International Hermelin Brain Tumor Center Symposium on Apoptosis

Henry Ford Hospital, Detroit, MI,USA 4th and 5th April 2003

Detroit, unfortunately known for its inner-city necrosis stemming from social turmoil in the late 1960s, is now undergoing a dramatic renaissance and formed the backdrop to the first in a new symposium series organized by the Hermelin Brain Tumor Center. The aim of these meetings is to explore the translational interface of new and promising disciplines with an eye on how they could be brought to bear on the daunting problem of nervous system malignancies sooner rather than later. Apoptosis was chosen for the inaugural meeting as few other research areas have so revolutionized our understanding of basic biology and hold out such strong hope for therapeutic applications. A public session of talks was held on April 4th at the main campus of Henry Ford Hospital, and, after a brief introduction on the role of apoptosis in glioma by Oliver Bögler¹ of the Hermelin Brain Tumor Center, provided a stimulating view of new molecules that need to be considered, such as p73, hsp70 and the calcium-regulated proteases, as well as new regulatory relationships between established players such as p53, BAX and BAK. Particular attention was paid to Apo2L/TRAIL, which because of its favorable profile of tumor cell cytotoxicity, and relative sparing of normal cells, is a particularly promising candidate for clinical development. The molecules most intensively discussed are shown in Figure 1 in a simplified schematic of currently known apoptotic pathways. This was followed by a colloquium on April, 5th which provided the opportunity for more extensive discussion leavened by unpublished results and highlighting issues ranging from the current challenges in the treatment of brain tumors, including a comparison of European and US American views and strategies, the practicalities of studying new approaches in clinical trials for these malignancies and what the potential of some of the powerful molecular analyses of cancer is in bringing new approaches in the coming months and years.

Figure 1

A simplified schematic of apoptosis pathways with emphasis on the molecules that were discussed at this meeting. Apoptosis can be triggered by activation of a death receptor at the start of the extrinsic pathway, or by stimuli that activate the p53 or p73 proteins at the start of the intrinsic pathway. Mitochondrial amplification of extrinsic signals is shown by the cleavage of tbid by caspase-8. Molecules in color are discussed in this report. For details see text

Apoptosis can be said, like many facets of cancer, to start with p53. This is not meant to imply that p53 plays a mechanistic role in every cell death event, but rather that the loss of its function in malignant cells profoundly alters their propensity to undergo apoptosis. This has been elegantly demonstrated in mouse models, such as the one for choroid plexus carcinoma, where the major consequence of p53 loss early in the evolution of tumors is the reduced sensitivity to spontaneous apoptosis.² Erwin Van Meir from the Department of Neurosurgery at Emory University, Atlanta, presented an interesting study investigating whether changes in apoptotic regulation were also an important element of p53-loss-driven gliomagenesis in humans. He presented a glioma patient with a germline p53 mutation, R283 H, in which the sequential inactivation of p53 had been examined in detail.³ While most commonly a point mutation in p53 is followed by loss of heterozygosity (LOH) eliminating the remaining wild-type allele, no LOH was detected in this patient. Instead, two mutant alleles of p53 were identified in the recurrent tumor: the one present in the germ line, which had also acquired a second mutation at R267W, and the other allele, which had suffered a different mutation, E258D. p53 transcriptional assays in yeast, used to trace the evolution of the tumor, surprisingly showed that the acquisition of the second mutation in the allele carrying R283 H preceded the mutation in the remaining wild-type allele. Once all three mutations were present, this led to rapid clonal dominance of p53^{R283 H R267W, E258D} cells. The need for a second point mutation in the R283 H allele was explained by the observation that the R283 H mutant retained its ability to activate the CDKN1A promoter and so inhibit the cell cycle. Interestingly, R283 H was significantly compromised in its induction of the bax promoter, suggesting an early and important contribution in a reduction in apoptosis to the survival of cells that may be proliferating abnormally.

Of course, p53 no longer stands alone, and its cousins may have an important hand in regulating cell survival. Gerry Melino⁴ from Tor Vegata University, Rome, and the Medical Research Council, Leicester, spoke about the expanded p53 gene family, focusing on p73.⁴ Phylogenetically, p53 is considered to be the latest addition to the gene family explaining its primary involvement in tumor suppression while the oldest, p63, is largely involved in development. The middle child, p73, appears to have a hand in both arenas. Thus DNA damage also induces p73 activity, which responds preferentially to ionizing radiation and cisplatin, and this occurs via MLH1 and ATM proteins and involves phosphorylation by c-Abl. Like p53, the stability of p73 is tightly regulated by ubiquitin-mediated degradation involving cullin proteins and is stabilized by PML with which it associates. Unlike p53, p73 is exclusively nuclear.

So how does p73 induce apoptosis? One mechanism is via the induction of the Scotin protein which mediates ER stress and affects calcium metabolism, recently recognized as an important regulatory point in apoptosis.⁵ Then p73 induces BAX to a similar degree as p53 at the protein level, and also upregulates PUMA which induces a conformational change in BAX, causing it to relocate to the mitochondrial membrane. In addition, p73 can induce the expression of the death receptor, CD95, albeit to a lesser extent than p53, but to levels sufficient to sensitize cells to CD95-mediated apoptosis.

Like the third member of this small group, p63, p73 has an extended C-terminus with regulatory regions, and exists in two forms that differ in the N-terminus: a TA form capable of transactivating promoters, and a ‘Δ’N isoform that lacks the transactivation domain and acts as a dominant-negative variant. The ΔNp73 isoform is activated by both p73 and p53 and is capable of inhibiting the effects of both of these proteins, and so forms a negative feedback loop on their activity as well as inhibiting their effects including apoptosis. For example, ΔNp73 inhibits Bax, PUMA and CD95 transcription in response to TAp73. That this can have significant clinical repercussions is suggested by the observation that patients with neuroblastoma whose tumors express ΔNp73 have a poorer prognosis.⁶

The strong connection between p53- and p73-mediated apoptosis and induction of the BAX protein is not reciprocated in that BAX, and its fellow proapoptotic member of the bcl-2 family, BAK, can suppress tumors in a p53-independent manner. This was discussed by Eileen White from the Howard Hughes Medical Institute and the Rutgers Center for Advanced Biotechnology and Medicine. The central question was whether inhibition of BAX and BAK is sufficient to inhibit apoptosis in oncogenesis, and the intellectual starting point was that of learning from viruses, adenovirus in particular.⁷ The infecting adenovirus attempts to trigger cell division and prevent apoptosis in order to cause its host to propagate for the purpose of creating more virus, and does this using its viral E1A and E1B proteins. Primarily, this is achieved by E1A targeting RB, E1B55 K targeting p53 and E1B19 K targeting BAX and BAK. If BAX and BAK were acting predominantly in a p53-dependent manner, then the two E1B proteins would be largely redundant, and this would represent an anomaly in the otherwise minimalist design typical of viruses. Direct experimental support for the p53-independent role of these apoptosis regulators came from experiments using bax^−/−, bak^−/− and bax^−/−bak^−/− cell lines. The loss of both genes caused a broad-ranging defect in apoptosis.⁸ In addition, it was sufficient to cause BMK cells to become tumorigenic and form highly invasive carcinomas. Although the double knockout cells were the most tumorigenic, loss of either gene promoted tumor formation, and if hetereozygous cells were implanted there was sufficient selective pressure to allow the formation of tumors that had lost the remaining wild-type allele. The central importance of BAX and BAK for tumor formation was underscored by the observation that acceleration of tumor formation by E1B19 K or BCL-2 was no longer observed in bax^−/−bak^−/− cells.

Members of the bcl-2 family of proteins are also involved in survival signaling, and the presentation by Frank Furnari from the Ludwig Institute for Cancer Research, San Diego Branch, addressed this aspect of cell death research. He discussed in detail the role of EGFR signaling in glioma survival. Overexpression, amplification and ligand-independent mutations, such as ΔEGFR, are common in gliomas, and have been recognized to enhance in vivo tumorigenesis. ΔEGFR has lower, but constitutive rates of phosphorylation, and, although the tyrosines that are phosphorylated on its C-terminus appear to be similar to those in wild-type EGFR, may be sending some qualitatively different signals. One clue to this is that overexpression of ΔEGFR confers a reduced apoptotic rate to U87MG glioma cells in vitro and in vivo while wild-type EGFR does not. Similarly, ΔEGFR, but not EGFR, confers resistance to apoptosis in response to treatment with cisplatin, and this is likely related to its ability to dampen the cisplatin-induced suppression of BCL-X_L expression, which EGFR does not do. Using tyrphostin concentrations that inhibit ΔEGFR but do not affect active EGFR, it was possible to show that synergistic effects with cisplatin were achieved both at the level of apoptosis rates of cultured cells and tumor size in vivo. The detailed analysis of this cancer-specific protein illustrates that targeting of a survival signal can effectively synergize with conventional therapy.

Not all modes of cell death involve caspases, or even the mitochondria, as was highlighted by Marja Jäättelä of the Danish Cancer Society, Kopenhagen. Death receptors and DNA-damaging agents, which are more commonly associated with conventional apoptosis, can activate lysosome- and endoplasmic reticulum-mediated cell death involving cathepsins and calpains, which is called apoptosis-like programmed cell death.⁹ Cathepsins appear to mediate this cell death by cleaving specific substrates, such as the survival kinase Akt, and can do so at neutral, physiologically relevant, pH. In the case of cathepsins, an intriguing connection to hsp70, which is overexpressed in transformed cells, and to the abnormal distribution of lysosomes in cancer can be made. Electron microscopy revealed that hsp70 is associated with the lysosomes in cancer cells, which are also present at the plasma membrane, and is probably chaperoning cathepsins there, as reduction in hsp70 levels by antisense or siRNA causes cathepsin-mediated apoptosis-like programmed cell death in a variety of malignant cells. Anti-hsp70 therapies in animals were able to prolong symptom-free survival in glioblastoma-bearing mice, highlighting a new tumor-associated target for therapy development.

An important consideration for the translational potential of any laboratory finding is its applicability to the real world. One element of this is the heterogeneity and complexity of cancers as they are encountered in the clinic, which contrasts with the carefully controlled and often homogeneous entities that are studied in the laboratory. This issue was directly addressed by David Ashley of the Royal Children's Hospital, Melbourne, who discussed some findings on the heterogeneity of brain tumors in the context of apoptosis. As an initial step in this direction, he presented data on a panel of glioma cell lines that had been tested for their sensitivity to CD95L, Apo2L/TRAIL and cisplatin.¹⁰ Mechanisms of resistance to CD95L and Apo2L/TRAIL differed, including, for example, low levels of FADD and caspase-8, or high levels of BCL-2 or BCL-X_L. Low caspase-8 expression was of particular interest as transfection of this gene could re-establish sensitivity. Some of the glioma cell lines were classified as being primarily type I in that they relied on the direct activation of caspase-3 from death receptors, or type II in that they availed themselves of the mitochondrial amplification loop via proapoptotic BCL-2 family proteins. Clearly, further characterization of how gliomas overcome apoptosis and an extension of this work to primary tumors will be invaluable to estimating the translational potential of any therapies that come from this field, and may play an important role in prescreening patients before apoptosis-based therapies are applied.

These days, only the most optimistic would propose that a new type of therapy is likely to be sufficient to control cancers, and many speakers proposed combination with existing modalities of treatment. This theme was particularly highlighted by two presentations. In the first of these Simone Fulda of the University of Ulm, Department of Pediatrics, presented data that showed that conventional therapies, such as gamma-irradiation and doxorubicin engaged the apoptotic machinery in that they elevated CD95 and its ligand, CD95L and caused formation of the death-inducing signaling complex (DISC). This suggests that direct modulation of the cell death pathways themselves could make conventional therapies more effective, a point that she illustrated convincingly by showing the potency of a novel approach of sensitizing cancer cells to apoptosis based on the functional neutralization of members of the inhibitor of apoptosis protein (IAP) family. This was achieved by using agonists of the proapoptotic mitochondrial protein, SMAC/DIABLO.¹¹ SMAC is released from the mitochondria along with cytochrome c, and binds to IAPs and releases them from caspases, which can then become activated. Using the HIV Tat leader sequence, SMAC peptides were introduced into cells and sensitized them to both Apo2L/TRAIL and doxorubicin treatment. A related theme, that new drugs are now commonly assessed in terms of their potential to modulate apoptosis, was highlighted by data on the discovery and preclinical assessment of betulinic acid, a compound derived from the birch tree, which exhibits promising in vitro and in vivo activity against various types of cancer cells and appears to act on the mitochondria.¹²

Wolfgang Wick of the University of Tübingen Department of Neurology also spoke on the synergy between conventional and apoptosis-based therapies. He illustrated how the standard treatment of malignant gliomas, fractionated external beam radiotherapy, might have some hitherto under-recognized biological effects: enhanced antiapoptotic BCL-2 family protein expression, enhanced metalloprotease activity, and enhanced migration and invasion.¹³ Needless to say that such changes would be relevant only if such cells also retained their ability to proliferate after irradiation, but there was good evidence that they do. Reassuringly, the motile phenotype of glioma cells afforded by sublethal doses of irradiation was nullified by coexposure to the novel alkylating agent, temozolomide,¹⁴ providing an a priori explanation for a possible positive result of EORTC trial 26981, which examines radiotherapy plus temozolomide compared with radiotherapy for newly diagnosed glioblastoma.

Three talks then addressed the possible role of Apo2L/TRAIL in the future management of human cancers, particularly gliomas, as this molecule currently has significant and near-term promise of impact in the clinic. Avi Ashkenazi¹⁵ of Genentech, San Francisco, reviewed the essentials of this death ligand, focusing on its potent antitumor activity in several in vitro and in vivo paradigms, its possible role in synergistic antitumor regimes involving radiotherapy and chemotherapy, and its favorable toxicity profile. Henning Walczak¹⁶ of the German Cancer Research Center in Heidelberg, Germany, confirmed the strong antitumor activity of Apo2L/TRAIL and examined in depth the role of caspase-10 in the Apo2L/TRAIL-mediated cell death signaling cascade. Both of these presentations emphasized the need to consider caspase-10 in Apo2L/TRAIL-mediated apoptosis rather than just caspase-8, which may be of clinical significance in view of data suggesting that caspase-8 is inactive in some nervous system cancers. Michael Weller¹⁷ of the University of Tübingen, Department of Neurology summarized the current knowledge of pathways of resistance to Apo2L/TRAIL-induced apoptosis in glioma cells and illustrated how different modifications of the Apo2L/TRAIL molecule, particularly the tails of the TRAIL, may have accounted for controversial observations regarding its activity on transformed versus nontransformed cells. Using amplifiers such as agonistic SMAC peptides, experimental gliomas can be safely eradicated by recombinant Apo2L/TRAIL in the absence of neurotoxicity,¹¹ still making Apo2L/TRAIL one of the hottest candidates for a major contribution of apoptosis research to clinical oncology in the future, with glioblastoma being a prime target.