2.1 Lessons from pathogens
Pathogens display a wide array of carbohydrates, (glyco)proteins and lipids that interact with the human immune system. Interestingly, our recent studies show that in schistosomes, lipids in particular are able to induce down-modulation of immune responses via the induction of Treg cells 19. Traditionally, many immunologists regarded lipids as inactive molecules embedded in membranes that at the most could carry bioactive headgroups. However, it is becoming more and more evident that lipids themselves are important molecules involved in immune responses. Although immunological activity was first attributed to the polar headgroup of lipids, it is now becoming increasingly clear that the apolar lipid moieties of glycolipids, phospholipids, lipoproteins and lipopeptides have a profound influence on the biological activity of these molecules.
There are already several excellent reviews on the importance of lipids in the control of granulocytes during inflammation 20, 21. In this review both endogenous and pathogen-derived lipids with immunomodulatory properties will be discussed, highlighting their anti-inflammatory activities on the antigen-presenting compartment and T cell responses (Fig. 1).
Figure 1. Anti-inflammatory lipids: an overview. Pathogens can induce anti-inflammatory responses either directly by lipids that are present in the pathogen, or by inducing the production of endogenous lipid mediators in the host. Some pathogens can induce apoptosis of host cells; apoptotic cells, which expose phosphatidylserine in the outer leaflet of the cell membrane, in turn can switch immune responses into an anti-inflammatory mode.
Download figure to PowerPoint
2.2 Induction of endogenous immunomodulatory lipids by pathogens
The best-characterized immunomodulatory lipids are naturally occurring endogenous eicosanoids derived from the 20C poly-unsaturated fatty acid arachidonic acid, which is released from membrane phospholipids and converted by the action of cyclooxygenases and lipoxygenases. It is now generally accepted that eicosanoids such as leukotrienes and some prostaglandins have pro-inflammatory activity whereas other prostaglandins and lipoxins exert anti-inflammatory signals. Here we will focus on certain prostaglandins and lipoxins (Fig. 2) that have been reported to exert anti-inflammatory effects.
PGE2, which can act through the G-protein-coupled receptors EP1, EP2, EP3 and EP4, is a potent inhibitor of cellular proliferation and regulates cytokine synthesis: it can inhibit IL-2 synthesis 22 and IL-2 receptor expression 23 on T cells. On antigen-presenting cells such as macrophages and DC it inhibits production of IL-1 24, TNF-α 25 and IL-12 26, whereas IL-10 production is increased 27. In contrast, PGD2, another prostaglandin for which anti-inflammatory effects have been reported, inhibits both IL-12 and IL-10 production by human DC 28. PGD2 can act through the DP receptor that is coupled to a Gαs-type G protein, and can activate the CRTH2 receptor (chemoattractant receptor-homologous molecule), a Gαi-type G protein that is selectively expressed on Th2 cells, basophils and eosinophils 29.
Prostaglandins were shown to be involved in the interaction of the parasite Trypanosoma cruzi with its host. High production of PGE2 was found in spleen cells from T. cruzi-infected mice, and blocking prostaglandin production by inhibiting cyclooxygenase activity markedly increased lymphocyte proliferation and enhanced host mortality upon experimental infection with this protozoan parasite, although parasite levels were not affected 30. Thus, production of prostaglandins is an important part of the host immune response, not to combat the infection per se, but rather to limit immune responses elicited by the parasite that would otherwise be destructive to the host. Parasites not only induce prostaglandin production by the host to suppress the immune response, but also are able to produce prostaglandins themselves. Schistosomes were shown to secrete PGD2, thereby inhibiting the migration of Langerhans cells upon exposure to schistosome cercariae 31, and Onchocerca volvulus was shown to express glutathione S-transferase on its surface that produces PGD2 directly at the host-parasite interface 32. The question regarding helminths is whether PGD2 production is involved in controlling anti-parasitic responses, which would justify the investment made by helminths to elaborate such anti-inflammatory molecules.
Another set of arachidonic-acid-derived molecules, termed lipoxins (lipoxygenase interaction products) are endogenous trihydroxytetraenes that are generated during inflammatory responses and are involved in the resolution of inflammation 33. LipoxinA4 (LXA4; Fig. 2) is involved in limiting parasite-induced inflammation and mortality during chronic Toxoplasma gondii infection. T. gondii was shown to inhibit IL-12 production in DC by inducing LXA4 generation 34, and mice deficient in 5-lipoxygenase, an enzyme critical for LXA4 synthesis, are more susceptible to mortality from T. gondii infection 35. This increased mortality was not the result of a higher parasite burden, but rather resulted from pro-inflammatory cytokine-induced tissue pathology. This indicates that LXA4 is not involved in combating the parasite, but has an important role in keeping the cell-mediated immune effector responses under control to avoid excessive damage to host tissues 33. The question of whether lipoxin induction is specific to Toxoplasma or represents a general anti-inflammatory feedback loop response to more pathogens needs to be investigated.
Related to this one should consider that lipoxin biosynthesis has been found in many human inflammatory diseases, including asthma. Interestingly, lipoxin levels were found to be higher in subjects with mild asthma compared with individuals with severe asthma 36. Indeed synthetic lipoxin analogues have been shown to block the development of allergen-induced airway hyperresponsiveness and prevent expression of IL-5, IL-13, eotaxin and prostanoids 37.
LXA4 can act on the G-protein-coupled receptor designated ALXR, also referred to as formyl peptide receptor-like 1 (FPRL1) 38, to inhibit NF-κB activation and inflammatory responses 39. In addition, LXA4 can also activate the aryl hydrocarbon receptor (AhR) 40, which is a ligand-activated transcription factor that is present in the cytosol and upon ligand binding translocates to the nucleus, where it can activate responsive genes, most of which are involved primarily in xenobiotic metabolism. Persistent activation of the AhR has been associated with immunosuppression 40, but the mechanism remains to be elucidated.
Figure 2. Eicosanoids that act on antigen-presenting cells to inhibit inflammatory pathways. PGE2, PGD2, LXA4 and 15dPGJ2 act on receptors that have been shown (solid lines) or are thought (dotted lines) to inhibit NF-κB activation and IL-12 production. As well as acting through peroxisome proliferator-activated receptor γ (PPARγ), 15dPGJ2 has been shown to inhibit NF-κB activation and its binding to DNA independently of a receptor 69.
Download figure to PowerPoint
2.3 Immunomodulatory lipids expressed by pathogens
Pathogens display a wide array of lipids and acylated proteins or peptides, and data on their importance in immunomodulation are only beginning to emerge.
Bacterial interaction with the immune system has been studied extensively at the level of the innate immune system. It has become clear that the presence of lipid moieties in bacterial products is essential for recognition by some pattern-recognition receptors, particularly by Toll-like receptor 2 (TLR2), which can be activated by a wide variety of acylated bacterial ligands (Fig. 3). For example, mutants of cloned Escherichia coli lipoproteins that lack acyl lipid moieties were shown to be unable to activate TLR2 41. Increasing details are becoming available on specific structural motifs that are required for the interaction of lipids with the immune system.
Many TLR2-activating bacterial lipoproteins contain the characteristic N-terminal lipopeptide Pam3CysSerLys4 (PAM3CSK) (Fig. 3), which consists of three acyl chains, all essential for TLR activation 42. The mycoplasmal lipopeptide MALP-2 (Fig. 3) differs from conventional bacterial lipoproteins in that it has a free N terminus. Thus, this structure contains only two ester-bound fatty acids, but nevertheless can potently activate TLR2 43. For MALP-2 it has been shown that not only the presence of two fatty acids is essential for TLR2 activation, but also the configuration of the lipid moieties appears to be important: the R-stereoisomer of MALP-2 is a 100-times more potent TLR activator than the S-stereoisomer 44. The situation is less stringent for TLR2 activation by lipoteichoic acid (LTA), a cell-wall component of Gram-positive bacteria. The lipid anchor of LTA is essential for biological activity, but the structural requirements appear to be less strict than for the bacterial lipopeptides discussed above: although the presence of two acyl chains within the lipid anchor is required for optimal TLR activation, anchors with only one acyl chain are still able to induce production of inflammatory cytokines 45.
The variable number of acyl chains necessary for TLR2 activation may be linked to the finding that TLR2 is active as a heterodimer 46. It was shown that TLR6 can function in association with TLR2 to confer responsiveness to MALP-2 47. In contrast, TLR2 activation by PAM3CSK was shown to require TLR1 but not TLR6 48. Thus, TLR6 seems to be essential for recognition of diacylated lipopeptides, but dispensable for TLR2 activation by triacylated lipopeptides. Indeed, synthetic MALP-2 containing an additional N-linked palmitoyl chain can activate TLR2 in the absence of TLR6 49.
For many of these ligand–receptor interactions that involve well-defined lipids, there is no information on their effect on the adaptive immune responses in terms of polarization towards Th1, Th2 or Treg responses. A few studies involving complex antigen mixtures have shown that in mice deficient for MyD88 — an adapter protein that is essential for most signaling of TLR molecules — Th1 responses are impaired, indicating that interaction of bacterial lipids with TLR molecules induces Th1 responses 50, 51. However, a few reports have also indicated that several bacterial acyl-containing ligands for TLR2 can inhibit activation of cells of the innate immune system. The 19 kDa lipoprotein of Mycobacterium tuberculosis and Treponema pallidum lipopeptides inhibit MHCII expression and antigen-processing by macrophages 52. Such down-regulation of events that are critical in activation of adaptive immune responses is expected to lead to immune hyporesponsiveness.
In addition to bacteria, intracellular protozoa harbor lipids that interact with the immune system. For example, GPI anchors of T. cruzi are able to activate TLR2. These GPI anchors can suppress immune responses by inhibiting maturation of human DC, as well as production of TNF-α, IL-10 and IL-12p40 in both DC and macrophages 53. Activation of the innate immune system by T. cruzi GPI anchors is mainly mediated by the lipid moiety, which can be a glycerolipid or a ceramide. Particularly the presence of an unsaturated fatty acid at the sn-2 position potentiates TLR-2 activation 54 (Fig. 3). In the protozoan parasite Leishmania major, the major parasite surface molecule lipophosphoglycan also has immunomodulatory properties, as it inhibits the migration of Langerhans cells out of the skin to the lymph node 55. This would be expected to curtail the immune response to the invading Leishmania parasite.
Recent studies involving helminth parasites have shown that these multicellular eukaryotes possess lipids with immunomodulatory properties. When lipids from schistosomes were fractionated into different classes and their effects on DC were analyzed, it became clear that the fraction containing phosphatidylserines could interact with DC and affect their T cell polarizing capacity. Further purification of the active components revealed that di-acylated phosphatidylserine promotes maturation of DC into a phenotype (termed DC2) that induces development of Th2 responses, whereas mono-acylated lysophosphatidylserine (lyso-PS) acts on DC to promote development of Treg cells by inducing a so-called DCreg phenotype (Fig. 4). Interestingly, although lyso-PS contains only one acyl chain, the effect of this schistosomal lipid was mediated by TLR2 19. Since not all TLR2 ligands have the capacity to induce development of DCreg cells (van der Kleij, unpublished observations), additional receptors and/or accessory molecules are likely to cooperate with TLR2 to define the outcome of pathogen recognition. For details on interactions between signaling of TLR molecules and other receptors, the reader is referred to a recent mini-review in this journal 56.
TLR activation by lyso-PS appears to be a unique property of lyso-PS derived from schistosomes, as mammalian lyso-PS was found to be inactive 19. Phosphatidylserines from schistosomes have acyl chains that are not commonly found in mammals; they contain long, unsaturated fatty acids that may well be important for biological activity. Schistosomes do not synthesize fatty acids de novo but rely on the host for fatty acid supply. Most lipids found in schistosomes have indeed been synthesized by the host on which they feed. However, the parasite has retained the capability to modify host-derived fatty acids by chain elongation, resulting in a fatty acid profile that is clearly distinctive from that of the host 57. For example, 20:1 and 22:4 are fatty acids that are prominently present in schistosomes but have not been found in mammals 58, 59.
In addition to chain elongation, schistosomes are capable of modifying host fatty acids in another way, resulting in the presence of octadec-5-enoic acid (18:1Δ5). This unusual fatty acid was found to be exclusively present on the tegument of the schistosome and is absent from the host 60. Schistosomes display a high rate of deacylation and reacylation, in contrast to host cells 61, thereby exposing the host cells to lysophospholipids (which are formed upon deacylation of phospholipids) and free fatty acids. In high concentrations, lysophospholipids and fatty acids have a general destabilizing effect on the host cell membranes and can induce cell lysis 62, 63, but are also active in low concentrations and modulate cells of the innate immune system 19.
Recent work on Ascaris lumbricoides has indicated that this intestinal helminth also contains lipids that stimulate TLR2 and induce the development of Th2 and Treg cells (van Riet, unpublished), but the structures involved have yet to be characterized.
In helminth lipids, immunomodulatory properties are not restricted to phosphatidylserines, but can also be found in glycolipid preparations. Glycolipids isolated from Ascaris suum stimulate the production of inflammatory cytokines in peripheral blood mononuclear cells. Within these structures, biological activity could be attributed to phosphocholine moieties that are covalently attached to the carbohydrate headgroup of the glycolipids 64. In contrast to glycolipids from nematodes (such as Ascaris), glycolipids from schistosomes (which are trematodes) do not contain such phosphocholine-containing headgroups. Nevertheless, schistosomal glycolipids were found to induce production of pro- and anti-inflammatory cytokines in monocytes 65, and polarize DC maturation into a phenotype that promotes development of Treg cells and Th2 cells (van der Kleij et al., unpublished observations), indicating that for glycolipids of trematodes the presence of certain carbohydrates and/or lipid moieties is sufficient for biological activity, independently of a phosphocholine headgroup.
The fatty acid composition of S. mansoni glycosphingolipids is completely different from that of other lipids in this parasite. Whereas phospholipids, tri-acylglycerols and free fatty acids are all dominated by 16:0, 18:0, 18:1, 18:2, 20:1, 20:4 and 22:4 acyl chains 66, many glycosphingolipids were shown to contain acyl chains such as 24:0, 24h:0, 25:0, 26:0, 26:1, 26h:0, 27:0 and 28h:0 67. Although the biological significance of these moieties has yet to be determined, it is possible that such unusually long fatty acid chains are particularly suitable for recognition by the mammalian innate immune system.
Figure 3. TLR2-activating pathogen-derived lipids for which the importance of a lipid moiety has been demonstrated. PAM3CSK is the bacterial lipopeptide PAM3CysSerLys4. MALP-2 is a mycoplasmal lipopeptide. The lipoteichoic acid (LTA) anchor is present in Gram-positive bacteria. Schistosomal lyso-PS can have a C20:1 acyl chain. Finally, the tGPI anchor is present in T. cruzi trypomastigotes.
Download figure to PowerPoint
Figure 4. Immune polarization by schistosomal phosphatidylserines. Schistosomal lyso-PS activates TLR2, which probably acts in concert with another, as-yet unidentified, receptor to polarize maturation of human immature DC (iDC) into a mature DC phenotype (mDCreg) that instructs naive T cells to polarize into Treg cells. These Treg cells are capable of suppressing proliferation of bystander T cells by producing IL-10. Schistosomal phosphatidylserine acts on an unknown receptor on iDC to induce maturation into a phenotype (mDC2) that polarizes the development of naive T cells into Th2 cells (which are characterized by the production of IL-4). Both lyso-PS and phosphatidylserine contain acyl moieties that are not commonly found in mammals.
Download figure to PowerPoint
2.4 Other relevant phosphocholine immunomodulatory lipids
In addition to lipids originating from pathogens or induced by pathogens, many other lipid molecules with immunosuppressive activity exist. Some of these lipids are structurally related to ones that have already been discussed and may turn out to play a role in host-pathogen cross-talk and therefore will be discussed below.
An important class of immunomodulatory lipids belonging to the eicosanoids are the cyclopentenone prostaglandins (cyPG), which are characterized by the presence of a highly reactive electrophilic carbon atom in the unsaturated carbonyl group of the cyclopentane ring, and have clear anti-inflammatory effects. A potent cyPG is the PGD2 metabolite 15deoxyΔ12,14PGJ2 (15dPGJ2) (Fig. 2), which inhibits macrophage activation and suppresses pro-inflammatory signaling pathways including NF-κB, AP-1 and STAT molecules 68, 69. Furthermore, 15dPGJ2 was shown to promote apoptosis in vitro70 and to inhibit the stimulatory capacity of DC by inhibiting maturation, expression of costimulatory molecules and cytokine production (IL-12 and IL-10) 28, 71. It is thought that 15dPGJ2 is a ligand for peroxisome proliferator-activated receptor γ (PPARγ), a receptor that has been implicated in down-regulation of inflammation 72, but several studies have shown that PGD2 metabolites also have modulatory effects independent of this receptor 69, 71, 73. The therapeutic potential of 15dPGJ2 has been tested in several animal models; it was shown to ameliorate adjuvant-induced arthritis in rats 74 and collagen-induced arthritis in mice 75, and moreover was shown to attenuate clinical signs of disease in experimental autoimmune encephalomyelitis 76.
Although the immunomodulatory effect of lyso-PS has so far only been found in schistosomes, it is possible that various classes of pathogens express such bioactive lysophospholipids. Therefore it is important to consider what is known about bioactive lysophospholipids so far. Whereas mammalian lyso-PS has no activity 19, other endogenous mammalian lysophospholipids do have modulatory functions 77 (Fig. 5). Lysophospholipids can be active at low concentrations through activation of G-protein-coupled receptors. These receptors can be divided into two subgroups, consisting of the EDG1–8 receptors encoded by endothelial differentiation genes, and the closely related receptors GPR4, OGR1, TDAG8 and G2A. Their main ligands are sphingosine-1-phosphate (S1P) (for EDG1, 3, 5, 6, and 8), lysophosphatidic acid (lyso-PA) (for EDG2, 4, and 7), sphingosylphosphorylcholine (SPC) (for OGR1), galactosylsphingosine (for TDAG8) and lysophosphatidylcholine (lyso-PC) (for G2A).
Recent findings demonstrate that lysophospholipids play a role in activation, function and trafficking of leukocytes 78. With respect to immune polarization, endogenous lysophospholipids have been shown to be active; lyso-PA and S1P suppress IL-12 and enhance IL-10 production in human DC, and polarize their maturation so that development of Th2 cells is favored 79, 80. In relation to this finding, it is interesting to note that in asthmatics, levels of lysophospholipids were found to be elevated upon antigen challenge 81. It has been proposed that lyso-PA contributes to the pathology of asthma by inducing aberrant airway repair and remodeling 82. In contrast to lyso-PA, lyso-PC affects DC maturation into a phenotype that stimulates the development of IFN-γ-producing T cells 83. However, lyso-PC appears to have an important immune down-modulatory function as well: deletion of the murine gene encoding its receptor G2A, that is normally constitutively expressed on immune cells, resulted in adult-onset autoimmune disease similar to human systemic lupus erythematosus 84. The discrepancy between these findings may indicate that other receptors are involved in recognition of lyso-PC, or that ligands other than lyso-PC act on the G2A receptor to induce immune suppression.
It is striking that lyso-PA and lyso-PC — molecules that are structurally very similar (Fig. 5) — have such opposite effects on immune polarization. These findings imply that whereas the presence of one acyl chain is required for receptor ligation (di-acylated phospholipids are not active), the headgroup determines which receptor is bound and what type of immune response is induced. In addition, other co-receptors that may act in concert with the receptors described above could be involved in initiation of immune polarization.
A relationship between exposure of cells of the innate immune system to mammalian diacylated phosphatidylserine and the induction of an anti-inflammatory state was found in the process of apoptotic cell clearance. Within normal cells, the presence of phosphatidylserine is restricted to the inner leaflet of the cell membrane. In apoptotic cells, however, phosphatidylserine is exposed in the outer leaflet, where it can be recognized by macrophages via receptors mediating the engulfment of apoptotic cells, such as CD36 and the phosphatidylserine receptor (Fig. 1)85, 86. Ingestion of apoptotic cells by macrophages induces TGF-β1 secretion, resulting in suppression of proinflammatory mediators 87.
In addition, the interaction of DC with apoptotic cells has been shown to inhibit maturation, resulting in DC that express low levels of MHC and secrete high levels of IL-10 88, 89. These cells are thought to induce tolerance by inducing the development of Treg cells 90. This pathway may be exploited by helminths in two ways. First, it has been found that Brugia malayi microfilariae induce apoptosis in host cells 91, 92, and this in turn may induce development of Treg cells and immune hyporesponsiveness as discussed above. Second, given that schistosomal lyso-PS also induces Treg cell development, it is tempting to suggest that schistosomes exploit the mechanism normally used by their host to prevent excessive inflammation and autoimmune reactions in response to apoptosis.
Figure 5. Immune polarization by endogenous lysophospholipids. Lyso-PA and sphingosine-1-phosphate (S1P) polarize the development of the immune response towards Th2, whereas lyso-PC skews towards Th1.
Download figure to PowerPoint