Chemical, molecular and structural studies of Boswellia species: β-Boswellic Aldehyde and 3-epi-11β-Dihydroxy BA as precursors in biosynthesis of boswellic acids

The distribution and biosynthesis of boswellic acids (BAs) is scarce in current literature. Present study aims to elucidate the BAs biosynthetic and its diversity in the resins of Boswellia sacra and Boswellia papyrifera. Results revealed the isolation of new (3β, 11β-dihydroxy BA) and recently known (as new source, β-boswellic aldehyde) precursors from B. sacra resin along with α-amyrin. Following this, a detailed nomenclature of BAs was elucidated. The quantification and distribution of amyrins (3-epi-α-amyrin, β-amyrin and α-amyrin) and BAs in different Boswellia resins showed highest amyrin and BAs in B. sacra as compared with B. serrata and B. papyrifera. Distribution of BAs significantly varied in the resin of B. sacra collected from dry mountains than coastal trees. In B. sacra, high content of α-amyrin was found in the roots but it lacked β-amyrin and BAs. The leaf part showed traces of β-ABA and AKBA but was deficient in amyrins. This was further confirmed by lack of transcript accumulation of amyrin-related biosynthesis gene in leaf part. In contrast, the stem showed presence of all six BAs which are attributed to existence of resin-secretory canals. In conclusion, the boswellic acids are genus-specific chemical constituents for Boswellia species albeit the variation of the amounts among different Boswellia species and grades.


Introduction
Over the centuries, humanity has known and utilized frankincense since ancient times for religious, social and therapeutic purposes [1]. Boswellia sacra Flueck (Burseraceae) is an endemic tree to Oman and an economically important species of genus Boswellia [2,3]. B. sacra is a good source of high quality frankincense and bioactive compounds having a widerange of vital biological activities. The frankincense, a yellowish-brown oleo-gum resin and its essential oil have been well-known for their ameliorative effects against human ailments PLOS

HPLC-DAD analysis
The BAs amount was quantified by a parallel liquid chromatography (1260 HPLC, Agilent Technologies; Japan) connected with a reverse phase C18 column (3.9 × 150 mm; Waters; Nova-Pak, USA). The samples were dissolved (10 mgÁmL -1 ) in HPLC-grade methanol, and then properly diluted. The chromatographic separation was carried out at a constant flow rate of 1 mL min -1 with the following conditions: Mobile phase A = 99.9% water with 0.1% acetic acid; Mobile phase B = 99.9% acetonitrile with 0.1% acetic acid; Total running time (stop time) = 30 mints for BAs and 35 mints for amyrins; Washing time (post time) = 1 minute; Injection volume = 20 μL; Column type = C18, Waters; Column temperature = 40˚C; DAD Signal = wave length of 254 for BAs and 210 for amyrins; Band width = 4.0; Reference wave length = 254 nm, Reference bandwidth = 100 nm. free falcon tubes having extraction buffer (NaCl-0.25 M, Tris-HCl-0.05M, pH 7.5; EDTA -20 mM, sodium dodecyl sulphate-1% and PVP-4% as described by Chan et al. [49]. RNA with good integrity and purity was used to synthesize cDNA through DiaStar™ RT kit (SolGent, Korea) according to the provided manufacturer's standard protocol. The gene expression of αamyrin synthase (Forward 5'-CTTTGTGGTCCTGCTGGTAA-3'; Reverse 3'-TGGCTTCAC ATTTGGAAGAG-5') and Squalene-synthase (Forward 5'-GAGGCACCCAGATGATCTTT-3'; Reverse 3'-GAGCGAAACTTCTGGAGACC-5') was further evaluated through RT-PCR. A 2x Real-time PCR Kit (BioFACT TM Korea) including 10nM of each gene specific primer with 100 ng of template cDNA in a final volume of 20 μL reaction mixture was used. The whole reaction was carried out according to manufacturer's standard protocol using Eco™ Real-Time PCR (Illumina™ USA) with a "no template control" (NTC) as a negative control. The expression of each gene was compared with relative expression of Actin as internal control and the experiment was repeated three times.

Microscopic analysis of stem
The bark samples collected from B. sacra tree (epidermal and cambium region; 20 mm 2 ) were fixed in Karnovsky solution (Karnovesky = 2.5% gluteraldehyde in sodium. cacodylate buffer and Gluteraldehyde) dehydrated in increasing ethanol series [30,50,70,90, and 100% (three times)] and then infiltrated with resin ethanol for polymerization [acrylic resin glycolmethacrylate and 100% ethanol at a ratio of 1:1 and then pure resin]. The blocks were cut on a microtome into 10um size. Similarly, samples for SEM were fixed, dried to the critical point with CO 2 and gold sputtered prior to observation. Details about light microscopy and scanning electron microscopy (SEM) are described by Peckys et al. [50].

Statistical analysis
The quantification of BAs and RT-PCR related analysis were performed in triplicate and represented with standard deviation. The Non-metric multidimensional scaling (NMDS) which analyses the quantities of BAs in different parts of the tree and different kinds of resin, was performed by PAST (v3.01; University of Auckland, New Zeeland). One-Way ANOVA analysis was performed using Graph Pad prism (v6.01) to identify the significant samples.
The 1 H NMR spectrum confirmed the presence of hydroxyl group at C-3 and was in β-orientation as evidenced by the doublet of doublet (11.4, 4.8 Hz) of the α-positioned geminal proton which appeared at δ H 3.24, an interpretation and β-orientation further substantiated by HMBC correlation between H-5 (δ H 1.54) and C-3 (δc 78.8) and NEOSY correlation between Chemical, molecular and structural studies of Boswellia H-3 and CH 3 -23 position. On the other hand, the singlet peak at δ 5.52 (H-12) correlated with C-11(δc 81.8), C-9 (δc 49.0), C-13 (δc146.3) and C-14 (δc 42.0) in the HMBC spectrum allowed its assignment to olefinic double bond between C-12 and C-13.
Besides olefinic double bond, the 1 H NMR spectrum of 1 displayed a peak at δ H 4.53 previously observed in boswellic acid derivatives [51]. The three bond long range correlations from this signal to C-9 (δc 49.0), C-12 (δc 124.5), and C-13 (δc 146.3) in the HMBC spectrum allowed its assignment to H-11. This proton was associated with a carbon signal at δ 81.8 in the HSQC spectrum and showed COSY correlation with a signal at δ H 1.81 (H-9) and 5.52 indicating the location of hydroxyl group at C-11. The configuration of H-11 in 1 was determined to be axial on the basis of coupling constants with H-9 and H-12 and further supported by NOESY correlation between H-11 and CH 3 -25which is in close agreement with the reported compounds 2-4 (Fig 2) [51]. One known analogue 15 has also been isolated from the same species having 13 C-NMR at δ 70.6, and 1 H-NMR at δ 4.24 (dd, J = 9.6, 3.0 Hz) corresponded to the C-11 position. However, the downfield shift observed in the 13 C-NMR of compound 1 at δc 81.8 further strengthens the β-orientation of hydroxyl group at C-11 position [54].
Similarly, the H-3 showed an HMBC cross-peak with a peak at δc 177.7 which further supported the presence of carboxylic group in ring A at either C-23 or C-24 positions. The characteristic NOESY correlation between CH 3 -23 and H-5 allowed the assignment of the functional group at C-24 as previously observed boswellic acids [55]. All the above data can be accommodated only on an urs-l2-ene triterpenoid structure for compound 1 with a hydroxyl group at the C-3β position [56] and another secondary hydroxyl group attached to the C-l lβ position [54]. Thus, on the basis of spectroscopic data and also comparison with literature data, structure of compound 1 was deduced as 3β, 11β-dihydroxy-12-en-24-oic acid commonly known as 3β, 11β-dihydroxy-β-boswellic acid.
It is noteworthy mentioning that the sole source of boswellic acids is Boswellia species where the oxidation takes place at C-24 by Cytochrome P450 enzymes [57] whilst ursolic acid exists abundantly in plant kingdom and the oxidation takes place at C-28. Thus, the boswellic acids can be considered as genus-specific and can be specific biomarkers for Boswellia species albeit the variation of the amounts among different Boswellia species and grades (Fig 2).

Isolation and characterization of boswellic aldehyde
Compound 5 was obtained as a colorless solid having melting point of 184-186˚C and assigned a molecular formula of C 30 H 48 O 2 as evidenced by HRESI-MS which showed an [M + H]+ ion peak at m/z 441.3722 (calculated: 441.3718). The ESI-MS molecular ion peak suggested seven double bond equivalents, six of which are assigned to the rings of the pentacyclic triterpene skeleton and one was attributable to the formyl group (δ C 204.6). The IR spectrum of 5 showed characteristic peaks at 3440, 2940, 2870 and 1730 cm -1 , which confirmed the presence of OH and CHO groups. An equivalent, assigned to a double bond consistent with Δ 12ursane skeleton, showed resonances at δc 124.3 (C-12) and 139.6 (C-13).
In the 1 H NMR spectrum (experimental) of compound 5, the presence of vinylic proton at 5.12 (1H, t, J = 3.0 Hz, H-12), one oxymithine proton at 4.11 (1H, br s, H-3), five tertiary methyls at 1.10, 1.08, 1.02, 0.82, and 0.78 (all singlets), and two secondary methyls at 0.90, (3H, d, J = 6.0 Hz) and 0.77 (3H, d, J = 6.0 Hz) were observed, unambiguously confirming the presence of Δ 12 -ursane skeleton [51]. The 13 C NMR spectrum (experimental part) exhibited a total of 30 carbon signals which were sorted out into nine CH 2 , seven CH 3 , seven CH including one olefinic, and one oxygen-bearing carbon as well as seven quaternary carbons. Comparison of the 1 H and 13 C NMR spectra of compound 5 with those of β-boswellic acids [55] displayed close similarities with notable difference being the substitution mode in ring A. The chemical shift at δc 69.2 in the 13 C-NMR spectra and HMBC correlations of H-3 with CH 3 -23 (δc 19.7) and formylcarbon C-24 (δc 204.7) confirmed the presence of hydroxyl group in ring A at C-3 position, while the alpha orientation was further confirmed by NEOSY cross peaks (Fig 2) [58].
The presence of formyl group was assumed to be located either at the C-23, C-24 or at the C-28 position. The key HMBC correlations of the formyl proton H-24 (δ H 9.71) with C-3 (69.2) and C-4 (52.2); CH 3 -23 (19.7) with C-3 (69.2) and C-24 (204.7) proved that aldehyde group located either at C-23 or C-24 positions. Since the formyl group was assumed to be formed in the reaction by oxidation of the CH 2 OH group of 14 ([59] ; Fig 3), it might be expected that the -CHO group will have the same location and orientation as the CH 2 OH group has in 9. Therefore, the formyl group in compound 5 located the position at C-24 (β-orientation) which was further confirmed by NEOSY correlations [60]. On the basis of the above results, compound 5 was deduced as3α-hydroxy-urs-12-ene-24-al commonly named as boswellic acid aldehyde. This compound, up to the best knowledge, is described for the first time from B. sacra and recently isolated from B. serrata as a first natural product [61].

Biosynthetic pathway of 3-epi-α-amyrin and its precursors
The biosynthesis of β-amyrin has been postulated [62] following the first proposal of the isoprene rule [63,64]. In order to understand the biosynthesis of BA and its consecutive oxidative and acetylated derivatives, the biosynthesis of their precursor β-amyrin is depicted in Fig 4. This historic precursor is commenced at the chair-chair-chair-boat conformation of (3S)-2,3-oxidosqualene A, which cyclized to the corresponding 6.6.6.5-fused tetracyclic dammarenyl C-20 cation B which at this stage led to the formation of the dammaranes isolated from Boswellia species [62]. A concerted hydride shifts from C-13 to C-17, methyl group (C-18) shift from C-14 to C-13 and methyl group (C-30) shift from C-8 to C-14, followed by enzymatically catalyzed base abstraction of C-9 proton, generating the double bond between C-8 and C-9, afforded tirucallanes isolated from Boswellia species [65]. After ring D enlargement, electrophilic addition on the terminal double bond of the tetracyclic baccharenyl cation C, a fivemembered ring E is constructed and the Lupanyl cation D is generated which leads to the formation of Lupanes isolated from Boswellia species [66]. This is followed by ring E enlargement which occurs via C-30 shift from C-20 to C-19 to afford the oleanyl cation E which can lead to the formation of Oleananes isolated from Boswellia species [67] to which β-amyrin (precursor for α-Boswellic acid) belongs. The oleanyl cation F then undergoes sequential 1,2-hydride shifts from C-18 to C-19 and from C-13 to C-18 followed by elimination of H-12 to afford the 6.6.6.6.6-fused pentacyclic Ursanes isolated from Boswellia species [68] to which α-amyrin (precursor for β-Boswellic acid) belongs.

Chemical diversity of boswellic acids
The four amyrins are known from different plant species including Boswellia [69], some of their corresponding boswellic acids remained unknown due to their existence in infinitesimal amounts in the resins which doesn't allow their isolation and hence characterization. It is imperative to highlight a few conventions adopted in the literature which can be confusing. In general, α-amyrins where C-29 and C-30 are in vicinal relationship at C-19 and C-20 respectively will lead to β-boswellic acids. Moreover, β-amyrins where C-29 and C-30 are in germinal relationship both at C-20 will lead to α-boswellic acids.
C-H activation and oxidation of the α-configured methyl group C-24 of the 3-epi-α-amyrin 6, the 3-epi-β-amyrin 8 (Fig 5) afforded the β-boswellic acid 5 and α-boswellic acid 7 respectively which were isolated from different Boswellia species. [70,71] Likewise C-H activation and oxidation of the α-configured methyl group C-24 of the α-amyrin 2 furnished the 3epi-β-boswellic acid 6 (where the hydroxyl group at C-3 and the methyl group at C-4 are both β-configured and the two methyl groups at C-19 and C-20 are in vicinal relationship) which was only reported once from B. carteri [59].
Similarly, C-H activation and oxidation of the α-configured methyl group C-24 of the βamyrin 9 should afford 3-epi-α-boswellic acid 8. However, as far as we are aware there is no report which describes the isolation of the 3-epi-α-boswellic acid. We found it vitally important to introduce the prefix "epi" to differentiate the configuration of the hydroxyl group at C-3 which has been a long-standing confusion in literature for decades. In this regard, 3β-amyrin where the β-OH at C-3 is above the plane will lead to 3-epi-boswellic acid again where the β-OH at C3 is above the plane. In contrast, 3-epi-amyrin where the α-OH at C-3 is under the plane will lead to boswellic acid where the α-OH at C-3 is under the plane (Fig 6).

Biosynthesis of boswellic acids
A series of enzymatic oxidations take place to convert C-24 into carboxylic acid at 3-epi-αamyrin 6 which undergoes a conversion to the corresponding alcohol 14 that is isolated from B. carteri [59]. Alcohol 14 is oxidized to the aldehyde 5 that is isolated as a new compound from Boswellia sacra with full spectrometric data [61]. The aldehydic group in 5 upon oxidation by Cytochrome P450 enzyme affords the carboxylic acid moiety of β-boswellic acid 10. The OH group is introduced at C-11 of the β-boswellic acid 10 which led to the formation of diol 15 which has also been isolated from Boswellia sacra, as a new source, by our group and its ethoxy analogue was reported earlier from B. neglecta [51] which upon oxidation of the OH at C11 should lead to ß-KBA 16. Acetylation of the OH at C-3 in β-boswellic acid 5 via Ac-CoA and an acetyl-transferase will give β-ABA 17. Oxidation of the OH at C-11 and acetylation of the OH at C-3 of diol 15 afforded the AKBA derivative 18 which appeared to be the final product of this enzymatically-driven cascade reaction. The reaction sequence of the acetylation and oxidation which led to formation of ABA and KBA is not known hitherto and lacks supported evidences. Certainly, comprehensive physiological investigations on the frankincense trees are mandatory to elucidate the exact order of steps for the biosynthesis of AKBA.
The boswellic acid derivative 21 (Fig 7) have resulted from the oxidation of alcohol 20. Isomer 22 was isolated from B. sacra earlier [72]. In a similar analogy to alcohol 14, alcohol 20 is a Chemical, molecular and structural studies of Boswellia result of the oxidation of the amyrin 19. All these four derivatives are isolated by our group from B. sacra [72].

Quantification of α-amyrin and BAs in different Boswellia tree parts and resins
Given the fact that there are no specific enzymes isolated from Boswellia species to elucidate the biosynthetic pathway, we have attempted quantification of boswellic acid derivatives using HPLC. This quantification is performed on the leaf, bark, roots and resin of B. sacra tree. Moreover, three resins are analyzed for their amyrins and BA contents namely B. sacra, B. serrata and B. papyrifera. In order to verify the influence of the geographical location of the Boswellia sacra tree, the resins of two grades (mountainous and costal) were analyzed for their  Chemical, molecular and structural studies of Boswellia boswellic acids composition. We realized that the quantities of the triterpenes vary significantly depending on the Boswellia species as well as on the environmental factors within the same species (Table 1). While in some species large amounts of a triterpenes exist, only infinitesimal quantities of the same compound can be detected which explains the huge chemical diversity of the triterpenes in Boswellia species (Table 1). B. serrata is distinguished from the other two species by higher amounts of β-KBA (1.29%) whereas B. papyrifera and B. sacra possess 0.46% and 0.73% respectively. This suggests β-KBA as a biomarker for B. serrata. B. papyrifera proved a complete absence of β-BA and lower amounts in case of B. serrata whilst higher amounts on average of the two grades in B. sacra. Therefore, β-BA can be identified as a biomarker for B. sacra. Comparable quantities of α-BA in all three species are found which makes it difficult to use α-BA as a biomarker to distinguish these species. The same conclusion can be drawn for α-ABA and β-ABA. The amounts of β-AKBA are comparable in B. serrata and B. papyrifera as well as in the costal grade of the B. sacra whereas higher amounts are observed in the mountainous type of B. sacra. The precursor of all β-boswellic acids (β-BA, β-ABA, β-KBA and β-AKBA) presented in Table 1 above is epiα-amyrin which was detected in higher amounts in B. sacra species whilst completely absent in B. serrata and B. papyrifera. The total β-Boswellic acids content is 7.04 and 9.68% in B. papyrifera and B. serrata respectively ( Table 2). Significantly higher amounts of β-boswellic acids are observed for the two B. sacra species. It is noteworthy mentioning that the amounts of α-ABA are lower in the case of B. sacra when compared with the other two species.
The HPLC results reported herein are in total agreement with our recently published quantification of KBA (keto-β-boswellic acid) and AKBA (3-acetyl-11-keto-β-boswellic acid) using NIRS coupled with PLS regression [73,40]. For the first time, NIR spectroscopy coupled with PLS regression as a rapid and alternative method was developed to quantify the amount of AKBA and KBA in different plant parts of Boswellia sacra and the resin exudates of the trunk. The quantification of various plant parts and the resin indicated that the MeOH extract of the resin has the highest concentration of AKBA (7.0%) followed by epidermis (1.37%), and essential oil (0.1%). Thus it can be concluded that AKBA is only present in the gum-resin exudate and epidermis of the stem. The existence of moderate amounts of AKBA in the epidermis is due to the presence of resin-producing canals [40]. A similar study was conducted for KBA and obtain results indicated that the MeOH extract of resin has the highest concentration of KBA (0.6%) followed by essential oil (0.1%). The MeOH extract of the resin was subjected to column chromatography to get various sub-fractions at different polarity of organic solvents. The sub-fraction at 4% MeOH/CHCl 3 (4.1% of KBA) was found to contain the highest percentage of KBA followed by another sub-fraction at 2% MeOH/CHCl 3 (2.2% of KBA; Table 2). The results also indicated that KBA is only present in the gum-resin of the trunk and not in the plant itself [73]. Interestingly, there are clear differences with regard to the quantities of amyrins and boswellic acids of the two grades of B. sacra which explain the influence of the geographical location and the environmental conditions of the two grades. This is in agreement with our recent genetic diversity work, which was based on the analysis of simple sequences repeat, and random-amplified polymorphic DNA genetic markers, suggesting a clear isolation of B. sacra populations in the eastern coastal regions (unpublished data). Thus, the current variation in the results can also be attributed to the tree's genetic diversity and climatic conditions. This is also in correlation with the results published [74][75][76] where it was shown that both genetic and climatic factors can influence the resin production and composition. To further confirm expression of amyrin related transcripts, primers were designed using known available genomic sets derived from Azadirachta indica. Since biochemical related genomic and molecular information are not available for any of Boswellia species, therefore, the homologues species at order level (Azadirachta indica) was selected [77]. The gene expression for both α-amyrin synthase was not detected in the leaf part of the B. sacra tree (Fig 8). However, Squalene-synthase related transcript was expressed in leaf part which suggests that BAs might be synthesized through this precursor. However, this needs a detailed transcriptomic analysis to obtain conclusive evidences on the biosynthetic pathway.
Interestingly, the roots contain α-amyrin (2.63%) and proved a complete absence of βamyrin and all boswellic acids. The leaves have only trace amounts of β-ABA and β-AKBA. Not unexpectedly, the stem proved to be completely free of amyrins and boswellic acids. This explains that the production of the resin is taking place at the cambium (total boswellic acids 1.0%) and epidermis (total of boswellic acids 4.93%) due to the existence of the resin-producing canals. This is supported by the scanning electron microscope images (Fig 9). The micrographs clearly demarcate the presence of radial and axil resin canals in the inner bark region (between epidermis and vascular cambium). The current results show an anatomical harmony with the presence of resinous ducts in the B. papyrifera [78].

Conclusion
The current study showed the isolation and characterization of new chemical constituents (βboswellic aldehyde and 3β, 11β-dihydroxy BA) from the B. sacra resins along with known αamyrin (3-epi-α-amyrin, β-amyrin and α-amyrin). A detailed elucidation of the BAs biosynthesis was made for the first time, providing chemical, molecular and structural details to reinforce the previously reported studies. A detailed quantification of BAs and their amyrin precursors in different parts of three Boswellia species (B. sacra, B. serrata and B. papyrifera) suggests variation of triterpenes on the basis of species, tree parts and geographical locations. However, BAs were more consistently available in the inner bark of the tree. This was also revealed by the micrographs of resin canals in the inner bark. The current study concludes that BAs are the major class of chemical constituents in Boswellia, where future studies should consider the characterization of enzymes responsible in its biosynthesis.