Cellular Thermal Shift Assay

As another approach to elucidate the cellular effects of BTB-1 and the most promising inhibitor 53 a cellular thermal shift assay (CETSA) was performed. In this assay cell lysate is applied and therefore the small molecules do not have to pass the cell membrane.¹⁵⁹ This assays relies on the thermal stabilization of the target protein upon Inhibitor binding. Mitotic HeLa cell extract was prepared and treated with 0.5%

DMSO as solvent control or compounds. After incubation and thermal treatment the samples were analyzed by SDS-PAGE followed by western blot analysis. First, the temperature at which Kif18A denatures was determined in the presence of 100 µM BTB-1, 53 or 0.5% DMSO. Unfortunately, no stabilization of Kif18A due to inhibitor addition was observed. No intensity differences of the samples were detectable (Figure 18). The DMSO control revealed that Kif18A is destabilized at temperatures between 50°-54°C (Intensity decrease, Figure 18) and therefore this temperature range was used to evaluate if higher inhibitor concentrations are able to stabilize Kif18A.

44

Figure 18: Western blot analysis of temperature gradient from 46°C to 62°C in the presence of 0.5% DMSO, 100 µM BTB-1 or 53. In line 10 (DMSO control 52°C) was less lysate loaded. No stabilization of Kif18A was observed. CycB, Cdc27 and tubulin served as mitotic markers and loading controls.

BTB-1 and 53 were applied in 500 µM, 200 µM, 100 µM and 50 µM concentration, but still no stabilization of Kif18A was observed (Figure 19). Since BTB-1 as well as 53 are not able to stabilize Kif18A even at concentrations up to 500 µM, it is possible that these inhibitors are not able to bind efficiently to Kif18A in a cellular environment.

Figure 19: Western blot analysis of compound titration of BTB-1 and 53 from 50°C to 54°C. The first and last line was loaded with random lysate. No stabilization of Kif18A was observed. CycB, Cdc27 and tubulin served as mitotic markers and loading control.

45 3.2.9 Immunofluorescence Imaging for Kif18A Localization

In order to analyze in detail the effect of BTB-1 and derivatives 53, 54, and 59 on cells and the architecture of the mitotic spindle, immunofluorescence imaging was performed. The same HeLa cell line as described for live cell imaging (chapter 3.2.7) was used.⁴⁹ Cells synchronized in S-phase using a high concentration of thymidine were released after nine hours and then treated with the solvent control DMSO or 30/50 µM of BTB-1 or derivatives 53, 54, and 59 (Figure 20). Thirty minutes after compound addition, cells were chemically fixed and tubulin (red) and DNA (blue) structures were visualized by immunostaining and treatment with the dye Hoechst 33342, respectively. Since initial immunofluorescence analysis of derivative 52 revealed that this small molecule caused unspecific cytotoxic effects on both dividing and non-dividing cells, it was excluded from further microscopic studies. The immunofluorescence images revealed that neither 53 nor 59 significantly affected spindle structures or alignment of chromosomes at 30/50 µM concentration (Figure 20, only 50 µM is shown). Additionally, the plus-end accumulation of GFP-Kif18A at microtubules – which as shown previously depends on the motor activity of Kif18A^44-45 – was not affected by the treatment with 53 or 59.

46

47 Figure 20: The upper panel shows the assay scheme for cell synchronization and compound treatment. First, cells were synchronized in S-phase using thymidine.

Nine hours after the release, cells were treated with DMSO as solvent control, 50 nM Nocodazole (Noc) or 30/50 µM BTB-1, 53, 54, 59 followed by formalin fixation and immunostaining. Representative immunofluorescence images of cells treated with DMSO, 50 nM Nocodazole (Noc) or 30/50 µM compounds BTB-1, 53, 54, 59 are shown. For each condition, merged images with DNA (blue), GFP-Kif18A (green) and microtubules (red) are shown on the right. In grey is shown from left to right DAPI (DNA), Kif18A, and tubulin (scale bar = 5 µm).

These, taken together with the normal spindle structures, suggest that treatment with derivative 53 or 59 does not result in efficient Kif18A inhibition in cells. HeLa cells treated with 50 µM BTB-1 revealed severe defects in spindle morphology and chromosome alignment, while 30 µM BTB-1 showed a mild effect (Figure 20), consistent with previous reports¹⁰⁶. Similarly, HeLa cells treated with 30 µM 54 showed multiple spindle poles, highly disorganized and fragmented microtubule structures and no detectable alignment of chromosomes comparable to the effect of 50 nM Nocodazole^160-161 (Figure 20). Since the observed phenotypes did not correlate with Kif18A depletion (elongated spindles with hyperstable microtubules) ^{42, 46}, it can be speculated that Kif18A is not the relevant binding partner of BTB-1 or 54 in cells.

48

3.2.10 Tubulin Polymerization Assay and IC50 Determination

The spindle phenotype induced by 54 was reminiscent of the phenotype caused by low doses of Nocodazole, a microtubule destabilizing compound (Figure 20). To analyze if the small molecule collection or BTB-1 target microtubules, a turbidity-based in vitro microtubule polymerization assay was performed.^162-163 The ability of tubulin to polymerize in the presence of GTP at elevated temperatures in a glutamate buffer and a turbidity based readout was used (Figure 21).

Figure 21: Schematic representation of tubulin polymerization. α-(light green) and β-(dark green or brown) tubulin heterodimers polymerize to microtubules.

After GTP hydrolysis the destabilized plus-end depolymerizes. Upon exchange of GDP to GTP in the tubulin dimers repolymerization occurs.

α- and β- tubulin heterodimers are able to polymerize in the presence of GTP. The GTP cap consists of tubulin dimers in the GTP bound state and stabilizes the highly dynamic plus-ends of microtubules.^164-166 After GTP hydrolysis, depolymerization, often called catastrophe, takes place. This process can be stopped by addition of GTP-bound tubulin dimers. After a certain time, the polymerization and depolymerization processes reach a dynamic equilibrium as long as enough GTP is present. The microtubule polymerization efficiency in the presence of the solvent

49 control DMSO was normalized to 100% and 1 µM Nocodazole was used as positive control. A small molecule selection was used at 50 µM concentration (Figure 22).

Figure 22: Screening results of the tubulin polymerization assay. DMSO was used as solvent control and its activity was set to 100%. 1 µM Nocodazole was used as positive control and the small molecules as well as BTB-1 were tested at 50 µM concentration. Average of three independent experiments and standard deviations are shown.

Notably, 52 and 61 showed the highest inhibitory effect on microtubule polymerization at 50 µM followed by BTB-1, 54, 60, 63 and 64 (Figure 22). BTB-1, 53, 54, and 59 were selected for further analysis due to their high potency against Kif18A (chapter 3.2.4) and 52 due to its high cytotoxicity (chapter 3.2.6).

Determination of the IC50 values showed that 52 has an IC50 value of 1.2 ± 0.1 µM, indicating that the high cytotoxicity of the compound might be due to its potent inhibitory effect on microtubule polymerization (Figure 23). BTB-1 and 54 showed a moderate effect on microtubule polymerization with IC50 values of 27.5 ± 3.3 µM and 9.2 ± 1.0 µM, respectively. 59 showed an IC50 value of 209.4 ± 37.3 µM on microtubule polymerization. Small molecule 53 did not show any significant effect on the tubulin polymerization up to a concentration of 1 mM. Thus, these studies revealed that the introduction of bulky substituents in phenyl moiety I at para position to the linker (54), especially the introduction of a nitro group (52), drastically increased the undesired inhibitory effect on microtubule polymerization.

0%

20%

40%

60%

80%

100%

120%

50

10^-7 10^-6 10^-5 10^-4 10^-3 0

20 40 60 80

100 BTB-1

52 53 54 59

M

% Tubulin polymerization

Figure 23: Dose-response curves of BTB-1, 52, 54, and 59 on microtubule polymerization in vitro with 52 being the most potent inhibitor. Derivative 53 did not show an effect on the tubulin polymerization up to 1 mM Average of three independent experiments and standard deviations are shown.

3.2.11 Structure Activity Relationships and Outlook

A small compound selection 52-71 was synthesized in one to two steps using BTB-1 as lead compound. The described synthesis route started from commercially available building blocks and allowed the robust and cost-efficient assembly of the small molecule collection. Several BTB-1 analogs, which inhibit the ATPase activity of Kif18A in the micromolar range were identified and enabled the establishment of SARs (Figure 24).

Figure 24: Scheme of SAR for Kif18A inhibition. Green colored groups are essential for activity. Yellow groups are tolerated but cause some activity loss.

Red groups are not tolerated and cause complete loss of inhibitory activity.

51 Comparison of identically substituted sulfone (52 and 59) and sulfoxide (64 and 67) derivatives highlight the importance of the sulfone functionality for the inhibitory activity of the compounds against Kif18A (Figure 24). Consistently, the replacement of the sulfone linker of BTB-1 with an ether linkage (71) resulted in an almost complete loss of inhibitory activity (Figure 24). Additional SAR studies revealed that the introduction of substituents in para position to the linker at phenyl moiety II was not tolerated whether they were electron withdrawing (62 and 63) or donating (57, 58, 66, 60, 61, 68 and 69). Likewise, the sterical expansion of the aromatic phenyl moiety II by replacing it with naphthalene (56) resulted in a complete loss of activity (Figure 24). Shifting the NO2 group from ortho (BTB-1) to meta (70) position to the sulfone linker within phenyl moiety I caused compound inactivity. In contrast, the replacement of chloride (BTB-1) at R¹ of phenyl moiety I with hydrogen (59) or different electron withdrawing groups like fluoride (53), NO2 (52), or CF3 (54) showed no drastic effect on compound activity (Figure 24). Since 53 and 59 – the most potent derivatives – did not interfere with the localization of Kif18A to the plus-ends of microtubules in mitotic cells, it might be that treatment with these small molecule inhibitors did not result in efficient Kif18A inhibition. One possible explanation could be that the compounds 53 and 59 are not cell membrane permeable, but the cellular effects received from structurally related compounds BTB-1 and 54 argues against this idea. Another possibility could be that association of Kif18A with binding partners or posttranslational modifications render Kif18A resistant to the binding of 53 and 59. Remarkably, BTB-1 and 54 revealed a cellular phenotype reminiscent of low doses of Nocodazole leading to the conclusion that they might interact with tubulin.

Additional in vitro studies confirmed that these compounds potently inhibited microtubule polymerization (Figure 25). For BTB-1 and 54 it is difficult to determine whether the altered Kif18A localization in cells treated with these compounds is due to their direct interaction with Kif18A or caused by the undesired effect on microtubule dynamics, which per se has an influence on Kif18A plus–end accumulation at microtubules⁴⁵. The in vitro microtubule polymerization assays revealed 52 as the most potent inhibitor (Figure 25). Therefore, it is likely that the high cytotoxicity of 52 is caused by its severe effect on microtubule polymerization.

52

Figure 25: Results of the SAR studies. The upper panel shows molecules sorted based on their inhibitory effect on microtubule polymerization, decreasing activity from left to right with 52 being the most potent inhibitor for microtubule polymerization. In the lower panel, compounds are sorted based on their inhibitory activity towards Kif18A with the most potent compound BTB-1 shown on the right side.

All these data suggest the introduction of electron withdrawing sterically demanding groups like NO2 (52) or CF3 (54) at R¹ of phenyl moiety I shift the inhibitory activity towards microtubules as well as the combination of a NO2 group at R¹ and a methyl group at the para position of phenyl moiety II (61). In contrast to that the introduction of small substituents at R¹ like hydrogen (59) or fluoride (53) influence the activity in favor of Kif18A inhibition (Figure 25).

The insights of the SAR studies might provide an important guideline for future design and synthesis of more potent Kif18A inhibitors. In order to establish even more powerful SARs the exact binding of BTB-1 and its analogs to Kif18A has to be resolved.

One possibility to achieve this would be the crystallization of Kif18A with bound inhibitor. The structure would reveal the interactions of the small molecules with the amino acids of Kif18A and allow the design of inhibitors that fit to the binding surface.

Such compounds would be invaluable tools for basic research, because Kif18A seems to have different functions at different times during mitosis and small molecules could be applied at specific time points. Further, the inhibitor would have the

53 potential to open up new strategies in the treatment of mitosis-related diseases such as Kif18A overexpressing tumors.

54

4. Summary

Kinesins are essential for the correct distribution of the genetic material into the daughter cells during mitosis.⁴ The functional dissection of individual kinesins involved in chromosome segregation is a challenging task, because of the complexity and dynamicity of mitosis. Due to their fast and often reversible mode of action, membrane permeable small molecules inhibitors are ideal tools to dissect the different functions of kinesins.

Paprotrain was discovered in 2010 by Tcherniuk et al.¹⁰⁹ as inhibitor for the kinesin Mklp2, whereas a small molecule named SH1, a 5-chloro-2-(3,4-dichlorophenyl)-7-(trifluoromethyl)quinoxaline, was identified and characterized in the laboratory of Prof. Thomas U. Mayer in 2009 ¹²². For the design of new Mklp2 inhibitors, SH1 served as the lead compound. The main drawback of SH1 was its low solubility in aqueous media. Thus, the aim was to develop new derivatives with enhanced solubility. After the establishment of a synthetic pathway towards SH1 using glyoxales, several compounds bearing polar substituents (7-26) as well as their precursors (1-6) were synthesized (chapter 3.1.3). In a second approach to enhance the water solubility the phenyl moiety of SH1 was exchanged to a thiophene (27-40, chapter 3.1.4).¹³⁴ Moreover, SH1 analogs linked to an enzymatic cleavable carbohydrate residue were synthesized (47-51, chapter 3.1.5).¹³⁸ All different synthetic approaches to enhance the water solubility of SH1 derivatives were successful. Characterization of the biological activity of the synthesized compounds revealed that only compounds barely soluble in aqueous media were active (chapter 3.1.6). Furthermore, addition of the detergent Triton X-100 suppressed the inhibitory activity of SH1 towards Mklp2 in vitro. Collectively, these results suggest, that aggregation of SH1 and its derivatives is necessary for their inhibitory activity (chapter 3.1.7 and 3.1.8).

For the kinesin Kif18A, a small molecule named BTB-1 was identified by Catarinella et al. in 2009 ¹⁰⁶ as a potent inhibitor. BTB-1 inhibited the ATPase activity of Kif18A in vitro and led to an increased percentage of cells in mitosis upon cellular treatment.

Nevertheless, cells treated with high concentration of BTB-1 did not show elongated spindles that are characteristic for the depletion of Kif18A. Here, the aim was to establish a synthetic pathway in order to synthesize a small molecule collection of

55 BTB-1 analogs. For the synthesis of the small molecule collection, the main reaction utilized was a nucleophilic aromatic substitution under basic conditions followed by an oxidation (52-63 and 64-69, chapter 3.2.3). The screening for Kif18A inhibition identified five active compounds (52-55 and 59), which showed IC50 values in the low micromolar range (<10 µM, chapter 3.2.4). Further, the selectivity of the newly found two most potent inhibitors (53 and 59) towards Kif18A was investigated. Both showed a higher selectivity towards Kif18A compared to BTB-1 (chapter 3.2.5). Next, the cellular effects of the novel small molecules were investigated. The cellular cytotoxicity was determined, followed by studies to detect Kif18A localization in cells (chapters 3.2.6-3.2.8). Fluorescence imaging revealed that the newly identified inhibitors did not alter the localization of Kif18A to the plus-ends of microtubules.

They might be not efficient enough to inhibit Kif18A in cells, due to binding partners or posttranslational modifications of Kif18A. Instead some compounds showed a severe effect on the microtubule network comparable to low doses of the microtubule poison Nocodazole (BTB-1 and 54, chapter 3.2.9). An in vitro microtubule polymerization assay confirmed that some compounds interfered with microtubule polymerization (BTB-1, 52 and 54, chapter 3.2.10). Most importantly, the undesired effect on microtubule polymerization was separated from the desired Kif18A inhibition with the help of in-depth SAR studies (chapter 3.2.11). The insights of the SAR studies provide a powerful guideline for future design and synthesis of highly potent Kif18A inhibitors, which would be invaluable tools for basic research and could open up new strategies in the development of novel drugs for mitosis-related diseases, like Kif18A overexpressing tumors.

56

5. Zusammenfassung

Mitotische Kinesine sind wichtig für die korrekte Verteilung des genetischen Materials in die beiden Tochterzellen während der Mitose.⁴ Sie haben verschiedene Funktionen zu unterschiedlichen Zeitpunkten des Zellzyklus. Spezifische niedermolekulare Inhibitoren sind daher ideale Werkzeuge, um die verschiedenen Funktionen der Kinesine zu analysieren, da sie in einer sehr schnellen und oft reversiblen Art wirken.

Paprotrain wurde 2010 von Tcherniuk et al.¹⁰⁹ als Inhibitor für das Kinesin Mklp2 entdeckt, während die niedermolekulare Substanz SH1, ein 5-Chlor-2-(3,4-dichlorphenyl)-7-(trifluormethyl)chinoxalin, im Labor von Prof. Thomas U. Mayer im Jahr 2009¹²² identifiziert und charakterisiert wurde. Für das Design neuer Mklp2 Inhibitoren diente SH1 als Leitstruktur. Der größte Nachteil von SH1 liegt in seiner schlechten Wasserlöslichkeit, weshalb neue Inhibitoren mit erhöhter Wasserlöslichkeit synthetisiert werden sollten. Nach der Etablierung eines Synthesewegs zur Darstellung von SH1 mittels Glyoxalen, wurden verschiedene Verbindungen mit polaren Substituenten (7-26) sowie deren Vorstufen (1-6) hergestellt (Kapitel 3.1.3). In einem weiteren Ansatz, um die Löslichkeit zu erhöhen, wurde der Phenylrest von SH1 durch Thiophen ersetzt (27-40, Kapitel 3.1.4).¹³⁴ Außerdem wurden SH1 Analoga verbunden mit einem enzymatisch spaltbaren Kohlenhydratrest synthetisiert (47-51, Kapitel 3.1.5).¹³⁸ Die verschiedenen strukturellen Modifikationen von SH1 führten erfolgreich zur Erhöhung der Wasserlöslichkeit. Die Charakterisierung der biologischen Aktivität zeigte jedoch, dass nur in wässrigen Medien schwer lösliche Verbindungen, aktiv waren (Kapitel 3.1.6). In in vitro Untersuchungen mit Triton X-100 wies SH1 keine Inhibition von Mklp2 auf, was zu dem Schluss führt, dass die Aggregation von SH1 notwendig für seine hemmende Aktivität ist (Kapitel 3.1.7 und 3.1.8).

Für das Kinesin Kif18A wurde ein niedermolekulares Molekül namens BTB-1 als potenter Inhibitor von Catarinella et al. 2009¹⁰⁶ identifiziert. BTB-1 inhibierte die ATPase-Aktivität von Kif18A in vitro und führte bei zellulärer Anwendung zu einer Erhöhung der Anzahl mitotischer Zellen.¹⁰⁶ Dennoch wiesen Zellen, die mit einer hohen Konzentration von BTB-1 behandelt wurden, keine verlängerten Spindeln auf,

57 die charakteristisch für die Depletion von Kif18A sind. Daher sollte ein Syntheseweg etabliert werden, um eine Sammlung von BTB-1-Analoga zu synthetisieren und Struktur-Aktivitäts-Beziehungen aufzustellen. Zur Synthese der BTB-1 Analoga wurde als Hauptreaktion eine nukleophile aromatische Substitution unter basischen Bedingungen verwendet, gefolgt von einer Oxidation (52-63 und 64-69, Kapitel 3.2.3). Im Screening zur Kif18A Inhibition wurden fünf aktive Verbindungen (52-55 und 59) identifiziert, die IC50-Werte im niedrigen mikromolaren Bereich zeigten (<10 µM, Kapitel 3.2.4). Ferner wurde die Selektivität der zwei potentesten Inhibitoren (54 und 60) gegenüber Kif18A im Vergleich zu anderen mitotischen Kinesinen untersucht und beide zeigten eine erhöhte Selektivität gegenüber Kif18A im Vergleich zu BTB-1 (Kapitel 3.2.5). Als nächstes wurden die zellulären Effekte der neuen Inhibitoren untersucht. Zuerst wurde die Zytotoxizität bestimmt, gefolgt von Studien zur Lokalisation von Kif18A in Zellen (Kapitel 3.2.6-3.2.8).

Fluoreszenz-Bildgebung zeigte, dass die neu identifizierten Inhibitoren nicht die Lokalisierung von Kif18A zu den Plus-Enden der Mikrotubuli veränderten. Trotz der Inhibition der ATPase-Aktivität von Kif18A in vitro sind die neu identifizierten Inhibitoren möglicherweise nicht effizient genug, um Kif18A in Zellen zu hemmen.

Dies könnte durch mögliche Bindungspartner oder posttranslationale Modifikationen von Kif18A verursacht werden. Stattdessen zeigten einige Verbindungen eine starke Wirkung auf das Mikrotubuli-Netzwerk, vergleichbar mit niedrigen Dosen des Spindelgifts Nocodazol (BTB-1 und 54, Kapitel 3.2.9). Ein in vitro Mikrotubuli-Polymerisationsversuch bestätigte, dass einige Verbindungen die Mikrotubuli-Polymerisation störten (BTB-1, 52 und 54, Kapitel 3.2.10). Jedoch wurde mit Hilfe intensiver Struktur-Aktivitäts-Studien die unerwünschte Wirkung auf die Mikrotubuli-Polymerisation von der gewünschten Kif18A Inhibition getrennt (Kapitel 3.2.11). Die Erkenntnisse der Struktur-Aktivitäts-Studien bieten einen sehr guten Leitfaden für das zukünftige Design und die Synthese von hochwirksamen Kif18A Inhibitoren, die nicht nur wertvolle Werkzeuge für die Grundlagenforschung wären, sondern auch neue Strategien in der Entwicklung innovativer Medikamente für Mitose-Erkrankungen, wie Kif18A überexprimierende Tumoren, darstellen könnten.^104-105

58

6. Materials and Methods

6.1 General

Chemicals, reagents and solvents

All solvents, reagents and fine chemicals are commercially available (Sigma‐Aldrich, Acros, Merck, Fluka, Roth, TCI, MCAT, ABCR, or Fluorochem) and are used without further purification. The used petroleum ether (PE) had a boiling point range of 35-80°C.

NMR‐Spectroscopy

1H‐, ¹³C‐ and ¹⁹F‐NMR spectra were recorded either on a Bruker Avance III 400 or Bruker Avance 600 spectrometer in the NMR facility of the University of Konstanz.

Chemical shifts are given in the δ‐scale and referenced to residual non deuterated solvent. A BBFOplus probe with actively shielded z‐gradient was used with its inner (BB‐) coil tuned to ¹⁹F. The following abbreviations were used for spin‐systems:

s (singlet); d (doublet); t (triplet); q (quartet). In case of multiple splittings, identifiers in the fashion of dt (doublet of a triplet) were used or m for undistinguishable multiplets. The spectra were processed using MestReNova.

Elementary Analysis

The elementary analysis was performed by the University of Konstanz micro

The elementary analysis was performed by the University of Konstanz micro

Im Dokument Novel Small Molecule Inhibitors for the Human Kinesins Mklp2 and Kif18A (Seite 57-0)