Discovery and Evaluation of Thiazinoquinones as Anti-Protozoal Agents

Pure compound screening has identified the dioxothiazino-quinoline-quinone ascidian metabolite ascidiathiazone A (2) to be a moderate growth inhibitor of Trypanosoma brucei rhodesiense (IC50 3.1 μM) and Plasmodium falciparum (K1 dual drug resistant strain) (IC50 3.3 μM) while exhibiting low levels of cytotoxicity (L6, IC50 167 μM). A series of C-7 amide and Δ2(3) analogues were prepared that explored the influence of lipophilicity and oxidation state on observed anti-protozoal activity and selectivity. Little variation in anti-malarial potency was observed (IC50 0.62–6.5 μM), and no correlation was apparent between anti-malarial and anti-T. brucei activity. Phenethylamide 7e and Δ2(3)-glycine analogue 8k exhibited similar anti-Pf activity to 2 but with slightly enhanced selectivity (SI 72 and 93, respectively), while Δ2(3)-phenethylamide 8e (IC50 0.67 μM, SI 78) exhibited improved potency and selectivity towards T. brucei rhodesiense compared to the natural product hit. A second series of analogues were prepared that replaced the quinoline ring of 2 with benzofuran or benzothiophene moieties. While esters 10a/10b and 15 were once again found to exhibit cytotoxicity, carboxylic acid analogues exhibited potent anti-Pf activity (IC50 0.34–0.035 μM) combined with excellent selectivity (SI 560–4000). In vivo evaluation of a furan carboxylic acid analogue against P. berghei was undertaken, demonstrating 85.7% and 47% reductions in parasitaemia with ip or oral dosing respectively.


Introduction
Natural products have historically played an important role in the discovery of new treatments for malaria [1]. From quinine and artemisinin [2,3] starting points, a diverse range of anti-malarials have been developed and have been the mainstay of frontline treatment for decades. Unfortunately with time has come loss of therapeutic efficacy due to the growing prevalence of drug resistant strains [4]. In the hunt for novel scaffolds from which to develop the next generation of anti-malarials, a structurally-diverse array of natural products, including those obtained from marine organisms, have been reported to exhibit activity towards Plasmodium falciparum [5][6][7].
As part of our own continuing search for new leads for the development of treatments for neglected human diseases [8][9][10][11][12] we have screened a library of synthesized and isolated marine natural products against a panel of four parasitic protozoa: Trypanosoma brucei rhodesiense, Trypanosoma cruzi, Leishmania donovani and Plasmodium falciparum K1 dual drug-resistant strain, with concurrent counter-screening for toxicity towards the non-malignant L6 rat myoblast cell line. We recently disclosed details of the first hit from this screen, the previously reported anti-inflammatory polyamine diamide ascidian metabolite orthidine F (1) [13][14][15] (Figure 1).

Figure 1. Structures of orthidine F (1), ascidiathiazone A (2) and analogues 3 and 4.
A second series of hits identified in this screening program were ascidiathiazone A (2), also previously reported by us as an anti-inflammatory alkaloid from a New Zealand ascidian, and synthetic analogues 3 and 4 [16]. The anti-protozoal evaluation of 2 (Table 1, entry 1) established the natural product to be a moderately potent in vitro growth inhibitor of P. falciparum K1 strain (IC 50 3.3 μM) and Trypanosoma brucei rhodesiense (IC 50 3.1 μM) while being effectively inactive towards T. cruzi and Leishmania donovani and exhibiting low levels of cytotoxicity against a mammalian cell-line (L6, IC 50 170 μM). Similar levels of potency and selectivity were observed for ester 3 (Table 1, entry 2), while Δ 2(3) analogue 4 ( Table 1, entry 3) exhibited more potent anti-malarial activity (IC 50 0.6 μM) with enhanced selectivity (SI Pf 300). Herein we report the results of a preliminary structure-activity relationship study investigating the influence of C-2 amide functionalization and thiazine-Δ 2 (3) oxidation on the biological activity of 2. In addition, we report that novel furan and thiophene analogues of 2 exhibit potent in vitro anti-malarial activity and that one analogue exhibits in vivo activity towards P. berghei.  [17,18]; i Melarsoprol, benznidazole, miltefosine, chloroquine and podophyllotoxin were used as positive controls.

Chemistry
We undertook a preliminary structure-activity relationship study to explore the effect of carboxylic acid functionalization and thiazine ring oxidation state towards the observed anti-protozoal activity of 2. Efforts to directly prepare amide derivatives of 2 by reaction of the synthesized natural product [16] with various amines in the presence of peptide coupling reagents, led to the formation of complex product mixtures and low yields (data not shown). Instead we made use of a longer four step reaction sequence (Scheme 1). Commercially available 8-hydroxyquinoline-2-carboxylic acid was converted to amides 5a-5j by reaction with the appropriate amine using PyBOP as the coupling agent in DMF. Subsequent oxidation using PIFA (phenyliodine bis(trifluoroacetate)) in MeCN/H 2 O yielded unstable quinones 6a-6j. Scheme 1. General reaction sequence for the preparation of analogues 7a-j and 8a-k. Reagents and conditions: (i) PIFA (2-3 equiv.), MeCN/H 2 O, 0 °C, 20 min; (ii) Hypotaurine (0.8 equiv.), CeCl 3 · 7H 2 O, MeCN/EtOH, rt, 2 days; (iii) 2 N NaOH, DMF, rt, 2 h; (iv) SOCl 2 , MeOH, 0 °C then rt, then 65 °C, 2 h, 93% yield.
Thiophene analogues of ascidiathiazone A were prepared (Scheme 2) starting from the literature quinone 9 [22]. Reaction with hypotaurine yielded two isomeric products 10a and 10b in a ratio of 1:0.3, as determined by NMR. Despite extensive attempts using chromatography, the isomers could not be separated and so were used as a mixture in the following steps. The regio-isomeric identity of 10a and 10b could not be established, as no relevant long range 1 H-13 C correlations were observed in HMBC data. Acid-mediated ester hydrolysis afforded carboxylic acids 11a and 11b, again characterized as an inseparable 1:0.3 mixture. HMBC data obtained for this isomeric mixture however was able to establish that the major regio-isomer was 11a as shown. Thus correlations observed between the major isomer H-8 resonance (δ H 7.84) to quinonoid resonance δ C 171.7 (C-9) and from the thiazine NH (δ H 9.31) to a second quinonoid resonance δ C 173.1 (C-5) confirmed the identity of 11a. In the case of base hydrolysis/autoxidation, reaction of the isomeric mixture 10a/10b with 1N NaOH in a biphasic reaction in EtOAc, yielded the expected Δ 2(3) product 12.
Column chromatography in this case was successful in affording the major regio-isomeric product in pure form. HMBC data analysis, in particular the observation of correlations from H-2 (δ H 6.57) and H-8 (δ H 7.82) to the same quinonoid carbon resonance at δ C 175.2 (C-9) established the dioxothiazine ring regiochemistry of 12 as shown.

In Vitro Biological Evaluation
The library of target analogues were screened against a set of four protozoa and for cytotoxicity towards the rat skeletal myoblast cell line L6 and the results are summarized in Table 1 (amide and oxidized analogues of 2) and Table 2 (thiophene and ester analogues). All of the amide analogues 7a-h,j evaluated were either equipotent or slightly more active against P. falciparum than the natural product 2. A similar observation was made for activities towards T. brucei rhodesiense, except for propargyl 7h (Table 1, entry 11) and glycyl ester 7j (Table 1, entry 12) both of which were significantly less active than 2. Notable in this series, unfortunately, was the lack of selectivity with most analogues exhibiting selectivity indices (SI) of 40 or less. Of this sub-set, only phenethyl amide 7e ( Table 1, entry 8) exhibited anti-protozoal activity and cytotoxic selectivity similar to those observed for 2. The corresponding Δ 2(3) analogues 8a-k while being typically equipotent or slightly more active against P. falciparum, were on the whole more cytotoxic with low selectivity. Significant amongst the series were the Δ 2(3) phenethyl amide 8e (Table 1, entry 17), which was the most active anti-T. brucei rhodesiense analogue, and ether 8i, ester 8j and carboxylic acid 8k (Table 1, entries [21][22][23] which maintained the anti-Pf activity of 2 but with modestly enhanced selectivity. There was no apparent correlation between calculated logP and observed biological activity (Table 1).  [17,18]; i Melarsoprol, benznidazole, miltefosine, chloroquine and podophyllotoxin were used as positive controls.
Thiophene and furan analogues 10a/10b, 11a/11b, 12, and 15-17 were evaluated against the same selection of protozoa and for cytotoxicity ( Table 2). Potent anti-Pf activity was observed for the thiophene examples, with the isomerically pure carboxylic acid 12 ( Table 2, entry 3) showing a desirable combination of nanomolar potency (Pf IC 50 35 nM) and excellent selectivity (SI Pf 4000). The furan analogues 15-17 (Table 2, entries 4-6) were slightly less active towards P. falciparum, exhibiting IC 50 ′s in the 110-340 nm range, with carboxylic acid 16 ( Table 2, entry 5) exhibiting the best selectivity (SI Pf 2400). It is interesting to note the broad-range activities of esters 10a/10b ( Table 2, entry 1) and 15 ( Table 2, entry 4): such pan-panel activity suggests the presence of an underlying general cytotoxic mechanism for these analogues. Once again, there was no apparent correlation between biological activity and calculated logP values.

In Vivo Anti-Malarial Evaluation
Furan carboxylic acid analogue 16 was selected for preliminary proof-of-principle in vivo evaluation in Plasmodium berghei infected mice. Preliminary ip acute toxicity of 16 showed no toxicity up to the highest test dose of 150 mg/kg. Using a standard test protocol [23], a repeated ip dose of 50 (mg/kg)/day for four days led to an 85.7% reduction in parasitaemia, and an increase in mean survival time from 4-6 days (untreated control) to 9.6 days. Switching to an oral dosing experiment (100 mg/kg once per day for 4 days) yielded a 47% reduction of parasitaemia. Although not considered significant, these levels of activity for both ip and po dosing clearly identifies heterocyclic dioxothiazinoquinone carboxylic acids to be a novel anti-malarial drug scaffold warranting further structure-activity relationship studies.

General
HRMS data were acquired on a Bruker micrOTOF-QII mass spectrometer. Infrared spectra were recorded on a Perkin-Elmer Spectrum 100 Fourier-transform IR spectrometer equipped with a universal ATR accessory. Melting points were obtained on an Electrothermal melting point apparatus and are uncorrected. NMR spectra were recorded using either a Bruker Avance DRX 300 or 400 spectrometer operating at 300 MHz or 400 MHz for 1 H nuclei and 75 MHz or 100 MHz for 13 C nuclei. Resonance assignments were made by interpretation of 2D data. NMR assignments marked by an asterisk are interchangeable. Proto-deutero solvent signals were used as internal references (DMSO-d 6 [16] and 9 [22] have been reported previously.