Heparin and Heparan Sulfate Binding of the Antiparasitic Drug

Feb 2, 2018 - *E-mail: [email protected]. Cite this:J. Phys. Chem. ... In contrast, when considering high molar excess of the heparin-binding si...
0 downloads 6 Views 3MB Size
Article Cite This: J. Phys. Chem. B XXXX, XXX, XXX−XXX

pubs.acs.org/JPCB

Heparin and Heparan Sulfate Binding of the Antiparasitic Drug Imidocarb: Circular Dichroism Spectroscopy, Isothermal Titration Calorimetry, and Computational Studies Ferenc Zsila,* Tünde Juhász, Gergely Kohut, and Tamás Beke-Somfai Biomolecular Self-Assembly Group, Institute of Materials and Environmental Chemistry, Research Centre for Natural Sciences, Hungarian Academy of Sciences, H-1117 Budapest, Hungary S Supporting Information *

ABSTRACT: This study is aimed to assess the binding interaction between the antiparasitic cationic drug imidocarb (IMD) and sulfated glycosaminoglycans (GAGs), the ubiquitious nonprotein macromolecules of living organisms. These complex, heterogeneous polyanions are the integral constituents of cell membranes and the extracellular matrix and display affinity toward basic compounds, the binding of which may affect their biological functions. Exciton-type circular dichroism (CD) spectroscopic features measured at low salt concentration verify the heparin and heparan sulfate binding of IMD, which occurs in a cooperative manner by association of several drug molecules to a disaccharide unit. Isothermal titration calorimetry (ITC) measurements reassured the heparin interaction, resulting in a Kd value in the low micromolar range. In contrast, when considering high molar excess of the heparin-binding sites, closer resembling in vivo conditions, an entirely different CD signature was induced, suggesting a shift from the oligo- to monomeric binding mode. This observation was also supported by ITC measurements using an identical sample setup. To better mimic in vivo conditions, several measurements were performed in physiological salt concentration ranges. On the basis of these, the inter- and intramolecular origin of CD activity observed under low- and high-salt conditions refer to electrostatically held oligomeric and intermolecular H-bonded monomeric drug−GAG adducts, respectively. To complement the experimental data, quantum chemical calculations were performed to assess the photophysical and conformational properties of IMD, indicating the existence of nonlinear, nonplanar interconverting conformer populations. Such a structural flexibility may be important in the multiple, cooperative binding of IMD to sterically adjacent GAG sites.



both the physiological and pathological organ functions.12 The majority of GAGs, such as heparin, heparan sulfate, dermatan sulfate, and chondroitin sulfate, are highly charged polyelectrolytes due to the presence of sulfate and carboxyl groups.13−15 As the alternating copolymers of uronic acids and amino sugars, these complex polysaccharides are ubiquitous on the surface of cells and in the extracellular matrix and have great importance in a vast array of physiological and pathological processes, including carcinogenesis,16 neurodegeneration,17 and inflammation.18,19 Distinctly from other sulfated GAGs, heparin is an intracellular product synthesized and stored exclusively in the secretory granules of mast cells. Its major repeating disaccharide unit contains α-L-iduronate-2-O-sulfate 1 → 4 linked to predominantly N-sulfated D-glucosamine.13,15 In heparan sulfate, the L-idunorate is replaced by its C5 epimer, β-Dglucuronate, and the glucosamine residues are predominantly N-acetylated (Scheme 2). Consequently, heparin bears more

INTRODUCTION Imidocarb (IMD) is a highly effective veterinary medicine approved for the treatment of tick-transmitted hemoparasitic diseases, such as bovine babesiosis and anaplasmosis.1 IMD is a cationic member of the carbanilide family of antiprotozoal compounds (Scheme 1) and is also effective against trypanosome infection.2 Despite its long therapeutic use started in the 1970s (Imizol), the mechanism of action of IMD is largely undefined. Some data point out its effect on the metabolism and/or utilization of polyamines.3 Besides the antiprotozoal action, the acetylcholinesterase inhibitory activity of IMD was also reported,4,5 and it augmented the endogenous interleukin-10 production in mice.6 Furthermore, analysis of the nuclear fraction of cell lysates suggested the nucleic acid affinity of IMD,7 which is in line with the similar binding preference of related aromatic diamidine derivatives, including berenil, stilbamidine, and pentamidine.8,9 Similar to DNA, the polyanionic glycosaminoglycans (GAGs) may be the potential targets of IMD, too. Of note, the drug preferentially accumulates in the liver and kidney,7,10,11 where GAGs as integral constituents of the glycocalyx have a pivotal role in © XXXX American Chemical Society

Received: September 6, 2017 Revised: January 12, 2018

A

DOI: 10.1021/acs.jpcb.7b08876 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

with berenil and the urea derivative surfen (Scheme 1), the GAG-binding ability of which has been demonstrated earlier.23,24 Considering these data, binding interactions between IMD and GAGs can be assumed. To test this hypothesis, circular dichroism (CD), UV absorption spectroscopy, and isothermal titration calorimetry (ITC) experiments were conducted on IMD samples treated with heparin and heparan sulfate. Quantum chemical calculations were also performed to explore the conformational behavior of IMD important in the GAG-binding process and for better correlation of its photophysical properties with the observed CD and UV spectroscopic changes.

Scheme 1. Chemical Structures of the Cationic Form of Imidocarb, Surfen, and Berenil (Diminazene)



MATERIALS AND METHODS Materials. Imidocarb (purity >97%) was obtained from Key Organics Ltd. (U.K.). Heparin sodium salt from porcine intestinal mucosa (Sigma), heparan sulfate sodium salt from porcine mucosa (Iduron), and chondroitin 6-sulfate (C6S) sodium salt from shark cartilage (Sigma) were used as supplied. All other reagents were of analytical grade. Preparation of Drug and GAG Solutions. Drug and GAG solutions were prepared freshly before each measurement in 10 mM, pH 7.0 phosphate buffer containing 15 mM Na+ from the buffer components. Due to the polydisperse nature of GAGs, their concentrations were calculated using the molecular weight of the average repeating disaccharide units: 665, 480, and 500 for heparin, heparan sulfate, and chondroitin 6-sulfate, respectively.27 Circular Dichroism and UV Absorption Spectroscopic Measurements. Spectroscopic experiments were conducted in 10 mM, pH 7.0 sodium phosphate buffer (15 mM Na+ as calculated from the buffer components) using a rectangular quartz cell of 1 cm optical path length (Hellma). CD and absorption spectra were recorded on a JASCO J-715 spectropolarimeter at 25 ± 0.2 °C and represent the average of three scans obtained by collecting data at a scan speed of 100 nm/min. Temperature control was provided by a Peltier thermostat equipped with magnetic stirring. Absorption spectra were obtained by conversion of the high voltage values of the photomultiplier tube of the CD equipment into absorbance units. CD and UV curves of drug−GAG mixtures were corrected by blank buffer solution. JASCO CD spectropolarimeters record CD data as ellipticity (“Θ”) in units of millidegrees (mdeg). The quantity of Θ is converted to molar circular dichroic absorption coefficient (Δε in M−1 cm−1) using the equation Δε = Θ/(33982cl), where “c” is the molar concentration of the ligand (mol/L) and “l” is the optical path length expressed in centimeter. Isothermal Titration Calorimetry (ITC). Thermodynamic parameters for the interaction of IMD and heparin were examined using a MicroCal iTC200 instrument (MicroCal, MA). Measurements were performed at 25 °C in low-salt (25 mM sodium phosphate, pH 7.0, no added salt) and high-salt phosphate-buffered saline (PBS) (10 mM sodium phosphate, pH 7.4, 137 mM NaCl, 2.7 mM KCl). Aliquots of heparin (250 μM, 750 μM, or 2 mM) were injected into the ITC cell containing IMD (60 μM, 200 μM, or 1 mM) in the same buffer. In the reverse experimental setup, 10 μM aliquots of IMD were added to concentrated heparin (2 mM). Titration curves were analyzed using the Origin for ITC software provided by MicroCal. Data were fitted to the one set of sites model. Computational Methods. All computations were carried out using the Gaussian 09 software package.28 Although the

sulfate groups and hence more acidic than related GAGs. Both heparin and heparan sulfate chains adopt a right-handed helical conformation.15,20 Scheme 2. Representative Repeating Disaccharide Units of GAGs Used in This Work (Redrawn from Ref 20)a

a Heparan sulfate (HS) is composed of alternating 4-linked β-Dglucuronic acid (GlcA) and 4-linked α-D-glucosamine (GlcN) rings. The glucosamine ring may be 6-O-sulfated, either N-sulfated or Nacetylated, and sporadically 3-O-sulfated, whereas the glucuronic acid may be 2-O-sulfated. The average degree of sulfation is 1.5 per disaccharide unit. Heparin (Hp) is formed by the action of C5 epimerase on GlcA giving rise to α-L-iduronic acid (IdoA). Most commonly, Hp contains a sulfate group at C2 of IdoA and additional ones with various frequencies at the GlcN unit.

Besides nucleic acids, these polyanions are also potential macromolecular targets of basic drug compounds.21−24 Inhibition of the binding of cell surface GAGs to their peptide and protein partners (e.g., chemo- and cytokines) by small molecules is a recently proposed pharmacological strategy for ameliorating deleterious autoimmune and inflammatory disorders.25,26 Imidocarb shares some structural characteristics B

DOI: 10.1021/acs.jpcb.7b08876 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Figure 1. (A) Structure of the three conformers of protonated imidocarb originating from different catchment regions with respect to dihedral angles β and γ. Definition of the four dihedral angles, α, β, γ, and δ, considered in this study are marked on CONF1/1. Left: side view; right: top view. Compared to CONF1/1 and CONF2/1, for CONF3/1, the two phenyl groups are significantly out of plane. (B) Top: the highest occupied molecular orbital (HOMO) of CONF1/1. Middle and bottom: the orientation of the transition dipole moment (tdm) allied to the π−π* transition at 253 and 267 nm, respectively.

Table 1. Structural and Energetic Properties of Imidocarb Conformersa,b ΔG (kcal/mol)c conformer

method

α

β

γ

δ

gas

H2O

CONF1/1

B3LYP M06-2X ωB97X-D B3LYP M06-2X ωB97X-D B3LYP M06-2X ωB97X-D B3LYP M06-2X ωB97X-D B3LYP M06-2X ωB97X-D B3LYP M06-2X ωB97X-D B3LYP M06-2X ωB97X-D B3LYP M06-2X ωB97X-D

9.3 20.1 15.9 11.7 16.5 15.3 −9.4 −11.2 −12.9 −8.1 −14.2 −13.1 0.2 −4.6 −3.3 −5.1 −11.0 −22.2 −155.3 −150.6 −152.0 −151.4 −146.4 −148.4

−178.6 −161.2 −175.8 −177.7 −176.4 −178.8 178.6 −160.9 −163.7 −173.7 −163.8 −162.8 0.2 −4.6 −3.3 5.1 0.3 −10.6 −155.3 −150.6 −152.2 −155.4 −148.8 −144.2

−27.2 −28.1 −29.4 −27.0 −28.1 −31.3 27.2 27.5 29.8 27.1 28.1 29.4 152.8 151.7 149.1 −152.4 −150.0 −149.2 −149.8 −149.1 −147.2 149.1 148.4 145.8

−149.0 −148.2 −145.0 148.8 148.4 146.5 149.0 147.6 146.5 −148.9 −149.2 −146.7 −28.0 −29.0 −31.7 −28.2 −29.5 −32.1 30.6 31.2 33.1 30.8 31.5 33.3

0.00 0.00 0.00 −0.09 −0.36 0.20 0.01 −0.39 −0.24 −0.27 1.22 −0.30 0.59 0.88 1.58 0.14 0.76 0.69 2.32 2.46 3.20 2.33 1.89 2.41

0.00 0.00 0.00 −0.06 −0.39 0.16 0.01 −0.52 −0.32 −0.26 1.14 −0.36 −0.28 −0.18 0.61 −0.80 −0.37 −0.18 0.71 0.96 1.53 0.69 0.36 0.71

CONF1/2

CONF1/3

CONF1/4

CONF2/1

CONF2/2

CONF3/1

CONF3/2

Torsional angles (α, β, γ, and δ in degrees) were obtained at the B3LYP/6-311++G(2d,2p), M06-2X/6-311++G(2d,2p), and ωB97X-D/6-311+ +G(2d,2p) levels of theory. bFree-energy values were obtained as detailed in the Materials and Methods Section. cValues are relative to CONF1/1 at all investigated levels of theory. a

structure of IMD renders its conformational flexibility relatively low, to minimize the chance of missing an important

conformation, an exhaustive systematic scan was performed. The initial exploration of the conformational space was carried C

DOI: 10.1021/acs.jpcb.7b08876 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

conformers. For the implicit water calculations, CONF1/3 and CONF1/4 are the most stable for M06-2X and ωB97X-D, respectively. However, for B3LYP, CONF2/2 appears to be the lowest energy structure. We note that these values have only minor differences especially if one considers that the three methods treat dispersion effects differently. The employed calculations consistently report highest energy for CONF3/1 and CONF3/2, rendering the most extended conformer class to the least stable one. To provide a quick overview on how one conformer is populated, we have combined the energy terms obtained from the point energy calculations as well as the normal mode vibration analysis, which resulted in enthalpy and entropy terms as well. By using these data, the Boltzmann distribution of the identified conformers is derived (Table 2).

out at the PM6 level of theory, where four dihedral angles were scanned with a resolution of 60° (Figure 1). These have resulted in 1296 constrained conformers. Selection procedures for higher-level calculations were performed on these conformers as follows. Initial clustering of the obtained conformers was based on the dihedral angles, by taking three catchment regions, 120° each, along the four torsional angles. The lowenergy conformers were selected on the basis of the relative energies of the neighboring conformer clusters, resulting in three conformers from separate catchment regions regarding their central β and γ dihedral angles, CONF1, CONF2, CONF3. By taking into account further symmetrical considerations on the relative positions of the imidazole rings, we finally concluded in a total of eight conformers, CONF1/1−4, CONF2/1−2, CONF3/1−2, from the three different clusters to be submitted for higher level of theory calculations (Figure 1). To test potential effect of the chosen theory, we employed Becke’s three-parameter functional with the Lee−Yang−Parr exchange functional (B3LYP),29,30 Head-Gordon’s ωB97X-D functional, which included dispersion correction and long-range electron correlation corrections,31 as well as M06-2X of the Minnesota functionals.32 For all of the functionals, the 6-311+ +G(2d,2p) basis set was employed. To better understand the molecular reason for the energetic preference of the conformers, CONF1/1, CONF2/1, and CON3/1 was also subject to natural bond orbital (NBO) analysis, based on calculations carried out at the B3LYP/6-31++G(2d,2p) level of theory. Solvent effects of water were considered for estimating the population of each conformer using the integral equation formalism for polarizable continuum model (IEFPCM).33 To calculate the thermal correction to Gibbs free energy, gas-phase optimizations and frequency calculations were performed in 298.15 K temperature and 1 atm pressure using the 631+G(d,p) basis set for all of the functionals. Finally, to compare the photophysical properties of the obtained conformers, with the experimentally observed spectra, excited-state energies and UV absorption spectrum of all conformers were calculated using time-dependent density functional theory (TD-DFT).

Table 2. Boltzmann Distribution of Imidocarb Conformers as Obtained from the Free-Energy Values at 25 °C conf. CONF1/1 CONF1/2 CONF1/3 CONF1/4 CONF2/1 CONF2/2 CONF3/1 CONF3/2

B3LYP

M06-2X

ωB97X-D

N%gas

N%water

N%gas

N%water

N%gas

N%water

16.9 19.7 16.6 26.5 6.3 13.3 0.3 0.3

9.4 10.3 9.2 14.6 14.9 35.8 2.8 2.9

18.3 33.7 35.3 2.3 4.2 5.1 0.3 0.8

10.6 20.4 25.5 1.6 14.3 19.7 2.1 5.8

19.0 13.6 28.5 31.2 1.3 5.9 0.1 0.3

13.5 10.3 23.2 24.6 4.9 18.4 1.0 4.1

The results clearly demonstrate that all conformers will be simultaneously present under standard room-temperature conditions in both gas phase and water. Further on, it can also be inferred that CONF3/1, CONF3/2, and CONF2/1 are less stable under the investigated conditions and thus have the smallest subpopulations. Photophysical Properties of Imidocarb. To facilitate the interpretation of spectroscopic observations, excited-state energies were calculated for all conformers employing TDDFT calculations at the B3LYP/6-311++G(2d,2p) level of theory, including the IEFPCM implicit solvent model for water. It has been demonstrated earlier that the B3LYP functional is suitable to accurately predict absorption spectral properties of charged, aromatic molecules.34−36 Our results indicate that there are no major differences within one conformer class; thus, these are not discussed separately. Considering conformers in CONF1, there are two π−π*-type excitations with high oscillator strengths at 253 and 267 nm, respectively. For CONF2, three similarly strong excitations can be identified at 254, 256, and 267 nm. Moreover, for CONF2, the oscillator strength at ∼255 nm is only half of the 267 nm one, which is the opposite of that observed for CONF1. In contrast to CONF1 and CONF2, for CONF3, only one main transition was found at 257 nm. All transitions have mixed excitations, involving HOMO−HOMO−2 and lowest unoccupied molecular orbital (LUMO)− LUMO+2 orbitals. The orientation of the transition dipole moments (tdms) was determined for CONF1, where both vectors lie in the plane of the conjugated aromatic moiety, the higher-energy tdm pointing more toward the charged imidazole group (Figure 1B). The reduced number of transitions obtained for CONF3 is probably due to the fact that these conformers have a nonplanar configuration around the central urea moiety. The β and γ torsion angles of this conformer (Table 1) show that the two phenyl rings are rotated



RESULTS AND DISCUSSION Structural and Energetic Properties of Imidocarb. Considering the conformation of IMD, the two central torsional angles, β and γ, are the most important in determining the relative orientation of the imidazole groups. On the basis of this, three major clusters can be identified, termed as CONF1, CONF2, and CONF3 (Figure 1A). These are mainly planar, however, for the case of α, γ, and δ, there is a small, ca. 10−30°, deviation from the perfectly planar values for all DFT methods (Table 1). This renders several distinguishable subconformers for each class, where the main difference arises from the relative sign of the α, γ, and δ dihedral angles (Table 1). When considering energies, as expected, the major changes are observed mainly between the three conformer classes. Due to the lack of strong intramolecular interactions, in principle, all conformers can be simultaneously present, as the energy difference between the lowest and highest energy conformers is within 3 kcal/mol. This is even more pronounced in water, where the polar environment further attenuates energetic differences, the largest one being 1.89 kcal/mol calculated between CONF1/4 and CONF3/1 at the ωB97X-D/6-311+ +G(2d,2p) level of theory. On the basis of the gas-phase values, CONF1/3 and CONF1/4 were found to be the lowest energy D

DOI: 10.1021/acs.jpcb.7b08876 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B relative to each other by about 50°. Although the HOMO orbitals are rather similar for all conformer classes (data not shown), the transition between the above-mentioned orbitals could be hindered, resulting in lower oscillator strength or forbidden transitions for CONF3. To study the electronic differences between the conformers, we have conducted natural bond orbital analysis on CONF1/1, CONF2/1, and CONF3/1, focusing on the potential changes in electronic structure underlying the relative stability as well as the different spectral properties of the conformers. As expected, the significant out-of-plane property of CONF3 structures (Table 1) results in the decrease of the extent of conjugation within the central urea core. This is primarily exemplified by the decrease of valence occupancy in lone pairs of the central carbonyl oxygen, O, coupled to the increase of valence occupancies on the two nitrogens, N1 and N2 (Table S1). For CONF3/1, the 2pz valence atomic orbitals of these atoms all decrease relative to those of CONF1/1 and CONF2/1, indicating that the rearrangement of electrons between the oxygen and nitrogen atoms is not coupled to increase of conjugation, rather spreads on orbitals not participating in conjugate bonds. We note, however, that even for CONF1 and CONF2 there is only a small cross conjugation rendering IMD colorless, irrespective of its conformation. Coupled to the electronic structure changes, all above-investigated orbitals of CONF3/1 have lower higher energies, which is the reason for its reduced overall stability compared to CONF1 and CONF2. On the basis of the relative energies of the conformers as well as their spectral properties, we conclude that the mixture of CONF1 and CONF2 dominates in solution, whereas CONF3 occurs in a minor fraction. Heparin- and Heparan Sulfate-Induced CD and UV Spectroscopic Changes of Imidocarb. Despite the high degree of inherent asymmetry, GAGs do not exhibit CD signals above 210 nm due to the lack of light-absorbing chromophores in that region. In contrast, IMD displays a broad, intense UV absorption feature between 210 and 280 nm consisting of at least three unresolved sub-bands around 220, 237, and 267 nm (Figure 2). The spectral positions of the most intense peaks are in a good correlation with the computational results that predicted the main π−π* transitions at ∼250 and ∼270 nm. Moreover, the experimental observation of these transitions is consistent with the prevalence of CONF1 and CONF2 because CONF3 possesses only a single principal excitation at 257 nm. Composed of two phenyl-imidazoline units connected by a central urea bridge, IMD is a symmetric molecule. Distinctly from berenil, the two halves are unconjugated and thus behave as identical but independent chromophores (IMD is colorless, whereas berenil is yellow). Accordingly, Figure 1B represents only the directionality, rather than the exact intramolecular position of the tdms. To our knowledge, no pKa value has been reported for IMD, but experimentally determined acid dissociation constants of related imidazoline drugs convincingly predict a pKa of ∼10 (Table S2). This implies that below pH 9 the drug molecules carry two positive charges. As expected from its structure, IMD itself shows no CD activity. However, addition of heparin or heparan sulfate into its buffer solution promptly induces a strong positive and a less intense, shorter-wavelength negative Cotton effect (CE). The correspondence of the peak positions to the respective absorption band indicates that they come from the π−π* transitions of IMD (Figure 3). A very similar CD pattern was

Figure 2. UV absorption spectrum of 60 μM IMD measured in pH 7.0 buffer solution (10 mM sodium phosphate, 15 mM Na+ as calculated from the buffer components, 25 °C). The blue line (right axis) is the second derivative of the UV curve displaying λmax of the unresolved components of the main band.

observed during the reverse titration, i.e., upon successive addition of the drug into GAG solution (data not shown). The development of these extrinsic CD signals suggests that upon binding to heparin or heparan sulfate, the ligand molecules adopt some sort of chirality. This could be due to the asymmetric perturbation of the electronic transitions of IMD by adjacent chiral carbons of the GAG chain, or it may be the result of dissymmetric orientation of the molecules along the polymer chain. In the former case, the induced CD band(s) has a typical Δε value of ca. ±10 or less. According to the CD curves obtained during the titration experiments, asymmetric splitting of the absorption band into opposite CD components occurs with a zero crossover point near the UV maximum (Figure 3). Of note, the λmax values of the CEs closely correlate with those of the masked absorption components centered around 242 and 263 nm (Figure S1). Taking these data together, it appears that intermolecular chiral exciton coupling occurs between the π−π* tdms of adjacent drug molecules bound in a helical array to the GAG template. In line with the exciton chirality rule,37,38 the sign order of the CD couplet is dictated by the dihedral angle between the interacting dipoles. If it is positive, i.e., the coupled tdms describe a right-handed screw sense, the sign of the first (longer wavelength) CE is induced. Accordingly, the induced CD spectrum of drug−GAG mixtures is indicative of the right-handed helical arrangement of the proximal IMD chromophores. It should be considered, however, that both CONF1 and CONF2 of IMD have two energetically close-lying excited states. Therefore, the various coupling combinations of the respective tdms, both degenerate and nondegenerate, may produce overlapping exciton CD couplets of varying amplitudes that could be the reason for the marked intensity difference of the experimentally recorded CD bands (Figure 3). In relation to heparan sulfate, the Δεmax values of heparininduced CEs were 2-fold higher, which, at least in part, can be ascribed to the lower dissociation constant of IMD−heparin complexes (vide infra). The exciton interaction is also manifested in the absorption spectrum. In GAG-bound state, the UV band of IMD displays a E

DOI: 10.1021/acs.jpcb.7b08876 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Figure 3. Difference CD and absorption spectroscopic changes of IMD measured upon consecutive addition of heparin and heparan sulfate (10 mM sodium phosphate buffer at pH 7.0, 15 mM Na+ as calculated from the buffer components, 25 °C). The arrows denote isosbestic points.

hypochromic as well as a small bathochromic effect (Figure 3). The lowering of the absorbance values referred to as hypochromism is the result of the Coulombic interaction between the tdms of drug molecules bound proximally to each other. It is an analogous phenomenon with the well-known but more pronounced hypochromism of nucleic acids of base stacking origin that depends critically upon the distance, relative orientation, and the strength of the interacting dipoles. We note that the overall shapes of the main UV band of the free and GAG-bound IMD do not differ significantly. Consequently, the GAG binding hardly affects the conformational equilibrium of the drug, suggesting that CONF1 and CONF2 remain the predominant species. Estimation of the GAG-Binding Affinity of IMD from CD Spectroscopic Data. To calculate the apparent dissociation constants (Kd) of IMD−GAG complexes, the magnitude of the first extrinsic CE was plotted against the increasing concentration of heparin or heparan sulfate in the sample solution. In both cases, the resulting titration curves showed a sigmoidal shape and the Hill coefficient (>1) suggests a positive cooperative binding interaction between the ligand molecules (Figure 4). This finding is in line with the

intermolecular exciton mechanism inferred from the CD spectra, which requires the accommodation of at least two drug molecules to proximal binding sites. According to the Kd values, IMD binds 2 times stronger to heparin than heparan sulfate, which might be related to the higher charge density of the heparin chain (Scheme 2). CD and UV Spectroscopic Features of IMD at High Molar Excess of Heparin. GAGs are ubiquitously found at the external surface of cells and in the extracellular matrix in all tissues, whereas the peak plasma concentration of IMD after intramuscular administration is about 2 μg/mL (∼6 μM).39,40 Consequently, we have also investigated the binding of IMD at high molar excess of heparin. Distinctly from the extracellular GAGs, heparin is typically found in secretory granules within mast cells, but due to structural similarities, it is commonly employed as a (albeit imperfect) substitute for heparan sulfate. When IMD was added to 2 mM heparin sample, a totally different CD pattern was measured. Instead of the intense positive CE, a low-amplitude negative−positive band pair can be seen on the red side of the spectrum (Figure 5). Below 255 nm, an additional couplet appears consisting of a negative and a more intense positive peak centered around 239 and 223 nm, F

DOI: 10.1021/acs.jpcb.7b08876 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

short-wavelength π−π* bands of IMD. The Δεmax values of this polyphasic CD spectrum are much lower than that observed at high drug/disaccharide molar ratios (Figure 3). An additional difference is that the absorption band exhibits no hypochromism. All of these spectral features suggest a fundamentally distinct binding mode of IMD. At high excess of the binding sites (2 mM heparin), the drug molecules are well separated from each other and thus no intermolecular interactions can occur among them. In such a case, the chiral perturbing effect of the immediate binding environment dominates, resulting in weak CEs allied to the π−π* transitions (Figure 5). Conversely, high drug loading of the heparin sites gives rise to intense exciton CD peaks as well as UV hypochromism due to the exciton coupling effect (Figure 3). It is to be noted that the exciton CD pattern may also contain some monomeric contribution, which, however, remains undetectable due to the strong masking effect of the much more intense exciton signals. In this sense, the three isosbestic points in the UV spectrum might be the indication for the coexistence of monomeric and oligomeric species (Figure 3). Impact of Sodium-Ion Concentration on the CD and Absorption Spectra of GAG-Bound IMD. Due to the polyanionic nature of GAGs, electrostatic interactions between their sulfate/carboxylate groups and cationic guest compounds often give a major contribution to the stabilization of the complexes. Therefore, GAG binding of positively charged molecules may be sensitive to the rise of salt concentration.21 The working buffer solution used for the CD experiments contained 15 mM sodium ion. At a fixed IMD/heparin binding ratio, this value was increased on a step-by-step basis with the consecutive recording of the CD/UV spectra. The magnitude of the induced CEs declined gradually and reached zero at around 90−100 mM Na+ (Figure 6). The UV hypochromism

Figure 4. CD titration data of 60 μM IMD plotted against the disaccharide unit concentration of the sample solutions (left axis: heparin; right axis: heparan sulfate). The red lines were obtained by nonlinear curve fitting analysis of the ellipticity values employing the “one site-specific binding with Hill slope” equation built in the GraphPad Prism software (ver. 6.01, San Diego, CA). Dissociation constants (Kd) and the Hill coefficients (h) are shown.

Figure 6. Effect of increasing sodium-ion concentration on the CD and UV absorption maxima of IMD−heparin mixture (10 mM sodium phosphate buffer at pH 7.0, 15 mM Na+, 25 °C). Figure 5. Difference CD and UV absorption spectrum of 40 μM IMD added into 2 mM heparin solution (10 mM sodium phosphate buffer at pH 7.0, 15 mM Na+, 25 °C). The blue line (right axis) is the second derivative of the UV curve.

was also canceled but only at significantly higher concentrations. Similar results were obtained by using heparan sulfate (data not shown). Such kind of deviation of the CD and UV displacement data points shows that in an optically inactive arrangement a fraction of IMD molecules may remain in heparin-bound state even at high sodium level.

respectively. According to the second derivative of the UV curve, the spectral positions of these CEs coincide with the G

DOI: 10.1021/acs.jpcb.7b08876 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B As mentioned in the previous section, samples with high molar excess of heparin may better represent the in vivo abundance of GAGs. To further improve this model, the effect of sodium load on the spectra was also studied at a large heparin excess. In parallel with the increasing salt concentration, the magnitude of the main CE gradually decreased but could be detected even at 140 mM Na+ (Figure 7). For

Figure 8. Difference CD and UV spectrum of 57 μM IMD added into 2.8 mM heparin solution prepared in 10 mM sodium phosphate buffer (pH 7.4) containing 137 mM NaCl and 2.7 mM KCl.

electrostatic forces are dominant in the multimeric GAG association of IMD allied with the intense, exciton CD signature (Figure 3). In vivo, some macromolecular constituents of the complex chemical environment of the human body may also perturb, in a various extent, the IMD−GAG interaction. The avid association of IMD to nucleic acids is well known for a long time.7,10 Due to nuclear and cytoplasmic localizations, however, neither DNA nor RNA can affect IMD−GAG binding. Furthermore, binding-capacity measurements made with selected bovine plasma components indicated the IMD affinity of α1-acid glycoprotein and albumin, but the binding constants were not determined.7 Comparison of Heparin-Binding Properties of IMD and Berenil. It is significant to note that at high drug loading the heparin-induced CD profiles of IMD and berenil24 are very similar. However, distinctly from IMD, which exhibits high sensitivity to the competing effect of sodium ions, this oligomeric binding mode of berenil is preserved even at 100 mM Na+, which may be related to structural differences. The terminal basic groups of berenil are in a para position related to the central linker moiety, whereas in IMD, the imidazoline rings are meta substituents. Additionally, the two amidine groups in berenil can form eight intermolecular H bonds with the anionic sites of heparin versus four ones of the imidazoline nitrogens in IMD (Scheme 1). These observations imply that the oligomeric IMD−heparin and berenil−heparin complexes are stabilized

Figure 7. Top: effect of increasing sodium-ion concentration on the Δεmax value of IMD. Bottom: induced CD curve of IMD−heparin mixture at 143 mM Na+.

additional corroboration of the IMD binding to heparin at physiological ionic strength, the drug was added into a heparin solution prepared in phosphate-buffered saline (10 mM sodium phosphate buffer at pH 7.4, 137 mM NaCl, 2.7 mM KCl). Similar, albeit less intense induced CD signals with better resolved vibrational sub-bands allied to the 1Lb transition of the benzene rings were obtained, verifying again the existence of IMD−heparin complexes (Figure 8). The shape, intensity, and λmax of the UV band are identical to those measured in GAGfree buffer solution, suggesting the monomeric heparin association of IMD molecules. Due to the complex, concentration-dependent changes of the CD profile, the monomeric binding affinity constant cannot be determined by employing conventional titration procedures. Overall, these findings refer to the decisive role of saltinsensitive intermolecular H bonds in the monomeric heparin binding of IMD. The two imidazoline nitrogens of IMD can make four intermolecular H bonds with the anionic sites of heparin (Scheme 1). Conversely, ionic-strength-dependent H

DOI: 10.1021/acs.jpcb.7b08876 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

Figure 9. ITC titration curves for the IMD−heparin interaction. (A) Aliquots of heparin (250 μM) were injected into the ITC cell containing IMD (60 μM) in 25 mM sodium phosphate buffer, pH 7, or PBS (low-salt and high-salt conditions, respectively). (B) IMD was titrated with heparin under high-salt condition (10 mM phosphate buffer, 137 mM NaCl, 2.7 mM KCl) using various concentrations in the syringe and the cell, as shown in the legend. (C) Aliquots (20 μM) of 1 mM IMD solution were added to concentrated heparin (2 mM) under conditions described in the (B). Heparin concentrations refer to the disaccharide unit (Mw: 665). Lines obtained from fitting the data to the one set of sites model are drawn to guide the eye.

mainly by electrostatic forces and intermolecular H bonds, respectively. Binding Determinants of the IMD−Heparin Interaction Studied with Isothermal Titration Calorimetry (ITC). CD titrations yielded an apparent dissociation constant of ∼10 μM for the IMD−heparin interaction and indicated the binding of several, at least two molecules to one disaccharide unit in a cooperative manner. To further clarify the binding scenario, ITC measurements were carried out. Affinity and binding determinants were tested under low-salt and high-salt conditions. Performing the ITC experiment using the setup as in CD titration, i.e., IMD in the cell was titrated with heparin under low-salt condition at pH 7, an almost identical binding curve was obtained (Figure 9A). Fitting the data to the one set of sites model assuming several identical and independent binding sites for IMD on heparin, a stoichiometry of ∼5 per disaccharide unit and an apparent Kd of 3.7 ± 0.7 μM were calculated. Considering the cooperativity of IMD binding as well as the polydispersity of heparin sample, the results obtained this way, especially the 5:1 binding stoichiometry, should be taken as approximates rather than exact values. The negative enthalpies (ΔH) obtained upon binding are indicative of electrostatic and hydrogen-bonding contributions, which is in agreement with the CD spectroscopic findings. In contrast, when the titration was carried out under physiological-like conditions, no heat could be measured (Figure 9A). The reason for this could be the masking of the electrostatic effects by the salt, resulting in the decrease of ΔH, a phenomenon similar to that observed for the induced CD signal as well. Supporting this idea, the interaction could be detected at higher compound concentrations (Figure 9B). To detect monomeric drug binding, the reverse experiment was performed adding small aliquots of IMD to concentrated heparin solution. In line with the results above, titration curves for low-salt and high-salt conditions showed remarkable differences compatible with the binding mode of IMD (Figure

9C). In the presence of salt, nearly constant small negative peaks were measured up to ∼120 μM IMD concentration, which could be associated with the monomeric binding of IMD. Addition of drug molecules producing higher IMD-todisaccharide ratios was accompanied by increasing signals referring probably to multiple ligand binding. In contrast, the latter phenomenon was dominated under low-salt condition for low IMD-to-disaccharide ratios up to reaching a plateau of large negative enthalpy changes at IMD concentrations >80 μM. The results of ITC titrations are in agreement with those obtained by CD spectroscopy, highlighting the salt sensitivity of IMD−GAG interactions. The concerted presence of the various binding modes precluded the determination of the affinity for monomeric IMD binding. Different Binding Mode of IMD with Chondroitin 6Sulfate (C6S). For comparing the binding mode and specificity of IMD among different GAGs, the effect of C6S on the CD and UV spectroscopic properties of the drug was studied. In this polymer, the glucosamine ring is replaced by Nacetylgalactosamine that is attached to a D-glucuronic acid. The average charge density of C6S lags behind that of heparin and heparan sulfate, and in contrast to heparin, its chain acquires a left-handed helical structure.41 Under the same experimental settings as applied in previous measurements, no extrinsic CD signals were observed upon mixing of C6S into an IMD sample (Figure S2). The εmax value, however, declined with the increasing GAG concentration although to a lesser extent than that found for heparin and heparan sulfate. Titration of the drug−C6S mixture with sodium chloride resulted in the gradual cancelation of the hypochromism and the full restoration of the original UV spectrum at around 140− 150 mM NaCl (not shown). Distinctly from heparin (see Figures 5 and 8), addition of IMD (57 μM) to excess concentration of C6S (0.9 mM) did not induce any CD band. The missing induced CD activity found in these experiments I

DOI: 10.1021/acs.jpcb.7b08876 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B

(4) Abdullah, A. S.; Sheikhomar, A. R.; Baggot, J. D.; Zamri, M. Adverse effects of imidocarb dipropionate (Imizol) in a dog. Vet. Res. Commun. 1984, 8, 55−59. (5) Panghal, R. S.; Rana, R. D.; Kumar, V. Effects of imidocarb on cholinesterase activity and grossly observable behaviour in dogs. Haryana Vet. 2009, 48, 26−28. (6) Katayama, T.; Hayashi, Y.; Nagahira, K.; Konishi, K.; Yamaichi, K.; Oikawa, S. Imidocarb, a potent anti-protozoan drug, up-regulates interleukin-10 production by murine macrophages. Biochem. Biophys. Res. Commun. 2003, 309, 414−418. (7) Moore, A. S.; Coldham, N. G.; Sauer, M. J. A cellular mechanism for imidocarb retention in edible bovine tissues. Toxicol. Lett. 1996, 87, 61−68. (8) Jenkins, T. C.; Lane, A. N.; Neidle, S.; Brown, D. G. NMR and molecular modeling studies of the interaction of berenil and pentamidine with d(CGCAAATTTGCG)2. Eur. J. Biochem. 1993, 213, 1175−1184. (9) Pilch, D. S.; Kirolos, M. A.; Liu, X.; Plum, G. E.; Breslauer, K. J. Berenil [1,3-bis(4′-amidinophenyl)triazene] binding to DNA duplexes and to a RNA duplex: evidence for both intercalative and minor groove binding properties. Biochemistry 1995, 34, 9962−9976. (10) Coldham, N. G.; Moore, A. S.; Dave, M.; Graham, P. J.; Sivapathasundaram, S.; Lake, B. G.; Sauer, M. J. Imidocarb residues in edible bovine tissues and in vitro assessment of imidocarb metabolism and cytotoxicity. Drug Metab. Dispos. 1995, 23, 501−505. (11) Lai, O.; Belloli, C.; Crescenzo, G.; Carofiglio, V.; Ormas, P.; Marangi, O.; Cagnardi, P. Depletion and bioavailability of imidocarb residues in sheep and goat tissues. Vet. Hum. Toxicol. 2002, 44, 79−83. (12) Tarbell, J. M.; Cancel, L. M. The glycocalyx and its significance in human medicine. J. Intern. Med. 2016, 280, 97−113. (13) Meneghetti, M. C. Z.; Hughes, A. J.; Rudd, T. R.; Nader, H. B.; Powell, A. K.; Yates, E. A.; Lima, M. A. Heparan sulfate and heparin interactions with proteins. J. R. Soc. Interface 2015, 12, No. 20150589. (14) Seyrek, E.; Dubin, P. Glycosaminoglycans as polyelectrolytes. Adv. Colloid Interface Sci. 2010, 158, 119−129. (15) Casu, B.; Naggi, A.; Torri, G. Re-visiting the structure of heparin. Carbohydr. Res. 2015, 403, 60−68. (16) Sasisekharan, R.; Shriver, Z.; Venkataraman, G.; Narayanasami, U. Roles of heparan-sulphate glycosaminoglycans in cancer. Nat. Rev. Cancer 2002, 2, 521−528. (17) Papy-Garcia, D.; Morin, C.; Huynh, M. B.; Sineriz, F.; Sissoeff, L.; Sepuveda-Diaz, J. E.; Raisman-Vozari, R. Glycosaminoglycans, protein aggregation and neurodegeneration. Curr. Protein Pept. Sci. 2011, 12, 258−268. (18) Taylor, K. R.; Gallo, R. L. Glycosaminoglycans and their proteoglycans: host-associated molecular patterns for initiation and modulation of inflammation. FASEB J. 2006, 20, 9−22. (19) Pomin, V. H. Sulfated glycans in inflammation. Eur. J. Med. Chem. 2015, 92, 353−369. (20) Rabenstein, D. L. Heparin and heparan sulfate: structure and function. Nat. Prod. Rep. 2002, 19, 312−331. (21) Zsila, F.; Gedeon, G. Binding of anti-prion agents to glycosaminoglycans: Evidence from electronic absorption and circular dichroism spectroscopy. Biochem. Biophys. Res. Commun. 2006, 346, 1267−1274. (22) Takemoto, N.; Suehara, T.; Frisco, H. L.; Sato, S.; Sezaki, T.; Kusamori, K.; Kawazoe, Y.; Park, S. M.; Yamazoe, S.; Mizuhata, Y.; et al. Small-molecule-induced clustering of heparan sulfate promotes cell adhesion. J. Am. Chem. Soc. 2013, 135, 11032−11039. (23) Zsila, F. Glycosaminoglycan and DNA binding induced intraand intermolecular exciton coupling of the bis-4-aminoquinoline surfen. Chirality 2015, 27, 605−612. (24) Zsila, F. Glycosaminoglycans are potential pharmacological targets for classic DNA minor groove binder drugs berenil and pentamidine. Phys. Chem. Chem. Phys. 2015, 17, 24560−24565. (25) Harris, N.; Kogan, F. Y.; Il’kova, G.; Juhas, S.; Lahmy, O.; Gregor, Y. I.; Koppel, J.; Zhuk, R.; Gregor, P. Small molecule inhibitors of protein interaction with glycosaminoglycans (SMIGs), a

suggests an altered, stereochemically less specific binding of IMD.



CONCLUSIONS The results presented herein demonstrate the heparin and heparan sulfate binding of the antiprotozoal imidocarb. The CD spectroscopic data obtained indicate the cooperative, electrostatic-driven association of the drug molecules to the polysaccharide chains. The intense π−π* exciton CD couplet and the UV hypochromism characteristic to this oligomeric binding mode are replaced by an entirely distinct spectral feature at high molar excess of the binding sites, suggesting the wide distribution of IMD molecules along the polymer chain. Such alterations of the CD/UV profiles refer to a concentration-dependent, monomer−oligomer binding equilibrium. Induced CD activity of monomeric type was also detected at physiological sodium-ion concentration, which, together with ITC data, verifies the formation of host−guest complexes held by intermolecular H bonds and provides a structural basis for in vivo IMD−GAG interactions. Quantum chemical calculations revealed the predominance of two kinds of nonlinear, nonplanar conformers of IMD, which are in a dynamic equilibrium with each other. This structural feature might be important in the conformational adaptation of IMD to various pharmacological targets.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcb.7b08876. Additional computational data (Table S1); pKa values of imidocarb-like compounds (Table S2); additional CD and UV curves of imidocarb (Figures S1 and S2) (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Ferenc Zsila: 0000-0002-4853-2810 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was funded by the Momentum program of the Hungarian Academy of Sciences (LP2016-2), the National Competitiveness and Excellence Program, Hungary (NVKP_16-1-2016-0007), and the BIONANO_GINOP2.3.2-15-2016-00017b project. T. B.-S. acknowledges the Marie Curie fellowship (MSCA-IF BARREL 660030). The heparan sulfate sample was gifted by Gismo Therapeutics Inc.



REFERENCES

(1) Vial, H. J.; Gorenflot, A. Chemotherapy against babesiosis. Vet. Parasitol. 2006, 138, 147−160. (2) Duch, D. S.; Bacchi, C. J.; Edelstein, M. P.; Nichol, C. A. Inhibitors of histamine metabolism in vitro and in vivo. Biochem. Pharmacol. 1984, 33, 1547−1553. (3) Bacchi, C. J.; Nathan, H. C.; Hutner, S. H.; Duch, D. S.; Nichol, C. A. Prevention by polyamines of the curative effect of amicarbalide and imidocarb for Trypanosoma brucei infections in mice. Biochem. Pharmacol. 1981, 30, 883−886. J

DOI: 10.1021/acs.jpcb.7b08876 J. Phys. Chem. B XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry B novel class of bioactive agents with anti-inflammatory properties. Biochim. Biophys. Acta 2014, 1840, 245−254. (26) Harris, N.; Koppel, J.; Zsila, F.; Juhas, S.; Il’kova, G.; Kogan, F. Y.; Lahmy, O.; Wildbaum, G.; Karin, N.; Zhuk, R.; Gregor, P. Mechanism of action and efficacy of RX-111, a thieno[2,3-c]pyridine derivative and small molecule inhibitor of protein interaction with glycosaminoglycans (SMIGs), in delayed-type hypersensitivity, TNBSinduced colitis and experimental autoimmune encephalomyelitis. Inflammation Res. 2016, 65, 285−294. (27) Ampofo, S. A.; Wang, H. M.; Linhardt, R. J. Disaccharide compositional analysis of heparin and heparan sulfate using capillary zone electrophoresis. Anal. Biochem. 1991, 199, 249−255. (28) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A. et al. Gaussian 03, revision C.02; Gaussian, Inc.: Wallingford, CT, 2004. (29) Lee, C.; Yang, W.; Parr, R. G. Development of the Colle-Salvetti correlation-energy formula into a functional of the electron density. Phys. Rev. B 1988, 37, 785−789. (30) Becke, A. D. A new mixing of Hartree-Fock and local densityfunctional theories. J. Chem. Phys. 1993, 98, 1372−1377. (31) Chai, J. D.; Head-Gordon, M. Long-range corrected hybrid density functionals with damped atom-atom dispersion corrections. Phys. Chem. Chem. Phys. 2008, 10, 6615−6620. (32) Zhao, Y.; Truhlar, D. G. The M06 suite of density functionals for main group thermochemistry, thermochemical kinetics, noncovalent interactions, excited states, and transition elements: two new functionals and systematic testing of four M06-class functionals and 12 other functionals. Theor. Chem. Acc. 2008, 120, 215−241. (33) Cancès, E.; Mennucci, B.; Tomasi, J. A new integral equation formalism for the polarizable continuum model: Theoretical background and applications to isotropic and anisotropic dielectrics. J. Chem. Phys. 1997, 107, 3032−3041. (34) Kitts, C. C.; Beke-Somfai, T.; Norden, B. Michler’s hydrol blue: a sensitive probe for amyloid fibril detection. Biochemistry 2011, 50, 3451−3461. (35) Fabian, J. TDDFT-calculations of Vis/NIR absorbing compounds. Dyes Pigm. 2010, 84, 36−53. (36) Jacquemin, D.; Perpete, E. A.; Ciofini, I.; Adamo, C. Accurate simulation of optical properties in dyes. Acc. Chem. Res. 2009, 42, 326−334. (37) Boiadjiev, S. E.; Lightner, D. A. Exciton chirality. (A) origins of and (B) applications from strongly fluorescent dipyrrinone chromophores. Monatsh. Chem. 2005, 136, 489−508. (38) Naito, J.; Yamamoto, Y.; Akagi, M.; Sekiguchi, S.; Watanabe, M.; Harada, N. Unambiguous determination of the absolute configurations of acetylene alcohols by combination of the Sonogashira reaction and the CD exciton chirality method − Exciton coupling between phenylacetylene and benzoate chromophores. Monatsh. Chem. 2005, 136, 411−445. (39) Su, D.; Li, X. B.; Wang, Z. J.; Wang, L.; Wu, W. X.; Xu, J. Q. Pharmacokinetics and bioavailability of imidocarb dipropionate in swine. J. Vet. Pharmacol. Ther. 2007, 30, 366−370. (40) Belloli, C.; Lai, O. R.; Ormas, P.; Zizzadoro, C.; Sasso, G.; Crescenzo, G. Pharmacokinetics and mammary elimination of imidocarb in sheep and goats. J. Dairy Sci. 2006, 89, 2465−2472. (41) Rodríguez-Carvajal, M. A.; Imberty, A.; Pérez, S. Conformational behavior of chondroitin and chondroitin sulfate in relation to their physical properties as inferred by molecular modeling. Biopolymers 2003, 69, 15−28.

K

DOI: 10.1021/acs.jpcb.7b08876 J. Phys. Chem. B XXXX, XXX, XXX−XXX