Toxicological Interactions among Arsenic, Cadmium, Chromium, and Lead in Human Keratinocytes

Dong-Soon Bae*, Chris Gennings{dagger}, Walter H. Carter, Jr.{dagger}, Raymond S. H. Yang* and Julie A. Campain*,1

* Quantitative and Computational Toxicology Group, Center for Environmental Toxicology and Technology, Department of Environmental Health, Colorado State University, Fort Collins, Colorado 80523; and {dagger} Department of Biostatistics, Virginia Commonwealth University, Richmond, Virginia 23298

Received February 26, 2001; accepted May 17, 2001

ABSTRACT

To evaluate health effects of chemical mixtures, such as multiple heavy metals in drinking water, we have been developing efficient and accurate hazard identification strategies. Thus, in this study, we determine the cytotoxicity of arsenic, cadmium, chromium, and lead, and characterize interactions among these metals in human epidermal keratinocytes. Three immortal keratinocyte cell lines (RHEK-1, HaCaT, and NM1) and primary keratinocytes (NHEK) were used. A statistical approach applying an additivity response surface methodology was used to test the validity of the additivity concept for a 4-metal mixture. Responses of the 4 keratinocyte strains to the metal mixture were highly dose-dependent. A growth stimulatory effect (hormesis) was observed in RHEK-1, NM1, and NHEK cells with the metal mixture at low concentrations (low ppb range). This hormesis effect was not significant in HaCaT. As the mixture concentration increased, a trend of additivity changed to synergistic cytotoxicity in all 4 cell strains. However, in NHEK, RHEK-1, and HaCaT, at the highest mixture concentrations tested, the responses to the metal mixtures were antagonistic. In NM1, no significant antagonistic interaction among the metals was observed. To explore a mechanistic basis for these differential sensitivities, levels of glutathione and metallothioneins I and II were determined in the keratinocyte cell strains. Initial data are consistent with the suggestion that synergistic cytotoxicity turned to antagonistic effects because at highest mixture exposure concentrations cellular defense mechanisms were enhanced.

Key Words: keratinocytes; toxicological interactions; additivity response surface; GSH; MT.

Both occupational and environmental exposures to hazardous metals, such as arsenic (As), cadmium (Cd), chromium (Cr), and lead (Pb), are significant toxicological concerns. Not only do these metals lead to acute toxicity at higher concentrations, but they may also mediate development of additional pathologic conditions in individuals exposed chronically to low levels. Environmentally relevant metals seldom occur alone. Rather, they most often occur in hazardous waste sites or ground water supplies in combination with other contaminants; this substantially complicates the risk assessment process for these chemicals. In general, within the scientific community, the concept of "additivity" is assumed for low level exposures to the component chemicals in a mixture (Svendsgaard and Hertzberg, 1994Go). The definition of additivity used here is that described by Berenbaum (1985, 1989). Whether this is a valid assumption for metal mixtures needs to be tested. This project is, therefore, an attempt to test such an additivity concept at low exposure levels. The metals chosen for our studies are highly relevant to human exposure. As, Cd, Cr, and Pb, are the top 4 metals in site frequency count by the ATSDR Completed Exposure Pathway Site Count Report (ATSDR, 1997Go); 3 of these, As, Pb, and Cd are among the Superfund Top 10 Priority Hazardous Substances (DeRosa et al., 1996Go), i.e., those considered to pose the greatest hazard to human health. In addition, as confirmed by ATSDR using the HazDat database, these metals most often occur together; they are present in 8 of 10 and 5 of 10 of the top 10 Binary Combinations of Contaminants in soil and water, respectively (Fay and Mumtaz, 1996Go).

As part of a larger study, we describe here an evaluation of the acute cytotoxicity of As, Cd, Cr, and Pb alone and together in a mixture in human epidermal keratinocytes, a highly relevant cell type. The skin is a critical target organ for As-mediated pathological effects, including proliferative disorders such as hyperkeratosis and, in many cases, carcinogenesis. Due to high As (and other metal) concentrations in many drinking water supplies, acute toxicity and development of neoplastic skin lesions have become health problems of global proportions. Both As and Cr, a well-known skin sensitizer, have substantial effects on epidermal keratinocytes in vitro, altering expression of several growth regulatory factors and inhibiting the normal process of differentiation (Cohen et al., 1993Go; Germolec et al., 1996Go, 1997Go; Kachinskas et al., 1997Go; Ye et al., 1995Go; Yen et al., 1996Go). Little is known, however, about the exact mechanism of toxicity in vivo of either metal alone or the effects of the 2 metals when they are present together or in complex mixtures with other metals on this cell type. To address this issue, an additivity response surface methodology was utilized to permit detailed statistical analysis of the cytotoxic interactions among As, Cd, Cr, and Pb when present together in a mixture. In addition, mechanistic studies are described in which the roles of glutathione (GSH) and metallothionein-I and -II (hMT) in the observed metal-metal interactions are explored.

MATERIALS AND METHODS

Chemicals.
Sodium metaarsenite (NaAsO2), cadmium chloride (CdCl2), chromium oxide (CrO3), chromium chloride (CrCl3), lead acetate ((C2H3O2)2Pb.3H2O), dimethyl sulphoxide (DMSO), 3-[4,5-Dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT), aprotinin, Nonidet P-40, and phenylmethanesulfonyl flouride (PMSF) were purchased from Sigma Chemical Co. (St. Louis, MO).

Cell lines and culture reagents.
Cryopreserved normal human epidermal keratinocytes (NHEK) were purchased from the Clonetics Corp. (San Diego, CA). NHEK cells were grown in defined Keratinocyte Growth Medium-KGMTM containing bovine pituitary extract (BPE), human epidermal growth factor (hEGF), insulin, hydrocortisone, transferrin, epinephrine, and an antibiotic, GA-1000 (Clonetics Corp.). Spontaneously immortalized HaCaT and NM1 keratinocyte cell lines were obtained from Dr. N. Fusenig, German Cancer Research Center (Heidelberg, Germany; Boukamp et al., 1988) and Dr. J. Kubilus, Mattek Corp. (Ashland, MA; Baden et al., 1987), respectively. The AD12/SV40 immortalized keratinocyte cell line (RHEK-1) was obtained from Dr. J. Rhim (Center for Prostate Disease Research, Rockville, MD; Rhim et al., 1985; Yang et al., 1992). Each of the keratinocyte cell strains used in our studies was isolated from different individuals; NHEK is, in fact, a pooled population of cells from multiple individuals. Early passage NIH3T3 cells were obtained from American Type Culture Collection (Manassas, VA). RHEK-1 and HaCaT were cultured in Dulbecco's Modified Eagle's medium (DMEM) supplemented with 100 U/ml penicillin, 0.1 mg/ml streptomycin, 10 mM L-glutamine, and 10% fetal bovine serum (FBS; Summit Biotechnology, Ft. Collins, CO). The NM1 cell line was cultivated in DMEM supplemented with 0.4 g/ml hydrocortisone, 10 ng/ml epidermal growth factor (EGF), 1 nM cholera toxin, 20% FBS, 100 U/ml penicillin, 0.1 mg/ml streptomycin, 10 mM L-glutamine, and 0.2 mM CaCl2 on an NIH3T3 feeder layer as described by Rheinwald and Green (1975, 1977). NIH3T3 cells for feeder layers were lethally irradiated with an MK-1-68A Mark 1 Cs-137 sealed source cabinet irradiator at 1466 rad/min (J. L. Shepard and Associates, Glendale, CA) and replated 24 h prior to use. For co-cultures, irradiated feeder cells were plated at a 1:3 ratio with NM1 cells. Feeder layers were removed from co-cultures prior to trypsinization of NM1 with a vigorous rinse of the flasks with 0.02 % EDTA. All keratinocyte strains were grown at 37°C in a humidified atmosphere of 5% CO2. Hydrocortisone and EGF were purchased from Sigma Chemical Co. (St. Louis, MO). Cholera toxin was purchased from Gibco BRL (Grand Island, NY).

Killing curves with individual metals or the metal mixture.
For toxicity studies, NHEK was plated at 2.4 x 104 cells/well in 6-well plastic tissue culture plates and grown for 24 h before metal treatment. RHEK-1 and HaCaT cell lines were plated at 1 x 104 cells/well. NM1 was plated at a density of 1 x 104 cells/well with 3 x 103 irradiated NIH3T3 feeder cells. Triplicate wells of cells were treated with increasing concentrations of As, Cr, Cd, Pb, or a mixture of the 4 metals for 24 h. Metal stock solutions were prepared in deionized distilled water and sterilized by filtration through 0.2 µm filters. Concentrations used for As and Cr were 0.3, 1, 3, 10, and 100 µM. The Cr stock solution was made by mixing a 1:1 ratio of trivalent and hexavalent chromium. Cd and Pb were administered at 3, 10, 30, 100, and 300 µM. To prepare the 4 metal mixtures, 50X stock solutions were made for the 4 individual cell lines that when diluted to their final 1X concentrations in cell cultures, contained the levels of each As, Cr, and Cd giving 50% cell killing in individual cytotoxicity assays (LD50). The concentrations of these 3 metals in the 1X mixture solutions for the 4 cell types were: 6.1 µM As, 7.1 µM Cr, 14 µM Cd for RHEK-1; 4.8 µM As, 6.8 µM Cr, 56 µM Cd for HaCaT; 9.0 µM As, 5.3 µM Cr, 55 µM Cd for NM1; 7.7 µM As, 4.9 µM Cr, 6.1 µM Cd for NHEK. Lead acetate concentration in these 1X solutions was 100 µM, as we were unable to get complete killing at any dose in HaCaT, NM1, and NHEK. Lead acetate concentration in 1X solution for RHEK-1 was 120 µM, the estimated LD50. To generate dose-response data, the mixture stock solutions were serially diluted 1:3 to get 0.333, 0.111, 0.037, 0.0123, 0.004, and 0.0014X dilution groups as a final concentration in cell cultures. Deionized distilled water was used as the vehicle control in all cases. After 24-h exposure to the individual metals or the 4-metal mixture, cells were refed with fresh metal-free medium and incubated for 3 days prior to viability analysis by the MTT assay (Mossman, 1983Go). At least 3 independent experiments were conducted in each of the 4 cell strains for individual metal and the metal mixture cytotoxicity curves.

MTT assay.
The MTT assay was carried out using a modification of the method of Mossman (1983). MTT was dissolved at 5 mg/ml in phosphate-buffered saline. This stock solution was filtered through a 0.2 µm filter and stored at 4°C for up to 2 weeks. Immediately before use, the stock solution was diluted to 0.5 mg/ml with serum-free culture medium to make a working solution. Medium was aspirated from cells to be analyzed and MTT working solution (1 ml) was added to each well. Cells were incubated at 37°C for 3 h, after which time, the MTT was removed by aspiration. Cells were subsequently lysed by addition of 0.5 ml of DMSO. The absorbance at 550 nm of samples as well as a DMSO control was read on a Microplate Autoreader (Bio-Tek Instruments, Inc., Winooski, VT). The absorbance of the DMSO blank was subtracted from all values. All absorbance values were expressed as a percent of water vehicle control. Absorbance measured in the MTT assay was converted to percent cell viability and analyzed by Minitab 11 (State College, PA) and SigmaPlot 4.0 (Chicago, IL) for determination of LD50 values.

Statistical analyses on mixture interaction.
The fundamental definition of additivity used in the construction of the nonlinear additivity model (fadd(x)) is that chemicals that combine additively do not change the slope of other chemicals in the mixture. That is, if the slope of chemical A changes in the presence of chemical B then the chemicals are said to interact. If the slope of chemical A does not change in the presence of chemical B then the chemicals are said to combine additively. Further, the nonlinear additivity model can be algebraically expressed in terms of Berenbaum's interaction index conditioning on the maximum and minimum responses (parameterized with {gamma} and {alpha}). It can thus be claimed that the nonlinear additivity model is consistent with dose addition as defined by Berenbaum's interaction index (Berenbaum, 1985Go, 1989Go). On the other hand, one can manipulate the response to a metameter that changes the nonlinear model to a linear model (here, -log(-log((y – {alpha})/{gamma})), which is called the complementary log-log transformation). On this response metameter, one can think of the model as being associated with response addition since the added effect of the ith chemical is ßixi. So, the additivity model used in the analysis is consistent with several concepts of additivity or zero interaction. The claim that "greater than additivity" implies synergism and "less than additivity" implies antagonism is based on the demonstration that the additivity model is (conditionally) related to Berenbaum's interaction index that makes the claim of synergism/antagonism depending on whether the index is less/greater than one.

Single chemical data were experimentally observed to support the estimation of an additivity surface (Berenbaum, 1985Go, 1989Go; Gennings and Carter, 1995Go; Gennings et al., 1997Go, 2001Go) to the single metal data. Separate analyses were conducted on the data for each cell line. At least 3 independent experiments were conducted and analyzed for each of the 4 cell lines. The endpoint of interest was percent viability. For the analyses considered here, the data from the individual experiments were grouped together and an experiment effect was not considered.

Seven mixture (M = 7) points were observed with a specified mixing ratio (here, based on the LD50s) with nm observations at each and M* = total mixture observations. Let µm be the mean at the mth mixture point, and µadd,m be the mean under additivity. The hypothesis of no interaction is equivalent to {delta}m = µmadd,m = 0, m = 1,..., M. Assume µadd,m = fadd(xm; {theta}) where Var(Y) = {sigma}2. Gennings et al. (2001) develop a test of the hypothesis {Delta} = [{delta}1,..., {delta}M]` = 0` based on a Wald-type test.

which follows an FM,N – p. Let fadd(x; {theta}1) be the additivity model with unknown parameter vector {theta}1; let fm(x; {theta}2) be the model along the fixed ratio ray using total dose for x and unknown parameters {theta}2. Suppose it is possible to express the parameter vector {theta}2 as a function of parameters in {theta}1 under the hypothesis of additivity, i.e., {lambda} = a({theta}1) – {theta}2 = 0. Then for large samples,

which follows an Fp – q, N-p + q in a test comparing the two curves. Gennings et al. (2001) also use a 100 (1 – {alpha})% simultaneous confidence band on the difference between the curves. These methods are illustrated in the analysis of 4 metals (As, Cd, Cr, and Pb) where the endpoint of interest is percent viability of treated NHEK. The mixing ratio is the LD50 ratio of the 4 metals. Seven dilutions along this ray were experimentally observed. The additivity model was

and the model for the mixture data along the fixed-ratio ray was

So, {theta}1 = [{alpha}, {gamma}, ß0, ß1, ß2, ß3, ß4]` and {theta}2 = [{alpha}*, {gamma}*, ß*0, ß*1, ß*2]`

Under additivity, {lambda} = [{alpha} - {alpha}*, {gamma} - {gamma}*, ß0 - ß*0, {Sigma}ßi - ß*1, ß*2]`= 0`. Unknown parameters were estimated using PROC NLIN in SAS (Cary, NC). Single chemical data were used to estimate fadd(x) and mixture data were used to estimate fM(x). Details of the methods are provided in Gennings et al. (2001).

Western blot for MT protein.
Cell homogenates were prepared by sonication of cells in 600 µl of ice-cold 50 mM Tris–HCl (pH 8.0), 150 mM NaCl, 0.02% sodium azide, 100 µg/ml PMSF, 1 µg/ml aprotinin, and 1% NP-40. Homogenates were clarified by centrifugation at 20,000 x g for 45 min at 4°C. Total protein concentration was determined using the BCA (Bio-Rad, Hercules, CA) assay. Samples (50 µg of total protein) from the 4 keratinocyte strains treated for 24 h with increasing concentrations of the 4-metal mixture were analyzed for human MT proteins using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS–PAGE; Laemmli, 1970) in 10–20% gradient gels. Proteins were transferred electrophoretically to nitrocellulose membranes. The resulting membranes were incubated in 2.5% glutaraldehyde for 1 h and then washed 3 times for 5 min in phosphate buffer (8.1 mM Na2HPO4, 1.2 mM KH2PO4, 2.7 mM KCl, pH 7.4). Monoethanolamine (50 mM) was added to the third wash solution to quench residual glutaraldehyde reactivity. MT proteins were detected by Immun-Star Chemiluminescent Protein Detection Systems (Bio-Rad, Hercules, CA). A monoclonal antibody to polymerized equine renal MT-1 and MT-2 (Dako Corp., Carpinteria, CA) was used for immunodetection. The chemiluminescent images were quantified using a Nikon AF camera and the NIH Image 1.55 b20 software (NIH, Bethesda, MD). In NM1, where we were unable to detect significant levels of MT expression, overall protein content and quality in cell lysates were confirmed by probing for cyclin G1 (Horne et al., 1996Go) expression using a rabbit polyclonal antibody from Santa Cruz Biotechnology (Santa Cruz, CA). Data are expressed as mean ± SEM. Statistical comparisons were performed using one-way ANOVA (Dunnett's multiple comparison test) (Tamhane and Dunlop, 2000Go). Values were considered to be significant when p < 0.05 (Minitab, Inc., State College, PA).

GSH assay.
A colorimetric assay was carried out to quantify intracellular levels of GSH (Anderson, 1989Go). Cell homogenates were prepared as described above except that the cell suspension was centrifuged at 3000 x g. The resulting cleared lysates were used for analysis. GSH concentrations in 100 µg samples from the 4 keratinocyte strains treated with increasing concentrations of the metal mixture were measured by BIOXYTECH GSH-400 as per the manufacturer's directions (OXIS International, Inc., Portland, OR). The concentration range of exposure in the 4 metal mixtures were: 0.08–6.1 µM As, 0.17–14 µM Cd, 0.09–7.1 µM Cr, and 1.48–120 µM Pb for RHEK-1; 0.06–4.8 µM As, 0.69–56 µM Cd, 0.08–6.8 µM Cr, and 1.23–100 µM Pb for HaCaT; 0.11–9.0 µM As, 0.67–55 µM Cd, 0.065–5.3 µM Cr, and 1.23–100 µM Pb for NM1; and 0.095–7.7 µM As, 0.075–6.1 µM Cd, 0.060–4.9 µM Cr, and 1.23–100 µM Pb for NHEK, corresponding to 0.0123 to 1X mixture dilutions in all cell lines. Data are expressed as mean ± SEM. Statistical comparisons were performed using one-way ANOVA (Dunnett's multiple comparison test; Tamhane and Dunlop, 2000). Values were considered to be significant when p < 0.05 (Minitab, Inc., State College, PA).

RESULTS

Human Keratinocytes Are Differentially Sensitive to As, Cr, Cd, and Pb
To characterize the effects of As, Cr, Cd, and Pb individually on human keratinocytes, we performed cytotoxicity studies in primary human epidermal keratinocytes (NHEK) and 3 immortalized human keratinocyte cell lines (RHEK-1, HaCaT, and NM1) using the MTT assay as a measure of cell viability. As would be expected, all 4 metals showed a dose-dependent cytotoxic effect, as expressed by decreased absorbance values of treated cells. The mean LD50 values for As, Cr, and Cd were 6.1, 7.1, and 14 µM for RHEK-1; 4.8, 6.8, and 56 µM for HaCaT; 9.0, 5.3, and 55 µM for NM1; and 7.7, 4.9, and 6.1 µM for NHEK, respectively. It can be seen from this data that the 4 keratinocyte strains showed similar sensitivities to As and Cr. In contrast, substantial differences in sensitivity to Cd were observed among the 4 cell strains. The LD50 value for Pb in RHEK-1 was determined to be 120 µM. However, Pb toxicity in NHEK, HaCaT, and NM1 was so low, we were unable to accurately determine the LD50 concentrations. In NM1, we were unable to get 50% lethality in cultures treated with Pb concentrations as high as 300 µM.

Toxicological Interactions among As, Cr, Cd, and Pb Are Present in Human Keratinocytes
To characterize potential cytotoxic interactions among As, Cd, Cr, and Pb in human keratinocytes, cell killing curves were generated for mixtures of the 4 metals. The approximate concentrations of As, Cr, Cd, and Pb in the mixture at the LD50 in each of the 4 keratinocyte strains were respectively, as follows: 2.1, 2.4, 4.7, and 40 µM in RHEK-1; 2.7, 3.8, 31, and 56 µM in HaCaT; 2.2, 1.3, 14, and 25 µM in NM1; and 4.8, 3.0, 3.8, and 62 µM in NHEK. Subsequently, thorough statistical analyses were carried out on the data from these cytotoxicity assays for individual metals as compared to the 4-metal mixture. An approach, developed by Carter and Gennings (Gennings and Carter, 1995Go; Gennings et al., 1997Go), using an additivity model permits testing the hypothesis that chemicals in a mixture act in an additive fashion. This model is described in detail in Gennings et al. (2001). Separate statistical analyses were conducted on the data from each keratinocyte strain.

Overall, the fit of the data seems adequate for the models associated with the 4 cell strains. The estimated model parameters as described in Materials and Methods and their associated sample standard errors and p values in 4 keratinocyte strains are provided in Table 1Go. All of the model parameters were significantly different from 0. The slope parameters associated with each of the 4 metals were negative and significant in all 4 models. This indicates that, as expected, as the concentration of the metal increases, the % viability decreases significantly regardless of the cell strain.


View this table:
[in this window]
[in a new window]
 
TABLE 1 Estimated Cell-Specific Parameters for the Additivity Model
 
The additivity model was fit to the single metal data to generate prediction intervals for each metal mixture group. Even though we are dealing with single chemical dose response curves, the purpose for using the additivity model was to generate proper {alpha}, ß, and {gamma} parameter values to be used in the statistical modeling of the chemical mixtures. A representative plot of the observed and predicted responses under additivity to the individual metals in RHEK-1 is shown in Figure 1Go. It can be seen from this figure that the individual chemical data points cluster in a band along the identity line, indicating an adequate fit of the single chemical data using the additivity model. The data from each of the other 3 keratinocyte cell lines, although not shown, also showed adequate fit. More variability was associated with the responses in NHEK as compared to the other 3 cell lines (Gennings et al., 2001Go); this may well be a function of the fact that NHEK are pooled populations of primary cells isolated from different individuals with intrinsic variations in sensitivity to toxic insult. Representative dose-response curves for the individual metals in RHEK-1 are graphically shown in Figure 2Go. The best-fit lines determined for the single chemical data in each cell strain were used to derive parameters {alpha}, ß, and {gamma} for the additivity model as described in Materials and Methods and in Gennings et al. (2001). The single chemical data demonstrated that the responses associated with exposure to lead have more variability than the other 3 metals, with levels of lead at 300 µM and percent viability between 30 and 60% for the 4 cell strains (data not shown).



View larger version (13K):
[in this window]
[in a new window]
 
FIG. 1. Observed versus predicted responses for the fit of the single metal data in RHEK-1. The asterisks (*) represent the individual data points plotted in a graph of the observed versus model predicted responses. The line is the identity line (y = x) so that a cluster of points in a band along this line indicates a reasonable fit of the data.

 


View larger version (17K):
[in this window]
[in a new window]
 
FIG. 2. Cytotoxicity of individual metals in RHEK-1. Exposure of RHEK-1 to increasing concentrations of As, Cr, Cd, and Pb and analysis of cell viability by the MTT assay were carried out as described in Materials and Methods. Three independent experiments with triplicate data points were run for RHEK-1. Individual data points measuring cell viability are plotted as a function of metal concentration (*). The line is the best-fit line. Single metal data was used to estimate the {alpha}, ß, and {gamma} parameters. These parameters were subsequently used in the model to predict responses to serial dilutions of a 4-metal mixture under the assumption of additivity.

 
The observed responses in the keratinocyte strains at the 7 mixture dilution points are provided in Tables 2A–2DGo. These tables also provided the predicted responses under the hypothesis of additivity. Based on the comparison of the observed responses and the prediction intervals under additivity, there appears to be interactions among the 4 metals with regard to cytotoxicity for all 4 cell strains. The corresponding prediction intervals (Table 2Go) can be used to determine the direction of the interaction. In the RHEK-1, NM1, and NHEK cell strains the lowest concentration of the metal mixture was associated with growth stimulation (Tables 2A, 2C, and 2DGo). This may be due to hormesis (Calabrese, 1997Go; Stebbing, 1982Go). The quadratic terms were initially included in the additivity model for NHEK to see if the single chemical data also demonstrated this "hormesis-like" effect. However, in this case, the quadratic terms were not jointly significant (p value > 0.05) and were removed from the final model. This indicates that the single chemical data do not support the occurrence of hormesis at the concentrations tested and growth stimulation was, thus, a function of binary or higher metal combinations. It is interesting to consider a hormesis effect of a mixture, but not the individual components. Higher concentrations (e.g., 0.037X through 0.333X dilutions) of the mixtures in the NM1 and NHEK cell strains were associated with significant departure from additivity (Tables 2C and 2DGo). The observed responses were synergistic or more extreme (i.e., further down the concentration effect curve) than that predicted under additivity. In NHEK, the measured cytotoxicity at the highest mixture concentration (1X) was less extreme than that predicted by the additivity model, i.e., was antagonistic. Moderate concentrations of the mixture in the HaCaT cell line (i.e., 0.0123–0.111X) were also associated with synergistic cytotoxicities (Table 2BGo). However, at the 2 highest concentrations (0.333 and 1X dilutions) tested in HaCaT, an antagonistic response was seen. In a similar trend, the 0.111X mixture dilution in the RHEK-1 cell line (0.111X) was associated with synergistic response, while the 0.333X dilution showed antagonistic responses. Only NM1 did not demonstrate a significant antagonistic response to the metal mixture at any concentration. Figures 3A–3DGo show the predicted response curve under the hypothesis of additivity and the observed response at the 7 mixture points in the 4 cell strains. The departures from additivity are clearly seen for each cell line in this figure.


View this table:
[in this window]
[in a new window]
 
TABLE 2 Observed Mean % Viability and Predicted Responses under the Hypothesis of Additivity in Keratinocytes
 


View larger version (18K):
[in this window]
[in a new window]
 
FIG. 3. Concentration effect curves along total concentration for the fixed ratio mixture ray. The solid line is the predicted response under the additivity model and the asterisks represent the observed sample mean responses at each of the mixture points in (A) RHEK-1, (B) HaCaT, (C) NM1, and (D) NHEK.

 
Potential Role for Reduced Glutathione and Metallothionein in the Toxicological Interactions of As, Cd, Cr, and Pb
As indicated above, the observed interactions among As, Cd, Cr, and Pb in RHEK-1, HaCaT, and NHEK switched from synergistic to antagonistic as the metal mixture increased from the 0.111X to 0.333X or 1X dilutions. We were interested in exploring potential mechanisms for this change and thus, analyzed levels of the important detoxifying molecules, GSH and hMT-I and hMT-II proteins in the keratinocyte strains treated with the metal mixture using a standard colorimetric assay and immunoblot analysis, respectively. Alterations in the levels of one or more of these molecules have been shown in other systems to modulate toxicity to metals. Table 3Go shows the levels of GSH in RHEK-1, HaCaT, NM1, and NHEK cells treated with increasing concentrations of the 4-metal mixture for 24 h. Interestingly, this analysis showed that in HaCaT and RHEK-1, the amount of intracellular GSH was fairly stable throughout the treatment range, except for at the 0.333X dilution, where the level increased approximately 2-fold over the control and preceding 0.111X treatment group. Only in RHEK-1 did the significant increase in GSH also occur in cells treated with the undiluted (1X) metal mixture; in HaCaT, GSH returned to control levels in cells treated with the 1X solution. In NM1 and NHEK, GSH was also elevated in response to treatment with the metal mixture; the highest level of GSH was measured in cells exposed to the 1X mixture concentration (4- and > 5-fold over untreated control cells for NM1 and NHEK, respectively). Levels of GSH were still 1.6-fold over control in NHEK cells treated with the 0.3X dilution and returned to control values at lower mixture concentrations. In contrast to the single peak in GSH in the other 3 keratinocyte cell strains, in NM1 2 peaks were consistently observed; approximately 2-fold elevations in GSH were also detected in cells treated with the lowest mixture concentration at 0.0123X. The differences in GSH were statistically significant as analyzed by Dunnett's multiple comparison analysis (p < 0.05; Tamhane and Dunlop, 2000). In contrast to our results with GSH, where in most cases, there appeared to be a fairly high threshold for induction, total MT levels in RHEK-1, HaCaT, and NHEK cells increased in a dose-dependent manner with increasing concentration of the metal mixture. As shown in Figure 4Go and Table 4Go, the highest levels of induction in the 3 cell strains was observed at the 1X dilution point, 4.0-, 5.3-, and 4.0-fold for RHEK-1, HaCaT, and NHEK respectively. In contrast to these 3 cell strains, untreated NM1 showed only a very weak to undetectable signal corresponding to hMT-I/-II. In addition, there was no significant induction of the proteins upon treatment of the cells with the metal mixture at any concentration.


View this table:
[in this window]
[in a new window]
 
TABLE 3 Levels of Reduced Glutathione in Human Keratinocytes Treated with the 4-Metal Mixture
 


View larger version (48K):
[in this window]
[in a new window]
 
FIG. 4. Human metallothionein (MT) expression in keratinocytes exposed to a 4-metal mixture. Total cellular protein (50 µg) from mixture treated and control RHEK-1 (A), HaCaT (B), NM1 (C), and NHEK (D) cells was analyzed by Western blotting for MT expression as described in Materials and Methods. Signal strength was quantified using NIH Image 1.55 b20. Equine renal MT-I was utilized as a positive control and runs at approximately 13 kDa (lane 1). The human MT isoforms migrate at a similar size range 13.5 kDa. Samples 2–7 are from cells exposed to: water control (lane 2), 1X (lane 3), 0.333X (lane 4), 0.111X (lane 5), 0.037X (lane 6), and 0.0123X (lane 7) dilutions of the metal mixture. Constitutive expression of cyclin G1 in NM1 cells was shown to confirm the overall protein content and quality in cell lysates.

 

View this table:
[in this window]
[in a new window]
 
TABLE 4 Human Metallothionein (hMT-I and -II) Protein Levels in Human Keratinocytes Treated with the 4-Metal Mixture
 
DISCUSSION

It is generally assumed that the concept of additivity is operative on low-level exposures to chemical mixtures (Svendsgaard and Hertzberg, 1994Go). Our studies are aimed at testing this hypothesis for environmentally relevant metal mixtures. The basic approach of the analysis conducted on cytotoxicity data includes fitting an additivity model to the single metal data (Berenbaum 1985Go, 1989Go; Gennings and Carter, 1995Go; Gennings et al., 1997Go). Using this model, comparisons are then made between the observed responses at mixture points of interest as compared to the predicted response under the additivity model. If significant departure from additivity is found, then an interaction can be claimed at the mixture levels tested. In our studies, all 3 types of responses to the metal mixture were seen (i.e., additivity, synergism, or antagonism), depending highly on both the keratinocyte cell line examined and the dose of the metal mixture. With some exceptions, the general trend of cytotoxicity by the 4-metal mixture appears to be hormesis to additivity to synergism and finally to antagonism as dose level increases. In that sense the additivity concept holds true for a narrow dose range.

In our studies 3 of the 4 lines, RHEK-1, NM1, and NHEK showed increased growth when exposed to very low concentrations (low ppb range) of the 4-metal mixture. One possible explanation for this observation is a hormesis effect. Much information has become available on the cellular and molecular basis of hormesis through studies on the biological effects of low levels of exposure to single chemicals or combinations of chemicals (Calabrese, 1997Go; Calabrese and Baldwin, 1997aGo,bGo; Mehendale, 1994Go; Stebbing, 1982Go, 1997Go). Hormesis may be the cumulative consequence of transient and sustained "overcompensation" by homeostatic mechanisms in response to low levels of inhibitory challenge (Calabrese and Baldwin, 1999Go; Stebbing, 1997Go). There is precedent for a hormetic effect of As on keratinocytes. Germolec et al. (1996, 1997) demonstrated that very low levels (0.001–0.005 µM) of As induced proliferation in primary NHEK. Although in our studies, we did not see growth stimulation in the keratinocyte strains with any of the metals alone, the concentrations tested were in a higher range (0.3 to 300 µM) than in these previous studies. In contrast, in the mixture dilutions where the hormetic effect was apparent in our studies, the concentration of As was 0.010–0.013 µM, depending on the cell strain; our finding of growth stimulation in these populations was, therefore, consistent with at least a partial effect of As. Whether or not the other 3 metals contributed is currently unknown. Of the 4 keratinocyte cell lines, only HaCaT did not show signs of hormesis at any concentration of the chemical mixture examined; it is possible that this keratinocyte strain displays differential induction of and/or sensitivity to growth regulatory molecules involved in the hormetic response as compared to RHEK-1, NM1, and NHEK.

Synergistic cytotoxicities of the 4-metal mixture were observed in all 4 cell types, albeit at different dose levels. Antagonistic responses were seen in RHEK-1, HaCaT, and NHEK at high treatment concentrations. Particularly interesting was the abrupt switch in RHEK-1 and HaCaT from synergistic to antagonistic interactions among the metals at the highest mixture concentrations. There are many types of metal-metal interactions that may be responsible for the synergism or antagonism we observed. Among these would be alterations in detoxifying or metabolizing pathways. To address this issue, we have explored expression of multiple molecules potentially involved in detoxification of metals, hMT-I and -II and GSH, in the keratinocyte cell lines treated with the metal mixture. One of the primary mechanisms conferring resistance to Cd is overexpression of the members of the metallothionein (MT) gene family; these proteins may be involved in protection against metal-induced toxicity. The MT genes are differentially regulated in response to heavy metals, cytokines, and reactive oxygen species (ROS). One possibility is that metals such as As or Cr increase expression of one or more of the MT genes indirectly through cytokine induction or generation of ROS, thus, having substantial impacts on Cd toxicity. The induction of MT expression by As, as well as this type of interactive effect, has been observed in numerous other studies (Albores et al., 1992Go; Hochadel and Waalkes, 1997Go; Kreppel et al., 1993Go; Liu et al., 2000bGo; Zhao et al., 1997Go). In addition, in some instances (although not all) MT expression has been correlated with decreased sensitivity to the cytotoxic effects of chronic As exposure (Liu et al., 2000aGo). As has also been shown to induce other genes such as those the encoding heat shock proteins (HSP), which may be involved in detoxification of As, itself, and Cd (Huot et al., 1991Go; Li, 1983Go; Wu and Welsh, 1996Go). Increased expression of HSP27, in turn, correlates with increased total cellular GSH (Mehlen et al., 1996Go). It has been demonstrated that the cellular levels of GSH and glutathione-S-transferase-pi (GST-pi) play important roles in resistance to As and other metals (Huang and Lee, 1996Go; Lee et al., 1989Go; Rosen, 1995Go; Shimizu et al., 1998Go; Wang, 1993). It is evident from these and other studies, that merely by alteration in expression levels of detoxifying molecules such as either MT or GSH, 1 metal may have substantial impact on the resultant toxicity of another when the 2 are present together in a mixture.

Several interesting findings resulted from our analysis of MT and GSH levels in the 4 keratinocyte strains. Somewhat contrary to our expectations, basal MT-IA levels in the 4 cell strains did not correlate with the differential sensitivity to Cd observed in our single metal cytotoxicity studies. However, these studies did suggest that either or both GSH or MT may play a role in the antagonistic interactions observed in cells treated with high mixture concentrations. The 2-fold increase in GSH at just the point where synergistic interactions become antagonistic in RHEK-1 and HaCaT would be consistent with this hypothesis. Additionally, in NHEK, GSH levels were highly elevated at the mixture concentration (1X) where antagonistic interactions were observed. In contrast to what we observed with GSH, MT proteins appear to be induced in a dose-dependent fashion with increasing mixture concentrations in 3 of the 4 cell strains. However, it may be that there is a critical intracellular level of MT required for detectable protection against the toxicity of 1 or more of the metals. Alternatively, both increased levels of GSH and MT may be required (Li et al., 1994Go; Susanto et al., 1998Go). Our inability to measure significant basal amounts or induction of hMT-I or -II in NM1 and to detect antagonistic interactions among the metals in the mixture at any concentration supports these hypotheses. It will be interesting to carry out these same types of studies on metal-metal interactions in the presence of the GSH-depleting agent, L-buthionine sulfoximine (BSO).

In summary, our studies are relevant and important to risk assessment of toxic metals in several regards. Our findings, via statistical analysis, that multiple heavy metals in a mixture do not necessarily act in an additive fashion at low doses as is commonly assumed is highly relevant in terms of developing accurate risk assessment strategies for these important environmental contaminants. The nature of the interaction between component metals in a mixture is extremely complicated, being both cell strain- and dose-dependent. Our single metal toxicity data supported these findings in that, for Cd and Pb, there are quite substantial differences in sensitivity among the same cell type isolated from different individuals. These findings suggest the involvement of a strong genetic component in susceptibility to the toxic effects of diverse chemicals. Under many conditions in each of the cell strains, we observed synergistic cytotoxicity of the mixture, clearly important findings for assessment of health effects in exposed populations. Given the complexity of chemical-biological interactions in the cells, these results again emphasize the need for computer technology and biologically based modeling in future risk assessment strategies of chemical mixtures.

ACKNOWLEDGMENTS

This study was supported by the Agency for Toxic Substances and Disease Registry (ATSDR) Cooperative Agreement U61/ATU881475, and the National Institute for Environmental Health Sciences (NIEHS) Superfund Basic Research Program Project P42 ES05949. The efforts of many colleagues at the Center for Environmental Toxicology and Technology at Colorado State University are gratefully acknowledged.

NOTES

1 To whom correspondence should be addressed. Fax: (970) 491-8304. E-mail: julie.campain{at}colostate.edu. Back

REFERENCES

Albores, A., Koropatnick, J., Cherian, M. G., and Zelazowski, A. J. (1992). Arsenic induces and enhances rat hepatic metallothionein production in vivo. Chem. Biol. Interact. 85, 127–140.[ISI][Medline]

Anderson, M. E. (1989). Enzymatic and chemical methods for determination of glutathione. In Glutathione (D. Dolphin, O. Avramovic, and R. Poulson, Eds.), pp. 339–365. John Wiley & Sons, New York.

ATSDR (1997). 1997 CERCLA Priority List of Hazardous Substances That Will be the Subjects of Toxicological Profiles & Support Document, Agency for Toxic Substances and Disease Registry, U.S. Department of Health and Human Services.

Baden, H. P., Kubilius, J., Kredar, J. C., Steinberg, M. L., and Wolman, S. R. (1987). Isolation and characterization of a spontaneously arising long-lived line of human keratinocytes (NM1). In vitro Cell Develop. Biol. 23, 205–213.[ISI][Medline]

Berenbaum, M. C. (1985). The expected effect of a combination of agents: The general solution. J. Theor. Biol. 114, 413–431.[ISI][Medline]

Berenbaum, M. C. (1989). What is synergy? Pharmacol. Rev. 41, 93–141.[ISI][Medline]

Boukamp, P., Petrusevska, R. T., Breitkreutz, D., Hornung, J., Markham, A., and Fusenig, N. E. (1988). Normal keratinization in a spontaneously immortalized aneuploid human keratinocyte cell line. J. Cell Biol. 106, 761–771.[Abstract]

Calabrese, E. J. (1997). Hormesis revisited: New insights concerning the biological effects of low-dose exposures to toxins. Environ. Law Report. 27, 10526–10532.

Calabrese, E. J., and Baldwin, L. A. (1997a). The dose determines the stimulation (and poison): Development of a chemical hormesis database. Int. J. Toxicol. 16, 545–559.[ISI]

Calabrese, E. J., and Baldwin, L. A. (1997b). A quantitatively-based methodology for the evaluation of chemical hormesis. Human Ecol. Risk Assess. 3, 545–554.[ISI]

Calabrese, E. J., and Baldwin, L. A. (1999). Chemical hormesis: Its historical foundations as a biological hypothesis. Toxicol. Pathol. 27, 195–216.[ISI][Medline]

Cohen, M. D., Kargacin, C. B., Klein, C. B., and Costa, M. (1993). Mechanisms of chromium carcinogenicity and toxicology. Crit. Rev. Toxicol. 23, 255–281.[ISI][Medline]

DeRosa, C. T., Johnson, B. L., Fay, M., Hansen, H., and Mumtaz, M. M. (1996). Public health implications of hazardous waste sites: Findings, assessment, and research. Food Chem. Toxicol. 34, 1131–1138.[ISI][Medline]

Fay, M., and Mumtaz, M. M. (1996). Development of a priority list of chemical mixtures occurring at 1188 hazardous waste sites using the HazDat database. Food Chem. Toxicol. 34, 1163–1165.[ISI][Medline]

Gennings, C., and Carter, Jr., W. H. (1995). Utilizing concentration-response data from individual components to detect statistically significant departures from additivity in chemical mixtures. Biometrics 51, 1264–1277.[ISI]

Gennings, C., Carter, Jr., W. H., Campain, J. A., Bae, D. S., and Yang, R. S. H. (2001). Statistical analysis of interactive cytotoxicity in human epidermal keratinocytes following exposure to a mixture of four metals. J. Agric. Biol. Environ. Stat. (in press).

Gennings, C., Schwartz, P., Carter, Jr., W. H., and Simmons, J. E. (1997). Detection of departures from additivity in mixtures of many chemicals with a threshold model. J. Agric. Biol. Environ. Stat. 2, 198–211.

Germolec, D. R., Spalding, J., Boorman, G. A., Wilmer, J. L., Yoshida, T., Simeonova, P. P., Bruccoleri, A., Kayama, F., Gaido, K., Tennant, R., Burleson, F., Dong, W., Lang, R. W., and Luster, M. I. (1997). Arsenic can mediate skin neoplasia by chronic stimulation of keratinocyte-derived growth factors. Mut. Res. 386, 209–218.[ISI][Medline]

Germolec, D. R., Yoshida, T., Gaido, K., Wilmer, J. L., Simeonova, P. P., Kayama, F., Burleson, F., Dong, W., Lang, R. W., and Luster, M. I. (1996). Arsenic induces overexpression of growth factors in human keratinocytes. Toxicol. Appl. Pharmacol. 141, 308–318.[ISI][Medline]

Hochadel, J. F., and Waalkes, M. P. (1997). Sequence of exposure to cadmium and arsenic determines the extent of toxic effects in male Fischer rats. Toxicology 116, 89–98.[ISI][Medline]

Horne, M. C., Goolsby, G. L., Donaldson, K. L., Tran, D., Neubauer, M., and Wahl, A. F. (1996). Cyclin G1and cyclin G2 comprise a new family of cyclins with contrasting tissue-specific and cell cycle-regulated expression. J. Biol. Chem. 271, 6050–6061.[Abstract/Free Full Text]

Huang, R. -N., and Lee, T. -C. (1996). Arsenite efflux is inhibited by verapamil, cyclosporin A, and GSH-depleting agents in arsenite-resistant Chinese hamster ovary cells. Toxicol. Appl. Pharmacol. 141, 17–22.[ISI][Medline]

Huot, J., Roy, G., Lambert, H., Chretien, P., and Landry, J. (1991). Increased survival after treatments with anticancer agents of Chinese hamster cells expressing the human Mr 27,000 heat shock protein. Cancer Res. 51, 5245–5252.[Abstract]

Kachinskas, D. J., Qin, Q., Phillips, M. A., and Rice, R. A. (1997). Arsenate suppression of human keratinocyte programming. Mut. Res. 386, 253–261.[ISI][Medline]

Kreppel, H., Bauman, J. W., Liu, J., McKim, J. M., Jr., and Klaassen, C. D. (1993). Induction of metallothionein by arsenicals in mice. Fundam. Appl. Toxicol. 20, 184–189.[ISI][Medline]

Laemmli, U. K. (1970). Cleavage of structural proteins during assembly of the head of bacteriophage T4. Nature 227, 680–685.[ISI][Medline]

Lee, T. -C., Wei, M. L., Chang, W. J., Ho, I. C., Lo, J. F., Jan, K. Y., and Huang, H. (1989). Elevation of glutathione levels and glutathione S-transferase activity in arsenic-resistant Chinese hamster ovary cells. In Vitro Cell Develop. Biol. 25, 442–448.[ISI][Medline]

Li, G. C. (1983). Induction of thermotolerance and enhanced heat shock protein synthesis in Chinese hamster fibroblasts by sodium arsenite and ethanol. J. Cell Physiol. 115, 116–122.[ISI][Medline]

Li, W., Kagan, H. M., and Chou, I. (1994). Alterations in cytoskeletal organization and homeostasis of cellular thiols in cadmium-resistant cells. Toxicol. Appl. Pharmacol. 126, 114–123.[ISI][Medline]

Liu, J., Liu, Y., Goyer, R. A., Achanzar, W., and Waalkes, M. P. (2000a). Metallothionein-I/II null mice are more sensitive than wild-type mice to the hepatotoxic and nephrotoxic effects of chronic oral or injected inorganic arsenicals. Toxicol. Sci. 55, 460–467.[Abstract/Free Full Text]

Liu, J., Liu, Y., Habeebu, S. M., Waalkes, M. P., and Klaassen, C. D. (2000b). Chronic combined exposure to cadmium and arsenic exacerbates nephrotoxicity, particularly in metallothionein-I/II mice. Toxicology 147, 157–166.[ISI][Medline]

Mehendale, H. M. (1994). Mechanisms of the interactive amplification of halomethane hepatotoxicity and lethality by other chemicals. In Toxicology of Chemical Mixtures (R. S. H. Yang, Ed.), pp. 299–334. Academic Press, San Diego, CA.

Mehlen, P., Kretz-Remy, C., Preville, X., and Arrigo, A. P. (1996). Human hsp27, Drosophila hsp27 and human alphaB-crystalline expression-mediated increase in glutathione is essential for the protective activity of these proteins against TNF alpha-induced cell death. EMBO J. 15, 2695–2706.[Abstract]

Mossman, T. (1983). Rapid colorimetric assay for cellular growth and survivals: Application to proliferation and cytotoxicity assays. J. Immunol. Methods 65, 55–63.[ISI][Medline]

Rheinwald, J. G., and Green, H. (1975). Serial cultivation of strains of human epidermal keratinocytes; the formation of keratinizing colonies from single cells. Cell 6, 331–343.[ISI][Medline]

Rheinwald, J. G., and Green, H. (1977). Epithelial growth factor and the multiplication of cultured human epidermal keratinocytes. Nature 265, 421–424.[ISI][Medline]

Rhim, J. S., Jay, G., Arnstein, P., Price, F. M., Sanford, K. K., and Aaronson, S. A. (1985). Neoplastic transformation of human epidermal keratinocytes by AD12-SV40 and Kirsten sarcoma viruses. Science 227, 1250–1251.[ISI][Medline]

Rosen, B. P. (1995). Resistance mechanisms to arsenicals and antimonials. Basic Clin. Physiol. Pharmacol. 6, 251–263.

Shimizu, M., Hochadel, J. F., Fulmer, B. A., and Waalkes, M. P. (1998). Effect of glutathione depletion and metallothionein gene expression on arsenic-induced cytotoxicity and c-myc expression in vitro. Toxicol. Sci. 45, 204–211.[Abstract]

Stebbing, A. R. D. (1982). Hormesis—the stimulation of growth by low levels of inhibitors. Sci. Total Environ. 22, 213–234.[ISI][Medline]

Stebbing, A. R. D. (1997). A theory for growth hormesis. BELLE Newsletter 6, 1–11.

Susanto, I., Wright, S. E., Lawson, R. S., Williams, C. E., and DeRosa, C. T. (1998). Metallothionein, glutathione, and cystine transport in pulmonary artery endothelial cells and NIH3T3 cells. Am. J. Physiol. 274, 296–300.

Svendsgaard, D. J. and Hertzberg, R. C. (1994). Statistical methods for the toxicological evaluation of the additivity assumption as used in the Environmental Protection Agency Chemical Mixture Risk Assessment Guidelines. In Toxicology of Chemical Mixtures (R. S. H. Yang, Ed.), pp. 599–642. Academic Press, San Diego.

Tamhane, A. C., and Dunlop, D. D. (2000) Multiple comparisons of means. In Statistics and Data Analysis from Elementary to Intermediate, pp. 475–476. Prentice-Hall, Inc., Upper Saddle River, NJ.

Wang, H. F., and Lee, T. -C. (1993). Glutathione S-transferase pi facilitates the excretion of arsenic from arsenic-resistant Chinese hamster ovary cells. Biochem. Biophys. Res. Commun. 192, 1093–1099.[ISI][Medline]

Wu, W., and Welsh, M. J. (1996). Expression of the 25-kDa heat shock protein (HSP27) correlates with resistance to the toxicity of cadmium chloride, mercuric chloride, cis-platinum(II)-diammine dichoride, or sodium arsenite in mouse embryonic stem cells transfected with sense or antisense HSP27 cDNA. Toxicol. Appl. Pharmacol. 141, 330–339.[ISI][Medline]

Yang, J. H., Thraves, P., Dritschilo, A., and Rhim, J. S. (1992). Neoplastic transformation of immortalized human keratinocytes by 2,3,7,8-tetrachlorodibenzo-p-dioxin. Cancer Res. 52, 3478–3482.[Abstract]

Ye, J., Zhang, X., Young, H. A., Mao, Y., and Shi, X. (1995). Chromium(VI)-induced nuclear factor-kB activation in intact cells via free radical reactions. Carcinogenesis 16, 2401–2405.[Abstract]

Yen, H. -T., Chiang, L. -C., Wen, K. -H., Chang, S. -F., Tsai, C. -C., Yu, C. -L., and Yu, H. -S. (1996). Arsenic induces interleukin-8 expression in cultured keratinocytes. Arch. Dermatol. Res. 288, 716–717.[ISI][Medline]

Zhao, C. Q., Young, M. R., Diwan, B. A., Coogan, T. P., and Waalkes, M. P. (1997). Association of arsenic-induced malignant transformation with DNA hypomethylation and aberrant gene expression. Proc. Natl. Acad. Sci. U.S.A. 94, 10907–10912.[Abstract/Free Full Text]