Telomeres as the Key to Cancer

The standard modus operandi for modeling human diseases in the mouse: Find an interesting gene, knock it out, and watch what happens. In theory, the approach makes perfect sense, and scientists have obtained countless subtle insights into the complexities of biology because of it. But mice, of course, are not humans, and many investigators have had to hastily rewrite otherwise elegant theories because of mouse data. One reason? Researchers have taken for granted that telomere length matters. But

May 27, 2002
Jeffrey Perkel
The standard modus operandi for modeling human diseases in the mouse: Find an interesting gene, knock it out, and watch what happens. In theory, the approach makes perfect sense, and scientists have obtained countless subtle insights into the complexities of biology because of it. But mice, of course, are not humans, and many investigators have had to hastily rewrite otherwise elegant theories because of mouse data. One reason? Researchers have taken for granted that telomere length matters. But new work is revealing just how critical these sequences are in human cancer biology.

Composed of both DNA and protein, telomeres are the specialized caps at the ends of linear chromosomes. The telomere's DNA contains hundreds of repetitions of a simple, short sequence (TTAGGG in humans), synthesized by a highly specialized enzyme called telomerase. From a teleological point of view, telomeres exist to solve the end replication problem, which arises from what Elizabeth Blackburn, professor of biochemistry and biophysics at the University of California, San Francisco, and telomerase's discoverer, calls a "glitch in the way the DNA replication machine is set up." The DNA polymerases that duplicate DNA are incapable of copying the very ends of a linear DNA molecule. To avoid the loss of critical genetic material, the cell caps its DNA with long stretches—10 to 20 kilobases (kb) or so in human cells—of noncoding sequence.

Telomeres have another function, too: their structure helps the cell distinguish the normal end of a linear chromosome from a DNA breakpoint. When DNA breaks, says Blackburn, "alarm bells go off." The cell stops dividing and throws its DNA damage repair machinery into the breach. Only when that break is mended will the cell resume its normal function. But a telomere is not a breakpoint, and if the cell makes that misinterpretation severe genetic problems arise because the chromosomes will rip apart during anaphase. Researchers have uncovered a number of proteins that bind the telomeres and help the cell differentiate them from damage breakpoints.1

Though telomerase is required to maintain telomere length, it is inactive in most human somatic cells. Only highly proliferative cells, such as skin cells, intestinal crypt cells, germ cells, and lymphoid cells, express the critical protein component of telomerase, called telomerase reverse transcriptase (TERT). These highly proliferative cells maintain their telomere lengths over time, while the rest lose about 100 base pairs per end per division.

Mice, however, have entirely different characteristics. Analysis of the telomeres of most laboratory strains of mice revealed that they are two to four times longer (30 to 40 kb) than human telomeres. And, says Ron DePinho, American Cancer Society Research Professor and professor of medicine and genetics at Harvard Medical School, murine cells more readily activate telomerase in their somatic tissues. So, while most human cells experience a gradual degradation of their chromosome ends, like the fraying of a well-worn piece of cloth, murine cells do not. And therein lies the problem with using mice as models of human cancer.

Size Matters

When primary human cells are placed into culture, they divide about 50 times and then stop, a threshold known as the Hayflick Limit. These cells have become senescent—alive, but not dividing. The current conventional wisdom holds that the Hayflick Limit results from gradual telomere attrition in these cells. In 1989, Jerry Shay and Woodring Wright, both at the department of cell biology at the University of Texas Southwestern Medical Center, Dallas, and coholders of the Southland Foundation Distinguished Chair in Aging Research, proposed a two-stage mechanism for cellular senescence, the M1/M2 model.2

In this model, M1 (mortality 1) is normal replicative senescence—the Hayflick Limit. "No human cells in vitro have ever been shown to spontaneously escape that blockade," says Shay. But, if the cell accumulates mutations in critical tumor suppressor genes such as p53 or Rb, then it can bypass M1 and continue dividing. Since such cells still lack telomerase, their telomeres continue to shrink. But the mutations these cells harbor keeps them alive at exactly the point when their genomes are becoming unstable, causing the cells to accumulate even more mutations. The cell begins to join telomeres end to end, and the cells enter M2, or crisis.

"As cells walk this telomere plank into cellular crisis, where there is massive cell death and genomic instability," says DePinho, "only a few would-be cancer cells rise from the ashes." These cells possess the right combination of mutations to make the transition from benign to malignant growth. The vast majority of cells die during crisis, says Shay, but one or two in 10 million will survive. And those that survive do so by reactivating telomerase.

Turn On Tert

Several events must occur for cancer to develop. In 1998, Shay and Wright, in collaboration with colleagues at Menlo Park, Calif.-based Geron, demonstrated that introduction of the human TERT gene into normal human cells causes these cells to bypass the normal Hayflick Limit and become immortal.3 But immortal is not the same as cancerous. Telomerase, says Shay, is permissive for cancer, but not causative.

A year later, Robert Weinberg, at the Whitehead Institute for Biomedical Research in Cambridge, Mass., demonstrated that ectopic expression of three genes—TERT, the simian virus 40 large T antigen, and a mutant allele of ras—is sufficient to convert normal human cells into cancerous ones.4 Immortal cells will grow indefinitely, but cancerous cells can induce tumor formation when implanted into mice.

Clinical data exist that support the link between telomerase and cancer. For one thing, the vast majority of—if not all—human cancers have activated telomerase; burgeoning cancers that do not will generally produce weak, nonprogressing tumors. The pediatric cancer, stage IV neuroblastoma, is a good example.

According to C. Patrick Reynolds, head of the developmental therapeutics section, division of hematology-oncology, Children's Hospital Los Angeles, infants and newborns with stage IV neuroblastoma present with metastases in the skin and bone marrow, and diffuse infiltration of the liver, but without the bone lesions characteristic of true stage IV neuroblastoma. Remarkably, in most of these children (about 85%) the neuroblastoma regresses without intervention, beginning at around 3 months of age.

Reynolds says the primary prognostic factor for this disease is N-myc amplification; those patients with amplification "do very badly" and are treated aggressively. Shay, along with Eso Hiyama, studied telomerase activity in stage IV neuroblastoma patients in Japan. They found that biopsies from those children who recovered all exhibited very low telomerase activity. When they examined these patients over time, the scientists observed that as the tumors regressed, the cells' telomere length progressively shortened. These data suggest that regression of stage IV neuroblastoma is caused by the absence of telomerase combined with telomere shortening, which induces cellular senescence.5

Reynolds has done additional analyses showing that, as with N-myc, telomerase is a good prognostic indicator of stage IV neuroblastoma outcomes. If both telomerase RNA expression and telomerase activity are negative, he says, a patient has an extremely good chance of a positive outcome; if either assay is positive, however, the patient may do well, but "it's a mixed bag." These studies, says Shay, demonstrate that telomerase is not required for cancer initiation, but its activation is a necessary step in cancer progression, because cells must be able to stabilize their genomes. Research in mice has since verified these conclusions.

Nice Mice

Unlike human cells, somatic murine cells have constitutive telomerase activity and do not experience crisis. As a result, says DePinho, mice don't suffer from the same kinds of cancers that plague aging humans, such as breast and colon carcinomas, which derive from epithelial cells. Instead, lab mice tend to get lymphomas and soft-tissue sarcomas. "As models of age-dependent epithelial carcinogenesis, mice have been somewhat of a disappointment," concludes DePinho.

Carol Greider, who discovered telomerase as Blackburn's graduate student and is now professor of molecular biology and genetics at the Johns Hopkins University School of Medicine, collaborated with DePinho to create mice that lack the telomerase RNA component. They showed that by patiently backcrossing the mice through six generations, the mice would eventually develop human-length telomeres. When DePinho crossed these "humanized" mice with p53 tumor suppressor heterozygotes, he observed a shift in the cancer spectrum of the mice to a more human constellation of diseases.6

DePinho says that when he and Greider started this work, they hypothesized that telomerase plays a simple, facilitatory role in cancer development. But now, scientists understand that the situation is much more complicated than that. "The telomere and crisis play opposing roles in the initiation versus the progression of cancer," he says. In other words, telomere attrition pushes cells toward genomic instability that can initiate the cancer process. But these are generally feeble, nonmetastatic cancers—carcinoma in situ, or stage IV neuroblastoma. The cells must turn on telomerase to enhance the progression phenotype.

Remodeling Time?

Because of the differences in telomere biology, some researchers are now questioning the utility of standard mouse strains as model systems for cancer research. "The mouse models that have been developed aren't fully representative of human tumorigenesis," says Titia de Lange, Leon Hess Professor of cell biology and genetics at The Rockefeller University in New York. "We need to either rewire the mouse, or just really be careful with what types of models we're using."

David Harrison, senior staff scientist at The Jackson Laboratory, Bar Harbor, Maine, concedes that if telomere shortening is a critical aspect of human cancer development, then telomerase-positive mice with lengthy telomeres are not an appropriate model system. "To the degree that telomere size and telomerase activity are important boundary conditions in the spontaneous development of a particular kind of cancer," he says, "to that degree, obviously you want to choose a mouse whose telomere size and telomerase activity are more like humans."

But, he also notes that like humans, all mice are not equal. When people talk about mice, they're "talking about a very limited number of laboratory mouse strains, a few genetic individuals." It is equivalent, he says, to making broad generalizations about human biology by looking at just a few individuals from the same part of the world. So, while some mice have long telomeres, others do not. For instance, the Mus spretus mouse strain has relatively human-sized telomeres. Greider says that she and DePinho have now migrated their telomerase knockouts into such a strain, and they are in the process of examining tumor biology in this genetic background.

DePinho and Greider's diligence notwithstanding, Harrison says that, in general, researchers need to be more careful with their models. "We're not looking at the whole mouse genome here; we tend to look at a very limited number of mouse strains, and that's probably a mistake," he says. Researchers must instead ask themselves, which kinds of mice are appropriate models for a given type of cancer? It may even be necessary to determine which mice make the best models for given groups of people, he adds.

Much genetic diversity has been captured by producing inbred mouse strains from previously unsampled, wild populations. These strains offer the genetic reproducibility that is so valuable in lab mice, but with a wider variety of genotypes and phenotypes. But Harrison stresses that using mice as models for cancer development has already been quite successful. For instance, every chemical that induces cancer in humans does so in mice as well, proving that the use of mice is an effective and powerful research tool. "If you lose the mouse as a tool, just because of some prejudice about telomeres," he concludes, "you take away a lot of the opportunity for advancement."

Jeffrey M. Perkel can be contacted at
1. T. de Lange, "Protection of mammalian telomeres," Oncogene, 21:532-40, Jan. 21, 2002.

2. W.E. Wright et al., "Reversible cellular senescence: Implications for immortalization of normal human diploid fibroblasts," Molecular and Cellular Biology, 9:3088-92, 1989.

3. A.G. Bodnar et al., "Extension of life-span by introduction of telomerase into normal human cells," Science, 279:349-52, 1998.

4. W.C. Hahn et al., "Creation of human tumour cells with defined genetic elements," Nature, 400:464-8, 1999.

5. E. Hiyama et al., "Correlating telomerase activity levels with human neuroblastoma outcomes," Nature Medicine, 1:249-57, 1995.

6. S.E. Artandi et al., "Telomere dysfunction promotes non-reciprocal translocations and epithelial cancers in mice," Nature, 406:641-5, 2000.

Gone Telomere Fishin'

IN FOCUS | Jeffrey M. Perkel
Courtesy of Peter Lansdorp

FRESH FISH: Quantitative fluorescence in situ hybridization (Q-FISH) analysis of a human lymphocyte cell at metaphase. Chromosomal DNA is stained blue (DAPI) and telomeres light up in yellow (Cy3-labeled telomere peptide-nucleic acid (PNA) probe).

Telomere length matters, of course, but just how does one measure that? In 1995, while on sabbatical in the laboratory of Hans Tanke at Leiden University, the Netherlands, Peter Lansdorp revolutionized the study of telomere biology when he developed a procedure called quantitative fluorescence in situ hybridization, or Q-FISH, which quantitatively measures telomere length.1

Prior to the Q-FISH method, the only way researchers could assess telomere length was using Southern blotting. According to Lansdorp, senior scientist at British Columbia Cancer Research Center and professor of medicine at University of British Columbia, Southern blot analysis of telomere length has two primary drawbacks. First, it requires a great many cells—about one million or so; and second, the analysis produces a smear on the blot, from which the investigator determines the cell population's average telomere length. Q-FISH, on the other hand, relays specific information on the length of each telomere in a cell, making it possible to determine, for example, whether each telomere in a given cell is the same length.

The key to Q-FISH, says Lansdorp, is the use of peptide-nucleic acid (PNA) probes. In typical (non-Q) FISH analysis, the fluorescent probes are made of either DNA or RNA. The hybridization conditions that enable binding of the probe with the target DNA also favor renaturation of the target DNA itself. The probe competes with the DNA strand that is complementary to the target, making binding events stochastic. In other words, FISH data is qualitative, but not quantitative.

In contrast to DNA and RNA, PNA (used in Q-FISH) has an uncharged backbone structure and can hybridize under extremely low-ionic conditions, which inhibit target DNA renaturation. Under these low-ionic conditions, fluorescence intensity correlates linearly with the number of bound fluorophores, and the technique becomes quantitative. "That was the quantum leap," says Lansdorp.

Carol Greider, professor of molecular biology and genetics at Johns Hopkins University School of Medicine, recently used Q-FISH to demonstrate that the cell's shortest telomeres—and not average telomere length—threaten cell survival.2 This finding would not have been possible with Southern blotting, says Greider. "When you are looking at an average you can never identify those [short] chromosome ends."

Despite its benefits, Q-FISH remains technically challenging. For instance, a typical fluorescence microscope, says Lansdorp, is not designed to be a quantitative instrument. "It's not terribly easy to do right," says Elizabeth Blackburn, professor of biochemistry and biophysics at the University of California, San Francisco. "But it can be done right, and it is powerful."

1. P.M. Lansdorp et al., "Heterogeneity in telomere length of human chromosomes," Human Molecular Genetics, 5:685-91, 1996.

2. M.T. Hemann et al., "The shortest telomere, not average telomere length, is critical for cell viability and chromosome stability," Cell, 107:67-77, Oct. 5, 2001.