Ethics statement

KM white mice were purchased from the Laboratory Animal Center of Southern Medical University. Mice were housed in the facility at the School of Public Health and Tropical Medicine according to the guidelines for laboratory animals approved by Guangdong Laboratory Animals Monitoring Institute. This research does not involve human participants, and it was approved by the Institutional Ethics Review Board of Southern Medical University.

Plasmids construction and site mutation

Animals, toxoplasma gondii strains and cell lines

Pathogens free (SPF) KM mice were bought from the Animal Institute of Southern Medical University (Guangdong Province, China). T. gondii RH and Pru strain were generous gift from Dr. Xi-Mei Zhan in the School of Medicine of Sun Yat-sen University. The COS-7 cell line was purchased from ATCC and the human bronchial epithelial (16-HBE) cell line was purchased from Shanghai Fuxiang Biotechnology Limited Company. Each cell line was grown in DMEM (Gibco) containing 10% (v/v) NCS (New born calf serum, Gibco) at 5% CO₂ and 37°C. For fluorescence microscopy and T. gondii infection rate counting experiments, COS-7 cells were grown on coverslips in the wells of 6-well plates (Corning). 16-HBE cells were used for RNAi and endogenous RhoA and Rac1 immunofluorescence experiments.

Toxoplasma gondii infection

Accumulation of Rho and Rac GTPases on the PVM

IRGs and Arf6 are members of large and small GTPase families, respectively, which accumulate on the PVM of T. gondii infected cells and play important roles during host cell invasion [14, 15]. However, the presence of these two GTPases is insufficient to explain the whole spectrum of cell signaling during infection. To determine whether other GTPases, namely RhoA and Rac1 are also recruited to the PVM, the tachyzoites of T. gondii RH strain were used to infect human 16-HBE cells, and Rho and Rac1 were localized by indirect immunofluorescence assay (IFA) using anti-Rho and -Rac1 antibodies. IFA revealed significant accumulation of these two small GTPases on the PVM. To further verify this observation, CFP-tagged RhoA and Rac1 were overexpressed in COS-7 cells, and 48 hr post-transfection, cells were infected with different virulent strains of RH and Pru tachyzoites, respectively. Regardless of the virulence of the parasite strains used, RhoA and Rac1 were recruited to the PVM (Figure 1).

Real-time observation of recruitment of RhoA GTPase to the PVM

To follow the events of RhoA GTPase recruitment to the PVM, COS-7 cells transfected with pECFP-RhoA WT were infected with T. gondii RH tachyzoites. The real-time photographs were taken at 0 min post-infection and every 5 min thereafter using a confocal fluorescence microscope (Figure 2).

We found that the CFP-tagged RhoA was recruited to the PVM at the very beginning of the invasion, probably through retention of the RhoA GTPase on the host cell membrane to the PVM, and the accumulation of RhoA on the PVM continued with the recruitment of the tachyzoite until it totally invaded into the host cell (Figure 2A-D: pink arrowhead). However, a focal point of RhoA was not seen at the immediate point of invasion (Figure 2A). After the tachyzoite was completely inside the host cell, the fluorescence intensity continually increased with time, suggesting continuous accumulation of the RhoA GTPase on PVM probably by diffusion from the host cell cytosol (E-H, white and yellow arrowhead). The green fluorescence and the DIC images showing the invasion processes are provided in Additional file 1: Data S1 and Additional file 2: Data S2.

The recruitment of Rho A and Rac1 GTPases into PVM is dependent on the GTPase activity

We next investigated if intact GTPase activity was required for PVM recruitment. We used dominant negative mutants of Rho and Rac1 (RhoA-N19 and Rac1-N17 respectively) in our study. These mutants tagged with CFP were overexpressed in COS-7 cells. At 48 hr post-transfection, the cells were infected with RH strain tachyzoites. Interestingly, the accumulation of these GTPases to the PVM was no longer seen when they were in these inactive forms, which constitutively bind only GDP (Figure 3). Thus, the recruitment of these Rho GTPases to the PVM only occurred when Rho GTPases retained normal activity.

The Rho A and Rac1 GTPases were activated upon T. gondiitachyzoite invasion

To determine if RhoA or Rac1 GTPases were actually activated following T. gondii tachyzoite invasion, we used GST-tagged Rhotekin-RBD protein on agarose beads specific for RhoA or GST-tagged PAK-PBD protein bound agarose beads specific for Rac only to bind the GTP-bound RhoA or Rac1 in the cell lysate, but not the GDP-bound form. Western-blot analyses detected increased amounts of GTP-bound RhoA and Rac1 from the infected cells compared with the uninfected cells (Figure 4), but no signals were detected in the negative control (16-HBE cells incubated with GDP) or the T. gondii infected groups. These results strongly suggest that T. gondii invasion results in the activation of RhoA and Rac1 GTPaes.

The recruitment of RhoA to T. gondiiPVM is dependent on different RhoA domains

In order to define what motifs are vital to the recruitment of Rho GTPases to the PVM, we concentrated on the study of Rho A as a representative protein. Sequential deletion of RhoA by 10 amino acids with site-directed mutation from the parental plasmid pECFP-RhoA-WT generated 19 RhoA mutants. The different CFP-tagged, truncated RhoA plasmids (M1-M19) were transfected into COS-7 cells grown on coverslips in 6-well plates and analyzed by immunofluorescence microscopy. M2 (RhoA^Δ11–20), M3 (RhoA^Δ21–30), M4 (RhoA^Δ31–40), M6 (RhoA^Δ51–60), M17 (RhoA^Δ161–170) could not be observed on the PVM (Figure 5), indicating the decisive motifs were potentially the GTP/Mg²⁺ binding site, the mDia effector interaction site, the G1 box, the G2 box and the G5 box. The other mutants were all similarly recruited to the PVM as in wild-type RhoA (Additional file 3: Data S3). These results show that the GAP (GTPase-activating protein) interaction site, the GEF (guanine nucleotide exchange factor) interaction site, the GDI (guanine nucleotide dissociation inhibitor) interaction site, the Rho kinase (ROCK) effector interaction site, the PKN/PRK1 effector interaction site, the Switch I region, the Switch II region, the G3 box and the G4 box were not the decisive motifs for the recruitment of RhoA to the PVM.

The PVM-localized Rho and Rac GTPases do not respond to epithelial growth factor (EGF) activation

Rho GTPases control cell motility by regulating the reorganization of the cytoskeleton in response to EGF [17]. Rho and Rac GTPases translocated from the cytosol to the cell membrane upon EGF activation [18]. To study whether the Rho and Rac GTPases accumulated on the PVM would translocate following EGF activation, the COS-7 cells overexpressing CFP-tagged Rho and Rac1 were starved overnight, infected with T. gondii RH tachyzoites and then activated with EGF. The result showed that the recruited Rho and Rac GTPases on the PVM did not change in fluorescence brightness, unlike the fluorescence brightness in the cytosol that became faint because of the translocation of RhoA and Rac1 from the cytosol to the cell membrane towards the EGF activation spot (Figure 6). More photographs showing the RhoA and Rac1 sequestered on the PVM regardless the activation of EGF are provided in Additional file 4: Data S4.

Interference with RhoA and Rac1 endogenous activity affects tachyzoite infection

To study the role of host cell RhoA and Rac1 GTPases during the tachyzoites invasion, COS-7 cells were over-expressed with RhoA-WT, RhoA-N19, Rac1-WT, and Rac1-N17. The endogenous expression of RhoA or Rac1 was inhibited by siRNA targeted towards either RhoA or Rac1 separately or towards both in human 16-HBE cells and then infected with RH tachyzoites. The infection rate was determined for each group. The results indicated that no significant difference was found between the infection rates of RhoA-WT or Rac1-WT overexpressed cells and the mock cells (Figure 7A-B, respectively). However, the dominant negative mutants RhoA-N19 and Rac1-N17 overexpressed in COS-7 cells inhibited the cell invasion by T. gondii tachyzoites significantly; the infection rates were approximately 60% of that of the mock cells (p < 0.01) (Figure 7A-B, respectively). Silencing RhoA, Rac1 or both RhoA and Rac1 in 16-HBE cells also showed a significant inhibition of cell invasion by tachyzoites (p < 0.01) Figure 7C-E). The infection rates of RhoA and Rac1 silenced cells were about 65% of that of the mock cells, while the infection rate of RhoA and Rac2 double-silenced cells was about 50% of that of the mock cells (Figure 7C).

The function of the Rho and Rac GTPases accumulated on PVM

Immunity-related GTPases (IRGs) also known as p47 GTPases, are key mediators of interferon-gamma-induced resistance to pathogens [19]. They cycle between GDP-GTP bound forms, and cooperatively oligomerize in the GTP-bound conformation on the T. gondii PVM [20]. Sequential recruitment of multiple IRGs to the PVM results in disruption of PVM and parasite digestion within 2 hr of infection [21]. Virulent type I strains resist recruitment and avoid clearance, while less virulent type II and III strains are effectively cleared by IRGs [22]. It was reported that a serine threonine kinase secreted by T. gondii, ROP18, binds to and phosphorylates IRGs on the PVM, and the phosphorylation of IRGs prevented clearance of T. gondii within inflammatory monocytes and IFN-γ-activated macrophages, conferring parasite survival in vivo[23]. ARF6 is a small GTPase of the ARF family that regulates membrane trafficking and actin cytoskeleton rearrangements at the plasma membrane. ARF6 was found recruited to the PV of T. gondii tachyzoites and ARF6 activity was necessary for cell invasion by tachyzoites of T. gondii[14]. These reports about the function of the GTPases on the PVM in T. gondii invasion urged us to hypothesize what is the function of the host cell Rho and Rac1 accumulating on the PVM.

Both the indirect immunofluorescence staining of the endogenous RhoA and Rac1 of the host cell, and the over-expressed CFP-tagged RhoA and Rac1 recombinant proteins in the host cell indicated the recruitment of RhoA and Rac1 in the PVM of T. gondii tachyzoites (Figure 1). From the real-time observation of the invasion of the host cell by T. gondii tachyzoites, we found that the recruitment of RhoA to the PVM happened at the very beginning of the invasion either from the membrane or from the cytosol (Figure 2). Those over-expressed CFP-tagged dominant negative mutants RhoA-N19 and Rac1-N17 did not accumulate to the PVM (Figure 3) implying the recruitment of RhoA and Rac1 is dependent on their GTPase activity. The GST-pull down assay detected greater amounts of GTP-bound RhoA and Rac1 in the infected host cells than in uninfected cells (Figure 4). Through CFP-tagged RhoA and Rac1 being visualized under the GFP filter, we found that RhoA and Rac1 GTPases in the host cell cytosol were translocated to the host cell membrane following EGF activation, while unlike the GTPases in the cytosol, RhoA or Rac1 on the PVM did not diffuse, translocate or respond to EGF activation. EGF activates RhoA and Rac1 through activation of the EGF pathway [24, 25]. This observation led us to hypothesize that the Rho and Rac1 GTPase recruited on the PVM probably was GTP-bound and could not be activated again by EGF, while most of the GTPases in the cytosol are in GDP-bound form and could be continually activated and translocated to the cell membrane upon EGF activation (Figure 6).

These observed results imply the invasion of the tachyzoites need the activation of RhoA and Rac1 GTPases; and the recruitment of these activated GTPases to the PVM is much more than a phenomenon as it may perform some as yet undefined but important function(s).

The decisive RhoA GTPases motifs for recruitment to parasitophorous vacuole membrane following T. gondiiinvasion

Wild-type Rho and Rac GTPases with normal GTPase activity were recruited to the PVM, but those mutants that constitutively bind only GDP (RhoA-N19 and Rac1-N17) lacked this ability. The 10 amino acid sequentially deleted RhoA mutants were used in the identification of the definitive motif(s) necessary for the recruitment to the PVM. M2 (RhoA^Δ11–20), M3 (RhoA^Δ21–30), M4 (RhoA^Δ31–40), M7 (RhoA^Δ61–70) and M17 (RhoA^Δ161–170) lacked the ability to be recruited to the PVM (Figure 5). Based on the analysis of the RhoA GTPase conserved domains[26], the GTP/Mg²⁺ binding site, the mDia effector interaction site, the G1 box, the G2 box and the G5 box are the most potential motifs that determine the recruitment of RhoA GTPases to the PVM (Figure 5).

Rho GTPases are molecular switches that cycle between an active GTP-bound and an inactive GDP-bound form, which regulate many essential cellular processes, including actin dynamics, gene transcription, cell-cycle progression and cell adhesion [27]. When in the active forms, Rho GTPases are able to interact with effector or target molecules to initiate downstream responses, signal transduction terminates when GTP is hydrolyzed to form GDP, and at which point the cycle is finished completely [27]. The GTP/Mg²⁺ binding site of Rho GTPases is used to bind GTP and Mg²⁺, which activates the GTPases [28]. The mDia effector interaction site is the domain that binds with mDia as a downstream Rho effector involved in microtubule stabilization. The mDia site induces stable microtubules that are capped and indicates that mDia may promote this microtubule capping by directly binding to microtubules. [29]. The G1-G5 boxes are the GDP/GTP-binding motif elements that comprise a ~ 20 kDa phosphate domain (G domain, Ras residues 5–166), which is conserved in all Ras super family proteins [30]. The decisive motifs are either related to GTP binding or with the effector regulating microtubules. This finding is consistent with our proposal that the recruitment of Rho GTPase to PVM depends on its enzymatic activity, and the invasion of T. gondii needs the rearrangement of host cell cytoskeleton.

Host cell RhoA and Rac1 activation is required for efficient cell invasion by T. gondiitachyzoites, which is a shared mechanism by many other intracellular pathogens infection

The major function of Rho GTPases activation is to regulate the dynamics and organization of the actin cytoskeleton [17], which is vital to the cell invasion of T. gondii tachyzoites. First, T. gondii tachyzoites invasion activates the reorganization of the microfilaments and microtubules of the host cell [31, 32]. Reorganization of host cell F-actin during entry of Toxoplasma tachyzoites has been visualized, and the entry was dependent on the actin dynamics [31]. Second, any treatment to cease the normal cytoskeleton reorganization of host cells will impair T. gondii invasion efficiency. Cell invasion by T. gondii tachyzoites is significantly inhibited in cells treated with colchicum (a MT inhibitor) [33], cytochalasin D (an actin inhibitor) [14, 33] and jasplakinolide (a chemical disrupting actin filaments, which induces actin polymerization) [31]. Maintenance of host cell actin cytoskeleton integrity is important to parasite invasion [14].

In our research, no significant difference was found in the infection rates of T. gondii tachyzoites in the cells overexpressed with RhoA or Rac1 wild-type proteins compared with those untransfected cells, while either overexpression of dominant negative mutants of RhoA or Rac1 or inhibition of endogenous RhoA or Rac1 in the host cell could significantly reduce cell invasion (Figure 7). This result indicated that it is the normal endogenous activity of RhoA and Rac1 that defines the efficiency of cell invasion by T. gondii tachyzoites, but not the amount of these proteins. This requirement is also reported in other intracellular pathogens. Shigella entry into HeLa cells induces membrane ruffling at the bacterial entry site, and the three Rho isoforms were recruited into bacterial entry sites. This membrane folding caused by invasion was abolished by using a Rho-specific inhibitor, and bacterial entry was impaired accordingly [34]. Hela cells transfected with the dominant negative versions of Rac1 or RhoA reduced group B Streptococcus invasion by 75% and 51%, respectively, suggesting that Rho GTPases are indispensable for efficient invasion of HeLa cells by this bacterium [35]. In MDCK cells, RhoA and Rac1were activated during Trypanosoma cruzi invasion and then triggered the reorganization of F-actin cytoskeleton, especially distinct in the invasion position on the cell membrane. The invasion of T. cruzi G strain extracellular amastigotes was specifically inhibited in Rac1-N17 dominant-negative cells [36, 37]. After the invasion of the rabbit corneal epithelial cells (SIRC) by Candida albicans, host cell actin filaments formed a rigid ring-like structure in the host cell. Immunochemical staining of actin and the expression of chimeric green fluorescent protein (GFP)-GTPases (RhoA, Rac1) showed the colocalization of the GTPases with actin at invasion and actin polymerization sites, but this colocalization was not seen in SIRC cells expressing a GFP-tagged dominant-negative mutant of GTPases. Inhibition of invasion was observed in SIRC cells expressing dominant-negative mutants of Rac1 and RhoA GTPases [38]. These findings suggest that many pathogens may employ conserved pathways for invasion.

The Rho and Rac cell signaling involved in the cytoskeleton reorganization triggered by T. gondiiinvasion

When epithelial cells are stimulated by EGF, c-Src is activated by EGF-induced EGF receptor activation [39]. After the activation of c-Src, Ephexin, VAV-2 and Tiam 1 are rapidly phosphorylated by c-Src [40, 41]. Phosphorylation of Ephexin promotes its GTPase activity toward RhoA [42, 43], and RhoA downstream effector Rho-associated kinase ROCK directly phosphorylates LIM-kinases LIMK1 and LIMK2, which in turn phosphorylates actin-depolymerizing factor destrin and actin-associated protein cofilin [44]. ROCK2 kinase phosphorylates CRMP2, and the phosphorylation of CRMP2 reduces its tubulin-heterodimer binding and the promotion of microtubule assembly [45, 46]. Activation of VAV-2 activates RhoA and Rac1 [47]. Downstream of Rac1, p21-activated kinase 1 (PAK1) activates LIMK1, and regulates the actin cytoskeletal reorganization through the phosphorylation of the actin-depolymerizing factors cofilin and destrin and their actin-depolymerizing activities [48, 49]. PAK1 also phosphorylates Arp2/3 complex to promote actin polymerization [50]. The F-actin-binding protein cortactin is a prominent target of various tyrosine kinases (c-Src) and regulates cytoskeletal dynamics [42, 50]. Tyrosine phosphorylation of cortactin has been suggested to reduce its F-actin cross-linking capability [51]. In our research, we are not clear about the upstream cell signaling component of the Rho and Rac GTPases involved in T. gondii infection, but we have witnessed the activation of RhoA and Rac1 of host cells and the reorganization of the cytoskeleton for PV formation during the infection of T. gondii. The cell signaling involved in this process is shown in Figure 8.