Introduction

Several U.S. military personnel participating in Operation Desert Storm were wounded in friendly fire accidents and have retained large fragments (approximately 2-20 mm) of depleted uranium (DU) in their bodies. DU, used in military munitions in the United States, was shown to be an effective material in kinetic energy penetrators in the 1991 Persian Gulf War. Chemically similar to natural uranium (1), DU is a low specific-activity heavy metal, with a density approximately 1.7 times that of lead (19 g/cm³ versus 11.35 g/cm³). DU differs from natural uranium in that it has been depleted of ²³⁵U and ²³⁴U. As a result, the specific activity of DU is significantly lower than that of natural uranium (0.4 µCi/g versus 0.7 µCi/g, respectively) (2).

Assessment of the carcinogenic risks from DU is complicated by the dual toxicity of uranium (i.e., chemical as well as radiological). Epidemiological studies have linked uranium mining and milling to human carcinogenesis (1), but there are no published studies to permit an accurate assessment of risks for carcinogenesis from DU. The radiological health risks (external exposure) to personnel handling DU munitions were evaluated and determined to be within allowed occupational exposure limits (1,2); the health risks from internalized DU, however, are more difficult to estimate due to the physical and chemical properties of the internalized DU (3). The chemical toxicity of acute, short-term exposures to uranium, primarily manifested as renal, pulmonary, and developmental toxicity, has clearly been demonstrated in animals and humans (4). In contrast, the long-term health risks associated with internal chronic exposure are not as clearly defined (1). In view of carcinogenesis risk estimates and medical management questions relevant to current and possible future incidents of DU internalization, an examination of molecular and cellular effects, including the potential transforming ability of DU, are necessary to understanding the potential carcinogenic effects of DU. The use of cell culture models to investigate potential or known carcinogens can provide important insights into the cellular and molecular mechanisms of carcinogenesis.

In spite of epidemiological studies that suggest that uranium is a carcinogen (1), there is no evidence that uranium of any type (depleted, naturally occurring, or enriched) can transform human cells to the tumorigenic phenotype (1). Furthermore, the in vitro transformation assay has not previously been used to study the transforming ability of any uranium compounds (depleted, naturally occurring, or enriched uranium). This assay has been widely used in conjunction with metal salts to assess the potential
carcinogenicity of metal compounds (e.g., nickel, chromium, lead) (5-8) and therefore we chose to use this assay to assess the potential carcinogenicity of DU. Investigations using metal salts have also been able to clarify the contradictions observed in some animal and human carcinogenicity studies (8). The HOS TE85 cell line, an immortalized, nontumorigenic osteoblastlike cell line, has been successfully used to demonstrate the transformation of nontumorigenic human cells to the tumorigenic phenotype by metals (7,8) and chemical (9) carcinogens. We chose the DU compound DU-UO₂Cl₂, which readily forms the UO₂²⁺ cation in solution, as the depleted uranium metal salt to be tested. We used nickel sulfate and lead acetate, which have previously been shown to transform cells in vitro, for comparison.

Our data demonstrate for the first time that a DU compound can transform human cells to the tumorigenic phenotype, similar to results observed with some nickel compounds (6-8). Isolated DU transformants form anchorage-independent colonies, produce tumors when injected into athymic mice, express the k-ras oncogene, and have an altered phosphorylation of the Rb tumor-suppressor protein. Based on these in vitro results, the carcinogenic potential of internalized DU remains a concern and warrants additional studies with experimental animals.

Materials and Methods

Cell lines and culture. Human osteosarcoma cell lines HOS (TE85, clone F-5) were obtained from the American Type Culture Collection (Rockville, MD). Cell cultures were propagated in Dulbecco's modified Eagle medium (D-MEM) supplemented with 2 mM glutamine, 10% heat-inactivated fetal calf serum (Gibco Laboratories, Grand Island, NY), 100 U/ml penicillin, and 100 µg/ml streptomycin (Sigma, St. Louis, MO). Cells were tested for mycoplasma by MycoTect Kit (Sigma), and only cells negative for mycoplasma were used.

Transformation and cell growth studies. DU-UO₂Cl₂ was initially dissolved in sterile water (freely soluble in water). Based on previous unpublished work with uranyl compounds which demonstrated that micromolar concentrations are relatively nontoxic to cultured cells, we chose a concentration of 10 µM for transformation and cell growth studies. Metal salts were added to sterile water and then filtered to prepare concentrated stock solutions that were used throughout these studies. Dilutions from concentrated stock solutions were dissolved into complete medium immediately before exposure. Briefly, 1 10⁶ cells were treated with DU-UO₂Cl₂ (10-250 µM), NiSO₄ (10 µM), or lead acetate (10 µM) for 24 hr in complete medium. After treatment, cells were rinsed, trypsinized, counted, and seeded in 100-mm tissue culture dishes at a density calculated to yield about 125-225 surviving cells per dish. Approximately 150 replicate dishes were seeded for each treatment group. The cultures were incubated for 5 weeks with weekly changes of nutrient medium.

For quantitative transformation assessment, at the end of the incubation period, cells were fixed, stained, and examined for the appearance of transformed foci (10). We assayed transformed foci using the criteria developed by Reznikoff et al. (10) and the IARC/NCI/EPA Working Group (11). The working group report indicated that because of the continuum of focus morphology, some foci can be intermediate (I/II, II/III) in character, and these foci should be scored conservatively and assigned to the category of lower aggressive behavior. With the HOS cells it was easy to distinguish between a type I and a type II foci, but somewhat difficult to differentiate between type II and type III foci. Therefore, in our experiments, only type II and type II/type III foci were scored as transformants. In Table 1, number of dishes with foci equals number of foci of type II plus type III. Details of the transformation frequency calculations and the standard error were described elsewhere (12) but are outlined below. Transformed foci are not contact inhibited and may be dislodged during refeeding forming satellite colonies. To avoid this source of errors, the average number of transformed foci per dish, , was computed from the proportion of dishes free of transformed colonies, f, i.e., = -ln f. To determine the transformation frequency, the following formula was used:

Transformation frequency =/number of surviving cells per dish

This calculation has been widely used in radiation-induced neoplastic transformation studies to quantitatively compare transformation frequencies. The transformation data are presented from three independent experiments unless otherwise noted.

We determined cell survival fraction (SF) using the conventional clonogenic assay (9). Cytotoxicity and survival assays were conducted in parallel with each transformation assay as described (9). Several transformed foci were picked by cloning cylinders and expanded by mass culture to establish transformed clonal lines. The transformation dose-response curve was determined as indicated above, except that increasing concentrations of DU-UO₂Cl₂ were used. We selected the optimal dose level for transformation following a preliminary toxicity test based upon colony-forming efficiency. The high dose selected resulted in >90% toxicity, and the low dose selected resulted in minimal toxicity. Three intermediate doses were also selected.

To determine saturation densities, cells were plated at 1 10⁵ cells per 100-mm diameter plate in complete growth medium and monitored for growth as previously described (9). For the soft-agar clonability assay, cells were plated at a density of 2 10³ cells per well in a six-well plate sandwiched by 1-ml bottom agar (0.6%) and 1-ml top agar (0.3%). Cells were fed weekly by adding a new layer of top agar. Colonies >0.5-mm diameter were scored using a microscope after 2 weeks. The plating efficiency value for each clone represents the mean number of colonies scored from three wells. For both saturation density determination and plating efficiency in soft agar, data represent three independent experiments.

Tumorigenicity assay. Experiments were performed with 4- to 5-week-old female athymic mice (Division of Cancer Treatment, NCI Animal Program, Frederick Cancer Research Facility). For this assay, 5 10⁶ or 1 10⁶ cells in a 0.2-ml sterile suspension were injected subcutaneously in the right scapula. Animals were then observed for tumor growth at the sites of injection. The appearance of tumor growth was monitored daily for 180 days. Tumor area was measured using a caliper measurement of two perpendicular diameters. When tumors were >100 mm², the animal was euthanized.

Sister chromatid analysis. The yield of sister chromatid exchange (SCE ) per cell in control and DU-UO₂²⁺ exposed cells following a 24-hr exposure was assessed in 100 second mitotic metaphase spreads (48-hr culture) using the conventional fluorescence-plus-Giemsa harlequin staining protocol (13). The SCE yield in selected transformed clones was similarly determined.

Microdosimetry. The term "specific energy" is used rather than "dose" because dose refers to macroscopic averages, whereas specific energy is a microdosimetric equivalent (i.e., dose to a single cell). We applied microdosimetric methods using Monte Carlo computer simulations (14). The code was modified to include the following parameters in the calculations: cell location and size, distribution of uranium (intracellular and extracellular), specific activity, alpha particle energies, media volume, and molar concentration of uranyl chloride. We calculated the trajectory of each alpha particle using vector analysis to determine whether it would hit a nucleus. For those that scored a hit, the amount of energy transferred to the nucleus was calculated by determining the residual energy retained by the alpha particle as it penetrated the nucleus and integrating the fractional energy loss under the Bragg curve from the entry point to the exit point.

Kinetic phosphorescence analysis of uranium. Kinetic phosphorimetry measurements were performed using a kinetic phosphorescence analyzer (KPA) (15). To determine the amount of uranium per cell following a 24-hr treatment with 10 µM DU-UO₂²⁺, cells were treated and then rinsed in phosphate-buffered saline. Using previously described procedures, cells were lysed and fractionated into subcellular fractions. Lysates were analyzed for uranium content as described (15).

Northern blot analysis and DNA probes. Cytoplasmic RNA was extracted from exponentially growing cells and separated by electrophoresis in 1% agarose-formaldehyde gels. RNA preparation and blotting onto nytran filters, hybridization with radiolabeled DNA probes, and autoradiography were as previously described (9). The ras probe, a Sac I 2.9 kb fragment of the human k-ras gene, was obtained from Oncor (Gaithersburg, MD). We prepared ³²P-labeled DNA probes using a random primed DNA labeling kit (Boehringer Mannheim, Germany).

Metabolic labeling and immunoprecipitation of pRb. Exponentially growing cells [1 10⁶ per T75-cm² flask (Costar)] were radiolabeled with ³²P-orthophosphate (250 µCi) for 24 hr and lysed in buffer (1% Triton X-100, 0.05 M Tris-HCl, pH 8.0, 5 mM MgCl₂ and 0.1 M NaCl). Lysates were incubated with polyclonal anti-pRb antibody-agarose conjugates (Oncogene Science, NY) for 2 hr at 4°C. Processing of immunoprecipitates was as previously described (9). The proteins were resolved on 7.5% polyacrylamide gels and analyzed by autoradiography.

Results

Characterization of model system: cytotoxicity, SCE induction, and microdosimetry. We initially assessed the effect of DU-UO₂CL₂ exposure on cell growth and survival, cytogenetic damage, and microdosimetry to characterize our cell model system. No effect on cell survival (surviving fraction, SF, 0.96 ± 0.04) was observed using the standard colony formation assay (9) following a 24-hr exposure to DU-UO₂²⁺ (10 µM). Growth rate analysis by cell enumeration and thymidine incorporation into DNA (9) similarly revealed that the DU-UO₂²⁺ treatment did not affect cell growth during the 24-hr exposure (inhibitory concentration, IC₅₀, 45 µM). Therefore, to preclude excessive cell killing due to heavy metal toxicity, the nontoxic, noncytostatic concentration chosen for the transformation experiment was 10 µM (24-hr exposure). The colony formation assay was also used to assess the cytotoxicity of NiSO₄ and lead acetate. Similarly, nontoxic concentrations of NiSO₄ and lead acetate were chosen (20 µM, SF = 0.94; 20 µM, SF = 0.095, respectively).

Cytogenetic analysis of the acute effect of the 24-hr exposure to 10 µM DU-UO₂²⁺ was assessed in the in vitro HOS model system using the SCE assay. As seen in Figure 1A, this acute DU exposure caused an approximately twofold increase in SCE induction (untreated: 8.3 SCE/cell; DU-UO₂²⁺: 17.9 SCE/cell). In comparison, lead acetate treatment did not result in a significant increase in SCEs, while nickel did induce a small but significant increase in SCEs (lead acetate, 20 µM: 8.9 SCE/cell; NiSO₄, 20 µM: 11.4 SCE/cell).

Figure 1. Characterization of an in vitro HOS cellular model to access DU-UO₂²⁺ exposure. The effect of DU-UO₂²⁺ exposure (24 hr, 10 µM) on sister chromatid exchange (SCE) induction and a microdosimetric assessment of the fraction of cell nuclei exposed to DU-induced alpha particles was determined. (A) Frequency distribution of SCE formation, measured as described in Materials and Methods. (B) Frequency distribution of nuclei hit by an alpha particle; microdosimetric assessment was determined using a Monte Carlo computer simulation based on measured levels of intracellular and extracellular uranium as described in the methods. F(hit) = 1.4 10^-3%.

Using a standard uranium analysis technique, KPA (15), the amount of uranium per cell following a 24-hr treatment with 10 µM DU-UO₂²⁺ was determined to be 1.8 (± 0.15) 10^-6 ng/cell with the average intercellular concentration being 2 µM.

To examine the stochastic fluctuations of alpha particle hits and energy deposition in the cell nuclei, microdosimetric methods using Monte Carlo computer simulations were applied (Figure 1B). The computer code, designed by Humm (14) to predict cell inactivation by internal alpha emitters, was modified to correspond to the experimental conditions as described in Materials and Methods. The trajectory of each alpha particle was calculated using vector analysis to determine whether it would hit a nucleus. The energy transferred to the nucleus was calculated by determining the residual energy retained by the alpha particle as it penetrated the nucleus. The calculations determined that only 0.0014% of the nuclei were hit and that the mean specific energy received by those cells was approximately 17 cGy. The results, plotted as a frequency distribution, are shown in Figure 1B.

Transformation of HOS cells by DU-UO₂²⁺: comparison to NiSO₄ and lead acetate. To assess for morphological cell transformation, the standard focus formation assay described by Reznikoff et al. (10) and Hill et al. (12) was used. After exposure to DU-UO₂²⁺, a morphological change in HOS cells was observed (Fig. 2). HOS cells exhibit a flat epithelial-like morphology and appear to grow in a monolayer (Fig. 2A). In contrast, after treatment with DU-UO₂²⁺ and weekly changes of nutrient medium for 5 weeks, diffuse type II foci appeared (Fig. 2B). The morphology of the focus is distinctly different from the surrounding cells (Fig. 2B), although the focus does not exhibit the "piled up" appearance seen in transformed C3H10T_1/2 cells (10). Table 1 shows measured values for the transformation frequencies (normalized per surviving cell) for HOS cells treated with DU-UO₂²⁺ (10 µM). The transforming abilities of NiSO₄ and lead acetate have been shown previously (8,16) but are repeated here to allow for a quantitative comparison to the DU compound. The data demonstrate that treatment with DU-UO₂²⁺ resulted in a transformation frequency of 40.2 10^-4, which is a 9.6 (± 2.8)-fold increase in transformation compared to untreated HOS cells. In comparison, the transformation frequencies induced by NiSO₄ and lead acetate treatment resulted in a 7.1 (± 2.1)- and 5.0 (± 1.6)-fold increase in transformation frequency compared to untreated cells.

Figure 2. Morphology of HOS cells and focus formation after DU-UO₂²⁺ exposure. (A) Morphology of control HOS cells; phase contrast micrograph, (40). (B) Focus formation in HOS cell; the edge of a focus of transformed cells is seen against a background of parental HOS cells with normal morphology (40).

Several foci were picked by cloning cylinders and expanded by mass culture to establish transformed clonal lines. Four DU-transformed clones (designated as DU1, DU2, DU3, and DU4) were selected for tumorigenic and growth analysis; a NiSO₄- and a lead acetate-transformed clone were also selected for similar analyses. A spontaneous focus arising from untreated HOS cells was also selected and expanded.

Biological characterization of the transformed phenotype. Alteration in growth control is critical to neoplastic transformation, and therefore the metal-transformed clones were further characterized by quantitative differences in growth properties associated with the neoplastic phenotype (e.g, saturation density and soft colony-forming efficiency). Additionally, to determine if cells transformed in vitro were capable of producing tumors in immunosuppressed mice, the nude mouse assay was used to test the tumorigenicity of DU-transformed cells (9,10). From Table 2 it can be seen that the saturation densities of DU-transformed cells were three to four times higher than that of the parental HOS cells. Data obtained for NiSO₄- and lead acetate-transformed cells were similar to that for DU. A comparison of the transformants' ability to grow in soft agar reveals that the DU transformants generated colonies within 1 week, with colony-forming efficiencies of 32-51%; the plating efficiencies were somewhat lower for NiSO₄- or lead acetate-transformed cells. Parental HOS cells did not form colonies in soft agar (Table 2). We have previously shown that MNNG and EJ-ras transformants formed soft agar colonies whose size and frequency were comparable to those observed with these DU transformants (9).

Inoculation of athymic nude mice with DU transformants resulted in the development of animal tumors within 4 weeks. All of the DU transformed cell lines analyzed for tumorigenicity appeared to be highly tumorigenic because as few as 1 million DU-transformed cells developed progressively growing tumors within 4 weeks. Similarly, nude mice inoculated with NiSO₄ or lead acetate transformants also formed tumors (Table 2). In contrast, parental HOS cells injected into nude mice did not form any tumors during a period of 6 months after cell inoculation. Histological analysis indicates that the tumors formed by DU-transformed cells resemble a carcinoma characterized by an undifferentiated, sheetlike growth. Tumors were reestablished in tissue culture and confirmed as human; their resemblance to the cells of origin was determined by karyological analysis (data not shown).

In contrast to the data presented in Figure 1A which demonstrate "acute" SCE elevation by DU, "persistent" cytogenetic changes, manifested as effects on the levels of SCEs per cell, were measured in the four transformed clones selected. Cytogenetic analysis of the DU-uranyl chloride-transformed clones revealed that all of the clones had a 2.14 (± 0.22)- to 3.47 (± 0.31)-fold higher level of SCE per cell compared to parental HOS cells (Table 2). The untreated, spontaneous HOS transformants did not exhibit a significant increase in SCE induction [1.07 (± 0.22)-fold increase] compared to parental HOS cells.

ras and pRb analysis in DU transformants. Neoplastic cell transformation is hypothesized to result from a multistep process involving activation of oncogenes and inactivation of tumor-suppressor genes (17). Both metal- and radiation-induced neoplastic transformation have been shown to be associated with genetic alterations in specific oncogenes and tumor-suppressor proteins, such as ras and pRb (6,9). Therefore, possible molecular changes in ras and pRb associated with DU-induced transformation were studied using the four DU-transformed cell lines. Northern blot analysis, shown in Figure 3A, revealed high levels of ras mRNA in each of the four DU clones tested. In contrast, the ras mRNA levels were undetectable in parental HOS. The amount of actin mRNA was similar in all clones tested (not shown). Heavy-metal-induced transformation (e.g., nickel) has been shown to cause a loss of phosphorylation of the retinoblastoma protein. We therefore used anti-pRb1 monoclonal antibodies to compare the phosphorylation of the pRb of the different DU-transformed cell lines to the parental cells by assessing the quantity of radioactive phosphate incorporated into the Rb protein. Each of the DU clones examined exhibited significantly lower amounts of ³²P incorporated into pRb relative to the parental HOS cells (Fig. 3B). The lack of phosphorylation of pRb could be a result of less protein available to phosphorylate, abnormalities of the cyclin-associated kinase systems responsible for phosphorylation of pRb, a mutation in the Rb gene, or a combination of these possibilities. Previous observations with nickel-transformed HOS cells demonstrated a mutation was induced in these transformants that affected the ability of the Rb protein to be phosphorylated and function normally (6).

Figure 3. Analysis of K-ras expression and Rb protein production in parental HOS and DU-UO₂Cl₂-transformed clones. (A) ras Oncogene expression; Northern blot analysis of cytoplasmic RNA (20 µg) from DU-1, DU-2, DU-3, DU-4, and HOS. Hybridization was with a ³²P-labeled K-ras probe (ras transcripts were undetectable in HOS cells). Hybridization with ³²P-labeled actin was used to indicate that the relative amounts of RNA loaded into each lane were the same (data not shown). (B) Rb protein level; immunoprecipitation analysis of Rb protein from DU-1, DU-2, DU-3, DU-4, and HOS. Anti-pRb1 was used to capture cellular Rb protein containing ³²P.

Dose response: morphological transformation by UO₂Cl₂. To further evaluate the ability of DU-uranyl chloride to morphologically transform cells, increasing concentrations of DU were tested for their transforming ability. The results in Table 3 show that there is a DU concentration-dependent increase in transformation frequency. A transformed clone was selected from each concentration tested and expanded to mass culture. An examination of the selected transformants' ability to grow in soft agar reveals that the DU transformants generated large colonies (>0.5 mm) within 1 week with colony-forming efficiencies ranging from 33% to 53%. Inoculation of athymic nude mice with these DU transformants resulted in the development of tumors within 4 weeks.

Discussion

Our studies demonstrate for the first time that the malignant transformation of immortalized human cells can be achieved by exposure to the depleted uranium compound UO₂Cl₂. These transformants showed morphological changes and anchorage-independent growth in soft agar, induced tumors when transplanted into nude mice, and exhibited alterations in ras oncogene expression and pRb phosphorylation. Based on equivalent concentrations and metal toxicities, the magnitude of this DU-UO₂²⁺ transformation is comparable to that seen here for NiSO₄ and approximately twofold higher than that observed for lead acetate, both known transforming metals (8,16). Furthermore, DU-UO₂²⁺ caused a dose-dependent frequency in the transformation of HOS cells. To our knowledge, this is the first report of human cell transformation to the tumorigenic phenotype by any uranium compound of any specific activity (depleted, naturally occurring, or enriched). Although specific heavy metals and alpha particle radiation have been shown individually to transform cultured cells and it is not unexpected that DU could transform cells, these results are important to the understanding of the mechanism of the potential carcinogenicity of DU.

Second, these studies confirm previous results that HOS cells can be used to assess morphological transformation (8). Morphological transformation studies have primarily involved rodent cell lines such as C3H and 10T1/2, and to our knowledge this is only the second time that HOS cells have been used for an assessment of morphological transformation. These studies with DU demonstrate another step in understanding the transformation process of HOS cells.

The precise mechanism(s) by which DU-UO₂²⁺ induces transformation in HOS cells is unknown. The possibility that alterations in specific oncogenes (e.g., ras) and/or inactivation of tumor-suppressor genes (e.g., p53, Rb) are involved in the conversion of these cells to the malignant phenotype has been considered. HOS cells contain a mutation at codon 156 resulting in a mutated form of p53 protein, which is believed to be partially responsible for their immortalization (18). As neoplastic conversion is postulated to result from a multistep process involving cell immortalization and gene alterations (17,19), the transformation of immortalized HOS cells by DU-UO₂²⁺ to the malignant phenotype may involve other cellular oncogenes in this process. Our data demonstrate that the ras oncogene was activated in the transformation process induced by DU exposure. The ras oncogenes have been implicated in both chemically and radiation-induced animal tumors (20-22) and spontaneous human tumors (23) and may play a prominent role in the multistage conversion process.

Because others have demonstrated that nickel-induced morphological transformation of HOS cells can affect the Rb gene and its encoded protein (6), we examined the status of this tumor-suppressor protein in our DU-transformed clones. Changes in the phosphorylation of the Rb tumor-suppressor gene-encoded protein in the DU-transformed clones were observed. The Rb gene has been shown to be mutated or inactivated in a broad spectrum of human tumors, including retinoblastoma, osteosarcoma, lung cancer, and breast cancer, and appears to be important in human carcinogenesis; furthermore, the inactivation of the Rb may be involved in the molecular mechanism of carcinogenesis of human osteoblasts (24). Recent in vitro studies with nickel-transformed HOS cells suggested that nickel exposure induced a mutation in the Rb gene that resulted in an alteration in the function of the Rb protein (6). It appears from our data that transformation of HOS cells by DU-UO₂²⁺ results from a conversion process which also involves activation of tumor-promoting gene(s) and/or alterations in tumor-suppressor proteins, including ras, p53, and Rb.

Without evidence that DU-UO₂²⁺ directly affects the DNA, we can only speculate about the mechanism(s) by which DU-UO₂²⁺ ultimately affects the ras gene and the Rb protein. The nature of this mechanism may involve direct damage to the DNA and to DNA repair processes. Transforming metals such as nickel can induce SCEs (25) and have been shown to directly damage DNA (26), whereas others such as lead acetate do not directly affect the genetic material and may induce transformation via an indirect mechanism such as changes in DNA conformation or enzymatic disturbances (16). An elevation in SCE levels has also been observed as the result of inhibition of DNA replication, rather than by direct damage of chromosomes (25). Our observation of an elevation in the level of SCE per cell, while consistent with these concepts, cannot distinguish between them. The mechanism of action at the genomic level for DU-induced transformation has not been determined and awaits further investigation; it is possible that DU induces transformation via mechanisms similar to those observed for nickel or lead, or by other by unknown mechanisms.

Elucidation of the mechanism by which DU transforms cells is further complicated by the concurrent exposure to both the chemical (heavy metal) component and the radiological (alpha particle) components from DU. The chemical toxicity of uranium, based on animal and epidemiology studies, is well known (27) and is believed to be responsible for cytogenetic damage observed in the lymphocytes from men occupationally exposed to insoluble uranium (28), but has not been linked to uranium carcinogenesis. However, our results provide evidence that the chemical/metal toxicity of DU plays a significant role in its ability to morphologically transform human cells. It is interesting to speculate that the observed chemical toxicity of DU-UO₂²⁺ may be attributed to the radiomimetic property of uranyl ions to act as a transition metal to produce free radicals via a Fenton-like reaction (29), similar to those chemical properties generally attributed to iron-II, copper-I, and recently demonstrated for plutonium by Claycamp and Luo (30). DNA is an effective chelator of metal ions, which suggest the possibility of a metal-mediated, site-specific free radical damage (31).

Increased lung cancer risks in uranium miners and uranium milling workers have been attributed to exposure to the alpha particle emitter radon and its daughter products (1,32). In our study with DU exposures in vitro, the biological effect from the radiological component may be significantly reduced since so few cells (0.0014%) are actually hit by alpha-particle emissions (1 alpha particle traversal/cell hit). Therefore, in our study, the defining question is whether the alpha particle dose distribution and its track could effectively produce a change in that target cell that would induce or promote morphological transformation. Preliminary data using an alpha particle microbeam have demonstrated that one alpha particle traversal of a cell is not enough to induce morphological transformation (33). Although these data from Miller's laboratory and our results argue for a negligible role for alpha radiation in the DU-UO₂²⁺-induced transformation, low levels of exposure to alpha radiation have been shown to induce genetic changes (34) and chromosomal instability (35) in the progeny of alpha-irradiated cells. The involvement of transmitted genetic instability in the transformation process is not fully understood and cannot be ignored because it could possibly be involved in the DU-UO₂²⁺-induced transformation. Whether by radiation or chemical mechanism, DU-UO₂²⁺ exposure is consistent with the generally accepted features associated with neoplastic transformation of cells by radiation or metal exposure including formation of sister chromatid exchanges (31,36,37) and genomic instability manifested as gene alterations (6,36,38).

In summary, we have used a model system of in vitro human osteoblast cells exposed for 24 hr to DU-uranyl chloride (10 µM), nickel sulfate (10 µM), or lead acetate (10 µM) to assess relative transforming potential of DU in an effort to better understand the potential health risks from long-term exposure to internalized DU. Despite the well-known low solubility of uranyl compounds, we found cellular uptake of uranium consistent with other transuranic compounds (39). As expected from the prompt chemical toxicity of heavy metal exposure, the levels of SCEs were elevated in DU-UO₂²⁺-treated cells. Microdosimetry studies demonstrate that few (<1.4 10^-3%) cells are actually hit by alpha particles emitted from DU. This argues for a negligible role for radiation effects from
DU-UO₂²⁺ exposure. DU-UO₂²⁺ appears to be 1.34 (± 0.24)- and 1.91 (± 0.42)-fold more potent that nickel sulfate and lead acetate, respectively, in inducing an elevation in neoplastic cell transformation frequency in vitro at equivalent concentrations. Hence, DU-UO₂²⁺ appears to have transforming ability slightly greater than that of many other trace heavy metals, which also induce neoplastic cell transformation in vitro, as well as cause tumor formation in animals (40). While additional animal studies are needed to address the effect of protracted exposure and tumor induction in vivo, the implication from our model system study is that the risk of neoplastic induction from internalized DU exposure may be similar to other biologically reactive and carcinogenic heavy metal compounds such as lead and nickel.