Free Radical Biology & Medicine 41 (2006) 1404–1412www.elsevier.com/locate/freeradbiomed

Original Contribution

Intragastric generation of antimicrobial nitrogen oxides fromsaliva—Physiological and therapeutic considerations

Håkan Björne a,b,⁎, Eddie Weitzberg a,b, Jon O. Lundberg a

a Department of Physiology and Pharmacology, Karolinska Institute, 177 76 Stockholm, Swedenb Section of Anesthesiology and Intensive Care, Karolinska Institute, 177 76 Stockholm, Sweden

Received 31 May 2006; revised 11 July 2006; accepted 25 July 2006Available online 27 July 2006

Abstract

Salivary nitrite is suggested to enhance the antimicrobial properties of gastric juice by conversion to nitric oxide (NO) and other reactivenitrogen intermediates in the stomach. Intubated patients exhibit extremely low gastric levels of NO, because they do not swallow their saliva. Thepresent investigation was designed to examine the antibacterial effects of human saliva and gastric juice. Furthermore, we studied a new mode ofNO delivery, involving formation from acidified nitrite, which could prevent bacterial growth in the gastric juice of intubated patients in intensivecare units. The growth of Escherichia coli ATCC 25922 and the formation of NO and nitroso/nitrosyl species were determined after incubation ofgastric juice with saliva from healthy volunteers that was rich (nitrate ingestion) or poor (overnight fasting) in nitrite. In a stomach modelcontaining gastric juice from intubated patients, we inserted a catheter with a silicone retention cuff filled with ascorbic acid and nitrite anddetermined the resulting antibacterial effects on E. coli and Candida albicans. Saliva enhanced the bactericidal effect of gastric juice, especiallysaliva rich in nitrite. Formation of NO and nitroso/nitrosyl species by nitrite-rich saliva was 10-fold greater than that by saliva poor in nitrite. Inour stomach model, E. coli and C. albicans were killed after exposure to ascorbic acid and nitrite. In conclusion, saliva rich in nitrite enhances thebactericidal effects of gastric juice, possibly through the generation of reactive nitrogen intermediates, including NO. Acidified nitrite inside a gas-permeable retention cuff may be useful for restoring gastric NO levels and host defense in critically ill patients.© 2006 Elsevier Inc. All rights reserved.

Keywords: Nitric oxide; Nitrite; Saliva; Ventilator-associated pneumonia; Ulcus; Antibacterial; Free radicals

The human stomach is constantly exposed to a variety ofnoxious stimuli, including bacteria, and the acidity of the gastricjuice is considered to constitute a major defense mechanismagainst ingested pathogens [1,2]. Furthermore recent studiessuggest that the antimicrobial effect of a low pH can be stronglyenhanced by the presence of inorganic nitrite [3–6], probably viapH-dependent conversion of this ion to nitric oxide (NO) andother nitrogen specieswith antibacterial properties. These findingsare of considerable interest because humans and other mammalscontinually swallow their own saliva, which contains nitrite.

Nitrite is produced by reduction of the nitrate both present inthe diet and formed from endogenous NO synthesized

Abbreviations: CFU, colony-forming units; NO, nitric oxide; NO2, nitrogendioxide; N2O3, dinitrogen trioxide; NO2

−, nitrite.⁎ Corresponding author. Fax: +46 8 307795.E-mail address: hakan.bjorne@karolinska.se (H. Björne).

0891-5849/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.freeradbiomed.2006.07.020

enzymatically and is actively taken up from the bloodstreamby the salivary glands and concentrated in the saliva. In order toobtain energy, commensal bacteria present in the oral cavityconvert this salivary nitrate to nitrite, which is reduced further toNO and other reactive nitrogen oxides in the acidic gastricenvironment [7].

NO is a pluripotent mediator of biological processes and,thus, this enterosalivary circulation of nitrate and its subsequentconversion to NO may be of significance not only in killingingested bacteria, but also in connection with other gastro-protective mechanisms, such as regulation of mucosal bloodflow and of gastric mucus production [8,9].

In addition, production of NO in humans independent of NOsynthase was first discovered in the stomach [3,10], but latershown to occur in the heart, skin, urinary tract, and mouth aswell [11–14]. It is now clear that the large pool of nitrite presentthroughout the body is a potential source of NO and other

1405H. Björne et al. / Free Radical Biology & Medicine 41 (2006) 1404–1412

nitrogen oxide species, a fact which may have majorimplications for cardiovascular physiology [15,16].

We have recently demonstrated that critically ill patientsreceiving mechanical ventilation in intensive care units haveextremely low levels of gastric NO because of their inability toswallow saliva [17]. Moreover such patients often receiveprophylactic treatment to prevent the development of stress-related ulcers, treatment which further increases the risk for gastriccolonization by bacteria and fungi. The stomach can then serve asa reservoir ofmicroorganisms that cause severe infections, such aspneumonia associated with being on a ventilator [18]. However,the level of NO in the stomachs of these patients can be restored tonormal by intragastric infusion of nitrite [17].

Although, mentioned above, nitrite has been shown toenhance the antimicrobial effects of various acidic solutionsagainst numerous pathogens, including enterobacter, yeast, andhelicobacter [3–6,19], the antibacterial effects of actual humansaliva and gastric juice have, to our knowledge, not yet beenexamined. Therefore, the aim of the present investigation was tocharacterize these effects of nitrite-containing human salivamixed with gastric juice. In addition, we have developed astomach model which allowed us to examine whether thegrowth of bacteria in gastric juice obtained from intensive carepatients could be influenced by a new mode of NO delivery.

Materials and methods

Bacterial cultures

The nonpathogenic Escherichia coli strain ATCC 25922(obtained from the Department of Clinical Microbiology,Karolinska University Hospital, Stockholm, Sweden) wascultured on nutrient agar plates and subsequently incubatedon Mueller–Hinton broth (pH 7.5) for 18 h at 37°C to give 108–109 colony-forming units (CFU)/ml the day before commence-ment of the experiments.

Collection of saliva and gastric juice

The saliva from 10 healthy, nonsmoking male volunteers,29–41 years of age and fasted overnight, was collected on threedifferent occasions. This collection was performed before and1 h after ingestion of sodium nitrate (10 mg/kg, correspondingto approximately 300 mg of spinach, dissolved to 150 mlwater), providing saliva that was poor and rich in nitrite,respectively. The saliva samples were pooled, divided intobatches, and stored at −20°C until use. In order to obtain salivathat was virtually depleted of nitrite, the saliva from fastingvolunteers was stirred in the open air for 24 h to allow oxidationof the nitrite remaining.

Gastric juice from these same volunteers was collected byaspiration through a nasogastric tube and then pooled and storedat −20°C. In order to prevent dilution of these samples thevolunteers were asked not to swallow any saliva for 30 minbefore collection of the gastric juice.

In experiments involving the stomach model, we employedgastric juice from 20 unconscious, intubated patients (19 male, 1

female; 22–74 years of age) under intensive care at aneurosurgery unit. All of these patients received sucralfate toprevent the development of stress ulcer. None of them wereadministered drugs that inhibit acid secretion or exhibitedclinical signs of gastric bleeding. Six hours after discontinuationof any enteral feeding, the gastric juice was obtained byaspiration through their nasogastric tubes. Samples from 10patients were thereafter stored at −20°C for experimentsemploying E. coli ATCC 25922. Before the actual experiments,these samples were cultured for 24 h on agar plates and, if nobacterial growth was detected, pooled and used for our studies.Samples from the remaining 10 patients were immediatelydivided into three aliquots for experiments involving existingpathogens.

Effects of gastric juice containing saliva on bacterial growth

Dose-dependent effects of sodium nitriteAll of these experiments were performed in sterile 10-ml

glass tubes. Two milliliters of nitrite-depleted saliva with anitrite content of <20 μM and 2 ml of gastric juice fromhealthy volunteers were mixed and this solution was thenadjusted to a final pH of 2, 3, 4, or 5 with hydrochloric acid.Thereafter, E. coli ATCC 25922 culture (40 μl) was added tothis mixture to obtain a final density of 1–1.9×107 bacteria/ml,after which various amounts of sodium nitrite were added togive final concentrations of <10, 100, or 1000 μM. Thetubes were then immediately sealed and incubated at 37°Cfor 10 and 30 min and 2 and 6 h, at which times 10 μl ofeach sample was removed, serially diluted with PBS (pH7.3), and transferred to standard blood agar plates.Additional experiments were performed using test tubesthat were not sealed, thus allowing the escape of NO. Afterfurther incubation for 24 h at 37°C, the numbers of colony-forming units were counted, with the lowest detectable countbeing 100 CFU/ml. A bactericidal effect was considered tobe present when at least 99.9% of the bacteria originallyinoculated were killed. All of these experiments wereperformed in quadruplicate.

The effects of saliva collected after ingestion of nitrateTwo milliliters of the saliva collected after oral loading with

nitrate ([NO2−] 3300 μM) or fasting overnight ([NO2

−] 340 μM)was added to sterile glass tubes and inoculated with 40 μl ofbacterial culture. Immediately afterward 2 ml of gastric juiceobtained from healthy volunteers and adjusted to a final pH of 3or 4 with hydrochloric acid was added, the test tubes weresealed, and incubation and determination of bacterial growthwere performed as described.

Quantitation of NO and nitroso/nitrosyl species from acidifiedsaliva

The influence of nitrate ingestion on the generation of NOand nitroso/nitrosyl species (defined here as the sum ofnitrosothiols, N-nitrosamines, and iron-nitrosyl species) byacidified saliva was also examined. Two milliliters of nitrite-

1406 H. Björne et al. / Free Radical Biology & Medicine 41 (2006) 1404–1412

rich (1220 μM) saliva or saliva low in nitrite (73 μM) wasmixed with 2 ml of gastric juice to a final pH of 3 or 5 (n = 4)and these samples were then incubated at 37°C in plasticsyringes with a headspace of 6 ml (room air with a NOcontent of <5 ppb). After 30 min, the headspace gas wasremoved, diluted 1:20 with NO-free air, and injected into achemiluminescence NO analyzer, from which peak valueswere recorded. For quantitation of the nitrite and nitroso/nitrosyl species in the liquid phase after incubation, themixtures were placed in Eppendorf tubes containing N-ethylmaleimide and EDTA at final concentrations of 5 and2 mM, respectively, and stored at −80°C until being analyzed.The levels of nitrite and nitroso/nitrosyl species in thissamples were determined by chemiluminescence assay of theNO released into the gas phase by reductive cleavage, asdescribed previously [16,20].

The kinetics of NO formation

In the case of kinetic experiments, 2 ml of saliva with a high(530 μM) or low (70 μM) nitrite content was mixed with 2 ml ofcitrate phosphate buffer, pH 3.2 (final pH 4), in a purged vessel(50 ml) with an outlet and an inlet, which was then placed on amagnetic stirrer. The headspace gas was sampled continuouslyat a rate of 140 ml/min for 4 h and analyzed online for NO bychemiluminescence. The inlet allowed free entry of room air(NO <5 ppb).

The stomach model

Effects of acidified nitrite on E. coli ATCC 25922First, airtight plastic bags were filled with 25 ml of pooled

gastric juice from patients (Fig. 1), after which the acidity of this

Fig. 1. The stomach model employed an all-silicone catheter fitted with aretention cuff filled with a nitrite (NO2

−)/ascorbic acid (AA) solution at a set pH(H+), which is inserted into an airtight plastic bag filled with gastric juiceinoculated with bacteria. The nitric oxide (NO) formed from the reduction ofnitrite diffuses through the silicone membrane of the cuff into the lumen of thebag.

juice was adjusted to a final pH of 3, 4, 5, or 6 with hydrochloricacid, and E. coli ATCC 25922 was inoculated to give a finaldensity of 1–2×107 CFU/ml. Next, an all-silicone urinarycatheter (Argyle, Sherwood Medical, Tullamore, Ireland) fittedwith a 10-ml retention cuff filled with a solution of ascorbic acid(10 mM) in saline adjusted to pH 2 with hydrochloric acid wasinserted into the bag and the opening sealed. Subsequently, afreshly prepared solution of sodium nitrite (0.2 M) was injectedinto the retention cuff to obtain a final nitrite concentration of10 mM, with ascorbic acid solution (10 mM, pH 2) alone beingutilized as the control. After 2, 6, and 24 h of incubation at37°C, the numbers of viable bacteria and levels of nitroso/nitrosyl species in the gastric juice were measured as describedabove.

Effects of acidified nitrite on pathogensThe pH of the sample was determined, whereafter each

sample was directly divided into three aliquots (10 ml), whichwere injected into airtight bags. The first aliquot was exposedto a solution of nitrite (10 mM) and ascorbic acid inside asilicone catheter retention cuff as described above, the secondwas mixed with a solution of sodium nitrite (100 μl, 0.1 M;final concentration 10 mM), and the third aliquot served ascontrol. The samples were incubated at 37°C and freshsolutions of nitrite were added after 12 h. After 24 h ofincubation the samples were processed by the ClinicalMicrobiology Laboratory, Karolinska University Hospital,according to standard procedures for quantification andidentification of species.

Formation of NO and NO2

In a separate set of experiments formation of NO and NO2

was measured in such airtight plastic bags filled with 25 ml ofeither gastric juice from patients or sterile water (both at pH 5)or 5 ml of human venous blood and insufflated with 1 liter ofambient air. After 5 min of exposure to a solution of ascorbicacid (10 mM) and sodium nitrite (10 mM) inside the catheterretention cuff, NO and NO2 in the gas phase were quantified bychemiluminescence (see below).

Determination of NO and NO2

For most NO measurements, a chemiluminescence analyzer(CLD 77 AM; Eco Physics, Dürnten, Switzerland) was usedwith a sample flow rate of 140 ml/min. In certain experimentsNO and NO2 were measured simultaneously using thechemiluminescence analyzer CLD 700 AL (flow rate700 ml/min; Eco Physics). Before each experiment, theseanalyzers were calibrated with NO (10 ppm; AGA AB,Lidingö, Sweden).

Statistical analysis

All data are expressed as medians (interquartile ranges). TheMann–Whitney U test was utilized for comparison betweengroups and the p value thus obtained corrected for multiplecomparisons. A p value of less then 0.05 was considered to bestatistically significant.

Fig. 3. Time-dependent growth of E. coli ATCC 25922 in a mixture of gastricjuice and saliva poor or rich in nitrite collected after overnight fasting or nitrateingestion (10 mg/kg), respectively, at pH 3 and 4. The data shown are medians(interquartile ranges) of four independent experiments. The dotted lines indicatea 99.9% (1000-fold) decrease from the initial number of colony-forming units,which was defined here as a bactericidal effect.

1407H. Björne et al. / Free Radical Biology & Medicine 41 (2006) 1404–1412

Results

The antibacterial effects of saliva and gastric juice

The dose-dependent effects of sodium nitrite (Fig. 2)In gastric juice containing saliva pH 2, E. coli bacteria were

killed within 10 min irrespective of the level of nitrite present(Fig. 2). In contrast, at pH 5 only a minor effect on bacterialgrowth was observed and only at the highest (1000 μM)concentration of sodium nitrite tested. Inhibition of the growthofE. coli in a mixture of saliva and gastric juice mixture at a pH of3 or 4 was dose-dependent. At pH 3, 100 μM nitrite resulted in abactericidal effect within 30 min, whereas at pH 4, 6 h ofincubation in the presence of 1000 μM sodium nitrite wasrequired to obtain such an effect. The bactericidal effect of gastricjuicemixedwith nitrite-depleted saliva (nitrite level <10μM)wassimilar to that of gastric juice alone, and therewas no difference inbacterial growth using sealed or open tubes (data not shown).

The influence of nitrate ingestion (Fig. 3)The nitrite concentrations in saliva collected after fasting

overnight and after nitrate ingestion were 340 and 3300 μM,respectively, and these concentrations decreased by approxi-mately 50% due to dilution, after mixing with gastric juice (Fig.3). Saliva poor or rich in nitrite exhibited bactericidal effectssimilar to those of nitrite-depleted saliva supplemented with 100or 1000 μM sodium nitrite, respectively. At pH 3, the salivacollected after fasting was bactericidal, whereas saliva collectedafter nitrate ingestion was required for bactericidal effect at pH 4.

Fig. 2. Time-dependent growth of E. coli ATCC 25922 in a mixture of gastric juice1000 μM) at pH 2, 3, 4, and 5. The data shown are medians (interquartile ranges) of foudecrease from the initial number of colony-forming units, which was defined here a

Formation of NO and nitroso/nitrosyl species by gastric juicemixed with saliva

NO and nitroso/nitrosyl species (Fig. 4)In these experiments, the nitrite concentrations in saliva after

fasting overnight and after nitrate ingestion were 73 and1220 μM, respectively, and these concentrations were decreasedby approximately 50% after mixing with gastric juice (Fig. 4).

and saliva depleted of (<10 μM) or supplemented with sodium nitrite (100 andr or five independent experiments. The dotted lines indicate a 99.9% (1000-fold)s a bactericidal effect.

Fig. 5. Continuous measurement of the headspace levels of nitric oxidegenerated by acidified saliva. Saliva (2 ml) collected after ingestion of nitrateload (nitrite concentration 530 μM) or after overnight fasting (73 μM) wasmixed with citrate phosphate buffer (2 ml; final pH 4) in a purged vessel (50 ml)and stirred continuously. Headspace gas was sampled at a rate of 140 ml/min andits content of NO determined online by chemiluminescence.

1408 H. Björne et al. / Free Radical Biology & Medicine 41 (2006) 1404–1412

After 30 min of incubation, the peak level of NO in theheadspace was 10-fold higher with nitrite-rich than with nitrite-poor saliva, both at pH 3 (319 (283–364) and 43 (41–50) ppm,respectively) and at pH 5 (23.6 (16.5–28.3) and 3.1 (2.3–4.5)ppm, respectively). The same relationship was observed withrespect to the formation of nitroso/nitrosyl species at both pH 3(875 (665–1140) and 71 (56–90) nM, respectively) and pH 5(71 (43–108) and 25 (10–39) nM, respectively).

The kinetics of NO formation (Fig. 5)In this case the salivary nitrite concentration was 530 μM

after nitrate ingestion and 70 μM after overnight fasting. At pH4, formation of NO by these samples was detected immediatelyin the headspace of the purged vessel, reaching a maximumwithin 5 min, after which this level declined exponentially (Fig.5). The maximal level of NO generated by nitrite-rich saliva(1700 ppb) was more than threefold higher than the correspond-ing value for saliva poor in nitrite (470 ppb) and the higher levelof NO in the former case was sustained throughout the 4-hobservation period.

The stomach model

Antibacterial effects on E. coli ATCC 25922 (Fig. 6)Exposure of E. coli ATCC 25922 in gastric juice to a mixture

of ascorbic acid and sodium nitrite resulted in effective killingwithin 6 h at pH 3–5, whereas ascorbic acid alone had nobactericidal effect under the same conditions (Fig. 6). Only after24 h of exposure and at pH 3 could a bactericidal effect be

Fig. 4. Levels of NO (top) and nitroso/nitrosyl species (bottom) after incubation(30 min, 37°C) of saliva poor or rich in nitrite (collected after fasting or nitrateingestion, respectively) and gastric juice at pH 3 and 5. The data presented aremedians (interquartile ranges) (n=4). *p < 0.05.

obtained with ascorbic acid alone. No effects were seen at pH 6in either case.

Antimicrobial effects on pathogens (Fig. 7)Gastric juice was colonized with Candida albicans in seven

patients and with E. coli in two patients. The pH of the samplesranged from 2.2 to 4.8. After 24 h of exposure to a solution ofascorbic acid and sodium nitrite inside a retention cuff,antimicrobial effects were seen in all samples (Fig. 7). Similarresults were obtained when a sodium nitrite solution wasdirectly mixed with gastric juice (data not shown).

Formation of NO, NO2, and nitroso/nitrosyl species (Fig. 8)The levels of NO and NO2 in the gaseous compartment of

our stomach model, after exposure to a mixture of ascorbic acidand sodium nitrite in the retention cuff, were dependent on thecomposition of the fluid compartment (Fig. 8). In all cases thelevel of NO exceeded that of NO2. The highest levels of thesetwo gases were observed in the presence of water, whichresulted in twice as much NO and 10 times as much NO2 as didgastric juice. The presence of venous blood in the liquidcompartment markedly reduced the levels of both NO and NO2.

The level of nitroso/nitrosyl species in the pooled gastric juicewas 0.7 μM before exposure to a mixture of nitrite/ascorbic acidin the retention cuff. After 24 h this level had increased to15.6 μM (8.7–32.7, p < 0.004), whereas it remained unchangedin the presence of ascorbic acid alone (0.5 μM (0.1–1.4)).

Discussion

Here we have demonstrated that nitrite in human salivaenhances the antibacterial effect of gastric juice in a dose-dependent manner. After ingestion of nitrate, salivary nitritelevels increased markedly and this nitrite was converted rapidlyto NO, NO2, and nitroso/nitrosyl species in gastric juice. Theantibacterial effect of nitrite-rich saliva was most evident at thepH levels normally present in the stomach after food intake, inconnection with which there is a risk for parallel ingestion ofbacteria. These findings suggest that a nitrate-rich diet including,

Fig. 6. Survival of E. coli ATCC 25922 in gastric juice with and without exposure to nitrite, as examined employing a stomach model. A silicone catheter fitted with aretention cuff was inserted into an airtight plastic bag (see Fig. 1) and the retention cuff then filled (10 ml) with a mixture of sodium nitrite (10 mM) and ascorbic acid(10 mM) (pH 2) or ascorbic acid alone (pH 2). The numbers of viable bacteria remaining after various periods of exposure (2, 6, and 24 h) at various pH (3, 4, 5, and 6)were determined. The data presented are medians (interquartile ranges) of four or five independent experiments. The dotted lines indicate a 99.9% (1000-fold) decreasefrom the initial number of colony-forming units, which was defined here as a bactericidal effect.

1409H. Björne et al. / Free Radical Biology & Medicine 41 (2006) 1404–1412

e.g., vegetables, may enhance gastric defense mechanismsagainst swallowed pathogens. We also describe a novel systemfor administration of nitrogen oxides into the gastric lumen thatmay have clinical implications.

The pH of the gastric juice in healthy fasting individuals isnormally 1.5–2.5 [1]. The bactericidal effects observed at pH 2in the present study were independent of salivary nitrite content,

Fig. 7. Survival of the existing pathogens C. albicans and E. coli in gastric juicefrom intubated patients, with and without exposure to acidified nitrite in astomach model. A silicone catheter fitted with a retention cuff was inserted intoan airtight plastic bag (see Fig. 1) and the retention cuff was filled (10 ml) with amixture of sodium nitrite (10 mM) and ascorbic acid (10 mM; pH 2) or ascorbicacid alone (control) at the beginning of the experiment and after 12 h ofexposure. The numbers of viable microorganisms remaining after 24 h ofexposure (37°C) were determined.

indicating that this acidity itself is sufficient to kill the bacteria.In contrast, at pH 5 E. coli survived exposure to gastric juiceeven in the presence of high levels of salivary nitrite. Afteringestion of food, the gastric pH may initially increase to >5, butwill then gradually decrease again. At pH values of 3 and 4, adose-dependent effect of salivary nitrite on bactericidal activitywas evident here.

In connection with a meal production of saliva is stimulated,which will probably lead to a net increase in the amount ofnitrite delivered to the stomach. Moreover, if the meal itself

Fig. 8. Levels of nitric oxide (NO) and nitrogen dioxide (NO2) in a stomachmodel containing gastric juice (25 ml, pH 5), water (25 ml, pH 5), or venousblood (5 ml) after exposure to a solution of sodium nitrite/ascorbic acid (10 mM/10 mM; pH 2). The data shown are medians (interquartile ranges) (n=4).

1410 H. Björne et al. / Free Radical Biology & Medicine 41 (2006) 1404–1412

contains nitrate, the nitrite content of the saliva will be elevatedeven further. In this way the amount of nitrite entering thestomach will be potently increased when it is most needed, i.e.,after food intake, which may involve simultaneous ingestion ofbacteria when the gastric pH is increased.

Humans produce 1–1.5 liters of saliva each day, most ofwhich continuously enters the acidic stomach. After ingestion ofnitrate, plasma and salivary levels of nitrate both increaserapidly, with peak concentrations being reached within 60 min[21]. For reasons that remain largely unknown, as much as 25%of the total plasma nitrate is actively absorbed by the salivaryglands, so that the nitrate concentration in the saliva is at least 10-fold higher than that in plasma [7]. Although not yet fullycharacterized, the mechanism for active transport of nitrate bythe salivary gland seems to be common with that of iodide andthiocyanate, which have also been shown to augment thebactericidal action of acidified media [5]. Salivary nitrate isreduced to nitrite by commensal anaerobic bacteria in the oralcavity, which can use energy released by this process. Thisrepresents a symbiotic relationship, because no known mam-malian enzymes are capable of reducing nitrate to nitrite [22].

The nitrite concentration in saliva is strongly dependent ondietary levels of nitrate. The oral nitrate load employed in thisinvestigation, which was able to induce a 10-fold increase insalivary nitrite levels, corresponds to that present in approxi-mately 150–300 g of lettuce or spinach [23]. However, thenitrite concentrations observed in saliva on three differentoccasions after ingestion of this same amount of nitrate variedconsiderably, which is consistent with earlier studies revealingthat salivary nitrate and nitrite concentrations have a highdegree of variability between subjects and in the same subject atdifferent times [24,25]. Differences in nitrate intake, salivaryflow rate, the types of bacteria present, and oral hygiene are allfactors that will affect the levels of nitrite in saliva [26,27].

When nitrite enters the acidic gastric environment nitrousacid (HNO2) is formed and subsequently decomposes to avariety of reactive nitrogen intermediates (RNIs) with anti-bacterial activity, including NO, NO2, dinitrogen trioxide(N2O3), dinitrogen tetraoxide, peroxynitrite (ONOO−), andnitroso/nitrosyl species [6]. Some of the chemical reactions areoutlined below:

NO�2 þ Hþ X HNO2ðnitrous acid; pKac3:2Þ;

2 HNO2 YN2O3 þ H2O;

N2O3 YNOþ NO2;

N2O3 YNOþðnitrosonium ionÞ þ NO�2 ;

NOþþ RSH ðreduced thiolÞY RSNOðS�nitrosothiolÞ þ Hþ;

NOþ þ RR VNH ðsecondary amineÞY RR VNNOðN � nitrosoamineÞ þ Hþ;

NOþ O�2 ðsuperoxideÞYONOO�:

The biological chemistry of RNIs is complex and isinfluenced by a variety of ambient factors, such as oxygentension, pH, redox state, presence of heme-containing proteins,and concentration of thiols [6]. We found here that the levels ofNO, NO2, and nitroso/nitrosyl species were all markedlyincreased when nitrite-rich saliva was added to gastric juice,although our experimental procedure does not allow identifica-tion of the particular RNI(s) responsible for this increase. Itseems probable that several different RNIs work in concert toprevent bacterial growth.

The bactericidal activity of saliva collected after fasting andafter ingestion of nitrate was similar to that of nitrite-depletedsaliva supplemented with sodium nitrite (100 and 1000 μM,respectively), indicating that nitrite was of major significancefor bacterial killing. However, saliva also contains several othercompounds with antibacterial properties, such as thiocyanate,iodide, lactoferrin, peroxidases, immunoglobulins, and defen-sins [28,29], the levels of which may vary between individuals.In order to reduce the influence of such variations we utilizedpooled batches of saliva in each set of experiments here.Moreover, saliva rich or poor in nitrite was collected from thesame individual before and after nitrate loading. Finally, theantimicrobial effect of nitrite-depleted saliva in gastric juice wasthe same as that of gastric juice alone, suggesting that otherantibacterial substances in saliva had a negligible effect on ourfindings.

The sensitivity of different microbial pathogens to differentRNIs varies significantly. For example, NO exerts only a weak,direct antimicrobial effect against E. coli [30], but is morestrongly bactericidal toward Mycobacterium tuberculosis andStaphylococcus aureus. On the other hand, E. coli is moresensitive to other RNIs, including S-nitrosothiols and peroxy-nitrite [6].

The microbial targets for RNIs are multiple and includeDNA, proteins, and lipids [31]. For instance, RNIs have beenshown to induce oxidative DNA damage, including strandbreaks, crosslinking, and deamination. Furthermore, the func-tions of both intracellular and cell-surface proteins can bealtered when RNIs interact with heme groups, reactive thiolmoieties, iron–sulfur clusters, tyrosine residues, etc. [31]. Thisexistence of multiple targets for NO and other RNIs shouldmake it more difficult for bacteria to develop resistance towardnitrosative stress. However, certain bacteria have developedmechanisms of resistance, involving, e.g., flavohemoglobin,flavorubredoxin, and cytochrome c nitrite reductase [32]. Weutilized E. coli ATCC 25922 as a representative entericmicroorganism in our present experiments, but further studieson clinical pathogens will provide additional informationconcerning the importance of salivary nitrite in connectionwith gastric defenses.

The stomach model we developed here was designed toprovide an alternative procedure for exposing an infected gastricenvironment to NO. NO is a small, uncharged molecule that is agas at physiological temperatures and can readily diffuse throughthe siliconemembraneof the retentioncuff employed. Inaclinicalsituation this mode of delivery could have clinical advantagecompared to direct intragastric administration of nitrite, i.e., NO

1411H. Björne et al. / Free Radical Biology & Medicine 41 (2006) 1404–1412

delivery would not be dependent on gastric pH. Similarly, thismode of NO delivery was recently shown to exert antibacterialeffects in a urinarybladdermodel thatwere independent of the pHof the surrounding urine [33].

In our model, E. coli ATCC 25922 inoculated in gastric juicefrom patients containingNOwere effectively killed within 2–6 hat pH values ranging from 3 to 5, whereas in the absence of NOthese bacteria survived for 24 h. However, at pH 6 the E. colisurvived for 24 h even in the presence of NO, so that in this case,the antibacterial action of NO was not entirely independent ofpH. This discrepancy in comparison to the findings with theurinary bladder model may be due to differences in the liquidenvironment examined (gastric juice versus urine), in thepreparation of bacteria utilized, etc. We could also show thatexisting pathogens, in this case C. albicans and E. coli, werekilled or inhibited in our stomach model, although they seemedslightly less sensitive compared to E. coli ATCC 25922.

Both NO and NO2 were detected outside of the catheter cuffas early as after 5 min of incubation. It seems likely that theformation of NO2 occurred outside this cuff because ascorbicacid promotes the preferential generation of NO from nitrite, atthe expense of most other nitrogen oxides [7]. Moreover, NO inthe headspace will dissolve in the gastric juice, where it can beoxidized to nitrite and nitrate. Under acidic conditions, nitriteforms nitrous acid, which subsequently decomposes to variousRNIs. Thus, at lower pH the formation of RNIs will increase, afact which provides a likely explanation for the pH dependencyof the antibacterial effects in our stomach model. Although theantibacterial actions of each RNI are difficult to pinpoint, theinitial formation of NO seems to be important. Indeed, a pureNO donor (DETA-NONOate) inside the cuff in a similar modelwas shown to have antibacterial properties [33].

The levels of NO and NO2 in the headspace of this modelwere dependent on the composition of the fluid compartment.Considerably more NO and NO2 was detected in the headspacegas in the presence of water than with gastric juice, whichprobably reflects on reactions of NO with components in gastricjuice such as thiol groups. As expected, blood scavengedmost ofthe NO released from the catheter and thereby reduced the levelof NO2 formed.

During the past few decades, major concerns have beenraised concerning the potential formation of carcinogenic N-nitrosamines from nitrite and secondary amines in the stomachin connection with a high intake of nitrate [34]. However,although N-nitrosamines have been shown to be carcinogenic inanimal models [35–37], epidemiological studies have notrevealed any such effect in humans. On the contrary, numerousinvestigations have indicated the existence of an inverserelationship between cancer and a diet rich in nitrate, such asprovided by vegetables and fruit [38–43]. In fact, recent animaland human studies indicate that NO exerts important protectiveactions in the stomach [44–46] and in addition emergingevidence suggests that dietary nitrate also has importantgastroprotective effects [8,9,47].

The stomachs of critically ill patients being treated withmechanical ventilation are readily colonized by bacteria andmay then serve as a reservoir of pathogens, such as E. coli and

other gram-negative bacilli, which can cause ventilator-associated pneumonia [18]. In addition, gastric colonizationwith Candida species is frequent in these patients and isassociated with VAP, longer ICU stays, and increased costs[48,49]. We have recently reported that gastric NO is almostdepleted in intubated patients, due to their inability to swallowsaliva, and it is reasonable to believe that the consequentabsence of the antibacterial actions of nitrite could contribute tothe gastric bacterial colonization of such patients [17]. In thissame article we described complete restoration of gastric NOlevels in these patients to those seen in healthy individuals bygastric infusion of nitrite. Further studies will demonstratewhether such restoration of gastric NO in critically ill patients inintensive care units has beneficial effects.

Such administration of nitrite to these patients may alsohave additional effects. Like increased nitrate intake, directintragastric infusion of nitrite will lead to elevated circulatinglevels of nitrite [17,50]. Recent investigations have revealedthat systemic nitrite may be converted into NO and othernitroso/nitrosyl species, especially under conditions ofischemia, by several different pathways [16,51,52]. Indeed,in animal models, administration of nitrite amelioratesdamage occurring in connection with ischemia/reperfusion[15]. Thus, enterosalivary circulation of nitrate and itssubsequent reduction to nitrite and NO may have profoundphysiological effects not only in the stomach, but alsoelsewhere in the body.

In conclusion, we have shown here that nitrite-containingsaliva enhances the bactericidal action of gastric juice in adose-dependent fashion. Salivary nitrite decomposes to avariety of nitrogen species with potential antibacterial effects.Ingestion of nitrate in amounts corresponding to a normalintake of vegetables has pronounced effects on the salivarylevel of nitrite, as well as on subsequent killing of bacteria inthe gastric juice, a process that is highly dependent on thereductive capacity of the commensal bacteria in the oralcavity. These findings shed new light on the relationshipbetween food intake and human disease, as well as motivatingstudies designed to determine whether intragastric supplemen-tation with nitrite may be of therapeutic value for critically illpatients.

Acknowledgments

We thank Carina Nihlén, Katarina Karlsson, and UllaKarlsson for technical assistance and Annelie Brauner forhelpful discussions.

References

[1] Smith, J. L. The role of gastric acid in preventing foodborne disease andhow bacteria overcome acid conditions. J. Food Prot. 66:1292–1303;2003.

[2] Laheij, R. J.; Sturkenboom, M. C.; Hassing, R. J.; Dieleman, J.; et al. Riskof community-acquired pneumonia and use of gastric acid-suppressivedrugs. J. Am. Med. Assoc. 292:1955–1960; 2004.

[3] Benjamin, N.; O’Driscoll, F.; Dougall, H.; Duncan, C.; et al. Stomach NOsynthesis. Nature 368:502; 1994.

1412 H. Björne et al. / Free Radical Biology & Medicine 41 (2006) 1404–1412

[4] Dykhuizen, R. S.; Frazer, R.; Duncan, C.; Smith, C. C.; et al.Antimicrobial effect of acidified nitrite on gut pathogens: importanceof dietary nitrate in host defense. Antimicrob. Agents Chemother.40:1422–1425; 1996.

[5] Fite, A.; Dykhuizen, R.; Litterick, A.; Golden, M.; et al. Effects ofascorbic acid, glutathione, thiocyanate, and iodide on antimicrobialactivity of acidified nitrite. Antimicrob. Agents Chemother. 48:655–658;2004.

[6] Lundberg, J. O.; Weitzberg, E.; Cole, J. A.; Benjamin, N. Nitrate, bacteriaand human health. Nat. Rev. Microbiol. 2:593–602; 2004.

[7] Weitzberg, E.; Lundberg, J. O. Nonenzymatic nitric oxide production inhumans. Nitric Oxide 2:1–7; 1998.

[8] Miyoshi, M.; Kasahara, E.; Park, A. M.; Hiramoto, K.; et al. Dietary nitrateinhibits stress-induced gastric mucosal injury in the rat. Free Radic. Res.37:85–90; 2003.

[9] Bjorne, H. H.; Petersson, J.; Phillipson, M.; Weitzberg, E.; et al. Nitrite insaliva increases gastric mucosal blood flow and mucus thickness. J. Clin.Invest. 113:106–114; 2004.

[10] Lundberg, J. O.; Weitzberg, E.; Lundberg, J. M.; Alving, K. Intragastricnitric oxide production in humans: measurements in expelled air. Gut35:1543–1546; 1994.

[11] Lundberg, J. O.; Ehren, I.; Jansson, O.; Adolfsson, J.; et al. Elevatednitric oxide in the urinary bladder in infectious and noninfectious cystitis.Urology 48:700–702; 1996.

[12] Zweier, J. L.; Wang, P.; Samouilov, A.; Kuppusamy, P. Enzyme-independent formation of nitric oxide in biological tissues. Nat. Med.1:804–809; 1995.

[13] Weller, R.; Pattullo, S.; Smith, L.; Golden, M.; et al. Nitric oxide isgenerated on the skin surface by reduction of sweat nitrate. J. Invest.Dermatol. 107:327–331; 1996.

[14] Allaker, R. P.; Silva Mendez, L. S.; Hardie, J. M.; Benjamin, N.Antimicrobial effect of acidified nitrite on periodontal bacteria. OralMicrobiol. Immunol. 16:253–256; 2001.

[15] Duranski, M. R.; Greer, J. J.; Dejam, A.; Jaganmohan, S.; et al.Cytoprotective effects of nitrite during in vivo ischemia–reperfusion ofthe heart and liver. J. Clin. Invest. 115:1232–1240; 2005.

[16] Lundberg, J. O.; Weitzberg, E. NO generation from nitrite and its role invascular control. Arterioscler. Thromb. Vasc. Biol. 5:915–922; 2005.

[17] Bjorne, H.; Govoni, M.; Tornberg, D. C.; Lundberg, J. O.; et al. Intragastricnitric oxide is abolished in intubated patients and restored by nitrite. Crit.Care Med. 33:1722–1727; 2005.

[18] Safdar, N.; Crnich, C. J.; Maki, D. G. The pathogenesis of ventilator-associated pneumonia: its relevance to developing effective strategies forprevention. Respir. Care 50:725–739; discussion 739–741; 2005.

[19] Xu, J.; Xu, X.; Verstraete, W. The bactericidal effect and chemicalreactions of acidified nitrite under conditions simulating the stomach. J.Appl. Microbiol. 90:523–529; 2001.

[20] Feelisch, M.; Rassaf, T.; Mnaimneh, S.; Singh, N.; et al. Concomitant S-,N-, and heme-nitros(yl)ation in biological tissues and fluids: implicationsfor the fate of NO in vivo. FASEB J. 16:1775–1785; 2002.

[21] McKnight, G. M.; Smith, L. M.; Drummond, R. S.; Duncan, C. W.; et al.Chemical synthesis of nitric oxide in the stomach from dietary nitrate inhumans. Gut 40:211–214; 1997.

[22] Doel, J. J.; Benjamin, N.; Hector, M. P.; Rogers, M.; et al. Evaluation ofbacterial nitrate reduction in the human oral cavity. Eur. J. Oral Sci.113:14–19; 2005.

[23] Chung, S. Y.; Kim, J. S.; Kim, M.; Hong, M. K.; et al. Survey of nitrate andnitrite contents of vegetables grown in Korea. Food Addit. Contam.20:621–628; 2003.

[24] Bos, P. M.; Van den Brandt, P. A.; Wedel, M.; Ockhuizen, T. Thereproducibility of the conversion of nitrate to nitrite in human saliva after anitrate load. Food Chem. Toxicol. 26:93–97; 1988.

[25] Stephany, R. W.; Schuller, P. L. Daily dietary intakes of nitrate, nitrite andvolatile N-nitrosamines in The Netherlands using the duplicate portionsampling technique. Oncology 37:203–210; 1980.

[26] van Maanen, J. M.; van Geel, A. A.; Kleinjans, J. C. Modulation of nitrate-nitrite conversion in the oral cavity. Cancer Detect. Prev. 20:590–596;1996.

[27] Granli, T.; Dahl, R.; Brodin, P.; Bockman, O. C. Nitrate and nitriteconcentrations in human saliva: variations with salivary flow-rate. FoodChem. Toxicol. 27:675–680; 1989.

[28] Johnson, L. R. Essential medical physiology, 2nd ed. NewYork: LippincottRaven; 1998.

[29] Tenovuo, J. Antimicrobial agents in saliva—Protection for the whole body.J. Dent. Res. 81:807–809; 2002.

[30] Brunelli, L.; Crow, J. P.; Beckman, J. S. The comparative toxicity of nitricoxide and peroxynitrite to Escherichia coli. Arch. Biochem. Biophys.316:327–334; 1995.

[31] Fang, F. C. Mechanisms of nitric oxide-related antimicrobial activity.J. Clin. Invest. 99:2818–2825; 1997.

[32] Poole, R. K. Nitric oxide and nitrosative stress tolerance in bacteria. Bio-chem. Soc. Trans. 33:176–180; 2005.

[33] Carlsson, S.; Weitzberg, E.; Wiklund, P.; Lundberg, J. O. Intravesical nitricoxide delivery for prevention of catheter-associated urinary tract infec-tions. Antimicrob. Agents Chemother. 49:2352–2355; 2005.

[34] Tannenbaum, S. R.; Sisnkey, A. J.; Weisman, M.; Bishop, W. Nitrite inhuman saliva: its possible relationship to nitrosamine formation. J. Natl.Cancer Inst. 53:79–84; 1974.

[35] Archer, M. C.; Milligan, J. R.; Skotnicki, S.; Lu, S. J. Reactive metabolitesfrom N-nitrosamines. Adv. Exp. Med. Biol. 283:521–524; 1991.

[36] Tricker, A. R.; Preussmann, R. Carcinogenic N-nitrosamines in the diet:occurrence, formation, mechanisms and carcinogenic potential. Mutat.Res. 259:277–289; 1991.

[37] Iijima, K.; Henry, E.; Moriya, A.; Wirz, A.; et al. Dietary nitrate generatespotentially mutagenic concentrations of nitric oxide at the gastroesopha-geal junction. Gastroenterology 122:1248–1257; 2002.

[38] Forman, D.; Al-Dabbagh, S.; Doll, R. Nitrates, nitrites and gastric cancer inGreat Britain. Nature 313:620–625; 1985.

[39] Al-Dabbagh, S.; Forman, D.; Bryson, D.; Stratton, I.; et al. Mortality ofnitrate fertiliser workers. Br. J. Ind. Med. 43:507–515; 1986.

[40] McKnight, G. M.; Duncan, C. W.; Leifert, C.; Golden, M. H. Dietarynitrate in man: friend or foe? Br. J. Nutr. 81:349–358; 1999.

[41] Knight, T. M.; Forman, D.; Pirastu, R.; Comba, P.; et al. Nitrate and nitriteexposure in Italian populations with different gastric cancer rates. Int. J.Epidemiol. 19:510–515; 1990.

[42] van Loon, A. J.; Botterweck, A. A.; Goldbohm, R. A.; Brants, H. A.; et al.Intake of nitrate and nitrite and the risk of gastric cancer: a prospectivecohort study. Br. J. Cancer 78:129–135; 1998.

[43] Hung, H. C.; Joshipura, K. J.; Jiang, R.; Hu, F. B.; et al. Fruit and vegetableintake and risk of major chronic disease. J. Natl. Cancer Inst.96:1577–1584; 2004.

[44] Lanas, A.; Bajador, E.; Serrano, P.; Fuentes, J.; et al. Nitrovasodilators,low-dose aspirin, other nonsteroidal antiinflammatory drugs, and the riskof upper gastrointestinal bleeding. N. Engl. J. Med. 343:834–839; 2000.

[45] Fiorucci, S.; Santucci, L.; Gresele, P.; Faccino, R. M.; et al. Gastrointestinalsafety of NO-aspirin (NCX-4016) in healthy human volunteers: a proof ofconcept endoscopic study. Gastroenterology 124:600–607; 2003.

[46] Wallace, J. L.; Miller, M. J. Nitric oxide in mucosal defense: a little goes along way. Gastroenterology 119:512–520; 2000.

[47] Larauche, M.; Anton, P. M.; Garcia-Villar, R.; Theodorou, V.; et al.Protective effect of dietary nitrate on experimental gastritis in rats. Br. J.Nutr. 89:777–786; 2003.

[48] Drakulovic, M. B.; Bauer, T. T.; Torres, A.; Gonzalez, J.; et al. Initialbacterial colonization in patients admitted to a respiratory intensive careunit: bacteriological pattern and risk factors. Respiration 68:58–66; 2001.

[49] Olaechea, P. M.; Palomar, M.; Leon-Gil, C.; Alvarez-Lerma, F.; et al.Economic impact of Candida colonization and Candida infection in thecritically ill patient. Eur. J. Clin. Microbiol. Infect. Dis. 23:323–330; 2004.

[50] Lundberg, J. O.; Govoni, M. Inorganic nitrate is a possible source forsystemic generation of nitric oxide. Free Radic. Biol. Med. 37:395–400;2004.

[51] Gladwin, M. T.; Schechter, A. N.; Kim-Shapiro, D. B.; Lundberg, J. O. Theemerging biology of the nitrite anion. Nat. Chem. Biol. 1:308–314; 2005.

[52] Cosby, K.; Partovi, K. S.; Crawford, J. H.; Patel, R. P.; et al. Nitritereduction to nitric oxide by deoxyhemoglobin vasodilates the humancirculation. Nat. Med. 9:1498–1505; 2003.