cell cycle inhibitors in normal and tumor stem cells

Cell cycle inhibitors in normal and tumor stem cells Tao Cheng*,1 1Department of Radiation Oncology, University of Pittsburgh School of Medicine and University of Pittsburgh Cancer Institute, Pittsburgh, PA 15213, USA Emerging data suggest that stem cells may be one of the key elements in normal tissue regeneration and cancer development, although they are not necessarily the same entity in both scenarios. As extensively demonstrated in the hematopoietic system, stem cell repopulation is hierarchically organized and is intrinsically limited by the intracellular cell cycle inhibitors. Their inhibitory effects appear to be highly associated with the differentia- tion stage in stem/progenitor pools. While this negative regulation is important for maintaining homeostasis, especially at the stem cell level under physiological cues or pathological insults, it constrains the therapeutic use of adult stem cells in vitro and restricts endogenous tissue repair after injury. On the other hand, disruption of cell cycle inhibition may contribute to the formation of the so- called ‘tumor stem cells’ (TSCs) that are currently hypothesized to be partially responsible for tumorigenesis and recurrence of cancer after conventional therapies. Therefore, understanding how cell cycle inhibitors control stem cells may offer new strategies not only for therapeutic manipulations of normal stem cells but also for novel therapies selectively targeting TSCs. Oncogene (2004) 23, 7256–7266. doi:10.1038/sj.onc.1207945 Keywords: adult stem cell; hematopoietic stem cell; tumor stem cell; tissue regeneration; self-renewal; cell cycle; cyclin-dependent kinase inhibitor; p21; p27; p18; p16; Rb protein Introduction The therapeutic efficacy of stem cells largely relies on their ability to replicate. Therefore, strategies to manipulate stem cells require an understanding of their cell cycle control. In this review, the distinct cell cycle kinetics of adult stem cells will be introduced and briefly followed by the current understanding of general cell cycle regulation in mammalian cells. The focus will be placed on the specific impact of the cell cycle inhibitors, namely the roles of cyclin-dependent kinase inhibitors (CKIs) on adult stem/progenitor cells. While many stem cell types have been or are being defined, the hemato- poietic stem cell (HSC) remains one of the best-studied stem cell types and therefore data have been largely obtained from HSCs partially in comparison with the other stem cell types such as the neural stem cell (NSC). The distinct impacts of different CKIs in stem cell populations underscore the crucial role that cell cycle inhibitors play in stem cell regulation and offer new insights into the mechanisms of cancer development, especially in light of the concept of ‘tumor stem cell’ (TSC) (or described as ‘cancer stem cell’ in general or ‘leukemic stem cells’ in leukemias by some other investigators in the field). The biochemical pathways of general cell cycle regulation will not be detailed in the current review, since they have been extensively reviewed elsewhere (Pardee, 1989; Sherr and Roberts, 1999; Sherr, 2000). Stem cell proliferation with a limited rate By definition, stem cells have the ability to reproduce themselves indefinitely while possessing the potential to differentiate into multiple lineages (or a single lineage in a few tissue types) via transit amplifying cells (progeni- tors). However, the proliferative rate of stem cells in vivo under physiological conditions is much slower com- pared to that of intermediate progenies, including the progenitor cells and proliferating precursors. In the hematopoietic system, an increase in stem cell divisions can take place (Pawliuk et al., 1996; Oosten- dorp et al., 2000) under activating conditions, such as after transplantation or the depletion of cycling cells using the S-phase toxin (5-fluorouracil or hydroxyurea) (Harrison and Lerner, 1991; Uchida et al., 1997). However, the relative quiescence of the HSC pool appears to be essential to prevent premature depletion under conditions of physiologic stress over the lifetime of the organism (Cheng et al., 2000a, b). The highly regulated proliferation of HSC occurs at a very limited rate under homeostatic conditions. It had once been hypothesized that the stem cell quiescence reflects a complete cell cycle arrest of the majority cells in the stem cell compartment, termed the ‘clonal succession model’ (Kay, 1965). While the retrovirus-based clonal marking studies revealed few active stem cell clones at a given time, which supports the clonal succession theory (Lemischka et al., 1986), this view has been challenged *Correspondence: T Cheng, Hillman Cancer Center, Research Pavilion, Office suite 2.42e, 5117 Center Avenue, Pittsburgh, PA 15213-1863, USA; E-mail: chengt@msx.upmc.edu Oncogene (2004) 23, 7256–7266 & 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc by the competitive repopulation model (Harrison et al., 1988) or 5-bromodeoxyuridine (BrdU) incorporation in the defined stem cell pools with a simulation methodol- ogy (Cheshier et al., 1999). Using BrdU labeling experiments, cell cycle lengths were estimated at approximately 30 days in small rodents (Bradford et al., 1997) and it was found that only about 8% of the cells are in cell cycle daily (Cheshier et al., 1999). Similar analyses using population kinetics estimated that stem cells replicate once per 10 weeks in cats (Abkowitz et al., 1996). In non-human primates, the rate of cell replication in the stem cell pool was estimated to be once per year (Mahmud et al., 2001). In contrast, the essential feature of the hematopoietic progenitor cell (HPC) population is that it irreversibly develops into maturing cells, and in the process undergoes multiple rapid cell divisions. The progenitor cell pool essentially operates as a cellular amplification machine, generating many differentiated cells from the few cells entering the system (Potten, 1997), and it is therefore directly responsible for the number of terminally differentiated cells. For this reason, the progenitor cell pool is also termed the transit, amplifying cell pool. The differences between the stem and progenitor cell populations have been regarded as phenotypic distinctions marking the stage of a cell within the hematopoietic cascade. However, an alternative model was recently put forward proposing that the specific position in a cell cycle determines whether a primitive cell functions as a stem or a progenitor cell (Quesenberry et al., 2002). In this hypothetical model, stimuli received at distinct positions in the cell cycle provoke proliferation/differentiation or not, and thereby yield either HSC or HPC outcomes. This seems to contradict the traditional view of ‘hierarchy’ within the hematopoietic differentiation. Nevertheless, in either model, ‘stemness’ is associated with the limited rate of the cell proliferation. The slow cycling feature seems to be a common feature in most adult stem cell types if not all (Potten, 1997). In the central nervous system (CNS), evidence suggests that the proliferative pools of adult neural progenitors are derived from a quiescent multipotent neural precursor or NSC (Bonfanti et al., 2001; Palmer et al., 2001). For example, if the proliferative zone containing the lineage-committed neuronal progenitor cells (NPC) is ablated, it can be repopulated from a small number of quiescent NSCs (Morshead et al., 1994; Doetsch et al., 1999). Perhaps largely owing to this quiescence, endogenous NSCs do not produce complete recovery in cases of severe injury, though they do participate in self-repair after brain damage (Horner et al., 2000). In the mouse dermal stem cell population, there are about fourfold fewer cells in S-G2/M phases in the stem cell population compared with the progenitor pool, though both cell populations constantly proceed through the cell cycle (Dunnwald et al., 2003). In addition, epithelial stem cells in mouse cornea (Cotsar- elis et al., 1989) or mouse hair follicle (Morris et al., 2004) appear to be slowly cycling as well. The relative quiescence of stem cells may prevent their premature exhaustion in vivo, but it has been considered to be one of the hurdles in the context of the in vitro expansion necessary for stem cell transplantation and gene therapy. Methods for inducing stem cell prolifera- tion have long been sought as a means to expand the population of cells capable of repopulating the bone marrow of ablated hosts and to render stem cells transducable with virus-based gene transfer vectors. Although great effort has been made to directly expand stem cells using different combinations of hematopoietic growth factors (cytokine cocktails), no culture system has been applied successfully in clinical settings. This is due, in part, to the lack of proof that any of the culture conditions support expansion of a long-term repopulat- ing HSC in humans (Verfaillie, 2002). Although data suggest that under certain specific conditions murine HSCs may divide in vitro, net expansion is achieved in limited fashion and is associated with and often dominated by cellular differentiation (Ema et al., 2000; Uchida et al., 2003). It remains unclear which combina- tion of hematopoietic cytokines is specifically selective for stem cell proliferation without differentiation. So far, there is no cytokine combination that can achieve the effect of stem cell expansion by the HoxB4 protein (Antonchuk et al., 2002). A recent promising study demonstrated a potent effect of purified Wnt3a protein in the expansion of mouse HSC in vitro. But the effect appeared to require a strong survival element since the net expansion could be only achieved on the HSCs from the Bcl2 transgenic mice (Willert et al., 2003). In short, the dichotomy of the relative resistance to proliferative signals by stem cells and the brisk respon- siveness by progenitor cells is generally believed to be a central feature of tissue maintenance, although the phenotypic distinctions between stem and progenitor cells in many nonhematopoietic organs have not been fully defined. Nevertheless, the functional difference in proliferative response between stem cells and progeni- tors represents a challenge for the specific manipulations of the stem cells for therapeutic purposes. While a complex array of extracellular signals and intracellular transduction pathways certainly participate in the distinct response, the cell cycle machinery, as a final step, must communicate with the specific regulatory cues (Steinman and Nussenzweig, 2002) and cell cycle regulators must play key roles in this process (D’Urso G and S, 2001). Cell cycle regulation and CDK inhibitors The molecular principles of cell cycle regulation have been defined largely in yeast and in orthologous systems applicable to the mammalian cell cycle (Hartwell and Weinert, 1989). A number of surveillance checkpoints monitor the cell cycle and halt its progression, mainly via the p53 pathway (Vogelstein and Kinzler, 1992), when DNA damage occurs. In mammalian cells, the cell cycle machinery that determines whether cells will continue proliferating or will cease dividing and differentiate appear to operate mainly in the G1 phase (Figure 1). Cell cycle progression is regulated by the Cell cycle inhibitors in stem cells T Cheng 7257 Oncogene sequential activation and inactivation of CDKs (Sherr, 1994; Sherr and Roberts, 1995). In somatic cells, movement through G1 and into the S phase is driven by the active form of the Cyclin D1, 2,3/CDK4, 6 complex and the subsequent phosphorylation retino- blastoma (Rb) protein (Classon and Harlow, 2002). Once Rb is phosphorylated, the critical transcription factor, E2F-1, is partially released from an inhibited state and it turns on a series of genes including cyclin A and cyclin E that form a complex with CDK2 as well as cdc25A phosphatase. Cdc25A is able to remove the inhibitory phosphates from CDK2 and the resultant cyclin E/CDK2 complex, then it further phosphorylates Rb, leading to a complete release of E2F and the transcription of a series of genes essential for S-phase progression and DNA synthesis. In parallel, the c-Myc pathway also directly contributes to the G1/S transition by elevating the transcription for cyclin E and cdc25A (Bartek and Lukas, 2001) (Figure 1). CDK activity is strictly dependent on cyclin levels that are regulated by proteosome-mediated proteolysis. Upon mitogenic stimulation, cyclin D serves as an essential sensor in the cell cycle machinery and interacts with the CDK4/6-Rb-E2F pathway. In addition to regulation by cyclins and phosphorylation/dephosphor- ylation of the catalytic subunit, CDKs are largely controlled by CKIs (Sherr and Roberts, 1999). Two families of low molecular weight CKIs, Cip/Kip and INK4, have been identified as capable of interacting with CDKs to suppress progression through G1. The Cip/Kip family, which includes p21Cip1/Waf1, p27kip1 and p57Kip2 (p21, p27 and p57 hereafter), may interact with a broad range of cyclin–CDK complexes, while the INK4 family, which includes p16INK4A, p15INK4B, p18INK4C and p19INK4D (p16, p15, p18 and p19 hereafter), specifically inhibits CDK4 and CDK6 kinases. Of note, p16 locus also encodes a structurally unrelated protein (p14ARF in humans or p19ARF in mice), through an alternative reading frame (ARF) (Sherr, 2001). Intriguingly, ARF protein is able to activate p53 function by removing the inhibitory effect of Mdm2 on p53 protein (Sherr and Weber, 2000). Therefore, the unique p16 locus serves as a bridging point between the two most frequently targeted networks in the cells of the majority of cancers: RB and p53 pathways. Both CKI families have been shown to be essential in arresting cell cycle progression in a broad spectrum of cell types (Morgan, 1995; Sherr, 1996). Studies using antisense of p21 or p27 have been able to release the cells from Go stage into the cell cycle (Nakanishi et al., 1995; Rivard et al., 1996). p27 and p18 have been proposed as intrinsic timers regulating animal organ size in an autonomous manner (Conlon and Raff, 1999). While we have a wealth of knowledge about the biochemical roles of CKIs in a variety of model systems (Sherr, 1996; Sherr and Roberts, 1999; Sherr, 2000), few experiments have been performed in the defined stem cell populations. There appears to be a distinct cell cycle control operating in stem cells in order to maintain their ‘stemness’. Mouse embryonic stem (ES) cells were shown to have a ‘defective’ Rb pathway and a nonresponsive p53 pathway (Aladjem et al., 1998; Prost et al., 1998; Burdon et al., 2002). In adult tissues, stem cells and progenitor cells share many common cytokine receptors (Berardi et al., 1995; Becker et al., 1999; Roy and Verfaillie, 1999; Shen et al., 1999), it is likely that the distinct cell cycle profile in stem cells is mediated by either distinct upstream intracellular mediators or unique combinatorial relationships of common biochemical mediators that limit the intensity of signals to enter into cell cycle. Defining the mechanisms in the stem cell response requires the analysis of individual cell cycle regulators and ultimately a systematic approach to define how these cell cycle regulators interact with one another and intersect various signaling pathways. Roles of CKIs in stem cell regulation Roles of CKIs in stem and progenitor cells have been shown in several species since Dipio, an analogue of p21/ p27, was reported to control embryonic progenitor proliferation in Drosophila (de Nooij and Hariharan, 1995; de Nooij et al., 1996). However, further study of their roles in stem cells from other higher species by conventional experiments in molecular biology is hampered due to the extremely limited availability of the primary stem cells. Fortunately, genetically engi- neered animal models and cell-sorting technologies allow us to vigorously test the gain or loss of gene function in the defined HSC and HPC populations in the hematopoietic system. CKI knockout rodent models have been particularly useful in providing a feasible approach for studying the roles of cell cycle inhibitors in stem cell biology (Brugarolas et al., 1995; Nakayama et al., 1996; Franklin et al., 1998). In the hematopoietic system, most CKI family members have been found to be differentially expressed in human CD34þ cells G1 G2 SM Rb E2F Rb- p E2F Cyclin A … G1 Cdc25A p16 INK4a p15 INK4b p18 INK4c p19 INK4d Cyclin D CDK4,6 p21cip1 p27kip1 p57kip2 Cyclin E CDK2 C-Myc G0 Figure 1 Cell Cycle Control of G1 Phase in Mammalian Cells. Description about each regulator during the G1 phase is detailed in the text. The CDK inhibitors are indicated in the black boxes Cell cycle inhibitors in stem cells T Cheng 7258 Oncogene (Taniguchi et al., 1999; Tschan et al., 1999; Yaroslavskiy et al., 1999; Marone et al., 2000; Cheng et al., 2001), indicating their distinct effects in hematopoiesis. In the following paragraphs, the HSC will be most used as an example to describe the roles of CKIs in stem cell regulation unless other adult stem cell types are specified. p21: a gatekeeper for quiescent stem cells p21 mRNA is abundant in quiescent human HSCs, while it is reduced in progenitor populations (Ducos et al., 2000; Stier et al., 2003). Functional assessment in hematopoietic cells been carried out using p21�/� mice (Cheng et al., 2000a, b). In the absence of p21, HSC proliferation tends to increase under normal homeo- static conditions. Exposing the animals to cell cycle- specific myelotoxic injury resulted in a higher rate of mortality due to hematopoietic cell depletion. Therefore, p21 governs cell cycle entry of stem cells, and its absence leads to increased proliferation of the primitive cells. However, self-renewal of HSCs was impaired in serially transplanted bone marrow from p21�/� mice (Cheng et al., 2000b), suggesting that restricted cell cycling is crucial to prevent premature stem cell depletion and hematopoietic death under conditions of stress. Such results might not be solely attributed to the stem cell effect and other possible defects in differentiation or apoptosis program, especially in the downstream of the hematopoiesis, cannot be completely ruled out based on the study. Nevertheless, the increased HSC cycling that results from p21 absence in mice has recently been extended to human cells and to a nondevelopmental context. Using postnatal CD34þCD38� human cells, it was shown that interrupting p21 expression with lentivectors ex vivo resulted in expanded stem cell number validated with the transplantation assay in irradiated NOD/SCID mice (Stier et al., 2003). This study demonstrated a proof-of-concept for an alter- native paradigm in which HSC numbers can be increased by releasing the brake on cell cycle entry rather than focusing on combinations of pro-prolifera- tive cytokines that can also induce differentation. These data further supported the notion that postnatal human stem cell proliferation can be uncoupled from differ- entiation in ex vivo settings. Given that relative quiescence is a common feature of different tissue stem cell types and that p21 maintains HSC quiescence, it was hypothesized that p21 might also participate in limiting NSC reactivity in vivo. Neural precursor cells from adults have exceptional proliferative and differentiative capabilities in vitro, yet respond minimally to in vivo brain injury due to constraining mechanisms that are poorly defined. The role of p21 in the regenerative response of neural tissue is evaluated in mice deficient in p21 following ischemic injury (Qiu et al., 2004). While steady-state conditions revealed no increase in primitive cell proliferation in p21-null brain, which is different from that found in the hematopoietic system, a significantly larger fraction of quiescent neural pre