A Key Motif in the Cholesterol-Dependent Cytolysins Reveals a Large Family of Related Proteins

Abstract

The cholesterol-dependent cytolysins (CDCs) are bacterial, β-barrel, pore-forming toxins. A central enigma of the pore-forming mechanism is how completion of the prepore is sensed to initiate its conversion to the pore. We identified a motif that is conserved between the CDCs and a diverse family of nearly 300 uncharacterized proteins present in over 220 species that span at least 10 bacterial and 2 eukaryotic phyla. Except for this motif, these proteins exhibit little similarity to the CDCs at the primary structure level. Studies herein show this motif is a critical component of the sensor that initiates the prepore-to-pore transition in the CDCs. We further show by crystallography, single particle analysis, and biochemical studies of one of these CDC-like (CDCL) proteins from Elizabethkingia anophelis, a commensal of the malarial mosquito midgut, that a high degree of structural similarity exists between the CDC and CDCL monomer structures and both form large oligomeric pore complexes. Furthermore, the conserved motif in the E. anophelis CDCL crystal structure occupies a nearly identical position and makes similar contacts to those observed in the structure of the archetype CDC, perfringolysin O (PFO). This suggests a common function in the CDCs and CDCLs and may explain why only this motif is conserved in the CDCLs. Hence, these studies identify a critical component of the sensor involved in initiating the prepore-to-pore transition in the CDCs, which is conserved in a large and diverse group of distant relatives of the CDCs.IMPORTANCE The cholesterol-dependent cytolysins' pore-forming mechanism relies on the ability to sense the completion of the oligomeric prepore structure and initiate the insertion of the β-barrel pore from the assembled prepore structure. These studies show that a conserved motif is an important component of the sensor that triggers the prepore-to-pore transition and that it is conserved in a large family of previously unidentified CDC-like proteins, the genes for which are present in a vast array of microbial species that span most terrestrial environments, as well as most animal and human microbiomes. These studies establish the foundation for future investigations that will probe the contribution of this large family of CDC-like proteins to microbial survival and human disease.

Keywords: Bacteroides; Bacteroidetes; Chryseobacterium; Deinococcus; Elizabethkingia; MACPF; Thalassiosira oceanica; pore-forming; toxin.

FIG 1

PFO structure showing critical residues for the mechanism of pore formation and the conserved CDC domain 3 motif. (A) The PFO crystal structure showing the core β-sheet (light blue) and the flanking αHBs (pink) in the soluble monomer (left) which unfurl into transmembrane hairpins (TMH) in the pore model (right). The αβ loop (red) in the soluble monomer refolds into an HTH upon conversion of the prepore to the pore. (B) Side view of PFO emphasizing the two interfaces which must be disrupted for the formation of TMHs and pore conversion. In TMH1, a water network is maintained around N197 (spheres) which stabilizes the interface. The TMH2 interface is maintained by the interaction of two methionine residues on the core β-sheet with a phenylalanine (spheres) in the αHB. (C) Representation of a growing PFO prepore following membrane binding. Following membrane binding, β5 of the αβ loop must be displaced to allow interactions to form between β1 and β4 of two monomers. Movement of β5 also increases the solvent accessibility of label-site residue V322 (green spheres). Upon completion of the prepore, K336 (coral spheres), located on the tip of the αβ loop, forms an electrostatic interaction with E183 (coral spheres) of the neighboring monomer. This interaction is necessary to supply the force to complete the flattening of the core β-sheet and to disrupt the water network around N197, resulting in pore conversion. (D) The conserved CDC D3 motif in the PFO soluble monomer. The conserved motif (orange spheres and sticks) resides on the severe bend in the core β-sheet on strands 2 and 3. R275 makes contacts with the beginning of the αβ loop at the turn between β-strands 4 and 5, and F230 forms a π-stacking interaction with a well-conserved aromatic, F351 (red sticks), at the end of the α-helix.

FIG 2

Residue R275 is necessary for αβ loop stability and conversion of the prepore to the pore. (A) Fluorescence scanning of NBD-labeled PFO^V322C shows β5 movement away from β4 upon membrane binding, as indicated by a decrease in fluorescence emission (14). Decrease in NBD fluorescence was due to an increased interaction with a polar solvent, i.e., water. (B) Fluorescence scanning of NBD-labeled PFO^R275A:V322C revealed β5 movement in the absence of liposomes, as indicated by the low fluorescence emission for both samples. (C) Kinetics of PFO mutants (5 μg) injected into HBS containing carboxyfluorescein-loaded liposomes. PFO wild type (WT) formed pores rapidly on liposomes, allowing release of entrapped dye. PFO^T319C:V334C, which forms a disulfide bond between β4 and β5, was allowed to form prepores prior to reduction of the disulfide bond with DTT, which led to rapid pore formation. PFO^{T319C:V334C:R275A} was allowed to form prepores, but upon reduction of the disulfide bond, pore formation could not be detected.

FIG 3

TEM images of membrane-bound oligomers of PFO^R275A and derivatives. (A) PFO^R275A incubated with cholesterol-containing liposomes prior to staining and imaging reveals linear oligomer complexes. (B) The boxed section shown in panel (A) digitally magnified 10×. (C) Oligomers of PFO^R275A:N197W. (D and E) Images of PFO^R275A oligomers in which an engineered disulfide bond between β-strands 4 and 5 was generated by substituting cysteines for T319 and V334, as previously described (14) under oxidizing (D) and reducing (E) environments. Only PFO^R275A:N197W exhibits pore-forming activity (Table 1).

FIG 4

The F230 and F351 stacking interaction must remain flexible during pore formation. (A) Kinetics of PFO mutants (250 ng) injected into HBS containing CF-loaded liposomes. PFO^F230C:F351C in its disulfide cross-linked form (ox) had no detectible activity relative to PFO WT. (B) Oxidized PFO^F230C:F351C (ox) was incubated with liposomes and allowed to form prepores prior to the addition of DTT (injected at 750 s) and displayed no detectible activity. Reduction to PFO^F230C:F351C (red) prior to addition to liposomes displayed activity greater than that for wild-type PFO. (C) PFO^F230C labeled with NBD was incubated in the absence (solid line) and presence (broken line) of liposomes representing the soluble monomer and membrane-bound pore complex, respectively. A very minor increase in fluorescence intensity was observed upon pore formation, indicating a shift to a slightly more nonpolar environment in the pore. (D) PFO^F351C labeled with NBD reveals a 2-fold increase in fluorescence intensity, indicating a change to a more nonpolar environment. Data are representative of a minimum of three replicates. The difference in fluorescence intensity maximum for (C) and (D) are due to differences in labeling efficiency between the two cysteine mutants.

FIG 5

Displacement of the α1β5 loop in the soluble PFO monomer blocks pore formation. PFO^V322C (A) labeled with NBD was incubated in the presence and absence of liposomes followed by fluorescence emission scanning. The decrease in emission profile for the sample incubated in the presence of liposomes (broken line) compared to the absence of liposomes (solid line) indicates β5 has moved away from β4, which is necessary for the HTH and pore formation. PFO^{F230C:F351C:V322C} was labeled in the oxidized form to prevent off-target NBD labeling. (B and C) The emission scans of the NBD-labeled PFO^{F230C:F351C:V322C} were obtained in the presence and absence of liposomes in its oxidized (B) and reduced (C) states. Both forms show decreased emission in samples without liposomes, indicating β5 movement. These data are representative of a minimum of three replicates.

FIG 6

Phylogenetic analysis of the CDCs with the CDCLs. Sequences were first aligned using MUSCLE (41, 53) and then IQ-tree (54–56) was used to perform a maximum likelihood analysis of the aligned primary structures of the current known CDCs and putative CDCLs using a bootstrap value of 200 and the default values set by the software. The tree figure was drawn using FigTree 1.4.4 (57). The single asterisk denotes a strain of Enterococcus faecalis derived from the marine environment, for which clinical isolates have not been identified with a gene encoding either a CDC or a CDCL. The double asterisks denote two Gram-negative species having the cognate D4 structure of the CDCs and shown to bind to and lyse cholesterol-containing membranes (1). The species within each clade are listed in Fig. S4 and each clade structure is shown.

FIG 7

Structure of a CDCL protein from Elizabethkingia anophelis. The recombinant purified E. anophelis CDCL expressed and purified from E. coli was crystallized and its structure solved to a 2.09 Å resolution. (A) Ribbon representations of the crystal structure of PFO (1PFO, left) (24) and the E. anophelis CDCL (right). For both PFO and the E. anophelis CDCL, the αHB1 and 2 that form the membrane-spanning β-hairpins (TMH1 and 2) are shown in magenta and the αβ loop is shown in dark blue. (B) Overlay of the region from both proteins (enclosed in squares in [A]) containing the F/Y-F/Y-X_n-YGR and GG motifs’ residues and the conserved Q228 of PFO (CDC residues in green and the CDCL residues are in orange). The numbers in parenthesis are the CDCL residues corresponding to those of PFO. (C) Residues of the CDCL that correspond to PFO motif residues F230, Y231, Y273, and R275 and the residues with which they have significant interactions mediated by their side chains compared to those made by the analogous residues in PFO (strongest interactions in bold) determined by their net pairwise interaction energies, as calculated by the interaction energy matrix analysis (21).

FIG 8

E. anophelis CDCL and CDCL^S pore formation and oligomeric complexes. (A) Kinetic release of marker (carboxyfluorescein, CF) from cholesterol-rich liposomes when treated with CDCL (1.4 μM), CDCL^S (1.9 μM), or CDCL (0.7 μM)+CDCL^S (0.95 μM). As CF is released from the liposomes, its emission is dequenched. No change in the emission is observed when the CF liposomes are incubated alone over the time frame of the experiment (not shown). (B) SDS-AGE of oligomer formation by PFO (15 μg), CDCL (15 μg), CDCL^S (15 μg), and CDCL⁺CDCL^S (15 μg total protein) in the presence of cholesterol-rich liposomes. The first lane of the CDCL, CDCL^S, and CDCL+CDCL^S samples was treated with just SDS sample buffer, whereas the second lane of each sample was treated with sample buffer and heated to 95°C for 3 min. O, oligomer; D, dimer; M, monomer. (C) Negative-stain transmission electron micrographs of cholesterol-rich lipid monolayers treated with CDCL, CDCL^S, and both (0.05 mg/ml total protein). (D) Selected classes from single-particle analysis of CDCL^S alone and in combination with CDCL from raw micrographs as shown in (C).