OUP user menu

★ Editor's choice ★

Toll-like receptor 3 and 4 signalling through the TRIF and TRAM adaptors in haematopoietic cells promotes atherosclerosis

Anna M. Lundberg, Daniel F.J. Ketelhuth, Maria E. Johansson, Norbert Gerdes, Sang Liu, Masahiro Yamamoto, Shizuo Akira, Göran K. Hansson
DOI: http://dx.doi.org/10.1093/cvr/cvt033 364-373 First published online: 15 February 2013

Abstract

Aims Members of the Toll-like receptor (TLR) family initiate innate immune responses and were recently shown to play a role in atherosclerosis. However, the mechanisms that link TLR ligation to vascular inflammation and atherogenesis remain unclear. To identify which signalling pathways downstream of TLRs in immune cells are pro-atherogenic, we analysed the role of the TLR-specific adaptors MyD88 adaptor-like (MAL), TRIF-related adaptor molecule (TRAM), and TIR-domain-containing adaptor-inducing interferon-β (TRIF) in atherosclerosis.

Methods and results Using a bone-marrow transplantation strategy into low-density lipoprotein receptor-deficient (Ldlr−/−) mice, we could specifically study the absence of the TLR adaptors in immune cells. We showed that haematopoietic deficiency of TRAM and TRIF, but not MAL, reduces atherosclerosis without affecting cholesterol metabolism. This was mediated by decreased aortic inflammation, indicated by lower aortic levels of pro-inflammatory mediators, and reduced influx of macrophages and T cells. Furthermore, by studying Tlr3−/− chimeric Ldlr−/− mice, we found that deleting TLR3 in immune cells significantly reduced both aortic inflammation and atherosclerotic burden.

Conclusions By studying hypercholesterolaemic mice with defects in TLR-signalling adaptors, we demonstrated that deleting either TRAM or TRIF in immune cells is sufficient to attenuate vessel inflammation and protect against atherosclerosis. In addition, these adaptors elicit partly different sets of inflammatory mediators and can independently inhibit the disease process. Furthermore, we identify TLR3 as a pro-atherogenic receptor in haematopoietic immune cells. The identification of these pro-atherogenic pathways downstream of TLR3 and TLR4 contributes to a better understanding of TLRs and their signalling pathways in the pathogenesis of atherosclerosis.

  • Atherosclerosis
  • TRAM
  • TRIF
  • TLR3
  • Vascular inflammation

1. Introduction

Atherosclerosis is a chronic inflammatory disorder in which metabolic and immune components interact to initiate and propagate arterial lesions.1 Immune cells derived from haematopoietic cells in the bone marrow have been identified as key players in driving the disease process. Retention and modification of lipoproteins in the vessel wall lead to attraction of monocytes that eventually differentiate into macrophages. Macrophages are the major immune cells in the lesion and they produce pro-inflammatory cytokines, participate in lipid uptake, and express pattern-recognition receptors, including Toll-like receptors (TLRs) that connect the innate and adaptive immune response during atherosclerosis. TLRs are important for the detection of both pathogenic infections and endogenous danger signals and several lines of evidence have identified them as initiators of inflammation in the lipid-laden artery wall.2 We have detected several TLRs on activated cells in human atherosclerotic plaques3 and experimental studies have shown pro-atherogenic effects of TLR2 and TLR4 in hyperlipidaemic mice.4,5 However, the pathways that link TLRs to pro-atherogenic effectors pathways in immune cells remain poorly characterized.

TLR signalling is controlled by four cytoplasmic receptor adaptors (as reviewed in Akira et al.6); myeloid differentiation factor 88 (MyD88), MyD88 adaptor-like (MAL), TRIF-related adaptor molecule (TRAM), and TIR-domain-containing adaptor-inducing interferon-β (TRIF) (Figure 1). MyD88 mediates signalling from the IL-1/IL-18 receptors and all TLRs except TLR3, whereas MAL facilitates the interaction between MyD88 and the TLR2 and TLR4 receptors. Together, MyD88 and MAL initiate a common pathway that activates nuclear factor (NF)κB. In contrast, TRIF mediates signalling from TLR3 and TLR4, while TRAM operates as a bridging adaptor between TRIF and TLR4. The TRIF- and TRAM-dependent pathways result in activation of both the transcription factor interferon regulatory factor (IRF)-3 and the alternate pathway to NFκB.

Figure 1

TLR signalling pathways. TLRs activate common and unique transcription factors, including nuclear factor (NF)κB and interferon regulating factor 3 (IRF-3). Stimulation with TLR ligands, including infectious material and products of tissue stress, results in recruitment of TLR adaptors. TLR3/7/8/9 are expressed in the endosomes (not shown). With the exception of TLR3, all TLRs and IL-1R/IL-18R recruit the adaptor myeloid differentiation factor 88 (MyD88), which signals to activate NFκB. In addition, TLR4 and TLR3 trigger a TIR-domain-containing adaptor-inducing interferon-β (TRIF)-dependent signalling pathway, which induces activation of IRF-3. The adaptors MyD88 adaptor-like (MAL) and TRIF-related adaptor molecule (TRAM) are important for recruiting MyD88 and TRIF to the TLR2 and TLR4 receptors. TLR4, TLR2, IL-1, IL-18, and MyD88 have all been found to mediate pro-atherosclerotic effects.

Since immune cells within the lesion express many TLRs, it is too simplistic to portray one specific TLR as the sole receptor triggering inflammation during atherogenesis. More likely is a combined involvement of several TLRs in disease development. It is therefore important to determine whether a common pathway downstream of these TLRs is essential for driving lesion inflammation. The role of MyD88 has been investigated and hyperlipidaemic mice deficient in this adaptor protein are protected against disease development.7,8 However, this could be a consequence of the crucial role of MyD88 in transducing signals from the receptors for IL-1 and IL-18, two well-known pro-atherogenic stimuli.

In the present study, we investigated the TLR-specific adaptor proteins MAL, TRAM, and TRIF in atherosclerosis in low-density lipoprotein receptor (LDLR)-deficient mice. By using a bone-marrow transplantation strategy, we could specifically study the absence of the TLR adaptors in immune cells. We demonstrate that MAL, which controls the signals from TLR2 and TLR4, has a redundant role in disease development. In contrast, TRAM and TRIF that regulate the signalling downstream of the TLR4 and TLR3 receptors were both found to be pro-atherogenic and that these two adaptors regulate lesion inflammation and atherogenesis through separate mechanisms. In addition, we identify TLR3 as a pro-atherogenic receptor in haematopoietic immune cells. Overall, our data indicate that signalling through the TRIF and TRAM pathways downstream of TLR3 and TLR4 in bone-marrow cells promote the development of atherosclerosis.

2. Materials and methods

2.1 Animals

LDLR-deficient mice were from Jackson Laboratory. MAL, TRAM, and TRIF-deficient mice were generated in the Akira laboratory (Department of Host Defense, Osaka University, Japan). TLR3 mice were kindly provided by Prof. Richard Flavell (Yale University, USA) and Dr Claudia Monaco (Imperial College, UK). The mice were weaned at 4 weeks, and all procedures were approved by the Stockholm North Committee for Experimental Animal Ethics. The investigations conform with the Directive 2010/63/EU of the European Parliament.

2.2 Bone-marrow transfer

Bone-marrow transfer (BMT) was performed on female Ldlr−/−-recipient mice that were irradiated with two doses of 700 rad, 3 h apart. Donor mice were euthanized with CO2 and bone-marrow cells were isolated from the femur and tibia. BMT was performed on the recipients in the absence of anaesthesia through i.v. injection in the tail vein of 5 × 106 bone-marrow cells per recipient in 50–100 µL. Matched recipients (6–9 weeks old for each study) were randomly assigned to receive bone marrow from Mal−/−, Tram−/−, Trif−/−, Tlr3−/− or corresponding C57Bl/6 control female donor mice (6–9 weeks for each study) (n = 7–11 recipients/group) in four separate experiments. Transplanted mice were left to recover for 6 weeks, and following feeding of atherogenic diet (1.25% cholesterol, 0% cholate) (D12108, Research Diets, USA) for 7–8 weeks, mice were terminated within 7 days in matched sets from both experimental and control groups.

2.3 Tissue preparation

Mice were euthanized by CO2, blood was saved for serum analysis, and organs were dissected after vascular perfusion. The aorta was cut between the heart and innominate artery and the heart frozen and cryosectioned. The aorta down to the bifurcation was dissected, carefully removing the adventitia, and snap-frozen for RNA isolation.

2.4 Lesion analysis and immunohistochemistry

Lesion analysis was performed as described.9 Briefly, 10 μm sections, collected at 100 μm intervals from 200-μm distance after aortic valve appearance, were fixed (4% formaldehyde) and analysed after oil red-O staining. Images were captured with a Leica DM-LB2 microscope (Leica, Germany) with a ×20/0.9 objective and a Leica DC300 camera. Lesion surface areas and the circumference of the vessel were measured and lesion size was calculated using Leica QWin software. For each mouse, a mean lesion area was calculated from seven sections. Primary rat anti-mouse antibodies to CD68 (BD Bioscience) were applied to acetone-fixed cryosections followed by detection with the ABC alkaline phosphatase kit (Vector Laboratories, USA) and analysed using Leica QWin software.

2.5 Cholesterol measurements, lipoprotein lipid profiles and oxLDL measurements

Total serum cholesterol was determined using enzymatic colorimetric kit according to the manufacturers' instructions (Randox Lab. Ltd, UK). Serum cholesterol lipoprotein profiles were determined using UFLC chromatography further described in Supplementary material, Methods. OxLDL serum levels were measured using ELISA (antibodies-online.com) using the manufacturer's' instructions. For oxLDL uptake, cells isolated by peritoneal PBS lavage were stimulated and uptake of FITC-oxLDL was measured as further described in Supplementary material, Methods.

2.6 RNA isolation, cDNA synthesis, and real-time PCR analysis

Aortic RNA, prepared using the RNeasy Lipid Mini kit (Qiagen, Germany) and analysed by BioAnalyzer (Agilent Technologies, Germany), were reverse transcribed with Superscript-II (Invitrogen), random hexanucleotide primers (pdN6), and RNasin (Life Technologies, France). cDNA was amplified by real-time PCR using primers and probes (Applied Biosystems, USA) for selected genes and the housekeeping gene hypoxanthine guanidine ribonucleosyltransferase (Hprt) in an ABI 7900HT Sequence Detector (Applied Biosystems). Specific mRNA was expressed as arbitrary units calculated as described previously.10

2.7 In vivo stimulation and cytokine measurement

Trif −/−, Tram−/−, or control mice were injected with 50 µg LPS (Serotype O111:B4 S-form TLRgrade™, Enzo Life Science) i.p. Blood samples (50 μL) were collected from the tail prior to and 3 and 6 h post LPS injection, and plasma were analysed for cytokines using the Cytokine Bead Array (CBA) detection kit (BD Biosciences, Sweden) or ELISA (R&D Systems, Inc., USA) according to manufacturer's instructions. For CBA, the mean fluorescence intensity (MFI) was measured by a CyanADP flow cytometer (Beckman Coulter, USA) and analysed by FCAP Array™ software (Soft Flow, Inc., USA). The MFI each cytokine was subsequently converted to picograms per millilitre based on MFI values from standards.

2.8 Statistical analysis

Results are expressed as mean ± SEM. The Mann–Whitney U test was used for pairwise comparisons and two-way ANOVA with Bonferroni post-test was used to compare differences between groups in lesion size plotted as distances from aortic root and lipoprotein lipid profile. Differences between groups were considered significant at P < 0.05.

3. Results

3.1 Haematopoietic deficiency of MAL fails to inhibit atherosclerotic development and lesion inflammation

To study the adaptor MAL in haematopoietic immune cells, which controls signalling from the two pro-atherogenic receptor TLR2 and TLR4, bone-marrow cells from Mal−/− or control mice were transplanted into Ldlr−/−-recipient mice and fed atherogenic diet to induce lesion formation. The resulting fibrofatty lesions in the aortic sinus did not differ in size between Mal−/− chimeric mice and controls (Figure 2A–C). In addition, no significant changes in either weight, total serum cholesterol (see Supplementary material online, Table S1), cholesterol levels in chylomicrons (CR)/VLDL, LDL and HDL (see Supplementary material online, Figure S1A), or serum levels of oxLDL (see Supplementary material online, Figure S2A) could be observed. Immunohistochemical characterization of the lesions showed similar levels of macrophages independent of MAL expression (Figure 2D). Furthermore, aortic expression levels of the key inflammatory cytokines TNF, IL-6, and IFNγ were unchanged between MAL-deficient Ldlr−/− chimeric mice and controls (Figure 2E–G),.

Figure 2

MAL deficiency does not affect atherosclerosis or aortic inflammation. Aortic root lesions, from Ldlr−/− mice transplanted with bone marrow of control or Mal−/− mice and fed an atherogenic diet, were evaluated by microscopic morphometry of serial sections after staining with Oil Red O. (A) Average lesion size at 200–800 μm from the aortic root is shown, calculated as size covered by lesion (μm2) related to the aortic perimeter (μm). Individual values are displayed by dots, and the mean for each group indicated by horizontal lines. (B) Representative micrographs showing cross sections of Oil Red O-stained lesion 400 μm from the start of the aortic root (black bar 25 µm). (C) Lesion size measured on serial sections throughout the proximal aorta. (D) Graph shows an average lesion area as proportion of total lesion area stained by CD68 antibody in aortic root sections from control or Mal−/− mice. Relative mRNA levels of (E) IFN-γ, (F) TNF, and (G) IL-6 mRNA normalized to Hprt in aorta from the two groups analysed with reverse-transcription real-time PCR. Data expressed as arbitrary units (AU). n.s., non-significant.

3.2 Disrupted TLR signalling through TRIF and TRAM leads to reduced atherosclerotic burden and decreased macrophage infiltration

Having established that MAL does not play a prominent role in atherogenesis, we investigated the two other TLR-specific adaptors, TRIF and TRAM. We observed that haematopoietic deficiency of either of these two TLR adaptors significantly reduced sinus lesion size compared with control Ldlr−/− mice that have intact TLR signalling (Figure 3A and B). Analysis of lesions in the aortic sinus revealed a decrease of 32.6% in Tram−/− chimeric Ldlr−/−mice, whereas TRIF deficiency led to an even greater reduction of 40.1% in lesion size compared with controls. Lesion size was uniformly reduced in cross sections throughout the entire aortic sinus in both Tram−/−and Trif−/− mice (Figure 3C). This effect was accompanied by a lower level of macrophage infiltration in both TRIF- and TRAM-deficient chimeric Ldlr−/−mice as examined by immunostaining of aortic root lesions (Figure 3D). No significant differences in weight, total serum cholesterol (see Supplementary material online, Table S1), cholesterol levels in CR/VLDL, LDL, or HDL (see Supplementary material online, Figure S1B and C), or circulating levels of oxLDL (see Supplementary material online, Figure S2B and C) were detected.

Figure 3

Atherosclerotic burden is reduced in TRIF and TRAM-defective mice. Quantitative Oil Red O lesion analysis in aortic root cross sections of control, Tram−/−, or Trif −/− chimeric Ldlr−/− mice fed an atherogenic diet using microscopic morphometry of serial sections. (A) Average lesion size at 200–800 μm from the aortic root is shown, calculated as size covered by lesion (μm2) related to the aortic perimeter (μm). Individual values are displayed by dots, and the mean for each group indicated by horizontal lines. (B) Representative micrographs showing cross sections of Oil Red O-stained lesion at 400 μm (black bar 25 µm). (C) Lesion size measured in throughout the proximal aorta. (D) Graph shows an average lesion area as the proportion of total lesion area stained by CD68 antibody in aortic root sections. n.s., non-significant. *P<0.05, **P < 0.01, and ***P < 0.001.

3.3 Differential involvement of TRAM and TRIF in controlling aortic expression of pro-inflammatory mediators and recruitment of macrophages and T cells

To determine whether disrupted TRAM and TRIF signalling in immune cells affected vessel inflammation, we evaluated the aortic expression levels from the adaptor-deficient chimeric Ldlr−/− mice and their controls (Figure 4). mRNA expression of the Th1-associated cytokine IFNγ and the pro-inflammatory TNF and IL-6 were all significantly decreased in the aorta of mice with Trif−/− haematopoietic cells (Figure 4A). In contrast, lack of TRAM reduced only expression of IL-6 in the aorta (Figure 4A).

Figure 4

TRIF and TRAM deficiencies differentially regulate aortic pro-inflammatory mediators and influx of macrophage and T cells. Aortic mRNA transcripts from control, Tram−/− or Trif −/− chimeric Ldlr/− mice were analysed with reverse-transcription real-time PCR and normalized to Hprt. Data are expressed as arbitrary units (AU). n.s., non-significant. *P<0.05, **P < 0.01, ***P < 0.001.

We further analysed whether reduction of atherosclerosis was accompanied by changes in cellular composition in the aorta. Aortic infiltration of macrophages as assessed by CD68 expression was significantly lower in Tram−/− and Trif−/− mice (Figure 4B), whereas the levels of the T-cell-specific CD3 transcript were affected only TRIF-deficient chimaeras (Figure 4B). These adaptor-specific effects were also reflected in the local production of chemokines that can attract both macrophages and activated T cells. mRNA levels of CCL2, a key chemokine that can regulate migration of monocytes/macrophages, were significantly reduced in both Tram−/− and Trif−/− mice (Figure 4C). In contrast, the expression of the chemokines CCL5 and CXCL10 that can recruit both macrophages and activated T cells during atherosclerosis11 were abrogated only in TRIF-deficient mice (Figure 4C). TRAM is commonly believed to function exclusively in combination with TRIF; however, our data suggest that TRAM and TRIF have independent roles in initiating aortic inflammation.

We assessed whether TRIF or TRAM deficiency could affect mediators of antigen presentation in the atherosclerotic aorta. mRNA expression of the important co-stimulatory B7.2 (CD86) molecule that ligates CD28 on the T cells and the Th1 polarizing cytokine IL-12 were both reduced in the aorta of mice with TRIF deficiency, whereas impaired TRAM-signalling had no effect (Figure 4D).

Next, the effect of TRIF and TRAM deficiency on the oxLDL uptake by macrophages was studied. No significant differences could be observed in the uptake of FITC-labelled oxLDL by peritoneal macrophages from Tram−/− and Trif−/− mice when compared with controls (see Supplementary material online, Figure S3A). Moreover, aortic expression of the cholesterol efflux transporter, ATP-binding cassette transporter A (ABCA)-1, were unchanged between Tram−/− and Trif−/− chimeric Ldlr−/− mice and their controls (see Supplementary material online, Figure S3B).

3.4 TRAM and TRIF regulate inflammatory gene expression in a stimulus-dependent manner

Having established that the TLR-pathways mediated by TRAM and TRIF are both pro-atherogenic, we next aimed to address the basis for their different activation patterns. The contributing roles between TRIF and TRAM in TLR4 signalling have not been directly compared, thus the dissimilarities could be the results of the adaptors activating different pathways downstream of TLR4. We tested this by injecting LPS in control, Trif−/−, or Tram−/− mice to determine the relative involvement of the two adaptors in TLR4-induced cytokine production. As expected, in the control mice, circulating levels of TNF and IL-12 peaked at 3 h post LPS injection and were declining at 6 h (Figure 5A and B). While these two cytokines could also be detected in plasma of both TRIF and TRAM-deficient mice, the levels were markedly reduced at 3 h and almost absent at 6 h post LPS stimulation. In contrast, secretion of IFNγ was similar in all mice at 3 h whereas both TRIF and TRAM deficiency caused a reduction at 6 h (Figure 5C), indicating that only the late phase of IFNγ secretion depends on the TRIF and TRAM pathways. Both adaptors were able to block the LPS-induced secretion of the chemokines, CCL2, CXCL10, and CCL5 (Figure 5D–F), to a similar extent. Taken together, both TRIF and TRAM were equally able to inhibit LPS-induced responses in vivo in a non-compensatory manner, in contrast to the results obtained in the atherosclerotic model.

Figure 5

TRIF and TRAM both control TLR4-dependent cytokine production in a non-compensatory manner. Tram−/− (squares), Trif −/− (inverted triangles), or control mice (circles) were injected with 50 µg LPS i.p. Blood samples were collected prior to, 3 and 6 h post LPS injection and plasma were analysed for (A) TNF, (B) IL-12, (C) IFNγ, (D) CCL2, (E) CXCL10, (F) CCL5 using CBA and ELISA. Values are represented as mean ± SEM from three to four mice per point.

3.5 TLR3 deficiency in haematopoietic cells protects against atherosclerosis in Ldlr−/− mice

While both adaptors are utilized by TLR4, TRIF is the only adaptor employed by TLR3. Therefore, TRIF deficiency might block both TLR4 and TLR3, while the absence of TRAM would only block TLR4, a possible explanation for the observed differences. Using BMT from Tlr3−/− donors into Ldlr−/− recipients, we observed that haematopoietic deficiency of TLR3 significantly reduced the lesion size in the aortic sinus by 43.9% (Figure 6A and B). The lesions were uniformly reduced in cross sections of the aortic sinus in the TLR3-deficient Ldlr−/− mice (Figure 6C). In addition, aortic transcript levels of the macrophage and T-cell markers CD68 and CD3 were decreased (Figure 6D and E), as well as mRNA for the pro-inflammatory cytokines, TNF and IFNγ (Figure 6F and G). There was no significant change in either weight, total serum cholesterol (see Supplementary material online, Table S2), cholesterol levels in CR/VLDL, LDL, and HDL (see Supplementary material online, Figure S1C), or serum levels of oxLDL (see Supplementary material online, Figure S2C) observed when compared with controls.

Figure 6

Haematopoietically expressed TLR3 is pro-atherogenic in LDLR-deficient mice. Quantitative lesion analysis in aortic root cross sections of control or Tlr3−/−chimeric Ldlr/− mice fed an atherogenic diet evaluated by microscopic morphometry of serial sections. (A) The average lesion size at 200–800 μm from the aortic root is shown, calculated as size covered by lesion (μm2) related to the aortic perimeter (μm). Individual values are displayed by dots, and the mean for each group indicated by horizontal lines. (B) Representative micrographs showing cross sections of Oil Red O-stained lesions at 400 μm (black bar 25 µm). (C) Lesion size throughout the proximal aorta. Relative mRNA levels of (D) CD68, (E) CD3, (F) TNF, and (G) IFNγ in aorta were analysed with real-time PCR and normalized to Hprt. Data are expressed as arbitrary units (AU). *P<0.05, **P < 0.01, and ***P < 0.001.

4. Discussion

Atherosclerotic lesions are characterized by cell death and accumulation of lipids, likely generating a co-ordinated response from several TLRs. This stimulation initiates both common and unique signalling pathways, which decide the pattern of gene expression elicited, and the recruitment and activation of immune cells. It is therefore important to identify the TLR adaptor proteins causing increased inflammation in the plaque.

Our data show that signalling through TRAM and TRIF, but not MAL, adaptors in haematopoietic immune cells are pro-atherogenic in hyperlipidaemic Ldlr−/− mice. This was not due to altered serum cholesterol, or changes in levels of CR/VLDL, LDL, HDL, but associated with reduced vascular inflammation. Bone-marrow chimaeras of Trif−/− and Tram−/− displayed reduced levels of macrophages and CCL2 in the lesions. CCL2 and CCL5, strikingly impaired only in Trif−/− chimeric mice, are both involved in recruitment of macrophages. This suggests that TRIF- and TRAM-dependent signalling in haematopoietic immune cells increases atherosclerosis by promoting recruitment of macrophages to the plaque.

Trif−/− and Tram−/− chimeric mice also displayed decreased aortic expression of pro-inflammatory cytokines. Interestingly, TRAM appeared to regulate a distinct signalling pathway. While TRIF deficiency reduced both TNF and IL-6, mice lacking TRAM only had lower IL-6 levels. Furthermore, Trif−/−chimeric mice had reduced mRNA levels for the T-cell-specific CD3 marker, probably a consequence of the diminished levels CCL5 and CXCL10 in the TRIF-deficient mice. These chemokines are involved in T-cell recruitment and mice deficient in CXCL10 or the CCL5 receptor CCR5 have impaired T-cell accumulation and reduced atherosclerosis.12,13

The Th1 subtype is the dominating T-cell population in the atherosclerotic plaque and can aggravate atherosclerosis in part through secretion of IFNγ. This Th1 cytokine pattern is predominantly induced by IL-12, which also contributes to the development of atherosclerosis. In our study, both IFNγ and IL-12 production was reduced in the Trif−/− mice paralleled by reduced aortic expression of CD86, a critical co-stimulatory molecule in the initiation of adaptive immunity. Mice lacking TRIF in haematopoietic immune cells displayed a more pronounced reduction of atherosclerosis which might be explained by the impaired recruitment and Th1 effector function of T cells in the lesion. In support of this, the TRIF pathway was recently shown to be critical for effector T-cell accumulation into non-lymphoid tissues following an induced immune response.14

In contrast to TRIF and TRAM, chimeric MAL-deficient mice showed no significant inhibition of any parameter measured in this study. Thus, our findings argue against a pro-atherogenic role for MAL signalling in haematopoietic immune cells, although a compensatory effect by MyD88, or the other adaptors, cannot be excluded in the absence of MAL. However, the discrepancy between MAL in comparison with TRIF and TRAM suggests that activation of the transcription factor IRF3, which is induced by TRIF and TRAM, is needed to mediate a more important pro-atherogenic signal. MAL which does not lead to IRF-3 activation would therefore not provide this signal. This would be in line with the recent finding that type I IFNs that are induced by IRF3 promote atherosclerosis by stimulating macrophage recruitment to lesions.15

Our finding that TRIF signalling in haematopoietic immune cells drives this disease is supported by a recent report published during the preparation of this manuscript, showing that a global deficiency of the Lps2 mutation in TRIF was atheroprotective in Ldlr−/− mice. We take this further by showing that a specific deletion of TRIF in bone-marrow immune cells is sufficient to inhibit atherogenesis. In addition, we demonstrate that haematopoetically expressed TRAM has an independent role in the disease process and blocking this protein is enough to inhibit the disease process.

The mechanism behind the differences between TRIF- and TRAM-deficient mice is unclear. TRIF and TRAM are proposed to cooperate in the induction of downstream signalling from TLR4.16 However, since TRAM has not been extensively studied, it cannot be excluded that it has a function separate from TRIF downstream of TLR4, therefore we cannot rule out that an athero-specific TLR4 ligand could selectively signal through only one adaptor. Recent evidence indicates that certain TLR4 ligands can specifically initiate only TRIF-selective signalling,17 and in the context of atherosclerosis stimulation by the two TLR4 ligands, LPS and minimally modified LDL result in differences in cell activation.18 It is unclear which TLR ligands are driving the disease in our study, however, using the prototypic TLR4 ligand LPS, we show that TRIF and TRAM are equally important for the secretion of pro-inflammatory cytokines and chemokines in vivo. This indicates that in the case of LPS they regulate the same pathway downstream of TLR4 and do not compensate for each other.

Alternatively, the differences between the two adaptors might be caused by activation of different TLR receptors in the lesion, where TRIF deletion would affect both TLR4 and TLR3 receptors, while absence of TRAM would only inhibit TLR4. TRIF is the sole adaptor utilized by the TLR3 receptor and, to further investigate this pathway, we analysed the role of TLR3 in experimental atherosclerosis. Similar to mice lacking TRIF, TLR3-deficient chimeric Ldlr−/− mice exhibited reduced atherosclerotic burden. Furthermore, the aortic levels of the macrophage and T-cell markers were inhibited along with the cytokines TNF and IFNγ, indicating a lower level of aortic inflammation. These results match the findings observed in the TRIF-deficient mice and suggest that TRIF mediates its pro-atherogenic role downstream of TLR3 in haematopoietic immune cells.

The TLR3-activating ligands in the atherosclerotic lesion remain to be identified. TLR3 is triggered by stimulation with dsRNA and reported to be an endogenous sensor of tissue necrosis,19 probably through the detection of host-derived RNA from necrotic cells.20 The presence of cell death material in the necrotic core makes this a plausible way for TLR3 activation within the lesion. However, even though there were signs of activity within the lesion in our chimeric Tlr3−/− mice, one cannot exclude a systemic effect.

Our study identifies TLR3 signalling in cells of the haematopoietic compartment as atherogenic. Both macrophages and dendritic cells express high levels of TLR3 and could be activated through stimulation via the receptor. Furthermore, TLR3 is expressed on certain T-cell subsets but cannot alone induce T-cell activation. Instead, TLR3 ligation can mediate co-stimulation with TCR in human γδ T cells21 and to a small extent in αβ T cells.22 Given that γδ T cells play only a minor role in the plaque,23 TLR3-induced γδ T-cell stimulation probably does not impact on the results in this study. However, based on our experimental set up, we cannot rule out a co-stimulatory effect on T cells.

Conflicting data regarding the role of TLR3 in atherosclerosis were recently reported.2426 Whole-body knockout of TLR3 in Apoe−/− mice exhibit increased atherosclerosis at early stages whereas this effect was lost at more advanced stages of the disease.24 In addition, similar results were reported during the preparation of this manuscript in compound mutant mice lacking Ldlr and Tlr3.26 In contrast to these findings and in line with our results, Zimmer et al. show that atherosclerotic plaque development is increased by TLR3 stimulation in Apoe−/− mice, suggesting a pro-atherogenic role for TLR3.25 It is likely that TLR3 exerts different responses depending whether the cell is of haematopoietic origin or not, similar to TLR2.4 This would also be in agreement with our previous report showing that human haematopoietic cells react differently to TLR3 ligands than non-haematopoietic cells.27 In addition, our finding of a pro-atherogenic role for bone-marrow-derived TLR3 is in line with the effects of its adaptor TRIF in haematopoietic cells, and also with several cell culture studies showing pro-atherogenic effects of TLR3 ligation on such cells, including impaired cholesterol efflux, increased inflammation, and enhanced protease secretion.2830

In summary, our results demonstrate that haematopoietic abrogation of the TLR-signalling pathways involving TLR3, TRIF, and TRAM reduces vascular inflammation and inhibits the development of atherosclerosis. These TLR-signalling pathways elicit partly different sets of pro-atherogenic mediators, and can independently inhibit the disease process. The identification of these pro-atherogenic pathways contributes to our understanding of the innate mechanisms at play during atherogenesis and points to TLR signal transduction in haematopoietic immune cells as a possible therapeutic target in atherosclerotic cardiovascular disease.

Funding

This work was supported by grants from the Swedish Research Council (grants 6816 and 8703); the Swedish Heart-Lung Foundation; the European Commission (AtheroRemo collaborative project); the Foundation for Strategic Research; Vinnova Foundation, Stockholm County Council; Konung Gustaf V:s 80-years fond; Loo and Osterman's foundation; O. E. and Edla Johansson's Foundation; Magnus Bergvall's Foundation; Prof. Nanna Svartz Foundation; Gun and Bertil Stohne's Foundation, and KI Foundations for Geriatric Research.

Acknowledgements

We thank Ingrid Törnberg, Anneli Olsson, Inger Bodin, and André Strodthoff for technical assistance and Prof. R. Flavell (Yale University, USA) and Dr Claudia Monaco (Imperial College, UK) for kindly providing the TLR3-deficient mice.

Conflict of interest: none declared.

References

View Abstract