Lesion complexity drives age related cancer susceptibility in human mammary epithelial cells

Dose dependent increase in centrosome aberrations

Centrosome aberrations were assayed as an indicator of genomic instability and a potential surrogate marker for increased cancer risk. Previous studies have noted an increase in centrosome aberrations following radiation exposure [24–26]. However, the effect of dose and lesion complexity on HMEC strains derived from women of different ages has not been previously defined. Cells in early passage (p4) were exposed to a low and high dose of Cs or Ti 300 MeV/n (herein referred to as Ti) radiation and the frequency of centrosome aberrations was defined for each strain. Centrosomes were visualized using fluorescent microscopy by detecting fluorescently tagged pericentrin, a major component of the pericentriolar matrix surrounding the centrosome. Representative images of cells with normal and aberrant numbers of centrosomes are shown in Fig. 1A. Cells with 1-2 centrosomes were characterized as normal (Fig. 1A: a, b) and >2 were considered aberrant (Fig. 1A: c, d). The frequency of cells containing aberrant centrosomes in each strain was graphed relative to its unexposed baseline (Fig. 1B). The black horizontal line indicates the mean frequency of aberrant cells in the population with each exposure. The relative proportion of aberrant cells did not notably change in cells exposed to low doses of radiation relative to controls (Fig. 1B). In contrast, when exposed to a high dose (HD), a 1.3-fold increase in the frequency of aberrant cells was noted (range 0.3-2.7) for Cs and a 1.6-fold (range 0.56-2.6) for Ti (Fig. 1B). The frequency of centrosome aberrations significantly increased following high doses in comparison to low dose (LD) following both Cs, which creates mainly simple lesions (p<0.05) and for Ti which creates primarily complex lesions (p<0.01).

Figure 1. Effect of dose on the frequency of centrosome aberrations. Fifteen HMEC strains derived from individuals of various ages were exposed to a low dose (LD) and high dose (HD) of two types of radiation namely Cs and Ti ions. Roughly equitoxic doses of Cs and Ti were used for exposures (LD-CS = 0.12 Gy; HD-CS = 0.8 Gy; LD-Ti = 0.05 Gy; HD-Ti = 0.5 Gy). Aberrant numbers of centrosomes per cell were visualized by indirect immunofluorescence microscopy using an anti-pericentrin antibody (A). Staining was carried out within days of fixation, and the number of cells containing aberrant numbers of centrosomes was counted in random regions of the slide. The majority of the cells in irradiated and unirradiated samples show one or two pericentrin foci (A: a, b). Cells with supernumerary centrosomes showed numbers of centrosomes ranging between 3 to as many as 7, and were scored as aberrant (>3P) (A: c, d). Cells were processed 9 days post-exposure and the percentage of aberrant cells in each strain was plotted relative to the unexposed population (B). Data are based on four independent experiments for high dose and two independent experiments for low dose. The frequency of cells with supernumerary (>3P) centrosomes was graphed against age of the individual from which the strain was derived (C-F). Regression analysis was used to model the relationship. The effects of lesion complexity on this relationship at a low (C) and high dose (D) were graphed. The effect of dose for Cs (E) and Ti ion exposure (F) were also graphed. Averaged data and the regression line for Cs and Ti ion exposed samples is shown with blue and red, symbols and lines respectively. In Fig. 1E and 1F, LD and HD are distinguished using dotted and full lines, respectively.

Impact of age of exposure on centrosome aberrations

To assess age as a prognostic factor in radiation induced cancer risk, we next investigated the relationship between the age of the individual who provided the mammary tissue and the frequency of aberrations induced using linear regression (Fig. 1C-F). For clarity, low and high doses are distinguished using dotted and full lines respectively and centrosome aberration data from Cs (blue) and Ti (red) exposures are presented. The baseline for centrosome aberrations in unexposed cells shows a slight increase with age that is not significant (black line Fig. 1C-F). The top panel (Fig. 1C and 1D) compares the frequency of aberrations induced by Cs and Ti exposures at low (left) and high doses (right) and reveals that the percentage of cells with supernumerary centrosomes increases linearly with age, particularly when exposed to Ti. The increase in centrosome aberrations after high dose Cs exposure is negligible and appears to be independent of age. Our data suggest that for every year increase in age, the centrosome aberration frequency increases by ~ 0.02 fold for both low and high dose Ti exposure. The linear trend is significant at low dose (p<0.05) where samples over the age of 50 years reveal an aberration frequency significantly higher than control. A comparison of dose effects within each radiation type suggests that the slopes are not significantly different for Cs. For Ti, the slopes for LD and HD appear parallel; with the high dose exposure exhibiting a 1.9-fold increase in centrosome aberrations relative to the low dose, for the same age group (Fig. 1D). As polyploidy is often observed in cells with higher levels of centrosome aberrations, ploidy was assessed following a high dose Ti exposure using flow cytometry analysis of Propidium Iodide (PI) stained samples. Fitting with an increase in centrosome aberrations, an age dependent increase in polyploidy (cells >4C DNA content) was observed with radiation exposure (data not shown).

Distribution of the number of centrosomes in aberrant cells after radiation exposure

Next we characterized how the nature of the centrosome aberrations was impacted by simple and complex damages. Aberrant cells (>2 centrosomes) identified while scoring a minimum of 400 cells were classified into two sub-groups; cells with 3 pericentrin foci (3P) and those containing 4 or more pericentrin foci (≥4P, Fig. 2). Representative images for cells with 3P and ≥4P foci/cell are shown in Fig. 1A: c and d, respectively. Cells containing greater numbers of centrosomes per cell (≥5) were noted with very low frequency and hence were pooled with the >4P subgroup. For both radiation types, the 3P subgroup was significantly increased with high dose as compared to cells exposed to low dose (Fig. 2A). The mean increase for 3P was not significantly different between Cs and Ti (Fig. 2A). In comparison, there is a notable increase in the ≥4P subgroup with high dose Ti exposure that is significantly higher than low dose (p<0.05 Fig. 2B). We next employed regression analysis to assess how the frequency of these sub-groups varies with age (Fig. 2C, 2D). The relationship between the number of centrosomes per cell and age revealed interesting trends. The proportion of cells with 3P showed an increasing trend with age for high dose exposure for both radiation types (Fig. 2C). At low doses, the fraction of cells with 3 centrosomes showed a significant linear increase with age (p<0.05) for Ti exposure, but not for Cs exposure (Fig. 2C). However this relationship was not noted for the ≥4P population exposed to low doses of either radiation type (Fig. 2D). In contrast, for high dose exposures, the ≥4P population displayed an age dependent increase for Ti but a decrease with Cs exposure (negative slope). With high dose Cs exposure, the inverse relationship between age and fold change in centrosome aberrations relative to control, appears to be a discrepancy at first glance. However, this age-associated decrease when calculating fold values can be attributed to the higher baseline centrosome aberration frequency in unexposed cells from older people (Sup. Fig. S1).

Figure 2. The number of centrosomes per cell varies with dose and age. Centrosome aberrations were further classified into cells containing 3 centrosomes (3P= 3 pericentrin foci) and those that had 4 or more (≥4P pericentrin foci). The fraction of cells with 3P and >4P were graphed relative to control. Roughly equitoxic doses of Cs and Ti were used for exposures (LD-CS = 0.12 Gy; HD-CS = 0.8 Gy; LD-Ti = 0.05 Gy; HD-Ti = 0.5 Gy). The dose effect on the change in 3P (A) and ≥4P (B) populations relative to control is shown. Wilcox test was used to determine statistical significance, *=p<0.05. The normalized proportion of aberrant cells with either 3P (C) or >4P (D) was plotted as a function of age. Regression lines analyzing the trends as a function of age were fitted to these data.

To further investigate the relationship between the number of aberrant centrosomes per cell and age, data was grouped into three age ranges, namely young (16-30 yo), middle (31-50 yo) and old (51-72 yo). The mean fold change in 3P and ≥4P frequency with Ti exposure was plotted for all three aged cohorts (Fig. 3). With low dose Ti exposure, we observed a significant increase in the 3P population in the older age group in comparison to both the middle (p<0.05) and younger age group (p<0.05), but no change in the proportion of cells with ≥4P centrosomes (Fig. 3A). The mean increase in 3P/cell for the different sub-groups was 0.86 ± 0.16 for the older subgroup (n=4), 0.27 ± 0.07 for the middle-aged group, (n=5) and 0.32 ± 0.12 for the younger cohorts (n=5). Upon exposure to high dose Ti, the mean frequency of cells containing both 3 and ≥4 centrosomes increases with age, although this increase is not significant (Fig. 3B).

Figure 3. Centrosome aberrations in age-grouped cohorts. Strains exposed to Ti at a low dose (LD= 0.05 Gy) or high dose (HD= 0.5 Gy), were sub-grouped into young, middle and old. The frequency of cells with 3P and ≥4P was compared between the different age ranges (A and B respectively). The horizontal black dotted line represents the mean frequency of centrosome aberrations in the younger age groups. Error bars represent maximum and minimum values within each age group. (C) Centrosome aberrations were assessed at 3 days (D3), 9 days (D9) and 12 days (D12) post-exposure. Regression lines that model the relationship between age and the normalized mean centrosome aberration frequency are plotted for the various days. Data represent two independent sets for D3 and D12 for Ti ion and D9 for CS. Data from 4 independent experiments are plotted for D9 Ti.

Time course of generation of centrosome aberrations following radiation exposure

There is strong evidence that centrosome aberrations can cause multipolar spindles that retard mitosis. The inherent growth advantage of normal healthy cells compared to cells containing centrosome aberrations thus poses a serious limitation in assessing frequency of this population at long times post-exposure. Centrosome aberrations have been observed as early as 3 days after radiation exposure [26] and can potentiate further genomic instability in the progeny of radiated cells. To define the time dependence of induction of centrosome aberrations in our aged population, we compared centrosome aberrations elicited by Ti exposure at Day 3 (D3) to those that were evaluated at later time points post-exposure; namely Day 9 (D9) and Day 12 (D12) (Fig. 3C). 4^th passage HMEC strains were irradiated with 0.5 Gy of Ti and plated on coverslips one day prior to fixation. Regression lines that model the relationship between age and the normalized mean centrosome aberration frequency (>2P) were plotted for various days. These time course experiments confirm that the numbers of cells with aberrant centrosomes increase as early as D3, but the level of increase relative to control was similar in all age groups. Fitting these data to a linear model showed that the centrosome aberration frequency shows a linear increase with age at D9, as noted in Fig. 2 but is significantly associated with age of the strain at D12 (P<0.05). Data represent an average of 2 experiments for each sample for D3 and D12 and 4 experiments for D9. Only a subset of 8 strains spanning the range of ages was assessed for D3 (16, 19, 21, 49, 51, 66, 68 and 72 years). Two HMEC strains, a 27 y and 72 y old, were not processed for D12 due to insufficient cell numbers. These data reveal that an age dependent increase in centrosome aberrations continues to be observed even ~2 weeks post-exposure. In order to test whether repair signaling is elevated at this time point, we also simultaneously analyzed pATF2 foci on the D12 samples. We noted that the fraction of cells exhibiting aberrant pATF2 foci after Ti exposure decreased with increasing age, though this decrease was not significant (data not shown).

Radiation exposure increases the number of stem/progenitor cells

Stem/progenitor (S/P) cells have been shown to express higher levels of aldehyde dehydrogenase (ALDH). This population described as ALDH positive (ALDH+) can be used to quantify S/P cell numbers by flow cytometry. Samples were exposed to both a low and high dose of Cs and Ti; subsequently cultured and processed 9 days post-exposure. Single cell suspensions were analyzed for ALDH+ cells to assess radiation-dependent changes in S/P cells. Only eleven HMEC strains were used for ALDH analysis, as we did not have sufficient numbers of viable cells for 4 strains (19, 27, 46 and 72 year old). In analyzing the entire cohort, we noted a radiation-induced increase in the mean proportion of ALDH+ cells relative to control for both types of radiation (Fig. 4). Of interest, this increase was noted even with low dose (5 cGy) exposures. A comparison of the effect of two different doses of Cs indicates that median increase in ALDH+ cells is significantly greater with high dose relative to low dose exposure. A similar although not significant trend was observed with Ti exposure (p=0.08). However, the number of stem/progenitor cells did not significantly differ between exposures that were due to simple as compared to complex damages.

Figure 4. The impact of dose on stem cells numbers. Eleven HMEC strains derived from individuals of various ages were exposed to a LD and HD of Cs and Ti ion. Roughly equitoxic doses of Cs and Ti were used for exposures (LD-CS = 0.12 Gy; HD-CS = 0.8 Gy; LD-Ti = 0.05 Gy; HD-Ti = 0.5 Gy). Cells were passaged and processed 9 days after radiation exposure. S/P cells were assessed based on ALDH+ signal using flow cytometry. DAPI negative cells defined the live population. Doublets were eliminated and the ALDH+ signal was assessed with reference to the DEAB control sample. The proportion of ALDH+ cells were plotted relative to the unexposed sham radiated control. Data are based on two independent experiments for high and low dose. Blue and red symbols represent strains exposed to Cs and Ti ion respectively. Empty symbols represent low dose and solid symbols high dose exposures.

Age of the cell strain impacts radiation-dependent changes in S/P population

Regression analysis was carried out to assess the relationship between age and radiation-dependent increase in S/P cells (Fig. 5). While there does not appear to be a correlation between the frequency of ALDH+ cells and age with exposure to low dose Cs, high dose exposure elicits a rising trend with increasing age (Fig. 5A). In contrast, when HMEC strains were exposed to complex damages from Ti exposure, the proportion of ALDH+ cells increased in an age-dependent manner at a low but not a high dose. High dose Ti exposure uniformly increased the proportion of S/P cells in all age groups (Fig. 5B). We further compared dose effects within each age group following exposure to both types of radiation. For the most part, age-grouped analysis confirmed the trends observed with linear regression with one exception. Although there was an age-related increasing trend in S/P numbers with high dose Cs exposure (Fig. 5A), age-grouping indicated that the mean frequency of S/P cells in the older cohort was higher than both the young and middle age groups (Sup. Fig. S2A). In addition, within each aged cohort (young, middle and old), we observed a relative increase in the proportion of ALDH+ population with high dose Cs exposure in comparison to low dose. A similar dose dependent increase has also been noted within each of the three age groups after Ti exposure (Sup. Fig. S2B). Thus an age and dose-dependent increase in numbers of S/P cells is observed following exposure.

Figure 5. The impact of age on stem cell numbers. S/P cells were assessed in HMEC strains based on flow cytometry assessment of ALDH+ signal. The fraction of ALDH+ cells relative to the unexposed control was plotted against age of the individual the strain was derived. Data are based on two independent experiments for high and low dose. Roughly equitoxic doses of Cs and Ti were used for exposures (LD-CS = 0.12 Gy; HD-CS = 0.8 Gy; LD-Ti = 0.05 Gy; HD-Ti = 0.5 Gy). Blue and red symbols represent strains exposed to Cs and Ti ion respectively. Empty circles represent low dose and solid circles high dose exposures. Dotted and solid lines represent the regression lines and note the relationship of stem cell numbers with age for each radiation type.

Mechanistic clues that underlie age effects

Cells activate various signaling networks in response to radiation [27]. These pathways play critical roles in regulating various cell fates, such as cell survival, proliferation, apoptosis, senescence and repopulation and cellular reprogramming. We hypothesized that the cellular response to the radiation would be unique to the nature of damage and show age dependence. To test this hypothesis, samples were processed at 9 days post-exposure to both types of radiation. We assessed the global phosphorylation profile of key nodes in three different stress-activated signaling pathways, namely the receptor tyrosine kinase signaling pathway, TNFα signaling pathway and TGFβ signaling. The fold change in phosphorylation relative to unexposed control was graphed from two independent experiments (Fig. 6). Regression analysis revealed a pronounced age effect in a subset of these proteins that highlights differences in response to simple and complex lesions. While, low dose Cs exposure appeared to show an inverse relationship for phosphorylation of ERK/MAP kinase ½ (Thr¹⁸⁵/Tyr¹⁸⁷) with age (Fig. 6A), we noted a uniform increase in phosphorylation with high dose Cs exposure that was independent of age (Fig. 6B). The level of pAKT (Ser⁴⁷³), pCREB (Ser¹³³) and total TGFβRII protein levels remain unchanged with Cs dose and did not show an association with age (Fig. 6A, 6B). In contrast, in response to complex lesions, we noted a linear relationship between age and the relative increase in level of phosphorylation of ERK/MAP kinase ½ (Thr¹⁸⁵/Tyr¹⁸⁷), pAKT (Ser⁴⁷³) and pCREB (Ser¹³³) following both doses (Fig. 6C). The increase was especially significant for pAKT with low dose exposure (P<0.05). While total protein levels of TGFβRII appear to reciprocally decrease with age at low doses of Ti exposure, we do not observe an association with age at high doses (Fig. 6D). Testing the association between these signaling nodes and previously assessed surrogate markers of cancer risk using Pearson’s correlation analysis revealed a significant positive correlation between ALDH+ and centrosome aberrations (P<0.05) and pATF2 and SMAD4 (P<0.05) (Sup. Table S1).

Figure 6. Phospho-protein patterns as a function of age. Phospho-profiles of key proteins in the RTK, TNFα and TGFβ signaling pathway were assessed in all fifteen of the HMEC strains 9 days post-exposure to Cs and Ti ions. Roughly equitoxic doses of Cs and Ti were used for exposures (LD-CS = 0.12 Gy; HD-CS = 0.8 Gy; LD-Ti = 0.05 Gy; HD-Ti = 0.5 Gy). Four proteins that showed a change with age were plotted. These include pERK, pCREB, pAKT and total levels of TGFβRII. Dotted lines and solid lines represent regression lines. Trends exhibited with a low (A) and a high (B) dose of Cs are shown. Similarly, trends observed at a low (C) and a high (D) dose of Ti ions are shown. Data from two independent experiments were plotted for each radiation type and dose.

Phospho-profiles provide information on processes that underlie strain specific variation

While a majority of the cell strains showed an increase for centrosome aberrations and stem cell numbers with high dose relative to low dose, for both Cs and Ti exposures, we noted some strain-specific exceptions to this pattern. For stem cell assessments, the strain from the 19 y old showed a dose-dependent decrease with Cs but not Ti exposure (Fig. 7A). In addition, the strain from the 68 y old revealed a dose-dependent decrease in the fraction of ALDH+ cells when exposed to Ti (Fig. 7C), and showed a dose-dependent decrease in the relative frequency of centrosome aberrations generated by both radiation types (Fig. 7D). To examine if protein phosphorylation profiles could provide clues to explain these strain-specific trends, we compared the phosphorylation signatures of these strains to another strain belonging to the same age group. In these strains, we assessed the fold change in the level of the phospho-protein induced by high dose relative to low dose (Fig. 7B, 7E). The data shows that while a HMEC strain from a 19 y old shows levels of pCREB, pNFKβ, pERK1/2, pAKT, p70S6K and pSTAT5 that are notably decreased in the high dose exposed cells relative to the low dose, the strain from the 16 y old shows an increasing trend with dose (Fig. 7B). Contrasting phospho-profiles between the strains from the 66 y old and 68 y old showed that the levels of pCREB, pNFKβ, p38MAPkinase, pERK1/2, pAKT, p70S6K and pSTAT5 are all increased in the 66 y old as compared to the 68 y old following a high dose exposure (Fig. 7E). Interestingly, the levels of pSmad2 (Ser⁴⁶⁵/Ser⁴⁶⁷), pSmad3 (Ser⁴²³/Ser⁴²⁵), as well as total protein levels of TGFβRII and SMAD4 were notably higher in the strain from the 66 y old exposed to low dose Ti relative to high dose Ti compared to inductions in a strain of a woman of similar age (68 y old) (Fig. 7E). These data reveal unique individual expression signatures that change based on dose and lesion complexity.

Figure 7. Strain-specific differences in the frequency of centrosome aberrations. The fraction of ALDH+ cells were plotted relative to control for cells exposed to low and high doses of Cs (LD = 0.12 Gy; HD = 0.8 Gy) (A). Colored lines identify the symbol representing the same strain at low and high dose for two strains of comparable ages (16 y-green and 19 y-pink). Phospho-profiles of key proteins in the RTK, TNFα and TGFβ signaling pathway were compared between the two strains after Cs and Ti exposure (B). Fold change of high relative to low dose phospho-protein expression was plotted for the strains. Data represent 2 independent experiments. The fraction of cells with aberrant centrosomes (>3P) (C) and ALDH+ cells (D) was plotted relative to control for cells exposed to low and high doses of Cs and Ti ion. Roughly equitoxic doses of Cs and Ti were used for exposures (LD-CS = 0.12 Gy; HD-CS = 0.8 Gy; LD-Ti = 0.05 Gy; HD-Ti = 0.5 Gy). Trend lines show that the strains from the 66 y old and 68 y old respond uniquely to low and high doses. Phospho-profiles of key proteins in the RTK, TNFα and TGFβ signaling pathway were compared between the two strains post-exposure to Cs and Ti (E). Fold change comparing the high to low dose was plotted for both strains following each radiation exposure (magenta solid line= Ti ion 66 y old strain, magenta dotted line = Cs ion 66 y old, green solid line = Ti ion 68 y old, green dotted line = Cs 68 y old). Data represent two independent experiments for each radiation type.

Cell strains

We have chosen fifteen cell strains from the Human Mammary Epithelial Cell Bank at LBNL to test from a cohort of well-characterized pre-stasis finite-lifespan normal human mammary epithelial cells (HMEC) derived from a wide range of aged individuals. These HMEC strains were produced in Dr. Martha Stampfer’s laboratory at LBNL, Berkeley CA, and originated from a large bank of uncultured organoids derived from reduction mammoplasty of normal breast tissue. Details on the derivation and culture of these HMECs can be found on http://hmec.lbl.gov. For some analyses we have classified strains into three age-defined cohorts based on the age of the individual from which the strain was derived. Four to five strains were chosen from various age ranges. The young cohort contained strains from women 16-27 years old (16, 16, 19, 27, 30 yrs), the middle-aged cohort contained strains from women 40 to 49 years old (40, 45, 46, 47, 49 yrs) and the old cohort contained strains from women 51-72 years old (51, 65, 66, 68, 72 yrs). These strains were previously used as part of a wider cohort in a study by Garbe et al who reported age dependent defects in the functional differentiation of multi potent progenitors [23].

Cell culture

Passage 2 (P2) cells were cultured and passaged to generate adequate P4 cell stocks for all experiments. These early passage (P4) mammary strains were grown to sub-confluence (~85%) in T75 flasks and exposed to radiation. Cell strains were maintained in serum-free M87A medium supplemented with 0.5 ng/ml cholera toxin (CT), 0.1 nM oxytoxin (Bachem) and 1X antibiotic-antimycotic (Invitrogen) [61]. Cultures were fed every two days. Cells were routinely tested for absence of mycoplasma contamination (Bionique Test Labs, NY) prior to use.

Irradiation

Strains plated in T25 flasks were exposed to two different radiation qualities at room temperature. Heavy ion exposures with Ti 300 MeV/n (nucleon) were carried out at the NASA Space Radiation Laboratory (NSRL) at Brookhaven National Laboratory, Upton, NY. To prevent particle fragmentation, cell culture flasks were placed perpendicular to the beam and oriented with the cell monolayer side of the flask facing the beam. Two to four independent experiments were carried out over the course of a two-year period at two different NSRL campaigns (NSRL14C and NSRL15B). At the NSRL14C campaign cells were exposed to just one dose (0.5 Gy), and centrosome aberrations were assessed at three different time points; day 3, day 9 and day 12 post-exposure. In the subsequent NSRL15B campaign, to assess low dose effects, we included a low dose (0.05 Gy) along with the high dose exposure and processed samples only at day 9. Dose rates ranging from (5.5-6.4) cGy/min for low dose exposures and (22.23-32.5) cGy/min for high dose exposures were used to maintain short exposure times (0.5-2.0 minutes). Cells were grown for 24 h following irradiation and either shipped back to LBNL as live cultures or frozen down at BNL for analysis at a later date. Upon arrival at LBNL, cells were allowed to acclimatize for a day and subsequently trypsinized, viability checked and reseeded at low concentrations in either 60 mm dishes or in 100 mm dishes for long-term culture to subsequently define centrosome analysis or to assess stem cell numbers. Cells were exposed to Cesium radiation at Lawrence Livermore National Laboratory at dose rates of ~25 cGy/min for low doses and ~100 cGy/min for high doses and processed similar to the Ti samples. Previous data, evaluating double strand break induction using ϒH2AX foci in a primary HMEC strain was used to calculate Cs doses that generate the same amount of initial damage as Ti (unpublished results). Based on these results, dose delivered with Cs exposures were ~1.5 fold higher than Ti for both the low and high doses to compare endpoints in response to approximately equitoxic doses. Two independent experiments were carried out for cesium exposures. Total population doublings for each culture was calculated beginning at passage 5 prior to the radiation exposure, using the formula PD= log₂ (recovered viable cell number/number of viable cells seeded)/0.301. Population doublings ranged from ~1-6, the proliferative potential of these strains is 10-15 PDs (personal communication J. Garbe).

Centrosome aberrations

For centrosome aberration analysis, cells were trypsinized 48 h after radiation exposure, and seeded at low density in 60mm dishes. In NSRL 14C campaign, Ti exposed cells were trypsinized on D2, and D8 and re-plated onto coverslips to analyze samples at D3, D9 and D12 after radiation exposure. Cesium samples were analyzed only at D9 post-exposure to compare with D9 Ti exposed samples. Coverslips were fixed in 100% methanol for 10 min at -20^oC, washed in PBS and subsequently stained for pericentrin. Coverslips were blocked in 1% BSA for 1 h at RT and centrosomes were detected by indirect immunofluorescence, using a primary rabbit antibody against pericentrin (1:1000) (Abcam, Cambridge MA). Following primary antibody incubation (1 h at RT), the cells were washed three times in blocking buffer, and subsequently incubated for 1 h with goat anti-rabbit Alexa Fluor 594 (Invitrogen, Waltham MA). Following secondary antibody incubation, cover slips were washed twice in PBS, counterstained with 0.1 µg/ml 4′, 6-diamidino-2-phenylindole (DAPI) in PBS, air-dried and mounted with Vectashield (Vector Laboratories, Burlingame, CA). Stained cells were imaged with a Zeiss Axiovert 200 M inverted fluorescence microscope equipped with a cooled CCD camera and Image-Pro®Plus software (MediaCybernetics, Rockville MD). Image acquisition was carried out within a week of staining.

Calculation of centrosome aberrations

Centrosome aberrations were scored blind and at representative areas throughout the slide. For each region we assessed the total number of cells based on DAPI stain, and the number of cells with aberrant centrosomes (>2). Images of cells with aberrant numbers of centrosomes were taken at 40 X and scored by two independent observers to evaluate the number of centrosomes in each aberrant cell. The type of centrosome aberration was characterized as either 3 centrosomes/cell (3P) or greater than 3 centrosomes/cell (>4P). Between 200-400 cells were examined for each sample.

Assessment of ploidy via flow cytometry

1x10⁵ cells from samples were fixed with 100% methanol, washed in PBS and subsequently stained with propidium iodide (PI). Ploidy was determined by flow cytometry based on DNA content and PI staining. DNA content of 10,000 cells was assessed on the FL2 channel using a BD FACS Calibur. Data was analyzed using Flowjo software Ver 8.6 (Tree Star Inc. OR).

Assessment of stem cell numbers

For assessing the proportion of stem cells we have used ALDEFLUOR reagent (STEMCELL Technologies), which has been shown to be specific for stem/progenitor cells in mammary epithelial cell populations. A previous study [62] using a derivative of one of our strains (184A1), revealed that pre-sorting cells using the ALDEFLUOR kit improves mammosphere generation by ~3 fold. Radiated and sham exposed cell strains were grown in 100 mm dishes and trypsinized 9 days following radiation exposure. Single cells in suspension were filtered using a 35 μ nylon mesh, counted using a hemocytometer and set up at the suggested concentrations for ALDEFLUOR labeling as per established protocols (STEMCELL technologies Inc., Vancouver). DAPI was included in the staining protocol for exclusion of dead cells that stain positive for DAPI (DAPI+). Cells expressing high levels of aldehyde dehydrogenase activity (ALDH+) were identified by their bright green fluorescence and enumerated from the total mammary cell population on a BD FACS Vantage SE flow sorter based on the FL1 channel. Gates for the ALDH+ population were defined using the supplied negative control that includes DEAB, an inhibitor of ALDH activity. Data are presented as proportion of stem cells in the radiated samples relative to sham-irradiated controls.

pATF2 immunostaining

For analysis of persistent pATF2 foci, coverslips processed on D12 following Ti exposure were co-stained with a mouse monoclonal antibody specific for pericentrin (Abcam, Cambridge MA) and a rabbit polyclonal antibody against pATF2 (Ser^490/498) (1:1000) (Rockland Inc, Gilbertsville PA). Following primary antibody incubation (1 h at RT), the cells were washed three times in blocking buffer, and subsequently incubated for 1 h with goat anti-mouse Alexa Fluor 594 (Invitrogen, Waltham MA) and goat anti-rabbit Alexa Fluor 488 (Invitrogen, Waltham MA). Following secondary antibody incubation, cells were washed twice in PBS, counterstained with 0.1 µg/ml 4′, 6-diamidino-2-phenylindole (DAPI) in PBS, air-dried and mounted with Vectashield (Vector Laboratories, Burlingame). Image acquisition was carried out within a week of staining. Data from two independent experiments were analyzed. Cells with >3 pATF2 foci were stratified as aberrant cells exhibiting persistent pATF2 signal. 200 cells were scored for each sample.

Multiplex analysis for phospho-protein expression

Two sets of samples were exposed to Ti (NSRL15B, BNL) and Cs (LLNL13A, 14C, 15B). Both irradiated and sham samples were trypsinized and grown in 100 mm dishes and processed 9 days post-plating. Six plates were processed in each batch. After a wash with cold PBS, 300 ul of ice-cold 1X Milliplex lysis buffer with protease inhibitors (Sigma) and phosphatase inhibitors (1 mM sodium orthovanadate, Sigma) was added to each plate on ice. Cells were scraped off the plate, and cell suspensions were transferred to a centrifuge tube and rocked for 15 mins at 4^oC. Particulates were removed by high-speed centrifugation 12000 rpm for 10 mins at 4C, and protein concentration was determined by the BCA assay on the nano-drop. Samples were aliquoted and stored at -70^oC. The MILLIPLEX® MAP 9-plex Multi-Pathway magnetic bead signaling kit was used to detect the phosphorylation profiles of key proteins in two different stress activated signaling pathways, namely the receptor tyrosine kinase (RTK) signaling pathway and TNF alpha signaling pathway. This kit was used to detect changes in phosphorylated ERK/MAP kinase ½ (Thr185/Tyr187), AKT (Ser473), STAT3 (Ser727), JNK (Thr183/Tyr185), p70 S6 kinase (Thr412), NF-κB (Ser536), STAT5A/B (Tyr694/699), CREB (Ser133), and p38 (Thr180/Tyr182) in cell lysates using the Lumina® system. The MILLIPLEX MAP Human TGFβ Signaling Magnetic Bead Panel 6-plex, was used to detect changes in phosphorylated Smad2 (Ser465/Ser467), Smad3 (Ser423/Ser425), as well as total protein levels of TGFRβII and Smad4 in cell lysates using the Luminex® system. 25 ugs of sample was used in each well and the immunoassay was carried out according to the published protocol for this kit (48-680 MAG, Millipore) and analyzed using the Luminex® system.

Statistical analysis

Linear regression was used to determine trends in surrogate endpoints for cancer relative to age. One-way ANOVA were used for multiple comparisons. Statistical significance of data was determined with p values <0.05. Pair wise comparison of differences in the percentage of cells with centrosome aberrations among dosed and unexposed controls were carried out using the Wilcoxon test. Differences in centrosome aberration frequency and stem cells amongst age-grouped cohorts were analyzed by the t test for independent samples. Pearson’s correlation was used to examine the correlation between frequency of centrosome aberrations, ALDH+, pATF2 persistent signal and the relationship of these parameters to the level of various phospho-proteins in the stress activated signaling pathways. Analysis of variance (ANOVA) was used to estimate the 95% confidence interval of the mean. All statistical analysis was performed using GraphPad Prism 6.0.

Research was conducted under Lawrence Berkeley National Laboratory Human Subjects Committee IRB protocols 305H002 and 108H004.