Introduction

Cyclodextrins (CDs) are among the most remarkable macrocyclic compounds with significant theoretical and practical impact in chemistry. These are widely used in analytical, pharmaceutical, agriculture, food chemistry and other fields as enzyme mimics, catalysts, drug carriers, masking agents for bad odor and taste. The importance of CDs in research as well as their industrial applications is due to their ability to form host–guest complexes via non-covalent interactions by engulfing molecules, ions or radicals into their hydrophobic cavity [1]. The basis of host–guest chemistry is molecular recognition which is the specific interaction between two molecules, complementary in their geometries and electronic features, resulting in an ensemble composed of two or more units. Molecular recognition is central for many chemical and biological processes and, therefore, understanding the fundamental principles that govern the binding of a guest molecule to its host and accurate prediction of their binding modes are important goals in host–guest chemistry. CD inclusion complexes are considered best model for understanding mechanisms of molecular and chiral recognition and their structure elucidation in solution state is of fundamental importance but a challenging task [2, 3]. A combination of experimental and computational methods is considered most appropriate for structural studies of CD complexes [4].

The structural, thermodynamic and kinetic properties of CDs alone or in complexes with small guest molecules have been explored extensively using a wide variety of experimental methods. Of various methods used, only NMR can yield atomic level information about the inclusion complexes in solution state [5]. CDs are doughnut shaped molecules, composed of 1-4 linked α-d-glucose units, called α-(six), β-(seven) and γ-CDs (eight). The narrow rim is lined with primary 6′–OH while 2′–OH and 3′–OH′ are located at the wider rim. The CDs exterior is fairly hydrophilic while cavity is relatively hydrophobic where H-3′ and H-5′ are positioned [6]. The formation of a CD inclusion complex is indicated by the observed chemical induced shift changes in the CD cavity as well as guest protons in the spectrum of their mixture compared to pure components. Stoichiometry and binding constants can be determined by NMR titrations [7]. 2D ROESY spectra are used to identify the part of the guest trapped in the cavity, the mode of penetration and the probable orientation of the guest in cavity [8]. Further structural details are obtained by computational methods.

Molecular modeling is a promising tool for structural studies and useful information, using well tested methods and established parameters, have been obtained in numerous cases for small and medium sized isolated molecules. However, molecular modeling of large flexible molecules and supramolecular systems, particularly CD complexes, is problematic due to the complexity of inclusion process. But CD complexes have been studied applying computational methods developed for small, rigid molecules. The accuracy of the calculations and dependence of the results on the assumed parameters have almost never been tested. It is believed that the methods used for CD modeling are improperly applied and those acceptable can only furnish qualitative results [9].

Measurement of interproton distances is a powerful tool for establishing three dimensional structure of a substance. ROESY peak intensities depend on internuclear distances and, in principle, can be used to determine interproton distances and vice versa but this is generally avoided because several other factors may affect intensities. Still, there are numerous examples where highly accurate structures of variety of rigid compounds have been established [10]. However, the use of this approach in the area of CD chemistry is totally neglected.

We are interested to develop method for analyzing computational models for their atom-accuracy and recently demonstrated that quantitative ROESY data analysis can provide valuable structural information on the atom-accuracy of CD inclusion complexes [11]. We have now studied the complexation of morniflumate hydrochloride (MOR) with β-CD in aqueous medium. MOR was selected for the study because it has aromatic rings which are considered best fit for inclusion by CDs. Moreover, complexation of pharmaceuticals with CDs is desired in the area of pharmaceutical science. The structure obtained by 1H NMR, molecular mechanics and molecular dynamics were studied by quantitative ROESY analysis. The results show that a comparison of calculated and experimental ROESY intensities can be used to analyze the atom-accuracy of the structure. The results again demonstrate that ROESY intensity ratios are not much affected by other factors except internuclear distances.

Results and discussion

1H NMR spectroscopic studies

Of several techniques used to study CD host–guest complexes, NMR spectroscopy is the most powerful tool which give an obvious distinction between inclusion complexation and other possible interactions. Entrapment of aromatic molecules by CDs into their hydrophobic cavity is indicated by upfield shifts of the internal CD protons (H-3′ and H-5′) and downfield shifts of included guest protons resonances in the spectra of mixture compared to pure CD and pure guest. Remaining CD protons are usually unaffected or show insignificant effect [12]. Since 1H NMR spectroscopy was first introduced to study CD complexes [13], there have been numerous studies involving aromatic compounds [14]. Stoichiometry and association constants (K a ) are easily determined using NMR titration experiment [7]. NMR spectra are obtained for several solutions containing a fixed concentration of one component and various concentrations of other component. The shift change data is graphically plotted against the concentration using modified Benesi-Hildebrand equation [15].

The foremost requirement for structural study of a CD complex is an unambiguous assignment of all the host and guest proton resonances, especially in the 1H NMR spectrum of mixture to identify, beyond any doubt, the included part of the guest. The assignment of each aromatic proton of MOR and all β-CD protons was made with the help of COSY spectrum of their mixture. The signals for two aromatic rings did not interfere with each other but signals for H-1 and H-2 were found partially merged. Parts of the COSY spectrum exhibiting assignment of aromatic protons of MOR and β-CD protons are shown in Fig. 1.

Fig. 1
figure 1

Parts of COSY spectrum showing a MOR aromatic and b β-CD protons assignment

Full size image

All NMR spectra were recorded at room temperature on a JEOL 500 MHz instrument in D2O and HDO signal at 4.800 ppm was used as internal reference. 1H NMR spectra were recorded for seven samples of MOR/β-CD mixture. MOR (3 mg) and calculated amounts of β-CD were dissolved in 0.7 ml D2O to prepare seven samples having molar ratios [β-CD/MOR] 0.25, 0.5, 0.75, 1, 1.5, 2 and 2.5 to obtain chemical shift (∆δ) data for MOR. Signals for H-2, H-5 exhibited downfield shift, H-1, H-4 and H-7 shifted highfield while remaining protons of MOR were unaffected in the presence of β-CD (Fig. 2). Moreover, β-CD cavity protons (H-3′, H-5′) exhibited highfield shift in the presence of MOR while exterior protons were not affected. This was clear evidence of the presence of MOR-β-CD inclusion complex in solution. The chemical shift (∆δ) data for aromatic protons of MOR was plotted against β-CD concentration in the form of [β-CD]/∆δ versus [β-CD] using Scott’s equation [16] for 1:1 complex which gave excellent linear fits confirming the 1:1 stoichiometry of the complex (Fig. 3).

Fig. 2
figure 2

NMR titration experiments of MOR against β-CD

Full size image
Fig. 3
figure 3

Scotts plot for 1:1 MOR-β-CD complex

Full size image

ROESY experiment is one of the most effective techniques to study CD inclusion complexes. Despite the challenges associated with binding dynamics, which may interfere with the development of cross relaxation interactions, the Nuclear Overhauser Effect (NOE) is widely used. The crosspeaks caused by NOEs between guest and β-CD protons provide useful structural information. The 1H NMR results, which merely indicate the presence of guest molecule inside β-CD cavity, are verified and the part of the guest and its orientation in β-CD cavity are established by the ROESY spectroscopy [17]. Figure 4 shows crosspeaks between aromatic protons of MOR and β-CD cavity protons in the ROESY spectrum recorded with a mixing time of 500 ms. As can be seen, all CF3-substituted ring proton signals display contacts with β-CD cavity protons but only H-5 of pyridine ring shows contact with either H-3′ protons. This confirmed not only the formation of MOR-β-CD inclusion complex but also encapsulation of only CF3-substituted ring by the β-CD cavity. However, it is not clear whether guest entry is from narrow or wider side.

Fig. 4
figure 4

ROESY spectrum showing intermolecular crosspeaks between β-CD cavity and MOR aromatic protons

Full size image

Molecular modeling studies

Molecular mechanics and molecular dynamics simulation studies were performed using Allinger’s force field (MM2) [18] as included in CS Chem3D Pro (Cambridge Soft Corp.) at 298 K in the vapor state because β-CD cavity is a hydrophobic microenvironment where no water molecule is present in the complexed state. Published neutron diffraction coordinates of hydrated β-CD were used, after removal of water coordinates [19]. The structure of MOR was drawn using CS Chem3D Pro and its geometry was minimized to a root mean square (RMS) value of 0.1 kcalmol−1 Å−1 and this geometry was used for all the studies. The β-CD cavity assumes a symmetrical configuration in water due to the presence of hydroxyl groups on the exterior but due to the presence of several free rotating hydroxyl groups C7 symmetry may be lost. This has been observed in several studies of complexes in solid as well as in solution states [20, 21]. To preserve the symmetry of β-CD in the complexes, computational studies were performed by allowing movement of only the MOR molecule while keeping bond lengths and bond angles of β-CD constant.

Molecular mechanics (MM) studies

MM studies were carried out for the inclusion process involving only the CF3-substituted phenyl ring, in light of ROESY data, from both the wide (W) and narrow (N) openings of the cavity since both are sufficiently large to allow the entry. The aromatic ring was manually placed in the center of β-CD cavity in different orientations followed by energy minimization. Figure 5 shows various studied orientations of phenyl ring, e.g. mode A means that C3–C4 bond was placed perpendicular to the z-axis in the cavity. Steric energy of various MM2 minimization studies are given in Table 1.

Fig. 5
figure 5

Schematic representation of various positions studied by MM

Full size image
Table 1 Steric energy (kcal/mol) of MM studies of MOR-β-CD complexes
Full size table

As can be seen, MM minimization studies suggests that entry of CF3-substituted ring into the cavity is energetically feasible from both sides since energy of complexes is less than the sum of the two components in all the cases. Also, wide side approach is more favorable than those for narrow side. The least energy conformation was obtained by mode WB (Fig. 6). The CF3 group prefers to stay outside the cavity and H-1 does not come in contact with β-CD cavity proton. However, since MM minimization affords local minima, MD studies were then performed which provide global minima.

Fig. 6
figure 6

MM least energy conformation (WB)

Full size image

Molecular dynamics (MD) studies

As mentioned earlier, methods developed for small rigid molecules are used for MD studies of CD complexes. One theme that emerges from their literature is the variation in initial structures of CD and guest molecule, methodologies and force field parameterizations. MD results also depend on the initial position of host and guest which is usually determined by MM minimization.

MD runs for simulating complexation process were performed starting with various positions of MOR (Fig. 5). Simulations were run for 5000 iteration steps with a step interval of 1.0 fs and frames were saved after each 10 fs interval resulting in 500 structures of each trajectory. The minimum energy conformation obtained from WB run was found to be the lowest steric energy of all. The significant feature of MD run (WB) was that CF3-substituted aromatic ring stayed in the β-CD cavity throughout. As the MD run was initiated, CF3-substituted aromatic ring moved out slightly but returned back going slightly deeper to give a global minimum after 3432 fs. Time-energy plot and few snapshots of the MD run (WB) are shown in Fig. 7.

Fig. 7
figure 7

Energy-time plot and few crucial snapshots (side view) of best MD run

Full size image

Quantitative ROESY analysis

Determination of the interproton distances in organic molecules is a potentially useful information to obtain stereochemical and conformational details [17]. NOE or ROE can be used, in principle, to measure interproton distances but are used in a qualitative or semi-quantitative manner due to the perceived inaccuracy in NOE-distance relationship because several factors may perturb NOE intensities. The transient NOE or ROE enhancements are only influenced by a single internuclear distance (r) with r−6 dependence within the Initial Rate Approximation limits. Under this condition, the two cross-relaxing spins initially behave as if they were an isolated spin pair and the growth of the NOE has a linear dependence on mixing time. Thus, for the initial rate approximation to be valid, mixing times significantly shorter than the T1 relaxation time of spin are used [10, 22]. There are many examples where quantitative use of NOEs has been made and highly accurate results are reported [23]. Assuming all the spins to be equally affected in a given experiment, the ratio of intensities for pairs of spins should be constant even with longer mixing time when the so-called initial rate approximation in not valid [24]. Using this approach, expected ROESY intensity ratios were calculated for all MM and MD minimum energy conformations by the following equation.

$$\frac{{I_{1} }}{{I_{2} }} = \frac{{r_{1}^{ - 6} }}{{r_{2}^{ - 6} }}$$

Referenced intensities for all intermolecular contacts between β-CD cavity and MOR aromatic protons were calculated and summed to obtain simple ratios, as reported, for all the MM models (Table 2). As can be seen, MM minimized conformations from the narrow side show no crosspeaks between H-5 of MOR and β-CD cavity protons which was observed in the ROESY spectrum. Also, calculated peak intensity of H-3′ of β-CD is very small compared to that of H-5′ (I H-3′ ≪ I H-5′) for all narrow side models which is opposite to what was observed. On the other hand, MM minimized models obtained by wide side entry can explain, at least qualitatively, all the contacts observed in the spectrum. Moreover, calculated intensities for MM models (WA and WB) are the best of the wide side models. We then analyzed the least energy MD conformation and found that calculated and experimental intensities were in better agreement compared to lowest energy MM conformations (WA and WB). We then manually refined the lowest energy MD model to get a model (Fig. 8) for which calculated and experimental intensities were found to agree further better but not exactly.

Table 2 Experimental and calculated relative intermolecular ROESY crosspeaks intensities
Full size table
Fig. 8
figure 8

Refined structure for MOR-β-CD complex

Full size image

As can be seen, MM2 minimized model (Fig. 6), MD least energy conformation (Fig. 7, Frame 343) and refined structure (Fig. 8) are qualitatively similar but they differ significantly in intermolecular interproton distances (Table 3). A comparison of calculated and experimental ROESY intensities clearly demonstrate that WA and WB are best among all MM models. The steric energy of WB model is also least among all MM models (Table 1). Also, MD model is better than WB which is also supported by their steric energy. However, quantitative ROESY analysis also demonstrates that these models are not atom-accurate and intermolecular interproton distances of refined model (Fig. 8) are closest, among all, to actual distances.

Table 3 A comparison of the most prominent interproton distances
Full size table

Conclusion

Structures of MOR-β-CD complex obtained by NMR, MM and MD studies were analyzed for atom-accuracy. ROESY intensities were calculated for several models from intermolecular interproton distances. Results demonstrate that atom-accuracy of molecular models can be tested by comparison of calculated and experimental intensities and this has potential to give highly atom-accurate structure of CD complexes. Also, ROESY experiment, with longer mixing time when initial rate approximation condition is not valid, can be used for quantitative purpose if intensity ratios instead of absolute intensities are used.