Regulation of Endotoxin-Induced Proinflammatory Activation in Human Coronary Artery Cells: Expression of Functional Membrane-Bound CD14 by Human Coronary Artery Smooth Muscle Cells¹

Atherosclerosis is increasingly recognized as a chronic inflammatory disorder (1, 2, 3), but the sources of vascular inflammation remain to be elucidated. One potentially important source of vascular inflammation is endotoxin (LPS), a unique glycolipid contained in the outer leaflet of the outer wall of Gram-negative bacteria (4, 5). Gram-negative organisms colonize the human gastrointestinal, genitourinary, and respiratory tracts, and generate endotoxin not only during overt infections, but also in subclinical conditions (e.g., periodontitis) that are common in patients with atherosclerotic disease (6).

Endotoxin can be detected in the plasma of apparently healthy individuals who are free of clinical infection. Epidemiological studies indicate that endotoxemia at levels as low as 50 pg/ml represents a strong risk factor for the development of atherosclerosis (7, 8). A variety of Gram-negative infections were associated with an increased risk of atherosclerosis (8), supporting the hypothesis that endotoxin may be pathogenically linked to the development of atherosclerosis.

Although endotoxin has been shown to evoke proinflammatory responses in cultured vascular cells (9), most prior studies were conducted in bovine aortic endothelial cells (EC)³ or HUVECs using very high (μg/ml) concentrations of endotoxin. Concentrations of endotoxin in this range are even higher than those observed in sepsis and may activate host cells by mechanisms independent of the CD14/TLR4-signaling pathway (10, 11, 12, 13). Also, very few studies have examined endotoxin responses in coronary artery cells, despite that fact that this vascular bed exhibits enhanced susceptibility to atherosclerosis in many patients. Thus, prior studies offer little insight into how low-level endotoxemia might contribute to atherosclerosis.

We have recently shown that human blood vessel explants are extremely sensitive to very low (picograms per milliliter) levels of endotoxin (14). The responses to endotoxin in intact blood vessels were far more potent than have been reported in cultured human EC. Because most of the vessel wall is composed of smooth muscle cells (SMC), these findings suggested to us that vascular SMC might be highly responsive to endotoxin, and, therefore, might play a previously unappreciated role in transducing immune responses in the vascular wall. Therefore, we initiated a series of experiments to compare the relative responses of human coronary artery (HCA)EC and HCASMC to very low levels of endotoxin, and to elucidate the mechanisms that regulate endotoxin responsiveness in vascular cells. In these studies, we examined the expression of membrane-bound CD14 (mCD14) and TLR4 in HCA cells. In addition, we investigated the modulatory effects of LPS-binding protein (LBP) and soluble CD14 (sCD14) on vascular cell activation by endotoxin. Our findings provide important insights into the mechanisms by which circulating endotoxin could contribute to atherosclerosis in humans, and by which vascular SMC may contribute importantly to innate immunity in the vasculature.

HCAEC and HCASMC were isolated from coronary arteries obtained at the time of heart transplantation at the University of Iowa using previously described methods (15, 16). Some HCAEC and HCASMC were purchased from Clonetics (San Diego, CA) and Cell Applications (San Diego, CA) and were maintained in the supplier’s medium. All cells were subcultured at a 1:3 ratio or less, and were used for experiments from passages 7 to 9. Chinese hamster ovary (CHO) cells stably transfected with human CD14 (CHO/CD14) or the vector pKoNeo (CHO/Neo) were a generous gift of Dr. D. Golenbock (17). To maintain selective pressure, CHO/CD14 and CHO/Neo cells were maintained in M199 with 5% FBS with 10 μg/ml ciprofloxacin and 400 μg/ml G418; however, both ciprofloxacin and G418 were withdrawn from cells plated for experiments. Preprepared medium and FBS used in these studies contained <3 pg/ml and <7 pg/ml endotoxin, respectively.

Cell viability was assessed by 3-(4,5-dimethalthiazol-2-yl)-5-(3-carboxymethyloxphenyl)-2-(4-sulfenyl)-2H-tetrazolium, inner salt (MTS) assay using a test kit (Promega, Madison, WI) according to the manufacturer’s instructions (18).

Cytokines were measured by ELISA, using matched Abs from R&D Systems (Minneapolis, MN), as described previously (19).

Expression of TLR4 in HCAEC and HCASMC was analyzed by RT-PCR, Western blot analysis, and immunohistochemistry, as described previously (14). Expression of CD14 (394 bp) in HCAEC, HCASMC, and U937 cells (as a positive control) was analyzed by RT-PCR, as described for TLR4 (14). The oligonucleotide primers used were: 5′-ACTTATCGACCATGGAGC and 5′-AGGCATGGTGCCGGTTA.

Cells were grown to confluence in 48-well plates; assays were performed on cells in their original plates. Each well was gently rinsed twice, then incubated for 1 h at 37°C with mouse anti-human CD14 mAb (MEM-18; Accurate, Westbury, NY) or UCH-M1 or SCA-1 (Santa Cruz Biotechnology, Santa Cruz, CA) at a dilution range from 1/8,000 to 1/512,000. For each well treated with primary Ab, there was a control well treated with buffer lacking primary Ab. The primary Ab was then removed; the plates were washed three times and then incubated 1 h at 37°C with biotinylated anti-mouse IgG (A85-1; BD Pharmingen, San Diego, CA) at a 1/2000 dilution. Plates were washed three times and then incubated with HRP-streptavidin (Pierce, Rockford, IL; 1/2000) for 30 min at 37°C. The plates were then rinsed again and incubated with 200 μl of TMB solution (Sigma-Aldrich, St. Louis, MO) for 5 min. The color reaction was halted with the addition of 0.5 M H₂SO₄. The contents of each well were then transferred to a 96-well plate and read at 450 nm on a Thermomax microplate reader (Molecular Devices, Sunnyvale, CA). Data for each experiment were averaged from triplicate wells exposed to primary Ab, and corrected for background by subtracting the average absorbance values from the control wells lacking primary Ab.

HCASMC monolayers were preincubated for 2 h with 0.2 U/ml PLC-γ (Sigma-Aldrich) in serum-free M-199 with 0.1% human serum albumin (HSA). Control HCASMC were preincubated with vehicle alone. Following this preincubation, the HCASMC were washed twice with HBSS containing 1 μM HSA to remove the cleaved CD14, along with other GPI-anchored proteins. The cells were then incubated for 24 h with 3 ng/ml LPS + 100 ng/ml LBP, and the medium was assayed for IL-8 release as previously described. To exclude the possibility of nonspecific cell damage, exogenous sCD14 (250 ng/ml) was added to some groups of cells.

LPS from Escherichia coli K12 LCD25 was purchased from List Biological Laboratories (Campbell, CA) and purified by phenol reextraction (20). Human recombinant LBP and sCD14 were a generous gift from Amgen (Thousand Oaks, CA). All other reagents, including lipoteichoic acid (LTA) from Streptococcus pyogenes, were purchased from Sigma-Aldrich.

Results are expressed as mean ± SEM. The data were analyzed by ANOVA followed by Bonferroni t testing.

In initial studies, we compared responses to clinically relevant concentrations of endotoxin on HCAEC and HCASMC. Experiments were routinely performed using E. coli K12 LCD25 endotoxin. Some experiments were also performed with Neisseria meningitidis endotoxin and E. coli K12 lipid A, with qualitatively similar results (data not shown).

Endotoxin has been reported to cause direct toxic effects and increased permeability in EC. In our model system, LPS at concentrations up to 10 ng/ml produced no cytotoxicity in HCAEC or HCASMC, as measured by a standard assay of cell viability (data not shown). Further, when HCAEC were grown as polarized monolayers on micropore filters, 10 ng/ml endotoxin caused no changes in electrical resistance of the monolayer through at least 6 h (data not shown).

Fig. 1 shows the effect of endotoxin on release of IL-8 and MCP-1, two important proinflammatory cytokines that are known to play a crucial role in recruitment of monocytes, neutrophils, and T lymphocytes to atherosclerotic lesions. Fig. 1 A shows that endotoxin produced potent dose-dependent increases in release of IL-8 and MCP-1 from HCAEC. Increases in cytokines were observed only after >4 h incubation with endotoxin, consistent with a transcriptional mechanism of induction, rather than release of preformed protein. As compared with similar studies in HUVEC, the HCAEC exhibited greater endotoxin responsiveness in regard to both the magnitude of IL-8 release (typically 6–8 ng/well, compared with 2–3 ng/well for HUVEC) and the fold increase over baseline (average 4-fold for HUVEC, 7-fold for HCAEC; data not shown).

Fig. 1 B shows that endotoxin produced extremely potent dose-dependent increases in release of IL-8 and MCP-1 from HCASMC. As compared with data obtained in HCAEC, HCASMC showed much greater endotoxin responsiveness with respect to release of both cytokines, both in terms of the absolute amounts of cytokine produced and in threshold sensitivity to very low LPS concentrations (10–30 pg/ml). Although fresh coronary arteries are difficult to obtain, one experiment was also performed in segments of nonatherosclerotic coronary arteries obtained at the time of cardiac transplantation. In this tissue, exposure to 10 ng/ml endotoxin resulted in an increase of ≅26-fold in IL-8 release and 14-fold in MCP-1 release as compared with vehicle (data not shown). These findings are consistent with those recently reported in human saphenous vein explains (14). Further, the magnitude of cytokine release from these tissues is consistent with potent activation of the SMC, the most abundant cell type in the vessel wall. These findings with both IL-8 and MCP-1 indicate that endotoxin is a potent proinflammatory mediator in the human coronary vasculature, consistent with its recent identification as a putative proatherosclerotic risk factor in clinical studies. Moreover, the data suggest that as compared with coronary artery EC, coronary artery SMC exhibit markedly enhanced responsiveness to endotoxin.

TLR4 has been proposed as a critical transmembrane component of the endotoxin receptor complex in a number of cell types, including human dermal microvascular EC (21). Because of the striking difference in LPS responsiveness of HCASMC compared with HCAEC, we looked for possible differential expression of TLR4 in coronary artery cells. Fig. 2 shows the expression of TLR4 in both HCAEC and HCASMC, as detected by both RT-PCR (A) and Western blot analysis (B). Similar findings were obtained by both RT-PCR and Western blot analysis for TLR2 (data not shown). Fig. 2 C shows the results of immunohistochemical analysis of a frozen section of a HCA stained for TLR4. The brown staining is consistent with strong TLR4 expression in the coronary endothelium and in the medial smooth muscle. TLR4 staining in the neointima of this atherosclerotic vessel appeared to be considerably reduced as compared with the endothelial and medial layers. Although none of these techniques is strictly quantitative, all three are consistent in suggesting that TLR4 is expressed by HCAEC at least as strongly as in HCASMC. Thus, these findings suggest that differential expression of TLR4 is unlikely to account for differences in endotoxin responsiveness between HCAEC and HCASMC.

In addition to TLR4, CD14 is involved in endotoxin signaling in inflammatory cells. The level of mCD14 can be an important determinant of endotoxin-induced cellular activation (10). Further, a CD14 promoter polymorphism has been linked to increased risk of myocardial infarction in individuals with otherwise low coronary risk (22). Considering the potent endotoxin responsiveness of HCASMC, we asked whether HCASMC might express mCD14. As shown in Fig. 3 A, mRNA for CD14 was highly expressed in HCASMC, at a level similar to that detected in U-937 cells.

To test for the expression of cell surface mCD14 in our cultured cells, we developed a cell surface ELISA (see details under Materials and Methods). To validate this assay, we first compared CHO cells stably transfected with CD14 with those cells transfected with an empty vector. The CHO/CD14 cells exhibited increasing absorbance with increasing concentrations of any of the three primary Abs tested (Fig. 3,B), with detectable Ab binding at dilutions of primary Ab up to 1/256,000. In contrast, control CHO cells transfected with empty expression vector showed no detectable binding of the MEM-18 Ab. When similar experiments were performed with HCAEC and HCASMC (Fig. 3 C), we found that the HCASMC showed high levels of specific Ab binding for all three anti-CD14 mAbs tested. However, parallel experiments on HCAEC show no evidence of specific Ab binding.

We then asked whether the mCD14 detected by the cell surface ELISA was functionally active in the HCASMC. We incubated HCAEC and HCASMC with increasing concentrations of LPS (0–10 ng/ml) in serum-free medium (SFM) containing albumin and LBP ± sCD14. Although expression of low levels of mRNA for mCD14 was detected in HCAEC (Fig. 3,A), LPS failed to consistently induce significant IL-8 release from HCAEC in the absence of exogenous sCD14 (Fig. 4,A). However, IL-8 release from HCASMC was similar with or without added sCD14 (Fig. 4 B). Similar sCD14-independent responses were observed in six other lines of passaged HCASMC (two lines purchased from commercial suppliers and four isolated in our laboratory) (data not shown).

To confirm a role for mCD14 in transducing endotoxin responses in HCASMC, cells were treated with MEM-18, an Ab that recognizes the CD14 epitope to which LPS binds. MEM-18 markedly reduced LPS-induced IL-8 release in the absence of sCD14, further demonstrating the presence of CD14 in HCASMC (Fig. 4,B). In contrast, the Ab had no effect on TNF-α-induced IL-8 release, confirming its specificity (data not shown). Moreover, treatment of HCASMC with PLC-γ to cleave GPI-anchored proteins reduced the LPS-induced IL-8 release by ∼85%; the reduction was completely reversed by the addition of exogenous sCD14 (Fig. 4 C). Taken together, these results suggest that LPS-induced IL-8 release by HCASMC is modulated by membrane-bound GPI-linked CD14 present on the surface of the HCASMC.

In inflammatory cells, CD14 also mediates responses to LTA, a product of Gram-positive bacteria that has recently been implicated in atherosclerosis (23, 24, 25). Therefore, we examined the effect of LTA on cytokine release in HCAEC and HCASMC. As was observed with endotoxin, LTA produced dose-dependent increases in IL-8 release in HCAEC only in the presence of sCD14 (Fig. 5,A). In contrast, LTA produced dose-dependent increases in IL-8 release in HCASMC, and these responses did not require the addition of sCD14 (Fig. 5 B).

Endotoxin signaling in EC is modulated by interaction with two serum proteins, LBP and sCD14. In healthy patients and those with low-grade infections, the molar ratio of LBP:sCD14 in serum is most commonly ≤1. Patients with chronic low-grade infections (e.g., periodontitis) appear to have chronically elevated levels of sCD14 (26), and increases in sCD14 are correlated with increased mortality in bacteremia (27). However, during the acute inflammatory response, while sCD14 levels increase slightly (2- to 3-fold), LBP levels increase by 10- to 30-fold (28). These high serum concentrations of LBP are believed to attenuate endotoxin signaling and limit the acute response, and have been shown to reduce mortality in experimental sepsis (29).

To determine whether alterations in the LBP:sCD14 ratio modulate the proinflammatory effects of low levels of endotoxin in coronary artery cells, cells were incubated in SFM containing 0.1% HSA, 60 ng/ml sCD14, 10 ng/ml LPS, and increasing concentrations of LBP (0–6 μg). Because of the similarity of the molecular masses of the two proteins (60 kDa for LBP, 55 kDa for sCD14), the ratio of the masses provides an approximation of the molar ratio. Fig. 6 shows the combined results from multiple experiments with HCAEC (not all LBP concentrations were tested in each experiment). The greatest IL-8 release was seen at low LBP:sCD14 ratios (∼0.1:1). Even a 1:1 molar ratio was somewhat less effective in promoting IL-8 release than the lowest ratios, whereas higher ratios (up to 10:1 in some experiments) were clearly inhibitory. A similar pattern was seen when the same HCAEC samples were assayed for MCP-1 levels (data not shown). A parallel series of experiments with HCASMC gave similar results (data not shown). These results indicate that the proinflammatory effects of endotoxin in HCAEC and HCASMC are maximal at the low LBP:sCD14 ratios characteristically seen in healthy patients or those with the chronic low-grade infections putatively linked to atherosclerosis.

High concentrations of endotoxin are known to evoke proinflammatory activation and membrane injury in cultured EC, which is pertinent to acute vascular dysfunction in septicemia (9). However, little attention has been paid to how the potential proinflammatory effects of low-level endotoxin could contribute to atherosclerosis. In this report, we present several novel findings regarding the proatherogenic effects of endotoxin in the vasculature. First, we show that HCASMC are profoundly sensitive to inflammatory activation by very low doses of LPS, which is likely due in part to functional expression of mCD14 by these cells. Next, we demonstrate by RT-PCR, Western blot analysis, and immunohistochemistry that both HCAEC and HCASMC express TLR4, the putative endotoxin-signaling receptor. Finally, we report that proinflammatory effects of low-level endotoxin are maximal at very low LBP:sCD14 ratios (i.e., <1:1). Because ratios in this range are typically seen in healthy individuals and in those with subclinical infections, these findings strongly suggest that the low levels of endotoxin observed in many patients may contribute to vascular inflammatory activation in atherosclerosis.

We demonstrate first that in HCAEC and especially in HCASMC, low levels of endotoxin potently induce release of the chemotactic cytokines IL-8 and MCP-1. These proinflammatory effects are thought to be critical to induction and progression of atherosclerosis in humans (30). IL-8 is chemotactic for neutrophils and activates NADPH oxidase in these cells, resulting in a local increase in reactive oxygen species (31). IL-8 also induces monocyte chemotaxis and converts monocyte rolling to firm adhesion on endothelial monolayers (32). MCP-1 has also been proposed to play a prominent role in atherosclerosis (30, 33), being highly expressed in human atherosclerotic plaques and likely playing a crucial role in monocyte recruitment into subendothelial lesions (33). In our studies, HCAEC were considerably more responsive than HUVEC in regard to cytokine release induced by endotoxin. In addition, we found that coronary artery SMC are remarkably sensitive to activation by endotoxin. Not only do the HCASMC release far greater amounts of inflammatory cytokines at any given LPS concentration than do HCAEC, but their threshold response occurs at ∼10–30 pg/ml LPS, as compared with ∼300–1000 pg/ml for HCAEC. These findings suggest that human vascular cells vary significantly in their responsiveness to endotoxin, which could have important implications in regard to the cellular sources of inflammation in atherosclerosis.

TLRs, a family of pattern recognition receptors involved in innate immune recognition of pathogen-associated molecules (e.g., endotoxin), are transmembrane proteins containing extracellular domains rich in leucine-repeat motifs and a cytosolic domain homologous to the IL-1R-signaling domain (34, 35). Among the TLR family, TLR4 is believed to be the specific endotoxin receptor (36, 37). TLR4 polymorphisms have been shown to play a role in atherogenesis, with the Asp²⁹⁹Gly mutation, which attenuates LPS signaling, associated with a decreased risk of atherosclerosis (38). TLR4 has been widely studied in monocytes and macrophages and was recently reported in human saphenous vein EC and SMC and tissues (14), human aortic EC (39), human microvascular EC (21, 39), coronary artery EC, (40), and saphenous vein SMC (41). Our finding of the presence of TLR4 in HCAEC and HCASMC (Fig. 2) is consistent with the potent endotoxin-induced inflammatory responses observed in the present study. However, differences in TLR4 expression in HCAEC vs HCASMC do not appear to account for differences in endotoxin responsiveness between these two cell types.

Monocytes and neutrophils respond to endotoxin through its interaction with mCD14, a 55-kDa GPI-anchored protein (42). Although it is generally believed that EC lack mCD14 and respond to endotoxin primarily through sCD14 circulating in blood, a recent report indicates that primary HUVEC express low levels of mCD14 (43). Also, Walton et al. recently reported the presence of mCD14 in human aortic EC, as detected by RT-PCR and a cell surface ELISA similar to ours (39). However, they did not demonstrate that the mCD14 conferred responsiveness to endotoxin in their aortic EC. It is conceivable that the lack of functional mCD14 expression in our HCAEC was because the cells were used at passage numbers 7–9. Because of the difficulty in procuring HCAEC, we were unable to perform experiments in primary cultured or early passage cells. However, our HCASMC lines, each of which expressed functional mCD14, were also used at passage numbers 7–9.

With respect to mCD14 expression in SMC, in the present study we found that seven different lines of HCASMC expressed mCD14 (Figs. 3 and 4), suggesting a possible mechanism for the striking endotoxin responsiveness in these cells. Loppnow et al. failed to find evidence of mCD14 expression in SMC isolated from human saphenous veins using RT-PCR, Northern analysis, cell-sorting analysis and immunoassay experiments (44), although these cells responded to 1–1000 ng/ml LPS in the presence of serum. It is uncertain whether these contrasting results are related to the vascular source of the SMC (saphenous vein vs coronary artery) or perhaps to differences in culture conditions.

Considering that HCASMC are capable of providing a vigorous defensive response to endotoxin by virtue of their expression of both mCD14 and TLR4, our findings suggest that HCASMC, the most abundant cell type in coronary arteries, likely play an important role in mediating vascular inflammation induced by endotoxin. Because CD14 recognizes a diverse array of inflammatory mediators and functions as a pattern recognition molecule in inflammatory cells (10), expression of functional CD14 in coronary artery SMC implies a potentially broader role for these cells in transducing innate immune responses in the vasculature. It may also be consistent with previous reports that have suggested a macrophage-like phenotype for some subpopulations of human intimal or neointimal SMC, based on their expression of the scavenger receptor CD36 (45, 46, 47). Our demonstration of functional mCD14 in HCASMC strongly implies a role for these cells in innate immunity, just as the presence of the scavenger receptor may imply a role for SMC in foam cell formation in atherosclerotic lesions.

LBP is a 60-kDa lipid/phospholipid binding and transfer protein (48, 49) that extracts endotoxin monomers from the bacterial membrane or from aggregates of circulating endotoxin, and subsequently delivers these molecules to CD14, resulting in target cell activation, or to lipoproteins, leading to hepatic clearance (50). The capacity of LBP to modulate endotoxin-induced activation of coronary artery cells has never been examined. Our results (Fig. 6) suggest that the optimal LBP:sCD14 ratio for HCAEC and HCASMC activation is <1:1, which is consistent with other reports showing that the transfer of endotoxin monomers by LBP to CD14 is catalytic in nature, with one molecule of LBP being able to transfer more than one LPS monomer to CD14 (48). Further, our results show that LBP:sCD14 ratios of 2:1 or greater inhibit activation, presumably promoting endotoxin clearance by delivering it to lower affinity binding sites. These findings are consistent with the recent report of Gutsmann et al. (51), who also found a concentration-dependent dual role for LBP in the activation of freshly isolated human mononuclear cells at low LBP levels, but an inhibition of activation at higher acute-phase LBP levels. To our knowledge, similar studies have never been reported in endothelial or vascular SMC. Although in vivo much of this inactivation produced by increased LBP levels would result from enhanced clearance by serum proteins and lipoproteins (50), our experiments were performed under serum-free conditions. This suggests that EC may themselves participate in endotoxin clearance, possibly through uptake and/or transcytosis mediated by caveolae or scavenger receptors. Because the chronic subacute infections implicated in atherosclerosis are associated with a lower LBP:sCD14 ratio, our findings suggest that the capacity of endotoxin to activate EC might be relatively greater in patients with these conditions.

In summary, we report in this study that low levels of endotoxin cause proinflammatory activation of HCAEC and HCASMC. HCASMC, which are remarkably responsive to endotoxin, likely contribute importantly to endotoxin-induced vascular inflammation. This striking responsiveness of HCASMC to low pg/ml levels of endotoxin may be due in part to expression of endogenous mCD14. We demonstrate that both HCAEC and HCASMC express TLR4, the putative endotoxin-signaling receptor. We also demonstrate that the degree of vascular cell activation by endotoxin is regulated by the molecular ratio of LBP and sCD14, two proteins thought to play a critical role in the initiation of the endotoxin-signaling pathway in vivo. Together, these findings provide novel insight into mechanisms by which low levels of circulating endotoxin might promote atherosclerosis in humans. More broadly, they suggest that vascular cells, especially vascular SMC, may play an active role in the innate immune response.

We thank Shao-Ping Xu for assistance with the RT-PCR experiments.