Hepatic phenotype of liver fatty acid binding protein gene-ablated mice

Materials.

Protease activity in liver homogenates was inhibited with a cocktail for mammalian tissues purchased from Sigma-Aldrich (St. Louis, MO). Protein content was determined by use of the Protein Assay Dye Reagent Concentrate from Bio-Rad Laboratories (Richmond, CA). Silica gel G thin-layer chromatography plates were from Analtech (Newark, DE), whereas standards for cholesterol, cholesterol ester, phospholipid, and fatty acids were from Nu-Chek Prep (Elysian, MN) or Sigma Chemical (St. Louis, MO). NBD-cholesterol was obtained from Molecular Probes (Eugene, OR). 6-Dansyl-cholestanol was synthesized as described earlier (Huang H, McIntosh AL, Atshaves BP, Kier AB, Schroeder F, unpublished observations). Free cholesterol benzophenone (FCBP) was generously provided by Dr. Thomas A. Spencer (Department of Chemistry, Dartmouth College, Hanover, NH). Custom synthesis of radiolabeled (³H at position C-3) free cholesterol benzophenone (³H-FCBP) was performed by Moravek Biochemicals (Brea, CA). Cholesterol (> 98% pure) was obtained from Avanti Polar Lipids (Alabaster, AL), and 2-hydroxypropyl-β-cyclodextrin (HPCD) was purchased from Sigma-Aldrich (St. Louis, MO). Rabbit polyclonal antiserum against recombinant rat L-FABP was produced as described (6). All reagents and solvents used were of the highest available grade.

Animals.

The experimental protocols for the use of laboratory animals were approved by the University Lab Animal Care Committee and met American Association for Accreditation of Laboratory Animal Care guidelines. N2 generation L-FABP^−/− C57Bl/6 mice, obtained as previously documented (56), were further backcrossed to C57Bl/6 mice from Charles River Laboratories (Wilmington, MA) obtained through the National Cancer Institute (Frederick Cancer Research and Development Center, Frederick, MD) to obtain the N4 (94% homogeneity), N6 (98% homogeneity), and N10 (99.9% homogeneity) backcross generations. Sex- and age- (≥8 mo) matched wild-type littermates of the same generation were used as controls. L-FABP expression was monitored as described earlier (56). Mice were allowed free access to low-fat (5% of calories from fat) standard rodent chow (Teklad Rodent Diet W8604) from Harlan Teklad (Madison, WI). The mice were housed in a temperature-controlled (25°C) facility on a 12-h light-dark cycle. Prior to euthanasia, mice were fasted for 12 h and euthanized by cervical dislocation. The livers were removed, divided into small portions, flash-frozen with dry ice, and stored at −80°C for subsequent analysis.

Lipid analysis.

Mouse liver was homogenized, followed by lipid extraction, resolution into individual lipid classes, and quantification as described earlier (5, 56). Protein was quantified by the Bradford protein assay (Bio-Rad Laboratories) (13). Commercially available kits were used according to the manufacturer's protocols to quantify serum lipids: total cholesterol, Wako no. 439-17501; free cholesterol, Wako no. 435-35801; nonesterified fatty acid, Wako no. 999-34691/991-34891; phospholipid, Wako no. 433-36201; triacylglycerol, Wako no. 461-08992/461-09092 (Wako Diagnostics, Richmond, VA).

Protein purification.

To perform cholesterol-binding assays, recombinant L-FABP and SCP-2 (used as a positive control) were isolated and purified as described earlier (15, 57). Purity of L-FABP and SCP-2 was determined by SDS-PAGE, followed by silver-staining, and correct molecular mass was confirmed by matrix-assisted laser desorption time-of-flight (MALDI-TOF) mass spectrometry and by electrospray ionization (ESI) mass spectrometry (Laboratory for Biological Mass Spectrometry, Department of Chemistry, Texas A&M University) as described (15, 57). The concentration of L-FABP and SCP-2 stock solutions was determined by amino acid analysis (Protein Chemistry Laboratory, Department of Biochemistry and Biophysics, Texas A&M University) as described earlier (29, 57).

NBD-cholesterol and dansyl-cholesterol binding.

The binding constants of NBD-cholesterol to L-FABP and SCP-2 were obtained by titrating a 2-ml sample of L-FABP (50 nM) or SCP-2 (50 nM) in PBS (pH 7.4) with small increments of NBD-cholesterol as described earlier (9, 57, 93, 98). Fluorescence intensities were corrected for signals from NBD-cholesterol and proteins in buffer. The binding curves were constructed by plotting NBD-cholesterol fluorescence intensity increase (F − F₀) vs. ligand concentration ([NBD-cholesterol]; brackets denote concentration), with F being fluorescence intensity of NBD-cholesterol in the presence of proteins, and F₀ being NBD-cholesterol fluorescence intensity in buffer. F_max, the maximum fluorescence intensity, was obtained from curve fitting and the dissociation constant K_d was calculated from the slope of double reciprocal plots: 1/(1 − (F − F₀)/F_max) vs. [NBD-cholesterol]/(F − F₀)/F_max, as described previously (9, 57, 93, 98).

Dansyl-cholesterol is highly fluorescent both in buffer and upon binding to proteins; therefore, dansyl-cholesterol binding affinity was examined by using a displacement assay described previously (38, 98) instead of direct binding assay. The displacement of NBD-cholesterol from L-FABP and SCP-2 by dansyl-cholesterol was determined by incubating L-FABP (50 nM in PBS) or SCP-2 (50 nM in PBS) with NBD-cholesterol (30 nM) for 8 min to obtain maximal fluorescence, followed by titration with increasing amounts of dansyl-cholesterol. Fluorescence intensities were corrected for signals from dansyl-cholesterol and proteins in buffer. From the displacement curve, the K_i value was calculated as described (38, 98). NBD-cholesterol fluorescence was recorded by utilizing a Cary Eclipse Fluorescence Spectrophotometer (Varian, Palo Alto, CA) at excitation wavelength = 465 nm and emission wavelength = 535 nm. All experiments were carried out at 24°C by utilizing a circulating water bath.

Binding and cross-linking with a photoactivatable cholesterol.

Cholesterol binding to recombinant L-FABP was determined by covalent attachment (i.e., cross-linking) of a photoactivatable radiolabeled cholesterol derivative (FCBP) basically as described (26, 57). Briefly, ³H-FCBP was cross-linked to L-FABP as detailed earlier for SCP-2 (57). Crosslinked ³H-FCBP/L-FABP unimers were separated from potential higher order multimers by SDS-PAGE, the ³H-FCBP/L-FABP was blotted onto nitrocellulose and probed with a polyclonal anti-L-FABP antibody, and ³H-FCBP/L-FABP was counted for ³H by liquid scintillation counting as described (57).

Mass spectral analysis of L-FABP cross-linked with photactivatable cholesterol.

The nonradioactive photoactivatable cholesterol derivative FCBP was cross-linked to L-FABP followed by mass spectrometric analysis essentially as described earlier for SCP-2 (57). Briefly, an Agilent (Santa Clara, CA) 1200 capillary HPLC system was interfaced to an API QSTAR Pulsar Hybrid QTOF mass spectrometer (Applied Biosystems/MDS Sciex, Framingham, MA) equipped with an ESI source. The L-FABP control and the FCBP-modified L-FABP samples were diluted to 0.05 and 0.1 mg/ml, respectively, in 0.1% formic acid. Liquid chromatography (LC) separation was achieved by using a Phenomenex C4 microbore column (150 mm × 0.50 mm) at flow rate of 10 μl/min. The proteins were eluted by using a gradient of 0.1% formic acid (solution A) vs. 0.1% formic acid in 85% acetonitrile (solution B). The gradient timetable was as follows: 2% (solution B) for 5 min, 2–30% (solution B) for 3 min, 30–60% (solution B) for 44 min, 60–95% (solution B) for 8 min, followed by holding the gradient at 95% (solution B) for 5 min, for a total run time of 65 min. The MS data were acquired in positive ion mode (500–2,000 Da) by using spray voltage of +4,900 V. BioAnalyst software (Applied Biosystems) was used for spectral deconvolution. A mass range of m/z 650–1,400 was used for deconvolution, and the output range was 8,000–30,000 Da by using a step mass of 0.1 Da and signal-to-noise threshold of 5.

Circular dichroism analysis of cholesterol binding to L-FABP.

Protein secondary structure of L-FABP in the presence or absence of cholesterol was determined by circular dichroism (CD) spectropolarimetry utilizing a J-710 spectropolarimeter (JASCO, Easton, MD) as described previously (36, 38, 57, 78) with the following modifications. The samples for CD analysis contained L-FABP at a final concentration of 4 μM in 10 mM potassium phosphate (pH 7.4). Because of its low aqueous solubility a stock solution of 1 mM cholesterol was prepared in 10 mM potassium phosphate (pH 7.4), 20% glycerol, 4.5% (32 mM) HPCD, 1 mM EDTA. For CD analysis of L-FABP in the presence of cholesterol, an aliquot of 1 mM cholesterol in the HPCD buffer was added to L-FABP at a final cholesterol concentration of 50 μM. L-FABP ± cholesterol was placed in the CD cuvette and incubated at 20°C for 10 min, and the sample was scanned from 185 to 260 nm under the following conditions: step resolution 1 nm, bandwidth 2 nm, sensitivity 10 m°, scan rate 50 nm/min, time constant 1 s. For each measurement, 10 scans were averaged, smoothed, background subtracted, and converted to mean residue molar ellipticity, [θ] (°·cm²·dmol⁻¹). Secondary structure analysis and percentage composition of α-helices, β-sheets, turns, and unordered structures for L-FABP ± cholesterol were determined as described (36, 38, 57). The change in percentage secondary structure was determined by subtracting the percentage secondary structure of L-FABP in HPCD buffer only from the percentage secondary structure of L-FABP in HPCD buffer + 50 μM cholesterol.

Isothermal titration calorimetry analysis of cholesterol binding to L-FABP.

Thermodynamic parameters of cholesterol binding to L-FABP or SCP-2 were determined by isothermal titration microcalorimetry utilizing an iTC₂₀₀ microcalorimeter (MicroCal, Northampton, MA). Cholesterol (300 μM) was prepared in a buffer containing 20 mM Tris·HCl (pH 8.0), 20% glycerol, and 4.5% (32 mM) HPCD as described above. Cholesterol was titrated (2.0 μl/injection) into a solution of protein [20 μM in 20 mM Tris·HCl (pH 8.0)/20% glycerol] at 25°C (298 K), and the amount of heat released or absorbed was measured. Three sets of controls were performed for each experiment: buffer injected into buffer, cholesterol injected into buffer, and buffer injected into protein. Each experiment was repeated five times (n = 5), and the data were analyzed by utilizing MicroCal Origin v7.0383 as supplied with the iTC₂₀₀ microcalorimeter.

Statistics.

Statistical analysis was performed by ANOVA combined with the Newman-Keuls multiple-comparisons test (GraphPad Prism, San Diego, CA). Data are expressed as means ± SE with n = 5–10 for all analyses, and P is indicated as described in the figure and table legends. Graphical analysis was accomplished by use of SigmaPlot 2000 for Windows version 6.10 (SPSS, Chicago, IL).

Effect of backcrossing on liver total cholesterol in L-FABP-null mice.

Liver total cholesterol mass was significantly increased twofold (P < 0.001) in female N2 generation L-FABP-null mice (Fig. 1). This effect was maintained at higher genetic homogeneity as shown by backcrossing to the N10 generation (99.9% homogeneity) (Fig. 1). In contrast, liver total cholesterol levels were not significantly altered by L-FABP gene ablation in male mice at any backcross generation number examined (Fig. 1). Interestingly, hepatic total cholesterol levels were higher in male than female wild-type L-FABP^+/+ mice at all backcross generations (Fig. 1). However, L-FABP gene ablation increased liver total cholesterol in female mice to levels equivalent to or higher than those in male wild-type L-FABP^+/+ mice (Fig. 1).

Effect of backcrossing on liver unesterified (free) cholesterol in L-FABP-null mice.

To determine whether the increased level of total cholesterol in livers of female L-FABP-null mice was due to increased unesterified (free) cholesterol, the free cholesterol was resolved as described in experimental procedures. The mass of free cholesterol was significantly higher (P < 0.001) in female N2 generation L-FABP-null mice (Fig. 2). This effect did not disappear with higher genetic homogeneity as shown by backcrossing to the N10 generation which exhibits 99.9% homogeneity (Fig. 2). In contrast, hepatic free cholesterol levels were not altered by L-FABP gene ablation in male mice, regardless of L-FABP backcross generation number (Fig. 2). The lack of hepatic free cholesterol phenotype in livers of male L-FABP-null mice correlated with the observation that the hepatic free cholesterol content of liver from male wild-type L-FABP^+/+ mice was already higher than that of female wild-type L-FABP^+/+. In female L-FABP-null mice the hepatic free cholesterol was only increased to levels similar to those observed in male wild-type L-FABP^+/+ and L-FABP^−/− mice (Fig. 2).

Effect of backcrossing on liver esterified cholesterol in L-FABP-null mice.

To show whether increased hepatic total cholesterol in female L-FABP-null mice was also associated with increased esterified cholesterol, the esterified cholesterol was resolved as described in experimental procedures. Hepatic cholesteryl ester content was significantly higher (P < 0.001) in female N2 generation L-FABP-null mice, regardless of L-FABP backcross generation number (Fig. 3). In contrast, cholesteryl-ester levels were not increased in male L-FABP-null mice at any backcross generation examined (Fig. 3). In female L-FABP-null mice the level of hepatic cholesteryl esters was increased to levels as high or higher than observed in male wild-type L-FABP^+/+ and L-FABP^−/− mice (Fig. 3).

Effect of backcrossing on liver phospholipid, triacylglycerol, and unesterified fatty acid content in L-FABP-null mice.

To show whether the L-FABP gene ablation-induced increase in all forms of cholesterol in livers of female mice was specific to cholesterol, the hepatic content of phospholipid and triacyglycerol was determined as described in experimental procedures. Phospholipid mass was significantly higher (P < 0.05) in female N2 generation L-FABP-null mice as well as in all subsequent backcross generations examined (Fig. 4A). In contrast, hepatic phospholipid levels of male mice did not display any consistent trend in response to L-FABP gene ablation or backcross generation number (Fig. 4A). Interestingly, L-FABP gene ablation generally increased the hepatic mass of phospholipids in female mice to levels as high as or higher than observed in livers of male wild-type L-FABP^+/+ or L-FABP^−/− mice (Fig. 4A).

With regard to hepatic triacylglycerol mass, in female mice this level was not significantly increased by L-FABP gene ablation regardless of backcross generation number (Fig. 4B). In contrast, hepatic triacylglycerol mass was markedly decreased in male L-FABP-null mice at all backcross generations examined (Fig. 4B).

Finally, the hepatic mass of unesterified fatty acids did not display any consistent pattern in response to sex, L-FABP gene ablation, or backcross generation number (Fig. 4C).

Effect of backcross generation on serum lipid levels in L-FABP null mice.

Hepatic free cholesterol, cholesteryl ester, and total cholesterol levels were increased in female L-FABP-null mice at all backcross generations examined as described above. To determine whether ablation of the L-FABP gene resulted in changes in serum lipid levels, the serum concentrations of total cholesterol, free cholesterol, cholesteryl ester, phospholipid, triacylglycerol, and nonesterified fatty acid were measured. L-FABP-null male and female mice exhibited no or minor changes in serum levels of total cholesterol (Fig. 5A), free cholesterol (Fig. 5B), cholesteryl ester (Fig. 5C), phospholipid (Fig. 5D), or nonesterified fatty acid (Fig. 5F) at all backcross generations examined. By contrast to the results observed for hepatic triacylglycerol concentrations (Fig. 4B), serum triacylglycerol levels did not change in L-FABP-null male mice; however, L-FABP-null female mice exhibited approximately twofold higher (P < 0.01) triacylglycerol levels than their wild-type counterparts at any backcross generation examined (Fig. 5E).

Affinity of L-FABP for cholesterol: direct binding.

Although the above findings confirm that female L-FABP-null mice exhibited elevated hepatic cholesterol levels at all backcross generations examined, as indicated in the introduction, it is also important to demonstrate that L-FABP binds cholesterol with high affinity. However, the poor aqueous solubility of cholesterol and the requirement for separation of L-FABP-bound and unbound radiolabeled cholesterol limit the usefulness of standard radiolabeled cholesterol binding assays. Therefore, determination of L-FABP affinity for cholesterol was addressed by two different approaches that took advantage of the intrinsic fluorescence of fluorescent cholesterol analogs and did not require separation of L-FABP-bound from unbound ligand.

First, L-FABP's binding affinity for NBD-cholesterol was determined by a sensitive direct fluorescence binding assay as described in experimental procedures. NBD-cholesterol is a fluorescent cholesterol analog previously shown to be bound with high affinity by other cholesterol binding proteins (9, 30, 92, 93, 98). Titration of L-FABP with increasing concentration of NBD-cholesterol resulted in a binding curve approaching saturation (Fig. 6A). Further analysis of the binding curves as described in experimental procedures showed that L-FABP bound NBD-cholesterol with high affinity as shown by K_d = 24.5 ± 1.5 nM (mean ± SE, n = 3). As a control, NBD-cholesterol binding to SCP-2, a protein known to bind cholesterol with high affinity (57, 93), was also examined. SCP-2 also exhibited a binding curve approaching saturation (Fig. 6C). SCP-2 bound NBD-cholesterol with high affinity as indicated by K_d = 38.1 ± 4.4 nM (mean ± SE, n = 3). Thus the NBD-cholesterol binding assay showed that L-FABP bound cholesterol with high affinity similar to that observed for SCP-2.

Second, a fluorescent dansyl-cholesterol displacement assay was used to confirm the high affinity of L-FABP for cholesterol. Since dansyl-cholesterol is fluorescent in buffer and when bound to protein, the affinity of L-FABP for dansyl-cholesterol was determined by a ligand-displacement assay as described in experimental procedures. Titration displacement of NBD-cholesterol bound to L-FABP with dansyl-cholesterol resulted in a monophasic decrease in fluorescence intensity (Fig. 6B); analysis of the displacement data yielded a K_i for dansyl-cholesterol/L-FABP of 20.8 ± 3.7 nM (mean ± SE, n = 3). Likewise, dansyl-cholesterol displacement of NBD-cholesterol bound to SCP-2 also resulted in a monophasic decrease in NBD-cholesterol fluorescence intensity (Fig. 6D); analysis of the displacement data yielded a K_i for dansyl-cholesterol/SCP-2 of 60.7 ± 3.0 nM (mean ± SE, n = 3). Thus the dansyl-cholesterol displacement assay confirmed that L-FABP bound cholesterol with high affinity, similar to that observed for SCP-2.

Stoichiometry of L-FABP binding of cholesterol: direct binding and photo-cross-linking.

Since the ligand binding cavity of L-FABP is sufficiently large to accommodate two long-chain fatty acids (29, 74, 84, 85), it was important to determine whether L-FABP bound one or two cholesterol molecules. This issue was addressed by the following two approaches.

First, the NBD-cholesterol binding assay supported the presence of a single cholesterol binding site in L-FABP. A double reciprocal plot of the binding data was linear (Fig. 6A, inset), indicating L-FABP bound NBD-cholesterol at a single binding site. As a control, a double reciprocal plot of NBD-cholesterol binding to SCP-2, known to bind cholesterol at a single binding site (57, 93), was linear (Fig. 6C, inset), consistent with a single cholesterol binding site in SCP-2. Thus both L-FABP and SCP-2 bound NBD-cholesterol at a single binding site.

Second, L-FABP bound a photoactivatable cholesterol derivative (FCBP) at a single site. FCBP binding and covalent cross-linking to L-FABP were shown by mass spectrometry as described in experimental procedures. In the absence of added FCBP, LC-MS analysis of non-cross-linked L-FABP indicated a parent ion of molecular mass = 14,349.7 Da (Fig. 7A). ESI analysis of higher-order ion fragments confirmed the molecular mass of the unreacted L-FABP (Fig. 7B). L-FABP bound 1.0 mol of FCBP per mole of protein such that the molecular mass of the L-FABP/FCBP complex was 14,862.5 Da (Fig. 7C). ESI analysis of the L-FABP complex indicated the presence of both unreacted L-FABP as well higher order ion fragments containing one FCBP per L-FABP molecule (Fig. 7D). Quantification of ³H-FCBP bound to L-FABP confirmed the covalent attachment of one molecule of FCBP per molecule of L-FABP (data not shown).

Secondary structure of L-FABP in the presence of cholesterol.

Since we have shown that cholesterol interacts with and binds to L-FABP with high affinity and high specificity, we wanted to examine the effect of cholesterol binding on the secondary structure of L-FABP. Cholesterol binding to L-FABP decreased the percentage of both regular α-helix and distorted α-helix such that there was an overall decrease in α-helix of 4.2 ± 0.3% (P < 0.05) relative to that exhibited by L-FABP in HPCD buffer only (Table 1). Likewise, cholesterol binding to L-FABP caused an 8.2 ± 0.7% decrease (P < 0.001) in regular β-sheet and a 4.7 ± 0.4% decrease (P < 0.05) in total β-sheet (Table 1). Interestingly, cholesterol binding induced a 3.5 ± 0.3% increase (P < 0.05) in distorted β-sheet structure in L-FABP (Table 1). Therefore, the overall effect of cholesterol interaction with L-FABP was to decrease total α-helix and β-sheet while increasing turn and unordered structural elements by 6.7 ± 0.5% (P < 0.01) and 6.6 ± 0.6% (P < 0.01), respectively (Table 1).

Thermodynamic binding parameters of cholesterol with L-FABP or SCP-2.

To further establish that the high binding affinity of L-FABP for fluorescent cholesterol analogs was not simply due to presence of the fluorophore tags, isothermal titration calorimetry was used to directly measure the binding of cholesterol to L-FABP and compared with that of a known cholesterol binding protein, i.e., SCP-2. In addition to resolving binding affinity without the use of a fluorescent or photolabel tag, isothermal titration calorimetry resolved the enthalpic and entropic contributions to ligand binding. Under the conditions used in these experiments, both L-FABP and SCP-2 bound ∼1 mol of cholesterol per mole of protein with high (nM) affinity, 140 ± 10 and 32 ± 8 for L-FABP and SCP-2, respectively (Fig. 8, A and B). Cholesterol binding to both L-FABP and SCP-2 was energetically favorable, as shown by the negative Gibbs free energy changes (Fig. 8E). Cholesterol binding to L-FABP was significantly (P < 0.001) more favorable enthalpically than the binding of cholesterol to SCP-2 (Fig. 8C). However, the binding of cholesterol to SCP-2 was significantly (P < 0.001) more favorable entropically than the binding of cholesterol to L-FABP (Fig. 8D). The net result of these contributions to ligand binding was a more favorable Gibbs free energy change for the cholesterol-SCP-2 interaction than was observed for the cholesterol-L-FABP interaction (Fig. 8E).