Motivations for this study

The final step for the target identification was to confirm binding to DPMS and to identify its exact binding mode. Therefore a rhodanine photo-affinity probe (34) was designed with the capability to covalently bind to proteins in close proximity upon UV-activation. In this chapter, the experimental setup and results of this target identification study is described. The photo-affinity probe34 showed activity against T. brucei at GI₅₀ 12.9 µM, however compared to the unsubstituted 3-benzyloxy analogue14v (GI₅₀ 4.4 µM), the activity decreased by a factor of 3.

Both34 and 14v did not show any toxicity against HL60 cells at 100 µM. Further details about the photo-affinity label are described in section 3.2.12. The labelling was carried out in 24 well plates with incubation times varying from 4 to 24 h in the presence of photo-affinity probe34 and T. brucei. Initially a tubular low-pressure mercury-vapour discharge lamp from Phillips with a UV maximum irradiation at 365 nm, a wavelength ideally for initiation of photo-cross-linking to proteins was used.^[179] After UV-activation, the parasites were lysed and proteins precipitated and subjected to Click-chemistry conditions with an alkyne-AlexaFluor488 tag. However, SDS-gel analysis did not show any fluorescent tagged protein bands (results not shown). After this initial experiment it was assumed that the photo-induced coupling to proteins is insufficient. In order to improve the experimental setup, the experiment was repeated as described above, but this time a portable UV-lamp at 254 nm maximum UV-irradiation was used for the photo-cross linking of the probe and the molecular target. Furthermore, this time the fluorescent tag was exchanged with an alkyne-biotin tag, in order to increase sensitivity by concentration of biotin-tagged proteins with streptavidin beads. However, SDS-gel analysis after Coomassie staining did not show any protein bands other than in the negative control without 34. It seemed that although rhodanine derivatives are described as promiscuous binders,^[82] 34 did not interact with any proteins. As this was rather unlikely considering its biological activity, the problem has to be the photo-cross-linking procedure. Indeed a recent communication with Prof. Keith A. Stubbs (The University of Western Australia) suggested that 24 well plates might not be suitable for the UV-cross linking step and that quartz spectrophotometric cuvettes might increase efficiency of the photo-labelling step due to better light exposure. In order to increase

3 Rhodanine-N-acetic acid derivatives

Figure 47: Representative docking result of rhodanine-N-acetic acid derivatives within the ac-tive site of homology model 4, upper figure shows the alignment of GDP-Man (lines) with the rhodanine inhibitor (stick); lower figure shows interactions within the active site; pictures were created with MacPymolX11 and MOE.2009.10.

Figure 48: Interactions of rhodanine-probe within the active site of homology model 4, overlaid with GDP-Man (lines); pictures were created with MacPymolX11 and MOE.2009.10.

3 Rhodanine-N-acetic acid derivatives

the sensitivity even further it is planned to lyse the parasite prior to addition of the photo-affinity probe and to initiate the cross-linking within the crude protein mixture.

3.6 Discussion

The rhodanine-N-acetic acid and the related thiazolidine-2,4-dione scaffold are an interesting starting point for the development of broad-spectrum anti-parasitic agents. Numerous biolog-ical applications of rhodanine and related compounds have been reported, often this scaffold is described as privileged for its ability to undergo multiple interactions with proteins.[81,83,91,224]

However, recently rhodanine derivatives were suggested to be promiscuous binders and there-fore it was crucial in this study to evaluate the anti-parasitic activity always based on previous toxicity studies against HL60 cells.^[82]While in many clinical applications it would be desirable to minimise side-effects, hitting multiple targets in parasites would be beneficial, as resistance is less likely to occur.

The investigations towards anti-parasitic agents started with a report of rhodanine-N-acetic acid derivatives as inhibitors of trypanosomal DPMS, an essential enzyme in GPI-anchor biosynthesis in T. brucei.^[68] However none of these derivatives inhibited DPMS completely (at best 10 % residual activity) or showed trypanocidal activity beyond ED5096 µM.^[68]Various derivatives of the previously found inhibitors were synthesised in order to increase in vitro anti-parasitic activity. A synthetically interesting derivative was the ortho-trifluoromethyl derivative 7e (Figure 49), as it showed a long range ⁵J_H,F coupling constant of 1.9 Hz in ¹H-NMR and

13C-NMR experiments. This long range coupling is most likely explained via orbital interactions (Figure 49) of the 1s orbital of H and the 2p orbital of F through-space. The long range coupling was used to determine the double bond configuration as the thermodynamically more stable Z-isomer.^[162] The interaction of the trifluoromethyl group and the CH group and therefore the

5J_H,F coupling could potentially be used as diagnostic tool for the measurement of intermolec-ular interaction of 7e and a target protein. Where disruption of the coupling would indicate binding of the protein towards the probe. A similar experiment has been performed previously with a 5-fluoropyrimidine substituted RNA, where the⁵J_{F ,H} long-range coupling was diagnos-tic for nucleic acid conformations.^[225] Isomerisation of the exo-cyclic double bond could be induced via a reaction sequence of esterification followed by a Knoevenagel reaction, giving access to various E- and Z-isomer mixtures with the Z-isomer as major product. This racemi-sation was most likely DMAP induced, as traces of DMAP have been observed.

Various modifications on the 5-benzylidene moiety in rhodanine-N-acetic acid derivatives have been explored, but none of these modifications resulted in improved anti-parasitic activity (GI50>100). Modifications on the N-3 position of the rhodanine scaffold, such as elongation of the side-linker, have been identified as essential for low µM activity against T. brucei ((GI50

9.0-N S S

O O

O

F F F 7e H

Figure 49: Long-range ⁵J_H,F coupling observed in compound 7e, possible induced by non-bonded orbital interactions

72.8 µM). Particularly remarkable was the effect of the esterification of the free carboxylic acid, resulting in low µM active compounds against T. brucei and T. cruzi (GI₅₀1.3-17.6 µM and 4.7-94.6 µM respectively), whereas the free carboxylic analogues showed no activity at 100 µM.

Reduction of the exo-cyclic double bond increased anti-trypanosomal activity, suggesting that Michael acceptor reactivity might not be part of the mode-of-action of these inhibitors.

Of particular interest were the rhodanine-N-acetic ester and thiazolidine-2,4-dione ana-logues displayed in Figure 50, as they showed low µM activity and selectivity against para-sites.

The rhodanine-N-acetic ester derivatives14b and 14v (Figure 50) both showed low µM ac-tivity against parasites, but furthermore the parent free carboxylic acid analogue of14v has been shown to inhibit DPMS, an essential enzyme in the GPI-anchor biosynthesis.^[68] Fur-ther modifications on the benzyloxy-substituent in 14v increased anti-trypanosomal activity, but also toxicity against HL60 cells. The meta-methyl substituted tert-butyl ester14b showed broad anti-parasitic activity in the lower µM range against T. brucei, T. cruzi and L. infan-tum (GI501.8, 5.1 and 2.4 µM respectively).

The para-tert-butyl derivative 14o (Figure 50) showed excellent anti-trypanosomal activity against T. brucei at GI₅₀1.7 µM, moderate activity at GI₅₀ 23.4 µM against L. infantum and no activity against T. cruzi.

A 2-pyridinyl substituent on position 5 of the rhodanine moiety resulted in trypannocidal activity against T. brucei and T. cruzi (GI₅₀ 1.5 µM and MIC 100 µM respectively) and had good selectivity indices for T. brucei (SI >67).

The ethyl esters of the meta-methyl (14a) and 3-benzyloxy (14v) derivatives were chosen for further mode-of-action studies. Using a myristic acid probe in combination with a fluores-cent tag, allowed the identification of GPI-anchor biosynthesis as possible target, indicated by decreasing intensities of GPI-anchored proteins using in-gel fluorescence. In order to further evaluate GPI-anchor biosynthesis as potential target, flow cytometry experiments with

FITC-3 Rhodanine-N-acetic acid derivatives

Figure 50: Rhodanine-N-acetic derivatives and analogues identified as promising anti-parasitic lead structures

labelled transferrin were performed on trypanosomes treated with14a and 14v. These deriva-tives caused a reduction of transferrin uptake compared to control cells, most likely mediated by decreasing amounts of GPI-anchored TbTf receptors. 14v caused a 1.6 fold reduction of transferrin uptake, whereas14a caused a 1.3 fold reduction, possible accounting for their try-panocidal activity in vitro. Fluorescence microscopy showed trypanosomes with diminished or missing FITC-labelled transferrin, confirming the FACS analysis experiment and missing up-take through GPI-anchored transferrin receptors. Lastly, cell-free assays of GPI-anchor syn-thesis in T. brucei were performed and revealed DPMS and/or the first mannosyltransferase (MT-1) as potential targets for14v and 14a. 14a was the stronger inhibitor, resulting in total vanishing of GPI-anchor intermediates after the synthesis of Dol-P-Man on HPTLC plates.

Analysis of logP and activity data against L. infantum showed that lipophilic substituents increased anti-parasitic activity. In the case of14a, this might explain the lack of activity of the ethyl ester, while the tert-butyl ester14b was active in the µM range against L. infantum.

It may be possible that increased lipophilicity aids plasma membrane diffusion, in this case 14b might have been the first GPI-anchor biosynthesis inhibitor with broad activity against all three protozoa. The next logical step to confirm this hypothesis would be screening against enzymes of the GPI-anchor biosynthesis in T. cruzi and L. infantum. An indirect confirma-tion would be the synthesis of a tert-butyl ester analogue of the 3-benzyloxy derivative14v,

as the increased lipophilicity might aid membrane diffusion and possible result in low µM ac-tivity against L. infantum. Alternatively the N-3 side chain could be elongated to increase lipophilicity. In order to further optimise the anti-parasitic activity and to confirm the molec-ular target, a photo-affinity probe has been designed, where two azides, a photo-activalible aromatic azide and an aliphatic azide for fluophore-tagging, were attached to the 3-benzyloxy-moiety of14v. However, attempts to identify the target protein failed, most likely due to missing photo-reactivity of the probe with the target in 24 well plates. Once the reaction conditions are optimised and cross-labelling between the probe and the target protein occur, further mass spectrometry studies would help to identify the protein-sequence, the mass and potentially the amino-acid residues where the labelling occurred.

Another interesting class of rhodanine and thiazolidine-2,4-dione derivatives had a cat-echol moiety on position 5. The derivatives 14t, 15f, and 7k were among the most potent inhibitors of T. brucei growth (GI₅₀ 0.9-1.6 µM).14ab also showed anti-leishmanial activity at GI5045.9 µM, but unfortunately high toxicity against HL60 was also observed (GI5031.2 µM).

However, 14t and its thiazolidine-2,4-dione analogue 7k showed no toxicity against HL60 at 100 µM. The free acid of14t, 2u showed no activity against any enzymes of the GPI-anchor biosynthesis, suggesting a different mode-of-action.2u has previously been reported as probe for many dehydrogenases, possible by mimicking the cofactor NAD(P)H.^[107,108]The aldose re-ductase 14-αdemethylase CYP51 plays an important role in sterol synthesis in T. brucei, T.

cruzi and possible L. infantum.^[111] It would be possible that 14t and 7k act as inhibitors for this or similar NAD(P)H dependent enzymes in parasites, explaining their low µM activity. A possible strategy to confirm this hypothesis would be the use of in-gel protein-affinity binding of 2u. A similar strategy has been employed previously, where 2u has been shown to be a weak fluorophore, and this property in combination with target-affinity has led to the identification of lactate dehydrogenase isozymes and 1-deoxy-D-xylulose-5-phosphate reductoisomerase as targets of2u in Escherichia coli.^[108] Therefore in future experiments, crude protein extracts of T. brucei would be separated on a Western-blot gel, this gel would be stained with2u, washed with buffer and the fluorescent proteins visualised (λ_ex=465 nm, λ_em=535 nm emission).^[108]

3 Rhodanine-N-acetic acid derivatives

3.7 Summary

130 rhodanine-N-acetic and related analogues have been synthesised and screened against T. brucei, T. cruzi and L. infantum in a whole cell activity assay.

At the same time all 130 derivatives were screened against HL60 cells for evaluation of toxicity against mammalian cells.

77 out of 130 derivatives were able to completely kill T. brucei in vitro at 100 µM (MIC 100 µM), a further 25 of them showed trypanocidal effects at 10 µM (MIC 10 µM).

27 out of 130 derivatives killed T. cruzi at 100 µM (MIC 100 µM).

8 out of 130 derivatives killed L. infantum at 100 µM (MIC 100 µM).

6 out of 130 derivatives showed anti-parasitic activity against T. brucei, T. cruzi and L. infan-tum in the whole cell activity screen, but only one compound (14b) had a good activity and se-lectivity profile. This tert-butyl ester analogue14b showed broad anti-parasitic activity against T. brucei, T. cruzi and L. infantum (GI50 1.8, 5.1 and 2.4 µM, SI 14, 5 and 10 respectively).

Primary whole organism phenotypic screens allowed identification of inhibitors with low µM anti-parasitic activity (GI₅₀0.9-17.6 µM) and good selectivity (SI >62.5 - 2).

Target identification studies, comprising metabolic labelling with myristate analogues, flow cy-tometry with FITC-labelled transferrin and cell-free radio-label assays for GPI-synthesis in T.

brucei allowed the identification of GPI-anchor synthesis as potential target, with DPMS and/

or MT-1 as molecular targets.

Taken together, these results provide further evidence that GPI anchor biosynthesis is a drug-able target for the development of novel anti-parasitic agents.

Catechol modified rhodanine-N-acetic ester analogues have been identified in a whole cell assay as low-µM growth inhibitors of T. brucei (GI500.9-1.4 µM, SI 21 - >75) with no activity in GPI-anchor biosynthesis and with unknown mode-of-action.