Toll-like receptor 3 signaling evokes a proinflammatory and proliferative phenotype in human vascular smooth muscle cells

Xin Yang, Vanishree Murthy, Kelly Schultz, Jeffrey B. Tatro, Katherine A. Fitzgerald, Debbie Beasley


Inflammation plays a key role in atherogenesis, perhaps promoted by bacterial and viral products present within the artery wall. Vascular smooth muscle cells (VSMC) can express certain bacterially responsive Toll-like receptors (TLR), which promote a proinflammatory and proliferative VSMC phenotype when activated, but it is unknown whether virally activated TLR can regulate VSMC phenotype. Here we tested the role in VSMC of TLR3, which is activated by double-stranded (dsRNA), a molecular signature of viruses. VSMC from multiple vessel types, including human coronary artery (HCoASMC) and mouse aorta (MAoSMC), expressed TLR3 constitutively, and HCoASMC were exquisitely sensitive to dsRNA-stimulated release of monocyte chemoattractant protein-1 (MCP-1) and interleukin-6. dsRNA-induced MCP-1 release was abolished by small interfering RNA-mediated TLR3 knockdown in HCoASMC and was absent in TLR3−/− MAoSMC but was unimpaired in TLR2−/− and in TLR4 signaling-deficient MAoSMC. Exposure to dsRNA also activated ERK1/2 and NF-κB in both human and murine SMC, but these effects were absent in SMC from TLR3-deficient mice, demonstrating a crucial role of TLR3 signaling. dsRNA also stimulated proliferation of HCoASMC, indicated by increased DNA synthesis, and induced persistent elevations in the intracellular levels of growth-promoting mediators, including interleukin-1α and phospho-ERK1/2. We conclude that exposure of HCoASMC to dsRNA elicits dramatic TLR3-mediated proinflammatory and proproliferative phenotypic changes, responses that could potentially be triggered by viral infection of cells within the arterial wall.

  • interleukin-1α
  • monocyte chemoattractant protein-1
  • cell proliferation
  • viruses
  • ribonucleic acid

chronic viral or bacterial infections may be involved in the pathogenesis of atherosclerosis (912, 26, 37, 41, 55). A link between innate immunity signaling pathways, which are activated by microbial pathogens, and atherogenesis has been suggested on the basis of recent findings that deficiency of Toll-like receptor (TLR) signaling decreased the size of vascular lesions in hypercholesterolemic mice and the extent of chemokine expression and macrophage accumulation therein (2, 31). Vascular smooth muscle cells (VSMC) play a pivotal role in atherogenesis, by first migrating into the intima and proliferating, and later producing excessive extracellular matrix proteins. A potential direct link between TLR signaling and these profound changes in VSMC phenotype has been posited on the basis of recent findings that VSMC exhibit a proinflammatory phenotype and proliferate when exposed to certain bacterial products that may be present in the artery wall, including LPS and Chlamydia pneumoniae antigens, and these effects involve the direct engagement of Toll-like receptors expressed by VSMC (39, 51, 52). Whether such dynamic interactions occur in the setting of vascular pathogenesis in vivo is not yet known. During atherogenesis or other forms of vasculopathy, VSMC may be exposed to products of invading and/or resident microbes. Therefore, it is important to understand the repertoire of microbial products capable of altering VSMC phenotype, and the mechanisms involved.

TLRs comprise a family of 10 or more transmembrane receptors that are expressed in a cell type-specific manner and are activated by distinct repertoires of specific, pathogen-associated molecular patterns. Each TLR family member recognizes a particular pathogen-specific molecular signature, the specificity for which is conferred by a leucine-rich extracellular domain, whereas the various TLRs share a conserved intracellular signaling domain. VSMC are known to express TLR2 and TLR4, two key members of the TLR family involved in the recognition of bacteria, and activation of TLR2 and TLR4 by the intracellular bacterium C. pneumoniae promotes VSMC chemokine release and proliferation, respectively (39, 51, 52). Viral DNA and RNA are thought to possess unique characteristics that can activate the related receptors TLR3, TLR7, TLR8, and TLR9 (32). To identify TLR family members having potential functional or pathological relevance in VSMC, here we first characterized the expression profiles of TLR mRNAs in human SMC isolated from anatomically and functionally distinct arteries. We found that the only viral nucleic acid-recognizing TLR family member expressed constitutively in human VSMC is TLR3, a receptor for viral double-stranded RNA (dsRNA) (19). Accordingly, considering the widely postulated pathogenic role of chronic viral infections in atherogenesis (912, 26, 37, 41, 55), we then tested the hypothesis that dsRNA can elicit proinflammatory and proproliferative phenotypic changes in human VSMC by activating a functional, constitutively expressed TLR3. We used two approaches: small interfering RNA (siRNA)-mediated knockdown of TLR3 in human arterial VSMC, and also TLR3-deficient VSMC derived from transgenic TLR3−/− mice, which allowed us to determine whether any TLR3-independent pathways are involved in dsRNA signaling by testing for residual effects of dsRNA.


VSMC culture.

VSMC obtained by enzymatic digestion of human coronary and pulmonary arteries of 17- to 31-yr-old male donors, and the aorta of a 5-yr-old female donor (Clonetics) were cultured in SMGM-2 medium supplemented with 5% FCS, basic fibroblast growth factor (2 ng/ml), epidermal growth factor (0.5 ng/ml), insulin (5 μg/ml), gentamicin (50 μg/ml), and amphotericin-B (50 ng/ml).

VSMC were isolated from aortas of 5- to 6-wk-old male mice of various strains, including wild-type (C57BL/6J and C3H/OuJ), TLR2−/− (45), TLR3−/− (17), and TLR4 signaling deficient [C3H/HeJ, having a dominant-negative mutation within the TLR4 gene (47)], as previously described (51). Animal studies were reviewed and approved by the Tufts-New England Medical Center Institutional Animal Care and Use Committee.

RT-PCR analysis.

VSMC were serum deprived in DMEM supplemented with 1% FCS and washed with PBS. Total VSMC RNA was isolated using an RNeasy kit (Qiagen) and then treated with DNase I. RNA (1 μg) was reverse transcribed for 1 h at 37°C with 200 U of MuMLV reverse transcriptase (GIBCO-BRL), using an oligo(dT) primer and the buffer supplied by the manufacturer, in a total volume of 20 μl. The reaction was terminated at 95°C for 5 min, and 2.5 μl of first-strand cDNA were added to each PCR reaction. PCR mixes contained 50 mM KCl, 10 mM Tris·HCl (pH 8.3), 1.5 mM MgCl2, 250 nM of each primer, and 1 U Taq DNA polymerase (GIBCO-BRL). TLR- or β-actin-specific primers (Table 1) were annealed at 55°C, and samples were amplified for 30 or 20 cycles, respectively. For semiquantitative analysis of TLR3 mRNA knockdown, we confirmed that amplification remained within the exponential range after amplifying each sample for 26, 28, and 30 cycles. cDNA prepared from freshly isolated human monocytes (Spring Bioscience) was used as a positive control for all RT-PCR assays except for TLR7, for which we used cDNA prepared from freshly isolated peripheral blood mononuclear cells.

View this table:
Table 1.

TLR primer sequences

Western blot analysis.

VSMC cultured in growth media were washed with PBS, rapidly frozen in liquid nitrogen, and then thawed on ice and scraped into buffer [50 mM Tris·HCl, pH 7.4, 150 mM NaCl, 1% NP-40, 5% glycerol, 10 mM EDTA, 1 mM sodium orthovanadate, and 1 mM sodium fluoride supplemented with Complete protease inhibitors (Boehringer)], incubated 30 min on ice, and centrifuged at 14,000 rpm for 10 min at 4°C. Cellular proteins (100 μg) were separated on SDS-PAGE gels and transferred to nitrocellulose, and the membrane was incubated sequentially with 5% nonfat dry milk, mouse anti-TLR3 (Clone 40C1285; directed against the amino acid sequence VLNLTHNQLRRLPAAN, corresponding to amino acids 55–70 of human TLR3; underlined sequence is homologous to mouse TLR3; Imgenex, San Diego, CA), followed by peroxidase-conjugated donkey anti-mouse IgG. Peroxidase bound to the blot was visualized by chemiluminescence (SuperSignal West Pico; Pierce Chemical). To validate comparable protein loading, blots were also probed with anti-smooth muscle α-actin antibody (Sigma).


Human aortic SMC (HAoSMC), human coronary artery SMC (HCoASMC), and mouse aortic SMC (MAoSMC) were plated in 48-well plates (10,000/well). Human SMC were serum deprived 72 h (1% FCS) and incubated 6 or 24 h in serum-free media with or without TLR agonist. Mouse SMC were serum deprived (0.25% FCS) 24 h and incubated 6 h in media supplemented with 0.25% FCS. Supernatants from the cells were collected for analysis of monocyte chemoattractant protein-1 (MCP-1by ) using specific sandwich ELISAs for mouse MCP-1 (BD Biosciences), human MCP-1 (R & D Systems), and human IL-6 (Cayman Chemical, Ann Arbor, MI). Human SMC were lysed in buffer for analysis of IL-1α (Cayman Chemical) (1).

To test a potential functional role of TLR3 in VSMC, we used a widely studied mimic of viral dsRNA, polyinosinic-polycytidylic acid [poly(I:C); Amersham]. Macrophage-activating lipoprotein 2 (MALP-2; Alexis Biochemicals) was used to activate TLR2. To activate TLR4, we used repurified Escherichia coli serotype 0111:B4 lipopolysaccharide (Sigma; repurified by phenol extraction and found to have no stimulatory activity on HEK-293 cells stably expressing TLR2) (39, 51, 52). All incubations with LPS also included human recombinant CD14 (100 ng/ml; R&D Systems) to compensate for potentially low CD14 expression in subcultured human coronary artery and mouse aortic SMC (51) and thereby enable efficient delivery of LPS monomers to the TLR4 signaling complex (14).

siRNA transfection.

HCoASMC were plated at subconfluent density and transfected the following day with either TLR3-directed siRNA (sense strand, UUAGAGUUGUCAUCGAAUCdTdT) (35) or nonspecific siRNA (sense strand, CGUACGCGGAAUACUUCGAdTdT) complexed with Lipofectamine 2000 (Invitrogen). The TLR-specific siRNA was selected for its efficacy in preliminary studies (not shown), from a set of six candidate siRNAs tested. After 5 h, RNA/lipid complexes were removed, and SMC incubated overnight in complete media. SMC were then trypsinized and replated in 48-well plates for analysis of MCP-1 release, as described above.

Kinase activation.

Mouse and human SMC were serum deprived 48 h (no FCS), then rapidly frozen in liquid nitrogen, either before or various times after addition of poly(I:C), and whole cell lysates prepared as previously described (51). Cellular protein (30 μg) was separated on SDS-polyacrylamide gels and transferred to nitrocellulose membranes, and the membranes were incubated sequentially with 5% nonfat dry milk, rabbit anti-phospho-ERK1/2 (Thr202/Tyr204), anti-phospho-Akt (Ser473), anti-phospho JNK(Thr183/Tyr185), anti-phospho-p38 (Thr180/Tyr182), or total ERK1/2 antibodies (Cell Signaling Technology), followed by peroxidase-conjugated donkey anti-rabbit IgG. Peroxidase bound to the blot was visualized as described above. In some cases, blots were reprobed with mouse anti-β-tubulin or anti-smooth muscle α-actin, as loading controls.


HCoASMC were serum deprived (0.25% FCS for 24 h) and stimulated with poly(I:C) (1 μg/ml) for various time points, and nuclear extracts were prepared as previously described (38). Nuclear proteins (8 μg) were preincubated in binding buffer containing 32P-labeled consensus oligonucleotide (AGTTGAGGGGACTTTCCCAGG; Promega), with or without a 25-fold excess of unlabeled competitor oligonucleotide, and oligonucleotide-protein complexes were resolved on nondenaturing acrylamide gels (38).

NF-κB reporter assay.

MAoSMC were plated in 12-well plates (50,000 cells/well) and transfected the following day with a reporter plasmid containing five NF-κB sites upstream of the firefly luciferase coding region (pNF-κB-luc, Stratagene). The NF-κB-luc plasmid (0.5 μg) was complexed with Gene Juice transfection reagent (3 μl; Novagen, Madison, WI) and added dropwise to the cells, which were then incubated 6 h. SMC were recovered overnight in fresh DMEM (10% FCS), serum deprived 24 h in DMEM (1% FCS), and then stimulated by incubating for 6 h in DMEM (1% FCS) with or without poly(I:C), MALP-2, or E. coli LPS. Cell lysates were prepared and analyzed using a commercial lysis buffer and assay kit (Luciferase Assay System, Promega) according to the protocol recommended by the manufacturer.

[3H]thymidine incorporation.

As an index of cellular proliferation, we measured the incorporation of [3H]thymidine into newly synthesized DNA. SMC were plated in 48-well plates (5,000/well) and serum deprived for 24 h (DMEM/1% FCS), and media were then replaced with DMEM containing 1% FCS, with or without added poly(I:C) or LPS/CD14. After 24 h, [3H]thymidine (1 μCi/well) was added, and the cells were incubated an additional 24 h, then washed three times in ice-cold PBS. Unincorporated [3H]thymidine was extracted from the cells with 5% ice-cold trichloroacetic acid, and incorporated [3H]thymidine was then extracted from the cellular precipitates into 0.5 N NaOH. Samples were then neutralized by the addition of 0.5 N HCl and added to scintillation fluid, and incorporated [3H]thymidine was determined by scintillation counting.

Statistical analysis.

Significant differences were analyzed by analysis of variance followed by Bonferroni post hoc tests where appropriate, using Prism 4.0 software.


Human arterial SMC express TLR3, TLR4, and TLR6 mRNAs.

To ascertain which TLRs are potentially functional in human arterial SMC, we determined whether mRNA encoding the TLR subtypes 1–10 is expressed constitutively in SMC derived from human coronary artery (HCoASMC), aorta (HAoSMC), or pulmonary artery (HPASMC), using semiquantitative RT-PCR (Fig. 1A). Coronary, aortic, and pulmonary VSMC were all found to express TLR3, TLR4, and TLR6 mRNAs. The respective RT-PCR products were similar in size to those obtained by using RNA from human monocytes, a prototypical TLR-expressing cell type. HCoASMC also expressed TLR1 mRNA and low levels of TLR5 mRNA, but mRNAs encoding TLR7, TLR8, TLR9, or TLR10 were not detected in any of the three human arterial SMC types tested (Fig. 1A; HPASMC and HAoSMC not shown). In contrast, TLR8, TLR9, and TLR10 mRNAs were readily detected in human monocytes, used as positive controls, and mRNA encoding TLR7, which is not expressed in human monocytes, was readily detected in human peripheral blood mononuclear cells that include TLR7-expressing B cells (Fig. 1A).

Fig. 1.

Constitutive expression of Toll-like receptor (TLR) subtype mRNA and TLR3 protein in arterial vascular smooth muscle cells (VSMC). A: messenger RNA encoding TLR3, TLR4,and TLR6 is seen in human SMC derived from coronary artery (CoA), pulmonary artery (PA), or aorta (Ao). Levels of mRNA encoding TLR subtypes 1–10 were analyzed by semiquantitative RT-PCR. TLR RT-PCR product obtained from human monocytes (Mc) or human peripheral blood mononuclear cells (Mn) are shown for comparison. B: immunoreactive TLR3 protein of ∼110–120 kDa is seen in VSMC derived from human arteries, as in A, and in VSMC derived from the aorta of TLR3+/+ (C57BL6/J) but not TLR3−/− mice. Shown are Western blots of whole cell lysates (100 μg protein) with molecular mass (M) marker proteins (120, 100, and 80 kDa).

Human arterial and mouse aortic SMC express TLR3 protein.

Among the TLRs known to be responsive to viral nucleic acids (TLR3, TLR7, TLR8, and TLR9) (32), mRNA encoding only TLR3 was found to be constitutively expressed in human arterial SMC. We therefore used Western blot analysis to determine whether VSMC express TLR3 protein. A TLR3-specific antibody bound to protein of the expected size (∼116 kDa) in whole cell extracts of nonstimulated VSMC derived from several distinct arterial sources, including human coronary artery, human pulmonary artery, human aorta, and wild-type mouse aorta (Fig. 1B), suggesting that VSMC constitutively express TLR3 protein. In contrast, TLR3-immunoreactive bands were not detectable in VSMC derived from TLR3−/− mouse aorta, confirming their identity as TLR3 protein (Fig. 1B).

dsRNA promotes MCP-1 and IL-6 release from HCoASMC.

HCoASMC were found to be exquisitely sensitive to low extracellular concentrations of dsRNA, as exposure to poly(I:C), a synthetic mimic of dsRNA, induced marked, concentration-dependent chemokine and cytokine release at concentrations as low as 0.25 μg/ml. These responses were maximal at 10 μg/ml (Fig. 2). In contrast, single-stranded homopolymers of either inosinic or cytidylic acids failed to stimulate MCP-1 release (not shown), indicating that the dsRNA motif is essential for VSMC activation by poly(I:C).

Fig. 2.

Double-stranded RNA (dsRNA) stimulates release of monocyte chemoattractant protein 1 (MCP-1) and IL-6 from human coronary SMC (HCoASMC). HCoASMC were incubated for 6 h with or without polyinosinic-polycytidylic acid [poly(I:C)] at concentrations shown. Cell supernatants were analyzed for MCP-1 and IL-6 levels by specific ELISA (n = 6 replicates/group). *P < 0.01 vs. VSMC without poly(I:C).

TLR3 mediates dsRNA-induced chemokine release in HCoASMC.

We tested the role of TLR3 in mediating dsRNA-induced chemokine release in human VSMC using RNA interference, an endogenous sequence-specific posttranscriptional gene silencing program that is triggered by the intracellular presence of dsRNA homologous in sequence to the target mRNA (8). Transfection of HCoASMC with TLR3-specific siRNA nearly abolished TLR3 mRNA, whereas transfection with nonspecific control siRNA had no effect, indicating the specificity of gene silencing (Fig. 3A). Transfection of HCoASMC with TLR3-specific siRNA also markedly reduced cellular TLR3 protein levels, but not those of β-tubulin, relative to the levels seen in control siRNA-transfected cells. Furthermore, dsRNA-induced MCP-1 release was typically abolished in HCoASMC transfected with TLR3-specific siRNA (Fig. 3B), and in a total of four experiments, poly(I:C) (5 μg/ml)-induced stimulation of MCP-1 release was inhibited by 76 ± 13%. In contrast, E. coli LPS-induced MCP-1 release, which is mediated by TLR4 in HCoASMC (51), was unaffected (Fig. 3B), indicating the specificity of TLR3 knockdown in abrogating dsRNA-induced MCP-1 release. Together, the results indicate that TLR3-directed siRNA induces a marked, specific knockdown of TLR3 expression and function in HCoASMC and indicate that dsRNA stimulates MCP-1 release by activating TLR3 signaling in human VSMC.

Fig. 3.

Specific knockdown of TLR3 expression and dsRNA-induced MCP-1 release in HCoASMC by TLR3-directed small interfering RNA (siRNA). A: TLR3 mRNA is nearly abolished and TLR3 protein substantially decreased in HCoASMC transfected with TLR3-specific siRNA but are unaffected by nonspecific siRNA. Top: total RNA was prepared 48 h after transfection and analyzed by RT-PCR (representative of 2 experiments). Middle and bottom: Western blot analysis of TLR3 and β-tubulin in whole cell lysates (100 μg protein). B: siRNA-mediated TLR3 knockdown abolishes poly(I:C)-induced, but not LPS-induced, MCP-1 release in HCoASMC. MCP-1 release was analyzed as in A, 48 h after transfection with nonspecific or TLR3-specific siRNA (n = 6 replicates per group, representative of 4 experiments). *P < 0.01 vs. nonspecific siRNA, without poly(I:C) or LPS.

dsRNA-induces MCP-1 release in MAoSMC via TLR3 signaling.

As an alternative approach to testing the hypothesis that TLR3 mediates dsRNA-induced MCP-1 release, we also tested whether selective TLR deficiency abrogates poly(I:C)-induced MCP-1 release in MAoSMC. As in human VSMC, poly(I:C) potently stimulated MCP-1 release from MAoSMC, inducing maximal stimulation at concentrations as low as 0.25 μg/ml (Fig. 4A). Whereas dsRNA-induced MCP-1 release was unimpaired in MAoSMC of TLR2−/− (data not shown) or TLR4 signaling-deficient mice (Fig. 4A), dsRNA failed to stimulate MCP-1 release from aortic SMC derived from TLR3−/− mice (Fig. 4B), indicating an essential role of TLR3 in this response. Furthermore, the abrogation of dsRNA-induced MCP-1 release by TLR3 deficiency was highly specific, because the MCP-1 responses induced by the TLR2 agonist MALP-2 was unaltered in TLR3−/− SMC vs. those in wild-type control SMC (Fig. 4B).

Fig. 4.

dsRNA stimulates MCP-1 release by aortic SMC derived from wild-type (WT) but not TLR3−/− mice. Poly(I:C)-induced MCP-1 release is similar in TLR4 WT and TLR4 signaling-deficient [mutant (Mut)] MAoSMC (A) and in C57BL/6 (BL6) and 129S (C) MAoSMC but is absent in TLR3−/− MAoSMC (B). In contrast, MCP-1 release induced by the TLR2 agonist macrophage-activating lipoprotein 2 (100 ng/ml) was unimpaired by TLR3 deficiency. MAoSMC were incubated 6 h with or without indicated stimuli. MCP-1 levels were determined in cell supernatants (n = 6 replicates/group). In B, results shown are representative of 3 experiments; *P < 0.001 vs. VSMC without TLR agonist.

The TLR3 null allele was generated in embryonic stem cells of 129-strain mice, a strain that is reportedly deficient in an undefined, but TLR3-independent, pathway of dsRNA recognition (16). Therefore, to rule out any potential contribution of the deficiency in this unrelated pathway to the observed impairment of dsRNA responses in TLR3-deficient VSMC, in additional control experiments, we compared dsRNA-induced MCP-1 release in aortic SMC derived from C57BL/6 and 129/Sv mice. dsRNA-induced MCP-1 release was similar in MAoSMC of C57BL/6 and 129/Sv mice (Fig. 4C), indicating that the abrogation of dsRNA responsiveness in TLR3−/− MAoSMC is not a function of the reported deficiency of the putative alternative dsRNA recognition pathway in 129/Sv mice. Furthermore, the results confirm that the impaired dsRNA responsiveness of TLR3−/− MAoSMC is not attributable to other genetic differences between C57BL/6 and 129/Sv mice but is specifically due to TLR3 deficiency.

A critical concern in this type of study is the possible presence of endotoxin (LPS) contamination of reagents, which could spuriously influence the results. In the present study, this concern was ruled out by the finding that poly(I:C)-induced MCP-1 release was similar in wild-type and TLR4 signaling-deficient SMC (Fig. 4A), indicating that the poly(I:C) preparation used was devoid of LPS concentrations sufficient to activate VSMC via the LPS receptor TLR4 (51). In contrast, we consistently found that poly(I:C) obtained from a different commercial source (Sigma Chemical) elicited robust increases in MCP-1 release by wild-type as compared with TLR4 signaling-deficient SMC, indicative of LPS contamination. Confirming this, we also found by limulus amebocyte lysate assay that the latter poly(I:C) preparation contained significant LPS levels (not shown). The results demonstrate the critical importance of excluding reagent contamination by LPS in studies of VSMC activation.

Together, these results validate the use of SMC from transgenic mice to assess the mechanisms involved in dsRNA-induced SMC activation and also provide evidence that VSMC express functional TLR3 that is activated by the dsRNA molecular motif to induce MCP-1 release.

dsRNA induces delayed, TLR3-mediated activation of NF-κB and ERK1/2 in VSMC.

MCP-1 expression is regulated at the transcriptional level by NF-κB, which is strongly activated by TLRs in many cell types. We therefore determined the time course of dsRNA-induced NF-κB activation in HCoASMC, by EMSA analysis (Fig. 5A). NF-κB was detectable within 3–6 h after exposure to dsRNA, was maximal after 24 h, and remained detectable for at least 72 h. We further tested whether dsRNA stimulates NF-κB by acting at TLR3, by comparing the abilities of poly(I:C) to stimulate NF-κB activity in SMC from TLR3−/− and wild-type mice, using a NF-κB-dependent luciferase reporter. Exposure to dsRNA for 6 h increased NF-κB activity more than threefold in wild-type MAoSMC but failed to stimulate NF-κB in TLR3−/− MAoSMC (Fig. 5B), indicating that dsRNA activates NF-κB via TLR3 signaling in MAoSMC.

Fig. 5.

dsRNA activates NF-κB and ERK1/2 in HCoASMC (A and C); role of TLR3 signaling in mouse aortic SMC (B and D) is shown. A: poly(I:C) induces persistent NF-κB activation in HCoASMC. Cells were serum deprived, incubated with poly(I:C) (1 μg/ml) for 0–72 h, and nuclear extracts were analyzed by EMSA. Addition of unlabeled κB oligonucleotide to the 72-h sample abolished binding of the labeled κB probe, demonstrating specificity of the bands (72c). B: WT and TLR3−/− MAoSMC were transfected with NF-κB-luciferase reporter plasmid, serum deprived, and incubated 6 h with or without poly(I:C) (10 μg/ml). Luciferase activity (n = 6 replicates/group) was normalized to the level in control SMC. Results shown are representative of 3 independent experiments. *P < 0.001 vs. control SMC. C: HCoASMC were serum deprived, incubated 1 h with or without poly(I:C) at concentrations shown, and analyzed for phospho-ERK1/2, phospho-Akt, and β-tubulin by Western blot analysis. D: WT and TLR3−/− MAoSMC were serum deprived, incubated 1 h with or without poly(I:C) at concentrations shown, and analyzed for phospho- and total ERK1/2 by Western blotting.

Another positive regulator of MCP-1 gene transcription is ERK1/2, which is activated by TLR3 engagement in some cell types. To determine whether dsRNA activates ERK1/2 in VSMC, we determined phospho-ERK1/2 levels in HCoASMC exposed to various concentrations of dsRNA for 1 h. Exposure to dsRNA (1–100 μg/ml) elicited a concentration-dependent increase in phospho-ERK1/2 levels, whereas phospho-Akt, phospho-JNK, and phospho-p38 levels were unaffected (Fig. 5C and data not shown). Similarly, dsRNA induced a robust concentration-dependent increase in phospho-ERK1/2 levels in aortic SMC from wild-type mice but failed to do so in SMC from TLR3−/− mice (Fig. 5D), indicating a crucial role of TLR3 signaling in this response.

dsRNA stimulates proliferation and persistent expression of IL-1α, a promitogenic cytokine, in HCoASMC.

We have previously found that C. pneumoniae, a microorganism that is frequently present in atherosclerotic lesions, stimulates human VSMC proliferation by a pathway involving TLR4 signaling, suggesting that TLR signaling can promote excessive cellular replication (39). To determine whether TLR3 signaling similarly stimulates VSMC proliferation, we tested the ability of dsRNA to stimulate DNA replication, by measuring [3H]thymidine incorporation, and compared its effects to that of E. coli LPS, the prototypical TLR4 agonist. In preliminary studies, we determined that exposure to poly(I:C) promotes DNA synthesis in HCoASMC incubated in the presence of 1, 2, or 3% FCS (Table 2), suggesting that poly(I:C) promotes FCS-induced proliferation. We then used 1% FCS for subsequent studies, because it yielded the largest fold increase in [3H]thymidine incorporation. DNA synthesis was increased threefold in HCoASMC after 24–48 h exposure to poly(I:C) and more than fourfold when incubated with E. coli LPS in the presence of the LPS transfer factor sCD14 (Fig. 6A). Furthermore, in HCoASMC, dsRNA induced persistent (≥48 h) increases in the levels of cell-associated IL-1α (Fig. 6B), an endogenous cytokine that, we have found, promotes proliferative responses of human VSMC to mitogenic stimuli (1, D. Beasley and K. Schultz, unpublished observations), further suggestive of a proproliferative role of TLR3 activation in human VSMC.

Fig. 6.

dsRNA stimulates proliferation and induces persistent increases in IL-1α and phospho-ERK1/2 levels in HCoASMC. A: HCoASMC were incubated 48 h with or without poly(I:C) at concentrations shown or with LPS/CD14 (each 100 ng/ml) (LPS) to activate TLR4. [3H]thymidine was added at 24 h, and its incorporation into newly synthesized DNA was analyzed at 48 h. *P < 0.001 vs. without TLR agonist. B: HCoASMC were incubated with poly(I:C) (1 μg/ml) for times shown, and IL-1α was analyzed in whole cell lysates by ELISA. C: HCoASMC and MAoSMC were incubated with poly(I:C) (10 μg/ml) for times shown. Shown are Western blots of phospho-ERK1/2 and smooth muscle α-actin in whole cell lysates, representative of 3 experiments, each with HCoASMC and MAoSMC.

View this table:
Table 2.

Effect of serum on poly(I:C)-induced [3H]thymidine incorporation

In contrast, poly(I:C) failed to stimulate proliferation in mouse aortic SMC at any of the concentrations tested (1–100 μg/ml). Consistent with this, it also failed to stimulate IL-1α expression as determined by highly sensitive real-time RT-PCR (not shown). Together, the results indicate that the proproliferative and IL-1α-stimulatory effects of dsRNA on VSMC are species-specific actions, occurring in human VSMC in particular. Alternatively, the retention of such properties after in vitro subculture may be species specific.

DsRNA induces biphasic increases in phospho-ERK1/2 in HCoASMC but only transient increases in MAoSMC.

To identify potential differences between dsRNA-induced signaling in murine and human VSMC, we compared the time courses of dsRNA-induced ERK1/2 activation in MAoSMC and HCoASMC. Poly(I:C)-induced ERK1/2 activation was delayed in both cell types: phospho-ERK1/2 was increased 1 h after exposure to poly(I:C) (Fig. 6C) but was not detectably increased after only 30 min (not shown). In HCoASMC, phospho-ERK1/2 levels then decreased at 2–4 h but then increased again by 6 h and remained elevated after 24 h, indicating a biphasic activation. In MAoSMC, phospho-ERK1/2 levels were also elevated 1–2 h after exposure to poly(I:C); however, in contrast to HCoASMC, the response was transient, and phospho-ERK1/2 levels returned to baseline 4 h after poly(I:C) exposure (Fig. 6C).


The results show that TLR3 is constitutively and prominently expressed in human arterial SMC and, furthermore, that it is the only TLR subtype expressed among those thought to be involved in the recognition of viral nucleic acids (32). Its presence in VSMC derived from anatomically and functionally diverse vessels, including aorta, coronary, and pulmonary arteries, suggests that its expression is an inherent property of arterial VSMC and further suggests a potential role of VSMC TLR3 in arterial inflammation. This notion is further reinforced by our recent findings that arterial SMC of both humans and mice acquire proinflammatory and proliferative phenotypes in response to relevant pathogenic stimuli, by pathways involving activation of other TLR subtypes (39, 51, 52).

The TLR3 agonist and viral dsRNA mimic poly(I:C) potently induced proinflammatory responses in human coronary artery and mouse aortic VSMC, including MCP-1 release and activation of NF-κB, a transcriptional activator of many proinflammatory genes. A crucial role of TLR3 in these responses was demonstrated by the finding that dsRNA-induced MCP-1 release was abolished by siRNA-mediated knockdown of TLR3 expression in human VSMC, whereas TLR4-induced MCP-1 release was unaffected. We also found that TLR3 similarly mediates dsRNA-induced MCP-1 release, NF-κB, and ERK1/2 activation in murine VSMC, because these responses were abrogated in aortic SMC from TLR3-deficient mice (17), whereas MCP-1 release elicited by a TLR2 agonist was unimpaired.

Arterial SMC were found to be exquisitely sensitive to dsRNA, because the threshold poly(I:C) concentrations for stimulation of MCP-1 release in both human and murine SMC (0.25 μg/ml; with maximal effects 1–10 μg/ml) were lower than those reported for activation of interferon-β and chemokine production in other cell types. By comparison, the effective concentration range reported for poly(I:C)-induced interferon-β release is 5–100 μg/ml in human endothelial cells (46), human dendritic cells (28), and mouse peritoneal macrophages and fibroblasts (49). Although TLR3 is not expressed ubiquitously, it is expressed in a number of cell types other than VSMC, including dendritic cells (33), fibroblasts (29), endothelial cells (46), and epithelial cells (5). Therefore, TLR3 signaling in VSMC may complement that in endothelial or other local cell types and thereby promote inflammatory responses within the arterial wall. Hence it will be critical to determine not only the roles of TLR3 but also its distinct functional contributions within these dynamically interacting cell types, in the context of vascular inflammation and pathogenesis in vivo, especially in models of viral infection.

Our findings suggest that persistent activation of ERK1/2 may be critical to the proproliferative effect of TLR3 activation. In HCoASMC, dsRNA promoted proliferation while inducing an initial transient activation of ERK1/2, followed by a persistent phase of activation. In contrast, in MAoSMC, dsRNA induced ERK1/2 activation only transiently and failed to induce proliferation. Because prolonged ERK1/2 activation is thought to be crucial to SMC proliferation (34), it seems likely that the persistent phase of ERK1/2 activation contributes to the proproliferative effect of dsRNA in HCoASMC. Similarly, we found earlier that stimulation of the type I IL-1 receptor, also a TLR family member, induces persistent (>72 h) activation of NF-κB in human VSMC and that NF-κB activation is likewise crucial to the proliferative response (38). Because activation of ERK1/2 is required for persistent NF-κB activation in rat aortic VSMC (20, 21), persistent ERK1/2 activation may contribute to dsRNA-induced proliferation in part by promoting persistent NF-κB activation. Nuclear localization of ERK1/2, and its phosphorylation of nuclear substrates, are also likely to be crucial contributors to the proproliferative effects of ERK1/2 activation (3). An interesting question that remains for future research is whether TLR3 activation directly and independently stimulates VSMC proliferation or alternatively enhances responsiveness to growth factors, including those present in the proliferation assay media (1% FCS), as well as VSMC-derived autocrine growth factors.

The expression in VSMC of dsRNA-activated TLR3 may be relevant to atherogenesis, because it has been widely postulated that chronic infection by viruses, including herpes simplex virus types 1 and 2 and cytomegalovirus (CMV), promote vascular pathogenesis (9, 26). Herpes viruses are single-stranded DNA viruses that may form dsRNA during viral replication (19). CMV is a prevalent virus that has been implicated in atherogenesis and its associated intimal hyperplasia. Prior exposure to CMV (41, 55) or to multiple Herpesviridae (37) is associated with increased risk of cardiovascular-related death, and prior infection with CMV may also be a significant risk factor for restenosis after coronary atherectomy (54). Transplant-associated atherosclerosis, a condition characterized by rapid, diffuse intimal proliferation in arteries and arterioles within the graft, is more severe in recipients who develop CMV infection (11). CMV antigens were found in SMC cultured from atherosclerotic lesions (30), and CMV mRNAs have been isolated from lesions, suggesting the presence of latent virus (12). CMV can infect and replicate within human arterial SMC in vitro (43), and systemic infection of mice with murine CMV increases lesion area and intra-aortic expression of MCP-1 and atherogenic genes (4) and also exacerbates vascular injury-induced intimal hyperplasia (24). It is unknown whether the proatherogenic effects of CMV infection are due to intravascular infection, but our findings of dsRNA-activated and TLR3-mediated proinflammatory and proliferative responses in VSMC support the possibility that VSMC may respond directly to CMV products known to be present in vascular lesions (12, 30).

In addition to the potential maladaptive role of TLR3 in mediating virus-induced vascular pathogenesis postulated here, a beneficial role in antiviral innate immunity has also been postulated (40). TLR3 was found to mediate antiviral responses elicited by rhinovirus (13), and TLR3-deficient mice were found to have increased susceptibility to CMV infection (44), although the latter finding was challenged by a report that the susceptibility of TLR3-deficient mice to infection by lower doses of CMV and by other viral infections is similar to that of wild types (7). Nevertheless, an antiviral role of TLR3 was supported by findings that mice deficient in the gene encoding Trif, a downstream adapter molecule that is recruited to the intracellular domain of TLR3 and crucial to its intracellular signaling, exhibited increased susceptibility to murine CMV infection, associated with increased lethality and higher splenic viral titers (15). TLR3 also mediates inflammatory responses to West Nile virus (48) and murine encephalomyelitis virus (42) and plays a role in proinflammatory cytokine and chemokine expression elicited by respiratory syncytial virus (36), suggesting an additional role of TLR3 in inflammatory responses.

Other cellular dsRNA recognition molecules known to exist, in addition to TLR3, are cytoplasmic enzymes rather than transmembrane receptors. These include dsRNA-activated protein kinase (PKR), RNaseL, and the RNA helicases RIG-1 (retinoic-acid-inducible protein I) and MDA5 (melanoma-differentiation-associated gene 5). In contrast with TLR3, which we found to be very efficiently activated by extracellular dsRNA, activation of these dsRNA-binding enzymes requires intracellular delivery of poly(I:C), either by transfection or by intracellular viral infection (18, 23, 53), and thus would not be expected to mediate the presently observed effects of low extracellular concentrations of dsRNA. Also, dsRNA-induced proliferation and MCP-1 release by VSMC are qualitatively different from the cellular responses elicited by PKR or RNAseL, which inhibit protein translation (18), and are thereby expected to inhibit, rather than stimulate, cellular proliferation and protein synthesis. RIG-1 and MDA5 do appear to be critical mediators of the antiviral responses elicited by several single-stranded RNA viruses, and presumably they interact directly with viral dsRNA that forms within the cytoplasm (23). However, the role of RIG-1 and MDA5 in virus-induced expression of proinflammatory cytokines and chemokines is unknown. We did not test whether RIG-1 or MDA5 is expressed by VSMC, but it is clear that neither of these helicases contributed significantly to the observed proinflammatory responses of VSMC to dsRNA, because siRNA-mediated knockdown of TLR3 abolished dsRNA-induced MCP-1 release in HCoASMC, and dsRNA failed to stimulate MCP-1 release or activate ERK1/2 or NF-κB in TLR3-deficient MAoSMC. We hypothesize that the role of TLR3 in responses of VSMC to viral infection may be most prominent during late stages of infection, when cellular lysis releases cytoplasmic viral RNA intermediates into the extracellular milieu, thereby activating neighboring cells (25, 27).

A major functional role of TLR3 is thought to be cellular activation by extracellular viral dsRNA, but TLR3 may also be activated when necrotic cells release host-derived RNA, which contains considerable secondary structure including dsRNA domains. This idea is based on the finding that in vitro-transcribed mRNA and RNA associated with necrotic cells activated dendritic cells via TLR3 (22). Because cellular necrosis is a common feature of advanced atherosclerotic lesions, this raises the possibility that VSMC TLR3 signaling could potentially exacerbate proinflammatory events in advanced lesions, when activated by RNA released by necrotic cells within the lesion. Hence TLR3 expressed in VSMC could conceivably be involved in vascular pathogenic responses triggered either by infection or by endogenous products arising from ongoing vasculopathies.

In addition to TLR3, which was found here to be expressed in VSMC and to elicit MCP-1 release, IL-6 release, and cell proliferation, VSMC also express bacterial LPS-activated TLR4, which elicits a similar repertoire of cellular responses (51), and TLR2, which is activated by a CMV envelope protein (6) as well as by C. pneumoniae, an intracellular bacterium that is often present within atherosclerotic lesions and thought to play a pathogenic role (50). In immune cells, activated TLRs are thought to elicit adaptive responses that reduce the replication and viability of invading organisms. In contrast, microbial activation of TLRs expressed in nonimmune vascular cells such as VSMC, perhaps complemented by those in endothelial cells, may play a pathogenic role in vascular disease progression. Therefore, the expression of microbially activated TLR2, TLR3, and TLR4 within VSMC, and the inflammatory and proliferative sequelae of their activation, represent novel molecular pathways that may provide a mechanistic basis for clinical findings that the number of infectious pathogens to which an individual has been exposed is positively related to carotid and coronary atherosclerosis progression (10, 37). Accordingly, a critical objective of future research will be to determine whether TLR3 activation in VSMC by viral dsRNA or, alternatively, by endogenous dsRNA released by necrotic intralesional cells contributes to the progression of vascular disease.


This work was supported by National Heart, Lung, and Blood Institute Grant HL-47569 (to D. Beasley).


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract