Nitric Oxide, Bacteria, and Host Defense in Sepsis: Who Needs What?

Nitric oxide (NO) is an important element of host defense, a central component of innate immunity, and an effective antimicrobial agent (1). It is synthesized by the NO synthase (NOS) family of enzymes that include the calcium-dependent, constitutive (cNOS) isoforms (neuronal [nNOS], NOS-1, endothelial [eNOS], and NOS-3) and the calcium-independent, inducible isoform (iNOS and NOS-2) (2). The constitutive forms are responsible for a constant, low rate of NO production, which can be increased acutely for a short time (in the nanomolar range) as part of acute response mechanisms. The inducible form is expressed on demand and needs a longer time (hours) to be activated (3), but once expressed, produces micromolar amounts of NO. There is also an important interdependency between the isoforms, including a mutual regulatory effect (4). It is now evident that NO plays a pivotal role as both “friend and foe” in many disease states, including sepsis and septic shock (5). The NO-generated by cNOS is an important messenger for body homoeostasis, including maintenance of vascular tone (6), and bears anti-inflammatory properties. The larger amounts of NO resulting from iNOS expression are at least partially responsible for sepsis-induced vasodilatation and cardiac depression, and therefore contribute to hypotension and multiple organ dysfunction (4). NO is also both an important anti-inflammatory regulator and powerful effector. The suggested mechanisms for direct microbicidal effects are inactivation of DNA-replicating and mitochondrial respiratory enzymes as well as antioxidant metalloenzymes due to induction of DNA stand breaks and formation of iron-nitrosyl complexes and peroxynitrite (ONOO−) (7). However, these mechanisms, especially ONOO− formation, may also be detrimental for the host. Thus, a large number of studies have addressed the role of NO production or NOS blockade, respectively, on bacterial clearance or direct antibacterial effects (i.e., phagocytosis or respiratory burst), but the results arising from these studies are unfortunately still controversial. Citrobacter rodentium-infected iNOS−/− mice showed a delayed bacterial clearance from the gut wall when compared with wild-type animals (1). Escherichia coli increased nitrate concentrations in blood in iNOS+/+ mice, but not in iNOS−/− mice, but there was no difference in bacterial clearance or persistence between the two genotypes (8). In a murine model of sepsis induced by inhalation of Klebsiella pneumonia, the pulmonary clearance of bacteria was even enhanced in iNOS−/− animals, but was decreased in nNOS−/− ones (9). Pharmacological NOS manipulation yielded similarly variable results: In mice, the nonselective NOS-inhibitor L-NAME decreased the number of circulating neutrophils and lymphocytes, which was partially reversed by l-arginine (10). A nonlethal rat model of intraperitoneal inoculation of Escherichia coli turned out to be lethal with administration of another nonselective NOS inhibitor (L-NMMA), which also caused higher peritoneal bacterial counts. The same effect was observed by adding erythrocytes into the peritoneum (11). Moreover, arginine supplementation improved survival affiliated with enhanced bacterial clearance in a mouse model of cecal ligation and puncture, and this effect was inhibited by nonspecific NOS blockade (12). Finally, NO inhalation improved bacterial clearance from bronchoalveolar fluid in rats after alveolar instillation of Pseudomonas aeruginosa, and NO was directly bactericidal in vitro, although ex vivo alveolar neutrophil functions were not altered by NO (13). In contrast, in mice infected with Legionella pneumophila L-NAME did not significantly change bacterial clearance from the lung (14), and bacterial transit time during intestinal bacterial translocation was decreased by NO supplementation with l-arginine and increased with selective iNOS or nonselective NOS blockade in rats (15). In the August issue of Shock, Stehr et al. (16) further address the controversial issue of NO in the host defense system. Thirty minutes after starting continuous infusion of the nonselective NOS inhibitor L-NAME, anesthetized and mechanically ventilated rabbits were challenged with a single dose of Escherichia coli and were observed for 3 h. L-NAME transiently maintained higher blood pressures, but ultimately increased mortality, which was associated with delayed bacterial clearance from the blood, enhanced organ bacterial counts, and decreased neutrophil phagocytic activity despite a rise in oxidative burst. The authors conclude that NO production is crucial for host defense during microbial invasion and, hence, that NOS inhibition is deleterious under these circumstances, thus at least in part explaining the failure of the recent double-blind, randomized, multiple-center trials using the nonselective NOS inhibitor L-NMMA. These studies provided data on improved shock resolution (17), increased vascular tone, but also reduced cardiac index, as well as oxygen delivery (18), and increased mortality in the treatment arm (19). The authors’ data on bacterial clearance are in sharp contrast to other, large animal studies mostly showing no deleterious effect of NOS inhibition on host defense, even when nonselective compounds were used. Dogs pretreated with stroma-free cross-linked hemoglobin, a well-known scavenger of NO (20), were challenged with i.v. Escherichia coli, and no changes were found in bacterial or endotoxin clearance (21). In sheep, L-NMMA treatment 24 h after induction of hyperdynamic sepsis with i.v. P. aeruginosa did not interfere with the pulmonary clearance of bacteria (22). In another report on this model, pulmonary bacterial clearance decreased over time irrespective of the treatment, and was only transiently lower in L-NMMA-treated animals when compared with controls (23). In the same model, pyridoxalated hemoglobin polyoxyethylene conjugate even decreased bacterial count in the organs, although pulmonary bacterial clearance did not change when compared with controls (24). The results relating to bacterial content of tissues are equally controversial. Stehr et al. (16) report increased tissue bacterial content, whereas Lingnau et al. (23) report just the opposite for the same tissue evaluated and after using the same drug (L-NAME). A possible explanation could be the different study design: While Stehr et al. used a pretreatment approach, a short-term bacterial challenge and only 3 h of observation period, Lingnau et al. investigated full-blown hyperdynamic sepsis with a posttreatment strategy as well as a markedly longer, 24-h observation period. This difference in design presumably affected the degree of inhibition of the different NOS isoforms: The pretreatment design probably fully blocked eNOS and nNOS and prevented iNOS activation, which was probably only activated at the end of the experiment even in the control group of the Stehr (25) study, and, therefore, may have also attenuated the antibacterial effect of NO (11). In contrast, during a late, posttreatment approach, all isoforms of NOS, especially iNOS, are active, and the constitutive forms could act already during the bacterial challenge, thus reducing tissue bacterial invasion (15). What other factors could explain the discrepancies between the study reported here and previous studies reported by Dr. Traber’s group (26) in Galveston? The different methods of bacteria administration (bolus administration by Stehr et al., whereas Traber’s group used continuous infusion), evaluation of bacterial clearance (arterial blood clearance by Stehr et al. and transpulmonary bacterial clearance by Traber’s group), or the different bacteria used to establish sepsis (Stehr et al. infused Escherichia coli, whereas Traber’s group administered P. aeruginosa) may have assumed importance in this context. Finally, despite the elegant design and the rigorous methodology applied, another important issue, e.g., the choice of the species, needs to be taken into account when evaluating the study by Stehr et al. The smaller the species (mouse, rat, or rabbit) used in sepsis research, the more concerns are raised regarding transferability of data obtained from these small animal models to human medicine without using large animal (sheep, swine, or monkey) investigations (26, 27), in particular when standard intensive care unit monitoring and supportive measures, e.g., fluid resuscitation, are not integrated into the experimental procedure. Unfortunately, Stehr et al. did not measure any hemodynamic parameters other than mean arterial pressure. Thus, they were unable to discriminate if the well-known vasoconstrictor effect of L-NAME lead to a marked reduction of cardiac output and oxygen delivery (28) and hence contributed to the observed increase in mortality. Clearly, the high lactate levels and pronounced metabolic acidosis together with the hypotension observed in both groups at the end of the observation period is strongly suggestive of tissue hypoperfusion affiliated with a hypodynamic shock state. Thus, a detrimental effect of the increased vascular resistance and, hence, decreased tissue perfusion in the L-NAME-treated animals may have been a confounding factor for the results on antibacterial processes. In the light of the interspecies variability of NO production between different rodents (29), but in particular with respect to higher species (26, 30) and the nonselectivity of the intervention with respect to the NOS isoenzymes, it appears prudent to be cautious when transferring the present data to the daily care of critically ill patients.