A Novel Galectin-like Domain from Toxoplasma gondii Micronemal Protein 1 Assists the Folding, Assembly, and Transport of a Cell Adhesion Complex*

Immediately prior to invasion Toxoplasma gondii tachyzoites release a large number of micronemal proteins (TgMICs) that participate in host cell attachment and penetration. The TgMIC4-MIC1-MIC6 complex was the first to be identified in T. gondii and has been recently shown to be critical in invasion. This study establishes that the N-terminal thrombospondin type I repeat-like domains (TSR1-like) from TgMIC1 function as an independent adhesin as well as promoting association with TgMIC4. Using the newly solved three-dimensional structure of the C-terminal domain of TgMIC1 we have identified a novel Galectin-like fold that does not possess carbohydrate binding properties and redefines the architecture of TgMIC1. Instead, the TgMIC1 Galectin-like domain interacts and stabilizes TgMIC6, which provides the basis for a highly specific quality control mechanism for successful exit from the early secretory compartments and for subsequent trafficking of the complex to the micronemes.

Immediately prior to invasion Toxoplasma gondii tachyzoites release a large number of micronemal proteins (TgMICs) that participate in host cell attachment and penetration. The TgMIC4-MIC1-MIC6 complex was the first to be identified in T. gondii and has been recently shown to be critical in invasion. This study establishes that the N-terminal throm-bospondin type I repeat-like domains (TSR1-like) from TgMIC1 function as an independent adhesin as well as promoting association with TgMIC4. Using the newly solved three-dimensional structure of the C-terminal domain of TgMIC1 we have identified a novel Galectin-like fold that does not possess carbohydrate binding properties and redefines the architecture of TgMIC1. Instead, the TgMIC1 Galectin-like domain interacts and stabilizes TgMIC6, which provides the basis for a highly specific quality control mechanism for successful exit from the early secretory compartments and for subsequent trafficking of the complex to the micronemes.
Toxoplasma gondii is a protozoan parasite of the phylum Apicomplexa, which infects virtually all warm-blooded animals and invades almost any cell type. Host cell invasion by this obligate intracellular parasite is an active process initiated by the formation of a tight association/junction with the host cell plasma membrane and leading to the creation of a parasitophorous vacuole. Contact with the host cell results in an increase in parasite intracellular calcium ions, which trigger apical organelles called micronemes to discharge their contents (1). Several micronemal proteins act as ligands for host cell receptors (2), while TgMIC2 and other transmembrane proteins establish a connection with the parasite actinomyosin system via their cytoplasmic tail (3), thus providing the motive force for penetration (4). It is becoming increasingly apparent that many microneme proteins are found in stable adhesive complexes, which are formed in the endoplasmic reticulum, and normally comprise an escorter protein, which is responsible for correct micronemal targeting, and one or more soluble effector proteins. The first such complex to be discovered in T. gondii was TgMIC4-MIC1-MIC6, in which TgMIC6 fulfils the role of the escorter protein, whereas TgMIC1 and TgMIC4 function as adhesins (5). Although TgMIC4-MIC1-MIC6 and the recently identified micronemal complex, TgMIC3-MIC8 (5,6), are individually dispensable, the generation of double knock-outs for TgMIC1 and TgMIC3 renders the parasites avirulent in vivo, demonstrating functional synergy between these complexes (7). Deletion of the mic1 gene in T. gondii also confirmed the specific and critical role played by TgMIC1 in host cell attachment and invasion in vitro.
Micronemal proteins have a modular structure with common themes in domain organization, for example many possess thrombospondin type-1 repeat domains (TSR1), 4 apple (or PAN) domains, and epidermal growth factor-like (EGF) domains (8). A schematic representation of the organization within the TgMIC4-MIC1-MIC6 complex is depicted in Fig. 1. TgMIC6 is a 34-kDa transmembrane protein possessing three EGF domains together with an acidic region and the targeting information for transporting the whole complex to the microneme (5). TgMIC4 is a soluble, adhesive 61-kDa protein containing six conserved apple domains, which is processed extracellularly, releasing the sixth apple domain (9). TgMIC1 is a 49-kDa, soluble protein and forms the core of the complex by simultaneously interacting with TgMIC4, TgMIC6, and host cells (9,10). Furthermore, TgMIC1 is required for the successful exit of TgMIC4 and TgMIC6 from early compartments of the secretory pathway, namely the endoplasmic reticulum (ER) and the Golgi apparatus (5,11).
Although earlier studies have clearly highlighted the importance of TgMIC1, the finer structural details are poorly understood. This is particularly apparent in sequence annotations for TgMIC1 (7); while the occurrence of two degenerate TSR1-like domains within the N terminus of TgMIC1 is well accepted, the architecture of the C-terminal 190 amino acids has remained unclassified and disregarded. This study addresses many of these issues and uncovers new features of the TgMIC4-MIC1-MIC6 complex. We have identified the regions within TgMIC1, as well as TgMIC4 and TgMIC6 that contribute to complex formation and binding to host cells. Combined NMR and biochemical experiments reveal a novel interaction surface in the previously unidentified Galectin-like fold of TgMIC1, which interacts with TgMIC6. A novel function is attributed to the Galectin-like fold of TgMIC1, in that it drives the correct folding and subsequent exit of TgMIC6 from the ER and Golgi, revealing a highly specialized type of quality control mechanism.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-Unless otherwise stated all chemical reagents were obtained from Sigma, and all restriction enzymes were purchased from New England Biolabs. Polyclonal rabbit serum reacting to TgMIC4, TgMIC6, and the monoclonal antibody 5B1 anti-TgMIC4 were described previously (5,9). TgMIC1 was detected with either a polyclonal rabbit serum (10) or using the anti-myc tag monoclonal antibody 9E10.
Pichia pastoris Cloning and Expression-The coding sequence corresponding to the amino acids 17-456 of TgMIC1 was amplified from cDNA by PCR. The PCR product was digested with EcoRI and NotI then ligated into the corresponding restriction sites within the multiple cloning site of pPICZ␣. pPICZ-TgMIC1 was used as a template from which all other TgMIC1 fragments were amplified. The coding sequences corresponding to amino acids 26 -580 of TgMIC4 were amplified from genomic DNA. The coding sequences corresponding to amino acids 29 -309 of TgMIC6 were also amplified from genomic DNA. All fragments were cloned directly into pPIC9K or via pGEM-T.
P. pastoris transformation and expression was performed using the Pichia expression kit (Invitrogen), according to the supplied protocols, and all yeast strains were maintained on yeast extract-peptone-dextrose (YPD)-rich media. Transformation of the supplied host strain GS115 was performed by electroporation following linearization of the plasmids with SalI or BglII for pPICZ␣-and pPIC9K-based vectors, respectively. Selection of transformants was then performed on YPD Zeocin (100 mg/ml) for pPICZ␣ or on minimal media lacking histidine in the case of pPIC9K. Expression was performed using BMGY and BMMY media according to the manufacturer's instructions.
Escherichia coli Cloning and Expression-TgMIC1-CT, spanning residues 320 -456 in TgMIC1, was expressed using the pET 21b plasmid (Novagen) in the BL21 (DE3) E. coli strain (Stratagene) as described previously (12,13). TgMIC6-EGF23 acid and TgMIC6-EGF3 acid were expressed using the pET 32 Xa/LIC plasmid (Novagen) in the Origami (DE3) E. coli strain (Novagen). 15 N-Labeled samples of TgMIC6-EGF3 acid were produced in minimal media, containing 0.07% 15 NH 4 Cl and 0.2% glucose, supplemented with 50 g/ml carbenicillin. Protein expression was induced by the addition of 500 M isopropyl ␤-D-thiogalactopyranoside. TgMIC6-EGF3 acid was purified using the binding of the internal hexahistidine tag to the nickel-nitrilotriacetic acid HISBind resin (Novagen) and cleaved from the thioredoxin fusion protein using thrombin. Samples were dialyzed into 20 mM sodium phosphate buffer, pH 7.0, 150 mM NaCl and concentrated to ϳ0.5 mM for NMR. In all cases only soluble fractions were used for subsequent purification.
Host Cells and Parasite Cultures and Transfection-Tachyzoites from RHhxprt Ϫ and derived mutant strains were propagated in African green monkey kidney (Vero) cells or human foreskin fibroblasts (HFFs) monolayers grown in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% fetal calf serum (Invitrogen), 1 mM glutamine, 10 g/ml gentamicin. The mic1ko mutant used in this study corresponds to the RHhxprt-strain in which the TgMIC1 gene was disrupted by homologous recombination (5). The transient transfection experiments were performed by electroporation as previously described (14).
Preparation of Excreted Secreted Antigen-For the preparation of excretory/secretory products, ϳ2 ϫ 10 9 tachyzoites were washed in HBBS, 10 mM HEPES, 0.1 mM EGTA (HHE) and resuspended in PBS, pH 7.4, 0.1% fetal calf serum, 1% ethanol to a cell density of 1 ϫ 10 6 cells/l and incubated at 37°C as described previously (2). Cells were removed by centrifugation and the excreted secreted antigen (ESA) product containing supernatant retained.
Co-immunoprecipitation-200 l of P. pastoris supernatant (0.002-0.01 mg/ml protein) or 100 l of ESA products (equivalent to 1 ϫ 10 8 tachyzoites) were incubated with an appropriate antibody overnight at 4°C with agitation. Supernatant or ESA products in the absence of antibodies, and antibodies in the absence of supernatant or ESA products, were included as controls. 200 l of 10% Protein A-Sepharose CL4B (Amersham Biosciences) slurry in PBS containing 1% BSA was added to each sample and incubated for 3 h at 4°C with agitation. The beads were washed five times for 10 min in PBS and boiled for 10 min in Cell Binding Assays-These were performed as described previously (9). Briefly, confluent monolayers of HFFs, Veros, or Chinese hamster ovary cells, grown in 6-or 12-well plates, were blocked for 30 min at 4°C with 1% BSA in PBS containing 1 mM CaCl 2 and 0.5 mM MgCl 2 (CM-PBS). Excess BSA was removed by two 5-min washes with ice-cold CM-PBS, after which the proteins to be assayed were then added either in the form of P. pastoris culture supernatant (ϳ0.25 g) or ESA products (equivalent to 2 ϫ 10 9 tachyzoites) diluted in CM-PBS to a total volume of 500 l. After incubation at 4°C for 1 h the unbound fraction was removed, and the cells were washed four times for 5 min with ice-cold CM-PBS. The cell-bound fraction (CBF) was collected either by the direct addition of 50 l 1ϫ SDS-PAGE loading buffer or by lysing the cells in 1 ml of RIPA (50 mM Tris, pH 7.5, 1% Triton X-100, 0.5% sodium deoxycholate, 0.2% SDS, 100 mM NaCl, 5 mM EDTA) prior to acetone precipitation and re-suspension in SDS-PAGE sample loading buffer with 0.1 M dithiothreitol.
Indirect Immunofluorescence Assay-All manipulations were carried out at room temperature. 24 -30 h after electroporation, tachyzoiteinfected HFF cells on glass coverslips were fixed with 4% paraformaldehyde, 0.05% glutaraldehyde for 20 min, followed by 3-min incubation with 0.1 M glycine in PBS. Fixed cells were permeabilized with 0.2% Triton X-100 in PBS for 20 min and blocked in 2% bovine serum albumin in PBS for 20 min. The cells were then stained with the primary antibodies followed by Alexa 594 goat anti-rabbit or Alexa 488-conjugated goat anti-mouse antibodies (Molecular Probes, Cappel, and Bio-Rad). Previously described rabbit polyclonal antibodies were used for the detection of TgMIC4 (9) and TgMIC6 (6). Confocal images were collected with a Leica laser scanning confocal microscope (TCS-NT DM/IRB) using a 100 ϫ Plan-Apo objective with NA 1.30. Single optical sections were recorded with an optimal pinhole of 1.0 (according to Leica instructions) with 16 times averaging.
NMR Spectroscopy and Structure Calculation-Backbone and sidechain assignment were completed using standard double and tripleresonance assignment methodology (12,13). H ␣ and H ␤ assignments were obtained using HBHA (CBCACO)NH (13). The side-chain assignments were completed using HCCH-total correlation (TOCSY) spectroscopy and (H)CC(CO)NH TOCSY (13). Three-dimensional 1 H-15 N/ 13 C NOESY-HSQC (mixing time 100 ms at 500 and 800 MHz) experiments provided the distance restraints used in the final structure calculation. Heteronuclear 1 H-15 N NOE data with minimal water saturation were acquired using the pulse sequence described previously (15).
A total of 42 long range NOEs, providing unambiguous three-dimensional information, were manually assigned from the NOESY data. The ARIA protocol (16) was used for completion of the NOE assignment and structure calculation. A total of 3156 NOE-derived distances were assigned from 13 C-and 15 N-edited spectra, which comprised 2275 unambiguous and 881 ambiguous restraints. Dihedral angle restraints derived from TALOS were also implemented (17). Using NOE data and characteristic side-chain chemical shifts a disulfide bond was imposed between cysteines 35 and 106. C␤ chemical shifts are 52 and 42 ppm, respectively. The frequency window tolerance for assigning NOEs was Ϯ0.03 ppm and Ϯ0.04 ppm for direct and indirect proton dimensions and Ϯ0.5 and Ϯ1.2 ppm for nitrogen and carbon dimensions, respectively. The ARIA parameters, p, Tv, and Nv, were set to default values. The 10 lowest energy structures had no NOE violations greater than 0.5 Å and dihedral angle violations greater than 5°. The structural statistics are presented in  13 Clabeled TgMIC1-CT were prepared in 20 mM sodium phosphate buffer at pH 7 at ϳ50 M in 0.25 ml. Unlabeled TgMIC1-CT or TgMIC6-EGF3 acid in the same buffer were introduced at several steps up to a 10-fold molar excess and two-dimensional 1 H-15 N HSQC spectra were recorded at each stage under identical experimental conditions. Perturbed amide resonances in the 15 N-TgMIC1-CT/TgMIC6-EGF3 acid complex were assigned to the nearest peaks to the 1 H-15 N HSQC spectrum of TgMIC1-CT. This is likely to represent an underestimation of the actual chemical shift perturbation. ESAs from several parasite knock-out strains and the corresponding CBFs using rabbit anti-TgMIC1. In the absence of TgMIC6 (mic6ko), TgMIC1 does not show altered cell binding. If TgMIC4 is totally absent (mic4ko), TgMIC1 still binds efficiently to host cells. Molecular mass markers are shown (in kilodaltons). B, Western blot analysis of ESAs from several knock-out strains and the corresponding CBFs using the rabbit anti-TgMIC4. TgMIC4 originating from wild-type or mic6ko parasites showed no reduced binding capacity, while TgMIC4 in ESAs lacking TgMIC1 (mic1ko) exhibits a dramatic reduction in binding. Molecular mass markers are shown (in kilodaltons). C, cell binding assays were performed using supernatants of P. pastoris cultures expressing TgMIC1 alone, TgMIC4 alone, or TgMIC6 alone. The P. pastoris strain used for each experiment is shown above each panel. The antibodies used for each Western blot are shown below each panel. Samples of supernatant (Sup), Wash (W), and CBF were run on each gel (see "Experimental Procedures"). Molecular mass markers are shown (in kilodaltons). These data confirm that recombinant TgMIC1 retains significant cell binding activity. wt, wild-type.

TgMIC1 Binds to Host Cells in the Absence of TgMIC4 and TgMIC6-
The individual contributions of each protein within the TgMIC4-MIC1-MIC6 complex to cell adhesion are poorly understood, particularly with regard to TgMIC1. As a first step to investigating the role of TgMIC1 and its interdependence within the TgMIC4-MIC1-MIC6 complex, the host cell binding ability of each protein was tested in the absence of the others. To do so, wild-type, mic1ko, mic4ko, and mic6ko mutant parasites were stimulated to secrete their micronemal contents, and the yielded ESAs were incubated with HFF cells and analyzed for the presence of bound TgMIC1 or TgMIC4. When produced in the absence of either TgMIC4 or TgMIC6, TgMIC1 bound HFF cells to the same extent as TgMIC1 released from wild-type parasites (Fig. 2A). The ability of TgMIC4 binding to host cells was not affected by the absence of TgMIC6, which is consistent with the lack of a direct interaction between these two proteins (5). In contrast, the absence of TgMIC1 resulted in a considerable decrease in binding of TgMIC4 to host cells (Fig. 2B). Since the absence of TgMIC1 also causes a significant accumulation of TgMIC4 early in the secretory pathway, it is plausible that the lack of TgMIC4 binding could be explained by its incorrect folding, or alternatively the cell binding properties of TgMIC4 may be predominantly conferred by its association with TgMIC1.
The N-terminal TSR1-like Region from TgMIC1 Performs Dual Functions, Cell Binding and Recruitment of TgMIC4-To dissect the functional regions of TgMIC1, fragments from TgMIC1, TgMIC4, and TgMIC6 were expressed in P. pastoris and subsequently tested for cell binding properties and complex formation. The cell binding observations using parasite-secreted antigens (ESAs) were first confirmed with Mouse anti-TgMIC1 precipitates TgMIC1 (panel c) and co-precipitates TgMIC4 (panel d). *, note that all TgMIC1 fragments were myc-tagged. The prominent band above 50 kDa indicates antibody cross-reactivity. B, cell binding assays were performed using supernatants of P. pastoris cultures co-expressing TgMIC1 and TgMIC4. The P. pastoris strain used for each experiment is shown above each panel. The antibodies used for each Western blot are shown below each panel. Samples of supernatant (Sup), Wash (W), and CBFs were run on each gel (see "Experimental Procedures"). Molecular mass markers are shown (in kilodaltons). In the presence of TgMIC1 the binding of TgMIC4 is substantially enhanced (compared with Fig. 2C). C, cell binding assays were performed with P. pastoris culture supernatants expressing full-length TgMIC1 (TgMIC1), both TSR1-like domains (TSR1), each TSR1-like domain individually (TSR1-1 and TSR1-2), and the C-terminal domain (CT). Western blots of the CBF and the supernatant were probed with anti-myc. Molecular mass markers are shown (in kilodaltons). *, note that all TgMIC1 fragments were myc-tagged. D, full-length TgMIC1 (MIC1), TgMIC1-TSR1 (TSR1), and the TgMIC1-CT were individually co-expressed with full-length TgMIC4 in P. pastoris. Immunopreciptation experiments were performed using the culture supernatant from each strain using rabbit anti-TgMIC4 (panel c) and mouse anti-myc (panel a). Panel a shows that full-length TgMIC1 and TgMIC1-TSR1 (precipitated by anti-myc) co-precipitate TgMIC4 but TgMIC1-CT does not. Panel c confirms the interaction between the TgMIC1-TSR1 and TgMIC4 as only full-length TgMIC1 and TgMIC1-TSR1 are co-precipitated by anti-MIC4. Panels b and d show the expression levels of TgMIC4 and the TgMIC1-derived fragments in the culture supernatants. The antibodies used for each Western blot are shown below each panel. *, note that all TgMIC1 fragments were myc-tagged. The prominent band above 50 kDa indicates antibody cross-reactivity. Co-IP, co-immunoprecipitation. recombinant TgMIC1 and TgMIC4, revealing that indeed recombinant TgMIC1 can bind to HFF cells independently of any other T. gondii factors (Fig. 2C). However, unlike TgMIC1, little or no recombinant TgMIC4, when expressed alone, could be detected in the cell binding fraction. This is consistent with TgMIC4 being incorrectly folded or having no independent cell binding property. As expected neither native nor recombinant TgMIC6 exhibit detectable cell binding activity (Fig. 2C).
Co-expression of TgMIC1 and TgMIC4 in P. pastoris followed by co-immunoprecipitation together with Western blot analysis revealed an intact TgMIC1-MIC4 sub-complex (Fig. 3A). There was no corresponding increase in the host cell binding efficiency of TgMIC1 in the presence of TgMIC4 indicating that there is no significant direct contribution of TgMIC4. Furthermore, the cell binding efficiency of TgMIC4 binding was markedly increased when expressed in presence of TgMIC1 (Fig. 3B), confirming findings from mic1ko parasites.
To define the region of TgMIC1 responsible for cell adhesion, fragments containing both TSR1-like domains (termed TgMIC1-TSR1 hereafter), individual TSR1-like domains, or the C-terminal region (termed TgMIC1-CT hereafter) were expressed in P. pastoris and tested separately. A stretch of ϳ60 amino acids between these two regions was shown previously to be unstructured (12). Only the fragment containing the entire TSR1-like region bound to host cells, indicating that the cell adhesion properties of TgMIC1 require the presence of both intact TSR1-like domains (Fig. 3C). Further experiments in which TgMIC4 was co-expressed with either TgMIC1-TSR1 or TgMIC1-CT revealed that the tandem TSR1-like domains are also responsible for the interaction between TgMIC1 and TgMIC4 (Fig. 3D).
The C-terminal Domain of TgMIC1 Contains a Novel Galectin-like Domain-Although the C-terminal domain possesses no sequence homology with any other protein, secondary structure prediction algorithms suggest the presence of a folded domain within the C-terminal 137 residues. To investigate the nature of TgMIC1-CT we embarked on a high resolution structure determination. Using a combination of manual and automated NMR assignment methods for analysis (16), a family of high resolution structures for TgMIC1-CT were calculated with excellent agreement with experimental data and structural quality (TABLE ONE). All areas of secondary structure are very well defined (Fig. 4A); the average pairwise root mean square deviation (r.m.s.d.) for the water-refined final structures is 0.47 Å for the backbone atoms and 0.92 Å for the heavy atoms of residues within secondary structure (Fig. 4A).
The final structure of the TgMIC1-CT exhibits a Galectin-like topology (Fig. 4) consisting of an 11-stranded ␤-barrel, formed from the association of a five-stranded ␤-sheet (␤K, ␤B, ␤G, ␤H, and ␤I) and a six-stranded ␤-sheet (␤A, ␤J, ␤C, ␤D, ␤E, and ␤F). The strands are connected by short loops, and with the exception of the C-terminal 5 residues, the entire structure forms a globular, rigid scaffold. Low 1 H-15 N heteronuclear NOE values (data not shown) for the five C-terminal residues (residues 133-137 in TgMIC1-CT) indicate that this region is highly flexible. The structure of TgMIC1-CT also reveals a short, additional strand at the N terminus (␤AЈ) that is not present in archetypal galectin domains (Fig. 4, B and C). Despite sequence identities of only 7%, TgMIC1-CT superimposes with an r.m.s.d. of 2.8 Å over 112 equivalent backbone C ␣ atoms of the human Galectin-3 (Protein Data Bank code 1a3k) (18,19).
The Galectin-like Domain of TgMIC1 Promotes Folding of TgMIC6-Co-immunoprecipitation and Western blot analysis performed with TgMIC1 fragments and TgMIC6 (Fig. 5A) revealed a specific interaction between TgMIC1-CT and TgMIC6. The extracellular portion of mature TgMIC6 possesses two EGF domains, namely EGF2 and EGF3, together with an acidic region that extends EGF3 at its C terminus (Fig.  1). To localize the region of TgMIC6 responsible for TgMIC1 binding, we expressed EGF2 and EGF3 together or EGF3 alone plus the C-terminal acidic portions (referred to TgMIC6-EGF23 acid and TgMIC6-EGF3 acid henceforth) together with TgMIC1 and successfully co-precipitated the complex in both cases, indicating a direct interaction between TgMIC1 and TgMIC6-EGF3 acid . It was not possible to produce EGF3 alone in a soluble form, which suggests the acid tail may be, at least in part, an elaboration to the canonical EGF domain or confer enhanced solubility.
High resolution detail for any micronemal protein interaction has yet to be reported. The availability of the high resolution structure for TgMIC1-CT provides the opportunity to perform NMR titration experiments as a means to investigate these interacting surfaces. An analysis of amide line widths and chemical shift changes for TgMIC1-CT in the presence of TgMIC6-EGF3 acid was carried out (Fig. 5B). Resonances from residues in ␤B, ␤G, ␤H, and ␤I on the six-stranded sheet and ␤C, ␤D, and ␤E from the opposite face experience significant chemical shift changes (Fig. 5C). The TgMIC6-binding region of TgMIC1-CT may be delineated, forming a large convex area covering both sides of the molecule, which overlays well with the hydrophobic surface displayed by TgMIC1-CT (Figs. 4E and 5C).
The two-dimensional 1 H-15 N NMR spectrum of TgMIC6-EGF3 acid in the absence of TgMIC1-CT is indicative of an unfolded protein (Fig.  5D, left panel). EGF domains often contain a Ca 2ϩ -binding site at their N termini that is important for stability and function. No effect was seen on the NMR spectrum of TgMIC6-EGF3 acid recorded in the presence of a large excess of Ca 2ϩ (100 mM), which confirms predictions based on the absence of any Ca 2ϩ recognition motifs. In contrast, the stepwise addition of TgMIC1-CT had a dramatic effect on the two-dimensional 1 H-15 N NMR spectrum of TgMIC6-EGF3 acid . A number of amide resonances appear at chemical shifts that are characteristic of the folding of a small, structured domain within TgMIC6-EGF3 acid (Fig. 5D, right  panel). The structures of classic EGF domains reveal two doublestranded ␤-sheets and six cysteine residues involved in three disulfide bonds. As residues on both sides of the Galectin fold of TgMIC1-CT are implicated in binding TgMIC6, it is likely that a portion of the acidic tail is also involved in the interaction. The Galectin-like Domain of TgMIC1 Rescues the Transport of TgMIC6-This, together with those describing the retention of TgMIC4 and TgMIC6 within the ER/Golgi of mic1ko parasites (5,11), implies that correct folding or stabilization of TgMIC6-EGF3 acid , assisted by the galectin-like domain of TgMIC1, could provide the necessary quality control mechanism for successful exit from the early compartments of the secretory pathway. To assess this, mic1ko parasites were complemented with constructs expressing full-length TgMIC1, TgMIC1-TSR1, or TgMIC1-CT (together with a myc epitope for detection). The complementation was analyzed by an indirect immunofluorescence assay and monitored by confocal microscopy. The expression of full-length TgMIC1 was able to rescue fully TgMIC4 and TgMIC6 targeting to the micronemes (Fig. 6B). In contrast, TgMIC1-TSR1 alone did not affect TgMIC6, which was retained in the early secretory compartment (Fig. 6C, bottom panels). Interestingly, however, the presence of the TgMIC1-TSR1 enabled TgMIC4 to exit the ER into the dense granules (Fig. 6C, top panels), which has been previously identified as the default pathway for secretion of soluble proteins in T. gondii (20). This result is consistent with a direct interaction between TgMIC4 and TgMIC1-TSR1 as neither the TSR1-like fragment nor TgMIC4 carry targeting information for the micronemes; the two proteins follow the default pathway and accumulate in the dense granules (Fig. 6C). The presence of TgMIC1-CT does not have any obvious impact on the localization of TgMIC4 (Fig. 6D, top panels). Strikingly, complementation with TgMIC1-CT restores the exit of TgMIC6 from the early secretory compartments and subsequent transport to the micronemes (Fig. 6D, bottom panels).

DISCUSSION
It is well accepted that the binding of micronemal proteins to host cells provides a "molecular bridge" to the parasite, thereby facilitating further steps of invasion. TgMIC1 supplies the platform for assembly of the TgMIC4-MIC1-MIC6 complex by making independent interactions to TgMIC4, TgMIC6, and the host cell (Fig. 1B). Our data reveal that the N-terminal TSR1-like domains from TgMIC1 are able to bind host cells independently of any other parasite factor. Furthermore, we highlight the multifunctionality of the TgMIC1-TSR1, in that it is able to recruit TgMIC4 to the complex and simultaneously anchor TgMIC1 to the host cell surface.
A role for the C-terminal region from TgMIC1 (TgMIC1-CT, Fig. 1) has yet to be revealed, and its annotation been has been neglected in literature studies. Specifically, no functional or sequence annotation for TgMIC1-CT has been reported, which can be presumably be attributed to the lack of detectable sequence homology with any other protein. The solution structure of the C-terminal domain from TgMIC1 represents the first atomic resolution insight into a T. gondii micronemal protein and reveals a novel member of Galectin-fold family. Galectins form part of a unique family of soluble, calcium-independent, carbohydrate-binding animal lectins (21,22). Structures of Galectin-carbohydrate complexes (Fig. 4, C and D) have revealed a consistent picture of carbohydrate recognition, in which critical side chains are located in a central region of the six-stranded ␤-sheet and comprise an array of hydrophilic residues (in the case of Galectin-3 these include His 158 , Asn 160 , Arg 162 , Glu 165 , Asn 174 , Glu 184 , and Arg 186 ; Fig. 4, C and D) together with the key aromatic side chain of Trp 181 (18,19). These positions are not conserved in the equivalent locations within TgMIC1-CT suggesting that carbohydrate recognition is not retained by TgMIC1-CT. Furthermore, the concave nature of the carbohydrate-binding pocket, formed by sur- FIGURE 5. TgMIC1-CT interacts with and assists the folding of TgMIC6-EGF3 acid . A, supernatants of P. pastoris cultures co-expressing TgMIC6 with full-length TgMIC1, TgMIC1-TSR1, or TgMIC1-CT were used for co-immunoprecipitation (Co-IP) experiments. All MIC1 Tg fragments were myc-tagged. Anti-myc precipitation of TgMIC1-CT, but not TgMIC1-TSR1 (panel a), co-precipitated TgMIC6. This was confirmed using anti-TgMIC6, which co-purified TgMIC1-CT but not TgMIC1-TSR1 (panel c). Panels b and d show the levels of expression of the recombinant proteins, all of which were expressed to similar levels in the different P. pastoris strains. The antibodies used to probe each Western blot are shown below each panel. B, region of two-dimensional 1 H-15 N HSQC spectra for 15 N-labeled TgMIC1-CT alone (black) and in the presence of unlabeled TgMIC6-EGF3 acid at the molar ratio 1:1 (red). Perturbed resonances are labeled. C, ribbon representations of TgMIC1-CT with residues colored in red, orange, and yellow for strong, medium, and weak perturbations upon TgMIC6-EGF3 acid binding, respectively. The surface is shown in the same orientation as Fig. 4E (left) and rotated by Ϯ90 o (right). D, two-dimensional 1 H-15 N HSQC spectra for 15 N labeled TgMIC6-EGF3 acid (left) and in the presence of unlabeled TgMIC1-CT at the molar ratio 1:1 (right). Examples of key resonances that shift away from the unfolded envelope are highlighted by green shading. Unstructured resonances remaining in the complex presumably originate from the large hydrophilic acidic domain (Fig. 1) rounding loops, is not present in the TgMIC1-CT fold (Fig. 4D). In stark contrast, TgMIC1-CT presents a predominantly hydrophobic environment in this region that is more reminiscent of a protein-protein interaction interface (Fig. 4E). In the light of literature reporting a lactose lectin activity of TgMIC1 (23) and the similarity of TgMIC1-CT with the galectin family, the lactose binding activity of TgMIC1-CT was monitored using NMR chemical shift mapping experiments. No detectable binding could be observed up to several hundred molar equivalents of lactose (100 mM), which substantiates the observation of an altered binding site. Furthermore, similar experiments revealed no interaction with glucose, maltose, mannose, heparin, fucose, L-arabinose, N-acetyl-D-galactosamine, and N-Acetyl-D-glucosamine (data not shown).
The absence of a traditional carbohydrate-binding site and its replacement with a large, hydrophobic surface (Fig. 4) that resembles a protein-protein interaction interface is intriguing. Moreover, it is loosely reminiscent of the substrate recognition by certain chaperones, for example the class I chaperones from bacterial type III secretion systems in which exposed hydrophobic grooves provide templates for the trapping of partially folded states within effector molecules (24). Indeed, NMR and biochemical experiments confirmed that not only does the Galectin domain recruit TgMIC6, but it assists in folding and stabilizing the C-terminal fragment encompassing the third EGF domain and the acidic region, TgMIC6-EGF3 acid . We also show that the Galectin-like domain is able to prevent TgMIC6 from being retained in the ER/Golgi of mic1ko parasites and rescue its transport to the microneme. The Galectin-like domain presumably interacts with and assists the folding of TgMIC6, thereby allowing the complex to pass the ER quality control checkpoint, exit, and continue its journey to the microneme via the targeting information within the C terminus of TgMIC6 (25). A similar role in the quality control checkpoint has been postulated for another soluble microneme protein, TgM2AP, which interacts with the transmembrane protein TgMIC2. Disruption of the TgM2AP gene also leads to the retention of TgMIC2 in the early compartments of the secretory pathway and dramatically alters the parasite ability to invade (5,11).
In summary, TgMIC1 plays a surveillance role that ensures neither TgMIC4 nor TgMIC6 progress through the early secretory pathways without the proper formation of folded, heterotrimeric complexes, which can successfully reach the micronemes where they perform their function.