Triterpene Acids from Frankincense and Semi-Synthetic Derivatives That Inhibit 5-Lipoxygenase and Cathepsin G

Age-related diseases, such as osteoarthritis, Alzheimer’s disease, diabetes, and cardiovascular disease, are often associated with chronic unresolved inflammation. Neutrophils play central roles in this process by releasing tissue-degenerative proteases, such as cathepsin G, as well as pro-inflammatory leukotrienes produced by the 5-lipoxygenase (5-LO) pathway. Boswellic acids (BAs) are pentacyclic triterpene acids contained in the gum resin of the anti-inflammatory remedy frankincense that target cathepsin G and 5-LO in neutrophils, and might thus represent suitable leads for intervention with age-associated diseases that have a chronic inflammatory component. Here, we investigated whether, in addition to BAs, other triterpene acids from frankincense interfere with 5-LO and cathepsin G. We provide a comprehensive analysis of 17 natural tetra- or pentacyclic triterpene acids for suppression of 5-LO product synthesis in human neutrophils. These triterpene acids were also investigated for their direct interference with 5-LO and cathepsin G in cell-free assays. Furthermore, our studies were expanded to 10 semi-synthetic BA derivatives. Our data reveal that besides BAs, several tetra- and pentacyclic triterpene acids are effective or even superior inhibitors of 5-LO product formation in human neutrophils, and in parallel, inhibit cathepsin G. Their beneficial target profile may qualify triterpene acids as anti-inflammatory natural products and pharmacological leads for intervention with diseases related to aging.

. Effects of triterpene acids on 5-lipoxygenase (5-LO) and cathepsin G. different Boswellia species by preparative high-performance liquid chromatography (HPLC), as described previously [9,22]. The R-and S-configured ally-alcohols 18 and 19 were synthesized from KBA and AKBA using a combination of LiBr and NaBH4 in diglyme at reflux temperature, as described [23,24]. The ABA and AKBA derivatives with modified C3 position were prepared as follows: β-BA 3 or KBA 2 were treated with oxalyl chloride to form the half-ester oxaloyl-BA 20 and oxaloyl-KBA 21; β-BA 3 or KBA 2 were treated with succinic or glutaric anhydride to form the half-ester succinoyl-BA 22 and succinoyl-KBA 23 as well as glutaroyl-BA 24 and glutaroyl-KBA 25; the carboxymethyl-BA 26 and carboxymethyl-KBA 27 were synthesized from β-BA 3 or KBA 2 using 2-chloro acetic acid and NaH in THF over night as described earlier [23,24]. All compounds were analyzed by 1 H-and 13 C-NMR, as well as by mass spectrometry.
Several studies reported that frankincense extracts, as well as AKBA 1 and KBA 2, are superior inhibitors of 5-LO in intact cells as compared to cell-based assays, for multiple possible reasons that remain unclear [6,12,13,26]. Thus, we studied the triterpene acids also against 5-LO activity in a well-established cell-based assay using human neutrophils stimulated with A23187 [25]. The reference 5-LO inhibitor BWA4C was about 4-fold less active in neutrophils (IC50 = 0.16 µM) as compared to cell-free assay conditions. Both AKBA 1 and KBA 2 were somewhat more potent in intact cells (IC50 = 3.0 and 3.5 µM) versus isolated 5-LO, and also for β-BA 3, but not for ABA 4, where slightly more efficient 5-LO inhibition was observed (47% inhibition at 10 µM, Table 1). Again, roburic acids 5 and 6 as well as DHNA 8 were hardly active, but 7 caused potent suppression of 5-LO with IC50 = 4.3 µM, being almost equally effective as AKBA 1 and KBA 2. Along these lines, also, the lupeolic acids 9, 11, and 12 (but not 10) Several studies reported that frankincense extracts, as well as AKBA 1 and KBA 2, are superior inhibitors of 5-LO in intact cells as compared to cell-based assays, for multiple possible reasons that remain unclear [6,12,13,26]. Thus, we studied the triterpene acids also against 5-LO activity in a well-established cell-based assay using human neutrophils stimulated with A23187 [25]. The reference 5-LO inhibitor BWA4C was about 4-fold less active in neutrophils (IC 50 = 0.16 µM) as compared to cell-free assay conditions. Both AKBA 1 and KBA 2 were somewhat more potent in intact cells (IC 50 = 3.0 and 3.5 µM) versus isolated 5-LO, and also for β-BA 3, but not for ABA 4, where slightly more efficient 5-LO inhibition was observed (47% inhibition at 10 µM, Table 1). Again, roburic acids 5 and 6 as well as DHNA 8 were hardly active, but 7 caused potent suppression of 5-LO with IC 50 = 4.3 µM, being almost equally effective as AKBA 1 and KBA 2. Along these lines, also, the lupeolic acids 9, 11, and 12 (but not 10) suppressed 5-LO product formation, with IC 50 = 4.0, 4.6, and 5.1 µM. Although tirucallic acids 13-17 failed to inhibit isolated 5-LO, 13-16 were quite effective in intact cells, with 14 being about 3-fold more potent (IC 50 = 1.1 µM) than AKBA 1, and with 13 and 15 displaying similar potencies (IC 50 = 2.9 and 3 µM). In conclusion, among the 17 triterpene acids, the tirucallic acid 14 is the most potent inhibitor of 5-LO product biosynthesis in neutrophils, thus outperforming AKBA, despite lack of pronounced effectiveness against isolated 5-LO.

Inhibition of Cathepsin G by Natural Occurring Triterpene Acids
Since cathepsin G is a pharmacological relevant target for BAs and frankincense with high affinities to BAs [17], we analyzed also the effects of the 17 triterpene acids for inhibition of cathepsin G activity in a cell-free assay. Since cathepsin G is a secreted protease that acts outside the cell upon degranulation, cell-based assays are not feasible for cathepsin G inhibitor studies as compared to 5-LO. JNJ-1031795 was used as reference drug [27] that blocked cathepsin G activity with IC 50 = 0.06 µM. The rank order for cathepsin G inhibition by the BAs was β-BA 3 > AKBA 1 > ABA 4 > KBA 2, with IC 50 = 0.5, 0.6, 2.0, and 4.1 µM, respectively. Among all other triterpene acids tested at 10 µM, only the lupeolic acid 10 exhibited >50% inhibition of cathepsin G with an IC 50 = 7.5 µM. Nevertheless, marked suppression of cathepsin G activity (i.e., 27-44% inhibition at 10 µM) was found for all roburic acids (5-7), DHNA 8, lupeolic acids 9 and 11, and tirucallic acid 13, 14, 16, and 17. Only 12 and 15 failed to markedly inhibit cathepsin G. Together, cathepsin G is a clearly preferred target for BAs (particularly β-BA 3 and AKBA 1), with much less susceptibility for all other triterpene acids tested.

Effects of Semi-Synthetic BAs against 5-LO and Cathepsin G
Next, we investigated a set of semi-synthetic BA derivatives for inhibition of 5-LO and cathepsin G to further explore the structure-activity relationships (SARs) of BAs. First, the 11-keto moiety of KBA was reduced to hydroxy, yielding the Rand S-configured ally-alcohols 18 and 19. Of interest, 18 was somewhat more efficient to inhibit 5-LO in neutrophils (IC 50 = 2.8 µM) versus parental KBA 2 while 19 was less active, and both compounds exhibited only weak effects against 5-LO in cell-free assays (<25% inhibition at 10 µM). Also, cathepsin G activity was less pronouncedly affected by 18 and 19. Then, we replaced the 3-O-acetyl moiety of AKBA 1 or ABA 4 by acidic moieties, such as oxaloyl, succinoyl, glutaroyl, and carboxymethyl groups. All these compounds were not or only weak 5-LO inhibitors, with <25% inhibition at 10 µM in cell-free assays. Similarly, in intact cells, 5-LO product formation was not or barely affected (IC 50 > 10 µM); the most potent compound was the carboxymethyl-BA 26 that inhibited 5-LO activity by 42% at 10 µM. By contrast, pronounced cathepsin G inhibition was observed for BA derivatives lacking the 11-keto moiety, such as oxaloyl-BA 20, glutaroyl-BA 24, and carboxymethyl-BA 26, with IC 50 = 4.7, 3.6, and 3.4 µM. Of interest, even though AKBA 1 (an 11-keto derivative carrying a 3-acetoyl moiety) was quite potent (IC 50 = 0.6 µM), the 11-keto derivatives oxaloyl-KBA 21, succinoyl-KBA 23, and carboxymethyl-KBA 27 were not active, and glutaroyl-KBA 25 (IC 50 = 8.6) was less efficient than the respective counterpart, 24, lacking the 11-keto moiety. Taken together, modification of the acetoyl group of AKBA 1 or ABA 4 towards acidic residues is strongly detrimental for inhibition of 5-LO, and to a varying extent, also for interference with cathepsin G, depending on the presence of the 11-keto moiety.

Discussion
Here, we report the SAR of various natural occurring triterpene acids from frankincense, and of novel semi-synthetic BA derivatives for inhibition of the two molecular targets, 5-LO and cathepsin G. For 5-LO, the SARs depend on the assay type: for direct interaction with 5-LO in the cell-free assay, AKBA 1 and KBA 2 were most potent, and neither natural nor semi-synthetic triterpene acids were comparably active. Modification of the acetoyl group of AKBA 1 towards acidic residues is clearly detrimental for interference with 5-LO. For inhibition of 5-LO product formation in neutrophils, however, several of the natural triterpene acids were efficient and outperformed AKBA 1, such as the tirucallic acids 13 and 14. Since tirucallic acids were shown to interfere with various signaling pathways [28,29], among others, Ca 2+ mobilization and mitogen-activated protein kinase kinase (MEK)-1/2, that could also activate 5-LO in intact cells, one may speculate that tirucallic acids and possibly also roburic and lupeolic acids interfere with the activation of 5-LO, and thus, with 5-LO product biosynthesis. In fact, lupeolic acids (e.g., 12) were shown to suppress the biosynthesis of various eicosanoids by interference with cytosolic phospholipase A2 [8] that provides substrate to 5-LO.
In analogy to interference with 5-LO, inhibition of cathepsin G activity was also most efficient for BAs, particularly for β-BA 3 and AKBA 1, with submicromolar IC 50 values. Among the natural triterpene acids, only the lupeolic acid 10 reached an IC 50 below 10 µM, even though several other triterpene acids significantly inhibited cathepsin G at 10 µM, albeit less than 50%. Among the semisynthetic BAs lacking the 11-keto moiety, oxaloyl-BA 20, glutaroyl-BA 24, and carboxymethyl-BA 26 were efficient cathepsin G inhibitors, indicating that the 11-keto moiety is dispensable for cathepsin G interference.
In summary, among the 17 investigated natural occurring triterpene acids from frankincense, the BAs provide the most potent inhibitors of 5-LO (i.e., AKBA 1) and cathepsin G (i.e., β-BA 3). Nevertheless, some lupeolic acids are efficient as well, and for inhibition of 5-LO product formation in intact cells, diverse representatives from all types of triterpene acids are active, even with superior potency (e.g., tirucallic acid 14) than BAs. Notably, our previous SAR study on the same 17 triterpene acids for inhibition of mPGES-1 revealed opposite findings: thus, all six tirucallic acids were more potent than the BAs, and also, the lupeolic acid 12 and the roburic acid 7 were clearly superior [9]. Together, dual interference with cathepsin G and 5-LO product formation in human neutrophils, on top of mPGES-1 inhibition, represents a beneficial pharmacological profile, and may qualify triterpene acids as anti-inflammatory natural products and pharmacological leads for intervention with inflammatory diseases.

Chemistry
General information. All reactions were run in dried solvents under N 2 in flame dried glassware. THF and diethyl ether were dried with Na/benzophenone under N 2 . Flash chromatography was done with silica gel (Merck, Darmstadt, Germany) with particle size 40-63 µm and with distilled solvents. Reagents were bought from chemical suppliers and used without further purification.
NMR spectra ( 1 H, 13 Kather, 2007 [24]. 11α-Hydroxy-β-boswellic acid 18 and 11β-hydroxy-β-boswellic acid 19. NaBH 4 (121 mg, 3.2 mmol) was dissolved in dry diglyme (3.5 mL). LiBr (278 mg, 3.2 mmol) was added with stirring at rt. After 30 min, β-KBA 2 (300 mg, 0.64 mmol) was added and the mixture was refluxed for 1 h. After cooling to rt, the mixture was diluted with cold water (approx. 5 mL) and acidified carefully to pH 4 with 1 N HCl. The mixture was extracted three times with diethyl ether (ca. 10 mL each). The combined ethereal extracts were washed with water and brine, dried with MgSO 4 , and filtered, and the solvent was evaporated in vacuo at rt. The crude product was purified by flash chromatography (pentane/diethyl ether 1:1 + 1% acetic acid) to yield 83 mg of 18 and 141 mg of 19 as white solids.  3-O-Succinoyl-β-boswellic acid 22. β-BA 3 was dissolved in dry pyridine (2.2 mL). Succinic anhydride (220 mg, 2.2 mmol) was added, followed by 4-pyrrolidino pyridine (36.2 mg, 0.22 mmol). The reaction mixture was refluxed for 7 h. After cooling to rt, the dark brown mixture was diluted with diethyl ether (ca. 20 mL) and was washed three times with 1 N HCl (ca. 10 mL each). The organic phase was washed with H 2 O and brine, and dried with MgSO 4 . After filtration, the solvent was removed in vacuo to yield an orange-brown solid which was purified by flash chromatography (pentane/diethyl ether 2:1 + 1% HOAc) to give 87 mg of 22 as a white solid.

Determination of 5-LO Product Formation in Neutrophils
5-LO product formation in intact human neutrophils was performed using 5 × 10 6 freshly isolated cells that were resuspended in 1 mL PGC buffer. Cells were pre-incubated with the test compounds or vehicle (0.1% DMSO) at 37 • C for 15 min prior to stimulation with 2.5 µM Ca 2+ -ionophore A23187 for 10 min (37 • C). 5-LO product formation was stopped by addition of one volume of ice-cold methanol, samples were subjected to solid phase extraction after addition of 200 ng PGB 1 as internal standard, and 5-LO products (LTB 4 , its trans-isomers and 5-HETE) were analyzed by RP-HPLC as described above.

Determination of Cathepsin G Activity in a Cell-Free Assay
Commercially available purified cathepsin G (0.2 µg) from human neutrophils (Calbiochem, La Jolla, CA, USA) was mixed with test compounds or DMSO (vehicle control) in 200 µL 0.1 M HEPES pH 7.4, 0.5 M NaCl, and 10% DMSO in a 96-well plate (Greiner Bio-One, Frickenhausen, Germany) and pre-incubated for 20 min at 25 • C. Then, N-Suc-Ala-Ala-Pro-Phe-pNA (Suc-AAPF-pNA) (1 mM) as substrate for cathespin G was added, and the absorbance was measured at 410 nm at 25 • C using a Victor 2 plate reader (PerkinElmer, Waltham, MA, USA). The enzymatic activity was determined by the progress curve method, and cathepsin G activity is given as percentage of the vehicle (DMSO) control.

Statistics
Results are presented as mean ± standard error of the mean (SEM) out of n independent experiments, where n represents the number of experiments performed on different days or with different donors. IC 50 values were calculated from at least 5 different concentrations using a nonlinear regression interpolation of semi-logarithmic graphs in GraphPad Prism (Graphpad Software Inc., San Diego, CA, USA). Statistical evaluation was performed by two-tailed student t-test for single comparisons. P-values < 0.05 were considered as significant.