Identification of TFPI as a receptor reveals recombination-driven receptor switching in Clostridioides difficile toxin B variants

Toxin B (TcdB) is a major exotoxin responsible for diseases associated with Clostridioides difficile infection. Its sequence variations among clinical isolates may contribute to the difficulty in developing effective therapeutics. Here, we investigate receptor-binding specificity of major TcdB subtypes (TcdB1 to TcdB12). We find that representative members of subtypes 2, 4, 7, 10, 11, and 12 do not recognize the established host receptor, frizzled proteins (FZDs). Using a genome-wide CRISPR-Cas9-mediated screen, we identify tissue factor pathway inhibitor (TFPI) as a host receptor for TcdB4. TFPI is recognized by a region in TcdB4 that is homologous to the FZD-binding site in TcdB1. Analysis of 206 TcdB variant sequences reveals a set of six residues within this receptor-binding site that defines a TFPI binding-associated haplotype (designated B4/B7) that is present in all TcdB4 members, a subset of TcdB7, and one member of TcdB2. Intragenic micro-recombination (IR) events have occurred around this receptor-binding region in TcdB7 and TcdB2 members, resulting in either TFPI- or FZD-binding capabilities. Introduction of B4/B7-haplotype residues into TcdB1 enables dual recognition of TFPI and FZDs. Finally, TcdB10 also recognizes TFPI, although it does not belong to the B4/B7 haplotype, and shows species selectivity: it recognizes TFPI of chicken and to a lesser degree mouse, but not human, dog, or cattle versions. These findings identify TFPI as a TcdB receptor and reveal IR-driven changes on receptor-specificity among TcdB variants.

Alignment of key positions for FZD interactions and CSPG4 interactions across all 206 TcdB sequences and 6 TcsL family sequences. Residues that are shared in TcdB1.1 are colored gray. Variable residues are colored blue (darkest blue = most common variant). Sequence ordering is based on phylogenetic analysis of full-length protein sequences. FZD-binding residues are based on PDB 6C0B; and the CSPG4-binding residues are based on PDB 7ML7.

Supplementary Figure 4: TcdB subtypes show variable dependency on FZD and CSPG4 receptors.
a-e HeLa-WT, FZD1/2/7-KO, CSPG4-KO, and UGP2-KO cells were exposed to either recombinant TcdB2.2 (a), TcdB4.2 (b), TcdB7.1 (c), or culture supernatant from native C. difficile strain expressing TcdB7.9 (d) for 24 h. The percentages of round-shaped cells were plotted over toxin concentrations or supernatant dilutions. The relative CR50 values in different cell lines were normalized to the WT and plotted as bar-chart (e). Error bars indicate mean ± s.d.; N = 3 (biologically independent experiments); *, p < 0.05; **, p < 0.01; NS, not significant (Student's ttest, two-sided). b TFPI, TFPI2, and mTFPI were expressed in HeLa or 5637 cells via lentiviral transduction. Expressed exogenous TFPI proteins in cells were confirmed via immunoblot detecting the triple HA tag fused to their N-termini. Actin was used as a loading control. Representative images were shown from two independent experiments. c-d Binding of 500 nM TcdB1.1, TcdB4.2, and TcsL to Fc-tagged TFPI (c) and mTFPI (d) was examined using BLI assays. Representative sensorgrams from one of three independent experiments are shown. e-h HeLa cells were exposed to either TcdB4.2 alone (4 pM) or TcdB4.2 pre-incubated with Fctagged TFPI (f), TFPI2 (g), or mTFPI (h) at the indicated molar ratios (1:250 ~ 1:20,000) on ice for 1 h. Representative images of the cell rounding effect at indicated conditions are shown (e). Scale bar, 20 µm. The percentages of cell rounding over time were recorded. Error bars indicate mean ± s.d.; N = 3 (biologically independent experiments).
Source data are provided as a Source Data file.

Supplementary Figure 7: Quantification of TcdB-TFPI interactions using BLI assays.
a-b Binding kinetics and affinity were determined using BLI assays for interactions between fulllength TcdB4.2 and TFPI-Fc (a) or mTFPI-Fc (b). Representative sensorgrams from one of two independent experiments are shown. c-g Binding kinetics and affinity were determined using BLI assays for interactions between TcdB4.21286-1805 and TFPI-Fc (c), mTFPI-Fc (d), cattle TFPI-Fc (e), chicken TFPI-Fc (f), and dog TFPI-Fc (g). Representative sensorgrams from one of two independent experiments are shown.
(m-n) Binding kinetics and affinity were determined using BLI assays for interactions between TcdB10.11285-1804 and mTFPI-Fc (m), and chicken TFPI-Fc (n). Representative sensorgrams from one of two independent experiments are shown.
o Summary of the binding kinetics between TcdB variants and TFPI across several species (mean ± s.d.).
s Summary of the binding kinetics between TcdB1.1-FBD-5M and the indicated proteins (mean ± s.d.).  HeLa cells were exposed to the culture supernatants of C. difficile strains expressing TcdB10.1 (ac), the recombinant TcdB7.2 (d-f), or the culture supernatants of C. difficile strains expressing TcdB11.2 (g-i), or TcdB12.1 (j-l) with or without preincubation with Fc-tagged TFPI (a, d, g, j) or mTFPI (b, e, h, k) at the indicated ratio on ice for 1 h. The percentages of cell rounding were recorded over time. The percentages of cell-rounding at 6 h incubation were plotted as bar-charts (c, f, i, l). Error bars indicate mean ± s.d.; N = 3 (biologically independent experiments); *, p < 0.05; **, p < 0.01 (Student's t-test, two-sided).

Supplementary
Source data are provided as a Source Data file.
b Binding of 500 nM TcdB4.2 full-length toxin to Fc-tagged TFPI, TFPI-K1, and TFPI-K2 was examined using BLI assays. Representative sensorgrams from one of three independent experiments are shown.
g-h The same amount of TcdB4.2 or TcdB1.1 (50 ng per 25 g bodyweight) was injected intraperitoneally into mice and the lung tissues were harvested and analyzed 15 h later. The volume of fluid in the thoracic cavity (g) and the dry-to-wet weight ratios of lung tissues (h) were shown. Injection of saline was included as a control. The range of boxes indicates ± s.e.m.; whiskers indicate ± s.d.; percentiles indicate median; *, p < 0.05; **, p < 0.01 (Student's t-test, two-sided).
i Experiments were carried out as described in panel g, except that the lung tissues were harvested 4 h after injection. Injection of saline was included as a control. The range of boxes indicates ± s.e.m.; whiskers indicate ± s.d.; percentiles indicate median; **, p < 0.01 (Student's t-test, twosided).
j Experiments were carried out as described in Fig. 4e-g, and the indicated lung tissues were harvested and subjected to histological analysis (H&E staining). Alveolar hemorrhage (upper panels) and widening of perivascular space (arrows in lower panels) were observed. These pathological changes were smaller in the TcdB4.2 + TFPI-Fc and TcdB4.2 + mTFPI-Fc groups. Scale bar, 200 µm. l Experiments were carried out as described in Fig. 4h- a HeLa-WT, two CSPG4 KO single clones (CSPG4#1 and CSPG4#8), and two CSPG4/TFPI double cells (CSPG4#1-TFPI-KO and CSPG4#8-TFPI-KO) were exposed to TcdB7.2 for 24 h. The percentages of rounded cells were plotted over toxin concentrations. Their relative CR50 values are plotted in a bar-chart (right panel). Error bars indicate mean ± s.d.; N = 3 (biologically independent experiments).
b-g HeLa cells were exposed to recombinant TcdB7.1 (b-d) or the culture supernatant of C. difficile strain expressing TcdB7.9 (e-g), with or without preincubation with Fc-tagged TFPI (b, e) or mTFPI (c, f) at the indicated ratio on ice for 1 h. The percentages of cell rounding were recorded over time. The percentage of cell-rounding at 6 h incubation was plotted as a bar-chart (d, g). Error