The mechanisms of lung inflammation and disease by asbestos have been widely studied (Haegens et al. 2007
; Janssen et al. 1997
; Mossman et al. 2000
) and its association with cardiovascular disease reported in occupational cohorts (Dement et al. 1983
; McDonald et al. 1993
; Sanden et al. 1993
; Sjogren 1997
). However, to our knowledge, no studies examining the mechanisms of the exacerbation of atherosclerosis by inhaled asbestos have been performed. Therefore, to test the hypothesis that airborne fibers exacerbate atherosclerosis and to elucidate the mechanisms involved, we used chrysotile asbestos fibers in a murine model of inhalation. Chrysotile asbestos is ubiquitous in the northern hemisphere and is the asbestos type historically used worldwide in > 95% of asbestos-containing products (Mossman et al. 1990
). Although significant efforts have been made to reduce occupational and environmental exposure to amphibole types of asbestos (crocidolite, amosite) that may be more pathogenic in mesothelioma (Mossman and Gee 1989
), the use of chrysotile asbestos continues worldwide (Nicholson 1997
). Moreover, analysis of airborne surface dust in residential areas of Lower Manhattan after the collapse of the World Trade Center in New York City revealed increased chrysotile asbestos [Centers for Disease Control and Prevention (CDC) 2003
], raising the concern about its short- and long-term pathogenic effects after inhalation by the general population.
In these experiments, we used the atherosclerosis-prone apolipoprotein E–deficient (ApoE−/−
) mouse and ApoE−/−
mice crossed with CD4−/−
[double-knockout (DKO)] mice to test the hypothesis that inhaled asbestos fibers exacerbate atherosclerosis and that the mechanism involves CD4+
T cells that are increased in lung after inhalation of asbestos (Shukla et al. 2007
). We exposed the mice to ambient air, fine titanium dioxide (TiO2
; a nonpathogenic control particle), or chrysotile asbestos in the University of Vermont (UVM) Inhalation Facility (Sabo-Attwood et al. 2005
). Our data are the first to show a direct relationship between inhaled pathogenic fibers and the exacerbation of atherosclerosis via a CD4+
T cell–dependent process. Moreover, they suggest an important role for activation of monocyte chemoattractant protein-1 (MCP-1) and early transcription factors activator protein-1 (AP-1) and nuclear factor-κB (NF-κB) in the development of atherosclerosis by inhaled pathogenic pollutants.
Materials and Methods
We maintained male and female breeding C57BL/6, ApoE−/−, and CD4−/−mice, purchased from Jackson Laboratories (Bar Harbor, ME), by brother/sister matings at UVM. We used the B6.129S2-Cd4tm1Mak strain of CD4−/− mice, the same background as the ApoE−/− mice used in these experiments. Female offspring were used in our experiments. Double knockout (DKO) animals (ApoE−/− and CD4−/−) were generated by producing F2 generation mice between ApoE−/− and CD4−/− and using polymerase chain reaction (PCR) to select mice that were homozygous for both gene knockouts. We then maintained homozygous ApoE−/−/CD4−/−DKO mice by brother/sister matings. Animals homozygous for both genes were used in the experiments. The DKO mice were selected in the F2 offspring by genotyping using PCR protocols and were backcrossed into the C57BL/6 background for more than eight generations. We maintained both strains as colonies at UVM and fed them normal chow. All procedures were approved by the UVM Institutional Committee on Use and Care of Animals. We treated mice humanely and with regard for the attenuation of suffering. We weighed all animals after anesthesia and found no differences between groups and treatments (data not shown).
Inhalation procedures and sample collection
or DKO mice at 4–6 weeks of age were exposed to chrysotile asbestos [Mg3
; National Institute of Environmental Health Sciences (NIEHS) reference sample at approximately 5 mg/m3
air; range, 4.7–5.7 mg/m3
air] for 6 hr/day, 5 days/week. This concentration is equivalent historically to concentrations of chrysotile asbestos in unregulated workplaces and levels that caused lung diseases, and in air during the World Trade Center disaster (CDC 2003
). The size dimensions (mean aerodynamic diameter, 0.34 μm) of aerosolized NIEHS chrysotile asbestos have been reported previously (BéruBé et al. 1996
We conducted whole-body inhalation exposures within our inhalation facility (accredited by the Association for Assessment of Laboratory Animal Care) as described previously (Sabo-Attwood et al. 2005
). We exposed control animals (sham groups) to clean, ambient air. We studied younger animals with early lesion development to optimize the likelihood of detecting differences in lesion size as a consequence of exposure. If we had used older animals with larger lesions, lengths of exposure to asbestos fibers may not have elicited easily detectable differences in lesion size.
Because the schedule of exposure was 5 days/week, simulating a workplace setting, 30 days of exposure is equivalent to 6 weeks. We used shorter exposures to determine whether differences in early responses to asbestos might reveal potential mechanisms responsible for triggering differential lesion sizes at the end of 30 days of exposure, a time point previously shown to be associated with lung fibrogenesis (Sabo-Attwood et al. 2005
We used fine TiO2 (0.2–2.5 μm diameter; Fisher Scientific, Pittsburgh, PA) as a non-pathogenic particle control at surface area concentrations approximately equal to those of asbestos (~ 28 mg/m3 air) in some of the experiments using ApoE−/− mice. After 3, 9, or 30 days of exposure, we collected bronchoalveolar lavage fluid (BALF) and blood, aorta, and lung tissues and prepared them for analyses. We performed surgical procedures after injecting the mice intraperitoneally with sodium pentobarbital. We collected blood via cardiac puncture into tubes with sodium EDTA as the anticoagulant, and saved the plasma for chemokine/cytokine assays. Tracheas of mice were cannulated with polyethylene tubing, and the lungs lavaged with sterile calcium- and magnesium-free phosphate-buffered saline (PBS) in a total volume of 1 mL.
Mouse lung, heart, and aorta were dissected and immersion fixed overnight in 3% paraformaldehyde/PBS at 4°C as previously described (Taatjes et al. 2000
; Wadsworth et al. 2002
). During fixation, we bisected the hearts with a cut parallel to both atria. The hearts and lungs were immersed in optimal cutting temperature compound (OCT; Tissue-Tek, Torrance, CA) in labeled embedding molds (hearts were oriented cut side up so that the first sections taken would reveal the sinus area), snap-frozen in liquid-nitrogen–cooled 2-methyl butane, and stored at −80°C until the time of cryostat sectioning, as previously described (Paigen et al. 1987
We defined the area for sectioning by three prominent valve cusps at the juncture of the aortic sinus region to the end of the valve region, when the valves disappeared and the aorta became more rounded in appearance.The sections on the slides were air-dried for 30 min to ensure proper adhesion before being stored in a slide box at −80°C. En face preparations were not done in these experiments because the thoracic and abdominal aortas were used for nuclear and cytoplasmic protein extraction for analyses of transcription factors.
Aortic lesion quantitation
We examined oil red O–stained sections (Pearse method) using an Olympus BX50 upright light microscope (Olympus America, Inc., Lake Success, NY) with an attached Optronics MagnaFire digital camera (Optical Analysis Corp., Nashua, NH). The sections were imaged with a 4 × objective lens, and we used MagnaFire software (version 2.0) to capture 1,280 × 1,024 pixel RGB digital images. We performed computer-assisted image analysis using MetaMorph software (Universal Imaging Corp., Downingtown, PA) essentially according to previously published protocols (Wadsworth et al. 2002
). We opened cropped digital images and set the appropriate (precalibrated) objective calibration by choosing “Measure/Calibrate Distance/Apply.” This allowed area measurements to be expressed in calibrated square micrometer values; pixel values were displayed at the bottom of the screen. The images were then assigned threshold values for pixel measurements. Once this was done, the “Integrated Morphometry” feature measured the thresholded area and logged the area values onto a Microsoft Excel spreadsheet (Microsoft Corp., Redmond, WA). The logged values were then converted into an Excel chart for presentation. Animal comparisons were calculated and expressed as area values.
Fiber deposition studies in lung and aorta
To compare the fiber burden in lung with that in aorta, we digested the left lobes and aortas of two control and two asbestos-exposed mice in hypochlorite, a process that does not affect fiber integrity, as previously described (BéruBé et al. 1996
). We transferred the digest to Nuclepore filters and examined them by scanning electron microscopy and X-ray energy-dispersive spectroscopy at 5,000× magnification. We evaluated 10 random fields per filter.
We measured total cholesterol in mouse plasma in the early experiments using a standard commercial cholesterol esterase enzymatic assay (Cayman Chemical Co., Ann Arbor, MI). No differences in cholesterol levels were found between ApoE−/− and DKO mice or in the different treatment groups.
Lung histopathology for inflammation and fibrosis
To determine whether the extent of lung inflammation and fibrosis in lungs correlated with the extent of atherogenesis, lungs were inflated after collection of BALF with a 1:1 mixture of OCT and PBS and fixed in 4% paraformaldehyde. We stained paraffin sections of lung (5 μm thickness) with hematoxylin and eosin (H&E) to quantify inflammation or Masson’s trichrome technique for detection of collagen (Sabo-Attwood et al. 2005
). A board-certified pathologist (K.J.B.) evaluated sections using a blind coding system, as previously described (Craighead 1982
; Haegens et al. 2007
We scored inflammation on a scale from 1 to 4: 1, no inflammation; 2, mild inflammation that was rarely peribronchiolar and consisted primarily of lymphocytes; 3, moderate inflammation with peribronchiolar neutrophils, eosinophils, lymphocytes, and abundant macrophages; and 4, severe inflammation with peribronchiolar neutrophils, eosinophils, and lymphocytes that extend to involve adjacent alveolar septa and abundant bronchiolar and intraalveolar macrophages. We also scored fibrosis on a scale from 1 to 4: 1, no fibrosis; 2, focal fibrosis; 3, moderate fibrosis; 4, severe fibrosis.
BALF differential cell counts
We enumerated total cells in BALF and centrifuged 2 × 104
cells onto glass slides at 600 rpm. We stained cytospins using the Hema3 kit (Biochemical Sciences, Swedesboro, NJ), and performed differential cell counts on 500 cells/mouse (Haegens et al. 2007
Cytokine concentrations in BALF and plasma
To determine whether altered cytokine profiles occurred in the lung and the systemic circulation, we initially measured cytokines using an enzyme-linked immunosorbent assay (ELISA). We used commercial kits for interleukin-6 (IL-6), MCP-1, macrophage inflammatory protein-2 (MIP-2), and interferon-γ (IFN-γ) (Endogen, Woburn, MA) according to the manufacturer’s directions. Briefly, we aliquoted 2-fold dilutions of plasma or BALF into 96-well microplates coated with antibody to the indicated cytokines for 2 hr at 25°C. Plates were then washed, incubated with biotin-conjugated secondary antibody, washed, incubated with streptavidin–horseradish peroxidase conjugate, washed again, and incubated with 3,3′,5,5′-tetramethylbenzidine substrate. The reaction was halted by adding the stop solution in the plate reader at 450 nm and analyzed using a Biotech EL808 ELISA Reader (BioTek, Inc., Winooski, VT). We determined cytokine concentrations from a standard curve using reference standards supplied in the kit.
In subsequent studies, we also measured cytokine and chemokine levels in BALF using the Bio-Plex Protein Array System and a mouse cytokine 22-plex panel (Bio-Rad, Hercules, CA), as previously described (Sabo-Attwood et al. 2005
). This method of analysis is based on Luminex technology and simultaneously measures IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-9, IL-10, IL-12 (p40), IL-12 (p70), IL-13, IL-17, tumor necrosis factor-α (TNF-α), regulated in activation, normal T expressed and secreted (RANTES; CCL5), MIP-1α, MIP-1β, MCP-1, CXCR2 ligand KC/GRO-α (KC), granulocytecolony–stimulating factor (G-CSF), granulocyte/monocyte-colony–stimulating factor (GM-CSF), IFN-γ, and eotaxin protein. We determined concentrations of each cytokine and chemokine using Bio-Plex Manager software, Version 3.0 (Bio-Rad).
Electrophoretic mobility shift assay (EMSA) and Western blot analyses for AP-1 and NF-κB in aortas
We used the EMSA to determine whether inhaled particulates affected the DNA binding of oxidant-associated transcription factors in aortic tissue. Isolated aorta was minced using sterile scissors and then homogenized in a prechilled homogenizer. Nuclear proteins were extracted as previously described (Janssen et al. 1997
; Schreiber et al. 1989
). Briefly, after homogenization, the cells were lysed in hypotonic buffer and 0.6% Nonidet P-40 (Sigma, St. Louis, MO). After centrifugation, the supernatant containing cytoplasmic proteins was collected and stored at −80°C for later analysis. Nuclear proteins were then extracted from the pelleted nuclei (Janssen et al. 1997
; Schreiber et al. 1989
Protein concentrations were measured using the Bio-rad Protein Assay (Bio-Rad). The AP-1 and NF-κB probes containing the consensus sequences of AP-1 and NF-κB binding sites, respectively, were synthesized commercially (Promega, Madison, WI). The oligos were then labeled with γ-32
P-ATP (NEN Life Science, Boston, MA) by T4
polynucleotide kinase (GIBCO-BRL, Gaithersburg, MD). We incubated 20 μL of each binding mixture, composed of 7 μg nuclear protein in binding buffer with 40 mM HEPES (pH 7.8), 4% Ficoll 400, 200 μg/mL poly(dI:dC) [poly(deoxy-inosinic:deoxycytidylic acid); Amersham Pharmacia, Piscataway, NJ], 1 mM MgCl2
, 0.1 mM dithiothreitol, and 1 μL of labeled probe (0.02 pmol), at room temperature for 15 min and then loaded it onto a 5% sodium dodecyl sulfate (SDS) polyacrylamide gel (nondenaturing), as previously described (Janssen et al. 1997
; Mossman et al. 2000
). The gel was run in 0.25 × Tris/borate/EDTA buffer at 120 V for 2.5 hr, then dried and exposed to X-ray film overnight at −80°C. The autoradiographic films were scanned by densitometry and analyzed with Quantity One Software, Version 4.2 (Bio-Rad).
We used cytoplasmic protein extracted from the aortas during the preparation of nuclear extracts for Western blot analysis to confirm the DNA binding of NF-κB by using an antibody to phosphorylated Kappa B inhibitor (IκB; Cell Signaling Technology, Inc., Danvers, MA). Briefly, 20 μg protein from each sample was electrophoresed on a 10% SDS-PAGE and electroblotted onto a nitrocellulose membrane, which was then incubated with the antibody overnight with shaking at 4°C. After incubation, the protein bands were visualized using a SuperSignal West Pico Trial Kit (Pierce, Rockford, IL) and exposed to radiographic film. The blots were reprobed with a β-actin antibody (Abcam, Cambridge, MA) to detect cytoplasmic β-actin as the loading control. The images and densities were captured with a GS-700 Imaging Densitometer (Bio-Rad), analyzed with Quantity One Software, and presented as the ratio of IκB to β-actin.
We used Student’s two-sample t-tests to compare groups when the data were normally distributed. We used the Mann-Whitney test to compare groups using transformed data to stabilize the variance or when results were not normally distributed. Two-way analysis of variance (ANOVA) was used to analyze transformed cytokine/chemokine data for each time point, with post hoc analysis to determine genotype effects within treatment or treatment effects within genotype. We used chi-square analysis to compare the histologic scores for fibrosis and inflammation. Nonparametric Kendall tau correlation coefficients were derived for the relationships between lesion size and MCP-1 concentrations in ApoE−/− mice. All analyses were performed using SPSS, version 14, and data are presented as mean ± SE.
These data are the first to show that inhaled chrysotile asbestos, a documented pathogenic airborne fiber, exacerbates atherosclerotic lesions in chow-fed ApoE−/−
mice. It should be noted that concentrations of airborne asbestos fibers generated in this study were high, as encountered episodically during the World Trade Center disaster, in contrast to mean ambient levels in buildings (0.00004 and 0.00243 fibers/mL air in rural and urban samples, respectively) or outdoor air (0.00001 and 0.0001 fibers/mL air in rural and urban samples, respectively) (Health Effects Institute 1991
). Atherosclerotic effects were attenuated significantly when animals also lack CD4+
T cells, showing their critical importance in the pathogenesis of disease. The fact that DNA binding of both AP-1 and NF-κB were increased in nuclear extracts from aortas of ApoE−/−
mice after 9 days of asbestos exposure, in the absence of asbestos fibers in the aortas, indicates that distal signaling mechanisms involving these two transcription factors are activated in atherogenesis. We found no detectable differences in AP-1 and NF-κB DNA binding activity in extracts from aortas of DKO mice exposed to either clean air or asbestos, which provides additional support for the important interactions among asbestos exposure, CD4+
T cells, and lesion development through redox-sensitive signaling pathways. Furthermore, our studies suggest that MCP-1, which has binding sites for both AP-1 and NF-κB in its promoter region, is an important chemokine in the signaling events leading to atherosclerosis. Recently, Sun et al. (2005)
reported that ApoE−/−
mice fed a high-fat diet and exposed to inhaled particulate matter had increased atherosclerotic lesions that were accompanied by increased macrophage infiltration and evidence of oxidative stress in the aorta. However, they did not identify the link between lung effects and these findings in aortic tissue and mechanisms of particulate matter–induced atherosclerosis.
Atherosclerosis is a complex disease with etiologies that involve both genetic and environmental factors. The present study confirms the importance of inflammation in the athero-genic process and supports a role for the CD4+
T cell, shown to be present in atherosclerotic lesions (Zhou et al. 1996
). Although we did not stain for the presence of CD4+
T cells in lesions in these experiments, we did find more circulating CD4+
T cells in the blood of a small subset of ApoE−/−
mice exposed to asbestos compared with a subset exposed to clean air for 30 days, although the differences were not significant (data not shown). Early work using ApoE−/−
mice crossed with IFN-γ receptor–deficient mice suggested that IFN-γ promoted atherosclerosis and was critical in modulating the balance between TH
1 (cellular immunity) and TH
2 (humoral immunity) sub-sets of T cells (Gupta et al. 1997
). Exogenous IFN-γ has also been shown to increase atherosclerosis in ApoE−/−
mice (Whitman et al. 2000
). IL-4 is reportedly necessary for invoking a TH
2 response (Feili-Hariri et al. 2005
) and has been shown to play a role in the progression of atherosclerosis (Davenport and Tipping 2003
). IL-4 levels were increased similarly in BALF of both ApoE−/−
and DKO mice exposed to asbestos for 3, 9, or 30 days, but levels were not detectable in plasma.
Elhage et al. (2004)
reported that DKO mice had increased lesions in the descending and abdominal aorta, but they found similar lesions in the aortic sinuses of ApoE−/−
mice. We recognize that the lesions observed in the present study are larger than reported in the literature for ApoE−/−
mice of similar age. The reasons for the differences between the present work and previous reports are unclear but may relate to the study design and housing conditions. We chose to start the exposures to asbestos at 4–6 weeks of age and to terminate the experiments after 30 days of exposure (5 days/week, for a total of 6 weeks) when animals were at most 12 weeks of age. Because we examined only the aortic sinus, extending the exposures to 18 weeks or 1 year or doing en face
analysis may have yielded different results. However, our objective was to determine whether the length of exposure to asbestos necessary to develop lung inflammation and early fibrosis was sufficient to elicit differences in aortic responses between ApoE−/−
and DKO mice. Of major significance here is that both ApoE−/−
and DKO mice appeared to respond similarly to asbestos exposure with respect to lung inflammation, fibrosis, and selected cytokine concentrations in BALF but differed significantly in their responses in plasma concentrations of MCP-1 after 30 days of exposure to asbestos. ApoE−/−
mice exposed to asbestos had an increase in plasma MCP-1 levels compared with clean-air–exposed animals. In contrast, MCP-1 concentrations in DKO mice, although constitutively higher, remained unchanged after clean air and asbestos exposure. In BALF, genotype differences in response became apparent at 9 and 30 days of exposure to asbestos in that MCP-1 and IL-6 were higher in ApoE−/−
than in DKO mice. The source of these cytokines remains unknown, and future studies will examine changes induced by asbestos in MCP-1–defi-cient mice crossed with ApoE−/−
Our data showing elevated levels of plasma MCP-1 in asbestos-exposed ApoE−/−
mice are consistent with growing evidence that cytokines participate as autocrine and paracrine mediators in the pathogenesis of atherosclerosis (Hansson 2001
; Upadhya et al. 2004
). Because elevated levels of MCP-1 are known to occur in BALF after inhalation of chrysotile asbestos in C57BL/6 mice (Haegens et al. 2007
), neutrophils elaborating MCP-1 may enter the systemic circulation. Alternatively, cells activated in the lung by asbestos [e.g., macrophages, dendritic cells (DCs), neutrophils, lymphocytes] may migrate to pulmonary lymph nodes, where they activate CD4+
T cells that are released into the circulation and contribute to atherogenesis. This latter explanation could account for the remarkably similar inflammatory and fibrotic patterns in the lungs of ApoE−/−
and DKO mice, but the dramatically different extent of atherosclerotic lesions in their aortas. The elevation of systemic levels of MCP-1 and the significant relationship between plasma MCP-1 concentrations and lesion size in the ApoE−/−
mice provide support for previous work showing the importance of MCP-1 in atherogenesis (Gosling et al. 1999
; Hansson 2001
). DKO mice had higher plasma MCP-1 levels than did ApoE−/−
mice, but asbestos had no effect compared with clean air in these mice. This suggests that constitutive levels may be higher in DKO mice but that higher levels are not necessarily associated with larger lesions. It appears that the induced change in concentrations is of greater importance in the pathogenesis of the disease, highlighting the complex interrelationships.
Our work also demonstrates that many classically defined inflammatory cytokines/ chemokines (e.g., KC, MIP-1α, MIP-1β,
IL-6) are elevated in BALF in this model, but these differences appear to be dissociated from effects seen in the aorta. Previous studies have revealed that MCP-1 is necessary for the activation of T lymphocytes (Taub et al. 1995
) and that MCP-1 produced by lung fibroblasts exerts an immunomodulatory effect on CD4+
T cells in vitro
(Hogaboam et al. 1998
). G-CSF and KC, both important in neutrophil recruitment, were elevated in our studies, but not GM-CSF, which is an important growth factor for lung DC maturation and proliferation (Vermaelen and Pauwels 2005
). However, because DCs are central to the initiation and orchestration of immunity and tolerance, we suggest that asbestos fibers initiate an inflammatory cascade in lung epithelial cells and macrophages, cells first encountering inhaled asbestos, thus leading to secretion of cytokines and chemokines with DC-attracting potential. DCs then could migrate to lymph nodes, where they activate T cells, which enter the systemic circulation and come into contact with a vulnerable aorta, an observation consistent with the detection of macrophages and CD4+
T cells in early aortic lesions (Zhou et al. 1996
). Further inflammatory cell infiltration would then exacerbate lesion development. Investigation is ongoing to document a role for DCs in the translation of asbestos-induced pulmonary inflammation to systemic responses.
In summary, exposure to inhaled chrysotile asbestos fibers was associated with an exacerbation of the development of atherosclerotic lesions in ApoE−/− mice. The induction of atherosclerosis appears to be CD4+ T cell dependent and dissociated from the magnitude of lung inflammation or fibrosis associated with inhaled asbestos. More important, our data also suggest the importance of a change in systemic MCP-1 in atherogenesis associated with altered signaling through the transcription factors AP-1 and NF-κB, and that the effects of inhaled asbestos extend beyond the lung. The data also raise the possibility of constitutive differences in MCP-1 levels between the different genotypes that warrant further investigation.