Involvement of Adenosine in the Antiinflammatory Action of Ketamine

To understand the mechanism of the protective effect of ketamine, we used two models of septic shock and followed the survival of treated animals. Initially, we used endotoxin (lipopolysaccharide)–induced sepsis to define the optimal dose and treatment time of ketamine. In the second model, sepsis was induced by intraperitoneal inoculation of lethal dose of Escherichia coli , a model more relevant to the clinical practice. The latter model, with sublethal doses of bacteria, was used to study the antiinflammatory effect of ketamine on leukocyte recruitment and cytokine production. We used sublethal doses of bacteria to reduce mortality at harvest time. We hypothesized that ketamine induces adenosine. Therefore, we defined adenosine concentrations in sera and peritoneal fluid after ketamine administration. For this purpose, blood samples were taken from rats at several time points after injection of ketamine. Rats were used for this experiment because we needed a large volume of blood for the multiple sampling. To rule out the induction of adenosine by hemorrhage and to confirm the induction of adenosine in our experimental model, we tested adenosine concentrations in mice. To overcome blood volume limitation, samples were taken once from separate groups of mice for each time point. To prove that adenosine is involved in the antiinflammatory effect of ketamine, we mimicked its effect on leukocyte recruitment and inhibition of cytokine production by an adenosine receptor agonist and blocked ketamine effects by adenosine receptor antagonists. Finally, we blocked the sepsis-protective effect of ketamine by an adenosine receptor antagonist in the bacterial septic model.

CD1 mice and Sprague-Dawley rats were bred and maintained in the animal laboratory of the Soroka Medical Center (Beer-Sheva, Israel). Experiments were done with the permission of the Ben-Gurion University of the Negev Committee for Ethical Care and Use of Animals in Experiments (Beer-Sheva, Israel). We used female mice aged 10–12 weeks.

Escherichia coli  were grown in Luria-Bertani broth (Conda Laboratories, Madrid, Spain) and harvested during the log phase. Aliquots were stored frozen in LB broth containing 30% glycerol.

Adenosine (Adenocor) was purchased from Sanofi Winthrop (Auckland, New Zealand). Adenosine receptor agonists and antagonists were purchased from Sigma (Rehovot, Israel). These included CGS-21680, which is an agonist of the A2 receptor (A2R); A2R antagonist 1,3-dimethyl-7-propargylxanthine (DMPX); A1 receptor (A1R) antagonist 8-cyclopentyl-1,3-dipropylxanthine (DPCPX); and A3 receptor (A3R) antagonist MRS-1220 (MRS). ZM 241385, an A2A receptor (A2AR) antagonist, was purchased from Tocris (Bristol, United Kingdom). The NMDA receptor antagonist ketamine was purchased from Parke Davis (Ketalar, Hampshire, United Kingdom).

Sepsis was induced in CD1 mice by intraperitoneal injection of a lethal dose of lipopolysaccharide (150 mg/kg, E. coli  055:B5; Sigma) or by inoculation with aliquots of E. coli  (3 × 10⁷colony-forming units). Peritonitis was induced in CD1 mice by inoculation of a sublethal dose of E. coli  (1.5 × 10⁷colony-forming units). At the end of each experiment, mice were killed by neck dislocation. Peritoneal leukocytes were harvested at indicated time points after inoculation by peritoneal lavage with phosphate-buffered saline containing 2% bovine serum albumin and 5 mm EDTA. Cells were washed once, and then total leukocyte counts were determined. Ketamine was injected subcutaneously; adenosine agonists and antagonists were injected intraperitoneally. Blood for bacteriologic culture was collected in culture vials (BD Bactec plus aerobic/F; Becton Dickinson, Spark, MA) and analyzed by the Clinical Bacteriologic Laboratory at the Soroka University Medical Center.

For continuous kinetic study, female Sprague-Dawley rats (weighing 315–345 g) were used. Rats were treated with ketamine (10 mg/kg) or saline at time 0. At time 0 and at 5- or 10-min intervals, blood was drawn by heart puncture for determination of plasma adenosine concentrations.

For kinetics experiments, CD1 mice were injected subcutaneously with saline or with ketamine (10 mg/kg), with or without an intraperitoneal lethal dose of E. coli . Groups of three mice were bled by heart puncture at 5-min intervals for plasma adenosine concentrations.

Adenosine receptor agonist and antagonists were injected intraperitoneally. Animals were treated with the A2R agonist CGS-21680 (2 mg/kg) and with the adenosine-receptor antagonists DMPX (10 mg/kg), ZM 241385 (1 mg/kg), DPCPX (10 mg/kg), and MRS (0.2 mg/kg) 1 h before inoculation.

At different time points after E. coli  inoculation, animals were killed and peritoneal lavage was performed. The lavage leukocytes were counted after trypan blue staining. After centrifugation at 400g  for 10 min, the cell-free supernatants were removed and kept frozen at −20°C until analysis. TNF-α and IL-6 concentrations were measured by enzyme-linked immunosorbent assay kits (R&D Systems, Minneapolis, MN). Cell counts and enzyme-linked immunosorbent assay were performed blindly on coded samples.

Peritonitis was induced by inoculation of E. coli . At different time points, animals were killed and a 1-ml syringe flushed with heparin was used to draw a blood sample from their hearts. The samples were stored on ice before centrifugation at 1,000g  and 4°C for 10 min. The cell-free supernatants were collected and frozen at −20°C for future analysis. TNF-α and IL-6 serum concentrations were quantified by enzyme-linked immunosorbent assay.

For adenosine determination in mice peritoneal lavage, 900 μl lavage fluid was placed in a polystyrene tube containing 100 μl stop solution (2.3 μm erytro-9(2-hydroxy non-3-yl)-adenine and 20 μm dipyridamole) designed to inhibit the metabolism and cellular uptake of adenosine.17For plasma adenosine determination, a 1-ml syringe containing a small amount of heparin and 50 μl stop solution was used to draw a 0.5-ml blood sample.

The method for measuring adenosine in deproteinated fluid was described previously.18Briefly, cell-free fluid was combined with methanol in a ratio of 2:3 and heated at 100°C for 3 min to extract proteins. After centrifugation at 13,000g  for 5 min, the supernatant was removed, dried with a centrifuge-vacuum concentrator, and reconstituted with 500 μl water. Then, the solution was mixed with 20 μl acetate buffer, 2.0 m, pH 4.5, and 10 μl (50% in water) chloracetaldehyde (Sigma), heated at 80°C for 2 h. The samples were protected from light during and after heating. After cooling to room temperature, the samples were briefly centrifuged and a portion (typically 50 μl) was directly injected onto a C18 high-performance liquid chromatography (HPLC) column for analysis (see HPLC Analysis). Samples could be stored at 4°C before HPLC analysis for at least 48 h without noticeable degradation. Adenosine determination was performed blindly on coded samples.

The HPLC system used in this study was equipped with a Quaternary pump, an Agilent 1100 autosampler, and an Agilent 1100 fluorescence detector. Samples were chromatographed on a Kromasil KR100-5C18 column (5 mm, 150 × 4.60 mm; cat. No. E8285; Agilent, Edison, NJ). For determination of 1-N  ⁶-ethenoadenosine, samples were eluted with 20% methanol in water in an isocratic mode with a flow rate of 1,500 ml/min. The fluorescent product 1-N  ⁶-ethenoadenosine was eluted at 2.4 min and was detected with an Agilent 1100 fluorescence detector set at an excitation wavelength of 270 nm and an emission wavelength of 411 nm. Adenosine concentrations were calculated by measuring peak areas and comparing these with peaks produced by adenosine standards of known concentrations.

For flow cytometry analysis, peritoneal leukocytes were harvested and rinsed with phosphate-buffered saline. Then, 1.5 × 10⁶cells were stained in 100 μl phosphate-buffered saline containing 1% bovine serum albumin and 0.05% sodium azide for 1 h on ice with a labeled leukocyte marker. The following monoclonal antibodies were used: fluorescein isothiocyanate–conjugated rat anti-mouse F4/80 monoclonal antibody (clone CI:A3-1; Serotec, Raleigh, NC); FITC-conjugated rat anti-mouse Ly-6G (GR-1) monoclonal antibody (clone RB6-8C5; Pharmingen, San Diego, CA); biotin-conjugated hamster anti-mouse CD3ε monoclonal antibody (clone 500A2; Pharmingen); and R-pycoerythrin–conjugated streptavidin (Jackson, West Grove, PE).

After staining, cells were washed and fixed in 0.2% paraformaldehyde–phosphate-buffered saline. Antibodies were diluted to recommended concentrations according to the manufacturer's instructions. Nonspecific binding of antibodies was adjusted with cells labeled with matching isotype control antibodies. At least 10,000 cells were analyzed per sample. Analyses were conducted by flow cytometry (FACS Calibur; Becton Dickinson, Mountain View, CA). Fluorescence data were analyzed by the Cell Quest Program (Becton Dickinson).

Survival rates of ketamine-treated and control rats were compared using a one-tailed log-rank test, and one-way analysis of variance (followed by Tukey test) was used to compare among the various treatment groups. P  values less than 0.05 were considered significant.

To determine the effect of ketamine on survival, three groups of mice were injected with a lethal dose of lipopolysaccharide. As shown in figures 1 and 2, a dose of 10 mg/kg given before lipopolysaccharide administration increased the survival rates of mice to 86% compared with 8% in untreated mice. However, when injected 1 h (not shown) or 2 h after lipopolysaccharide administration (fig. 1), the same dose did not improve survival rates as compared with untreated animals. In experiments aimed at determining the optimal dosage of ketamine (fig. 2), we found a positive dose response to 0.5–10 mg/kg ketamine (fig. 2). Therefore, we used 10 mg/kg ketamine for the rest of the experiments.

To investigate whether the antiinflammatory properties of ketamine are mediated by release of adenosine, we measured adenosine concentrations in biologic fluids. As shown in figure 3A, the adenosine concentration peaked at 2.4 min in an HPLC analytical method developed here. Pretreatment of the same sample with adenosine deaminase completely eliminated the adenosine peak (fig. 3B), indicating the specificity of the assay. Integration of adenosine standard gave a linear dose response from 0.01 to 5 μm (not shown). To assess the direct effect of ketamine on blood adenosine concentration, one rat was injected with ketamine and a control rat was injected with saline; blood samples were taken from them at 5-min intervals (fig. 4A). In contrast to the control saline-injected rat in the ketamine-injected animal, a sharp peak of adenosine reaching 5 μm was seen 25–30 min after injection. To ascertain that the increase of adenosine in the ketamine-treated animals was indeed due to ketamine treatment and not to the stress of massive bleeding, we killed mice at 5-min intervals and analyzed their blood for adenosine concentration (fig. 4B). Similar to our findings in rats, adenosine serum concentrations in mice peaked at 25 min after ketamine injection and reached a concentration of approximately 0.4 μm. In the experiments with mice, we were able to see a significant increase of adenosine serum concentrations as early as 15 min after ketamine administration. The same increase in adenosine concentration was observed when mice were inoculated with a lethal dose of E. coli  in addition to ketamine administration (fig. 4B). To test whether a similar increase in adenosine concentration after ketamine administration occurs in the peritoneal cavity, we measured adenosine concentration in peritoneal lavage after ketamine administration. As shown in figure 4C, similar to the increase of adenosine in mice blood, we found a significant increase in adenosine concentrations in the peritoneum starting from 10 min, and then a decline at 30 min. To reduce distress of the experimental animals caused from the long mortality experiments, we continued the study in a shorter and less painful model of bacterial peritonitis. For this reason, two groups of mice were inoculated intraperitoneally with a sublethal dose of E. coli  and treated subcutaneously with saline or ketamine (10 mg/kg). Cell exudates were collected at various time points by peritoneal lavage. As shown in figure 5, a half lethal dose of E. coli  inoculation to the peritoneum induced massive leukocyte recruitment, which peaked at 12 h and slowly declined. At every time point tested, after ketamine administration, leukocyte recruitment was reduced by approximately 50% compared with saline-treated animals. Ketamine had no effect on the leukocyte profile in the peritoneal lavage as analyzed by flow cytometry using specific markers (data not shown).

To support our finding that the increase in adenosine concentrations after ketamine administration is responsible for the antiinflammatory activity of ketamine, we examined the effects of adenosine receptor antagonists and agonists in a mouse peritonitis model. Mice were injected with saline or inoculated with a sublethal dose of E. coli . Inoculated mice were pretreated with saline; with ketamine; with an A2R antagonist, DMPX; with DMPX and ketamine; with an A2AR agonist, CGS-21680; with a specific A2AR antagonist; with ZM 241385; or with ketamine and ZM 241385. Similar to the effect of ketamine, the adenosine A2AR agonist CGS-21680 significantly reduced peritoneal leukocyte recruitment after E. coli  inoculation. The A2R antagonist DMPX and the specific A2AR antagonist ZM 241385, however, effectively reversed the inhibitory effect of ketamine (fig. 6). We then investigated the effect of ketamine and the A2R agonist and antagonist on TNF-α and IL-6 concentrations. As shown in figure 7, ketamine and the A2AR agonist CGS-21680 reduced TNF-α and IL-6 concentrations in the serum (figs. 7A and B) and peritoneal fluid (figs. 7C and D) of the mice with peritonitis. The effect of ketamine and CGS-21680 reached statistical significance except to the inhibitory effect of CGS-21680 on TNF-α concentration in the peritoneum (fig. 7C). Like the effect on leukocyte recruitment, the inhibitory effect of ketamine on TNF-α and IL-6 concentrations in blood and peritoneal fluids was blocked by DMPX (figs. 7A–D). In addition, ZM 241385 effectively reversed the inhibitory effect of ketamine on TNF-α and IL-6 secretion (figs. 7A–D). In contrast to A2R antagonists, the A1 and A3 adenosine receptor antagonists DPCPX and MRS had no effect on ketamine inhibition (figs. 7A and B). Similar to the effect on survival, late ketamine injection (2 h after inoculation) did not significantly reduce cell recruitment or TNF-α and IL-6 concentrations in the serum (data not shown). DPCPX, DMPX, ZM 241385, and MRS without ketamine had no effect on serum and peritoneal TNF-α and IL-6 concentrations induced by E. coli  inoculation (data not shown). Finally, to prove that the beneficial effect of ketamine on survival is mediated by adenosine and is not purely circumstantial, we inoculated animals with a lethal dose of E. coli  and tested the effect of the A2A receptor antagonist ZM 241385. Similar to the beneficial effect of ketamine treatment in the lipopolysaccharide studies, we found that ketamine significantly increased survival rate of E. coli  inoculated mice to 60%, compared with untreated animals, which had only a 6.7% survival rate (fig. 8). Similar to the effect on leukocyte recruitment and cytokine secretion, ZM 241385 blocked the beneficial effect of ketamine and reduced survival to 20%. The survival rate of mice inoculated and treated with ZM 241385 was not significantly different compared with the untreated control group inoculated with E. coli . Blood bacteriologic culture indicated that all four groups had development of bacteremia at 24 h, and their peritoneal lavage was positive for E. coli  (data not shown).

Our objective was to elucidate the mechanism of the antiinflammatory action of ketamine. We found that reduced mortality of mice treated with ketamine is dose related, with a maximal effect at the subanesthetic dose of 10 mg/kg (the anesthetic dose of ketamine in mice and rats is approximately 100 mg/kg). In addition, we showed that ketamine is effective only when given concurrently with lipopolysaccharide but not when given 1–2 h after lipopolysaccharide administration. These findings suggest that ketamine suppresses the initial steps of the inflammatory cascade. Previous findings5–7and our current data show that ketamine suppresses TNF-α secretion, known to be one of the main inducers of uncontrolled inflammatory cascade observed in septic patients.19 

Because in a previous in vitro  study we had found that therapeutic concentrations of ketamine had no effect on cytokine secretion from leukocytes,13we hypothesized that an antiinflammatory mediator, such as adenosine, is induced in vivo  after ketamine administration. In the current study, we clearly show that ketamine administration to rats and mice induced a significant increase in plasma and peritoneal adenosine concentrations, which occurs within 20–30 min after ketamine administration. The concentrations measured in sera (peak concentrations approximately 5 μm) and peritoneal concentrations (estimated concentrations 2–3 μm) lie within the physiologic range of activation of the A2AR (dissociation constant = 1.2 nm).20The concentrations of adenosine in the peritoneum are estimated concentrations because we have no data on the volume of intraperitoneal fluid in mice. We measured adenosine concentrations in 5 ml lavage fluid which peak at a range of 0.1 μm. We estimate that the actual peritoneal fluid volume is approximately 0.2 ml; therefore, adenosine peak concentrations are 20- to 30-fold higher (2–3 μm).

Theoretically, adenosine can be produced by any cell in the body; however, the target of ketamine is the NMDA receptor in the central nervous system, which suggests that in this case, adenosine is produced by nerve endings. Strongly supporting this possibility is the work of Bong et al. ,15which demonstrated the role of adenosine in inhibition of neutrophil migration by AP-5, a glutamate NMDA receptor antagonist. Because ketamine and AP-5 are NMDA receptor blockers, it is possible that their action on this receptor in the central nervous system induces the release of adenosine by nerves in the periphery. The obtained peak of adenosine was short and sharp, a fact that can be explained by the rapid clearance of adenosine in sera (half-life = 1.5 s) by cellular uptake and/or metabolism through phosphorylation to adenosine monophosphate or through deamination to inosine.

Adenosine activities are mediated by at least four different receptors; A1 and A3 receptors are Gi protein–coupled receptors, whereas A2A and A2B receptors are Gs protein–coupled receptors that activate adenylate cyclase and cause accumulation of intracellular cyclic adenosine monophosphate. To demonstrate that adenosine mediates the antiinflammatory action of ketamine, we used different agonists and antagonists of adenosine receptors. We found that A2AR activation by the agonist CGS-21680 mimics the effect of ketamine on peritonitis. CGS-21680 significantly reduced peritoneal leukocyte recruitment as well as the increase of cytokine concentrations in blood and in the peritoneum. More direct evidence pointing to adenosine as a ketamine mediator was obtained using the A2R antagonist DMPX and the specific A2AR antagonist ZM 241385. DMPX injected before ketamine blocked the ketamine inhibition of TNF-α and IL-6 release as well as leukocyte recruitment. Similarly, ZM 241385 reversed the inhibitory effect of ketamine on leukocyte recruitment and on TNF-α and IL-6 secretion. In contrast to DMPX, A1R- and A3R-specific antagonists DPCPX and MRS did not block the effects of ketamine. The results of the survival experiments strongly support the involvement of adenosine and its A2AR in the ketamine effect. In these experiments, we demonstrated that the A2AR antagonist ZM 241385 blocked the beneficial effect of ketamine on survival from bacterial sepsis. Ketamine mimicry by the A2AR agonist and the specific blocking of ketamine effects by the A2AR antagonists therefore suggest that ketamine suppresses inflammation through the A2AR. It is important to emphasize that the antiinflammatory activity of adenosine, including inhibition of leukocyte migration and cytokine production, were shown to be predominantly mediated by the A2AR.6Consistent with these data, our findings suggest that A2AR is the dominant receptor of adenosine involved in ketamine activity.

Because ketamine effectively reduces cytokine secretion and mortality from sepsis only when injected immediately or shortly after lipopolysaccharide administration, we hypothesized that the ketamine-induced short adenosine surge halts the inflammatory cascade at its initial steps, perhaps during pre–TNF-α stages. Adenosine and its agonists have been shown to effectively reduce lipopolysaccharide-induced TNF-α release in vitro  by peritoneal macrophages and by dendritic cells21,22and in vivo  in mice injected with lipopolysaccharide.23The narrow timing of the inhibitory effect of adenosine is clearly illustrated in the work of Hasko et al. ,22who showed that adenosine inhibits lipopolysaccharide-induced IL-12 secretion from macrophages only when given shortly after lipopolysaccharide administration. Additional evidence for the short-term action of adenosine is provided by a recent study conducted by Majumdar et al. ,24who showed that TNF-α activation of nuclear factor κB could be blocked only when cells were treated with adenosine at least 30 min before TNF-α stimulation. When adenosine was given at the same time as TNF-α or later, no inhibition of nuclear factor κB was observed. The antiinflammatory effect of adenosine has been demonstrated in many other studies, which have shown that the inhibition of secretion of various chemokines and inflammatory mediators, such as TNF-α, IL-6, and IL-12, is due to activation of A2AR (reviewed in Sitkovsky16).

In addition to ketamine, adenosine is also considered to be the mediator of the antiinflammatory effect of methotrexate, a drug commonly used for the treatment of rheumatoid arthritis25,26and for the antiinflammatory drugs salicylates and sulfasalazine.27 

Given that the antiinflammatory effect of ketamine is effective only at the initial stage of sepsis and because diagnosis is usually made at later stage, after the clinical symptoms of inflammation become manifest, patients with sepsis will probably have a limited benefit from the antiinflammatory effect of ketamine. However, administration of subanesthetic doses of ketamine might be advantageous for patients who have undergone elective surgery or other invasive procedures with a high risk of bacteriemia. It should also be noted that ketamine is beneficial in preservation of cardiovascular function of septic patients as previously described.9 

To conclude, we propose that the sepsis-protective antiinflammatory effects of ketamine are mediated by adenosine. Ketamine administration causes release of adenosine in the periphery, and adenosine through A2AR reduces the systemic inflammatory response by inhibition of secretion of proinflammatory cytokines as well as leukocyte activation and recruitment.

The authors thank Mrs. Valeria Frishman and Robert Yulzari, M.D., Ph.D. (Technicians, Laboratory of Nephrology, Soroka University Medical Center, Ben-Gurion University of the Negev, Beer-Sheva, Israel) for their excellent technical assistance and Mrs. Rita Soullam (Secretary, Department of Nephrology, Soroka University Medical Center, Ben-Gurion University of the Negev) for her invaluable help in writing this article.