NMR spectroscopy has been utilized as a useful method for analyzing macromolecular complexes and screening compounds with affinity to target macromolecules. Various NMR-based screening methods have been developed using chemical shift perturbation [1-3], transferred NOE [4,5] and diffusion and relaxation editing methods [6,7]. In the development of NMR techniques for screening compounds, one of the essential requirements is the selective detection of the ligand signals while suppressing signals from target proteins. It has been shown that NOE-pumping [8], reverse NOE-pumping [9], saturation transfer difference (STD) [10] and water-ligand observed via gradient spectroscopy (WaterLOGSY) [11,12] experiments could directly detect bound ligands. These methods were designed to detect ligands with proton detection, and furthermore, the NMR-based methods have been extended to fluorine detection [13,14]. An incorporation of fluorine into drugs provides simultaneous modulation of electronic, lipophilic and steric parameters, indicating that inclusion of fluorine atoms in a drug molecule can alter its chemical properties and biological activities, and also influence the interaction with its target [15]. In analysis of fluorinated compounds, some useful NMR methods, which are applicable to the NMR-based screening, are required to be developed. Although ¹⁹F NMR spectroscopy is feasible to analyze the fluorinated compounds [16-19], the pervasive NMR spectrometer consoles consist of a single high and a few low band amplifiers, which are incapable of performing 1H-19F heteronuclear experiments. In the recent study, an efficient technique to achieve ¹H-¹⁹F heteronuclear experiments using a conventional NMR spectrometer equipped with a ¹H/¹⁹F /¹³C-double tuned probe was developed in our study [20]. Here, we propose an effective approach to discriminate binding of the fluorinated compounds to proteins with ¹⁹F detection, using the complex of diflunisal (Figure 1a) and human serum albumin (HSA).

Figure 1: Chemical structures of (a) diflunisal and (b) enoxacin.

Instrumentation and chemicals

All of the spectra were recorded at 20°C on a Varian 600 MHz NMR system or JEOL ECZ-400S spectrometer. Diflunisal, enoxacin (Figure 1b) and HSA were purchased from Sigma-Aldrich. Two NMR samples were prepared; (i) a 5 mm tube containing a solution of 4.0 mM diflunisal and 4.0 mM enoxacin, (ii) double NMR tubes, which comprised a 3 mm tube used as an inner tube containing 10 mM enoxacin and a 5 mm tube used as an outer tube containing 5.0 mM diflunisal and 0.1 mM HSA. The enoxacin solution, which was prepared in the absent of HSA, was used as a negative control. Each solution contained 90% ¹H₂O and 10% ²H₂O.

NMR spectroscopy

The experimental parameters of ¹⁹F{1H} STD experiment at ¹H frequency of 600 MHz were as follows; data points=8192, spectral width of ¹⁹F=16026 Hz, number of scans=8192, d1=0.1 s, d2=1.5 ms, on-resonance frequencies of ¹H=0.9 or 4.8 ppm, and off-resonance frequency=-20 ppm. In the reference ¹⁹F experiment, number of scans was 64. Those of W5-WaterLOGSY experiment were as follows; data points=16384, spectral width of ¹H=8012 Hz, number of scans=128, recycle time=1.0 s, mixing time=1.5 s. The Gaussian window function was used with zero-filling by a factor of 2.

Pulse sequence optimization

Share of one high band amplifier without circuitry changes enabled acquiring the ¹H-¹⁹F heteronuclear spectra, although the pervasive NMR spectrometer, consisting of a single high and a few low band amplifiers, is incapable of performing ¹H-¹⁹F experiments. In the present research, the ¹⁹F{¹H} STD pulse sequence was modified to be applicable to the WaterLOGSY type experiment for screening of fluorinated ligands bound to protein. In comparison with the former type ¹⁹F{¹H} STD pulse sequence (Figure 2a), the present pulse sequence was rather simplified as shown in Figure 2b. The consecutive ¹⁹F and ¹H 90° pulses between the gradient pulses G1 were omitted, and a single ¹H pulse at the low power was incorporated for the selective excitation of proteins. In the ¹⁹F{¹H} STD experiment, two type of spectra were acquired depending on the ¹H resonance of selective excitation. In the first experiment, the ¹H resonance at 0.9 ppm, corresponding to the methyl region of protein, was selectively excited. During the subsequent process, the proton magnetization of the protein was transferred to fluorine of the ligand for detection. This experiment corresponds to the conventional STD experiment with 19F detection, which can also be acquired using a sample solution of 100% ²H₂O. The ¹⁹F{¹H} STD spectra acquired using the above pulse sequences are shown in Figure 3. The ¹⁹F chemical shifts of diflunisal were -112.5 and -115.0 ppm, and that of enoxacin was -129.4 ppm. Enoxacin in a 3 mm tube, corresponding to the inner tube, was used as a free ligand in the absence of HSA. Although the ¹H resonance of a methyl group of enoxacin (1.4 ppm) was close to that of the selective excitation (0.9 ppm), an intensity of the ¹⁹F signal of enoxacin resonating at -129.4 ppm in the ¹⁹F{¹H} STD experiment was minute (Figure 3b). This result indicated that the ¹H selective excitation at 0.9 ppm unexcited ¹⁹F of enoxacin and the ¹⁹F signals of diflunisal bound to HSA were selectively detected. The present optimization of the pulse sequence resulted in the increase of sensitivity by a factor of 1.7.

Figure 2: Pulse sequences of 19F{1H} STD experiment. (a) The former type pulse sequence and (b) the present 19F{1H} STD pulse sequence. The thin bars represent 90 degree pulses. All pulses were along x unless otherwise shown. The experimental parameters were; d₁=0.1 s, d₂=1.5 ms, G₁=7.2 G/cm, gradient pulse width=2.0 ms, 1H soft pulse width in (b)=2.5 s. Phase cycling: φ1=x, -x, -x, x, y, -y, -y, y; φr= x, x, -x, -x, y, y, -y, -y.

Figure 3: (a), (b) The 19F{1H} STD spectra acquired using the pulse sequences shown in Figure 2a and 2b, respectively. The 1H resonance at 0.9 ppm was selectively excited. Double NMR tubes, which comprised a 3 mm tube containing 10 mM enoxacin and a 5 mm tube containing 5.0 mM diflunisal and 0.1 mM HSA, were used.

¹⁹F - WaterLOGSY type experiment

In the second experiment, the ¹H resonance at 4.8 ppm, corresponding to the water resonance, was selectively excited. The ¹H magnetization of water, which was transferred to the protein, was again transferred to the bound ligand via direct or indirect relay processes. The signals of enoxacin were expected to be observed with the opposite phase with respect to those of the bound ligand. This experiment is conceptually identical to the WaterLOGSY. In a sample solution without HSA, all ¹⁹F signals of difunial and enoxacin were observed with the same phase as the negative signals (Figure 4a). In the experiments using a sample of double NMR tubes, the ¹⁹F signals of diflunial which bound to HSA and that of enoxacin as a negative control were observed as opposite phase (Figure 4b). The present methods are feasible to distinguish the bound molecules from the unbound molecules. In comparison of the conventional ¹⁹F spectra in the presence and absence of HSA, (Figure 4c and 4d), the signals of diflunial, resonating at -112.8 and -115.0 ppm, were clearly broadened in the presence of HSA (Figure 4c), indicating the complex formation.

Figure 4: (a) The ¹⁹F{¹H} STD spectrum and (d) the reference 19F spectrum acquired using a sample solution of 4.0 mM diflunisal (■) and 4.0 mM enoxacin (●). (b) The ¹⁹F{¹H} STD spectra and (c) the reference ¹⁹F spectrum acquired using double NMR tubes, which comprised a 3 mm tube containing 10 mM enoxacin and a 5mm tube containing 5.0 mM diflunisal and 0.1 mM HSA. In (a) and (b), 1H resonance at 4.8 ppm was selectively excited.

It was demonstrated that the WaterLOGSY type experiment with ¹⁹F detection was an effective method to selectively detect the fluorinated compounds bound to macromolecules. The ¹⁹F detection was advantageous in setting the NMR experiments, because suppression of the protein signals before acquisition using the T2- filter was unnecessary and sample solutions containing 10% ²H₂O can be used for the present experiments. Solvent exchange to 100% ²H₂O often causes loss of the valuable samples. The proposed approach for discrimination of binding is expected to be a useful NMR-based screening method applicable to the fluorinated compounds.

This work was supported by two Grant-in-Aids for Scientific Research (No. 20580108 for KF and 15K05550 for MT) from the Ministry of Education, Culture, Sports, Science, and Technology.