Most human tumors develop and progress toward malignancy through
accumulation of multiple genetic alterations including activation of
oncogenes and inactivation of tumor suppressor genes [1-4]. The proteins
which are encoded by these genes interact with one another and construct
elaborated networks involved in cell growth regulation such as signal
transduction, cell cycle control and programmed cell death [1,4]. Over- or
under expression and abnormal structures of these proteins are involved in
oncogenesis. Patterns of genetic alterations are different not only among
types of tumors, but also among tumors of the same organ from different
patients. Thus, understanding of the genetic alterations accumulated in
tumors in each patient is essential to determine the molecular mechanisms
underlying tumorigenesis. To detect genetic changes including small
alterations such as point mutations, we developed a method of single-strand
conformation polymorphism (SSCP) analysis [5]. This method was applied to
analyses of DNAs from the same set of surgical specimens of human tumors
including non-small cell lung cancers, pancreatic cancers, hepatocellular
carcinomas and gliomas. Studies of the DNA from each type of human cancer
cell lines could also provide useful information about the multiple genetic
alterations accumulation of which resulted in tumorigenesis. This article
provides an overview of the mutated genes accumulated in tumors from a
variety of patients and also in a number of human cancer cell lines
detected both by ourselves and by other groups using a variety of DNA
analysis methods including SSCP.
Results
I. Altered regulation of cell growth by genetic changes
Numerous pathways for the control of cell growth have been demonstrated in
normal cells, where the products of oncogenes and tumor suppressor genes
are involved in several independent cascades or within the same cascade
[1-4].
p16 interacts with CDK4 and inhibits its activity, while CDK4 associates
with cyclin D1 to inactivate pRB resulting in cell cycle progression. Thus
p16, CDK4/cyclin D1 and pRB share, at least in part, the same cascade (the
RB cascade). In fact, inactivation of the tumor suppressor p16 or RB genes,
or amplification of the proto-oncogenes CDK4 or cyclin D1 were demonstrated
in various types of human cancer [4,6].
p53, p21 and MDM2 protein are also known to form a network, where p53 can
induce p21 production for the induction of cell cycle arrest, while MDM2
can bind and inactivate p53. Inactivation of the tumor suppressor gene p53
as well as amplification of the mdm2 gene have been reported in various
human cancers [7-9]. The gene encoding p21 is also a possible tumor
suppressor gene [10].
RAF protein has been identified as the downstream effector of the signal
transduction pathways for RAS protein the genes encoding which, i.e. the
K-, H- or N-ras genes, are often activated by point mitations in human
cancers [11]. Thus, the raf gene is a possible proto-oncogene because
activating rearrangement or mutation could enhance the RAS cascade [12,13].
AKT2 was shown to be activated by the signal from the platelet-derived
growth factor receptor (PDGFR)[14], while oncogenic activation occurs by
amplification of the gene in a subset of tumors [15,16].
Thus, aberrations of genes and their accumulation in each type of cancer
cell should be examined from the view-point of the altered regulation
pathways of cell growth.
II. Multiple genetic alterations in human tumors.
We have been analyzing alterations in various genes in the same set of
surgical tumor specimens and in human cancer cell lines from several
organs, including the lung, pancreas, liver and bain. The results of
analyses of surgical specimens as well as those from studies on cell lines
by both ourselves and other groups are summarized in Tables 1 to 4.
(1) Non-small cell lung cancers
We analyzed mutation of the ras, p53, RB, p21, raf and c-erbB2 genes in
human non-small cell lung cancers (NSCLCs) by PCR-SSCP. As summarized in
Table 1A, mutations in the ras and p53 genes were observed in 22 (19%) and
60 (52%) of 115 NSCLC tumors, respectively, while only 9 tumors carried
both aberrations [26,27]. These results suggested that mutations in the p53
and ras genes are frequent but independent events in NSCLC tumors.
Moreover, alterations in these two genes were not restricted to any
particular clinical stage or to histological differentiation [26-29]. On
the other hand, DNA analyses demonstrated that most cell lines established
from NSCLCs carried mutations in both K-ras and p53 genes, suggesting that
cells with both genetic alterations could be selected preferentially to
grow in vivo (Table 1B).
In connection with the RAS cascade, aberration of the raf gene was
analyzed in the same set of 115 NSCLC tumors. However, no mutation was
observed in the raf gene, suggesting that the oncogenic activation of RAS
protein might not be replaced by RAF activation in NSCLC tumors [30].
With regard to the p53 cascade in NSCLCs, absence of mutations in the p21
gene in 67 tumors suggested that dysfunction of p21 was not equivalent to
that of p53. Multiple disorders caused by p53 inactivation including
disruption of apoptotic pathways could be important in the development of
this type of tumor [31].
Inactivation of the RB gene was detected in only 2 of 43 NSCLC tumors
[32,33]. A subset of NSCLC cell lines also carried a mutated RB gene [34].
Inactivation of the p16 gene was frequently observed in NSCLC cell lines
indicating that disruption of the p16 - cyclin D1/CDK4 - pRB cascade is
involved in the majority of these tumors [32-35]. It should be noted that
alterations of the p16 and RB genes are mutually exclusive 8 (Table 1B).
Six cell lines carrying mutated p16 genes expressed normal RB proteins. On
the other hand, in the cell line NCI-H596 with the wild-type p16 gene,
homozygous deletion of the RB gene was observed. In contrast, no
amplification of the cyclin D1 or CDK4 genes was detected [36].
No possible activating mutations in the transmembrane domain of c-erbB-2
protein were detected in NSCLCs tumors [37]. While most of the NSCLC cells
showed multiple genetic alterations, a large cell lung cancer cell line,
Lu99, did not carry mutations in any of the genes analyzed [34,38].
(2) Pancreatic cancers
Mutation of the K-ras gene, inactivation of the p53, p16, DPC4, RB, BRCA2
and p15 genes and amplification of the AKT2 and c-myc genes are known to be
involved in human pancreatic cancers [39]. Rozenblum et al. reported
comprehensive analyses of first-passage xenografts from 42 primary
pancreatic cancers and demonstrated the accumulation of multiple genetic
alterations in these samples [40]. We also detected frequent mutations at
codon 12 of the K-ras and p53 genes, relatively rare inactivation of the
APC gene and no mutation of the p21 gene by SSCP analysis (Table 2A)
[31,41,42,43]. The incidence of K-ras mutation (14/27, 52%) in our SSCP
analysis of primary pancreatic cancers was lower than those in other
reports [44,45]. We confirmed the lack of mutations by sequence analysis of
all the 12 tumors which showed no mobility shift. Contamination by normal
cells in tumor specimens did not seem to be a severe problem because some
of the tumors without K-ras mutation showed the mobility shift on the SSCP
analysis of the p53 gene [Table 2A, K. Yashima, unpublished results]. These
findings in primary pancreatic cancers as well as the presence of
pancreatic cancer cell lines without K-ras mutation suggested that a
significant proportion of pancreatic cancers do not require the activated
K-ras gene.
We also demonstrated amplification of the AKT2 gene in a subset of
pancreatic tumors and cell lines [16]. Since the AKT2 gene encodes a
cytoplasmic serine-threonine kinase which is activated by PDGF and
overexpression of PDGF has been reported in pancreatic cancers, aberrations
in the cascade in which AKT2 is involved may play an important role in
pancreatic carcinogenesis [14].
The combined results of studies on genetic alterations in pancreatic
cancer cell lines indicated that multiple genetic alterations were
accumulated in the majority of the cells (Table 2B). As observed in NSCLC
cell lines, inactivation of the p16 gene is frequently observed in
pancreatic cancer cells and could be complementary to the aberration of the
RB gene, as 5 cell lines carrying the mutated p16 gene did not show any
abnormalities in the RB gene (Table 2B). The only exception was a Capan-2
cell line carrying aberrations in both the p16 and RB genes. In this cell
line, inactivation of p16 and pRB may act synergistically in malignant cell
growth.
DPC4 and p15 proteins are involved in a growth inhibitory cascade from the
signals of transforming growth factor beta. The genes for these proteins
were also inactivated in a subset of pancreatic cancer cell lines [46].
However, homozygous deletion of the p15 gene was always accompanied by that
of the adjacent p16 gene and no mutation specific to the p15 gene was
detected. These observations indicate that inactivation of the p15 gene
might be an accessory event in pancreato-carcinogenesis.
It is noteworthy that no alterations were observed in the genes analyzed
in a small subset of pancreatic cancers, even in advanced stages including
two cell lines, SW979 and SW850, or the xenograft QGP-1.
(3) Hepatocellular carcinomas
Chronic hepatitis and liver cirrhosis resulting from chronic infection by
hepatitis C virus (HCV) and hepatitis B virus (HBV) are the major
etiological factors of hepatocellular carcinoma (HCC) in Japan [47]. We
have demonstrated that p53 and RB gene inactivation are frequent events in
HCCs [28,48,49]. As shown in Table 3A, RB mutation was observed in
well-differentiated adenocarcinomas, while that of the p53 gene was not,
suggesting involvement of the RB mutation in an earlier stage than that of
the p53 mutation in multi-stage carcinogenesis of HCCs. Functional
inactivation of p16 protein and amplification of the cyclin D1 gene
reported by other groups indicated that disruption of the p16 - cyclin
D1/CDK4 - pRB cascade was also important in HCCs [50-52]. However, neither
amplification of the mdm2 gene nor mutation of the p21 gene was detected in
the same set of HCCs, suggesting that disorder of the pathways specifically
induced by p53 such as those of apoptosis might play a pivotal role in HCC
formation [Y. Ohyama, Y. Murakami, unpublished results].
Overview of the results of genetic analyses in HCC cell lines confirmed
the finding that inactivation of p53, RB and p16 genes were the major
genetic alterations. Several reports suggested that HBX protein could
associate with and inactivate p53 at least at a very early stage of HCC
formation [53]. Observations in two cell lines, HuH1 and HuH2, supported
this suggestion (Table 3B). However, some HCC cell lines show both
inactivation of the p53 gene and integration of the HBX gene [54].
Furthermore, in some cell lines, neither of these aberrations were
observed. Mutation of the ras gene was quite rare in HCCs, while it was
observed in the hepatoblastoma cell line HepG2 [55].
(4) Gliomas
We have demonstrated p53 mutation, homozygous deletion of p16, CDK4
amplification as well as LOH at polymorphic loci on chromosomes 11q and 10p
in advanced type human gliomas [56-60]. In glioma cell lines, inactivation
of the p16 gene is observed frequently, while aberrations of other genes
involved in the RB cascade such as mutation of the RB gene or amplification
of the cyclin D1 or CDK4 gene were not detected in cells with p16
alterations [61]. Mutations of the PTEN and p53 genes were also observed at
high frequency in glioma cells, while mutation of the ras gene has not been
reported in gliomas [61-63].
Back to the top.