Bacterial superantigens

Superantigens (SAgs) are the most powerful T cell mitogens ever discovered. Concentrations of less than 0·1 pg/ml of a bacterial superantigen are sufficient to stimulate the T lymphocytes in an uncontrolled manner resulting in fever, shock and death [1–3]. SAgs bind, as intact molecules to the class II major histocompatibility complex (MHC) antigens expressed on professional antigen presenting cells (APCs) outside the peptide-binding groove then sequentially bind the T cell receptor (TcR) via the variable region of the TcR β-chain [3–6]. Every SAg binds a subset of TcR Vβ domains and as the number of different Vβ regions in the human T cell repertoire is restricted to approximately 50, comprising about 24 major types of Vβ elements, a substantial number of T cells are activated by SAgs. This can be as high as 20% compared with only 1 in 105−106 naive T cells that are responsive to conventional peptide antigen. This results in massive systemic release of pro-inflammatory cytokines, such as tumour necrosis factor-alpha (TNF-α) and interleukin-beta (IL-1β), and T cell mediators, such as IL-2, which can lead to fever and shock [3,5,7,8]. Over the last 4 years the number of known bacterial SAgs has increased sharply, due mainly to various microbial genome sequencing projects [9–14]. There are now 41 bacterial SAgs described in the literature (Table 1) and the number is growing steadily. In addition, a new family of SAg-related proteins has been identified in Staphylococcus aureus that show sequence and structural homology to the ‘classical’ SAgs, but appear to have a quite different role. This review provides a summary of the field to date with some of the more recent discoveries that shed light on the how superantigens are able to trigger such strong T cell responses and the diseases related to SAg intoxication. Functional properties of superantigens and their associated diseases Functional properties of superantigens and their associated diseases SAgs are characterized by their ability to bind both MHC class II molecules and T cell receptors [5,6]. This occurs in a sequential fashion. The sole purpose of SAgs appears to be to bring these two critical molecules together in order to activate as many T cells as possible. The net result is the release of a large and sudden bolus of cytokines which causes the acute condition toxic shock [3,5,7,8] The histocompatibility class II molecule, despite its polymorphism, is the principal cell receptor for all SAgs but the affinity for MHC class II varies depending on the class II molecule and the SAg [15–17]. All SAgs examined so far display higher affinities towards human MHC class II molecules than mouse class II, which explains partly why SAgs are several orders of magnitude more potent on human T cells than mouse T cells. A variety of binding modes exist to both MHC class II and TcR (described in more detail below) which indicates the lengths that the bacteria have gone to target these two critical molecules of the adaptive immune response. The first bacterial SAg was isolated in the late 1960s by Bergdoll and coworkers as a secreted toxin of S. aureus and was named staphylococcal enterotoxins A (SEA) for its potent enterotoxic properties. The staphylococcal enterotoxins (SE) are the causative agent in staphylococcal food poisoning and induce vomiting and diarrhoea within 1–2 h following ingestion. The mitogenic activity of SEs was discovered many years later, but the term ‘superantigen’ was not coined until 1989 when Marrack and coworkers found that the mitogenic activity was a result of a massive expansion of T cells that all shared the same T cell receptor Vβ chain domains [2]. Today, 18 different SEs have been described in the literature (Table 1) and all are potent T cell mitogens with half maximum stimulation values as low as 0·1 pg/ml. The SAg family also includes the S. aureus toxic shock syndrome toxin (TSST) which is the causative agent in toxic shock syndrome [18]. Twelve SAgs have been identified in Group A Streptococci (GAS), predominantly but not exclusively produced by S. pyogenes. These are the streptococcal pyrogenic exotoxins (SPEs) A, C, G-M, the streptococcal superantigen (SSA) and the streptococcal mitogenic exotoxin (SMEZ) 1 and 2. Many new SAgs have been identified by screening the completed S. pyogenes genomes; serotypes M1 (Oklahoma University, USA), M3 and M18 (Rocky Mountain Laboratories, NIH, USA) with conserved sequence motifs [10–13]. The sudden explosion of new superantigen sequences has resulted in confusing nomenclature where some sequences have been given two different names. For example, SPE-K identified in a serotype M3 strain from the United States is identical to SPE-L found in an M3 strain from Japan and an M89 strain from New Zealand [10,19,20]. The superantigen gene spe-l found in an M18 strain from the United States is identical to a gene named spe-m found in New Zealand [13,20]. For the purposes of this review, spe-m* is the gene described in the United States as spe-m. The mitogenic potency varies between both the streptococcal SAgs and the staphylococcal SAgs. The least potent of all superantigens so far examined is SPE-H, which produces a 50% maximal response (P50) in human PBL of 50 pg/ml, while SMEZ-2 is the most potent SAg known thus far with a P50 of 0·08 pg/ml. This is equivalent to 8 × 10−14 gm/ml or 21 000 molecules/ml [12]. By structural comparison, the staphylococcal and streptococcal superantigens build a large protein family (Fig. 1), indicating that they have all evolved from a single primordial superantigen. Primary amino acid sequence homologies vary greatly from as low as 15·5% sequence identity, e.g. between SEB and SEK to over 90% (SEA versus SEE). Nevertheless, all SAgs possess a characteristic PROSITE amino acid sequence signature K-X(2)-[LIVF]-X(4)-[LIVF]-D-X(3)-R-X(2)-L-X(5)-[LIV]-Y (PS00278). So far, 11 superantigens have been crystallised and all show remarkable similarities in their overall structure despite very different primary amino acid sequences (see below). An alignment of all streptococcal and staphylococcal superantigens based on amino acid sequence. Note that TSST-1 is a clear outlier of this group and that several clusters are formed based loosely on whether they are streptococcal or staphylococcal superantigens. The streptococcal SAgs SPE-A, SPE-H, SPE-I and SSA are related more closely to the staphylococcal SAgs than to any other streptococcal SAg (Fig. 1) and the genes for these toxins are all located on mobile elements, so it is likely that this SAg subgroup in S. pyogenes arose through the horizontal transfer from S. aureus rather than evolving from existing streptococcal superantigen genes. For many years the streptococcal proteins SPE-B and SPE-F were considered to be SAgs but have since been shown to be due to contamination from the potent SAg SMEZ-2. SPE-B is a cysteine protease and SPE-F (also known as mitogenic factor or MF) is in fact streptococcal DNase. SAgs have also been found in two different group C streptococci. The Streptococcus equi pyrogenic exotoxins (SePE) H, I, L and M are homologous to their S. pyogenes counterparts SPE-H, I, L and M (>98% sequence identity) indicating another horizontal transfer from S. pyogenes to S. equi or vice versa [20,21]. Another two SAgs have been identified from S. dysgalactiae called SDM [22] and SPE-Gdys[23]. SDM is most similar to SPE-M* and SPE-Gdys is most similar to SPE-G. Amino acid exchanges are outside the MHC class II and TcR binding sites suggesting that the GAS toxins and the non-GAS toxins are orthologues with identical functions. SPE-I and SPE-H are both located on the defective prophage SF370·2, SPE-L is located on the active prophage ΦNIH1·1 and SPE-M* was found on ΦspeL/speM[11,13,19]. This suggests that horizontal gene transfer between GAS and non-GAS occurred more recently than between GAS and S. aureus. SAgs have also been isolated from the Gram-negative bacteria Yersinia pseudotuberculosis and Mycoplasma arthritidis. The Y. pseudotuberculosis mitogens (YPM) A and B are 21 kDa proteins, which both target human TcR Vβ3, 9, 13·1 and 13·2 regions [24]. The M. arthritidis mitogen (MAM) is a 25-kDa protein that targets T cells bearing the Vβ6 and Vβ8 TcR [25]. These Vβ profiles differ from any profile of the ‘classical’ SAgs (Table 1). MAM and YPM-A/B are unrelated by amino acid sequence to the ‘classical’ SAgs and also lack the SAg family signature sequence. The protein structures of these SAgs have yet to be solved, so their mode of action remains a mystery. However, functional studies have shown that MAM binds preferentially to murine I-E or its human equivalent HLA-DR. The DR4, DR7 and DR12 subtypes present MAM most efficiently [26]. A study published in 1998 on the interaction between MAM and TcR indicated that MAM might contact not only the germ-line encoded TcRVβ region, but also the hypervariable CDR3 region [27]. This has also been shown to be the case for SPE-C (see below). The crystal structure of seven staphylococcal SAgs (SEA, SEB, SEC2, SEC3, SED, SEH and TSST) and four streptococcal SAgs (SPE-A, SPE-C, SPE-H and SMEZ-2) have been solved and revealed a common core-fold based on two globular domains: a smaller N-terminal pseudo β-barrel domain, which is most similar to the classical oligosaccharide/oligonucleotide-binding fold (OB-fold) found in many bacterial proteins that bind oligomeric molecules and a larger C-terminal β-grasp domain. The two domains are separated by a long solvent-accessible α-helix that extends down the centre of the molecule [28–35]. Several co-crystal structures of SAgs bound to MHC-II have also been solved (Fig. 2). The first was the crystal structure of SEB bound to HLA-DR1 followed closely by a structure of TSST-HLA-DR1 [36,37]. Both SEB and TSST possess an exposed hydrophobic loop region in the smaller N-terminal domain to bind to a hydrophobic groove located in the distal region of the invariant α1 domain of HLA-DR. However, the structures are not identical. SEB binds out to the side of MHC-II away from the peptide binding groove while TSST sits over the top of the groove and makes considerable contacts with peptide residues. As a consequence, TSST prevents any contact between MHC-II and the TcR while SEB relies on continued contacts between MHC and TcR to strengthen the interaction. A comparison of the co-crystal structures of three superantigens bound to MHC class II. The first is SEB which binds to the invariant α-chain of HLA-DR out to one side [36]. The TcR interacts with both SEB and MHC class II residues. The second shows TSST-1 bound to the same chain but is positioned further over the peptide groove and interacts with peptides [37]. The third structure is SPE-C bound to the polymorphic β-chain of HLA-DR2 via a zinc atom. It sits clearly over the top of the peptide groove and prevents any interaction between the TcR and the MHC molecules [42,43]. The affinity of SAgs towards MHC class II varies considerably. SAgs that utilize the generic HLA-DR α-chain (such as SEB and TSST) bind with relatively low affinity (KD∼10−5m). Other SAgs, such as SEA and SEE, have in addition to the generic low affinity α-chain binding site a high-affinity zinc-mediated binding site (KD∼10−7m) for the polymorphic HLA-DR β-chain. The zinc cation forms a tetrameric coordination complex with three residues from the C-terminal SAg domain and with a conserved histidine residue (H81) from the HLA-DR β-chain. This group of SAgs can bind to both sites of the molecule, cross-linking MHC-II molecules on the surface of APC and resulting in increased secretion of IL-1 and TNF-α[34,38–41] Yet another subgroup of the bacterial SAgs, such as SEH, SPE-C, SPE-H and SMEZ, lack the generic low-affinity α-chain site and only bind the polymorphic HLA-DR β-chain. The recently solved crystal structures of SPE-C bound to HLA-DR2a and of SEH:HLA-DR1 show that the C-terminal domain of the SAgs contacts the α-helix of the MHC-II β1-domain, as well as the N-terminal part of the bound peptide [42,43]. Peptide-restricted binding to MHC-II has also been shown in biochemical studies for other SAgs and raises the possibility that certain bound peptides could enhance the potency of the SAg by promoting high-affinity binding [44,45]. Yet another mechanism of MHC-II binding is seen with SED and SPE-C. Both SAgs are capable of forming zinc-mediated homodimers that can cross-link MHC molecules on the cell surface of APCs [32,46]. The SPE-C crystal structure revealed that the dimer structure is formed through an interface where the low-affinity N-terminal binding-site is normally located [32]. Less is known about the way in which SAgs bind TcR. In 1996, Mariuzza and colleagues succeeded in crystallizing the SAg SEC3 with a soluble form of the murine Vβ8·1 TcR β-chain [47]. In this complex, SEC3 makes multiple contacts with residues from the complementarity determinining region 2 (CDR2), the third framework region (FR3) and the 4th hypervariable loop (HV4) of the TcR β-chain. Furthermore, all the hydrogen bonds involve only main-chain atoms of the TcR β-chain, so that side-chain amino acid variation is less likely to affect the affinity of binding. In an attempt to correlate binding affinity to the potency to activate T cells, Leder et al. performed mutational analysis of the TcR binding site in SEC3. Most interestingly, their results showed a proportional relationship between binding affinity and potency, which is in sharp contrast to conventional peptides recognition, where even the smallest reduction in affinity results in complete loss of T cell stimulation [48]. In 1998, the crystal structure of SEB bound to murine Vβ8·1 TcR was solved and showed little difference from the SEC3-Vβ8·1 TcR complex [49]. However, in 2002 two additional complexes were determined − that of SPE-A bound to the same Vβ8·1 TcR and the streptococcal superantigen SPE-C bound to human Vβ2·1. These showed striking differences in SAg/TcR interaction. The SPE-A binds Vβ8·1 complex via many hydrogen bond-mediated contacts between the side-chain atoms of both molecules, suggesting that Vβ sequence specificity is required to restrict SPE-A reactivity, whereas SEC3/TcR interactions solely depend on conformation. The SPE-C/hVβ2·1 complex revealed a completely different binding mode involving not only residues in the CDR2, but also the CDR1 and the hypervariable CDR3 regions, including numerous specific electrostatic interactions between side-chain atoms of SPE-C and hVβ2·1 [50]. One question that has yet to be answered adequately is why there so many different binding modes for molecules intent on doing the same thing − that is bringing MHC and TcR together. One possible answer is that the different modes of MHC-II and TcR binding are a result of different immunological responses that might be related to different TcR β-chain subsets of T cells. This functional targeting of SAgs might also why a single more than one In a gene in the completed staphylococcal genome was identified that related genes with the characteristic PROSITE SAg family signature (PS00278). The gene were named staphylococcal toxins The most closely the TSST amino acid which to their a single isolated of the SAg family (Fig. 1). The addition of the to the SAg family a new that now includes The are all secreted from S. aureus and most strong One has been and the structure determined a SAg with the N-terminal pseudo β-barrel domain and the C-terminal β-grasp separated by a long solvent-accessible α-helix A structural comparison of to TSST is in A structural comparison between the superantigen TSST-1 and the secreted toxin a structure with TSST-1 with a larger C-terminal domain of the β-grasp and smaller N-terminal β-barrel domain. are both shared and the larger α-helix the conserved superantigen PROSITE K-X(2)-[LIVF]-X(4)-[LIVF]-D-X(3)-R-X(2)-L-X(5)-[LIV]-Y (PS00278). However, so far of the have any of the functional of all SAgs, such as MHC class II binding or T cell while they are they appear to have very different functions. The of the complete genome sequence of S. aureus and revealed additional all on were found within on strain and within the homologous on strain while were found within on strain The of is but their within suggests a as and it is most likely they have some in immune of both the adaptive and immune a causes murine The SAgs were discovered first by in and were to as The T cell response to antigens is similar to the response to bacterial SAgs with expansion of TcR Vβ subsets The SAg gene was identified within the of the genome and not show any homology to the bacterial SAg genes The gene is a II protein with a amino acid polymorphic region the which is for the TcR Vβ is present in and of only a mouse The SAg molecule is an of the in B cells by Vβ T cell and promoting B cell The SAg is in and causes T cell as a result of in the As a the of an the identical SAg be by the lack of T cells, the mouse from For many was the only that was known for certain to a In 1996, et al. that human B cell the with a B cell mitogen that stimulate bearing T cells suggesting the of a encoded SAg the same group showed that the described SAg activity is in fact encoded by of the human which is activated by is located on 1 within the first of which an Furthermore, of is by of were identified and all of mitogenic activity towards and T cells. is identical to the identified associated The that SAgs might the of the similar to in and could to e.g. the T cell in The staphylococcal superantigens and are potent toxins for staphylococcal food of less than 1 of toxin are sufficient to trigger vomiting in This appears to be from the SAg activity but this remains A within the N-terminal domain has been with the but the mechanism that to the or a specific receptor molecule have not yet been identified toxic shock syndrome by S. aureus can be considered as a syndrome and includes such as fever and major is through TSST is as the primary causative agent for which is associated with the of certain of high that the of S. aureus. In contrast to other staphylococcal SAgs, TSST has the ability to the TSST and other staphylococcal SAgs have been associated with which can in any This is by the that these toxins induce in in the and in [18]. by S. is the most form of streptococcal with of to The are very similar to in but is associated with or SAgs have been in and includes the The and genes were found higher in from compared to lack of was found to be associated with an increased for and SAgs were found in several from a is the of It occurs in and with S. pyogenes. is a immune response to the and it has been that the T cells might be by SAgs. several streptococcal SAgs have been identified from The genes for were found in high on serotypes M3 and and on M89 while and SPE-M* were found in M18 et al. showed that and SPE-M* were more common in from compared to with a common target of the SAgs and SPE-M* are T cells bearing the with is an acute of that mainly and is now as the of in in the is associated with of T cells and and there is a remarkable and fever in the is when given suggesting that the causative agent is a Several the expansion of T cells bearing the which towards a SAg in the A between and the Y. pseudotuberculosis mitogenic factor (YPM) has also been It has been that SAgs might to the of by T cells that are specific for there is of SAg it has been that SAgs the the or of T cell and induce a of for this from an of multiple where it was shown that of SEB to from stimulation of the peptide specific T cells resulting in a of the and colleagues found a TcR in T cells from towards suggesting the activity of a SAg showed that the mitogenic activity was encoded by a gene on an named and that occurred only in It was shown by the same group that is identical to one of the the SAg (see One of the most is why some diseases GAS while show only such as Several have shown an between streptococcal and or of the GAS However, most were out the of streptococcal SAgs were discovered and more recent has not the genes for some potent SAgs, such as and are present in all GAS A lack of SAgs has also been as a possible factor for the of to of responses in have been by the number of known SAgs that most of the SAgs have been identified from the various completed staphylococcal and streptococcal a larger of be examined to this differ in their ability to present different SAgs and recent structural analysis revealed that the bound peptide also a in SAg binding to the MHC molecule (see an MHC with to SAg and colleagues showed that the of the the of streptococcal human strong from systemic by streptococcal whereas other increased the of This was the first clearly between MHC class II and the activity of SAgs. It now remains to the MHC with SAgs expressed by the Superantigens have a of since the of their in a of about their structure and has been yet little has been on their in other than the food poisoning and toxic Nevertheless, there has been considerable that they are in other are a family of molecules, to the adaptive immune response by targeting the two most antigen molecules the TcR and MHC class II. are clearly as a a immune of this are is not certain is how this stimulation of many T cells results in for the Both S. aureus and S. pyogenes are in so the fact that they have the to activate the immune response in such a that their be and that the immune with their this T cell is of some to the as