Numerous proteins mediate the intracellular and extracellular binding and transport of fatty acids, which both facilitate the physiological functions of fatty acids while minimizing cellular exposure to high (and potentially toxic) concentrations of these compounds. Albumin is the most abundant fatty acid binding protein and is also the best characterized in terms of structure and function (Hamilton, 2004). Apart from functioning as the primary plasma transporter of unesterified fatty acids, serum albumin also binds a myriad of endogenous and exogenous compounds, including drugs with acidic or electro-negative features.

Like albumin, members of the intracellular fatty acid binding family of proteins (FABPs), which are among the most abundantly expressed cytosolic proteins, bind saturated and unsaturated long-chain fatty acids with high affinity (Veerkamp and Maatman, 1995; Zimmerman and Veerkamp, 2002; Chmurzyńska, 2006). Within the cell, FABPs solubilize and transport fatty acids and function more generally as lipid chaperones that coordinate the biological actions of fatty acids (Furuhashi and Hotamisligil, 2008). To date, nine human FABPs, each with a characteristic tissue distribution, have been identified (Chmurzyńska, 2006; Furuhashi and Hotamisligil, 2008). The intestinal FABP (IFABP) is expressed along the entire length of the small intestine, although most abundantly in the medial segment, and accounts for approximately 3% of enterocyte cytoplasmic protein (Vassileva et al., 2000; Agellon et al., 2002). Rat IFABP binds a single molecule of fatty acid in a slightly bent conformation (Sacchattini et al., 1989; Zhang et al., 1997). The carboxylate head group of the fatty acid is deeply buried in the ligand binding site, where it hydrogen-bonds with Arg106 (Hamilton, 2004). Although it is known that rat IFABP and liver FABP (LFABP) bind a number of acidic and neutral drugs with varying affinities (Velkov et al., 2005, 2007; Chuang et al., 2008), the binding of drugs to human IFABP has not been characterized.

In addition to its endogenous “binding” roles, albumin has been included as a constituent of microsomal incubations in experiments to generate kinetic parameters for drug and xenobiotic metabolism. In vitro approaches for the prediction of drug pharmacokinetic parameters in vivo [in vitro-in vivo extrapolation (IV-IVE)], particularly intrinsic clearance (CL_int) and hepatic clearance (CL_H), have attracted widespread interest in recent years (Houston, 1994; Iwatsubo et al., 1997; Obach et al., 1997; Miners et al., 2006). However, there is a bias toward underprediction of CL_int and CL_H for drugs metabolized by UDP-glucuronosyltransferase (UGT) and cytochrome P450 (P450), particularly when human liver microsomes (HLMs) are used as the enzyme source (Boase and Miners, 2002; Miners et al., 2004, 2006; Ito and Houston, 2005; Brown et al., 2007).

Recent studies have shown that this discrepancy arises to a large extent from the release of membrane long-chain unsaturated fatty acids that act as potent competitive inhibitors of several UGT and P450 enzymes, resulting in overestimation of the Michaelis constant (K_m) and hence underprediction of microsomal CL_int (Tsoutsikos et al., 2004; Rowland et al., 2007, 2008a,b). Addition of bovine serum albumin (BSA) or essentially fatty acid free human serum albumin (HSA-FAF) to HLM sequesters the inhibitory long-chain unsaturated fatty acids and improves prediction accuracy of the in vivo CL_H values for substrates of UGT1A9, UGT2B7, and CYP2C9 (Rowland et al., 2007, 2008a,b). For example, the K_m value for the glucuronidation of the UGT2B7 substrate zidovudine (AZT) by HLM was decreased by an order of magnitude in the presence of BSA, with a corresponding increase in CL_int.

However, many drugs and nondrug xenobiotics that are metabolized by these enzymes bind extensively to albumin. Characterization of the kinetics of metabolite formation of such drugs by incubations of HLM supplemented with albumin presents significant difficulties given the requirement to measure the low unbound concentration of substrate in microsomal incubations. Thus, the availability of a protein that binds drugs to a lesser extent than fatty acids would be a valuable development in IV-IVE. Here, we report for the first time the binding of acidic, basic, and neutral compounds (Fig. 1) to purified human IFABP and, for comparison, to BSA, including measurement of the binding affinities of the “model” acidic drugs phenytoin and torsemide to each protein. Furthermore, comparable enhancement of human liver microsomal AZT CL_int by IFABP and BSA was shown, indicating that IFABP may be substituted for BSA in microsomal kinetic studies.

Materials and Methods

Materials. Alamethicin (from Trichoderma viride), 1-anilino-8-naphthalene sulfonate (ANS), BSA (product no. A7906), caffeine, β-estradiol, frusemide (furosemide), lignocaine, naproxen (S-enantiomer), nortriptyline, propofol, propranolol, sulfinpyrazone, UDP-glucuronic acid (trisodium salt), AZT, and AZT β-d-glucuronide were purchased from Sigma-Aldrich (Sydney, NSW, Australia). Diazepam and torsemide were obtained from Hoffmann-La Roche (Basel, Switzerland) and Boehringer Mannheim International (Mannheim, Germany), respectively. Homo sapiens fatty acid binding protein 2 (intestine) TrueClone cDNA (reference sequence NM_000134.2) was purchased from Origene (Rockville, MD), and Tetra-His horseradish peroxidase-conjugate kit was from QIAGEN (Melbourne, VIC, Australia). Solvents and other reagents were of analytical reagent grade.

HLMs. Pooled HLMs were prepared by mixing equal protein amounts of microsomes from five human livers (H7, 44-year-old woman; H10, 67-year-old woman; H12, 66-year-old man; H29, 45-year-old man; and H40, 54-year-old woman) obtained from the human liver “bank” of the Department of Clinical Pharmacology (Flinders University, Adelaide, SA, Australia). Approval for the use of human liver tissue in xenobiotic metabolism studies was obtained from both the Clinical Investigation Committee of Flinders Medical Centre and from the donors' next of kin. HLMs were prepared by differential centrifugation as described by Bowalgaha et al. (2005).

Expression of Recombinant Human IFABP. cDNA encoding IFABP was polymerase chain reaction-amplified from the human IFABP cDNA. To facilitate the purification of the IFABP, six histidine residues were added to the C terminus of wild-type IFABP cDNA using the primers IFABP-His5, 5′-ATTAGGATCCAAATGAGTTTCTCCGGCAAGTAC-3′ and IFABP-His3, 5′-ATTATCTAGAGATCAGTGATGGTGATGGTGATGATCCTTTTTAAAGATCCTTTTGGCTTC-3′. The IFABP insert was ligated into the pCWori(+) bacterial expression plasmid. pCW-IFABP was transformed into DH5α Escherichia coli cells, and colonies were screened for the correct plasmid by restriction enzyme analysis. Plasmid DNA was purified, and the nucleotide sequence was confirmed on both strands by sequencing (ABI Prism 3100; Applied Biosystems, Foster City, CA). Overnight subcultures (4 ml) grown in Luria-Bertani broth with 100 mg/l ampicillin at 37°C were used to inoculate 400-ml cultures of modified Terrific broth containing 100 mg/l ampicillin. Cultures were grown at 37°C with shaking (180 rpm) for 3 h or until an optical density of approximately 0.7 at 600 nm was attained, at which time the temperature was reduced to 30°C and 1 mM isopropyl-β-d-thiogalactopyranoside was added. Cultures were grown at 30°C with shaking (150 rpm) for an additional 40 h. Cells were then harvested by three passages through a French press cell at a cell pressure of 8 to 11 kPa. IFABP was recovered in the soluble fraction after centrifugation at 45,000g for 90 min and then purified by chromatography on a pre-equilibrated Ni²⁺ affinity (Ni-NTA agarose) column according to the general procedure of Johnson et al. (2005). The concentration of purified IFABP protein was determined according to the Lowry procedure.

Immunoblotting of IFABP. Denatured purified protein (1 μg) was separated by SDS-polyacrylamide gel electrophoresis and then transferred onto nitrocellulose membranes. Membranes were subsequently washed and blocked in 1% (w/v) blocking reagent (QIAGEN) for 2 h at room temperature, then washed (5 min) with Tris-buffered saline and probed (120 min at room temperature) with a commercial anti-His antibody (Tetra-His horseradish peroxidase-conjugate; 1:1000 dilution). Before development, the membrane was washed in Tween 20. The BM chemiluminescence (Roche Diagnostics GmbH, Mannheim, Germany) blotting substrate was used for immunodetection. The membrane was exposed to Kodak (Rochester, NY) X-Omat films for 10 s, and films were processed manually using AGFA G153 developer and G354 fixer (Agfa Gevaert, Dubendorf, Switzerland).

AZT Glucuronidation Assay. Incubations, in a total volume of 0.2 ml, contained 0.1 M phosphate buffer (pH 7.4), 4 mM MgCl₂, 0.05 mg HLM, 0–2.5% BSA or IFABP, and 10–3000 μM AZT. HLMs were fully activated by the addition of the pore-forming polypeptide alamethicin (50 μg/mg protein) with incubation on ice for 30 min (Boase and Miners, 2002). After a 5-min preincubation at 37°C, reactions were initiated by the addition of 5 mM UDPGA. Incubations were performed at 37°C in a shaking water bath for 60 min. Reactions were terminated by the addition of 6 μl of perchloric acid (70%, v/v). Samples were subsequently centrifuged at 4000g for 10 min, and a 30-μl aliquot of the supernatant fraction was injected directly into the high-performance liquid chromatography (HPLC) column. AZT β-d-glucuronide formation was quantified by HPLC as described by Rowland et al. (2007).

Measurement of ANS Binding and Competitive Displacement by Ligands. The binding of ANS to IFABP and BSA was quantified by fluorescence spectroscopy using a modification of the method of Norris and Spector (2002). Spectrofluorometer (PerkinElmer 300; PerkinElmer Life and Analytical Sciences, Waltham, MA) excitation and emission wavelengths were set at 370 and 475 nm, respectively, with respective excitation and emission slit widths of 2.5 and 5 nm. Incubations to measure ANS binding to IFABP and BSA (3-ml total volume) were performed at 20°C in a quartz cuvette (10-mm path length) and contained 0.1 M phosphate buffer (pH 7.4) and 500 nM IFABP or BSA. The incubation sample was titrated with 14 10-μl aliquots of ANS (100 μM), resulting in an ANS concentration range of 500 nM to 10 μM. ANS fluorescence was not detected in the absence of protein. Incubations to measure the displacement of ANS by arachidonic acid contained 0.1 M phosphate buffer (pH 7.4), 500 nM IFABP, and 1 μM ANS. The incubation samples were titrated with eight 10-μl aliquots of arachidonic acid, resulting in an ANS concentration range of 25 to 1000 nM. In drug displacement studies, incubation samples containing 0.1 M phosphate buffer (pH 7.4), 500 μg/ml IFABP or BSA, and 50 μM ANS were titrated with competitor ligand (phenytoin and torsemide; 100 nM to 50 μM).

Measurement of Drug Binding to IFABP and BSA. The binding of drugs to IFABP and BSA was measured using Microcon centrifuge filter devices (Millipore Corporation, Billerica, MA), which comprise reservoir and filtrate compartments separated by a 3-kDa cellulose membrane. The cellulose membranes were conditioned with 500 μl of phosphate buffer (0.1 M, pH 7.4). Incubation samples (total volume, 110 μl) contained the drug of interest, 0.1 M phosphate buffer (pH 7.4), and protein (IFABP or BSA, 0–0.5%). Samples were incubated in a shaking water bath at 37°C for 120 min, after which time a 100-μl aliquot was added to the reservoir compartment, and the device was centrifuged at 14,000g for 2 min. Under these conditions, less than 15% of the sample passed from the reservoir compartment to the filtrate compartment. A 10-μl aliquot was collected from each compartment, and protein was precipitated with ice-cold methanol containing 4% glacial acetic acid (10–20 μl) or ice-cold methanol (20 μl) alone in the case of lignocaine and propofol. Samples were cooled on ice, centrifuged at 4000g for 10 min at 4°C, and aliquots of the supernatant fraction (5 μl) were analyzed by HPLC. Drug and protein concentrations investigated are shown in Supplemental Table 1. Fraction unbound (f_u) was calculated as 1 – ([drug_reservoir] – [drug_filtrate])/[drug_reservoir].

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Chemical structures of compounds used in IFABP and BSA binding studies.

It should be noted that conditioning of the membrane before use overcame the nonspecific binding and poor recovery reported in certain previous studies that have measured protein binding using centrifugation filter devices. Recovery, assessed by mass balance against an unfiltered sample, was ≥95% for all the drugs. In addition, the f_u values for the binding of AZT, lignocaine, phenytoin, and propofol to BSA measured here (see under Results) were in close agreement with those determined previously in this laboratory using equilibrium dialysis (Rowland et al., 2007, 2008a,b; J. O. Miners and D. J. Elliot, unpublished data).

Quantification of Drug Binding to IFABP and BSA. Drug present in the filtrate from binding experiments was separated on a Waters (Sydney, NSW, Australia) NovaPak C18 analytical column (3.9 × 150 mm, 4 μm) using a mobile phase comprising 10 mM triethylamine (pH adjusted to 2.5 with perchloric acid; mobile phase A) and acetonitrile (mobile phase B) at a flow rate of 1 ml/min (see Supplemental Table 2 for proportions). Column eluant was monitored at the optimal wavelength (determined by spectroscopic analysis) for each drug (Supplemental Table 2). The concentration of drug in ultrafiltration samples was determined by comparison of peak areas with those of authentic standards using calibration curves that spanned the concentration ranges used in binding studies.

Data Analysis. All the data points represent the mean of duplicate estimates (<10% variance). Kinetic constants for AZT glucuronidation by HLM, in the absence and presence of BSA and IFABP, were obtained by fitting experimental data to the Michaelis-Menten equation. The stoichiometry of ANS, arachidonic acid, phenytoin, and torsemide binding to IFABP and BSA was determined by Scatchard ([S]_bound/[P] versus [S]_bound/[S]_total) analysis. When K_d values for binding of substrate molecules at all the binding sites are equivalent, a plot of [S]_bound/[P] versus [S]_bound/[S]_total gives a straight line with an x-intercept of n, where n is the number of substrate molecules bound to each protein molecule (Moller and Denicola, 2002). K_d values for ANS and arachidonic acid binding to IFABP and BSA, as well as torsemide binding to IFABP, were determined using a single binding mode equation, whereas K_d values for phenytoin binding to IFABP and BSA, as well as torsemide binding to BSA, were determined using a two-binding mode equation: where ΔF is the percentage change in fluorescence on addition of ANS or competitive ligand, C_n is the capacity of the nth binding mode, K_dn is the dissociation constant for the nth binding mode, [S] is the concentration of added substrate, and n is the number of binding modes.

Results

IFABP Expression and ANS and Arachidonic Acid Binding. Expression of IFABP was shown by immunoblotting the purified protein. A single band corresponding to the molecular mass of His-tagged IFABP (approximately 16 kDa) was detected by chemiluminescence after immunoblotting with a Tetra-His antibody (Fig. 2). No other bands were observed. Coomassie staining of the SDS-polyacrylamide gel electrophoresis gel before transfer onto nitrocellulose was also consistent with the presence of a single protein with an approximate molecular mass of 16 kDa.

For characterization purposes, the binding capacity of IFABP was confirmed with the fluorescent probe ANS and by measuring the displacement of ANS by arachidonic acid. Titration with ANS resulted in a saturable increase in fluorescence in the presence of IFABP, which was not observed in the absence of protein (Fig. 3A). The binding of ANS was well described by the single binding site equation. The derived dissociation constant (2.9 μM) is comparable with K_d values reported previously for the unmodified rat protein (6.9 μM, Kirk et al., 1996; 3.6–7.4 μM, Velkov et al., 2005). Scatchard analysis indicated that each molecule of IFABP bound one molecule of ANS. Addition of arachidonic acid to samples containing IFABP and ANS resulted in a decrease in ANS fluorescence (Fig. 3B), which was best fit to the single binding site equation. The derived K_d value for arachidonic acid binding to IFABP was 122 nM, which is within the range reported for the binding of saturated and unsaturated long-chain fatty acids to rat IFABP (Richieri et al., 1994; Velkov et al., 2005).

Comparative Binding of Drugs to IFABP and BSA. Preliminary experiments were performed to assess the binding of phenytoin (an acid), propofol (a neutral compound), and propranolol (a base) to IFABP and, for comparison, to BSA. Binding was measured at three drug concentrations that spanned the known K_m for each substrate at each of four protein concentrations, ranging from 0.05% (0.5 mg/ml) to 0.5% (5 mg/ml). Values of f_u for phenytoin, propofol, and propranolol in the presence of IFABP and BSA are shown in Table 1. As observed previously (Rowland et al., 2008a), phenytoin bound appreciably to BSA; the binding was independent of ligand concentration but increased with increasing BSA concentration. Phenytoin also bound appreciably to IFABP. Again, binding was independent of ligand concentration but increased with increasing protein (IFABP) concentration. Whereas the binding of propofol to IFABP was negligible, <10% at the protein concentrations used (0.05–0.5%), binding to BSA was appreciable. Binding to BSA did not vary with propofol concentration but increased with increasing albumin concentration. Binding of propranolol to IFABP was minor (≤13%) at all the protein concentrations (0.05–0.5%). Likewise, the binding of propranolol to BSA was minor at the low protein concentrations (0.05 and 0.25%), although binding was >20% at the highest BSA concentration (0.5%).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Western blot of purified His-tagged human intestinal fatty acid binding protein. The sample contained 1 μg of purified protein.

Based on the data for phenytoin, propofol, and propranolol, the binding of 10 further compounds to IFABP and BSA (0.5%) was assessed. The compounds (Fig. 1) were classified as acids (frusemide, naproxen, sulfinpyrazone, and torsemide), bases (lignocaine and nortriptyline), or neutrals (caffeine, diazepam, β-estradiol, and AZT) based on the charge state at pH 7.4. Binding measurements were performed at three or four ligand concentrations that spanned the known K_m for the major human liver microsomal metabolic pathway of each substrate (and which included 100 μM) at a protein concentration of 0.5% (5 mg/ml).

The binding of all the acidic compounds to both proteins exceeded 10% over the ligand concentration ranges investigated (Table 2). The f_u was independent of drug concentration but generally varied between compounds and proteins. Frusemide, naproxen, sulfinpyrazone, and torsemide bound extensively to BSA; mean f_u values ranged from 0.05 to 0.24. Phenytoin binding to both proteins and sulfinpyrazone binding to IFABP were moderate, with mean f_u values of approximately 0.4 to 0.6. Whereas frusemide, naproxen, and torsemide bound strongly to BSA, mean f_u values for the binding of these compounds to IFABP were approximately 0.8. The binding of the basic compounds lignocaine, nortriptyline, and propranolol to IFABP was negligible (<10%). The binding of lignocaine to BSA was also minor, but nortriptyline and propranolol bound appreciably to BSA. As observed for the acidic compounds, binding of nortriptyline and propranolol to BSA was concentration-independent but varied for each drug (Table 2). Differing patterns of binding were observed for the neutral compounds: diazepam and β-estradiol bound appreciably to both proteins, although binding was much higher with BSA; binding of AZT and caffeine was minor with both BSA and IFABP; and, as observed in the screening experiment, propofol bound appreciably only to BSA.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

A, plots for the binding of ANS to IFABP. B, percentage of ANS displacement versus ligand concentration ([S]) and Scatchard plots for the binding of arachidonic acid to IFABP. Points show experimentally determined values, whereas curves are from model fitting.

Binding Affinities of Phenytoin and Torsemide to IFABP and BSA. The binding affinities of phenytoin and torsemide to IFABP and BSA were determined by ANS (10 μM) displacement at a protein concentration of 500 μg/ml (0.5% w/v). Experimental data for the binding of phenytoin to IFABP and BSA and torsemide to BSA were best described by a two-binding mode equation (Table 3; Fig. 4). In contrast, data for the binding of torsemide to IFABP were best fit to the single binding mode equation (Fig. 4).

Effect of IFABP on AZT Glucuronidation by HLM. As indicated under Materials and Methods, binding of AZT to BSA, IFABP, and HLM was accounted for in the calculation of kinetic constants. The conversion of AZT to AZT β-d-glucuronide by HLM, in the absence and presence of IFABP and BSA (0.5–2.5%), was well described by the Michaelis-Menten equation (Fig. 5). Derived K_m and V_max values for AZT glucuronidation by HLM in the absence of added protein and in the presence of BSA (0.5, 1.5, and 2.5%) were similar to previous reports from this laboratory (Boase and Miners, 2002; Rowland et al., 2007). Addition of BSA and IFABP (0.5, 1.5, and 2.5%) to incubations increased the rate of AZT glucuronidation by decreasing the K_m for this pathway without an appreciable effect on V_max (Table 4; Fig. 5). K_m values for AZT glucuronidation in the presence of 0.5 and 1.5% IFABP were approximately 2- to 3-fold higher than those observed in the presence of the same concentration of BSA. However, K_m values determined in the presence of 2.5% of each protein were similar (Table 4; Fig. 5).

Discussion

This study reports for the first time the binding, both extent and concentration dependence, of model acidic, basic, and neutral drugs to purified recombinant human IFABP and, for comparison, to BSA. Acidic drugs and two neutral compounds bound to IFABP to a minor or moderate extent (based on f_u values), but binding was consistently lower than to BSA. The utility of IFABP as an alternative fatty acid sequestrant to BSA in incubations of HLM, and hence its potential application in IV-IVE, was confirmed with the glucuronidated drug AZT, although the content of IFABP required to produce a 10-fold reduction in the K_m for AZT glucuronidation in incubations of HLM was higher than for BSA.

The comparative binding of drugs to IFABP and BSA was assessed as f_u at three or four concentrations (always including 100 μM) that spanned the known K_m for the principal route of metabolism of each compound. After screening studies with phenytoin, propofol, and propranolol, which indicated that acidic drugs were likely to bind most extensively to IFABP, studies (at 0.5% w/v protein) were conducted with an additional four acids, two bases, and three neutrals. Like propranolol, the two other bases (lignocaine and nortriptyline) bound negligibly to IFABP, despite moderate binding of nortriptyline to BSA. Of the four neutral compounds, only diazepam (mean f_u, 0.80) and β-estradiol (mean f_u, 0.56) bound appreciably to IFABP. Binding of the neutrals to IFABP was lower than to BSA.

As with the neutral compounds, the binding of acidic drugs to BSA (f_u range, 0.04–0.49) was higher than to IFABP (f_u range, 0.41–0.85). Mean f_u values of frusemide, naproxen, sulfinpyrazone, and torsemide were 4- to 10-fold higher with IFABP. Binding affinities and capacities measured for phenytoin and torsemide at 0.5% w/v protein were in broad agreement with these observations. Based on f_u values, phenytoin binds “moderately” to both BSA and IFABP, whereas torsemide binds much more avidly to BSA (Table 2). Binding of torsemide to BSA occurs at two sites, with high total capacity compared with the single IFABP site (Table 3). In contrast, K_d values and capacities for the binding of phenytoin to BSA and IFABP were of a similar order (Table 3). The observation of two modes for the binding of phenytoin to IFABP is consistent with the docking of ibuprofen and bezafibrate into the binding cavity of the rat IFABP X-ray crystal structure, which suggests that, unlike fatty acids, some acidic drugs may bind to IFABP in two orientations (Velkov et al., 2005). It should be noted that the molecular masses of IFABP (≈15 kDa) and albumin (≈66 kDa) differ. Thus, the drug binding capacity of BSA substantially exceeds that of IFABP on a molar basis.

The human IFABP drug binding data reported here are in agreement with two recent studies with rat IFABP. Velkov et al. (2005) found that ibuprofen and bezafibrate bound strongly to rat IFABP, whereas binding of the neutral drugs nitrazepam and diltiazem was weak or absent. More recently, the same group characterized the binding of a larger group of compounds, mostly organic acids (Velkov et al., 2007). Highest binding to rat IFABP was generally observed with compounds possessing a carboxylic acid group, although binding affinities varied by an order of magnitude. In addition to drugs, rat IFABP is also known to bind butylated hydroxytoluene and phthalate esters (Kanda et al., 1990).

Although the primary ligands for IFABP are fatty acids, our work and that of Velkov et al. (2007) show that the presence of a carboxylic acid function is not a requirement for binding. Only two of our study drugs, naproxen and frusemide, contain the carboxylate group. Sulfinpyrazone and β-estradiol (a phenol essentially uncharged at pH 7.4) exhibited the lowest f_u values. The acidity of sulfinpyrazone arises from the presence of a C-H bond α to two carbonyl groups, whereas phenytoin and torsemide contain hydantoin and sulfonylurea features, respectively. The binding of bulky multiring structures (e.g., β-estradiol and sulfinpyrazone; Fig. 1) is consistent with the known volume of the ligand binding domain of rat IFABP (≈850 ų; Hamilton, 2004) and with a flexible ligand entry portal (Hodsdon and Cistola, 1997). Binding data for a broader range of substrates will be necessary to generate quantitative structure-activity relationships, particularly the comparative importance of polar and hydrophobic features for drug binding to IFABP.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Percentage of ANS displacement versus ligand concentration ([S]) and Scatchard plots for the binding of phenytoin and torsemide to IFABP and BSA. Points show experimentally determined values, whereas curves are from model fitting.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Eadie-Hofstee (V versus V/[S]) plots for AZT glucuronidation by HLM in the absence and presence of IFABP and BSA (0.5, 1.5, and 2.5%). Points show experimentally determined values, whereas curves are from model fitting.

The pharmacological significance of the binding of acidic and neutral drugs to IFABP, which together with LFABP comprises up to 6% of the cytosolic protein of enterocytes, is unknown. However, it has been proposed that binding to FABPs enhances the transcellular delivery of drugs to the enterocyte basolateral membrane, with subsequent uptake into the intracellular space (Velkov et al., 2007). Thus, IFABP may assist the absorption of acidic drugs and xenobiotics.

IV-IVE is a major research interest of this laboratory. Recent work has shown that the addition of BSA or HSA-FAF to incubations of HLM markedly increases the prediction accuracy of in vivo CL_int and CL_H for substrates of UGT1A9, UGT2B7, and CYP2C9 (Rowland et al., 2007, 2008a,b) and the prediction of drug interactions arising from inhibition of UGT2B7 (Rowland et al., 2006; Uchaipichat et al., 2006). The “albumin” effect arises from sequestration of long-chain unsaturated fatty acids released from the microsomal membrane during the course of an incubation. In the absence of BSA or HSA-FAF these compounds act as potent competitive inhibitors of several P450 and UGT enzymes. Thus, experiments conducted in the presence of BSA or HSA-FAF provide a true estimate of K_m and K_i. However, the limitations of the use of albumin were shown in a recent study involving the UGT1A9 substrate frusemide. Whereas the K_m and CL_int values generated in the absence of BSA were almost certainly overestimated, the extensive albumin binding of frusemide (>97%) precluded in vitro kinetic studies in the presence of BSA (Kerdpin et al., 2008). In fact, most acidic and many neutral drugs bind extensively to albumin at concentrations added to incubations of HLM, which is apparent from data presented in Table 2.

The lower binding observed here with IFABP suggested that this protein may represent an alternative to BSA for enhancing the prediction accuracy of IV-IVE based on human liver microsomal kinetic data. To confirm this, the kinetics of AZT glucuronidation by HLM was characterized in the absence and presence of BSA and IFABP (0.5–2.5% w/v). Both proteins decreased the K_m for AZT glucuronidation without affecting V_max, as reported previously with BSA and HSA-FAF (Rowland et al., 2007). However, a higher content of IFABP (2.5% versus 1%; Table 4; Rowland et al., 2007) was necessary to reduce the K_m for AZT glucuronidation by an order of magnitude. The increased requirement for IFABP is consistent with the known fatty acid binding capacity of albumin; whereas each molecule of IFABP binds only a single fatty acid molecule, albumin contains at least seven binding sites for long-chain unsaturated fatty acids (Hamilton, 2004). However, this difference is partially compensated by the 4.4-fold lower molecular mass of IFABP.

The K_ms generated for microsomal AZT glucuronidation in the presence of 2.5% IFABP (74 μM) and BSA (69 μM) are in accord with that reported with human hepatocytes (87 μM; Engtrakul et al., 2005) as the enzyme source and thus seem to represent true hepato-cellular K_m. Based on IV-IVE scaling factors for AZT given in Uchaipichat et al. (2006), predicted in vivo hepatic clearances for AZT glucuronidation were approximately 27 l/h. Although this still represents an underestimation of known in vivo CL_H (82 l/h), the prediction bias is substantially less than that normally associated with the scaling of CL_int derived using HLM and human hepatocytes as the enzyme sources (Boase and Miners, 2002; Miners et al., 2006; Brown et al., 2007).

In summary, it has been shown that human IFABP has the capacity to bind acidic and neutral drugs to a minor or moderate extent. Binding is not dependent on the presence of the carboxylate group. At equivalent protein concentrations, binding to BSA exceeds that of IFABP. Based on AZT glucuronidation kinetic data, IFABP is able to substitute for BSA as a fatty acid sequestrant in incubations of HLM, although there is a requirement for a higher content of IFABP in incubations. The utility of IFABP in IV-IVE is being investigated further with drugs that bind extensively to BSA.