Syntheses and biological investigations of kirkamide and oseltamivir hybrid derivatives

Abstract The C7N-aminocyclitol kirkamide was recently isolated from the plant obligate symbiont Candidatus Burkholderia kirkii and was hypothesized to be beneficial to the plant host due to its cytotoxic activity against insects and arthropods. To study its mechanism of action and inspired by its structural similarity with N-acetylglucosamine (GlcNAc) and oseltamivir, kirkamide-oseltamivir hybrid derivatives were synthesized and investigated for their biological activity. Interestingly, kirkamide analogues were reasonably potent against a known bacterial neuraminidase.


Introduction
Natural products remain a prime source of inspiration for the development of new lead structures in drug discovery [1,2]. Recently, two novel natural products, kirkamide (1) and streptol glucoside (2) have been isolated from an obligate symbiosis system composed of the plant Psychotria kirkii and the bacterium Candidatus Burkholderia kirkii [3,4]. Synthetic routes towards accessing these natural products and derivatives have been already identified in the literature [3,5,6]. The discovery of kirkamide (1) was guided by the genome-based prediction of 2-epi-5-epi-valiolone [7,8], which is known to be involved in the biosynthesis of the C 7 N aminocyclitol family, a rich source of bioactive compounds [9e13]. The cytotoxicity of kirkamide (1) was evaluated against several organisms, such as insects and arthropods [3]. The structural similarity between N-acetylglucosamine (3, GlcNAc), oseltamivir (4) [14], and kirkamide (1) suggested that the natural aminocyclitol 1 could act as a glycoside hydrolase (glycosidase) inhibitor (Fig. 1).
The aim of this study was to (a) prepare hybrid compounds between kirkamide (1) and oseltamivir (4), (b) investigate plausible enzyme targets, and (c) evaluate the potential of oxidation at the methylene allylic position of kirkamide (1) under biologically relevant conditions. In order to reach these goals, we employed a divergent synthetic route allowing for late stage modification of a synthetic intermediate.  (1), streptol glucoside (2), N-acetylglucosamine (3,GlcNAc), and oseltamivir (4).

Results and discussions
To achieve our aims, triflate 5 [3] was selected as the key intermediate for the synthesis of all compounds (Scheme 1). Evaluation of different electrophilic precursors for the subsequent carbonylation, both by us [3,6] and others [5], all provided evidence for the superiority of enol triflates over enol halogenides. Consequently, using the standard Pd-mediated carbonylation/esterification conditions, the ethyl ester 6 was synthesized in good yield (68%) and smoothly Si-deprotected using TBAF in THF to obtain the alcohol 7. The benzyl deprotection was found to be challenging: In contrast to the hydroxymethylene derivative in the kirkamide synthesis [3], the ester 7 proved to be recalcitrant under classical Birch conditions, eventually leading to the decomposition of the starting material. After screening of several Lewis acids, only BCl 3 in CH 2 Cl 2 delivered the triol 8 in low yield. Pleasingly, the addition of pentamethylbenzene [15] to this reaction mixture significantly increased the yield; and the ester was subsequently saponified with KOH solution to obtain the kirkamide carboxylic acid derivative 8.
The free hydroxyl group of the available ester intermediate 7 (Scheme 2) was activated with mesyl chloride to obtain mesylate 9 (71%, purified by flash chromatography on silica) and its configuration at position 1 was subsequently inverted, using KNO 2 , to the C(1) epimer 10 [16]. Using the optimized deprotection conditions developed earlier, the second kirkamide derivative 11 was smoothly synthetized via debenzylation.
Taking advantage of the mesyl intermediate 9( Scheme 3), the amine 12 was synthesized in two steps composed of a S N 2 type reaction with azide following by a Staudinger reaction using PPh 3 as reductant. We were pleased by the good yield of this two-step procedure and the full deprotection was performed to obtain the amine kirkamide derivative 13.
Synthetic kirkamide 1 and its derivatives 8, 11, and 13 were evaluated as potential inhibitors of various N-acetyl-D-glucosaminidases due to their structural similarity to the valienamine class of compounds and oseltamivir (4) [14,17,18]. We used a panel of enzymes (HexB, EcNagZ, NAGLU, Jack bean b-hexosaminidase, chitinase from Streptomyces griseus (S. griseus), human chitotriosidase and BbLNBase) that are known N-acetyl-Dglucosaminidases (Table 1). Interestingly kirkamide (1) was found to be weakly active against human NAGLU (IC 50 ¼ 240 mM) and was modestly active against the other N-acetylglucosaminidases that process single GlcNAc residues. From all the hybrid derivatives, 13 possessing an amine functional group at the C(1) position exhibits the highest inhibition for HexB, EcNagZ, Jack bean b-hexosaminidase and BbLNBase. In comparison, kirkamide (1) displays generally better activity than the oxidized derivatives 8 and 11, and surprisingly, replacement of the OH group by the basic amine (/ 13) resulted in better activity against three of the enzymes tested (HexB, EcNagZ, NAGLU).
In addition, we assessed kirkamide (1) and the synthesized derivatives against a neuraminidase from Clostridium perfringens (C. perfringens) due to the structural similarity of 1 with oseltamivir (4). Interestingly compounds 11 and 13 possess potency towards the enzyme even though they lack the 3-pentyl sidechain of 4, which is generally required for good activity against bacterial and viral neuraminidases [14,19]. Interestingly, the des-3-pentyl derivative of oseltamivir (4) also displayed only micromolar activity and the introduction of the 3-pentyl chain boosted activity by three orders of magnitude [14].
Another possible mode of action, and potentially more interesting, is that kirkamide (1) acts as a N-acetylglucosamine surrogate and is incorporated into the UDP-GlcNAc (14) salvage pathway found within eukaryotes [20]. Kirkamide (1) would be metabolized into UDP-kirkamide (15) (Fig. 2) and potentially act as an inhibitor of O-GlcNAc transferases, critical enzymes in chitin biosynthesis [21], a common polysaccharide found in arthropods and insects [22]. Indeed, this type of incorporation of unnatural GlcNAc-based compounds has literature precedent [23e25].
We decided to use a model system taking advantage of the O-GlcNAc modification which is a post-translational modification found in eukaryotes with the enzyme facilitating this modification, O-GlcNAc transferase (OGT) using UDP-GlcNAc (14) as its substrate. Utilizing a per-acetylated version of kirkamide 16, prepared by treatment of kirkamide (1) with excess acetic anhydride, we treated   Table 1 Inhibitory activity (IC 50 , mM) of kirkamide (1) and derivatives. the model cell line COS-7 using various concentrations of 16. A similar strategy has been applied for other systems and showed an increase in the uptake of carbohydrate-based compounds into cells [26]. Unfortunately, we found that there was no clear decrease in O-GlcNAc modified proteins even at high concentrations of 16 ( Fig. 3) and concentrations of 16 above 1.5 mM were toxic. This result does not rule out that kirkamide (1) is converted to UDP-kirkamide (15) to act as a protective compound against organisms affecting Psychotria-Burkholderia leaf nodule symbiosis but further studies are required to evaluate this hypothesis. Finally, we have investigated the possibility that the oxidized kirkamide derivative 8 is present in planta, for example as a byproduct by oxidation of the allylic methylene group of 1. In addition, we wanted to test if plant extracts are able to mediate the oxidation of 1 to 8. However, under several conditions evaluated, no oxidation products such as 8 could be detected. While this does not rule out that no oxidation takes place in planta, these observations together with the decreased biological activity against some of the enzymes tested in this study, render the carboxylate 8 less likely to be an active material in the Burkholderia/Psychotria symbiosis.

Conclusions
In conclusion, the mechanism of action of kirkamide (1)w a s investigated by testing its activity against enzymes involved in the metabolism of structurally related compounds. Surprisingly, kirkamide (1) did not exhibit a strong inhibition against these potential biological targets. Furthermore, hybrid oseltamivirkirkamide derivatives were synthesized using a divergent approach and the compounds were evaluated. Interestingly some of the derivatives showed a low micromolar activity against neuraminidases which could result in the development of kirkamidebased compounds towards this important class of enzyme. Overall, based on this and previously published data and the strong phenotype activity of kirkamide (1) on insects and arthropods, we hypothesize that the natural product target(s) is/are related to other metabolic pathways that await discovery.

General information, materials and equipment
All chemicals were purchased from Acros, Fluka or Sigma-Aldrich and were used without further purification (except for Et 3 N, which was freshly distilled before use). All reactions were carried out in heat gun-dried glassware (unless aqueous reagents were used) and reactions involving air sensitive compounds were performed under an argon or nitrogen atmosphere. Solvents applied for chemical transformations were either puriss quality or HPLC grade solvents, which had been dried by filtration through activated aluminum oxide under nitrogen (H 2 O content <10 ppm, Karl-Fischer titration). For work-up and purification solvents were distilled from technical grade. All synthetic transformations were monitored either by thin layer chromatography (TLC) or 1 H NMR spectroscopy. Yields refer to purified, dried and spectroscopically pure compounds. TLC was performed on Merck silica gel 60 F254 plates (0.25 mm thickness) pre-coated with a fluorescent indicator. Concentration under reduced pressure was performed by rotary evaporation at 40 C. Flash chromatography was performed using silica gel 60 (230e400 mesh) from Sigma-Aldrich with a forced flow eluent at 0.1e0.3 bar pressure. Preparative HPLC was carried out on a Shimadzu HPLC system and a fraction collector. All 1    Assays against the neuraminidase from C. perfringens were conducted in buffer (50 mM sodium citrate, 50 mM sodium phosphate, pH 5.0) using 2'-(4-methylumbelliferyl)-a-D-N-acetylneuraminic acid as substrate. Output of activity after enzymatic reactions were quenched was recorded using a Spectramax spectrometer (Molecular Devices) at 400 nm for those using 4-nitrophenolate as an output and a Varian CARY Eclipse Fluorescence Spectrophotometer 96-well plate system using excitation and emission wavelengths of 368 and 450 nm respectively, with 5 mm slit openings for those using 4-methylumbelliferonate as an output. Assays contained substrate at the previously determined K m value of the substrate for the enzyme, and the enzyme was at a concentration of 20 nM.

Cell culture and inhibition
COS-7 cells (Sigma Aldrich ECACC) were cultured in Dulbecco's modified Eagle's medium (DMEM) containing high glucose and no glutamine (Gibco) supplemented with 5% fetal bovine serum (Bovogen) and 2 mM Glutamax (Gibco). Cells were maintained in a humidified 37 C incubator with a 5% CO 2 atmosphere. Aliquots of inhibitor (stock in 95% ethanol) were delivered onto 100 mm tissue culture plates, and the ethanol was evaporated before cells were seeded onto the plate. The cells were incubated at 37 C for 24 h at which time they reached~80% confluence. The cells were washed twice with DPBS (Gibco), removed from the plates by trituration and pelleted (800Âg, 10 min). Cells were gently washed again with cold DPBS, (10 ml), and pelleted. The cells could be frozen at À80 C at this point. Control cultures without inhibitor were treated in the same manner.

Western blot analyses
Frozen cells pellets were thawed on ice, and lysed using cell lysis buffer (1 ml of 50 mM Tris, pH 8.0, containing 150 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1% Triton X-100, 0.5% sodium deoxycholate) and by inversion rotation at 4 C for 20 min. Cell debris were removed by centrifugation at 14,000 rpm in an Eppendorf 5430R microcentrifuge for 20 min at 4 C. The cell lysates were then either used immediately or aliquoted and stored at À80 C until required. SDS-PAGE loading buffer was added to an aliquot of each sample, and after heating at 95 C for 10 min, aliquots were loaded onto 10% Tris$HCl polyacrylamide gels. After electrophoresis, the samples were electroblotted to nitrocellulose membrane (0.45 mm, Bio-Rad). The membranes were blocked using 2% BSA (fraction V, Sigma) in PBST (PBS pH 7.4, 0.1% [v/v] Tween-20) and then incubated with blocking buffer containing the appropriate primary antibody, washed with PBST and then incubated with a horseradish peroxidise (HRP) conjugated secondary antibody. The membranes were washed again and detection of secondary HRP conjugate was done using a SuperSignal West Pico chemiluminescent detection kit (Pierce) and Chemidoc MP Imaging system (Biorad) (software Image Lab 5.2). For detection of O-GlcNAc modified proteins, mouse anti-O-GlcNAc IgM (clone CTD 110.6, BioLegend) was used as the primary antibody at a dilution of 1:2500, and goat anti-mouse-IgM-HRP conjugate (Santa Cruz Biotechnology) was used as the secondary antibody at a dilution of 1:2500. For detection of b-actin levels, mouse monoclonal anti-bactin IgG (clone AC-40, Sigma) was used as the primary antibody at a dilution of 1:4000, and rabbit anti-mouse-IgG-HRP conjugate (Jackson Immunoresearch) was used as the secondary antibody at a dilution of 1:10,000.

Dedication
This paper is dedicated to Professor Nuno Maulide on receipt of the Tetrahedron Young Investigator Award.

Declaration of competing interest
There are no conflicts of interest to declare.