Tissue Demography and the Distribution of Tumors

Roughly 90% of cancers arise as carcinomas in epithelial (surface) tissues. The epithelium may be the external surface of an organ, such as the skin or outer lining of the intestine, or internal surfaces of the bladder, prostate, breast, and so on. The other 10% of cancers arise mostly as leukemias (blood) and sarcomas (connective tissues, bone, etc.).

Cairns (1975) listed the tissue distributions from the Danish Cancer Registry, as shown in Table 12.1. Peto (1977) estimated that for fatal cancers in Britain, 20% derive from sex-specific epithelial cells (breast, prostate, ovary), 70% derive from other epithelial cells (lung, intestine, skin, bladder, pancreas, etc.), and 10% derive from non-epithelial cells (blood, bone, connective tissues, etc.).

Table 12.1

Cancer incidence in Denmark, 1943–1967.

The age-specific rate of cell division explains part of the relative risk for different tissues. Rare childhood cancers concentrate in tissues that undergo cell division early in life followed by relative cellular quiescence (see Section 2.3). Common adult-onset cancers occur in surface epithelia that renew throughout life, such as in the skin and intestine.

Renewing Tissues and Epithelial Risk

The epithelium of the human colon turns over at least once per week throughout life. As cells die at the surface, they are replaced by new cell divisions. By age 60, a person has been through at least 3,000 replacement cycles, which means that some cell lineages must pass through many generations. Those renewing lineages would be at high risk for accumulating mutations and progressing to cancer.

Cairns (1975) recognized the importance of tissue renewal in the distribution of cell divisions, and the key role that cell division plays in cancer progression. He wrote:

We may ... expect to find, especially in animals which undergo continual cell multiplication during their adult life, the evolution of mechanisms that protect the animal from being taken over by any "fitter" cells arising spontaneously during its lifetime—that is mechanisms for minimising the rate of production of variant cells and for preventing free competition between cells ... Because most of the cell division is occurring in epithelia, that is where we may expect to find the protective mechanisms most highly developed.

Hematopoietic Renewal

The numerous distinct blood cell types derive from hematopoietic stem cells via a complex transit hierarchy (Weissman 2000; Kondo et al. 2003). Figure 12.3 shows the differentiation hierarchy. Only the long-term (basal) stem cell lineage survives over time. The other cell lineages divide a limited number of times, differentiate, and die, to be replaced by new daughter cells derived from the basal stem lineage. I could not find any clear statement about the typical number of cell divisions from the basal lineage to extinction of a transit lineage.

Figure 12.3

The transit lineage of hematopoietic differentiation in adult mice. Long-term hematopoietic stem cells (LT-HSC) renew throughout life. Short-term hematopoietic stem cells (ST-HSC) self-renew over a 6–8 week period. Multipotential progenitor (MPP) (more...)

The long-term stem cells of young mice appear to divide roughly every 10–20 days. No evidence suggests different rates of division between stem cells (Bradford et al. 1997; Cheshier et al. 1999).

Gastrointestinal Renewal

Studies of mice and humans show that the epithelial surface of the intestine sloughs off continually and is renewed by fresh cells (Bach et al. 2000). Renewal occurs by a flow of cells from numerous invaginations—crypts—throughout the intestinal surface (Figure 12.4). Cells flow from the base of each long, narrow crypt to the surface.

Figure 12.4

The morphology of normal colon tissue. Labels show surface epithelium (SE), colon crypts (CC), goblet cells (GC), lamina propria (LP), and muscularis mucosa (MM). The crypts open to the surface epithelium—in this cross section, some of the crypts (more...)

The small intestine of the mouse has about 15 cell layers from the epithelial surface to the base of the crypt (Figure 12.5). In the small intestine, stem cells reside around the fourth cell position from the bottom. Those stem cells produce daughters that flow either down to the lowest layers, where they differentiate into Paneth cells, or upward where the daughter cells continue to divide and differentiate into the functional goblet cells and enterocytes of the intestinal epithelium.

Figure 12.5

Schematic of a small intestine crypt of a mouse. The crypt has about 15 cells from the epithelial surface at the top to the base, as numbered along the right. In three dimensions, the cylindrical lining of the mouse small intestine crypt has about 200–250 (more...)

Figure 12.6 shows the cell lineage hierarchy of the mouse small intestine. The active stem cells divide to give rise to daughter cells. One-half of the daughter cells must remain active stem cells to continue future renewal. The other half of the daughters begins the transit pathway to differentiation.

Figure 12.6

Cell lineage hierarchy in a small intestine crypt of a mouse. Active stem cells give rise to daughter stem cells that remain near the base of the crypt and the first generation of the transit lineage pathway (T₁), the potential clonogenic stem cells. (more...)

In the first few transit divisions, T₁–T₃, the cells retain the potential to return to fully active stem cells in order to replace stem cells that die or to contribute to tissue renewal after injury. Some of those early transit lineage cells differentiate into Paneth cells and flow downward; the others continue to flow upward, divide, and eventually differentiate into the mature epithelial cells. Within a week or so, the daughters of the stem cells have flowed to the surface and died, to be replaced by the continual flow from below. Figure 12.7 gives a rough idea of the three-dimensional crypt architecture.

Figure 12.7

Three-dimensional schematic of a crypt in the mouse small intestine. The positions of the individual cells show how things might look in a typical crypt. The Paneth cells tend toward the bottom, where they contribute to innate immunity by responding to (more...)

Gastrointestinal stem cells remain difficult to identify unambiguously. Through various indirect studies, Bach et al. (2000) conclude that each mouse small intestine crypt has 4–6 active stem cells. Those stem cells divide about once per day; each crypt produces about 300 new cells per day. There are about six transit divisions, so it takes about one week for a daughter cell of the stem lineage to move up, differentiate, and die at the surface. The mouse small intestine has about 7 × 10⁵ crypts, so the whole small intestine of the mouse produces about 2 × 10⁸ cells per day.

The large intestine (colon) has a similar architecture but lacks Paneth cells. Cancer occurs more often in the large intestine than in the small intestine, in spite of the similar tissue architecture and pattern of cellular renewal. Probably the colon suffers greater concentrations of carcinogens that result from digestion and excretion. The human large intestine has around 10⁷ crypts that each renew about once per week. If a stem lineage in the human colon divided once every six days for 80 years, it would divide about 5,000 times. However, the actual history of stem lineages and the number of divisions over time remains unknown.

Epidermal Renewal

The epidermal layer of the skin turns over about every 7 days in mice (Potten 1981; Ghazizadeh and Taichman 2001) and approximately every 60 days in humans (Hunter et al. 1995); however, those numbers must be taken only as rough estimates.

Several lines of indirect evidence suggest that the skin renews by a stem-transit architecture (Watt 1998; Janes et al. 2002). For example, about 60% of basal epidermal cells are progressing through the cell cycle, but in mice only about 10% of those cells can continue through several rounds of cell division after irradiation. Human epidermal cells plated in cell culture also show a distinction between rare cells that have a high capacity for self-renewal and common cells that divide only a few times. Those cycling cells with limited capacity for self-renewal are thought to be the transit population (Watt 1998).

Figure 12.8 shows Potten's model of the epidermal proliferation unit for mice (Potten 1974, 1981; Potten and Booth 2002). Each approximately hexagonal unit of surface skin renews from a basal layer comprising about ten cells, of which only one basal stem cell renews the unit.

Figure 12.8

Architecture of skin renewal in the mouse based on Potten's model of epidermal proliferative units. The top cross-sectional view shows the epidermal layers. About one in ten basal cells are stem cells (S). The neighboring basal cells and all cells in (more...)

Human skin is more complex: it has variable thickness in different locations, often has more layers than mouse skin, and has an undulating basal layer. Most authors agree that stem cells reside at the basal layer and give rise to an upward-migrating transit lineage. Controversy continues over the location of the stem cells in the basal layer, the frequency of stem cells among basal cells, and the architecture of stem-transit lineages and proliferative units (Potten and Booth 2002; Ghazizadeh and Taichman 2005).

The hairs in the epidermis renew by a different process. Figure 12.9 shows the hair cycle, in which each follicle alternates between rest and growth phases. During hair growth, there seems to be a stem-transit type of architecture: stem-like cells replace themselves in the follicular germ and simultaneously initiate transit lineages that move up and continue to divide. After the growth phase, the lower part of the follicle regresses.

Figure 12.9

Life cycle of a mammalian hair follicle. As the follicle moves from the rest phase to the growth phase, the follicular germ region moves downward and becomes an active site of cell division. Transit cells from the follicular germ move upward to form the (more...)

It remains unclear where the stem cells come from to reseed the follicular germ at the start of the next growth phase. Those stem cells may come from cells in the follicular germ of the rest phase, shown as FG(s?) in Figure 12.9, or the next round of stem cells may migrate down from daughter cells produced by the stem cells in the bulge region. Potten and Booth (2002) emphasized the difficulty of interpreting various studies on this issue. Two recent studies favor the bulge stem cells as the progenitors for each new round of follicular growth (Morris et al. 2004; Kim et al. 2006).

In development, the stem cells of the bulge region appear to be the ultimate source for the interfollicular stem cells (those, for example, in Figure 12.8) and at least for the initial seeding of the follicular germ. After injury, the bulge stem cells can regenerate the hair follicle, sebaceous gland, and interfollicular proliferative units (Cotsarelis et al. 1990; Taylor et al. 2000; Potten and Booth 2002).

So far, I have discussed the keratinocyte lineages that produce the hair and the epidermal surface. In those tissues, melanocyte cell lineages provide pigmentation. Recent studies suggest that, in the hair follicles, the bulge region contains melanocyte stem cells (Nishimura et al. 2002; Lang et al. 2005; Sommer 2005). In each hair cycle, the melanocyte stem cells produce some daughters cells that migrate to the base of the follicle where the active keratinocyte transit lineages will be generated. Melanocytes in each new hair cycle seem to derive from the melanocyte stem cells in the bulge region.

Cancer risk concentrates in long-lived cell lineages—the stem lineages. Morris (2004) recently summarized evidence that various skin cancers derive from keratinocyte stem lineages. Similarly, melanomas probably descend from transformed melanocyte stem cells. Alternatively, transformed transit cells may de-differentiate into cancer cells with stem-like properties of renewal.

Other Tissues

The blood, intestine, and skin renew frequently and have clear stem-transit architectures. Several other tissues also appear to have stem lineages that may provide a source for regular renewal, a reservoir for tissue repair, or daughter cell lineages that terminally differentiate (Lajtha 1979; Watt 1998).

Mammalian spermatogenesis has a clearly defined stem-transit architecture of renewal and differentiation (de Rooij 1998). In other tissues, the details of lineage history are less clear at present. Clarke et al. (2003) discuss a model of breast epithelium renewed by a stem-transit hierarchy of differentiation. Numerous recent articles describe the properties of breast stem cells (reviewed by Dontu et al. 2003; Liu et al. 2005; Villadsen 2005). Rizzo et al. (2005) discussed a stem-transit pathway of renewal for the normal prostate, but at present we have only limited understanding of tissue architecture in the prostate. Cells with some stem-like properties may occur in many tissues, but cell lineage architectures probably vary according to demands for cell turnover and regeneration.

Immortal Stranding

In every mitosis, the DNA duplex splits, each strand acting as a template for replication to produce a new complementary strand. It is possible that most mutations during replication arise on the newly synthesized strand. A stem lineage could reduce its mutation rate if each stem cell division segregated the oldest template strands to the daughter destined to remain in the stem lineage and the newer strands to the daughter destined for the short-lived transit lineage.

Figure 12.10a shows Cairns' hypothesis for segregation of DNA template strands. The DNA duplex at the lower left begins with identical DNA strands. The duplex splits as shown, and each strand serves as a template for replication. Suppose, each time a stem cell copies its DNA, that during replication one new mutation arises on the new strand. The "X" marks the new mutation. In the figure, the first round of replication shows the original templates without mutations and the newly replicated strands, each new strand with one mutation.

Figure 12.10

Cairns' (1975) hypothesis of asymmetric DNA segregation in stem cell divisions. (a) Immortal stranding, in which the stem lineage along the bottom always receives the older strand of the DNA duplex in each round of cell division. (b) Segregation of the (more...)

With each subsequent round of replication in Figure 12.10a, the older template without mutations segregates to the stem lineage along the bottom, and the younger strand with one new mutation segregates up to the transit lineage. This pattern reaches a steady state, in which the stem line retains the original template strand and a strand replicated once off the template with one new mutation. At the steady state, the transit lineage always receives a strand copied from the template that carries one new mutation; replication in the transit cell adds another mutation.

Figure 12.10b shows the opposite pattern, in which the newest strand always segregates to the stem lineage along the bottom. The newer strand always has one additional mutation, so the stem lineage accrues one new mutation in each generation.

By the standard view of DNA replication and mitosis, strands segregate randomly to daughter cells. If so, then the pattern by which mutations accumulate would follow a stochastic process between case (a), in which the stem lineage always gets the older strand, and (b), in which the stem lineage always gets the newer strand. Stochastic segregation would, on average, cause mutations to accumulate in the stem lineage at one-half the rate at which mutations arise on newly copied strands.

Cairns (1975) called the pattern in Figure 12.10a "immortal stranding." Any tendency away from purely random segregation and toward immortal stranding would lower the rate at which mutations accumulate in the stem line.

Immortal stranding requires asymmetric stem cell division, in which the fate of the daughters is determined during mitosis, before segregation occurs. Any evidence for immortal stranding also provides evidence for asymmetric stem cell division.

Several recent studies support Cairns' hypothesis of immortal stranding in stem cell lineages. Potten et al. (2002) marked DNA strands in mouse small intestine crypts with tritiated thymidine, then labeled newly synthesized strands with a different label, bromodeoxyuridine. Over time, only a few cells in crypt positions 3–7 retained the initial label; those cell positions delineate the crypt location in which stem cells reside (Figure 12.7). When the second label was removed, the putative stem cells that retained tritiated thymidine lost the second label, bromodeoxyuridine, showing that those cells did pass through the mitotic cycle.

Smith (2005) similarly showed that cells with stem lineage properties in mouse mammary glands retain immortal strands through epithelial tissue renewal.

Studies of asymmetrically dividing cells in tissue culture also demonstrate conditions under which immortal stranding occurs (Merok et al. 2002; Karpowicz et al. 2005). Interestingly, both asymmetric division and immortal stranding may be regulated by p53 and IMP dehydrogenase, the rate-determining enzyme in ribonucleotide biosynthesis (Rambhatla et al. 2005).

Stem Cell Sensitivity to DNA Damage

Mutations in the template strand of a stem cell carry forward through the stem lineage and the renewing tissue. Cairns (1975) suggested that if mutagens or other processes caused significant DNA damage to a stem cell, the cell might undergo apoptosis rather than risk repair. Apoptosis would reliably remove the mutations from the tissue. In particular, Cairns predicted that stem cells would be exceptionally prone to apoptosis in response to DNA damage when compared with other cells. Most other cells have a relatively short expected life for their descendant lineage; for those short-lived cell lineages, DNA damage does not impose such severe risks as for stem cell lineages.

Several studies suggest that stem cells have extreme sensitivity to damage, such that even a single radiation-induced hit can trigger apoptosis (Potten 1977; Hendry et al. 1982; Potten et al. 1992; Potten and Grant 1998). Those studies demonstrated sensitivity in gastrointestinal crypts near where stem cells reside, but it remains difficult to identify the exact location of stem cells in vivo.

We are left with an association between extreme radiosensitivity of a small fraction of cells and the expected location of stem cells. Potten et al. (2002) used the methods described above to label DNA strands and identify label-retaining cells as stem cells. They then found some evidence for an association between those cells that retain label and those cells that undergo apoptosis in response to mild radiation-induced damage.

Tissue Repair and Risk of Symmetric Division

We can measure the age of a DNA strand as the number of strand replications back to some ancestral template. In Figure 12.10 each "X" on a strand measures age back to the ancestral template on the left.

If a stem cell dies, it may be replaced by another stem cell (Cairns 2002). The replacement requires a symmetric mitosis, because both daughters must be retained as stem cells in order to increase by one the number of stem cells in the pool. In a symmetric mitosis, the age of the DNA strands increases in one of the new daughter stem cells. This increase in age can be seen on the right side of Figure 12.10a. In a steady-state stem cell division, the top daughter that would normally segregate to the transit lineage has templates that have ages one and two relative to the initial template of age zero that the main stem lineage has retained.

The lost stem cell may alternatively be replaced by a daughter transit cell (Cairns 2002). If, for example, the most recent daughter transit cell on the right side of Figure 12.10a reverted to a stem cell, strand age would increase by one relative to the lost ancestral stem cell.

Mitogenesis caused by wounds, chemical carcinogens, or irritation increases the rate of cancer progression (reviewed by Peto 1977; Cairns 1998). Presumably wounds and other forms of tissue damage often kill stem cells; repair requires that those stem cells be replaced.

The interesting comparison is: How much of the increased risk comes from the accumulation of mutations in the stem line caused by symmetric mitoses, and how much of the enhanced risk comes from an increased rate of mitosis independently of the distinction between symmetry and asymmetry in DNA strand segregation?