Nitric oxide is negatively correlated to pain during acute inflammation

Background The role that nitric oxide (NO) plays in modulating pain in the periphery is unclear. We show here, the results of two independent clinical studies (microdialysis and gene expression studies) and a pilot dose finding study (glyceryl trinitrate study), to study the role of NO in the early phase of acute inflammatory pain following oral surgery. The effect of ketorolac on NO production and nitric oxide synthase (NOS) gene expression was also studied. Results Microdialysis samples showed significantly higher levels of NO at the first 100 min compared to the last 80 minutes in the placebo treated group. In the ketorolac group, on the other hand, NO levels gradually decreased over the first 60 min but were similar to placebo over the later 100-180 min, with no significant change in NO level over time. The levels of NO were negatively correlated to pain intensity scores. Local infusion of the NO donor glyceryl trinitrate at the site of surgery, showed a small analgesic effect that did not reach statistical significance in the sample size used. While the gene expression of iNOS and eNOS were not up-regulated, 3 hours after surgery, nNOS was downregulated in both treatment groups and eNOS gene expression was significantly lower in the ketorolac group compared to the placebo group. Further, there was a positive correlation between the change in gene expression of nNOS and eNOS in the placebo goup but not in the ketorolac group. Conclusion We suggest that at this early stage of inflammatory pain in man, NO is analgesic in the periphery. Further, ketorolac down-regulates eNOS gene expression.


Background
Nitric oxide (NO) is involved in numerous physiological processes in the peripheral and central nervous system [1]. It is produced intracellularly by the catabolism of Larginine to L-citrulline by NO synthase enzyme (NOS), which is present in three isoforms, the neuronal (nNOS), endothelial (eNOS) and inducible (iNOS) isoforms. The role that NO plays in pain is not simple, since it may show pro-or anti-nociceptive effects depending on the circumstances [2]. The majority of data, in preclinical studies support a pronociceptive role of NO at the spinal level [3][4][5][6]. Yet, other studies show inconsistent results; for review see [7].
The effect of NO on pain and analgesia in the periphery is further conflicting. Sodium nitroprusside, which releases NO non-enzymatically, has an antihyperalgesic effect in the rat paw pressure test [8]. Conversely, intradermal administration of the NOS substrate L-Arg or the NO donor SIN-1, both of which elevate NO levels, cause a dose-dependent mechanical hyperalgesia [9]. Conflicting clinical effects are also reported; as intracutaneous injection of NO in healthy volunteers evokes pain in a dose dependent manner [10], while transdermal application of the NO donor glyceryl trinitrate improved pain in patients with shoulder pain syndrome [11]. Associations between NO levels and pain intensity also show conflicting results. While there was no association between NO concentration in the perifacetal region and pain duration or pain level in patients with chronic low back pain [12], NO correlated with pain in the polyarticular subtype of juvenile idiopathic arthritis with active disease [13].
The role of different NOS isoforms in inflammatory pain process is derived primarily from animal experiments [2]. While nNOS is mainly observed at the spinal level or in neuropathic pain models, iNOS is up-regulated in inflamed tissues [14] and is involved in the development of hyperalgesia in inflammatory and neuropathic pain models [15]. Knockout mice, providing another tool for the identification of the specific role of each isoform in different pain models, confirm the role of nNOS and iNOS isoforms [16][17][18]. However, studies performed on knockout mice as well as those using different selective and non-selective NOS inhibitors point to a possible role for eNOS isoform in pain processing [16,17].
NSAIDs are among the most prescribed medications, yet their efficacy is compromised by a ceiling analgesic effect. The inhibitory effect of NSAIDs on NO production has been reported in many studies both clinically [19] and experimentally [5,20] and the involvement of the NO-cGMP pathway in the antinociceptive effects of NSAIDs has been suggested in several experimental studies [21,22].
The aim of the present study was to investigate the role of NO and NOS isoforms in acute inflammatory pain induced by a clinically well-defined model of clinical pain, third molar extraction. The effect of the NSAID ketorolac on NO production and NOS gene expression was also studied. To achieve this goal, two independent studies as well as a third pilot dose finding study were carried out: (1) a microdialysis study to estimate changes in NO levels at the site of tissue injury and to correlate these levels to pain intensity reported at the same time points; (2) a gene expression study to determine change in gene expression of the three NOS isoforms at the site of tissue injury, and finally (3) the glyceryl trinitrate glyceryl trinitrate study, which aimed at finding an efficacious dose to study the effect of the NO donor glyceryl trinitrate on pain intensity following tissue injury.

I. Association between NO level at the site of tissue injury and pain intensity A. Pain intensity following tissue injury
Pain intensity as evaluated by visual analogue scale (VAS) showed a continuous increase over the 3 hours observation period (Figure 1a). The repeated measure two way ANOVA model, controlling for time and the random effects of patients, detected a significant difference in pain intensity between placebo and ketorolac treated patients (p = 0.018, n = 24 & 20 for placebo and ketorolac, respectively).

B. NO levels at the site of tissue injury
NO levels at the surgical site were evaluated every 20 min following tissue injury using microdialysis collection. Patients that had more than one sample missing from the set of 9 serial samples due to failure of microdialysis probe, blood contamination of the sample or a low sample volume that did not allow analysis of replicates were excluded from this analysis. As seen in Figure 1b, NO levels gradually increased over the first 80 min in the placebo-treated group before declining gradually over the last 100 min of the observation period, reaching its lowest level by the end of the 3 hours collecting period. Contrast analysis under the linear mixed effects model controlling for time shows that NO levels over the first 100 min were significantly higher than those over the last 80 min (p < 0.0001) in the placebo treated group. NO level in microdialysate in subjects administered ketorolac before surgery gradually decreased over the first 60 min but was similar to placebo over the 100-180 min observations. Although, Figure 1 Microdialysis Study. (a) Effect of tissue injury and ketorolac treatment on pain intensity following third molar tooth extraction (p = 0.018; repeated measure two way ANOVA, n = 24 & 20 for placebo and ketorolac treatment groups, respectively). (b) Effect of tissue injury and ketorolac treatment on NO production at the site of oral surgery. Data are presented as mean ± SEM; n = 19 & 10 for placebo and ketorolac treatment groups, respectively; *indicates the first five time points were significantly different from the last four time points, p < 0.0001; contrast analysis under the mixed linear model. Raw data were not normally distributed so were transformed to ln(VAS+1) and statistical analysis were conducted on transformed data.
there was no significant change in the NO levels over time in the ketorolac group, the difference between the two treatment groups was not statistically significant (p = 0.48).

C. Correlation between pain intensity and NO levels
Because both pain intensity and NO levels were reported repeatedly in the same patients, simple correlation analysis would have ignored this fact and would have tested the correlation considering each time point as a different patient. Further, it would have ignored the difference in the time of collection or reporting of pain. Therefore linear mixed effects model was fitted to overcome both problems and to further test interaction of treatment and NO levels on pain intensity.
Only pain intensity scores for patients with successful microdialysis sampling (n = 19 & 10 for placebo and ketorolac, respectively) were included in the analysis. Under the linear mixed model with covariates time, treatment, NO, and the interaction of treatment and NO, NO was found to be negatively correlated to pain scores (regression coefficient = -0.6782, p = 0.0014, Table 1).

II. Effect of the NO donor glyceryl trinitrate on pain intensity
This study was a pilot dose finding study that aimed at finding the appropriate dose and sample size to further confirm the analgesic role of NO by administering the NO donor glyceryl trinitrate following tissue injury. Placebo or glyceryl trinitrate was infused directly into the site of surgery using PE 50 tubing. Three doses of glyceryl trinitrate were tried, none of which produced a significant analgesic effect ( Figure 2a).
However, when two groups (150 and 200 μg) were pooled, there was a significant difference between the placebo and glyceryl trinitrate treated patients over the 1 hour observation period (p = 0.038, repeated measure two way ANOVA; n = 20 & 37 for placebo and glyceryl trinitrate groups, respectively; Figure 2b). Further, when the sum of pain intensity reported by each patient over the one hour observation period were plotted against the dose of glyceryl trinitrate, considering placebo as zero μg, this resulted in a linear dose response that significantly deviated from zero (p = 0.044) and that showed a negative slope of -0.60 ± 0.29; linear regression; Figure 2c).
Power analysis of the data, showed a required sample size of 47 patients per group in order to achieve an 80% power to show a difference between the sum of VAS over one hour observation period of the placebo and glyceryl trinitrate groups. Given the small effect size, the large sample size required and the increased rate of headache with the use of these doses of glyceryl trinitrate, a larger study was not conducted for ethical consideration.

III. Effect of tissue injury and ketorolac treatment on the gene expression of NOS isoforms A. Change in gene expression
We further studied the role of NO on acute inflammatory pain by measuring the change in gene expression of the 3 NOS isoforms using qRT-PCR. nNOS was significantly down-regulated in both the placebo group (2.4 fold; p = 0.02; paired t-test) and the ketorolac treatment group (3.4 fold; p = 0.0003; paired t-test). However, there was no significant difference between the two treatment groups (Figure 3a).
On the other hand, iNOS did not show any significant change compared to the pre-surgery level neither in the placebo group (p = 0.7) nor in the ketorolac treatment group (p = 0.98).
As for eNOS gene expression levels, there was a small insignificant change in both treatment groups when compared to the pre-surgery biopsies (p = 0.17 and 0.23 for the placebo and ketorolac treatment groups, respectively). However, these small changes being in opposite directions resulted in a significant difference between the two treatment groups (p = 0.026, independent twosample t-test).

B. Correlation between the gene expression of NOS isoforms
The correlation between the changes in gene expression of the three NOS isoforms following tissue injury was examined using Pearson's correlation coefficients (Fig.  3b&3c). The gene expression of nNOS and eNOS were significantly correlated in the placebo group (r = 0.56; p = 0.038; n = 14; Fig. 3b), but not in the ketorolac  Fig. 3c). There was no significant correlation between the changes in gene expression of iNOS and either of the two other isoforms (data not shown).

Discussion
As mentioned earlier, NO has been mostly considered as a pain mediator. However, we show here in a clinical study that NO is not only devoid of this pronociceptive property in the periphery, at this early stage of acute inflammatory pain, but it is also negatively correlated to pain. The unresolved role of NO in nociception at the periphery [8,9] has been attributed in animal studies to several factors, including the level of NO. Low levels of NO at the site of injury show antinociceptive effects, whereas higher concentrations have pronociceptive effects. In a rat model of incisional pain, the NO donor, S-nitroso-N-acetylpenicillamine (SNAP) applied inside the surgical wound, reduced the incision allodynia at concentrations between 1-10%. On the other hand, higher concentration (30%) intensified the allodynia [23]. The same finding was seen in the formalin test [24].
Baseline level of neuronal excitability is also thought to affect the role of NO in nociception [25].
Electrophysiological recordings of dural nociceptive afferents showed that neurons with higher mechanical activation thresholds were activated by NO donors, while neurons with lower mechanical activation thresholds were inhibited by NO donors [25]. Further, in PGE 2 induced pain, the NO donor SIN-1 induced a dose dependent hypernociception when injected intradermally, while inducing an antinociceptive effect when injected subcutaneously, which suggests that different subsets of primary nociceptive neurons may respond differently to NO [26].
The NOS isoform expressed at the site of injury might also contribute to identifying the role of NO in nociception. The role of different NOS isoforms in pain processing is still subject to debate. It is known that nNOS and eNOS, though found primarily within the nervous system and endothelial tissue respectively, are also found in other tissues, whereas iNOS is expressed in various cell types, such as macrophages and neutrophils [6]. In the present study, the three NOS isoforms were expressed in the gingival mucosa, following tissue injury. While both nNOS [16,17,27] and iNOS [27,28] are frequently suggested to play an important pronociceptive role in several pain models, eNOS is not often studied. However, it was reported that eNOS was the only isoform involved in both phases of the rat formalin test [17] and eNOS -/mice showed a more rapid recovery from thermal hyperalgesia compared to wild type mice after intraplantar injection of complete Freund's adjuvant [16]. Interpretation of findings in NOS -/mice is compromised by the compensation of other isoforms for the deficient isoform [16,29,30], at least at the spinal level. The analgesic role of NO at the periphery may be explained by the fact that NO generated by eNOS may modulate leucocyte adherence, which contributes to inflammation [31] and pain [32]. NO is also responsible for vasodilatation with subsequent increased blood supply to the injured region and ultimately increased clearance of local inflammatory mediators [33], another possible mechanism for the analgesic effect of NO at the periphery.
The down-regulation of nNOS in both treatment groups seen here is an interesting finding that may reflect a feedback effect to the higher NO levels at earlier time points. However, the present data cannot indicate such a conclusion. Since the increase in NO production seen in the microdialysis samples occurred during the first 80 min, this may explain the minimal change noticed at the gene expression level of NOS isoforms at 3 hours. Therefore, further investigation to study the effect of tissue injury on NOS isoforms at earlier and latter time points is warranted. However, the gene expression results suggest a relatively low level of NO at the site of injury, given the down regulation of nNOS and insignificant change of the other two isoforms.
Taking into account the negative correlation between NO levels and pain intensity reported here, it is plausible that at lower levels, NO released at the periphery plays an analgesic role in this clinical model of inflammatory pain (Fig. 4).
Another interesting finding in the present study is the correlation between nNOS and eNOS expression in the placebo, but not in the ketorolac treatment group, which suggests an inhibitory effect of ketorolac on eNOS gene expression. To our knowledge, the association between the expression of these two isoforms has not been reported earlier. It is known that both nNOS and eNOS are constitutive isoforms of NOS that are calcium-dependent [2]. We have recently reported the upregulation of kinin B 1 and B 2 receptors at the same pain model and time frame [34], both of which cause elevated free intracellular calcium upon activation [35] that may contribute to the activation of both NOS isoforms. The down regulation of eNOS by NSAIDs, on the other hand, is in line with the suppression of eNOS expression by high-dose aspirin or meloxicam, in retinas of diabetic rats [36].
The negative correlation between NO levels and pain intensity that we show here, is in line with the analgesic effect of topical glyceryl trinitrate that was reported earlier in several tendinopathies [11,33,37,38]. Unfortunately, we could not show the same analgesic effect of glyceryl trinitrate in the present study as the infusion of a rather small dose of glyceryl trinitrate (150 or 200 μg) into the surgical site did not show a significant analgesic Figure 4 Tissue injury increases NO production possibly via eNOS or nNOS in the early phase following surgery. This causes a small increase in NO level that lead to an analgesic effect. However, in the late phase iNOS isoform is up-regulated, producing a marked increase in NO level that would have a hyperalgesic effect. cNOS, constitutive NOS. effect at the sample size used. However, combining both groups resulted in a small analgesic effect. It was not possible to further increase the dose of glyceryl trinitrate to examine a dose dependent nature of NO effect on pain, nor to extend the study due to ethical concerns, given the small analgesic effect and the associated increased rate of headache with glyceryl trinitrate infusion.
It is not clear, why infusion of glyceryl trinitrate in the present study did not achieve a clear analgesic effect, in contrast to the earlier studies [11,33,37,38]. However, it is possible that to reach a detectable analgesic effect, longer exposure to NO is required, given in our study glyceryl trinitrate was infused for 20 min only, while in earlier studies, patients received transdermal glyceryl trinitrate patch for at least three days and up to 24 weeks.
There is a great body of evidence that supports an inhibitory role of NSAIDs on NO production both in clinical and experimental settings [19,[39][40][41]. However, some studies suggest a stimulatory role e.g. [42,43] and others did not show any effect [44]. This variability might be due to different experimental settings, including selectivity of NSAID, concentration used, stimulants and cell type, which may reflect on the NOS isoform expressed. Nevertheless, most of the studies showing an inhibitory effect of NSAIDs on NO production also show downregulation of iNOS expression. Since in the present study iNOS isoform was not up-regulated, 3 hours following tissue injury, this may explain the failure of ketorolac to decrease NO production at this early time point. Even though ketorolac down-regulated eNOS expression, this did affect NO levels, possibly because eNOS generate small amounts of NO, compared to iNOS [2], which makes the effect of ketorolac on NO production rather small.
We therefore suggest that at this early phase of inflammatory pain, NO plays an analgesic role in the periphery. Further studies are needed to identify the contribution of NOS isoforms to this analgesic effect.

Subjects and procedures
The present work shows the results of two independent studies: a microdialysis study, and a gene expression study. A third, dose finding, pilot study was also conducted, glyceryl trinitrate intervention study. Participants in these three independent studies were healthy volunteers (n = 147) aged between 16 and 36 years, who required extraction of impacted third molars. Demographic data of participants in each study are shown in Table 2, 28 other subjects, who did not develop moderate pain following surgery, were excluded from the glyceryl trinitrate study. Protocols for the three studies were approved by the Institutional Review Board of NIDCR/NIH. Written informed consent was obtained from all participants before treatment. Pregnant or lactating females or patients with the presence of clinical signs of infection or inflammation at the extraction sites were not included in the study. All subjects received intravenous midazolam (4.1 ± 1 mg) and local anesthesia by 2% lidocaine (151.3 ± 25.8 mg) with epinephrine 1:100,000 prior to surgery. Following ethical regulations, a rescue analgesic drug was provided upon patient's request in both the microdialysis and gene expression studies. The observation period in case of the glyceryl trinitrate study was one hour, so rescue medicine was not permitted during this hour.

Microdialysis procedure
Following satisfactory local anesthesia, subjects randomly received either ketorolac (30 mg) or placebo Data presented as mean ± S.D. * Extraction difficulty is the sum calculated by assigning a score of (2) for soft tissue impactions, (3) for partial bony impactions and (4) for full bony impactions.
intravenously immediately before surgery. Two mandibular third molars were extracted and a surgical difficulty score was assigned for each tooth. After extraction, a microdialysis probe (CMA/20 Microdialysis Probe; CMA/Microdialysis, North Chelmsford, MA) was placed along the buccal aspect of the mandible, beneath the mucogingival flap elevated for the surgical procedure [45,46]. The probe fiber consists of a 10-mm flexible, nonmetallic, semipermeable dialysis membrane with a molecular cutoff ranging from 3000 to 20,000 Da. The probes were secured to an adjacent tooth with silk suture and the flap closed using 3-0 chromic gut suture. Sterile lactated Ringer's solution was pumped at 10 μL/ min and samples collected at 20-min intervals after the completion of surgery. The samples were immediately placed on dry ice after each collection period. Subjects remained under observation for the first 3 hours after surgery to evaluate pain and adverse events and for collection of samples by microdialysis. Patients rated their pain intensity every 20 min for the first 3 h post-operatively using 100-mm visual analog scale (VAS). At the conclusion of the observation period, the microdialysis probes were removed and the samples were stored at -70°C.

Glyceryl trinitrate administration
Only patients who developed moderate pain (≥ 35 on the VAS) following surgery and disappearance of local anesthetic effect were included in this protocol. A PE50 tubing was placed under the flap elevated for tooth extraction. Three different doses of glyceryl trinitrate (100 μg, 150 μg or 200 μg) prepared in a final volume of 1 ml or a placebo were delivered over 20 min to the surgical site. These doses were selected based on the analgesic effect of transdermal glyceryl trinitrate in shoulder pain syndrome due to supraspinatus tendinitis [11]. Since this was a pilot study, with no previous experience of the possible effect size, there was no sample size calculation prior to conducting this study. The number of patients enrolled was based on previous experience with the oral surgery model. Pain was evaluated using 100 mm VAS for 1 hour following drug delivery.

Gene expression analysis by qRT-PCR
In this protocol, two impacted third molars were extracted. A 3 mm punch biopsy was taken from the oral mucosa overlying one impacted third molar immediately prior to surgery and a second biopsy was taken from the opposite surgical site 3 hours post-surgery. All biopsies were immediately frozen in liquid nitrogen and stored at -70°C until ready for RNA extraction. Patients received a placebo or ketorolac 30 minutes prior to surgery.
Oral mucosal biopsies (n = 64) were used to detect gene expression using ABI Prism 7900 HT Sequence Detection System (Applied Biosystems, Foster City, CA) as described previously [45,47,48]. All reagents were purchased from Applied Biosystems and 2 μg of DNasetreated RNA was used to synthesize cDNA using random primers from the High-Capacity cDNA Archive Kit according to the manufacturer's instruction. Polymerase chain reaction was performed with cDNA template using the PCR Master Mix with AmpErase UNG. Sequence-specific primers and TaqMan MGB probes were purchased from Assays-on-Demand Gene expression product. Quantification of gene expression was performed in a 20-μL reaction (384-well plate) and each sample was run in triplicate. The housekeeping gene 18S rRNA was used as endogenous control and negative controls were processed under the same conditions without a cDNA template. Data acquisition was conducted using User Bulletin #2 software (v1.6, Applied Biosystems). The threshold cycle (Ct) of 18 rRNA was used to normalize target gene expression (ΔCt) to correct for experimental variations. The relative change in gene expression (ΔΔCt) was used for comparison of the gene expression in post-surgery tissue versus that in pre-surgery tissue.

Measurement of NO degradation products
NO has a very short half-life and is quickly degraded invivo to its stable end-products nitrite and nitrate. Nitrite and nitrate (NOx) as an index of total NO production were measured in microdialysis samples using a commercially available kit (Nitrate/Nitrite Fluorometric Assay Kit, Cayman Chemical Company, Ann Arbor, MI), following the manufacturer's instructions. Samples containing blood were excluded as blood interfers with accurate evaluation of nitrite and nitrate levels using fluorometry. The samples were run in triplicate and fluorescence was read using a Victor 3 Perkins Elmer spectrofluorometer (excitation = 370 nm; emission = 420 nm).
Fluorometric assay for NOx measurement is more sensitive than the conventional colourimetric assays based on Griess reaction and both methods were found to produce essentially the same results [49].

Statistical analysis
Statistical analysis was carried out using SAS (v. 9.1, SAS Institute Inc., Cary, NC) or SPSS (v. 16.0, SPSS, Chicago, IL). Demographic data, such as age and gender, were summarized for different treatment groups. Mean and SEM of NO and VAS were plotted for ketorolac and NTG treatment at 9 time points. Glyceryl trinitrate dose response was evaluated using linear regression. Comparisons between effects of ketorolac or glyceryl trinitrate treatment and placebo on VAS were performed using repeated measure two-way ANOVA. Linear mixed effects models were fitted to investigate the effect of time (at 9 levels), treatment, NO and the interaction effect between treatment and NO on VAS for the microdialysis study. Patient ID was assumed to be a random effect and each patient had a random intercept in all the linear mixed models since the samples were pulled out from a large population. The effect of treatment on NO production was also examined using linear mixed effect models. Data that did not follow normal distribution was transformed to ln(x+1). The relative change in gene expression was assessed by paired t-test, and the comparison between treatment groups was assessed by independent two-sample t test and the association among these gene expressed was examined by Pearson's correlation coefficients. Results were considered significant at α = 0.05.