Anti-Inflammatory Activity of Boswellia serrata Extracts: An In Vitro Study on Porcine Aortic Endothelial Cells

This study is aimed at investigating the cytotoxicity, anti-inflammatory, and angiogenic activities of two Boswellia serrata extracts on primary culture of porcine aortic endothelial cells (pAECs). Chemical characterization of a dry extract (extract A) and a hydroenzymatic extract (extract G) of B. serrata was performed by HPLC using pure boswellic acids (BAs) as standard. In cultured pAECs, extract G improved cell viability, following LPS challenge, in a dose-dependent manner and did not show any toxic effect. On the other hand, extract A was toxic at higher doses and restored pAEC viability after LPS challenge only at lower doses. Pure BAs, used at the same concentrations as those determined in the phytoextracts, did not contrast LPS-induced cytotoxicity. Extract A showed proangiogenic properties at the lowest dose, and the same result was observed using pure AKBA at the corresponding concentration, whereas extract G did not show any effect on the migration capacity of endothelial cells. In conclusion, an anti-inflammatory activity of B. serrata extracts on endothelial cells was reported, though cytotoxicity or proliferative stimulation can occur instead of a protective effect, depending on the dose and the formulation.


Introduction
The endothelium, uniquely positioned at the interface between the vascular wall and the blood, regulates multiple functions such as maintenance of normal vascular tone, modulation of coagulation, and immune responses [1]. It is widely demonstrated that the exposure of endothelial cells to proinflammatory stressors results in the production of molecules correlated with a proadhesive, prothrombotic, and proinflammatory phenotype that contributes to vascular disorders [2,3], including cardiovascular diseases (CVDs).
Since ancient times, the extracts from the oleo-gum resin of Boswellia serrata Roxb. ex Colebr. (family Burseraceae), also identified as Indian frankincense or Salai Guggal, have been used in traditional Ayurvedic medicine for the treatment of inflammatory diseases, including osteoarthritis and chronic bowel diseases [4][5][6][7][8].
Differences in the relative amount of BAs and other components of the phytocomplex are related to the existence of different species of the genus Boswellia, to environmental conditions (e.g., soil composition, season, and air humidity), and to the extraction procedure [13] leading to herbal products of different composition and quality. In a previous study, seven B. serrata extracts were compared for their AKBA content and antioxidant power, highlighting wide variations [14]. In particular, one of the extracts obtained by bioliquefaction based on enzyme biocatalysis (hydroenzymatic extract) [15] showed interesting peculiarities. A lower content of AKBA and antioxidant power but higher activity in ex vivo tests on peripheral blood mononuclear cells (PBMCs) was determined in comparison with the dry extract [14]. In recent years, attention has also been focused on the role of other BAs, namely, KBA and βBA [16,17], suggesting a possible pharmacological activity also for these BAs. Preliminary data showed wide variability in the concentration of BAs in different extracts [18]; therefore, the present research is aimed at deepening the chemical characterization of the two extracts previously studied, focusing on HPLC quantification of KBA and βBA. The effect of different formulations will be evaluated in comparison with the individual pure BAs in an interesting in vitro model: primary culture of porcine aortic endothelial cells (pAECs). With pig as an excellent model for translational medicine in the cardiovascular field [19,20], we have previously isolated and cultured endothelial cells from thoracic aortas [21]. These primary cultures maintain a stable phenotype, and they prove to be an excellent model of study for the vascular response to different stressors [22,23]. Therefore, pAECs were chosen as an ideal in vitro model to study the anti-inflammatory and angiogenic properties of the two B. serrata extracts in comparison with pure AKBA, KBA, and βBA, either individually or mixed together.

Qualitative and Quantitative
Characterization of B. serrata Extracts. Qualitative and quantitative analyses of B. serrata extracts were performed by a reversed-phase highperformance liquid chromatography (HPLC) method using the HPLC system (Beckman Coulter, Brea, CA, USA), comprising a 116 pump, a 507 automatic autosampler, a UV-Diode Array 168 detector, and integration software 32 Karat as reported by Beghelli et al. [14]. Seven samples (A-G) were analyzed for KBA and βBA concentrations and were prepared by dissolving extracts in methanol. KBA and βBA standard stock solutions were prepared by dissolving 5 mg of analytical standard in methanol (5 mL). The calibration curves were obtained by analyzing six serial dilutions (50 ppm, 25 ppm, 10 ppm, 5 ppm, 2.5 ppm, and 1 ppm) of the stock solution and by plotting the peak area measured at 260 nm against KBA concentrations and at 210 nm against βBA concentrations. The following equations of the curves were obtained: KBA = 77361x + 44918, r 2 = 0 999, βBA = 26532x + 721 54, r 2 = 0 999 1 The KBA and βBA peaks in the samples were identified on the basis of the retention time on the chromatogram at 260 nm and 210 nm, respectively. All measurements were performed in triplicate and data were reported as mean ± SD.
2.3. Cell Culture and Treatment. Porcine aortic endothelial cells (pAECs) were isolated and maintained as previously described by Bernardini et al. [21]. All experiments were performed with cells from the third to the eighth passage. The first seeding after thawing was always performed in T-25 tissue culture flasks (3 × 10 5 cells/flask) (T-25, BD Falcon, Franklin Lakes, NJ, USA), and successive experiments were conducted in 24-well plates (scratch test) or 96-well plates (cell viability) with confluent cultures. Cells were cultured in human endothelial SFM medium, added with FBS (5%) and antimicrobial/antimycotic solution (1x) in a 5% CO 2 atmosphere at 38.5°C. Extract A was dissolved in DMSO at 10 mg dry extract/ml (stock solution) and then diluted in culture medium to obtain four doses containing 0.1, 1, 10, and 100 μg of dry extract/ml, respectively. Extract G, which is an aqueous solution, was directly diluted in culture medium to obtain four doses referring to 2.4, 24, 240, and 2400 μg of resin/ml. These doses were chosen and normalized on the basis of AKBA concentration in extracts as reported in [14]: for both extracts, the lowest dose contained 3.8 ng/ml of AKBA and the highest dose contained 3.8 μg/ml of AKBA.
Pure analytical grade BAs (KBA, AKBA, and βBA) were dissolved in methanol (stock solution 1 mg/ml) and then in culture medium to obtain the required concentrations. Two doses were chosen: low, corresponding to 3.8 ng/ml AKBA, 3 ng/ml KBA, and 8 ng/ml βBA, and high, corresponding to 380 ng/ml AKBA, 300 ng/ml KBA, and 800 ng/ml βBA. For each treatment, the same concentration of the specific vehicle was used as control.
2.4. Effect of B. serrata Extracts on pAEC Viability. pAECs were seeded in a 96-well plate (6 × 10 3 cells/well) and exposed to four increasing doses of B. serrata extracts for 24 h. Cell viability was measured using tetrazolium salt (MTT assay). The formazan absorbance was measured at a wavelength of 570 nm, using Infinite® F50/Robotic absorbance microplate readers from TECAN (Life Sciences). The background absorbance of multiwell plates at 690 nm was also measured and subtracted from the 570 nm measurements.

Effect of B. serrata Extracts on LPS-Induced pAEC Death.
pAECs seeded in a 96-well plate (6 × 10 3 cells/cm 2 ) were exposed to lipopolysaccharide (LPS) (25 μg/ml) for 24 h either in the presence or in the absence of extracts A and G or pure BAs at the concentrations reported above. Cell viability was evaluated by MTT assay.
2.6. Effect of B. serrata Extracts on pAEC Migration Capacity. pAECs were seeded in a 24-well plate (4 × 10 4 cells/well). When cells reached confluence, a wound was induced scratching the surface by a pipette tip, then the detached cells were removed by washing with DPBS. Complete medium containing low and high doses of extract A (0.1 μg dry extract/ml and 10 μg dry extract/ml) and extract G (2.4 μg resin/ml and 240 μg resin/ml) and pure BAs at low (3 ng/ml KBA, 3.8 ng/ml AKBA, and 8 ng/ml βBA) and high (300 ng/ml KBA, 380 ng/ml AKBA, and 800 ng/ml βBA) concentrations were added. Microscopic phase-contrast pictures and three measurements of the damaged areas were taken immediately after the scratches (T0) and after 6 h (T1) and 24 h (T2). Images were acquired using a Nikon epifluorescence microscope equipped with digital camera (Nikon, Yokohama, Japan).

Statistical Analysis.
Each treatment was replicated three times (migration capacity) or eight times (cell viability and LPS challenge). Data were analyzed with a one-way analysis of variance (ANOVA) followed by the Tukey post hoc comparison test or Student's t-test. Differences of at least p < 0 05 were considered significant. Statistical analysis was carried out using R software (http://www.R-project.org).

KBA and βBA Quantification by HPLC-DAD Analysis.
Representative chromatograms of KBA, AKBA, and βBA analytical standards as well as extracts A and G analyzed at 210 and 260 nm are reported in Figure 1.
Both extracts presented two major peaks at 260 nm: the first one, at Rt of 13.2 min, identified as KBA by the use of the analytical standard, and the second one, at Rt of 26 min, previously identified as AKBA. Other components of the B. serrata phytocomplex were only visualized at 210 nm, and the peak at Rt of 49 min was identified as βBA by the use of the analytical standard. KBA, AKBA, and βBA concentrations, calculated based on the peak area and the calibration curve, are shown in Table 1.
Quantitative and qualitative differences were present. The concentrations of BAs in extract G were two orders of magnitude lower than in extract A, and the chromatogram of extract G was characterized by a major number of peaks resolved at 210 nm. Data on KBA and βBA concentrations in other additional five dry extracts (B-F) are reported in Table S1 in the Supplementary Material.

3.2.
Effect of B. serrata Extracts on pAEC Viability. Extract A was cytotoxic at higher concentrations, resulting in a reduction in cell viability of 12 and 47%, respectively, while lower concentrations did not affect cell viability (Figure 2(a)). Extract G did not show any toxic effect on pAECs (Figure 2(b)). In the presence of pure BAs, a significant (p < 0 05) cytotoxic effect was detected at the concentrations studied (Figure 2(c)). Only AKBA presented a dose-dependent effect.

Effect of B. serrata Extracts on LPS-Induced pAEC Death.
LPS challenge determined a significant 30% reduction of cell viability. Extract A significantly (p < 0 05) reduced the cytotoxicity induced by LPS at the lower concentrations (Figure 3(a)). The highest concentration elicited a significant exacerbation of LPS cytotoxicity resulting in 70% reduction of cell viability, while the lowest concentration showed a significant proliferative effect, resulting in a 40% increase in cell viability. Extract G significantly (p < 0 05) restored pAEC viability after LPS treatment at all the concentrations analyzed (Figure 3(b)), without a dose-dependent effect. None of pure BAs, individually or mixed together, was able to contrast LPS cytotoxicity (Figure 3(c)).

Effect of B. serrata Extracts on pAEC Migration Capacity.
Extract A reduced the damaged area at T1 (6 h) and restored completely the monolayer at T2 (24 h) at the lower concentration, while at 10 μg dry extract/ml no significant effect on cell proliferation was measured (Figure 4(a)). The incubation with extract G did not determine the recovery of the damage (Figure 4(b)). Pure BAs showed a significant wound-healing effect at the end of the incubation at the lower concentration (Figure 4(c)). In particular, AKBA at 3.8 ng/ml completely restored the monolayer.

Discussion
The gum resin obtained from B. serrata, used in Ayurvedic medicine for the treatment of a variety of diseases, is considered a promising natural source of anti-inflammatory molecules, in particular BAs [4,9].
The quantification of these active molecules is a prerequisite for testing any biological effect of a phytoextract from B. serrata. Therefore, the first aim of this study was to better characterize the BA profile through the quantification of KBA and βBA in addition to AKBA. The concentrations of BAs determined in extract A are in the range of those reported by other authors [24][25][26]. AKBA and KBA are used as markers to ensure the quality of B. serrata dry extracts, but their concentrations show wide variability in commercial products, which in general claim 65% of BAs. In general, BAs represent only a percentage of total organic acids, whose concentrations are determined by unspecific titration methods and, as a consequence, the claimed content of 65% BAs is absolutely unrealistic as recently pointed out also by other authors [24,25]. Very low percentages of KBA and βBA were found in extract G compared to extract A. This aqueous extract was also characterized by low AKBA and low polyphenol concentrations [14], confirming again the importance of the extraction procedure on the phytocomplex composition.
To evaluate the possible biological effects of these different formulations, extracts A and G, normalized on the basis of AKBA content, were used for in vitro analyses to assess cytotoxicity, anti-inflammatory activity, and angiogenic properties in comparison with pure BAs. Cytotoxic effects of B. serrata dry extracts and BAs were reported in several studies in different cancer cell lines, such as leukemia cells, prostate cancer cells, and gastrointestinal cancer cells [7,[27][28][29][30]. As regards the biochemical mechanism of cell death, Liu et al. [31] reported that BAs are able to induce apoptosis in Hep-G2 cells through the activation of caspase-8, while Bhushan et al. [32] found that a triterpendiol derived from BAs induced apoptosis in HL-60 cells through the activation of Bcl-2 and caspase-3.
The anti-inflammatory activity of Boswellia extracts was demonstrated in microvascular endothelial cells by preventing TNFα-induced expression and activity of MMP-3, MMP-10, and MMP-12 [33]. Moreover, previous studies have shown that B. serrata extracts and BAs antagonize the inflammatory effect of LPS in human and mouse macrophages, monocytes, and PBMCs [34][35][36]. Our results demonstrated for the first time the protective effect of B. serrata extracts against LPS inflammatory stimulus in endothelial cells. In particular, extract G was the most effective, restoring completely cell viability at all the doses studied without any cytotoxicity. On the contrary, increasing concentrations of extract A lead to opposite results ranging  Table 1: KBA, βBA, and AKBA quantification in Boswellia serrata extracts. Data are reported as mean ± SD (n = 3). Concentration is expressed in mg/g of dry extract (extract A) or mg/ml of hydroenzymatic extract (extract G). For each BA, significant differences between extracts are indicated by * (p < 0 05, Student's t-test) and by * * (p < 0 001, Student's t-test).  [14].  1, 1, 10, and 100 μg dry extract/ml) (a), extract G (2.4, 24, 240, and 2400 μg resin/ml) (b), and pure BAs (low, corresponding to 3.8 ng/ml AKBA, 3 ng/ml KBA, and 8 ng/ml βBA, and high, corresponding to 380 ng/ml AKBA, 300 ng/ml KBA, and 800 ng/ml βBA) (c) on pAECs. Cell viability was measured by MTT assay. Data are reported as mean ± SD of 8 independent replicates. Different letters above the bars indicate significant differences (p < 0 05 ANOVA post hoc Tukey's test).  1, 1, 10, and 100 μg dry extract/ml) (a), extract G (2.4, 24, 240, and 2400 μg resin/ml) (b), and pure BAs (low, corresponding to 3.8 ng/ml AKBA, 3 ng/ml KBA, and 8 ng/ml βBA, and high, corresponding to 380 ng/ml AKBA, 300 ng/ml KBA, and 800 ng/ml βBA) (c) on pAEC viability, in the presence of LPS (25 μg/ml), measured by MTT assay. Data are reported as mean ± SD of 8 independent replicates. Different letters above the bars indicate significant differences (p < 0 05 ANOVA post hoc Tukey's test).  from hyperproliferative effect (the lowest dose) to cytotoxic effect (the highest dose). Interestingly, in our model the use of pure KBA, βBA, and AKBA, either individually or mixed together, failed to protect endothelial cells from LPS toxicity and are only partially in accord with data reported by Henkel et al. [35]. In a cell-free assay, those authors suggested a direct molecular interaction between LPS and BAs lacking the keto moiety, in particular βBA, underlying the anti-inflammatory effect of Boswellia extracts.

Extract
Our results support the hypothesis that the antiinflammatory effect of Boswellia extracts is not strictly dependent on the presence of the most studied BAs, but it can be related to other bioactive molecules. Other triterpenes, as incensole, could be considered interesting candidates for the pharmacological properties of frankincense, accordingly to suggestions previously reported by other authors [9,37,38]. Beyond these bioactive terpenes, the gum resin does contain polysaccharides. These molecules are likely to be minor components in dry extract A, whereas they can be more concentrated in extract G, due to the different polarity of the extraction medium. A water-soluble fraction extracted from the gum resin of B. serrata containing galactose, arabinose, and D-glucuronic acid was suggested to act as a potent enhancer of humoral and cellmediated immune response [39], while the potential antiinflammatory activity of these polysaccharides has not yet been explored. We cannot exclude that the polysaccharide fraction present in extract G can develop additional modulatory effects on pAECs.
The migration ability of endothelial cells is critical in the physiological and pathological angiogenesis [40]. Our results obtained with an in vitro model of physiological angiogenesis showed proangiogenic activity of extract A at the lowest concentration, in agreement with a proliferative effect of the same dose recorded in LPS challenge. In addition, incubation with pure AKBA at the same concentration as that measured in extract A determined the same proangiogenic effect, indicating a possible involvement of this BA in promoting angiogenesis. In contrast, incubation in the presence of extract G containing the same concentrations of AKBA did not show any effect on endothelial cell migration capacity, indicating one more time the existence of complex molecular interactions, which can modify the biological effect of the phytoextract. Contrasting results are also reported in literature. Lulli et al. [41] observed that AKBA reduced proliferation, migration, and tube formation in human retinal microvascular endothelial cells (HRMECs) stimulated with exogenous vascular endothelial growth factor (VEGF). On the other hand, Wang et al. [17] reported that β-BA can attenuate endothelial cell injury in a blood stasis model and protect human umbilical vein endothelial cells (HUEVCs) against cell death induced by oxygen and glucose deprivation. Different regulation pathways could be involved in the repairing activity of Boswellia extracts, and further investigations will be necessary to explain why different formulations determine different effects on endothelial cells pathophysiology.
How extracts of B. serrata gum resin should modulate the cardiovascular system has been scarcely investigated, so far. Kokkiripati et al. [42] reported that antioxidant and antithrombotic activities of extracts from B. serrata gum resin determined the inhibition of human monocytic cell activation and platelet aggregation. However, recently Siemoneit et al. [43] pointed out the complex agonizing and antagonizing effects of BAs on human platelet aggregation and prompted for careful evaluation of B. serrata extract safety in cardiovascular disease-risk patients.
In conclusion, our results demonstrate that different formulations (e.g., dry and hydroenzymatic extracts) obtained from the same botanical species show significantly different biological effects on endothelial cells. The antiinflammatory activity of B. serrata extracts on endothelial cells suggests a potential pharmaceutical application for cardiovascular health, though cytotoxicity or proliferative stimulation can occur instead of a protective effect, depending on the dose and the formulation. This aspect should be carefully considered when these herbal products are used in human and animal phytotherapy.

Disclosure
Preliminary data has been presented as a poster at the 21th Congress Phytopharm 2017, Graz, Austria, 2-5 July 2017.

Conflicts of Interest
The authors declare that there is no conflict of interest regarding the publication of this paper.