Human salivary secretions are usually supersaturated with the basic calcium phosphate salts that form dental enamel. This property is considered important to protect the teeth as it suppresses dissolution of enamel into the fluid environment and prevents or slows down demineralization by bacterial acids. It has also been suggested that a high salivary phosphate may be associated with lower caries rates. On the other hand, saliva supersaturated with calcium phosphate might be expected to cause undesirable precipitation in the salivary glands and on the teeth. The stable but supersaturated state of salivary secretions that creates a protective and reparative environment is considered to be due to several inhibitors of precipitation found in saliva. However, the exact mechanism of this inhibitory action for primary (spontaneous) and secondary precipitation (crystal growth) of calcium phosphate has not yet been fully elucidated.

NMR variables, in particular relaxation times, often provide useful information about the kinetics and dynamic structure of molecules. The results of several proton NMR studies on saliva suggest that high-resolution proton NMR could be useful diagnostically. To the best of our knowledge ³¹P NMR spectroscopy has never been applied to saliva, although it may be expected to provide direct information about the state of phosphate in saliva that might assist in elucidating the mechanism of the inhibitory action for the precipitation of calcium phosphate. In order to investigate the physicochemical properties of phosphate compounds contained in saliva, we have now measured ³¹P NMR spectra of healthy human whole, submandibular and parotid salivas, and the spin–lattice and spin–spin relaxation times for the phosphorus of healthy human parotid saliva, aseptically collected, free from contamination by food, bacteria leucocytes etc.

Acid-stimulated human whole, submandibular/sublingual and parotid salivas (from individuals aged 26–42 years) were collected according to the procedure described by Yamada-Nosaka et al. (1991). About 2 ml of saliva was collected during 10–30 min. There was no significant difference between the NMR spectra of the first and second millilitre. The pH of saliva, measured just after collection, was 6.9±0.6. Samples from eight different individuals (four healthy females, four healthy males) were tested. The collected saliva was stored at 5°C in a plastic test-tube (8 cm high, 1.0 cm dia) sealed with a plastic stopper and plastic laboratory film for 2 days before measurement, as the proton NMR spectra measured 1 hr, 9 hr and 2 days after collection were similar except for a peak at 1.2 parts/10⁶ that disappeared in the spectrum measured 2 days after collection, as reported previously. No precipitation was observed in the samples 2 days after collection.

The NMR measurements were made in 10-mm O.D. NMR sample tubes with a Bruker WM-360 FT spectrometer operating in the Fourier transform mode at 145 MHz. Spin–lattice and spin–spin relaxation times were measured by the inversion recovery and Carr–Purcell–Meiboom–Gill methods, respectively; 0.2 ml of D₂O (99.5%) was added to 1.8 ml of each sample as a spectrometer lock compound. All the NMR measurements except for NOE were done without proton decoupling at room temperature to avoid temperature rises in the sample during measurement. Phosphoric acid (85%) was used as an external standard for chemical-shift and signal-intensity measurements. Forty transients were collected unless otherwise stated. NOE of ³¹P occurs only in the case of dipolar relaxation with protons and is observed upon saturation of proton signals. NOE of ³¹P resonance upon proton irradiation was obtained here as a relative integrated intensity of ³¹P resonance between conditions with the proton noise-decoupler on and off.

As shown in Fig. 1, the phosphorus NMR spectra of human whole, submandibular/sublingual and parotid salivas all presented a single broad line. The signal intensities increased in the order: parotid> whole> submandibular/sublingual, with the ratio 2.3, 1.2 and 1.0. The chemical shifts were almost the same, although they changed slightly as a function of saliva pH. Although earlier work had shown the three types of saliva to have significantly different proton NMR spectra due to viscosity effects, little difference was here observed in their ³¹P NMR line-widths.

Fig. 1. ³¹P NMR spectra of human (a) parotid, (b) submandibular/sublingual, and (c) whole saliva measured at 145 MHz and at 303 K. Chemical shifts and signal intensities were referenced to 85% phosphoric acid; 40 transients were collected. ppm=parts/10⁶.

Fig. 2 shows the semilogarithmic plots of (a) spin–lattice and (b) spin–spin relaxation of parotid saliva measured at 303 K. Mz(?) and MZ(t) represent the magnetization in the equilibrium state and at the time t after the flipping pulse. The magnetizations are directly proportional to the peak intensities measured at the time t after the flipping pulse. When relaxation is described by a single time constant, a straight line is obtained. The straight line for spin–lattice relaxation obtained here indicates that the decay of signal intensity follows the expected single exponential, whereas the spin–spin relaxation could not be described by a single constant. Three major mechanisms are considered to be important in ³¹P relaxation: (1) dipolar interaction with neighbouring protons; (2) chemical-shift anisotropy; and (3) interaction with paramagnetic species. These interactions are coupled with molecular motions to cause phosphorus relaxation in solution. Paramagnetic metal ions and/or dissolved oxygen contained in saliva can often affect the relaxation. However, as we actually observed a large NOE (50%) for parotid saliva, the contribution of the paramagnetic species and chemical-shift anisotropy could be neglected as compared with that of the dipolar interaction.

Fig. 2. Semilogarithmic plots for (a) the spin–lattice and (b) the spin–spin relaxation results obtained for ³¹P in human parotid saliva measured at 145 MHz and at 303 K. Mz(?) and Mz(t) represent the magnetization in the equilibrium state and at the time t after the flipping pulse, respectively.

Fig. 3 shows the temperature dependence of T₁ and T₂ of ³¹P of parotid saliva. Working on the assumption that the relaxation plot displayed in Fig. 2(b) consists of two components (a and b), it was estimated that there were two sets of T₂ (T_2a and T_2b) at each temperature, wherein each set corresponds to one of the components. When relaxation is determined by a dipolar interaction, T₁ goes through a minimum at ?_c?_p1, where ?_c and ?_p are a correlation time and an observing frequency, respectively. In the region of short ?_c (?_c?_p<<1) T₁ decreases with increasing ?_c, while in the region of long ?_c (?_c?_p>>1) it increases with increasing ?_c. On the other hand, T₂ is observed to decrease with increasing ?_c in the whole region. As ?_c is known to increase with decreasing temperature, the results shown in Fig. 3 for ³¹P of parotid saliva indicate that the T₁ are in the short ?_c region (?_c<<10^?9 sec) within the temperature region measured. Most of the phosphate in saliva is inorganic. It is noteworthy that organic phosphates such as phosphoproteins are also present in saliva. However, by examination of the rapid rotational motion (?_c<<10^?9 sec), the observed ³¹P resonance can be confidently attributed to inorganic phosphorus compounds of small molecular weight, as opposed to the phosphoproteins.

Fig. 3. Semilogarithmic plots of T₁, T_2a and T_2b of ³¹P in human parotid saliva against the reciprocal of the absolute temperature (?). =T₁, •=T_2a, ?=T_2b, (303 K corresponds to ?=3.3×10^?3/K).

T₂ decreases with increasing temperature, which is the opposite tendency of that stated above. In the case of slow chemical exchange, this contributes to the spin–spin relaxation rate. When the contribution of the chemical exchange rate to T₂^?1 is dominant, the temperature dependence is the reverse of that when the dipolar term is dominant. Hence T₂ of phosphorus in parotid saliva appears here to be determined by the chemical exchange rate. Using the T₂ values deduced from Fig. 3, we estimate the apparent exchange rates at 303 K to be 0.2/sec (a) and 1.5/sec (b), respectively, and from the slopes of T_2a and T_2b, we estimate the apparent activation energy ?E to be 20.2 and 5.8 Kcal/mol, respectively.