Phosphorylation plays a role in the stabilization of p53 following DNA damage [12,20]. p53 is modified by a range of kinases some of which overlap the kinases that target Mdm2 and Mdmx (see below). Phosphorylation of p53 in response to DNA damage occurs mainly in the amino terminal transactivation domain [12] (Table 1). Phosphorylation of p53 usually drives p53 transcriptional activation since these modifications stabilize p53. In human cells, ionizing radiation (IR) and ultraviolet light (UV) lead to extensive phosphorylation in the transactivation domain of p53 (serines 6, 9, 15, 20, 33, 37, 46, and threonines 18 and 81) [21]. IR and UV also induce phosphorylation at the carboxy terminus of p53 (serines 315 and 392) [22,23]. Adding to the potential for complexity in regulation, threonines 55, 150, 155 and serine 149 in the central region of p53 [24,25,26] and serines 376 and 378 [27,28] of p53 are phosphorylated under homeostatic conditions and may become hypo-phosphorylated following genotoxic stress. Interestingly, several kinases are capable of phosphorylating the majority of target sites of p53. This redundancy indicates the importance of p53 in tumor suppression and allows a mechanism for fine-tuning the control of p53 responses by various signaling pathway inputs [12].

Phosphorylation of serine residues near the p53 amino terminus (especially serines 15 and 20) is important for stabilization of p53 by decreasing association with Mdm2 and possibly Mdmx [29,30]. However, it does not appear that these residues are solely responsible for stabilization since mouse knock-in mutations of the corresponding murine sites (serines 18 and 23) show limited affect in certain tissues. This indicates that phosphorylation of these sites may not be a universal requirement for stabilization of p53 [31]. ATM is the primary kinase for p53 serine 15 leading to enhanced transcriptional activation. The importance of this modification has been shown by in vitro methods [32] and through expression of phospho-mimetic substitutions [33,34]. ATM also activates the checkpoint kinase Chk2 [35]. Chk2 phosphorylates p53 at serine 20 and interferes with the p53-Mdm2 interaction serving to stabilize p53 (see above). While ATM and Chk2 seem to be most important following IR, ATR is required for efficient response to UV damage in human cells through phosphorylation of p53 at serines 15 and 37 [36].

DNA damage also leads to phosphorylation of p53 by additional kinases (Table 1). Notable are, casein kinase 1 delta (CK1) that phosphorylates p53 at serine 9 and threonine 18 in a cascade of events that depends on the upstream phosphorylation of p53 at serines 6 and 15 [37,38,39,40]. The activity of CK1 serves to stabilize p53 by blocking interaction with Mdm2. Mass spectrometric and antisense experiments have shown that c-Jun N-terminal kinase (JNK) phosphorylates p53 at threonine 81 in response to DNA damage [41]. Homeodomain-interacting protein kinase 2 (HIPK2) has been shown to phosphorylate p53 at serine 46 both in vitro and in response to DNA damage in vivo [42,43]. These and other studies have shown that differences in the phosphorylation pattern of p53 exist in response to various sources of DNA damage. These complex and interconnected signaling mechanisms give some indication to the versatility and adaptability of the p53 response.

Phosphorylation of Mdm2 is localized to four main regions that are induced either by mitogenic signals or DNA damage (Table 2). Mitogenic signals lead to phosphorylation of a group of four serine residues near the nuclear localization and nuclear export sequences (serines 157, 166, 186 and 188). These sites will not be considered further in this article but have been reviewed elsewhere [11]. In response to DNA damage, Mdm2 is modified at the amino terminus, within the central acidic domain and within a disperse group near the carboxy terminal RING domain. Mdm2 serine 17 near the amino terminus is phosphorylated by DNA-PK in vitro [44]. More recent biochemical studies have shown that this site is responsible for dictating the dynamic equilibrium of Mdm2-p53 interactions [45]. Under homeostatic conditions, a large group of serine residues (including: serines 240, 242, 246, 253, 256, 260, 262 and 269) in the acidic domain are phosphorylated. This region becomes hypo-phosphorylated under stress conditions [11,46,47]. The acidic domain is important for target recruitment and ubiquitination [48]. DNA damage also leads to phosphorylation of a more disperse group of serine and tyrosine residues mainly residing near the RING domain (tyrosine 394, serines 386, 395, 407, 425, 428 and threonine 419) with an additional site adjacent to the acidic domain (tyrosine 276).

DNA damage activates cell cycle checkpoints that lead to the robust activation of ATM and ATR kinase pathways. ATM is activated by DNA double strand breaks while ATR is activated by stalled replication forks [49]. Direct phosphorylation of Mdm2 at serine 395 by ATM blocks nuclear export of p53 and leads to stabilization of p53 protein [32,50]. ATM phosphorylation of Mdm2 at serine 386, 395, 425 and 428, and at threonine 419 cooperatively lead to stabilization of p53 by preventing poly-ubiquitination, a consequence of preventing Mdm2 RING domain homo-dimerization [51]. ATR phosphorylates Mdm2 at serine 407 in response to specific types of DNA damage blocking nuclear export of p53 [52]. ATM also activates the downstream c-Abl kinase through direct phosphorylation in response to DNA damage [53,54,55]. c-Abl phosphorylates Mdm2 at tyrosines 276 and 394 [56,57]. Phosphorylation of Mdm2 tyrosine 276 leads to increased levels of nucleolar Mdm2 and increases binding of Mdm2 to its negative regulator, ARF. Thus ARF protects p53 through re-localization of Mdm2 [56]. Phosphorylation of Mdm2 tyrosine 394 stabilizes p53 and inhibits the negative regulation of Mdm2 on p53 transcriptional and apoptotic activities [57]. An additional c-Abl target site at Mdm2 tyrosine 405 has been identified but a physiological role has not been determined [56]. These events support a multi-factorial model of Mdm2 regulation based on varied signaling events.

As with Mdm2, Mdmx is also phosphorylated at multiple sites in response to DNA damage (Table 2). ATM phosphorylation of Mdmx at serine 403 leads to rapid degradation of Mdmx alleviating repression of p53 activity [58,59]. ATM-dependent Chk2 phosphorylation of Mdmx at serine 367 increases binding to the adapter protein 14-3-3, which has been suggested to compete with the de-ubiquitinating enzyme HAUSP leading to destabilization of Mdmx [59,60,61]. In addition, Mdmx serines 161, 342, 365 and 391 are also phosphorylated in response to DNA damage but their relative contribution to Mdmx regulation is not known [11]. Wang YV and co-authors have generated a mouse that harbors a series of three conserved serine-to-alanine mutations in Mdmx, sites that become phosphorylated in response to DNA damage. The authors report that these mice lack robust Mdmx degradation in response to DNA damage and that this compromises p53 activity [62]. This result highlights the in vivo importance of these modifications in control of the cellular response to stress. Recently it has been shown that c-Abl phosphorylates Mdmx at tyrosines 55 and 99. Phosphorylation of Mdmx at tyrosine 99 inhibits Mdmx-p53 complex formation, which frees p53 to activate gene expression [63]. Also, casein kinase 1 alpha (CK1α) has been shown to phosphorylate Mdmx at serine 289 in the acidic domain. Knockdown of CK1α or ionizing radiation leads to the activation of p53 and apoptosis but the molecular mechanism remains to be determined [64]. Thus varied responses to DNA damage have the potential for multiple levels of control with regard to the Mdmx response.

Protein kinases are able to catalyze the transfer of the terminal phosphate of ATP to a target substrate. Protein kinases either target serine and threonine residues or tyrosine residues around some amino acid sequence specificity or structural specificity motif. ATP binding is typically in a deep pocket of the kinase active site. The majority of kinase inhibitors target the ATP binding site for competitive binding [90]. Four different classes of kinase inhibitors have been identified. Type I kinase inhibitors represent the largest class of kinase inhibitors and are competitive inhibitors of the kinase active conformation. Type II kinase inhibitors recognize the inactive conformation of the kinase typically through a hydrophobic patch near the ATP binding site that is only exposed in the inactive conformation. In addition to compounds that target the ATP binding site, a third type, the allosteric kinase inhibitors have been developed that modulate kinase activity. These compounds exhibit the highest degree of selectivity since their binding sites are independent of the well-conserved kinase active site. This class of compounds also includes inhibitors that bind accessory molecules that are required for kinase activity. The fourth type of inhibitor is covalent inhibitors that form irreversible crosslinks to the kinase active site rendering it inactive [65]. In addition to the current compounds in development or trials, a large group of analogues that have modifications to the basic chemistry of the original lead compound are being designed to provide enhanced selectivity or lower toxicity.

Over the past decade pharmaceutical and academic researchers have begun to understand and target kinase-signaling pathways that are involved in cancer development and metastasis. Much work has led to the appreciation that targeting kinases in cancer will likely require some rationalization of drug selection based on individualized patient criteria. Patient specific differences arise from the type of tumor and the tumor microenvironment. Understanding the tumor in the context of its kinase-dependent growth characteristics will aid selection of treatment regimens. Understanding the kinase signaling pathways involved in loss of growth control affords the clinician some therapeutic rationale for treatment.

Understanding the interplay of Mdm2 and Mdmx with p53 in tumor cells would aid drug selection. Dysregulation of p53 function plays a critical role in tumor development by side stepping p53-dependent responses. Inactivation of p53 in tumors is achieved through two main mechanisms. First, inactivation of p53 function by direct mutation of p53 and second, by disrupting signaling pathways that lead to p53 activity. For tumors harboring wild-type p53, re-activating p53 in established tumor cells represents an effective intervention scheme [91,92]. In more than half of tumors with non-functional p53, the p53 protein is wild type. In these cases, affecting p53 activity directly or through modulation of Mdm2 and/or Mdmx to re-activate p53 activity would likely lead to therapeutically favorable responses. Of particular interest are therapies that might exert less selective pressure on cells while exerting their effects on multiple targets (e.g. Mdm2 and Mdmx). There is little doubt that drugs that activate a functional p53 pathway would have wide applications in the treatment of cancer.

Modulating Mdm2 and Mdmx levels has profound effects on p53 activity. Low expression levels of Mdm2 or Mdmx is lethal whereas an excess of either can be oncogenic [93]. Many human tumors express high levels of either Mdm2 or Mdmx [94]. In fact, a modest two-fold increase in Mdm2 protein is sufficient for tumorigenesis [95,96]. Additionally, a single nucleotide polymorphism (SNP) in the Mdm2 promoter that increases Mdm2 mRNA and proteins levels on the order of two- to four-fold is a strongly correlated with poor prognosis [97]. Further, deletion of one allele of Mdm2 or Mdmx in mice suppresses B-cell lymphoma development induced by the oncogene c-Myc [98]. These data taken with the fact that signal transduction pathways: (1) are responsible for the nuclear import and export of Mdm2, (2) alter Mdm2 ubiquitin ligase activity, (3) affect Mdm2 binding partners and (4) affect Mdm2 regulatory functions suggests that selectively targeting the kinases that modulate Mdm2 and Mdmx activity would protect and activate p53. Thus providing novel therapeutic targets.

The classic example of a rationally designed kinase inhibitor is the Abelson tyrosine kinase (Abl) inhibitor imatinib [66]. The use of imatinib to treat chronic myelogenous leukemia (CML) presents a case study of the need for a careful understanding of the disease mechanism and drug action in predicting drug applicability for other indications. Imatinib inhibits the Abl kinase activity of the constitutively active mutant BCR-Abl fusion kinase protein by blocking ATP binding. In addition, imatinib is minimally toxic to non-disease cells [99]. BCR-Abl is the result of a gene fusion between the break-point cluster region (BCR) gene and c-Abl kinase (translocation t(9;22) or Philadelphia chromosome). BCR-Abl is present in 95% of patients diagnosed with CML [100]. BCR-Abl functions as an oncogene by dysregulating intracellular signaling leading to aberrant proliferation and resistance to apoptosis. The clinical outcome of the BCR-Abl fusion gene product is an abundance of myeloid progenitor and differentiated cells. At the time of diagnosis, CML patients typically have peripheral blood counts nearly 20-fold higher than normal [101]. Blood cells harboring the BCR-Abl fusion gene product initially maintain their normal activity but eventually lose their ability to differentiate leading to blast crisis. Imatinib is much less effective after blast crisis presumably due to the presence of multiple “hits” to the cell. Imatinib provides positive cellular response in 65–90% of patients with CML and up to 80–90% response when patients are in early chronic phase [102,103,104]. Imatinib is generally well tolerated with few side effects compared to standard cytotoxic chemotherapy. Low peripheral blood counts are a common side effect with imatinib treatment while non-hematologic reactions are minor [101]. Imatinib is a success story of rationalized drug design but also illustrates a need for multifaceted approaches in cancer treatment [66].

The initial excitement of imatinib’s success was dampened by the early identification of resistance mutations mainly in the BCR-Abl kinase domain [105]. Resistance to imatinib in CML is usually by the reactivation of BCR-Abl signal transduction. Imatinib resistance in CML develops quickly, and some argue inevitably, since the selective pressures on cells treated with single target therapies is high. Since cells exposed to single target therapies only need to overcome a single source of inhibition, a further point mutation is often sufficient to develop resistance. And due to the rapid proliferation of cancer cells, the rise of resistance mutations often occurs in a clinical setting.

Imatinib has also been used on a limited basis for treatment of other tumors with mixed success. Imatinib exhibited a lack of response in at least one study with metastatic Leydig cell tumor [106]. Further, in a mouse model of mammary adenocarcinoma cells (4TI, p53 null), imatinib treatment lead to accelerated tumor growth [107]. These results suggest that the reported in vitro and animal model findings for imatinib may not be directly applicable for additional indications [108,109]. These disparate results suggest that a more complicated signaling cascade is at play in various tumor models. Since CML is typified by hyperactive Abl kinase activity, imatinib is useful in reducing the level of Abl kinase activity in the cell to a more normal physiological level. However, pressures for tumor growth eventually overtake the action of the drug and resistance mutants develop. The action of imatinib in cells that have normal Abl signaling would produce a whole different range of signaling events that may or may not be advantageous as cancer therapeutics. In this context, treatment of tumors harboring wild-type p53 with imatinib would not likely provide benefit since p53 levels would be negatively impacted through inhibition of Abl kinase activity. Additionally, blocking Abl phosphorylation of Mdmx would cause the formation of Mdmx-p53 complexes, rendering p53 transcriptionally inactive [63].