All mice expressed a conditional Lox-Stop-Lox Kras^G12D/+ (LSL-Kras) allele (Jackson et al., 2001). Expression of oncogenic Kras^G12D is controlled by a Cre-regulatable transcriptional Stop element. Because this conditional oncogenic Kras allele was targeted into the endogenous locus (Tuveson et al., 2004), endogenous levels of oncogenic Kras^G12D protein are expressed following removal of the stop element upon expression of Cre-recombinase. LSL-Kras mice for this study were bred with mice with conditional alleles for the tumor suppressors p53 (Jonkers et al., 2001) or Ink4a/ARF (Aguirre et al., 2003). In these conditional alleles, LoxP sites flank important coding regions in these tumor suppressor genes and are termed floxed alleles: p53^FL or Ink4a/ARF^FL. These conditional alleles are targeted to their respective endogenous loci so that prior to deletion, expression of the respective WT tumor suppressor gene is normal. However, in the presence of Cre-recombinase the LoxP sites recombine, thereby creating a deletion with loss of tumor suppressor gene function. Mice were bred to contain one allele of the conditional oncogenic Kras gene while simultaneously containing tumor suppressor genes p53 (LSL-Kras; p53^FL/FL) or Ink4a/ARF (LSL-Kras; Ink4a/ARF^FL/FL) with both alleles flanked by loxP sites.

Adult mice expressing the conditional alleles described above were infected with adenovirus expressing Cre-recombinase (Adeno-Cre) intranasally as previously described (Kirsch et al., 2010a). Adeno-Cre preferentially infects the lung epithelium and Cre-recombinase acts on the conditional alleles, allowing the mice to develop multiple lung tumors. Tumors become visible on micro-CT, a non-invasive small animal imaging modality, approximately 60 days after Adeno-Cre infection.

In order to isolate lung tumors for gene expression and histologic studies, mouse lungs along with other tissues including small intestine, thymus, and spleen were collected at the time of sacrifice. For studies in which bromodeoxyuridine (BrdU) immunohistochemical staining was performed, mice were injected with 5 mg/mL of BrdU 2 h prior to sacrifice to allow for BrdU incorporation to take place. Tissue used for gene expression analysis was stored in RNAlater (Ambion #AM-7021). Tissue utilized for immunohistochemical analysis was placed in a histology cassette in formalin overnight prior to paraffin embedding. Slides were cut from paraffin embedded tissue to generate unstained slides for immunohistochemical staining.

RNA was isolated from tissue stored in RNAlater using a traditional organic solvent precipitation technique followed by RNA cleanup using the RNeasy MinElute Cleanup kit (Qiagen #74204). RNA 260/280 ratios were consistently better than 2.00 and 260/230 ratios were generally better than 1.50. After converting isolated RNA to cDNA using the iScript cDNA Synthesis Kit (Bio-Rad #170-8890), TaqMan probes directed against p53, p21, and PUMA were used to measure gene expression relative to the house-keeping gene GAPDH using a real-time thermal cycler (Bio-RAD, DNA Engine). Student’s t-test (two-tailed) was performed to compare relative gene expression among different treatment cohorts.

Immunohistochemical staining for BrdU (1:200, Becton Dickinson, Cat #3457580), phospho-histone H3 (1:200, Cell Signaling, Cat#9701), and cleaved-caspase 3 (1:500, R&D Systems, Cat#AF835) was performed on unstained slides cut from formalin-fixed paraffin embedded tissue blocks according to standard immunohistochemistry protocols. Hematoxylin counterstaining was used for all immunohistochemical studies. Analysis of histological specimens was performed by calculating the number of positively staining cells/400× powered field within regions of tumor for each slide. When possible, a two-tailed Student’s t-test was performed to compare tumors among different treatment cohorts.

Mice were placed in radiation treatment cradles with lead shielding above and below the thorax to minimize normal tissue injury. Single-dose radiation treatment was delivered using an X-RAD 320 biological irradiator (Precision X-Ray) at a dose output of 320 kV, 10 mA to deliver 11.6 Gy at a dose rate of 69 cGy/min. In the two-fraction arm of the study, radiation was delivered using the same setup and dose rate with 24 h of separation between fractions. To provide an estimate of equivalent dose, we calculated the biologically effective dose (BED) (Fowler, 2010) in two fractions for a single-dose of 11.6 Gy, we estimated that the α/β ratio for lung cancer in mice was 10 and used the following formula for BED or E/α (Fowler, 2006): BED = E/α = nd(1 + d/α/β), where E is the log cell kill from n fractions of d grays.

Using an α/β ratio of 10 Gy, the BED for a single 11.6 Gy fraction is 25 Gy₁₀. Therefore, the dose required for each of the two fractions to deliver 25 Gy₁₀ would be 7.3 Gy. Animals that did not receive any RT still underwent the stress of placement in the radiation treatment cradles for the same length of time as mice that underwent single-dose radiation treatment. Dose rate calibrations were performed by the Duke University Office of Radiation Safety.

Computed tomography techniques developed especially for small animal imaging (Micro-CT) allowed us to perform this study. Mice were imaged at 2-week intervals before and after radiation treatment. We used a custom built micro-CT system that has been previously described (Badea et al., 2008). Mice were anesthetized using isoflurane and then were subsequently immobilized in a cradle during CT image acquisition with isoflurane anesthesia via a nose cone delivery system. Projections were acquired using a single tube/detector over a circular orbit of 195° with a step angle of 0.65°. Reconstructions were performed using a commercially available CT reconstruction program (COBRA Exxim, v5.0), with a filtered back projection technique. A resolution of approximately 88 μ/pixel was achieved. The cumulative radiation dose from all three micro-CT scans was 0.3 Gy, which represents about 2% from the 14.6 Gy treatment dose. Using this approach, we have been able to achieve reasonably high-throughput imaging of up to approximately 60 mice in a single day.

Following CT reconstruction, images were analyzed using an advanced image analysis software suite (Visage Imaging, Amira v5.0). Tumor volumes were calculated by contouring individual tumor areas on each slice of the reconstructed CT scans. In most cases, multiple tumors were contoured for each mouse. The individual assigned to contouring tumor volumes was blinded to the experiment and therefore not aware of mouse genotype or treatment regimen at the time of contouring. Tumors were followed over time before and after radiation treatment and tumor growth rate was determined relative to initial tumor volume calculations. GraphPad Prism software (v5.02) was utilized to perform linear regression and fit growth curves based on tumor volumes from each cohort (LSL-Kras; p53^FL/FL or LSL-Kras; Ink4a/ARF^FL/FL) with 0, 1, and 2 radiation treatments. Student’s t-tests were used to analyze differences in relative growth at 14 days after radiation treatment within each mouse genotype.

Intranasal delivery of Adeno-Cre into LSL-Kras; p53^FL/FL or LSL-Kras; Ink4a/ARF^FL/FL mice led to the development of multiple, aggressive adenocarcinomas in the lungs bilaterally. Primary lung cancers in mice were generated and imaged with micro-CT as previously described (Kirsch et al., 2010a). The mice were imaged with a thoracic micro-CT at 8 weeks after infection (baseline scan), at 10 weeks (1 day before whole lung RT), and at 12 weeks (to assess treatment response) (Figure 1). Lung tumors from both genotypes were histologically similar to previously characterized mouse tumors that are known to resemble human lung adenocarcinomas (Calbo et al., 2005) (Figures 2A,B). We used real-time quantitative RT-PCR to confirm that primary tumors excised from lungs of LSL-Kras; Ink4a/ARF^FL/FL mice (Figure 2C – blue bar) express p53, while tumors from LSL-Kras; p53^FL/FL mice do not (P = 0.0001, Figure 2C – red bar). Residual p53 expression in the LSL-Kras; p53^FL/FL tumors is likely from non-epithelial stromal cells or possibly from epithelial cells with an unrecombined p53^FL allele (Jackson et al., 2005). Tumor growth rates from both genotypes in the absence of RT (Figure 2D) were similar and the volume doubling time of about 14 days (Figure 2E) was similar to previous reports (Kirsch et al., 2010a; Oliver et al., 2010).

The bar graphs depicted in Figures 3A,B show relative volume increase over two time points: time = 0 days (relative tumor volume at the time of RT, 10 weeks after Adeno-Cre infection), and time = 14 days (relative tumor volume 14 days after RT, 12 weeks after Adeno-Cre infection). Relative tumor volume was assessed by normalizing the tumor volume at 14 days after RT to the tumor volume at the time of RT. Lung cancers in LSL-Kras; Ink4a/ARF^FL/FL mice that received a single fraction of 11.6 Gy had decreased growth compared to unirradiated tumors [factor of 1.53 (SEM – 0.03), n = 26 tumors vs. factor of 2.19 (SEM – 0.03), n = 27 tumors, P = 0.003], but two fractions of 7.3 Gy led to an even better response [factor of 0.88 (SEM – 0.01), n = 14 tumors, P = 0.002, Figure 3A]. Although lung cancers in LSL-Kras; p53^FL/FL mice also responded well to two fractions of 7.3 Gy [factor of 0.96 (SEM – 0.04), n = 14 tumors], this was not statistically different than the response to a single fraction of 11.6 Gy (factor of 1.2, n = 28 tumors, P = 0.23, Figure 3B). Therefore, these results suggest that lung cancers respond differently to one vs. two fractions of RT depending on the genotype of the lung cancer.

To investigate how p53-mediated signaling influences the radiation response of lung tumors, we examined the mRNA expression of p53 targets, which regulate p53-mediated cell-cycle arrest and apoptosis, in lung tumors after a single fraction of 11.6 Gy whole lung RT. Primary tumors of each genotype were excised from the lungs 4 h after RT, and the mRNA levels of p53 transcriptional targets p21 and PUMA were measured by quantitative reverse transcription-polymerase chain reaction (qRT-PCR). In tumors with WT p53 (LSL-Kras; Ink4a/ARF^FL/FL) a nearly 20-fold increase in p21 mRNA levels was noted after RT (P = 0.003, Figure 4A – blue bars). Tumors from LSL-Kras; p53^FL/FL mice showed significantly lower levels of p21 activation after RT (P = 0.09, Figure 4A – red bars). Similarly, RT-induced PUMA mRNA expression was robust in tumors from LSL-Kras; Ink4a/ARF^FL/FL mice, but not in tumors from LSL-Kras; p53^FL/FL mice (P = 0.0001, Figure 4B – blue bars). Thus, these data indicate that p53-mediated DNA damage response is significantly more active in LSL-Kras; Ink4a/ARF^FL/FL tumors as compared with LSL-Kras; p53^FL/FL tumors.

Because the induction of cyclin-dependent kinase inhibitor p21 contributes to the G1 cell-cycle arrest after irradiation (Kastan and Bartek, 2004), we evaluated the progression of cell-cycle after RT in the tumors from LSL-Kras; Ink4a/ARF^FL/FL and LSL-Kras; p53^FL/FL mice by the labeling index of BrdU, which is a thymidine analog that is incorporated into replicating DNA. Immunostaining of the lungs with an antibody against BrdU (Figure 5) revealed that tumors from LSL-Kras; Ink4a/ARF^FL/FL mice had significantly decreased BrdU uptake 4 h after RT (P = 0.005, Figure 5A), as compared with unirradiated tumors, indicating the presence of an intact G1 cell-cycle checkpoint. Thus, lung cancers with an intact p53-dependent induction of p21 appear to undergo a radiation-induced G1 cell-cycle arrestin vivo. In contrast, lung tumors from LSL-Kras; p53^FL/FL mice incorporated high levels of BrdU after irradiation, indicating that these tumors lacked an intact G1 cell-cycle checkpoint (Figure 5A). Interestingly, immunohistochemical staining for phospho-histone H3, which is present in cells during mitosis, revealed a significant reduction in tumors from both genotypes after RT (Figure 5B). Finally, cleaved-caspase 3 staining was performed to evaluate the extent of radiation-induced apoptosis. At 4 h after RT, apoptosis was an uncommon event in both tumor types (Figure 5C).Taken together, these assays provide functional data to demonstrate that only in primary lung tumors from LSL-Kras; Ink4a/ARF^FL/FL mice, which retain p53, is the radiation-induced p53 response functionally intact.