Human capillary morphogenesis protein 2 functions as an anthrax toxin receptor

Abstract

Bacillus anthracis secretes two bipartite toxins thought to be involved in anthrax pathogenesis and resulting death of the host. The current model for intoxication is that protective antigen (PA) toxin subunits bind a single group of cell-surface anthrax toxin receptors (ATRs), encoded by the tumor endothelial marker 8 (TEM8) gene. The ATR/TEM8-PA interaction is mediated by the receptor's extracellular domain related to von Willebrand factor type A or integrin inserted domains (VWA/I domains). A metal ion-dependent adhesion site (MIDAS) located within this domain of the ATR/TEM8 protein chelates a divalent cation critical for PA binding. In this report, we identify a second PA receptor encoded by capillary morphogenesis gene 2 (CMG2), which has 60% amino acid identity to ATR/TEM8 within the VWA/I domain, as well as a conserved MIDAS motif. A recombinant CMG2 protein bound PA and mediated toxin internalization when expressed on receptor-deficient cells. Binding between the CMG2 VWA/I domain and PA was shown to be direct and metal-dependent, although the cation specificity of this interaction is different than that observed with ATR/TEM8. Northern blot analysis revealed that CMG2 is widely expressed in human tissues, indicating that this receptor is likely to be relevant for disease pathogenesis. Finally, a soluble version of the CMG2 VWA/I domain inhibited intoxication of cells expressing endogenous toxin receptors when it was added to PA at a 3:1 ratio. These studies distinguish CMG2 as a second anthrax toxin receptor and identify a potent antitoxin that may prove useful for the treatment of anthrax.

Figure 1

(A) The VWA/I domains of CMG2 and ATR/TEM8 are highly related. The aligned amino acid sequences of the VWA/I domains of each protein (GenBank accession nos. AY233452 and NP_444262, respectively) are shown. Identical residues are indicated by an asterisk, and the five MIDAS motif residues of each protein are indicated by shading. Putative N-linked glycosylation sites found exclusively in ATR/TEM8 are indicated by underlined text. (B) Four different CMG2 proteins are identified or predicted to result from six alternatively spliced CMG2 mRNA transcripts. CMG2⁴⁸⁹, described in this article, has a putative signal peptide (SP), extracellular VWA/I domain, and putative transmembrane region (TM). CMG2⁴⁸⁸ (GenBank accession no. BAC03731) is identical to CMG2⁴⁸⁹ except that the last 12 aa of the cytoplasmic tail diverge (indicated by differences in shading). CMG2³⁸⁶ (GenBank accession no. AAK77222) is identical to CMG2⁴⁸⁹ except that it lacks amino acid residues 213–315. CMG2³²² is predicted to be a secreted protein with amino acid residues 1–290, matching those of CMG2⁴⁸⁸ and CMG2⁴⁸⁹ but lacking a transmembrane domain (see NCBI's AceView, locus 118429, sv 4 (g).

Figure 2

CMG2 binds PA in a divalent cation-dependent manner and supports intoxication. (A) PA receptor-deficient CHO-R1.1 cells expressing CMG2⁴⁸⁹-EGFP (blue line) were incubated with PA, a PA-specific antiserum, and an allophycocyanin-conjugated secondary antibody and then analyzed by flow cytometry. For control purposes, these experiments were also performed with receptor-negative parental CHO-R1.1 cells (red line), CHO-R1.1 cells expressing ATR/TEM8 sv2-EGFP (green line), and wild-type CHO-K1 cells that express endogenous receptors (orange line). (B) PA was bound to wells of a microtiter plate and then incubated with varying amounts of purified CMG2^VWA/I-MycHis protein in the absence or presence of 1 mM CaCl₂, MnCl₂, MgCl₂, or ZnCl₂. The bound CMG2 protein was then detected by adding a horseradish peroxidase-conjugated anti-His antibody and a chemiluminescent substrate. (C) CHO-R1.1 cells expressing CMG2⁴⁸⁹-EGFP (blue circles) were incubated with a constant amount of LF_N-DTA (2 × 10⁻¹⁰ M) and with increasing amounts of PA (shown). The level of toxin-mediated killing was measured by using a cell viability assay (WST-1 assay; Roche Molecular Biochemicals). The assay was performed in triplicate (standard deviations of the data are indicated with error bars), and the results were normalized to those obtained from cells treated with LF_N-DTA in the absence of PA (100% viable). For control purposes, these experiments were also performed with parental CHO-R1.1 cells (red triangles), CHO-R1.1 cells expressing ATR/TEM8 sv2-EGFP (green diamonds), and CHO-K1 cells (orange squares).

Figure 3

CMG2 is widely expressed in human tissues. (Upper) A radiolabeled CMG2-specific DNA probe was hybridized to a 12-tissue Northern blot (CLONTECH) that contained 1 μg of poly(A)⁺ mRNA per lane. Three predominant CMG2 mRNA transcripts were detected (indicated by arrowheads) that correspond in size to the 5.23-, 4.06-, and 3.53-kb transcripts predicted by NCBI's acembly method (see locus 118429). (Lower) A radiolabeled β-actin control probe (CLONTECH) was hybridized to the same blot, and it detected the expected 2-kb mRNA transcript in all tissues as well as the expected 1.6- to 1.8-kb transcript variant in heart and skeletal muscle (17).

Figure 4

Soluble CMG2 VWA/I domain protein inhibits intoxication of CHO-K1 cells. CHO-K1 cells, which express endogenous anthrax toxin receptors (), were incubated with 10⁻⁹ M PA and with 10⁻¹⁰ M LF_N-DTA in the presence of increasing amounts of CMG2^VWA/I-MycHis protein (0–10 μg/ml). Cell viability after intoxication was then measured (as in Fig. 2) and is shown as a percentage of that obtained with cells treated with PA alone (100% viable). For control purposes, the same experiments were performed with PA receptor-deficient CHO-R1.1 cells (□). (Inset) Coomassie blue-stained gel of the purified CMG2^VWA/I-MycHis protein preparation (indicated by an arrowhead).