In the current study, mitotic kinesin KIF11, which is required for the separation of duplicated centrosomes and for the spindle formation [
17], was deemed to be a promising target for glioblastoma treatment. The expression profile of KIF11 mRNA in glioblastoma cells versus normal astrocytes was first assessed. KIF11 mRNA expression is reported to be elevated in tumor samples compared with adjacent normal tissue in tumors derived from breast, colon, lung, ovary, rectum and uterus [
18]. We confirmed this trend in GBM cell-lines (U87MG and DBTRG-05-MG) versus normal human astrocytes (data not shown).
Forward chemical genetics were applied to investigate the phenotypic effect of KIF11 inhibitors on GBM proliferation, apoptosis and cell cycle. This implied the selection BBB-permeating compounds known to specifically inhibit the target without affecting normal brain function.
To perform this study, we had access to a panel of existing preclinical efficacious KIF11 inhibitors. Monastrol (
3), the first reported small molecule inhibitor, the quinazolinone derivative inhibitor from Cytokinetics SB-715992 and MKI-833, the Merck reported inhibitor (
2) were demonstrated to be potent inhibitors of cell proliferation in several human tumor cell-lines (lung NCI-H460, A549; breast MDA-MB-231, MCF-7; colon HT29; ovarian SKOV-3, OVCAR-3; leukaemia HL-60, K-562, CNS SF-268; renal A498; osteosarcoma U2-OS; cervical HeLa) [
14]. Ispinesib is being advanced to
Phase II clinical trials as a general cancer therapeutic agent for cancers such as breast, ovarian and others. Moreover Monastrol (
3) and other monastrol analogues are reported to be specific inhibitors of human GBM cells inducing growth inhibition and affecting spindle formation [
13] without affecting the other kinesin-driven motor functions. Moreover, we selected a small set of Merck compound analogues whose smaller size meant they had greater probability of being brain penetrant. Since reaching the tumor in the brain was considered a critical criterion, only those compounds predicted to be BBB permeant were further investigated. The ability to pass the BBB is dependent on multiple factors, including lipophilicity, ionization profile, molecular size, polar surface area and molecular flexibility [
19,
20]. Relatively lipophilic drugs can cross the BBB by passive diffusion while polar molecules normally do not cross it unless they are substrates of specific active transport systems. There are several computational
in silico tools that help chemists and biologists understand the complex physico-chemical features of compounds, and hence to predict the BBB properties of a molecule. In this study, two computational-statistical suites have been used for that purpose: VolSurf (VOLSURF, version 4.0; available from Molecular Discovery Ltd.: London, U.K.
http://www.moldiscovery.com) and Cerius2 (Cerius2 version 4.11, available from Accelrys Inc.
http://www.accelrys.com). These data provided a robust base for assigning the probability of compounds for crossing the BBB based on their physico-chemical profile. Only compounds
2,
3,
4,
5 and
6, having passed the BBB selection filter, were further analyzed for their capacity to affect cell proliferation, to block the cell cycle and to induce apoptosis. We showed that the Ispinesib analogue compound
1 (Figures
2 and
3) has a higher anti-proliferative activity against human GBM cell lines when compared to Monastrol (
3) and to the Merk fragments (
4,
5 and
6). The effect of compound
1 on GBM cell-lines was also reflected by an increase of caspase 3 activity and by cell cycle block in G2/M phase.
In the neurotoxicity experiments carried against normal human astrocytes and rat cortical neurons, compound 1 revealed to be characterized by a relatively broad therapeutic window. This could at least partially be attributed to compound 1 selectivity for KIF11 over transport kinesins.
Monastrol (
3), Ispinesib MKI-833 (
2) and the majority of KIF11 inhibitors have been shown to have the same mechanism of action; they allosterically alter the ability of KIF11 to bind to microtubules and inhibit their movement by preventing the release of ADP without preventing the release of the KIF11-ADP complex from the microtubule [
21]. This non-ATP binding, allosteric site, which is formed by helices α2 and α3 and Loop 5 appears to be specific for KIF11. However, several studies have shown that loop 5 mutations may induce resistance problems such as those demonstrated in colorectal cancer cells [
22]. Should such mutability be identified in GBM patients as well, a need for KIF11 inhibitors that bind away from loop 5 may arise. The design of such novel KIF11 inhibitors, should take into account the recent evidence [
23] that ATP-competitive compounds can/should not interfere with microtubule dynamics.
Overall, although specific KIF11 inhibitors are of great value to GBM, mechanism-based toxicity of kinesin inhibitors in general may limit the development of specific mitosis inhibitors.