1 Introduction
To date, many studies concerning the clinical evaluation of ozone (O3) have, unfortunately, been based on assessments of its harmful effects rather than demonstrating any therapeutic benefits that it may offer 1-4. In the Earth’s biosphere, the role of O3 is controversial. Indeed, it is one of nature’s most powerful oxidants and is reactive towards many biomolecules 5 This ‘reactive oxygen species’ (ROS) can be produced either by ss ultra-violet rays of the sun, or artificially with an ozone generator. In the former case, O3 is the major constituent of the atmosphere responsible for shielding off ultra-violet light from the sun and oxidising pollutants in the air. O3 can also be usefully employed by man for purposes of water purification and sewage treatment.

O3 was first suggested as a disinfectant for drinking water in the 19th Century in view of its powerful ability to inactivate micro-organisms 6 Indeed, many researchers have emphasised that the utilisation of this oxidant over limited time periods can effectively disinfect water supplies 7-9 It acts rapidly and in lower concentrations when compared to chlorine, and has no side effects such as taste and odour which are characteristic of other disinfecting agents 10. However, high levels of O3 were found to be insufficient for disinfection in view of the limited solubility of this gas in water and adverse toxicological problems arising from the employment of such high concentrations 11-13 O3 is an agent that disinfects by destroying, neutralising or inhibiting the growth of pathogenic micro-organisms, and has been found to be an effective alternative biocidal agent to chlorine 14 The treatment of drinking water with ozone has been shown to be more efficient against spores of Bacillus subtilis, a 0.35 mg/l concentration giving rise to the reduction of at least 5 log values in populations of approximately 1 ? 106 cells/ml of Escherichia coli, Vibrio cholerae, Salmonella typhi, Yersinia enterocolitica, Pseudomonas aeruginosa, Aeromonas hydrophila, Listeria monocytogenes and Staphylococcus aureus when compared with 0.50 mg/l chlorine (with the exception of Vibrio cholerae). Both disinfectants were probably consumed during the treatment period 12 Indeed, Broadwater et al.7 agreed with Morris 13 and indicated that O3 is an effective bactericide against both vegetative cells, and the B. megaterium and B. cereus spores of E. Coli ATCC 9677. However, on consideration of the level of organic components present in water, they suggested that this ROS can be applied at higher dosages (0.5-10 mg/l) and for longer contact periods (2-10 min.) for practical applications since, having a potent oxidising activity ( redox potential ca. +1.6 V ) ozone is consumed by such agents, a phenomenon that prevents full utilisation of the applied dose as a disinfectant 13.

Historically, O3 has been the agent of choice for the disinfection of public water supplies rather than chlorine in the United States of America. The reason for the preferential application of O3 here is primarily to remove iron and manganese ion colourations, and to avoid adverse tastes and odours 9. However, one major disadvantage is that O3 is stable only for a short period of time and decomposes to form molecular oxygen which is utilised for sustenance by aquatic life 13 and hence it disappears very quickly from the system 15. Consequently, it is impossible to maintain a residual level of this ROS in a water distribution system. However, Burleson et al.16 found the application of a simultaneous treatment of O3 with sonication to be more effective in a bactericidal treatment process designed for the disinfection of waste-water. Sonication reduced or altered oxidisable organic material as measured by biological oxygen demand and chemical oxygen demand determinations, and therefore the O3 demand of the secondary effluent was also reduced 11. This may also enhance the O3-mediated inactivation of micro-organisms by ozone via dissipation of particulate organic material and clusters of bacteria, a process facilitating their exposure to the oxidant 14 With pre-treated water, ozone can kill bacteria by rapidly rupturing their cell membranes (within 2 seconds). Chlorine, on the other hand, simply diffuses into the cell and requires 30 min. to achieve bactericidal effects. The rate of lysis is dependent upon the ozone level: higher ozone concentrations are used for highly contaminated systems, whereas lower concentrations are used for maintenance. Following O3-mediated lysis of the micro-organisms, the cytoplasmic constituents are oxidised by this agent, which is then removed by UV irradiation at a wavelength of 254 nm 15 O3 has also been employed commercially to remove trace organic contaminants from water.

In view of its powerful oxidising actions, ozone has a very rich chemistry and the oxidation of critical biomolecules undoubtedly accounts for its broad-spectrum biocidal properties. Indeed, ozone can attack a very wide variety of biomolecules 5, 17, for example free or protein-incorporated amino acids such as cysteine, methionine, histidine and tyrosine, carbohydrates, phenolic adducts and, of course, its well-characterised ozonylation of carbon-carbon double bonds, e.g., those of polyunsaturated fatty acids (PUFAs). Oxidation of PUFAs by ozone gives rise to the production of conjugated hydroperoxydiene and subsequently aldehydic species 18, the latter apparently serving as biomarkers of ozonylation 19, 20. Interestingly, reaction of water-soluble, single electron-donors with ozone primarily generates the ozone radical anion, a transient adduct which, on protonation, decomposes to hydroxyl radical and dioxygen 21. Hence, some of the reaction products which putatively arise from the interactions of ozone with molecules present in tissues and biofluids are, at least in principle, identical to those produced from the attack of hydroxyl radical on such ROS scavengers.

Currently, root caries represents a challenging problem to the dental profession in view of a substantial increase in the population of elderly patients during the late 20th century. This condition is primarily ascribable to tooth demineralisation processes induced by organic acids (e.g., lactic and pyruvic acids) generated by bacteria, predominantly Strptococcus mutans 22, and a recent investigation conducted by Baysan et. al 23 revealed that ozone exerts a powerful bactericidal action towards this pathogen, together with further micro-organisms associated with primary root carious lesions. Indeed, the application of ozone in the dental practices may serve as a viable, cost-effective and convenient means of treating dental caries, and it is conceivable that it could eventually replace conventional ‘drilling-and-filling’ procedures currently employed by dental surgeons.

In view of the clear indications for the therapeutic application of ozone in the treatment of periodontal conditions, a clinical trial involving this agent has recently been approved by the US Food and Drug Administration (FDA). Therefore, in this investigation we have employed high resolution proton (1H) nuclear magnetic resonance (NMR) spectroscopy to achieve a multicomponent evaluation of the oxidising actions of ozone towards biomolecules present in intact human saliva. The therapeutic, aesthetic and biochemical significance of the results acquired are discussed in detail.

2.1 Sample Collection and Preparation
Unstimulated human saliva samples were obtained from a total of 12 healthy volunteers (7 male, 5 female). Subjects were seated comfortably and then asked to collect all saliva into a cup for a period of 10 min. Immediately after collection, all samples were centrifuged at 16,000 ? g for 30 min. (4?C) to remove debris.

Aliquots (5.00 ml) of the above salivary supernatant samples were removed and each of them was divided into two equivalent portions (2.50 ml). The first of these was treated with O3 generated by the HealOzone Unit (CurOzone, USA) for a period of 10 s (equivalent to a delivery of 4.484 mmol. of this oxidant) and then equilibrated at a temperature of 37 C for a 30 min. period prior to NMR analysis ; the second group of portions served as untreated controls.

2.2 Proton NMR Measurements
Proton (1H) NMR measurements on the above samples were conducted on a Bruker AMX-600 spectrometer (ULIRS, Queen Mary, University of London facility, U.K.] operating at a frequency of 600.13 MHz and a probe temperature of 298 K. Typically, 0.60 ml of sample was placed in a 5-mm diameter NMR tube, and 0.07 ml of 2H20 was added to provide a field frequency lock.

The intense water signal (? = 4.80 ppm) was suppressed by presaturation via gated decoupling during the delay between pulses. Pulsing conditions for one-dimensional (1-D) spectra acquired on salivary supernatant samples were: 64 or 128 free induction decays (FIDS); 16,384 data points; 3-7 ?s pulses; 1.0 s pulse repetition rate. Line-broadening functions of 0.30 Hz were utilised for the processing of experimental NMR data. For selected salivary supernatant specimens, the broad protein resonances present in control and dentifrice supernatant-treated salivary supernatant samples were suppressed by the Hahn spin-echo sequence (D[90?x-t-180?y-t-collect]) 24 which was repeated 128 times (t = 68 ms). Chemical shifts were referenced to external sodium 3-trimethylsilyl [2,2,3,3-2H4] propionate (TSP; ? = 0.00 ppm). Where present, the methyl group resonances of alanine (? = 1.487 ppm) and lactate (? = 1.330 ppm) served as secondary internal references for the saliva samples examined.

The identities of biomolecule resonances present in the salivary 1H NMR spectra acquired were routinely assigned by a consideration of chemical shift values, coupling patterns and coupling constants. The relative intensities of selected signals therein were determined by electronic integration, and the concentrations of components detectable were determined by comparisons of their resonance areas with that of a 42.0 mM standard solution of TSP located within a coaxial NMR tube insert. This procedure was employed to avoid broadening of the TSP resonance which arises from its binding to salivary proteins or alternative macromolecules. Where required, salivary metabolite concentrations were determined by electronic integration of their 1H resonances and expressing their intensities relative to that of the ‘internal-but-isolated’ TSP standard. In view of the protein concentration of human saliva (1.40-6.40 g/l in resting specimens, mean value 2.20 g/l) 25, which is much lower than that of human blood serum (65-83 g/l) 26, the minimal broad macromolecule resonance envelope in single-pulse (one-dimensional) 1H NMR spectra was found not to interfere with the observation and integration of the sharp pyruvate and actetate signals, and hence all salivary organic acid anion concentrations documented here represent those obtained from such spectra. Of course, the levels given reflect only the non-macromolecule-bound fraction of these biomolecules and therefore are expected to be somewhat lower than the total salivary concentrations. Indeed, it has been established that an ‘NMR-invisible’ pool of protein-bound metabolites such as lactate can be liberated from macromolecular binding sites by the addition of high levels of ammonium chloride (? 0.5 M) to biofluids 27.

Two-dimensional shift-correlated 1H-1H NMR (COSY) spectra of human salivary supernatants were acquired on the Bruker AMX-600 facility using the standard sequence of Aue et. al 28 , with 2,048 data points in the t2 dimension, 256 increments of t1, a 3.00 s relaxation delay, and 24 transients.

Total correlation spectroscopy (TOCSY) spectra were acquired on these samples using 2,048 data points in the t2 dimension, 512 increments of t1, a spin lock of 50 ms, a recycle delay of 1.22 s, and 64 transients ( also Bruker AMX-600 facility).

Aqueous solutions containing sodium pyruvate, L-methionine, L-cysteine and B-D-glucose (1.00 or 5.00 mM) (Sigma Chemical Co., Poole, Dorset, UK) were prepared in 40.0 mM phosphate buffer (pH 7.00) which was rigorously deoxygenated with argon gas prior to use (30 min. at ambient temperature). 5.00 ml. aliquots of these solutions were treated with ozone and then equilibrated as described above for salivary supernatant specimens. Matching de-oxygenated solutions of these agents untreated with ozone but equilibrated in the same manner served as controls.

For one-dimensional (1-D) 600 MHz proton NMR spectra acquired on the simple chemical model systems described above (section 2.1), typical pulsing conditions were 64 FIDs using 32,768 data points, 72 degree pulses and a 3s pulse repetition rate to allow full spin-lattice relaxation of the hydrogen nuclei in the samples investigated. Exponential line-broadening functions of 0.30 Hz were routinely employed for purposes of processing.