Aurora kinase inhibitors: which role in the treatment of chronic myelogenous leukemia patients resistant to imatinib?

At present, there are no compounds in clinical development in the field of chronic myeloid leukemia (CML) or Philadelphia-positive (Ph+) acute lymphoblastic leukemia (ALL) that have been documented to harbor significant activity against the imatinib-resistant T315I mutation. Recent reports on the pre-clinical activity of some emerging tyrosine kinase inhibitors such as ON012380, VX-680 and PHA-739358 promise possible clinical efficacy against this specific Bcr-Abl mutant form. Here, we focus on the role of aurora kinase inhibitor VX-680 and PHA-739358 in blocking the leukemogenic pathways driven by wild-type and T315I-Bcr-Abl in CML or Ph+ ALL by reviewing recent research evidence. We also discuss the possibility of employing aurora kinase inhibitors as a promising new therapeutic approach in the treatment of CML and Ph+ ALL patients resistant to first and second generation TK inhibitors.


Introduction
The molecular signature of chronic myeloid leukemia (CML) and Philadelphia-positive (Ph + ) acute lymphoblastic leukemia (ALL) is the Bcr-Abl hybrid gene, originating from a reciprocal t(9;22) chromosomal translocation on the 22q-derivative, commonly referred to as the Philadelphia chromosome. 1 The resulting fusion protein, Bcr-Abl, displays deregulated tyrosine kinase activity and drives CML. 2 The disease begins with an indolent chronic phase (CP) marked by the expansion of myeloid cells with normal differentiation, and then inexorably proceeds to advanced phases, i.e., accelerated phase (AP) and the terminal blastic phase (BP). Imatinib (Gleevec, Glivec; STI571), a relatively selective tyrosine kinase inhibitor that blocks the catalytic activity of Bcr-Abl, is now the first-line treatment for all newly diagnosed CML patients. Despite excellent clinical results, there is still a need to improve therapy for patients with CML and Ph + ALL. More than 80% of newly diagnosed CML patients treated with imatinib in CP achieve a complete cytogenetic remission, as typified by the absence of the Philadelphia chromosome at the examination of 20 bone marrow meta-phases. 3 However, residual Bcr-Abl transcripts persist in the majority of these patients, as assessed by sensitive assays such as nested reverse transcription-polymerase chain reaction, and represent the potential pool from which disease recurrence may originate. While responses in CML in CP patients have been shown to last more than five years, 3 most responding patients with AP-and BP-CML, as well as those with Ph + ALL, relapse early despite continued therapy. Resistance to imatinib is most commonly mediated by Abl kinase domain mutations. 4 We and other authors have reported that approximately half CML patients have evidence of point mutations within the Abl kinase domain at the time of resistance to imatinib. Mutations target critical contact points between imatinib and Bcr-Abl or, more often, induce a conformation to which imatinib is unable to bind. 5 In the remaining patients, the reasons for imatinib resistance have to be traced to Bcr-Abl gene amplification or overexpression, clonal cytogenetic evolution, or altered levels of transport molecules responsible for imatinib influx and efflux (ABC transporters, hOCT1). 4 Abl mutations are at present the most extensively investigated and best characterized mechanism of resistance to imatinib. So far, at least 90 different point mutations have been isolated from relapsed CML patients who are resistant to imatinib. [6][7] The clinical and pathogenetic impact of mutations varies according to their different degree of residual sensitivity to imatinib. Indeed, while certain Bcr-Abl mutations retain in vitro sensitivity to imatinib at physiologically relevant concentrations and therefore may not be clinically meaningful, others require increased doses of imatinib, and some confer a highly resistant phenotype (Table 1)

. 9
The T315I mutation is highly resistant to imatinib An amino acid substitution occurring at the so-called "gatekeeper" residue, i.e. threonine 315, has attracted particular inter-est since it confers a high level of resistance not only to imatinib therapy but also to all of the newly developed tyrosine kinase inhibitors entered in clinical trials. Co-crystal structure analysis indicates that, on binding, the hydroxyl group of threonine 315 forms a crucial hydrogen bond with imatinib. 10 Moreover, the side chain of threonine also sterically controls the binding of the inhibitor to hydrophobic regions adjacent to the ATPbinding site. 11 In 10-15% of imatinib-resistant patients, especially those in more advanced phases of disease, a threonine to isoleucine amino acid substitution may be observed. The T315I abrogates imatinib binding because it disrupts the above mentioned hydrogen bond and introduces a bulkier isoleucine side chain into the gatekeeper position. 12 However, this explanation is not the most up-to-date. In fact, as recently demonstrated, the T315I resistance to imatinib mainly results from the breakdown of interactions between imatinib and both E286 and M290. 13 As a result, biochemical and cellular IC 50 values of imatinib for the T315I-Bcr-Abl have been shown to be >6400 times higher than those of wild-type Bcr-Abl (Table 1). 9 Some authors have suggested that the T315I is associated with highly aggressive disease phenotype and poor outcome if no timely therapeutic reassessment is made. 14,15 However, the effects of the T315I mutation on kinase activity in vitro and transforming efficiency of Bcr-Abl in vitro and in vivo have been very recently investigated, suggesting that in the absence of imatinib, there is nei- [Hematology Reviews 2009; 1:e1] ther increased kinase activity nor any growth advantage for cells carrying T315I-Bcr-Abl as compared to wild-type Bcr-Abl. 5 The two second-generation inhibitors in clinical development, dasatinib and nilotinib, are ineffective against the T315I mutant To counteract the problem of resistance due to point mutations, several second-generation inhibitors have been synthesized and tested in pre-clinical assays: nilotinib (AMN107), 8 33 Two of them are currently being evaluated in phase II clinical trials -the dualspecificity Src/Abl inhibitor dasatinib and the imatinib derivative nilotinib. Dasatinib is a novel, dual Src and Abl inhibitor entered in clinical trials. It has been shown to be ~300 times more potent than imatinib in Bcr-Abl inhibition assays. Excellent results in terms of hematologic and cytogenetic response in CML and Ph + ALL patients resistant to imatinib have been reported after dasatinib administration. 34 Pre-clinical studies have demonstrated that dasatinib is active against at least fourteen imatinib-resistant Bcr-Abl mutants (M244V, G250E, Q252H/R, Y253F/H, E255K/V, F317L, M351T, E355G, F359V, H396R, F486S). 19 The only imatinib-resistant Bcr-Abl isoform that was clearly insensitive to dasatinib was the T315I mutant, which retained kinase activity even in the presence of micromolar concentrations of the compound (Table 1). 19 Accordingly, imatinib-resistant patients harboring the T315I mutation have been shown not to benefit from dasatinib in the recent phase I trial. 34 Nilotinib is a close relative of imatinib with more than 20-fold improved affinity for wildtype Bcr-Abl. 16 It is highly efficacious in patients with imatinib-resistant Ph + CML. In vitro experiment with cell lines transformed with mutated forms of Bcr-Abl showed IC 50 proliferation inhibition for most mutations with the exception of the T315I, which remains refractory to nilotinib 8 (Table 1). Accordingly, clinical responses have been observed in patients with various imatinib-resistant Bcr-Abl mutations but not in patients positive for the T315I in the recent phase I trial. 35 Despite the pressing need for a clinically effective T315I-Bcr-Abl inhibitor, relatively few pre-clinical candidates have been reported. A potential pitfall might be the tendency to screen initially for Abl kinase inhibition rather than for T315I-specific inhibition. A promising approach is to design inhibitors targeting other regions of Bcr-Abl. For example, ON012380, a putative substrate-competitive inhibitor, exhibited low nanomolar activi-ty against imatinib-resistant Bcr-Abl mutants, including the T315I, in biochemical and cellular assays. 33

Aurora kinases as targets for cancer
Between these new promising drugs, VX-680 and PHA-739358, two aurora kinase A, B and C inhibitors, have a leading place. The aurora kinases are a family of serine/threonine kinases involved in many cellular functions, including progression through mitosis, by regulating spindle formation, chromosome segregation and cytokinesis. [35][36][37] The overexpression of aurora kinases has been reported in many human solid tumors, leading to defects in centrosome function, aberrant spindle assembly, misalignment of chromosomes, abnormal cytokinesis and genetic instability, determining the activation of oncogenic pathways. [38][39][40] Many authors reported an aberrant expression of the aurora A and B kinases also in leukemia cells, suggesting a potential role of these molecular targets in the treatment of CML and ALL. [41][42] Aurora kinase function is mediated by the phosphorylation of several substrates that have important roles in cell division, such as proteins survivin, CENP-A and serine 10 on histone H3. 37 The aurora kinases range in size from 309 to 403 amino acids. They have a C terminal domain that is responsible for regulation of the protein levels via proteasomal degradation; a highly conserved catalytic domain; and a short N-terminal domain that varies in length between the kinases and contributes to the differing locations of the kinases within cells. 43 (Figure 1). The aurora A isotype (also known as aurora, Aurora-2, AIK, AIR-1, AIRK1, AYK1, BTAK, Eg2, MmIAK1 and STK15) is widely expressed in proliferating normal tissues, with expression being cell-cycle-dependent and peaking at the G2/M point of the cell cycle. During mitosis, the kinase is virtually confined to the spindle poles, where it is needed for centrosome separation and maturation. 44 An overexpression of aurora A causes an increase in centrosome numbers and aneuploidy, 45 leading to the transformation of mammalian cells and also causes resistance to apoptosis induced by taxol in human cancer cell lines. Moreover, this kinase is a key regulatory component of the p53 pathway as its overexpression leads to increased p53 degradation, which facilitates oncogenic transformation. 46 Human aurora A has been proposed as a "drugable" target in several tumors including pancreatic, 46,47 hepatocellular, 48 breast, 49 nonendometriod, 50 and ovarian carcinomas, 51 gliomas 52 and aggressive non-Hodgkin's lymphoma. 53 Aurora B (also known as Aurora-1, AIM-1, AIK2, AIR-2, AIRK-2, ARK2, IAL-1 and STK12) activity is required for bipolar chromosome orientation and condensation. 54 Aurora B kinases are chromosomal passenger proteins, which are found in cells in a complex with inner centromere protein (INCENP) and survivin. The overexpression of an aurora B kinase-dead mutant (K-R) causes multiple defects in the mitotic machinery, including the loss of kinetochore attachment to microtubules and the exit from mitosis without anaphase or cytokinesis. 55 Increased Aurora B expression correlates with increased grade in glioma and colon cancer 56-57 and with anaplasia in thyroid carcinoma. 58 Aurora C (also known as AIK3) expression plays a role in spermatogenesis at the time when cells assemble the two meiotic spindles and also cooperates with aurora B to regulate mitotic chromosome dynamics in mammalian cells. 59 Aurora C overexpression has been detected in tumor cell lines in vitro 60 Tables 3 and 4). In fact, contrasting results related to this compound have been published.

Review
In particular, BIRB-796 binds with good affinity to T315I-Bcr-Abl (Kd = 40 nM), but has significantly weaker affinity for wild-type and other imatinib-resistant forms of Abl, with Kd values >1 μM. 21 In contrast, as reported by other authors, the compound fails to inhibit the proliferation of cells expressing T315I, suggesting a lack of clinical benefit for patients harboring such a mutation. 22 In a recent phase I-II study, MK-0457 was shown to be active in patients with T315I phenotype-refractory CML or Ph-positive ALL, with no significant extramedullary toxicity. 62 Because of a potential heart safety issue revealed in one patient who experienced QTc prolongation, the enrolment on phase II protocol was halted in November 2007. Furthermore, an innovative phase I clinical study of sequential and concomitant treatment with dasatinib and MK-0457 has been conducted, based on the suggestion that such a combinatory approach would suppress the emergence of T315I and other resistant clones, improving upon the response rate for dasatinib and the durability of response. To date, 3 patients with wild-type chronic myeloid leukemia (CML) or Ph-positive acute lymphoblastic leukemia (ALL) have been enrolled, and this innovative therapeutic combination showed a relevant hematologic activity and a good safety profile. PHA-739358 is a small molecule that selectively inhibits the ATP site of Aurora-A (IC 50 =13 nM) and Aurora-B (IC 50 =79 nM) kinases. 63 Starting from the rationale that aurora kinases play an important role in mitosis and that the interruption of their function has significant potential in the treatment of cancer, the drug, formulated for intravenous infusion, is being developed for therapeutic use in solid tumors and in patients with Philadelphia positive leukemias. Interestingly, PHA-739358, when tested against a panel of more than 30 kinases, has shown a strong cross-reactivity with c-Abl (IC 50 =25 nM). Its inhibitory activity on ABL in cells was confirmed in K562 leukemia cells which bear the Philadelphia chromosome related translocation Bcr-Abl. Furthermore PHA-739358 inhibits phosphorylation of Tyr412, which is located in the kinase activation loop of Abl and is also active against the T315I mutant of Abl, which is resistant to other ATP competitive inhibitors in the clinic, such as gleevec, and second generation TK inhibitors. A multicentric phase I/II study, aimed to test PHA-739358 in patients with chronic, accelerated or blast phase CML relapsing on gleevec or c-Abl therapy and preferably with T315I mutation in Bcr-Abl kinase is ongoing.

Binding mode of VX-680 and PHA-739358 to Abl
The compound VX-680, developed by Vertex Pharmaceuticals as an inhibitor of the aurora kinases, is a Y-shaped molecule, with a N-methyl-piperazine group forming the base or leg of the "Y", a pyrimidine group at the fork, and a methylpyrazole group at one arm and a substituted phenyl group at the other arm (Figure 2). A recent study 25 showed that VX-680 forms a hydrogen bond with the strictly conserved Asp381 of the Asp-Phe-Gly (DFG) motif in the Abl kinase domain and maintains it in an orientation close to one that is normally seen in active kinases, although the activation loop of Abl is not phosphorylated in this structure ( Figure 3). Furthermore, VX-680 does not deeply penetrate into the kinase domain as imatinib does and it is anchored to it by four hydrogen bonds. Three of these are formed between two carbonyl groups (Glu316 and Met318) and an amide nitrogen (Met318) in the "hinge region" of the kinase and three nitrogen atoms, one in the linker between the pyrimidine group and the methylpyrazole group, and the other two in the methylpyrazole group. These bonds are a common feature of kinase inhibitors and are independent of the sequence of the kinase. 59 Likewise, the fourth hydrogen bond, made by VX-680 to the side chain of Asp381 of the DFG motif, is to a strictly invariant catalytic residue. Using these four anchors, the inhibitor makes contact with 14 side chains within the kinase domain, eight of which are identical between Abl and aurora.

Review
One of the non-conservative substitutions is at the gatekeeper position, where Thr315 in Bcr-Abl is replaced by Leu210 in aurora A kinase ( Figure 4). The side chains of isoleucine (at position 315 of Bcr-Abl) and leucine (at position 210 of aurora A kinase) can be accommodated readily between the two sides of the "Y" of VX-680. For this reason, VX-680, in contrast to imatinib, is able to inhibit the kinase activity of both wild-type Bcr-Abl and T315I-Bcr-Abl. To understand the structural basis of the capability of PHA-739358 to bind and inhibit the T315I mutant, the crystal structure of the inhibitor-protein complex was determined 63 ( Figure 5). The protein is in the typical conformation of active kinases, with the activation loop in the extended DFG ''in'' conformation. Indeed, Asp381 points into the active site and interacts with Mg 2+ ion that occupies a position similar to the one usually seen in the structures of kinases in complex with ATP. The glycine loop adopts an extended conformation, in contrast to the other publicly available Abl structures where the loop is more distorted, which could be due to the specific binding mode of our inhibitor. The purified T315I Abl kinase domain used for crystallization experiments is predominantly phosphorylated on the activation loop at Tyr393, whereas Tyr253, Tyr257, and Tyr264 are phosphorylated at lower levels. These interactions probably stabilize the active conformation of the activation loop, which is, however, very similar to the structures reported for dasatinib in complex with the WT Abl kinase domain 64 and of MK-0457 in complex with the Abl mutant H396P. 25 The mutation of the threonine to the more bulky isoleucine does not seem to cause any widespread conformational changes but creates a steric hindrance that would interfere with the binding of inhibitors, such as imatinib, nilotinib, and dasatinib, which make use of the hydrophobic pocket. The binding mode of PHA-739358 is very similar to that reported for the complex of the same compound with aurora A ( Figure 5B and D), although the conformation of the proteins around the ATP-binding site shows some differences because in the aurora A structure the DFG motif is more similar to the ''out'' conformation. However, all of the essential contacts between PHA-739358 and Abl T315I involve highly conserved elements. The molecule makes three hydrogen bonds with the protein  backbone of the hinge region: the two nitrogen atoms of the pyrrolopyrazole core interact with the carbonyl oxygen of Glu316 and with the amide nitrogen of Met318, whereas the nitrogen of the amide group hydrogen bonds to the carbonyl oxygen of Met318. In addition, the side chain nitrogen of the conserved Lys271 is within hydrogen bonding distance of the oxygen of the carbonyl group and the oxygen of the methoxy group. As in the aurora structure, the benzyl group packs against Leu370 (Leu263 in aurora), whereas the Nmethyl-piperazine points toward the solvent accessible area of the kinase pocket. The gatekeeper residue in the aurora kinases is Leu210, a large and hydrophobic residue very similar to isoleucine, and we have observed that PHA-739358 binds in the ATP-binding pocket of aurora A without any steric hindrance with the gatekeeper residue. Indeed, the co-crystal structure reported here reveals that the compound is bound to the Abl T315I kinase domain in a way that accommodates the substitution of isoleucine for threonine. Figure 6 shows the structure of the Abl T315I complex with PHA-739358 superimposed on those of the Abl WT with imatinib and Abl H396P with MK-0457. In the T315I mutant, the isoleucine side chain causes a steric clash with imatinib and the hydrogen bond between imatinib and the side chain oxygen of threonine is lost ( Figure 6A and B). On the contrary, both PHA-739358 and MK-0457 bind in such a way to avoid the gatekeeper residue ( Figure 6C) and this provides an explanation for the ability of both compounds to accommo-date the isoleucine substitution. Furthermore, the pyrrolopyrazole scaffold of PHA-739358 is situated within van der Waals distance of the side chain of Ile315 mimicking the interaction between the inhibitor and Leu210 in aurora A. It is possible that this favorable hydrophobic packing interaction may explain why PHA-739358 is more active against the mutant than the WT protein. PHA-739358 could represent a valuable novel agent to target the T315I Bcr-Abl mutation, and preclinical and clinical data are coming through to support this concept.

Conclusions
The T315I is responsible for approximately 15% of the cases of relapse in CML and Ph + ALL patients on imatinib therapy. The clinical relevance of this mutant is likely to increase considerably as to date it seems to represent the main mechanism of resistance to dasatinib and nilotinib, the second-generation inhibitors already being developed clinically. Structural analyses indicate that the substitution of threonine with isoleucine at residue 315 eliminates a crucial hydrogen-bonding interaction and introduces a steric clash which abrogates binding and effective inhibition of Bcr-Abl by imatinib as well as by several novel inhibitors. A possible approach to the development of second-line strategies overcoming resistance induced by the T315I mutation is to design inhibitors binding regions of Bcr-Abl other than the ATP binding pocket. An intriguing alternative is to explore the possibility of whether molecules that have been developed as inhibitors for other protein kinases and are already undergoing clinical trials might include the T315I-Bcr-Abl mutant among their off-targets. Although off-target activity may lead to undesirable side effects, it has to be recognized that focusing on compounds that are already being tested in clinical practice may speed up the development of successful therapeutic strategies. Recent studies have shown that MK-0457 (VX-680) and PHA-739358, two small-molecule aurora kinase inhibitors, have in vitro activity against the T315I-Bcr-Abl. Moreover, preliminary data showed promising clinical efficacy in patients affected by Philadelphia positive leukemias, relapsing or resistant to first and second generation TK inhibitors. Such a remarkable efficacy raises the question of whether aurora kinases may also harbor some pathogenetic significance in CML and/or Ph + ALL or may be selectively deregulated by the T315I-Bcr-Abl, and whether auroras may be a suitable secondary target for inhibition. Thr 315