Chromosomal Radiosensitivity in Two Cell Lineages Derived from Clinically Radiosensitive Cancer Patients Free

Around 1% to 5% of cancer patients suffer significant side effects in normal tissues exposed to ionizing radiation during radiotherapy. Radiotherapy is an effective therapy for cancer treatment, but the treatment dose intensity is generally restricted to minimize the incidence of these severe reactions. This imposes tumor cure limitations on most patients. Therefore, a major goal of radiation biology research is to develop efficient predictors that could identify these radiosensitive individuals before treatment. Such predictors might involve single genes or proteins, but given the complexity of mammalian ionizing radiation responses (1), assays that test a functional process may be more likely to yield positive identifiers of radiosensitivity. Ultimately, successful predictive assays may consist of a number of functional assays and possibly include more global approaches such as microarrays and proteomics. In turn, this should allow individualization of radiotherapy (ref. 2; reviewed in ref. 3) and improvement of tumor control rates and/or a reduction in the incidence of severe radiotherapy side effects.

There are a number of lines of evidence pointing to a genetic basis to clinical radiosensitivity (2, 3). In humans, DNA double-strand break repair proteins such as ataxia telangiectasia mutated (ATM) and DNA ligase IV, when compromised, confer a clinical radiosensitive phenotype (3–5). Abrogation of the function of these same proteins also confers radiosensitivity at the cellular level (3, 6, 7). Animal loss-of-function models also support the idea that other genes involved in DNA repair, such as DNA-PKcs, may cause clinical radiosensitivity in humans (8). Clearly, DNA repair proteins are an important gene category involved in mammalian organismal radiosensitivity. Consistently, other DNA double-strand break repair proteins, when dysfunctional, cause cellular radiosensitivity (e.g., Ku70, Ku80, Artemis, XRCC2, XRCC4, BRCA1, and BRCA2; see refs. 9–16, 3 for review). Therefore, it is reasonable to predict that DNA repair protein dysfunction is involved in eliciting a clinical radiosensitive phenotype, at least in some cases.

A good measure of cellular radiosensitivity is chromosomal aberration frequency. Chromosomal aberrations are commonly associated with altered DNA repair function (17). Aberrations can result in misrepaired DNA that can lead to DNA lesions such as dicentric chromosomes and gene amplification, which are implicated in carcinogenesis (18, 19). Gene amplification products and other acentric chromatin fragments are commonly found in small nuclear membrane–bound structures called micronuclei (20–22). The numbers of micronuclei are reported to increase, usually as an approximate linear function of dose, when cells are exposed to DNA-damaging agents such as ionizing radiation or other clastogens (23–25). The micronucleus assay is highly suited to application in the clinic because it is relatively easy, reproducible, and results can be obtained in a timely manner to facilitate clinical decision making (26–28).

Micronucleus frequency in cell lines from radiosensitive individuals has been determined in a variety of contexts. For example, lymphoblastoid cell lines (LCL) with mutant ATM protein revealed a significant increase in micronucleus frequency using the micronucleus assay (29). Furthermore, radiosensitivity in lymphocytes from normal individuals versus ATM heterozygotes following irradiation were distinguishable using the micronucleus assay (30, 31). Correlations between micronucleus frequency and acute or late clinical radiosensitivity in fibroblasts have also been observed (32). Recently, a strong correlation between clinical acute and late reactions and micronucleus frequency in lymphocytes from cervical cancer patients treated with radiotherapy was observed (33). In addition, ionizing radiation treatment of peripheral blood lymphocytes of prostate cancer patients who showed late radiosensitivity reactions compared with normal reactors have also shown a clear difference in micronucleus frequency (34). Additional studies have attempted to correlate the radiosensitivity between two cell lineages using various radiosensitivity assays, some showing a good correlation (35–37) and others showing weak or no correlation (38–41).

Here we have made use of different cell lines from our bank of radiosensitive and control cancer patients. Specifically, primary fibroblast cell lines and EBV-transformed blood B cell (lymphoblast) cell lines from highly clinically radiosensitive patients (42–44) were used to assess the relationship between clinical radiosensitivity and micronucleus formation following ionizing radiation. We show that cell lines from some radiosensitive patients have significantly higher frequencies of total and/or multiple micronuclei per cell compared with controls who had normal radiotherapy reactions. Additionally, cells from one individual showed significantly higher micronucleus frequencies in both cell lineages, strongly suggesting a constitutional radiosensitivity in this patient. This patient's LCLs had high micronucleus frequencies comparable with an ATM-deficient LCL at particular radiation doses. This same patient showed a significantly higher micronucleus frequency compared with controls in both cell lineages tested. In some cases, subclassifying our cohort according to tumor type increased the observed difference between radiosensitivity and controls, consistent with recent studies (33, 32). Our observations support the idea that the micronucleus assay in some instances may have clinical use in predicting radiosensitivity.

To obtain samples from truly clinically radiosensitive patients, a careful selection was made based on detailed discussion with the treating doctor. Patients were excluded if there was any chance of dosimetry errors/hotspots or alternative causes of apparent radiotherapy toxicity. Additionally, a detailed personal and family history was obtained to exclude any known genetic syndromes. Furthermore, in an effort to further avoid any misdiagnosis, we elected to eliminate any reactions that were classified as Radiation Therapy Oncology Group grade 2 therefore making a clear demarcation of symptoms between the controls and radiosensitive individuals (42–44). Primary fibroblasts were derived from 21 cancer patients who had normal (n = 7) or severe tissue reactions (acute reactors, n = 4; late reactors, n = 10; Radiation Therapy Oncology Group grade 3 or 4; Table 1). A 4-mm skin biopsy was minced and put into DMEM containing 15% fetal bovine serum and 20 μg/mL gentamicin in a 10-cm² plate and cultured at 37°C, 5% CO₂ in a humidified incubator. The tissue was incubated for 2 weeks at which time the cells were detached with trypsin, plated in 150-cm² flasks and grown to confluency. LCLs were established as previously described (42). Briefly, lymphocytes were purified from fresh whole blood using a Ficoll separation method (42). LCLs were subsequently immortalized using EBV transformation (45). Thirty-eight LCLs (controls, n = 6; acute reactors, n = 8; both acute and late reactors, n = 2; late reactors, n = 20; ATM heterozygote, n = 1; ATM homozygote, n = 1) were used for this study. Six controls, two acute reactors, and five late reactors were assessed in both cell-type systems. The cells were prepared and micronuclei were carefully scored according to standardized criteria (28, 46). Briefly, log-phase cells were irradiated at 0, 1, 2, or 4 Gy using a ¹³⁷Cs source delivering 1 Gy/1.5 minutes at room temperature. Cells were incubated for 44 hours at 37°C before addition of cytochalasin B, which blocks cells at cytokinesis, to a final concentration of 2 μg/μL. Twenty-four hours later, the cells were fixed and stained with 4′,6-diamidino-2-phenylindole and propidium iodide and micronuclei were scored using a fluorescent microscope at 100×. For each cell line, at least two separate counts of 100 binucleated cells were made except in the following cases where only one count was made: fibroblasts at 0 Gy: control 1 (C1), C2, C7, acute 2 (A2), A4, late 7 (L7), L8; fibroblasts at 2 Gy: L3; LCLs at 0, 1, and 2 Gy: C6. Multiple micronuclei were defined as micronuclei found two or more times in a single binucleated cell. Error bars were determined by calculating the SD on two or more replicates of any one cell line or by using the mean of each sample for calculating the SDs of the average values for entire subgroups. Statistical significance was determined using the unpaired t test to generate two-tailed P values. The average follow-up time for patients categorized in the control and late reactive groups was 45 and 46 months, respectively. All patients gave written informed consent and the study was approved by the Ethics Committee at the Peter MacCallum Cancer Centre (approval no. 96/39).

A cytokinesis block micronucleus assay (28) was done to determine the frequency of micronuclei in primary fibroblasts and LCLs derived from patients who had severe reactions from radiotherapy for cancer treatment. Cell lines were treated with ionizing radiation and compared with basal micronucleus frequencies. Of particular interest was the testing of two different cell lineages to assess the generalizability of radiosensitivity data across cell lineages. It was also of interest to subclassify into cell lines from patients with acute and late reactions, because the underlying mechanisms of these two reaction classes are presumably different (47). Classification according to tumor type was also of interest.

Primary fibroblasts were treated with 0, 2, or 4 Gy of ionizing radiation. As expected, micronucleus frequency increased with increasing dose (Fig. 1A). The mean number of micronuclei found in the combined radiosensitive (acute and late) cohort trended towards being higher than the controls at every dose including basal levels (no ionizing radiation; Fig. 1A). Certain fibroblasts from radiosensitivity reactors such as A4 or L7-10 had a significantly (P < 0.05) higher frequency of micronuclei compared with the mean of the control samples at 4 Gy (Fig. 1C). Due to the high amount of interindividual heterogeneity observed (Fig. 1C), the differences between the means of micronuclei frequencies of individual cell lines at particular doses were not significant (Fig. 1A). Subclassification into acute and late reaction types revealed that 40% of the late reactors had a statistically significantly greater micronucleus number than the mean control value (Fig. 1C). There was also a significant difference (P < 0.05) at 2 Gy for these same late reactors compared with controls (Fig. 1C). Nearly 20% of the radiosensitive fibroblast cell lines had a mean value of ≥25 micronuclei per 100 binucleated cells at 4 Gy, whereas no control cell line showed this (Fig. 1C).

Radiosensitivity may be tissue specific (48, 49). We therefore tested the chromosomal radiosensitivity of cell lines from an additional cell lineage (i.e., LCLs). LCLs were treated with 0, 1, and 2 Gy of ionizing radiation and the micronucleus frequency determined. Heterogeneity of micronucleus frequency in LCLs was again apparent (Fig. 1B and D). Across the different doses, the combined mean micronucleus frequency consistently trended towards being higher in radiosensitive versus the controls. Twenty-eight percent of LCLs (A10, A11, L6, L10, L23, L24, L5, and L25) showed significantly (P ≤ 0.05) increased differences in micronucleus frequency relative to the mean of the controls at 2 Gy (Fig. 1D). Some cell lines (A11, L6, L5, and L25) showed high micronucleus frequencies, with one cell line (L10) derived from a late reactor having a very high micronucleus frequency, similar to ataxia telangiectasia cells (Fig. 1D). Moreover, 42% of the LCLs derived from radiosensitive patients had a higher micronucleus average than any of the controls (Fig. 1D).

It is possible that a more discriminative radiosensitive phenotype could be obtained from cells that had multiple micronuclei (micronuclei found two or more times in a single binucleated cell). That is, multiple micronuclei may represent a cellular condition that may be more prevalent in highly radiosensitive cells, reflecting greater chromosomal damage, or reduced repair, per unit ionizing radiation dose. To test this, cells that contained micronuclei were also scored for the presence of multiple micronuclei (Fig. 2). Certain cell lines (fibroblasts: L7, L9, and L10; LCLs: A3, A7, A10, A11, L6, L10, L17, L18, L20, L24, L5, and L25) had significantly higher frequencies of multiple micronuclei (P < 0.05) compared with the mean of the controls. The multiple micronucleus values obtained generally reflected the results for total micronucleus frequencies. However, the LCL group had an increase in significantly different micronucleus frequencies for cells with multiple micronuclei, compared with the number of cells having micronuclei. Compared with 27% of LCLs with one or more micronuclei, 41% of the radiosensitive late reactors showed significantly higher levels of LCLs containing multiple micronuclei compared with the average of the control group (Fig. 2D).

We had seven pairs of cell lines (i.e., both fibroblast and LCL) from radiosensitive cases and five pairs of control cell lines available. To determine if chromosomal radiosensitivity was, or was not, conserved across cell lineages, we observed the micronucleus frequency to see if it correlated between the two cell lineages. Of the five cell lines derived from late reactors (L1, L2, L5, L6, and L10) and the two cell lines derived from acute reactors (A1 and A3), one of these, L10, had a significant (P < 0.05) increase in micronuclei and multiple micronucleus frequency compared with the mean values of controls at 4 Gy (in fibroblast) or 2 Gy (in LCLs) in both cell lineages (Fig. 1C-D and Fig. 2C-D).

We also tested LCLs derived from an ataxia telangiectasia patient. As expected (29, 32, 50), these cells had a very high frequency of ionizing radiation–induced micronuclei, although basal (0 Gy) levels were not elevated compared with other cancer patient cell lines (Fig. 1D and Fig. 2D). One of the LCLs from a severe reactor (L10) had as many micronuclei as did the ATM-deficient cell line at 2 Gy (Fig. 1D and Fig. 2D). However, at a lower dose of 1 Gy, the ataxia telangiectasia LCLs were distinguishable by having the highest numbers of micronuclei, especially evident in the class of multiple micronuclei (Fig. 2D). We also separately examined a subgroup of LCLs derived from patients who had breast cancer, because this was the largest cancer-type subset (Table 1), and recent studies have shown good correlations between radiosensitivity and micronucleus frequency in some tumor types (e.g., in prostate and cervical cancer patient lymphocytes; refs. 33, 34). The severe acute reactors in this group showed a significant difference (P = 0.01) of the average multiple micronucleus values compared with controls (Fig. 3). Although a trend was seen with the late reaction group, statistical significance was not shown (data not shown).

We found considerable heterogeneity in micronucleus frequency after ionizing radiation between both primary fibroblasts and LCLs derived from radiosensitive and control cancer patients. However, a substantial subset of cell lines derived from acute and late radiosensitive patients showed a significantly higher frequency of micronuclei than their control counterparts in both cell types after ionizing radiation. One individual late reactor (L10) had a very high micronucleus frequency in both primary fibroblasts and LCLs. This observation suggests that this patient may have constitutional radiosensitivity (i.e., radiosensitivity of different tissues) and, furthermore, that chromosomal radiosensitivity as assessed by the micronucleus assay is replicated across very different cell types. This is consistent with a report that found a moderate correlation between LCL and fibroblast radiosensitivity on a growth assay to assess radiosensitivity (35).

Furthermore, a correlation was found between fibroblast and lymphocyte radiosensitivity using the comet assay (37) and between epidermal skin cells and lymphocytes using a DNA double-strand break assay (36). Not all of our paired cell lines derived from clinically radiosensitive patients for which we have tested post-ionizing radiation micronucleus induction showed a significant correlation between the two. This suggests that the radiosensitivity of some individuals may be due to different factors, some of which may be cell type specific. The interindividual variation in micronucleus frequency was large, as has been observed in a number of other studies (32, 33). However, the average micronucleus frequency of pooled acute and late radiosensitive cell lines versus controls did show a trend that was higher in the radiosensitive cohort. As expected, the ataxia telangiectasia cell line showed a very high number of micronuclei at different doses, contrasting with a patient LCL that showed high levels of micronuclei at 2 Gy but not as high as the ataxia telangiectasia cell line at 1 Gy. This suggests that some cell lines (e.g., mutant ATM cells) may have a lower threshold for damage that causes genomic instability in the form of chromosome fragmentation.

Previous studies have provided discordant results in attempts to correlate micronucleus frequency with normal tissue reactions. Some studies have used samples derived from patients with different cancer types and have not found a correlation (51, 52). Recent studies of a more homogeneous selection of cell lines derived from radiosensitive patients of one cancer type (cervical or prostate cancer) and treatment scheme resulted in a high concordance among radiosensitive patients using the post-ionizing radiation micronucleus assay (33, 34). Therefore, it seems that some additional variation can be introduced by the use of cell lines from patients having a variety of cancer types. A significant number of our samples are from patients of one cancer type. Cell lines from seven of eight of our controls and 18 of 37 of our radiosensitive cohort were obtained from patients that had breast cancer. Interestingly, some studies have found that cancer patients have a higher percentage of baseline micronuclei than the general population (53, 54). Furthermore, cells from breast cancer patients as assessed by the micronucleus assay have been found to have significantly higher micronucleus levels than the general population (53). Given the high number of breast cancer controls in our cohort, it may be that we are underestimating the difference between some severe reactors and the control population in our study, because breast cancer patients have higher basal levels of micronuclei. Our breast cancer patients had significant differences for the average micronucleus frequency in acute reaction patient LCLs versus controls (Fig. 3). The LCLs and fibroblasts from late reactors also showed a trend towards higher micronucleus frequencies in breast cancer patient cell lines compared with controls (P = 0.085 and 0.07, respectively; data not shown). This means that doing the micronuclei on specific groups (e.g., breast cancer) may allow for a better discrimination between radiosensitive and nonradiosensitive patients using the micronucleus assay.

Some of the heterogeneity of micronucleus frequency from the radiosensitive patient cell lines could be due to the many potential and proven genetic contributors to radiosensitivity (25, 29, 55). Other factors, such as tumor type, tissue environment, radiotherapy regimen, and variation from different micronuclei counters could also play a role in the variation we observed.

Some patients that exhibit clinical radiosensitivity may have genetic determinants that will not be picked up by the micronucleus assay. If this is the case, any particular assay such as the micronucleus assay will only be able to detect a proportion of the total population of individuals susceptible to radiosensitivity. Our experiments indicate that there is a high likelihood that a significant number of radiosensitive patients could be predicted to be susceptible to severe reactions using the micronucleus assay and therefore could potentially have their course of radiotherapy treatment individualized, or another therapy delivered, to suit their possible tissue sensitivity. Additionally, the micronucleus assay could be used as part of a multiassay screen for radiosensitivity to facilitate the individualization of radiotherapy courses.

We thank Dr. Martin Lavin, Johnathan Ramsay, and Geoff Birrel for sharing some of the LCLs.