Analysis of the Binding of Aripiprazole to Human Serum Albumin: The Importance of a Chloro-Group in the Chemical Structure.

Aripiprazole (ARP), a quinolinone derivative, is an atypical antipsychotic drug that is used in the treatment of schizophrenia. ARP has an extensive distribution and more than 99% of the ARP and dehydro-ARP, the main active metabolite, is bound to plasma proteins. However, information regarding the protein binding of ARP is limited. In this study, we report on a systematic study of the protein binding of ARP. The interaction of ARP and structurally related compounds with human serum albumin (HSA) was examined using equilibrium dialysis, circular dichroism (CD) spectroscopy, fluorescent probe displacement, and an X-ray crystallographic analysis. The binding affinities (nK) for ARP and its main metabolite, dehydro-ARP with HSA were found to be significantly higher than other structurally related compounds. The results of equilibrium dialysis experiments and CD spectral data indicated that the chloro-group linked to the phenylpiperazine ring in the ARP molecule plays a major role in the binding of these ligands to HSA. Furthermore, fluorescent probe displacement results indicated that ARP appears to bind at the site II pocket in subdomain III. A detailed CD spectral analysis suggests that the chloro-group linked to the phenylpiperazine ring may control the geometry of the ARP molecule when binding in the site II binding pocket. X-ray crystallographic analysis of the ARP-HSA complex revealed that the distance between the chlorine atom at the 3-positon of dichlorophenyl-piperazine on ARP and the sulfur atom of Cys392 in HSA was 3.4-3.6 Å. A similar halogen bond interaction has also been observed in the HSA structure complexed with diazepam, which also contains a chloro-group. Thus, the mechanism responsible for the binding of ARP to a protein elucidated here should be relevant for assessing the pharmacokinetics and pharmacodynamics of ARP in various clinical situations and for designing new drugs.

Introduction

Aripiprazole (ARP), 7-(4-(4-(2,3-dichlorophenyl)-1-piperazinyl)buthoxy)-3,4-dihydro-2-(1H)-quinolinone (Figure ), is a novel antipsychotic agent with a different pharmacological profile from other antipsychotics.[1−3] ARP is a dopamine–serotonin system stabilizer with potent partial agonist activity with respect to dopamine D2 and 5-HT1A receptors, but with antagonist activity for 5-HT2A receptors.[1] ARP is metabolized by two human cytochrome P450 (CYP) isozymes, CYP 2D6 and CYP 3A4, to dehydro-ARP (Figure ) being the main metabolite.[4] Interestingly, dehydro-ARP, the main metabolite of ARP, contributes to about 40% of the ARP AUC in plasma.[4] It is also known that when used at therapeutic drug concentrations, more than 99% of ARP and dehydro-ARP are bound to plasma proteins.[4] Spectroscopic and molecular modeling studies have been carried out to identify its binding site on human serum albumin (HSA), the most abundant protein in human plasma.[5,6] However, in spite of the fact that ARP is widely used to treat various psychiatric diseases, only limited information is available concerning the binding of ARP to plasma proteins.

Figure 1

Chemical structures of ARP, its metabolite (dehydro-ARP), and structurally related compounds (deschloro-, 2-deschloro-, and 3-deschloro-ARP).

The direct measurement of the concentration of a free drug can provide meaningful information, provided a reference therapeutic range for the free concentration has been established. In the case of drugs that are highly bound to plasma proteins, the unbound fraction is likely to show wide variations among patients, thus making them suitable candidates for monitoring their free concentrations. In the present study, we examined the protein binding of ARP. We initially determined the binding parameters of ARP as well as its metabolites (dehydro-ARP) and structurally related compounds (deschloro-, 2-deschloro- and 3-deschloro-ARP; Figure ) to HSA, using an equilibrium dialysis technique. The interactions of ARP, its metabolite, and structurally related compounds to HSA were examined by CD. In addition, the binding site involved in ARP binding was also identified using site-specific fluorescent probe displacement. Finally, an X-ray crystallographic analysis of an ARP–HSA complex was performed to evaluate the binding mode at the molecular level.

Results

Determination of the Binding Parameters of ARP, Its Metabolite, and Structurally Related Compounds with HSA

The binding parameters of ARP, its metabolite, and related compounds with HSA were determined by an equilibrium dialysis technique. The number of high-affinity binding sites and the respective binding affinity constant, n and K values, obtained by fitting experimental binding data to the binding equation are summarized in Table . ARP and its metabolite, dehydro-ARP, showed markedly greater binding affinities (nK), compared with other compounds. In contrast, the nK values were substantially smaller for three other deschloro-ARP derivatives (deschloro-, 2-deschloro- and 3-deschloro-ARP) that were studied. In particular, deschloro-ARP had the smallest nK value. Furthermore, it was noteworthy that the position of the chlorine group in the molecule had a significant effect on the nK values, as shown in the cases of 2-deschloro-ARP and 3-deschloro-ARP. These results indicate that the chlorine group plays an important role in the binding of ARP to albumin.

Table 1

Binding Parameters of ARP, Its Metabolite, and Structurally Related Compounds to HSA at pH 7.4 and 25 °Ca

	K (×106 M–1)	n	nK (×106 M–1)
ARP	7.51 ± 1.36	0.80 ± 0.05	5.94 ± 0.79
dehydro-ARP	3.19 ± 0.55	0.98 ± 0.10	3.08 ± 0.27
deschloro-ARP	0.17 ± 0.01	0.63 ± 0.03	0.10 ± 0.01
2-deschloro-ARP	0.65 ± 0.22	0.97 ± 0.16	0.61 ± 0.11
3-deschloro-ARP	1.55 ± 0.09	0.84 ± 0.02	1.30 ± 0.07

The results are means ± SD (N = 3).

CD Spectroscopic Study of the Interaction of ARP and Structurally Related Compounds with HSA

The extrinsic CD spectra of the ligands bound to HSA are shown at longer wavelength than 250 nm because the influence of the intrinsic CD of HSA on the induced CD spectra was relatively large. The induced CD spectra of ARP bound to HSA are shown in Figure . A positive Cotton effect was observed at around 265 nm, and a negative Cotton effect was observed at around 300 nm. It should be noted that the magnitude of these two peaks, with positive and negative signs, increased with increasing ligand concentration. In addition, an isosbestic point was observed at around 292 nm, suggesting that ARP formed a 1:1 complex with HSA. The sign and magnitude of the induced Cotton effects for ARP-related compounds bound to HSA were different and varied with their chemical structures. Contrary to the induced CD for the ARP–HSA complex, a peak with a negative sign for deschloro-ARP was observed at around 265 nm. The CD characteristics for 2-deschloro-ARP were similar to those of ARP, where a positive Cotton effect at around 260 nm and a negative Cotton effect at around 295 nm were induced by the binding to HSA. A small Cotton effect with a negative sign was found for HSA with bound 3-deschloro-ARP (Figure ).

Figure 2

CD spectra of ARP–HSA systems at pH 7.4 and 25 °C. The concentration of HSA was 40 μM, and ARP concentrations were 5 (a), 10 (b), 20 (c), and 40 μM (d).

Figure 3

CD spectra for the interaction of ARP derivatives with HSA at pH 7.4 and 25 °C. The concentrations of HSA and ligands were 40 and 20 μM, respectively. (a–e) ARP-, dehydro-ARP-, deschloro-ARP-, 2-deschloro-ARP-, and 3-deschloro-ARP-HSA systems.

The effect of fatty acids on the induced CD of the ARP–HSA complexes was also investigated. Octanoic acid and myristic acid were selected as typical examples of medium- and long-chain saturated fatty acids, respectively. Octanoic acid caused a decrease in the induced CD ellipticities of the ARP–HSA complex, as shown in Figure , suggesting that octanoic acid may displace the ARP molecule that is bound to HSA. However, myristic acid showed a different effect, with the positive Cotton effect at around 265 nm was inverted to become a negative Cotton effect. This observation can be attributed to allosteric conformational changes in the HSA molecule that are induced upon the addition of myristic acid.

Figure 4

Effects of fatty acids on CD spectra for the interactions of ARP (A) and deschloro-ARP (B) with HSA at pH 7.4 and 25 °C. The concentrations of HSA, ligands, and fatty acids were 40, 20, and 120 μM, respectively. (a,d) Spectra without fatty acid. ((b,e) and (c,f)) Spectra with octanoic acid and myristic acid.

Solution pH (pH 6.5–9.0) had little effect on the induced ellipticities of the ARP–HSA complex, indicating that the involvement of an N–B transition for the conformational change found here can be excluded. It is also possible that the protonation state of the ligand molecule was not critical for the change in the induced CD spectra (data not shown). On the other hand, in the case of deschloro-ARP, the enhancement in CD intensities in the presence of fatty acids was observed.

Identification of ARP Binding Site on HSA

To identify the binding site of ARP, we performed site-specific marker displacement experiments using fluorescent probes, warfarin (a site I probe), and dansylsarcosine (a site II probe), as reported by Sudlow et al.[7,8] The addition of ARP caused a decrease in the fluorescence of dansylsarcosine, whereas only a slight change was observed in the fluorescence of warfarin (Figure ). This result suggests that ARP binds to site II on HSA. Although we did not examine the binding of ARP to mutant HSA using equilibrium dialysis in this study, Tyr411 appears to generally play an important role in ARP binding, similar to most site II drugs. In fact, the induced CD spectrum for the binding of ARP to Y411A was quite small, compared with that for native HSA (Figure S1).

Figure 5

Effects of ARP on the fluorescence of warfarin and dansylsarcosine when bound to HSA. The concentration of HSA, warfarin, and dansylsarcosine was 2 μM. Closed and open circles are the effects on the fluorescence of warfarin and dansylsarcosine, respectively.

Crystal Structure of the ARP–HSA Complex

The crystal structure of ARP–HSA complex was determined at a resolution of 2.28 Å and refined to final R and Rfree factors of 20.7 and 25.5%, respectively. This is the first crystal structure of ARP bound to a biomacromolecule. Data collection and structure refinement statistics are summarized in Table . A clear electron density map for the ARP molecule in the crystal structure of the ARP–HSA complex can be observed, in which one ARP molecule is bound to subdomain IIIA of the HSA structure (Figure a,b). It is well known that there are three ligand binding sites, referred to as drug site II, FA sites 3 and 4 in subdomain IIIA of HSA.[7,9,10] The ARP binding region appears to include both drug site II and FA site 4 in subdomain IIIA of HSA (Figure a; stereo-view). The interface area between HSA and ARP is 513 Å2, as calculated by the PISA program.[11] The binding region for the dichlorophenyl-piperazine group of the ARP structure coincides with drug site II, which is surrounded by Leu387, Ile388, Asn391, Cys392, Phe395, Phe403, Tyr411, Leu430, Val433, Gly434, Cys438, Ala449, Leu453, Arg485, and Ser489 (Figure a,b). Furthermore, the distance between the chlorine atom at the 3-positon of the dichlorophenyl-piperazine group on ARP and the sulfur atom of Cys392 in the HSA molecule is 3.4–3.6 Å. This result suggests a halogen bond formation, which is observed for the interaction between the sulfur atom of HSA and the chlorine atom of diazepam[12,13] (Figure b). The binding sites of the dihydro-quinolin group and the butoxy linker of the ARP structure coincide with FA site 4, which is surrounded by Tyr411, Lys414, Val415, Val418, Leu423, Ser427, Leu430, Leu460, Phe488, and Leu491 (Figure a,c).

Table 2

Data-Collection and Structure Refinement Statisticsa

Data set		ARP–HSA complex
Data-collection
	source	Photon factory BL-17A
	wavelength (Å)	0.9800
	Space group	P21
	Unit-Cell Parameters
	length (Å)	a = 59.2, b = 184.9, c = 59.3
	angle (deg)	β = 106.7
	Resolution range (Å)	50.0–2.28 (2.42–2.28)
	No. of observed reflections	361 307 (58 618)
	No. of unique reflections	54 292 (8793)
	Multiplicity	6.7 (6.7)
	Completeness (%)	97.7 (99.1)
	Rmerge (%)b	8.3 (46.8)
	⟨I/σ(I)⟩	14.2 (3.3)
Refinement
	Resolution (Å)	48.4–2.28 (2.32–2.28)
	Reflection used	54 291 (2793)
	Rwork (%)c	20.7 (27.3)
	Rfree (%)d	25.5 (35.2)
	Completeness (%)	97.8 (94.0)
	No. of nonhydrogen atoms	8920
	protein	8466
	ligands	80
	solvent	374
	r.m.s.d. from ideality
	bond length (Å)	0.002
	bond angle (deg)	0.424
	Average B-factor	51.1
	protein	51.7
	ligands	44.9
	solvent	38.7
	Ramachandran Plot
	favored region (%)	93.63
	allowed region (%)	6.28
	outlier region (%)	0.09
	Clashscore	7.0
	Twin operators	(l, −k, h)
	Twin fractions	0.48

Values in parentheses denote the highest resolution shell.

Rmerge = 100 × ∑∑|I(hkl) – ⟨I(hkl)⟩|/∑∑I(hkl), where ⟨I(hkl)⟩ is the mean value of I(hkl).

Rwork = 100 × ∑||Fo| – |Fc||/∑|Fo|, where Fo and Fc the observed and calculated structure factors, respectively.

Rfree is calculated as for Rwork, but for the test set comprising 5% reflections not used in refinement.

Figure 6

Overall structure of the ARP–HSA complex. (a) Overall structures of the ARP–HSA complex. The HSA molecule is shown as a cartoon representation, and the sub-domain structures are colored in magenta (IA), pink (IB), green (IIA), palegreen (IIB), blue (IIIA), and cyan (IIIB). The ARP molecule (yellow) is shown as a CPK (Corey–Pauling–Koltun) representation. (b) 2mFo–DFc electron density map of ARP is shown as a blue mesh control at 1.5σ.

Figure 7

ARP binding at subdomain IIIA in HSA. (a) Stereo-view of the binding of ARP at subdomain IIIA. The ARP molecule (yellow) is shown as a ball-and-stick representation. (b) Comparison of the binding position of ARP (yellow) with that of diazepam (magenta stick, PDB: 2BXF(14)) at drug site II. Chlorine atoms are colored in green and hydrogen bonds are shown as orange dashed lines. (c) Comparison of the position of binding of ARP (yellow) with myristic acid (orange stick, PDB: 1BJ5(9)) at FA site 4.

Discussion

Antidepressants are one of the fastest growing classes of drugs because of a variety of interrelated factors. Mental health issues, including depression and mood disorders, have become far more frequent in recent decades than in the past.[15] The rapid development of new antidepressants with minimal side effects is a desirable effort. In fact, new generations of antidepressants with improved side effect profiles, such as ARP, are becoming increasingly available for the treatment of emotional disorders.[4,15]

ARP is rapidly absorbed after oral administration (bioavailability: 87%). This drug is extensively metabolized in the liver by CYP 450 3A4 and 2D6 enzyme systems. The major metabolite, dehydro-ARP, is pharmacologically active with an activity that is about 40% equivalent to the parent drug. Although more than 99% of ARP and its major active metabolite are bound to plasma proteins at therapeutic drug concentrations, it is extensively distributed in the body.[4] However, our understanding of the nature of the protein binding, including the actual binding site remains controversial.

ARP is sometimes co-administered with a variety of drugs in cases where patients have two or more concurrent diseases. For example, patients with hypertension normally will be given an angiotensin receptor blocker and diuretics drugs are known to strongly bind to plasma proteins. This led us to predict that drug–drug interactions of ARP might involve the CYP enzyme system and/or binding to plasma proteins. The focus of this study was on elucidating the protein-binding mechanism of ARP, with the aim of precisely predicting the pharmacokinetics and therapeutic effects of ARP. Equilibrium dialysis results indicated that ARP preferentially binds to HSA, as expected from the results of a previous report.[4] The parameters for the binding of ARP to HSA obtained here (n = 0.80, K = 7.51 × 106 M–1) indicated that ARP forms a strong 1:1 complex with HSA. Pharmacokinetic data also indicated that dehydro-ARP, the main metabolite, also has a high affinity for binding to HSA. Compounds that are structurally related to ARP have different binding affinities that appear to be related to the presence and position of a chlorine group on the phenylpiperazine part of the molecule. The high affinities of ARP and dehydro-ARP for HSA that were obtained using equilibrium dialysis are in reasonable agreement with the bound fraction (more than 99%) observed in human plasma.[4] Yan et al., using a fluorescence quenching method, reported a much lower association constant for ARP (K = 2.03 × 104 M–1 at 25 °C and 0.57 × 104 M–1 at 37 °C).[5] Although the reason for this discrepancy is not entirely clear at present, it appears that the fluorescence quenching method may underestimate the affinity of ARP for HSA.

That chlorine is important in the binding of ARP to HSA was also supported by a CD spectroscopic study. ARP and dehydro-ARP, both possess two chlorine atoms on the phenylpiperazine moiety, and these groups induce a relatively large positive Cotton effect at around 265 nm. In contrast, deschloro-ARP induced a negative Cotton effect at around 265 nm. The CD spectral characteristics for 2-deschloro-ARP were similar to those for ARP, in that a positive Cotton effect was found at around 260 nm and a negative Cotton effect was found at around 295 nm, which were induced by binding to HSA. A small negative Cotton effect was observed for 3-deschloro-ARP.

In ethanol, ARP shows a maximum absorbance at around 255 nm based on the tetrahydoquinolin ring and a shoulder peak around 295 nm due to the dichlorophenyl-piperazine ring.[16] The positive Cotton effect at around 265 nm can be explained by the electronic transition of the tetrahydoquinolin moiety and the small negative peak can be attributed to the electronic transition of the dichlorophenyl-piperazine moiety. The positive Cotton effect at around 265 nm may depend on π–π interactions of the tetrahydoquinolin ring with Tyr411 and Phe488, amino acid residues that contain an aromatic ring, based on the findings of the X-ray crystallographic analysis (see Figure ). In fact, the CD spectra of ARP bound to the single-mutant Y411A was markedly different from that for the wild-type HSA (Figure S1). This spectral change can be attributed to the change in the geometry of the ARP molecule through the weakening of π–π interactions between ARP and HSA. Interestingly, negative Cotton effects at around 260 nm were observed for deschloro-ARP and 3-deschloro-ARP. These dramatic findings may be also explained by conformational changes in ligand molecules within the binding pocket through the disappearance of a halogen bond between a chlorine atom of ARP and the sulfur atom of Cys392 in HSA. It is also possible that a change in the dihedral angle of the C–N bond in the ARP molecule accompanying the microenvironmental change may also affect the CD spectra (different conformation from ARP molecule). The differences in CD spectra observed here may reflect the molecular structures of the ligands. In addition, extrinsic Cotton effects depend on the spatial relationship between the asymmetric center and the perturbed chromophore of the ligand as indicated in a previous report.[17−19]

The binding affinity for 3-deschloro-ARP was found to be significantly higher than that for 2-deschloro-ARP. The difference in CD spectral characteristics between 2-deschloro-ARP and 3-deschloro-ARP can be explained by differences in the geometry of the ligand within the binding pocket in HSA. Equilibrium dialysis and CD spectral data indicate that the chlorine atom of ARP plays an important role in the binding of the molecule to HSA. This finding is further supported by the CD spectroscopic study of the effects of fatty acids on the binding of ARP to HSA.

To collect further information regarding ARP–HSA binding, the effects of fatty acids on the induced CD on the ARP–HSA system were examined. Octanoic acid and myristic acid were used in this study because it is known that they bind to HSA (i.e., the affinities, binding sites, or effects on ligand binding and HSA conformation)[9,20−24] and that these characteristics have been studied extensively. The presence of octanoic acid decreased the induced CD ellipticities of the ARP–HSA complex, suggesting the existence of a displacement mechanism. On the other hand, the presence of myristic acid resulted in the reversal of the sign of the Cotton effect for the ARP–HSA complex. This observation can be due to allosteric conformational changes in the HSA molecule induced by the binding of myristic acid. In fact, no change in the free concentration of ARP was found upon the addition of myristic acid under these experimental conditions (Figure S2). Interestingly, the pattern of induced CD spectra for deschloro-ARP was generally similar to that of the ARP–HSA–myristic acid system. Therefore, the reversed sign of the Cotton effects for ARP–HSA–myristic acid system can most likely be explained by the geometry of ARP within the binding pocket being reoriented following that of deschloro-ARP through an allosteric conformational change of HSA that is induced by the binding of myristic acid. Moreover, the increase of the observed ellipticities of deschloro-ARP upon the myristic acid can be owing to the enhanced rigidity of deschloro-ARP molecule accompanying microenvironmental changes of site II binding pocket generated by myristic acid binding.

Myristic acid was found to have more than one high affinity binding site and several low affinity binding sites on HSA.[20] The fact that myristic acid did not increase the free concentration of ARP (Figure S2) suggests that the high affinity sites for ARP and myristic acid do not overlap.[25] This hypothesis can be clarified by an X-ray crystallographic analysis of the ARP–HSA–myristic acid complex and is currently under way in our laboratory.

Fluorescent probe experiments were carried out to identify the binding site of ARP on the HSA molecule. It has been suggested that ARP binds to site II in the HSA molecule. Indeed, equilibrium dialysis data indicated that ARP forms a 1:1 complex with HSA and it competitively inhibits the binding of diazepam (a site II ligand) but not warfarin (a site I ligand) (Figure S3). Er et al. classified ARP as a site II drug by monitoring the color change in a fluorescence dye cocktail–HSA system after adding ARP.[6] Meanwhile, Yan et al., using a spectral and molecular modeling approach, proposed that ARP binds to site I.[5] In spite of this discrepancy, our crystallographic data for the ARP–HSA complex clearly indicated that ARP is located in subdomain IIIA (site II). On the basis of crystallographic analysis of the ARP–HSA complex, the binding interaction region of the dichlorophenyl-piperazine group in the ARP structure is surrounded by amino acids including Asn391, Cys392, Phe395, Tyr411, and Leu430. It should be noted that the distance between the chlorine atom at the 3-positon of the dichlorophenyl-piperazine group and the sulfur atom of Cys392 in HSA is 3.4–3.6 Å. This observation further supports the conclusion that the chloro-group in this molecule plays an important role in the binding of ARP to HSA. The effect of such a halogen bond has also been observed in the interaction between HSA and diazepam, which also contains a chloro-group.[12,13]

Conclusion

In this study, the binding of ARP, a quinolinone derivative and an atypical antipsychotic drug, to HSA was extensively investigated. ARP binds to subdomain IIIA (site II) of HSA with a high affinity. As in the case of diazepam,[12,26] Tyr411 is a key amino acid that is involved in the high affinity binding of ARP. On the other hand, the chlorine atom on the phenylpiperazine ring of ARP plays an important role in the binding of ARP to HSA. It is known that endogenous substances, including fatty acids and uremic toxins as well as exogenous substances such as drugs also bind to site II.[27] Therefore, it is highly possible for drug displacement to occur under diseased conditions such as chronic renal failure. Such a displacement on ARP which is 99% bound to plasma protein would undoubtedly lead to clinically significant consequences. The findings regarding the binding of ARP to plasma proteins obtained in this study will allow further clarifications of the pharmacokinetics and pharmacodynamics of ARP in various clinical situations and will lead to the design of new drugs.

Experimental Section

Materials

Recombinant human albumin was a gift from Nipro Co. (Shiga, Japan). Warfarin and dansylsarcosine were purchased from Sigma Chemical Company (St. Louis, MO). Using a modification of the procedure reported by Chen,[28] albumin was defatted with activated charcoal at 0 °C in an acidic solution, deionized, and then freeze-dried. ARP was purchased from Tokyo Chemical Industry Co Ltd. (Tokyo, Japan). Dehydro-ARP, the main active metabolite of ARP, was synthesized according to the method of Zeidan et al.[29] 2-Deschloro-ARP was purchased from Toronto Research Chemicals (Toronto, Canada). 3-Deschloro-ARP was synthesized according to the published procedure.[30] Deschloro-ARP was synthesized following a method reported by Banno et al.[31] Octanoic acid and myristic acid were purchased from Tokyo Chemical Industry Co Ltd. (Tokyo, Japan). All other chemicals were purchased from commercial sources and were of the highest grade available. About 67 mM sodium phosphate buffer (pH 7.4) was used in the equilibrium dialysis and spectroscopic (CD and fluorescence) experiments. Stock solutions of ARP, its derivatives, warfarin, and dansylsarcosine (2.5 mM) were prepared in methanol.

Equilibrium Dialysis

Equilibrium dialysis experiments were carried out using 2 mL Sanko plastic dialysis cells (Fukuoka, Japan). The same volume of samples and buffer solutions (0.5 mL) were inserted into the cell compartments which were separated by Visking cellulose membranes and shaken gently at 25 °C for 12 h. After equilibrium was achieved, ARP concentrations in the buffer (Cf; unbound drug concentration) and albumin compartments (Cb+f; sum of bound and unbound drug concentrations) were determined by HPLC. Bound concentration (Cb) was calculated by subtracting Cf from Cb+f. The experimental data were fit to the following equation using GraphPad PRISM Version 4 (GraphPad Software, Inc, CA, U.S.A.).where r is the number of moles of drugs bound per mole protein. Pt is the protein concentration. K and n are the association constant and the number of binding sites.

CD Measurements

CD measurements were carried out using a Jasco model J-720 spectropolarimeter (Tokyo, Japan), using a 10 mm cell at 25 °C. Induced ellipticity was defined as the ellipticity (in degrees) after subtracting the ellipticity of HSA alone from that of a drug–HSA mixture within the same wavelength region. No induced ellipticities for fatty acid–HSA mixtures were observed at wavelengths longer than 240 nm.

Fluorescent Probe Displacement

Warfarin and dansylsarcosine were used as site-specific fluorescent probes for site I and site II of HSA, respectively. Steady-state fluorescence measurements were made on a Hitachi F-2500 fluorescence spectrophotometer (Hitachi, Ibaragi). All measurements were performed at 25 °C using 5 nm excitation and emission bandwidths. The excitation wavelengths for warfarin and dansylsarcosine were 320 and 350 nm, respectively. The probe–HSA ratio was maintained at 1:1. Fluorescence spectra from the probe–HSA systems were monitored after the addition of ARP, its metabolite and structurally related compounds as displacers. The percentage of displacement of the probe was determined using the following equation.where F1 and F2 represent the fluorescence of the probe plus HSA without or with displacers, respectively. Although HSA concentrations were maintained at low levels to minimize inner-filter effects, the fluorescence intensities were further corrected according to the method described by Lakowicz.[32]

Crystallization of the ARP–HSA Complex

Preparation of the HSA solution for crystallization was performed as described previously.[33] A stock solution of 50 mM ARP for crystallization was prepared by dissolving in dimethyl sulfoxide (DMSO). The ARP–HSA complex was formed by mixing the HSA solution and the ARP stock solution at a 1:5 HSA–ARP molar ratio in 50 mM potassium phosphate pH 7.0 and 10% (v/v) DMSO, after which it was incubated at 20 °C overnight. After incubation, the excess unbound and insoluble ARP that was deposited was removed by centrifugation (20 400g for 1 h at 20 °C), and the ARP–HSA complex was washed with 50 mM potassium phosphate pH 7.0 by performing four cycles of dilution and concentration using a Vivaspin 500 (MWCO 10 000, Sartorius) centrifugal concentrator. The ARP–HSA complex solution was finally concentrated to an HSA concentration of 1.4 mM. Co-crystallization of the ARP–HSA complex was performed using the hanging–drop vapor diffusion method, and the ARP–HSA crystals for the X-ray analysis were obtained by multiple rounds of streak-seeding with droplets prepared by mixing 2 μL of the ARP–HSA complex solution and 2 μL of the reservoir solution containing 32% (w/v) polyethylene glycol 3350 and 50 mM potassium phosphate pH 7.0 at 4 °C and pre-equilibrated for 1–3 days.

Data-Collection, Structure Determination and Refinement

The ARP–HSA complex crystals were directly frozen in liquid nitrogen. Synchrotron experiments were performed at Photon Factory BL-17A (Tsukuba, Japan). Diffraction data sets were collected at −173 °C using a Pilatus3 S 6M detector and the data sets were processed and scaled using XDS.[34] The initial phase of the ARP–HSA complex structure was determined by the molecular replacement method using MOLREP(35) from the CCP4 program suite,[36] with the coordinate (PDB: 5YOQ(37)) serving as the search model. Further model building was performed with COOT.[38] Structure refinements including the twin refinement with twin law (l, −k, h) and the refinement of atomic displacement parameters by the translation, liberation, and screw (TLS) method was performed with phenix.refine.[39] Twin law and TLS groups were determined by using phenix.xtriage and phenix.find_tls_groups from the PHENIX package,[40] respectively. The stereochemical quality of the final structure was evaluated by MolProbity.[41] All molecular graphics were prepared using PyMOL.[42] The atomic coordinates of the HSA–ARP complex have been deposited in the Protein Data Bank under the accession code 6A7P.