RESULTS AND DISCUSSION

ERDC TN-DOER-R4

September 2004

In the current results, any given dose for any given compound resulted in differential expression

(altered expression as compared to that of the solvent control) in approximately 3-6 percent of all

the genes on the array. This result is consistent with observations of other groups who have

conducted gene expression experiments with HepG2 cells using various compounds. For

example, Ishida et al. (2002) found 0.2 to 0.8 percent of genes responded to exposures below

toxic concentrations on the Affymetrix gene chip containing 12,626 oligomers. Compounds

tested included model inducers such as α- and β-naphthoflavone and omeprazole. Most other

studies have been conducted at toxic doses, which resulted in more genes being differentially

expressed. In experiments conducted with Clontech Human Toxicology II arrays containing 588

genes, concentrations of chemicals causing 20 to 30 percent cell death resulted in significant

alterations in gene expression for 4 percent (carbon tetrachloride) and 15 percent (ethanol) of the

monitored genes (Harries et al. 2001). In another study, etomoxir, a mitochondrial enzyme

inhibitor, elicited significant alterations of 9.4 percent of the genes on the Clontech Human

Stress Toxicology array containing 234 genes (Merrill et al. 2002). Detailed analysis of

individual gene expression alterations and how they change with increasing dose can provide

insight into the mechanisms of toxicity of specific compounds. However, to apply this

technology to environmental monitoring, it is only necessary that compounds possess unique

gene expression profiles that are sustained over a large dose range.

In order to determine whether gene responses were characteristic for the different compounds, all

genes that were expressed at a significantly different level than their corresponding control were

combined for a given compound regardless of dose, and the resultant list of genes compared

between the compounds. The Vin diagram in Figure 3 displays the overlap in responding genes

between the treatments; the box in each circle defines the compound the circle represents, with

the total number of responding genes in parentheses; overlapping sectors represent genes in

common for the compounds, with the number enclosed in the overlapping sectors representing

the number of genes in common. Some overlap in differentially expressed genes was observed,

which was expected since selected PCBs and PAHs are known to interact with the same receptor

system responsible for toxicity of TCDD. However, the bulk of the genes were unique to each

treatment, with only two genes responding to all three treatments; only 32 of the combined 320

genes were differentially expressed by more than one compound, supporting the application of

this technique for identifying contaminants and/or contaminant classes. However, for application

as a screening technique for environmental samples it is also desirable that the genes respond in a

consistent manner across a wide range of doses. This requirement reduces the number of selected

genes, as some genes that respond at lower doses may not respond at higher doses, and vice

versa. The wide dose range over which the response was required to be consistent explains the

lack of expected response of some genes, such as cyp1A1 for TCDD. Frueh et al. (2001) used

RT-PCR to determine the concentration-dependent gene response to 2,3,7,8-TCDD in HepG2

cells, and found that many of the genes are not induced until concentrations reach at least 100

pM TCDD; the doses in this study were equivalent to 311, 78, and 23 pM TCDD. The gene was

thus eliminated, as the selection criteria depended on a consistent response over the entire dose

range, and cyp1A1 was only induced in two of the three tested doses.