37-kDa laminin receptor precursor modulates cytotoxic necrotizing factor 1-mediated RhoA activation and bacterial uptake.

Cytotoxic necrotizing factor 1 (CNF1) is a bacterial toxin known to activate Rho GTPases and induce host cell cytoskeleton rearrangements. The constitutive activation of Rho GTPases by CNF1 is shown to enhance bacterial uptake in epithelial cells and human brain microvascular endothelial cells. However, it is unknown how exogenous CNF1 exhibits such phenotypes in eukaryotic cells. Here, we identified 37-kDa laminin receptor precursor (LRP) as the receptor for CNF1 from screening the cDNA library of human brain microvascular endothelial cells by the yeast two-hybrid system using the N-terminal domain of CNF1 as bait. CNF1-mediated RhoA activation and bacterial uptake were inhibited by exogenous LRP or LRP antisense oligodeoxynucleotides, whereas they were increased in LRP-overexpressing cells. These findings indicate that the CNF1 interaction with LRP is the initial step required for CNF1-mediated RhoA activation and bacterial uptake in eukaryotic cells.

Escherichia coli K1 is a major cause of neonatal Gramnegative bacillary meningitis. Despite advances in antimicrobial chemotherapy and supportive care, the mortality and morbidity associated with E. coli meningitis remain significant because of incomplete understanding of the pathogenesis of this disease. We have previously shown that E. coli invasion of human brain microvascular endothelial cells (HBMEC) 1 is a prerequisite for penetration into the central nervous system in vivo, and identified several E. coli determinants contributing to invasion of HBMEC, including Ibe proteins, AslA, TraJ, and CNF1 (1)(2)(3)(4)(5)(6). We have also demonstrated that E. coli invasion of HBMEC requires host cell actin cytoskeleton rearrangements and activations of RhoA (7,8). CNF1, a bacterial toxin known to induce host cell cytoskeleton rearrangements, activates Rho GTPases such as RhoA, Cdc42, and Rac1 (9, 10), which regulate various cellular processes involving actin filaments. The con-stitutive activation of Rho GTPases by CNF1 has been shown to induce stress fiber formation, membrane ruffling, and phagocytosis in epithelial cells (11)(12)(13). We have shown that CNF1 contributes to E. coli invasion into HBMEC, in part through activation of RhoA (8).
CNF1 is a 113-kDa single chain toxin molecule that consists of a N-terminal cell surface receptor binding domain, a Cterminal catalytic domain, and a transmembrane domain in the middle (14). After translocation into the cytosol, the enzymatic domain of CNF1 activates Rho GTPases by deamidation of glutamine 63 of RhoA (9,10), or glutamine 61 in Rac1 and Cdc42 into glutamic acid (11). The glutamine residue is essential for GTP hydrolysis, and its modification results in constitutively activated Rho GTPases by arresting the Rho GTPases cycle between the GDP-bound inactive and GTP-bound active forms (14). However, it is unclear how CNF1 enters the eukaryotic cells and activates Rho GTPases. CNF1 has been suggested to be internalized via receptor-mediated endocytosis upon binding to a cell surface receptor (15,16), but the identity and characteristics of the CNF1 receptor are not known.
In this study, for the purpose of identifying CNF1 receptor, the cDNA library of HBMEC was constructed and screened by the yeast two-hybrid system using the N terminus of CNF1 (nCNF1) as bait. We identified for the first time that 37-kDa laminin receptor precursor (LRP) functions as the receptor for CNF1. Exogenous LRP or LRP antisense oligodeoxynucleotides (ODN) inhibited CNF1-mediated RhoA activation and bacterial uptake, whereas overexpression of LRP increased binding and effects of CNF1.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Antibodies-E. coli K1 strain E44 is a spontaneous rifampin-resistant mutant derived from the cerebrospinal fluid isolate of a neonate with meningitis, strain RS218 (serotype O18:K1:H7). The isogenic cnf1 deletion mutant of strain E44 was previously reported (8). E. coli BL21 (Invitrogen, Carlsbad, CA) or XL1-Blue (Stratagene, Ceda Creek, TX) was used for expression of the recombinant proteins. E. coli DH10B (Invitrogen) was used for the construction of the cDNA library. Plasmids pGBKT7 and pGADT7 were purchased from Clontech (Palo Alto, CA), and used for construction of the bait CNF1 and prey cDNA libraries, respectively. Anti-LRP polyclonal antibody was purchased from Abcam (Cambridgeshire, UK). Anti-CNF1 monoclonal antibody NG8 (17) was a generous gift from Dr. Allison O'Brien (Uniformed Services University of the Health Sciences, Bethesda, MD). Plasmid pGEX-2T (Amersham Biosciences) was used for construction of GST fusion proteins.
cDNA Construction and Yeast Two-hybrid Screen-For bait construction, the N-terminal segment of cnf1 (nCNF1) was amplified by PCR from E44 chromosomal DNA using primers, 5Ј-aagaattcatgggtaaccatggc-3Ј and 5Ј-acggatccattgctaagtgtcttattgg-3Ј. The amplified fragments (amino acids 1-299) were cloned into pGBKT7 vector via EcoRI and BamHI sites (underlined) to create the bait plasmid pGBKT7:nCNF1. Full-length CNF1 (fCNF1, amino acids 1-1007) and the C-terminal * This work was supported by grants from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In Vitro Translation of Recombinant Proteins and Ligand Overlay Assay-The N-terminal domain of CNF1 (nCNF1), C-terminal domain of CNF1 (cCNF1), and full-length CNF1 (fCNF1) were in vitro translated from pGBKT7:nCNF1, pGBKT7:cCNF1, and pGBKT7:fCNF1, respectively, using TNT quick coupled transcription/translation systems (Promega, Madison, WI) according to the manufacturer's protocol. pGADT7:LRP-(89 -295) was used for in vitro translation of LRP-(89 -295) following the same protocol. For ligand overlay assay, HBMEC membrane proteins or the bacterial lysates from E. coli BL21 containing pGADT7:LRP-(89 -295) were separated in a SDS-PAGE gel and transferred to a PVDF membrane. After washing, the membrane was blocked overnight with 5% skim milk in phosphate-buffered saline at 4°C, and incubated with in vitro translated [ 35 S]CNF1 for 1 h at room temperature. After extensive washing, radiolabeled CNF1 was detected with a Cyclone PhosphorImager (Packard, Boston, MA).
Co-immunoprecipitation and Western Blot-30 g of total protein of bacterial lysates from E. coli K1 strain E44 or its isogenic cnf1 deletion mutant (E44⌬CNF1) was combined with 10 l of in vitro translated [ 35 S]LRP-(89 -295), and the mixture was incubated at 37°C for 1 h. After incubation, the mixture was precipitated with anti-CNF1 monoclonal antibody NG8 in the presence of protein A-Sepharose (Roche Diagnostics, Indianapolis, IN) in 0.5 ml of co-immunoprecipitation buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM dithiothreitol, 5 g/ml aprotinin, 0.5 mM phenylmethylsulfonyl fluoride, 0.1% Tween 20 (v/v), and 1% Nonidet P-40) at 4°C for overnight. After washing, immunoprecipitates were separated by SDS-PAGE in duplicate. One set was dried and subjected to scan by PhosphorImager and the other was transferred to PVDF membrane, blocked, and incubated with anti-CNF1 monoclonal antibody NG8. The membrane was washed and incubated with horseradish peroxidase-conjugated secondary antibody. Protein detection was performed with the enhanced chemiluminescence (ECL) system (Amersham Biosciences).
Construction and Purification of Recombinant LRP-For construction of GST-LRP, full-length LRP was amplified from the HBMEC cDNA library by PCR using primers, 5Ј-acggatccatgtccggagcccttgat-3Ј and 5Ј-aagaattcttaagaccagtcagtggtt-3Ј. For GST-CNF1, full-length CNF1 was PCR-amplified from E44 genomic DNA using primers 5Јacggatccatgggtaaccatggc-3Ј and 5Ј-aagaattccaataccgatatttcgg-3Ј. Amplified fragments were cloned into pGEX-2T (Amersham biosciences) via BamHI and EcoRI sites, and introduced into E. coli XL1-Blue. Expression and purification of the recombinant proteins were performed with a GST purification kit (Clontech) according to the manufacturer's protocol.
RhoA-GTP Assay-To examine the effects of LRP in CNF1-mediated RhoA activation, RhoA-GTP assays were performed as described previously (8) with the following modifications. Briefly, HBMEC or HEp-2 cells were lysed in lysis buffer (50 mM Tris-HCl, pH 7.4, 0.1% SDS, 0.5% sodium deoxycholate, 10 mM sodium pyrophosphate, 25 mM ␤-glycerophosphate, 150 mM NaCl, 2 mM EDTA, 2 mM EGTA, 1% Triton X-100, 1 mM Na 3 VO 4 , 50 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml leupeptin, and 1 g/ml pepstatin). Glutathione S-transferase-rhotekin beads (Upstate Biotechnologies, Inc., Lake Placid, NY) were incubated with cytosolic fractions of the cells for 1 h at 4°C by head over head rotation to collect active forms of RhoA, i.e. RhoA-GTP. After washing three times with lysis buffer, the protein complex was resolved by 12% SDS-PAGE and transferred to PVDF membrane. The blots were blocked with 25 mM Tris (pH 7.4), 150 mM NaCl, and 0.1% Tween 20 containing 4% skim milk for 30 min at room temperature. After blocking, the membranes were incubated overnight at 4°C with mouse monoclonal antibody against RhoA (Santa Cruz Biotechnology, Santa Cruz, CA) and subsequently incubated for 60 min at room temperature with horseradish peroxidase-linked secondary antibody against mouse. Antibody-bound RhoA was visualized using an ECL system (Amersham Biosciences). A single band of RhoA in the 21-kDa range was detected. The density of RhoA-GTP bands were quantitated using an imaging densitometer.
In Vitro HBMEC Invasion Assay-Invasion assays were performed as described previously (8) with the following modifications. Briefly, confluent HBMEC in 24-well plates were incubated with 10 7 E. coli (multiplicity of infection of 100) in experimental medium (M199-Ham's F-12 (1:1) containing 5% heat-inactivated fetal bovine serum, 2 mM glutamine, and 1 mM pyruvate). Plates were incubated at 37°C in 5% CO 2 incubator for 90 min. Monolayers were washed with RPMI 1640 and extracellular bacteria were killed by 1 h incubation in experimental media containing gentamicin (100 g/ml). The monolayers were washed again and lysed in 0.5% Triton X-100. The released intracellular bacteria were enumerated by plating on sheep blood agar plates. To determine the role of LRP on CNF1-induced E. coli K1 invasion, monolayers were incubated with a mixture of 3 g/ml CNF1 and 30 g/ml GST-LRP or GST in experimental medium for 2 h before addition of bacteria and HBMEC invasion assays were carried out as described above.
Enzyme-linked Immunosorbent Assay-Full-length LRP was amplified from the HBMEC cDNA library by PCR using primers, 5Ј-acggatccatgtccggagcccttgat-3Ј and 5Ј-aagaattcttaagaccagtcagtggtt-3Ј, and cloned into the vector pcDNA3.1/His A (Invitrogen) via BamHI and EcoRI sites (underlined) to create pcDNA3.1:LRP. HBMEC stably transfected with pcDNA3.1 or pcDNA3.1:LRP was seeded in 96-well plates. Upon confluence, the cells were washed with phosphate-buffered saline containing 0.05% Tween 20 and different concentrations of GST-CNF1 or no ligand in binding buffer (phosphate-buffered saline containing 0.05% Tween 20 and 2% bovine serum albumin) were added to each well. After 1 h incubation at room temperature, the ligand was aspirated and the cells were washed five times with binding buffer. After washing, anti-GST polyclonal antibody (Amersham Biosciences) was added, and the cells were further incubated for 1 h at room temperature. After washing, the cells were incubated for 1 h at room temperature with anti-goat IgG conjugated with alkaline phosphatase. To detect binding of GST-CNF1, p-nitrophenyl phosphate was added to each well and the color change was quantitated at 405 nm on a microplate spectrophotometer.
Antisense ODN Treatment-Based on the human LAMR1 cDNA sequences, antisense and sense phosphorothiate oligodeoxynucleotides, respectively, were designed against nucleotides 1-20. They were antisense, ACATCAAGGGCTCCGGACAT, and sense, ATGTCCGGAGC-CCTTGATGT. HBMEC were transfected with antisense and sense phosphorothiate ODNs with OligofectAMINE (Invitrogen) followed by manufacturer's protocol and 48 h was allowed for HBMEC cells to uptake the ODNs.

Identification of LRP as a CNF1 Receptor by the Yeast Twohybrid System and Coimmunoprecipitation-
In an attempt to identify the CNF1 receptor, we screened a cDNA library of HBMEC by the yeast two-hybrid system using the N terminus of CNF1 (nCNF1) as bait. We screened more than 2 ϫ 10 6 individual clones from a HBMEC cDNA library and identified nine positive clones at the highest stringency selection conditions (Trp-, Leu-, His-, Ade-). Sequence analysis revealed that one of the nine colonies contained a putative transmembrane domain, which showed a 100% match to the partial sequence (amino acids 89 -295) of LAMR1 encoding laminin receptor precursor (LRP). LRP is comprised of a cytoplasmic N-terminal domain (amino acids 1-85), and an extracellular C-terminal domain (amino acids 102-295) that are separated by a transmembrane domain (amino acids 86 -101) (19). This clone was further characterized for its interaction with CNF1 by retransformation. A yeast colony co-transformed with pGADT7:LRP-(89 -295) and pGBKT7:nCNF1 was positive in both growth on the selection medium (Leu-, Trp-, His-, Ade-) and 5-bromo-4chloro-3-indolyl-␤-D-galactopyranoside (X-gal) assays (Fig. 1a). Similarly, interaction between fCNF1 and LRP-(89 -295) in yeast was observed (Fig. 1a). In contrast, co-transformation with pGADT7:LRP-(89 -295) and pGBKT7:cCNF1 or pGBKT7 vector alone did not restore growth ability or enzyme activity (Fig. 1a), suggesting that the LRP-(89 -295) binds to full-length CNF1 and nCNF1, but not to cCNF1 in vivo.
Identification of LRP by Ligand Overlay Assays-In parallel with the yeast two-hybrid screening, a [ 35 S]CNF1 ligand overlay assay was performed to identify and confirm the cell surface receptor for CNF1. Membrane proteins of HBMEC were subjected to SDS-PAGE, and the overlay assay was performed. In this experiment, a protein of 37 kDa was identified as an interacting protein with either [ 35 S]nCNF1 or [ 35 S]fCNF1, but no membrane protein was detected with [ 35 S]cCNF1 (Fig. 2a). To verify the specific interactions between CNF1 and the putative receptor, we performed the overlay assay after preincubation of the membrane with non-radiolabeled nCNF1, which totally blocked binding of radiolabeled nCNF1 to the membrane (Fig. 2a). Because LRP was identified as a CNF1 receptor with yeast two-hybrid screening, we speculated that this 37-kDa protein might be LRP. We next recovered the prey vector containing tagged hemagglutinin (pGADT7:LRP-(89 -295)) from the yeast and introduced it into E. coli. Bacterial lysates containing HA-tagged LRP-(89 -295) were subjected to [ 35 S]nCNF1 ligand overlay assay to verify that CNF1 indeed binds to LRP-(89 -295) in vitro. [ 35 S]nCNF1 bound LRP-(89 -295) when the protein was expressed (Fig. 2b, left panel, lane  2), whereas bound [ 35 S]nCNF1 was not detected in the absence of LRP-(89 -295) (Fig. 2b, left panel, lane 1). Western blot using anti-hemagglutinin monoclonal antibody indicates that the protein detected with the [ 35 S]CNF1 overlay assay migrates at the same molecular weight as HA-tagged LRP-(89 -295) (Fig.  2b, right panel). These results taken together indicate that the extracellular domain of LRP binds to CNF1 and thus is likely to be the receptor that mediates the CNF1 biologic activity, such as RhoA activation and enhancement of bacterial uptake.
LRP Inhibits CNF1-induced RhoA Activation-We and others have previously shown that CNF1 activates RhoA in HB-MEC and other eukaryotic cells such as HEp-2 cells (8 -10). To determine the effect of LRP on CNF1-induced RhoA activation, HBMEC or HEp-2 cells were treated with CNF1 with or without GST-LRP, and total lysates were compared for the activated GTP-bound form of RhoA as described previously (8). The amount of RhoA-GTP was quantitated from the blot using an imaging densitometer, and normalized to the values of controls where the cells were treated with only GST. CNF1, as expected, increased RhoA activation in HBMEC and HEp-2 cells by 60 and 91%, respectively (Fig. 3). However, when CNF1 was added in the presence of GST-LRP, the levels of RhoA activation decreased by 44 and 50%, respectively, in HBMEC and HEp-2 cells when compared with CNF1-treated HBMEC or HEp-2 cells in the presence of GST. GST alone did not have any inhibitory effect on CNF1-induced RhoA activation (Fig. 3). These results suggest that exogenous LRP competes with endogenous LRP for CNF1 binding, and acts as an inhibitor of RhoA activation by CNF1 in both HBMEC and HEp-2 cells.
LRP Decreases CNF1-enhanced Bacterial Invasion-We have previously shown that CNF1 increases E. coli K1 invasion of HBMEC (8). We next examined whether exogenous LRP inhibited CNF1-induced enhancement of bacterial uptake. CNF1-treated cells showed an approximately 400% increase in E. coli E44 invasion compared with non-treated cells (Fig. 4). Consistent with the results obtained from RhoA-GTP assays where GST-LRP decreased CNF1-mediated RhoA activation in FIG. 1. Interaction of CNF1 and LRP. a, yeast two-hybrid system. Yeast cells were cotransformed with the prey plasmid pGADT7:LRP-(89 -295) and the bait plasmids pGBKT7, pGBKT7:cCNF1, pGBKT7: nCNF1, and pGBKT7:fCNF1. Strains were grown on selective media, diluted to A 600 of 0.5, and 10-fold serial dilutions were spotted onto plates (Leu-, Trp-) either with or without histidine (His) and adenine (Ade), which were incubated for 4 days at 30°C. b, co-immunoprecipitation. The 35   response to CNF1, the enhancement of E. coli uptake into HBMEC was significantly (p Ͻ 0.05) decreased when CNF1 was co-incubated with GST-LRP compared with CNF1 alone or with GST incubation (Fig. 4). These results suggest that GST-LRP effectively inhibits E. coli invasion by sequestering CNF1.
Binding and Effects of CNF1 in LRP-overexpressing HB-MEC-To confirm the binding of CNF1 to LRP in HBMEC, HBMECs were stably transfected with pcDNA3.1:LRP and the binding of GST-CNF1 to HBMEC was analyzed by enzymelinked immunosorbent assay. The binding of CNF1 to HBMEC was dose-dependent, but was markedly greater when LRP was overexpressed, compared with control vector-transfected cells (Fig. 5a). These results suggest a specific interaction between CNF1 and HBMEC via LRP, thus supporting the concept that LRP is the receptor for CNF1. In addition, CNF1-mediated bacterial uptake was significantly (p Ͻ 0.05) greater in LRP overexpressing cells (Fig. 5b). Also, CNF1-mediated RhoA activation was markedly greater in LRP-overexpressing HBMEC (Fig. 5c). Quantification of the RhoA-GTP amount with densitometry showed 42% higher activation of RhoA by CNF1 in LRP overexpressing cells than in the control cells (Fig. 5c). These results support our finding that LRP acts as the CNF1 receptor, thus mediating CNF1-mediated bacterial uptake and RhoA activation in HBMEC.
LRP-Antisense ODN Treatment Decreased CNF1 Effects on RhoA Activation and Bacterial Uptake-To confirm correlation between CNF1 effects and level of LRP expression, HBMEC were treated with LRP antisense ODNs before RhoA-GTP or invasion assays. Treatment of HBMEC with LRP antisense ODNs decreased LRP expression by 62% compared with the sense ODN-treated HBMEC (Fig. 6a). CNF1 effects on RhoA activation were decreased when the expression of LRP was reduced by antisense ODN treatment, whereas sense ODN treatment did not show noticeable changes in CNF1 effects, when compared with wild type HBMEC (Fig. 6b). For example, RhoA activation by CNF1 in HBMEC was 74% decreased with LRP antisense ODNs compared with non-treated cells, whereas the effect of LRP sense ODNs was negligible (Fig. 6b). CNF1-mediated bacterial uptake was also significantly (p Ͻ 0.01) decreased in LRP antisense ODN-treated HBMEC compared with sense ODN-treated HBMEC (Fig. 6b). These results taken together indicate that LRP acts as the CNF1 receptor, thus mediating CNF1-mediated bacterial uptake and RhoA activation. DISCUSSION We have previously shown that CNF1 contributes to E. coli K1 invasion of HBMEC in vitro and traversal of the blood-brain barrier in the experimental hematogenous meningitis animal model (8). CNF1 is a protein belonging to the group of dermonecrotizing toxins, produced by pathogenic E. coli (20,21). It is described as AB toxins, such as diphtheria, cholera, and tetanus toxins, which are comprised of catalytic domain, cell binding domain, and membrane translocation domain (22). Entry of CNF1 into eukaryotic cells includes binding to a host receptor and internalization via an endocytic mechanism (16). Endocytosed CNF1 is routed to the degradative pathway, i.e. fusion with the late endosome compartment and, its catalytic activity is translocated into the cytosol in an acid-dependent manner (16). Previous competition experiments suggested the presence of the receptor for CNF1 and the specific interaction between CNF1 and the cellular receptor. For example, the activity of CNF1 was inhibited when the full-length CNF1 was co-incubated with the N-terminal domain of CNF1 while coincubation with the C-terminal domain of CNF1 had no effect on the CNF1 activity (14). However, the identity and characteristics of the receptor have not been determined. In the present study, we identified 37-kDa LRP as the CNF1 receptor by the yeast two-hybrid system, co-immunoprecipitation, ligand overlay, and correlating the activity of CNF1 with the expression levels of LRP.
LRP has been known to be cell surface non-integrin lamininbinding protein receptors in several cell types, and involved in a variety of cellular mechanisms such as cell migration, adhesion, angiogenesis, and metastasis (23)(24)(25). According to the previous study, the N-terminal domain of LRP (amino acids 1-85) is a cytoplasmic domain, and the C-terminal domain of LRP (amino acids 102-295) is an extracellular domain, which HBMEC monolayers were incubated with a mixture of 3 g/ml CNF1 and 30 g/ml GST-LRP or GST in experimental medium for 2 h before addition of bacteria and invasion assays were carried out as described under "Experimental Procedures" (*, p Ͻ 0.05 compared with CNF1 or CNF1 ϩ GST, calculated using two-tail paired t test). Experiments were performed in triplicate. Error bars represent standard deviation.
are separated by a transmembrane domain (amino acids 86 -101) (19). Because the cDNA clone identified in this study matches the extracellular domain of LRP containing the transmembrane domain (amino acids 89 -295), it is likely that LRP is a receptor for CNF1. In vivo interaction between the Nterminal binding domain of CNF1 and extracellular domain of LRP was verified by cotransformation experiments where yeasts expressing LRP-(89 -295) could grow on selection medium only when they are co-expressing fCNF1 or nCNF1, but not cCNF1. This finding is consistent with the previous demonstration that the receptor binding domain is located at the N terminus of CNF1 (14). Co-immunoprecipitation experiments showed specific interaction between LRP-(89 -295) and CNF1 in vitro, and the ligand overlay assay indicated that the CNF1 binding to LRP (or host cell) is mediated by direct interaction between CNF1 containing N-terminal domain and the extracellular domain of LRP. For example, LRP-(89 -295) bound to fCNF1 and nCNF1, but not to cCNF1. The interaction between LRP-(89 -295) and [ 35 S]nCNF1 was abolished by preincubation with non-labeled nCNF1, proving the specific interaction between nCNF1 and LRP-(89 -295). Indeed, exogenous recombinant LRP could inhibit CNF1 activities such as RhoA activation and enhancement of bacterial uptake in HBMEC, by competing with cellular LRP in binding to CNF1. The inhibitory effect of LRP on CNF1 activation of RhoA also occurs in HEp-2 cells, where CNF1 has been shown to induce cytoskeletal rearrangements and bacterial uptake (12,13) After 48 h, whole lysates were separated by SDS-PAGE and transferred to PVDF membrane, and LRP was detected by polyclonal anti-LRP antibody. The membrane was stripped and reprobed with monoclonal anti-tubulin antibody for loading control. 48 h after treatment of HBMEC with ODNs, RhoA assays (b) and invasion assays (c) were performed as described under "Experimental Procedures" (*, p Ͻ 0.01, calculated using two-tail paired t test) (Ⅺ, sense; f, antisense ODN). Quantification of RhoA activation was performed as described above with densitometry. WT, wild type.
pcDNA3.1:LRP-transfected HBMEC. HBMEC stably transfected with pcDNA3.1:LRP (f) or empty plasmid pcDNA3.1 (ࡗ) were incubated with different concentrations of GST-CNF1. Absorbance was normalized by subtracting the background value where no CNF1 was added. b, invasion assays with E. coli K1 strain E44 in pcDNA3.1:LRPtransfected HBMEC (f). As control, invasion assays in pcDNA3.1transfected cells (Ⅺ) were included (*, p Ͻ 0.05, calculated using twotail paired t test). c, CNF1-induced RhoA activation in pcDNA3.1:LRPtransfected HBMEC. pcDNA3.1 or pcDNA3.1:LRP-transfected HBMECs were treated with 3 g/ml GST-CNF1 for 30 min and RhoA assays were performed as described above. For control, an equivalent level of GST alone was added to transfected HBMEC. The densities of RhoA-GTP bands were quantitated using an imaging densitometer, and normalized to the values of controls, where only GST was treated to the cells, to calculate -fold increases. that LRP is a receptor for CNF1 in a variety of eukaryotic cells. Enzyme-linked immunosorbent assay experiments showed higher binding of CNF1 to HBMEC transfected with pcDNA3.1:LRP than control vector-transfected HBMEC, illustrating the specific interaction between CNF1 and HBMEC via LRP. The effects of CNF1 on bacterial uptake and RhoA activation were, as expected, much higher in LRP-overexpressing HBMEC, possibly because of higher binding of CNF1 to LRP. In contrast, the CNF1 effects were inhibited when the expression of LRP was decreased by antisense ODNs, showing correlation between effects of CNF1 and levels of LRP expression in host cells.
LRP has also been identified as the receptor for the cellular prion protein (26 -28) and certain alphaviruses including Sindbis and Venezuelan equine encephalitis virus (29,30). However, this is the first demonstration that LRP acts as a cellular receptor for bacterial toxin in activation of RhoA and enhancement of bacterial uptake. Identification of receptors for bacterial toxins has been a challenge, and only a limited number of these receptors have been identified to date (31,32). The identification of a toxin receptor is essential for elucidating structure-functional analysis of a toxin as well as understanding the pathogenesis of toxin-induced diseases. In addition, LRP has been shown to be associated with the progression of a wide variety of carcinomas. For example, the interactions between tumor cells and LRP have been shown to play an important role in tumor invasion and metastasis (33)(34)(35)(36). p53 has been reported to down-regulate LRP expression levels by repressing an AP-2 cis-acting element localized in the first intron of the LRP gene in ovarian carcinoma cells (37). Our demonstration of LRP as the receptor for CNF1 suggests that CNF1 protein or CNF1-expressing bacteria may interact with mammalian cells exhibiting higher expression of LRP, and contribute to their transformation to cancer cells.
In summary, LRP was identified as the specific receptor for CNF1 by the yeast two-hybrid system and in vitro ligand binding assays. Exogenous LRP competes with endogenous LRP in binding with CNF1, thus decreasing effects of CNF1, such as RhoA activation and bacterial uptake. CNF1-mediated RhoA activation and bacterial uptake were markedly increased when LRP is overexpressed and the effects of CNF1 were decreased when the expression of LRP is reduced by antisense ODN, indicating that LRP functions as the receptor for CNF1. The demonstration that the excess amount of recombinant LRP or LRP antisense ODN could not totally inhibit the effects of CNF1 on RhoA activation or bacterial uptake, however, suggests that there may be an alternative pathway for CNF1 entry into eukaryotic cells, and this issue is currently being investigated in our laboratory.