DNA repair and cell cycle biomarkers of radiation exposure and inflammation stress in human blood

Abstract

DNA damage and repair are hallmarks of cellular responses to ionizing radiation. We hypothesized that monitoring the expression of DNA repair-associated genes would enhance the detection of individuals exposed to radiation versus other forms of physiological stress. We employed the human blood ex vivo radiation model to investigate the expression responses of DNA repair genes in repeated blood samples from healthy, non-smoking men and women exposed to 2 Gy of X-rays in the context of inflammation stress mimicked by the bacterial endotoxin lipopolysaccharide (LPS). Radiation exposure significantly modulated the transcript expression of 12 genes of 40 tested (2.2E-06<p<0.03), of which 8 showed no overlap between unirradiated and irradiated samples (CDKN1A, FDXR, BBC3, PCNA, GADD45a, XPC, POLH and DDB2). This panel demonstrated excellent dose response discrimination (0.5 to 8 Gy) in an independent human blood ex vivo dataset, and 100% accuracy for discriminating patients who received total body radiation. Three genes of this panel (CDKN1A, FDXR and BBC3) were also highly sensitive to LPS treatment in the absence of radiation exposure, and LPS co-treatment significantly affected their radiation responses. At the protein level, BAX and pCHK2-thr68 were elevated after radiation exposure, but the pCHK2-thr68 response was significantly decreased in the presence of LPS. Our combined panel yields an estimated 4-group accuracy of ∼90% to discriminate between radiation alone, inflammation alone, or combined exposures. Our findings suggest that DNA repair gene expression may be helpful to identify biodosimeters of exposure to radiation, especially within high-complexity exposure scenarios.

Figure 1. Radiation-induced transcriptional responses of DNA repair genes in the human ex vivo radiation blood model.

Relative transcript level responses using human blood from 5 healthy human donors measured by quantitative RT-PCR analysis 24 hrs after 2 Gy exposure with respect to sham (0 Gy) transcript levels. Expression of the sham (0 Gy) and 2 Gy transcript responses were calculated relative to the average expression of ACTB (β-Actin). The delta Ct for β-Actin between sham and 2 Gy irradiated samples was <0.3 for all but one sample, which was excluded from this analysis. The fold-change for each gene between sham and irradiated samples was calculated using the delta-delta Ct method. Similar results were obtained when normalized using GAPDH expression as endogenous control (data not shown).

Figure 2. Independent human ex vivo and in vivo confirmation of the radiation response of the 8-gene panel.

The robustness of our panel of 8 non-overlapping radiation biomarkers was confirmed using two published expression array data sets: (A) ex vivo irradiated (0, 0.5, 2, 5, 8 Gy) human blood samples obtained from five independent donors 6 and 24 hrs after radiation exposure (GSE8917; [10]) and (B) human in vivo irradiated blood samples obtained from patients undergoing total body irradiation (GSE20162; [12]). A. Shown is the average of the summed expression for the samples in each exposure group (+/− standard error) normalized to the average expression of the 0 Gy samples for each time-point. B. Shown is the plot of the summed expression of the 8-gene panel of each blood sample in the in vivo study, normalized to the average of the healthy donor samples.

Figure 3. Effects of LPS treatment on radiation responsive DNA repair and cell cycle genes.

Transcript level responses measured by quantitative RT-PCR analysis 24 hrs after LPS treatment of whole blood of two apoptosis, three cell cycle and three DNA repair genes with respect to transcript levels in untreated blood cultures. CDKN1A was strongly upregulated (∼8.2-fold) by LPS treatment alone in the absence of radiation exposure with little variation among donors. BBC3 and FDXR expression was downregulated (∼3-fold and ∼1.5-fold, respectively) by LPS treatment. LPS treatment did not modulate expression levels of GADD45a, PCNA, XPC, DDB2 and POLH (<1.5-fold change in expression compared to untreated samples). ACTB was used to normalize gene expression in samples in which the delta Ct of LPS treated vs untreated was less than 0.3 (donor 1.1, 3, 4.1, 5 and 5.1). GAPDH was not used to normalize since its levels varied depending on the presence of LPS (the average Ct difference between GAPDH in LPS treated and untreated samples was 0.54).

Figure 4. Radiation-induced transcript responses of CDKN1A, BBC3 and FDXR are confounded by LPS treatment.

Transcript level responses measured by quantitative RT-PCR analysis 24 hrs after exposure to 2 Gy, LPS treatment, and combined LPS and 2 Gy of whole blood of CDKN1A (A), BBC3 (B), and FDXR (C) genes with respect to transcript levels in un-treated blood cultures. A. A radiation exposure of 2 Gy in the absence of LPS (left panel) or LPS treatment alone (middle panel) induced CDKN1A to approximately the same level at 24 hrs: 7.3 vs 8.2-fold, respectively (T-Test p = 0.47). LPS treatment in the presence of a 2 Gy radiation exposure induced CDKN1A expression ∼10.2-fold (right panel), which is a 1.4-fold increase compared to 2 Gy alone (T-test p = 0.03). B. In the absence of LPS, radiation induced BBC3 ∼2.7-fold (left panel). LPS treatment alone (middle panel) suppresses BBC3 ∼2.9-fold. LPS treatment in the presence of a 2 Gy radiation exposure induced BBC3 expression ∼1.7-fold (right panel), a ∼1.6-fold decrease in BBC3 expression when compared to 2 Gy alone (T-test p = 0.03). C. In the absence of LPS, radiation induced FDXR ∼17-fold (left panel). LPS treatment alone (middle panel) suppressed FDXR ∼1.5-fold. LPS treatment in the presence of a 2 Gy radiation exposure induced FDXR expression ∼10-fold (right panel), a ∼1.7-fold decrease in FDXR expression when compared to 2 Gy alone (T-test p = 1.2E-04).

Figure 5. Radiation-induced transcript responses of GADD45a, PCNA, XPC, POLH and DDB2 are minimally affected by LPS.

Transcript level responses measured by quantitative RT-PCR analysis 24 hrs after exposure to 2 Gy, LPS treatment, and combined LPS and 2 Gy of whole blood of GADD45a, PCNA, XPC, POLH and DDB2 genes with respect to transcript levels in un-treated blood cultures. Transcript levels of none of these five genes were significantly modulated 24 hrs after 2 Gy exposure in the presence of LPS compared to 2 Gy alone (fold-change <1.4-fold or p>0.03). Interestingly, the radiation response of all five genes was slightly suppressed by LPS treatment suggesting that LPS had a small effect on the 2 Gy response of these genes.

Figure 6. LPS mediated suppression of phosphorylated CHK2-thr68 protein at 24 hrs after 2 Gy exposures.

Protein levels of phosphorylated CHK2-thr68 in protein lysate from cultured whole blood in the presence or absence of LPS (50 ng/ml) were measured by ELISA. Absorbance values were normalized with respect to the average pCHK2-thr68 level in unirradiated donors. In the absence of LPS, radiation induced CHK2-thr68 levels ∼1.6-fold (±0.1) relative to sham irradiated samples, whereas in the presence of LPS, CHK2-thr68 levels were indistinguishable from sham irradiated samples (p>0.4).

Figure 7. Transcript and protein panel discriminates 2 Gy exposure and unirradiated samples, independent of inflammation stress.

In comparison to untreated sham samples, inflammation in the absence of radiation exposure upregulates CDKN1A (red) and downregulates FDXR and BBC3 (green). Samples exposed to 2 Gy radiation only exhibit increased expression of all nine biomarkers, whereas subjects exposed to 2 Gy plus inflammation stress show modified induction of CDKN1A, FDXR and BBC3 and abrogation of the phosphorylation of CHK2 protein. The arrows in the radiation and inflammation combined treatment group indicate the direction of expression relative to the radiation alone group.

Figure 8. Classification (4-class) accuracy of the transcript and protein panel.

Classification accuracy based on ten 10-fold cross-validation as a function of the number of markers considered, based on order determined during filtering with the Gini index. The four classes used in this analysis are: radiation only (R), inflammation stress only (L), combined exposures involving both radiation and LPS (RL), and samples with no radiation exposure and no LPS treatment (N). Marker order is: PCNA, CDKN1A, pCHK2-thr68, BBC3, FDXR, DDB2, XPC, POLH, and GADD45a. Maximum classification accuracy was 0.88 for the top 5-marker set.