Infection by Trypanosoma cruzi. Identification of a parasite ligand and its host cell receptor.

The infective trypomastigote stage of Trypanosoma cruzi expresses a set of surface glycoproteins that are known collectively as Tc85 and belong to the gp85/trans-sialidase supergene family. A member of this family, Tc85-11, with adhesive properties to laminin and cell surfaces was recently cloned. In this report, the Tc85-11 domain for cell binding and its corresponding receptor on epithelial cell LLC-MK(2) are described. Using synthetic peptides corresponding to the Tc85-11 carboxyl-terminal segment, we show that the mammalian cell-binding domain colocalizes to the most conserved motif of the trypanosome gp85/trans-sialidase supergene family (VTVXNVFLYNR). Even though Tc85-11 binds to laminin, the 19-residue cell-binding peptide (peptide J) does not contain the laminin-binding site, because it does not bind to laminin or inhibit cell binding to this glycoprotein. The host cell receptor for the peptide was characterized as cytokeratin 18. Addition of anti-cytokeratin antibodies to the culture medium significantly inhibited the infection of epithelial cells by T. cruzi. Tc85-11 is a multiadhesive glycoprotein, encoding at least two different binding sites, one for laminin and one for cytokeratin 18, that allow the parasite to overcome the barriers imposed by cell membranes, extracellular matrices, and basal laminae to reach the definitive host cell. This is the first description of a direct interaction between cytokeratin and a protozoan parasite.

entry is the ocular conjunctiva that is put in contact with contaminated insect feces by involuntary scratching from nearby bites on a sleeping person's face, leading to a periorbital swelling known as Romañ a's sign. Other forms of transmission such as blood transfusion, congenital transmission, and breast feeding are also important, particularly in northern hemisphere regions that received intense migratory currents from Ibero-American countries. In recent years, 85-90-kDa parasite surface proteins have been implicated in host cell invasion by different investigators (2)(3)(4)(5)(6)(7). Our laboratory was the first to describe trypomastigote-specific 85-kDa surface glycoproteins, suggesting their role in host cell invasion by the parasite (6 -10). These proteins, collectively denominated Tc85, form a population of heterogeneous glycosylphosphatidylinositol-anchored surface glycoproteins with similar molecular masses but different electric charges (7)(8)11). Tc85 proteins belong to the gp85/trans-sialidase gene superfamily (12) and share common motifs with bacterial neuraminidases (1,(12)(13). Interestingly, all members of the superfamily contain a conserved sequence (VTVXNVFLYNR) (12) upstream from the carboxyl terminus and absent in bacterial neuraminidases. The involvement of at least one member of the Tc85 family in parasite-host cell interactions is indicated by the observation that the monoclonal antibody H1A10, which specifically recognizes Tc85 glycoproteins, inhibits host cell invasion by the parasite in vitro by 50 -90% (6,7). An acidic 786-amino acid member of the Tc85 family (Tc85-11) and a recombinant fusion protein of the monoclonal antibody H1A10 epitope-containing carboxyl-terminal segment of Tc85-11 (Tc85-1) both showed adhesive properties to isolated laminin and to entire cells (10).
The high plasticity of the cytoskeleton is often exploited by pathogens to enter non-phagocytic cells. Increasing evidence has been provided for the expression of cytoskeletal proteins on cell surfaces that serve as receptors for different ligands. For example, intermediate filament proteins belonging to the cytokeratin family are expressed on the cell surface and act as receptors for bacteria as well as for plasminogen and tissue plasminogen activator, high molecular weight kininogen, and thrombin-antithrombin complexes (14 -20).
The present work demonstrates that the conserved common sequence VTVXNVFLYNR of the gp85 glycoprotein/trans-sialidase supergene family is a mammalian cell-binding domain. Its host cell receptor for this motif was purified and characterized as cytokeratin 18 (CK18) 1 present on the surface of LLC-MK 2 cells (monkey kidney epithelial cells). Because Tc85 also binds to laminin (10), the results presented herein suggest that the Tc85 family is composed of multiadhesive glycoproteins that bind to different receptor molecules either located on the cell surface or belonging to components of the extracellular matrix.

EXPERIMENTAL PROCEDURES
Parasite Strain and Culture-T. cruzi strain Y was used throughout. Culture conditions for parasites and mammalian cells are described elsewhere (21).
Peptide Synthesis-Peptides were synthesized in an automated bench top simultaneous multiple solid phase peptide synthesizer (PSSM 8 system from Shimadzu) using the Fmoc (N-(9-fluorenyl)methoxycarbonyl) procedure. The synthesized peptides were deprotected and purified by semipreparative HPLC using an Econosil C-18 column (10 m, 22.5 ϫ 250 mm) and a two-solvent system: (A) trifluoroacetic acid/H 2 O (1:1000) and (B) trifluoroacetic acid/MeCN/H 2 O (1:900:100). The peptides were separated at a flow rate of 5 ml/min and a gradient from 10 (or 30) to 50 (or 60)% of solvent B. Analytical HPLC was performed using a binary HPLC system (Shimadzu) with an SPD-10AV Shimadzu UV-visible detector and a Shimadzu RF-535 fluorescence detector, coupled to an Ultrasphere C-18 column (5 m, 4.6 ϫ 150 mm), which was eluted with solvent systems A1 (H 3 PO 4 /H 2 O, 1:1000) and B1 (MeCN/H 2 O/H 3 PO 4 , 900:100:1) at a flow rate of 1.7 ml/min and a 10 -80% gradient of B1 over 15 min. The HPLC column eluates were monitored by their absorbance at 220 nm and by fluorescence emission at 420 nm following excitation at 320 nm. The purity of obtained peptides was checked by matrix-assisted laser desorption ionizationtime of flight (MALDI-TOF) spectroscopy in the reflectron mode (Tof-Spec-E from Micromass, Manchester, UK) and by amino acid sequencing, performed with a Shimadzu sequencer, model PPSQ-23 (22).
Binding of Cells to Synthetic Peptides-In a 24-well plate, 40 g of each peptide in 200 l of 10% Me 2 SO were dried overnight at 37°C with agitation, washed with PBS (140 mM NaCl, 2.7 mM KCl, 10 mM phosphate buffer, pH 7.3), and incubated for 2 h with 1% BSA/PBS. LLC-MK 2 cells were cultured as described (6) in a 75-cm 2 bottle, removed by trypsin, and resuspended in 5 ml of DME medium supplemented with 10% FCS. The cells were incubated for 1 h at 37°C in 50-ml polyethylene tubes (Corning) and washed twice with DME medium to remove FCS. Then, 1 ϫ 10 5 cells in 0.5 ml of DME medium were added to the peptide-coated wells and incubated for 1 h at 37°C. The wells were washed three times with DME medium and analyzed using an inverted microscope. In binding competition assays, LLC-MK 2 cells were preincubated for 15 min with the peptide acting as a competitor and added to peptide-coated wells. After incubation and washing as described, the number of bound cells was quantified following staining with crystal violet (23).
Binding of 125 I-Peptide J to Mammalian Cells-Peptide J was radiolabeled with 125 I (Amersham Pharmacia Biotech) using the chloramine-T method (24) and purified by reverse-phase HPLC, resulting in a specific activity of 2 ϫ 10 7 cpm g Ϫ1 . LLC-MK 2 cells were collected as described above, and 6 ϫ 10 5 cells were incubated for 2 h on ice in the presence of increasing concentrations of the radiolabeled peptide. Nonspecific binding was determined in the presence of a 100-fold excess of unlabeled peptide J. The reaction mixtures were washed three times to remove unbound radiolabeled ligand, and cells were lysed in 1% SDS for 10 min and directly assayed for radioactivity by scintillation counting. Each experiment was performed in triplicate.
Isolation and Biotinylation of Cell Surface Proteins-LLC-MK 2 and K562 cells were collected as described, washed three times with PBS, and biotinylated with the EZ-Link-Sulfo-NHS-biotinylation kit (Pierce) as recommended by the manufacturer. Plasma membranes were prepared (25) and solubilized in 100 mM ␤-D-n-octyl glucoside for 2 h at 4°C.
Affinity Chromatography-Peptide J was synthesized with an additional cysteine at the amino terminus. One mg of peptide J was coupled to a solid matrix (UltraLink TM iodoacetyl, Pierce) and used for affinity chromatography experiments. The supernatants from solubilizations in ␤-D-n-octyl glucoside were incubated overnight with the peptide J affinity gel at 4°C with agitation. The gel was loaded into a column, and the column was washed with 40 volumes of 25 mM ␤-D-n-octyl glucoside in incubation buffer. The gel was then washed with 1 M NaCl, followed by agitation for 1 h at room temperature. The column was washed again with 40 column volumes of PBS and then incubated with 8 M urea as above. The collected fractions were dialyzed, concentrated, and analyzed by SDS-PAGE (26) in 9% gels.
Western Blot-Following analysis by SDS-PAGE, proteins were transferred to a supported nitrocellulose membrane using 25 mM Tris, 150 mM glycine, and 20% methanol (pH 8.3) as transfer buffer. The blots were blocked with 3% BSA in TBSTT (Tris-buffered saline (TBS; 10 mM Tris, pH 7.5, 150 mM NaCl) containing 0.05% Tween 20 and 0.03% Triton X-100) and incubated for 2 h at room temperature with ExtrAvidin-peroxidase (Sigma) or anti-PAN-cytokeratin antibody (Sigma), as recommended by the manufacturer. The latter recognizes cytokeratins 4, 5, 6, 8, 10, 13, and 18. The membrane was washed with TBSTT and, when necessary, incubated with the secondary antibody conjugated to peroxidase. The reaction was developed with the ECL kit (Amersham Pharmacia Biotech).
Cross-linking of 125 I-Peptide J with Cell Surfaces-Membrane fractions from LLC-MK 2 and K562 cells were prepared as described (25) and incubated for 2 h on ice with 125 I-peptide J in the presence and absence of a 100-fold excess of unlabeled peptide. The receptor-ligand complexes were separated from unbound ligands by centrifugation and washed once with PBS, 0.01% Triton X-100. Chemical cross-linking was initiated by addition of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide, 100 mM final concentration (Sigma). After 30 min of incubation on ice the reaction was stopped with 150 mM glycine, pH 6.8. Samples were then washed twice with PBS and solubilized for 2 h at 4°C in PBS containing 100 mM ␤-D-n-octyl glucoside, 1 mM EDTA, 2 g/ml aprotinin, 1 mM N-␣-p-tosyl-L-lysine chloromethyl ketone, and 1 mM phenylmethylsulfonyl fluoride. The proteins were separated by SDS-PAGE, and 125 I-peptide J-protein complexes were detected by autoradiography.
Binding of 125 I-CK18 to T. cruzi-Trypomastigotes were washed twice with DME medium and purified by centrifugation over a Lymphoprep gradient (Nycomed Pharma AS) to eliminate contaminating host cells and cell debris. CK18 (Research Diagnostics) was radioiodinated using the chloramine-T method (24) and purified on a G-25 Sepharose column. Purified trypomastigotes (9 ϫ 10 6 ) in a total volume of 200 l were incubated with 1 ϫ 10 6 cpm of 125 I-CK18 (8.6 ϫ 10 5 cpm g Ϫ1 ) in the presence and absence of a 20-fold excess of unlabeled CK18 for 2 h on ice. The incubation mixtures were separated by filtration over nitrocellulose filters previously saturated with 0.1% BSA, and the filterbound radioactivity was quantified using scintillation counting.
Infection of Mammalian Cells by Trypomastigotes-T. cruzi trypomastigotes were cultured as described (6). LLC-MK 2 cells grown in an 8-well dish were washed three times with DME medium, and each well was incubated for 15 min in DME medium supplemented with 2% FCS containing 200 M alanine-substituted peptide (peptide J-Ala), 100 or 200 M peptide J, 10 M Tc85-11, mouse IgG (Sigma), or a 1:20 dilution of an anti-CK18 antibody (Research Diagnostics). These cells were infected with trypomastigotes in a 1:100 ratio for 2 h at 37°C and washed twice with PBS. The nonadherent parasites were removed by addition of Lymphoprep to the cell layers, followed by two washes with PBS. The cells were incubated with DME medium supplemented with 2% FCS for 48 h at 37°C, fixed with 100% methanol, and stained with chromomycin A (Molecular Probes). All experiments were performed in triplicate, and 12 photos of each replicate were made using a digital video-imaging fluorescence microscope, enabling the counting of infected and non-infected cells in samples containing 200 cells each. The data were compared for statistical significance using the unpaired Student's t test.
Immunofluorescence Microscopy-LLC-MK 2 cells were adhered to peptide J or FCS-coated 30-mm glass coverslips as described above and then fixed in 4% paraformaldehyde. Cells were washed with PBS and incubated for 30 min at 37°C with either anti-PAN-cytokeratin antibody (Sigma) or specific anti-CK18 antibody (Research Diagnostics), diluted as recommended by the manufacturers, washed five times with PBS, and then incubated under the same conditions with fluorescein isothiocyanate-labeled goat anti-mouse IgG (1:30 dilution) and rinsed five times. In experiments where K562 cells, which do not bind to peptide J, were used, cells were washed by centrifugation at 1,000 ϫ g. The same protocol was used for saponin-treated cells, except that 0.05% saponin was added during incubation with the antibodies.
Characterization of CK18 -Fractions of the affinity column-purified and biotinylated 45 kDa-protein from LLC-MK 2 cells were determined to be cytokeratin 18 by mass spectrometry of tryptic peptides at the HHMI Biopolymer Laboratory and W. M. Keck Foundation Biotechnology Resource Laboratory at Yale University (New Haven, CT).

RESULTS
Definition of the Tc85-1 Cell-binding Site-To define the Tc85-1 cell-binding site, we synthesized 11 peptides with five amino acid overlaps that spanned the carboxyl-terminal segment of the recombinant Tc85-11 protein. These peptides were used to coat the surface of 24-well plates and mediate LLC-MK 2 cell adhesion. As shown in Fig. 1A, the cells only adhered significantly to the well coated with peptide J (19 amino acids) that contained the VTVTNVFLYNR motif. This motif is highly conserved and present in all members of the gp85/trans-sialidase superfamily. This observation implicates the importance of that common sequence in the binding of members of the gp85/trans-sialidase supergene family to their host cell receptors (Fig. 1B).
Other cell lines that are invaded by T. cruzi were tested for their affinity for peptide J. In addition to LLC-MK 2 cells, tumor cells (B16F10), human umbilical cord endothelial cells (ECV), macrophage-like cells (J774), mouse fibroblast cells (3T3), and mouse pheochromocytoma cells (PC-12) bound to peptide J (data not shown). Consistent with a crucial function of peptide J in cell invasion, mouse erythrocytes and K562 erythroleukemia cells (27) that are not invaded by T. cruzi did not bind to peptide J.
Additional evidence for the physiological relevance of peptide J binding to mammalian cells is that a 10 M concentration of this peptide inhibited the binding of the recombinant Tc85-11 protein to the host cell. As opposed to Tc85-11, peptide J does not bind to laminin; nor does it inhibit cell binding to laminin. Furthermore, it was established that peptide J does not bind to cells at the same receptor used by laminin or to laminin on its cell-binding domain, because different concentrations of this glycoprotein did not affect LLC-MK 2 adhesion to peptide J. The combined data strongly suggest that the Tc85-11 recombinant protein is a molecule with multiple adhesion sites, specific for different ligands of the vertebrate host cell.
Radioiodinated peptide J binds to LLC-MK 2 cells in a specific, saturable manner, as shown by nonlinear saturation analysis (Fig. 2). The data suggest 1.66 Ϯ 0.16 ϫ 10 6 binding sites with a K D of 175 Ϯ 56 nM. The number of binding sites is comparable with that determined for plasminogen binding to CK8 (15). Interestingly, cytokeratin 8 associates with CK18 to form an intermediate filament heteropolymer in several cell types.
Mapping of the Amino Acids Required for the Binding of The most conserved sequence of the gp85/trans-sialidase superfamily binds to LLC-MK 2 cells. A, each well of a 24-well plate was coated with 40 g of peptides A-K corresponding to the 131 amino acids of the carboxyl terminus of Tc85-11 in the presence of BSA to diminish unspecific binding and then incubated with 1 ϫ 10 5 LLC-MK 2 cells in 0.5 ml of DME medium for 1 h at 37°C. The wells were washed three times with DME medium, and analysis by inverted microscopy revealed that cells adhered significantly only to the well coated with peptide J but not to wells coated with BSA or other peptides. The Peptide J to the Host Cell by Alanine Scanning-To identify the minimal sequence that is relevant for the binding of peptide J to the host cell, truncated peptides were constructed spanning the whole sequence of peptide J. Cells were layered on peptide J-coated plates, and truncated peptides were checked for their competing ability for cell adhesion. The minimal inhibitory sequence was VTNVFLYNRPL (data not shown). To identify the residues responsible for this binding, each amino acid of the minimal inhibitory sequence was consecutively substituted by alanine, and the modified sequences were tested for their inhibitory effects in adhesion assays of peptide J to cells. It was observed that substitution in some positions resulted in the loss of the inhibitory effect on the binding of cells to peptide J (Fig. 3). These experiments strongly indicate that the amino acid sequence VTXVFLYXR, conserved in most members of the 85-kDa trypomastigote surface glycoprotein family, is essential for parasite-cell interaction. In 40 analyzed sequences (Fig.  1B), LYXR was present in all members of the family, whereas the first Val residue that is found in 80% of the sequences was substituted in the remaining sequences by Leu or Ala, which are also apolar amino acids. Threonine, at position 2, showed a smaller degree of conservation (38%), most often being replaced by other polar residues: Ser (15%), Asn (20%), and Lys (15%). The valine at position 4 is again highly conserved (95%), and Leu substitutes Phe at position 5 in 38% of the molecules.
Identification of the Host Cell Receptor for the Truncated Common Sequence of Tc85-To characterize the host cell receptor for peptide J, the peptide was coupled to an affinity matrix and used for purification of the receptor, employing chromatographic methods. The affinity matrix was incubated with solubilized membranes from biotinylated LLC-MK 2 cells. The 8 M urea eluates of the LLC-MK 2 cell extracts revealed a biotinylated 45-kDa molecule that was detected by Western blot (Fig. 4).
To obtain further evidence that the 45-kDa molecule is the host cell receptor, radioiodinated peptide J was chemically cross-linked with LLC-MK 2 cells. For a negative control, we also performed the experiment using K562 cells, which were neither infected by T. cruzi nor adhered to surfaces coated with peptide J. As an additional control, an alanine-substituted peptide, peptide J-Ala (VTNVFAYNRPL), that does not inhibit cell binding to peptide J was radiolabeled and cross-linked to LLC-MK 2 cells. Following solubilization and separation of plasma membrane proteins by SDS-PAGE, a protein migrating with a molecular mass of 45 KDa was detected only in the LLC-MK 2 cell extract (Fig. 5). The labeling of the 45-kDa protein was specific, because it could be inhibited by a 100-fold molar excess of unlabeled peptide. As expected, no specific labeling of K562 cells by 125 I-peptide J was observed, and the peptide J-Ala did not bind to LLC-MK 2 cells. These results strongly suggest that a 45-kDa molecule present on LLC-MK 2 cells is involved in adhesion of the parasite to these cells.

Cytokeratin 18 Is a Host Cell-binding Site for the Most Conserved Domain of the Tc85
Family-Purified 45-kDa biotinylated protein fractions, as described in Fig. 4, were digested with trypsin, and the peptides were analyzed by mass spectrometry. The identified peptides from three independent experiments indicated that the 45-kDa molecule was biotinylated CK18. In agreement with the mass spectrometry analysis, the isolated protein comigrated with authentic cytokeratin 18 with an apparent molecular mass of 45 kDa and a pI of 5.4 in a two-dimensional SDS-PAGE (data not shown). To further confirm these results, LLC-MK 2 and K562 plasma membrane extracts were incubated with the peptide J affinity column, and after elution with 1 M NaCl and 8 M urea, the eluates were analyzed by SDS-PAGE and tested by Western blot with anti-PAN-cytokeratin antibody. As shown in Fig. 6A, a cytokeratin molecule of 45 kDa is present only in 8 M eluates of LLC-MK 2 cells. As expected, no K562 cytokeratin could be eluted from the peptide J column. Furthermore, 125 I-CK18 bound to the peptide J affinity column and showed the same elution pattern as CK18 from LLC-MK 2 cells (Fig. 6B). Control BSA columns did not bind CK18. The fact that the anti-PAN-cytokeratin antibody, which recognized many proteins in the cell extract, was able to recognize only CK18 in the column eluate suggests a highly specific binding.

CK18 Is Present on the Surface of Intact LLC-MK 2 Cells and
Binds to Trypomastigotes-Intact LLC-MK 2 and K562 cells were tested for the presence of cytokeratin by immunofluorescence microscopy with fluorescent anti-CK18-specific antibody (Fig. 7). Whereas CK18 is present in the cytoplasm of both cell lines, only LLC-MK 2 cells express cytokeratin on the surface. Moreover, 125 I-labeled CK18 binds in a specific manner to trypomastigotes (Fig. 8) but not to epimastigotes, the noninvasive developmental form of T. cruzi (data not shown).
CK18 and Anti-CK18 Antibody Inhibit Binding of LLC-MK 2 Cells to Peptide J-As shown in Fig. 9, previous incubation of peptide J-coated wells with CK18 (100 g/ml) completely inhibited cell adhesion to the wells. Furthermore, previous incubation of LLC-MK 2 cells with anti-CK18 antibody inhibited cell adhesion to peptide J by 75%. Addition of 100 M peptide J completely inhibited the binding, whereas 100 M peptide J-Ala had no effect.

Effects of Peptide J and CK18 on Host Cell Infection by T. cruzi in Vitro-
The effects of peptide J, Tc85-11, and CK18 on the invasion of LLC-MK 2 cells by trypomastigote forms were analyzed by invasion assays in the presence of these molecules. The data show statistically significant (p Ͻ 0.05) differences among the number of infected cells in the absence and presence of peptide J, recombinant Tc85-11, and anti-CK18 antibody and when both peptide J and anti-CK18 antibodies were added simultaneously (Fig. 10). Thus previous incubation of LLC-MK 2 cells with peptide J and Tc85-11 increases cell invasion by T. cruzi, whereas the anti-CK antibody inhibits invasion by more than 60%. The effect of peptide J on cell invasion depends on the concentration used, suggesting a role for the conserved sequence in Tc-85 as a signaling molecule that will prepare the epithelial host cell for parasite invasion. When host cells were previously incubated with anti-CK18 antibody, the increase in infection promoted by peptide J could not be observed (Fig. 10). As controls, invasion assays were performed in the presence of mouse IgG and peptide J-Ala, both of which had no effect on invasion.

DISCUSSION
T. cruzi invades non-phagocytic cells in an energy-dependent manner (28) by a mechanism different from phagocytosis. Invasion is preceded by an adhesion step involving surface molecules from both the parasite and the host cell. Members of the Tc85 glycoprotein family, expressed on the surface of the infectious form of T. cruzi, were first suggested to be involved in the adhesion step necessary for parasite invasion of host cells (8,11,29) because monoclonal antibodies against Tc85 molecules were able to partially inhibit host cell invasion (6), and a cloned member of the Tc85 family (Tc85-11) was shown to bind to laminin (10). Other laboratories have also implicated the involvement of several ϳ85-kDa proteins in parasite-host cell interaction (30 -32), including an 85-kDa protein, probably from the same superfamily but different from Tc85-11, that was described as a fibronectin receptor (33). The Tc-85 family of glycoproteins belong to the gp85/trans-sialidase supergene family, which comprises ϳ1,000 genes (34). The whole super-family represents 1-2% of the T. cruzi genome, with highly redundant and simultaneously expressed members, a stumbling block that eliminates the possibility of employing genetic approaches to their functional analysis. It is our working hypothesis that the gp85/trans-sialidase gene family, in addition to members coding for trans-sialidase activity, comprises a family coding for adhesion proteins, with several of its members interacting with specific ligands. It is worth noting that the 3.5-h half-life of the Tc85 family is considerably short (35). This fast turnover could facilitate the progression of the parasite from blood vessels to the cells if different continuously expressed subsets of the family bound to different ligands on cell surfaces, extracellular matrices, and basal laminae. It would seem likely that cruzipain (36,37) and other proteases (38) could be operative in breaking the successive proteinprotein interactions, thus facilitating parasite progression. An additional, but not exclusive, possibility is that individual members of the Tc85 family may interact with two or more ligands on the cell surface.
The latter hypothesis is favored by our data, because Tc85-11 has two different binding sites. One of these sites for laminin binding is under investigation and is known to be located somewhere in the 100 amino acids upstream from the peptide J conserved region of the Tc85-1 clone. The other binding site to CK18 comprises the most conserved region of all members of the gp85/trans-sialidase family. This is the first time that multiple adhesion sites on a T. cruzi molecule, a common feature in other systems (39,40), have been defined.
The discovery that the most conserved sequence of the gp85/ trans-sialidase family, the trypomastigote adhesion peptide TXVFLYXR, is a CK18-binding domain for T. cruzi adhesion on vertebrate host cells expands our understanding of the function of this protein family in the molecular mechanisms of cell invasion. Our findings may also contribute to the unraveling of the yet unknown function of trans-sialidases in invasion, because antibodies against the catalytic site of this enzyme, which is distant from the conserved peptide J sequence, did not inhibit parasite internalization into host cells (41). The search for similar motifs in other pathogen-host cell interactions unrelated to T. cruzi infection may elucidate our general understanding of cell infection.
Primary infection of human beings by T. cruzi naturally occurs through skin lesions from the insect bite, by direct contact of contaminated feces with the dermal layer, or by contact with epithelial and endothelial tissue. Our work with epithelial cells suggests that the interaction of T. cruzi with cytokeratin may be important for the parasite to cross the eye epithelial mucosa (as an early event leading to the typical Romañ a's sign) or the trophoblast epithelium (explaining the congenital transmission of Chagas' disease).
CK18 associates with CK8 to form a component of intermediate filaments in simple epithelia and many epithelial cellderived neoplasms. CK18 can be aberrantly expressed in many non-epithelial cancers, including lymphomas, melanomas, gliomas, and sarcomas, and in many cases, CK18 has been correlated with increased tissue invasion in vitro and in vivo (14). Additionally, recent data indicate the presence of CK18 and CK7 in endothelia of normal veins, venules, lymphatics, and capillaries in the skin; subcutaneous soft tissues; mucosal sites; skeletal muscle and smooth muscle cells in the placenta (42); and the synovial microvasculature (43). This is consistent with the hypothesis that trypanosomes may bind to CK18 to traverse the endothelial barrier.
The view that cytokeratins and other intracellular proteins are confined solely to the cytosol has recently undergone revision. The fact that various intracellular proteins are also expressed on the cell surface and there exercise specific functions is now well established (14 -20, 44). CK1 was described as a putative kininogen-binding protein on the surface of endothelial cells (18), CK13 binds to cable-piliated Burkholderia cepacia (19), and CK8 is a plasminogen and tissue plasminogen-activator receptor on the surface of hepatocytes and breast cancer cells (14 -16) as well as a receptor for group B streptococci and other Gram-positive cocci (20). In particular, CK18 was identified as the binding site for thrombin-antithrombin III complexes on the plasma membrane of rabbit hepatocytes (17).
In summary, the results herein presented favor the hypothesis that the conserved common sequence of the gp85/transialidase family is an important docking domain to the host cell surface, although other sites should not be ruled out (10,27). Along with growing evidence for surface expression of intracellular proteins and the expression of cytokeratins also in epithelial and other tissues, our data indicate CK18 as a putative mammalian cell receptor for T. cruzi and/or a binding protein that is necessary for further receptor activation. Other studies have shown that transforming growth factor ␤ receptors are required for the infection of mammalian cells by T. cruzi (45), but the parasite ligand is unknown. Interestingly, cells treated with epidermal growth factor plus transforming growth factor ␤ express higher levels of CK18 (46). The fact that CK18 can be phosphorylated (47) suggests an involvement of cytokeratin in the intracellular signaling induced by T. cruzi (48).
Previous incubation of LLC-MK 2 cells with peptide J or Tc85-11 increases cell invasion by T. cruzi, suggesting a role for the gp85/trans-sialidase family as signaling molecules that enhance receptiveness of the host cell for the parasite by a yet unknown mechanism. Members of this family possessing a relatively short half-life (35) are constitutively shed into culture medium (49), suggesting that contact between the surface of the parasite or shed Tc85 proteins and CK18 on the mammalian cell may promote signaling events in the host cell, thus facilitating T. cruzi infection.
T. cruzi internalization requires host cell lysosome recruitment, inducing localized clustering and fusion of host cell lyso-somes with the plasma membrane at the site of trypomastigote attachment. This process requires host cell [Ca 2ϩ ] i transients and transient rearrangement of actin microfilaments, which might facilitate lysosome access to the plasma membrane during parasite invasion (48). Because other filaments are connected to actin microfilaments in the cytosol, it is possible that parasite binding to CK18 may also influence the lysosome migration process. It is worth noting that thrombin-anti-thrombin complexes, which are internalized via the CK18 receptor on the surface of hepatocytes, are degraded by lysosomes (17).
T. cruzi infection is a complex process involving several host and parasite molecules in the recognition process as well as the involvement of enzymatic reactions and bivalent ions. The present study indicates a new and physiologically relevant role for the most conserved sequence of the gp85/trans-sialidase super gene family. We have shown that this sequence is involved in host cell binding during the infection process and that CK18 is a putative trypomastigote receptor on epithelial cells.