Cancer Immunotherapy Targeting Survivin: Commentary re: V. Pisarev et al., Full-Length Dominant-Negative Survivin for Cancer Immunotherapy. Clin. Cancer Res., 9:6523–6533, 2003.

The Survivin protein regulates both cell division and cell survival and is overexpressed in the vast majority of human cancers (1). In fact, the gene encoding Survivin is notable for its high degree of tumor-specific expression, ranking among the top five most tumor-specific genes in the human genome based on comparisons of the number of times survivin transcripts appear in tumors compared with normal cells and tissues (reviewed in Ref. 2). In normal cells, Survivin is produced only in small amounts and only briefly during mitosis. However, in tumors, the Survivin protein is continuously present at excess levels, suppressing apoptosis and aiding in cell division (reviewed in Ref. 3). Consequently, Survivin has emerged as a hot target for cancer therapy, with a variety of strategies already articulated for nullifying Survivin in tumors, including (a) small-molecule antagonists that block interactions of Survivin with critical partner proteins (4, 5, 6); (b) antisense oligonucleotides that reduce survivin expression; (c) ribozyme-mediated inhibition of survivin expression (7); and (d) gene therapy using dominant-negative mutants of Survivin that induce cell cycle arrest and apoptosis (8). In this issue of the Clinical Cancer Research, the Gabrilovich group [Pisarev et al. (9)] describe preclinical studies aimed at yet another strategy for exploiting Survivin as a cancer target, namely, using survivin to create a tumor vaccine.

Inducing cell-specific immunity against tumors represents a long-standing goal of cancer immunology (10). To be successful, several conditions must be met, including identifying tumor-associated antigens (TAAs) that are capable of being presented on human leukocyte class I or II antigens (HLA-I or -II) and stimulating host T cells using the available repertoire of T-cell antigen receptors (reviewed in Ref. 10). Ideally, such TAAs should be displayed only on tumor cells and not normal cells, or, at least, such antigens should be expressed only on nonessential normal tissues if tumor specificity is impossible (e.g., breast, prostate, ovary, testes). Furthermore, the TAA should be an essential gene, so that tumors cannot avoid immune surveillance simply by refusing to express it.

Gene ablation studies in mice have confirmed that survivin is an essential gene (11). Knocking-down survivin expression in human tumor cell lines using antisense methods, or interfering with the function of the endogenous Survivin protein by ectopic expression of dominant-negative Survivin mutants, results in arrested cell division and apoptosis, implying that tumor cells cannot replicate and survive without this important and nonredundant protein. Consequently, it stands to reason that tumors could not easily escape therapeutic interventions targeting Survivin by simply shutting-off its expression.

With these issues in mind, the Gabrilovich group set out to ask whether peptides-derived from the Survivin protein could stimulate cytolytic T-cell responses in vitro, using peripheral blood lymphocytes (PBLs) derived from normal volunteers and prostate cancer patients. The strategy they used involved infecting autologous dendritic cells (DCs) with recombinant adenovirus expressing the full-length Survivin protein in the hope that at least some of the endogenously processed peptides from Survivin could be displayed in immunogenic forms on HLA-A2, the most common of the HLA class I antigens expressed in Caucasian populations. To avoid any potential pro-oncogenic side effects, a dominant-negative mutant of Survivin [containing a single amino acid substitution (T34A)] was used instead of the wild-type protein for in vitro immunizations (8).

Elegant studies of the mechanisms of antigen processing and presentation have revealed that immunogenic peptides of 8–9 amino acids in length are displayed on cell surface HLA class I molecules after degradation of intracellular proteins via specialized proteasomes, and transport of the resultant peptides to the endoplasmic reticulum where they associate with class I antigens and then are transported to the plasma membrane (Ref. 12; Fig. 1). Thus, the 142-amino-acid human survivin protein could theoretically generate 136 nonamer peptides. Rather than testing all those possible peptide sequences to see whether they are reflected in the repertoire of T-cell responses made to full-length Survivin, the Gabrilovich group took some educated guesses. Using data from the literature (13, 14) and predictive computer algorithms, they surmised that three peptide sequences contained within Survivin ought to be generated in antigen-presenting cells and should be capable of binding HLA-A2. The strategy thus involved in vitro stimulation of autologous PBLs with Survivin-adenovirus-infected DCs, followed by testing of the resulting activated T cells for their ability to recognize HLA-A2-expressing cells pulsed with one of the three synthetic Survivin 9-mer peptides. The question asked was whether T cells could be enticed, at least in vitro, to respond to a self-antigen (Survivin), thus breaking tolerance. Secondary questions involved addressing the issues of whether such Survivin-specific T cells could kill either peptide-pulsed target cells or a tumor cell line that endogenously expresses Survivin and HLA-A2, and whether normal hematopoietic stem cells would be affected inasmuch as they are among the nontransformed types of cells known to express at least low levels of Survivin in vivo (15).

PBLs derived from all four normal subjects and the four cancer patients resulted in T-cell populations capable of responding to at least one of the three synthetic Survivin peptides, after three rounds of successive stimulation with Survivin-adenovirus-infected DCs. Moreover, seven of the eight PBL specimens responded to two of the three Survivin peptides. Thus, these findings argue that, at least in vitro, it is possible to break tolerance and encourage T cells within the available repertoire to respond to Survivin peptides in the context of HLA-A2. Although an encouraging observation, these findings only begin to scratch the surface and many questions remain to be addressed toward the goal of establishing whether Survivin-based tumor vaccines are likely to be safe and effective in cancer patients.

First, regarding efficacy, from the information provided, it is unclear how efficiently Survivin-transduced DCs activated T cells within the available repertoire of PBLs of normal subjects and prostate cancer patients. One would like to know some of the details about the frequency of responding T cells in peripheral blood and whether that frequency changes in cancer patients, to make comparisons of the merits of Survivin relative to other TAAs that might have been used as an alternative (e.g., prostate-specific antigen; prostate-specific membrane antigen for prostate cancer or telomerase for many types of cancer). According to the data shown, when the Survivin-stimulated T-cell lines were tested on peptide-pulsed DCs, using the frequency of lymphokine-producing cells as a measure of response (“ELISPOT assay”), five of six HLA-A2-positive healthy donors and all four HLA-A2-positive cancer patients showed only an approximate doubling in the number of Survivin-responsive T cells, compared with background controls. Similarly, cytolytic killing of peptide-pulsed cells by the stimulated T cells was relatively feeble, with only a ∼10% increase in specific lysis, sometimes requiring high E:T ratios of 100:1. Of course, one of the limitations of these experiments is that the peptides chosen to measure responsiveness to Survivin are only guesses of what the T-cell repertoire might produce. Of the 136 theoretically possible 9′mer peptides represented in the human Survivin protein sequence, only 3 were tested. On the basis of the SYFPEITHI algorithm, which assesses peptide sequences for HLA binding,³ probably at least five peptides in the Survivin sequence are candidate HLA-A2-binders. Thus, the actual frequency of Survivin-responsive T cells in the normally HLA heterozygous human population might be significantly higher than estimated from best-guess approaches, speaking to the need for a more systematic survey of the possible “peptide space” in which the use of Survivin as a TAA is concerned. It should be emphasized that only T-cell responses directed toward antigens displayed on HLA-A2 were examined, and, therefore, the contributions of other HLA-A allele products and members of the other HLA families (e.g., HLA-B and HLA-C) remain to be defined.

Indeed, an issue to consider in any tumor vaccine strategy that relies on cell-mediated immunity is the question of whether the target tumor cells are competent to process and express antigen on class I antigens. In this regard, mechanisms of tumor resistance to CTLs can include reduced expression or activity of the proteins responsible for digesting whole tumor-associated proteins into small peptide antigens and transporting them into the lumen of the endoplasmic reticulum, as well as loss of expression of the HLA antigens required for presenting TAAs to T cells (reviewed in Refs. 16 and 17). The study by Pisarev et al. used, in part, PBLs from prostate cancer patients. Advanced prostate cancers are known for their loss of expression of HLA-A antigens, whereas HLA-B and HLA-C seem to be more commonly preserved (18). Thus, it might be necessary to exploit a combination of HLA antigens for presentation of survivin-derived peptides to ensure adequate T-cell stimulation. An ideal TAA for vaccines, therefore, ought to be capable of generating several endogenous peptides that can be effectively presented in the context of various different HLA antigens. Thus far, only responses in the context of HLA-A2 have been explored. Thus, although it is unlikely that prostate cancers or other types of tumors will completely lose expression of all types of HLA-class I antigens because of ensuring natural killer (NK) cell-mediated attack that would result (19), a successful tumor vaccine might require that more than one type of HLA antigen be capable of displaying immunogenic peptides that break tolerance to self.

Also at issue with respect to efficacy is the question of whether the Survivin-stimulated T cells can respond robustly to tumor cells expressing endogenous Survivin (as opposed to peptide-pulsed cells). Thus far, the Gabrilovich group has examined only one human tumor cell line that intrinsically expresses elevated levels of Survivin and that displays HLA-A2 (e.g., MCF7 breast cancer cell line), using cell competition experiments to indirectly infer a response to endogenous Survivin in the context of HLA-A2. To definitively demonstrate a response to endogenous Survivin, one would need to knock-down Survivin expression using small interfering RNA (siRNA), antisense, or some other method, showing that T cells fail to lyse tumor cells when the putative TAA is no longer made. Alternatively, specifically neutralizing HLA class I antigens with an antibody might at least indicate whether the lysis observed is HLA-restricted, as opposed to occurring through the non-class-I-restricted mechanisms typically seen with lymphokine-activated killer and NK cells that are likely to be present in the same cell preparations. Finally, it would be interesting to contrast the lytic activity against tumor cells expressing endogenous Survivin using T cells derived from cancer patients, as opposed to the normal donors used, asking whether cancer patients can mount a response, and also using autologous tumor cell lines or at least tumor cell lines derived from the same tissue of origin (in this case, prostate) as the patients’ cancer.

Second, regarding safety, most likely, Survivin is expressed in all types of normal cells during division, but only transiently and only at low levels, compared with tumor cells. Although Pisarev et al. have attempted to address the issue of T-cell reactivity toward normal hematopoietic CD34+ stem cells, finding essentially none, this initial foray into the subject does not exclude the possibility of autoimmune reactions against other types of Survivin-expressing cells such as those seen in the gut, skin, and other tissues in which cell replication occurs. Even activated lymphocytes express endogenous Survivin when stimulated to proliferate (20), raising the possibility of inducing populations of Survivin-responsive T cells that attack themselves in an autoantigen response that could be detrimental to immune-defense and counterproductive to mounting an effective antitumor response. Clearly, extending the preclinical studies to animal models is necessary to gain a better understanding of the relative risks of efforts to exploit Survivin as a TAA.

Finally, it is interesting to contemplate anticancer strategies targeting Survivin with respect to its dual role as a possible TAA versus its function as an antiapoptotic protein. Although most efforts to develop tumor vaccines focus on breaking tolerance and triggering tumor recognition by T cells, scant attention is generally given to the question of whether the responding T cells can effectively kill the tumor target cells once recognized. The cytolytic mechanism induced by T cells involves activation of specific apoptosis pathways, involving caspase-family cell death proteases (21). Survivin is a member of the inhibitors of apoptosis protein (IAP)-family of apoptosis suppressors, a group of antiapoptotic proteins known for their ability to directly bind and suppress caspases (reviewed in Ref. (22). Although Survivin appears to block caspase activation through a distinct mechanism relative to better-studied IAPs (6), the possibility that Survivin-responsive T cells will be unable to overcome roadblocks to apoptosis in tumor target cells containing high levels of Survivin remains a distinct possibility.

In this regard, apoptosis of mammalian cells can be induced via at least three pathways: (a) a mitochondria-dependent pathway, wherein cytochrome c is released from these organelles into the cytosol, in which it binds the protease-activator Apaf1, thus leading to activation of caspase-9; (b) a death receptor pathway involving members of the tumor necrosis factor/Fas-family, where ligand-bound receptors recruit adapter proteins that bind and activate caspase-8; and (c) a pathway exclusive to CTLs and NK cells in which granzyme B is injected into target cells, with granzyme B then cleaving and activating several members of the caspase family and some of their cellular substrates (reviewed in Ref. 23; Fig. 2). Recent observations regarding the antiapoptotic mechanism of overexpressed Survivin suggest that it binds to pro-caspase-9 in association with a cofactor, the HBXIP protein, and blocks cytochrome c-mediated activation of this cell death protease (6). Thus, given that CTLs possess weapons for activating caspase-9-independent apoptosis pathways, particularly Fas-ligand and granzyme B, Survivin ought not to be a major barrier to CTL-mediated apoptosis. However, cross-talk among apoptosis pathways can complicate matters, and abundant examples exist in the literature in which the killing of tumor cells by Fas or granzyme B was found to be dependent on the mitochondria-dependent pathway that activates caspase-9 (see, for example, Refs. 24 and 25). Consequently, even if T cells can be provoked to respond to tumor cells that express Survivin, there is no guarantee they will be able to kill them.

This problem of effective killing of tumor targets by tumor-specific T cells is not unique to Survivin and may apply to a variety of situations in which malignant cells overexpress antiapoptotic proteins that squelch the cell death mechanisms that immune cells rely on for eradicating their targets. Not until both the afferent (tumor recognition by T cells) and efferent (tumor killing by T cells) limbs of the immune response are mastered (Fig. 3), can we expect to see a major leap forward in our ability to invoke the immune system as an ally in the treatment of established cancers. But, as an ancient Chinese philosopher once said, “The journey of a thousand miles begins with a single step.” We are encouraged to see the first steps taken toward a possible tumor vaccine targeting Survivin.