CELLULAR PROLIFERATION

Neoplasia is the ultimate result of the disruption of exquisite mechanisms regulating normal cell growth. Growth is determined by the balance of cellular proliferation, differentiation, senescence, and programmed cell death. Proliferation occurs as cells traverse the cell cycle (Fig. 3-1). In preparation for cell division, there is a period of deoxyribonucleic acid (DNA) synthesis, designated the S phase. After an intervening gap period, designated the G₂ phase, actual mitosis occurs in the M phase. After another intervening gap period, the G₁ phase, DNA replication can begin again.

Figure 3-1. Regulation of the cell cycle by cyclins, cdks, and cdk inhibitors. In the normal cell cycle, DNA synthesis (in which chromosomal DNA is duplicated) occurs in the S phase, whereas mitosis (in which nuclei first divide to form a pair of new nuclei, followed by actual cellular division to form a pair of daughter cells) takes place in the M phase. The S and M phases are separated by two gap phases, the G1 phase after mitosis and before DNA synthesis, and the G2 phase following the S phase. During these gap phases, the cell is synthesizing proteins and metabolites, increasing its mass, and preparing for the S phase and M phase. Cell cycle progression is regulated primarily at two points, the G2/M and G1/S boundaries, through the coordinated activities of cyclins and cyclin-dependent kinases (cdks), which in turn are negatively regulated by cdk inhibitors (Ink4 and Cip/kip families). The mid-G1 phase is characterized by the interaction between cyclin D1 and cdk4/6. This complex hyperphosphorylates the retinoblastoma protein (pRb) and its family members (e.g., p130). Another important complex at the G1/S boundary is that of cdk2 and cyclin E (cyc E). The result is to release transcription factors such as E₂F that are complexed with pRb. In turn, E₂F binds to and activates the promoters of genes important in DNA synthesis.

The commitment to proceed through DNA replication and cell division occurs during the G₁ phase at the so-called start or restriction (R) point. Cells may exit this cycle of active proliferation before reaching the R point and enter a quiescent phase, G₀. Cells can subsequently re-enter the cell cycle from the G₀ state (see Fig. 3-1). The duration of each cell cycle phase as well as the overall length of the cycle vary among cell types.

Regulation of cell cycle progression appears to be achieved principally by cyclins and cyclin-dependent kinase activity at the G₁/S and G₂/M phase transitions. Cyclin proteins are classified on the basis of their structural features and temporal expression patterns during the cell cycle (see Fig. 3-1). Cyclins A and B are expressed predominantly during the S and G₂ phases. In contrast, cyclins D and E proteins are most active during the G₁ phase.¹ Overexpression of cyclin D1 in fibroblasts results in more rapid entry of cells into the S phase. Cyclin D1 is frequently overexpressed in a number of GI and non-GI malignancies, including those originating from the oral cavity, esophagus, breast, and bladder.²

Each cyclin forms a complex with a cyclin-dependent kinase (cdk) in a cell cycle–dependent fashion. Cyclins function as catalysts for cdk activity (see Fig. 3-1). Cdks physically associate with cyclins through their catalytic domains. The cyclin-cdk complexes regulate cell cycle progression through phosphorylation of key target proteins, including the retinoblastoma gene product (pRb) as well as the Rb family members p130 and p107.³ The final result is progression out of G₁ into the S phase of the cell cycle.

The cell cycle is also regulated by multiple cdk inhibitors; p21^CIP1/WAF1 and p27^KIP1 are inhibitors of cyclin E/cdk2. Originally discovered to be part of the complex containing cyclin D1 and cdk4/6, p21^CIP1/WAF1 is transcriptionally activated by the TP53 tumor suppressor gene product (see Fig. 3-1).⁴ p16^INK4A is another cdk inhibitor that specifically inhibits cdk4 and cdk6⁵ and is part of a larger family of related inhibitors that includes p14, p15, and p18. p16^INK4A is frequently inactivated in esophageal squamous cell cancers and pancreatic ductal adenocarcinomas, a finding that is consistent with its function as a tumor suppressor gene.^6,7 p16^INK4A disrupts the complex of cyclin D1 and cdk 4/6, thereby freeing p21^CIP1/WAF1 and p27^KIP1 to inhibit the activity of cyclin E/cdk2.⁸

PROGRAMMED CELL DEATH AND SENESCENCE

Apoptosis (or programmed cell death) is an important mechanism that counterbalances cell proliferation, and escape from normal apoptotic mechanisms plays a critical role in oncogenesis. Apoptosis is characterized by distinctive features that include chromatin compaction, condensation of cytoplasm, and mild convolution of the nucleus and cytoplasm. These changes are followed by nuclear fragmentation and marked convolution of the cell surface. Eventually, membrane-bound apoptotic bodies that represent the cellular residue are produced and phagocytosed. Apoptosis is distinguished biochemically by cleavage of double-stranded DNA, which results in fragmented DNA.

Studies of the roundworm Caenorhabditis elegans have led to the initial identification of the gene ced-3, a protease that is the major effector of apoptosis. Two key regulators of ced-3, designated ced-9 and ced-4, were found to prevent or induce apoptosis, respectively.⁹ The mammalian oncogene bcl-2 shares homology with ced-9 and protects lymphocytes and neurons from apoptosis¹⁰; bcl-2 complexes with bax, a protein that by itself contributes to apoptosis.¹¹ Of note, both bcl-2 and bax are part of larger gene families, and the stoichiometric relationships among different combinations of the encoded proteins can determine the balance between cell survival and cell death.¹²

Two well-defined pathways that trigger apoptosis have been described in detail. One pathway is mediated through membrane-bound death receptors, which include tumor necrosis factor (TNF) receptors, Fas, and DR5, whereas the other pathway involves activation of TP53 expression by environmental insults such as ionizing radiation, hypoxia, or growth factor withdrawal, with a subsequent increase in the bax-to-bcl-2 ratio. Both pathways converge to disrupt mitochondrial integrity and release of cytochrome c (Fig. 3-2). The so-called apoptosome complex (cytochrome c, caspase 9, and Apaf1) then activates downstream caspases, such as caspase 3, eventuating in cell death. Activation of caspases, intracellular cysteine proteases that cleave their substrates at aspartate residues, is a key step in programmed cell death in mammalian cells.

Figure 3-2. Apoptosis (programmed cell death) counterbalances cellular proliferation to regulate overall tissue growth. A complex interplay of proapoptotic and antiapoptotic molecules results in the downstream activation of caspases that mediate cell death. Some of these signals are initiated through environmental insults that activate the TP53 tumor suppressor gene, and some are initiated through death receptors, including TNF-R1, Fas, and DR5. Death receptors activate caspase 8, which in turn activates BID. In addition, there is an interplay between proapoptotic (bax, bak) and antiapoptotic (bcl-2, bcl-xl) molecules. Both pathways converge on the mitochondria, resulting in the release of cytochrome c and formation of the apoptosome (Apaf1, caspase 9, and cytochrome c). This leads to activation of multiple caspases, in particular caspase 3, and ultimately to cell death. BID, bcl-2 interacting domain; DD, death domain; FLICE, FADD-like IL-1β-converting enzyme; FLIP, FLICE (also known as caspase 8) inhibitory protein; TNF-R1, tumor necrosis factor receptor 1.

Replicative senescence also plays a role in determining overall growth in cell populations. Most primary cells when grown in vitro have a limited replicative potential and eventually undergo senescence.¹³ In contrast, malignant cells can replicate indefinitely. Up-regulation of the telomerase enzyme is essential to escape from replicative senescence. Telomeres are repetitive DNA sequences at the ends of all chromosomes that regulate chromosomal stability. Telomeres shorten with each cell division and, when they have been reduced to a certain critical length, senescence occurs. Cancer cells are able to maintain their telomere length despite multiple cell divisions through the reactivation of telomerase enzyme activity.¹⁴

SIGNALING PATHWAYS THAT REGULATE CELLULAR GROWTH

Cellular proliferation is achieved through transition of cells from G₀ arrest into the active cell cycle (see Fig. 3-1). Although progression through the cell cycle is controlled by the regulatory proteins just described, overall proliferation is modulated by external stimuli. Growth factors that bind to specific transmembrane receptors on the cell surface may be especially important. The cytoplasmic tails of these transmembrane receptor proteins produce an intracellular signal after ligand binding.

In addition to peptide growth factors, extracellular matrix and cell-cell adhesion molecules have a significant impact on cell proliferation. Although the full spectrum of molecules that play a role in cell-matrix and cell-cell adhesion is still not defined, it is known to include integrins, cadherins, selectins, and proteoglycans. Interactions with these adhesion molecules lead to changes in the cell cytoskeleton, indirectly modulating external growth stimuli. Alterations in cell-matrix or cell-cell interactions are particularly important in contributing to the invasive phenotype characteristic of malignant cells.

Interaction of ligands with their receptors at the cell surface induces intracellular signals that ultimately result in alterations in gene transcription. Three important receptor subtypes appear to initiate cellular signaling through ligand-receptor interaction at the cell surface: (1) tyrosine kinases; (2) serine and threonine kinases; and (3) G protein–coupled receptors.

The receptors for many peptide growth factors contain intrinsic tyrosine kinase activity within their intracellular tail. After ligand binding, tyrosine kinase activity is stimulated, leading to phosphorylation of tyrosine residues in target proteins within the cell. The full spectrum of proteins phosphorylated by each tyrosine kinase remains to be determined. Most receptors also autophosphorylate tyrosine residues present in the receptors themselves to initiate signaling and, in some cases, this also causes attenuation of their own activity to effect an intramolecular feedback regulatory mechanism. The receptors for many peptide growth factors, including epidermal growth factor (EGF), belong to this receptor class.

Other receptors on the cell surface possess kinase activity directed toward serine or threonine residues rather than tyrosine. These receptors also phosphorylate a variety of cellular proteins, leading to a cascade of biological responses. Multiple sites of serine and threonine phosphorylation are present on many growth factor receptors, including the tyrosine kinase receptors, suggesting the existence of significant interactions among various receptors present on a single cell.¹⁵ The transforming growth factor-β (TGF-β) receptor complex is one important example of a serine-threonine kinase–containing transmembrane receptor.

Many receptors are members of the so-called seven-membrane–spanning receptor family. These receptors are coupled to guanine nucleotide binding proteins, and are designated G proteins. G proteins undergo a conformational change that is dependent on the presence of guanosine phosphates.¹⁶ Activation of G proteins can trigger a variety of intracellular signals, including stimulation of phospholipase C and the generation of phosphoinositides (most importantly, inositol 1,4,5-triphosphate) and diacylglycerol through hydrolysis of membrane phospholipids, as well as modulation of the second messengers cyclic AMP and GMP.¹⁷ Somatostatin receptors exemplify a G protein–coupled receptor prevalent in the GI tract.

Binding of growth factors and cytokines to cell surface receptors typically produces alterations in a variety of cellular functions that influence growth. These functions include ion transport, nutrient uptake, and protein synthesis. However, the ligand-receptor interaction must ultimately modify gene expression within the nucleus to affect cell proliferation. The regulation of the content and activity of transcriptional factors within the nucleus is the final step in pathways that translate an external stimulus to a change in cell proliferation. These transcriptional factors modulate the expression of genes that control cell proliferation and phenotype.

The Wnt pathway is one important example of a signaling pathway that regulates the cell cycle machinery to control the proliferation of intestinal epithelial cells (Fig. 3-3). Although the details of the specific interactions between the Wnt ligand and its receptor Frz, a member of the seven-membrane receptor family, in the GI tract are not fully clarified, an active Wnt signal ultimately results in the accumulation of β-catenin in the nucleus, where it binds with the transcription factor TCF-4 to activate a set of target genes.¹⁸ Inhibition of the Wnt signal in mice can be achieved by deletion of TCF-4 or overexpression of a Wnt inhibitor designated Dickkopf1, which results in dramatic hypoproliferation of the intestinal epithelium.^19,20 This hypoproliferation appears to be mediated by decreased expression of the TCF-4 target gene c-MYC, which directly represses p21^CIP1/WAF1.²¹ Thus, a Wnt signal stimulates proliferation of intestinal epithelial cells by repressing the cell cycle inhibitor p21^CIP1/WAF1.

Figure 3-3. The Wnt signaling pathway is an important regulator of intestinal epithelial cell proliferation and tumorigenesis. In the absence of a Wnt signal (left panel), β-catenin forms a cytoplasmic complex with APC, Axin, and glycogen synthase kinase-3β (GSK-3β). This β-catenin destruction complex phosphorylates (-P) β-catenin and targets it for degradation via the ubiquitin-mediated proteasomal pathway. In the presence of an active Wnt signal (right panel), β-catenin is stabilized, and excess cytoplasmic β-catenin is translocated to the nucleus, where it interacts with the TCF-4 transcription factor to regulate the expression of many key target genes. VEGF, vascular endothelial growth factor.

Cyclin D1 has an extremely short half-life (<20 minutes) and is a rate-limiting factor for progression through the G₁ phase of the cell cycle (see Fig. 3-1). Consequently, it is one of the most tightly regulated of all cell cycle proteins. Extracellular signals from growth factors, including EGF, colony-stimulating factor 1 (CSF-1), platelet-derived growth factor (PDGF), and insulin-like growth factor (IGF), can regulate cellular proliferation by rapidly inducing the expression of the cyclin D1 gene.²²

Tissue homeostasis is also maintained by growth-inhibiting signals that counterbalance proliferative signals. TGF-β is a potent growth-inhibiting factor that mediates arrest of the cell cycle at the G₁ phase. TGF-β not only induces the transcription of the cell cycle inhibitors p15^INK4B and p21^CIP1/WAF1, but also enhances the inhibitory activity of p27KIP1 on the cyclin E/cdk2 complex (see Fig. 3-1).²³ These effects are mediated intracellularly through the Smad family of proteins.

ONCOGENES

Typically, oncogenes are genes that encode a normal cellular protein expressed at inappropriately high levels or mutated genes that produce a structurally altered protein that exhibits inappropriately high activity. For example, several genes that encode tyrosine kinase–containing growth factor receptors become oncogenes after a mutation results in unregulated tyrosine kinase activity that is no longer dependent on the presence of the appropriate ligand. The normal cellular genes from which the oncogenes derive are designated proto-oncogenes or cellular oncogenes.

More than 80 oncogenes have been isolated, and additional oncogenes continue to be identified. Most of these genes are widely expressed in many different types of tumor cells. Multiple oncogenes (usually in combination with altered tumor suppressor genes) are commonly found within a single tumor.

Several mechanisms can lead to oncogene activation. These include gene transduction or insertion, point mutation, gene rearrangement, and gene amplification. Gene transduction and insertion generally result from retroviral infection. Point mutations result in constitutively active oncogene products. Gene rearrangements can result in oncogenic fusion proteins, and gene amplifications lead to uncontrolled overexpression of a normal gene product.

The proteins encoded by oncogenes comprise at least four distinct groups—peptide growth factors that may be secreted into the extracellular milieu, protein kinases, signal-transducing proteins associated with the inner cell membrane surface (membrane-associated G proteins), and transcriptional regulatory proteins located in the nucleus.

TUMOR SUPPRESSOR GENES

The products of tumor suppressor genes prevent the acquisition of the transformed phenotype in vitro and have similar functional properties in vivo. Mutations that disrupt the biological function of these genes are associated with all GI cancers. Germline mutations of this class of gene underlie most of the known inherited cancer syndromes in which a specific gene has been implicated. A number of these genes and their products have been identified and characterized (Table 3-2).

Table 3-2 Chromosomal Localization and Function of Several Key Tumor Suppressor Genes in Gastrointestinal (GI) Cancers

* Clinical GI disorders associated with defects in some of these genes are listed in Table 3-4.

Initial recognition of the existence of tumor suppressor genes was derived from analysis of families with a markedly increased incidence of specific tumors. Almost all types of tumors have been found to occur in an inherited form. In the GI tract, hereditary colon cancer, gastric cancer, and pancreatic cancer syndromes are the best described and are discussed elsewhere in this text. Within these relatively rare kindreds, as many as half of first-degree relatives of a proband (and each subsequent generation) develop specific tumors, consistent with a single-gene disorder with an autosomal dominant mode of transmission.

Despite the variation in the type of tumor found in different inherited cancer syndromes, a number of features are common to all inherited GI cancer syndromes. Most importantly, the marked increase in risk for a particular tumor is found in the absence of other predisposing environmental factors. In addition, multiple primary tumors often develop within the target tissue, and tumors in these affected members typically arise at a younger age than they do in the general population. Finally, affected individuals are sometimes at risk for some types of tumors outside the GI tract.

These observations led Knudson to hypothesize that tumors in familial cancer syndromes might derive from independent mutations in the two alleles of a specific tumor suppressor gene (Fig. 3-5)—that the first mutation was present in one copy of the gene inherited in the germline and therefore present in all cells in affected family members.⁴⁶ A somatic mutation of the remaining normal allele of the tumor suppressor gene that might occur in any cell would then lead to tumor development, explaining the high incidence of cancer and multiple tumors. The same gene might play a role in the development of the same tumor type in the general population (sporadic cancer), but two independent somatic mutations of each of the two alleles would be required. However, this combination of events should be uncommon and would explain the lower frequency and later age of diagnosis of similar tumors in the general population. Comings was the first to suggest that the relevant gene in a familial cancer syndrome might encode a tumor-suppressing gene product.⁴⁷ Although this two-hit hypothesis has been generally accepted, there are exceptions. For example, there are data suggesting that a single alteration in just one allele of the Lkb1 tumor suppressor gene that underlies the Peutz-Jeghers syndrome may be sufficient for intestinal polyp formation.⁴⁸

Figure 3-5. Knudson’s two-hit hypothesis of tumor suppressor genes. In an inherited cancer syndrome, one chromosome has an inactive tumor suppressor gene locus because of a germline mutation. The counterpart tumor suppressor gene on the remaining paired chromosome is subsequently inactivated by a somatic mutation, leading to formation of carcinoma. In contrast, in a sporadic cancer, the two alleles of the tumor suppressor gene become inactivated through two independent somatic mutations, an unlikely event within a single cell. (Ref. 46).

Loss of Heterozygosity, Allelic Deletion, and Tumor Suppressor Gene Inactivation

Some tumor suppressor genes were first cloned through detection of regions of gene deletion in tumor samples from cancer-prone kindreds by screening of DNA for markers scattered throughout the genome. These deletions targeted the second wild-type allele and served to pinpoint the chromosomal location of the disease-causing gene present on the other allele. Polymorphisms that distinguish between the two different paternal and maternal alleles on a molecular level made these discoveries possible.

Three types of polymorphisms are recognized—single nucleotide polymorphisms (SNPs), restriction fragment length polymorphisms (RFLPs), and microsatellite polymorphisms. SNPs represent single base pair alterations that are typically silent; these are the most abundant type of polymorphism, occurring in approximately every 1000 base pairs throughout the genome.⁴⁹ RFLPs are a unique type of SNP in which the single nucleotide change alters a recognition site for a restriction endonuclease. Thus, digestion of DNA with a restriction endonuclease allows the two different alleles of the same genetic segment inherited from the subject’s two parents to be distinguished (Fig. 3-6). A more widely applicable approach uses polymorphisms present within DNA microsatellite markers. Microsatellite DNA sequences are short repetitive mononucleotide or dinucleotide repeats, such as a poly-A or poly-CA sequence. Microsatellite polymorphisms are much more common than RFLPs in the genome. There is a wide variation in the number of repeats in different alleles, and these differences can be detected through a polymerase chain reaction (PCR; see Fig. 3-6). Although SNPs or microsatellites can be used to detect deletions, SNPs will likely supersede other techniques. Either approach provides a means of assessing whether a specific region of a chromosome is deleted in tumor tissue when compared with normal tissue from the same individual.

Figure 3-6. Genomic polymorphisms facilitate the identification of allelic deletions. A, Single nucleotide polymorphisms (SNPs). SNPs are single base pair changes at two identical positions of a specific chromosome. Most are silent and do not result in alterations in gene products. B, Restriction fragment length polymorphism (RFLP) analysis. Normal homologous chromosomes have sequences that can be recognized by bacterial enzymes, designated restriction endonucleases (REs). These REs cleave DNA at the site of these specific sequences to yield a pattern of restriction fragments. If segments of DNA inherited from the two parents differ by the presence or absence of an RE site, the RE fragments will migrate as bands of different size on gel electrophoresis. After transfer of the restriction fragments to a membrane and hybridization with a specific gene probe, different bands can be detected by autoradiography. In this manner, it can be determined whether both the paternal- and maternal-derived alleles of a gene or a DNA segment are present. C, Loss of heterozygosity (LOH) analysis using microsatellites. Polymorphic microsatellite markers vary in the number of repeats—that is, CA_n—between the two alleles. These markers can be amplified using PCR primers that recognize conserved sequences flanking the microsatellite DNA. When tumor DNA is compared with a normal DNA sample, tumors may exhibit a deletion in one of the alleles. This is LOH. CA, cytosine adenine dinucleotide; PCR, polymerase chain reaction.

These losses, termed loss of heterozygosity or an allelic deletion (loss of an allele from one parent), represent an important mechanism of inactivation of one copy of tumor suppressor genes (see Fig. 3-6). When coupled with a preexisting germline mutation, such allelic deletions provide the second hit, which results in a loss of function of both tumor suppressor gene copies. Other mechanisms of tumor suppressor gene inactivation include point mutation or small intragenic deletions that result in premature truncation of the protein product, or promoter hypermethylation. Transcriptional silencing can result from methylation of CpG islands in gene promoters; this has been demonstrated to occur in the gene encoding p16^INK4A in esophageal and pancreatic cancers and the gene encoding E-cadherin in gastric cancer.⁵⁰

Tumor suppressor genes do not function identically in every tissue type. Consequently, inactivation of a particular tumor suppressor gene is tumorigenic only in certain tissues. For example, the tumor suppressor genes RB, BRCA1, and VHL play crucial roles in retinoblastomas, breast cancer, and renal cell cancer, respectively, but are rarely mutated in GI malignancies. Three tumor suppressor genes shown to have a critical role in the pathogenesis of GI malignancies, APC, TP53, and SMAD4, are described below.

DNA REPAIR GENES

Cellular mechanisms have evolved to preserve the fidelity of DNA. Errors can be introduced into the genome through the spontaneous mismatching of nucleotides during normal DNA replication. This occurs most commonly from slippage in microsatellite DNA, which involves regions of mononucleotide (e.g., poly-A) or dinucleotide (e.g., poly-CA) repeats.⁶⁸ The DNA mismatch repair system corrects these errors. The components of this system have been studied most extensively in prokaryotes and lower eukaryotes, most notably yeast. The enzymes bind mismatched DNA, cut the DNA strand with the mismatched nucleotide, unwind the DNA fragment, fill in the gap with the correct nucleotide, and finally reseal the remaining nick. The human homologs of these DNA mismatch repair genes include hMSH2, hMSH3, hMSH4, hMSH5, hMSH6, hMLH1, hMLH3, hPMS1, and hPMS2, and likely others.

The genes hMSH2 and hMLH1 are the two DNA mismatch repair genes that are most frequently mutated at the germline level in the hereditary nonpolyposis colorectal cancer (HNPCC) syndrome, also known as Lynch syndrome.^69,70 Mutations can lead to functional alterations that allow strand slippage during replication. Affected cells are called replication error (RER)–positive, in contrast to the RER-negative phenotype.^71,72 Because microsatellite DNA sequences are primarily affected by this type of genetic instability, the tumor cells are said to display microsatellite instability (MSI). DNA mismatch repair genes are mutated not only in Lynch syndrome, but also in a subset of sporadic GI cancers, including many arising in the esophagus, stomach, pancreas, and colon. Mechanistically, the absence of DNA repair does not cause cancer directly. Rather, the DNA repair defect creates a milieu that permits the accumulation of mutations in a variety of other genes that contain microsatellite DNA sequences, such as the TGF-β type II receptor, BAX, IGF type II receptor, and E2F-4. This MSI pathway represents a novel mechanism for the accumulation of mutations within a tumor (see Fig. 3-4). It is characteristic of all Lynch-related tumors and is observed in approximately 15% of all sporadic colon cancers.

Errors can also be introduced when individual nucleotides are damaged by chemical factors; the base excision repair system corrects these types of errors. 8-Oxoguanine residues can result from oxidative DNA damage, and these altered bases will inappropriately pair with adenines, ultimately leading to G:C→T:A mutations if uncorrected. MYH is a DNA glycosylase that participates in the repair of these oxidized guanine nucleotides. An autosomal recessive adenomatous polyposis syndrome caused by germline mutations in the MYH repair gene has been identified.^73,74 Interestingly, G:C→T:A mutations in the APC gene were almost universally found in the polyps of patients with germline MYH mutations, indicating that there are important similarities in the molecular pathogenesis of polyps in the MYH and FAP syndromes.

ONCOGENIC SIGNALING PATHWAYS

Individual oncogenes or tumor suppressor genes do not necessarily induce cellular transformation directly but typically function as components of larger oncogenic signaling pathways. Some of the pathways that are particularly relevant for gastrointestinal tumorigenesis include the Wnt and ras signaling pathways. These are pathways that regulate normal tissue homeostasis but become oncogenic when the signals are transduced in an aberrant or amplified manner. The key features of Wnt signaling are illustrated in Figure 3-4. β-Catenin is translocated from the plasma membrane to the cytoplasm. There, it forms a macromolecular complex with the APC protein, glycogen synthase kinase-3β (GSK-3β), and Axin. Phosphorylation of β-catenin by GSK-3β triggers its degradation. In the presence of an active Wnt signal, β-catenin is stabilized, and it enters the nucleus where it interacts with the transcription factor TCF-4 to upregulate a number of key target genes, including c-Myc, cyclin D1, and VEGF. As discussed earlier, Wnt signaling is essential for regulating proliferation of normal intestinal epithelium, and dysregulated Wnt signaling is an almost universal feature of all colon cancers. The latter can result from a mutation in the APC, Axin, or β-catenin genes, but alterations in the APC tumor suppressor gene are the most common. An alteration in just one of these components is sufficient to activate the entire pathway. Thus, it is essential to consider individual genetic alterations in the context of the overall signaling pathway in which they function.

Because pathways are typically not linear, additional levels of complexity arise. There is frequent overlap among pathways, and the distinction between pathways can be somewhat arbitrary. For example, mutations in the K-ras oncogene result in the activation of multiple distinct signaling pathways, including Raf/ERK/MAPK, PI3K/Akt, and NF-κB, all of which play an important role in tumorigenesis (Fig. 3-7). Crosstalk between these effector pathways serves to modulate the cellular responses further. Akt, a target of PI3K, can phosphorylate Raf and thereby regulate signaling through the MAPK pathway.⁷⁵ Finally, each of these signaling pathways regulates multiple biological processes related to tumorigenesis,⁷⁶ including cell cycle progression, apoptosis, senescence, angiogenesis, and invasion.

Figure 3-7. Diversity of signaling through K-ras. Oncogenic K-ras can activate multiple signaling pathways. The mechanisms that determine which pathway may be preferentially activated in a given cell type are not fully defined. Crosstalk between these pathways increases the complexity of the signaling networks. These effector pathways can influence cellular biological processes including proliferation, apoptosis, differentiation, and motility.

Another pathway that plays a particularly important role in gastrointestinal tumors is the cyclooxygenase-2 (COX2) pathway. The enzyme COX-2 is a key regulator of prostaglandin synthesis that is induced in inflammation and neoplasia. Although no mutations of COX-2 have been described, overexpression of COX-2 in colonic adenomas and cancers is associated with tumor progression and angiogenesis, primarily through the induction of synthesis of prostaglandin E2. Inhibition of COX-2 with a variety of agents (aspirin, nonsteroidal anti-inflammatory drugs, or COX-2 selective inhibitors) is associated with a reduced risk of colorectal adenomas and cancer.⁷⁷

CHEMICAL CARCINOGENESIS

Metabolic activation by the host is a key determinant of the carcinogenic potential of many compounds. The initial compound, the procarcinogen, is converted by host enzymes to an electrophilic derivative, which then chemically modifies DNA. Mutations result from errors that occur during DNA replication as a result of distorted base pairs. These mutations, in conjunction with other tumor-promoting factors, facilitate or cause the development of malignancy. Factors that influence the potency of any chemical carcinogen include the equilibrium between the activation of the procarcinogen and deactivation or degradation of the carcinogen.⁷⁸ Deactivation typically occurs through a conjugation reaction, usually in the liver.

These principles are exemplified by experimental colonic carcinomas that arise in rodents fed cycasin, a glucosylated compound present in the cycad nut. The glucose residue of cycasin is cleaved in the rat liver by β-glucosidase to form methylazoxymethanol (MAM), which is subsequently deformylated by enzymes in the liver and colon to give rise to methyldiazonium, a carcinogen. These same metabolites are formed through hepatic enzymatic modification of the compound dimethylhydrazine and result in colon cancer in the rat.

DIETARY FACTORS

Chemical mutagenesis may be especially important in the development of cancers within the GI tract and related organs. The mucosal surfaces from which most primary cancers in the GI tract develop are exposed to a complex mixture of dietary constituents that are potential carcinogens or procarcinogens. The ability of dietary factors to act as mutagens in humans was demonstrated directly in 1995. The frequency of contamination of foodstuffs with aflatoxins, a fungal metabolite, parallels the incidence of hepatocellular carcinoma in various areas of the world.⁷⁹ Studies demonstrating that aflatoxins cause mutations in the TP53 gene in hepatocellular carcinoma have provided a compelling link between genes and the environment.⁷⁹

Nitrates present in many foods appear to be additional dietary constituents that may act as procarcinogens in the GI tract. Diet-derived nitrates can be converted by bacterial action in a hypochlorhydric stomach to nitrites and subsequently to mutagenic nitrosamines.⁸⁰ These events may underlie the documented correlation between dietary intake of foods high in nitrates and the incidence of gastric cancer in different populations.

Other dietary factors may modulate the biological potency of dietary procarcinogens. Variations in the relative and absolute amounts of dietary fats may lead to alterations in the composition of the colonic microflora and their metabolic characteristics, resulting in modulation of the production of enzymes that convert dietary constituents into potentially mutagenic compounds. Changes in dietary fiber content can alter the transit time of luminal contents in the bowel, thereby changing the duration of exposure of the mucosa to potential mutagens. Bile salt content may be an additional luminal factor that can modulate the biological effect of procarcinogens. Deconjugated bile salts may promote carcinogenesis through mucosal injury and enhanced epithelial proliferation.

These mechanisms could explain well-documented correlations between the intake of various dietary constituents and the incidence of colon cancer in certain populations. Populations that have a high fiber intake and resulting fast colonic transit times generally exhibit a lower incidence of colon cancer than populations with low fiber intake and delayed transit. The incidence of colon cancer in Japanese immigrants to the United States who consume a Western diet is much higher than that of native Japanese who consume a traditional Japanese diet.⁸¹

EPITHELIAL-MESENCHYMAL TRANSITION

Modulation of tumor cell interactions with adjacent cells and with the extracellular matrix is an essential step as tumor cells invade through the basement membrane and ultimately metastasize to distant sites. A similar process occurs during normal embryogenesis, when polarized epithelial cells no longer recognize the boundaries imposed by adjacent epithelial cells or their basement membrane and adopt features of migratory, mesenchymal cells. This phenomenon, designated epithelial-mesenchymal transition (EMT), has provided a new model for understanding tumor progression (Fig. 3-8). E-cadherin is a critical component of adherens junctions that maintain cell-cell interactions, and loss of E-cadherin is one of the key features of the EMT phenotype.⁸³ Mutations in E-cadherin are common in many GI cancers, particularly gastric cancer. E-cadherin gene expression can be down-regulated by the transcriptional repressors Snail, SIP1, and Twist,^84–86 but it is not yet clear whether these are relevant in GI cancers.

Figure 3-8. An epithelial-mesenchymal transition (EMT) provides a model for tumor progression and invasion. Epithelial cells maintain their polarity and boundaries with adjacent cells through many junctional proteins, including E-cadherin. The loss or down-regulation of E-cadherin is a key feature in EMT, wherein epithelial cells can adopt a migratory mesenchymal phenotype. In tumor progression, EMT can occur at multiple levels, including the transition from early carcinoma in situ to invasive cancer, as well as the invasion of a tumor cell into blood and lymphatic vessels.

The epithelial basement membrane consists of a dense matrix of collagen, glycoproteins, and proteoglycans and normally does not permit passive penetration of cells. The transmigration of tumor cells through the basement membrane likely involves production of key proteolytic activities. Alternatively, the tumor cell may produce factors capable of activating proenzymes present in the extracellular matrix. For example, the tumor may produce urokinase, itself a protease, or plasminogen activator. Having gained access to the interstitial stromal compartment, tumor cells can then enter lymphatic and blood vessels and metastasize.

ANGIOGENESIS AND LYMPHANGIOGENESIS

Angiogenesis is essential to sustain continued growth of the primary tumor. If new vessels are not developed as the primary tumor expands, cells most distant from available vessels are deprived of an adequate source of nutrition and central necrosis occurs. Neovascularization is also an important permissive factor in facilitating metastatic dissemination of tumors.⁸⁷ A number of protein growth factors, produced by malignant tumor cells and stromal cells, have been found to be potent stimuli of angiogenesis, including vascular endothelial growth factor A (VEGF-A), basic fibroblast growth factor (bFGF) and TGF-β. VEGF-A is perhaps the most critical factor that is up-regulated in most tumor types, including colon cancer. Multiple genetic pathways modulate VEGF-A expression, including Wnt and mutant ras.⁸⁸ Therapeutic strategies that inhibit VEGF-A have demonstrated some promise for patients with advanced colon cancer.⁸⁹

Angiogenesis occurs in an ordered series of events. Endothelial cells in the parent vessel are stimulated to degrade the endothelial basement membrane, migrate into the perivascular stroma, and initiate a capillary sprout. The sprout develops into a tubular structure that in turn develops into a capillary network. In vitro models that recapitulate the early events of angiogenesis indicate that this process involves a balance between proteases and protease inhibitors in a manner similar to that during tumor invasion. Indeed, functional parallels between tumor invasion and angiogenesis are evident in their mutual requirement for cellular motility, basement membrane proteolysis, and cell growth.

In addition to angiogenesis, lymphangiogenesis plays an important role in tumor metastasis. Some important clues into the molecular basis of tumor lymphangiogenesis have been obtained. VEGF-C or VEGF-D bind to the VEGF receptor-3 on lymphatic endothelial cells to stimulate the formation of new lymphatic vessels.⁹⁰ This results in the development of new lymphatic channels within the tumor mass and, consequently, the enhanced dissemination of tumor cells to regional lymph nodes.⁹¹ Strategies to inhibit tumor lymphangiogenesis are being actively pursued.

METASTASIS GENES

It is likely that properties important to the development of metastasis reflect the effects of genes distinct from oncogenes involved in the initial formation of the tumor. Although no gene specifically associated with metastasis has yet been identified, one gene, designated nm-23, may be a potential metastasis suppressor gene. Levels of nm-23 are reduced in a variety of metastatic tumors and in cell lines with high metastatic potential, compared with primary tumors and cell lines with low metastatic potential, respectively.⁹² The function of nm-23 is uncertain but may be inferred from its predicted sequence homology to nucleoside diphosphate (NDP) kinases. NDP kinases are involved in microtubule assembly and signal transduction through G proteins, functions that may be important in the formation of metastases.

DNA-BASED APPROACHES

Progress in the identification of cancer-associated genes coupled with the inherent power of molecular biological techniques to analyze exquisitely small amounts of DNA and protein are leading to more effective diagnostic markers (Table 3-3). The most immediate application is assessment of cancer risk in members of cancer-prone kindreds. Strategies have been developed to identify germline mutations in patients with a variety of inherited GI cancer syndromes, including FAP, HNPCC, and hereditary gastric cancer (Table 3-4). In most of these conditions, there is no consensus hot spot mutational site, so these tests analyze the full gene through a variety of analytic techniques (see Table 3-3). Genetic testing is a powerful tool to identify high-risk families and to define the cancer risk for individual family members. Application of genetic testing must take into consideration the sensitivity and specificity of the assay as well as issues of patient confidentiality and potential impact on medical insurability. For these reasons, genetic counseling is an essential component of the genetic testing process.

Table 3-3 Molecular Diagnostic Techniques for Detection of Cancer-Associated DNA Mutations or Altered Proteins

Table 3-4 Applications of Molecular Diagnostics for Gastrointestinal Cancers

APC, adenomatous polyposis coli; FAP, familial adenomatous polyposis; AFAP, attenuated FAP; HNPCC, hereditary nonpolyposis colorectal cancer; MEN1, multiple endocrine neoplasia, type 1; MSI, microsatellite instability.

Improved detection of sporadic GI cancers and their precursor lesions has also been the focus of research studies. Small numbers of shed cells obtained from stool can be assessed for the presence of mutations in specific tumor-associated genes (K-ras, APC, and TP53) using the PCR assay.⁹³ Detection of ras mutations in DNA extracted from the pancreatic ductal fluid obtained at the time of endoscopic retrograde cholangiopancreatographic evaluation for pancreatic cancer has also been reported.⁹⁴

The MSI test can be performed on archived colon tumor samples and serves as a useful screening test to identify individuals whose colon cancers may have developed as a manifestation of the Lynch syndrome.⁹⁵ Loss of hMSH2, hMLH1, or hMSH6 protein by immunohistochemical staining may provide similar information. Studies have suggested that the MSI status of a colon tumor may be predictive of the response to 5-fluorouracil–based chemotherapy.⁹⁶ Therapies that target specific signaling pathways are likely to increase as our molecular understanding of GI cancers increases. Antibodies that target EGF receptors and block the EGF receptor signaling pathway have proven therapeutic benefit in colon cancer, and their role in treatment strategies has been evolving.⁹⁷ In addition, small molecule tyrosine kinase inhibitors of the c-KIT oncogene now constitute routine treatment of gastrointestinal stromal tumors (see Chapter 30).⁹⁸ Molecular techniques may also find a role in the staging of disease. For example, the PCR assay has been used to detect lymph node micrometastases from colon cancer.⁹⁹ Finally, as more tests for genetic markers become available, monitoring for disease recurrence after surgery may become another important application.

ONCOFETAL PROTEINS

Characterization of malignant and transformed cells has led to the identification of markers that may be useful for the early detection and diagnosis of GI cancers. The most productive approaches have exploited the antigenicity of distinctive cell surface glycoconjugates to prepare antisera or monoclonal antibodies directed against tumor-associated determinants. The first useful marker developed through this approach was the carcinoembryonic antigen (CEA), which was identified by Gold and coworkers after immunization of rabbits with colorectal cancer tissue.¹⁰⁰ The resulting antisera were found to recognize a determinant present in tumor tissue and circulating in blood from patients with colorectal cancer but largely absent from normal colonic mucosa and normal serum. This oncofetal determinant is also expressed in nonmalignant mucosa in association with increased proliferation. On a practical level, however, the CEA concentration is falsely elevated in a variety of inflammatory conditions associated with increased cell turnover, such as ulcerative colitis. In addition, it was noted that CEA could be produced by tumors arising from many sites, particularly those elsewhere in the GI tract (e.g., gastric and pancreatic cancers). This finding underscores the relatively limited tissue specificity of transformation-associated alterations in cell surface determinants. Future strategies to identify new protein markers that may be useful for diagnosis, therapy, or prognosis will rely on emerging proteomic techniques.