Novel hydroxyl carboximates derived from β-elemene: design, synthesis and anti-tumour activities evaluation

Abstract A series of novel N-alkyl-N-hydroxyl carboximates derived from β-elemene were fortuitously discovered. Most of them showed more potent anti-proliferative activities than their lead compound β-elemene (1). Notably, compound 11i exhibited significant inhibitory effects on the proliferation of three lung cell lines (H1975, A549 and H460) and several other tumour cell lines (H1299, U87MG, MV4-11, and SU-DHL-2). Preliminary mechanistic studies revealed that compound 11i could significantly induce cell apoptosis. Further in vivo study in the H460 xenograft mouse model validated the anti-tumour activities of 11i with a greater tumour growth inhibition (TGI, 68.3%) than β-elemene and SAHA (50.1% and 55.9% respectively) at 60 mg/kg ip dosing, without obvious body weight loss and toxicity. Thus, such N-alkyl-N-hydroxyl carboximate class of compounds exemplified as 11i demonstrated potent anticancer activities both in vitro and in vivo, and should warrant further investigation for potential anticancer therapy.


Introduction
Elemene extract obtained from the rhizome of Curcuma wenyujin composes of a mixture of sesquiterpene. Among these elemene isomers, b-elemene is the most abundant one and is responsible for elemene's anti-tumour activities (Figure 1) 1,2 . In fact, elemene liposomal injection and elemene oral emulsion have been approved by the Chinese Food and Drug Administration (CFDA) for the treatment of various human cancers (Figure 1) 3,4 . Pharmacologically, b-elemene suppresses tumour cell growth via diverse effects including induction of apoptosis, autophagy and cell cycle arrest, and intervention of cell proliferation and migration [5][6][7][8] . Scientists also discovered that the combination of b-elemene with other immune drugs could enhance the body's immune response to tumours 9 .
Despite these attractive anticancer properties, the poor water solubility and moderate anti-tumour activities limit the maximisation of their clinical applications.
To improve the water solubility and the anti-tumour efficacy of b-elemene, scientists from various research groups including us have worked on the structural modifications of b-elemene from then to now [10][11][12][13][14][15][16][17] . Based on the characteristics of functional groups, b-elemene derivatives could be divided into the following classes: amines, esters, amino acid derivatives, ethers, alcohols, glycosides and organometallic compounds, etc 18 , as shown in Figure 2. In recent years, the dimer derivatives 19,20 (Figure 2, Ii) and NO donor derivatives 21,22 (Figure 2, Ij) of b-elemene were investigated and showed promising biological activities. However, the hybrid drugs derived from b-elemene and histone deacetylase inhibitors (HDACi) were rarely studied and reported.
Our previous work in multi-targeting drug discovery 23,24 prompted us to design the hybrid molecules of b-elemene by incorporating HDACi pharmacophore (hydroxamic acid). Thus, compound 2 was designed for such a purpose ( Figure 3). Based on the retro-synthesis analysis, the key precursor 3 could be obtained from allylic bromide 4 and THP-protected hydroxyl carboximate 5a (Figure 3). Unfortunately, the displacement reaction of 4 and 5a produced two products; none of them was consistent  with compound 3 despite of having the same molecular weight. Further investigation of the displacement products leads us to identify novel hydroxyl carboximate class compounds exhibiting potent anti-tumour activities. Herein, we wish to report this class of compounds, including their discovery, synthesis, and structureactivity relationship as well as in vivo anti-tumour efficacy.

Discovery of novel N-alkyl-N-hydroxyl carboximates
The discovery of N-alkyl-N-hydroxyl carboximate 11a (Figure 4) was an accidental process. The hybrid compound, such as 2 was originally designed from b-elemene (1) and Suberoylanilide hydroxamic acid (SAHA or Vorinostat, a histone deacetylases (HDACs) inhibitor approved in 2006). The intention of reacting the amino group at the 2-position of pyridyl with allylic bromide 4 apparently did not go in the desired direction as we expected ( Figure 3). Indeed, the alkylation produced two products with very close polarity on the TLC plate. They were separable only after multiple cycles of flash column chromatography in about 1:2 ratio (more polar product vs less polar compound). Structure elucidations of these two products were not straightforward even though several 2D NMR experiments were performed. This was due to many overlapping 1 H NMR signals for both of them. Therefore, they were independently taken into the THP-deprotection step, and the products were characterised respectively. The less polar compound was deprotected to afford 1 and 10a while the more polar compound to afford 11a. The structure elucidations of 1, 10a and 11a helped us to postulate the two products in the alkylation step. This became explainable that 9a (R f values 0.3 in TLC plate (petroleum ether-ethyl acetate 2:3, v/v)) was derived from the alkylation at the nitrogen atom of carboximate (more polar compound), while 8a (R f values 0.4 in TLC plate (petroleum etherethyl acetate 2:3, v/v)) was derived from the alkylation at the oxygen atom of the tautomer of carboximate (enol tautomerization form, less polar compound). The novel compound 11a showed potent anticancer activities against three lung tumour lines (H1975, A549 and H460) in vitro anti-proliferative assay.

Chemistry
The interesting biological activities and the easy accessibility of novel compound 11a encouraged us to investigate more on Nalkyl-N-hydroxyl carboximate class analogs of b-elemene. In order to accomplish the work, we first needed to prepare 13-Br-b-elemene (4) in a workable amount (Scheme 1). Compound 4 was previously prepared from the corresponding allylic alcohol by Xu et al. 19 using NBS/Ph 3 P condition. This route not only added one extra step of alcohol preparation but also required tedious separation of close related isomer in allylic alcohols. We worked out an alternative route of direct bromination at the 13-position (one of the allylic positions) of 1. By following the procedure in our patent application 25 , compound 4 was prepared in 30.1% yield with good purity, enough for the next step displacement reaction. The  minor isomer 14-Br-b-elemene (6) did not interfere with the following step reaction.
As mentioned above, the displacement of THP-protected Nhydroxyl carboximate with allylic bromide 4 yielded two isomers (i.e. 8 and 9), close to each other on the TLC plate due to the similarity in polarity. Attempt to separate them using dichloromethane (DCM) and methanol (MeOH) mixed solvent in flash column chromatography did not give a satisfactory result. After several tries, we figured out the separation of these two isomers could be achieved by using petroleum ether (PE)-ethyl acetate (EA) as eluent in silica gel column for 2-3 separation cycles.
The removal of the THP group of 9 to afford compound 11 would be easily achieved using acid-catalyzed deprotection conditions. An attempt to purify 11 in silica gel column chromatography using a methanol-dichloromethane system proved to be unsuccessful. Finally, reversed-phase (C18) column chromatography was applied for the separation to afford pure compound 11 in good yields.
Thus the proposed synthetic route for N-alkyl-N-hydroxyl carbomixates 11 (Scheme 1) is now practical for medicinal chemistry purposes. The synthesis steps for the target compound 11 were detailed in Scheme 2. The bromination of b-elemene (1) was achieved using NBS in acetic acid in 30.1% yields with good selectivity of 13-position over 14-position (4 : 6 ¼ 17 : 3). Small amount of isomer 6 did not interfere with the following step reaction. Subsequently, the intermediates 5a-5j could be easily prepared from the corresponding carboxylic acids 7a-7j and O-(tetrahydro-2H-pyran-2-yl)hydroxylamine (NH 2 OTHP) under the standard amide coupling reaction conditions such as HOBt/EDCI/DIPEA. The group "R" on different positions of the aromatic ring of carboxylic acids 7a-7j were designed to examine the substituent effects based on the principle of bioisosterism in medicinal chemistry (eg: ÀNH 2 vs ÀOH, ÀCH 3 ; ÀCN vs ÀCF 3 ) (Scheme 2). The displacement reaction between 5a-5f and 5h-5j and allylic bromide 4 could be achieved under the condition of Cs 2 CO 3 /DMF at 60 C overnight. Finally, the THP-deprotection of 9 under acidic condition TsOHÁH 2 O/MeOH at room temperature, afforded the target compounds 11a-11f and 11h-11j in 50.6% to 94.0% yields.
In addition, compound 15 was designed for comparison with compound 11i. The synthetic route for 11 was modified in the allylic bromide displacement step using N-Boc-cyclopropylamine (12) instead of 5, as shown in Scheme 1. The resulting intermediate 13 underwent a Boc-deprotection process, yielding compound 14. Subsequent condensation of 14 with 7i afforded the desired product 15.
Intermediate 5g (R ¼ para-OH) possesses at least three reactive centres which could potentially react with intermediate 4. In fact, the designed analog 11g was unable to obtain using the above synthetic route. During the displacement step, double alkylation occurred and compound 16 was isolated. This was explainable that Cs 2 CO 3 was capable to deprotonate both phenolic proton and nitrogen proton of THPO-NH-C(O)-group. The resulting dianion attacked the Br-bearing allylic carbon of 4 to facilitate the double alkylation. The final deprotection of 16 gave the novel dimer derivative 17, which was also subjected to biological testing. It should be noted that compound 19, an analog of 11i with saturated a,bposition of carboximate group, could not be synthesised from 18 and 4 using the above synthetic route. Apparently, the alkylation at the nitrogen of THPO-NH-C(O)-did not proceed as expected. By comparing with the majority of substrates 5 (no matter n is 0 or 1), one could postulate that the conjugated component (CÀC double bond or aryl group) directly attaching to THPO-NH-C(O)-was necessitated for N-alkylation (Scheme 3) to generate novel N-alkyl-N-hydroxyl carboximate derivatives of b-elemene.

In vitro anti-tumour activities
The in vitro anti-tumour activities of all the compounds against three human lung tumour cell lines H1975, A549, and H460 were determined (Table 1). To investigate the effects of substituted positions of the phenyl ring on anti-proliferative activities, compounds 11b-11d were synthesised. The results suggested that the substitution position on the phenyl ring has little effect on the activities (i.e. methyl group at ortho-, meta-, and para-positions). We chose para-substituted on the phenyl ring for further analogs exploration. Compounds 11e-11f and 11h were synthesised and compared with 11b. The results revealed that replacing the methyl group on the phenyl ring of 11b with CN, CF 3 and NH 2 did not have much influence on tumour inhibition effects (11e, 11f and 11h). Considering the price of precursor drugs with CN and CF 3 , the 11h displayed 1-3 fold stronger inhibitory activities against tumour cell lines compared to 11b, so we used compound 11h as a new starting point for further structural modifications. It should be noted that the para-OH analog 11g was designed but we were unable to obtain the compound. The displacement step of 5g with 4 yielded 16, a double alkylation product at both N atom and phenolic oxygen, despite of the presence of the excess amount of 4 (2 equivalent). The deprotection of 16 gave compound 17. Unfortunately, 17 showed poor inhibitory effects against three human lung tumour cell lines (IC 50 > 30 lM). As mentioned above, compound 11h has better activity than 11b, and could serve as a new starting point for further optimisation. Since the aniline group is a known functional group for carcinogenicity, we decided to incorporate a nitrogen atom on the phenyl ring to mitigate such liability. Analog 11i was designed and synthesised for head-to-head comparison in biological testing. Notably, 11i exhibited more than 6-fold potency enhancement compared to 11h in anti-proliferative activities against A549 cells.
To determine which portion of 11i contributed more to the activity, we designed two analogs 11j and 15. The absence of NH 2 on pyridyl ring (i.e. 11j) tenuate the activity significantly (17.56 mM vs 1.67 mM to H1975 cell lines, 2.5 mM vs <1 mM to A549 cell lines, 5.71 vs 1.37 mM to H460 cell lines) compared to 11i, while the replacement of N-hydroxyl group with N-cyclopropyl group (i.e. 15) significantly reduce the anticancer activities in both H1975 (IC 50 > 30 mM) and H460 cells lines (IC 50 > 30 mM). These results pointed to a conclusion that both N-hydroxyl carboximate moiety and NH 2 group on pyridyl ring were very important contributors for good biological activities of 11i.
Another important factor for biological activity could be the CÀC double bond between N-hydroxyl carboximate and the pyridyl ring. This is supported by the biological data of 11i and 11a. Compound 11a, a close analog of 11i without a CÀC double bond next to N-hydroxyl carboximate moiety, exhibited more than 15-fold activity loss compared to compound 11i.
In summary, compound 11i possessed several important elements including N-hydroxyl carboximate, CÀC double bond, and 2pyridyl amine group, all contributing to its potent biological activity. This compound also exhibited potent anti-proliferative activities against several tumour cell lines, including H1299 (lung cancer cells), U87MG (malignant glioma cells), MV4-11 (hematogical cells), and SU-DHL-2 (lymphoma cells) (Table 2). Thus, compound 11i progressed to further preliminary mechanistic and in vivo studies.

In vitro induced cell apoptosis
To investigate the effect of the selected representative compound on the induction of apoptosis, compound 11i was evaluated by annexin VFTIC/propidium iodide (PI) assay. H460 cells were incubated with vehicle alone or tested compounds at 10 lM for 72 h. As shown in Figure 5, compound 11i could significantly induce H460 cell apoptosis. The percentage of apoptotic cells for compound 11i at the concentration of 10 lM is greater than 50%, which was obviously higher than lead compound b-elemene 1 (10.94% apoptotic cells at 10 lM, p < 0.05).

In vivo anti-tumour activities against H460 xenografts
To investigate the in vivo anti-tumour effects, xenograft model of H460 was set up. First, compound 11i was tested in the H460 xenograft model. H460 cells (1 Â 10 6 ) were implanted on right flanks subcutaneously in female nude mice. When the implanted tumour reached a volume of 80-100 mm 3 , the animals were randomly divided into groups of 4 and compound 11i was administered intraperitoneally at 60 mg/kg once a day for 21 consecutive days. Compounds 1 and SAHA were used as the positive controls. As shown in Figure 6(A,C,D) and Table 3, treatment with compound 11i caused significantly reduction in tumour growth, which showed higher tumour growth inhibition (TGI) values (68.3%) and lower treatment-to-control (T/C) values (31.18%) than positive drugs 1 (b-Elemene, (TGI ¼ 50.1%; T/C ¼ 50.25%)) and SAHA (TGI ¼ 55.9%; T/C ¼ 45.28%) in the xenograft H460 model. Moreover, compound 11i was observed to be well tolerated during the test and no significant loss of body weight was observed ( Figure 6B), indicating that its toxicity is low.

The solubility and solubility levels of the target compounds
Accelrys Discovery Studio (DS) software was used to assess the solubility level of all final products. It's not a surprise that the solubility of many target compounds (i.e. 11b-11d, 11f, 17 and 15) was as poor as that of b-elemene due to the highly lipophilicity of the parent. But it's encouraging that compounds 11a and 11h-11j have slightly improved solubility compared to b-elemene ( Table 4). As expected, compounds 11h, 11i and 15 exhibited much improvement of solubility level when they were in their hydrochloride forms (Table 4). Among all compounds examined in the DS software, compound 11i is the best in term of solubility level whether it is in the free state or in its HCl salt form.

Conclusion
Structure modifications of natural products, especially those with interesting biological activity profiles such as b-elemene have been one of the hot areas in modern drug discovery. We reported herein a novel class of N-alkyl-N-hydroxyl carboximate derivatives of b-elemene with attractive anticancer activities both in vitro and in vivo. The representative compound 11i not only exhibited marked anti-tumour effects against 6 tumour cell lines, but also possessed potent proapoptotic activity. In in vivo study, compound 11i showed strong anti-tumour efficacy in the H460 xenograft mice model without observable toxicity. Although the b-elemene analogs described in this article exhibited much improved anti-tumour activity, their precise biological targets are yet to be uncovered. Nevertheless, this study expanded the structure scope of b-elemene modifications beyond those reported and provided further optimisation for medicinal research around natural products such as b-elemene.

General information
All of the chemicals were purchased from commercial suppliers. The melting points of the compounds were determined using B€ uchi B-540 capillary melting point instrument. 1 H NMR (500 MHz) and 13 C NMR (126 MHz) were recorded on a 500 MHz Bruker NMR spectroscopy using CDCl 3 , CD 3 OD or DMSO-d 6 as the deuterated solvent. Chemical shifts (d) were reported in parts per million (ppm) relative to residual solvent as an internal reference. Lowresolution mass spectra were recorded with Agilent 1260 Infinity II/1625. High-resolution mass spectra (HRMS) were measured on a 1.34 ± 0.29 1.82 ± 0.04 1.63 ± 0.05 a Data represent the mean value from at least three independent experiments. SAHA was used as positive control for in vitro assay. Bruker MICR OTOF-Q II instrument or Shimadzu LCMSIT-TOF mass spectrometer using the ESI technique. High performance liquid chromatography (HPLC) was determined on a Shimadzu LC-2030Plus instrument.  monitored by TLC. The mixture was then neutralised with saturated NaHCO 3 solution and extracted with petroleum ether (3 Â 50 mL). The combined organic layers were washed with water and brine, and dried over Na 2 SO 4 . The drying agent was filtered off. The filtrate was concentrated under reduced pressure, and the residue was purified via flash column chromatography (petroleum ether as eluent) to give 13-Br-b-elemene 4 (2.51 g, yield 30.1%) as a light yellow liquid. 1 40 mmol), HOBt (7.20 mmol) and the corresponding acids 7 (5.54 mmol) in DMF (20 mL) was stirred at room temperature for 6 h. The reaction was monitored by TLC. Upon completion, the mixture was quenched with water and extracted with ethyl acetate (3 Â 100 mL). The combined organic layers were washed with water and brine and dried over Na 2 SO 4 . The drying agent was filtered off. The filtrate was concentrated under reduced pressure and the residue was purified via flash column chromatography (dichloromethane/methanol 9:1, v/v) to give compounds 5.