Search

Tumor stem cells are quiescent and, therefore, resistant to therapy, yet harbor the capacity to replenish a tumor after therapy. Therefore, it is tempting to explain all therapeutic failures by the persistence of tumor stem cells. Yet, this explanation is relevant only to initial stages of stem-cell-dependent tumors (such as chronic myeloid leukemia) that, actually, are well controlled by therapy. In advanced cancers that poorly respond to therapy, quiescent tumor stem cells play a negligible role. Instead, proliferating cells determine disease progression, prognosis, therapeutic failures, and resistance to therapy. And therapy fails not because it eliminates only proliferating tumor cells, but because it does not eliminate them. With noticeable exceptions, it is the proliferating cell that should be targeted, whereas resting cancer cells including stem and dormant cells need to be targeted only when they 'wake up'. Finally, I discuss a strategy of selectively killing dominant proliferating clones, including proliferating stem-like and drug-resistant cancer cells, while sparing normal cells.

Geldanamycin (GA), herbimycin A and radicicol bind heat-shock protein-90 (Hsp90) and destabilize its client proteins including v-Src, Bcr-Abl, Raf-1, ErbB2, some growth factor receptors and steroid receptors. Thus, Hsp90-active agents induce ubiquitination and proteasomal degradation of numerous oncoproteins. Depending on the cellular context, HSP90-active agents cause growth arrest, differentiation and apoptosis, or can prevent apoptosis. HSP-active agents are undergoing clinical trials. Like targets of most chemotherapeutics, Hsp90 is not a cancer-specific protein. By attacking a nonspecific target, HSP-90-active compounds still may preferentially kill certain tumor cells. How can this be achieved? How can therapeutic potentials be exploited? This article starts the discussion.

Seemingly disappointing, the Bcr-Abl kinase inhibitor STI-571 shares an 'unfortunate' characteristic with conventional cancer drugs: the development of drug resistance. I argue that the resistance must develop even faster to STI-571 than to conventional drugs, because STI-571 is so effective. This is predictable, but is it inevitable? And how do mechanisms of resistance in relapse depend on a degree of remission. In addition to mutation rate and number of tumor cells, one additional factor determines relapse vs. 'extinction' of the leukemia cell population.

The Bcr-Abl fusion protein drives leukemogenesis and can render leukemia cells resistant to conventional chemotherapy. Geldanamycin (GA), a drug which destabilizes Hsp90-associated proteins, depletes cells of Bcr-Abl, an Hsp90 client, but not of Abl. Both HL60 cells transfected with Bcr-Abl and naturally Ph1-positive K562 leukemia cells are resistant to most cytotoxic drugs, but were found to be sensitive to GA. Furthermore, GA sensitized Bcr-Abl-expressing cells to doxorubicin (DOX) and paclitaxel (PTX). In contrast, in parental HL60 cells, 90 nM GA inhibited PARP cleavage, nuclear fragmentation, and cell death caused by 500 ng/ml DOX. Like GA, STI 571 (an inhibitor of the Abl kinase) sensitized Bcr-Abl-expressing cells to DOX. Unlike GA, STI 571 did not antagonize the cytotoxic effects of DOX in parental HL60 cells. These results indicate that sensitization of Bcr-Abl-expressing cells, but not desensitization of HL60 cells, depends on inhibition of Bcr-Abl. Thus, GA differentially affects leukemia cells depending on their Bcr-Abl expression and selectively increases apoptosis in Bcr-Abl-expressing cells.

Many chemotherapeutic agents induce apoptosis in tumor cells, but killing of normal cells remains a major obstacle. Development of multidrug resistance further limits chemotherapy in cancer. Here, I show that multidrug resistance can be exploited for selective killing of multidrug-resistant cells by a combination of an apoptosis-inducing agent that is not a substrate of either Pgp or MRP (e.g. flavopiridol) with a caspase inhibitor that is a substrate (e.g. Z-DEVD-fmk). In normal cells, treatment with caspase inhibitors prevented PARP cleavage, nuclear fragmentation, and cell death caused by flavopiridol or epothilone B. In contrast, Pgp- and MRP-expressing cells were not rescued by caspase inhibitors. Furthermore, reversal of drug resistance renders Pgp cells sensitive to caspase inhibitors abolishing therapeutic advantage. Thus, caspase inhibitors, that are inactive in multidrug-resistant cells, protect normal but not multidrug-resistant cells against chemotherapy, permitting selective eradication of multidrug-resistant cells. Clinical application of this approach may diminish the toxic side-effects of chemotherapy in patients with multidrug-resistant tumors.

Recent evidence indicates that anti-apoptotic functions of BcI-2 can be regulated by its phosphorylation. According to the 'mitotic arrest-induced' model, multi-site phosphorylation of the BcI-2 loop domain is followed by cell death. In contrast, in cytokine-dependent cell lines, cytokines mediate phosphorylation of BcI-2 on S70, preventing apoptosis. As discussed in this review, these models are not mutually exclusive but reflect different cellular contexts. During mitotic arrest, signal transduction is unique and is fundamentally different from classical mitogenic signaling, since the nucleus membrane is dissolved, gene expression is reduced, and numerous kinases and regulatory proteins are hyperphosphorylated. Hyperphosphorylation of BcI-2 mediated by paclitaxel and other microtubule-active drugs is strictly dependent on targeting microtubules that in turn cause mitotic arrest. In addition to serine-70 (S70), microtubule-active agents promote phosphorylation of S87 and threonine-69 (T69), inactivating BcI-2. A major obstacle for identification of the mitotic BcI-2 kinase(s) is that inhibition of putative kinase(s) by any means (dominant-negative mutants, antisense oligonucleotides, pharmacological agents) may arrest cycle, preventing mitosis and BcI-2 phosphorylation. The role of BcI-2 phosphorylation in cell death is discussed.

Though the term apoptosis was originated in pathology and developmental biology as an alternative to necrosis, the tissue necrosis with inflammation is irrelevant to cell culture conditions where apoptosis is mostly studied. Furthermore, no one single morphological feature is either necessary or sufficient to define apoptosis. The emerging biochemical definition, a cell death with caspase activation, allows the distinction of alternative forms of cell death. Thus, inhibition of caspases delays but does not prevent cell death. Slow cell death without caspase activation may nevertheless be associated with DNA fragmentation. Oncogenic Ras, Raf, and mitogen-activated kinases inhibit apoptosis by affecting the cytochrome C/caspase-9 pathway but may arrest growth and cause slow cell death with delayed DNA fragmentation. Such 'slow' cell death without caspase activation is often caused by chemotherapeutic drugs. Whether a cell will undergo apoptosis or slow death depends not only on a chemotherapeutic agent but also on the readiness of cellular caspases. Therefore, one can distinguish apoptosis-prone (eg leukemia) vs apoptosis-resistant cells. Cell susceptibilities to spontaneous, starvation-induced and drug-induced apoptosis are correlated and characterize an apoptosis-prone phenotype. Finally, distinction of slow cell death allows rephrasing of a question regarding the goal of cancer therapy: apoptosis vs slow cell death, or cancer cell-selectivity regardless of the mode of cell death.

Inhibitors of proteases are currently emerging as a potential anti-cancer modality. Nonselective protease inhibitors are cytotoxic to leukemia and cancer cell lines and we found that this cytotoxicity is correlated with their potency as inhibitors of the proteasome but not as inhibitors of calpain and cathepsin. Highly selective inhibitors of the proteasome were more cytotoxic and fast-acting than less selective inhibitors (PS341>>ALLN>>ALLM). Induction of wt p53 correlated with inhibition of the proteasome and antiproliferative effect in MCF7, a breast cancer cell line, which was resistant to apoptosis caused by proteasome inhibitors. In contrast, inhibitors of the proteasome induced apoptosis in four leukemia cell lines lacking wt p53. The order of sensitivity of leukemia cells was: Jurkat>HL60> or =U937>>K562. The highly selective proteasome inhibitor PS-341 induced cell death with an IC50 as low as 5 nM in apoptosis-prone leukemia cells. Cell death was preceded by p21WAF1/CIP1 accumulation, an alternative marker of proteasome inhibition, and by cleavage of PARP and Rb proteins and nuclear fragmentation. Inhibition of caspases abrogated PARP cleavage and nuclear fragmentation and delayed, but did not completely prevent cell death caused by PS-341. Reintroduction of wt p53 into p53-null PC3 prostate carcinoma cells did not increase their sensitivity to proteasome inhibitors. Likewise, comparison of parental and p21-deficient cells demonstrated that p21WAF1/CIP1 was dispensable for proteasome inhibitor-induced cytotoxicity. We conclude that accumulation of wt p53 and induction of apoptosis are independent markers of proteasome inhibition.

Drug resistance is a well recognized problem in cancer therapy. Despite the current dogma that drug resistance is always an obstacle for treatment, here I show that it provides opportunities for selective protection of non-resistant cells with killing of drug-resistant cancer cells. According to the proposed 'two-drug' strategy, the first drug should be ineffective against a target drug-resistant cell (ie the drug is a substrate of MRP or Pgp pumps). In addition, it must be cytostatic but not cytotoxic. The second drug, which is applied in sequence, must be a cycle-dependent apoptotic drug to which the target cell is not cross-resistant. Thus, low doses of adriamycin, etoposide and actinomycin D, used as the first drugs, were cytostatic to parental HL60 cells. Therefore, these drugs precluded Bcl-2/Raf-1 phosphorylation, PARP cleavage and cell death which are otherwise induced by paclitaxel, a mitosis-selective apoptotic drug for HL60 cells. In contrast, HL60/ADR cells which express MRP, a transporter which pumps out the first drugs from a cell, were insensitive to the first drugs and therefore readily underwent apoptosis following the second drug. This strategy also allowed a selective killing of HL60/TX cells which express MDR-1, with the only difference being that the second drug, paclitaxel, was substituted for epothilones, non-Pgp substrates. Lack of protection by the first drug, a Pgp substrate, resulted in HL60/TX killing by the second drug, whereas parental HL-60 cells were fully protected. Therefore, drug resistant cells can be selectively killed by a combination of drugs not killing sensitive cells. Lack of toxicity against normal cells will be clinically translated in reduction of adverse side-effects of chemotherapy against drug-resistant malignancies.

Raf-1 activation and Bcl-2 hyperphosphorylation following treatment with paclitaxel (Taxol) or other microtubule-active drugs is associated with mitotic arrest. Here we show that microtubule-active drugs do not activate the mitogen-activated protein kinase (MAPK) pathway in leukemia cells. PD98059, a MEK inhibitor, and SB202190, a p38 MAP kinase inhibitor, do not abrogate Bcl-2 phosphorylation nor apoptosis. Simultaneously with PARP cleavage, paclitaxel induces cleavage of Bcl-2 protein yielding a potentially pro-apoptotic 22 kDa product. In comparison, the stimulation of Raf-1 by phorbol ester (TPA) activates the MAPK pathway, causes MAPK-dependent p21WAF1/CIP1 induction, Rb dephosphorylation and growth arrest without Bcl-2 phosphorylation or apoptosis. Like TPA, cAMP induces p21WAF1/CIP1 but does not cause Bcl-2 phosphorylation. MEKK1 and Ras, upstream activators of JNK and ERK MAPK, also fail to induce Bcl-2 hyperphosphorylation. Although Lck tyrosine kinase has been recently implicated in Raf-1 activation during mitotic arrest, microtubule-active drugs induce Raf-1/Bcl-2 hyperphosphorylation and apoptosis in a Lck-deficient Jurkat cells. Therefore, microtubule-active drugs induce apoptosis which is associated with Raf-1 and Bcl-2 phosphorylation and Bcl-2 cleavage but is independent of the MAPK pathway. In contrast, TPA-activated MAPK pathway causes p21WAF1/CIP1-dependent growth arrest without apoptosis.