Rational Targeting of Notch Signaling in Cancer

P Rizzo1, C Osipo1, K Foreman1, T Golde2, B Osborne3 and L Miele1

1Breast Cancer Program, Cardinal Bernardin Cancer Center, Loyola University Chicago, Chicago, IL, USA; 2Department of Neurology, Mayo Clinic at Jacksonville, Jacksonville, FL, USA and 3Department of Veterinary and Animal Sciences, University of Massachusetts at Amherst, Amherst, MA, USA

Accumulating preclinical and clinical evidence supports a morbidity and mortality from cancer is the result of such pro-oncogenic function for Notch signaling in several solid recurrences. tumors, particularly but not exclusively in breast cancer. For decades, we have been searching for a ‘magic Notch inhibitory agents, such as c-secretase inhibitors, are bullet’ that could hit an elusive target only cancer cells being investigated as candidate cancer therapeutic agents. depend upon. Yet, to date most of pharmacological Interest in therapeutic modulation of the Notch pathway cancer treatments still rely on the relatively blunt has been increased by recent reports, indicating that its instruments of traditional chemotherapy. Targeted role is important in controlling the fate of putative ‘breast agents are slowly but steadily being added to our cancer stem cells’. However, as is the case for most therapeutic arsenal. These are directed largely at growth targeted therapies, successful targeting of Notch signaling factor receptors or their associated kinase activities, as in cancer will require a considerable refinement of our well as angiogenic growth factors. Monoclonal anti- understanding of the regulation of this pathway and its bodies (mAbs)such as trastuzumab, pertuzumab, effects in both normal and cancer cells. Notch signaling rituximab, bevacizumab and others, as well as tyrosine has bidirectional ‘cross talk’ interaction with multiple kinase inhibitors, such as imatinib, gefitinib, lapatinib other pathways that include candidate therapeutic targets. and others have gained clinical acceptance as part of Understanding these interactions will greatly increase combination regimens. Some of these agents have our ability to design rational combination regimens. To proven to be remarkably effective. Yet, even with these determine which patients are most likely to benefit from agents resistance is observed. treatment with Notch inhibitors, it will be necessary to Our increasing understanding of cancer biology is develop molecular tests to accurately measure pathway beginning to explain the reasons for these therapeutic activity in specific tumors. Finally, mechanism-based failures (Hanahan and Weinberg, 2000). First, decades toxicities will have to be addressed by a careful choice of signal transduction research have revealed that the of therapeutic agents, combinations and regimens. This receptors, enzymes and transcription factors that article summarizes the current state of the field, and regulate cell fate are virtually all connected into an briefly describes opportunities and challenges for Notch- amazingly complex network of cross-regulatory interac- targeted therapies in oncology. tions. The Internet rather than an old-fashioned Oncogene (2008) 27, 5124–5131; doi:10.1038/onc.2008.226 machine may be an apt metaphor for the workings of cell signaling. Even if one or several ‘routers’ are taken Keywords: Notch; cancer; targeted therapy out of commission, Internet traffic can be rerouted around them. To completely stop Internet traffic, a very large number of routers or cables need to be disabled. Second, we have learned that the cell-fate control system is not only interconnected but also highly redundant, Introduction: why target the Notch pathway? such that if a gene or protein is disabled, another can perform a similar function. Connectivity and redun- For those of us involved in the pharmacological dancy mean that if the system is stressed (for example, treatment of human cancers, whether experimental or by a targeted drug that blocks a growth factor receptor), standard of care, the most frustrating feature of the system can reset itself to a new status that works neoplastic diseases is the inherent ability of cancer cells around the block or replaces the disabled receptor to respond to therapy by adapting and/or selecting with a similar one. All this makes evolutionary sense. resistant subclones. In clinical practice, this results in Eukaryotic cells are very complex and delicate objects drug-resistant recurrence of disease, often in a meta- with tens of thousands of moving parts. To allow static form even after apparent eradication. Much of the survival in an often forbidding environment, evolution has created a robust system of cell-fate regulation that is highly resistant to the loss of one or even a few Correspondence: Professor L Miele, Breast Cancer Program, Cardinal Bernardin Cancer Center, Loyola University, 2160 South First components. Unfortunately, this adaptability is hijacked Avenue, Room 236, Maywood, Chicago, IL 60163, USA. by cancer cells, making the search for ‘magic bullets’ a E-mail: [email protected] very frustrating task. Even though cancers can become Targeting Notch in the treatment of cancer P Rizzo et al 5125 ‘addicted’ to specific pathways, they are remarkably bulk population of cancer cells through a process of adept at weaning themselves from their addictions if aberrant differentiation that recapitulates that of the need arises, and find alternatives to pathways hit normal tissues (Song and Miele, 2007). These ‘cancer by therapeutics. This is, in part, because the genomic stem cells’ are characterized by properties of normal instability resulting from loss of DNA repair check- stem cells, such as indefinite self-replication through points, in transformed cells makes these cells capable of asymmetric cell division, very slow proliferation rates selecting drug-resistant mutants by utilizing the built-in and resistance to toxic agents due in part to high-level connectivity and redundancy in signaling. Thus, the expression of ABC transporters. Whether cancer stem reason why ‘magic bullets’ have been so hard to find is cells are derived from the malignant transformation of that there are very few, if any, individual, irreplaceable normal tissue stem cells or from the ‘dedifferentiation’ targets that are uniquely associated with cancer cells. of normal non-stem cells is a matter of considerable Is there any hope of finding an Achilles’ heel in debate. What seems likely is that these cells are uniquely cancer? Perhaps. Not all pathways are created equal. capable of resisting anticancer agents, surviving for a Some of the most ancient cell-fate control pathways long time in a nearly quiescent status and produce which evolved with the initial appearance of multi- recurrences and metastases. Thus, a complete eradica- cellular organisms are conserved in all living organisms. tion of these cells will be necessary to attain a cure. This These pathways function as ‘Internet service providers’ will require targeting of pathways that participate in the in the signaling network. That is, they are multi- survival, replication and differentiation decisions in functional and control key nodes in the traffic of undifferentiated, pluripotent cells, such as the Wnt, signaling information that regulates differentiation, Hedgehog and Notch pathways (Song and Miele, 2007). proliferation and survival. The Wnt, Hedgehog and Indeed, there is significant evidence that Notch is Notch pathways belong to this ‘aristocracy’ of signaling relevant to the survival of breast cancer ‘stem cells’ systems. Not surprisingly, they are important in devel- (Farnie et al., 2005, 2007; Farnie and Clarke, 2007; opmental biology and come into play whenever critical Sansone et al., 2007). cell-fate decisions are made. Learning how to safely and Finally, the interaction between cancer cells and the effectively manipulate these pathways in cancer cells surrounding stroma is receiving increasing attention as a may bring us the next quantum leap in cancer therapy. key factor in tumor progression. ‘Tumor stroma’ The Notch pathway, the subject of this article, is a short- includes endothelial cells, necessary for tumor angiogen- range communication system in which contact between esis, fibroblasts that can produce growth factors and a cell expressing a membrane-associated ligand and a cytokines, as well as many subtypes of immunocytes, cell expressing a transmembrane receptor sends the from T cells to dendritic cells to NK cells that can affect receptor-expressing cell (and quite possibly both cells)a tumor progression either favorably or unfavorably. cell-fate regulatory signal. This signal takes the form of There is significant evidence that bidirectional inter- a cascade of transcriptional regulatory events that cellular communication involving Notch signals takes affects the expression of hundreds if not thousands of place between tumor cells and stromal cells in some genes, and has profound phenotypic consequences that malignancies (Jundt et al., 2002, 2004; Houde et al., are context dependent. The basic features of the 2004; Zeng et al., 2005), suggesting that targeting the pathway and the possible biological roles of Notch Notch–ligand interaction in endothelial cells can have signaling in human malignancies have been discussed in therapeutic applications (Yan and Plowman, 2007). several recent reviews (Allenspach et al., 2002; Miele, Moreover, Notch signaling has multifaceted functions 2006; Miele et al., 2006; Berman and Look, 2007; Koch in the immune system that need to be taken into account and Radtke, 2007; Roy et al., 2007; Shih and Wang, when planning therapeutic interventions (Dallman et al., 2007). In most cases, its deregulation has oncogenic 2005; McKenzie et al., 2005; Minato and Yasutomo, effects, with the notable exception of epidermal kerati- 2005; Minter et al., 2005; Tu et al., 2005). This article nocytes where Notch-1 functions as a tumor suppressor. discusses current efforts to develop Notch-targeted In several malignancies going from T-cell acute lympho- cancer therapeutics by both small molecules and blastic leukemia (Roy et al., 2007)to breast cancer biologics, including pros and cons of different targeting (Reedijk et al., 2005; Dickson et al., 2007)to melanoma strategies as well as identifying challenges to overcome. (Pinnix and Herlyn, 2007)to lung adenocarcinoma (Chen et al., 2007)and others, the inappropriate activation of Notch signaling results in signals that stimulate proliferation, restrict differentiation and pre- Targeting Notch in cancer vent apoptosis in cancer cells. As a result, Notch signaling inhibitors are being actively investigated for For targeting purposes, some features of the Notch the treatment of a variety of malignancies. pathway have unique relevance. First, the fact that the An additional reason for focusing on Notch and other signaling cascade triggered by Notch–ligand interactions ancient developmental pathways is that in recent years, does not include an enzymatic amplification step (for a distinct cellular hierarchy has been identified in example, a nucleotide cyclase or a kinase)means that hematopoietic and some solid tumors. Many cancers ‘signal intensity’ can be modulated very precisely by appear to contain a small population of pluripotent cellular regulatory mechanisms. As a result, the down- ‘stem cells’ or ‘tumor-initiating cells’ that give rise to the stream effects of Notch activation are exquisitely dose

Oncogene Targeting Notch in the treatment of cancer P Rizzo et al 5126 dependent (Miele et al., 2006). This means that complete Neuralized, EC shutdown of the pathway may not always be necessary Ligand-N Mindbomb Trans-endocytosis UQ to achieve a therapeutic effect. A second key feature is that the intracellular half-life of the active form of Notch is generally very short, in the order of minutes, * Receptor decoy * Ligand-blocking Fucosylation inh. though it may be longer in transformed cells (Weijzen mAb et al., 2002). The Notch signal is essentially a short pulse of gene regulation (Miele et al., 2006). This implies that Conformation-stabilizing mAb * ADAM * ADAM inh. sustained inhibition may not be always necessary and Numb α -adaptin UQ Deltex that intermittent inhibition may be successful. A third important feature to keep in mind is that the effects of UQ ligase Kurz Notch are remarkably context dependent. This means that Notch signals can be used for different purposes in Activation Arrestin- different cell types, and for each cell type the effects of Endocytosis mediated Degradation Degradation Endocytosis Notch manipulations need to be investigated without γ -secretase GSIs, GSMs preconceived assumptions. Systemic inhibition of Notch * signaling is likely to have a multitude of effects in

different cell types. Thus, for therapeutic purposes we Cross-talk, shall have to determine whether there is a level (or non-canonical timing)of Notch inhibition that is sufficient to attain pathways ? efficacy in disease control without causing intolerable adverse effects. An alternate strategy may be to identify NTC, monomeric Notch expression or dimeric miRNAs control siRNAs more context-specific targets within the Notch pathway, SKIP p300 MAML1 Notch selectively effective drug combinations (based on cell- target specific cross talk)or designing more selective delivery CBF-1 genes strategies for Notch inhibitors. Inhibition of Notch signaling can be achieved Figure 1 Diagram of putative therapeutic targets in the Notch theoretically at many different levels. Agents targeted pathway. (Red)Flat-tipped arrows indicate inhibitors. (Green) to some of these levels have been described, whereas Sharp-pointed arrows indicate stimulation of Notch degradation others remain hypothetical possibilities (Figure 1). It is by the Numb pathway (e.g., by inducing Numb expression)or the b-arrestin/Kurz pathway (e.g., by inducing expression of b-arrestin possible to interfere with Notch–ligand interactions by or Kurz). Asterisks indicate targets for which specific agents have using receptor decoys (Nickoloff et al., 2002), blocking been developed to date. Question marks indicate potential targets, ligand ubiquitination/trans-endocytosis (Pitsouli and yet unverified. Theoretically, inhibition of Notch signaling could be Delidakis, 2005)or Notch receptor fucosylation achieved by targeting ligand ubiqutination (UQ), ligand-receptor (Okajima and Irvine, 2002). It is possible to interfere interaction, NEC/NIC dissociation, a disintegrin and metalloprotease (ADAM)-mediated cleavage, receptor ubiquitination/endocytosis, with receptor activation by blocking ligand-induced g-secretase cleavage, assembly of the co-activator complex with conformational changes in Notch receptors (Gordon Notch and C-promoter binding factor 1 (CBF-1)and heterodimer- et al., 2007), receptor cleavage by ADAM proteases ization of DNA-bound Notch transcriptional complexes (NTCs). (Brou et al., 2000)or g-secretase (Kopan and Ilagan, In addition, expression of receptors and ligands as well as expression of selected downstream Notch targets could be 2004; Miele et al., 2006), as well as Notch mono- conceivably targeted (e.g., by modulating pathways that regulate ubiquitination (Gupta-Rossi et al., 2004). Finally, it is Notch or ligand expression, or by using RNA-targeted therapies to conceivable that Notch signaling could be inhibited by modulate translation of Notch targets. disrupting protein–protein interactions involved in Notch-dependent nuclear events (Nam et al., 2006, 2007), including assembly of co-activators with the of administration, oral bioavailability and low cost. In Notch transcriptional complex (NTC)and formation of general, small molecules can be dosed more precisely higher order DNA-bound complexes. As of this writing, than biologics because of their relatively short biological g-secretase inhibitors (GSIs)are in early clinical trials in half-life and simpler dose–response relationships, and various institutions, and mAbs that ‘lock’ Notch this may be important in a field where the therapeutic receptors in an inactive conformation by binding to window may be small. An additional potential advan- the ‘negative regulatory region’ (NRR)are in preclinical tage is the fact that a single agent can block the development (Li et al., 2008). Moreover, mAbs that activation of all four Notch homologues. In our target Notch ligand DLL4 (Ridgway et al., 2006)have experience and that of others, some solid tumors such been shown to inhibit Notch signaling in endothelial as breast cancers and melanomas co-express several cells and cause disorganized angiogenesis. These mAbs Notch homologues, and it is conceivable that redun- are currently being developed as antineoplastic agents dancy between them could blunt the effects of more (Noguera-Troise et al., 2007; Thurston et al., 2007; Yan selective inhibitors and lead to resistance. On the other and Plowman, 2007). hand, there is some evidence that different Notch Each of these approaches has potential advantages homologues may have opposite effects. For instance, and disadvantages. GSIs are active in several experi- Notch-2 appears to counteract the pro-oncogenic effects mental models and have the advantage of relative ease of Notch-1 and Notch-4 in breast cancer cells (O’Neill

Oncogene Targeting Notch in the treatment of cancer P Rizzo et al 5127 et al., 2007), and its expression correlates with better The potential disadvantages of biologics in this setting differentiated tumors (Parr et al., 2004)unlike that of include their generally complex dose–response curves in Notch-1 and Jagged-1, which correlate with poor vivo and their long biological half-lives. If intermittent prognosis (Reedijk et al., 2005; Dickson et al., 2007). inhibition of Notch signaling is desirable to minimize Similar differences have been noted in embryonal brain adverse effects, using an mAb that will remain in tumors (Fan et al., 2004), where conversely it is Notch-2 circulation for days or weeks may prove challenging in that has an oncogenic role. Another obvious potential terms of regimen design. Of course, the biological disadvantage of GSIs is off-target effects. Most GSIs in half-lives of mAbs can be modulated recombinant development are competitive inhibitors that mimic the engineering or generation of F(ab)2s, F(ab)s or even structure of short hydrophobic peptides. Given the fact single chain Fvs. that g-secretase has numerous targets other than Notch Indirect mechanisms of modulating Notch signaling receptors and a rather promiscuous cleavage specificity without directly engaging pathway members with drugs (Kopan and Ilagan, 2004), it is to be expected that GSIs or biologics can be envisioned. Modulating the expres- will not specifically inhibit the cleavage of Notch sion of Notch receptors, ligands or downstream receptors. It should be noted that off-target effects are mediators is conceivable. This approach will require a the rule rather than the exception for most approved detailed understanding of the transcriptional, transla- small molecule drugs. The mere fact that an agent can tional and post-translational regulation of Notch have off-target effects should not discourage its pathway component expression in specific cellular therapeutic development, unless it can be demonstrated contexts. At the moment, our knowledge in this area is that a more selective agent offers safety or efficacy still limited, with few exceptions. Expression of Jagged-1 advantages. The recent discovery that some nonsteroidal was reported to be increased by nuclear factor (NF)-kB antiinflammatory drugs and structurally related com- (Bash et al., 1999). Transcription of Notch-4 in pounds can allosterically modify the substrate specificity endothelial cells is induced by AP-1 (Wu et al., 2005) of g-secretase and decrease or increase the production of and the glucocorticoid receptor (Wu and Bresnick, Ab42 amyloid peptide (Kukar and Golde, 2008)raises 2007), which uncharacteristically act synergistically at the prospect that a class of g-secretase modifiers (GSM) the Notch-4 promoter. Notch-1 has been reported to can be developed that are capable of selectively upregulate its own transcription (Deftos et al., 1998) modulate the cleavage of Notch receptors as opposed and that of Notch-4 (Weijzen et al., 2002). The Ras to other substrates, though specificity for individual (Weijzen et al., 2002), AKT (Liu et al., 2003)and Notch homologues is unlikely. mitogen-activated protein kinases (Zeng et al., 2005) Is there a role for biologics (mAbs or recombinant pathways have been reported to stimulate Notch activity decoys)in Notch modulation? If specificity for an in some experimental models, suggesting that inhibitors individual receptor or ligand is desirable, biologics are of these pathways may affect Notch signaling in some more likely to deliver such specificity than small situations. Some isoforms of transcriptional regulator molecules. Particularly mAbs, if epitopes are carefully Ikaros inhibit Notch-dependent transcription in normal selected, can achieve exquisite specificity and very high and neoplastic T cells (Beverly and Capobianco, 2003; affinities. One situation in which a specific biologic may Bellavia et al., 2007; Kathrein et al., 2008). Corepressor be preferable is if the target has a relatively restricted SHARP (Oswald et al., 2002, 2005)associates with the expression pattern compared to other Notch pathway Notch transcriptional complex and recruits CtIP/CtBP members and/or can be replaced through redundancy in to block Notch-dependent transcription. If expression of nontarget tissues. This may limit mechanism-based Ikaros or SHARP could be modulated in specific cell toxicities. For example, the expression pattern of types, this would result in selective Notch inhibition. Notch-4 appears to be considerably more restricted In Drosophila, various microRNAs (miRNAs)regulate than that of Notch-1, suggesting that agents targeting Notch signaling through the 30-untranslated sequences Notch-4 may have less systemic toxicity than agents of numerous Notch target genes (Kwon et al., 2005; targeting Notch-1. The mAbs to DLL4 (Noguera-Troise Lai et al., 2005), and conversely miRNAs have been et al., 2007; Thurston et al., 2007; Yan and Plowman, suggested to mediate some Notch effects in Caenorhab- 2007)represent a promising approach in this direction. ditis elegans (Yoo and Greenwald, 2005), raising the These mAbs exploit the role of DLL4–Notch interac- possibility that RNA-based therapeutics (siRNAs or tions in endothelial cell activation. In xenograft models, miRNAs)could be used to modulate Notch signaling by they appear to target tumor angiogenesis by causing targeting specific subsets of Notch targets. At the post- disorganized blood vessel development but do not have transcriptional level, Notch-1 protein levels and activity the same side effects caused by systemic Notch are regulated by several ubiquitin ligases that mediate inhibitors. An additional potential advantage of bio- either the degradation of Notch by polyubiquitination logics and mAbs in particular is the ability to conjugate or conversely, its activation by monoubiqutination them with radionuclides or toxins to selectively target (Miele et al., 2006). One of these ligases (SEL10/Fbw7/ cells that overexpress their targets. In the case of Notch Ago/hCDC4), responsible for the polyubiquitination pathway components, this approach would require a and degradation of nuclear Notch-1 is mutated in some target that is at least relatively specific to cancer cells or T-cell acute lymphoblastic leukemia (T-ALL)cases, significantly overexpressed by them (for example, resulting in prolonged Notch-1 signals (Malyukova Jagged-2 in multiple myeloma (Houde et al., 2004). et al., 2007; O’neil et al., 2007; Thompson et al., 2007).

Oncogene Targeting Notch in the treatment of cancer P Rizzo et al 5128 Desensitization of Notch-1 signaling by nonvisual of clinically relevant toxicities and the development of b-arrestin and Kurz (Mukherjee et al., 2005)or strategies to prevent or ameliorate them. endocytosis and degradation of Notch-1 mediated by Early clinical experience with GSIs indicates that the Numb (Santolini et al., 2000; Pece et al., 2004)are also main adverse event in patients is dose-limiting secretory potential avenues to regulate Notch signaling if expres- diarrhea caused by goblet cell metaplasia of the small sion of Kurz or Numb can be modulated in cell-specific intestine, which was first observed in preclinical models fashions. (Wong et al., 2004). The absence of myelotoxicity is Are there situations in which stimulation of Notch welcome news in the setting of cancer chemotherapy. In signaling can be envisioned as a therapeutic strategy in mice, other adverse effects of systemic GSI treatment cancer? The strongest evidence for a tumor-suppressive include reversible thymic suppression (Wong et al., 2004) role of Notch-1 is in epidermal keratinocytes (Nicolas and, in our hands, reversible hair depigmentation. Hair et al., 2003; Koch and Radtke, 2007), because kerati- loss in dose-escalation experiments is an indication that a nocyte-targeted Notch-1 knockout in mice increases toxic dose has been reached and is associated with chemical carcinogenesis in the skin. However, it is diarrhea and weight loss. Skin tumors have not been difficult to envision chronic treatment of basal cell observed in humans and we have not observed them in carcinomas with Notch ligands, given that these lesions mice treated for several weeks, though patches of mild, can be generally cured by excision. In the context of reversible hyperkeratosis are observed in nude mice. It is tumor immunology, it is conceivable that Notch possible that life-long and/or complete shutdown of activation in T cells or other immunocytes may promote Notch signaling is required for more severe skin tumor rejection induced by a therapeutic tumor vaccine. proliferative phenotypes. In both preclinical models This may be achieved, for example, by expressing a and clinical studies, intermittent rather than continuous Notch ligand in antigen-presenting cells such as oral administration of GSIs greatly ameliorates the dendritic cells. We still do not have a clear under- intestinal toxicity, presumably because it allows at least standing of the multiple functions of Notch signaling in some intestinal stem cells to correctly differentiate as the peripheral immune system, where it could modulate enterocytes. Parenteral administration of GSIs in mouse Th1-type (Minter et al., 2005)or Th-2-type (Tu et al., xenograft models in our hands was associated with less 2005)T-cell responses. It is conceivable that slightly severe side effects, and doses that caused significant different strategies for stimulation or slightly different antineoplastic effects did not cause diarrhea or weight signal strengths could produce either effect. Systemic loss (Nickoloff et al., 2005), (Rizzo et al., submitted for delivery of a Notch stimulator may pose pharmacolo- publication). In mice, GSIs have immunosuppressive gical challenges, because physiologically Notch ligands effects that may be undesirable under some circum- are transmembrane proteins. The mechanism of Notch stances. However, these effects may find clinical applica- activation requires trans-endocytosis of the Notch tions of their own in autoimmune disorders such as extracellular subunit NEC into the ligand-expressing cell, multiple sclerosis (Minter et al., 2005), and could which in turn requires monoubiquitination of ligand potentially be indirectly useful in oncology in the intracellular tails (Miele et al., 2006). This is because the treatment of graft-versus-host disease after bone marrow dissociation of NEC from the transmembrane subunit transplantation (Minter et al., 2005). GSIs are not NTM requires significant mechanical strength to disrupt significantly myelotoxic, making such a potential appli- hydrophobic interactions in the NRR at the interaction cation at least theoretically feasible. Whether prolonged site between the two subunits (Gordon et al., 2007). Notch inhibition by agents that pass the blood–brain Thus, cell-associated or at least solid phase-bound barriers is associated with neurological toxicity is ligands (for example, ligand-coated beads)are expected unknown at this time, but possible based on mouse data to be more effective than soluble ligands. Having said (Wang et al., 2004). At this time, it would appear that that, oligopeptides mimicking the conserved DSL region relatively short-term (days to weeks)systemic suppres- of Notch ligands can activate Notch receptors, albeit at sion of Notch signaling is consistent with a reasonable high concentrations (Nickoloff et al., 2002). Fc fusion safety profile, especially in the context of cancer therapy. constructs that may associate with Fc receptors may be Perhaps the best way to use systemic Notch inhibitors in another strategy, particularly if the objective is immuno- cancer may turn out to be as multiple cycles separated by modulation. ‘drug holidays’. Chronic, long-term suppression of Notch signaling is likely to require more selective agents, such as DLL4 antibodies in the context of antiangiogenic therapy, or more targeted delivery strategies of Notch Toxicity inhibitors (for example, encapsulated within tumor- targeted nanocarriers). Although there is wide agreement that targeting cell-fate modulatory pathways is one of the most attractive new avenues in experimental cancer therapy, it would be Looking to the future: rational combinations including naive to expect the modulation of such ancient, Notch inhibitors and individualized medicine pervasively important pathways to have no adverse effects. Thus, key components of studying the possibility Long-term therapeutic success in cancer is rarely of targeting Notch in cancer must be the identification achieved with monotherapy, and even targeting

Oncogene Targeting Notch in the treatment of cancer P Rizzo et al 5129 developmental pathways such as Notch will most likely effective in Her2/Neu overexpressing, trastuzumab- require the development of combination regimens. resistant tumors. An unexpected but potentially useful Traditionally such regimens have been produced observation was that co-treatment with tamoxifen through a process of ‘clinical trial and error’, often greatly alleviated the intestinal side effects of orally based on limited mechanistic information. Clinical administered GSIs, suggesting that this combination experimentation will always be necessary, because no may be not only more effective but also safer than single preclinical model completely recapitulates a patient. agent GSI treatment. However, the more complete and accurate our mechan- To design the best combination regimens including istic understanding of how the pathways we target cross Notch inhibitors, indication-specific studies will have to talk with each other, the less guesswork will be involved be performed. These studies will need to include in designing future therapeutic regimens. The high a whole range of approaches, from simple model evolutionary conservation of developmental pathways organisms such as Drosophila or zebrafish to in vitro means that information from simpler model organisms and in vivo studies in mammalian models to identify is likely to be reliably predictive of human pathophy- which genetic and epigenetic factors interact with siology. On the other hand, the context dependence of Notch signaling. These findings will require validation Notch signaling will require each specific cancer type to by studies of primary clinical specimens. Ultimately, be studied independently, without preconceived notions. the best use of these new therapeutic targets, as is the Our knowledge is still considerably incomplete, but case for most new targeted agents, will be in the context evidence accumulated so far suggests that some combi- of ‘individualized medicine’. It will be necessary to nation regimens involving Notch inhibitors deserve identify groups of patients and/or subtypes of further investigation. The examples that follow are not cancer who are most likely to benefit from Notch meant to be all-inclusive. inhibitors. To that end, we will have to determine: Inhibitors of the PI3-kinase–AKT–mTOR pathway (1)which cancers and specific subtypes are characterized may be useful in combination with Notch inhibitors, by active Notch signaling; (2)what role do specific and there is evidence that this strategy may components of Notch signaling perform in these cancers reverse resistance to GSIs in T-ALL that carry PTEN (for example, Notch-2 versus Notch-1), and whether inactivating mutation (Palomero et al., 2007). Whether global or selective Notch modulation is most desirable this strategy can be successful in other cancers and (3)what genes or pathways cross talk with Notch in characterized by loss of PTEN is still unclear. The specific cancers, indicating targets for combination complex cross talk between Notch and NF-kB suggests regimens. High-throughput system biology and bio- that at least in some circumstances drugs that inhibit informatics will be important in this task. Simply NF-kB activity directly or indirectly could be success- determining expression levels of receptors and ligands fully combined with Notch inhibitors (Wang et al., in clinical specimens will not necessarily identify 2006a, b; Vilimas et al., 2007; Osipo et al., 2008). As prospective responders. Because of extensive cross talk intracellular Notch is degraded by the proteasome, and between Notch and other pathways, there is no simple accumulates in cells treated with proteasome inhibitors, correlation between expression and Notch activity. it is possible that these agents may benefit from the Thus, it will be important to develop accurate molecular addition of a GSI. As DLL4 mAb appear to be effective tests that measure the level of pathway activity in vivo, independently of VEGF, they may be useful in possibly based on expression levels of multiple genes in combination with agents that block the VEGF pathway the pathway. With the tools of today’s cancer biology, such as bevacizumab. Our data (Rizzo et al., submitted these tasks are not as daunting as they would have been for publication; Osipo et al., submitted for publication) a few years ago. And the payoff for these efforts may be indicate that: (1)Notch signaling is prominently multiple new treatments for a whole range of human regulated by estrogen and Her2/Neu; (2)estrogen malignancies. receptor-a (ERa)-negative and Her2/Neu-negative cancers have higher Notch activity and may respond to Notch inhibitors; (3)combinations of selective estrogen Acknowledgements receptor modulators (SERMS)plus GSIs are particularly effective in ERa-positive tumors; (4)Combinations This work was supported by NIH grant P01 AG2553101 of trastuzumab plus GSIs or receptor tyrosine kinase and DOD IDEA grant W81XWH-04-1-0478. We are grateful inhibitors plus GSIs are synergistic in vitro and may be to Antonio Pannuti for helpful discussions.

References

Allenspach EJ, Maillard I, Aster JC, Pear WS. (2002). Notch signaling Bellavia D, Mecarozzi M, Campese AF, Grazioli P, Gulino A, in cancer. Cancer Biol Ther 1: 466–476. Screpanti I. (2007). Notch and Ikaros: not only converging players Bash J, Zong WX, Banga S, Rivera A, Ballard DW, Ron Y et al. in T cell leukemia. Cell Cycle 6: 2730–2734. (1999). Rel/NF-kappaB can trigger the Notch signaling pathway by Berman JN, Look AT. (2007). Targeting transcription inducing the expression of Jagged1, a ligand for Notch receptors. factors in acute leukemia in children. Curr Drug Targets 8: EMBO J 18: 2803–2811. 727–737.

Oncogene Targeting Notch in the treatment of cancer P Rizzo et al 5130 Beverly LJ, Capobianco AJ. (2003). Perturbation of Ikaros isoform Liu ZJ, Shirakawa T, Li Y, Soma A, Oka M, Dotto GP et al. (2003). selection by MLV integration is a cooperative event in Notch(IC)- Regulation of Notch1 and Dll4 by vascular endothelial growth induced T cell leukemogenesis. Cancer Cell 3: 551–564. factor in arterial endothelial cells: implications for modulating Brou C, Logeat F, Gupta N, Bessia C, LeBail O, Doedens JR et al. arteriogenesis and angiogenesis. Mol Cell Biol 23: 14–25. (2000). A novel proteolytic cleavage involved in Notch signaling: the Malyukova A, Dohda T, von der LN, Akhondi S, Corcoran M, role of the disintegrin-metalloprotease TACE. Mol Cell 5: 207–216. Heyman M et al. (2007). The tumor suppressor gene hCDC4 is Chen Y, De Marco MA, Graziani I, Gazdar AF, Strack PR, Miele L frequently mutated in human T-cell acute lymphoblastic leukemia et al. (2007). Oxygen concentration determines the biological effects with functional consequences for Notch signaling. Cancer Res 67: of NOTCH-1 signaling in adenocarcinoma of the lung. Cancer Res 5611–5616. 67: 7954–7959. McKenzie GJ, Khan M, Briend E, Stallwood Y, Champion BR. Dallman MJ, Smith E, Benson RA, Lamb JR. (2005). Notch: control (2005). Notch: a unique therapeutic target for immunomodulation. of lymphocyte differentiation in the periphery. Curr Opin Immunol Expert Opin Ther Targets 9: 395–410. 17: 259–266. Miele L. (2006). Notch signaling. Clin Cancer Res 12: 1074–1079. Deftos ML, He Y-W, Ojata EW, Bevan MJ. (1998). Correlating Notch Miele L, Golde T, Osborne B. (2006). Notch signaling in cancer. signaling with thymocyte maturation. Immunity 9: 777–786. Curr Mol Med 6: 905–918. Dickson BC, Mulligan AM, Zhang H, Lockwood G, O’Malley FP, Minato Y, Yasutomo K. (2005). Regulation of acquired immune Egan SE et al. (2007). High-level JAG1 mRNA and protein predict system by Notch signaling. Int J Hematol 82: 302–306. poor outcome in breast cancer. Mod Pathol 20: 685–693. Minter LM, Turley DM, Das P, Shin HM, Joshi I, Lawlor RG et al. Fan X, Mikolaenko I, Elhassan I, Ni X, Wang Y, Ball D et al. (2004). (2005). Inhibitors of gamma-secretase block in vivo and in vitro Notch1 and Notch2 have opposite effects on embryonal brain T helper type 1 polarization by preventing Notch upregulation of tumor growth. Cancer Res 64: 7787–7793. Tbx21. Nat Immunol 6: 680–688. Farnie G, Brennan K, Clarke RB, Bundred NJ. (2005). Ductal Mukherjee A, Veraksa A, Bauer A, Rosse C, Camonis J, Artavanis- Carcinoma in situ (DCIS)mammosphere formation effect of Tsakonas S. (2005). Regulation of Notch signalling by non-visual epidermal growth factor (EGF)and Notch signaling pathways on beta-arrestin. Nat Cell Biol 7: 1191–1201. self-renewal capacity. Breast Cancer Res Treat 94(Suppl. 1): S14. Nam Y, Sliz P, Pear WS, Aster JC, Blacklow SC. (2007). Cooperative Farnie G, Clarke RB. (2007). Mammary stem cells and breast cancer— assembly of higher-order Notch complexes functions as a switch to role of Notch signalling. Stem Cell Rev 3: 169–175. induce transcription. Proc Natl Acad Sci USA 104: 2103–2108. Farnie G, Clarke RB, Spence K, Pinnock N, Brennan K, Anderson NG Nam Y, Sliz P, Song L, Aster JC, Blacklow SC. (2006). Structural basis et al. (2007). Novel cell culture technique for primary ductal for cooperativity in recruitment of MAML coactivators to Notch carcinoma in situ: role of Notch and epidermal growth factor transcription complexes. Cell 124: 973–983. receptor signaling pathways. J Natl Cancer Inst 99: 616–627. Nickoloff BJ, Hendrix MJ, Pollock PM, Trent JM, Miele L, Qin JZ. Gordon WR, Vardar-Ulu D, Histen G, Sanchez-Irizarry C, Aster JC, (2005). Notch and NOXA-related pathways in melanoma cells. Blacklow SC. (2007). Structural basis for autoinhibition of Notch. J Investig Dermatol Symp Proc 10: 95–104. Nat Struct Mol Biol 14: 295–300. Nickoloff BJ, Qin JZ, Chaturvedi V, Denning MF, Bonish B, Miele L. Gupta-Rossi N, Six E, LeBail O, Logeat F, Chastagner P, Olry A et al. (2002). Jagged-1 mediated activation of Notch signaling induces (2004). Monoubiquitination and endocytosis direct gamma-secre- complete maturation of human keratinocytes through NF-kappaB tase cleavage of activated Notch receptor. J Cell Biol 166: 73–83. and PPARgamma. Cell Death Differ 9: 842–855. Hanahan D, Weinberg RA. (2000). The hallmarks of cancer. Cell 100: Nicolas M, Wolfer A, Raj K, Kummer JA, Mill P, Van Noort M et al. 57–70. (2003). Notch1 functions as a tumor suppressor in mouse skin. Houde C, Li Y, Song L, Barton K, Zhang Q, Godwin J et al. (2004). Nat Genet 33: 416–421. Over-expression of the NOTCH ligand JAG2 in malignant plasma cells Noguera-Troise I, Daly C, Papadopoulos NJ, Coetzee S, Boland P, from multiple myeloma patients and cell lines. Blood 104: 3697–3704. Gale NW et al. (2007). Blockade of Dll4 inhibits tumour growth by Jundt F, Anagnostopoulos I, Forster R, Mathas S, Stein H, Dorken B. promoting non-productive angiogenesis. Novartis Found Symp 283: (2002). Activated Notch1 signaling promotes tumor cell prolifera- 106–120. tion and survival in Hodgkin and anaplastic large cell lymphoma. O’neil J, Grim J, Strack P, Rao S, Tibbitts D, Winter C et al. (2007). Blood 99: 3398–3403. FBW7 mutations in leukemic cells mediate NOTCH pathway Jundt F, Probsting KS, Anagnostopoulos I, Muehlinghaus G, activation and resistance to {gamma}-secretase inhibitors. J Exp Chatterjee M, Mathas S et al. (2004). Jagged1-induced Notch Med 204: 1813–1824. signaling drives proliferation of multiple myeloma cells. Blood 103: O’Neill CF, Urs S, Cinelli C, Lincoln A, Nadeau RJ, Leon R et al. 3511–3515. (2007). Notch2 signaling induces apoptosis and inhibits human Kathrein KL, Chari S, Winandy S. (2008). Ikaros directly represses the MDA-MB-231 xenograft growth. Am J Pathol 171: 1023–1036. Notch target gene Hes1 in a leukemia T cell line: implications for Okajima T, Irvine KD. (2002). Regulation of Notch signaling by CD4 regulation. J Biol Chem 283: 10476–10484. O-linked fucose. Cell 111: 893–904. Koch U, Radtke F. (2007). Notch and cancer: a double-edged sword. Osipo C, Golde TE, Osborne BA, Miele LA. (2008). Off the beaten Cell Mol Life Sci 64: 2746–2762. pathway: the complex cross talk between Notch and NF-kappaB. Kopan R, Ilagan MX. (2004). Gamma-secretase: proteasome of the Lab Invest 88: 11–17. membrane? Nat Rev Mol Cell Biol 5: 499–504. Oswald F, Kostezka U, Astrahantseff K, Bourteele S, Dillinger K, Kukar T, Golde TE. (2008). Possible mechanisms of action of NSAIDs Zechner U et al. (2002). SHARP is a novel component of the Notch/ and related compounds that modulate gamma-secretase cleavage. RBP-Jkappa signalling pathway. EMBO J 21: 5417–5426. Curr Top Med Chem 8: 47–53. Oswald F, Winkler M, Cao Y, Astrahantseff K, Bourteele S, Knochel Kwon C, Han Z, Olson EN, Srivastava D. (2005). MicroRNA1 W et al. (2005). RBP-J{kappa}/SHARP recruits CtIP/CtBP inﬂuences cardiac differentiation in Drosophila and regulates Notch corepressors to silence Notch target genes. Mol Cell Biol 25: signaling. Proc Natl Acad Sci USA 102: 18986–18991. 10379–10390. Lai EC, Tam B, Rubin GM. (2005). Pervasive regulation of Drosophila Palomero T, Sulis ML, Cortina M, Real PJ, Barnes K, Ciofani M et al. Notch target genes by GY-box-, Brd-box-, and K-box-class (2007). Mutational loss of PTEN induces resistance to NOTCH1 microRNAs. Genes Dev 19: 1067–1080. inhibition in T-cell leukemia. Nat Med 13: 1203–1210. Li K, Li Y, Wu W, Gordon WR, Chang DW, Lu M et al. (2008). Parr C, Watkins G, Jiang WG. (2004). The possible correlation of Modulation of Notch signaling by antibodies speciﬁc for the Notch-1 and Notch-2 with clinical outcome and tumour clinico- extracellular negative regulatory region of Notch3. J Biol Chem 283: pathological parameters in human breast cancer. Int J Mol Med 14: 8046–8054. 779–786.

Oncogene Targeting Notch in the treatment of cancer P Rizzo et al 5131 Pece S, Serresi M, Santolini E, Capra M, Hulleman E, Galimberti V Vilimas T, Mascarenhas J, Palomero T, Mandal M, Buonamici S, et al. (2004). Loss of negative regulation by Numb over Notch is Meng F et al. (2007). Targeting the NF-kappaB signaling pathway relevant to human breast carcinogenesis. J Cell Biol 167: 215–221. in Notch1-induced T-cell leukemia. Nat Med 13: 70–77. Pinnix CC, Herlyn M. (2007). The many faces of Notch signaling in Wang Y, Chan SL, Miele L, Yao PJ, Mackes J, Ingram DK et al. skin-derived cells. Pigment Cell Res 20: 458–465. (2004). Involvement of Notch signaling in hippocampal synaptic Pitsouli C, Delidakis C. (2005). The interplay between DSL plasticity. Proc Natl Acad Sci USA 101: 9458–9462. proteins and ubiquitin ligases in Notch signaling. Development Wang Z, Banerjee S, Li Y, Rahman KM, Zhang Y, Sarkar FH. 132: 4041–4050. (2006a). Down-regulation of Notch-1 inhibits invasion by inactiva- Reedijk M, Odorcic S, Chang L, Zhang H, Miller N, McCready DR tion of nuclear factor-kappaB, vascular endothelial growth factor, et al. (2005). High-level coexpression of JAG1 and NOTCH1 is and matrix metalloproteinase-9 in pancreatic cancer cells. Cancer observed in human breast cancer and is associated with poor overall Res 66: 2778–2784. survival. Cancer Res 65: 8530–8537. Wang Z, Zhang Y, Banerjee S, Li Y, Sarkar FH. (2006b). Inhibition of Ridgway J, Zhang G, Wu Y, Stawicki S, Liang WC, Chanthery Y et al. nuclear factor kappab activity by genistein is mediated via Notch-1 (2006). Inhibition of Dll4 signalling inhibits tumour growth by signaling pathway in pancreatic cancer cells. Int J Cancer 118: 1930–1936. deregulating angiogenesis. Nature 444: 1083–1087. Weijzen S, Rizzo P, Braid M, Vaishnav R, Jonkheer SM, Zlobin A Roy M, Pear WS, Aster JC. (2007). The multifaceted role of Notch in et al. (2002). Activation of Notch-1 signaling maintains the cancer. Curr Opin Genet Dev 17: 52–59. neoplastic phenotype in human Ras-transformed cells. Nat Med 8: Sansone P, Storci G, Giovannini C, Pandolﬁ S, Pianetti S, Taffurelli M 979–986. et al. (2007). p66Shc/Notch-3 interplay controls self-renewal and Wong GT, Manfra D, Poulet FM, Zhang Q, Josien H, Bara T et al. hypoxia survival in human stem/progenitor cells of the mammary (2004). Chronic treatment with the gamma-secretase inhibitor LY- gland expanded in vitro as mammospheres. Stem Cells 25: 807–815. 411,575 inhibits beta-amyloid peptide production and alters Santolini E, Puri C, Salcini AE, Gagliani MC, Pelicci PG, Tacchetti C lymphopoiesis and intestinal cell differentiation. J Biol Chem 279: et al. (2000). Numb is an endocytic protein. J Cell Biol 151: 12876–12882. 1345–1352. Wu J, Bresnick EH. (2007). Glucocorticoid and growth factor Shih I, Wang TL. (2007). Notch signaling, gamma-secretase inhibitors, synergism requirement for Notch4 chromatin domain activation. and cancer therapy. Cancer Res 67: 1879–1882. Mol Cell Biol 27: 2411–2422. Song LL, Miele L. (2007). Cancer stem cells—an old idea that’s new Wu J, Iwata F, Grass JA, Osborne CS, Elnitski L, Fraser P et al. again: implications for the diagnosis and treatment of breast cancer. (2005). Molecular determinants of NOTCH4 transcription in Expert Opin Biol Ther 7: 431–438. vascular endothelium. Mol Cell Biol 25: 1458–1474. Thompson BJ, Buonamici S, Sulis ML, Palomero T, Vilimas T, Basso G Yan M, Plowman GD. (2007). Delta-like 4/Notch signaling and its et al. (2007). The SCFFBW7 ubiquitin ligase complex as a tumor therapeutic implications. Clin Cancer Res 13: 7243–7246. suppressor in T cell leukemia. J Exp Med 204: 1825–1835. Yoo AS, Greenwald I. (2005). LIN-12/Notch activation leads to Thurston G, Noguera-Troise I, Yancopoulos GD. (2007). The delta microRNA-mediated down-regulation of Vav in C. elegans. Science paradox: DLL4 blockade leads to more tumour vessels but less 310: 1330–1333. tumour growth. Nat Rev Cancer 7: 327–331. Zeng Q, Li S, Chepeha DB, Giordano TJ, Li J, Zhang H et al. (2005). Tu L, Fang TC, Artis D, Shestova O, Pross SE, Maillard I et al. (2005). Crosstalk between tumor and endothelial cells promotes tumor Notch signaling is an important regulator of type 2 immunity. J Exp angiogenesis by MAPK activation of Notch signaling. Cancer Cell Med 202: 1037–1042. 8: 13–23.

Oncogene