ARF inhibits the growth and malignant progression of non-small-cell lung carcinoma

Abstract

Non-small-cell lung carcinoma (NSCLC) is among the deadliest of human cancers. The CDKN2A locus, which houses the INK4a and ARF tumor suppressor genes, is frequently altered in NSCLC. However, the specific role of ARF in pulmonary tumorigenesis remains unclear. KRAS and other oncogenes induce the expression of ARF, thus stabilizing p53 activity and arresting cell proliferation. To address the role of ARF in Kras-driven NSCLC, we compared the susceptibility of NIH/Ola strain wild-type and Arf-knockout mice to urethane-induced lung carcinogenesis. Lung tumor size, malignancy and associated morbidity were significantly increased in Arf(-/-) compared with Arf(+/+) animals at 25 weeks after induction. Pulmonary tumors from Arf-knockout mice exhibited increased cell proliferation and DNA damage compared with wild-type mice. A subgroup of tumors in Arf(-/-) animals presented as dedifferentiated and metastatic, with many characteristics of pulmonary sarcomatoid carcinoma, a neoplasm previously undocumented in mouse models. Our finding of a role for ARF in NSCLC is consistent with the observation that benign adenomas from Arf(+/+) mice robustly expressed ARF, while ARF expression was markedly reduced in malignant adenocarcinomas. ARF expression also frequently colocalized with the expression of p21(CIP1), a transcriptional target of p53, arguing that ARF induces the p53 checkpoint to arrest cell proliferation in vivo. Taken together, these findings demonstrate that induction of ARF is an early response in lung tumorigenesis that mounts a strong barrier against tumor growth and malignant progression.

Figure 1. Arf loss increases lung tumor growth, malignancy, and morbidity in urethane-exposed mice

[A] Arf loss led to accelerated lung tumor-associated morbidity after urethane exposure (logrank test for trend, P < 0.0001). [B] At necropsy, Arf^−/− mice (bottom) bore massive lung tumors compared to Arf^+/+ mice (top). H&E images shown on right; scale bar 1 mm. [C] At 17 weeks post-exposure, an equivalent number of tumors were visible on the surface of Arf^+/+ (6.300 ± 0.7461) and Arf^−/− (5.111 ± 0.7718) lungs (P = 0.2840; n = 10 Arf^+/+ and 9 Arf^−/−). [D] Mean tumor diameter. Tumors were measured with calipers; largest diameter was used for analysis. Mean values differed significantly between genotypes (**** P < 0.0001; n = 351 Arf^+/+ (1.205 ± 0.02899 mm), 196 Arf^+/− (1.306 ± 0.07118 mm), and 211 Arf^−/− tumors (2.815 ± 0.1883 mm)). [E] Mass of lungs from non-tumor bearing control and urethane-exposed male mice. Arf^−/− mice (1.103 ± 0.106 g) carried a significantly greater lung tumor burden than Arf^+/+ (0.514 ± 0.209 g) (*** P = 0.0007; n ≥ 4 animals each condition). [F] Adenomas occurred primarily in Arf^+/+ mice (top panel). Tumors had discrete borders and were composed of a uniform cell population with small, round nuclei and moderate amounts of eosinophilic cytoplasm. Mitotic figures were rare. Adenocarcinomas predominated in Arf^−/− mice (bottom panel). Invasion (arrow) and mitotic figures (arrowheads) were frequent. [G] Incidence of lung adenomas (Ad) and adenocarcinomas (AC) across three genotypes (**** P < 0.0001; n = 44 Arf^+/+, 40 Arf^+/− and 39 Arf^−/− tumors examined). [H] Intravasation of Arf^−/− adenocarcinoma. Tumor cells have broken through the basement membrane (arrow) to invade the neighboring blood vessel (BV). An endothelialized thrombus, in pink, has formed at the site. Scale bars 100 μm.

Figure 2. Poorly differentiated, sarcomatoid lung tumors develop in Arf-deficient animals

[A] Pulmonary sarcomatoid carcinoma-like lesions formed in the lung. [B] Magnification of boxed region in [A]. Arrowheads point to well-differentiated, epithelial regions. [C] Renal metastasis (M) of mixed epithelial and sarcomatoid components. Arrows indicate normal kidney glomeruli. [D] Lung adenocarcinomas (top) and sarcomatoid metastases (bottom) from the same mouse exhibited immunoreactivity for pro-surfactant protein C (pro-SPC), a marker of lung epithelial cells. Bottom panel displays invasion of the pleura and chest wall. Arrow indicates rib. The metastatic tumor maintained regions of epithelial differentiation (boxed region, magnified in middle panel) that expressed pro-SPC (arrowheads). [E] Metastasis of sarcomatoid carcinoma to peritoneal cavity. Whereas both the sarcomatoid (box S) and epithelioid (box Ep) compartments were positive for cytokeratin 8, vimentin expression was restricted to the sarcomatoid regions. Scale bars 100 μm, except panel [D, bottom left] (1 mm).

Figure 3. RAS pathway signaling, cell proliferation, and DNA damage

[A] Arf^−/− adenocarcinomas (AC) displayed a higher proliferation index than tumors from Arf^+/+ littermates, as shown by phosphorylated histone H3 staining (**** P < 0.0001; n = 119 adenoma and 220 AC fields counted from Arf^+/+; n = 22 adenoma and 829 AC fields from Arf^−/− mice). [B] Phosphorylated histone H2A.X staining illustrates a marked increase in DNA damage in tumors from Arf-deficient mice (**** P < 0.0001; *** P = 0.0001; n = 85 adenoma and 307 AC fields counted from Arf^+/+; n = 31 adenoma and 788 AC fields from Arf^−/− mice). In addition to H2A.X⁺ cells occurring more frequently per field, damage foci were also larger and more numerous per cell in Arf^−/− mice compared to Arf^+/+. Note that the paucity of adenomas in Arf^−/− animals prevented inclusion of additional fields in the statistical analysis of phospho-H3 and phospho-H2A.X staining. [C] Nuclear cyclin D1 and phospho-ERK1/2 expression were strongly upregulated in adenomas (top) and adenocarcinomas (bottom) from individual Arf-deficient mice (lanes 4–6 and 10–12) compared to Arf^+/+ mice (lanes 1–3 and 7–9). β-actin is provided as loading control. β-tubulin and lamin B1 are included as markers of cytoplasmic and nuclear compartments, respectively. [D] Relative intensities of nuclear versus cytoplasmic cyclin D1 bands from [C] compared between genotypes at each tumor stage. All samples were normalized to β-actin (* P = 0.0114 adenomas, * P = 0.0308 ACs). [E] Expression of Ccnd1 mRNA is equivalent in adenocarcinomas from Arf^+/+ and Arf^−/− animals. Gapdh was used as endogenous control. Data is plotted as fold change compared to mean Arf^+/+ value (n = 6 tumors per genotype; P = 0.6600).

Figure 4. The ARF-p53 signaling pathway

[A] Adenomas (Ad) from Arf^+/+ mice robustly expressed ARF, but adenocarcinomas (AC) frequently lost ARF expression. Inset shows nucleolar staining of ARF in adenomas. [B] Quantification of ARF positive cells (**** P < 0.0001; n = 46 Ad and 145 AC fields examined). [C] ARF and p21 proteins colocalize by IHC in adenomas (top) and regions of low-grade adenocarcinomas (bottom) in wild-type mice. All scale bars 100 μm. [D] Western blot analysis of p53 reveals expression of p53 protein in adenomas and adenocarcinomas isolated from both genotypes. Irradiated spleen is positive control. β-actin is provided as loading control. [E] PCR amplification of genomic DNA shows that the Trp53 locus has not been deleted in adenocarcinomas from either genotype. GAPDH is provided as loading control. A dilution series is included to demonstrate that the PCR was performed in the exponential range. NL = normal lung.