The Cell membrane and cell signals: New targets for novel anticancer drugs

Annals of Oncology 1: 100-111, 1990. © 1990 Kluwer Academic Publishers. Printed in the Netherlands.

Special article The cell membrane and cell signals: New targets for novel anticancer drugs P. Workman MRC Clinical Oncology Unit, Medical Research Council Centre, Cambridge and EORTC Pharmacokinetics and Metabolism (PAM) Group, UK

Key words: cell membranes, signal transduction, new targets, novel drugs

Introduction

Why cell membranes and signals?

There is a widespread consensus that we have reached a plateau of effectiveness in the chemotherapy of the major solid tumours. Although significant improvements in therapeutic index may be achieved by better use of existing agents or the development of improved analogues, there is no doubt that a quantum leap in activity is unlikely to arise by these conventional means. Scientists and clinicians closely involved in new drug discovery and development are looking increasingly towards the identification and pharmacological exploitation of novel molecular features of cancer cells which are qualitatively or more likely quantitatively different from critical normal tissues. The search is on for new targets. Where might these be found? Traditional antitumour agents have in most cases been identified by, and certainly selected for activity against, the increasingly discredited, fast-growing, anaplastic rodent tumour screens which are of dubious relevance to the biochemistry of human solid tumours. Almost without exception these drugs exert antineoplastic activity by chemical reaction with genomic DNA, by inhibiting the synthesis of DNA, or by interfering in some other way with the mechanics of DNA replication and cell division. Not surprisingly, dose-limiting toxicity is normally seen in the rapidly proliferating normal tissues. Nevertheless, valuable gains are likely to arise through an improved understanding at the molecular level of the factors which control the interaction of drugs with particular DNA sequences, leading eventually perhaps to the development of drugs targeted to specified key genes [1]. Other approaches attracting current interest include targeting of drugs or toxins to tumour cells using specifically engineered antibodies [2] and the application of "biological" agents such as the cytokines [3]. But in recent years there has been a phenomenal growth in our understanding of the ways in which cells receive and process information, including that relating to proliferation, differentiation, transformation and mitogenesis. This has generated enthusiasm to search for new targets for anticancer drug design amongst the plethora of biochemical elements that control these fundamental processes.

The cell membrane has proved to be the cellular location for many of these key regulatory control elements [4, 5]. Information is received by cells at their exterior surface in the form of hormones, growth factors, neurotransmitters, antigens and so on. Although hydrophobic lipid soluble steroid hormones are an exception in that they can easily penetrate the cell membrane to interact with cytosolic receptors, most information delivery molecules are large and/or hydrophilic (water soluble) in character and cannot therefore traverse the protective, hydrophobic phospholipid bilayer. Their information must be transduced through binding to exterior membrane receptors so as to elicit the appropriate biological response form the wide range of functions available in the cell's repertoire. The translation of the signal from exterior membraneinteractive ligands or "first messengers" to the inside of the cell utilizes an initially limited number of fundamental biochemical effector systems located within the membrane. The initiated signal is then amplified via the triggering of a serfs of more complex metabolic cascades in the membrane, cytosol and nucleus, leading in rum to the ultimate cellular response including the induction of transcription from specific "early" response genes. These cascades involve small effector molecules or "second messengers". The four basic primary signalling systems are: 1) the admission of extracellular ions, especially calcium; 2) the breakdown of membrane-bound inositol lipids generating the currently very fashionable second messengers diacyl glycerol (DAG) and inositol (1, 4, 5) trisphosphate [Ins(l, 4, 5)?$ (6); 3) the formation of cyclic AMP and GMP messengers by the corresponding cyclase enzymes; and 4) the activation of receptor-linked tyrosine kinases. Further complexity is provided by interaction of these different signal cascades with one another - a phenomenon known as "cross-talk". This is facilitated by the action of "integrator" proteins such as phospholipase C and protein kinase C. Also focal to the majority of transmembrane signalling mechanisms are the guanine nucleotide-binding G proteins. In the cancer field, the products of an increasing number

101 of transforming cellular oncogenes are turning out to be growth factors or their membrane receptors (e.g. plateletderived growth factor, PDGF, and epidermal growth factor, EGF); these are either structurally modified or over-expressed, leading to excessive stimulation and subversion of normal regulatory control mechanisms. This excessive or pathological stimulation of cell-surface receptors in turn triggers the inappropriate activation of various signalling pathways and gives rise to early gene transcription, mitogenesis and malignancy [5]. Products of other oncogenes such as myc, fos and jun are themselves nuclear proteins which are transcribed early in the mitogenic response and are involved in the regulation of the cell cycle and proliferation [5, 7, 8]. New oncoproteins will no doubt continue to be found amongst the gene transcription factors and the signal transduction effector proteins, and an example of the latter is the recently identified mutant version of GQS, the GTP-binding subunit of the G protein G s which normally transmits signals from the p-adrenergic receptor to adenyl cyclase; the constitutive activation in the mutant leads to accumulation of cAMP [9]. The increasing appreciation that the biochemistry of cell membranes and cell signal transduction might offer fertile ground for new anticancer drug discovery led the EORTC PAM Group to hold a small one day symposium on "Membrane Targets in Cancer Chemotherapy" in Nice, France in December 1987 [10]. Arising in part from the success of that meeting, together with the continuing explosion in understanding of the mechanisms of signal transduction and the mounting frustration with conventional DNAdirected drugs, a much larger conference (around 220 delegates) was convened in Queens' College, Cambridge, September 14-16, 1989 wthe title "The Cell Membrane and Cell Signals as Targets in Cancer Chemotherapy". Also providing a stimulus to hold the symposium was the growing appreciation that some conventional cancer drugs, including doxorubicin and alkylating agents, do themselves have effects on membranes and signalling, and the belief that the molecular and therapeutic specificity of these effects might be greatly increased by rational design. This first truly international meeting on the subject was sponsored jointly by the American and British Associations for Cancer Research in addition to the EORTC PAM Group. The object was to bring together leading researchers working on membrane structure and function and cell signalling, together with chemists, cancer pharmacologists and oncologists interested in (or perhaps sceptical of?) exploiting this new area of knowledge for anticancer drug development The agenda focussed on cross-fertilisation of ideas between different disciplines, with particular emphasis directed towards assessing progress, defining problems and identifying the most profitable areas for future advances. Abstracts of the 32 invited oral presentations and the 88 proferred posters were published [11] at the time of the meeting and a review based on the chairpersons' summaries will appear elsewhere [12]. The present report represents a somewhat more personal view of the meeting, from my standpoint as one of the conference organizers and more especially as a cancer pharmacologist involved in

European new drug development. As we review the proceedings of the symposium, which was certainly viewed as a success judging from the response of most participants, the following questions might be borne in mind: 1. How much do we know and how much have we yet to learn about the structure and function of membranes and signal transduction? 2. What changes occur in cancer and what are the differences between normal and malignant cells in this respect? 3. Do we know enough to interfere pharmacologically with membranes and cell signalling in cancer cells? 4. How selective will the therapeutic effects be, and what sort of toxicities might we expect? 5. Do we have any lead compounds for drug design? 6. How can these be improved and what are the new and most promising avenues for novel drug development? 7. How should provising drugs be evaluated and developed? 8. When should the drugs enter clinical trial? The sessions were structured in such a way that state of the art basic science overviews of key areas were followed by presentations dealing more directly with anticancer drug development. Both basic researchers and pharmacologists responded to the invitation to speculate on prospects for selective pharmacological intervention, and this was certainly a subject for lively debate in the discussion and poster sessions.

Membranes and signalling targets Opening the scientific proceedings, Michael Berridge (Cambridge University, UK) reviewed membrane-initiated signalling systems with emphasis on phosphoinositol and calcium as second messengers. Rozengurt has categorized steps in mitogenesis as either regulatory or obligatory [5, 13]. The latter are those molecular events which must take place. The former are signals which can be elicited or bypassed by multiple pathways and agonists, and Berridge regarded this redundancy as a problem in the study of proliferation. He saw the unravelling of the complex interactions of the different mitogenic pathways as the key to understanding "how cells grow". In the bifurcating phosphoinositol signalling pathway, occupation of the cell surface receptor by ligand results in a G protein-linked activation of phospholipase C within the cell membrane. Phospholipase C hydrolyses phosphatidulinositol 4, 5-bisphosphate [Ptdlns (4,5)P2] to yield diacuyl glycerol (DAG) and inositol - 1,44 trisphosphate [Ins (1, 4, 5) Pj]. Both are second messengers, the lipid soluble DAG activating protein kinase C in the membrane (14 and see later) and the water soluble Ins (1, 4, 5) P 3 releasing calcium from internal stores [6]. Berridge pointed out that although much has been learned since the initial demonstration of Ins(l, 4, 5)P 3 - mediated calcium release in 1983, the molecular mechanisms were by no means finalised. However, in the short time since the meeting

102 clear evidence has been presented that Ins (1, 4, 5)P3 acts predominantly in the endoplasmic reticulum (ER) and not the calciosome (the latter responds to caffeine at least in chromaffin cells [15]). In addition, the Ins (1,4, 5)P 3 receptor has been cloned and the cDNA sequence predicts a membrane-spanning structure similar to the ryanodine receptor in the sarcoplasmic reticulum [16]. Finally, the purified receptor has been shown by itself to be sufficient to mediate calcium release [17]. These discoveries illustrate the astonishing pace of this field. Berridge went on to discuss a unifying hypothesis to describe the distribution of intracellular calcium signals over space and time, calcium waves and oscillations respectively [6]. A relatively slow and modest entry of extracellular calcium follows the large initial mobilisation of internal stores. Calcium influx into the cytosol may be direct through membrane channels or indirect via the ER. It appears to be regulated by Ins (1, 4, 5)P3 together with the tetrakiphosphate Ins (1, 3, 4, 5)P 4 , which may exert a helper function. This may provide a steady input of calcium which is released at intervals to generate transient spikes. The calcium signal is thereby "digitized", the frequency varying with agonist concentration. Turning to topology, ligand-gated calcium signals may be highly localised within the cell and may propogate in waves, either by calcium-induced Ins (1, 4, 5)P 3 production or, Berridge prefers, a process of calcium-induced calcium release through adjacent Ins(l, 4, 5)P3-primed stores and even into neighbouring cells via gap junctions. He also illustrated how calcium waves can sweep across the nucleus, a possible means of activating intranuclear signalling. A bewildering number of inositol phosphates are found in eukaryotic cells [6]. Ptd Ins (4, 5)P 2 is synthesized by the sequential actions of PtdIns-4-kinase (type II), which converts phosphatidulinositol (Ptdlns) to phosphatidulinositol 4-phosphate [Ptd Ins(4)P], followed by PtdIns(4)P-5kinase. Berridge pointed out how until recently we were "blissfully unaware" of phosphatidulinositol 3-phosphate [PtdIns(3)P2]. This is formed from Ptdlns by Ptdlns 3kinase (type I). Of special interest in oncology is the exciting observation that the product of the c-src oncogene in association with the middle T antigen of SV40 acts to phosphorylate and probably activate PtdIns-3-kinase [18]. Moreover, removal of the tyrosine kinase insert from the PDGF receptor prevents the stimulation of this kinase and blocks proliferation [19]. It is proposed that Ptdlns (3)P may regulate the inositol lipid cascade in a similar fashion to fructose-2,6-biphosphate in glycolysis. Berridge hypothesized further that the 3-isomer may increase the availability of the bulk, hormone-insensitive pool of inositol lipid, and that its promotion may underlie the mitogenic syngergism between insulin and tyrosine kinaselinked ligands like bombesin (see Rozengurt). While key members of the lower inositol lipids are involved in signalling, the higher relatives such as the inositols pentakiphosphate and hexakisphosphate (IP5 and IPg) are agonist-insensitive and probably have a "housekeeping" function. They may help to regulate the structure of the cytoskeleton. Apart from acting as a

precursor for signalling lipids, Ptdlns performs a structural role in the bilayer and also acts to anchor proteins to the cell surface. Speculating on the best targets for therapeutic intervention, Berridge favoured the "multimodal integrator" proteins, like phospholipase C, protein kinase C and various oncoproteins, to be returned to later in the report. Tim Rink (Smith Kline and French, Welwyn, UK) was asked to focus specifically on the future prospects for pharmacological intervention with signal transduction in the development of oncolytics. He pointed out that malignancy may be sustained by an "overdrive" from external and internal positive signals or through an "underdrive" of negative elements. In support, he cited the successful blockade of the oestrogen and androgen drive in breast and prostate cancer respectively. For example, tamoxifen binds to and antagonized the cytoplasmic oestrogen receptor. Rink looked forward to the development of new agents affecting the signalling cascades - he termed this "intracellular endocrinology". We have the option to inhibit the multimodal integrators common to many cancers in search of broad spectrum activity, or to attack targets more specific to individual tumour types for greater selectivity. He considered additional challenging questions to be: are we looking for cytocidal or cytostatic drugs?; do we envisage acute or chronic therapy?; how do we find the required molecular specificity?; do we need sophisticated targeting?; and what model systems do we use for testing? Good questions 1 Although cells use the same basic signal transduction mechanisms for normal functions such as secretion and contraction as in mitogenesis and malignancy, Rink counselled against taking a nihilistic view. He pointed to the plethora of isoforms of various signalling proteins which should give grounds for optimism in drug design, and suggested further that signalling inhibitors might have much greater impact on pathologically overactive pathways. In support of these arguments he cited the encouraging precedent set by the calcium channel antagonists which, against rational expectation, express enormously valuable activity through the structural and functional diversity of voltagegated calcium channels and preferential action in hyperactive tissue. Lithium is a second example. This acts to inhibit enzymes hydrolyzing inositol phosphates, thus reducing the supply of inositol for the signal cascade. Its therapeutic value in manic depression appears to derive from the "damping down" of overactive signal transduction, perhaps because inhibition is uncompetitive giving rise to bigger effects at elevated inositol phosphate levels. Concluding with examples from his own work on calcium. Rink first indicated that there was no clear picture as to whether levels of this ion are consistently higher or lower in cancer cells compared to normal. He speculated nevertheless on the therapeutic role for blockade of receptor-mediated calcium entry, and queried the feasibility of reducing a sustained pathological calcium overdrive while leaving normal acute oscillations unaffected. Finally, he described a new compound SKF 96365 which suppresses calcium influx but not internal release and which

103 exhibits activity against skin oedema and neutrophil accumulation.

Membrane structure, biophysics and response In reviewing the structure of membranes in relation to function, Dennis Chapman (University of London, UK) emphasised the changing perspective resulting from the application of a range of powerful biophysical techniques. Membrane fluidity - the ability of the bilayer to tolerate movement - is extremely important [20]. Membrane proteins must move to function. Chapman stressed our ignorance of the precise molecular nature of, for example, lipid-protein interactions and membrane protein structure and function, even for well studied proteins such as bacteriorhodopsin. Most structures are predicted from sequence data, but a variety of biophysical methods can be applied to obtain valuable information. Tom Tritton (University of Vermont, USA) discussed the structural basis for doxorubicin interaction with membranes so as to change their structure and function [21]. The association of doxorubicin with membranes generally increases fluidity, except where the membrane contains cardiolipin in which case it becomes more solid. Since cancer cell membranes but not normal cell membranes usually contain cardiolipin, this provides a basis for antitumour specificity. Unfortunately, the cardiolipin content of mitochondrial membranes may be responsible for the cardiotoxicity of doxorubicin. So it is clear that the lipid make up of a membrane can control its interaction with drugs. Tritton showed that the anthacycline penetrates deep into the phospholipid bilayer and binds at a "particular ugly" angle of 55°, severely disrupting the ordered packing of the fatty acyl chains. Current studies are revealing structureactivity relationships for membrane interaction among doxorubicin analogues. Ben de Kruiff (University of Utrecht, Netherlands) described his studies using nuclear magnetic resonance spectroscopy (NMR) to demonstrate the specific affinity of the positively charged amphipathic doxorubicin molecule with negatively charged membrane lipids like cardiolipin. Doxorubicin binding by cardiolipin prevents the formation of nonbilayer structures, resulting in the inhibition of mitochondrial enzymes and a decline in energy-rich phosphates in the heart. The drug also interacts strongly with the negatively charged lipid phosphatidic acid [22], almost certainly contributing to effects on signal transduction (see Hickman). The ether lipids are an interesting class of membraneactive drugs which are related in structure to platelet-activating factor (PAF). Allesandro Noseda (Mario Negri Institute, Milan, Italy) showed how ether lipids such as ET18-0-Methyl cause membrane perturbations such as ruffling and blebbing, and the formation of holes of 1.5 urn mean diameter covering up to 13% of the cell surface. Interestingly, lysophosphatidyl choline was reported to induce membrane damage but not holes. Membrane fluidity increases prior to cytotoxicity. Biophysical changes in

membranes were correlated with cytotoxicity for a series of ether lipids. Intriguing new results indicated that ether lipid uptake is slow for model membranes rich in cholesterol, and preloading membranes with cholesterol led to an attenuation of drug effects. Analogous to the situation with cardiolipin content and doxorubicin, this provides a clear basis fo4r selectivity.

Growth factor signalling In case we hadn't noticed, Hal Moses (Vanderbilt University, Nashville, USA) reminded us that "growth factors are everywhere". Over 50 growdi factors are known to stimulate or block proliferation (excluding neuropeptide-like factors) and it was predicted that there are a lot more we have yet to discover. He emphasised that the tight control of cell proliferation is achieved through the action of both positive and negative signals from diffusible peptides. This was illustrated by reference to the transforming growth factors (TGFs) a and P using the skin keratinocyte as a model. TGFa is a 5.6kD peptide mitogen which mediates its effects via the EGF receptor. As well as being produced in tumours, TGFa may have a regulatory role in embryonic tissue and normal adult skin keratinocytes. TGFpi is the prototype of a superfamily of genes regulating growth control, extracellular matrix production and development Expression of TGFp mRNAs varies during embryological progression. TGFPs exhibit a wide range of effects depending on cell type, including growth stimulation, inhibition and differentiation [23]. They have essentially no sequence homology with TGFa. Post-translational activation involves dissociation of non-covalent bonding between the mature peptides and the glycosylated N-terminal peptides of pre-pro-TGFps. Removal of the N-terminal glycopeptide by plasmin activates mammalian TGFpi. The powerful growth inhibition by TGFfJs in most epithelial cells arises via inhibition of the transcription of a few specific genes, for example c-myc but not c-fos or c-jun. This may be brought about by die synthesis of a protein which affects an upstream element of the c-myc gene. Clearly both autocrine stimulation by TGFa or reduced autocrine inhibition by TGFp could contribute to malignancy. Pharmalogical interference with these pathways should take into account the involvement of TGFp in wound healing and also its immunosuppressive properties. Mike Waterfield (Ludwig Institute, London, UK) reviewed recent progress in elucidating the structure - function relationships for signal transduction by EGF and TGF a. The shared receptor for these factors has been crystallized, and the 3D structure should soon be available. Transformation may arise via hypercxpression of normal receptors or the production of a truncated receptor encoded by the oncogene erb B [24]. The intact receptor has three domains - an extracellular ligand-binding domain, a single short transmembrane domain and a cytoplasmic domain. Ligand binding or truncation stimulates the tyrosine kinase activity located in the cytoplasmic domain. This brings about autophosphorylation of four sites on the receptor, as

104 well as tyrosine phosphorylation of extrinsic proteins. The latter may include a specific phospholipase C, although phosphoinositol breakdown is not required for the transformation signal and the later steps of transformation are unknown. In terms of molecular models for activation of the receptor tyrosine kinase, Waterfield felt the evidence was now overwhelming in favour of ligand-induced receptor dimerization and intermolecular phosphorylation as a consequence of apposition of the internal domains. He went on to describe studies designed to test this model and to elaborate potential inhibitors. Techniques have been developed which should produce large amounts (20-100 mg) of native or mutant receptors and various domains thereof in insect cells for biophysical studies. Although the isolated external domain binds EGF and TGFa strongly it fails to oligomerize on ligand binding, but conformational changes are detected. EGF and TGFa can be shown to adopt similar structures when complexed to the receptor. Molecular modelling approaches are underway, akin to those described for the rat neu gene product, which is equivalent to the human crbB-2 and closely related to the EGF receptor. In neu, receptor aggregation or packing is caused by a single amino acid (Glud 664) in the transmembrane a-helix caused by two interhelical hydrogen bonds [25]. Sequence motifs analogous to that required for packing in neu were found in 18 from 20 growth factors receptors. This led to the suggestion of using transmembrane peptide sequences as possible inhibitors to form a non-productive complex [25]. Discussion centred on the merits of antagonizing the unchanged but overexpressed receptor which will also be expressed on non-malignant cells, as against the mutant receptors where greater opportunities for selectivity might be envisaged. The following two presentations were concerned with guanine nucleotide binding regulatory G proteins (David Clapham, Mayo Foundation, Rochester, USA) and the related p21 ras oncoproteins (Chris Marshall, Institute of Cancer Research, London, UK). The G proteins couple in excess of 70 receptors to adenylate cyclase, phospholipase C, cGMP phosphodiesterase or ion channels. The classical G proteins are heterotrimers of a, p and y subunits. In contrast to an earlier doctrine, it is now known that both a and py subunits can regulate cardiac potassium channels [26]. Occupation of the receptor leads to the inhibition of the G protein cycle, involving binding of the GTP to the a subunit and its subsequent self-limiting hydrolysis. Multiple inhibitory and stimulatory G proteins are known, and some are uncoupled from the physiological response by toxincatalysed ADP-ribosylation. Marshall began with a commercial for the forthcoming crystal structure of the ras p21 oncoprotein, since published [27]. This should greatly facilitate the rational design of inhibitors. The ras protein has particular similarities to the non-classical low molecular weight G proteins described by Clapham, which probably also exist as a monomers. p21 requires interaction with the GTPase activating protein GAP to accelerate hydrolysis [28]. The GTP-bound form is active in signal transduction and strongly transforming p21

mutants are activated by their reduced intrinsic GTPase activity. Membrane localization of p21 is also required for transformation, and Marshall's presentation highlighted the group's recent elegant elucidation of the post-translocational modifications leading to the targeting of p21 to the inner leaflet of the cell membrane [29]. An initial increase in p21 hydrophobicity arises from a sequence of cytosolic processing changes involving removal of the three C-terminal amino acids, followed by polyisoprenylation and carboxyl methylation at cystein 186. This involves recognition of the C-terminal CAAX motif. Polyisoprenylation is essential for stable membrane association. Blockade of this process by inhibitos of mevalonic acid synthesis (mevalonin and compactin) prevents membrane binding. Harvey and N-ras, but not Kirsten ras, proteins are palmitoylated upstream of cystein 186 - this contributes to high avidity membrane binding, but is not essential for membrane localization or transformation. Inhibition of Malitoylation and especially isoprenylation clearly represents a target for new anti-p21 drugs, but the problem may once again be specificity since so many proteins contain CAAX boxes for membrane integration. Previous evidence from yeast implicated adenylate cyclase activation in transformation by ras but the nature of the oncogenic signal is unknown. Transfection with ras in the presence of insulin-like growth factor 1 (IGF-1) induces both DNA synthesis and transformation. However, Marshall described how activation of protein kinase C is required for DNA synthesis but not morphological transformation, implicating two distinct signalling mechanisms.

Membrane lipids and signalling Continuing a theme begun in earlier sessions, Peter Downes (Smith Kline and French Research, Welwyn, UK) described approaches to the development of medicinal chemistry in the complex area of inositol phospholipid miutogenic signalling pathways. Having surveyed various potential therapeutic wrecking sites in phosphoinositol metabolism, Downes identified the branch point created by the recently identified Ptd Ins 3-kinases (see earlier) and Ptd Ins 4-kinases as a particularly exciting target for the development of anticancer agents. Using partially purified placental and erythrocyte Ptd kinases, his team is exploring the structure-activity relationships for the ATP and phosphatidylinositol molecular recognition sites. Differences exist between the two enzymes. Downes described the group's recent progress in the development of ATP-site antagonists of Ptdlns^4-kinase. ATP has a Km of 47 urn while the Ki for adenosine is higher at 86 uM. However, inhibitory potency is improved by substitution in the 8-position, as in 8-(2-pyridyl) adenine with a Ki of 36 uM. Enhanced activity could also be attained by modification at the 9-position, and a low Ki of 3.7 uM was achieved with 9-cyclohexyl adenine (SKF 97495). Pharmacological activity was observed in intact erythrocytes and macrophages with this agent Linking together the earlier presentation by Noseda on

105 the biophysical perturbation of membranes by ether lipids and subsequent experimental and clinical papers on these drugs by Modest, Berdel and Unger (as well as numerous proferred posters) Ewa Ninio (INSERM, Paris, France) reviewed the biochemistry and biology of PAF [30] and related compounds. She emphasised the important range of cell responses which are mediated through the binding of PAF to specific membrane receptors. PAF antagonists, including natural products (ginkgolides, kadsurenone), structural analogues (CV 6209, SRI 63-072) and triazolobenzodiazopine (WEB 2086) [31], allow receptormediated effects to be distinguished from alternative, potentially less specific pathways. The pleiotropic molecular effects of PAF include elevation of intracellular calcium, protein kinase C activation and stimulation of DNA synthesis. Ninio pointed out that although lyso-phospholipids had been regarded somewhat perjoratively as "just detergents" because of their lyric properties, recent evidence also supported protein kinase C modulation by lyso PAF and lysophosphatidyl choline. The latter stimulates the enzyme at low doses, but inhibits at high doses. Lyso-phospholipids are not stored in the cell but are (re)acetylated by acyltransferases and CoA-dependent transacylases. Some "escape" acylation with long chain fatty acids and are converted to PAF by a specific acetyltransferase. Stimulation of appropriate cells (neutrophils, monocytes, mast cells) was shown to generate both lyso PAF and PAF itself. Acetylation of lyso PAF to form PAF seems to require a phosphorylation-dependent stimulation of the specific acetyltransferase. A second route of PAF production involves de novo synthesis via CDPcholine phosphotransferase acting as alkyl-acetyl-glycerol: this pathway governs constitutive manufacture in tissues such as kidney, lung and uterus. Synthesis is regulated variously in different cell types. Control is exerted through activation of analbolic enzymes phospholipase A2 and acetyltransferase, and also via the recently identified PAFdegrading enzyme acetylhydrolase. New data implicate protein kinase A in the phosphorylation of acetyltransferase, and protein kinase C and G-proteins may also regulate PAF synthesis. The wide ranging functions of PAF in health and disease are becoming increasingly manifest. PAF appears to play an important role in asthma, shock, ischemia, allergy and various other inflammatory diseases. It can be isolated from both tumour and peritumoural tissue [32]. Ninio speculated on the potential for the phospholipid evironment to regulate numerous receptors and enzymes, in addition to the phospholipid breakdown products which participate as intercellular and intracellular mediators of signal transduction. Chemical analogues of these products are also likely to exhibit multiple pharmacological effects. Numerous proferred posters at the meeting dealt with the mechanism of action of potential anticancer ether lipids. It is not clear that this time to what extent these agents require functional PAF receptors, and this will be an important issue to clarify as PAF-related drugs continue to enter early clinical trial in cancer patients. It is, however, important to bear in mind that selective antitumour activity might be attained by

mechanisms not involving specific PAF receptors. This is now being determined using PAF antagonists [31]. In addition to the biophysical effects of ether lipids described earlier, the multiple biological actions of these agents, reviewed by Ed Modest (Boston University, USA), include induction of differentiation, macrophage activation, inhibition of protein kinase C, increased intracellular calcium and interaction with the EGF receptor. They do not damage DNA. Modest described two classes of ether lipids, type A containing phosphorus and type B which have positively charged substituents in place of the phosphate at position 3. Both fluidize cell membranes and interact with protein kinase C, as well as having activity in ADDS screens. Promising results for combinations of ether lipids with conventional cytotoxic drugs or hyperthermia were described. Clinical studies with ether lipids were reported by Wolfgang Berdel (Technical University of Munich, FRG) and Clemens Unger (University Medical Clinic, Gottingen, FRG). These agents have a range of antitumour properties in preclinical models [33]. Berdel described the successful use of ET-18-O-Methyl for selective purging of residual malignant myeloid and lymphoid cells from bone marrow. A Phase 1 study established the safety and efficacy of autologous bone marrow transplantation using remission marrows cleaned up with ether lipid in vitro, and further studies are in progress. Unger reviewed clinical studies with alkylphosphocholines, particularly hexadecylphosphocholine (HPC). Like the other ether lipids, these are membrane-active and induce differentiation. Antitumour activity was noted in some preclinical screens but not in the rather unpopular P388 mouse leukaemia. Responses have been seen with topical application of HPC in patients with skin metastases from breast cancer. Phase 1 studies with an oral formulation are in progress. Returning once more to doxorubicin, John Hickman (Aston University, Birmingham, UK) reviewed his group's work concerning the effects of this agent on inositol lipid metabolism in relation to regulation of the cytoskeleton in erythrocytes [34]. Low levels of doxorubicin prevented the calcium-induced echginocyte to discocyte shaoe change and this was accompanied by a modulation of phosphoinositol breakdown. The drug both stimulates and inhibits calcium-induced Ins(l, 4, 5)P3 production, possibly via effects on putative G s and Gj proteins. It is not clear how those signals are coupled to cytoskeleton remodelling.

Ions and signal transduction Enrique Rozengurt (Imperial Cancer Research Fund, London, UK) reviewed his group's elucidation of the early signals in the mitogenic response [5, 13]. He described the requirement for the esynergistic effects of different growth factors acting on quiescent Swiss 3T3 mouse fibroblasts grown in serum-free medium. He focussed especially on the role of bombesin, hastin-releasing peptide (GRP) and related peptides, an area of considerable activity at the moment because of the potential to develop bombesin an-

106 tagonists to block autocnne stimulation in the treatment of small cell lung cancer. Rozengurt reported the new observation that whereas the bombesin receptor has no intrinsic tyrosine kinase activity, it does promote the phosphoinositol cascade, mobilising calcium and activating protein kinase C via a Pertussis toxin-insensitive G protein. This leads to sodium and potassium fluxes, transmodulation of the EGF receptor, accumulation of cAMP and transcription of c-fos and c-myc. A different Pertussis toxin-sensitive G protein mediates cross-talk. Rozengurt criticized the proposal of Nishizuka [14] that the negative action of protein kinase C induced by growth factors was important for mitogenesis. He showed instead that these effects can be divorced from each other, favouring a forward action for protein kinase C in growth promotion (see later). Early mitogenic signals induced by bombesin, vasopressin, bradykinin and similar peptides can be antagonised by multiple specific antagonists [35] and also the less selective substance P antagonists in both 3T3 and small cell lung cancer lines. The results suggested action of substance P antagonists at both ligand-specific and non-specific domains. Such broad-spectrum antagonists would be expected to have an advantage over more specific ones in situations where multiple synergistic signals are driving mitogenesis - an interesting new thought Continuing the synergy theme, Wouter Moolenaar (Netherlands Cancer Institute, Amsterdam) proposed that, in contrast to normal cells, histamine can act as a cancer cell mitogen because tumours generally manufacture their own EGF-like factors which are required synergistically for histamine activity. He considered that ligands acting through tyrosine kinase were more effective mitogens in normal cells than those, like histamine, acting via phosphoinositol metabolism. Moolenaar went on to describe recent experiments which identified luysophosphatidic acid, originally present as an impurity in phosphatidic acid, as a potent mitogen [36]. The growth stimulus is inhibited by pertussis toxin, implicating a G protein. Although the phosphoinositol cascade is activated, this is not required or sufficient for mitogenesis. Instead, the critical event is the activation of a Gj protein mediating inhibition of adenylate cyclase. Michael Cahalan (University of California, Irvine, USA) discussed the three types of voltage-gated potassium channels in lymphoid cells. These were studied using elegant patch-clamp recording and video imaging of fluorescent ion-sensitive dyes. Potassium channels appear to be involved in the mitogenic activation of T and B lymphocytes. Related to this, Hans Grunicke (University of Innsbruck, Austria) showed how anticancer alkulating agents cause a rapid decrease in the influx of the potassium congener rubidium. Nitrogen mustard was shown to inhibit ouabain and furosemide-sensitive rubidium uptake. The effects were in proportion to cytotoxicity, and were much less in mustard-resistant cells. Nitrogen mustard alsod epresses the sodium/proton antiport, suggesting interference with signal transducdon. This work may be seen in two ways: a means to improve the use of alkulating agents

[37] or a basis upon which develop new types of drugs. From the attention it has received, here and elsewhere, calcium signalling is clearly an extremely important target for pharmacological intervention. Intracellular free calcium is normally maintained at around 10~7M, which is 10,000 fold lower than the extracellular level, rising only to 10^M during signalling. This allows the cell to exploit calcium as a very sensitive signalling mechanism. For this reason Garth Powis (Mayo Clinic, Rochester, USA) was optimistic about possible rational approaches to pharmacological manipulation of calcium signalling in cancer. This might be achieved by inhibiting calcium influx channels, blockade of calcium efflux to give a sustained rise, or preventing calcium uptake or export from internal stores, perhaps by antagonizing the Ins(l, 4, 5)P 3 receptor. Like heparin which blocks Ins(l, 4, 5)P3-activated calcium release from internal stores, polyionic dextran sulphate (500 kD) also has this property. The sulphate groups are required for activity, but the size and charge conspire to prevent intracellular access. Another polysulphate substance which does have antiproliferative activity is the experimental cancer and AIDS drug suramin. Suramin and dextran sulphate also antagonize PDGF receptors as well as blocking calcium mobilisation. Powis showed how prolonged exposure to quite high concentrations of ether lipid ET-18-O-Methyl can block both uptake into the ER as well as Ins(l, 4, 5)P3mediated calcium release in permeabilized 3T3 cells. In contrast posters by Lazenby and Dive (Aston and Cambridge Universities, UK) showed a phorbol ester-sensitive rise in intracellular calcium by low levels of ehter lipid SRI 62-834 in intact tumour cells, apparently implicating protein kinase C in the drug's effect

Protein kinases as targets Opening a particularly important session for current drug development initiatives, John Haley (Oncogene Science, Manhasset USA) took up the story of structure-function relationships for the EGF receptor begun by Waterfield. He showed, as expected, that in immortalized rodent flbroblasts the truncated receptor lacking the ligang binding domain induced transformation through activation of the intrinsic tyrosine kinase. Not only that, but transformation was enhanced by further truncation of the C-terminus containing two autophosphorylation sites. In contrast C-terminal truncation of hyperexpressed intact receptor decreases transformation driven by EGF, perhaps suggesting a role for this domain in receptor dimerization. Haley concluded by speculating on possible approaches to inhibiting EGF receptor-induced cell proliferation. The key signalling pathways activated by the receptor tyrosine kinase are unknown, although the PDGF (J-receptor (also a tyrosine kinase) has been shown to activate the serine/threonine kinase of the raf-1 oncogene product through tyrosine phosphorylation. For the present Haley considered modifying the EGF receptor itself, and mentioned inhibition of tyrosine kinase by pseudosubstrate peptide sequences, a successful approach with sequences

107 present in cAMP-dependent protein kinases and protein A series of fibroblast lines transfected so as to stably exkinase C [38]. He also suggested the use of serine and press a high level (up to 53 times control) of a protein threonine kinase inhibitors, since phosphorylation was also kinase subspecies exhibited both disordered cell growth seen at these sites in the EGF and related receptors. Such and tumorigencity [41, 42]. The effects were more like antagonists might be used in combination with inhibitors of those resulting from myc and fos transfection than with the the receptor tyrosine kinase itself. more frankly malignant phenotype seen with ras. These On that precise theme, Yosef Graziani (Ben Gurion results are consistent with the known tumour promoter role University, Beer Sheva, Israel) summarised what we know for protein kinase C-activating phorbols and confirm a funabout tyrosine-specific protein kinase inhibitors. The damental involvement of the enzyme in multistage carbioflavonoid quercetin has been a prototype compound cinogenesis. But 3T3 cell tumours induced by initial [39]. Additional leads in this area include other natural protein kinase C overexpression failed to maintain this high products like genistein, orobol, erbstatin and herbimycin. enzyme level [42]. This may mean that alternative transThese are often quite potent but specificity can be a forming pathways bypass the protein kinase C signal in the problem, with not only tyrosine-independent kinases but final malignant state in this particular example. The "dual also various other enzymes being affected. This is because function" [14] oif protein kinase C is also relevant to this these agents generally compete with the ATP-binding site discussion. Short-lived agonists like growth factors produce which is likely to be similar, though not perhaps identical, a "forward action" through temporary activation of protein across different classes of kinases. Nevertheless, quite kinase C by DAG and also calcium. This positive action surprising selectivity can be achieved. For example, the includes induction of interleukin-2 receptor and oncogene trihydroxy isoflavone genistein inhibits the EGF receptor activation. The "negative action" involves the negative tyrosine kinase with an ID50 of 6.5 ug/ml, without any feedback control over various steps in cell signalling. These effect on the cAMP-dependent serinc/threonine protein effects include reduced Pdtlns(l, 4, 5)P3-induced calcium kinase, protein kinase C, phosphorylase kinase or 5'- release. This may perhaps occur by blocking or decoupling nucleotidase event at 100 ug/ml [40]. Inhibitors which bind the receptor-induced inositol lipid cascade or by stimulatat the peptide substrate site might be expected to show fur- ing a Ptdlns(l, 4, 5)P3 phosphatase, or protein kinase C ther improvements in specificity. A lead compound here may promote calcium removal by stimulating a calciummay be erbstatin. Most interesting are the tyrphostin ben- ATPase or sodium/calcium exchanger. Also, protein kinase zylydene compounds, from Levitzki's group, which are C phosphorylation of EGF receptor reduces ligang binding very specific inhibitors of the EGF receptor tyrosine kinase. and inhibits ligang-induced tyrosine phosphorylation to Tyrphostin 8 exhibits a Ki of 11 um against this enzyme produce a functionally "down-regulated" receptor [43]. compared to 1200 um for the insulin receptor equivalent. Although their early effects mimic DAG, phorbol esters Enzyme inhibition is associated with antiproliferative (which bind at the same regulatory site) produce sustained activity against EGF-stimulated cells. activation of protein kinase C, prolonging its association Turning now to another important kinase, protein kinase with the membrane (rather than promoting its translocation C - referred to at the meeting as the Clapham Junction (or there, according to Parker) and initiating enzyme perhaps it should be the Schiphol Airport) of cellular sig- degradation by calcium-dependent calpain 1 [14]. nalling traffic. Peter Parker (Ludwig Institute, London, UK) The permissive or obligatory role of the various protein explained how cDNA cloning has led to the identification kinase C molecules will emerge in part from studies on of a family of these calcium and phospholipid dependent various new exogenous modulators. Because of the comserine/threonine protein kinases [14]. Activated by both plexity, drug development interest is currently focussed on diacyl glycerol liberated through inositol lipid metabolism, both activation and inhibition [44,45]. Dan Dexter (E.I. du as well as by phorbol esters, protein kinase C phosphory- Pont de Nemours, Wilmington, USA) described the substanlates many proteins including the EGF receptor, ppoXF^, tial progress in the discovery of specific regulatory and the insulin receptor and ras p21. As an important integrator catalytic domain ligands. Computer modelling of the posin signal transduction it is an especially attractive target for sible pharmacophores was carried out using regulatory drug development We should be encouraged here by the domain ligands like phorbol esters and especially the inmultiplicity afforded by six distinct genes generating at hibitor dihydrotelocidin B. Diacylglycerol was not used least seven polypeptides via alternative splicing of the {$ because it is "too floppy". This led to a "triangle gene, and the fact that these are differentially expressed hypothesis" for pharmacophore binding. At the apices of across tissues. It is not clear, however, to what extent the the triangle are a hydroxyl group, a hydrogen bond donor different forms (?isoenzymes) have different functions and and a hydrogen acceptor. Based on this model a series of regulatory features. Nevertheless, early data look promis- benzyl alcohols, including compound X-6789, were syning. Thus, the y and a subspecies are activated much less thesized and shown to exhibit potent competitive binding at readily by DAG in the presence of phosphatidyl serine than the regulatory domain with respect to phorbols. They are the p" subtypes, while the latter are more poorly ac- proved to be partial or full agonists and stimulated plastivated by physiological arachidonate levels than the minogen activator in endothelial cells. Also shown to be former. This is clearly good news, suggesting that selective regulatory domain inhibitors were sphingosine sulphate, pharmalogical modulators might be found. W7, tamoxifen and dibucaine. The alkaloid N.Ndimethylstaurosporin was shown to bind reversibly with Should we be activating or inhibiting protein kinase C?

108 high affinity (dissociation constant 3.8nM) to the catalytic domain, and an assay based on competition for binding with the tritiated compound was used to search for ligands at this site. Inhibitors included staurosporin (Ki 4.7nM), the dimethyl analogue itself (4.7nM), K-252a (66nM) and H-7 (79 uM), together with doxorubicin (54 uM) and the nuclcosidc analogue sangivamycin (15 uM). Interestingly, magnesium-ATP was only weakly competitive, suggesting that staurosporins do not bind at the ATP site itself. In a third approach, a series of bis-naphthalene sulphonamides was made in which competitive binding at the regulatory site with respect to phorbol and the resulting inhibition of catalytic activity and were found to correlate with cytostatis in a human melanoma cell line in vitro. Analogues like X8003 showed modest activity against the B16 mouse melanoma in vivo and are interesting lead compounds. In a poster, Japanese workers (Nakano et al., Kyo Hakko, Tokyo) described their search for protein kinase C inhibitors by screening microbial products for inhibition of various kinases, and identified calphostin (UCN 1028) as a regulatory domain inhibitor. Moreover, it was very selective with an IC50 of 50 nM for C-kinase compared to > 50 uM for A-kinase and pp60v"src tyrosine kinase. Respective values were 4.1nM, 42nM and 45nM for 7-hydroxystaurosporine; and for staurosporine these were 2.7nM, 8.2nM and 6.4nM, with 630nM for the tyrosine kinase of the EGF receptor. In vitro and in vivo antitumour activity was reported for calphostin in some relevant models. Andy Gescher (Aston University, Birmingham, UK) continued with the theme of pharmacological modulation of protein kinase C. He discussed the tight structure-activity relationship for phorbol esters, with 12-0tetradeconoylphorbol-13-acetate being the most active. Other natural but unrelated terpenoids like mezerein, teleocidin, aplysiatoxin and ingenol 3-tetradecanoate were also potent agonists. Exogenous DAGs mimic some but not all effects of TPA on cells. In a poster Bradshaw and Gescher showed that serum factors may act in concert with TPA to produce growth arrest of A549 cells. However, Gescher chose to concentrate on a remarkably interesting and potent class of protein kinase C agonist - the bryostatins [44, 46]. These are complex macrocyclic lactones isolated by Pettit from marine bryozoans which grow on the bottom of boats. Nanomolar concentrations inhibit TPA binding to the regulatory domain. Like TPA they stimulate protein kinase C activity and also promote membrane localization and down-regulation of the enzyme. Growth arrest for both types of agent is seen at lower doses than those required for the above effects. Gescher showed that the bryostatins differed from TPA in that high concentrations abolish the drugs' growth inhibitory properties and antagonize growth arrest by TPA. They also antagonize TPA-induced differentiation of HL60 leukaemia cells as wellas TPA-induced tumour promotion in mouse skin. In vivo antitumour activity was seen in mouse models, including antimetastatic effects in the B16 melanoma. Interestingly, bryostatins can be mitogenic to 3T3 cells in the presence of insulin. Bryostatin 1 will shortly enter Phase 1 clinical trial in the UK under the

auspices of the CRC Phase I/n Clinical Trial Subcommittee. Important mechanistic as well as clinical dats will be obtained. Mike Dexter (Paterson Laboratories, Manchester, UK) reviewed the complex orchestration of normal hemopoietic cell differentiation by growth factors such as interleukins 1 and 3, granulocyte colony stimulating factor (G-CSF), macrophage-CSF (M-CSF) and granulocyte/macrophageCSF (GM-CSF). Some growth factors promote proliferation and differentiation of multipotent stem cells but others are more confined, supporting only the development of lineage-restricted progenitor cells unless used in combination [47]. For example, IL3 sustains growth and differentiation of early stage progenitors common to all cell lineages. IL 1 or G-CSF can synergize with GM-CSF and M-CSF on primitive stem cells. We now know that loss or modulation of receptors regulates differentiation. Cell-cell contact is important, as are effects of the stromal extracellular matrix. Growth factor molecules can be localized by human marrow matrix; for example both GM-CSF and IL3 are absorbed by the major sulphated glycosaminoglycan, heparin sulphate [48]. Several of the above factors translocate protein kinase C to, or stabilize the enzyme within, the cell membrane. Accordingly, their effects can be blocked by TPA. Phosphoinositide and calcium signalling may not be involved but effects on the sodium/proton anitport appear to cause alkalinization. Many of the important growth factors have now been cloned and sequenced, and their recombinant forms are available in large amounts for experimental and clinical studies. For example, several clinical trials using hemopoietic growth factors to protect bone marrow during chemotherapy are now in progress. Dexter went on to describe recent work on the remarkable selctive toxicity of ether lipids against leukemic versus normal marrow progenitors, an effect which is blocked by PAF antagonists [49]. Although this implicates PAF receptors, Dexter's group has also shown that diverse agents including chloroquin, monensin and vinblastine also protect leukemic cells by reducing ether lipid uptake. He proposed the involvement of vesicular endosomal traffic rather than PAF receptors, and queried the effects of oncogene expression on this as well as the relationship to multidrug resistance (MDR). Several posters were corcerned with membrane aspects of MDR. One by Workman (Cambridge University, UK) showed that there was minimal resistance to the membrane-active ether lipid SRI 62-834 in a classic MDR variant of the EMT6 mouse mammary tumour cell line which hyperexpresses the P-glycoprotein drug efflux pump in the membrane. The effects of retinoids on leukemic cell signalling were reviewed by Jean-Pierre Abita (Hfipital Saint-Louis, Paris, France). In the HL60 model, phospholipase C and various ion transport mechanisms were implicated. Alkalinization appears to be important, together with a reduction in inositol phosphates and DAG probably via a direct effect on the membrane which decouples a stimulatory G protein. Abita reported promising results from clinical studies in Shanghai, China, and more recently in Paris, France, where patients with acute myelocytic leukaemia were treated with

109 all-fra/w-retinoic acid. In one of a series of three talks on metastasis, Garth Nicolson (M.D. Anderson Cancer Centre, Houston, USA) presented evidence that the organ specificity of cancer metastasis is regulated by molecular changes at the cell surface affecting specific adhesion, invasion and growth [50]. These studies are providing a molecular understanding of the old "seed and soil" scenario, as well as a rational basis for therapeutic intervention. For example, molecules involved in the sticking of tumour cells to microvessel endothelial cells and invasion of basement membranes were discussed. Inhibition of degradative enzymes on tumour cell membranes is a particularly attractive target. Heparinlike agents and suramin (see earlier) inhibit tumour membrane heparanase which degrades heparan sulphate proteoglycan in the basement membrane. This is effective against invasion and metastasis. Nicolson predicted paracrine growth stimulation would be an especially fruitful area of investigation, and pointed to a 66kD growth factor isolated from lung which binds to a specific receptor on tumour cells metastatic to this organ. He also advised using membrane-active agents in combination with more traditional ones for optimal effect. Marc Mareel (University Hospital, Ghent, Belgium) looked at the effects of various agents interacting with cytoskeletal elements on invasion in his chick heart model. Microtubule inhibitors may be anti-invasive through interference with directional migration. Pyrimido-pyrimidines and flavonoids may modify tumour adhesion mediated by laminin. Ether lipids such as ET-18-0-Methyl inhibited invasion of certain tumour cells while Pertussis toxin and retinoic acid antagonised some and inhibited others. Mareel suggested that these results support the possibility of modulating metastasis at specific biochemical stages. Concluding this section, Ian Hart (Imperial Cancer Research Fund, London, UK) hypothesized that signalling mechanisms will vary according to the stage of tumour development, as may the therapeutic response. He provided evidence to support this from his recent work on normal and oncogene-transfected melanocytes [51]. Normal melanocytes persist in culture in the presence of protein kinase C-activating phorbol esters and mezerein, as well as sapinotoxins A and D. Elevation of cAMP further enhances proliferation in normal cells, but inhibitsmalignant melanoma cells. Tumorigenic and metastatic clones were derived from TPA-treated normal melanocytes after transfection with active Uz-ras or polyoma middle T oncogenes. Hart proposed that molecular lesions in the protein kinase C pathweay could be especially important in the malignant and metastatic cascades. The relationship between the cell membrane and calcium homeostasis with toxicity and programmed cell death (or apoptosis) was the subject of the final presentation, by Sten Orrenius (Karolinska Institute, Sweden). Cell damage by toxic chemicals, drugs and cytotoxic cells is correlated with sustained increases in intracellular free calcium [52]. As examples, alkylating agents and redox-active chemicals like the quinonc menadione affect calcium homeostasis by inhibition of transport ATPases while cytotoxic T cells and

natural killer cells stimulate calcium influx. Sustained calcium elevations appear to mediate killing by activation of calcium-dependent protease, phospholipase and endonuclease enzymes. The latter produces a characteristic pattern of DNA damage called "laddering" which appears to be characteristics of apoptosis [53]. This can be seen in recent experiments in which immature thymocytes are treated with glucocorticoid hormones, where production of a novel pore-forming protein factor is thought to permeabilize the cell membrane to extracellular calcium influc. Orrenius also reported that a similar pattern of DNA laddering arose in these cells on exposure to oxidative stress or the environmental toxin 2, 3, 7, 8-tetrachlorodibenzo-p-dioxin [53]. The connection between cell damage and killing and prolonged, elevated calcium is clearly very widespread and the answer to the cause or effect question is becoming clearer. The precise nature of the molecular events common to the pathological calcium rise and normal calcium signalling remain to be defined.

Summary In the concluding Discussion session, emphasis focussed on the potential for interfering selectively with cell membranes and cell signalling in tumour as against normal tissues. There could be no doubt that tremendous advances are being made in our understanding of the molecular changes associated with malignancy and that the information available for the rational design of inhibitors of particular signalling pathways is increasingly sophisticated. There was a consensus that we need more information on the qualitative and quantitative differences in the structure and function of membranes and the signalling machinery in various normal tissues as compared to their cancerous counterparts. Ideally we will develop drug against, for example, specific forms of, let us say, protein kinase C or tyrosine kinase which are found to be predominantly active in neoplastic cells. This may well prove possible, at least in some instances, in which case a safe therapeutic margin will be assured. But differences may in other situations turn out to be in the level of expression rather than purely qualitative in nature, and the scale of the disparate expression may not always be great. Even in such situations, adequate therapeutic selectivity may still be achieved. This may derive from a "damping down" of signalling in the hyperactive tumour. Although there are legitimate concerns regarding the possible toxic effects of administering signalwrecking molecules in man, we should not be pessimistic as there are clear precedents elsewhere in medicine for drugs acting on membrane signals proving to be safe and effective against expectation informed by hindsight. There may also be concerns about new forms of drug resistance. But this will be so for any new agent or novel target. And with mechanism of action clearly to the fore we should be able to predict resistance pathways in advance and devise appropriate circumvention strategies or targeted second line therapies. There was a palpable buzz at the meeting that this is a valid, different and above all rational approach.

110 Not only that, but the new therapeutic molecules which we discover will themselves prove to be valuable tools with which to probe further into the mechanisms of malignancy and signal transduction. We had expected to see a bewildering amount of new information from the basic sciences of molecularbiology and cell physiology, and we got it But it was also impressive to witness the number of new compounds coming through which look like real drugs or at least exciting lead compounds. The membrane-active ether lipids are in clinical trial. Bryostatin 1 will shortly join them. A number of other novel chemicals, many of them natural products, are also moving through clinical and preclinical development. Many of these are likely to have new mechanisms of action, quite distinct from existing drugs, and some will be found to affect cell membranes signalling pathways, as is now recognized for the bryostatins. However, our real optimism must be based soundly in rational design. The prospect for a battery of highly selective protein kinase inhibitors surely does not seem as remote as it did before the advent of genistein and the tyrphostins. Although in the last few years clinical results with new but often basically traditional drugs have been disappointing, there have been real gains. The machinery for preclinical and clinical drug development is firmly in place. We are now benefiting from fast-track toxicology, improved formulation facilities, pharmacologically-guided dose escalation and so on. Sensible screening is also of paramount importance. It is pointless to employ sophisticated modem drug design techniques to develop exciting new drugs with novel mechanisms of action, only to test them on oldfashioned, fast-growing undifferentiated and biochemically irrelevant tumours like the P388 mouse leukaemia. In many cases the primary screen will involve biochemical assays (enzyme inhibition, receptor-binding) or even the computer graphics display. Where antirumour testing is required human cell lines, xenografts and even more biologically advanced systems should be used. An important question is when to take a given new approach to the bedside. There will always be a dilemma here - between the need to test new compounds which may have improved activity as early as possible, and the desire to wait until the ideally selective agent can be identified. The danger of going too early is that unacceptable toxicity or lack of activity may reduce enthusiasm for a particular class of new agent or target Some or probably many problems will be encountered in early clinical trials. In order to maximize the information gleaned from such trials it is important to derive as much mechanistic and pharmacologic data as possible. This will be an exciting era of cancer chemotherapy with our imaginations directed towards the membrane and signalling targets discussed here, together with other new loci yet to be identified. It will require the concerted interdisciplinary collaboration of basic molecular biologists and cell physiologists with innovative chemists, bom-again cancer pharmacologists and biologically sophisticated experimental chemotherapists, working closely with interested clinicians. This first international meeting showed

how specialists in these different areas can be brought together successfully to assess progress, define problems and indicate vanues in which real advances are to be made.

References

1. Proceedings of the meeting Drug - DNA Interactions. Anticancer Drag Design (in press). Z Rodwell JD. Engineering monoclonal antibodies. Nature 1989; 342: 99-100. 3. BaDcwill FR. Cytokines in Cancer Therapy. Oxford University Press, Oxford 1989. 4. Cohen P, Houslay M (eds). Molecular mechanisms of transmembrane signalling. Vol. 4. Molecular aspects of cellular regulation. New York: Elsevier, 1985. 5. Rozengurt E. Early signals in the mitogenic response. Science 1986; 234: 161-6. 6. Berridge MJ, Irvine RF. Inositol phosphates and cell signallingJiature 1989; 341: 197-205. 7. Sharp PA. Gene regulation and oncogenes. Cancer Research 1989; 49:2188-94. 8. Murray AM. The cell cycle as a ede 2 cycle. Nature 1989; 3/2: 14-15. 9. Landis CA, Masters SB, Spada A, Pace AM, Bourne HR, Vallar L. GTPase inhibiting mutations activate the t chain of Q s and stimulate adenyl cyclase in human pituitary tumours. Nature 1989; 340: 692-6. 10. Hickman JA for the EORTC Pharmacolrinetics and Metabolism Group. Membrane targets in cancer chemotherapy. Eur J Cancer 1988; 24:1385-9. 11. Proceedings of the meeting The Cell Membrane and Cell Signals as Targets in Cancer Chemotherapy. Cancer Chemother Pharmac 1989; 24: Supplement 2. 12. Powis G, Hickman JA, Workman P, Tritton TR, Abita J-P, Berdel WE, Gescher A, Moses HL, Nicolson GL. The cell membrane and cell signals as targets in cancer chemotherapy. Cancer Research (in press). 13. Rozengurt, E. Signal transduction pathways in mitogenesis. Br Med Bull 1989; 45: 515-28. 14. Nishizuka Y. The molecular heterogeneity of protein kinase C and its implications for cellular recognition. Nature 1988; 334: 661-5. 15. Burgoyne RD, Cheek TR, Morgan A, Sullivan AJ, Moreton RB, Berridge MJ, Mata AM, Colyer J, Lee AG, East JM. Distribution of two distinct Ca2+-ATPase-like proteins and their realtionsrdps to the agonist-sensitive calcium store in adrenal chromaiTia cells. Nature 1989; 342: 72-4. 16. Fimrichi T, Yoshikawa S. Miyawaki A, Wada K, Maeda N, MikosHba K. Primary structure and functional expression of the inositol 1, 4, 5-trisphosphate-binding protein P400. Nature 1989; 342: 32-8. 17. Ferris CD, Huganir RL, Supattapone S, Snyder SH. Purified inositol 1, 4, 5-triphosphate receptor mediates calcium flux in reconstituted lipid vesicles. Nature 1989; 342: 87-9. 18. Whitman M, Dowries CP, Keeler M, Keller T, Cantley L. Type 1 phosphan'dulinositol kinase makes a novel inositol phospbolipid, phosphatidylinositol-3-phosphate. Nature 1988; 332:644-6. 19. Coughlin SR, Escobedo, JA, Williams LT. Role of phosphatidulinositol kinase in PDGF receptor signal transduction. Science 1989; 243: 1191^*. 20. Sdnnitzky M. Physiology of membrane fluidity, Vols. 1 and 2. Boca Raton: CRC Press, 1984. 21. Tritton TR, Hickman JA. Cell surface membranes as a chemotherapeutic target In: Muggia FM, ed. Experimental and Clinical Progress in Cancer Chemotherapy. The Hague, Martinus Nijhoff 1985; 81-131. 22. Nicolay K, Sautereau AM, Tocanne JF, BrasseuT R, Huart P, Ruysschaert JM, de Kruijff B. A comparative model study on struc-

Ill

23.

24.

25. 26.

27.

28. 29.

30. 31.

32.

33.

34.

35.

36.

37.

38. 39.

40.

41.

42.

43. 44. 45.

tural effects of membrane-active positively charged anti-tumour drugs. Biochem Biophys Acta 1988; 940: 197-208. Spom MB, Roberts AM, Wakefield L, Assoian R. Transforming growth factor-0: Biological function and chemical structure. Science 1986; 233: 532-A Downward J, Yarden Y, Mayes E, Scrace G, Totty N, Stockwell P, Ullrich A, Schlessinger J, Walerfield MD. Close similarity of the epidermal growth factor receptor and v-erb B oncogene protein sequence*. Nature 1984; 307: 521-7. Steinberg ME, Gullick WJ. Neu receptor dimerization. Nature 1989; 339: 587. Logothetis DE, Kurachi Y, Galper J, Neer EJ, dapham DE. The ft subunits of GTP-binding proteins activate the muscarirdc K* channel in heart Nature 1987; 325: 321-6. Pai EF, Kabsch W, Krengel U, Holmes KC, John J, Wittinghofer A. Structure of the guanine-nudeotide-binding domaine of Ha-rar oncogene product p21 in the triphosphate conformation. Nature 1989;341:209-14. McCormick F. ras GTPase activating protein: signal transmitter and signal terminator. Cell 1989; 56: 5-8. Hancock JF, Magee AL Childs JE, Marshall CJ. All ras proteins are polyisoprenylated but only some are palmitoylated. Cell 1989; 57: 1167-77. Hanahan DJ. Platelet activating factor a biologically active phosphoglyceride. Ann Rev Biochem 1986; 55:483-509. Casals-StenzelJ, Muacevic G, Weber KH. Pharmacological actions of WEB 2086, a new specific antagonist of platelet activating factor. J Pharmac Exp Ther 1987; 241:974-81. Pitton C, Lanson M, Besson P, Fetissoff F, Lansac J, Benveniste J, Bougnoux P. Presence of PAF - acether in human breast carcinoma: relation to axillary lymph node metastases. J Nat] Cancer Inst 1989; 81: 1298-1302. Berdel WE, Munder PG. Antincoplastic actions of ether lipids related to platelet-activating factor. In: Snyder F, ed. Platelet-activating factor and related lipid mediators. New York Plenum 1987; 449-67. Thompson MG, Chahwala SB, Hickman JA. Inhibition of human erythrocyte inositol lipid metabolism by adriamycin. Cancer Res 1987; 47: 2799-803. Woll PJ, Rozengurt E Multiple neuropeptides mobilise calcium in small cell lung cancer effects of vasopressin, bradykinin, cholecystokinin, galanin and neurotensin. Biochem Biophys Res Comm 1989; 164: 66-73. van Corven EJ, Groenik A, Jalink K, Eicholtz T, Mollenaar WH. Lysophosphatidale-induced cell proliferation: identification and dissection of signalling pathways mediated by G-proteins. Cell 1989; 59:45-54. Grurdcke H, Doppler W, Hoffman J. New strategies for the improvement of alkylating anlitumour agents. In: Cory JG, Sventivanyi A, eds. Cancer biology and therapeutics. New York Plenum 1987; 127-39. Hardie G. Pseudosubstrates turn off protein kinases. Nature 1988; 335: 592-3. Graziani Y, Chayoth R, Kamy N, Feldman B, Levy J. Regulation of protein kinases activity by quercetin in Eriich asdics tumour cells. Biochem Biophys Acta 1981; 714:415-21. Alayama T, Ishida J, Nakagawa S, Ogawara H, Watanabe S-I, Shibuya M, Fukami Y. Genistein, a specific inhibitor of tyrosinespecific protein kinases. J Biol Chem 1987; 262: 5592-5. Housey OM, Johnson MD, Hsiao WL, O'Brian CA, Murphy J, Kirschmeier P, Weinstein IB. Overproduction of protein kinase C causes disordered growth control in rat fibroplasts. Cell 1988; 52: 343-54. Persons DA, Wilkinson WO, Bell RM, Firm OJ. Altered growth regulation and enhanced tumorigencity of NIH 3T3 fibroplasts transfected with protein kinase C-l cDNA. CeU 1988; 52:447-58. Schlesxinger J. Allosteric regulation of the epidermal growth factor receptor kinase. J Cell Biol 1986; 103: 2067-72. Blumberg PM. Protein kinase C as the receptor for the phorbol ester tumor promoters. Cancer Res 1988; 48: 1-8. Nakadate T, Jeng AY, Blumberg P. Comparison of protein kinase C

46.

47. 48.

49.

50.

51. 52.

53.

functional assays to clarify mechanisms of inhibitor action. Biochem Pharmac 1988; 37: 1541-5. Smith JB, Smith L, Pettit GR. Bryostatins: potent, new mitogens that mimic phorbol ester promoters. Biochem Biophys Res Comm 1985; 132: 939-45. Dexter TM. Haetnopoietic growth factors. Br Med Bull 1989; 45: 337-^9. Roberts R, Gallagher J, Spooncer E, Allen TD, Bloomfteld F, Dexter TM. Heparin sulphate bound growth factors: a mechanism for stromal cell mediated haemopoiesis. Nature 1988; 332: 376-8. Barill GW, Dexter TM. An antagonist to platelet activating factor counteracts the tumouricidal action of alkyl lysophospholipids. Biochem Pharmac 1989; 38: 374-7. Nicolson GL. Organ specificity of tumor metastasis: role of preferential adhesion, invasion and growth of unique malignant cells at specific organ sites. Cancer Metast Rev 1988; 7: 143-88. Hart IR, Goode NT, Wilson RE. Molecular aspects of the metastatic cascade. Biochim Biophys Acta 1989; 989: 650-84. Orrenius S, McConchey DJ, Jones DP, Nicoteria P. Calcium-activated mechanisms in toxicity and programmed cell death. ISI Atlas of Science Pharmacology 1988; 319-24. McConchey DJ, Hartzell P, Duddy SK, Harkansson H, Orrenius S. 2, 3, 7, 8-tetrachloTodibenzo-p-dioxin kills immature thymocytes by calcium-mediated endonuclease activation. Science 1988; 242: 256-8. Correspondence to: Paul Workman, PluD., MRC Clinical Oncology Unit, Medical Research Council Centre, Cambridge CBN2 2QH, UK.