Insight into intermolecular binding mechanism of apatinib mesylate and human alpha-1-acid glycoprotein: combined multi-spectroscopic approaches with in silico

Abstract Apatinib mesylate (APM), an oral tyrosine kinase inhibitor, has a good anti-tumor activity in the treatment of various cancers, particularly in advanced non-small cell lung cancer. In this study, the intermolecular binding mechanism between APM and human alpha-1-acid glycoprotein (HAG) was investigated by combining multi-spectroscopic approaches with in silico techniques. The findings revealed that APM gave rise to the fluorescence quenching of HAG by forming a ground-state complex between APM and HAG with a stoichiometric ratio of 1:1, and APM has a moderate affinity for HAG as the binding constant of APM and HAG of approximately 105 M−1, which was larger than the APM-HAG complex. The findings from thermodynamic parameter analysis indicated that the dominant driving forces for the formation of the APM-HAG complex were van der Waals forces, hydrogen bonding and hydrophobic interactions, which were also verified with site-probe studies and molecular docking. The findings from in silico study indicated that APM inserted into the opening of the hydrophobic cavity of HAG, leads to a slight conformational change in the HAG, which was verified by circular dichroism (CD) measurements, that was, the beta sheet level of HAG decreased. Additionally, the results of synchronous and 3D fluorescence spectroscopies confirmed the decline in hydrophobicity of the microenvironment around Trp and Tyr residues. Moreover, some common metal ions such as Cu2+, Mg2+, Fe3+, Ca2+, and Zn2+ could cause the alteration in the binding constant of APM with HAG, leading to the change in the efficacy of APM. It will be expected that these study findings are to provide useful information for further understanding pharmacokinetic and structural modifications of APM. Communicated by Ramaswamy H. Sarma


Introduction
Apatinib mesylate (APM, Figure 1a), an oral tyrosine kinase inhibitor, aims to inhibit tumor growth by selectively targeting vascular endothelial growth factor receptor 2 (VEGFR2) and inducing cell apoptosis (Geng & Li, 2015;Liu et al., 2017;Qiu et al., 2018).Numerous studies have demonstrated that APM has a good anti-tumor activity, thus being used in the treatment of non-small cell lung cancer (NSCLC), liver cancer, breast cancer, and colon cancer (Maroufi et al., 2020;Xie et al., 2021).In clinical cases, NSCLC accounted for about 85% of all lung cancer (Chen et al., 2016;Siegel et al., 2020;Zappa & Mousa, 2016).And, both single dose and combined use of APM have good therapeutic effects (Liang et al., 2020;Zhou et al., 2021).In addition, APM has also demonstrated good therapeutic effect and safety in advanced gastric cancer (Song et al., 2020).Accordingly, APM has a promising application for the treatment of various cancers.However, as far as we know, the binding of drugs to plasma proteins, particularly serum albumin (SA) and human alpha-1-acid glycoprotein (HAG), acts as a major player in its transport and distribution.In addition to its own toxic effects and resistance mechanisms, the effectiveness of a drug may also be limited by interactions with plasma proteins.Overly strong interaction is detrimental to the influence on the concentration of available drugs in the body, while moderate interaction with certain proteins has the advantage of stabilizing the drug and facilitating its transport to the target site (Chaves et al., 2015).Recently, the study of the intermolecular interaction mechanism between drugs and plasma proteins has aroused considerable interest (Alanazi et al., 2022;Almehizia et al., 2020;Alsaif et al., 2020;Moradi et al., 2018;Shamsi et al., 2020;Siddiqi et al., 2018).
HAG, as an important glycoprotein, is one of the plasma proteins, which is primarily produced in the liver.HAG consists of 183 amino acid residues and 5 N linked oligosaccharide chains with a molecular weight of 41 � 43 kDa, the content of glycosylation is about 45% of molecular weight), and the isoelectric point is about 2.8 � 3.8 (Hochepied et al., 2003;Weimer et al., 1950).Simultaneously, HAG is also a member of the immunocalcin family, which regulates immune and inflammatory responses (Logdberg & Wester, 2000).As an acute phase protein, the normal level of HAG in plasma is 0.6 � 1.2 mg/mL, but in patients with chronic or acute inflammation, the level is increased by a factor of 2 � 5 mg/mL (Ajmal et al., 2016).The exact function of HAG remains unclear.However, it has two important attributes including the binding with basic and neutral drugs and its immunomodulation, which functional diversity arises from the variability in attachment site of the oligosaccharide moiety (Ajmal et al., 2016).In other words, the function of HAG can be affected due to binding with drugs.The binding interaction between drugs and HAG has important significance.Structurally, HAG with three tryptophan residues  has a barrel-shaped hydrophobic cavity, which is the main binding region for various ligands (Figure 1b) (Almutairi et al., 2022;Tang et al., 2006).In recent years, more than 300 drugs have been reported to interact with HAG, most of which are basic drugs (Zsila et al., 2009).For example, many basic tinib drugs such as imatinib (Fitos et al., 2012), ceritinib (Wang et al., 2020a), brigatinib (Wang, Kou, et al., 2020), cabozantinib and tofacitinib (Ajmal et al., 2016) act with HAG have been reported successively.Additionally, it can be also found that the affinity of some basic drugs like olmutinib on HAG is greater than that on SA (Kou et al., 2021a(Kou et al., , 2023)).However, as far as we know, the intermolecular interaction between APM and HAG remains to be investigated.
Reportedly, stressful stimuli such as cancer and inflammatory diseases cause a several-fold increase in hepatic HAG secretion.The concentration of HAG in the body will increase several-fold compared to levels in normal conditions.In this clinical situation, the binding of drugs to transporter proteins may be affected, thus affecting the amount of drug reaching the target site.Considering that the concentration of HAG can be affected by diseases such as cancer, it is essential to study the interaction of the anti-cancer drug APM with such proteins.Consequently, this work was undertaken aiming to disclose the intermolecular interaction characteristics between APM and HAG in vitro at the molecular level and to provide theoretical support for further understanding its pharmacodynamic and pharmacokinetic studies.To achieve this goal, multi-spectroscopic techniques and in silico.It is certain that the results of this work will contribute to further understand its transport and distribution in vivo and provide insights into its stability and efficacy upon binding to HAG.

Reagents
HAG were sourced from Sigma Aldrich, United States.APM were sourced from Shenghong Biotechnology, China.8-Aniline-1-naphthalenesulfonic acid ammonium salt (ANS) was supplied by Tokyo Chemical Industry Co., Ltd.D -(þ) -Sucrose were provided by Adamas Reagent.All other reagents were of analytical grade.
The stock solution of HAG was prepared in Tris-HCl buffer solution (50 mM, pH 7.40).The stock solution of APM was dissolved in DMSO to obtain the stock solution.Distilled water is used for the preparation of the ANS and sucrose solutions.

UV-vis spectroscopy measurements
HAG (7.2 lM) was titrated with APM solution from 0 to 0.06 lM at 298 K in quartz cuvette.The UV spectrum of the APM-HAG system from 200 nm to 500 nm was documented on a Shimadzu UV-1601 spectrophotometer.The background was corrected with APM solution with the corresponding concentration.

Fluorescence spectral measurements
All fluorescence experiments were performed on an F97 Pro spectrofluorometer aiming to explore the binding information including the fluorescence quenching mechanism of HAG upon binding to APM, the affinity of APM on HAG, and the structural changes in HAG induced by APM.
In the steady-state fluorescence measurements, the excitation and emission slit widths were fixed at 5 nm and 10 nm, respectively, and the excitation wavelength was set at 285 nm.An equal amount of APM was gradually added to the HAG solution with thorough mixing.The fluorescence spectra of HAG solutions were recorded at 293, 298, 303 and 308 K, respectively.To eliminate the error caused by the inner filter effect, the raw fluorescence data is corrected using the following formula [28].
Here, F obs and F cor represent the observed value and corrected value of the fluorescence intensity, A ex and A em represent the absorbance of the APM-HAG system at k ex and k em , respectively.
In synchronous fluorescence experiments, the excitation and emission slit widths were set at 10 nm.The changes in the microenvironment around Tyr and Trp residues of HAG were investigated by fixing the difference between excitation wavelength (k ex ) and emission wavelength (k em ) at 15 nm and 60 nm, respectively.
In terms of 3D fluorescence experiments, setting the excitation wavelength to be changed in the range of 200À 350 nm at 2 nm intervals and the emission wavelength to be set at 200 -600 nm.The 3D fluorescence spectra of HAG (1.2 lM) were recorded with the APM concentration of 0, 2.4, 4.8 and 7.2 lM, respectively.

Circular dichroism spectral measurements
The far UV-CD spectra of HAG solutions (3.0 lM) that was prepared in a 25 mM phosphate buffer solution (pH ¼ 7.4) with and without APM in the 200-250 nm were measured on a JASCO J-815 Spectrophotometer.The slit width was set at 2 nm, sampled at 1 nm intervals and the background interference was subtracted with the corresponding concentration of APM solution.

Molecular docking
To better elucidate the intermolecular interaction characteristics of HAG with APM, molecular docking, one of in silico approaches (Abraham et al., 2015), was carried out with the software AutoDock 4.2.The initial geometry of HAG (PDB ID: 3KQ0) and APM were retrieved from the Protein Data Bank (https://www.rcsb.org/)and PubChem (https://pubchem.ncbi.nlm.nih.gov/),respectively.To acquire stable structures of receptor and ligand for use in molecular docking, the retrieved structures of HAG and APM were optimized.That was, the energy minimization of HAG was performed by using the amber 99SB-ILBN force field and the steepest descent algorithm (50000 steps) on GROMACS (Goodsell et al., 1996).The conformation of APM was initially optimized with the MM2 method with the aid of ChemBio3D Ultra 13.0 software and then further optimized on Gaussian 03 software using the DFT method at the base group of B3LYP/6-31 þ g (d, p).
In the docking process, the semi-flexible docking strategy, that was HAG as a rigid molecule and APM as a flexible molecule, was carried out in a grid box of 70 � 70 � 70 grid points with a grid interval of 0.375 Å, which could cover the whole simulation system.The docking pose of the APM-HAG complex was searched for with the help of Lamarckian Genetic algorithm method.The maximum number of evaluations and generations was fixed at 2500000 and 27000 and the number of docking times was set at 2000.Ultimately, the best docking pose of the HAG-APM complex was obtained by the clustering analysis.

Molecular dynamics
To further investigate the binding characteristics of the APM-HAG complex, MD simulation was performed using GROMACS software.The original structures of the APM-HAG system that was used for MD simulation were obtained from molecular docking.Then, the RESP charge of the ligand APM was fitted using the antechamber program in Amber Tools.Subsequently, the topology and coordinates of the ligand were obtained with the aid of the ACPYPE program and the GROMACS software.The APM-HAG complex was placed in the center of a dodecahedral box at least 1.0 nm away from the edge of the box, followed by solvation using the TIP3P water model.The charge of the system is neutralized by the addition of an appropriate amount of Na þ , followed by the steepest descent method (50000 steps) for energy minimization of the system.Next, under restricting APM and HAG, the temperature and pressure of the simulation system were equilibrated at 300 K and 1 bar using the Berendsen thermostat and the Parrinello-Rahman methods, respectively.Finally, the molecular dynamics were simulated for 50 ns under unrestricted APM and HAG.The bonds associated with hydrogen atoms are constrained by a linear constraint algorithm.Van der Waals interactions were calculated using the cut-off method by setting the cut-off distance to 1.2 nm.The simulation trajectory was sampled at the interval of 10 ps.

Data analysis
All spectral measurements were done three times.The results of all measurements were expressed as average value.The obtained data were analyzed with origin software.

UV-vis spectroscopy
It is well documented that the protein-ligand interaction is commonly analyzed using the UV spectrum (Wang, Kou, et al., 2020).It has been investigated that the UV spectrum of HAG shows significant absorption at near 205 and 280 nm, respectively.The former is allocated to the n!p � transition of the carbonyl group in the peptide bond that presents the backbone of the protein, and the latter is allocated to the p!p � transition of the aromatic amino acids such as Tyr, Trp, and Phe residues (Wang et al., 2020a, Makarska-Bialokoz, 2018).Generally, for the n!p � transition absorption band, the red shift of the band occurs with the decrease in the polarity of the environment surrounding the chromophore.The UV spectra of HAG solutions after mixing with different concentrations of APM (0-0.06 lM) were illustrated in Figure 2. It was found that upon increasing the concentration of APM, the absorbance of HAG at near 207 nm was gradually reduced and the peak was slightly red-shifted by 3 nm, while the absorbance at near 278 nm was also reduced slightly, but no change in position.To our knowledge, the absorbance mainly depends on the type and probability of electron transition.For a certain absorption band, the transition type is certain.Then, for the corresponding absorption band, the greater the transition probability, the greater the absorbance.In the presence of APM, the decrease in the absorbance of HAG suggests that thee APM-HAG interaction causes the decline in the transition probability.The observed results suggested that the addition of APM resulted in a slight change in the conformation of HAG and formed the APM-HAG complex and the overall polarity of the environment surrounding the chromophores decreased with adding APM (Makarska-Bialokoz, 2018).

Fluorescence spectra of HAG
Different fluorescence spectroscopic techniques are widely used for the study of the intermolecular interaction characteristics.The fluorescence emission spectrum of HAG upon increasing a concentration of APM were shown in Figure 3a, which is responsible for Trp and Tyr residues (Gandhi & Roy, 2019).As seen from Figure 3a that the APM was added into the HAG solution led to an obvious decline in the endogenous fluorescence of HAG with the blue shift of the maximum excitation wavelength (1 nm).This illustrates the fact that APM effectively quenched the endogenous fluorescence of HAG and reduced the hydrophobicity of the microenvironment surrounding Trp and Tyr residues.Meantime, an obvious emission peak at around 424 nm was observed when adding APM, which belongs to the emission of APM.Thus, the effect of HAG on the fluorescence of APM was investigated and the findings were charted in Figure 3b.Obviously, APM exhibits only a very weak fluorescence emission band at around 424 nm in the buffer solution, while its strength is enhanced with increasing amounts of HAG, further suggesting that the interaction between APM and HAG happened and produced a fluorescent complex.
Many studied results have confirmed that the fluorescence quenching mechanism can be divided into three different manners: static, dynamic and a combination of dynamic and static (Wang et al., 2020b, Khan et al., 2022).Static quenching is due to the formation of a ground-state complex between the protein and the quencher, the relationship between F 0 /F and the concentration of the quencher is linear, and its K sv value decreases as increasing temperature.The dynamic quenching results from the collisions between the protein and the quencher, the relationship between F 0 /F and the concentration of the quencher is also linear, and the K sv value rises as increasing temperature.However, in the case of the combination quenching, the relationship between F 0 /F and the concentration of the quencher is non-linear.Thus, to clarify the quenching style, the emission spectra results obtained from the experiments were treated usually utilizing the classical Stern-Volmer equation (Yeggoni et al., 2022).
where F 0 and F are the fluorescence values of HAG in the absence and presence of APM.K sv and K q represent the classical Stern-Volmer constant and the bimolecular quenching rate constant, respectively.s 0 denote the average lifetime of the protein with a quencher, which is about 10 ns.
[Q] is the concentration of APM.Subsequently, the fluorescence quenching mechanisms can be identified by monitoring the response of quenching constant to temperature change (Al-Shabib et al., 2020).Nevertheless, it is easy to observe from Figure 3c that the correlation between F 0 /F and [Q] is not linear but a downward curved curve.This is owing to the fact that HAG has three tryptophan residues and has some residues in the binding process that are not accessible to APM.Therefore, the quenching process between APM and HAG can be described utilizing the modified Stern-Volmer equation (Mohammadi et al., 2017).
Here, f a is the percentage of fluorophores quenched by the quencher and K a denotes the Stern-Volmer quenching constant for quenchable fluorophores.
Figure 3d displays the modified curves of the APM-HAG complex at 293, 298, 303 and 308 K, the quenching constants can be obtained by the ratio of the intercept and the slope.F 0 /(F 0 -F) show well linear relationships with 1/[Q], which is suggestive of a single dynamic or static quenching mechanism.The K a and f a for the APM-HAG system based on the intercepts and slopes of these lines are calculated in Table 1.As seen, K a reduced with increasing temperature and the K q is well above 2.0 � 10 10 L�mol À 1 �s À 1 , indicating that the type of fluorescence quenching of HAG was a static quenching (Mohammadi et al., 2017).

Measurement of binding constants
In the study of intermolecular interaction between APM and HAG, the binding constant is an important index for assessing the affinity between drug and protein.Recently, many studied results confirmed that the stoichiometric ratio of HAG and drug in the drug-HAG complex is 1:1 (Fitos et al., 2012;Wang et al., 2020a;Zsila et al., 2009).The assumption about the stoichiometric ratio of 1:1 for the complexation of APM and HAG, and there is the following equilibrium.
The binding constant (K b ) is obtained from the following equilibrium constant equation where [P], [D], and [PD] denote the equilibrium concentrations of HAG, APM and APM-HAG complex, respectively.[D 0 ] is the total concentration of APM and [P 0 ] is the total concentration of HAG.Eq. ( 5) is collated to give: ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi From Figure 3a, the fluorescence intensity of APM at 334 nm is negligible, so Here, F P and F PD are the fluorescence strengths of the free HAG as well as APM-HAG complex once equilibrium is reached.k p and k c denote the fluorescence value per unit concentration of the free HAG as well as APM-HAG complexes, respectively.The fluorescence strength (F) of the mixture solution of HAG and APM at near is generated by the free HAG and APM-HAG complex.that is, Substitutes Eq. ( 7) into Eq.( 8), that is, ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi where F 0 is the initial fluorescence intensity of the HAG, F is the fluorescence intensity of the HAG solution after the addition of APM.
The steady-state fluorescence data in this work were processed using Eq. ( 9) and the curves of F 0 -F against [D 0 ] were displayed in Figure 4.The findings indicated that the relationship between F 0 À F and [D 0 ] was non-linear and had good non-linear correlation (r 2 > 0.97) according to Eq. ( 9), indicating that the assumption was reliable and feasible, that was, the complex is formed between APM and HAG in a 1:1 ratio, the K b value for the APM-HAG complex was calculated through a non-linear fitting to F 0 -F and [D 0 ] and the findings were presented in Table 2.The K b values of approximately 10 5 M À 1 was seen and decreased with increasing temperature, indicating the existence of a strong affinity between APM and HAG Table 1.Binding parameters in the APM-HAG complexation process at various temperatures.as well as a static quenching mechanism.Meanwhile, the affinity of APM on HAG (4.98 � 10 5 M À 1 , 298 K) was stronger than that on BSA (7.59 � 10 4 M À 1 , 298 K) (Kou et al., 2021b).

Assaying interaction force of APM with HAG
As is well known, the type of the main driving forces during the intermolecular interaction of drugs with biological macromolecules can be determined according to the signs and magnitudes of the thermodynamic parameters (DH 0 , DS 0 and DG 0 ), which are calculated as follows (Kou et al., 2023).
where R and T represent the gas constant and experimental temperature (293, 298, 303 and 308 K).The curve of lnK b against 1/T for the intermolecular interaction of APM with HAG were shown in Figure S1.The DH 0 and DS 0 values were computed based on intercept and slope from Figure 5a and the findings were tabulated in Table 2. Based on the viewpoint of Ross and Subramanian interpretation (Ross & Subramanian, 1981), the negative DH 0 and DS 0 values in the APM-HAG system implied that the intermolecular interaction between HAG and APM was predominantly driven by van der Waals forces and hydrogen bonding.Furthermore, jTDS 0 j < jDH 0 j indicated that the APM binding to HAG was enthalpy-driven.The sucrose is known to be a probe for monitoring hydrogen bonding forces, the competitive experiment was conducted using sucrose as a competitive agent (Momeni et al., 2017;Roy et al., 2018).The effect of sucrose on the process of quenching HAG fluorescence by APM was shown in Figure S2.There is no doubt that the addition of sucrose significantly inhibited the quenching behavior of APM towards HAG, implying a competition between sucrose and APM, also further confirms the presence of hydrogen bonds in the interaction between APM and HAG, because main interacting force between protein and sucrose is hydrogen bonding (Wang, Pan, et al., 2020).
Due to its specificity to the hydrophobic sites of protein molecules, ANS is considered an effective fluorescent probe for exploring whether hydrophobic interactions are involved in intermolecular binding processes (Wang, Pan, et al., 2020).The fluorescence of ANS can be negligible alone in the water but becomes strongly fluorescent when combined with the hydrophobic cavity of HAG (Ota & Takano, 2019).For the ANS-HAG system, the fluorescence intensity was significantly reduced due to the addition of APM (Figure S3).The facts illustrate that APM can displace ANS to combine into the hydrophobic cavity of HAG, that is, hydrophobic interactions also contribute to the complexation of APM and HAG.
According to the above experimental results, it can be ascertained that van der Waals forces, hydrogen bonding, and hydrophobic interactions dominate the complexation of HAG with APM.

Binding distance of APM with HAG
Static quenching is present in the complexation of APM with HAG, implying that energy migration may occur between the two.F€ orster resonance energy transfer (FRET) theory could frequently be useful for estimate the binding distance between APM and HAG through energy transfer efficiency (Mousavi & Fatemi, 2019).There is a clear overlap between the fluorescence emission spectra of HAG and the UV spectra of APM, which was evident from Figure S4.This means that HAG and APM can be seen as donors and acceptors, respectively.The energy transfer efficiency (E) and the binding distance (r) between HAG and APM have the following relationship (Epps et al., 2010) Table 2. Binding and thermodynamic parameters in the APM-HAG complexation process at various temperatures.where R 0 denotes the critical distance between HAG and APM when the value of E is 50%, which can be given by the following equations.
Here, k 2 represents the spatial orientation factor (k 2 ¼ 2/3), n represents the average refractive index of the medium (n ¼ 1.336), U represents the quantum yield of the fluorophore (U ¼ 0.118) (Epps et al., 2010)Epps et al., 2010.J is the area of the overlapping part of the spectrum, which can be calculated by Eq. ( 14), where F(k) is the fluorescence intensity of the HAG at k and e(k) is the molar absorption coefficient of the APM at k.
For the APM-HAG system, the area of overlap between the two spectra was 5.99 � 10 À 15 cm 3 �L�mol À 1 while the R 0 value was 2.34 nm.It was easy to calculate E ¼ 31.8% and r ¼ 2.66 nm by using the obtained R 0 value.Clearly, r is in the range of 0.5 R 0 À 1.5 R 0 and 2 < r < 8, suggesting that non-radiative energy transfer occurs during the complexation of HAG and APM.At the same time, the fact of R 0 < r 4.27 � 10 5 À 14.3 0.9812 a Rate (%) ¼ (K b, M À K b )/ K b �100%.K b and K b, M are binding constants in the presence of metal ions.b r 2 is the correlation coefficient.
reconfirms the existence of a static quenching mechanism (Louis-Jeune et al., 2012).

Circular dichroism experiments
For the study of conformational changes in protein before and after binding with ligand, far-UV circular dichroism is a well-established tool that can detect subtle changes in the secondary structure of HAG.To explore the effect of APM on the secondary structure in HAG, the CD plots of APM, HAG, and APM-HAG in the far-UV region (200-250 nm) were presented in Figure 5.It is clear from the graph that the negative peak at around 218 nm for HAG protein alone, which is the characteristic band of the b-fold structure (Abdelhameed et al., 2019), and the b-folded content of HAG alone was calculated to be 41.7%.mixing APM with HAG, the b-fold content of HAG was reduced to 40.8%, meaning that the slight change in the secondary structure happened after the formation of the APM-HAG complex.

Synchronous fluorescence spectra measurements
For obtaining information about some aromatic amino acid residues in proteins, in particular Trp and Tyr residues, synchronous fluorescence spectroscopy is a convenient and efficient tool (Sharma et al., 2014).The synchronous fluorescence spectra of HAG at various concentrations of APM were measured for characterizing the change in the microenvironment around Tyr and Trp residues and the findings were charted in Figure 6.It was seen that the fluorescence level of both Tyr and Trp residues decreased with the addition of APM, but the peak position of the Tyr residue was unchanged, while the peak position of the Trp residue was slightly red-shifted (about 1 nm).According to the above results, it can be inferred that the addition of APM caused an increase in polarity as well as a decrease in hydrophobicity around the Trp residues while not affecting the microenvironment around the Tyr residues.

3D fluorescence spectra measurements
As is known, 3D fluorescence spectrometry is also a more sensitive means of determining the conformational changes in proteins caused by binding with small molecules.The 3D fluorescence spectra of HAG in the presence and absence of APM were charted in Figure 7 and some characteristic parameters were illustrated in Table 3.In Figure 7, peak 1 is the characteristic fluorescence peak of HAG which characterizes the changes in hydrophobicity of the microenvironment around Trp and Tyr residues, while peak 2 is the expression of the Rayleigh scattering peak (Zhao et al., 2011).From Table 3, it can be found that the level of peak 1 decreased as the increasing of APM, suggesting that APM quenched the fluorescence of HAG and formed a ground-state complex with HAG, which also supports the previous experimental results.Furthermore, the maximum emission wavelength of peak 1 enlarged from 333 to 338 nm when the amount of APM was increased to 7.2 lM.This result implies that the binding reaction changes the conformation of the HAG and reduces the hydrophobicity of the microenvironment around the chromophore group.

Influences of metal ions on the affinity of APM on HAG
It is common knowledge that various metal ions are present in vivo.These metal ions may act with protein or drug and then influence the affinity between drug and protein (Li et al., 2011), so there is reason to believe that the presence of metal ions may influence the affinity of APM with HAG.In this work, the influences of some common metal ions on the binding constants between APM and HAG were studied, the results are presented in Table 4.The results indicated that with  the involvement of Ca 2þ and Zn 2þ , the K b values increased from 4.98 � 10 5 M -1 to 5.29 � 10 5 M -1 and 5.39 � 10 5 M -1 , respectively, indicating Ca 2þ and Zn 2þ promoted the binding of APM to HAG.In contrast, several other metal ions caused a certain degree of decline in the K b value, indicating that these ions weakened the affinity between APM and HAG.The promotive influence of Ca 2þ and Zn 2þ may be related to the formation of APM-Ca 2þ /Zn 2þ -HAG complexes or changes in the conformation of HAG (Zhao et 2011).The inhibition of binding between APM and HAG by some metal ions may be related to competitive interactions, thus reducing the affinity of the HAG for the APM.

Molecular docking
&Molecular docking is based on the recognition relationship between drugs and protein macromolecules in the body, as receptors, which is similar to keys and locks (Salmaso & Moro, 2018;� Sled� z & Caflisch, 2018).To investigate the intermolecular interaction of APM with HAG protein, molecular docking was performed on Autodock 4.2 software.The docking posture of APM on HAG for the minimum level of energy (-37.30kJ�mol À 1 ) is illustrated in Figure 8a and the corresponding energy data are listed in Table 5.It is not difficult to find that APM binds into the hydrophobic cavity of HAG, and 1-cyanocyclopentyl locates at the outer end while 4-pyridylmethylamino-3-pyridyl inserts deep into the cavity.Energetically, the sum of van der Waals forces, hydrogen bonding, and desolvation free energy is greater than electrostatic energy, meaning that they play a predominant position in the binding between APM and HAG, while the presence of electrostatic interactions can be excluded.
The docking results also revealed the intermolecular interaction of APM with a number of amino acid residues including conventional hydrogen bonds with Ser-125 and Tyr-37 residues with bond length of 2.10 Å and 2.86 Å, hydrophobic interactions with His-97, Leu-112, and Phe-114 residues, van der Waals forces with Arg-90, Thr-47, and Tyr-27, etc. (Figure 8b).

Molecular dynamics simulations
In order to further explore the interaction characteristics between APM and HAG, the APM-HAG complex obtained from molecular docking with the minimum binding energy and free HAG was simulated using MD approach and the simulated findings were depicted in Figure 9.The root mean square deviation (RMSD) of atoms in the protein backbone is generally utilized for assessing the conformational stability of the system.The fluctuations of RMSD values for both systems tend to stabilize after 35 ns (Figure 9a) and the mean values for free HAG and the APM-HAG complex within 15 ns of stabilization were 0.2445 nm and 0.2417 nm, respectively.This stated clearly that the conformation of free HAG and the APM-HAG complex has stabilized and the molecular docking findings were reliable.The root mean square fluctuation (RMSF) stands for the deviation of residues in protein from a reference position, which reflects the changes in the flexibility of the residues during the whole simulation period.It can be seen that the RMSF values for residues from 11 to 23 and 50 to 62 increased significantly upon binding to APM (Figure 9b), suggesting that the rigidity of the residues was reduced, in other words, their flexibilities were enhanced due to binding APM Meanwhile, the radius of gyration (Rg), which reflects the compactness of protein structure, for two systems was also assessed and shown in Figure 9c.Obviously, compared to the HAG alone system, the average Rg value of HAG in the APM-HAG complex increased from 1.625 nm to 1.655 nm after binding to APM, indicating that the binding reaction reduced the compactness of the HAG structure.Additionally, the hydrogen bonding interactions during the binding process were evaluated and the finding was charted in Figure 9d.The findings demonstrated the involvement of hydrogen bonds in the binding process between APM and HAG.
To further evaluate the interaction forces and the contribution of individual residue in the binding process of APM to HAG, the binding free energy, which is consisted of van der Waals energy, electrostatic interaction energy, polar solvation energy, and nonpolar solvation energy, was analyzed using the MM-PBSA method (Kumari et al., 2014) and the results were reported in Table 6.The results showed that the binding energy of system is À 142.504 kJ mol À 1 .In addition, the van der Waals energy has the most negative value, meaning that it plays a dominant role in driving the complexation between HAG and APM.And, the contribution of each energy was of the order of van der Waals energy > electrostatic interaction energy > nonpolar solvation energy > polar solvation energy.However, the polar solvation energy (>0) is not conducive to binding reactions.
The energy contribution of individual residue in the binding process of APM to HAG was illustrated in Figure 10 and Table 7.The findings indicated that residues Phe-51, Leu-62, Leu-79, Ile-88, Val-92, Leu-112 and Phe-114 had more negative binding energy values, suggesting that these residues contribute to the binding reaction was relatively large.The main source of energy for these residues was van der Waals energy.In contrast, residues Arg-90 and Tyr-110 had positive binding energy values, implying that they are not conducive to the binding process.In combination with the previous fluorescence experiments, it was determined that van der Waals forces, hydrogen bonding and hydrophobic interactions drive the binding between APM and HAG.

Conclusion
The purpose of this work was to disclose the intermolecular interaction characteristics between APM and HAG in vivo at the molecular level with the help of multi-spectroscopic approaches as well as computational methods.The analytical findings of fluorescence spectra and UV-vis spectra showed clearly that APM formed a complex with HAG and had a stronger affinity (K b ¼4.98 � 10 5 M À 1 at 298 K), and the fluorescence of HAG was quenched in a static mode by APM.The findings of both spectroscopic analysis and molecular docking confirmed that the dominance interaction was owing to van der Waals forces, hydrogen bonding and hydrophobic interactions in the binding process.Based on FRET theory, the binding distance between APM and HAG was calculated to be 2.66 nm.The findings of circular dichroism suggested that the b-fold level of HAG was decreased upon interaction with APM.The changes in the  respectively refer to van der Waals energy and electrostatic energy, DE pol refers to the electrostatic contribution to solvation free energy, DE apol refers to the free energy of nonpolar solvation, DG binding refers to the binding free energy.
synchronous and 3D fluorescence spectra demonstrated that the formation of the APM-HAG complex influences the microenvironment of Trp residues.The participation of certain metal ions can promote or inhibit the binding reaction, thereby affecting the binding capacity of APM with HAG.Molecular docking and MD simulations provided evidence and a valid complement to the results of spectroscopic experiments.This study was to better understand the pharmacodynamic and pharmacokinetic information of APM in humans and to provide a basis for designing optimal therapeutic dosage forms.In addition, it is expected that this binding information may assist in the structural modification of relevant drugs.

Figure 2 .
Figure 2. UV spectrum of HAG (7.2 lM) at 298 K as the amount of APM was increased from 0 to 0.06 lM.

Figure 3 .
Figure 3. (a) Steady fluorescence spectrum of HAG (1.2 lM) when the amount of APM was added stepwise from 0 to 14.4 lM.(b) Steady fluorescence spectrum of APM with the addition of HAG from 0 to 4 lM.(c) The plot of F 0 /F vs C for the APM-HAG system.(d) The modified Stern-Volmer curves of APM interaction with HAG at 293, 298, 303, and, 308 K, respectively.

Figure 5 .
Figure 5.The Far-UV CD spectra of HAG, APM and APM-HAG.

Figure 6 .
Figure 6.Synchronous fluorescence spectroscopy of HAG when increasing amounts of APM from 0 to 3.6 lM.(a) Dk ¼ 15 nm, (b) Dk ¼ 60 nm.The Dk value is a difference between k ex and k em .

Figure 8 .
Figure 8.(a) The finest docked conformation of APM with HAG.(b) Interaction forces involved between APM and HAG.

Figure 9 .
Figure 9. (a) Changes in the RMSD values of the system during the simulation.(b) Changes in the RMSF values of residues during the simulation.(c) Changes in the R g values of the system during the simulation.(d) Data of hydrogen bonding during the complexation between APM and HAG.

Figure 10 .
Figure 10.Contribution of per-residue to the binding free energy during the APM-HAG complexation process.

Table 3 .
Characteristic parameters of APM-HAG system obtained from threedimensional fluorescence experiments.

Table 4 .
Binding constants of the APM-HAG complexation process in the presence of metal ions.

Table 7 .
The energy contribution (kJ�mol À 1 ) of key residues in the complexation of APM and HAG a .DE MM refers to the vacuum potential energy, including van der Waals energy (DE vdw ) and electrostatic energy (DE elec ).