Cadmium Is a Novel and Independent Risk Factor for Early Atherosclerosis Mechanisms and In Vivo Relevance
Objectives— Although cadmium (Cd) is an important and common environmental pollutant and has been linked to cardiovascular diseases, little is known about its effects in initial stages of atherosclerosis.
Methods and Results— In the 195 young healthy women of the Atherosclerosis Risk Factors in Female Youngsters (ARFY) study, cadmium (Cd) level was independently associated with early atherosclerotic vessel wall thickening (intima-media thickness exceeding the 90th percentile of the distribution; multivariable OR 1.6[1.1.–2.3], P=0.016). In line, Cd-fed ApoE knockout mice yielded a significantly increased aortic plaque surface compared to controls (9.5 versus 26.0 mm2, P<0.004). In vitro results indicate that physiological doses of Cd increase vascular endothelial permeability up to 6-fold by (1) inhibition of endothelial cell proliferation, and (2) induction of a caspase-independent but Bcl-xL-inhibitable form of cell death more than 72 hours after Cd addition. Both phenomena are preceded by Cd-induced DNA strand breaks and a cellular DNA damage response. Zinc showed a potent protective effect against deleterious effects of Cd both in the in vitro and human studies.
Conclusion— Our research suggests Cd has promoting effects on early human and murine atherosclerosis, which were partly offset by high Zn concentrations.
- cadmium, zinc
- risk factor
- intima media thickness
- cell death
Since the use of Cd in manifold industrial applications, sources for and the amount of Cd uptake by humans has increased dramatically. Cd is, for example, released into the air through the burning of fossil fuels (coal, oil) and the incineration of municipal waste (Environmental Protection Agency, 2000). The most relevant sources for Cd uptake by humans are, however, cigarette smoking (one cigarette contains ≈1 to 2 μg; daily uptake of Cd ≈1 to 3 μg per pack smoked) and food for nonsmokers (daily intake ≈30 μg; daily uptake ≈1 to 3 μg), as well as exhaust gases (Agency for Toxic Substances and Disease Registry, 1999). After inhalation or ingestion of Cd, it is transferred into the bloodstream (whole blood and serum Cd concentrations range between ≈0.2 and ≈20 nmol/L1,2), where Cd is transported either as a free ion or protein-bound, eg, attached to albumin or metallothioneins. Cd is taken up by cells of Cd target organs (liver, kidneys, and testis) via solute carriers, calcium and manganese channels, and iron transporters.3–5 In 2001, Abu-Hayyeh et al6 demonstrated that the aortic vessel wall is another under-recognized target organ for Cd accumulation (aortic wall concentrations of Cd are up to 20 μmol/L). Epidemiologically, high Cd level was found to be associated with hypertension, stroke, and cardiac arrest,7–9 but confirmatory data are sparse and the mechanistic basis for these interactions remains unclear. Houtman et al observed a higher than expected frequency of atherosclerosis in a Cd-contaminated area in the Netherlands.10 Coronary arteries in Cd-exposed rabbits showed enhanced atherosclerosis,11 but the precise role of Cd in the initiation of disease remained unresolved. The first report on an interaction of Cd with endothelial permeability stems from Alsberg and Schwartze who observed a purple discolouration of testis after subcutaneous injection of Cd.12 In 1983 Sacerdote et al suggested a potential reason for this phenomenon by demonstrating that a subcutaneous injection of Cd in rats causes disruption of endothelial adherence junctions of capillaries.13 More recent findings on the effects of Cd on ECs were summarized by Prozialeck et al.14 Especially, Cd effects on cell structure and induction of cell death have been described. Structural changes like breaking down of cell-cell contacts and reorganization of intermediate fibers15,16 were ascribed to an interaction of the metal with Ve-cadherins. A variety of mechanisms underlying Cd-induced cell death have been suggested including JNK-, p38/MAPK-, p53, or bcl2 family member-dependent pathways, but the data available are not consistent and suggest a high level of cell-type specificity.17–19 Interestingly, several in vitro studies reported on the protective interaction of elements like manganese or Zn with Cd-mediated processes,19 but again CVD-relevant interactions in vivo remain largely unclear.
For details and further methods see the supplemental materials.
Association Between Intima Media Thickness and Serum Metal Concentrations in Healthy Young Females: the ARFY Study
Classical vascular risk factors, lifestyle behaviors, and family history were assessed in 195 female participants aged 18 to 22 years as detailed elsewhere,20 and characteristics of study participants are summarized in supplemental Table I (available online at http://atvb.ahajournals.org). Mean maximum IMT of the right and left common (CCA) and internal carotid artery (ICA) was quantified by high resolution B-mode ultrasound (supplemental Figure I). High IMT was predefined as exceeding the 90th percentile of the site-specific IMT distribution (CCA, ICA, or both)20,21 (supplemental Figure II). In addition, serum metal concentrations of 11 different metals including Cd were measured by induced-coupled plasma mass spectrometric analyses (ICP-MS).22 The association between Cd and high IMT was analyzed by means of multivariate logistic regression analysis adjusted for classical risk factors. Differential effects of Cd on high IMT according to Zn levels were tested by inclusion of an appropriate interaction term.
The isolation and culture of human umbilical vein ECs (HUVECs) has been described elsewhere.23
Quantification of Cell Death
The detection and quantification of cell death with the Annexin/PI method and light scatter analyses were performed as previously described.23
Lactate Dehydrogenase Release Assay
The amount of lactate dehydrogenase (LDH) released from cells was quantified using the LDH cytotoxicity kit II (Biovision) according to the manufacturer’s instructions.
Monolayer Permeability Assay
Analyses of endothelial permeability were performed on a Transwell-based assay system that was developed by our group. After the incubation times with different concentrations of Cd, endothelial permeability was determined by the amount of horseradish peroxidase permeation through the cell layer.
Detection of DNA Strand Breaks
The detection and quantification DNA strand breaks was performed with the in situ cell death detection kit, POD (Roche) according to the manufacturer’s instructions.
Western Blotting was performed as previously described.23
Caspase 3 Activity Assay
Caspase-3 activity was performed as described elsewhere.24
Analysis of the Number of Viable Cells
Quantification of the number of viable cells was done by the XTT assay (Biomol GmbH).
Treatment of Animals and Assessment of Atherosclerotic Plaque Area
Female ApoE KO mice were divided randomly into 4 groups. Group 1 received normal water; group 2 100 mg/L of CdCl2 in drinking water; group 3 400 mg/L ZnCl2; and group 4 100 mg/L CdCl2 plus 400 mg/L ZnCl2. In addition, all mice were fed a Western type diet. After 12 weeks of treatment blood samples were taken and the aorta was excised and subjected to staining and analysis of atherosclerotic plaques.
Fixation and Scanning Electron Microscopy of Mouse Aortas
After anesthesia, rinsing of the vasculature, and fixation of animals, aortas were carefully removed, tissues dehydrated and desiccated, mounted, sputtered, and examined with a Zeiss DSM 982 Gemini scanning electron microscope. The images were taken from the central nonplaque containing parts of the aortas.
SEM Analysis of Cultured Cells
For SEM analyses, cells were grown on glass coverslips, and treated as indicated. After incubation, cells were fixed by replacing the medium with 2.5% glutaraldehyde (in PBS). Dehydration, desiccation, etc were performed as described above.
Analysis of Serum Lipid Profiles
Lipoprotein profiles were analyzed by FPLC using 2 Superose-6 columns (Amersham) connected in series as described elsewhere.25
Where indicated primary data were tested for a Gaussian distribution and equality of variances. Further analyses were performed using ANOVA (Bonferroni correction for multiple comparisons), followed by pair-wise comparisons.
Elevated Serum Cadmium Levels Are Associated With an Increased Risk for High Intima Media Thickness in Healthy Young Adults
Distribution of CCA and ICA IMT in the 195 young healthy females (ARFY Study) is shown in supplemental Figure II. A total of 33 participants (16.9%) formed the “high IMT” group. Of the various metals measured, cadmium yielded an independent significant association with high IMT after adjustment for a broad array of vascular risk factors (multivariable odds ratio per standard deviation unit increment [95%CI] 1.6 [1.1.–2.3], P=0.016). Risk steadily increased over tertile groups for cadmium concentration (multivariable OR [95%CI] 1.0, 5.2[1.2 to 22.4] and 6.4[1.2 to 33.4], P=0.026 for linear trend). Of note, Zn appeared to abrogate Cd-mediated effects on IMT (Figure 1, P for interaction 0.052). No further interactions with other metals were observed. Serum Cd concentration was not associated with standard cardiovascular risk conditions. In a sensitivity analysis of the nonsmoking subpopulation (n=121), the association between Cd and high IMT remained robust: multivariable OR per standard deviation unit increment [95% CI] 1.9 [1.1 to 3.1], P=0.014) and for tertile groups (multivariable OR [95%CI] 1.0, 5.6[0.9 to 33.9] and 14.3[1.7 to 120.7], P=0.013 for linear trend).
Physiological Doses of Cadmium Increase Endothelial Permeability In Vitro: Inhibition by Zinc
To determine the impact of Cd on vascular endothelial permeability, a new transwell-based test system was set up (see supplemental Methods), and effects of Cd on endothelial permeability were analyzed after 1, 2, and 7 days, as well as after 3 weeks. The histogram in Figure 2 shows that already on a short-term basis (24 hours) endothelial permeability is significantly increased by Cd at 15 μmol/L, but also that long-term application of 1.5 μmol/L of Cd significantly increases the permeability of a vascular endothelium in culture. The left part of the histogram in Figure 2 (addition of Cd to the luminal side) compared to the right (addition to the adhesion substrate site) shows that Cd affects endothelial permeability in a cell polarity-dependent fashion. (P=0.01).
Scanning electron microscopic analyses (images A through D) revealed that the integrity of the endothelium was dramatically affected by the addition of 15 μmol/L of Cd for 96 hours. Starting with 48 hours and in good agreement with the data from the permeability assay, a progressive loss of endothelial integrity could be observed (data not shown). Zn potently prevented Cd-induced endothelial permeability. Among many elements and compounds tested, eg, metal chelators, antioxidants, metal ions (eg, manganese), and polyphenols, Zn proved to be the most powerful agent in inhibiting Cd effects (data not shown).
Cadmium Inhibits the Proliferation of Vascular ECs and Induces Cell Death
To uncover the reasons for increased endothelial permeability in response to Cd treatment, we tested for potential changes in the number of viable cells and for the induction of cell death by Cd. XTT-based time course analyses (Figure 3A) revealed that Cd reduces the number of viable cells significantly already 24 hours after the addition of 15 μmol/L or 100 μmol/L of Cd, and reaches a maximal effect between after 48 and 72 hours. Coapplication of 60 μmol/L Zn (added 24 hours before Cd addition) potently inhibited the Cd effect. Because the XTT assay measures only the number of viable cells and cannot differentiate between the inhibition of proliferation and the induction of cell death, we also tested directly for the induction of cell death by annexinV staining and FACS analyses (apoptosis) as well as by LDH-release assays (necrosis). Figure 3B shows that Cd causes a significant time-dependent increase in annexinV-positive cells not starting before 72 hours, which was slightly preceded by LDH release, arguing for the induction of a necrotic form of cell death. Figure 3C shows that contracted round cells (dying or dead cells) have holes in their membranes, suggestive of an early presence of necrotic cells in the cultures.
Cadmium-Induced DNA Strand Breaks Precede Cellular DNA Damage Response, and Caspase-Independent, Partly Calpain-Dependent, and Bcl-xL-Inhibitable Endothelial Cell Death
Figure 4A shows that the presence of Cd in the cultures significantly increases the number of cells with DNA strand breaks compared to the controls, Zn-treated cells, and Zn plus Cd-treated cells already after 12 hours of incubation. To determine the cellular response to these strand breaks, we performed Western blot-based analyses of DNA-damage response proteins ie, phospho-ATM, p53, and p21(WAF1/Cip1). Western blots (Figure 4B) show that the addition of Cd to ECs leads to the phosphorylation of ATM on serine 1981 already 24 hours after Cd addition, and to the accumulation of p53 starting at 48 hours. In addition, the p53 downstream target, cell cycle inhibitor p21(WAF1/Cip1), was also increased on the protein level after 24 and 48 hours. To test for the involvement of mitochondrial signaling in Cd-induced cell death, ECs were infected with retroviruses containing a Bcl-xL expression vector. Figure 4C shows that the overexpression of Bcl-xL significantly inhibited Cd-induced cell death, arguing for the opening of the permeability transition pore complex (PTPC) in the course of Cd-induced cell death. Although the final outcome of Cd-induced cell death was necrosis (Figure 3B and 3C), the data in Figure 4D and 4E show that apoptosis-executing enzymes ie, caspases are cleaved to their active forms, and show a small increase in activity compared to the control in response to the treatment of ECs with 15 μmol/L Cd, but also that high concentrations (100 μmol/L) inhibit the activation, and possibly also the activity of caspases. To functionally test for the relevance of caspases activity in Cd (15 μmol/L)-induced cell death, we inhibited caspases with the pan-caspase-inhibitor zVAD-fmk and performed Western blots of caspase target proteins. Figure 4C shows that the presence of zVAD had no effect at all on the extent of Cd-induced cell death. Similar results were obtained after 24, 48, 72, and 96 hours, with other caspase inhibitors (ie, DEVD-fmk, and LEHD-fmk), and with 15 and 100 μmol/L Cd (data not shown). In addition, Figure 4B shows that neither the caspase-3 target PARP, nor the capase-8 target BID are cleaved in cells treated with Cd (also note the downregulation of BID by Zn). Further inhibitor experiments (Figure 4C) revealed that the inhibition of calpains partially protects cells from Cd-induced cell death, that Zn not only inhibits the reduction in viable cell numbers by Cd (Figure 3) but also cell death induction, and that Cd-induced cell death is not dependent on PARP activity. The SEM image (Figure 4F) shows that contracting ECs have small membrane blebs with a rough surface. This finding may be interpreted as the presence of imperfect membrane blebbing.
Cadmium Causes Endothelial Damage, Atherogenic Alterations of the Lipid Profiles, and Accelerates Atherosclerotic Plaque Formation in ApoE KO Mice
Because increased endothelial permeability is a key component of our hypothesis on atherogenesis-initiation by Cd, we tested for the presence of morphological alterations of the aortic endothelium by Cd in ApoE KO mice in vivo. Figure 5 shows that the vascular endothelium of control mice (Figure 5A) has a flat surface, Cd treatment resulted in structural changes which may be interpreted as contraction and detachment of ECs ie, endothelial damage (Figure 5B). We also observed holes between ECs (Figure 5C) which were probably caused by the contraction of dying ECs, and found signs of endothelial necrosis (Figure 5D). In controls or Zn-only treated animals, no contraction or necrotic cells could be detected. The endothelium of animals that were treated with Cd plus Zn showed essentially the same pattern as Cd-only treated animals. Along with the analysis of endothelial morphology, lipid profiles of ApoE KO mice with or without Cd and with or without Zn in drinking water were analyzed (Figure 5E), and the plaque area of mouse aortas was determined (Figure 5F). The addition of Cd to the drinking water of mice changed the lipid profiles of ApoE KO mice toward an even more atherogenic profile (increased VLDL particles; VLDL at EV ≈15; LDL ≈18, HDL ≈27), and Zn reduced Cd effects. The plaque surface area increased significantly in Cd-treated mice compared to the controls (P=0.004). Although the coapplication of Zn reduced the median plaque area compared to Cd-only treatment, the changes did not reach significance. About 75% of control animals were free of atherosclerosis (AHA Grade 0), whereas about 90% of Cd-treated littermates already showed atherosclerotic wall changes (AHA Grade I through VIII) including atheroma formation (AHA Grade IV through VIII) in about 40% of the cases (P<0.001).
Although Cd has long been known for its carcinogenic and toxic activities, the role of Cd in CVD is still not clear. To address this issue we conducted a study on 195 healthy young female subjects. Epidemiological data clearly indicated that high serum levels of Cd increases the risk of high IMT, a well-established marker for early atherosclerosis. Because the average age of the study population was 20.6 years, these results suggest that Cd may play an important role in the initiation of atherosclerosis. The generally accepted response-to-injury hypothesis postulates a disruption of endothelial barrier function as the initial step in atherogenesis.26 This hypothesis is supported by our finding that Cd causes endothelial damage in vitro and in animals in vivo, where also accelerated plaque formation could be observed. Analysis of serum lipid profiles revealed an increase in VLDL particles in response to Cd treatment. Based on the relation between alterations in lipid profiles and their impact on plaque formation we calculated that in our model the altered lipid profile by Cd could only be made responsible for 20% of the Cd effect on plaque formation. In addition, no evidence for an effect of Cd on the lipid profiles in our study on human subjects could be obtained. Based on the results of this study we hypothesize that Cd primarily exerts its atherogenic activity by causing endothelial damage.
In contrast to previous reports, serum Cd levels were not associated with the smoking status of the study participants. We speculate that the low exposure of smoking individuals to cigarette smoke may account for this finding (frequency of smokers with a cumulative exposure of >3.0 pack-years was 7.8%).
The final outcome of Cd exposure to ECs is necrosis (Figure 3), but a large number of intracellular signaling processes seem to be involved, the inhibition of which abrogates Cd-induced cell death. Our in vitro data suggest that Cd, which is taken up by ECs via solute carriers or ion channels clustering on the luminal side of ECs (see Figure 2), causes DNA strand brakes. An involvement of oxidative stress, as has been suggested by others27,28 seems unlikely in our model because we were not able to detect oxidative stress by various methods (eg, 123 Di-hydro-rhodamine, H2-DCF-DA staining, Oxyblotting, and no shift in the GSH:GSSG ratio toward GSSG; data not shown). The phosphorylation of ATM on serine 1981 clearly indicates that ECs sense DNA damage and react by upregulation and stabilization of p53 and its downstream target, cell cycle inhibitor p21/WAF1/Cip1 (note the drop in the number of viable cells in the absence of cell death after 24 and 48 hours in response to Cd treatment; Figure 3A and 3B). The fact that viral overexpression of Bcl-xL29 potently inhibited Cd-induced cell death argues for the essential involvement of mitochondrial signaling in Cd-induced cell death. However, other classical apoptotic processes like caspase-activation and -activity (Figure 4D and 4E) as well as surface changes that may be interpreted as membrane blebbing (Figure 4F), although present, were not completed or are not relevant in Cd-mediated cell death. Because of the strict dependence of Cd-induced cell death on cell signaling, we hypothesize that Cd-induced cell death rather represents an atypical form of apoptosis that has the necrotic rupture of the plasma membrane as end point, than classical necrosis.
Because necrotic endothelial cells could also be found in vivo, we speculate that this process may also contribute to fatty streak and plaque formation in humans. Necrosis is known to cause inflammation and the attraction and activation of macrophages, both well known contributors to the atherosclerotic process.26,30 In addition, disruption of the endothelial barrier function may enhance lipid deposition and infiltration of the vessel wall by macrophages.
The capacity of Zn to interfere with Cd toxicity in vitro is well established in the literature,14,19,31 and our in vitro and human data clearly support this view. However, the potential consequences of an altered balance between serum Cd and Zn levels (in favor of the former) in atherogenesis have not been reported. In the human study part, the association between Cd and high IMT was confined to individuals in the low and medium Zn tertile groups and showed a dose-response effect. In brief, odds ratios [95%CI] for the low (≤200 ng/kg), medium (201 to 300 ng/kg), and high (300 ng/kg) Cd tertile group in individuals with low and medium Zn level amounted at 1.0, 2.9[0.8 to 8.3], 7.7[2.1 to 28.0] in unadjusted and 1.0, 7.4[1.3 to 41.4], 19.6[2.2 to 173.2] in multivariable analysis (Figure 1). It must be kept in mind, however, that Zn supplementation showed only a nonsignificant tendency to reduce Cd-mediated plaque formation in ApoE KO mice. The fact that Zn administration via drinking water led only to a 12% to 32% increase in serum Zn levels may explain this finding. Changes in formulation or a different route of application may improve the Zn effect.
Cd is abundantly present in our environment. A potential proatherogenic effect—even if modest compared to other traditional risk factors—has a significant impact population wide. Possible pathogenic effects of Cd on the vascular endothelium have been described and could form the basis for medical interventions. Most interestingly Zn seems to abolish the deleterious effect of Cd and might—warranting results from further studies—prove to be an effective preventive measure for people exposed to Cd.
The authors thank Prof Dr Frank Thevenod for discussions, Dr Christian Ploner for help with virus-based expression of bclXL, Dr Martin Hermann for confocal microscopy, Rajam Csordas-Iyer for correction of the English language, and Angelika Flörl and Sandra Frotschnig for excellent technical assistance.
Sources of Funding
This project was supported by the Austrian National Bank (Project #12697 to D.B.).
Received December 11, 2008; revision accepted June 10, 2009.
Dalton TP, He L, Wang B, Miller ML, Jin L, Stringer KF, Chang X, Baxter CS, Nebert DW. Identification of mouse SLC39A8 as the transporter responsible for cadmium-induced toxicity in the testis. Proc Natl Acad Sci U S A. 2005; 102: 3401–3406.
Abu-Hayyeh S, Sian M, Jones KG, Manuel A, Powell JT. Cadmium accumulation in aortas of smokers. Arterioscler Thromb Vasc Biol. 2001; 21: 863–867.
Bhatnagar A. Environmental cardiology: studying mechanistic links between pollution and heart disease. Circ Res. 2006; 99: 692–705.
Alsberg CL SE. Pharmalogical action of cadmium. J Pharmacol Exp Ther. 1919.
Prozialeck WC, Edwards JR, Nebert DW, Woods JM, Barchowsky A, Atchison WD. The vascular system as a target of metal toxicity. Toxicol Sci. 2008; 102: 207–218.
Lee WK, Abouhamed M, Thevenod F. Caspase-dependent and -independent pathways for cadmium-induced apoptosis in cultured kidney proximal tubule cells. Am J Physiol Renal Physiol. 2006; 291: F823–F832.
Szuster-Ciesielska A, Stachura A, Slotwinska M, Kaminska T, Sniezko R, Paduch R, Abramczyk D, Filar J, Kandefer-Szerszen M. The inhibitory effect of zinc on cadmium-induced cell apoptosis and reactive oxygen species (ROS) production in cell cultures. Toxicology. 2000; 145: 159–171.
Knoflach M, Kiechl S, Penz D, Zangerle A, Schmidauer C, Rossmann A, Shingh M, Spallek R, Griesmacher A, Bernhard D, Robatscher P, Buchberger W, Draxl W, Willeit J, Wick G. Cardiovascular risk factors and atherosclerosis in young women: atherosclerosis risk factors in female youngsters (ARFY study). Stroke. 2009; 40: 1063–1069.
Knoflach M, Kiechl S, Kind M, Said M, Sief R, Gisinger M, van der Zee R, Gaston H, Jarosch E, Willeit J, Wick G. Cardiovascular risk factors and atherosclerosis in young males: ARMY study (Atherosclerosis Risk-Factors in Male Youngsters). Circulation. 2003; 108: 1064–1069.
Bernhard D, Rossmann A, Henderson B, Kind M, Seubert A, Wick G. Increased serum cadmium and strontium levels in young smokers: effects on arterial endothelial cell gene transcription. Arterioscler Thromb Vasc Biol. 2006; 26: 833–838.
Bernhard D, Pfister G, Huck CW, Kind M, Salvenmoser W, Bonn GK, Wick G. Disruption of vascular endothelial homeostasis by tobacco smoke: impact on atherosclerosis. Faseb J. 2003; 17: 2302–2304.
Yang MS LD, Lin T, Zheng JJ, Zheng W, Qu JY. Increase in intracelluar free/bound NAD[P]H as a cause of Cd-induced oxidative stress in the HepG(2) cells. Toxicology. 2008; 2: 6–10.
Ploner. Role of the Bcl-2 Rheostat in Glucocorticoid-Induced Apoptosis of Acute Lymphoblastic Leukemia Cells. Innsbruck, Austria: Medical University of Innsbruck; 2006.