Interactions between Phytochemical Components of Sutherlandia Frutescens and the Antiretroviral, Atazanavir In Vitro: Implications for Absorption and Metabolism

– Purpose. African traditional medicinal plants, such as Sutherlandia frutescens have the potential to interact pharmacokinetically with the protease inhibitor class of antiretrovirals, thereby impacting on their safety and efficacy. The effects of extracts and phytochemical components of Sutherlandia frutescens , on the in vitro absorption and metabolism of the protease inhibitor, atazanavir were thus investigated. Methods. Aqueous and methanolic extracts of Sutherlandia frutescens were prepared by freeze-drying of hot water and methanol decoctions of Sutherlandia frutescens plant material respectively, whilst crude triterpenoid glycoside and flavonol glycoside fractions were isolated by solvent extraction and subsequent column chromatography. Atazanavir was quantitated in the absence or presence of these compounds as well as commercially available purported constituents of Sutherlandia frutescens , namely, L-canavanine, L-GABA and D-pinitol, after a one hour co-incubation in Caco-2 cell monolayers and human liver microsomes. Results. The triterpenoid and flavonol glycoside fractions were found to be present in the aqueous and methanolic extracts of Sutherlandia frutescens and were shown to contain the sutherlandiosides and sutherlandins known to be present in Sutherlandia frutescens . The aqueous extract and D-pinitol significantly reduced atazanavir accumulation by Caco-2 cells, implying a decrease in atazanavir absorption, whilst the opposite was true for the triterpenoid glycoside fraction. Both the aqueous and methanolic extracts inhibited atazanavir metabolism in human liver microsomes, whilst enhanced atazanavir metabolism was exhibited by the triterpenoid glycoside fraction. Conclusions. The extracts and phytochemical components of Sutherlandia frutescens influenced the accumulation of atazanavir by Caco-2 cells and also affected ATV metabolism in human liver microsomes. These interactions may have important implications on the absorption and metabolism and thus the overall oral bioavailability of atazanavir.


INTRODUCTION
Several surveys detail the use of African traditional medicines (ATMs) by HIV/AIDS patients in South Africa (1)(2)(3)(4). Moreover, there have also been anecdotal reports on the use of the indigenous Southern African plants, Hypoxis hemerocallidea (African potato) and Sutherlandia frutescens (SF) for the treatment of HIV/AIDS (5), which was even recommended at a consultative meeting of representatives of the Ministries of Health of the member states of the Southern African Development Community (SADC), in 2003 (6).
The efflux transporter, P-glycoprotein (P-gp) and CYP450 isoenzymes are expressed in the small intestine and the liver and act on several drug substrates, thereby limiting their oral absorption and facilitating their metabolism (7)(8)(9).
Through modulation of CYP isoenzymes and/or P-gp activities, many in vitro studies (10)(11)(12)(13)(14)(15)(16) allude to potentially significant effects of ATMs on the absorption and metabolism of ARVs. This in turn suggests that concomitant oral administration of the ATMs with ARVs may increase the risk for clinical adverse reactions, such as toxicity as a result of increased absorption and/or reduced metabolism, and lack of efficacy consequent to decreased absorption and/or enhanced metabolism.
The protease inhibitor (PI), atazanavir (ATV) has less severe adverse effects in comparison to lopinavir, therefore it has been included in the South African clinical guidelines for the management of HIV/AIDs in adults and ________________________________________ adolescents (17), as an alternative to lopinavir in patients who experience intolerable gastrointestinal problems, hyperlipidaemia and/or hyperglycaemia. Like other PIs, ATV is a substrate of P-gp (18,19) as well as CYP3A4 and CYP3A5 (20) and may thus be susceptible to pharmacokinetic (PK) interactions with agents able to modulate the activities of this transporter and family of CYP enzymes.
SF is a Southern African plant which has a long history of use in the practice of traditional medicine, particularly by the isiZulu, isiSotho, isiXhosa and Khoi-San people of South Africa (21,22). Aqueous decoctions prepared from this medicinal plant have been used to treat a variety of symptoms and illnesses, such as pain and fever, wounds, stomach ailments, stress and anxiety, cancer and diabetes (21,23). With the advent of the HIV/AIDS crisis in SA, SF has become popular as an "adaptogenic tonic", which anecdotal reports claim to be useful for alleviating the cachexia (muscle-wasting) (21,22) commonly observed in patients at the end stages of the disease. Pharmaceutical dosage forms, such as tablets and capsules which contain powdered SF leaves are currently readily available in pharmacies, health shops and even online. Triterpenoid glycosides known as Sutherlandiosides A -D (24) and flavonol glycosides, kaempferol and quercetin glycosides known as sutherlandins A -D (25) have been isolated from SF plant material and are known to be present in the plant (26). Moreover, SF is also purported to contain the non-protein amino acid, L-canavanine, the inhibitory neurotransmitter, L-GABA and the sugar, D-pinitol (22). Aqueous and methanolic extracts (100 mg/ml) of SF demonstrated ≥ 50% CYP3A4 inhibitory activity, but ≤ 50% inhibition of P-gp activity (12).
The potential impact of SF on the safety and efficacy of ATV was investigated by determining the effects of aqueous and methanolic extracts as well as phytochemical constituents known or purported to be present in SF, on the accumulation of ATV by Caco-2 cells, an established in vitro model of human intestinal epithelium and the metabolism of ATV in human liver microsomes.

Chemicals and Reagents
Caco-2 cells were obtained from the American Type Culture Collection (Manassas, Virginia, USA) and pooled human liver microsomes were purchased from BD Biosciences (Woburn,

Preparation of SF Extracts
SF is traditionally prepared for oral administration by the infusion of 2.5 − 5 g dry plant material in a cup of boiling water (22). To be able to examine the effect of the phytochemical constituents which may be present in such a decoction, an aqueous extract of SF was prepared by boiling SF leaves (5 g) in 250 ml water for 20 minutes on a hot plate. The brew was left to cool for 3 hours and was then centrifuged at 10 000 rpm for 15 minutes. The supernatant was transferred to a round-bottomed flask, shell-frozen in liquid nitrogen and freeze-dried. A methanolic extract was similarly prepared, except that SF leaves (5 g) in 250 ml methanol were heated at 50ºC. The rationale for the use of the methanolic extract was to be able to determine the effects of phytochemical constituents which are less polar and unlikely to be present in the aqueous extract in any significant quantities. These constituents along with those present in the aqueous extract are likely to be amongst the various phytochemical components in commercial dosage forms which contain SF plant material.

Preparation of SF Flavonol Glycoside and Triterpenoid Glycoside Fractions
Prior to the isolation (24,25) and thus availability of sutherlandiosides and sutherlandins present in SF plant material, crude flavonol and triterpenoid glycoside fractions were prepared for use as indicators of the presence of flavonol and triterpenoid glycosides in SF plant material and extracts thereof.
To prepare the flavonol glycoside and triterpenoid glycoside fractions, SF plant material (100 g) was extracted with ethanol by soxhlet extraction and concentrated under vacuum. The concentrated mass was subsequently extracted with diethyl ether. The ethereal fraction was diluted with alcohol and chromatographed on a silica gel column, where dichloromethane was used to elute an organic fraction which contained the triterpenoid glycosides. The fraction which precipitated on the column and which contained the flavonol glycosides was eluted with ethanol. These crude fractions were further fractionated using a stationary phase of silica gel in dichloromethane and eluted with varying concentrations of ethanol in dichloromethane.

Confirmation of the Presence of Sutherlandiosides and Sutherlandins in SF Triterpenoid and Flavonol Glycoside Fractions and SF Aqueous and Methanolic Extracts
Qualitative HPLC analyses of SF aqueous and methanolic extracts (10 mg/ml), the triterpenoid glycoside fraction (500 µg/ml) and flavonol glycoside fraction (500 µg/ml) were conducted, using a method modified from Prevoo et al. 2008 (27). These were the final concentrations of the respective extracts and components used in the in vitro studies. Such analyses were also conducted on either individual or combined Sutherlandiosides and sutherlandins at a concentration of 50 µg/ml for each. These qualitative analyses served to confirm that the triterpenoid and flavonol glycoside fractions contained the relevant sutherlandiosides and sutherlandins and that the triterpenoid and flavonol glycoside fractions were present in the methanolic and aqueous extracts.

In Vitro Intestinal Absorption Studies Cell Culture
Caco-2 cells are human colon adenocarcinoma cells that can undergo spontaneous enterocytic differentiation and polarization once they reach confluency in culture. Completely differentiated monolayers of Caco-2 cells display microvilli and brush border hydrolases on the apical surface, and express endogenous transport systems that are similar to human intestinal epithelium. Because of this structural and functional homology between intestinal epithelium, the primary use of Caco-2 cells has been for transepithelial flux / uptake / accumulation experiments to evaluate transcellular and/or paracellular absorption of drugs. Furthermore, the expression of efflux transporters, such as P-gp, MRPs, and BCRP, and members of the SLC family of influx transporters, such as OCTs and OATP-B have been well characterized (28)(29)(30).
Caco-2 cells used from passages 65 to 80 were grown as a monolayer at a density of 2000 cells/cm 2 on 75 cm 2 tissue culture flasks in DMEM supplemented with 10% foetal bovine serum, 1% non-essential amino acids, 2 mM Lglutamine, 100 units/ml penicillin and 0.1 mg/ml streptomycin, which was replaced every 2 to 3 days, in an incubator set at 37ºC, 95% humidity and 5% CO 2 . Cells from confluent cultures were harvested using trypsin-EDTA (0.05%) in DMEM and were seeded into 48 well plates at a density of 60 000 cells/cm 2 . Cells were cultured in the plates under the same conditions described above for 21 to 24 days to allow cell differentiation into a tight epithelial monolayer.

Preparation of Conditioning and Incubation Buffers
Conditioning and incubation buffers were prepared using Hank's buffered salt solution (HBSS) containing 1.3 mM CaCl 2 , 0.49 mM MgCl 2 , 0.41 mM MgSO4, 5.3 mM KCl, 0.44 mM KH 2 PO 4 , 138 mM NaCl, 0.34 mM Na 2 HPO4, and 5.6 mM D-glucose supplemented with 25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) and 0.01% BSA, which was adjusted to pH 7.4. Aqueous and methanolic extracts of SF were prepared in water and methanol respectively to give a final concentration of 10 mg/ml in the conditioning and incubation buffers. This concentration was used to determine whether more dilute extracts of SF than the 100 mg/ml previously investigated (12) were still potent enough to exert an effect on the in vitro system of human intestinal absorption. The flavonol and triterpenoid glycoside fractions were similarly prepared in water and methanol respectively. Since these were crude fractions and not individual flavonol and triterpenoid glycosides, the concentration was determined by mass rather than molarity. A final concentration of 500 µg/ml was expected for each of the two classes of compounds assuming that all 5% w/w of flavonol and of triterpenoid glycosides estimated to be present in SF plant material (26) was extracted when a traditional aqueous decoction of 2.5 g SF plant material in a cup (250 ml) boiling water was prepared. L-canavanine, L-GABA and D-pinitol content in SF appears to be highly variable (22), therefore an estimated average final concentration of 100 µM was chosen for L-canavanine, L-GABA and D-pinitol, which were all dissolved in water. PSC833 (1 µM) dissolved in dimethylsulphoxide (DMSO) was used as the positive control for increased accumulation of ATV in Caco-2 cells. Solvent content was limited to 1% since preliminary experiments showed that at this concentration, water, methanol and DMSO had no effect on ATV accumulation in the cells. In addition, the incubation buffer (but not the conditioning buffer) contained [ 3 H] ATV (0.1 µCi/ml) supplemented with unlabelled ATV to give a final concentration of 1 µM ATV.

Cellular Accumulation Experiments
These experiments were performed according to well established methods in Dr. R. Bendayan's laboratory (31). To maintain maximum viability of the cells, culture media was replaced 24 hours before the experiment. Accumulation studies were conducted at 37ºC and were performed in triplicate for each group on the same cell culture plate. Culture media was aspirated and cells were initially washed and conditioned for 15 minutes with 250 µl warm (37ºC) conditioning buffers. Preliminary experiments showed that 1 hour incubation of Caco-2 cells with SF extracts and components, at these concentrations, under light microscopy at a magnification of 400 X, had no effect on the cell morphology, therefore the viability of the cells was not likely affected by the extracts and constituents of SF for the duration of the incubation time used. The pre-incubation buffers were aspirated and the cells were then incubated with the incubation buffers at 37ºC for 1 hour. At the end of the incubation period, the reaction was stopped and the cells were washed twice by the addition of ice-cold (4ºC) 0.1 M PBS (500 µl) after aspiration of the incubation buffer. The cells in each well were solubilised with 1% Triton-X (500 µl) at 37ºC for 30 minutes before transfer to scintillation vials. Scintillation fluid (2 ml) was added to each vial and the radioactivity of each sample was measured. Three wells of each plate which were only exposed to blank HBSS buffer throughout the experiment were set aside for the determination of protein concentration. Standard solutions of BSA (0.1, 0.25, 0.5, 0.75 and 1 mg/ml) were also prepared. The Bradford method adapted for compatibility with detergents was applied to the samples and standard solutions. Absorbances were measured on a UV spectrophotometer at 750 nm. ATV cellular accumulation was then expressed in pmol/mg protein.

In Vitro Liver Metabolism Studies Preparation of Microsomes
Pooled human liver microsomes (20 mg/ml protein as purchased) were thawed in a water bath at 37ºC and diluted to 2 mg/ml protein with PBS.

Preparation of Incubation Buffers
A stock solution of ATV was prepared by dissolving an accurately weighed amount of ATV sulphate in a mobile phase, viz acetonitrileammonium formate buffer (10 mM, pH 3) (45: 55, v/v) to provide the equivalent of 5 mM ATV. Similarly, a stock solution of NADPH (10 mM) was prepared in PBS. Using these solutions, an incubation buffer containing 0.05 mM ATV and 2 mM NADPH was prepared in PBS. Incubation buffers containing aqueous and methanolic extracts of SF were prepared at concentrations of 20 mg/ml. The flavonol and triterpenoid glycodside fraction buffers had concentrations of 1 mg/ml and those of L-canavanine, L-GABA and D-pinitol had a concentration of 200 µM.
Each incubation mixture was prepared in triplicate and incubated for 1 hour at 37ºC. The reaction was stopped by addition of ice cold (4 ºC) mobile phase (250 µL), viz, acetonitrileammonium formate buffer (10 mM, pH 3) (45: 55, v/v) containing DIAZ (2 µg/ml) used as an internal standard for the HPLC analysis.

HPLC Analysis of ATV in Microsomes
Calibration standards of ATV were prepared by serial dilution of the ATV stock solution with 50:50 PBS: mobile phase, containing 1 µg/ml DIAZ, as the IS. The chromatographic conditions used were as previously described for the quantitative analysis of ATV in plasma (32). The Bradford method for protein assay was applied to samples of the microsomes and BSA standard solutions as described above and amount of ATV was expressed in nmol/mg protein.

DATA ANALYSIS
The results were reported as the mean ± SD of values from three independent experiments using different passages of cells and batches of pooled human liver microsomes. To determine the statistical significance of differences observed between control and treatment groups, an unpaired two tailed Student's t-test was computed using Graphpad Prism, version 3 (San Diego, California, USA). A p-value < 0.05 was considered significant.

RESULTS
The aqueous and methanolic extracts of SF prepared by hot water and methanol decoctions were obtained at yields of 25 % and 15 %, respectively, whilst that of the flavonol and triterpenoid glycoside fractions was 5 % and 2.3 % respectively. Figures 1 and 2 show that the isolated sutherlandiosides and sutherlandins, either alone or in combination were all likely present in the triterpenoid glycoside and flavonol glycoside fractions, respectively. In Figure 1 A, Sutherlandioside A eluted at a retention time of 16.6 minutes which corresponded to the presence of a group of unresolved peaks in the triterpenoid glycoside fraction. A similar observation was made for Sutherlandioside B, which eluted at 13.9 minutes (Figure 1 B). In Figure 1 C, a single peak represented the combination of Sutherlandiosides C and D. Nevertheless, the retention time of the peak which represents both of these compounds fell within the same cluster of peaks (13-17 minutes). A similar trend was observed in Figure  2 A, for the combination of Sutherlandins A and B, where one peak between 10 and 11 minutes was present which is within the retention time range of a cluster of peaks (9-11 minutes) in the flavonol glycoside fraction. Two unresolved peaks were evident in Figure 2 B for the combination of Sutherlandins C and D between 9 and 10 minutes which also eluted within the same retention times of the group of peaks in the flavonol glycoside fraction.
From Figure 3, it is evident that both the aqueous (A) and methanolic (B) extracts contain flavonol (peaks at 9-11 minutes) and triterpenoid (peaks at 13-17 minutes) glycosides present in the triterpenoid and flavonol glycoside fractions.
The aqueous extract (10 mg/ml) of SF reduced ATV accumulation (p < 0.05) by Caco-2 cells (Figure 4). On the other hand, no effect was exhibited in the presence of the methanolic extract. Figure 5 indicates that both aqueous (p < 0.01) and methanolic (p < 0.05) extracts (10 mg/ml) of SF significantly increased the amount of ATV present in human liver microsomes after co-incubation with these extracts.
Despite the effects of the aqueous extract of SF, the flavonol glycoside fraction (500 µg/ml) present therein had no effect on the amount of ATV present in Caco-2 cells (Figure 4) or human liver microsomes ( Figure 5).  The triterpenoid glycoside fraction significantly increased the accumulation of ATV (p < 0.001) by Caco-2 cells (Figure 4) but significantly reduced the ATV present (p < 0.001) in human liver microsomes ( Figure 5).
Overall, the purported constituents of SF, viz L-canavanine, D-pinitol and L-GABA had no effect on the amount of ATV present in Caco-2 cells (Figure 4) and human liver microsomes ( Figure 5) respectively, except for a significant reduction in ATV accumulation ( p < 0.01) in Caco-2 cells observed in the presence of D-pinitol (Figure 4).

DISCUSSION
The presence of the individual or combined sutherlandiosides and sutherlandins in the triterpenoid and flavonol glycoside fractions, as well as the aqueous and methanolic extracts of SF suggest that these phytochemical constituents of SF (24,25) may have contributed at least in part to any effects on the amount of ATV present after co-treatment of Caco-2 cells and human liver microsomes with ATV and each fraction and/or extract.  The accumulation of ATV by Caco-2 cells may have been curtailed by the aqueous extract of SF by reducing the activity of influx transporters of ATV or enhancing the activity of efflux transporters of ATV, such as P-gp, which are present in Caco-2 cells and in the human intestinal epithelium. Through one or both of these mechanisms, the extract may have had the potential to limit ATV absorption and thus ATV bioavailability. Brown et al. 2008 (33) found that a higher concentration of an aqueous extract (20 mg/ml) of SF than that used in this study (10 mg/ml)   Unlike ATV, nevirapine is however not a substrate of P-gp (34), which may possibly indicate that the aqueous extract of SF elicited its effect on drug absorption, in Caco-2 cells and possibly in the human intestinal epithelium too, via P-gp. Mills et al. 2005 (12) found that a 100 mg/ml aqueous extract of SF exhibited inhibition of efflux of a fluorescent probe via inhibition rather than activation of basal human P-gp ATPase activity, further indicating that the effect of the aqueous extract of SF on P-gp may be substrate-dependent. The aqueous and methanolic extracts inhibited ATV metabolism as evidenced by the increase in ATV after co-treatment with each extract in human liver microsomes, indicating elimination was decreased and the bioavailability of ATV may be increased by these extracts. It is hypothesized that the effect of the SF extracts on ATV metabolism may be CYP3A4-mediated since observations by others (12) demonstrated that 100 mg/ml aqueous and methanolic extracts of SF inhibited CYP3A4 activity by ≥ 50% and ATV is a known CYP3A4 substrate (20).
The observation that the flavonol glycoside fraction exhibited no effect on the amount of ATV in the Caco-2 cells and human liver microsomes indicates that the flavonol glycosides, such as Sutherlandins A − D present in SF (25) were not likely responsible for the effects demonstrated by the aqueous and methanolic extracts of SF. The transepithelial flux of cimetidine in Caco-2 cells was altered by 4 flavonol aglycones, but not by 3 out of 4 of the corresponding flavonol glycosides (100 µM) (35) and the flavonol aglycones, quercetin and kaempferol are known to inhibit CYP3A4mediated metabolism (36)(37)(38). It may thus be concluded that glycosylation of flavonols may limit the capability of these compounds to modulate transepithelial flux in Caco-2 cells and CYP3A4-mediated metabolism in microsomes. Since deglycosylation of flavonol glycosides may occur in the gastro-intestinal tract (39), before these reach the site of intestinal absorption and the liver, changes in the bioavailability of ATV which may be observed in vivo after concurrent administration of an aqueous decoction of SF may potentially be related to the flavonol glycosides present therein. However deglycosylation of flavonol-3-glycosides appeared to be impeded in cell-free extracts from human intestines and livers (40), thus it is not clear whether the effects of the SF flavonol glycosides on ATV absorption and metabolism would be significant in vivo. A flavonol glycoside fraction from SF plant material similarly demonstrated no effect on progesterone and pregnenelone metabolism in sheep adrenal microsomes (27).
The increase in accumulation of ATV by Caco-2 cells in the presence of the triterpenoid glycoside fraction alludes to the potential for the sutherlandiosides and/or sutherlandins present therein to have enhanced absorption and thus bioavailability of ATV, either by inhibition of efflux transporters of ATV, such as P-gp or increased activity of influx transporters of ATV, which are present in Caco-2 cells and human intestinal epithelium. Madgula et al. 2008 (41) found that Sutherlandioside B, the most abundant of the known triterpenoid glycosides (Sutherlandiosides) in SF, is a potential P-gp and MRP substrate. This triterpenoid glycoside and possibly one or more of the others, may therefore have exhibited competitive inhibition of P-gp and/or MRP2-mediated transport of ATV, thereby enhancing ATV accumulation. Reduced ATV levels in human liver microsomes after treatment with this fraction implies that ATV metabolism was increased and the bioavailability of ATV was therefore limited. Activation of CYP3A4 occurs via positive co-operativity (42,43), and since ATV is a CYP3A4 substrate (20), it is likely through this CYP450 isoenzyme that activation of ATV metabolism was elicited. Interestingly, Sutherlandioside B has been shown to inhibit CYP3A4 metabolism of a fluorescent substrate with an IC 50 of 20 µM (41) and a three-fold lower concentration of the triterpenoid glycoside fraction than that used in this study inhibited CYP450 17 α-hydroxylase and CYP450 21 hydroxylase involved in steroidogenesis in the adrenal gland (27). The discrepancies between the studies may, as previously suggested for the aqueous extract, be due to differences in the substrates utilised.
A clinical study in healthy human volunteers has shown that the bioavailability of a single dose of atazanavir was decreased by chronic dosing of tablets containing Sutherlandia frutescens (44). This indicates that SF may have the potential to reduce the efficacy of ATV during routine administration to patients in the clinical setting. These findings corroborate the effect of the aqueous extract and D-pinitol on ATV accumulation in the Caco-2 cells and the effect of the triterpenoid glycoside fraction on ATV metabolism in the human liver microsomes, indicating that these constituents and extract of SF may primarily be involved in the drug-drug interaction.

CONCLUSION
The aqueous extract and D-pinitol significantly reduced ATV accumulation in Caco-2 cells, an established model of human intestinal epithelial cells, implying a decrease in ATV absorption and bioavailability, whilst the opposite effect would be expected after treatment with the triterpenoid glycoside fraction, which increased ATV accumulation. Both the aqueous and methanolic extracts inhibited ATV metabolism in human liver microsomes, suggesting enhanced ATV bioavailability, whilst the converse was true for the triterpenoid glycoside fraction. The effects of the aqueous extract and D-pinitol in the Caco-2 cells and the triterpenoid glycoside fraction in the human liver microsomes were in agreement with the results of a clinical study in healthy human volunteers (44), therefore, it may be concluded that SF has the potential to alter the absorption and metabolism of ATV in the clinical setting.