RTX toxins

RTX toxins are exoproteins secreted by Gram-negative bacteria, grouped into this family due to shared structural characteristics: i: RTX toxins are encoded in an operon along with other genes necessary for their post-transcriptional activation and secretion. Figure 1 provides a general schematic of this operon. The rtxA genes encode the inactive protoxin (ProRTX), which is activated intracellularly by acylation of internal Lys [29]. This process is catalyzed by an acyltransferase encoded by the rtxC genes that transfers the fatty acids from an acyl-carrier protein (ACP) to ProRTX. The rtxB and rtxD genes encode bacterial membrane proteins that form a “channel-tunnel” facilitating toxin export via a Type I secretion system, which is ATP-dependent [30–33]; ii: RTX toxins possess a secretion signal peptide towards the C-terminal end [9, 34, 35]; iii: RTX toxins contain the characteristic glycine- and aspartate-rich repeats (GGXGXDXUX) that bind numerous calcium ions, typically in the carboxy-terminal portions of the molecule [9, 36]. The large multifunctional autoprocessing RTX (MARTX) toxins also bear similar repeats in the N-terminal segments. RTX toxins thus require physiologically high (> 1 mM) Ca²⁺ concentrations for proper folding and biological activity on host cells. This so-called RTX motifis the one that defines this class of toxins.

Display full size

Figure 1. 

Scheme of the synthesis, activation and secretion of RTX toxins. rtxC gene encodes for an acyltransferase that transfer fatty acids from the acyl-carrier protein (ACP) to the ProRTX protein encoded by rtxA gene. Rtx B and Rtx D are part of the secretory machine with tolC. AlphaFold structures were included in the scheme using as example for the RtxA polypeptide, HlyA from E. coli (UniProtKB P09983), and for RtxC, the acyltransferase ApxC from Actinobacillus pleuropneumoniae (PDB 4WHN). Image HlyA from E. coli was adapted from UniProt of UniProtKB P09983. CC BY, and the image acyltransferase ApxC from Actinobacillus pleuropneumoniae was adapted from RCSB PDB (RCSB.org) of PDB 4WHN [37]. CC0. RTX: Repeat in ToXins; HlyA: alpha-hemolysin

It is important to highlight that the crystallization or exact structural determination of RTX toxins has not yet been achieved, neither in solution nor when bound to membranes. Current knowledge is based on the interpretation of biophysical experiments, molecular dynamics (MD), and potential structures predicted by AlphaFold.

The main members of the RTX toxin family are: HlyA from uropathogenic E. coli (UPEC), CyaA from B. pertussis, leukotoxin (LtxA) from Aggregatibacter actinomycetemcomitans, EhxA from enterohemorrhagic E. coli, ApxIA from Actinobacillus pleuropneumoniae, VcRtxA from Vibrio cholerae, among others [35].

RTX toxins interact with a wide range of cells, leading to extensive research on the existence of specific protein receptors for these toxins in target cells. Traditionally, RTX toxins have been subdivided in two categories (namely, hemolysins and LtxAs) due to their range of activity against different cell types. The so-called hemolysins were initially found to be active toward many different cell types from a wide variety of species, while cytotoxins and LtxAs display toxicity toward a narrower range of cell types and species [38–40]. The hemolytic effect, although significant for the biochemical and genetic characterization of these toxins, is not highly relevant physiologically. More significant are the sublytic effects produced by these toxins on both erythrocytes and leukocytes [21, 41–43]. The physiologically relevant target cells of most RTX toxins are leukocytes that express β2 integrins, which serve as specific receptors for many RTX toxins. Although recently, Ristow et al. [44] have demonstrated that both HlyA and LtxA interact with the extracellular domain of β2 integrins, downstream signaling from this interaction is not implicated in the sublytic effects on cells expressing these proteins. At high concentrations, RTX toxins can cause direct lysis of the target leukocytes. Alternatively, when present at sublytic concentrations (which is most likely the case during in vivo infection), they can alter membrane permeability, resulting in calcium entry, potassium efflux, or both. This can trigger various signaling cascades that may alter cell physiology, eventually leading to apoptosis. Ultimately, these damages result in inflammation and contribute to tissue or organ lesions, consequently enhancing the pathogenicity of the bacterial hosts.

On the other hand, glycophorin had been characterized as a protein receptor for HlyA in erythrocytes [45], and it was demonstrated that the region between amino acids 914–936 was involved in this interaction [46]. Recently, our group has shown that human erythrocytes lacking glycophorin A and B (GPA^null/GPB^null variant) are equally sensitive to the toxin as normal erythrocytes, indicating that the presence of glycophorin is not essential for the lytic effect [26].

Bearing in mind the concept introduced by Skals et al. [47], that none of the RTX cytolysins meet the strict pharmacological requirement of specific binding to a membrane protein to exert their effect, it is noteworthy that HlyA, CyaA, and LtxA, for instance, can disrupt protein-free liposomes [48–50]. Therefore, the authors propose the existence of “proximal molecules” in the membrane that facilitate the approach of the toxins to the cells, allowing them to insert into the membrane in the correct conformation to exert their effect. These “proximal molecules” could include glycophorins, integrins, etc.; but it can also include lipids, such as cholesterol [47].

Role of cholesterol in the mechanism of action of the RTX toxins

In recent years, it has been demonstrated that the activity of many RTX toxins depends on the cholesterol content in membranes. For example, the hemolytic activity of HlyA decreases when erythrocytes are depleted of cholesterol using methyl-β cyclodextrin. Furthermore, a reduction in the oligomerization of this toxin in the membrane was observed [51]. Additionally, the release of fluorophores from liposomes varies according to their cholesterol content [52]. This was observed for both HlyA and CyaA [50, 52, 53]. Similarly, it was demonstrated that reducing the cholesterol content in J774A.1 macrophages decreases the translocation capacity of the adenylate cyclase (AC) domain of CyaA [54].

In the case of LtxA from A. actinomycetemcomitans, its ability to kill Jurkat clone Jn.9 and THP-1 cells (monocyte and T lymphocyte cell lines, respectively) decreases when these cells are depleted of cholesterol. Once cholesterol is restored to their membranes, these immune cells regain sensitivity to the toxin [55]. The hemolytic activity of RtxA from Kingella kingae (K. kingae) also showed a dependence on cholesterol content [56]. All these findings suggest that cholesterol binding may be another structural characteristic of this toxin family. Moreover, cholesterol interaction domains have been identified in the sequences of all these toxins. The cholesterol recognition/interaction amino acid consensus (CRAC) motifs have the consensus sequence: L/V-(X1-5)-Y-(X1-5)-R/K. This sequence consists of a branched apolar residue (L or V) followed by a variable segment containing 1 to 5 residues of any amino acid, then an aromatic residue Y, followed by another 1 to 5 residues of any amino acid, and finally a basic residue (K or R) [57, 58]. The CARC sites are similar to CRAC motifs but have the opposite orientation, with the sequence K/R-(X1-5)-Y/F-(X1-5)-L/V [59]. This domain contains basic residues that ensure the correct positioning of the CARC motif at the polar/apolar interface of a transmembrane domain, exactly where cholesterol is presumed to be located and can accept tyrosine (T), phenylalanine (F) or tryptophan (W) as the central amino acid residue [58]. Table 2 shows a list of the CARC and CRAC found in several RTX toxins, together with the number of residues of each protein. This information was obtained from [60].

Antimicrobial peptides

Antimicrobial peptides (AMPs) are a diverse group of natural or synthetic molecules capable of targeting a broad spectrum of pathogens, including antibiotic-resistant bacteria [62, 63]. AMPs typically exert their antibacterial effects by disrupting the bacterial cell membrane, leading to cell lysis and death. This is primarily driven by electrostatic interactions between the cationic peptides and the negatively charged components of the bacterial envelope. Once the peptide reaches the bacterial membrane, it can penetrate through various models, including the barrel-stave, carpet, annular pore, aggregation channel, and sinking raft models [63]. The five main proposed mechanisms of action include the interruption of quorum sensing, disruption of membrane potential, inhibition of alarm systems, degradation of biofilm matrix and polysaccharides, and inhibition of transporter expression [64]. Additionally, the role of AMPs in inhibiting biofilm formation is significant. Bacteria within biofilms are able to evade the immune system and survive, often causing chronic infections. Conventional antibiotics are frequently unable to penetrate these biofilms; hence, the development of peptides that inhibit biofilm formation and enhance antibiotic penetration is promising. AMPs with these properties have been tested against biofilms of Staphylococcus aureus (S. aureus) [64] and Pseudomonas aeruginosa (P. aeruginosa) [65].

In recent years, AMPs have emerged as potential alternatives to traditional antibiotics due to their unique and complex antibacterial mechanisms, offering a promising solution to combat multi-drug resistant bacteria [61]. Additionally, the probability of drug resistance to peptides generated by gene mutation is low; therefore, drug resistance is more difficult to develop in AMPs than antibiotics [62, 65].

Antivirulence peptides

As mentioned earlier, in recent years, the design of peptides and molecules inhibiting virulence factors of various bacteria has increased to reduce antibiotic resistance [2]. The synthesis of peptides that inhibit these factors is not as widely disseminated as AMP. Therefore, I will describe some of these peptides, particularly those targeting RTX toxins.

The main goal of these therapies is to inhibit the interaction of the toxin with its target cell, either by preventing its interaction with a membrane protein or a lipid. As described above, the presence of a protein receptor for RTX toxins is quite controversial. However, CRAC and CARC motifs that interact with cholesterol have been detected in several of them [60]. For three of them, including LtxA from A. actinomycetemcomitans [66], HlyA from E. coli [67], and CyaA from B. pertussis [68], the possibility of synthesizing peptides that inhibit the interaction of these domains with cholesterol in target cells has been studied. In fact, the application arose from studying various peptides synthesized from CRAC/CARC sequences of the toxins to understand the molecular mechanism of these toxins’ binding to the membrane. Once this interaction was characterized, they were tested as potential inhibitor.

CyaA

B. pertussis is the etiological agent causative of whooping cough, a highly contagious, acute respiratory illness in humans. Among its virulence factors is CyaA, a distinct member of the RTX toxin family, due to its bifunctional activity. It possesses an AC domain in addition to the core structural features common to other RTX hemolysins [72]. CyaA exhibits the same hemolytic activity as other RTX toxins [72, 73], and can deliver its catalytic AC domain into the cytosol of target cells, where it is activated by calmodulin (CaM).

Figure 2 shows the AlphaFold structure of the three toxins. In HlyA structure the main charateristic domains of the RTX toxins are marked, but as well as the LtxA, the binding calcium domain is the region predicted with high confidentiality (blue region). Instead the insertion region presents less confidentiality (orange-yellow region).

Display full size

Figure 2. 

AlphaFold structure of LtxA (UniProtKB P16462), HlyA (UniProtKB P09983) and CyaA (UniProtKB J7QLC0). The main structural domains of HlyA were highlighted with different colours as in [67]. HlyA image was adapted with permission from [67]. © 2023 American Chemical Society, and LtxA and CyaA images were adapted from UniProt (UniProtKB P16462 and UniProtKB J7QLC0). CC BY. LtxA: leukotoxin; HlyA: alpha-hemolysin

In the case of LtxA, 44 CARC and CRAC motifs were identified in its sequence (Table 2). Brown and colleagues [49] demonstrated that the domain located between amino acids 334–340 is specifically involved in the interaction with cholesterol and inhibits LtxA’s ability to lyse Jurkat cells (Jn.9). Based on this finding, they synthesized a peptide containing this domain (display in bold):

327-FDRARMLEEYSKRFKKFGY-346

Using isothermal titration calorimetry (ITC), they showed that the peptide binds with high affinity to cholesterol-containing membranes and does not disrupt membrane packing, suggesting that it primarily resides near the membrane surface [74]. The peptide undergoes a conformational change upon binding to cholesterol. In solution, the peptide consists of 25% α-helices and 25% β-sheets, with the remaining structure being random coils or disordered regions. When bound to a cholesterol-containing membrane, there is a large decrease in helicity and a significant increase in β-sheet structure [74].

Preincubating giant unilamellar vesicles (GUVs) containing cholesterol with the peptide showed a decrease in LtxA binding. Additionally, pretreating THP-1 cells with the peptide resulted in reduced toxin internalization, demonstrating its efficacy in mitigating the toxin’s effects.

In HlyA, 26 CRAC/CARC motifs were identified according to Table 2, while Vazquez et al. [52] found 20. That is because the two groups used different consensus motif sequences. Researchers who compiled the data in Table 2 used the following sequence: (L/V)-X1,5-(Y/F)-X1,5-(R/K) [60], whereas Vazquez et al. [52] required that the central aromatic amino acid has to be only tyrosine (Y). Based on the location of these domains in the HlyA sequence, Cané et al. [67] synthesized two peptides:

341-RFKKLGYDGDSLL-353 (PEP 1), located in the pore-forming domain, and

639-VVYYDK-644 (PEP 2), located between the two acylated lysines (K563 and K689).

The authors explored the interaction of both peptides with membranes of different lipid compositions (pure POPC and POPC/CHOL at 4:1 and 2:1 molar ratios) using surface plasmon resonance (SPR) and MD simulations. They showed that both peptides preferentially interact with cholesterol-containing membranes, presenting a higher affinity PEP 2 than PEP 1 in all the membranes tested. However, only the peptide located between the fatty acids (PEP 2) was able to inhibit hemolysis caused by HlyA (Figure 3A) and did not exhibit lytic activity per se. Instead, PEP 1 presented hemolytic activity at high peptide concentration, and it did not inhibit toxin activity. These differences might be due to the location of the peptides in the membrane as predicted in MD snapshots in Figure 3B. In this case, two inhibition strategies were tested: 1) pre-treating erythrocytes with the peptide, and 2) adding the peptide and toxin together. The latter strategy would be more akin to an in vivo situation if these peptides were to be used as a treatment. Unfortunately, inhibition of toxin binding to erythrocyte ghosts was only observed when the erythrocytes were pre-treated with the peptide Figure 3C [67].

Display full size

Figure 3. 

These results were extracted from the Cané et al. [67]. (A) Inhibition of alpha-hemolysin (HlyA) lytic activity by PEP 1 and PEP 2. Hemolysis percentage of 2% v/v erythrocytes as a function of different toxin/peptide molar ratios. Black bars correspond to erythrocytes preincubated with PEP 1, and gray bars correspond to erythrocytes preincubated with PEP 2. (B) Representative snapshots of the final state of POPC/Cho (2:1 molar ratio) membranes in the absence and presence of PEP 1 and 2. T_f = 200 ns. The polar headgroups of POPC (phosphate and choline groups) and Cho are represented as VDW. Peptides are represented in licorice, colored by the polarity of their residues. Water and hydrocarbon tails of POPC are not shown to improve visualization. (C) The binding percentage of HlyA was calculated as the optical density ratio between HlyA and β-actin bands of the Western blot membranes showed below. Figure A and B were reprinted with permission from [67], Figure C was adapted with permission from [67]. © 2023 American Chemical Society

CyaA contains 40 CRAC/CARC motifs in its sequence (Table 2). Masin and colleagues [75] reported that Y/F or Y/A substitutions in the respective central tyrosine residues of the four CRAC motifs (627–638, 654–661, 721–728, and 732–741) predicted in the pore-forming domain of this toxin had no effect on the translocation or hemolytic activity of the toxin. However, recently, Amuategi et al. [76] found four CRAC/CARC motifs: two located in the transmembrane helices (413–420 and 481–487) and the other two in the first helices of the membrane insertion domain (518–527 and 527–534). Surprisingly, none of these four functional CRAC/CARC motifs are conserved in the rest of the RTX toxins [60]. Studies with point mutants where F residues 521 and 532 were substituted by A in the CRAC 518–527 and CARC 527–534 motifs caused a potent inhibition of the ACT activity, while two mutants in the central F residue of the CARC 413–420 and CRAC 481–487 motifs caused a prominent inhibition of the ACT capacity to generate cAMP in the target cell cytosol. At the same time, these mutations augmented the toxin’s lytic capacity. This suggests that in cholesterol-rich membranes, such as the eukaryotic plasma membrane, binding to cholesterol through these CARC 413–420 and CRAC 481–487 drives h1 and h2 to embed into the membrane, and further, that this insertion is essential for AC translocation [76].

Additionally, Amuategi et al. [68] showed that a peptide containing the CRAC 481–487, 454-ASAHWGQRALQGAQAVAAAQRLVHAIALMTQFGR-487, undergoes a conformational change upon binding to cholesterol-containing membranes and inhibits the hemolytic effect of CyaA and the generation of AC in J774A.1 cells.

As mentioned before, some of these toxins also interact with surface proteins such as β2 integrins and glycophorins [45, 77]. It has been demonstrated that these toxins’ activity diminishes when cells are depleted in cholesterol. However, their activity remains largely unaffected when the toxin-protein receptor is compromised [26, 51]. Furthermore, it was demonstrated that many RTX toxins interact with β2 integrins, but intriguingly, the sublytic effects of the toxin do not involve β2 subunit downstream signaling [44].

Considering this evidence leads to a hypothesis proposing that these toxins first interact with proteins, which are in high concentration on membranes, acting as “proximating proteins”. Subsequently, they interact with membrane cholesterol through CRAC and CARC motifs, which stabilizes active conformations for further oligomerization or clustering in lipid rafts or facilitates toxin translocation. The mimicking CRAC peptides inhibit toxin interaction with cholesterol, affecting the adoption of the active conformation, altering in this way the action of the toxin.