange nucleotides readily; as such, they do not rely onGAPs or guanine nucleotide exchange factors (GEFs) to regu-late their functions (26). Instead, IRGM proteins are thought toregulate IRGs by preventing their assembly through guanylatedissociation inhibitor activity that maintains them in a GDP-bound state (28, 29). Because GRA7 binds only to oligomers ofIrga6-GTP and not to Irga6-GDP monomers, it is unlikely to actas a GEF. Binding of GRA7 to Irga6 oligomers slightly enhancedGTP hydrolysis and phosphate release, consistent with the in-creased amount of polymer formed. In addition, GRA7 greatlyincreased the rate of depolymerization of Irga6, resulting infaster turnover. The precise interactions between GRA7 andIrga6 that drive increased assembly and faster turnover remain tobe defined by future studies.The in vivo significance of enhanced Irga6 polymerization in

the presence of GRA7 is presently unclear, although severalmodels can be considered based on our in vitro findings. IfGRA7 on the PVM were recognized by the host as a target forIRGs, we would expect to see higher recruitment to vacuolessurrounding WT parasites. This model can be ruled out becausethere was little change in IRG recruitment in the Δgra7 mutant,whereas the Δgra7Δrop18 mutant showed much greater re-cruitment of Irga6 and Irgb6. Instead, it seems likely that GRA7disrupts IRG function by altering the kinetics of polymerization,which might occur in several different ways. First, binding ofGRA7 to small oligomers or dimers may prevent them fromassembling correctly on the PVM. Second, GRA7 binding toIRGs may lead to enhanced assembly but then drive rapid dis-assembly into inactive GDP-bound monomers. Such activity byGRA7 would be expected to make substrates available toROP18, which phosphorylates residues in switch region I (SWI)(15, 16), or to ROP5, which is thought to bind to monomericIRGs to prevent their assembly (19). Further studies are neededto define how these various virulence factors work cooperativelyin vitro and in vivo to thwart the IRG system and protect theparasite from destruction.IRGs target a variety of intracellular pathogens, including

T. gondii, where they oligomerize onto the PVM and result in itsvesiculation and destruction (13). The virulence factor ROP18phosphorylates key residues in SWI of IRGs, and this is associ-ated with preventing their accumulation on the PVM (15, 16).The function of ROP18 is augmented by the pseudokinaseROP5, which can directly bind to IRGs and prevent assembly invitro (18, 19), while also enhancing the activity of ROP18 (17).Our findings suggest that GRA7 acts in vitro to disrupt IRGs bya unique biochemical mechanism of increasing the turnover ofIrga6 oligomers, thus potentially complementing the functions ofother known virulence determinants ROP18 and ROP5. Thisexample reveals multiple layers of defense used by the parasite tocounteract innate immunity in the host.

Materials and MethodsParasite Propagation. Parasite strains (referred to in detail in Table S3) werepassaged as tachyzoites in human foreskin fibroblast (HFF) monolayers in DMEMcontaining 10 mM Hepes, 10 μg/mL gentamicin, 6 mM L-glutamine, 1 mM py-ruvate, and 10% (vol/vol) FBS, cultured at 37 °C in 5% (vol/vol) CO2. Plaque assayswere carried out on confluent monolayers of HFF cells as described in SI Materialsand Methods.

Antibodies and Microscopy. Rabbit antisera that were specific to recombinantGRA7 or ROP2 were generated as described in SI Materials and Methods.Immunoprecipitation and Western blotting assays were performed as de-scribed previously (15) (SI Materials and Methods). Immunofluorescencemicroscopy to localize parasite proteins in infected cells and EM to evaluatethe morphology of infected cells were performed as described previously

Fig. 4. Interaction of GRA7 and Irga6 in vitro. (A) Oligomerization of 50 μMIrg6 ± 20 μM GRA7 in a 90° light scattering assay, in which 1 mM GTP wasadded at the location indicated by the arrow at 37 °C. AU, arbitrary units; PK,proteinase K-treated GRA7. (B) Oligomerization of 50 μM Irg6 ± GRA7 ina 90° light scattering assay in which 1 mM GTP was added at the locationindicated by the arrow at 37 °C. (C) GTP-dependent binding of WT (GRA7) vs.double-point mutant (GRA7 MUT) to Irga6. Supernatants (S; 10%) and pel-lets (P; 50%) were resolved by SDS/PAGE and stained with SYPRO Ruby. His-tagged ALD (His-Ald) was included as a negative control. (D) Nucleotide-dependent binding of GRA7 to Irga6. Recombinant His-tagged GRA7 (His-GRA7) was incubated with WT or mutant Irga6 and various nucleotides for 10min at 37 °C and then immunoprecipitated with Ni beads. The supernatantswere resolved by SDS/PAGE and stained with SYPRO Ruby. (E) Phosphate re-lease during oligomerization of Irga6 (50 μM); incubation ± 20 μM GRA7,20 μMALD (Ald), or PK-treated GRA7 (+PK); and addition of final 1 mM GTP(+GTP). [Pi], cumulative phosphate released (micromolar) (n = 3). **** P <0.0001, two-way ANOVA test.

1130 | www.pnas.org/cgi/doi/10.1073/pnas.1313501111 Alaganan et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

021

(15) (SI Materials and Methods). Immunofluorescence microscopy to evalu-ate the recruitment of Irga6 and Irgb6 to parasite-containing vacuoles andthe clearance of parasites in IFN-γ–activated RAW cells was performed asdescribed previously (15) (SI Materials and Methods).

MS. Proteins were eluted from protein G Sepharose beads by incubation in2% RapiGest SF surfactant (Waters) with 20 mM DTT at 95 °C for 10 min,followed by 5 min at room temperature. Eluates were collected by centri-fugation at 1,000 × g and digested with 2 μg of trypsin for 16 h at 37 °C.Digested samples were dried and resuspended in 10 mL of 5% (vol/vol)acetonitrile in 0.1% formic acid and analyzed by liquid chromatography MS/MS using the LTQ-Orbitrap Velos (Thermo Scientific) with a 2-h gradient. Them/z spectra data were searched using MASCOT (Matrix Science) against theGenBank nr (nonredundant) and ToxoDB 8.1 databases.

Generation of Plasmids and Transgenic Parasites. Primers used for generationof plasmids for KOor complemented strains used in this study are available onrequest. To generate the KO and complemented lines, plasmids were line-arized and transfected into the WT parasites, and single-cell clones wereisolated (Table S3) as described in SI Materials and Methods.

In Vivo Infection. CD-1, IFN-γR−/−, and C57BL/6 mice were obtained froma commercial vendor (Charles River Laboratories) and housed in an AmericanAssociation for Assessment and Accreditation of Laboratory Animal Care-approved facility. The Irgm3−/− mice were obtained from Greg Taylor (DukeUniversity, Durham, NC) (30) and bred locally. Female mice between 8 and 10wk of age were challenged with freshly lysed tachyzoites and monitored for30 d postinfection, as described previously (15). Infections were done ingroups of five mice per strain and repeated two (n = 10) or more times.Animal studies were approved by the Institutional Animal Studies Commit-tee (School of Medicine, Washington University in St. Louis).

Irga6 Oligomerization Assays. Recombinant Irgs6 was expressed as GST-fusionprotein, cleaved, and purified as described in SI Materials and Methods.Oligomerization of Irga6 was monitored by 90° light scattering in a totalreaction volume of 150 μL containing 40 mM Tris·Cl (pH 8.0), 5 mM MgCl2,and 2 mM DTT (buffer B). Polymerization of 50 μM Irga6 alone was com-pared with that of Irga6 combined with different concentrations of GRA7 orproteinase K-treated GRA7 (SI Materials and Methods). Proteinase K-treatedbuffer A [40 mM Tris·Cl (pH 8.0), 15 mM NaCl] was added to the Irga6-onlyreaction as a control. The protein solutions were first equilibrated to 37 °C,

followed by the addition of GTP (final concentration of 1 mM), and moni-tored over time with a PTI Quantmaster spectrofluorometer (Photon Tech-nology International). Curves were processed using Prism (GraphPad).

In Vitro Binding Assay. Recombinant GRA7 and T. gondii aldolase (ALD) wereexpressed in Escherichia coli and purified as described in SI Materials andMethods. Ni beads (Sigma–Aldrich) were either chargedwithHis-taggedGRA7or His-tagged ALD or left uncharged in buffer A. Irga6 (50 μM) was added tothe Ni beads in a final volume of 50 μL of buffer B and various nucleotides(final concentration of 1 mM) and incubated at 37 °C for 10 min. Beads werewashed in buffer B, and 10% (vol/vol) of the unbound supernatant fractionand 50% (vol/vol) of the bound-bead fraction were resolved on 10% (vol/vol)acyrlamide gels by SDS/PAGE and stained with SYPRO Ruby (Invitrogen).

GTP Hydrolysis Assays. To measure phosphate release, 50 μM purified Irga6was incubated with either 20 μM GRA7, 20 μM proteinase K-treated GRA7,20 μM ALD, or proteinase K-treated buffer A in the presence of 1 mM GTP inbuffer B at 37 °C. As negative controls, Irga6 and GRA7 without GTP andGRA7 alone were incubated similarly. At time intervals, reactions weresampled and phosphate release was measured using a GTPase colorimetrickit (Innova Biosciences). Pi release from incubation of GTP alone was sub-tracted from the samples containing GTP.

Statistics. Student t tests were performed under the assumption of equalvariance and with a two-tailed test, where P ≤ 0.05 was considered signifi-cant. Graphs were plotted with mean ± SD of the population or SD as noted.Scatter plots were analyzed using the Mann–Whitney test for statisticalsignificance under nonparametric conditions not assuming Gaussian distri-bution and a two-tailed test, where P ≤ 0.05 was considered significant.Kaplan–Meier plots that were analyzed using the Gehan–Breslow Wilcoxontest, where P ≤ 0.05 was considered significant. For the phosphate releaseassay, curves were analyzed by a two-way ANOVA test, where P ≤ 0.05.

ACKNOWLEDGMENTS. We thank Vern Carruthers and Mae Huynh (Univer-sity of Michigan) for the Δku80 strain, Greg Taylor (Duke University) forIrgm3−/− mice and antibodies, and members of the Sibley laboratory forhelpful comments. Sophie Alvarez (Danforth Plant Sciences Center) per-formed the MS studies. Wandy L. Beatty (Microbiology Imaging Facility,Washington University in St. Louis) performed the EM studies. This studywas supported by the National Institutes of Health (Grants AI34036 andAI36629).

1. Dubey JP (2010) Toxoplasmosis of Animals and Humans (CRC, Boca Raton, FL).2. Sibley LD (2004) Invasion strategies of intracellular parasites. Science 304:248–253.3. Carruthers VB, Sibley LD (1997) Sequential protein secretion from three distinct or-

ganelles of Toxoplasma gondii accompanies invasion of human fibroblasts. Eur J CellBiol 73(2):114–123.

4. Hunter CA, Sibley LD (2012) Modulation of innate immunity by Toxoplasma gondiivirulence effectors. Nat Rev Microbiol 10(11):766–778.

5. Peixoto L, et al. (2010) Integrative genomic approaches highlight a family of parasite-specific kinases that regulate host responses. Cell Host Microbe 8(2):208–218.

6. Fentress SJ, Steinfeldt T, Howard JC, Sibley LD (2012) The arginine-rich N-terminaldomain of ROP18 is necessary for vacuole targeting and virulence of Toxoplasmagondii. Cell Microbiol 14(12):1921–1933.

7. Reese ML, Boothroyd JC (2009) A helical membrane-binding domain targets theToxoplasma ROP2 family to the parasitophorous vacuole. Traffic 10(10):1458–1470.

8. Saeij JPJ, et al. (2007) Toxoplasma co-opts host gene expression by injection ofa polymorphic kinase homologue. Nature 445(7125):324–327.

9. Taylor S, et al. (2006) A secreted serine-threonine kinase determines virulence in theeukaryotic pathogen Toxoplasma gondii. Science 314(5806):1776–1780.

10. Saeij JPJ, et al. (2006) Polymorphic secreted kinases are key virulence factors intoxoplasmosis. Science 314(5806):1780–1783.

11. Reese ML, Zeiner GM, Saeij JP, Boothroyd JC, Boyle JP (2011) Polymorphic family ofinjected pseudokinases is paramount in Toxoplasma virulence. Proc Natl Acad Sci USA108(23):9625–9630.

12. Behnke MS, et al. (2011) Virulence differences in Toxoplasma mediated by amplifi-cation of a family of polymorphic pseudokinases. Proc Natl Acad Sci USA 108(23):9631–9636.

13. Howard JC, Hunn JP, Steinfeldt T (2011) The IRG protein-based resistance mechanismin mice and its relation to virulence in Toxoplasma gondii. Curr Opin Microbiol 14(4):414–421.

14. Haldar AK, et al. (2013) IRG and GBP host resistance factors target aberrant, “non-self” vacuoles characterized by the missing of “self” IRGM proteins. PLoS Pathog 9(6):e1003414.

15. Fentress SJ, et al. (2010) Phosphorylation of immunity-related GTPases by a parasitesecretory kinase promotes macrophage survival and virulence. Cell Host Microbe 16:484–495.

16. Steinfeldt T, et al. (2010) Phosphorylation of mouse immunity-related GTPase (IRG)resistance proteins is an evasion strategy for virulent Toxoplasma gondii. PLoS Biol8(12):e1000576.

17. Behnke MS, et al. (2012) The polymorphic pseudokinase ROP5 controls virulence inToxoplasma gondii by regulating the active kinase ROP18. PLoS Pathog 8(11):e1002992.

18. Niedelman W, et al. (2012) The rhoptry proteins ROP18 and ROP5 mediate Toxo-plasma gondii evasion of the murine, but not the human, interferon-gamma re-sponse. PLoS Pathog 8(6):e1002784.

19. Fleckenstein MC, et al. (2012) A Toxoplasma gondii pseudokinase inhibits host IRGresistance proteins. PLoS Biol 10(7):e1001358.

20. Selleck EM, et al. (2013) Guanylate-binding protein 1 (Gbp1) contributes to cell-au-tonomous immunity against Toxoplasma gondii. PLoS Pathog 9(4):e1003320.

21. Boudeau J, Miranda-Saavedra D, Barton GJ, Alessi DR (2006) Emerging roles ofpseudokinases. Trends Cell Biol 16(9):443–452.

22. Zeqiraj E, van Aalten DM (2010) Pseudokinases-remnants of evolution or key allostericregulators? Curr Opin Struct Biol 20(6):772–781.

23. Dunn JD, Ravindran S, Kim SK, Boothroyd JC (2008) The Toxoplasma gondii densegranule protein GRA7 is phosphorylated upon invasion and forms an unexpected as-sociation with the rhoptry proteins ROP2 and ROP4. Infect Immun 76(12):5853–5861.

24. Coppens I, et al. (2006) Toxoplasma gondii sequesters lysosomes from mammalianhosts in the vacuolar space. Cell 125(2):261–274.

25. Henry SC, et al. (2009) Balance of Irgm protein activities determines IFN-gamma-in-duced host defense. J Leukoc Biol 85(5):877–885.

26. Uthaiah RC, Praefcke GJ, Howard JC, Herrmann C (2003) IIGP1, an interferon-gamma-inducible 47-kDa GTPase of the mouse, showing cooperative enzymatic activity andGTP-dependent multimerization. J Biol Chem 278(31):29336–29343.

27. Pawlowski N, et al. (2011) The activation mechanism of Irga6, an interferon-inducibleGTPase contributing to mouse resistance against Toxoplasma gondii. BMC Biol 9:7.

28. Hunn JP, et al. (2008) Regulatory interactions between IRG resistance GTPases in thecellular response to Toxoplasma gondii. EMBO J 27(19):2495–2509.

29. Papic N, Hunn JP, Pawlowski N, Zerrahn J, Howard JC (2008) Inactive and active statesof the interferon-inducible resistance GTPase, Irga6, in vivo. J Biol Chem 283(46):32143–32151.

30. Taylor GA, et al. (2000) Pathogen-specific loss of host resistance in mice lacking theIFN-γ-inducible gene IGTP. Proc Natl Acad Sci USA 97(2):751–755.

Alaganan et al. PNAS | January 21, 2014 | vol. 111 | no. 3 | 1131

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Janu

ary

24, 2

021