C. perfringens enterotoxin-claudin pore complex: Models for structure, mechanism of pore assembly and cation permeability

Abstract

The pore-forming Clostridium perfringens enterotoxin (CPE), a common cause of foodborne diseases, facilitates Ca²⁺ influx in enterocytes, leading to cell damage. Upon binding to certain claudins (e.g., claudin-4), CPE forms oligomeric pores in the cell membrane. While the mechanism of CPE-claudin interaction is well understood, the structure and assembly of the pore complex remain elusive. Here, we used AlphaFold2 complex prediction, structure alignment, and molecular dynamics simulations to generate models of prepore and pore states of the CPE/claudin-4 complex. We sequentially addressed CPE-claudin, CPE-CPE, and claudin-claudin interactions, along with CPE conformational changes. The CPE pore is a hexameric variant of the typical heptameric pore stem and cap architecture of aerolysin-like β-barrel pore-forming toxins (β-PFT). The pore is lined with three hexa-glutamate rings, which differ from other β-PFTs and confer CPE-specific cation selectivity. Additionally, the pore center is indicated to be anchored by a dodecameric claudin ring formed by a cis-interaction variant of an interface found in claudin-based tight junction strands. Mutation of an interface residue inhibited CPE-mediated cell damage in vitro. We propose that this claudin ring constitutes an anchor for a twisting mechanism that drives extension and membrane insertion of the CPE β-hairpins. Our pore model agrees with previous key experimental data and provides insights into the structural mechanisms of CPE-mediated cytotoxic cation influx.

Keywords: Claudin; Clostridium perfringens enterotoxin; Molecular dynamics simulations; Pore-forming toxins; Protein complex modeling.

Graphical abstract

Fig. 1

The combination of ColabFold/AlphaFold2 and structural alignment with CPE and cCPE/CLDN4 complex crystal structures (PDB 3AM2 and 7KP4) enables modeling of CPE prepore complex. (A) Binding of monomeric CPE to CLDN4 in the membrane (black lines) leads to state I-CPE₁R₁. (B-C) Oligomerization via state II-CPE₃R₃ into prepore state III-CPE₆R₆. Side and top views are shown. CPE (CPE1-CPE6) and claudin receptor (R1-R6) subunits are labeled. Inset: β3- and β5-strands forming the inner β-barrel (marked area, left), and β2-, -β6- and β7- strands forming the outer β-barrel (marked area, right) are shown. (D) State IV-nCPE₆ corresponding to prediction AF-I (for CPE_1–202), used as reference for alignment of the other states. Side and top views. The inner β-barrel is depicted by the clipped side view and is highlighted in the top view (dashed red line). Segment 77–111 is colored orange in one subunit (yellow cartoon). (E) Superposition of prepore states III-CPE₆R₆ (gray) and IV-nCPE₆ (colored). Inset: Differences between IV-nCPE₆ (green) and III-CPE₆R₆ (gray). These differences (red arrows) result in the absence of clashes in IV-nCPE₆.

Fig. 2

Prediction of the sequential transition from prepore to pore states of the nCPE hexamer. (A-D) Transition from prepore state IV-nCPE₆ (prediction AF-I, for CPE_1–202) to prepore state V-nCPE₆ (prediction AF-II, for CPE_1–202). See Fig. 1D for IV-nCPE₆ overview. (A, B) The direction of movement/swapping of the α1-region between CPE subunits is indicated by arrows in A. The segment 77–111 including α1 is highlighted for one subunit in orange. (C, D) Completion of the β-barrel, as indicated by arrows. (E, F) Side views showing stepwise β-barrel extension in prepore states V-nCPE₆ (AF-II), VI-nCPE₆ (AF-III, for CPE_26–319) and pore state VII-nCPE₆ (AF-IV for CPE_1–202), representing potential intermediates and the final pore. (F) Clipped side view of the extended β-barrel. Three hexa-glutamate rings (E80, E101 and E115) forming constrictions along the β-barrel pore are indicated by arrows. The hexa-glutamate ring formed by E94 at the intracellular opening of the pore is also shown. Negatively (red) and positively (blue) charged side chains are shown as spheres, hydrophobic residues facing lipids are shown as sticks, membrane boundaries are indicated as black lines. (G) Clipped top view on ring formed by E101 in the middle of the transmembrane region. The six E101 are shown as sticks, other residues as lines, and the distance (12 Å) between two opposing E101 O-atoms as a dotted line.

Fig. 3

Comparison of the (A) nCPE hexamer model (state VII-nCPE₆, AF-IV) with the β-pore forming toxins (B) aerolysin, PDB 5JZT, homo-heptamer, (C) C. perfringens ε-toxin, PDB 6RB9, homo-heptamer, (D) lysenin, PDB 5GAQ, homo-nonamer, and (E) the HA3 subcomponent of C. botulinum type C progenitor toxin, PDB 2ZS6, hetero-hexamer. Side view, top view and close-up of pores in cap region of the toxins are shown. Charged residues are shown as sticks (red and blue arrows for negative and positive charges, respectively) and other residues as lines.

Fig. 4

Conformational change in linker region between cCPE and nCPE (residues 193–205, blue) allowing full transmembrane penetration of the β-hairpin tips of the pore. (A) Comparison of CLDN4-bound CPE hexamer pore complexes without (state VII-CPE₆R₆) and with downwards shift of nCPE (state VIIIa-CPE₆R₆). E94 at β-hairpin tip is shown as stick. (B) Clipped view of state VIIIa-CPE₆R₆ focused on one claudin with superimposed state I-CPE₁R₁ to visualize the conformational difference between the monomeric CPE structure (PDB ID 3AM2 in I-CPE₁R₁, red) and CPE in VIIIa-CPE₆R₆ (cyan, beige). Individual CPE-claudin dimers are shown separately for VII-CPE₆R₆ (C), I-CPE₁R₁ (D) and VIIIa-CPE₆R₆ (E). The region 73–116 that changes conformation is labeled yellow. In I-CPE₁R₁ it includes the α1-helix, in VII-CPE₆R₆ and VIIIa-CPE₆R₆ it forms the main part of the pore β-barrel. The manual shift of nCPE in VIIIa-CPE₆R₆ relative to the position in I-CPE₁R₁is labeled by dashed arrows in (B, D-G). (F, G) Close-ups of (D, E) highlighting the move of H146 at tip of nCPE from vicinity to L278 of cCPE in I-CPE₁R₁ toward W234 of cCPE in VIIIa-CPE₆R₆.

Fig. 5

MD simulation of the CPE/CLDN4 pore complex state VIIIa-CPE₆R₆. Snapshots after 100 ns of free simulation are shown with protein as cartoon, relevant residues as sticks, lipid acyl chains as gray lines, phosphate head groups as gray spheres, sodium ions as blue. (A) VIIIa-CPE₆R₆ shown as top view with labeled CPE (CPE1-CPE6) and CLDN4 receptor (R1-R6) subunits. (B) Details of the stable interaction between cCPE domain and CLDN4. Key residues are labeled and key electrostatic interaction shown as dashed lines. (C, D) VIIIa-CPE₆R₆ shown as side view (clipped in (D)). The pore barrel and the pore cap are well preserved. The pore complex is well embedded in the membrane. (E) Clipped side view of pore to illustrate membrane embedding and strong presence of sodium ions in the pore lumen. (F) The inner and outer β-barrel in the cap region (upper arrow in (E)) are held together mainly by hydrophobic interactions. Some key residues are labeled. In the center, a ring of six E115 residues strongly attracts cations (spheres). Rings formed by six E80 residues (G) slightly above membrane plane (middle arrow in (E)) and by six E101 residues (lower arrow in (E)) within the membrane plane (H) also strongly attract cations.

Fig. 6

Modification of relative cCPE/nCPE positions for modeling of state VIIIc-CPE₆R₆R´₆. (A) State VIIIa-CPE₆R₆: CPE pore hexamer (red) bound to six individual claudins (blue). (B) State VIIIc-CPE₆R₆R´₆: CPE pore hexamer (green) bound to a dodecameric claudin ring. Claudin subunits (R) primarily bound to CPE are colored cyan, and claudin subunits (R´) serving as secondary receptors are colored yellow. (C) Superposition of VIIIa-CPE₆R₆ and VIIIc-CPE₆R₆R´₆. Top views of complex (top), single CPE subunits of the complex bound to claudins in top (middle) and side view (bottom). Arrows indicate direction of movement for cCPE (red) and claudin (blue) as they transition from state VIIIa-CPE₆R₆ to state VIIIc-CPE₆R₆R´₆. 100 ns snapshots of MD simulations are shown (see Fig. 5 and below in 3.8).

Fig. 7

Comparison of the CPE monomer crystal structure (A; top, side view, bottom, top view) with state VIIIa-CPE₆R₆ (B), state VIIIc-CPE₆R₆R´₆ (C). In (D), a superposition of the three structures is shown. Conformational changes: Yellow: residues 70–116, region forming the pore barrel (straight yellow arrow); blue: residues 191–204, linker between cCPE and nCPE. Note the shift and turn of cCPE relative to nCPE between states (curved black arrow). For VIIIa-CPE₆R₆ and VIIIc-CPE₆R₆R´₆, snapshots after 100 ns MD simulation are shown (see Fig. 5 and in 3.8).

Fig. 8

MD simulation of CPE pore complex state VIIIc-CPE₆R₆R´₆. Snapshots are shown after 100 ns of simulation, with the protein as cartoon, relevant residues as sticks, lipid acyl chains as gray lines, phosphate headgroups as gray spheres, and sodium ions as blue spheres. (A) Top view with CPE (CPE1-CPE6, alternating green and red), primary receptor CLDN4 (R1-R6, blue) and secondary receptor CLDN4 (R´1-R´6, cyan) subunits. (B, C) Side view (clipped in (C)). The pore β-barrel and cap are well preserved throughout the simulation. The CPE transmembrane region and the bound CLDN4 are well embedded in the membrane. (D) Clipped side view of membrane-spanning pore region illustrating hydrophobic residues facing lipids while hydrophilic ones line the pore lumen. (E) Root-mean-square deviation (RMSD) of protein backbone over the simulation time (100 ns) with respect to the starting structure (0 ns) is plotted. RMSD for whole complex reached plateau at ∼ 3.0 Å after ∼ 50 ns. In addition, RMSD for individual subunits and mean RMSD for six CPE and six claudins are shown.

Fig. 9

Pore diameter and interaction of pore-lining residues with sodium ions in state VIIIc-CPE₆R₆R´₆. Snapshots after 100 ns of simulation are shown. (A) Clipped side view of pore barrel to illustrate pore lining, strong presence of sodium ions and position of hexa-glutamate rings (arrows). (B) Ring formed by six E80 residues slightly above membrane plane. (C) Ring formed by six E101 residues within the membrane plane. (D) Pore path and dimensions (gray surface) of the pore β-barrel, as obtained by quantitative analysis of simulation trajectory data using HOLE . CPE and CLDN4 subunits are shown as colored cartoons. The positions of pore-lining glutamates are shown as spheres in the cartoon. (E) Diameter along the pore coordinates calculated by HOLE. Constrictions of the pore diameter near the rings formed by E80, E101, and E115 were observed. Mean minimal pore diameter at E101: 7.8 Å. (F) Normalized contact times of pore-lining residues with Na⁺ ions over simulation time (last 50 ns). E80, E101 and E115 interacted more (> 40 % of time) than neighboring polar residues, as well as E94 and E126 residing at the pore periphery (all ∼ 5–10 % of time).

Fig. 10

Analysis of intra- and inter-subunit interfaces after MD simulation of the CPE pore complex state VIIIc-CPE₆R₆R´₆. Structure images show snapshots after 100 ns. (A) Top view of the complex with CPE (alternating green and red), primary receptor CLDN4 (blue), and additional secondary CLDN4 receptor subunits (cyan). Close-up regions shown in (B-F) are highlighted. (B) Interactions within the cap region of a CPE subunit. The inner and outer β-barrels are held together primarily by hydrophobic interactions. Key residues are shown. (C) CPE-CPE interactions in the cap region of the pore. The E67-K165 interaction between inner and outer β-barrel is shown. (D) Top: Electrostatic network at the backside of the bottom of the outer β-barrel. Bottom: Mean interaction count ( ± SD) for D48 with S60, Y129, and K131 within the last 50 ns of simulation for the six CPE-CPE subunit interfaces. (E) Key CPE-CLDN4 interface. Left: Cartoon of the interface, key residues are shown as sticks, and the binding pocket in CPE as semi-transparent surface. The perspective is turned by ∼ 180° with respect to (A). Top: Distance between L151 (Cγ) and Y306 (Cγ) over last 50 ns of simulation for individual CPE-CLDN4 dimers (CLDN4 subunits A, C, E, G, I, K; CPE subunits U, V, W, X, Y, Z) and average of all six. Bottom: SASA values (mean for last 50 ns) for the individual subunits (letters) and mean. Free: Mean value of L151 of the six claudins not interacting with this pocket in cCPE. (F) Cis-interface between two CLDN4 subunits. Top left: Cartoon of the interface, key residues are shown as sticks and interacting hydrophobic residues also as semi-transparent surface. CLDN4 subunits are colored cyan with ECS2 shown and blue with ECH shown, CPE is green. Top right: Distance between L71 (Cγ) and F147 (Cγ) over the last 50 ns of simulation for individual CLDN4 dimers (subunits given as letters) and average of all six. Bottom: SASA values (mean for last 50 ns) for the individual subunits (letters) and mean. Free: Mean value of corresponding residues in VIIIa-CPE₆R₆ simulation not consisting of CLDN4-CLDN4 interfaces.

Fig. 11

Cell viability and cCPE binding assays with HEK293 cells expressing either mouse CLDN3 wild type (mCLDN3-WT) or mCLDN3-S68E. (A) Viability of the HEK293 cells after 1 h incubation with different CPE concentrations tested by MTT assay. Normalized values (n = 3) were plotted against log of concentration (nM). (B) EC50 calculation from the data in (A) revealed a significantly higher EC50 value for HEK293 cells expressing the mCLDN3-S68E mutant. Mean ± SEM, n ≥ 6, *** p < 0.0003. (C) Cellular binding assay. HEK293-mCLDN3-WT, HEK293-mCLDN3-S68E cells, or non-transfected, claudin-free HEK293 cells were incubated for 30 min with different concentrations of GST-cCPE. After fixation, binding was detected by anti-GST antibodies and normalized to cell number using the Hoechst signal. From the binding data (n = 2), the unspecific binding (i.e., binding to non-transfected control cells) was subtracted. (D) Determination of K_D from the data in (C) and two replicates. The K_D was significantly higher for the mCLDN3-S68E mutant (38 ± 3 nM) than for WT (14 ± 5 nM). Mean ± SEM, n ≥ 3, * p < 0.05.

Fig. 12

Schematic model of CPE/claudin pore complex assembly. Top views (top) and respective side views (bottom) of different stages. (A) Claudin dimers form dynamically. (B) Monomeric CPE binds to claudin dimer containing primary (CLDN4) and secondary receptor (e.g., CLDN1 or CLDN4). The trimeric complex (small complex) is stabilized by high affinity CLDN4-cCPE-, CLDN-CLDN cis-, and low affinity CLDN-cCPE interactions. (C) 6 small complexes oligomerize with minor conformational changes, only, into a large prepore complex containing 12 claudin and 6 CPE subunits. (D) α1-helix swaps clockwise (red arrows) from domain II in same CPE subunit to domain II in neighboring subunit. Concomitant anticlockwise movement (yellow arrows) of β-sheet in domain II, but not of the claudin receptors, support change of angle between domain II and cCPE. (E, F) α1-helix dissociates from the β-sheet in domain II leading to twisted extension of inner β-barrel. nCPE (domains II and III) starts to move downwards to membrane. cCPE angles further relative to nCPE. Claudins bound to cCPE move more to center leading to formation of a continuous claudin ring. (G) Further twisting movement of complex center against complex periphery drives full extension of the β-barrel, membrane penetration and formation of the active pore. For clarity, modeling intermediates (nCPE₆ states, II-CPE_2–5R_2–5 and CPE₆R₆) are not depicted.