Telomerase and Cancer: Kirk A. Landon - AACR Prize for Basic Cancer Research Lecture

Telomeres, the ends of the chromosomes, are required to protect chromosomal ends. A complex set of biological functions is encompassed in the word “capping” of the ends of the chromosomes. The telomere, after all, represents a DNA end, which is at the end of the linear chromosomal DNA, but it has to be treated in a different way from an accidental DNA break in a chromosome. Primarily, capping is accomplished through formation of a specific DNA-protein complex at the chromosomal terminus.

A popular metaphor for what is involved in capping is that the telomere acts in a way comparable to an aglet, which is the end of a shoelace. Like an aglet, the telomere prevents that end from fraying away. What this means at the molecular level is that the telomeric DNA contains binding sites for proteins that protect the end of the chromosome. This was first shown in a model organism, the ciliate Tetrahymena, and subsequently in many other organisms. The telomeres consist, first, of a DNA scaffold composed of a tract of very simple repeated DNA sequences. In the case of humans, the repeat sequence T₂AG₃ is repeated, in tandem, thousands of times at the end of every chromosome. That sequence tract provides a molecular scaffold on which binds a set of proteins, about which much is known. These proteins include DNA sequence–specific binding proteins and other proteins that together, in effect, sheathe the chromosome end. The resulting structure is very dynamic. The proteins come on and off the telomere with surprising speed. Hence, the telomeric structure is continually being assembled and disassembled. Furthermore, the underlying DNA is also dynamic, for two reasons. One reason is that, as was predicted in the 1970s, the known mechanism of DNA polymerase prevents complete DNA replication all the way to the ends of the chromosome at every round of replication. Thus, incomplete replication of the chromosome ends occurs, which, if left unchecked by any compensatory mechanism, leads to a problem of maintaining the full-length of chromosomal DNA molecules. The chromosome ends become progressively shorter, which was predicted eventually to lead, by some unknown mechanism, to cells ceasing to divide. The solution to this theoretical problem was the enzyme telomerase (1). Carol Greider and I identified the telomerase enzyme first in the ciliated protozoan Tetrahymena. We used this organism because it has a large number of telomeres and is a relatively rich source of telomerase. Telomerase contains an RNA component. The RNA has a short template sequence that is copied into DNA, which extends, and thus lengthens, the chromosomal DNA. We were able to show that this also occurs in vivo. It is this addition of telomeric DNA in increments to the ends of chromosomes that offsets and counterbalances the shortening of chromosome ends.

Telomerase is by nature a reverse transcriptase by virtue of its action of copying the short RNA template sequence within the telomerase RNA into DNA; an enzyme that copies RNA into DNA is by definition a reverse transcriptase. Unlike viral or retroviral reverse transcriptases, such as that of HIV-1, the cellular enzyme telomerase specializes in making the multiple short tandem repeats that are at the ends of chromosomes. The protein component of telomerase, telomerase reverse transcriptase (TERT), is indeed a protein enzyme and its amino acid sequence includes reverse transcriptase motifs. However, the RNA is also critically important to the enzyme action, and not only because it provides the template. The template is only a minor part of the entire telomerase RNA molecule. The telomerase RNA is built into the structure of the core ribonucleoprotein complex of telomerase, which contains the protein TERT and the telomerase RNA component. After much study, it has emerged that the RNAs from evolutionarily diverse eukaryotes share a common core structure (2). The core structure includes a conserved pseudoknot, which is one of the parts of RNA that interact tightly and specifically with the protein TERT in the enzyme complex. The pseudoknot is also critically important for the action of telomerase. Thus, the RNA and the protein of telomerase collaborate in the enzymatic action. That property also makes telomerase a very unusual reverse transcriptase.

The ciliated protozoan Tetrahymena thermophila normally has constitutive telomerase and can continue to proliferate indefinitely like a cancer cell. We were originally able to show that if we interfered with the action of telomerase in this organism, the telomeres gradually became shorter and cells eventually ceased to divide. Thus, we converted an effectively immortal, perpetually growing organism into a mortal one by the simple action of making its telomerase nonfunctional, thereby showing that telomerase was important for the continued maintenance of the telomeres and of cell proliferation (3).

Humans also have telomerase. With its different RNA template sequence, it makes the sequence at the ends of our chromosomes, T₂AG₃ repeats, which is very much like that of some ciliated protozoans, T₂G₄ repeats. Those T₂AG₃ repeats are the critical binding sites for sequence-specific telomeric protective proteins. Telomerase levels are very highly regulated in normal human cells. As you might expect, telomerase is active in human stem cells and in germ line cells. At various levels, it is also active in other adult human cells, but the levels are often very low and not enough to sustain telomere length over a lifetime. In humans, gradual shortening of telomeres is observed throughout life in many cell types in every part of the body although the cells have a low level of telomerase. In cell culture, telomere shortening leads to cell senescence as the telomeres no longer can sustain a capping structure at the ends of the chromosomes. In 1998, Bodnar et al. (4) showed in a sufficiency experiment that forced ectopic overexpression of telomerase is sufficient to overcome cellular senescence in culture and maintain the telomeres at a steady length as cells continue to divide. Although that experiment showed that with enough telomerase the telomeres can be maintained, the length at which they are maintained in a cell is the result of a complex interplay among many different factors. Today, many of those factors are known, including proteins that aid in binding to the telomere and protect it from increased by telomerase activity.

In the context of the human body, the fact that the action of telomerase allows cells to keep multiplying can have different kinds of consequences. Although in many normal adult human cell types telomerase is often expressed at very low, sometimes undetectable, levels, it is clear that telomerase has some protective role in such normal cells. In the context of cancer cells, particularly those that are well on the way to a malignant state, telomerase has cancer-promoting properties. It is in that latter context that we are interested in intervening in telomerase action.

As mentioned previously, if telomerase is present in sufficient amounts, it permits cells to keep multiplying. And, indeed, many labs and many groups have now shown that the apparent increase in telomerase is a common feature of human tumor cells. This increased expression of telomerase makes it an attractive target for intervention. We investigated the effect of telomerase inhibition on cancer cells using two different approaches.

The first way of intervening in telomerase action in human cancer cells exploited the fact that the telomeric DNA makes the molecular scaffold for the binding of telomere-protective proteins, which include DNA sequence–specific binding proteins. As described above, the telomeric DNA sequence is specified by copying the telomerase RNA template. We mutated the telomerase RNA template so that now a mutated DNA sequence was added to the chromosomes, directed by the mutated template. The mutated sequence could not bind the DNA sequence–specific protective proteins. Hence, the DNA at the ends of the chromosomes was effectively naked. This had the consequence of leaving the telomeric tip uncapped (Fig. 1). We found that the cellular response to this way of uncapping of telomeres in cancer cells was a very robust apoptotic response (Fig. 2). For example, in the earliest experiments we did, we transfected the mutant template telomerase RNA constructs into human cancer cell lines and selected for cells that stably expressed the mutant-template telomerase RNA (5). These cancer cells already had high levels of telomerase, including high levels of the normal wild-type telomerase RNA. We were never able to obtain cells that stably expressed more than just a low fraction of the mutant-template kind of telomerase RNA. Yet that small fraction of the total telomerase that contained the mutant-template telomerase RNA was sufficient to elicit a robust apoptotic response.

Hence, all one needed to do was spike the endogenous telomerase in the cancer cells with a little mutant-template telomerase RNA that assembled into the telomerase to make mutant repeats. The population of cells stably expressing the various different mutant-template RNAs at low levels could be cultured, although they grew very slowly. Interestingly, despite being continually serially passaged for months, we never observed any reversion or any fast-growing resistant subpopulations growing out from 80 to 100 different clonal lines that were followed. We interpret these findings to mean that as the cells grew, with some probability they would die by apoptosis. Such cell death is, in effect, a one-way street—once they make the decision to enter apoptosis, the cells cannot recover to generate any further daughter cells. Similar findings were also made in the setting of a simple xenograft model: When we xenografted human cancer cells expressing the mutant-template telomerase RNA into a mouse model, we saw greatly decreased tumor growth and some regression of the tumors, accompanied by high rates of apoptosis in the tumors (5).

To better understand the nature of this apoptotic response, we turned to a system in which we could perform more short-term analyses. The problem was that previously, the only stable lines that grew out were those that had expressed extremely low levels of mutant-template telomerase RNA. Therefore, Shang Li developed a short-term expression system for mutant-template telomerase RNA (6). We expressed the telomerase RNA under a promoter that allowed telomerase to be expressed and active, even in cell lines that normally do not express any telomerase. These experiments demonstrated that the telomerase RNA was being expressed in a way that supported its assembly into functional telomerase. We chose as examples for study two mutant-template telomerase RNAs, which were mutated in such a way that they dictated the synthesis of DNA that would be unable to bind the sequence-specific telomere-protective proteins. Within a few days, the cell growth rate was markedly decreased. In contrast, the controls, which received either a wild-type-template telomerase RNA construct or just the empty lentiviral vector, grew normally. We did these experiments in a manner that made the test of the effects as stringent as possible because we did not include a drug selection step after infecting the cells with the lentiviral vectors. Not all cells—only about 90% to 95% of the cells—were infected. Various lines of evidence showed that the cells that grew out were those that either did not receive the lentivector or failed to overexpress the mutant-template telomerase RNA. For example, we could reinfect those outgrowing cells and recapitulate the same growth kinetics compared with the controls for at least three cycles of reinfections. Thus, again, no evidence was found for any resistant subpopulation of cells that evade the effect of mutant-template telomerase RNA expression.

An intriguing aspect of these results was the rapidity of the cell growth response, which we could now examine and analyze. The response to mutant-template telomerase RNA is highly specific for cells that have high levels of telomerase. This response does not require the telomeres to shorten. We used a particular cell line, the LOX human melanoma cell line, which happens to have particularly long telomeres, about 40 kb. DNA Southern blot experiments, done to examine the length of the telomeric DNA, showed that there was no bulk shortening of the telomeres. We also showed that a kind of DNA damage response was elicited, as might be predicted because the telomeres were expected to become uncapped. For example, up-regulation of p21 expression was seen, consistent with a DNA damage response. To determine whether the telomeres had become uncapped, we looked for the appearance of DNA damage foci at telomeres (Fig. 3). For these immunostaining experiments, we used proteins that are known to accumulate in DNA damage foci at sites of DNA damage, such as the protein hRif1. Shang Li and Lifeng Xu showed that when we expressed the mutant-template telomerase RNA, even as early as 3 or 4 days after infection, DNA damage foci appeared at the telomeres, as visualized by immunostaining of the telomere-specific protein TRF2. The mutant-template telomerase RNA caused uncapping of the telomeres. The slowing of cellular proliferation did not require the protein p53, which is ordinarily involved in the responses of cells to DNA damage. One line of evidence for the lack of p53 protein requirement came from the use of a p53-null human colon cancer cell line kindly provided by Dr. Burt Vogelstein. When compared with the parent cell line, in which the p53 locus is left intact, the responses between the lines were similar. The p53 response pathway is commonly abrogated in human cancers. Hence, such a p53-independent apoptotic response is desirable if one is interested in intervening in telomerase action as an anticancer approach. Finally, we observed the same kinds of rapid growth inhibitory effects in the context of a more in vivo–like setting for tumor cells. In collaboration with Jerry Cunha's lab at University of California, San Francisco, human bladder tumor cells were xenografted under the renal capsule in a mouse. In control cells, the wild-type telomerase RNA was introduced, and big robust tumors formed. The tumors that expressed the mutant-template telomerase RNA were smaller and much less vascularized, and the tumors weighed less (6).

This work comprises one approach by which one might exploit the increased expression of telomerase, which is so common in cancer cells by forcing them to make toxic telomeres. Encouragingly, this is effective even with very low levels of expression of the mutant-template RNAs. We are very interested in understanding the signaling pathways leading to the apoptosis, which is under active investigation right now.

Thus far, everything discussed has related to telomerase acting in its familiar mode: making telomeric DNA longer. Curiously, however, the high level of telomerase in the majority of cancer cells—which is typically a couple of orders of magnitude higher in cancer cells than in the differentiated normal cells around it—usually does not lead to the cancer cells having long telomeres. Their telomeres are mostly maintained at a rather short length; the 40 kb telomeres in LOX melanoma cells are the exception rather than the rule. Thus, there exists a puzzle that the abundant telomerase activity is apparently not there to make long telomeres. We therefore investigated what happens when you knock down the high levels of telomerase in human cancer cells. Through this work, we learned, somewhat to our surprise, that we might also be able to incur a more rapid effect on cancer cell growth than expected and even change their cancerous properties in ways that were hitherto unsuspected (7).

Inhibition of telomerase activity leads to a very simple expectation: Failure to counteract telomere shortening, the result of incomplete DNA replication, would cause the telomeres to get shorter and shorter and eventually they would become too short to sustain capping, and some signaling process would direct the cells to cease multiplying or even undergo apoptosis. Indeed, that expected result is exactly what is seen when the telomerase activity in cells is inhibited by, for example, expressing an excess of a catalytically dead version of the protein TERT, which, although it can still assemble into the telomerase ribonucleoprotein complex, forms a complex that is not catalytically active. This inactive complex swamps out the endogenous telomerase. A delay ensues, during which telomeres shorten, before the cells finally cease to grow. Similarly, using inhibitors of telomerase activity, which prevent telomeric DNA synthesis but do not affect the level of telomerase ribonucleoprotein, elicits the same expected time course of events (Table 1).

We did something somewhat different: We knocked down the telomerase RNA levels so that now, not only was there less telomerase enzymatic activity but also the cell was depleted of telomerase ribonucleoprotein itself. The component of the telomerase ribonucleoprotein we targeted for knockdown by RNA interference (short interfering RNA or siRNA) was the telomerase RNA. The siRNA was directed against either the wild-type template sequence or the conserved pseudoknot structure of human telomerase RNA. We could achieve up to about 90% knockdown either by using short hairpin siRNAs expressed in the cells from a lentiviral vector expression cassette or by treating the cancer cells with the short synthetic siRNA oligonucleotides with or without modified backbones (Fig. 4). In all cases, we obtained a surprising result: rapid inhibition of the growth of human telomerase-positive cancer cells.

The unexpectedly rapid response to telomerase knockdown prompted us to carry out many controls to rule out possibilities such as targets other than the telomerase RNA being responsible for the effects on cell growth. Satisfied that the effects indeed resulted specifically from knockdown of the intended telomerase RNA, we analyzed the effects in multiple cell lines. In contrast to the effects of mutant-template telomerase RNA expression on telomerase RNA knockdown, no telomere shortening or telomere uncapping, or DNA damage response, accompanied the cell growth rate decrease. Once again, the effects did not require p53 (7).

The rapid response of telomerase-positive cancer cells to an abrupt depletion of their telomerase despite the lack of telomere uncapping prompted the hypothesis that these effects were unrelated to telomere maintenance or telomere function. Supporting this model, Shang Li, in collaboration with Chris Haqq and his group at University of California, San Francisco, showed that global gene expression profile changed within 4 days after the telomerase RNA was knocked down. The effects were different from the gene expression effects elicited by mutant-template telomerase RNA expression (7). This new gene expression profile defined an interesting signature for telomerase knockdown, marked by diminished expression of 73 genes. These included some intriguing genes, such as cyclin G2 or CDC27—genes that have been implicated in cell cycle progression. Integrin αV, which has been implicated in promoting tumor metastasis, was also suppressed.

Clinically, metastasis is arguably the most important factor in cancer. That telomerase knockdown produced down-regulation of a specific subset of genes that included those implicated in tumor progression properties, such as metastasis, was intriguing. In collaboration with Mohammed Kashani-Sabet and his group at University of California, San Francisco, we examined the effect of telomerase RNA knockdown on metastasis in two in vivo models. The telomerase RNA of mouse melanoma cells (B16 cells) was targeted using ribozymes, and the human telomerase RNA in human melanoma cells was targeted using the siRNA approach described previously (8). Each ribozyme was delivered systemically after B16 cells had been introduced into mice by tail-vein injection. After 30 days, the tumor burden in the lungs, representing metastatic tumors in the lungs, was quantified. The ribozyme treatment decreased the metastatic burden.

The rapid inhibition of cell growth seen on knockdown of telomerase indicated that cancer cells are particularly susceptible to its depletion. This implies that they have become adapted to their high levels of telomerase. As Bernard Weinstein (9) has proposed for other signaling pathways in tumors, cancer cells might be “addicted” to high levels of telomerase, such that their physiology becomes habituated to it that when telomerase is withdrawn, a “cold turkey” response is deleterious to the cell. Clearly, our challenge is to understand how such a mechanism might work.

An important conclusion of this work is that telomerase participates in cell responses in ways that do not seem to involve the telomere itself. Hence, our attempts to intervene in telomerase action have led us unexpectedly down a new avenue in which we find telomerase doing things that we never expected to see. We are hopeful that these avenues will lead us to a greater understanding of the biology that is being played out in the progression to a more malignant state of cancer cells, particularly metastasis.