Atherosclerosis was once thought to be a simple lipid storage disease which caused pathology by arterial obstruction. It certainly leads to arterial obstruction, but not simply by accumulation of lipid. Now, it is clear that atherosclerotic lesions are foci of vessel wall inflammation [1–4]. Lesions begin very early in life as fatty streaks, progressing to pathological lesions under the influence of both genetic and lifestyle insults [5]. The final obstructive event is more likely to be a sudden thrombotic event rather than a slow, inexorable constriction by accumulation of lipid. During the major part of a human lifetime that may elapse between fatty streak formation and overt disease, there will be many events that can accelerate lesion progression. Genetic diseases, notably hypercholesterolemias, lead to very severe disease [6]. Lifestyle choices which include smoking, obesity, and excessive quiescence can also make significant contributions to disease progression. Finally, inflammatory disease elsewhere in the body, ranging from periodontitis to autoimmune disease, can have an exacerbating influence as well [7–10]. However, the factors which promote atherosclerotic lesion development do not do so uniformly throughout the arterial tree. Lesions form first at sites of disturbed flow [11]. Such sites are at vessel bifurcations and at the lesser curvature of the aortic arch. These sites display an inflamed phenotype even in the absence of any of the exacerbating disease factors. This is due to the altered flow patterns at these sites and the endothelial cell’s response to this disturbed flow [12–14]. Although the biochemical mechanisms by which endothelial cells detect flow patterns are beginning to be understood [15–17], the need to detect flow patterns is not. Evolutionarily, TLR expression and inflammation are certainly driven in part by infectious diseases [18]. However, no particular disease risk has been identified that would promote a pro-inflammatory phenotype only at arterial bifurcations and the lesser curvature of the aortic arch.

If inflammation is a hallmark of the disease, then it is only logical to study the roles of Toll-like receptors (TLRs) in atherosclerosis. TLRs are initiators of inflammation driven by agonists which may be either endogenous, such as sterile tissue damage [19], or exogenous, such as pathogens. Probably, given the vicissitudes of daily life, TLRs normally initiate inflammatory responses everyday. The majority of these are no doubt beneficial, promoting healing and homeostasis, but occasionally, they go awry and cause pathology. Atherosclerosis is one of the pathologic consequences, although TLRs are by no means the only cause or promoter of this disease.

Our laboratory studies TLRs and atherosclerosis in C57Bl/6J mice. There are two C57Bl/6J models in common use, each involving a single gene deletion that leads to plasma hypercholesterolemia. When wild-type C57Bl/6J mice are fed a high-fat, high-cholesterol diet, they do not develop atherosclerosis because they do not develop hypercholesterolemia. However, mice deficient in the gene encoding the low-density lipoprotein receptor (LDLr), a gene the lack of which causes human familial hypercholesterolemia and whose discovery and study won the Nobel prize for Michael Brown and Joe Goldstein [20], leads to hypercholesterolemia because synthesis and dietary intake of fat continues, but clearance from the plasma for catabolism in the liver is lost. The plasma cholesterol accumulates in the intima of the arteries in foam cells and leads to lesion development [13]. The other common model for the study of murine atherosclerosis is the ApoE-deficient mouse. ApoE deficiency leads to atherosclerotic lesion development because its absence causes massive hypertriglyceridemia, again leading to foam cell formation. For study of the TLRs in atherosclerosis, we have used the LDLr-deficient (LDLr−/−) mouse exclusively [21, 22]. The reason for this is a disadvantage of the ApoE model. Bone-marrow-derived cells (BMDC) synthesize and secrete significant quantities of ApoE. Thus, transplantating BMDC from ApoE-expressing mice into ApoE-deficient mice reverses the ApoE deficiency and the phenotype of the host mouse. However, transplanting bone marrow from an LDLr-expressing mouse into an LDLr−/− mouse does not reverse the phenotype of the host mouse [23]. The utility of these BMDC chimeras will become obvious below.

In addition to our experiments, reviewed below, others have studied disease development in MyD88-, TLR4-, and CD14-deficient mice [24, 25]. These studies showed that TLR4 and MyD88 deficiency lessened disease development, while CD14 deficiency had no effect on disease development.

We focused on TLR2 because in vitro experiments with human coronary artery endothelial cells showed that cells exposed to disturbed flow culture conditions to simulate flow patterns at sites of lesion development upregulated responsiveness to TLR2 agonists [26]. We found this to be due to an enhanced expression of TLR2 in response to disturbed flow. We did not observe that disturbed flow or no flow enhanced responses to TLR4 or TNF receptor ligands (see Fig. 1).

Fig. 1
figure 1

Responses of human coronary artery endothelial cells to ligands of a TLR4, b TNF receptor, or c TLR2 under either laminar flow conditions (open circles) or no flow conditions (solid circles) which mimics disturbed flow. Note that only TLR2 responses are modified by flow vs no flow

Full size image

Initial results with LDLr−/−TLR2−/− mice showed that whole body deficiency of TLR2 results in decreased lesion burden after 10 or 14 weeks of high-fat diet consumption (see Fig. 2) [21]. Thus, functional TLR2 promotes disease. In these experiments, no agonist of TLR2 was deliberately administered to the mice, and thus, we refer to the TLR2 agonists involved as “endogenous” agonists. It should be understood that the mice studied were healthy, but not germ-free. Thus, the endogenous agonist may come from the genome of a commensal organism [27] or from the genome of the mouse itself. However, Wright et al. [28] have shown that germ-free ApoE-deficient mice do develop atherosclerosis, and thus, we favor the hypothesis that the endogenous TLR2 agonists are derived from the mouse itself. Others had shown previously that administration of exogenous LPS, the TLR4 agonist, promoted disease in rabbits and mice [29, 30]. In additional experiments, we showed that administration of Pam3CSKKKK, a synthetic TLR2 agonist, promoted additional disease in the LDLr−/− model and that this was abrogated in LDLr−/−TLR2−/− mice [21]. Thus, both endogenous and exogenous agonists promote disease by activation of TLR2 signaling, and TLR2 is one route for both the genetic and inflammatory pathways to atherosclerotic disease. These findings were recently confirmed in ApoE-deficient mice [31].

Fig. 2
figure 2

Aortic sinus lesion volumes of LDLr−/− mice fed a high fat diet are lessened by deficiency of TLR2 at either of two time points

Full size image

The technique of bone marrow transplantation allowed us to ask whether TLR2 expressed by BMDC or non-BMDC was disease-promoting. To our surprise, we found that BMDC and non-BMDC were both disease-promoting, but which cells were involved depended on whether the agonist was endogenous or exogenous. Non-BMDC-expressed TLR2 was sufficient to promote disease in response to the endogenous agonists, while BMDC-expressed TLR2 was sufficient to promote disease only in response to exogenous agonists (see Fig. 3) [21]. Given the known role of endothelial cells in regulating transport into the intima, it seems likely that the non-BMDC responding to endogenous TLR2 agonists are endothelial cells. Similarly, given the known relevance of monocytes or macrophages in responding to exogenous TLR2 agonists, it seems likely that these are the cells driving lesion development in response to the exogenous agonists.

Fig. 3
figure 3

Aortic sinus lesion volumes of LDLr−/− or LDLr−/−TLR2−/− mice after undergoing bone marrow transplantation with bone marrow from a wild type or a TLR2−/− donor. Note that the lesion volume phenotype correlates with the recipient’s TLR2 expression, not the donor’s TLR2 expression

Full size image

The mechanism by which these specificities for endogenous and exogenous agonists come about is unknown. In this volume, Lee and Raz describe the differential effects of TLR9 expression in intestinal epithelial cells when TLR9 is expressed apically as opposed to basolaterally. In a similar vein, the work reported by Gewirtz notes that TLR5 is expressed only basolaterally in intestinal epithelial cells. We hypothesize that something similar could be happening with TLR2 on vascular endothelial cells that might explain our findings. Thus, if TLR2 were expressed by endothelial cells not at the apical membrane facing the vascular lumen but only at the basolateral membrane exposed to the intima, it would seem to be correctly located to detect endogenous agonists in the intima, but not available to detect exogenous blood-borne agonists. At present, there are no data to show where TLR2 is expressed in vascular endothelial cells, but endothelial cells are well known to express other proteins in a polarized fashion (e.g. [32]).

Using confocal microscopy of aortic arch tissue fragments, we have examined the distribution of TLR2 expression on the luminal face of the murine aortic arch (Fig. 4) [33]. These images show very little TLR2 expression in either wild-type C57Bl/6J mice or in LDLr−/− mice fed a chow diet. Importantly, even the minimal TLR2 expression that is present is seen in the lesser curvature region. Within 1 to 2 weeks of feeding a high-fat diet to LDLr−/− mice, TLR2 expression is significantly increased in the lesser curvature region of the aorta. However, it is expressed only by the endothelial cells, and not by the myeloid cells, in the intimal layer just below the endothelial cells [33]. Why these myeloid cells are not expressing TLR2 is open to conjecture. Brunn et al. [34] have shown that TLR4 signaling is strongly inhibited by extracellular matrix components. Something similar could be happening with TLR2. Certainly, the intimal macrophages are residing in a space containing much extracellular matrix material. Further, when one studies which cells are accumulating lipid at these early stages of lesion development, it is not the myeloid CD68-expressing cells, but rather the CD31-expressing endothelial cells [33]).

Fig. 4
figure 4

LDLr−/− and LDLr−/−TLR2−/− mice were fed a high-fat diet for the number of weeks as indicated in the figure. After sacrifice, expression of TLR2 and CD31, an endothelial cell membrane marker was visualized by laser scanning confocal microscopy. The figure shows that alterations in TLR2 expression and endothelial morphology are visible in LDLr−/− mice as early as one week after initiation of the diet. The LDLr−/−TLR2−/− mice show no staining for TLR2, as expected, and much less perturbation of endothelial morphology

Full size image

A recent study concluded that overexpression of endothelial TLR2 in rabbits fed a high-fat diet led to enhanced progression of atherosclerosis. Thus, both diminution and enhancement of TLR2 have effects that point to a role for TLR2 in atherosclerosis [35].

What are the endogenous TLR2 agonists to which endothelial cells are sensitive ? Given the important role of oxidized lipoproteins in lesion development, it is tempting to point to these moieties, but no oxidized lipoprotein has yet been shown to be a TLR2 agonist. Other possibilities include HMGB1, biglycan, hyaluronic acid fragments, and SAA [19, 36–40]. All of these have been shown to be TLR2 agonists in vitro. We hypothesize that these agonists, and possibly others, may activate endothelial cells at lesion prone sites through TLR2 because it is only the endothelial cells that express TLR2.

If the aortic bone-marrow-derived myeloid cells do not express TLR2, how is it that exogenous agonists such as Pam3CSKKKK promote lesion development? There are at least two possibilities which we are considering. First, that the exogenously administered agonist stimulates a mediator response which drives TLR2 expression on the intimal myeloid cells within the intima. Alternatively, the exogenous agonist stimulates mediators that promote lesion-prone site-specific endothelial activation, which in turn serves to promote recruitment of monocytes that then enter the lesion and become activated by other means. These two hypotheses should not be difficult to distinguish experimentally.

In as much as inflammation is a hallmark of atherosclerosis, anti-inflammatory therapies will probably become more important as therapeutic measures. Statins are thought to have important anti-inflammatory properties, although their particular targets are not well defined [41]. Inhibitors of TLRs have not been tested in models of atherosclerosis, but there are some which should be tested. Youn et al. [42, 43] have found that curcumin and auranofin inhibit TLR4 by interfering with TLR4 homodimerization. Others have found that endotoxin analogs are also effective inhibitors of TLR4 [44, 45]. These reagents all interact with the extracellular domains of TLR4. We have recently utilized an high throughput screening assay to identify a group of low molecular weight, water-soluble reagents that interfere with the intracellular signal initiating mechanism of TLRs 2 and 4 [46]. Any of these approaches could lead to the development of important new drugs to modify the activity of TLRs.

In conclusion, there are abundant data to indicate that inflammatory processes play a pivotal role throughout the development of atherosclerosis. The idea that TLRs participate in these processes is now valid, although only two, TLR2 and TLR4, have been specifically shown to do so. Multiple cell types within the atheroma, including endothelial cells, dendritic cells, macrophages, vascular smooth muscle cells, and perhaps even adventitial fibroblasts, could participate in this inflammatory response. TLR2 does not cause atherosclerosis, but it certainly participates in its development. Given that the endogenous agonists that drive atherosclerotic disease are likely to be involved in normal surveillance and homeostatic processes, atherosclerosis can be considered as an innate autoimmune disorder.