Anthrax toxin receptor 1/tumor endothelial marker 8: mutation of conserved inserted domain residues overrides cytosolic control of protective antigen binding

Abstract

Anthrax toxin receptor 1 (ANTXR1)/tumor endothelial marker 8 (TEM8) is one of two known proteinaceous cell surface anthrax toxin receptors. A metal ion dependent adhesion site (MIDAS) present in the integrin-like inserted (I) domain of ANTXR1 mediates the binding of the anthrax toxin subunit, protective antigen (PA). Here we provide evidence that single point mutations in the I domain can override regulation of ANTXR1 ligand-binding activity mediated by intracellular signals. A previously reported MIDAS mutant of ANTXR1 (T118A) was found to retain normal metal ion binding and secondary structure but failed to bind PA, consistent with a locked inactive state. Conversely, mutation of a conserved I domain phenylalanine residue to a tryptophan (F205W) increased the proportion of cell-surface ANTXR1 that bound PA, consistent with a locked active state. Interestingly, the K(D) and total amount of PA bound by the isolated ANTXR1 I domain were not affected by the F205W mutation, indicating that ANTXR1 is preferentially found in the active state in the absence of inside-out signaling. Circular dichroism (CD) spectroscopy and (1)H-(15)N heteronuclear single-quantum coherence (HSQC) nuclear magnetic resonance (NMR) revealed that structural changes between T118A, F205W, and WT I domains were minor despite a greater than 10(3)-fold difference in their abilities to bind toxin. Regulation of toxin binding has important implications for the design of toxin inhibitors and for the targeting of ANTXR1 for antitumor therapies.

Figure 1. ANTXR1(F205W) is locked into the active conformation

CHOR1.1 cells expressing the indicated ANTXR1-EGFP fusion proteins were incubated with AlexaFluor 647 labeled PA_SSSR(K729C) for 6 h on ice, and PA binding was measured by flow cytometry. ANTXR1 expression was normalized by gating on equivalent EGFP signal (115 – 155 relative fluorescence units) from each sample. (A) PA was titrated in the presence of 2 mM MgCl₂ and binding measured via flow cytometry. Data shown are representative of two independent experiments with each point corresponding to the geometric mean fluorescence (GMF) of >300 individual cells. (B) The relative contribution of specific cations was determined by incubating 100 nM PA with receptor expressing cells in the presence of 2 mM CaCl₂, MgCl₂, or MnCl₂ and analyzing as in (A). Data points represent the mean ± standard deviation (SD) for three independent experiments. (C) A model depicting the differences in ANTXR1-sv1, -sv2, or the I domain point mutant -sv1(F205W) ability to bind anthrax protective antigen (PA). Two types of receptor activation states exist, relative to PA binding ability. The membrane bound short-tailed sv2 receptors are in the active state and bind PA. The long-tailed sv1 receptors exist in an equilibrium of active and inactive receptors where active receptors are capable of binding PA (arrow) and inactive receptors do not bind PA (line-bar). The I domain point mutation F205W increases the percent of active sv1 receptor capable of binding PA and this overrides inside-out signaling through the cytoplasmic tail, which restricts binding to a proportion of the ANTXR1-sv1. Closed spheres represent inactive receptors whereas open spheres represent active receptors.

Figure 2. PA binding of the soluble ANTXR1 I domain is not affected by the F205W mutation

(A, B) PA or BSA was adsorbed to a 384-well plate and the indicated GST-sANTXR1 proteins were titrated in the presence of 1 mM Ca²⁺, Mg²⁺, Mn²⁺, or EDTA. Bound GST-ANTXR1 proteins were detected using anti-GST-HRP and developed with 1-Step TMB as described in methods. Results shown are representative of at least two independent experiments performed in triplicate. Data represent mean ± the standard error (SE). (C) sANTXR1 proteins were incubated with 5 mM EDTA to remove residual cations, buffer exchanged to remove EDTA, and incubated with 1 mM MnCl₂ or ddH₂O at RT. Free metal was removed and proteins digested in Optima trace metal grade nitric acid. The molar ratio of bound Mn²⁺ per receptor molecule was determined via ICP-OES at wavelengths 257.610 nm, 259.373 nm, and 260.569 nm. An asterisk indicates a value below the detection limit. Data represent the mean ± SD for three independent experiments. (D) The F205W point mutation does not increase PA binding as measured by response units (RU) compared to WT sANTXR1 I domain. The binding curves for sANTXR1(WT) and sANTXR1(F205W) are nearly identical and overlap. PA_SSSR(K729C) was conjugated to a CM5 chip via thiol exchange and sANTXR1(WT) and (F205W) flowed over chip using a Biacore T-100. Sensorgrams are shown with concentrations of sANTXR1(WT) (black lines) and sANTXR1(F205W) (grey lines) indicated.

Figure 3. Point mutations in the soluble ANTXR1 I domain do not affect secondary structure

CD spectra were recorded in the far-UV range at 25°C for 0.2 mg/mL sANTXR1(WT), (F205W), and (T118A), as well as sANTXR2 (solid-black, black-dash, solid-grey, and long black dash, respectively). Analysis of the spectrum from sANTXR2 reveals different relative proportions of alpha helix compared with sANTXR1 as listed in Table 1.

Figure 4. ¹H-¹⁵N HSQC NMR of sANTXR1

¹H-¹⁵N HSQC NMR indicates proper structural folding of sANTXR1 WT and mutants as judged by the linewidths and dispersion of the amide cross peaks in their NMR spectra. (A) An overlay comparison of the ¹H-¹⁵N HSQC spectrum of sANTXR1(WT) and (F205W) shown in red and blue, respectively. The majority of the resonances in the spectra are superimposable, indicating similar conformations. (B) Overlay of sANTXR1(WT) and (T118A) shown in red and black, respectively. As in (A), the vast majority of resonances overlap.