Extracts from Boswellia serrata have been reported to have anti-inflammatory activity, primarily via boswellic acid-mediated inhibition of leukotriene synthesis. In three small clinical trials, boswellia was shown to improve symptoms of ulcerative colitis and Crohn's disease, and because of its alleged safety, boswellia was considered superior over mesalazine in terms of a benefit-risk evaluation. The goal of this study was to evaluate the effectiveness of boswellia extracts in controlled settings of dextran sulfate- or trinitrobenzene sulfonic acid-induced colitis in mice. Our results suggest that boswellia is ineffective in ameliorating colitis in these models. Moreover, individual boswellic acids were demonstrated to increase the basal and IL-1β-stimulated NF-κB activity in intestinal epithelial cells in vitro as well as reverse proliferative effects of IL-1β. We also observed hepatotoxic effect of boswellia with pronounced hepatomegaly and steatosis. Hepatotoxity and increased lipid accumulation in response to boswellia were further confirmed in vitro in HepG2 cells with fluorescent Nile red binding/resazurin reduction assay and by confocal microscopy. Microarray analyses of hepatic gene expression demonstrated dysregulation of a number of genes, including a large group of lipid metabolism-related genes, and detoxifying enzymes, a response consistent with that to hepatotoxic xenobiotics. In summary, boswellia does not ameliorate symptoms of colitis in chemically induced murine models and, in higher doses, may become hepatotoxic. Potential implications of prolonged and uncontrolled intake of boswellia as an herbal supplement in inflammatory bowel disease and other inflammatory conditions should be considered in future clinical trials with this botanical.
- dextran sulfate
- trinitrobenzene sulfonic acid
complementary and alternative medicine (CAM), and herbal remedies in particular, are increasingly used by patients with chronic diseases, including patients with inflammatory bowel diseases (IBD). In IBD, poor quality of life correlates with increased use of CAM (21). IBD patients turn to alternative therapies for various reasons, including side effects or lack of effectiveness of conventional therapies, fear of surgery, presumed safety and effectiveness of CAM treatments, or the simple desire to regain control of their deteriorating health (15, 27, 30, 31, 39, 44). Whereas certain practices such as acupuncture, chiropractics, massage, or reflexology are generally considered safe, increased use of herbal and homeopathic medicine, often manufactured and marketed without solid scientific basis, should be approached with caution because herbal preparations contain many bioactive compounds with beneficial as well as potentially deleterious effects.
One such herbal supplement widely advertised by dietary supplement manufacturers to treat ulcerative colitis is Boswellia serrata (Roxb. ex Colebr.; Burseraceae). The therapeutic value of dried resinous gum derived from tapping the B. serrata tree, which grows in hilly areas of India, has been known since antiquity. Boswellia gum, mentioned in the ancient Ayurvedic texts, has been used for the treatment of the inflammatory disease in the traditional Ayurvedic medicine in India (18) and was more recently demonstrated as beneficial in bronchial asthma (10), but it had a limited or no effect in rheumatoid arthritis (37) or osteoarthritis (20). In a small clinical trial, extract from B. serrata offered improvement of ulcerative colitis symptoms similarly to that of sulfasalazine (11, 12). Similar findings were reported in patients with Crohn's disease (9). Acetyl-11-keto-β-boswellic acid, a constituent of boswellia resin, also has been shown to attenuate experimental ileitis in rats (19). The gum oleoresin consists of sesquiterpenoid essential oils (5–9%), an ether-soluble fraction (alcohols, esters, boswellic acids; 60–70%), and an ether-insoluble fraction (25–30%) containing polysaccharides (3). It is the boswellic acids, ursane types of pentacyclic triterpenes, that are believed to be the biologically active components of boswellia extract. Indeed, boswellic acids have been reported to possess anti-inflammatory and anti-tumor activity, which may be at least partially due to inhibition of leukocyte elastase (35), 5-lipoxygenase (5-LO) (2, 34), and topoisomerase (42), leading to apoptosis-related tumor cell death (17). The orally administered ethanolic extract of B. serrata or its constituent, acetyl-11-keto-β-boswellic acid, a reported potent inhibitor of leukotriene synthesis (36), has been shown to significantly attenuate inflammatory features of indomethacin-induced ileitis in rats (19). In animal models, leukotriene synthesis inhibitors markedly accelerated healing of colonic ulcers and resolution of colonic inflammation (45, 46). In humans, the involvement of leukotrienes in the pathophysiology of IBD is controversial. Leukotriene B4 (LTB4) levels in rectal dialysate specimens and colonic mucosal specimens are increased in patients with active ulcerative colitis but normal in patients with quiescent disease (22, 40). An in vitro study performed in healthy human peripheral blood neutrophils in Boyden chambers showed that LTB4 was responsible for some of the chemotactic activity in involved ulcerative colitis mucosa (25). Also, current therapies for ulcerative colitis such as corticosteroids and mesalamine produce significant inhibition of LTB4 synthesis (5, 22). Recent clinical studies in ulcerative colitis patients, however, showed that inhibition of 5-LO did not correlate with remission and that 5-LO inhibitors did not differ significantly from placebo in clinical efficacy (14, 23). It would therefore seem that targeting leukotriene synthesis pathway is not sufficient to treat active ulcerative colitis.
Considering the above findings, the wide availability of boswellia extracts as an over-the-counter dietary supplement, and the fact that human trials are proceeding without sufficient base in animal studies or without more extensive toxicology evaluations, we aimed to test the role of boswellia extracts in chemically induced models of murine colitis. Our studies suggest that dietary supplementation with either hexane or methanolic boswellia extracts is ineffective in dextran sulfate sodium (DSS)- or trinitrobenzene sulfonic acid (TNBS)-induced colitis. Moreover, mice fed boswellia-supplemented diets exhibited hepathomegaly and evidence of steatosis, a phenomenon confirmed in vitro with HepG2 cells. Microarray analysis of hepatic gene expression in mice fed a boswellia-supplemented diet identified a number of genes involved in xenobiotics metabolism and detoxification and in steroid metabolism that were significantly induced or inhibited by the botanical extracts.
MATERIALS AND METHODS
Preparation of B. serrata extracts, chromatographic analysis, and diets.
B. serrata was purchased in powder form (lot no. 10417) from San Francisco Herb and Natural Food on June 24, 2002, and was assigned the identification number B9. One kilogram of B9, taken in a 6-liter stainless steel beaker, was treated with 3 liters of n-hexane, stirred to form a homogeneous mixture, and allowed to stand at room temperature (RT) for 24 h. The supernatant was filtered by decantation through a fritted, medium porosity funnel under vacuum. The highly resinous marc was washed by being suspended in 1 liter of fresh n-hexane for 1 h and filtered by decantation as described above. The solvent from the combined filtrate and washing was stripped off under reduced pressure and left under vacuum overnight to yield 113 g (11.3%) of highly resinous hexane extract (B9SDJ3_90F01, or fraction 1). The marc that remained after extraction with hexane was left in the hood overnight for drying. The dried marc was triturated with methanol (3.5 liters) until it was dispersed completely to form a homogeneous mixture. The mixture, after being stirred mechanically at RT for 24 h, was left to stand at RT to allow the residue to settle down and was then filtered by decantation under vacuum as described above. The marc was washed with 1 liter of fresh methanol. The combined filtrate and washing, when evaporated to dryness under vacuum in a 10-liter round-bottomed flask with a LABOROTA 20 rotary evaporator, presented a light yellow, foamy material of 472.4 g (47.2%; B9SDJ3_90F02, or fraction 2). Both final fractions were evaporated under elevated temperatures and under reduced pressure until a constant weight was obtained to ensure complete removal of the extraction solvent.
Chromatographic quantitative analyses of six boswellic acids in the hexane and methanol extracts of B. serrata were performed by using an Agilent 1100 series HPLC system (Palo Alto, CA) consisting of a quaternary pump, degasser, thermostated autosampler, thermostated column compartment, and photodiode-array detector. A Luna C18 (2) column (5 μ, 250 × 4.6 mm) with C18 SecurityGuard guard column (4.0-mm length × 3.0-mm ID) from Phenomenex (Torrance, CA) was used. ChemStation for LC 3D Rev. A.09.03  by Agilent Technologies (1990–2002) was used to process the data and calculate the quantitative levels of boswellic acids. The HPLC separation of the samples was performed using a gradient elution with 0.01% trifluoroacetic acid in Nanopure water (mobile phase A) and HPLC-grade acetonitrile (mobile phase B) at a flow rate of 1.0 ml/min. The mobile phases were filtered under vacuum through a 0.45-μm nylon Whatman filter (Whatman International, Maidstone, UK). The gradient elution had the following profile: 0–15 min, 60–70% solvent B; 15–45 min, 70–80% solvent B; 45–70 min, 80–100% solvent B; 70–80 min, 100% solvent B. The column temperature was set at 30°C, the injection volume of the samples was 20 μl, and the eluent was monitored at 203 (signal A) and 242 nm (signal B). UV spectra were taken in the 190- to 700-nm region. Authentic standards (11-keto-β-boswellic acid, 3-acetyl-11-keto-β-boswellic acid, α-boswellic acid, β-boswellic-acid, 3-acetyl-α-boswellic acid, and 3-acetyl-β-boswellic acid) were obtained from ChromaDex (Santa Ana, CA).
Powdered modified NIH-31 mouse diet (Harlan-Teklad, Madison, WI) was supplemented with 0.1 or 1% of the respective boswellia fraction. Fraction 1 (resinous) was first dissolved in ether and then mixed to homogeneity with the powdered chow and evaporated to dryness under vacuum. The more powdery fraction 2 was mixed with the diet in a mortar to visual homogeneity. Both diets were stored at 4°C for the duration of the study and were prepared daily by being mixed with water and apple juice to 18.75% each to improve the taste and were then served to mice as paste in glass containers.
Induction and evaluation of colitis.
Six-week-old Swiss-Webster mice (29 ± 2.7 g initial body weight; Harlan, Indianapolis, IN) were given 4% DSS (MW 40,000–50,000; USB, Cleveland, OH) in drinking water for 7 days. Animals were killed on day 7 or allowed a 14-day recovery period with normal drinking water without DSS. Age-matched C57BL/6 mice (22 ± 1.4 g initial body weight; Harlan) were administered TNBS (2 mg/mouse in 50% ethanol) or 50% ethanol in a total volume of 100 μl as an enema into the colonic lumen (∼3.5 cm from the anal verge) with the use of a 1-ml syringe fitted with a polyethylene cannula. Diets were changed to control or boswellia-supplemented on day 0 (beginning of DSS or TNBS treatment) and continued till the time of death.
For histological analysis, the colon was removed, rinsed with phosphate-buffered saline (PBS, pH 7.4), and opened longitudinally. Two segments of 2 cm were taken from the proximal and distal part of the colon, fixed in 10% buffered formalin, embedded in paraffin, and sectioned longitudinally. Sections (4 μm) were cut and stained with hematoxylin-eosin. Colon sections were interpreted semiquantitatively in a blinded manner by a veterinary pathologist according to the criteria depicted in Table 1. The final histological score is the sum of the scores from the proximal and distal segments of the mouse colon. Livers also were removed and weighed, and sections were snap-frozen in liquid nitrogen in OCT embedding medium and stored at −80°C until used. Sections (10 μm) were subsequently cut and stained with Sudan black B for visualization of lipids.
All animal studies were approved by the University of Arizona Institutional Animal Care and Use Committee.
Cell culture and transfections.
The HTB-37 clone of the human colonic adenocarcinoma cell line Caco-2 (ATCC, Manassas, VA) was cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 2 mM glutamine, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 100 U/ml penicillin G, and 100 μg/ml streptomycin (all from either Invitrogen, Carlsbad, CA, or Irvine Scientific, Irvine, CA). Cells were transiently transfected with the pNFκB-Luc reporter plasmid (Clontech, Palo Alto, CA) at 70–80% confluency in 24-well plates by using Effectene (Qiagen, Valencia, CA) according to the manufacturer's protocol. Twenty-four hours after transfection, cells were washed with PBS (pH 7.4), and the medium was replaced with one supplemented with 2 ng/ml human recombinant interleukin-1β (IL-1β; Endogen, Rockford, IL), DMSO, and/or individual boswellic acids (ChromaDex). Human liver cells, HepG2 (ATCC), were maintained in minimum essential medium (Invitrogen) supplemented with 10% fetal bovine serum and penicillin/streptomycin (Irvine Scientific) at 37°C and 5% CO2.
Reporter gene and proliferation assays.
Control, IL-1β-treated, and boswellic acid-treated Caco-2 cells were rinsed with PBS and lysed in passive lysis buffer (Promega, Madison, WI) and then assayed for firefly luciferase activity with the luciferase assay system (Promega) according to the manufacturer's recommendations by using a tube luminometer (FB12; Zylux, Oak Ridge, TN). Because all vectors tested as internal controls were to some extent regulated by either IL-1β or boswellic acids, luciferase activity (relative light units, RLU) was normalized by protein concentration instead (RLU/μg protein). Cell proliferation and viability were assessed in Caco-2 cells seeded in 96-well plates by using the CellTiter-Glo luminescent cell viability assay (Promega), a homogeneous method for determining the number of viable cells in culture based on quantitation of the ATP present, an indicator of metabolically active cells.
Lipid accumulation and viability assay in HepG2 cells.
Nile red, resazurin, and Pluronic F-127 were purchased from Sigma Chemical (St. Louis, MO). Cells were seeded at 6 × 104/well in 96 white, clear-bottomed plates and were incubated the following day with boswellia fraction 1 or fraction 2 for 24, 48, 72, and 96 h. Mifepristone (100 μM) was used as a positive control. All compounds were diluted in medium to 0.1% DMSO final concentration. Nile red binding assay was prepared as previously described (26). After the indicated exposure, cells were washed with Hanks' balanced salt solution (HBSS; Sigma) and background fluorescence was read (excitation 544 nm, emission 590 nm) in a Fluoroscan Ascent FL (Labsystems). Cells were then incubated in 100 μl of 1 μM Nile red with 1% Pluronic F-127 and 0.1% DMSO. Incubation was carried for 4 h at RT in the dark, and cells were washed once with HBSS (Invitrogen). Fresh HBSS was added into wells, incubation continued for 16 h, and fluorescence was read as described above. To detect the cytotoxic effect of used compounds, we added 10 μl of resazurin in HBSS to cells and read the background fluorescence twice: immediately and after 1 h of incubation at RT. All values obtained from the cytotoxicity assay were ∼10 times higher than those from the Nile red binding assay; therefore, results obtained from the same cells were not affected.
Microarray analysis of hepatic gene expression.
Livers from three mice in each dietary group (control diet, 1% fraction 1, and 1% fraction 2) were pooled for RNA isolation. Two pooled preparations from each group were used for microarray analyses. Total RNA was isolated with TRIzol reagent (Invitrogen) according to the manufacturer's protocol but modified to include LiCl precipitation to increase RNA purity. RNA was subsequently processed, and biotin-labeled cRNA was produced essentially according to the manufacturer's instructions (Affymetrix; Expression Analysis Technical Manual) by using reagents provided by Affymetrix [GeneChip sample cleanup module, T7 oligo(dT) primer, Enzo BioArray High Yield RNA transcript labeling kit] and Invitrogen (dNTP mix, Superscript II reverse transcriptase, Escherichia coli DNA ligase, E. coli DNA polymerase I, E. coli RNase H). Fragmented cRNA was mixed with control oligonucleotide B2 (Affymetrix), eukaryotic hybridization controls (Affymetrix), herring sperm DNA (Invitrogen), bovine serum albumin (Invitrogen), 2× hybridization buffer, and RNase-free water. This hybridization cocktail was then applied to mouse MOE430A arrays (Affymetrix) and hybridized at 45°C for 16 h while spinning at 53 rpm. The MOE430A array contains probe sets against 13,672 well-annotated genes and 432 EST clones. Chips were immediately washed and stained with the GeneChip Fluidics Station 400 (Affymetrix). Streptavidin phycoerythrin and antibody solutions were prepared according to the manufacturer's recommendations (Affymetrix), and chips were washed and stained using the EukGE-WS2v4 fluidics protocol. After chips had been washed and stained, they were scanned with the Agilent GeneArray scanner (Affymetrix). Data obtained have been subsequently normalized to edited normalization mask file (Microarray Suite, v.5.0; Affymetrix) and exported for analysis to GeneSpring v.6.2 (Silicon Genetics, Redwood City, CA). Stringent empirical analysis was employed to compare gene expression profiles between mice on control diet and on diet supplemented with respective boswellia fractions with a cross-gene error model based on replicates. Data were filtered in the following order: select genes flagged as present or marginally present in two of four samples; select genes with raw expression values greater than double the acceptable background (>200); and select genes with a change equal or greater than threefold (up or down) compared with the control dietary group. Venn diagrams were used to identify genes in which expression was altered by either one of or both the boswellia fractions.
Hexane extraction was aimed at eliminating highly resinous content from the subsequently prepared alcoholic extract. Because liquid chromatography analyses of the two fractions obtained indicated a substantial carryover of the potentially bioactive boswellic acids into the hexane fraction (fraction 1), both fractions were used as dietary supplements to evaluate their potential in reducing tissue damage in chemically induced colitis. The calculated total content of the six boswellic acids was 29.9 and 34.4% in fractions 1 and 2, respectively, with some boswellic acids partitioning preferentially to either fraction (Fig. 1). The prepared supplemented diets were palatable to the experimental mice, and no differences in food consumption were noted as demonstrated by body weight of animals on different diets without colitis (Figs. 2A and 3A). One exception was noted in mice fed 1% fraction 2, which showed a tendency toward lower body weight after 21 days of treatment (Fig. 2A). Both fractions, but particularly fraction 2, had a negative impact on the overall health status of the mice receiving them, regardless of the DSS treatment. Mice fed a higher dose (1%) of fraction 2 exhibited abnormal posture, hypoactivity, lack of grooming, and rough coat.
Boswellia does not prevent weight loss or mucosal damage in chemically induced colitis.
The well-established murine model of DSS-induced colitis (28) is commonly used to screen pharmacological agents. The developed histopathological scoring criteria provide a reliable means of quantifying disease severity that correlates with histological healing. The maximum colonic inflammation usually developed at day 7, and after termination of DSS administration, clinical and histopathological parameters slowly improved. Neither fraction of boswellia used in this study improved survival rate or prevented body weight loss observed in DSS-treated mice at both the acute (7 days) and recovery phase (21 days) of colitis (Figs. 2A and 3A). In fact, a larger body weight loss was observed in animals fed boswellia-supplemented diets after 7 days of DSS treatment. This body weight loss was especially dramatic in mice fed 1% fraction 2 for 21 days (Fig. 3A). Histological assessment also showed that during the acute phase of DSS colitis, feeding diets supplemented with either of the two boswellia fractions at 0.1% resulted in a further exacerbation of symptoms and an increased histopathology score (Figs. 2B and 4) with mucosal ulcerations, multifocal transmural necrosis, and more significant edema than in DSS-treated mice on control diet. Although aberrant crypts were still present in boswellia-fed, DSS-treated mice, we observed a tendency toward an increased number of goblet cells compared with DSS-treated mice on control diet. Feeding boswellia-supplemented diets offered no significant improvement during the recovery phase of colitis (21-day protocol; Fig. 3B). Both boswellia extracts also were found to be similarly ineffective in improving mortality, body weight loss, or histology of the colon in TNBS-treated C57BL/6 mice (data not shown).
Effect of boswellic acids on NF-κB activity and proliferative response of intestinal epithelial cells to IL-1β.
Because both fractions appeared ineffective in reducing the symptoms of colitis, and in some cases increased their severity, we investigated the influence of individual boswellic acids on the activity of NF-κB, a major transcription factor regulating expression of inflammatory mediators. As demonstrated in Caco-2 cells transiently transfected with NF-κB reporter construct, not only did four of six boswellic acids, namely, α-boswellic acid, 3-acetyl-β-boswellic acid, 11-keto-β-boswellic acid, and 3-acetyl-11-keto-β-boswellic acid, increase basal activity of the NF-κB-dependent promoter, but the latter three acids also increased IL-1β-stimulated activity of NF-κB (Fig. 5, A and B). The concentration used (50 μM) was consistent with the reported IC50 values for inhibition of 5-LO product formation (34).
The epithelium of the gastrointestinal tract is rapidly renewing, and it can greatly increase its proliferative rate in response to inflammation and certain inflammatory mediators, such as IL-1β. This crucial process of mucosal repair and regeneration maintains the epithelial integrity necessary for gut homeostasis. Consistent with the in vivo observations of the two fractions of B. serrata and the effects of boswellic acids on NF-κB activity in vitro, we observed that the proliferative response of Caco-2 cells to IL-1β was reversed in the presence of 3-acetyl-11-keto-β-boswellic acid, the reportedly active component of boswellia gum resin with the highest potency to inhibit leukotriene synthesis (48) and to attenuate indomethacin-induced ileitis in rats (19) (Fig. 5C).
Hepatotoxicity of boswellia extracts in experimental mice.
A striking feature of mice fed higher doses (1%) of the two extracts of B. serrata for 21 days was a pronounced increase in the liver size. This was especially evident in mice fed 1% fraction 2, in which the liver-to-body weight ratio increased by 48 ± 5% compared with that in mice fed a control diet (Fig. 6, A and C). The increase in liver size corresponded with an increased deposition of intracellular lipids, visible macroscopically in the form of white speckles and microscopically in Sudan black-stained frozen sections of the livers (Fig. 6, B and D).
The Nile red binding assay was used as a screen for steatosis-inducing compounds to confirm the effect of the two boswellia fractions in vitro with human hepatoma HepG2 cells. This assay, combined with resazurin reduction assay of viability of the same cells, was developed by McMillian et al. (26), yielding comparative results in HepG2 cells and isolated primary dog and rat hepatocytes. Both fractions induced progressive steatosis in HepG2 cells in concentrations of 50 μg/ml and higher as expressed by the Nile red/resazurin fluorescence ratio, representing a measure of lipid accumulation in viable cells (Fig. 7). The induced steatosis also could be observed under a confocal microscope in HepG2 cells stained with 4,6-diamidino-2-phenylindole and Nile red (Fig. 8).
Microarray analysis of hepatic gene expression.
Of the ∼14,000 genes analyzed, application of the stringent criteria identified 24 genes (including 1 RIKEN clone not annotated) regulated by both boswellia fractions, 58 genes regulated exclusively by fraction 1 (including 15 RIKEN and IMAGE clones not annotated), and 20 genes regulated exclusively by fraction 2 (including 7 RIKEN clones not annotated). Because the resinous hexane-soluble fraction (fraction 1) would not normally be considered of therapeutic value and was used here only for comparative purposes, and also for the sake of brevity, only well-annotated genes regulated by both fractions and genes regulated exclusively by alcohol fraction 2 are presented in this report (Tables 2 and 3, respectively). More detailed results of the analyses, including raw and normalized expression values, can be viewed at the National Center for Biotechnology Information Gene Expression Omnibus microarray depository web site (http://www.ncbi.nlm.nih.gov/geo/; GEO accession no. GSE1847). All hybridization parameters were within acceptable range [raw Q <3.5, background <100, 3′/5′ ratio of hybridization controls <3, and spike controls present 70% (Bio-B) and 100% of the time (Bio-C,D and Cre)].
In Table 2, 22 of 23 genes listed were regulated by both fractions in a similar manner as far as the direction of regulation (up- or down-) and relative change, suggesting the influence of components present in both fractions. One exception was noted for Fmo3, flavin-containing monooxygenase 3, which was upregulated eightfold in the livers of mice fed a fraction 1-supplemented diet and reduced to nondetectable levels in mice fed a fraction 2-supplemented diet.
A number of genes listed in Tables 2 and 3 could be described as phase I and phase II metabolizing enzymes, typical responders to hepatotoxic xenobiotics and/or oxidative stress, including four members of the cytochrome P-450 family (with the strongest induction of Cyp2c55 and Cyp2b20) and three members of the glutathione S-transferase (GST) family (with the strongest being 28- to 38-fold induction of GST mu3). Several genes identified using microarray analysis are involved in steroid metabolism, such as aldo-keto reductase/Akr1b7, sterol-C5-desaturase/Sc5d, fatty acid translocase/CD36, and hydroxysteroid dehydrogenase/Hsd3b5. Expression of the latter of these enzymes, Hsd3b5, a male liver-specific NADPH-dependent 3-ketosteroid reductase catalyzing the inactivation of steroid hormones, such as dihydrotestosterone (1), was dramatically reduced by both boswellia fractions. Although not directly involved in steroid metabolism, insulin-induced gene 2/Insig2 also should be acknowledged, because it can regulate lipid synthesis by blocking the proteolytic activation of Sterol regulatory element binding proteins (49). Despite visible signs of ongoing inflammation, such as lymphocytic or neutrophilic infiltration, several genes related to immune response were also induced (H2-D1 and H2-K1 histocompatibility genes, serum amyloids A1 and A2), but a surprising reduction in expression of anti-inflammatory SOCS3 gene was noted in the livers of mice fed fraction 2-supplemented diet.
Unconventional therapies are utilized by ∼25% of patients seeking medical attention for a serious medical problem, and 70% of these patients do not disclose their use to their physicians (6, 7). In a general adult population, 42% of those polled reported using alternative therapies, according to a survey conducted in 1997 (6). The author of this survey estimated that 15 million adults were at some risk for potential drug-supplement interaction. In the population of patients diagnosed with IBD, a comparable percentage (34–51%) reported the use of alternative and complementary medicine (16, 30, 31, 44). In a recent analysis of a Canadian IBD population, the frequency of herbal therapy use came in second (17% of patients) after diet (45%) (4). The primary reason for IBD patients to reach for alternative therapies is a perception that the conventional treatment is not effective or a fear of serious side effects related to corticosteroid use (39).
Patients actively seek credible information on complementary therapies from various sources. Most frequently, physicians, pharmacists, and pharmacy technicians are not adequately equipped to provide satisfactory advice, largely because of a scarcity of peer-reviewed scientific information on this subject. Patients are therefore often left to turn to herbal supplement manufacturers' advertising, leading to a number of prevalent misconceptions and myths, such as that herbs are equally or more effective and have fewer side effects than conventional pharmaceuticals, or that being natural, they are pure and can be taken safely without consulting a physician. Many consumers do not realize that herbal products do not undergo the same scrutiny from the Food and Drug Administration as do conventional medications and that herbal product manufacturers are not required to demonstrate the safety or efficacy of an herb before marketing it.
One such herbal medication widely advertised as being effective in reducing the symptoms of ulcerative colitis and other inflammatory conditions (osteoarthritis, rheumatoid arthritis, bursitis, asthma) is B. serrata, or more precisely, an alcoholic extract from the gummy oleoresin. Boswellia is typically referred to as generally safe when used as directed, with rare side effects such as diarrhea, skin rash, and nausea. Determining the reported doses of boswellia and their direct comparison with our experimental model is complicated by the fact that the methods used to obtain them, as well as the composition of the extracts, are never disclosed and are bound to vary greatly not only between manufacturers but also among lots. We also have found that, in certain cases, doses used in clinical trials are misreported on the web sites of herbal product distributors. For example, in the setting of ulcerative colitis, doses of 300 mg (11) or 350 mg (12) three times daily have been published, but 550 mg three times daily are listed by a distributor referring to one of the two above-cited studies. Also, in a polyarthritis study, doses as high as 3.6 g daily were administered to patients according to Sander et al. (37). One also has to consider the limited duration of our study as opposed to the chronic nature of ulcerative colitis or Crohn's disease and the fact that there is virtually no control over the amount or quality of the ingested herb.
The presented data from the chemically induced models of murine colitis suggest a lack of effectiveness of boswellia extracts in attenuating clinical signs of colitis. Dietary supplementation with 0.1% or 1% boswellia extract did not improve the pathological score in either of the two models employed: DSS colitis, representing a model of mucosal damage and restitution, or the TNBS model, representing a T-cell-mediated reaction to a haptenizing agent. The available evidence seems to support the hypothesis that the major mechanism of the alleged anti-inflammatory activity of boswellia is inhibition of leukotriene synthesis by boswellic acids. The efficacy of targeting the leukotriene synthesis and signaling pathways in inflammatory bowel disease is, however, questionable. Although leukotriene expression, and LTB4 in particular, is undoubtedly associated with the severity and relapse of ulcerative colitis, inhibitors of leukotriene synthesis have been repeatedly reported to be ineffective in the treatment or maintenance of remission in ulcerative colitis (13, 32). From this perspective, therefore, the lack of effect of boswellia extracts in our experimental models was not unanticipated, although it is not supported by the limited clinical data available. The discrepancy between our results and those of published clinical trials may indicate that the results from chemically induced models of murine colitis (both DSS and TNBS) are not always easily translated to human pathology. It is also possible that absorption and/or metabolism of components of B. serrata varies between species. One also has to consider the small scale of the poorly controlled trials, especially in patients with ulcerative colitis (11, 12).
The exacerbation of the histological score in the colon during the acute phase of DSS-induced colitis in mice fed boswellia extracts was surprising. This phenomenon could be explained, at least partially, by the observed effect of individual boswellic acids on basal and IL-1β-stimulated activity of NF-κB in the intestinal epithelial cells as well as reversing the proliferative effects of inflammatory mediators. The mechanism by which the selected boswellic acids activate NF-κB and potentiate the effect of IL-1β are not known but may involve oxidative stress, as suggested by liver microarray analyses (see below). The intestinal epithelium is a rapidly renewing cellular compartment that can greatly increase its proliferative rate in response to inflammation (24). This crucial process of mucosal repair and regeneration maintains the epithelial integrity necessary for gut homeostasis. Among the cytokines responsible for this effect is IL-1β, which can influence the proliferation of a variety of cell types, including intestinal epithelial cells (IEC) (43), fibroblasts (29), keratinocytes (38), and thymic epithelial cells (8). It was therefore intriguing to find that 3-acetyl-11-keto-β-boswellic acid (AKBA), the same compound reported as attenuating experimental ileitis in rats (19), reversed the effects of IL-1β on proliferation of Caco-2 cells in a dose-dependent manner. This finding suggests that AKBA and perhaps more components of boswellia extracts may impair epithelial restitution in the course of colitis. Induction of NF-κB activity by AKBA may be an underlying mechanism for inhibition of IL-1β-mediated cell proliferation. Activation of NF-κB by other proinflammatory cytokines, such as TNF-α, results in inhibition of proliferation and induction of apoptosis in IEC. This explanation would require the assumption that an IL-1β-mediated increase in proliferation is indirect and NF-κB independent. Although the involvement of NF-κB in proliferative responses of IEC to IL-1β is not known, indirect mechanisms of IL-1β actions on cell proliferation have been documented. An example of such a mechanism involves IL-1β induction of secretion of IGF-binding proteins by IEC (41), which would increase the bioavailability of IGF, a mitogenic growth factor. This hypothesis is supported by the fact that IL-1β does not induce proliferation of IEC under serum-free conditions (43), an environment devoid of growth factors that at the same time does not preclude activation of NF-κB.
The most striking finding in our studies was increased liver size and clear evidence of steatosis, particularly in mice fed the higher dose of methanolic extract of B. serrata. This observation was further confirmed in vitro with human HepG2 cells as a model. Although microarray analyses of hepatic gene expression cannot fully explain this phenomenon, it gives important clues to the potential hepatotoxic effects of Boswellia. The liver is the major organ responsible for the biotransformation of xenobiotics, including procarcinogens and drugs, to more hydrophilic products and for their elimination through bile secretion. Biotransformation results from the activities of phase I and phase II metabolizing enzymes that are mainly, although not exclusively, located within hepatocytes. Often, these enzymes facilitate detoxification of the parent compounds, but they also may occasionally lead to the formation of toxic metabolites and of autoantigens. The observed induction of selected detoxifying enzymes also points to a potential risk of herb-drug and/or drug-drug interaction when boswellia extracts complement conventional medication. This may result in either increased elimination of the pharmaceutical, leading to its decreased potency and effectiveness, or increased production of toxic metabolites. The potential physiological or pathophysiological consequences of induction or inhibition of expression of the identified genes can only be speculated upon by extrapolation of the results to their known downstream products and/or targets. For example, Cyp2c55 gene coding for a P-450 epoxygenase, induced 38-fold by hexane fraction 1 and over 28-fold by methanolic fraction 2, is involved in the metabolism of arachidonic acid and linoleic acid to epoxyeicosatrienoic acids, hydroxyeicosatetraenoic acids, and epoxyoctadecenoic acids and to hydroxyoctadecadienoic acids, respectively (47). These compounds can accumulate in the liver and serve as cyclooxygenase substrates to produce prostaglandin and prostaglandin analogs and can generate reactive oxygen species (33), thus participating in the regulation of liver hemodynamics and metabolic activity. They also appear to be involved in liver diseases such as cirrhosis and play a key role in the pathophysiology of portal hypertension and renal failure. Interestingly, this P-450 isoform is primarily expressed in the colon (47). Several other genes may potentially be linked to the observed steatosis, including CD36 fatty acid translocase. Induction of CD36 by overexpression of peroxisome proliferator-activated receptor (PPAR)-γ1 in PPAR-α−/− mice parallels adipogenic transformation of hepatocytes (50). Sc5d, a sterol-C5-desaturase, is a critical enzyme in cholesterol synthesis, catalyzing the conversion of lathosterol into 7-dehydrocholesterol. Its deficiency results in lathosterolosis characterized by impaired cholesterol synthesis and lathosterol accumulation (OMIM no. 607330). Although we were unable to find reports of increased expression of Sc5d in published literature, by analogy, one could anticipate increased synthesis of 7-dehydrocholesterol, which, if not metabolized promptly by DHCR7, may display teratogenic effects as observed in Smith-Lemli-Opitz syndrome (OMIM no. 270400) or may result in increased cholesterol synthesis and accumulation. Other genes that did not meet the stringent selection criteria and as such were not listed in Tables 2 and 3 but are available through the GEO repository (see materials and methods) also may provide clues to the hepatotoxicity of B. serrata in mice.
Although the results presented in this short report may or may not directly translate into human pathophysiology, this compelling evidence should be a cause for concern and will hopefully prompt more scrupulous toxicological assessment of liver functions in patients participating in clinical trials with B. serrata. Contrary to the widespread popular view that “because it is natural, it is safe,” herbal therapy carries more risks and produces more serious side effects than any other form of alternative therapy. Unfortunately, there are no formal data on the ineffectiveness of certain compounds and on the incidence even of acute, severe side effects, such as liver failure, after taking certain herbal medications. This is primarily due to the approach of investigators to negative data and to the skewed peer review process, which favors positive and promising results. Although the pharmaceutical industry recently agreed to give physicians and patients full access to both positive and negative results of clinical trials, a systematic reporting of the collection of ineffective trials and adverse responses to herbs is still missing.
This work was supported by National Institutes of Health Grants P50 HL-61212-01, R01 DK-067286-01, and P50 AT-000474-04.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society