molecular biology of plants cancer

7/27/2019 Molecular Biology of Plants Cancer

1/57

Review

The molecular biology of cancer

John S. Bertram *

Cancer Research Center of Hawaii, University of Hawaii at Manoa, 1236 Lauhala Street,

Honolulu, HI 96813, USA

Abstract

The process by which normal cells become progressively transformed to malignancy is now

known to require the sequential acquisition of mutations which arise as a consequence of

damage to the genome. This damage can be the result of endogenous processes such as errors

in replication of DNA, the intrinsic chemical instability of certain DNA bases or from attack

by free radicals generated during metabolism. DNA damage can also result from interactions

with exogenous agents such as ionizing radiation, UV radiation and chemical carcinogens.

Cells have evolved means to repair such damage, but for various reasons errors occur and

permanent changes in the genome, mutations, are introduced. Some inactivating mutations

occur in genes responsible for maintaining genomic integrity facilitating the acquisition ofadditional mutations. This review seeks rst to identify sources of mutational damage so as to

identify the basic causes of human cancer. Through an understanding of cause, prevention

may be possible. The evolution of the normal cell to a malignant one involves processes by

which genes involved in normal homeostatic mechanisms that control proliferation and cell

death suer mutational damage which results in the activation of genes stimulating prolifer-

ation or protection against cell death, the oncogenes, and the inactivation of genes which

would normally inhibit proliferation, the tumor suppressor genes. Finally, having overcome

normal controls on cell birth and cell death, an aspiring cancer cell faces two new challenges: it

must overcome replicative senescence and become immortal and it must obtain adequate

supplies of nutrients and oxygen to maintain this high rate of proliferation. This review ex-amines the process of the sequential acquisition of mutations from the prospective of Dar-

winian evolution. Here, the ttest cell is one that survives to form a new population of

genetically distinct cells, the tumor. This review does not attempt to be comprehensive but

identies key genes directly involved in carcinogenesis and demonstrates how mutations in

these genes allow cells to circumvent cellular controls. This detailed understanding of the

process of carcinogenesis at the molecular level has only been possible because of the advent of

modern molecular biology. This new discipline, by precisely identifying the molecular basis of

the dierences between normal and malignant cells, has created novel opportunities and

provided the means to specically target these modied genes. Whenever possible this review

Molecular Aspects of Medicine 21 (2001) 167223 www.elsevier.com/locate/mam

* Tel.: +1-808-586-2957; fax: +1-808-586-2970.

E-mail address: [email protected] (J.S. Bertram).

0098-2997/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved.

PII: S 0 0 9 8 - 2 9 9 7 ( 0 0 ) 0 0 0 0 7 - 8


2/57

highlights these opportunities and the attempts being made to generate novel, molecular based

therapies against cancer. Successful use of these new therapies will rely upon a detailed

knowledge of the genetic defects in individual tumors. The review concludes with a discussion

of how the use of high throughput molecular arrays will allow the molecular pathologist/

therapist to identify these defects and direct specic therapies to specic mutations. 2001

Elsevier Science Ltd. All rights reserved.

Keywords: Cancer; Carcinogenesis; Mutations; DNA damage; DNA repair; Oncogenes; Tumor

suppressor genes; Growth control; Angiogenesis; Apoptosis; Senescence

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

2. Carcinogenesis: the conversion of normal cells responsive to homeostatic

feedback mechanisms to cells capable of autonomous growth and invasion . . . . . 170

2.1. Mutations require proliferation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171

2.2. DNA is subject to chemical damage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173

2.2.1. Induction of spontaneous DNA damage . . . . . . . . . . . . . . . . . . . . . . . 173

2.3. Induction of DNA damage by exogenous agents . . . . . . . . . . . . . . . . . . . . . 174

2.3.1. Chemical carcinogens. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174

2.3.2. Physical carcinogens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177

2.3.3. Many cancer chemotherapeutic agents are carcinogenic . . . . . . . . . . . . 177

2.4. Most DNA damage is repairable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178

2.4.1. Defects in DNA repair are responsible for many familial cancersyndromes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178

2.5. Cell-cycle checkpoints restrict replication of damaged DNA . . . . . . . . . . . . . 179

3. Pathways to cancer: overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181

4. Cancer cells are independent of external growth signals . . . . . . . . . . . . . . . . . . . 182

4.1. Inappropriate synthesis of growth factors . . . . . . . . . . . . . . . . . . . . . . . . . . 183

4.2. Inappropriate expression of growth factor receptors. . . . . . . . . . . . . . . . . . . 183

4.2.1. Erb-B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

4.3. Activation of downstream signal transduction pathways. . . . . . . . . . . . . . . . 1844.3.1. c-abl. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184

4.3.2. ras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185

4.4. Inappropriate activation of nuclear transcription factors . . . . . . . . . . . . . . . 188

4.4.1. Inappropriate expression of c-myc, a transcription factor . . . . . . . . . . . 188

4.4.2. Mutation of a nuclear hormone receptor leads to blocked dierentiation 189

5. Cancer cells become refractory to growth inhibitory signals: the discovery

of tumor suppressor genes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190

5.1. The retinoblastoma gene RB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191

5.1.1. RB functions to restrict entry into S-phase . . . . . . . . . . . . . . . . . . . . . 192

5.1.2. RB gene therapy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1935.2. p53 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

5.2.1. p53 mutations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

168 J.S. Bertram / Molecular Aspects of Medicine 21 (2001) 167223


3/57

5.2.2. p53 monitors genomic integrity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

5.2.3. p53 is a transcription factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

5.2.4. Loss of p53 alters response to chemotherapeutic agents . . . . . . . . . . . . 197

5.2.5. p53 gene therapy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

5.2.6. Human papilloma virus (HPV) can inactivate both p53 and RB . . . . . . 198

5.3. Mutations in the APC gene link cell surface receptors with the nucleus . . . . . 198

5.3.1. The APC gene oers many targets for intervention. . . . . . . . . . . . . . . . 199

6. Cancer cells are decient in intracellular communication mediated by gap

junctions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201

7. Cancer cells evade apoptosis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202

7.1. Overexpression of bcl-2 protects lymphoma cells from apoptosis . . . . . . . . . . 202

7.2. Tumor cells evade apoptosis by modied FAS and FAS-L interactions . . . . . 203

7.3. The induction of apoptosis is an important target in cancer therapy . . . . . . . 204

8. Cancer cells must avoid senescence and achieve immortality: role of telomeres . . . 205

8.1. Many cancer cells reactivate telomerase . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

8.2. Telomerase oers an exciting novel target for cancer prevention and therapy . 206

9. Cancer cells require adequate supplies of nutrients and stimulate angiogenesis . . . 207

9.1. Inhibitors of angiogenesis exert potent anti-tumor aects . . . . . . . . . . . . . . . 208

9.2. Inhibitors of pro-angiogenic signals are eective anti-tumor agents . . . . . . . . 209

9.3. Conventional chemotherapy can be targeted to endothelial cells . . . . . . . . . . 210

10. Putting it all together: prospects for molecular medicine in the 21st century . . . . 21010.1. Use of genomic arrays in the molecular proling of cancers . . . . . . . . . . . . 211

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213

1. Introduction

The last two decades have seen enormous advances in our understanding of

cancer at the molecular level. This understanding has revealed large numbers of

exciting new targets for the development of eective therapies, some of which havealready entered clinical practice. These new targets identify both early and late events

in the carcinogenic process and thus oer opportunities for treatment and for pre-

vention surely the most exciting goal in conquering this dreaded disease. By al-

lowing the direct targeting of the genetic defects that are responsible for malignancy,

it is a realistic expectation that increasing numbers of tumor-specic drugs will soon

be available which will spare normal cells from the devastating eects of conven-

tional cytotoxic therapeutic agents. To be eective, conventional agents must be used

at dosages which are acutely life-threatening to the patient. Furthermore many

currently available drugs also induce genetic damage which can itself be carcino-

genic. New molecular therapies should allow the physician an unprecedented abilityto treat the cancer without harming the patient. In order to fully exploit these new

opportunities, it is becoming apparent that the wide diversity of genetic aberrations

J.S. Bertram / Molecular Aspects of Medicine 21 (2001) 167223 169


4/57

present in tumor cells will make it necessary to genotype individual tumors just as

they are currently phenotyped by standard pathological procedures. High

throughput screening tests which can simultaneously measure the expression pattern

and presence of specic mutations in thousands of individual genes will make thispossible. Their use is currently restricted to major research centers as research tools

rather than diagnostic instruments; however, as new therapeutic agents become

available whose use depends upon specic genetic information, it seems inevitable

that this technology will be a necessary requirement for most cancer diagnoses. It is

the purpose of this review to outline our current knowledge of cancer genetics and in

so doing draw attention to the enormous possibilities for future research in the

design of specic cancer therapeutic agents.

2. Carcinogenesis: the conversion of normal cells responsive to homeostatic feedbackmechanisms to cells capable of autonomous growth and invasion

The adult human is composed of approximately 1015 cells, many of which are

required to divide and dierentiate in order to repopulate organs and tissues which

require cell turnover. Obvious examples are cells in the basal layer of the skin

which divide, dierentiate and are nally sloughed, cells composing the epithelial

layer of the intestines which turnover and must be replaced approximately every

10 days, and cells in the bone marrow which divide and dierentiate to produce

white and red cells whose life-time varies from 24 h in the case of some leukocytes

to 112 days for mature red cells. Cells which have the capacity for division and

replenishment are called stem cells. It can be calculated that there are approxi-

mately 1012 divisions per day in these stem cell compartments. Even in organs

which normally exhibit low levels of cell division, the liver being the prime ex-

ample, massive proliferation can be initiated by events such as trauma or infec-

tion. Yet in spite of this enormous production of new cells, the human body

maintains a constant weight over many decades. Even obesity is not primarily the

result of increased cell multiplicity but of increased volume and thus mass of

adipocytes. This exquisite control over cell multiplicity is achieved by a network of

overlapping molecular mechanisms which govern cell proliferation on one handand cell death, termed apoptosis when the result of a programmed event, on the

other. Any factor which alters this balance between birth and death, just as it

would in an isolated species of individuals, has the potential if not corrected to

alter the total number of cells in a particular organ or tissue. After many cell

generations this increased cellular multiplicity would be clinically detectable as

neoplasia, literally new growth.

As will be described below, it is genes that alter the birth rate or the death rate of

individual cells that have now been rmly implicated as causative in the carcinogenic

process. Just as Darwinian evolution depends upon random mutations giving rise to

a selective advantage to individuals, it now seems clear that random mutations in thegenes which control proliferation or apoptosis are responsible for cancer. To take the

analogy further, just as evolution allows the survival of the ttest individual, so too



5/57

in the case of carcinogenesis, it is those mutations in individual genes which render

cells most capable of evading normal homeostatic mechanisms that are the muta-

tions detected in successful cancer cells. Success in this scenario must be viewed from

the perspective of the individual cell, not from the perspective of the individualpatient who harbors that cell. Clearly, this success is generally short-lived since if

untreated it leads to the death of the host. Other cells have been more fortunate;

HeLa cells, derived from a cervical carcinoma which killed their host in 1956, can be

found in thousands of research institutes throughout the world. From an evolu-

tionary perspective, in this particular environment, clearly HeLa cells have been

highly successful.

The vast majority of mutations that give rise to cancer are not inherited, but arise

spontaneously as a consequence of chemical damage to DNA resulting in altered

function of crucial genes. In a few specic cancers, the cervical cancer that gave rise

to HeLa cells would be a prime example, genes encoded by the HPV virus directlyinterfere with gene action and perform the same function as mutations. However, as

will be seen, mutations which inactivate these same genes in non-infected cells have

the same carcinogenic consequences. Thus parallel evolution also occurs during the

genesis of a cancer cell.

In discussing mutations in the context of carcinogenesis we will be using the

broadest denition: the change in the genome of a particular cell. This includes:

point mutations which cause amino acid substitutions; frame-shift mutations or

mutations to stop codons which either truncate the protein product or scramble its

sequence; chromosomal imbalance or instability resulting in amplication, over-

expression or inappropriate expression of a particular gene; loss of a gene or itsfusion with another gene as a result of chromosomal breakage and rearrangement

resulting in a chimeric protein with altered function; epigenetic modications to

DNA of which the most important is the methylation of cytosine in CpG islands

leading to gene silencing. Developing cancer cells select mutations having two basic

functions: mutations which increase the activity of the proteins they code for; this

class of genes are called oncogenes; or mutations which inactivate gene function in

the case of genes classed as tumor suppressor genes. However, regardless of ultimate

eect, the types of chemical damage causing these mutations are believed identical. A

broad understanding of these chemical events are important for two reasons: rst

since these initial events are causative of the whole process of carcinogenesis, their

inhibition would be an eective preventive measure; secondly, several genetic dis-

eases which predispose to cancer have as their origin mutations in genes whose

purpose is to protect DNA from mutational events. Thus the understanding of these

events has direct clinical relevance.

2.1. Mutations require proliferation

It is important to note that chemical damage to DNA itself is not a mutagenic

event. DNA replication and subsequent cell division is necessary to convert chemicaldamage to an inheritable change in DNA that we call a mutation. Thus, proliferation

is a vital factor in the formation of mutations and in the expansion of clones of cells



6/57

bearing these mutations. This is illustrated in Fig. 1 and has been eloquently dis-

cussed by Ames et al. (1993).

Because of the multiple checks and balances that exist in stem cells to limit inap-

propriate proliferation, with few exceptions, malignant human cells must accumulate

multiple mutations in crucial cellular genes that allow their autonomous replication

and invasion. Yet mutation at a particular genetic locus is a relatively rare event. Even

after deliberate chemical damage to a cell in a laboratory situation, the frequency of

mutations at a particular allele is of the order of 106, i.e., only one cell in one million

is mutated. Mutation rates in human stem cells may be expected to be of the order of

1010/cell division, a very low probability, yet because of the large number of pro-

liferating stem cells it appears likely that initiation is a common event and all adults

probably contain many mutated cells. Fortunately, a successful human cancer cell is

required to have mutations in at least ve genes, as elegantly shown in the case ofcolon carcinoma, with each mutation creating a cell increasingly well adapted for

autonomous growth in the host organism (Cahill et al., 1999; Cho and Vogelstein,

Fig. 1. Role of proliferation in the sequential acquisition of cancer-causing mutations. Because of the

large number of normal stem cells there is a high probability of unrepaired DNA damage causing a single

mutation in a critical gene leading to the formation of an initiated cell. Additional proliferation is nec-

essary to produce a clone of at least 106 cells in order that a second, third, etc. mutation has a nite

probability of occurring. As described in the text, each mutation results in a cell progressively better

adapted to avoid normal controls on proliferation and apoptosis. Mutations in genes such as p53, and

chromosome instability resulting from telomere erosion, will act to increase the mutation rate in cells

progressing to neoplasia.



7/57

1992). Because the probability of a single cell simultaneously acquiring these muta-

tions is vanishingly small, this sequential process of acquisition of mutations can only

be achieved if cells bearing the initial mutation, the so-called initiated cells, clonally

expand until the population increases to many millions. In this population theprobability of a second mutation at a critical locus in one of the cells again reaches or

exceeds unity. This process of clonal expansion must then be repeated so that sub-

sequent mutations can be amassed and cells become progressively better adapted to

an independent life. The sequence of mutations is shown in Fig. 1. This process is

observable clinically as disease progression characterized by an increased growth rate,

acquisition of the ability to invade neighboring normal tissue and to metastasize and,

after application of chemotherapeutic agents, to become progressively drug-resistant.

2.2. DNA is subject to chemical damage

Although endowed with almost magical properties, DNA nevertheless is a mol-

ecule whose chemical bonds obey the same laws as other chemicals and which exists

in an aqueous environment at 37C in the middle of a cell whose very existence

depends upon making and breaking chemical bonds. Thus it is perhaps not sur-

prising that DNA constantly suers chemical damage, some as a consequence of

spontaneous thermal eects, some as a consequence of chemical attack by other

reactive molecules. It has been estimated that approximately 70% of cancer in

Western populations can be attributed to diet and lifestyle with exposure to tobacco

products the major contributor at 30% (Doll and Peto, 1981). However, much of the

remaining increased risk appears to be associated with deciencies in dietary factors,principally fruits and vegetables, which exert a protective role on cancer induction.

When chemical damage occurs as a consequence of exposure to exogenous agents,

either chemical or physical, these agents are generally carcinogenic and the type of

damage and mutations they induce can act as a molecular ngerprint indicating

exposure to these environmental carcinogens (Greenblatt et al., 1994; Multani et al.,

2000). It is clear however that many human cancers occur in individuals without

obvious exposure to environmental carcinogens and many human cancers occur in

organs for which no environmental or genetic causes have yet been identied. It must

be deduced then that spontaneous DNA damage does occur which gives rise to

carcinogenic mutations.

By understanding the causes underlying the genetic damage that results in cancer

we are in a position to reduce its incidence. 20th-century medicine has made great

strides in reducing the incidence of infectious diseases through eective vaccination

programs, as for example with smallpox and polio, and creating eective public

health programs providing for example, safe drinking water. It is hoped that 21st

century medicine will place equal emphasis and have equivalent success in the

reduction of cancer rates through focused preventive measures.

2.2.1. Induction of spontaneous DNA damageSpontaneous DNA mutations can occur directly as a consequence of errors in

replication, or indirectly as a consequence of chemical damage to DNA leading to



8/57

errors in the correct reading of the damaged DNA by DNA polymerase during the

process of replication. Fortunately, cells have evolved highly ecient mechanisms

for replicating their DNA which combine high-delity DNA polymerases with

subsequent proofreading capabilities. As a consequence, the direct error rate duringnormal replication of DNA is of the order 1X3 1010 mutations/base pair/cell di-

vision in a human genome of approximately 2 109 base pairs. Thus, in a single

stem cell, one miscoding error would be introduced every 10 divisions. Because

approximately 97% of DNA is non-coding and because of the redundancy of codon

recognition, many base changes do not give rise to amino acid substitutions. Thus,

the functional mutation rate must be several orders of magnitude below the actual

mutation rate. Nevertheless, with an estimated 1016 cell divisions occurring in an

individual's life span, a total of 1015 base-pairs can accumulate, perhaps 103 base pair

changes in each of the estimated 1012 cells capable of replication in an adult human.

There thus seems a low probability that any one of these could eect the oncogenesor tumor suppressor genes known to be mutated in cancer. Other mechanisms

therefore must exist that cause the observed mutations. Mutations as a result of

chemical damage to DNA appear to be a major factor in initiating a cascade of

events, one of which is an increased mutation rate, the so-called mutator phenotype,

as a result of damage to genes whose function is to ensure the delity of DNA

replication (Jackson and Loeb, 1998; Loeb, 1991) (see also Section 2.4.1).

Spontaneous DNA damage is a frequent event as a result of the inherent insta-

bility of the DNA molecule: depurination from breakage of the N-glycosidic bond

connecting purines to deoxyribose occurs at the rate of 104 events/cell/day (Lindahl

and Nyberg, 1972); deamination of cytidine to uridine occurs about 20 times/cell/day, while deamination of methylcytosine to form thymidine is probably the most

frequent spontaneous chemical event with mutagenic potential (Jones et al., 1992).

Mutations occur during replication because: apurinic sites can result in random base

insertions; uridine when in DNA will base-pair with adenine leading to a G 3 T

mutation, while deamination of methylcytosine will lead ultimately to a C 3 T

transition, a mutation frequently observed in human cancers (Jones et al., 1992). In

addition to these spontaneous changes, DNA damage occurs as the result of

chemical attack, in large part by products of oxidative metabolism, and is probably

the most frequent potentially mutagenic event. Although estimates vary, production

of 8-hydroxydeoxyguanosine, perhaps the most dangerous of these mutagenic

products (Cheng et al., 1992), occurs to the extent of 2 104105 lesions/cell/day

(Shigenaga et al., 1989). Fortunately, none of these lesions accumulate as evolution

has developed a number of DNA-repair enzymes, which can rapidly restore the

damaged sequence.

2.3. Induction of DNA damage by exogenous agents

2.3.1. Chemical carcinogens

DNA is also subject to damage from exogenous agents both chemical andphysical, most of which are now recognized as environmental carcinogens. For both

types of agent the most frequent chemical reaction giving rise to DNA damage can



9/57

be characterized as an electrophilic attack upon a tissue nucleophile (Miller and

Miller, 1975). The most signicant of tissue nucleophiles to be damaged by chemical

attack of this type is guanine, and the chemical changes induced are now known to

interfere with base-pair recognition during replication.Perhaps the earliest example of environmental carcinogenesis was reported in

1775 and involved tumor induction in workers exposed to coal tar. This lead ulti-

mately to the identication of the polycyclic aromatic hydrocarbon 3,4-benzpyrene

and other polycyclic hydrocarbons in coal tar and the discovery of their action as

skin carcinogens in laboratory animals. Similarly, the discovery of a high frequency

of bladder carcinogenesis in workers in the rubber and chemical industries lead to

the identication of 2-naphthylamine as a bladder carcinogen. With the growing

realization that some human cancers have an environmental origin that could be

linked directly to chemical exposure, the list of carcinogenic chemicals rapidly ex-

panded (Doll and Peto, 1981). What was immediately apparent was the greatchemical diversity of these structures, and for many of them such as the polycyclic

aromatic hydrocarbons, their great chemical stability. How can we explain their

similarity of actions in causing cancer and their ability to cause profound changes in

cell behavior? Major insights to this question came from the work of the Millers in

the 1960s and '70s with their discovery that these stable chemical carcinogens un-

derwent a process of metabolic activation by enzymes normally involved in the

detoxication of xenobiotic compounds, to yield highly reactive chemical species

the electrophiles mentioned above (Miller and Miller, 1975).

The sequence of events leading to DNA adduct formation and carcinogenesis can

be best exemplied by reference to one of the simplest chemical carcinogens, dim-ethylnitrosamine. This compound was widely utilized as a chemical solvent and was

investigated because of suspicions that it caused liver damage in exposed workers.

This suspicion was conrmed when laboratory rats developed a similar pathology

after exposure. Its carcinogenic potential was discovered serendipitously when ani-

mals surviving acute doses later were found to develop liver carcinomas (Magee,

1972). This accidental discovery had major repercussions: not only was a new in-

dustrial carcinogen discovered but this class of carcinogen, the N-nitrosamines, were

found to be present in a large number of consumer items from beer, to tobacco

smoke to cosmetics (Hecht, 1997). In addition, it was found that nitrosamines could

be formed in the acid environment of the stomach after ingestion of primary and

secondary amines, found in high levels in sh, and of sodium nitrite, also found in

salted sh as a preserving agent. It is now believed that this endogenous production

of nitrosamines explains the particularly high incidence of gastric cancer in Japan

and Iceland where salt-preserved sh is a major dietary item (Mirvish, 1995). From

the perspective of the cancer researcher striving to understand the nature of the

interaction of chemical carcinogens with the cell, perhaps the greatest benet was the

chemical simplicity of the electrophile a methylcarbonium ion CH3 generated by

metabolic activation of dimethylnitrosamine. With the use of C-14-labeled carcin-

ogens it was soon discovered that 06 methylguanine was a product of reaction ofactivated nitrosamines with DNA and that this base, if not repaired, could introduce

point mutations which were potentially carcinogenic (Lawley, 1980; Lawley and



10/57

Shah, 1973). The sequence of events from the activation of dimethylnitrosamine and

production of the potentially mutagenic lesion 06 methylguanine is shown in Fig. 2.

Shortly after this discovery, the metabolic conversion of polycyclic aromatic hy-

drocarbons to diol-epoxides and the subsequent reaction of the unstable epoxide

group with the N-2 position of guanine was discovered, a lesion also with potential

mutagenic properties (Jerey et al., 1977; Tucker et al., 1988). The signicance of

these ndings cannot be understated as they represented the beginning of our mo-

lecular understanding of cancer. Prior to these discoveries the role of DNA damage

in the process of carcinogenesis was unclear and many competing hypotheses existed.

The development of rapid in vitro assays for the detection of environmental

mutagens was an additional repercussion of the realization that carcinogens cause

potentially mutagenic DNA adducts. Principal among these was the Ames test

conducted in Salmonella bacteria which allowed the rapid and semi-quantitative

assessment of the mutagenic ability of test chemicals in the presence or absence ofmammalian metabolic activation (Ames, 1984). As a result of this and other tests

there was a growing awareness of the presence of potential carcinogens in food and

Fig. 2. Metabolic activation of dimethylnitrosamine and reaction product with DNA. Metabolic acti-

vation by cytochrome P450 enzymes occurs mainly in the liver and results in the formation of the

potentially mutagenic adduct 06 methylguanine. If unrepaired, this adduct can base-pair with adenine

instead of cytidine during DNA replication to form a point mutation.



11/57

the environment (Ames, 1986). The frequent lack of correlation between the ability

to induce mutations in bacteria and the positive carcinogenic potential of the same

chemical when tested at high concentrations in experimental animals lead to erce

debates regarding the validity of classication of mutagens as carcinogens. With thediscovery of the genetic basis of cancer with an absolute requirement for mutation,

this debate has largely subsided. Reasons for the lack of correlation between the

positive long-term animal tests for carcinogenicity and the negative short-term

bacterial tests have been persuasively explained by Ames as being due to the use of

toxic concentrations of the test agent in animals leading to excessive regenerative

proliferation with the consequences outlined in Fig. 1 (Ames and Gold, 1991). The

need for adequate testing of chemicals for carcinogenic potential in humans remains

a vital public health concern without as yet a totally satisfactory solution. Principal

among the problems is that metabolic activation of carcinogens is species and tissue

specic. Perhaps genetic engineering will allow the production of a humanized mousein which the human pattern of metabolic activation of xenobiotics is faithfully

replicated (Wolf and Henderson, 1998).

2.3.2. Physical carcinogens

Here will briey be discussed the carcinogenic potential of ionizing radiation, both

particulate and photon, and of ultraviolet radiation. Although the chemical

reactions dier, both classes of physical carcinogens produce DNA damage which, as

with the chemical carcinogens, can lead to mutations. Ionizing radiation can cause

direct damage to DNA by causing single and double-strand breaks to the DNA

helix, and can also induce indirect damage as a consequence of radiolysis of water to

yield free radicals (Hall and Angele, 1999). It is of interest that the most biologically

damaging radiation produces ionizations that are spaced approximately 2 nm apart

the diameter of the DNA double helix (Hendry, 1991). Ultraviolet irradiation,

though of insucient energy to produce ions, is absorbed by DNA bases and is

suciently energetic to induce chemical reactions. The most relevant of these occurs

between two adjacent thymidines in the DNA helix and results in covalent cross

linking to form a cyclobutane-linked thymine dimer. This disrupts normal base

pairing and presents a formidable obstacle to DNA polymerase, which if not

repaired can give rise to mutations. It is no coincidence that approximately 90% ofskin cancers arise in sun-exposed areas. A rare inherited disease, xeroderma

pigmentosum, which results in acute sensitivity to ultraviolet rays and if not rec-

ognized early, an extremely high incidence of skin cancer, is a result in defects in the

genes responsible for removal and repair of this DNA damage (Lehmann et al., 1977;

van Steeg and Kraemer, 1999). The demonstration that ultraviolet causes DNA

damage and that failure to repair this damage results in carcinogenesis, was the rst

unequivocal evidence that damage to DNA was directly implicated in human cancer.

2.3.3. Many cancer chemotherapeutic agents are carcinogenic

An unfortunate consequence of the DNA damage caused by many chemothera-peutic agents is that patients surviving therapy are at an increased risk of iatrogenic

cancer (reviewed in Fraser and Tucker, 1989). Clearly, this concern also exists with



12/57

radiotherapy, however the localized nature of the delivered dose, in comparison to

the systemic therapy with chemotherapeutic agents, limits the overall risk of sec-

ondary cancer. The cancer chemotherapeutic agents of concern include alkylating

agents such as cyclophosphamide, which chemically reacts with DNA in a mannersimilar to the carcinogens discussed above, and antibiotics such as Doxorubicin

which interacts non-covalently with DNA and induces free-radical damage to the

genome. As would be expected the risk is proportional to the cumulative dose, with

younger patients being more susceptible. The most extensive data for increased risk

has been accumulated in survivors of Hodgkin's disease in which the risk for de-

veloping any secondary cancer, excluding nonmelanoma skin cancer, was 17.6% vs

2.6% in the general population. It is of interest that the most rapidly developing

tumors were leukemias, whose incidence peaked approximately eight years post-

therapy, in contrast to solid tumors which rst appeared some ten years post-therapy

and continued to increase in incidence with time (Tucker et al., 1988). This is con-sistent with evidence to be presented later that leukemias require fewer mutations

than do solid tumors. It is to be hoped that the new opportunities presented by our

increased understanding of the molecular biology of cancer will lead to specic

therapies which do not themselves increase cancer risk.

2.4. Most DNA damage is repairable

Although we have only recently become aware that man-made chemicals and

ionizing radiation induce DNA damage, the genome has been constantly exposed to

chemical damage, both endogenous and exogenous, since life began. In order toprotect against the immediate and long-term eects of excessive mutational rates,

genes, such as p53 discussed in detail below, have evolved whose sole purpose is to

survey the genome for damage and/or to repair this damage. In addition genes exist

whose function is to repair errors introduced during the replication process.

Repair mechanisms dier according to the type of damage for example, the re-

moval of thymine dimers formed as a consequence of UV radiation involves removal

of a whole stretch of DNA followed by resynthesis using the opposing strand as

template; alkylated bases such as 06 methylguanine can be directly removed

without breaking the phosphate backbone; single-strand breaks in the DNA mole-

cule formed as a consequence of damage from ionizing radiation can be directly

repaired (reviewed in Frosina, 2000). Perhaps the only type of DNA damage, which

is not repairable, consists of DNA double-strand breaks. Here, since both strands

are damaged, the cell has no unmodied template that can provide the information

necessary to reconstitute the damage strands. Depending upon the site of the double-

strand break this type of damage can lead to cell death, or, of signicance to the

carcinogenic process, chromosome breakage and recombination with resulting

activation or inactivation of crucial genes.

2.4.1. Defects in DNA repair are responsible for many familial cancer syndromesIt is not proposed to exhaustively deal with these repair mechanisms in this sec-

tion, however, the knowledge that repair capacity exists is vital to our understanding



13/57

that defects in these repair processes play a vital role in carcinogenesis by increasing

the rate of mutation and thus the rate of neoplastic progression. Several inherited

diseases which predispose to cancer have as their genetic origin defects in DNA

repair capacity. These include ataxia-telangiectasia, in which cells are sensitive toX-radiation (Lavin and Khanna, 1999) and the UV sensitivity disease xeroderma

pigmentosum referred to above (van Steeg and Kraemer, 1999). The breast cancer

susceptibility gene BRCA1 appears essential for repair in response to DNA damage

and inactivation of BRCA1 in mouse cells results in increased cell sensitivity to

DNA-damaging agents (Chen et al., 1999a). As discussed in more detail later, one of

the most frequently mutated genes in human solid tumors is p53. p53, which has

been called ``the guardian of the genome'' has among its functions the monitoring of

the integrity of the genome and has the capacity to either delay replication until

repair has been completed, or, if damage is too extensive, to induce a series of events

leading to the programmed death of the cell by a process called apoptosis (Lakin andJackson, 1999). Mutations in one allele of the p53 gene results in cancer susceptible

individuals with the LiFraumani syndrome (Malkin et al., 1990). An additional

syndrome, has been demonstrated to be responsible for the non-polyposis form of

inherited colon cancer. Here, mutations in enzymes involved in mismatch repair,

cause increased genomic instability, The role of mismatch repair deciencies has

been recently reviewed (Lynch and de la Chapelle, 1999).

2.5. Cell-cycle checkpoints restrict replication of damaged DNA

As discussed above, chemical damage to DNA is itself not a mutagenic event, butif unrepaired can be converted to a mutagenic event during the process of DNA

replication. Because DNA synthesis itself is a tightly controlled, highly coordinated

process, delays in progression through S-phase as a consequence of DNA damage or

insucient availability of protein or DNA precursors frequently result in cell death,

chromosomal abnormalities or mutations. Since these latter two events are inti-

mately associated with carcinogenesis, it is not surprising that many of the genes

found to be damaged in cancer cells have actions that relate to cell cycle checkpoint

control. An overview of this G1/S checkpoint is shown in Fig. 3.

To most eectively decrease the probability of mutations, the genome should be

damage-free before the onset of replication. To achieve this, and to also ensure that a

cell has all the nutritional support required for the synthesis of the new strands of

DNA and the protein matrix to allow packaging of the newly synthesized DNA into

chromatin, mammalian cells have devised elaborate checkpoints to prevent prema-

ture entry into the division cycle. The most signicant checkpoint occurs in late G1,

approximately four hours prior to the cell's entry into S-phase. This restriction point,

rst identied by Arthur Pardee (1974), represents the nal checkpoint after which

the cell is irrevocably programmed to begin DNA synthesis. Activation of this

checkpoint control in response to DNA damage, delays entry into S-phase and

provides the cell the time necessary for repair. Many years ago, we demonstratedthat mouse 10T1/2 cells are acutely sensitive to chemical carcinogenesis when

damage occurs just as they exit this checkpoint and progress into S-phase (Bertram



14/57

and Heidelberger, 1974). This suggests that the most dangerous DNA lesions occur

in cells damaged in late G1 and early S-phase after the restriction point, pointing

again to the role of proliferation in carcinogenesis.

Central to the function of this restriction point is the interaction between theretinoblastoma protein RB and the E2F family of transcription factors. In its un-

phosphorylated form, RB tightly binds E2F to form a silencing complex restricting

Fig. 3. Mammalian cell cycle checkpoints. Cells possess multiple mechanisms to prevent inappropriate

passage from G1 into S-phase of the cell cycle were DNA synthesis occurs. Central to this is the phos-

phorylation of RB and RB family members such as p107 by cyclin dependent kinases (CDKs). Phos-

phorylation releases and activates the transcription factor E2F which in turn initiates the transcription of a

number of genes required for S-phase entry and additional cyclins which maintain the phosphorylated

state of RB making continued progression through S-phase a mitogen-independent event. Also shown are

other cell cycle checkpoints which can be activated in G2 or M phase of cell cycle in response to DNA

damage. (For additional discussion see Sherr, 1996, and Reed, 1997). (Figure courtesy of Biocarta.com,

``cyclins and cell cycle regulation'').



15/57

transcription of genes required for cell cycle entry (Weintraub et al., 1995). In re-

sponse to mitogen stimulation, D-type cyclins are synthesized together with their

associated kinases, the cyclin dependent kinases (CDKs) 4 and 6. Initially the activity

of these kinases is inhibited by specic inhibitory factors, the so-called INK4 pro-teins. Sustained mitogenic stimulus results in the release of kinase inhibition and

phosphorylation of RB. This alters its conformation so that it no longer binds E2F

which is released and initiates transcription of two major classes of genes: genes such

as thymidine kinase, dihydrofolate reductase, thymidylate synthase and DNA

polymerase whose actions and products are essential to DNA synthesis, and genes

such as cyclin E and CDK-2 whose actions are to maintain the phosphorylated state

of RB and allow mitogen-independent passage through the remainder of S-phase

cycle. Thus RB phosphorylation constitutes the molecular basis of the restriction

point control (reviewed in Reed, 1997).

The presence of DNA damage induces an independent block to passage throughthis restriction pathway. In response to damage the tumor suppressor gene p53

becomes a transcription factor and induces expression of a series of CDK inhibitors,

p21, p27 and p57 which function to maintain RB in its unphosphorylated state even

in the face of mitogenic stimulation (reviewed in Colman et al., 2000). This control is

released once the cell has eectively repaired its damaged DNA.

3. Pathways to cancer: overview

In the previous discussion I have attempted to present a brief overview of how

mutations are introduced into the genome. The following chapters will present an

overview with key examples of the consequences of these mutations to the devel-

opment of a cancer cell. These examples were chosen to illustrate key genes whose

involvement in human cancer has been most clearly demonstrated and, as an added

criterion, I have chosen examples of genes that show great promise as targets for

molecular intervention. This list of genes is by no means comprehensive but is in-

tended to illustrate the many discrete pathways utilized by cancer cells in order to

achieve unlimited replication.

The incidence of most human cancers increases dramatically with age, and tobriey repeat for emphasis what was discussed above, in these cancers which are

predominantly tumors of epithelial origin, some 47 independent events must take

place before such a cell can be considered malignant. From a functional per-

spective these mutations have two distinct consequences: they allow the inappro-

priate expression or activation of genes, or conversely, they result in the functional

inactivation of the gene or its protein product. Genes which are activated by

mutation are called oncogenes; those inactivated by mutation are called tumor

suppressor genes. As may be deduced, oncogenes are involved in signaling path-

ways which stimulate proliferation, while most human suppressor genes code for

proteins which normally act as checkpoints to cell proliferation or cell death. Indiscussing these genes and the mutations responsible for their altered function, it

will be apparent that ``Murphy's Law'', which states: ``anything that can go wrong



16/57

does go wrong'', is applicable to the genesis of a cancer cell. It must also be re-

membered that the mutations which are available for analysis represent only those

successful mutations allowing uncontrolled proliferation. This situation again

mirrors that found in the Darwinian evolution where only benecial changes

survive as new species.

At the risk of over-simplication, ve major pathways must be activated or in-

activated in the genesis of a cancer cell. These are presented in Fig. 4 and are listed

below:

development of independence in growth stimulatory signals; development of a refractory state to growth inhibitory signals;

development of resistance to programmed cell death, i.e., apoptosis;

development of an innite proliferative capacity, i.e., overcoming cellular senes-

cence;

development of angiogenic potential i.e. the capacity to form new blood vessels

and capillaries.

4. Cancer cells are independent of external growth signals

Normal cells proliferate in response to an array of external, mostly locally pro-

duced, growth factors produced by one cell type to activate a second. These factors

Fig. 4. Pathways to cancer. As a cell accumulates carcinogenic mutations, it progresses through pre-

neoplastic stages characterized by the acquisition of properties, listed under the heading ``progression''

required for its survival. At each stage it must overcome control mechanisms, listed under ``protection'',

which would act to eliminate the mutated cell from the host. The nal control of therapeutic intervention is

becoming much more selective with the identication of crucial targets which distinguish cancer cells from

their normal counterparts.



17/57

include, epidermal growth factor (EGF), broblast growth factor (FGF), tumor

growth factor alpha (TGF-a) and platelet derived growth factor (PDGF) produced by

platelets at sites of wounding. These growth factors exert their proliferative action

after binding to appropriate receptors and induce a cascade of responses most ofwhich involve phosphorylation events. Tumor cells have found mechanisms to enable

constant activation of these proliferative signals. These mechanisms dier from cancer

to cancer depending upon cell type, and within a specic tumor type by pure chance,

but the end result is continued mitogenic stimulation, centering as discussed above, on

cyclin D. Examples are given below and are organized in terms of their position in the

signal transduction pathway from the plasma membrane to the nucleus.

4.1. Inappropriate synthesis of growth factors

A major growth factor in mesenchymal cells is PDGF. When malignantlytransformed these cells give rise to sarcomas, meningiomas and gliomas and other

connective tissue tumors. Inappropriate expression of PDGF can be demonstrated

to induce neoplastic transformation in rodent cells. Various isoforms of PDGF are

expressed in gliomas and in sarcomas whereas expression cannot be demonstrated in

the normal cells giving rise to these tumors (Westermark et al., 1995). At present it is

not clear if these forms of PDGF need to be secreted in order to activate the receptor

for PDGF, or whether this receptor can be activated internally. In any event,

autocrine stimulation by inappropriately expressed PDGF can be demonstrated to

activate downstream signaling pathways leading to mitosis (Black et al., 1994).

4.2. Inappropriate expression of growth factor receptors

4.2.1. Erb-B

There is currently considerable interest in the role of an overexpressed plasma

membrane receptor for heregulin, a growth factor related to EGF. Erb-B is over-

expressed in approximately 30% of breast carcinomas and is associated with a worse

clinical outcome. In most cases analyzed, over-expression is a consequence of gene

amplication, i.e., an increased copy number, usually as a result of end-to-end

replication of this gene at the same chromosome location. Amplication can fre-

quently be detected microscopically as homogeneously staining regions at the

chromosomal location 17q12. The erb-B receptor when activated by its ligand

heregulin, becomes an active Tyr kinase as do many cell surface receptors, and this

Tyr phosphorylation stimulates downstream events resulting in mitotic activation

(Neve et al., 2000). In addition to its mitogenic eects, overexpression of erb-B in

breast carcinoma cells has been shown to lead to increased secretion of vascular

endothelial growth factor (VEGF) which stimulates the angiogenesis necessary for

progressive growth of the tumor (Yen et al., 2000). It is of interest that the proto-

oncogene ras, (see below) which is very infrequently mutated in breast carcinomas

becomes strongly activated in the presence of overexpressed erb-b, perhaps makingthis additional mutation unnecessary (von Lintig et al., 2000). It is at present unclear

why overexpression alone should give rise to activation of this receptor, as no ac-



18/57

tivating mutations have been described and there seems to be no concomitant in-

crease in expression of its ligand. However, there is recent evidence that simultaneous

secretion of prolactin occurs in erb-B over-expressing breast carcinoma cells and that

this hormone may be responsible for receptor activation (Yamauchi et al., 2000). Ifsubstantiated, prolactin and its receptor would provide another exciting target for

breast cancer therapy.

Recently, a humanized monoclonal antibody directed against erb-B has been

developed and has received FDA approval. Its use has been associated with dramatic

tumor responses especially when combined with conventional chemotherapy (Pe-

gram et al., 2000; Pegram and Slamon, 1999; Stebbing et al., 2000). This antibody,

Herceptin, represents only the second monoclonal antibody approved for human use

against cancer and represents an exciting prelude to the increasing numbers of

therapeutically useful antibodies being developed against targets identied by

molecular biology. Reports that erb-2 is expressed in other human tumors, includ-ing gastric carcinoma (Allgayer et al., 2000) and in 21% of several types of lung

carcinoma (Scheurle et al., 2000), suggests that Herceptin may also be useful at these

organ sites.

4.3. Activation of downstream signal transduction pathways

4.3.1. c-abl

Chronic myelogenous leukemia (CML) is always associated with an abnormal

chromosome, called the Philadelphia chromosome, formed by the fusion of a

portion of chromosome 22 to chromosome 9. This chromosomal translocationoccurs between a small 5.8 kb break point cluster region (BCR) within a gene

called BCR on chromosome 22 and a region containing portions of the c-abl gene

on chromosome 9. This fusion results in the transcription of a chimeric mRNA

that is translated into a chimeric fusion protein BCR-ABL of 210 kb size. While

the exact molecular consequences of this fusion protein are not yet clear, in

murine models it is sucient to induce leukemia (Daley et al., 1990) and it may

also be sucient in humans to induce the aberrant proliferation characteristic of

CML. A similar translocation occurs in many cases of acute lymphocytic leukemia

(ALL); here the break point in the BCR gene is in the rst exon leading to a

shorter mRNA transcript size of 7.0 kb. However, the contribution of the c-abl

gene is identical.

c-abl, i.e., the normal cellular gene, is a Tyr kinase with structural similarities to

many similar kinases involved in signal transduction. The BCR gene has serine

kinase activity and it is believed that serine phosphorylation by BCR on the ABL

portion of the chimeric protein results in constitutive activation of the Tyr kinase

ability of this protein. However, it is at present unclear what is the substrate for this

kinase and how subsequent events lead to aberrant proliferation. However, there is

persuasive evidence that ABL signaling results in the inappropriate expression of

many growth factors in CML cells. This topic has recently been reviewed (Sattler andSalgia, 1997). The presence of the BCR-ABL transcript is a sensitive indicator of

disease state and the use of PCR to measure its expression in blood oers an at-



19/57

tractive and rapid means of measuring disease progression or response to chemo-

therapy without the necessity for bone marrow aspiration (Branford et al., 1999).

The apparent central role of the ABL kinase activity has led to eorts to syn-

thesize specic inhibitors. One such inhibitor, STI 571 has been shown to be sur-prisingly specic in its inhibition of this kinase and not to be a general inhibitor of

Tyr kinase. These kinases are used in many signal transduction processes, and if

inhibited would be toxic to normal as well as neoplastic cells. Studies using X-ray

crystallography have recently shown this specic activity is due to the ability of STI

571 to bind the catalytic site of the kinase only when the kinase is in its inactive state

(Schindler et al., 2000). Pre-clinical studies in the mouse have shown that STI 571

can eradicate BCR-ABL positive human leukemia cells (le Coutre et al., 1999).

Reports of a number of clinical trials of this compound have recently been reported

in abstract form and the results look very encouraging: 96% of CML patients

showed complete responses (Druker, 1999) with a 55% response rate in patients withALL (Talpaz, 2000). These studies clearly indicate the crucial importance of an

active BCR-ABL Tyr kinase to the continued proliferation of these leukemia cells

and again indicate the power of rational drug design against a molecularly dened

target.

4.3.2. ras

This gene represents a family of signal transduction molecules which are plasma-

membrane associated and which interact with a large series of downstream signal

molecules with multiple functions including the stimulation of proliferation. The

discovery of mutated ras was important for two reasons: (1) it represented the rst

oncogene to be discovered in human cancer cells; (2) it represents the oncogene

most widely activated in human cancers with incidence levels ranging from 90% in

the pancreas, 50% in the colon, to 30% in the lung with comparable levels being

found in most other solid tumors (Bos, 1989). Mutant ras was discovered by

Weinberg's group (Parada and Weinberg, 1983) in mouse 10T1/2 cells which had

been neoplastically transformed by Heidelberger's group a group in which this

author was a member (Rezniko et al., 1973). Its presence was conrmed in human

bladder carcinoma cells (Parada et al., 1982). In these studies fragments of DNA

from the malignant cells were transfected into growth-controlled immortalizedmouse broblasts. It was found that a small fraction of cells themselves became

neoplastically transformed and that these transformants contained human DNA

sequences encompassing the mutated ras gene. The importance of this nding

cannot be overstated; just as the discovery of a DNA repair deciency for UV-

induced lesions had identied DNA as a target for human carcinogens (Cleaver

and Bootsma, 1975), now some 9 years later, a gene had been identied which

could induce neoplastic transformation. This discovery, and the later discovery of

the rst tumor suppressor gene RB, provided the groundwork for the enormous

explosion in our understanding of human cancer that occurred from that date

forward.ras is one of a large family of proteins that can bind GTP and act as a signal

transduction molecule. The active state of ras is produced by the binding of GTP



20/57

which produces a conformational change and allows interactions of ras with other

downstream signaling molecules. The non-mutated, proto-oncogene form of ras

binds then hydrolyses GTP to GDP which is then released and returns ras to its

inactive state. Mutations in the proto-oncogene c-ras act to decrease the ability ofthis molecule to act as a GTPase. Since GTP is not released by mutated ras it now

acts as a permanently activated signal transduction molecule.

Studies of ras mutations have revealed mutational hotspots centered on codons

12, 13 and codon 61. At codon 12 the most frequent alteration is a G 3 T

transversion causing a glycine 3 valine amino acid substitution (Minamoto et al.,

2000). It is of interest that the induction of ras mutations appears to be an early

event in the carcinogenic process. For example, in the lung approximately 39% of

hyperplastic lesions, considered a carcinogenic precursor lesion, vs 42% of ade-

nocarcinomas had codon 12 mutations. Furthermore, only infrequently was the

same mutation found in geographically separate samples taken from the samepatient indicating that independent events had given rise to these lesions (Westra

et al., 1996) a result consistent with the eld cancerization theory of tobacco

carcinogenesis. The presence of codon 12 mutations in the ras gene has been

exploited recently as a sensitive indicator for the presence of pre-neoplastic cells in

samples as diverse as feces, for the detection of early colon cancer, in bronchial

washings for lung cancer, and duodenal samples for pancreatic cancer (Minamoto

et al., 2000).

The molecular basis for the hotspots on codons 12, 13 and 61 became clear when

the ras protein was crystallized and its three-dimensional structure made apparent.

As shown in Fig. 5, the binding pocket for GTP is dened at one edge by the aminoacids coded by codons 12 and 13, while a crucial region of the protein involved in

hydrolysis of GTP to GDP is encoded by codon 61. Thus both mutated portions of

this protein are intimately involved in either binding or hydrolysis, thus explaining

the biochemical observations that mutated ras lacks GTPase activity. In spite of

intensive research, the full range of cellular and molecular consequences of activated

ras are still not understood completely. Indeed, because of the plethora of signaling

pathways that exist between plasma-membrane associated ras and the nucleus, the

situation appears to be becoming more, rather than less complex (Campbell et al.,

1998; Shields et al., 2000). ras is now known to associate with a second GTPase

necessary for signal transduction activity called p120GAP, and its normal associa-

tion with these proteins critically depends upon the amino acids coded by codons 12

and 61, again underlining the essential nature of these domains (Schezek et al.,

1997). Recent studies have indicated that ras transformation is dependent upon the

third protein Rho, which is a member of another large family of GTPases. These

studies demonstrated that a dominant negative mutant of Rho was capable of

blocking ras transformation in cell culture, while an activated form of Rho enhanced

ras transformation (Prendergast et al., 1995). These Rho family members are known

to be regulators of the actin cytoskeleton, to activate kinase cascades, and to regulate

gene expression, thus making them important players in the overall regulation of cellhomeostasis (McCormick, 1998; Zohn et al., 1998). There is also evidence that a

third GTPase, RAC is also involved in ras signaling (Qiu et al., 1995). This signaling



21/57

Fig.

5.Carcinogenicmutationsinras.AsdeterminedbyX-raycrystallogr

aphy,themutagenichotspotsinrasarelocatedincodons12,

13and61,are

localizedtoregionsoftheproteininvolvedininteractionswithGTPandth

eaccessoryproteinp120.

(Reprod

ucedfrom

Schezek,

1997,withp

ermission.)



22/57

cascade therefore interacts with two important regulatory pathways: (1) in organi-

zation of the cytoskeleton, a structure essential for maintaining normal morphology

and perhaps also for nuclear functions, and (2) via activation of the Raf-1/mitogen-

activated protein kinase pathway in causing activation of transcription factors suchas C-jun by way of a series of signal transducing kinases cumulating in MAP kinase

(Campbell et al., 1998).

Because of the apparent central role of ras mutations in many human solid tu-

mors, there have been many eorts to develop specic therapies directed against this

oncogene. The most promising of these appears to be the development of drugs

which inhibit the association of ras with the plasma membrane. This association is a

result of the addition of a farnesyl isoprenoid moiety in a reaction catalyzed by the

enzyme protein farnesyltransferase. Several inhibitors of this enzyme have been

developed but unfortunately they appear to possess unacceptable side eects (Ro-

winsky et al., 1999). Some of these side eects may be related to inhibition of Rhofunction (Lebowitz and Prendergast, 1998).

4.4. Inappropriate activation of nuclear transcription factors

The ultimate target of the oncogenes discussed above is to achieve activation of

transcription factors such as cyclin D, primarily through protein phosphorylation. A

more direct means would achieve direct activation of these transduction factors thus

circumventing the complexity and feedback controls which exists in upstream signal

transduction pathways. As suggested in the introductory section, Murphy's Law is

obeyed in carcinogenesis and a number of tumors have evolved means to cause directactivation. This is achieved by overexpression of the transcription factor or the

production of mutated proteins with altered functions. Important examples are given

below.

4.4.1. Inappropriate expression of c-myc, a transcription factor

C-myc is a transcription factor whose expression is tightly regulated in normal

cells and is only expressed in S-phase of the cell cycle. In a large number of human

tumor types this regulated expression is lost, and c-myc becomes inappropriately

expressed and/or overexpressed throughout the cell cycle driving cells continuouslytowards proliferation. If this inappropriate expression occurs in epithelial cells and is

the only genetic alteration then growth regulatory genes, in general the tumor sup-

pressor genes to be discussed below, restrict this proliferation and in many cases will

cause apoptosis. However, if these genes are themselves mutated inappropriate

proliferation occurs. In contrast, hematopoietic cells appear to have fewer controls

on their proliferation, perhaps explaining the early onset of many cancers aecting

these cells, and c-myc expression can be oncogenic. Other family members of these

transcription factors are N-myc, overexpressed in neuroblastoma, and L-myc, which

is overexpressed in small cell lung cancer.

One of the most interesting examples by which the regulation of c-myc expressionis perturbed is in Burkitt's lymphoma. Here, a characteristic chromosomal translo-

cation fuses the c-myc gene on chromosome 8q24 with either the heavy chain, j or k



23/57

locus of the immunoglobulin genes on chromosomes 14q23, 2p12 and 22q11, re-

spectively. This results in the removal of c-myc from normal cell cycle control and

places it under the control of genes normally activated by infection. There is a strong

association between prior infection with EpsteinBarr virus (EBV) and Burkitt'slymphoma. EBV gene products increase the translocation rate of c-myc (Li and

Maizels, 1999) and viral infection itself could be expected to increase the tran-

scription from immunoglobulin loci. There is also evidence that translocation may

give rise to additional mutations in the c-myc gene by the antibody hypermutation

mechanism. This has been demonstrated in a B cell line which mutates the c-myc

allele that is translocated into the IgH locus whilst leaving untranslocated c-myc

allele intact (Bemark and Neuberger, 2000).

C-myc functions as a heterodimer with a second transcription factor, max, and

while it seems clear these two function together to facilitate neoplastic transforma-

tion the exact sequence of events has yet to be discerned. A major problem is thecomplexity of cellular events modied by c-myc. It is now clear that this gene par-

ticipates in many aspects of cellular function, including replication, growth,

metabolism and dierentiation (Liao and Dickson, 2000). One central confusing

feature of c-myc overexpression in many cells is that it induces apoptosis, apparently

by increasing transcription of the cyclin-dependent kinase cdc25A, which can induce

apoptosis in cells depleted of growth factors (Galaktionov et al., 1996). It would

seem that in order to overcome apoptosis, tumor cells must also possess other mu-

tated pathways (reviewed in Homan and Liebermann, 1998). Most breast cancers

overexpress c-myc, and this overexpression acts to facilitate the ability of erb-B to

cause proliferation (Neve et al., 2000). There is also suggestive evidence that thepromoting eect of estrogen in estrogen receptor (ER) positive breast tumors may in

part be due to the ability of the estrogen receptor to cause increased transcription of

the c-myc gene and also of telomerase, an enzyme required for cell immortalization,

a topic to be discussed below (Neve et al., 2000).

4.4.2. Mutation of a nuclear hormone receptor leads to blocked dierentiation

A second example of how a chromosome translocation can give rise to a

chimeric protein with altered function is the translocation between the nuclear

receptor for all-trans retinoic acid (RAR-a

) and one of two other chromosomallocations. The result is a nuclear receptor with altered signaling properties mani-

fested by strongly enhanced activity as a transcriptional repressor leading to

arrested dierentiation.

In the 1970s, there was intense interest in the development of retinoids, the

natural and synthetic derivatives of the locally produced hormone, retinoic acid, as

cancer chemopreventive agents. One test for retinoid function was the production

of terminal dierentiation in a human promylocytic leukemia cell line, HL-60

(Strickland et al., 1983). The success of retinoids in model cell culture systems of

dierentiation led Chinese physicians to access the ability of all-trans retinoic acid

to induce remissions in human promylocytic leukemia, PML. This clinical trial wasa dramatic success and was the rst example of dierentiation therapy in cancer

(Huang et al., 1988). The studies were followed up by the French group led by



24/57

Degos who, in 1990, conrmed these dramatic ndings in the majority of PML

patients (Castaigne et al., 1990). Molecular and cytogenetic studies soon revealed

the molecular basis for this disease; a chromosomal translocation which always

involved the RAR-a locus (Castaigne et al., 1992; Chen et al., 1991). The trans-location was found to involve one of two other chromosomes that gave rise to

chimeric proteins: PML-RAR-a and PLZF-RAR-a. PML cells containing the

former translocation were found to be sensitive to pharmacological doses of reti-

noic acid, whereas those expressing the PLZF-RAR-a fusion protein were found

resistant. The molecular basis for both the block to dierentiation caused by the

presence of these fusion proteins and the refractory nature of the PLZF-containing

protein has now been resolved.

Retinoic acid nuclear receptors are now known to act both as transcriptional

silencers and transcriptional enhancers depending upon the binding of retinoic

acid. In the absence of the ligand, RAR binds as a heterodimer with RXR, aclosely related receptor, to retinoic acid responsive elements found in the promoter

regions of retinoid responsive genes and silences transcription from these genes.

This is achieved by the binding to the RAR/RXR complex of repressor proteins

such as N-CoR, which possesses histone deacetylase activity. This enzyme removes

acetyl groups from core histones and promotes the tight binding of DNA to hi-

stones thereby preventing access to transcriptional factors. Upon binding of reti-

noic acid to RAR-a a conformational change is induced which causes release of

the repressor and its exchange for proteins with transcriptional activity. One of the

actions of these transcription factors is to acetylate the core histones releasing

DNA and allowing access to the transcription complex (Chen et al., 1999b; Wole,1997). In studies of the PML fusion proteins it was found that PML-RAR-a would

release the repressor in the presence of high doses of retinoic acid, whereas the

PLZL-RAR-a fusion protein had an additional site of interaction which prevented

release. In the presence of an inhibitor of histone deacetylase such as Trichostatin

A, even PLZL-RAR-a containing PML cells regained sensitivity to retinoic acid

and terminally dierentiated (Grignani et al., 1998; He et al., 1998). The studies

very nicely illustrate successful transitional research from the laboratory to the

clinic resulting in a novel, highly eective therapy for a previously refractory

disease.

5. Cancer cells become refractory to growth inhibitory signals: the discovery of tumor

suppressor genes

The discovery of the ability of oncogenes to induce neoplastic transformation

when transfected into immortalized mouse cell lines, initially seemed to answer

many basic molecular questions about the molecular origins of cancer. However, it

soon became clear that this was not the whole picture and that there existed other

genes that could suppress transformation. For example, most of the studies dem-onstrating oncogene activity of ras and other oncogenes had been performed in a

mouse NIH/3T3 cell line that was already immortal, easy to transfect and, from



25/57

personal experience, was unstable and on the brink of spontaneous transformation.

When oncogenic ras was instead transfected into hamster cells which had not

undergone immortalization, transformation did not occur. However, after spon-

taneous immortalization in culture, ras was capably of inducing transformationindicating that ras did not act alone (Newbold, 1985; Newbold and Overell, 1983).

These and other discoveries led to renewed interest in earlier observations by

Harris that the fusion of normal cells with malignant cells frequently resulted in

loss of tumorigenicity in the hybrids; results which strongly suggested that normal

cells possessed genes which could suppress tumorigenicity an activity lost in

tumor cells (Harris et al., 1969). The studies were taken further by Stanbridge's

group in particular who developed techniques for transfer of single chromosomes

and showed that some, but not all chromosomes derived from normal cells could

achieve suppression of the neoplastic phenotype (Anderson and Stanbridge, 1993).

The stage was thus set for the identication of these genes; a process which has ledto completely new insights into cell control processes. This search is still

continuing.

In contrast to the mutagenic activation of oncogenes, where, because of the

dominant nature of this activation step, mutation of a single allele is sucient to

induce some aspects of the neoplastic phenotype, oncogenic mutations in tumor

suppressor genes result in a lack of function. There are two important conse-

quences of these dierences: rst, because in most cases the normal suppressor

allele can function in the presence of the damaged allele, both copies must be

inactivated before loss of function is manifested; second, again in contrast to

oncogenes whose dominant eects would preclude normal embryonic development,loss of one allele of a suppressor gene is generally silent and allows germ-line in-

heritance of the damaged allele. Familial inheritance of mutated tumor suppressor

genes has tragic results in leading to cancer-prone individuals, however, the study

of such individuals has allowed signicant breakthroughs in identication of the

genes responsible.

5.1. The retinoblastoma gene RB

Retinoblastoma is a childhood cancer which is the most common malignant eye

tumor and is responsible for 1% of cancer deaths in children. Approximately 40% of

cases are familial and the remainder sporadic. In familial disease, retinoblastoma

may be present neonatally or develops shortly after birth. It usually presents uni-

laterally in which multifocal tumors may be present. There is a high probability of

the second orbit becoming involved within approximately four years. Survivors have

an increased chance of secondary malignancies particularly osteosarcoma, bro-

sarcoma and Wilm's tumor. In contrast, most of the sporadic cases have only a low

incidence of involvement of the second orbit and a low incidence of secondary

malignancies. This pattern of inherited disease led Knudson to hypothesize a two-hit

theory of carcinogenesis. This hypothesized that in familial cases children are bornwith one damaged and one normal allele which itself becomes damaged shortly

after birth. This explains the frequent occurrence of bilateral retinal tumors in the



26/57

inherited form of this disease and the high frequency of tumors in other organs.

Individuals with spontaneous disease, characterized by a low frequency of bilateral

events and of distant tumors, were hypothesized to be born with two normal alleles

resulting in a low probability that both alleles will be damaged in both retinas(Knudson et al., 1975; Knudson, 1978). This theory received support from the ob-

servation of frequent chromosomal abnormalities in retinal tumors involving lesions

in both copies of chromosome 13q14 (Lemieux et al., 1989).

5.1.1. RB functions to restrict entry into S-phase

The retinoblastoma gene was cloned and found to encode a nuclear protein, RB,

which acts to control entry into the cell cycle. As discussed above in the section on

cell cycle control, RB is normally not phosphorylated and associates with the

transcription factor E2F. This combination acts as a silencing complex (Weintraub

et al., 1995), whose mechanism has recently been elucidated. Just as described for theretinoic acid receptor, the complex maintains core histones in a non-acetylated form

and restricts access to transcription factors (Brehm et al., 1998). After mitogen

stimulation cyclin D/cdk4 phosphorylates RB in the C-terminal region of the protein

which disrupts the binding region for E2F and causes its release. This altered con-

formation also appears to allow access of cdk2 during S-phase which produces

additional phosphorylation further inhibiting E2F binding (Harbour et al., 1999). As

described above, this disruption of RB/E2F allows transcription of crucial genes

required for cell cycle entry.

In view of the apparent central role of the RB proteins as a gatekeeper to S-phase

entry, it is rather surprising that a wider spectrum of tumors is not observed in cases

of familial retinoblastoma. There is evidence that only a sub-set of cells in the de-

veloping retina is susceptible to RB deletion and it has been suggested that these cells

would be normally programmed for apoptosis during development, thus restricting

disease to early childhood. Interruption of this program, perhaps through continued

stimulation into the cell cycle as a consequence of RB deletion, may well be re-

sponsible for this tumor (Gallie et al., 1999). It is curious also that the human retina

appears abnormally sensitive to this mutation. In mice, targeted heterozygous dis-

ruption of the RB locus allows the development of the embryo until about the day 14

when death occurs because of deciencies in blood-forming elements. Homozygousdeletion, the genetic equivalent of human RB carriers, results primarily in pituitary

adenomas, not retinoblastomas (reviewed in Vooijs and Berns, 1999). The ability of

mouse embryos to live to day 14, and the lack of more extensive tumor development

in both mouse and human heterozygotes suggests that other genes may well be able

to substitute for RB. Two such genes have now been discovered, p107 and p130, and

it seems likely that they have overlapping, but not always equivalent functions in the

cell (Classon et al., 2000).

Analysis of inactivating mutations in the RB gene indicate that most are the result

of CT transitions at CpG dinucleotides (CpGs). Such recurrent CpG mutations, are

likely the result of the deamination of 5-methylcytosine within these CpG islands. Amajor proportion of these mutations result in truncated proteins as the result of the

premature termination of protein synthesis either through the introduction of chain-



27/57

termination sequences or altered spice sites resulting in changes in the processing of

mRNA (Lohmann, 1999; Mancini et al., 1997). These deletions in the RB protein

primarily aects sites involved in nuclear localization and phosphorylation. They

also disrupt the sites involved in binding by certain tumor viruses (Templeton et al.,1991), a topic to be discussed below.

5.1.2. RB gene therapy

The realization that mutations in tumor suppressor genes result in a loss of

function which is recessive and requires both copies of the gene to be damaged,

opens up the possibility that gene therapy may be used to reintroduce one or more

copies of the damaged gene. This has been successfully achieved in cell culture ex-

periments utilizing RB negative human tumor lines. In one such experiment, rein-

troduction of RB into several human carcinoma cell lines led to a loss of invasion

capacity but not necessarily loss of tumorigenicity in immunocompromised mice (Liet al., 1996). In a second experiment, reintroduction of RB function in human

prostate carcinoma cells led to decreased tumorigenicity in mice but not to altered

growth rates in cell culture (Bookstein et al., 1990). That other RB family members

may also function as tumor suppressor genes has recently received conrmation in

studies of human lung carcinoma lines which lack expression of functional p130;

reintroduction of this gene strongly suppressed tumorigenicity in nude mice (Claudio

et al., 2000).

5.2. p53

p53 is a tumor suppressor gene which monitors stress and directs the cell towards

an appropriate response. The types of stress to which p53 is responsive include:

anoxia; insuciency of nucleotides for DNA synthesis; the inappropriate activation

of oncogenes; and DNA lesions as diverse as single-strand breaks and covalent

adducts. There is also growing evidence that p53 monitors telomere length and thus

is critically involved in cell senescence. Upon activation, p53 induces either cell cycle

arrest or apoptosis. For these reasons, p53 has been called ``the guardian of the

genome''. Its central role in eliminating the genomic damage so central to the suc-

cessful genesis of the cancer cell is reected by the fact that over 70% of humancancers have defects in this gene, and virtually all have defects in genes upstream or

downstream of p53 function (reviewed in Levine, 1997). As with RB, cancer-prone

families have been identied which possess p53 mutations in one allele (the Li

Fraumani syndrome). The importance of p53 function is indicated by the incidence

of cancer in such individuals of approximately 100% (Malkin et al., 1990).

5.2.1. p53 mutations

Analysis of p53 mutations revealed mutational hotspots localized in evolutionary

conserved regions indicating that these regions were central to p53 function. When

the crystal structure of p53 in a complex with a p53-specic DNA sequence waselucidated by X-ray methods, reasons for these mutational hot spots became ap-

molecular biology of plants cancer

Documents