Main

The immune system has been divided traditionally into an INNATE and ADAPTIVE component, each of which has different roles and functions in defending the organism against foreign agents, such as bacteria or viruses. The innate immune system has developed a series of conserved receptors, known as pattern-recognition receptors (PRRs), that recognize specific pathogen-associated molecular patterns (PAMPs), thereby allowing the innate immune system to distinguish self-molecules from pathogen-associated non-self structures and initiate the host defence response1. Recognition of PAMPs by PRRs results in the activation of signalling events that induce the expression of effector molecules, such as cytokines, chemokines and co-stimulatory molecules, which subsequently control the activation of an antigen-specific adaptive immune response. An important advance in our understanding of early events in microbial recognition and the subsequent development of immune responses has been the identification of Toll-like receptors (TLRs) as key PRRs of the innate immune system1.

The gene that encodes the Toll protein was shown to be involved in embryonic development in Drosophila, and was identified subsequently as being an essential molecule for driving the immune response. TLRs are mammalian homologues of this protein that can detect PAMPs2. The first report of a mammalian TLR and its involvement in host defence — TLR-4 as a receptor for LIPOPOLYSACCHARIDE (LPS)3 — was followed rapidly by the discovery that the human genome contains several TLRs — ten have been found so far, TLR-1–TLR-10. Members of the TLR family share characteristic extracellular and cytoplasmic domains. Their extracellular domains include several leucine-rich repeats (LRRs), whereas the cytoplasmic domain is similar to the cytoplasmic portion of the interleukin-1 receptor (IL-1R), which is commonly known as the Toll/IL-1R homologous region (TIR domain)4. Consistent with their function as PRRs, TLRs are expressed mainly in the cell types that are involved in the first line of defence, such as DENDRITIC CELLS, MACROPHAGES, NEUTROPHILS, mucosal epithelial cells and dermal endothelial cells5. Distinct subsets of dendritic cells and dendritic-cell precursors express different subsets of TLRs that enable them to induce specific patterns of immune response to different pathogens. For example, plasmacytoid dendritic cells strongly express TLR-9, whereas myeloid dendritic cells do not. Instead, they express other TLRs, including TLR-4, which cause them to be activated by a different set of PAMPs, including LPS6. Activation of dendritic cells by stimulation of TLRs results in the production of cytokines, such as IL-12, that induce T-helper type 1 (TH1) CELLS, which tends to direct the adaptive immune response to differentiate towards a TH1 phenotype6. The identification of TLRs as PRRs that are involved in controlling dendritic-cell activation led to substantial interest in these receptors as possible regulators of adaptive immune responses and, by inducing a specific cytokine environment, their involvement in the pathogenesis of several chronic inflammatory diseases. Therefore, we focus our discussion on the emerging role of TLRs in the pathogenesis of inflammatory diseases and how agonists or antagonists of these receptors might be used to treat various pathological conditions. We describe the basic mechanism of an allergic inflammatory response as an example to show how targeting TLRs might offer promising new therapeutic benefits. In addition, the current and future use of TLR agonists and antagonists in other chronic inflammatory diseases, and the activation of TLRs by exogenous or endogenous ligands are discussed.

TLRs: members and ligands

Several reviews describe elegantly the known members of the TLR family, their ligands and the signalling pathways that are induced by stimulation of the receptors4,5,6. Therefore, we summarize only briefly the current knowledge, focusing on the most important information that is required to understand the function of TLRs (Table 1). Different TLRs have been shown to be crucial for the selective recognition of specific PAMPs that are common to a range of either GRAM-NEGATIVE or GRAM-POSITIVE BACTERIA. The observation that a point mutation in the TIR domain of the mouse Tlr4 gene abolished the response to LPS provided the first evidence that this particular receptor might be involved in the innate immune response to Gram-negative bacteria3,7,8. By contrast, Tlr-2-deficient mice have a normal inflammatory response to LPS, but macrophages from these animals are less responsive to Gram-positive bacterial cell walls and PEPTIDOGLYCAN. Indeed, TLR-2 recognizes LPS from Porphyromonas gingivitis9 and Leptospira interrogans10, which differs in structure from the LPS of Gram-negative bacteria. These results are evidence of TLR selectivity in PAMP recognition, although other TLRs can recognize the same components of both Gram-negative and Gram-positive bacteria. Indeed, flagellin, the principal element of bacteria flagella, is a highly virulent molecule that is recognized by TLR-5 (Ref. 11), whereas TLR-9 is required for the inflammatory response that is triggered by hypomethylated bacterial DNA12. TLRs also trigger an innate immune response to viruses that produce double-stranded RNA, and a reduced response to the dsRNA analogue polyinosine–polycytidylic acid was observed in Tlr-3-deficient mice13. Recently, Takeuchi et al.14 showed that TLR-1 is involved in the recognition of mycobacterial lipoprotein and triacylated lipopetides. In addition, TLRs can form multimeric complexes (homodimers or heterodimers) to increase the spectrum of molecules that they recognize. For example, the cytoplasmic domain of TLR-2 can form functional pairs with TLR-6 and TLR-1, leading to signal transduction and cytokine expression after ligand activation4.

Table 1 TLR-family members and exogenous and endogenous ligands
Full size table

TLR signal transduction

Different TLRs can exert distinct, but overlapping, sets of biological effects, and increasing evidence indicates that this can be attributed to both common and unique aspects of the signalling mechanisms (Fig. 1). TLRs activate signalling pathways that are similar to those engaged by IL-1 because of the presence of the TIR domain. The TIR domain can interact with the adaptor protein MYD88 (Ref. 15), and activation through TIR domains also leads to the activation of the IL-1R-associated kinase (IRAK)15,16, a serine kinase that activates another adaptor molecule; tumour-necrosis factor (TNF)-receptor-associated factor 6 (TRAF6)15,17. Recruitment of TRAF6 leads to the activation of inhibitor of NF-κB (IκB) kinase (IKK). This event frees NUCLEAR FACTOR-κB (NF-κB) from IκB, and allows its nuclear translocation and the subsequent transcriptional activation of many pro-inflammatory genes, which encode cytokines, chemokines, proteins of the complement system, enzymes such as cyclooxygenase-2 (COX-2) and the inducible form of nitric oxide synthase (iNOS), adhesion molecules and immune receptors. All of these molecules are involved in engaging and controlling the innate immune response, which is essential for pathogen elimination, and in orchestrating the transition to an adaptive immune response4.

Figure 1: Signalling pathways induced by TLR ligation.
figure 1

The Toll-like receptor (TLR)-family members share many signalling components. These include the adaptor MYD88, the Toll-interacting protein (TOLLIP), interleukin (IL)-1-associated kinase (IRAK) and TRAF6 (tumour-necrosis factor (TNF)-receptor-associated factor 6). By contrast, recent evidence indicates that some TLRs might use other receptor-specific signalling components. For example, TLR-4 signals through another adaptor molecule known as MAL or TIRAP (Toll/IL-1 receptor (TIR)-domain-containing adaptor protein) that is not used by other TLRs. Protein kinase R (PKR) functions downstream of TIRAP, but its importance has not been established yet. IκB, inhibitor of NF-κB; NF-κB, nuclear factor-κB; NIK, NF-κB-inducing kinase.

Full size image

Recently, another molecule, the Toll-interacting protein (TOLLIP), has been proposed to interact with the TIR signalling domains and participate in signal propagation18,19. Another TLR signalling pathway has been revealed by the observation that certain LPS-induced responses do not require MYD88. This MYD88-independent TLR-4 signalling was identified because NF-κB activation was delayed, but not absent, in MYD88-deficient cells, whereas it was completely suppressed in TLR-4-deficient cells20,21. These studies led to the identification of a new molecule known as the TIR-domain-containing adaptor protein (TIRAP, also known as MAL, for MYD88-adaptor-like protein), which interacts specifically with TLR-4 and not with other TLRs and is probably responsible for this MYD88-independent signalling. A biological consequence of this signalling pathway was found by Kaisho and co-workers22, who observed that the LPS-induced maturation of dendritic cells does not require MYD88 but does require TLR-4. These recent findings of differences in relation to the signalling proteins that are used after TLR stimulation could open up possibilities to selectively interfere with specific signalling components that are unique to particular TLR-family members.

TLRs as therapeutic targets in sepsis

Our current understanding of the pathogenesis of sepsis indicates that bacteria and bacterial products activate an uncontrolled network of host-derived mediators, such as pro-inflammatory cytokines, that can lead ultimately to multi-organ failure, cardiovascular collapse and death23,24,25. On the basis of the key role of TLRs in the recognition of microbial components, it is clear that an inappropriate TLR response to bacterial signals could have important consequences during infection, leading to exaggerated responses, such as sepsis23,24. Antagonists for TLR proteins might, therefore, be useful tools to counteract the harmful pro-inflammatory response that is associated with systemic microbial infections. Three basic strategies have been proposed that could reduce signal transduction through TLRs in sepsis26: first, soluble TLRs that bind and neutralize the microbial ligands; second, small molecules or antibodies that block the binding of ligands or protein–ligand complexes to the receptors; and third, specific small-molecular-mass inhibitors of the common signalling pathways that are induced by ligand–TLR interaction.

However, in addition to the potential therapeutic effect of blocking TLRs in sepsis, these receptors have also provided new opportunities to design therapeutics for the treatment of chronic inflammatory diseases or human diseases that are caused by dysregulation of the immune system. The development of TLR agonists or antagonists has the potential to create pharmacological tools to modulate these responses, and some examples of this role are discussed below.

TLRs as therapeutic targets in asthma

How do TLR-driven innate immune processes regulate and control a disease-associated adaptive immune response? Allergic asthma is chosen as an example of a chronic inflammatory disease that has a relatively well-understood adaptive immune response to show how TLR agonists or antagonists might offer possibilities for therapeutic intervention.

The initial acquired immune response that is responsible for the development of allergic diseases such as asthma is the generation of allergen-specific CD4+ TH2 CELLS (Fig. 2). Analysis of blood, bronchoalveolar lavages and bronchial mucosal biopsies from patients with allergic asthma showed a predominant activation of TH2-like T cells that produced IL-4, IL-13 and IL-5, but no IL-2 or interferon-γ (IFN-γ)27,28. IL-4 is required for TH2-cell development and, together with IL-13, is intimately involved in the regulation of IMMUNOGLOBULIN (Ig)E production, which is fundamental to the pathogenesis of allergic asthma29,30. IL-5 is the principal, and possibly the only, TH2 cytokine to be involved in the accumulation of EOSINOPHILS, another characteristic feature of asthma31. Crosslinking of allergen-specific IgE on MAST CELLS and activation of T cells and eosinophils during subsequent encounters with antigens stimulates the release of various preformed and newly synthesized products, including histamine, cytokines and chemokines, which together lead to characteristic airway changes that contribute to obstruction, airway hyperresponsiveness, goblet-cell metaplasia, mucus overproduction and mucosal oedema32,33,34. However, despite the relatively well characterized cellular mechanisms and the introduction of potent and effective drugs, the prevalence, severity and mortality rates of asthma have increased markedly over the past few decades and, so far, no effective preventive measure exists.

Figure 2: Asthma pathogenesis and potential involvement of TLRs.
figure 2

T-helper type 2 (TH2)-driven inflammation leads to characteristic airway changes that contribute to airway obstruction, airway hyperresponsiveness and mucus overproduction. Activation of Toll-like receptors (TLRs) by bacteria or bacterial products might shift the predominantly TH2-driven inflammation into a more protective TH1 response, or might lead to the induction of regulatory T cells, which subsequently might prevent the development of allergic asthma. By contrast, activation of TLRs by viruses or bacteria might induce disease exacerbations. APC, antigen-presenting cell; Ig, immunoglobulin; IL, interleukin.

Full size image

The mechanism that underlies the preferential activation of TH2 cells in allergic asthma is still not known. However, the key event that determines the cytokine phenotype of allergen-specific T cells seems to occur early in life. The current view is that the divergence of pre-existing, foetally primed TH2 immunity towards TH1 fails in ATOPIC individuals, leading instead to the persistence and boosting of foetal TH2 responses after allergen contact35. Indeed, it has been shown that the rate of postnatal maturation of TH1 function is slower in children with a high genetic risk of atopy36. Moreover, it has been proposed that the increase in the incidence of atopic disorders is linked to a decrease in the prevalence of infections that induce TH1 responses early in life. This theory is referred to as the 'hygiene hypothesis', which proposes that the relatively sterile environment in industrialized western countries has contributed to the recent epidemic of asthma and atopy, and that the overall load of infectious agents encountered early in life is an important factor that influences the maturation of the immune system from a TH2 bias at birth towards a predominantly TH1-type response37,38,39. This is supported by recent reports that show a clear relationship between atopy and bacterial infection or antibiotic use40,41,42. It has been shown that certain mycobacterial strains can reduce the formation of specific IgE, eosinophilia and allergen-induced bronchial hyperresponsiveness, and increase IFN-γ production43, which is a powerful suppressor of TH2 activity44. So, exposure to TH1-inducing mycobacterial infections was proposed to cause a shift from TH2 to TH1 immune responses, which would subsequently prevent the development of allergy (Fig. 2). It is reasonable to believe that these bacterial infections are linked closely to the activation of several TLRs, as this family of receptors recognize conserved molecular patterns that are shared by large groups of microorganisms and might therefore have an essential role in the activation of the adaptive immune system. Moreover, the potential role of TLRs in the hygiene hypothesis is supported further by recent findings in Myd88-deficient mice, which showed that a lack of TLR signalling leads to a TH2 response with IgE production by default and no induction of TH1 responses45. However, the existence of PAMPs that are recognized by PRRs other than TLRs that actively stimulate TH2-type responses cannot be ruled out.

In addition to the potentially important role of TLRs in the induction phase of an allergic phenotype, a TLR-driven innate immune response might also have an essential role in disease exacerbations, which are episodes of worsening shortness of breath, coughing, wheezing or chest tightness associated with acute, mild-to-life-threatening airflow limitation. In addition to pollution, the main cause of exacerbations in asthma are viral or bacterial infections46 (Fig. 2). Virus infections are detected in up to 85% of such episodes. Rhinovirus is common in all age groups, whereas respiratory syncytial virus (RSV) is most important in infants and young children47. The innate immune response to the fusion protein of RSV is mediated by TLR-4 and CD14, and it has been shown that RSV persists longer in the lungs of infected Tlr-4-deficient mice than of normal mice48. Bacterial infections are also inducers of asthma exacerbation, and endotoxin is well known for its ability to exacerbate existing allergy and asthma symptoms49. The mechanisms of how viral or bacterial infections induce disease exacerbations are not well understood, but probably include the biased production and effector function of specific cytokines. For example, it has been shown that the exposure of peripheral-blood mononuclear cells from asthmatic individuals to rhinovirus induced a different cytokine pattern, with increased levels of IL-4 and lower levels of IL-12 and IFN-γ, compared with the exposure of cells from normal individuals50. Similarly, it was shown that Escherichia coli LPS — through the activation of TLR-4 — induced abundant IFN-γ, but little or no IL-13, IL-5 or IL-10 in vivo. By contrast, Porphyromonas gingivalis LPS — through the activation of TLR-2 — strongly induced IL-13, IL-5 and IL-10, but resulted in lower levels of IFN-γ, indicating that the production of a cytokine pattern that is compatible with the allergic phenotype might further amplify the underlying inflammatory response and thereby contribute to disease exacerbation51.

TLR-based treatment strategies in asthma

The above described observations associated with the hygiene hypothesis indicate that prospective human vaccines will probably aim to induce strong TH1 responses, leading to the induction of expression of cytokines such as IFN-γ, IL-12 and IL-18, and the induction of allergen-specific tolerance. The most promising approaches to achieve this objective include the induction of systemic or local immune responses through the use of bacterial vaccines, which probably activate several TLRs, or the use of selective TLR-specific ligands, such as CpG oligonucleotides. A different strategy needs to be applied when dealing with exacerbations in asthma, for which the selective induction of a TLR-mediated antiviral response or the blocking of TLR docking sites for viruses or bacteria might be more beneficial.

Bacterial vaccines. Mycobacteria are known to be highly immunostimulatory and, through recognition, uptake and presentation by macrophages and antigen-presenting cells, favour the (at least partly) TLR-dependent stimulation of TH1-type reponses by the production of IL-12 and IFN-γ52,53. The Mycobacterium bovis bacillus Calmette–Guerin (BCG) vaccination has been associated with a reduction in atopic diseases in Japan54. Moreover, BCG inoculation in mice administered 14 days before allergen sensitization reduced the formation of specific IgE in response to allergen, as well as the eosinophilic response and BHR responses, and increased the production of IFN-γ55,56. The application of heat-killed BCG to the lungs of mice inhibited the development of allergen-induced airway eosinophilia for up to two months after treatment57. However, other clinical studies have found no protective effect of BCG vaccination on the development of atopy. The reason for this discrepancy is not clear.

Similar results have been obtained in mice after a single injection of heat-killed Mycobacterium vaccae, another potent inducer of TH1 responses, and Listeria monocytogenes58. Indeed, there is some evidence that the fast-growing M. vaccae, which induces a vigorous cell-mediated immune response and shares several immunodominant epitopes with other mycobacteria, might be a potential candidate for an improved anti-asthma vaccine. Preliminary experiments with heat-killed M. vaccae in adults with asthma and rhinitis or in children with atopic dermatitis showed clinical benefits, as measured by a reduction in the use of rescue medication and severity of disease or inhibition of the allergen-induced, late-phase response59,60. By contrast, however, another study that used lower doses of M. vaccae in asthmatics failed to show any statistically significant beneficial clinical effects61. All patients in this study were prescribed inhaled GLUCOCORTICOSTEROIDS, and it is reasonable to speculate that the effector functions induced by M. vaccae treatment are suppressed by this treatment through antagonizing NF-κB function.

Targeting TLR-4: agonists or antagonists? Human studies have variously reported an association of LPS exposure with an increased risk of asthma-like respiratory symptoms or with a decreased risk of sensitization to aeroallergens62. In this regard, a promoter polymorphism in the human gene that encodes CD14 was found to correlate inversely with IgE levels, as individuals with low levels of IgE had high levels of CD14, and vice versa63. These data indicate that genetic variation in CD14 might modulate the effect that exposure to endotoxins has on the development of TH2 responses. Indeed, other studies have indicated that the CD14 polymorphism might result in the expression of a more severe allergic phenotype64,65. Taken together, these data support the use of TLR-4 agonists, rather than antagonists, for the treatment of asthma. Recently, several compounds, known as aminoacyl glucosaminide phosphates (AGPs), have been developed for monotherapeutic use in manipulating innate-immune mechanisms. These compounds can bind TLR-4 with agonist or antagonist characteristics, depending on the acyl-chain composition of the individual family member66. Several of these AGP TLR-4 agonists have shown pro-inflammatory activity profiles, including the induction of IFN-γ, TNF and IL-12, and activation of cytotoxic T lymphocytes. The ability of these compounds to activate TH1 responsiveness might, therefore, be useful for downregulating or dampening the characteristic TH2 response in atopic diseases such as asthma. By contrast, it has also been reported that allergic airway inflammation and production of IgE were enhanced when LPS was given after sensitization67. Moreover, some of the known disease-exacerbation-inducing stimuli, such as endotoxin or RSV, bind to TLR-4 (Ref. 48). These observations indicate that the development of TLR-4 antagonists might be more beneficial, in particular to prevent or treat episodes of asthma exacerbations.

Targeting TLR-7 and TLR-8. The natural ligands for TLR-7 and TLR-8 are not known. However, synthetic compounds with antiviral activities have now been described as ligands for both receptors68,69. These antiviral imidazoquinoline compounds, imiquimod (R-837) and resiquimod (R-848), are low-molecular-mass immune-response modifiers that can induce the synthesis of IFN-γ and other cytokines in various cell types (Fig. 3). Imiquimod has been used successfully for the local treatment of genital warts caused by human papilloma virus in the clinic. Resiquimod also shows promise for the treatment of genital herpes70. Recently, it has been shown that these compounds probably exert their antiviral properties by activating immune cells through the TLR-7 and TLR-8 signalling pathways. Macrophages from Myd88 or Tlr-7-deficient mice no longer respond to these imidazoquinolines, and both compounds could activate signalling pathways in Tlr-7- or Tlr-8-transfected cell lines68,69. Considering the role of viral infections in disease exacerbations of asthma, TLR-7 or TLR-8 agonists, such as imiquimod or resiquimod, could therefore be extremely valuable for the treatment of these conditions. In addition, the induction of a predominantly TH1-type cytokine profile by these compounds might have further benefits by redirecting the dominant TH2-type response in asthma to a more protective TH1 response.

Figure 3
figure 3

Structures of imiquimod (R-837) and resiquimod (R-848).

Full size image

Targeting TLR-9 by CpG oligonucleotides. Bacterial DNA, but not vertebrate DNA, has a direct immunostimulatory effect on immune cells in vitro71. The immunostimulatory effect is due to the presence of unmethylated CpG dinucleotides, which are under-represented and almost always methylated in vertebrate DNA72. Evidence from the literature shows that bacterial DNA containing unmethylated CpG motifs binds to TLR-9 and has a direct immunomodulatory effect on immune cells. It is well known that CpG DNA can induce the proliferation of almost all B cells and further protects B cells from apoptosis. It also triggers polyclonal immunoglobulin, IL-6 and IL-12 secretion from B cells. In addition to its effects on B cells, CpG DNA also directly activates dendritic cells to secrete IFN-α, IL-6, IL-12, granulocyte–macrophage colony-stimulating factor (GM-CSF), chemokines and TNF. These cytokines stimulate natural killer (NK) cells to secrete IFN-γ73. Overall, CpG DNA induces a TH1-like pattern of cytokine production that is dominated by IL-12 and IFN-γ, with little secretion of TH2 cytokines74. Indeed, treatment of immunized mice with CpG DNA, before or after the airway challenge, redirects the immune response from a TH2-like response towards a TH1-like response, leading to a reversal of established airway eosinophilia and bronchial hyperreactivity. More recently, Tighe et al.75 reported that the chemical conjugation of a CpG oligodeoxynucleotide to the short ragweed allergen Amb resulted in an enhanced immunotherapeutic effect compared with the CpG oligodeoxynucleotide alone. It was also reported that the allergen–CpG conjugate caused less histamine release from basophils than the allergen itself. So, these data indicate that CpG DNA, through activation of TLR-9, might be an effective new method of inducing prophylactic and therapeutic protection against atopic disorders. Furthermore, the combination of enhanced immunogenicity and reduced allergenicity observed for the allergen–CpG conjugate might offer a more effective and safer approach for allergen immunotherapy compared with conventional methods. Although most of the work on CpG DNA has been carried out in mice, it is clear that human cells respond in a similar manner. CpG oligonucleotides are now in clinical trials for various indications, including asthma and allergy, and it will be interesting to see whether the results that have been obtained in animal models can be confirmed in patients with atopic diseases. Moreover, patent applications indicate that several companies are developing TLR-9 agonists76.

In addition to asthma and allergic diseases, therapeutic applications of CpG DNA have been shown for the activation of innate immune defences against infections, as vaccine adjuvants and finally in DNA vaccines. Moreover, the efficacy of CpG DNA in preventing or treating tumour development or metastasis, as well as for tumour immunotherapy, has been shown in several experimental models, extending the potential for TLR-9 agonists far beyond their use in asthma77.

Issues associated with TH1-inducing strategies. Recent advances in our understanding of the hygiene hypothesis indicate that strategies to induce a strong TH1 response might be dangerous. Treatment that pushes responses non-specifically towards a TH1-type response are just as likely to initiate TH1-mediated pathology as to cure the allergy78. Moreover, although the reconstitution of the TH1/TH2 balance by bacterial vaccines is an attractive theory, it is unlikely to explain the whole story, as autoimmune diseases that are characterized by TH1 responses, such as type 1 diabetes, can also benefit from treatment with mycobacteria, and the prevalence of these diseases has increased in parallel with the prevalence of allergies79. This raised the question of whether mycobacteria modulate inflammatory responses through effects on REGULATORY T CELLS, which have been shown to have anti-inflammatory properties80,81. We have shown that treatment of mice with a heat-killed M. vaccae suspension can successfully induce the generation of allergen-specific regulatory T cells that can suppress the allergen-mediated inflammatory response82 (Fig. 2). It is, therefore, no longer reasonable to propose that a lack of exposure to TH1-inducing infectious agents leads to a bias towards TH2 responses. It seems more probable that a general deficiency of regulatory-cell activity is responsible for the increased prevalence of several groups of disease that involve immunodysregulation. However, it still remains to be shown whether activation of TLRs is one of the key elements in the induction of such allergen-specific regulatory T cells.

TLR agonists and antagonists in other diseases

As described for asthma, TLRs provide a mechanism by which exogenously generated signals can markedly affect the initiation, maintenance and progression of inflammatory diseases. This is a logical consequence of Janeway's 'extended self, non-self' theory, which predicts the interaction of the innate and acquired immune responses through the sensing of pathogen-derived motifs83,84. Similar arguments as for asthma can be made for another chronic respiratory disease, chronic obstructive pulmonary disease (COPD), for which LPS and bacterial infections are well known to induce disease exacerbations85, again indicating that TLR-4 or TLR-2 antagonists might have therapeutic benefits. Targeting of TLR-2 and TLR-4 can also be applied to other diseases, as an increased expression of TLR-2 and TLR-4 was observed in intestinal inflammation, indicating that the presence of TLRs might contribute to the inflammatory process observed in gastrointestinal diseases86. Finally, in acne lesions, TLR-2 is expressed on the cell surface of macrophages that surround pilosebaceous follicles, and these macrophages produce inflammatory cytokines after TLR-2 activation by Propionibacterium acnes. As such, TLR-2 could be a new target for the treatment of this common skin disease87.

A more subtle role for PRRs in chronic inflammatory disease is indicated by the role of NOD2 in Crohn's disease, in which polymorphisms in this intracellular LPS receptor lead to increased susceptibility to disease88,89. It has been proposed that NOD2, expressed in intestinal monocytes, acts as a negative regulator of inflammation, sensing the bacterial population of the gastrointestinal tract and possibly producing regulatory cytokines, such as IL-10 (Ref. 90), to dampen inflammatory responses and ensure that the normal commensal state in the gastrointestinal tract is maintained. The polymorphisms are believed to affect the expected signal from NOD2, either by reducing the association with LPS or decreasing the direct interaction of NOD2 with NF-κB88,89. This aberrant signalling stimulates an inflammatory response that leads to chronic inflammation in at least some patients with this disorder. So, PRR signalling can act as a factor to drive responses that occur in many diseases.

The extension of Janeway's original theory to the 'danger' theory of Matzinger91 gives the opportunity to broaden the origin of the stimulating factors. It is now proposed that, in addition to pathogen-derived signals, some of the endogenously released danger signals could interact directly with the PRRs that drive immune responses92. If these molecular danger signals are liberated by internal cellular activity at sites of tissue damage that are associated with inflammatory responses, then it is possible that the PRR-driven aspects of the innate immune response could be activated without pathogenic intervention. This implies that — in chronic diseases — there might be alternative stimuli maintaining or driving the responses that are seen. This hypothesis is dependent on crossreactivity of endogenously generated ligands for the various PRRs that are discussed in the introduction. Only a few endogenous ligands for PRRs have been identified so far. The mannose receptor and the scavenger receptor are good examples of PRRs that recognize exogenous and endogenous ligands. Some of the literature93,94,95,96,97,98,99 indicates that there are examples of endogenous stimulators of the TLRs in mammalian systems, which include HEAT-SHOCK PROTEINS (HSPs) 60 and 70, saturated and unsaturated fatty acids, hyaluronic-acid (HA) fragments, dsDNA, and surfactant protein-A (Table 1).

Role of endogenous TLR ligands. HSPs have a well-described role in stimulating various aspects of the acquired immune response100. Furthermore, theories have been advanced that propose that antibody crossreactivity between microbial and endogenous HSPs might have an important role in many inflammatory or autoimmune diseases101,102. Recent data show that HSPs can stimulate TLRs directly93,94,95, indicating alternative potential roles for these molecules in inflammatory events, with the potential for positive-feedback loops. An example of this inflammatory activation has been described after open-heart surgery103. Patients who undergo coronary-artery bypass grafting (CABG) by cardiopulmonary bypass are known to suffer an inflammatory response104,105. This is typified by the increased expression of HSP70 by myocardial cells106. Through protein-expression profiling, Dybdahl et al.103 showed elevated levels of HSP70 in the circulation of CABG patients. They also showed that expression of TLR-2 and TLR-4 on monocytes increased on days one and two after the operation. In a corresponding set of in vitro experiments, the ability of recombinant HSP70 to stimulate monocytes in a TLR-4/CD14-dependent manner was shown103. These data typify the extent of proof of concept of the endogenous activation of TLRs in inflammatory conditions. However, there are other concerns related to these types of study. In a recent review of HSP activation of the innate immune response, Wallin107 discussed several confounding factors that could influence the conclusions regarding the involvement of TLR-2 and TLR-4 in the process. They specifically discuss the potential for LPS contamination of recombinant protein preparations, the potential for certain HSPs to act as LPS-binding proteins, and that polymyxin B does not neutralize all species of LPS. These are all issues that need to be considered when interpreting the results.

The ability of fatty acids to induce the expression of COX-2 through TLR-4-mediated activation of the NF-κB pathway has been comprehensively reviewed by Hwang108. One of the most interesting points about fatty-acid activation of TLRs is that saturated and unsaturated fatty acids have different properties. Lauric acid, a saturated fatty acid that is structurally similar to the lipid A component that is found in LPS of the Gram-negative bacterium E. coli105, is a strong stimulator of TLR-4 (Ref. 96). By contrast, long-chain, unsaturated fatty acids, such as docosahexaenoic acid, not only fail to stimulate COX-2 induction, but also inhibit LPS-induced COX-2 induction, indicating that such fatty acids might be antagonists of TLR-4-mediated lipid signalling108. These functional differences between the fatty-acid ligands need to be understood with respect to the structure–activity relationships (SARs) that are being developed for LPS binding to TLR-2 and TLR-4. It has been proposed that the conformation of the lipid A component explains which TLR is favoured or even whether the LPS molecule is antagonistic109,110. The above observations set out the potential for endogenously derived fatty acids to modulate inflammatory responses. Perhaps of more significance to disease management is the opportunity to modify inflammatory responses through the provision of exogenous fatty acids or their derivatives in dietary or therapeutic regimens, providing the receptor–ligand SAR can be understood and exploited.

HA is a principal component of the extracellular matrix, which can be degraded at sites of inflammation111,112. Furthermore, the low-molecular-mass products of this degradation are pro-inflammatory and activate macrophages113,114. Termeer et al.97 have extended their previous observations on the maturation of human dendritic cells by HA fragments115 to show that HA fragments act through TLR-4 both in vitro and in vivo.

The synthesis of autoreactive antibodies specific for DNA or the nucleosome is a hallmark of some autoimmune diseases116. It has been shown recently in a mouse model that B cells can be directly stimulated to proliferate in response to immune complexes isolated from the sera of autoimmune mice117. This observation has been refined to show that Myd88−/− animals are unresponsive, indicating that autoimmune serum stimulates a TLR, which was identified as TLR-9 (Ref. 98), the hypomethylated DNA receptor77. Interestingly, this observation, although in an animal model, points to several aspects of autoimmune disease: first, the dual stimulation of the TLR and the B-cell receptor provides a mechanism for direct antigen stimulation of autoreactive B cells118; second, autoantibodies specific for DNA or the nucleosome are a feature of systemic lupus erythematosus (SLE)116; and third, the inhibition of the process by chloroquine (an endosome inhibitor that is assumed to disrupt the presentation of DNA to TLR-9) is mirrored by its use as a therapeutic for the treatment of SLE119.

So, there are data to indicate that HSPs, fatty acids, HA fragments and endogenous DNA can act as endogenous ligands for TLRs, and thereby modulate the inflammatory response by inducing inflammatory mediators108, sensitizing or activating cells such as monocytes103 or dendritic cells97, or stimulating mature immune cells, such as antibody-producing B cells98. The importance of these pathways in specific diseases is unknown, and, therefore, the potential for targeting them in therapeutic interventions has not yet been validated. However, it is reasonable to speculate that more evidence for a role of endogenous TLR ligands in the pathogenesis of autoimmune diseases, transplant rejection and many chronic inflammatory diseases will accumulate in the near future, showing the real potential of TLRs as therapeutic targets in these (chronic) inflammatory diseases.

Conclusion

The discovery of TLRs has provided us with an important insight into the mechanisms whereby the innate immune system senses and responds to pathogens and how an adaptive immune response is controlled. These new insights not only provide a basis for the development of new therapies for diseases such as sepsis, but also offer the potential to develop disease-modifying therapies that result in immune deviation from a TH2-to a TH1-dominated immune response or the induction of regulatory T cells (Table 2). Moreover, reagents that enhance TLR signalling pathways can be powerful adjuvants for fighting pathogens or cancer. More recent data extend the role of TLRs by showing that, in addition to pathogen-derived signals, endogenous ligands, such as surfactant protein-A, HSPs, HA fragments and dsDNA, can activate TLRs. These data indicate that these receptors might also have an important role in the maintenance of the inflammatory process that leads to chronic inflammation. Together with the emerging data on the role of TLRs in chronic inflammation, there are equally fascinating insights to show that molecules that are tractable to the pharmaceutical industry, such as oligonucleotides, fatty acids and imidazoquinolines, can elicit responses at various TLRs. Indeed, the demonstration that low-molecular-mass compounds, such as resiquimod, bind to TLR-7 and TLR-8 and have antiviral and anti-allergic properties, supports the concept that targeting TLRs could offer a promising new therapeutic strategy for the treatment of human inflammatory diseases that are caused by dysregulation of the immune system. However, the potential for adverse effects when targeting such fundamental pathways of the host defence mechanism needs to be carefully assessed, especially for diseases that are not usually life-threatening. Nevertheless, the data that are available at present clearly indicate that, in the near future, it is possible that TLRs will offer validated druggable targets for anti-inflammatory therapeutics.

Table 2 Potential use of TLR agonists and antagonists in various diseases
Full size table