Michel Goldman MD, PhD, Institute for Medical Immunology, 8, rue Adrienne Bolland, B-6041 Charleroi, Belgium. E-mail: firstname.lastname@example.org
Characterization of the Toll-like receptor (TLR) family and associated signalling pathways provides a key molecular basis for our understanding of the relationship between exposure to microbial products and susceptibility to immune-mediated disorders. Indeed, ligation of TLR controls innate and adaptive immune responses by inducing synthesis of pro- as well as anti-inflammatory cytokines and activation of effector as well as regulatory lymphocytes. TLRs are therefore considered as major targets for the development of vaccine adjuvants, but also of new immunotherapies. Herein, we review the potential of TLR ligands as a novel class of pharmaceuticals for the prevention or treatment of allergic disorders.
The prevalence of allergic asthma and other atopic disorders has increased dramatically during the last 25 years in developed countries and asthma is currently the most frequent chronic disorder in children in the western world. As proposed initially by Strachan , improved hygiene and health policy regulations might have favoured the blossoming of allergies by modifying the immunological stimulations provided by the microbial environment. Indeed, the ‘hygiene hypothesis’ is supported by a number of epidemiological observations, including the lower risk of developing allergy in children exposed in early life to other infants or animals . Furthermore, exposure to certain microbes or microbial products also appeared to ensure some level of protection against allergic disease [2–7]. Interpretation of these epidemiological associations has sometimes been controversial, and might be complicated by confounding factors. For example, after the pioneer Japanese study showing an inverse association between tuberculin responses and atopic disorders , a protective effect of neonatal vaccination with Mycobacterium bovis bacille Calmette–Guérin (BCG) against allergy was observed in Guinea-Bissau  but not in Sweden , suggesting that genetic and/or environmental factors might interfere with the action of BCG vaccine. In fact, results of an Australian study suggested that the benefit of neonatal BCG vaccination might depend on the prevalence of tuberculosis infection in the considered population . Furthermore, the anti-asthma effect of neonatal BCG vaccination could be influenced negatively by subsequent infection with respiratory syncytial virus, as documented in a murine model .
Among microbial products which might affect the occurrence and severity of allergic asthma, bacterial lipopolysaccharide (LPS), the endotoxin of Gram-negative bacteria, elicited the greatest interest. Well-performed epidemiological and immunological studies have established that endotoxin levels in house dust were related inversely to allergen sensitization  and development of atopic asthma  in children. Interestingly, the consequences of endotoxin exposure seem to depend upon the time of exposure as the results of two independent epidemiological studies sustain the initial observation of Michel et al., suggesting that endotoxin exposure might exacerbate asthma in patients with established disease [13–15]. This double-edged effect of endotoxin was established clearly in a rat model of asthma in which endotoxin was protective when administered before allergen sensitization, whereas it exacerbated lung inflammation when given in sensitized animals . There is suggestive epidemiological evidence that other microbial products might regulate the course of allergic diseases in humans, including fungal polysaccharides, muramic acid (a bacterial cell wall peptidoglycan), bacterial DNA and schistosomal phosphatidylserine [17–21]. Most of these microbial products act on immune cells through membrane receptors belonging to the family of Toll-like receptors (TLR) . Endotoxin uptake also depends on the molecule CD14, either membrane-bound or in soluble form . Interestingly, polymorphisms of the genes encoding TLR4 and CD14 were found to influence the severity of asthma and its relation with endotoxin exposure [24–26]. Similarly, a genetic variation in TLR2 promoter was shown to be a major determinant of susceptibility to asthma and atopy in farmers' children .
The T helper type 1 (Th1)/Th2 paradigm was initially proposed as the immunological concept underlying the hygiene hypothesis. This was supported by the evidence that Th1 responses are compromised in early life, because of both an intrinsic T cell defect in interferon (IFN)-γ production  and an impaired capacity of newborn antigen-presenting cells to secrete the bioactive form of interleukin (IL)-12 [29,30]. However, the idea that infections prevent Th2-driven allergy by promoting Th1 responses did not fit with the evidence that parasitic diseases which induce strong Th2 responses were also protective . This was especially well documented in schistosoma-infected children who were found to be protected against allergy and to lose this protection after anti-helminthic therapy [31,32]. Furthermore, there is evidence that the hygiene hypothesis also applies to autoimmune diseases caused by Th1-type immune reactions as the incidence growth of juvenile autoimmune diabetes, multiple sclerosis and Crohn's disease parallel that of allergic disorders . The demonstration that regulatory T cells are potent suppressors of experimental allergic inflammation  and the observation that deficient T cell regulatory activities are associated with atopic diseases in humans [35–37] led to the proposal that infectious agents might actually prevent allergic disorders by enhancing regulatory T cell activities [6,33,38]. Indeed, infection with the nematode Heligmosomoides polygyrus or injection of heat-killed M. vaccae were shown to protect mice against allergic airway inflammation by inducing regulatory T cells producing IL-10 and transforming growth factor (TGF)-β[34,39].
The hygiene hypothesis has obvious implications for the prevention of allergic diseases, as it suggests that enhanced exposure to microbes in early life might represent a valuable strategy to reduce the burden of atopic diseases. Indeed, perinatal oral administration of the probiotic Lactobacillus rhamnosus was shown to have a preventive effect on the development of atopic eczema in a randomized placebo-controlled trial , and a single intradermal injection of heat-killed M. vaccae exerted a short-term beneficial effect in children with atopic dermatitis over 5 years old . The pharmaceutical development of natural and synthetic TLR ligands offers new perspectives for more selective targeting of the immune system and several of them are currently under trial in allergic disorders . Herein, we review the experimental evidence that TLR ligands modulate allergic responses and the putative mechanisms involved, focusing on the induction of regulatory T cells. We then address key questions to be solved in order to translate into clinical practice the promising results obtained in experimental settings. We will limit our review to TLR2, TLR4, TLR7 and TLR9 ligands because TLR-mediated modulation of Th2-type allergic responses has been best documented for those four TLRs.
Dual effects of TLR4 ligands in allergic inflammation
Bacterial LPS is the prototypical TLR4 ligand. In murine models of allergic lung inflammation, divergent enhancing or suppressive effects of LPS were reported. Although LPS was shown initially to enhance IgE synthesis in vitro, subsequent in vivo studies established that LPS inhibits the development of allergen-specific IgE when given before sensitization . It appeared that the dose of LPS and the period of administration might represent critical factors determining the net effect of TLR4 ligation in the course of allergic responses. Low doses of inhaled LPS were found to promote Th2 responses to the sensitizing antigen and eosinophilic inflammation, whereas high doses of LPS induced Th1 responses without eosinophilic lung inflammation . However, increased numbers of neutrophils accumulated in lungs of mice having received high-dose LPS . Interestingly, the pro-allergic effect of low doses of LPS was found to depend on the Myd88-dependent signalling pathway, at least when administered by the airway route . These observations were confirmed in a model of cockroach antigen-induced asthma, where high doses of inhaled LPS suppressed expression of Th2-type cytokines and decreased eosinophilic inflammation and airway hyperresponsiveness but induced neutrophil inflammation . The influx of neutrophils promoted by high-dose LPS might be clinically relevant, as neutrophils are thought to play an important role in the pathogenesis of severe asthma . Furthermore, in a small cohort of asthmatic patients, LPS inhalation was shown to enhance bronchial obstructive response in association with a neutrophilic response . In the model of ovalbumin-induced lung inflammation in mice, it has been possible to avoid lung influx of neutrophils while maintaining the suppressive effect on the Th2 response by administering LPS intravenously instead of intranasally . When given systemically, LPS was shown to act by inducing nitric oxide synthase activity . The timing of LPS administration in relation with allergen sensitization and challenge represents another important parameter to consider as early exposure to LPS might be required to achieve optimal prevention of the allergic response [16,50]. Two recent studies suggested strongly that the duration and cadence of exposure is also critical. First, intranasal LPS was found to protect mice from Th2-type allergic responses when administered daily but not weekly . Secondly, TLR4- deficient mice display exacerbated asthmatic responses when exposed for a prolonged but not for a short period to aerosols of ovalbumin contaminated with low doses of LPS . The fact that Th1 and Th2 responses were modified in a similar manner in this setting would be consistent with the involvement of regulatory T cells, induced possibly by repeated exposure to endotoxin.
Several studies indicate that the impact of LPS exposure on allergic responses might be greater in early life than in adult individuals. Experimentally, repeated administration of LPS to newborn mice by the nasal route resulted in the emergence of regulatory T cells which prevented the occurrence of a Th2-type response upon exposure to allergen . Similarly, ex vivo studies on the nasal mucosa of atopic children revealed that LPS was able to inhibit local allergic inflammation by skewing immune responses from Th2 to Th1 with a concomitant induction of IL-10, an effect which was not observed in adults .
Modulation of atopic disorders by TLR4 ligands involve several cell types engaged in the induction or effector phases of allergic responses. Most studies focused on the action of TLR4 ligands on dendritic cells (DC) because of their central role in the instruction of Th2 cells, which orchestrate the allergic reaction [55,56]. TLR4 ligands are known to induce DC maturation resulting in the up-regulation of major histocompatibility complex (MHC) and co-stimulatory molecules and in the production of cytokines which govern the polarization of the CD4+ T cell responses. As shown in Fig. 1, a number of factors present in the DC microenvironment influence the type of T cell responses elicited by TLR4-stimulated DC. In the context of strong inflammatory stimuli, especially when other TLR ligands are present and IFN-γ is produced, DC will secrete high levels of bioactive IL-12p70, express the Delta-4 notch ligand, and as a consequence will instruct CD4+ T cells to differentiate into Th1 cells [56–58]. When histamine and/or thymic stromal lymphopoietin (TSLP) produced by epithelial cells, mast cells and stromal cells are present at high level, DC promote Th2 polarization in relation with a low production of IL12p70 and the expression of the Jagged notch ligand [56,57,59,60]. The presence of TGF-β together with IL-6 and other inflammatory mediators favours the IL-23/IL-17 axis leading to the emergence Th17 cells that are stabilized by DC-derived IL-23. Th17 cells, which produced IL-17 preferentially, promote neutrophil-mediated inflammation and are involved in major T cell-mediated autoimmune disorders . In the context of asthma, Th17 cells might play a pathogenic role not only by promoting neutrophil influx but also by inducing the production of pro-fibrotic cytokines by bronchial fibroblasts  and the release of the eosinophil chemoattractant eotaxin/CCL11 ligand by airway muscle cells . Regulatory T cells (Treg) represent another major pathway of differentiation of naive T cells upon interaction with DC. Indeed, we documented induction of regulatory T cells upon interaction of naive CD4+ T cells with autologous LPS-exposed DC in the absence of exogenous antigen . Several factors that enhance DC with the capacity to induce Treg responses have been identified recently. These factors include TGF-β in the absence of inflammatory cytokines, IL-10, glucocorticoids, neuropeptides as vasointestinal neuropeptide (VIP), vitamin D3 and thrombospondin [65–69]. A recent study indicates that induction of regulatory T cells by TLR4-stimulated DC is favoured by blockade of the RP105–MD1 complex, which acts as a competitor of the TLR4–MD2 complex for LPS binding . Although there is one report suggesting that direct engagement of TLR4 on regulatory T cells can induce their activation , further studies failed to document such a direct effect .
In the context of allergic inflammation, epithelial cells and mast cells also represent potential targets for TLR ligands. There are important differences in the expression of TLR receptors on these cell types across species, so that human studies are essential to explore clinical implications . With regard to human mast cells, there is an indication that TLR4 ligands could promote the release of cytokines associated with Th2 responses, including IL-5 and IL-13 .
Because LPS is not suitable for clinical formulations, natural and synthetic ‘non-toxic’ derivatives of LPS have been developed by the pharmaceutical industry, primarily as vaccine adjuvants. Monophosphoryl lipid A (MPL) is the lead compound of this category of TLR4 ligands. Natural and synthetic versions of MPL are currently under development for applications in allergic disorders, essentially as prophylaxis of allergic flares. Indeed, a seasonal allergy vaccine which includes either grass pollen or tree pollen formulated with MPL (PollinexR Quattro; Allergy Therapeutics, Worthing, Essex) was shown to be efficient in patients with seasonal allergic rhinitis; it reduces symptoms and allows decreased use of palliative medications when given as an ultra-short course of four injections before the allergy season . This vaccine was usually well tolerated and its efficacy was confirmed by the reduction in the seasonal boost of allergen-specific serum IgE levels and by changes in the levels of Th1 and Th2 cytokines secreted in response to allergen challenge in vitro. A sublingual version of a similar MPL-based product is under development (MPL-103; Allergy Therapeutics). Another strategy proposed initially to tackle allergic rhinitis using MPL analogues is based on intranasal administration of TLR4 ligands alone, in the absence of antigen (CRX675 & CRX527; GlaxoSmithKline, Hamilton, MT, USA). In view of the data available on sublingual immunotherapy , the sublingual route certainly deserves further consideration both for TLR4 ligand monotherapy and TLR4-based anti-allergy vaccines.
TLR2 ligands: optimal inducers of regulatory T cells?
With regard to TLR4 ligands, the divergent effects of TLR2 ligands were reported in experimental murine models of allergic lung inflammation. On one hand, the addition to ovalbumin of either Pam3Cys or peptidoglycan at the time of subcutaneous or intranasal immunization was found to result in enhanced Th2 responses and increased airway hyperresponsiveness upon airway ovalbumin challenge [78,79]. This pro-allergic effect of TLR2 ligands was attributed to activation of a distinct extracellular-regulated kinase (ERK)-dependent pathway in DC leading to expression of genes promoting Th2 responses  and to activation of mast cells [74,81]. On the other hand, TLR2 ligands such as lipopeptide CGP 40774, lipoprotein I from Pseudomonas aeruginosa or Pam3CSK4 were found to be protective against allergic airway inflammation in other settings [82–85]. Interestingly, when TLR2 ligands were administered in the airways, eosinophilic inflammation was replaced occasionally by neutrophilic inflammation [83,84]; this might be related to local activation of the IL-23/IL-17 axis because TLR2 ligands were shown to induce high levels of IL-23 .
Several mechanisms could contribute to the protective action of TLR2 ligands against allergic responses. First, TLR2 ligands are able inhibit the production of Th2 cytokines by allergen-specific T cells [82,87]. Secondly, TLR2 ligands act directly on Treg, resulting in Treg expansion and increased Treg cell activity after extinction of TLR2 signalling (Fig. 2) [88–90]. Thirdly, certain TLR2 ligands such as zymosan and schistosomal phosphatidylserine induce a tolerogenic programme in DC characterized by high production of IL-10 and low production of IL-6 (Fig. 2) [91,92] In the case of zymosan, the induction of this programme depends on the simultaneous engagement of dectin-1 with subsequent activation of ERK/mitogen-activated protein (MAP) kinase and also involves induction of TGF-β production by macrophages . Once loaded with antigenic peptides, TLR2-induced tolerogenic DC promote the emergence of adaptive Treg with potent suppressive activities on both Th1 and Th2 responses. This sequence of events has been proposed as a basis for the inhibition of allergic responses during schistosomiasis [21,92].
Although the conditions under which TLR2 ligands elicit regulatory circuits able to prevent or control allergic inflammation in humans remain to be specified, several observations in populations protected against allergic disorders suggest that TLR2 is an appropriate target for future immunotherapy in atopic individuals, including the identification of a TLR2 promoter gene polymorphism and the increased expression of TLR2 in farmers' children [27,93], and the modified responses to TLR2 ligands in schistosome-infected children . A triacylated lipid A derivative, which signals through TLR2 and TLR4, is currently in preclinical development for the prevention of allergic inflammation (OM-174; OM-Pharma, Meyrin, Switzerland). Repeated exposure to TLR2 ligands at short time intervals and simultaneous engagement of other DC receptors might represent important factors for the efficient suppression of allergic responses.
TLR9 and TLR7 ligands: plasmacytoid DC as key targets for immunotherapy?
TLR9 and TLR7 are intracellular endosomal receptors for nucleic acids . In humans, TLR9 is expressed on B cells and plasmacytoid DC . The latter cells represent a major source of type 1 IFN, which acts on many cell types and promotes directly and indirectly Th1 polarization of CD4+ T cell responses . Oligodeoxynucleotides containing unmethylated cytosine-guanine motifs (CpG) are potent TLR9 activators which induce the release of high levels of type 1 interferon and stimulate strong Th1 responses in vivo[95,96]. CpG oligodeoxynucleotides (ODN) are therefore developed as adjuvants for vaccines against intracellular pathogens and cancer, but are also considered as candidates for immunotherapy in atopic disorders [95–97]. In fact, CpG-ODN were shown to redirect Th2 responses towards Th1 responses both in mouse models [98,99] and in clinical trials [100,101]. Interestingly, CpG-ODN suppressed allergic inflammation even when administered after the sensitization period [99,102]. Moreover, they were shown to prevent the chronic process of airway remodelling associated with Th2 responses, both in rodents and non-human primates [103,104]. The beneficial action of CpG-ODN in allergic disorders cannot be attributed solely to redirection of CD4+ T cell responses, as it was also observed in mice genetically deficient in IL-12 and IFN-γ. Inhibition of IgE production by B lymphocytes [106,107], suppression of the production of chemokines involved in trafficking of naive and Th2 cells [100,108], and induction of indoleamine 2,3-dioxygenase (IDO) activity in pulmonary epithelial cells  might represent important additional mechanisms by which CpG-ODN might protect against allergic lung inflammation. It is likely that most of these effects are related to the production of type 1 and type 2 IFNs consecutive to the activation of plasmacytoid DC. Furthermore, there is suggestive evidence that CpG-activated DC can induce the emergence of Treg[110,111]. Whatever the mechanisms listed in Table 1 involved, early clinical trials with CpG-ODN for atopic disorders provided encouraging results, both in terms of biomarkers of Th2 responses and symptoms [100,101]. Indeed, in patients with seasonal allergic rhinitis to ragweed, six subcutaneous injections of CpG-ODN linked to Amb a 1 (the major ragweed allergen) administered at weekly intervals suppressed local eosinophilia and decreased ragweed-induced symptoms in the following season . Results of a large clinical trial with such a product (Tolambatm; Dynavax Technologies, Berkeley, CA, USA) should be available in 2007. Another product in which oligonucleotides are packaged into virus-like particles decorated with house dust mite allergen extract also showed preliminary evidence of efficacy in patients with allergic rhinoconjunctivitis and asthma enrolled in a phase 2 trial (CYT005-AllQbG10; Cytos Biotechnology, Schlieren, Switzerland). Interestingly, as CpG-ODN alone (without concomitant administration) was shown to be efficient in animal models, CpG-ODN are also developed as monotherapy for atopic disorders (AVE0675, AVE7279; Sanofi-Aventis/Coley, Paris, France).
Table 1. Mode of action of Toll-like receptor 9 (TLR9) and TLR7 ligands in allergic disorders.
DC: dendritic cells; IDO: indoleamine 2,3-dioxygenase; Th: T helper; Treg: T regulatory cells.
The effects of TLR7 ligation are similar to those elicited by TLR9 ligation, including the stimulation of plasmacytoid DC to produce high levels of type 1 IFN . Synthetic TLR7 ligands such as imiquimod, resiquimod and related imidazoquinolines were found to inhibit Th2 responses [112–114], IgE production by B cells  and experimental lung allergic inflammation [114,116], with an efficiency usually comparable to that of CpG-ODN. A potential advantage of TLR7 ligands is their capacity to stimulate IFN-γ production directly by memory CD4+ T cells .
TLR ligands offer exciting perspectives for the development of new immunotherapeutic strategies for allergic disorder, but important questions remain to be answered before TLR-based treatments enter routine clinical practice. We still have to define which are the TLR to be targeted for optimal prevention or treatment of allergy, to determine whether or not TLR ligands have to be combined with antigen, and the best cadences and routes of administration. Safety issues obviously represent a major concern, especially for products designed to be administered in early life. In several models of allergy, TLR ligands appear as double-edged swords, detrimental or beneficial effects being observed depending on the dose of the product, and the timing and route of administration. Furthermore, TLR ligands used to redirect CD4+ T cell responses may favour autoimmune reactions, as shown experimentally for CpG-ODN . TLR2 ligands which elicit the production of high levels of IL-23 could have a similar downside . Although clinical data available so far suggest that TLR ligands are safe in the context of allergy, careful follow-up of patients from phase 1 to post-marketing studies will be required to establish TLR ligands as pharmaceutical substitutes for environmental microbes to protect against allergic disorders.
The Institute for Medical Immunology is supported by the government of the Walloon Region, GlaxoSmithKline Biologicals, the Fonds National de la Recherche Scientifique, and an Interuniversity Attraction Pole of the Belgian Federal Government. Michel Goldman serves as consultant for GlaxoSmithKline Biologicals (Rixensart, Belgium) and OM-Pharma (Meyrin, Switzerland).