Cholera toxin (CT) from Vibrio cholerae and the related heat-labile enterotoxin (LT) from toxinogenic strains of E. coli are both A: B5 ADP-ribosylating exotoxins that cause abundant secretory diarrhea and enhance bacterial pathogenicity. They are also extremely potent immunogens and mucosal and parenteral adjuvants of immunization that potentiate mucosal and systemic B and T cell responses against unrelated co-administered antigens (Ags) in many experimental models (reviewed in [1,2,3,4,5]). Both toxins comprise a single A subunit responsible for the ADP-ribosyl transferase activity and five identical B subunits, responsible for cell binding. Another group of enterotoxins that are expressed by E. coli strains and have the same structure as CT and LT includes LT-IIa and LT-IIb (type II subfamily, CT and LT representing the type I subfamily) [6,7]. To overcome the enterotoxicity and use them as adjuvant in humans, non toxic mutants of the A subunit that retain adjuvant properties have been developed [1,2,3,4,5] and in some cases tested in clinical trials [8,9,10,11,12].

The B subunit of these toxins is devoid of toxicity and thus could also be used as adjuvants in humans. However, the adjuvant activity of the B subunit has been early a source of controversy and the question remains confusing. Moreover, the B subunits of CT (CTB) and LT (LTB), paradoxically, have been shown to promote tolerance to heterologous antigens. CTB has been the most studied, especially as conjugates with antigens and it has been proposed as a strategy to prevent auto-immune and allergic diseases [13] and has been tested in clinical assays [14].

On the whole, these molecules (the whole toxins, their non toxic mutants or the B subunits) are potent immune modulators which are involved in complex interactions with the immune system, leading either to increase or decrease immune responses.

Many studies have reported about the effects of the whole toxins or their mutants on different innate or adaptive immune cells that could explain the adjuvant effect (reviewed in [1,2,3,4,5,15]). However, the precise mechanism of action of these adjuvants has not been completely elucidated.

Different effects of the B subunit of these toxins on immune cells have also been reported but, conversely to the whole toxins and, because they potentiate tolerance and activation of regulatory T cells (Tregs) is a major mechanism involved in tolerance induction, the impact on Tregs has often been hypothesized, notably for CTB.

The aim of this review is to make a point on what is known about Tregs and the cholera-like enterotoxins, including the whole molecules or their mutants as well as their subunits.

First, the structures of these molecules will be described to underline similarities and differences which may impact the immunomodulatory properties, as well as the different regulatory T cell types.

Whole toxins. CT and LT are highly homologous (80% AA homology). They are made of a single A subunit which is associated to a non covalently linked pentameric ring of B subunit (A:B5). The A subunit is enzymatically active and consists of two chains A1 and A2 joined by a proteolytically sensitive peptide (Arg192) joined by a disulphide bond which has to be reduced for full enzymatic activity [5]. The main receptor for the B subunit of both CT and LT is GM1-ganglioside [16] that is found ubiquitously on the surface of mammalian cells, including enterocytes and immune cells, dendritic cells, macrophages, B and T cells. Other interactions have been reported which may have importance in immunological functions. Notably CTB and LTB also bind to GD1b-ganglioside but with a lower affinity and LTB, that binds to GM1 with a lower affinity than CTB, has been shown to have wider receptor specificity, as it binds notably to GM2 and asialo-GM1 [17]. The A1 fragment enters the cell cytosol and catalyses ADP-ribosylation of Gsa (GTP-binding protein family) which irreversibly activates adenylate cyclase and leads to an elevation in cyclic adenosine monophosphate (cAMP) levels. cAMP causes protein kinase A (PKA) to phosphorylate and open the cystic fibrosis transmembrane regulator (CFTR) Cl- channel, thus water and Cl- are flushed out causing diarrhea. The A2 fragment represents an adaptor molecule which interacts with the B subunit and may be involved in delivery of the A1 fragment into the correct cellular compartment and may also have a role in modulating the immunological effects.

The type II subfamily (LT-II) includes two variants, LT-IIa and LT-IIb, that show high homology to CT and LT but recognize different ganglioside receptors, notably the B subunit of LT-IIb binds to Gd1a ganglioside [18].

Mutants. Site-directed mutagenesis has permitted the generation of mutants that have reduced toxicity. Mutants have been constructed to dissociate the enterotoxic effects from their adjuvant activity. Mutations in both the active and protease sites of CT and LT have been performed including peptide extension and amino acid deletion [1,3,5]. These mutants have been shown to possess adjuvant activity in animal models and some of them, the protease site mutant of LT, LT-R192G [8,9,10] and the enzymatic site mutant LTK63 [11,12] have been tested in humans. Another approach to detoxify CT has been proposed by Agren et al. [19] by linking the enzymatically active A subunit domain of the toxin to a synthetic analog of the Staphylococcus aureus protein A. The fusion protein devoid of the B subunit cannot bind to ganglioside receptors. CTA1-DD binds B cells, is non toxic despite enzymatic activity and has been shown to have adjuvant activity in animal models, notably after mucosal administration with chlamydial Major Outer Membrane Protein (MOMP), rotavirus VP6 chimeric protein and HIV1 envelope glycoproteins or as a fusion protein containing tandem repeats of the influenza matrix protein 2 (M2e) ectodomain epitope (CTA1-3M2e-DD) [20,21,22,23]. Of interest, two recent studies have reported that enzymatically inactive mutants fusion proteins of CTA1-DD containing either a type II collagen peptide CTA1R7K-COL-DD or an OVA peptide CTA1R7K-OVA-DD induced tolerance [24,25].

B subunits. To avoid toxicity, CTB and LTB have been examined for their ability to augment immune responses to heterologous antigens. Devoid of enzymatic activity, they are responsible for cell binding. However, CTB and LTB have been shown to promote tolerance to coadministered antigens. They have been used either coadministered with antigens or as antigen-conjugates, especially CTB. The binding of LT-IIb B subunit to GD1a ganglioside facilitates TLR2 recruitment and activation and probably confers on the molecule its specific proinflammatory potential and its adjuvant activity [18]. This observation highlights the importance of cell binding in immunomodulatory properties of these molecules.

Immunosuppression is an intrinsic property of the immune system and is partially mediated by T cells. The best defined T-cell population with immunosuppressive activity is the naturally activated subset of CD4⁺ T cells (nTregs), that constitutively express the interleukin (IL)-2Rα chain CD25 and Foxp3 (forkhead box p3) transcription factor, which controls the development and function of nTregs. These thymus-derived suppressor cells are present in normal individuals, contribute to the maintenance of self tolerance and protect from a variety of autoimmune diseases. They are also implicated in immune responses against pathogens [26,27]. nTregs exert suppressive effects via cell-cell contact or via cytokines, IL-10 and transforming growth factor β (TGFβ).

Regulatory T cells can also be induced from naïve peripheral CD4⁺CD25^- T cells after antigenic exposure. Induced Tregs include three subtypes, CD4⁺CD25⁺Foxp3⁺, Tr1 and Th3 regulatory T cells and their induction depends on the nature of the antigen, the context of antigen presentation, especially the type of dendritic cells, and the specificities of the tissue in which the antigen is delivered [28].

CD4⁺CD25⁺Foxp3⁺ iTregs have been generated in vitro and in vivo after TCR stimulation in the presence of IL-2 and TGFβ, under a variety of conditions. The gut-associated lymphoid tissue (GALT), which is a TGFβ-rich environment and contains a particular subset of dendritic cells producing retinoic acid, favors the development of CD4⁺CD25⁺Foxp3⁺ iTregs. Moreover, studies have shown that nTregs and iTregs work in synergy to protect the host [29].

Tr1 cells have been generated in vitro from naïve T cells in the presence of IL-10 and antigen. They have been shown to prevent colitis induced in SCID mice [30] and to be involved in suppressing immune responses towards food antigens. Classically, they produce high levels of IL-10 and no IL-4, but depending on experimental conditions, they can also secrete high amounts of TGFβ and IL-5 and low amounts of IFNγ and IL-2. Tr1 cells have no specific marker but are Foxp3^-CD25^low. They exert their suppressive effect via IL-10 and TGFβ [31].

Th3 cells are specifically triggered during oral tolerance. They express Foxp3 and secrete high amounts of TGFβ and some Th3 clones can produce IL-10 and IL-4. Moreover, they seem dependent on IL-4 (rather than IL-2) for growth. TGFβ plays a fundamental role in the normal intestine where it is abundant and TGFβ1 is an important mediator of epithelial cell differentiation and IgA class switching and it has immunosuppressive effects on lymphocytes. Moreover, TGFβ seems to be the link between the three induced Tregs mentioned above [32].

Besides CD4⁺ Tregs, several types of CD8⁺ Tregs have also been reported [33]. As CD4⁺ Tregs, some CD8⁺ Tregs express Foxp3 and CD25, others exert a suppressor effect via TGFβ or via IL-10. CD8⁺ Tregs recognizing peptides associated with the MHC class Ib molecule Qa-1 and implicated in protection from experimental auto-immune encephalomyelitis (EAE) have also been reported. Moreover, in the intestine, intestinal intraepithelial lymphocytes (IEL) have been shown to have suppressive properties as TCRγδ IEL that secrete TGFβ and TCRαβ CD8αα IEL that secrete IL-10. Finally, CD8⁺ Tregs with suppressive properties are also present in the intestinal lamina propria [28].

Many studies have reported on the mechanisms that could explain the adjuvant properties of CT and LT (reviewed in [1,2,3,4,5,15]). Effects on the epithelium (as a mechanical barrier and as a source of mediators) and on antigen presenting cells (APC) (dendritic cells (DC), macrophages and B cells) as well as the more recently described effect of LT-IIb B on TLR2 [18] are easy to correlate with adjuvanticity. Other effects such as CT or LT-induced cell death are more difficult to interpret because of a great complexity (different in vivo/in vitro results, importance of the state of maturation or activation of lymphocytes [34]). Finally, two recent studies suggested additional mechanisms for adjuvanticity. Lee et al. have reported that CT induced strong Th17-type responses through intranasal delivery [35], whereas CT was thought to induce anti-inflammatory responses (by promoting Th2 and inhibiting Th1 responses). Moreover, the CTB subunit may be responsible for the CT’s ability to induce Th17-dominated responses which is in contradiction with the tolerogenic role of CTB. Another study by Tamayo et al. [36] demonstrated that LT upregulates the expression of the glucocorticoid-induced TNFR-related protein (GITR) in CD4⁺ T cells, that may contribute to adjuvanticity by acting as an activation-induced costimulatory signal in CD4⁺CD25^- cells. Of note, this effect was observed after systemic but not mucosal administration of LT, thus demonstrating that the mechanisms of adjuvanticity of LT are influenced by the route of administration.

The B subunit of both CT and LT has been shown to promote tolerance to heterologous antigens. Tolerogenicity induced by CTB or LTB may result from a direct depletion of effector T cells as the B subunits of CT and LT have been shown to induce CD4⁺ and CD8⁺ T cell apoptosis [5]. Another mechanism has been recently proposed by Wang et al. who showed that GM1 cross-linking by CTB resulted in the transient receptor potential canonical (TRPC) 5 channel activation which mediated Ca²⁺ influx in effector T cells and contributed to autoimmune suppression in an EAE model [37]. This effect mimics that of activated Tregs via Gal-1 on effector T cells.

As a matter of fact, the most studied tolerogenic mechanism is the involvement of Tregs directly or via modulating APC.

As previously indicated, the generation and involvment of Tregs in immune modulation after administration of cholera-like molecules has been mainly hypothezised and studied with CTB or LTB coadministered with or conjugated to antigens, based on the tolerogenic properties of such Ag delivery systems. However, despite the great number of studies reporting on the effects of CT and LT on immune cells, few studies have looked at their impact on Tregs.

Both the whole toxins and the B subunits have been shown to modulate Tregs. Moreover, different mechanisms may account for their effects and, except for the mechanism involving cAMP, it is difficult to associate one mechanism with a type of molecule, i.e., whole toxin or mutant vs B subunit, thus confirming the great complexity of these molecules and the difficulty to categorize them into adjuvants on the one hand and tolerance inducers on the other hand. However, the recent report by Hasselberg et al. showing that ADP-ribosylation controls the outcome of tolerance or active effector T cell immunity to an internal peptide p323–339 from OVA inserted into the cholera toxin (CT)-derived CTA1-OVA-DD [24] demonstrates that it is possible to decipher at the molecular level the critical determinants that impact on the orientation of immune responses.

Nevertheless, the variability of the effects depending on the type of antigen, the dose, the route and the time of administration (prime/boost or prevention/treatment) makes it imperative to test these molecules in different models.

As these molecules are potent immunomodulating agents with therapeutic potential in humans, future research to improve understanding of their effects is necessary.