This study sought to determine the influence of tumor necrosis factor-α (TNF-α) on intimal hyperplasia (IH) and characterize the mechanisms of transcriptional regulation after vascular injury. A murine model of wire carotid artery injury was employed to induce IH in wild-type (WT) and TNF-α-deficient [TNF(−/−)] animals. Three days after injury, TNF-α and nuclear factor-κB (NF-κB) protein expression was markedly increased in the injured WT carotid artery compared to control. Injury increased TNF-α and NF-κB mRNA expression 100- and 7.5-fold, respectively. Compared with WT specimens, injury in TNF(−/−) animals decreased both NF-κB mRNA and protein nearly 7.5- and 4-fold, respectively. Expression of the NF-κB-dependent cytokine monocyte chemotactic protein 1 was markedly diminished in injured TNF(−/−) animals. Finally, TNF(−/−) animals demonstrated a sevenfold reduction in IH compared with WT animals. Cumulatively, these data mechanistically link TNF-α and NF-κB in vivo and suggest an important influence of TNF-α on postinjury IH.
- nuclear factor-κB
accumulating evidence links inflammation and intimal hyperplasia (IH; Refs.9, 19, 22). Luminal endothelial injury, both indirectly as with nicotine and directly as with angioplasty, initiates inflammatory cell infiltration and subsequent expression of cytokines, growth factors, and chemoattractants. Vascular smooth muscle cells (VSMCs) proliferate and migrate to the intima thus forming the nidus for IH. Within this paradigm, numerous investigators have attempted to define specific inflammatory mediators.
Tumor necrosis factor-α (TNF-α) is a multifunctional cytokine that has been identified in human atherosclerotic plaque specimens (1). Although macrophages are classically identified as the source of TNF-α after arterial injury, other cells, including the endothelium and VSMCs, also produce and respond to TNF-α (4,29). In vitro, TNF-α mediates proinflammatory events including VSMC proliferation, migration, and endothelial cell adhesion molecule expression (13, 21). TNF-α exerts these effects by activating the transcriptional regulator nuclear factor-κB (NF-κB; Ref. 23). In vivo, balloon injury of rabbit aortas results in increased expression of TNF-α in actively proliferating VSMCs (28). Balloon injury also results in increased NF-κB activity and production of the NF-κB-dependent chemokine monocyte chemotactic protein 1 (MCP-1; Ref. 10). To date, no study of vascular injury has linked expression of TNF-α and NF-κB in vivo.
The pathogenic mechanisms that relate TNF-α and IH remain controversial. Some suggest that TNF-α actually has antiproliferative effects by promoting VSMC apoptosis (7, 12). Although there is a microanatomic correlation between cytokines and IH formation, clinical data suggest a weaker association between TNF-α and symptomatic cardiovascular occlusive disease (24). To date, one study has demonstrated attenuation of IH in a TNF-α-deficient mouse (16). However, these investigators utilized a model of low shear stress. Herein, we employ a model of direct mechanical endoluminal injury. The purposes of the present study are to 1) determine the influence of TNF-α on the in vivo development of IH and 2) examine the effect of TNF-α-mediated IH on NF-κB and MCP-1 expression after vascular injury.
Murine model of carotid injury.
This study utilized 6- to 8-wk-old, 25- to 30-g, male, wild-type (WT) mice of strain B6,129SF2/J and TNF-α-deficient [TNF(−/−)] mice of strain B6,129S-Tnftm1Gkl (Jackson Laboratory, Bar Harbor, ME) (15). Experimental groups for morphometric analysis at 28 days were as follows: WT injured (n = 7), WT sham (n = 6), and TNF(−/−) injured (n = 5). All experimental protocols were approved by the University of Colorado Animal Review Committee. General anesthesia was administered (80 mg/kg ketamine, Fort Dodge Laboratories, Fort Dodge, IA and 20 mg/kg xylazine, Phoenix Pharmaceutical, St. Joseph, MO) via intraperitoneal injection. Surgical procedures were assisted with a StereoZoom 7 microscope (Bausch and Lomb, Rochester, NY). The left carotid artery was exposed via a midline neck incision. The common, external, and internal carotid arteries were identified and controlled (using 7–0 Surgilene suture). A 30-gauge needle was then employed to make an arteriotomy at the external carotid. A 0.014-in flexible wire (ACS Hi-Torque Floppy II, Advanced Cardiovascular Systems, Temecula, CA) was thrice passed into the common carotid with a simultaneous twisting motion as previously described (11). The external carotid was ligated and hemostasis was achieved. The skin was then closed with 7–0 Surgilene suture. The sham-operated animals did not undergo arteriotomy and wire injury. Euthanasia was performed according to the guidelines set by the American Veterinary Medical Association Panel on Euthanasia with general anesthesia and pentobarbital sodium (100 mg/kg). For morphometric analysis, animals were killed at 28 days and received intracardiac injection of 500 μl of heparinized saline followed by 4% paraformaldehyde. For all other studies, animals were euthanized at 3 days. Animal care was provided by the University of Colorado animal facility.
Three days after injury, bilateral common carotid arteries were isolated and removed. The right common carotid arteries were examined as uninjured contralateral controls. All tissue was immediately preserved in Tissue Freezing Medium (Triangle Biomedical Sciences, Durham, NC). Slides were fixed in 70% acetone-30% methanol solution for 10 min. After air drying, slides were washed three times in PBS for 10 min.
Specimens were blocked with 10% donkey serum (TNF-α and MCP-1) or 10% goat serum (NF-κB) for 1 h at room temperature. Sections were then incubated at 4°C overnight with either goat anti-mouse TNF-α polyclonal IgG, goat anti-mouse MCP-1 polyclonal IgG, or rabbit anti-mouse NF-κB p65 polyclonal IgG (Santa Cruz Biotechnology, Santa Cruz, CA). After PBS wash in triplicate, sections were incubated for 1 h in the dark at room temperature in either indocarbocyanine (Cy3)-labeled donkey anti-goat IgG (TNF-α and MCP-1) at a 1:150 dilution or goat anti-rabbit IgG (NF-κB). Negative controls were run parallel with all sections by substituting a goat anti-mouse nonspecific IgG for the primary antibody in the samples stained for TNF-α and MCP-1. On sections stained for the NF-κB p65 subunit, a nonspecific rabbit anti-mouse antibody was employed.
Cell-wall glycoproteins and nuclei were, respectively, stained green with Alexa Fluor 488/wheat-germ agglutinin (WGA) (Molecular Probes, Eugene, OR) and bis-benzimide (Sigma Chemical, St. Louis, MO). Fluorescent images were evaluated and photographed with appropriate filter cubes using an automated Leica DMRXA confocal microscope with full software control (Intelligent Image Innovations). Differences in immunofluorescence are expressed semiquantitatively as mean integrated Cy3 intensity per unit of vessel area (μm2).
RT quantitative PCR.
RNA was isolated from individual arterial specimens with an isolation kit (Ambion, Austin, TX). One microgram of total RNA was reverse transcribed using random hexamer primers (2.5 μM) in a final concentration of 5.5 mM MgCl2, 50 mM KCl, 10 mM Tris · HCl (pH 8.3), 0.5 mM of each dNTP, 20 U of RNase inhibitor, and 50 U of MultiScribe RT (Perkin-Elmer, Foster City, CA). The sample set at each amplification included a negative control represented by the mixture minus the template. Probes were designed across intronic sequences. After amplification, the product size was tested by electrophoresis in 1.8% agarose gel. The primers for TNF-α were obtained from Applied Biosystems. The primers for the NF-κB p65 subunit were provided by Source Precision Medicine (J. Emmick, Boulder, CO). These sequences, which have not been previously published, were forward primer: AGATCTTCTTGCTGTGCGACAA, and reverse primer: GTGCCTCCCAGCCTGGT.
Probes for murine TNF-α and NF-κB were labeled at the 5′ end with the reporter dye molecule 6-carboxy fluorescein (FAM). A ribosomal RNA 18S target probe labeled at the 5′ end with the reporter dye molecule VIC (Applied Biosystems) was employed as the internal control. TaqMan real-time quantitative PCR was performed by amplifying mixtures containing 100 nM of selected probes, 200 nM of primers, and the target cDNA template at 95°C for 20 s and 60°C for 1 min for 40 cycles using the ABI PRISM 7700 sequence detector (Applied Biosystems). During each PCR cycle, the 5′ to 3′ exonuclease activity of DNA polymerase cleaves the TaqMan probe, thereby releasing the fluorescence of the reporter dye at the appropriate wavelength. The increase in fluorescence is proportional to the concentration of template in the PCR (2). The threshold was determined according to the manufacturer's protocol for the TaqMan PCR kit. By using 18S as an internal control, the relative number of amplified target DNA copies was calculated. Data are expressed as a relative percent increase of mRNA in the injured arteries versus the contralateral control.
A transcription factor assay was employed to investigate the presence of unbound NF-κB p65 subunit in individual arterial specimens (17). This assay is based on the specific binding of the active form of NF-κB from tissue extract to a NF-κB consensus oligonucleotide attached to an ELISA plate. The primary antibody used to detect NF-κB recognizes an epitope on the p65 subunit, which is accessible only when NF-κB is bound to its target DNA. A secondary horseradish peroxidase-conjugated antibody provides a colorimetric readout quantified by spectrophotometry. Positive controls for the NF-κB p65 subunit were provided from cellular extracts previously evaluated by both transcription factor assay and electrophoretic mobility-shift assay (EMSA, Active Motif, Carlsbad, CA). To monitor the specificity of the assay, both WT and mutated consensus oligonucleotides were employed in each reaction.
Carotid specimens from both WT and TNF(−/−) mice were skeletonized, harvested bilaterally, and fixed at −70°C. The tissue was diced into small pieces with a cooled razor blade and placed in lysis buffer. A mechanical homogenizer was then applied for 30 s, which maintained a temperature of 4°C. Samples were incubated on ice for 30 min and centrifuged at 10,000 g for 10 min. Supernatants were transferred to separate tubes and recentrifuged. Tissue protein content was measured via the Coomassie protein assay (Pierce, Rockford, IL). Total protein (20 μg) was loaded to each well and assayed according to the manufacturer's directions (Active Motif). Quantification of the NF-κB p65 subunit was expressed in mean absorbance (λ) per arterial sample.
IH and morphometric analysis.
The right and left common carotid arteries were harvested, embedded in paraffin, and sectioned for hematoxylin and eosin staining. Serial sections were taken along the length of the vessel at 150-μm intervals. Qualitative review of these specimens revealed the area of greatest luminal stenosis. At this point, 20–30 sections were taken at 4-μm intervals of which multiple six to eight samples underwent quantitative morphometric analysis. Plain images were taken on the confocal microscope and the following structures were identified: lumen, internal elastic lamina (IEL), external elastic lamina (EEL), and neointima. Intimal (specimen from lumen to IEL) and medial areas (specimen from IEL to EEL) were measured using Slidebook software (version 184.108.40.206). Intimal to medial ratios were also calculated.
Data are presented as means ± SE. ANOVA with Bonferroni-Dunn post hoc analysis was used to analyze differences between experimental groups. Statistical significance was accepted within 95% confidence limits.
Vascular injury and TNF-α expression.
Several methods were utilized to demonstrate the increased expression of TNF-α in WT animals. Three days after injury, TNF-α mRNA expression was increased 100-fold compared with the uninjured contralateral arteries (n = 3; P < 0.05; Fig. 1 A). Immunohistochemistry utilizing Cy3-labeled TNF-α demonstrated increased TNF-α in injured specimens. Localized predominantly to the tunica media of the injured arteries (Fig. 1 B), the uninjured control vessels demonstrated essentially no signal in any portion of the specimen. Furthermore, no differences were noted between the noninjured vessels and vessels from sham WT-operated animals (n = 3). The reported immunofluorescence consistently represents data from multiple arterial sections of three separate animals. Together these results suggest a strong association between injury and TNF-α expression.
Vascular injury, TNF-α, and IH.
Endoluminal injury promoting IH has been well documented in various animal models ranging from sheep to rats (20, 26). Carotid injury in mice has previously been described (11, 27). To validate the effectiveness and reproducibility of our model, direct wire injury was performed on seven consecutive WT mice. Twenty-eight days after injury, arteries were harvested. Qualitatively, the intimal size was consistently larger in the injured versus the noninjured contralateral vessels (Fig. 2).
Although demonstrating that TNF-α is upregulated after vascular injury, it remains unclear whether TNF-α is an important factor in the development of IH. We therefore studied the effects of vascular injury in a TNF-α-deficient animal. Twenty-eight days after injury, TNF(−/−) mice had less IH than WT mice. Intimal and medial areas were calculated and compared from multiple sections with the highest degree of stenosis from each animal (Fig. 3). The intimal area of the WT injured vessels was substantially greater than that of the noninjured vessels [(14.1 ± 2.5) × 103 vs. (0.7 ± 0.1) × 103μm2; P < 0.05]. Compared with WT animals (n = 7), injured TNF(−/−) mice (n = 5) demonstrated a sevenfold decrease in intimal area [(14.1 ± 2.5) × 103 vs. (2.0 ± 0.9) × 103 μm2; P < 0.05; Fig. 3]. Qualitatively, although vascular injury in TNF(−/−) animals resulted in less IH than WT, these animals still exhibited some intimal change. However, compared to the uninjured side, injury in the TNF(−/−) animals did not have a significant increase in IH [(2.0 ± 0.9) × 103 vs. (0.6 ± 0.1) × 103 μm2 contralateral; P = 0.57]. To verify that these differences were exclusively related to intimal changes, we also measured medial areas and calculated intimal-to-medial ratios. There were no differences in medial areas between injured groups. Furthermore, intimal-to-medial ratios support exclusive intimal proliferation. The WT injured specimens maintained a sevenfold increase in the intimal-to-medial ratio compared with the injured TNF(−/−) vessels (1.29 ± 0.16 vs. 0.18 ± 0.06;P < 0.05). No differences in intimal area were noted between the sham-operated group (n = 6) and the uninjured controls.
Vascular injury, TNF-α, and NF-κB.
We utilized several methods to characterize the influence of TNF-α on NF-κB activation in our in vivo model. Three days postinjury, NF-κB mRNA (Fig. 4 A) in the injured WT was increased sixfold compared with contralateral control and TNF(−/−) (P < 0.05). Immunohistochemically, NF-κB was identified in the intima and media of both WT and TNF(−/−) animals (Fig. 4 B). Mean integrated Cy3 intensity/vessel area (μm2) of the injured WT demonstrated a threefold increase versus the injured TNF(−/−) sections. Intranuclear signal could be identified at the site of injury. Results were consistent across multiple sections in three separate animals.
We next measured unbound NF-κB p65 subunit protein in individual arteries (Fig. 4 C). WT injured specimens maintained a fivefold increase in unbound p65 absorbance versus contralateral control (1.87 ± 0.4 vs. 0.33 ± 0.1; P < 0.05). Compared with injured WT mice, injury in TNF(−/−) mice decreased NF-κB protein fourfold (1.87 ± 0.4 vs. 0.49 ± 0.08; P < 0.05). Interestingly, injured TNF(−/−) mice still demonstrated an increase in unbound p65 compared to uninjured TNF(−/−) control mice (0.49 ± 0.08 vs. 0.14 ± 0.06; P < 0.05). To monitor the specificity of the assay, both a WT and mutated p65-specific consensus oligonucleotide were used. When added to the reaction, the WT oligonucleotide consistently prevented p65 binding to the plate and resulted in zero absorbance at 450 nm. Mutated consensus oligonucleotide had no effect.
Vascular injury and MCP-1 expression.
To evaluate the downstream effect of TNF-α-mediated IH, we investigated the expression of the NF-κB-dependent protein MCP-1 after vascular injury. MCP-1 was identified by immunofluorescence after wire injury in both WT and TNF(−/−) animals (Fig.5). Qualitatively, fluorescence intensity and distribution were markedly diminished in the injured TNF(−/−) mice compared with injured WT specimens. Quantitatively, the mean integrated Cy3 intensity/vessel area (μm2) was approximately fivefold higher in the injured WT samples versus the control. TNF(−/−) samples revealed minimal MCP-1-labeled signal in the intima and media. These data are consistent across multiple samples from three different WT and TNF(−/−) animals.
The present study demonstrates that 1) TNF-α is abundantly expressed locally after mouse carotid injury; 2) these observations are associated with activation of NF-κB; and3) TNF-α-deficient animals have significantly lower levels of MCP-1, NF-κB, and IH. Despite data suggesting a fundamental role for TNF-α in the inflammatory fibroproliferative response to vascular injury, several groups conversely report that TNF-α may not be an important component in the promotion of IH. Elhage and colleagues (5) demonstrated a decrease in atherosclerotic area when blocking the action of interleukin-1 in apolipoprotein E-deficient mice, whereas blocking TNF-α had no effect. Clinical studies that measured plasma levels of TNF-α in patients with unstable angina showed no difference compared with healthy controls (25). Furthermore, TNF-α promoted apoptosis in both human and rat VSMCs in vitro (7). Whereas VSMC apoptosis in the mature plaque may predispose patients to acute coronary syndromes, apoptosis of postinjury VSMCs should theoretically decrease plaque mass and quite possibly inhibit IH (9).
The well-known proinflammatory profile of TNF-α suggests its key role in the vascular response to injury. However, few reports characterize the relative role of TNF-α in IH. Over a decade ago, Barath and colleagues (1) identified TNF-α in human atherosclerotic specimens. Numerous studies have characterized the proinflammatory effects of TNF-α in vitro. VSMCs produce and respond to TNF-α by accelerating proliferation and growth factor production (13,29). Several in vivo studies have identified and temporally characterized TNF-α expression after vascular injury (6,28). Although these data suggest a change in VSMC phenotype in vitro and temporal association of TNF-α expression in vivo, they do not demonstrate a causative role of TNF-α in IH.
First reported by Lindner and colleagues (11), the murine model of wire carotid artery injury denudes the vascular endothelium. Although VSMC proliferation and migration to form a neointima are a central component of postangioplasty restenosis, it would be inappropriate to make direct associations between IH in mice and the fibroproliferative response in humans. The majority of in vivo studies are performed on normal native vessels. As such, we are unable to extrapolate results from normal murine arteries to diseased atherosclerotic vessels of patients with acute coronary syndromes. The absence of an atherosclerotic lesion before injury may lead to variations in the degree of resultant IH. Although hyperlipidemic mouse models exist, they do not fully mimic the complexities of atherosclerosis or postangioplasty restenosis. Furthermore, the stretch and denudation injury of angioplasty in humans may differ from the denudation of the vascular endothelium alone in mice (8). Finally, in the present study, we did not characterize the specific cellular subtypes involved in the hyperplastic response. As such, we are unable to draw conclusions as to influence of TNF-α on monocytes, lymphocytes, or endothelial cells and the relative contribution to the development of IH.
A recent study by Rectenwald and colleagues (16) addressed the role of TNF-α and IH in TNF-α-deficient mice. Using a low-shear-stress carotid-ligation model, they observed a consistent lack of IH in TNF-α-deficient animals. Furthermore, TNF-α mRNA was present in ligated arteries but not in normal or sham-operated mice. Interestingly, mice that were able to produce only membrane-bound TNF-α demonstrated an increased hyperplastic response. Our data corroborate and extend the Rectenwald study by demonstrating an attenuation in hyperplasia in TNF-α-deficient animals after mechanical injury. Similarly, we explored this relationship by comparing intimal-to-medial ratios. Instead of pooling samples, we chose to quantify TNF-α and NF-κB mRNA and unbound NF-κB p65 protein in individual arterial specimens. It must be noted that the vascular response to injury in TNF(−/−) mice was not completely eliminated; IH, albeit minimal, can be detected in these specimens. This leads us to believe that other molecular mechanisms contribute, at least in part, to this fibroproliferative response.
TNF-α is one of many NF-κB-dependent cytokines. Functioning as a transcriptional activator, NF-κB mediates the overexpression of other proinflammatory genes (18, 24) and has been identified in human atherosclerotic lesions (3). Although we found an increase in unbound NF-κB p65 subunit protein after vascular injury concurrent with TNF-α expression and IH, these data must be interpreted cautiously. The current study is limited to a temporal association between TNF-α expression and unbound NF-κB. The current dogma that transcription of NF-κB mRNA is regulated by NF-κB itself is contrary to our findings that transcription can actually be induced by vascular injury alone. Furthermore, it is surprising that removal of a single proinflammatory cytokine in the transgenic animal results in such a dramatic decrease in transcribed message. We selectively investigated a limited number of time points, with 72 h postinjury revealing the most striking difference in mRNA and free NF-κB p65 protein. With regard to the TNF(−/−) animal, it is possible that these animals are deficient in other proinflammatory mediators aside from TNF-α. Finally, we acknowledge the use of EMSA as a well-known standard for measuring NF-κB. Unfortunately, with the small amount of tissue in our model, we were unable to reproducibly perform EMSA analysis on individual specimens. As such, a transcription factor assay was utilized to quantify unbound NF-κB p65 subunit in individual arteries, which maintains a reported 10-fold higher sensitivity compared to EMSA (17).
MCP-1 is a powerful chemotactic agent produced by several resident vascular cells including monocytes, endothelium, and VSMCs. Expression of MCP-1 is induced by TNF-α (14) and has been linked to NF-κB activation after vascular injury in vivo (10). We investigated the effect of vascular injury on expression of MCP-1 as a downstream surrogate marker for NF-κB activation. Our results suggest a decrease in MCP-1 expression in the TNF(−/−) animal. Cumulatively, these observations suggest that the influence of TNF-α on IH is twofold. Although TNF-α does appear to have direct mitogenic and chemotactic effects on various vascular cells, the influence of TNF-α might be more global via its downstream effects on NF-κB-dependent mitogens.
This work is supported by National Institutes of Health Grants GM-49222 and GM-08315 (to A. H. Harken).
Address for reprint requests and other correspondence: C. H. Selzman, Division of Cardiothoracic Surgery, Box C-310, Univ. of Colorado Health Sciences Center, 4200 East Ninth Ave., Denver, CO 80262 (E-mail:).
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.
March 29, 2002;10.1152/ajpregu.00033.2002
- Copyright © 2002 the American Physiological Society