Review |
From the Center for Molecular Medicine and Department of Medicine (G.K.H., Z.-Q.Y.), Karolinska Institute, Stockholm, Sweden; and Leducq Center for Cardiovascular Research (P.L., U.S.), Department of Medicine, Brigham and Womens Hospital, Harvard Medical School, Boston, Mass.
Correspondence to Dr Göran K Hansson, CMM L8:03, Karolinska Hospital, SE-17176 Stockholm, Sweden. E-mail Goran.Hansson@ cmm.ki.se
| Abstract |
|---|
|
|
|---|
Key Words: lymphocytes macrophages antibody inflammation cytokines
| Introduction |
|---|
|
|
|---|
| Innate Immunity: Fast but Blunt |
|---|
|
|
|---|
B pathway, and culminate in the expression of a set of highly active gene products.9 Thus, innate immunity constitutes a rapid first line of defense that can mobilize in minutes to hours.
|
Innate immunity involves several different cell types, most importantly those of the mononuclear phagocyte lineage9,10 (Figure). Macrophages express receptors that recognize a broad range of molecular patterns foreign to the mammalian organism but commonly found on pathogens. These pattern-recognition receptors include various scavenger (ScRs) and Toll-like receptors (TLRs).12 Their ligands contain PAMPs such as lipopolysaccharides (LPS), surface phosphatidylserine, and aldehyde-derivatized proteins. Whereas ligation of scavenger receptors leads to endocytosis and lysosomal degradation of the recognized particles,13,14 engagement of TLR transmits transmembrane signals that activate NF-
B and mitogen-activated protein kinase (MAPK) pathways.15 17 TLR ligation therefore induces expression of a wide variety of genes such as those encoding proteins involved in leukocyte recruitment, production of reactive oxygen species, and phagocytosis.17,18 Activation of TLRs will also elicit the production of cytokines that augment local inflammation. Finally, TLR ligation may directly induce apoptosis, which is probably of key importance in the first line of defense.19 Several PAMPs can ligate TLRs and/or ScRs (Table 1). Some of these ligands are implicated not only for microbial pathogenesis, but in atherogenesis as well.
|
| Adaptive Immunity: High Specificity Through Somatic Rearrangement and Clonal Expansion |
|---|
|
|
|---|
Initial activation of "naïve" T cells requires strong activating stimuli best provided by the dendritic cell (DC), a specialized macrophage cell.20,21 DCs express on their surface major histocompatability complex (MHC) class II molecules at high density as well as a set of costimulatory factors needed to instigate the adaptive immune response, ie, the activation of naïve T cells. Once successful activation has occurred, the remaining memory T cells have a lower activation threshold. Subsequent rounds of stimulation therefore require lower amounts of antigen. Regular macrophages, not just DCs, can accomplish this less stringent function and reactivation can therefore occur in nonlymphoid tissues such as the vessel wall.
Most of our antibodies are encoded by genes that have undergone somatic rearrangement. Activation of a specific B cell by antigen causes hypermutations in its immunoglobulin genes. Together with an evolutionary pressure caused by the antigen, this leads to an affinity maturation that generates antibodies of increasing specificity.
Certain B cells called B1 cells do not undergo affinity maturation. Instead, they produce germline-encoded immunoglobulins that are usually low-avidity antibodies.22 Some of these "natural antibodies" recognize microbial components such as phosphorylcholine of pneumococci. Interestingly, this particular natural antibody also recognizes oxidized phospholipids of low-density lipoprotein (LDL).23
Sophisticated control mechanisms reduce the risk for inappropriate activation of the immune system. However, such activation can still occur, due to dysregulation or molecular mimicry. In the former case, a lower general threshold for activation can lead to systemic autoimmune disease such as systemic lupus erythematosus. In the case of antigenic mimicry, endogenous molecules form that resemble foreign antigens. This situation can lead to organ-specific autoimmunity in the tissues containing such autoantigens. Finally, production of endogenous molecules that can bind to pattern recognition receptors on macrophages may lead to inappropriate activation of innate immunity and pathological inflammation. All these situations cause human diseases, and at least the two latter types seem to contribute to atherogenesis.
| Innate and Adaptive Immune Mechanisms Operate Together and in Sequence During Host Defense |
|---|
|
|
|---|
|
| Vasculature as Part of the Immune System |
|---|
|
|
|---|
Endothelial cells (ECs) express TLRs,24,25 whose ligation induces expression of leukocyte adhesion molecules, inducible NO synthase 2 (NOS2), endothelin, interleukin-1, and other inflammatory molecules. These cells also express the scavenger receptors CD36 and LOX-1, and can internalize ligands such as modified LDL particles.12,26 Strategically located at the interface of blood and tissues, ECs play a pivotal role in the inflammatory response.27 Their activation causes leukocyte recruitment, increased permeability, edema, and other characteristic features of inflammation. Furthermore, ECs can activate adaptive immunity by presenting foreign antigens to specific T cells.28,29 Although they present antigen less efficiently than DCs or macrophages, the unique interfacial location of the endothelium renders ECs particularly important in recall responses to blood-borne antigens.
The blood vessels also occupy a central role in innate immune responses. Consider the extreme example of septic shock as a host response to a microbial invader. LPS from the microbes cell wall combine with soluble or cell-associated CD14 to ligate TLR4 on ECs, perivascular macrophages, and possibly vascular smooth muscle cells (SMCs). TLR signaling induces NF-
B activation and expression of genes such as NOS2. Production of abundant NO ensues, causing smooth muscle relaxation, vasodilation, and hypotension.30 NF-
B activation of tissue factor gene expression produces a hypercoagulable state. The combination of circulatory failure and disseminated intravascular coagulation often causes multiple organ failure and death. Other microbial components such as fimbriae on Escherichia coli can cause similar reactions, also probably through the TLR-NOS2 pathway.31 TNF-
and other cytokines released from activated macrophages act as amplifiers during septic shock in a positive feedback loop, probably via NF-
B activation,32 intensifying and disseminating the inflammatory response.
| Mediators of Innate and Adaptive Immunity |
|---|
|
|
|---|
Although T and B lymphocytes, the detector cells of adaptive immune responses, differ entirely from those of innate immunity, the effector pathways overlap to a great extent. Thus, T cell activation leads to secretion of the cytokine interferon-
(IFN-
), which primes macrophages, lowering their threshold for TLR-dependent activation. In addition, T cells can produce TNF-
, a proinflammatory cytokine with NF-
B activating capacity. Moreover, the activated T cells express CD40 ligand (CD40L or CD154), which ligates its receptor, CD40, on macrophages; B cells; and many other cells including DCs, ECs, and SMCs.33,34 By involving inflammatory cells in the effector phase, T cells with the T helper-1 (Th1) phenotype tend to promote and amplify the same kind of inflammatory responses also induced when innate immune cells recognize PAMPs through their pattern-recognition receptors.
| Initiation of Atherosclerosis |
|---|
|
|
|---|
The animal studies, as well as cell culture experiments and studies of human tissue samples, support the view that the initiation of atherosclerosis often represents a response of the innate immune system to the accumulation and modification of lipoproteins in the arterial intima. Extracellular accumulation of lipids occurs very early in response to increased plasma lipoprotein levels in animals.35,36 Proteoglycan and protein-bound lipoprotein particles, perhaps in microenvironments shielded from plasma antioxidants, can undergo modification.37 39 Such modifications include oxidation of the lipid or protein moieties40 as well as nonenzymatic glycation of lipoproteins.41,42 Furthermore, the products of hypochlorous acid-induced modification of lipoproteins localize in atheroma.43 Of note, macrophages within atheroma can contain myeloperoxidase, the enzyme that produces hypochlorous acid.44,45
Several lines of evidence suggest that microbial products may promote plaque growth and/or activation. Bacterial products such as LPS and heat shock proteins (HSPs) may act on vascular cells, DNA and proteins of certain microorganisms can be detected in lesions, and seroepidemiological data show correlations between antibody titers to microbes and progression of cardiovascular disease.46,47 The recent finding of TLR expression in atherosclerotic plaques25,48 offers a possible mechanism by which microbial products may activate plaque cells.
Excellent in vitro evidence supports an important effector role for variously modified lipoproteins and their constituents in triggering the production of the mediators of innate immunity.49,50 In addition, nonlipid mediators implicated in vascular disease may also elicit cytokine gene expression. For example, angiotensin II, previously regarded primarily as a vasoconstrictor molecule, can induce the elaboration of cytokines from atheroma-associated cells.51,52
Cytokines elicited by such atherogenic stimuli augment the expression of the genes encoding various leukocyte adhesion molecules, increasing their expression on the surface of ECs in regions of nascent atheroma formation. Candidate molecules for endothelial-leukocyte adhesion in early atherogenesis validated by experiments in genetically altered mice include vascular cell adhesion molecule-1 (VCAM-1) and E- and P-selectin.53,54 In addition, lipid components of modified lipoproteins can directly induce adhesion molecules. Lysophosphatidyl choline and other phospholipid species generated during lipid peroxidation act as proinflammatory stimuli, inducing VCAM-1 in ECs.55 Reactive oxygen species may also induce VCAM-1 due to their capacity to activate NF-
B.56
Once adherent, the leukocytes migrate into the underlying intima in response to chemoattractant stimuli. The chemoattractant cytokines (chemokines) probably participate in this process. Interruption of signaling due to monocyte chemoattractant protein-1 (MCP-1), for example, will retard lesion formation in hypercholesterolemic mice.57 MCP-1 attracts mononuclear leukocytes bearing the chemokine receptor CCR-2.58 Different categories of chemokines may participate in recruitment of distinct leukocyte classes to the atheroma. For example, a trio of CXC chemokines (IP-10, Mig, and I-Tac) selectively attract T and B lymphocytes, which bear the CXC R3 receptor.59 Recent work has localized eotaxin in human atherosclerotic plaques.60 This CC chemokine, in addition to recruiting mononuclear phagocytes, may participate in mast cell accumulation within atheroma.
Once resident in the arterial intima, monocytes differentiate into macrophages, which accumulate intracellular lipid. This process depends on the expression of scavenger receptors,61 including SR-AI and II, CD36, MARCO, SR-PSOX, and CD68, also known as macrosialin.26,62,63 As discussed above, these receptors recognize structural motifs shared by a wide variety of microbial macromolecules, as well as apoptotic cells and modified lipoproteins. Uptake through SR-A can lead to presentation of processed ligands to specific T cells, and this receptor therefore links innate and adaptive immunity.64
Other potentially important features of the lipid-laden macrophage include proliferation65 and the elaboration of certain cytokines and growth factors. In addition to MCP-1, macrophage-colony stimulating factor (M-CSF) appears to play a key role in the activation of various macrophage functions implicated in atherogenesis. Studies of human and experimental atherosclerosis have documented overexpression of M-CSF within lesions.66,67 Moreover, mutant mice lacking the ability to produce M-CSF display delayed atheroma development in a gene dosage-dependent fashion.68
Complement constitutes an additional family of effector proteins involved in innate immunity. In experimental atheroma, complement activation can actually precede the development of lesions.69 Products of the complement cascade including anaphylatoxin can attract leukocytes. In addition, the terminal membrane attack complex of complement can promote damage to cell membranes and eventual cell death. Sublethal injury may permit the release of growth factors, such as fibroblast growth factor, from cells.70 C-reactive protein, an acute phase reactant whose serum and lesional levels appear elevated during atherogenesis, may activate complement.71 However, the role of complement in atherosclerosis remains controversial.72,73
| Mobilization of Adaptive Immunity in the Atheroma |
|---|
|
|
|---|
Most of the T cells in atherosclerotic lesions bear the CD3 and CD4 markers and the T-cell antigen receptor (TCR
ß+). This pattern of surface molecule expression implies that they recognize protein antigens presented to them by macrophages or DCs after uptake and processing through the endosomal pathway. Such cells represent approximately two-thirds of all CD3+ T cells in advanced human lesions74,76 and more than 90% of T cells in lesions of apoE knockout (KO) mice.77
Lesional T cells largely have properties of the Th1 subtype and secrete the cytokines IFN-
, IL-2, TNF-
, and -ß.78 These cytokines cause activation of macrophages and vascular cells, promote inflammation, and also participate in cellular immunity. In addition, many lesional cells produce the Th1-stimulatory cytokines, IL-1279,80 and IL-18.80,81 In contrast, human atheroma contain only modest quantities of Th2 cytokines such as IL-4, IL-5, and IL-10.78 In apoE-KO mice, arterial lesions contain Th2 cytokines only in conditions of extreme hypercholesterolemia.82
Notably, Th1 and Th2 cytokines exhibit cross-regulation: IL-10 inhibits the Th1 pathway, whereas IL-12 reciprocally inhibits Th2 responses. The low level of Th2 activity in lesions may therefore result at least in part from local IL-12 secretion. Interleukin-10 may correspondingly dampen the Th1 response. Accordingly, IL-10-deficient mice have increased fatty streak development.83,84 In addition to regulatory cytokines, the level of antigen expressed in the vicinity of the antigen-presenting cell and the T cell may also influence the Th1/Th2 bias. Therefore, the level of autoantigens such as modified lipoproteins in the lesions and regional lymph nodes may act in concert with IL-10/IL-12 regulation to control the balance between the two Th effector pathways.
In addition to helper T cells, atherosclerotic plaques harbor moderate numbers of CD8+ T cells as well as occasional B cells.74,85 By exhibiting cytotoxic activity and producing antibodies, respectively, both these cell types may wield an importance greater than indicated by their sheer numbers.7
As explained, antigen presented by macrophages or ECs readily activate memory-effector T cells. Naïve T cells, however, require presentation by DCs. Interestingly, such cells have been detected in human as well as experimental atherosclerotic lesions.86,87 DCs have a high migratory capacity and may " patrol" tissues such as the artery wall in search for antigens. Foreign materials encountered during such surveillance when endocytosed, transported to regional lymph nodes, and presented by DCs, could activate both naïve and memory T cells.20,21
In addition to T effector cells and macrophages, atherosclerotic lesions contain another immune effector cell, the mast cell.88 Although macrophages and T cells by far outnumber mast cells, they may nonetheless function importantly in plaque activation and acute coronary syndromes because they produce a host of proteases (including some not made by macrophages) and accumulate at sites of plaque rupture.89 91 Factors released from mast cells may degrade the extracellular matrix and could also influence the functions of surrounding cells and modify locally deposited lipoproteins.92
| Specific Antigens Initiate Adaptive Immunity in Atherosclerosis |
|---|
|
|
|---|
The early lesions of apoE-deficient mice show evidence for such clonal T cell expansion.94 Human lesions, studied of necessity at later stages and by nature less uniform than their experimental counterparts, present a more complex situation. Advanced human atheroma contain a heterogeneous population of TCRs and therefore T cells.9598 Clonal expansions may occur in earlier phases of disease; however, the limited availability of early lesions makes it difficult to assess this possibility.
Autoimmune conditions are often linked to certain MHC alleles. In type I diabetes, for example, one specific HLA-DQ allele favors an autoimmune response that leads to ß-cell destruction and disease. However, the ubiquity and multifactorial nature of human atherosclerosis make a simple relationship with a single MHC determinant unlikely. Again, the access to mouse models of human disease has made it possible to study immunogenetic aspects of atherosclerosis. Such studies have demonstrated a disease-promoting role for (MHC class II) I-Ab restricted Th1 cells, contrasting to an antiatherosclerotic effect of I-Ek and I-Ak in the development of fatty streaks in fat-fed C57BL/6 mice.99
Evidence from studies of human disease supports the involvement of autoantigens in atherosclerosis. T cells can be isolated from fresh human plaques, cloned and expanded in culture, and challenged with candidate antigens. Such experiments identified oxidized LDL as a major autoantigen in the cellular immune response of atherosclerosis.100 This finding, together with the detection of anti-oxLDL antibodies in atherosclerotic patients and experimental animals, 101 supports the concept that immune responses to oxLDL operate in atheroma. Lymph nodes and spleens of apoE-deficient mice can give rise to oxLDL-specific T and B cell lines that display strong humoral as well as cellular immune responses to such modified lipoproteins.102105
HSPs comprise further candidate antigens in atherosclerosis. These proteins produced in large amounts by injured cells act as chaperones to limit denaturation of other cellular proteins. HSPs serve as targets for autoimmune responses in many inflammatory diseases, including rheumatoid arthritis and Crohns disease.106
Immunization with HSP65/60 induces vascular inflammation, with infiltrates of HSP60-reactive T cells.107109 Peripheral blood of atherosclerotic animals contains anti-HSP60 antibodies, and immunization with HSP60 can aggravate disease in rabbits and mice.107,110 Interestingly, HSP60 can activate TLR4, in a CD14-dependent manner, similar to bacterial endotoxin.111 Therefore, HSP60 release may not only induce specific antibodies and T cells but also directly activate innate immunity.
A third proposed autoantigen, ß 2-glycoprotein Ib (ß2GpIb), is present on platelets and under some circumstances on ECs. In several inflammatory disorders, including atherosclerosis, lupus, and the antiphospholipid antibody syndrome, plasma contains autoantibodies to ß2GpIb.112,113 The immune response to ß2GpIb appears to promote atherosclerosis; however, the mechanism involved remains obscure.114,115
Some evidence has implicated microbial pathogens in atherogenesis, and bacteria may induce innate immunity, molecular mimicry, and autoimmunity as well as direct infection of tissues. Several studies suggest a role for Chlamydia pneumoniae in atherosclerosis. Interestingly, HSP60 of this microbe resembles human HSP60 and can elicit inflammatory responses.116 Another putative vascular pathogen, cytomegalovirus, encodes a chemokine receptor that renders infected SMCs susceptible to CC chemokine-induced migration.117 It remains uncertain whether immune reactions to microbes and/or molecular mimicry between microbes and autoantigens contribute to atherosclerosis.
| Immune Cytokines Regulate Vascular Cells |
|---|
|
|
|---|
Immune cytokines also influence the development and maintenance of differentiated properties in the vasculature. In SMCs, the gene for the contractile protein,
-actin, is stimulated by TGF-ß but inhibited by IFN-
.120 Similarly, TGF-ß strongly promotes the synthesis of interstitial collagens (types I and III) by human SMCs, whereas IFN-
powerfully inhibits their synthesis of collagen as well as
-actin.121 ECs respond to proinflammatory cytokines and chemokines by expressing adhesion molecules (see earlier) but also by reducing the continuity of interendothelial junctions, leading to increased permeability across the endothelial barrier.122
As lesions progress, they frequently accumulate calcium mineral, which is a tightly regulated process. Cytokines control the expression of osteopontin/Eta-1, a protein implicated in lesion calcification but also in Th1 immunity.123,124 Members of the TGF-ß family of cytokines, including the bone morphogenetic proteins (BMPs), may also participate in this process.124 Mice lacking M-CSF exhibit exaggerated calcification of atherosclerotic lesions in response to hyperlipidemia.68 This support the view that intralesional macrophages may function as osteoclasts during atherogenesis. The steady-state level of calcium in a lesion at any given time probably reflects the balance between mineralization and dissolution due to this osteoclastic activity of the M-CSF- activated macrophage.
Importantly, vascular cells do not merely respond to cytokines but also can produce large amounts of these molecules. Both ECs and SMCs respond to stimulation with IL-1, TNF-
, or CD40 ligand by producing large amounts of IL-6 and also by augmented expression of IL-1 and CD40.125 They also express PTX-3, a member of the pentraxin gene family, which also includes CRP.126 PTX3 is found in atherosclerotic plaques127 and elevated serum levels of this cytokine is an early marker of myocardial infarction.128 In summary, vascular cells participate in and propagate the inflammatory response at sites of microbial challenge or pathological processes. Table 3 summarizes some important cytokine-vascular interactions.
|
| Importance of Immune Mechanisms in Atherosclerosis Deduced From Genetically Altered Mice |
|---|
|
|
|---|
|
Several studies demonstrate a modulating role for adaptive immunity. The complete lack of adaptive immunity caused by the RAG or SCID mutations results in a 40% to 70% reduction of atherosclerosis in apoE-/- mice.132,133 Although these findings imply that atherosclerosis develops also in the absence of adaptive immunity, they point to an important modulating role of the arm of host defense. When conducted under conditions of extreme hypercholesterolemia, the modulatory effect of this immune defect is diminished.132,134 Importantly, reconstitution of SCIDx apoE-/- mice with CD4+ T cells from immunocompetent apoE-/- mice accelerates disease almost to the level of the fully immunocompetent apoE-/- mouse.133 This result pinpoints CD4+ T cells as playing a proatherosclerotic role, a conclusion consistent with the substantial reduction of atherosclerosis observed in apoE-/- mice lacking interferon-
signaling.135,136 CD40 ligation also plays an important proatherogenic role because interruption of CD40 signaling substantially reduces murine atheroma formation137,138 or evolution.139,140 Moreover, interruption of CD40/CD40L interaction yields lesions that express features of plaque stabilization, including diminished lipid and enhanced collagen content.139,140
Immunization with oxidized LDL reduces atherosclerosis in hypercholesterolemic rabbits and mice141144 and transfer of immunoglobulins also inhibits disease development.145 These observations point to atheroprotective immunity that may modulate the proatherogenic immune pathways. Support for this notion was recently obtained when protection against atherosclerosis was achieved by transferring B cells from atherosclerotic apoE KO mice to young apoE KO mice that had not yet developed disease.105 It is not yet known whether such atheroprotective immunity depends on circulating antibodies or on T-B cell interactions. However, the demonstration of transferable, atheroprotective immunity encourages further studies of immunization and immunomodulation as possible means for treatment of atherosclerosis.
| Immune Mediators Regulate the Thrombotic Complications of Atherosclerosis |
|---|
|
|
|---|
, IL-1, interferon-
, and PTX-3 all induce tissue factor gene expression in human ECs.146,147 CD40 ligation can also induce tissue factor gene expression in human monocyte/macrophages.148 In atheroma, macrophages, SMCs, and ECs, can all express CD40 Ligand (CD154).125 In addition, activated platelets express functional CD154.149,150
The accumulation of thrombus depends not only on its formation due to tissue factor-induced activation of the coagulation cascade, but also on clot breakdown due to thrombolysis. The principal atheroma-associated fibrinolytic mediators, tissue type, and urokinase plasminogen activators vary depending on the milieu of mediators of innate immunity. Soluble cytokines such as interleukin-1 and TNF-
can alter the activity of plasminogen activators, as well as thrombomodulin, an antagonist of coagulation.151 Cytokines can also augment the expression of inhibitors of the endogenous plasminogen activators.151 In this manner, mediators of innate immunity can modulate the delicate balance between clot formation and dissolution.
A physical disruption of the atherosclerotic plaque precipitates thrombus formation in a majority of cases.152154 Rupture of the plaques fibrous cap allows contact of the blood coagulation proteins with the tissue factor procoagulant found within the intima. It is likely that a dynamic balance between collagen synthesis and degradation determines the fragility of the plaques fibrous cap, and hence the tendency to rupture and cause thrombosis.154 Interferon-
inhibits collagen synthesis and may therefore link the immune response to weakening of the fibrous cap. Matrix metalloproteinases, cathepsins, and mast cell proteases can impair the integrity of the fibrous cap by degrading its collagen cap. Proinflammatory cytokines can regulate the release of these matrix-degrading proteinases.
Because the SMC produces most of the interstitial collagen that lends strength to the plaques fibrous cap, paucity of these cells may render a plaque weak and vulnerable to rupture. Indeed, death, including death by apoptosis, of SMCs in the advanced atheroma might impair the ability of this cell type to repair and maintain the extracellular matrix that regulates the integrity of the plaques all-important fibrous cap.155 IFN-
, not only inhibits smooth muscle proliferation,120,156 but may also, together with TNF-
and IL-1, promote apoptosis of SMCs.119 Loss of SMCs, in part governed by mediators of immunity, can influence the susceptibility of a plaque to rupture. The findings of activated macrophages, T cells, and mast cells and of proteolytic enzymes at sites of plaque rupture in human coronary atheroma support the link between immunity and thrombosis.
| Conclusion |
|---|
|
|
|---|
Despite the plethora of mediators and pathways that prevail during this prolonged period of lesion evolution, a unifying principle can simplify the fundamental concepts. The dynamism of plaque biology emerges as a major simplifying concept. Balances between positive and negative signals, between synthetic and degradative processes, between life and death, regulated by the alphabet soup of mediators ultimately determine the tempo of lesion evolution, complication, and clinical manifestations. By evoking elements of host defense reaction, atherosclerosis shares much with other inflammatory and/or fibrotic diseases. Future work will no doubt add to the list of mediators involved in regulating these processes. As we fill in the molecular details, new potential targets for therapies will doubtless emerge.
| Acknowledgments |
|---|
Received December 17, 2001; revision received July 3, 2002; accepted July 3, 2002.
| References |
|---|
|
|
|---|
B and JNK/SAPK activation upstream of tumor necrosis factor receptor-associated factor 6 (TRAF6). J Exp Med. 1998; 187: 20972101.
B through toll-like receptor 4 (TLR-4) in cultured human dermal endothelial cells: differential expression of TLR-4 and TLR-2 in endothelial cells. J Biol Chem. 2000; 275: 1105811063.