|Home | About | Journals | Submit | Contact Us | Français|
Alterations in DNA copy number contribute to the development and progression of cancers and are common in epithelial tumors. We have used array Comparative Genomic Hybridization (aCGH) to visualize DNA copy number alterations across the genomes of lung tumors in the KrasLA2 model of lung cancer. Copy number gain involving the Kras locus, as focal amplification or whole chromosome gain, is the most common alteration in these tumors, and with a prevalence that increased significantly with increasing tumor size. Furthermore, Kras amplification was the only major genomic event among the smallest lung tumors, suggesting that this alteration occurs early during the development of mutant Kras driven lung cancers. Recurring gains and deletions of other chromosomes occur progressively more frequently among larger tumors. These results are in contrast to a previous aCGH analysis of lung tumors from KrasLA2 mice on a mixed genetic background, in which relatively few DNA copy number alterations were observed regardless of tumor size. Our model features the KrasLA2 allele on the inbred FVB/N mouse strain, and in this genetic background there is a highly statistically significant increase in level of genomic instability with increasing tumor size. These data suggest that recurring DNA copy alterations are important for tumor progression in the KrasLA2 model of lung cancer, and that the requirement for these alterations may be dependent on the genetic background of the mouse strain.
The development of cancer is largely driven by progressive accumulation of genetic and epigenetic alterations, resulting in concerted deregulation of biological functions that together contribute to the cancer phenotypes. Alterations in DNA copy number are common in cancers, and especially in epithelial tumors. The gain or loss of DNA can occur focally and more often involves whole chromosomes or chromosomal arms (1, 2). DNA amplification commonly occurs in regions containing proto-oncogenes whereas deletion frequently targets regions with tumor suppressor genes. In consequence, genomic regions with recurring focal alterations have emerged as genomic landmarks for discovery of genes with important functions in tumorigenesis and that are commonly altered in tumors (1). The relevance of genomic imbalance is further underscored by the association between aneuploidy with disease aggressiveness in cancers of many tissues. Because molecular analyses have mostly focused on tumor tissues at relatively advanced stages, it has been difficult to formally study the evolution of genomic alterations during the natural course of tumor development.
Lung cancer is the second most common cancer in both men and women, and is responsible for the highest number of cancer-related deaths in the U.S. Adenocarcinoma is the most common subtype of lung cancer, accounting for 30–40% of the cases. Activating mutations in KRAS occur in ~30% of human lung adenocarcinomas and are associated with poor clinical outcome (3). Studies in the mouse have demonstrated that oncogenic activation of Kras alone is sufficient for the initiation of lung tumors (4, 5). In the KrasLA2 model, spontaneous somatic activation of the latent KrasLA2 allele initiates lung tumor development with complete penetrance (5). The majority of these tumors are benign lung adenomas, some of which progress to invasive adenocarcinomas in a process that mirrors lung tumor development and progression in humans. Adult KrasLA2 mice develop multiple lung adenomas with a wide range in size that likely reflects the variable timing of the Kras activation and tumor initiation.
Previous analysis of genomic changes in tumors from the KrasLA2 mice were carried out on tumors from animals of mixed C57BL6 and 129/SvJae backgrounds (6). These studies demonstrated some whole chromosome changes in a relatively small proportion of the tumors (27%) and noted the absence of any focal amplicons or deletions. We have transferred the KrasLA2 allele onto the FVB/N background, and noted that on average, tumor development was delayed relative to the previous data on the mixed background. Using aCGH we show that on the FVB/N background, there is a significant progressive increase in the degree of genomic instability associated with increasing tumor size. Some of the smallest tumors show focal amplification of Kras as the only detectable change, suggesting that this may be the earliest genetic event during lung tumor development in the KrasLA2 model.
KrasLA2 mice on a C57BL6/129/SvJae background were backcrossed with FVB/N mice to create an KrasLA2 inbred line on the FVB/N strain background. The mice used in this study are in between generation N11 and N15. Animals were sacrificed at 6 months of age and their lung tissues were collected for molecular studies.
We isolated 83 lung tumors of different sizes from KrasLA2 mice. To avoid normal cell contamination, only the tumor mass inside the tumor capsule was used for DNA isolation. Genomic DNA was extracted using the Dneasy Tissue kit (Qiagen) and further purified by phenol/choroform extraction. DNA concentrations were measured using a Biorad fluorometer.
aCGH arrays were prepared and printed as reported previously (7, 8). Each array consisted 1056 BAC/P1 clones printed in quadruplicate (2 spatially separated duplicates) onto 3D-Link activated slides (Motorola Life Sciences) using a custom DNA arraying device developed at the UCSF Cancer Center. Two arrays, each 12 mm × 12 mm in size, were printed per slide. The identities and genomic coordinates of the BAC/P1 clones were based on the February 2002 freeze of the assembled mouse genome through the UCSC Genome Browser at http://genome.ucsc.edu. A detailed list of the clones including their associated markers and genomic coordinates were previously published online (7). We labeled 1 μg of test (tumor) DNA and reference genomic DNA (normal lung) with CY3 and CY5 (Amersham), respectively. DNA labeling, hybridization, slide imaging and data analyses were performed essentially as previously described (8). The 16-bit TIFF images were collected using a custom-built CCD camera through CY3, CY5 and DAPI filters (9). Images were segmented and quantified using custom software (10). Data were normalized to the median log2 ratio of Cy3/Cy5. All clones except those mapping to chromosome six, known to be frequently altered in KrasLA2 lung tumors, were used for normalization. The mean and standard deviation (s.d.) of the normalized log2 ratios of the quadruplicate spots were calculated. Clones were declared missing if their ratio was based only on a single spot or their s.d. exceeded 0.2. Clones were called amplified or deleted if their log2 ratio exceeded +/− 0.3. Percentage of the genome altered was calculated by dividing each chromosome into 1,000,000 equally spaced bins and calling each bin amplified or deleted depending on the status of the most physically proximal probe. Association between tumor size category and percentage of altered loci was assessed with the Kruskal-Wallis rank sum test. Statistical analysis was performed with the R package (10). CGH data have been deposited in the NCBI GEO archive under accession GSE29230.
The KrasLA2 allele, originally on a C57BL6/129/SvJae mixed background, was backcrossed into the FVB/N background for more than 10 generations in order to minimize the effect of genetic heterogeneity on lung tumor development. Mice were sacrificed at 6 months of age, and a total of 83 surface lung tumors from more than 25 animals were obtained for molecular analyses. The size of the tumors was estimated based on their surface area measured at the time of resection. We categorized the tumors into four groups based on size (Table 1). Genomic DNA was isolated from the tumors and genome-wide DNA copy number alterations were quantified using aCGH. Histological analysis was performed on 27 of the tumors. The majority of these tumors (22 of 27) had a papillary histology (Fig. 1A), as has been previously described (5). Four tumors had a solid histology (Fig. 1B), but these were relatively small (Categories 1 and 2) and we could not ascertain whether this pattern of growth was uniform throughout the tumor. Interestingly, one tumor had a mixture of both papillary and solid components.
Somatic oncogenic activation of Kras initiates tumor formation in the KrasLA2 mouse model of lung cancer (5). We and others have demonstrated that oncogenic ras mutations are often accompanied by DNA copy number alterations that result in extra copies of the mutant allele in tumors, often through gain of the whole chromosome (11, 12). The mouse Kras gene is located on distal chromosome six, and gain of this region through focal amplification or trisomy of chromosome six was the most frequent alteration in the lung tumors analyzed. Focal amplification involving the Kras locus was observed in only two cases (Fig. 2A), but both were among tumors of the smallest size (Category 1; < 6 mm2). The level of focal amplification in these tumors was as high as fifteen-fold, and with very few, if any, copy number alteration in other parts of the genome. In three additional tumors in Category 1, more than 30% of the probes on chromosome six, including those overlapping with the Kras locus, showed a gain of a single copy that we interpreted to reflect a gain of the entire chromosome (Fig. 2B). We conclude that gain of DNA copy number involving the Kras locus, either through focal amplification or whole chromosome gain, is the earliest detectable somatic alteration in oncogenic Kras-driven lung tumors.
Trisomy of chromosome six is the most common DNA copy number alteration across the four size categories of lung tumors. Of the 83 lung tumors analyzed, 49 (59%) had amplification of more than 30% of chromosome six probes (Fig. 3A and Table 1). Importantly, the prevalence of chromosome six gain increased with increasing tumor size. The frequencies of trisomy of chromosome six ranged from 20% of Category 1 tumors to more than 80% of Category 4 tumors (P = 0.003, Chi-squared trend test; Table 1). When the threshold for calling trisomy was increased to 50% of probes, the number of tumors in each of the categories meeting this criterion decreased, but the trend between increasing tumor size and increasing incidence of trisomy six was maintained (P = 0.07, Chi-squared trend test). Probes overlapping the Kras locus were amplified in all tumors classified as having trisomy of chromosome six at both levels of stringency. These data suggest that gain of chromosome six, likely as a mechanism to up-regulate the level of mutant Kras, is an important event in the development and progression of lung tumors driven by oncogenic Kras.
While the genomes of the smallest tumors (Category 1) were relatively stable apart from changes on chromosome six (Fig. 2, ,3B),3B), frequent gains and losses of DNA copy number were evident among larger tumors (Fig. 3C, D, E). Considering all tumors together, we observed gains on chromosomes 1, 2, 8, 15, 17 and 19 as well as focal deletions or whole chromosomal loss on chromosomes 4, 5, 9 and 11 (Fig. 3A). Because these events commonly involve whole chromosomes or relatively large chromosomal regions, it was not possible to determine with confidence the targeted gene(s). However, there was a clear correlation between tumor size and level of genomic alterations as indicated by the progressive increase in percentage of aCGH probes altered across Categories 1 to 4 tumors (P value = 8.3 ×10−7, Kruskal-Wallis rank sum test for association between group rank and amount of variation, Fig. 4A). This strong correlation was not driven exclusively by the alterations on chromosome six because the relationship was still statistically significant when chromosome six probes were excluded from the analysis (P value = 1 × 10−5, Kruskal-Wallis rank sum test, Fig. 4B). These data demonstrated the contribution of accumulating genomic alterations to the growth of otherwise histologically similar lung tumors.
We have used aCGH to generate a genome-wide view of DNA copy number alterations during the development and progression of mouse lung tumors in the KrasLA2 model. We dissected surface lung tumors from multiple KrasLA2 mice, and used the surface area of the tumors as an approximate measurement of overall size. Somatic activation of the KrasLA2 allele can occur throughout the lifetime of the mouse, and we reason that this variable timing in tumor initiation accounted for the differences in size of lung tumors within the same animal. We therefore used tumor size as an indicator of tumor progression, with larger tumors being further along the tumorigenic process. However, we cannot rule out the possibility that acquisition of particular alteration(s) during the early stage of tumor formation could have blunted or rapidly accelerated the progression of some tumors.
The major DNA copy number alterations in lung tumors from KrasLA2 mice involved large chromosomal regions and whole chromosomes. Duplication of chromosome six is the most frequent event, but recurring gains of chromosomes 1, 2, 8, 12, 14, 15, 17, 18 and 19 as well as recurring loss of on chromosomes 4, 5, 9 and 11 were also observed. Recurring regions of focal deletion (e.g. proximal chromosome 4) were detected in a significant proportion of the largest tumors (Category 4; Fig. 3E) but even here the size of the region limited our ability to pinpoint the driver gene(s). A previous study also showed that lung tumors from the KrasLA2 mice have frequent trisomy of chromosome six, but the overall proportion of tumors with evidence of DNA copy number change was relatively small (6). Importantly, the authors reported a lack of association between tumor size and genomic alterations (6). In contrast, we observed a clear and significant increase in DNA copy number alterations with increasing tumor size (Fig 4A). This correlation remained highly statistically significant even after chromosome six was removed from the analysis (Fig. 4B). These data demonstrate the association of DNA copy number alterations with lung tumor progression in this model.
A likely explanation for the difference between our results and those of the prior study is that we have backcrossed the KrasLA2 allele into the FVB/N background for more than 10 generations whereas the allele was originally carried on a C57BL6/129/SvJae background (6). On a C57BL6/129/SvJae background, approximately 30% of animals carrying the KrasLA2 allele developed skin papillomas or carcinomas and thymic lymphoma (5) but the incidence of these tumors on a FVB/N background is dramatically reduced (data not shown). This result was surprising, as FVB/N mice are known to be highly susceptible to development and progression of skin tumors initiated by activating mutations in the Hras gene (13, 14). These observations suggest that although FVB/N mice are susceptible to Hras-driven skin tumors, they may be more resistant generally to activating mutations in Kras, including Kras mutant lung tumors. Consequently, more genomic or genetic alterations may be required to drive Kras mutant lung tumor formation in this background. The relationship between frequency of genomic alterations and genetic background has also been observed in other murine models of cancer (15). Data from analyses of mouse (16) and human (17) squamous cell carcinomas also support the notion of genetic background affecting the pattern of somatic alterations in tumors. The potential interactions between germline variants and somatic alterations could influence not only cancer susceptibility (18), but can also affect tumor progression and disease prognosis.
Activating mutations in the ras family of genes, particularly Kras, are capable of initiating and driving tumorigenesis in many tissues. In spite of its oncogenic potency, tumors containing ras mutations have often been observed to have DNA copy alterations at the ras locus that result in a genomic imbalance in favor of the mutant allele. Mouse skin tumors harboring an activating Hras mutation commonly have gross chromosomal alterations that result in an increase in copy number of mutant Hras or loss of the wild-type (WT) Hras allele (11, 12). Furthermore, the imbalance at the Hras locus contributes to the progression of squamous carcinomas to more invasive tumors (12). We identified trisomy of chromosome six as the most common genomic alteration in lung adenomas from the KrasLA2 mice, particularly in the largest tumors, suggesting that genomic imbalance on this chromosome plays an important role in tumor progression. Interestingly, loss of heterozygosity spanning the KRAS locus has been reported in human lung cancer that correlates with KRAS mutation and preferentially targets the WT KRAS allele (19). Genetic and in vitro studies have compellingly demonstrated that WT Kras can effectively attenuate the oncogenic effect of mutant Kras through mechanisms that remain to be elucidated (18, 20, 21). We have further shown that the balance in expression levels of mutant and WT Kras is a major determinant of susceptibility to lung cancer development in the mouse (18). Therefore, the imbalance in favor of mutant Kras in lung tumors, through gain of the mutant allele or loss of the WT allele, is compatible with the requirement of tumor cells to overcome the tumor suppressor effect of WT Kras.
The occurrence of focal amplification involving the Kras locus supports the notion that Kras is the major driver of the genomic imbalance on chromosome six. The level of focal amplification was as high as 15-fold. However, focal amplification and especially high level of focal amplification of distal chromosome six was rare and occurred only in the smallest tumors (Category 1). Instead, the majority of tumors including many of the small tumors (Categories 1 and 2) have gained only a single copy of chromosome six. These data suggest that while Kras may be the main driver for genomic imbalance on chromosome six, the accompanying dosage increase of other chromosome six gene(s) may also contribute to lung tumor progression.
Interestingly, both Braf and Raf1 (cRaf), well known downstream effectors of Ras signaling (22), are also located on chromosome 6 and must therefore undergo copy number gains in lung tumors. Alternatively, a subtle increase in copy number, rather than high levels of amplification, is more compatible with the growth and progression of the tumors. Cellular senescence is an important barrier against tumorigenesis, and high levels of oncogenic Ras (23) or Braf (24) can trigger this process. In addition, the cyclin-dependent kinase inhibitor p27/Kip1 is located near Kras, and high levels of p27/Kip1 as a result of co-amplification with Kras would potently inhibit tumor growth. Therefore tumor cells must strike a balance between overcoming the suppression effect of WT Kras and avoiding the triggering of intrinsic mechanisms that have evolved to protect cells against conditions of high oncogenic activity. Further analysis of these tumors on the FVB/N background by high resolution CGH, gene expression and tumor genome sequencing may reveal additional genetic targets that contribute to progressive genomic instability and tumor progression.
This work was supported by NCI grants CA111834-01 and CA84244 to AB. MDT acknowledges the support from the Nan Tucker McEvoy Research Fund in Thoracic Oncology. AB acknowledges support from the Bruce and Davina Isackson Foundation and from the Barbara Bass Bakar Chair in Cancer Genetics.
Disclosure of Potential Conflicts of Interest: No potential conflicts of interest.