Derivatisation of parthenolide to address chemoresistant chronic lymphocytic leukaemia

A parthenolide-derivative with favourable pharmacokinetic properties and good activity against drug-resistant chronic lymphocytic leukaemia is reported.

Ele na G. M er isor, ( b ) M ich ae l J. Mo rto n, ( i ) Al ex S . Qu y, ( a) Ru th Rober ts, ( i , j) R osa nn a Sc arl l, ( e) Timot hy S chu lz-Ut erm oeh l, ( b ) Tat jana S tanko vic, ( e) Br ett S tev ens on, ( b ) Jo hn S . F ossey, ( a † ) * Ang elo Ag at hanggel ou, (     automatic polarimeter at room temperature. The X-ray crystal structure included was previously reported; for information, it was determined using an Agilent SuperNova X-ray diffractometer with an Atlas detector (wavelength 1.5418 Å). Column chromatography was carried out using standard flash column chromatography and a Combiflash Rf 200i (stationary phase silica), chromatograms were recorded by evaporative light scattering detector (ELSD) and absorbance at two wavelengths (254 nm and 280 nm). Reactions were monitored by thin layer chromatography (TLC) on Merck silica gel 60 F254 plates. TLC plates were visualised by either UV light with 254 nm / 365 nm, a methanolic solution of ninhydrin or with potassium permanganate. Where dry solvents were required, pre-treatment with molecular sieves 3 Å was sufficient to permit reactions to proceed with good conversations.

Chemistry efforts at Sygnature Discovery (where differing to University of Birmingham)
NMR spectra were recorded on a Bruker AVIII 400MHz NMR spectrometer at room temperature. Mass spectra were recorded on Waters Acquity UPLC, CSH C18, 1.7 μm, 2.1 x 30 mm column, 3 minutes acidic method as described follows: Analytical UPLC/MS was carried out using a Waters Acquity UPLC, CSH C18, 1.7 μm, 2.1 x 30 mm column eluting with a gradient of 0.1% formic acid in MeCN in 0.1% formic acid in water. The gradient is structured with a starting point of 5% MeCN held from 0.0-0.11 minutes. The gradient from 5-95% occurs between 0.11-2.15 minutes with a flush from 2.15-2.56 minutes. A column re-equilibration to 5% MeCN is from 2.56-2.83 minutes. UV spectra of the eluted peaks were measured using an Acquity PDA (scan from 150-800 nm) and mass spectra were recorded using an Acquity QDa detector with ESI pos/neg switching.

Biological efforts at University of Birmingham
Tissue culture, MEC1 cells were obtained from the American Type Culture Collection (Manassas, VA 20110 USA) and were cultured in RPMI 1640 medium (Sigma-Aldrich, Irvine, UK) with 10% fetal bovine serum (Sigma-Aldrich).
The alamarBlue® cytotoxicity assay was conducted by seeding MEC1 cells in triplicate at density of 25000 cells/well in a 96 well plate, final volume of 200 μL. Following treatment with test compound, viability was determined by measuring the reduction of resazurin. Resazurin solution was added to each well at a final concentration of 50 μg/mL and incubated for 3 hours at 37 °C with 5% CO2. Reduction of resazurin was determined by measuring absorbance at 590 nM using a PheraSTAR FS plate reader (BMG Labtech). Cell viability was calculated as a fraction of the untreated cells after subtracting background fluorescence of resazurin in media only. Data is presented as the mean of five independent experiments and significance was determined by Student's t-test. MEC1 cells were obtained from the American Type Culture Collection (Manassas, VA 20110 USA) and were routinely cultured in RPMI 1640 medium (Sigma-Aldrich) supplemented with 1% L-glutamine (Sigma-Aldrich), 1% Penicillin/Streptomycin (Sigma-Aldrich) and 10% FBS (Gibco). The MEC1 proliferation assay was conducted by performing a re-feed of the MEC1 cells 24 h prior to cell seeding. On the day of the assay, cells were seeded into a 384-well assay plate at a density of 12,500 cells/well in 25 μL of complete MEC1 medium. Compound treatment was performed on the day of cell seeding, with compounds prepared as duplicate compound response curves (final DMSO concentration of 0.6%). Post compound treatment, MEC1 assay plates were returned to a 37 °C/5% CO2 incubator for 24 h of incubation. Following compound treatment, cellular viability was determined by measuring reduction of alamarBlue® reagent (Invitrogen). alamarBlue® reagent (2.5 μL) was added to each assay well to give a 1:10 dilution of the neat stock and assay plates were incubated for 3 h at 37 °C/5% CO2. Reduction of alamarBlue® was determined using an Envision Xcite plate reader (Perkin Elmer) with an excitation wavelength of 560 nm set and an emission wavelength of 590 nm set. Cell viability was determined by normalising raw fluorescent data to a cell only high control and Parthenolide treated low control to provide percentage viability readout. Data was plotted and EC50 values extracted and reported EC50s represent an average of at least two independent assays.

Growing medium
Seeds were sown in Petersfield peat-free supreme compost; a general seed and potting compost produced from well composted finely screened bark. Compost produced by W E Hewitt & Son.

Seed sources
A variety of sources of feverfew, seeds have been exploited over the period 2013-2018. The commercial supplier and the products purchased are listed below, the suppliers' name for the seeds are given. Feverfew of the family: asteraceae; genus: Tanacetum can be annuals, evergreen or herbaceous perennials or sub-shrubs, with simple or pinnately divided leaves and solitary or clustered, terminal, daisy-like or button-like flower-heads; detail: T. parthenium is a short-lived bushy perennial with pungently aromatic, ovate, pinnately lobed leaves and daisy-like flowerheads 2.5 cm across, with white rays and yellow disk florets in summer; plant range: Balkans, 1 is also known by alternative common names including Altamisa, Bachelor's Buttons, Featherfew, Featherfoil, Flirtwort Midsummer Daisy, Grande Camomille and Santa Maria. Tanacetum parthenium is recognised as having had 26 synonyms. 2 Seeds purchased were listed by seed merchants as Tanacetum parthenium (feverfew), which has daisy-like flowers. Since 2017 seeds sown consistently produced button-like flowers (Tanacetum parthenium 'Flore Pleno') instead of the single flowers described for the species (Tanacetum parthenium), regardless of the supplier. Seed was grown in an unheated glasshouse. The sowing date was determined by temperature. Tanacetum germinates in 7-10 days at temperatures between 15-22 °C. 3 Sowing dates varied between January and April. Seed was sown into plastic modular trays of 10 x 5 modules (each tray 51.5 cm long x 29.5 cm wide by 4.5 cm in depth). Three to four seeds were sown per module and left uncovered. Mains tap water was used for watering the seed trays, which were checked daily. Watering was carried out using a Haws 170/1.5 watering can with fine brass rose. Following germination, seedlings were thinned out using forceps. One seedling was left per module to grow on.

Potting on
When the seedlings had developed roots that filled each module, they were pricked out into 10 cm pots (Teku VDC 10 plastic pots) containing Petersfield peat-free supreme compost. The feverfew in 10 cm pots were moved to a sheltered position in between two glasshouses to harden off (subject to weather conditions) once they had put on growth. Growth was assessed by increase in height/width of plant and tapping individual plants out of their pots to check root growth.

Planting out
From the outset of the project, once plants had a well-developed root system, they were planted out on to a west facing, sloping and free draining site. The site was previously used to grow another member of the compositae; Anthemis. The soil had previously had grit added but no feeding or mulching had taken place for several years. Plants were spaced 50 cm apart, with 50 cm between rows. They were watered in once planted and checked for watering while they were establishing roots, and in periods of hot weather.
Weeds were a problem on the site, so plants were mulched with woodchip to reduce weeding. A large number of plants would grow for two years on this site-feverfew tends to be grown as an annual but will grow as a short-lived perennial in ideal conditions.
In 2017 the growing area changed to accommodate more plants. The new site was also on a west facing slope. It was less free draining than the previous site and has been regularly improved with feed and compost, for use growing vegetables in the past. Feverfew plants were grown at 30 cm spacings to provide dense cover with the intention of shading out weed competition. Plants grew well but did not come through the winter of 2017/18 making it necessary to replant the field.
Parthenolide extraction procedure Plant material (feverfew) was chopped into 1 to 5 cm pieces by hand and allowed to dry, laid out on plastic sheets in a glass house. Once dry to touch, and taking care not to discard any seeds, the dried material was transferred to vacuum sealed containers, and stored until an appropriate time to conduct the next, extraction step. To the dried plant material hot water (10 mL per 1 g plant matter) was added and the mixture stirred at 80 °C for 10 minutes. The solution was allowed to cool to room temperature and the cool broth thus obtained decanted, this procedure was repeated twice more on the same batches of plant material. The combined aqueous broth extracts were filtered. The filtrate was extracted three times with chloroform (3 x vol 1:1). The majority of the solvent from the combined organic extracts was removed in vacuo to afford a dark green sticky residue. The residue thus obtained was purified by flash chromatography (hexane/ethyl acetate) and recrystalised to purity from with hexane/ethyl acetate to afford colourless crystals of parthenolide 1. A typical procedure used 6.5 kg of dried plant material delivered 8.53 g of analytically pure parthenolide (0.13 wt%).
Parthenolide was isolated following the above extraction and isolation procedure. 1

Synthesis of N-Boc-piperidine-4-carboxylic acid (8) 12
To an ice slush bath-cooled aqueous solution of sodium hydroxide (50 mL, 1 M) a solution of piperidine-4-carboxylic acid (6.46 g, 50 mmol) in THF (30 mL) was added dropwise. To which, a solution of di-tertbutyl carbamate (19.10 g, 87.5 mmol) in THF (30 mL) was next added, dropwise. After stirring for 30 min, the solution was allowed to warm to RT and was stirred for ~18 h for convenience. The resulting mixture was concentrated in vacuo to half its original volume. The resulting mixture was adjusted to pH 5-6 (universal indicator paper) by addition of aqueous HCl (1 M). This resulted in the formation of a white precipitate that was collected by filtration and washed with water. In order to remove any insoluble material, the precipitate was dissolved

Synthesis of tert-butyl 4-(3-ethoxy-3-oxopropanoyl)piperidine-1-carboxylate (9) 12b, 14
To an ice slush bath-cooled 2:1 mixture of tetrahydrofuran and acetonitrile (60 mL) compound S4 (0.85 g, 5 mmol), magnesium chloride (0.95 g, 10 mmol, 2.0 eq.) and DMAP (61 mg, 0.8 mmol, 0.1 eq.) were added. The mixture was allowed to warm to room temperature and was stirred for 5 h. In an additional ice slush bath-cooled flask containing tetrahydrofuran (40 mL), compound 8 (1.14 g, 5 mmol, 1.0 eq.) and carbonyl diimidazole (1.05 g, 6.5 mmol, 1.3 eq.) were added. This flask was also allowed to warm to room temperature and was stirred for 4 h. The mixtures were cooled to 0 °C and the latter was added to the former, the resulting mixture was stirred for 18 h (for convenience). The resulting mixture was concentrated in vacuo to half its original volume. The resulting mixture was adjusted to pH 4 (universal indicator paper) by addition of

Synthesis of tert-butyl-4-(6-hydroxy-2-methylpyrimidin-4-yl)piperidine-1-carboxylate (10) 15
Acetamidine hydrochloride (260 mg, 2.75 mmol) and potassium carbonate (633 mg, 4.58 mmol) were dissolved in water (7 mL). To which a solution of compound 9 (549 mg, 1.83 mmol) in methanol (7 mL) was added. The mixture was stirred at 70 °C for 12 hours, after which time a saturated aqueous solution of ammonium chloride (7 mL) was added. The resulting mixture was extracted with dichloromethane (3 x 10 mL) and washed with water. The combined organic fractions were dried over anhydrous sodium sulphate, 16     In order to explore some plausible mechanisms for the activity of the selected amino parthenolide derivatives, docking studies were performed. One aspect of interest is whether the compounds act as prodrugs, that function to deliver parthenolide to a site of action, and this was explored by docking derivatives 2a, 5a and 5f and also the drug they all feasibly liberate, parthenolide 1 (Supplementary Scheme 2). One possibility that has been considered is that the prodrugs will fragment in a protein binding site yielding the drug if there are appropriate conditions to facilitate the proton-shuffling required for the elimination of the amine. Further reactivity is likely if this elimination occurs in the presence of a cysteine thiol. When reactions occur in the diffusion limiting environment in protein binding sites, the eliminated amine could act as a base to deprotonate the thiol and promote its addition to the Michael-acceptor that had been revealed by loss of the amine.

Supplementary Scheme 2. Proposed retro-Michael-type reaction that delivers parthenolide (1) from amino parthenolide derivatives.
Two biological mechanisms by which the active compounds might act have been considered computationally; one or both of these might contribute to the activity reported above. The first mode of action is inhibition of the binding between DNA and the p65 domain of NF-B. This mechanism has been shown to contribute to the anti-inflammatory activity of 1 and other reactive compounds, it is dependent upon Cys 38, presumably alkylation of its thiol lies at the origin of this observation. 20 Double reactivity involving a second reaction with Cys 120 had been proposed but mutation studies supported only a minor role for interaction with this residue. In order to investigate and illustrate this mode of action, a relevant protein structure was sought. The highest resolution structure of the p65 subunit of NF-B in which Cys 38 is present was found by searching the protein data bank and has the structure code 2RAM. 21 The receptor structure was prepared by deleting the DNA and selecting only protein chain A. Docking employed the Autodock Vina code, 22 with protein structures prepared using the suite of Autodock tools using default settings. 23 The binding site was defined by a box centred on the sulfur atom in Cys 38. The box was selected to have length 18 Å based on a longest atom-atom distance of 9 Å in the crystal structure of 1 (detailed in Supplementary Figure 1) Amino parthenolide pro-drug protected from unwanted side reactions with glutathione by amine which could also engender favourable pharmacological properties conformation found in this crystal structure was used as the input geometry for this ligand while the other structures were obtained by editing this structure using the Gaussview 24 program and converting them to the appropriate input format with Open Babel. 25 The docking pose that was scored best for each compound (1, 2a, 5a and 5f) is shown in Supplementary Figure 3 (all compounds achieved docking scores in the range -4.0 to -5.0 where more negative scores are preferred). This reveals that parthenolide 1 can bind in close proximity to Cys 38 in such a way that it is primed to react; the relevant intermolecular S-C distance is highlighted and is 4.1 Å. There is a salt bridge between Glu 39 and Arg 187 that is in close proximity to Cys 38 and the three amines that have been docked (2a, 5a and 5f) are all found to bind in identical binding modes in which the protonated amine is proximal to Glu 39. This mode could plausibly permit the sidechain carboxylate of Glu 39 to deprotonate at the -carbon adjacent to the carbonyl in each of these derivatives and therefore promote elimination to form 1 in situ. The second mode of action that was considered is inhibition of the Inhibitor of B Kinase, IKK . IKK phosphorylates I B targeting it for proteosomal degradation, thus releasing the NF-B heterodimer p50/p65 which translocates to the nucleus. Parthenolide 1 and related compounds have been shown to reduce phosphorylation of p65 in cells taken from acute myelogeous leukaemia (AML) patients and inhibition of IKK would contribute to this effect. 26 A structure of this kinase in complex with a pyrimidine derivative inhibitor is reported, protein data bank entry code 3RZF, the inhibitor's structure is depicted in the supporting information (compound S5, Supplementary Figure 5). 27 This structure reveals that despite having a hinge-binding motif (adjacent donor and acceptor, a typical binding mode for kinases), the inhibitor in structure 3RZF does not exploit this. The pyrimidine core of that literature compound (S5) binds adjacent to Cys 99, which raised the possibility that parthenolide (and derivatives thereof) described above might act covalently. In order to investigate these possibilities, docking was performed using the coordinates of the sulfur atom in Cys 99 as the centre of the box with a box side of length 18 Å. The docking scores for the poses discussed below for 1, 2a and 5a were in the range -5.8 to -6.1 while that for 5f was -4.5.
The nine best scoring poses of 1 were examined to identify the one that places the Michael acceptor in closest proximity to the cysteine thiol of Cys 99 (Supplementary Figure 4: upper left). The cysteine in this case has its sidechain rotated in towards the rest of the protein and so long as its motion is not hindered, it could presumably swing out. The distance between the sulfur atom and the -carbon of the Michael acceptor is 6.5 Å. For the amino parthenolide derivatives considered (2a, 5a and 5f), one docking pose was identified that is common to all of the compounds studied The docking studies support a mechanism of action via liberation of parthenolide because compound 5f is unlikely to bind better than 1 to either of the proteins considered. Compound 1 could plausibly bind in a covalent fashion to the p65 domain of NF-B and to IKK . Both of these interventions in the same pathway would be consistent with the observed activity.

Docking scores
Docking scores pertaining to the main text computational study are given in Supplementary Table 3.

Mass spectrometry of lead compound 5f
Included as a representative example of the MS data reported in this manuscript, scanned versions of the MS and HRMS data obtained for 5f are included herein. Scanned from A4 printouts obtained at the time of sample preparation. The scans of the black and white print outs have been cropped and brightness adjusted. Note the processing software for generating the HRMS report uses neutral formal as predicted mass, this ignores charge and mass of the electron.