INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

Respiratory failure is a common and often lethal complication of severe peritonitis (1, 2). In this inflammatory process, the peritoneal cavity is invaded by activated leukocytes including neutrophils and macrophages that produce and release a variety of inflammatory mediators (3, 4). These include tumor necrosis factor (TNF), interleukin-1 (IL-1), reactive oxygen species (ROS), nitric oxide (NO), arachidonic acid metabolites, and other cytokines that have been shown to impair function of the respiratory muscles (5).

It is possible that direct effects of inflammation on the diaphragm might contribute to the frequent occurrence of respiratory failure in peritonitis. In pilot studies (18), we observed that diaphragm fiber bundles develop contractile dysfunction after being incubated with activated neutrophils. We reasoned that prolonged exposure to peritoneal inflammation might similarly compromise the diaphragm in situ. This postulate was supported by a preliminary report from Motauoakkil and coworkers (19) that peritonitis caused by abdominal inoculation with feces plus barium sulfate caused impairment of the rat diaphragm. The present experiments were designed to formally test the hypothesis that peritonitis causes loss of contractile function in the diaphragm. This paper documents the findings of our study, extending a preliminary report published in an earlier issue of this journal (20).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

Animal Conditioning

Sixty-five male Sprague-Dawley rats (220-306 g) were maintained according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All procedures were approved in advance by the Institutional Review Board of Baylor College of Medicine. Peritonitis was induced by intraperitoneal injection of 10 ml of 6% oyster glycogen (type II; Sigma Chemical, St. Louis, MO) 16 h prior to peritoneal lavage (21). Control animals were not given intraperitoneal injections. On the day of the experiment, all animals were anesthetized by intramuscular injection (0.7 ml/kg) of a combination anesthetic (containing in mg/ml, 42.8 ketamine, 8.6 xylazine, and 0.7 acepromazine). Following induction of anesthesia, a tracheotomy was performed, and each animal was mechanically ventilated with 100% O₂. Peritoneal fluid was obtained by peritoneal lavage with 50 ml of cold Dulbecco's phosphate-buffered saline (Life Technologies, Grand Island, NY). The left hemidiaphragm was excised with ribs and central tendon attached. The hemidiaphragm was placed in Krebs-Ringers solution (containing, in mM, 137 NaCI, 5 KCl, 2 CaCl₂, 1 MgSO₄, 1 NaH₂PO₄, and 24 NaHCO₃) equilibrated with 95% O₂-5% CO₂. Fiber bundles were isolated from the lateral costal diaphragm. A portion of rib and central tendon was retained on each bundle to which suture (4-0 silk) could be tied without compromising constituent fibers.

Leukocyte Analyses

White blood cell differential counts were performed on the lavage fluid. Neutrophils were isolated from the lavage fluid and analyzed for activation by morphologic assay (22); cells were fixed using glutaraldehyde, counted under oil immersion, and classified as normal (inactive), ruffled (active), or bipolar (late stage active). Isolated neutrophils also were tested for extracellular release of reducing equivalents using an adaptation of the cytochrome c reduction assay.

Cytochrome c Reduction Assay

Extracellular release of reducing equivalents was determined using the cytochrome c reduction assay, which detects redox-active molecules including superoxide anions and nitric oxide derivatives (23). Cytochrome c (C-2506; Sigma Chemical) was added to either phosphate-buffered saline (neutrophil assays) or Krebs-Ringer solution equilibrated with 95% O₂-5% CO₂ (diaphragm assays) to achieve 10⁵ M. The test medium was incubated in an opaque container at 37° C with either isolated neutrophils (average concentration 1.6 × 10⁵ cells/ml; incubation time 120 min) or with passive diaphragm fiber bundles (incubation time 90 min). Neutrophils were removed from the test medium by 5 min centrifugation at 1,800 × g; muscle bundles were removed manually; absorbance of the cell-free medium was measured at 550 nm using a spectrophotometer (model 260; Gilford, Oberlin, OH). For comparison, time-dependent absorbance changes were determined using cell-free control solutions incubated under identical conditions. Peak absorbance was determined daily to confirm that maximal range of the assay did not limit detection of biological signals.

Myeloperoxidase Assay

Myeloperoxidase (MPO) catalyzes the conversion of hydrogen peroxide to hypochlorous acid by activated neutrophils; MPO activity is a standard index of tissue neutrophil content (24). We measured MPO activity using a modification of the method described by Mullane and coworkers (25). Neutrophils were isolated from lavage fluid and resuspended in 5 ml phosphate buffer at pH 7.4. Aliquots of 0.5-1 ml were centrifuged (Sorvall RT6000B, Wilmington, DE) at 675 g for 5 min and the supernatant was discarded. Pellets were resuspended in 1 ml 50 mM sodium phosphate buffer (pH 6.0) containing 0.5% hexadecyltrimethylammonium bromide (HTAB; Sigma H5882) and were homogenized for 10 s using a polytron (Brinkman PCU-2; Westbury, NY). Solid tissue samples were added to the same solution to achieve 50 mg tissue/ml; these samples were homogenized × 10 s, were sonicated for three 10-s intervals, underwent three freeze/thaw cycles (cycle = 2 min in liquid N₂, 3 min in 37° C water bath), were sonicated for a final 10 s, then were centrifuged at 1,300 g × 15 min. The supernatant was assayed for MPO activity by mixing 0.01 ml sample with 0.99 ml 50 mM phosphate buffer (pH 6.0) containing 167 µg/ml o-dianisidine hydrochloride and 0.0005% hydrogen peroxide. The change in absorbance at 460 nm was measured at 30-s intervals for 2 min using a spectrophotometer (Shimadzu UV160U; Columbia, MD). The change in absorbance from 30 to 90 s was used to calculate MPO activity where one unit was defined as that which degraded 1 µmol hydrogen peroxide/ min at 25° C (25). All samples were measured in duplicate.

Immunohistochemistry

Solid tissues (diaphragm, limb muscles, heart, spleen) were excised, coated with OCT compound (Miles, Elkhart, IN), snap-frozen in isopentane cooled with liquid nitrogen, and stored at 70° C. Cryostat sections were fixed in 2% paraformaldehyde and processed for immunostaining as described previously (26). We used a commercial polyclonal antibody to inducible nitric oxide synthase (iNOS) (Upstate Biotech, NY) or control rabbit IgG (Sigma) as primary antibodies. After detection of immunolabeling using diaminobenzidine as chromagen, slides were counterstained with hematoxylin, dehydrated, and mounted for microscopy.

Contractile Measurements

Contractile properties were measured in vitro using one fiber bundle from the lateral costal diaphragm of each animal (mean bundle length = 13.3 mm, weight = 18.5 mg, cross sectional area = 1.4 mm², n = 24). The bundle was mounted in a temperature-controlled muscle bath containing 0.025 mM D-tubocurarine chloride in Krebs-Ringers solution equilibrated with 95% O₂-5% CO₂. With use of the attached sutures, the bundle was secured between platinum plate electrodes (4 × 37 mm) and was attached to an isometric force transducer mounted on a micrometer by which muscle length could be adjusted. Transducer output was amplified and displayed on a storage oscilloscope from which contractile properties were recorded. Each bundle was stimulated directly using supramaximal current density (measured characteristics: 550 mA at 13 V), stimulus duration of 0.2 ms, and a train duration of 300 ms. At the beginning of each experimental trial, the bundle was maintained at 25° C to minimize temperature-dependent deterioration. Fiber bundle length was adjusted to maximize twitch force (optimal length). The temperature of the muscle bath was then changed to 37° C and 30 min were allotted for thermal equilibration. Twitch characteristics including twitch force (P_t), time to peak twitch force (TPT), and twitch half-relaxation time (1/2 RT) were measured. Tetanic forces then were measured at 2-min intervals using stimulus frequencies of 200, 15, 200, 30, 200, 40, 200, 50, 200, 80, 200, 120, 200, 160, and 200 Hz; these data were used to compute the relationship between force and stimulus frequency and to monitor stability of the preparation. Acute fatigue was induced over 10 min using repetitive submaximal tetanic stimulations (30 Hz, 500 ms train duration, 0.5 trains/s) during which force was measured. At the end of each experiment, bundle length was measured at optimal length. The muscle was trimmed of bone and connective tissue, blotted dry, and weighed.

Computations

Extracellular release of reducing equivalents was determined by the rate of cytochrome c reduction which was calculated using:

where  absorbance is absorbance of test medium incubated with either neutrophils or diaphragm fiber bundles minus absorbance of test medium incubated without biological material; k is 163,000 pmol/unit  absorbance, and time is the duration of incubation in min. Using methods described previously (27), muscle fiber bundle cross-sectional area was estimated and force measurements were expressed in Newtons per square cm (N/cm²). Relative tetanic forces were corrected for time-dependent deterioration during measurement of the force-frequency relationship using the equation:

where P_f is submaximal tetanic force developed at a given frequency (1-160 Hz) and P_pre and P_post are the P_o (200 Hz) values measured 2 min before and 2 min after P_f, respectively. Relative tetanic force was plotted against stimulation frequency.

Statistics

Results are expressed as means ± SE. Leukocyte responses, bundle characteristics, and baseline contractile properties were compared by one-way analysis of variance (ANOVA). Peritonitis effects on the force-frequency curve and on fatigue characteristics were estimated by two-way repeated-measures ANOVA using mixed-effect models. Muscle was handled as a random effect; fixed effects included frequency (force-frequency data) or time (fatigue data), treatment (control versus peritonitis), and frequency-by-treatment or time-by-treatment interactions. Type III sums of squares were computed for the fixed effects and were estimated using least-squares means. Pairwise comparisons to test a priori hypotheses were performed by linear contrasts. Statistical significance was assigned to p values < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

Peritoneal Inflammation

Prior to anesthetic induction, those animals that had received the oyster glycogen injection appeared to be normal; cage behavior (e.g., posture, exploratory activities, escape maneuvers) was unremarkable and abdominal contact did not elicit an exaggerated withdrawal response. Nevertheless, these animals did develop significant peritonitis. Data in Figure 1 show that the total number of leukocytes present in peritoneal lavage fluid was markedly increased (p < 0.005). This primarily reflected an 800% increase in neutrophil content although macrophage content also doubled. Increases in the numbers of lymphocytes (364,100 cells/ml peritonitis versus 221,700 control) and eosinophils (607,300 versus 265,600) were not significant nor did basophil levels change (12,257 versus 204,000). Morphologic assays indicated that approximately half the neutrophils were either active (45.2 ± 4.0%) or late-stage active (3.2 ± 0.6%); 51.7 ± 4.4% of neutrophils were classified as inactive. Over a 2-h incubation period, neutrophils isolated from animals with peritonitis released reducing equivalents at a mean rate of 8.1 pmol/million cells/min (Figure 2).

View larger version (23K): [in this window] [in a new window]   Figure 1.   Leukocyte composition of peritoneal lavage fluid from control animals (n = 3, open bars) and from animals with peritonitis (n = 14, hatched bars); differences from control: * p < 0.05; ** p < 0.01.

View larger version (20K): [in this window] [in a new window]   Figure 2.   Cytochrome c reduction by neutrophils isolated from animals with peritonitis. Absorbance at 550 nM for cytochrome c solution alone (closed circles) and for cytochrome c solution with neutrophils added (n = 3); see text for details.

Inflammation of Diaphragm and Other Tissues

Immunolocalization studies showed that peritonitis caused superficial adherence of macrophages to the abdominal surface of diaphragm (Figure 3A); leukocytes did not obviously accumulate within the muscle, however. MPO activity of diaphragm homogenates was not significantly increased (1.03 ± 0.10 U/mg peritonitis [n = 22] versus 0.93 ± 0.07 control [n = 6]) nor did diaphragm fiber bundles release reducing equivalents at accelerated rates (3.5 ± 0.2 pmol/mg/min peritonitis [n = 30] versus 3.7 ± 0.5 control [n = 6]). Figures 3C and 3D illustrate the further observation that peritonitis increased inducible NO synthase (iNOS) levels in diaphragm; iNOS labeling was localized to the vascular compartment; diaphragm muscle fibers did not stain perceptibly. Similar patterns of iNOS distribution were observed in limb skeletal muscles, i.e., soleus (Figure 3E-F), extensor digitorum longus, and gastrocnemius (data not shown). iNOS staining was not detected in cardiac muscle or spleen from animals with peritonitis (Figure 3G-I) or in any tissue obtained from control animals (e.g., Figure 3B).

View larger version (195K): [in this window] [in a new window]   Figure 3.   Immunohistochemical labeling for iNOS in (A) diaphragm after peritonitis shows positive anti-iNOS labeling of adherent macrophages, ×200 original magnification; (B) diaphragm after peritonitis is not labeled by incubation with control lgG, ×200; (C ) diaphragm after peritonitis shows positive labeling of microvessels between muscle fibers, ×1,000; (D) microvessels in diaphragm from a control animal were not labeled by anti-iNOS, ×1,000; (E ) soleus after peritonitis; anti-iNOS labeling evident in microvessels at low magnification, ×200; and (F ) at higher magnification, ×1,000; (G) myocardium after peritonitis was not labeled by anti-iNOS, ×200; (H ) spleen after peritonitis was not labeled by anti-iNOS, ×200; (I ) spleen from endotoxemic rat shown as positive control for iNOS labeling, ×200.

Diaphragm Dysfunction

Peritonitis weakened the diaphragm significantly. Maximal tetanic force was depressed at the beginning of in vitro measurements (Table 1), a decrement that persisted throughout the experimental protocol (Figure 4). Twitch characteristics were relatively insensitive to peritonitis (Table 1). Only half-relaxation time was altered; a slight reduction in twitch force was statistically insignificant and time-to-peak tension was unchanged. Figure 5 illustrates peritonitis effects on tetanic force across the full range of diaphragm activation; absolute forces were diminished at all stimulation frequencies above 30 Hz. Peritonitis had little effect on acute fatigue during 10 min of repetitive, submaximal tetanic contractions (Figure 6); force output was depressed during the first 2 min of the protocol but was identical to control thereafter.

View larger version (16K): [in this window] [in a new window]   Figure 4.   Repeated measurements of maximal tetanic force (Po) developed by diaphragm fiber bundles from animals with peritonitis (n = 17, open circles) and control animals (n = 7, closed circles); the overall rate of decline of Po was not different between groups (3.4 ± 1.5%/28 min control versus 3.2 ± 0.8% peritonitis).

View larger version (17K): [in this window] [in a new window]   Figure 5.   Force/frequency relationships of diaphragm fiber bundles from control animals (n = 7, closed circles) and animals with peritonitis (n = 17, open circles); the relationships differed significantly overall (p < 0.001) and at each stimulus frequency > 30 Hz; * p < 0.05.

View larger version (16K): [in this window] [in a new window]   Figure 6.   Acute fatigue of diaphragm bundles from control animals (open circles) and animals with peritonitis (closed circles) during repetitive submaximal tetanic (30 Hz) stimulation; relative force differed between groups at times 15-120 s (p < 0.0001 overall); * p < 0.05, ** p < 0.01.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

Intraperitoneal injection of oyster glycogen induced acute peritonitis in rats as demonstrated by gross accumulation of macrophages and activated neutrophils within the peritoneal space. We found no evidence of leukocyte accumulation within the diaphragm. Peritonitis caused upregulation of iNOS in the vasculature of diaphragm and other skeletal muscles. Peritonitis also caused generalized diaphragm weakness; force production was diminished during both maximal and submaximal tetanic contractions.

Peritonitis Model

The current method for inducing peritonitis has been used previously to elicit activated neutrophils for ex vivo study (21). The oyster glycogen injection is a relatively modest insult; our experimental animals did not appear to be sick and did not exhibit withdrawal behavior in response to abdominal touch. Nevertheless, glycogen injection did produce an inflammatory response. The numbers of macrophages and neutrophils detectable in peritoneal lavage fluid were markedly increased over control values. Neutrophils isolated from the lavage fluid were activated, based on morphologic appearance, and released reducing equivalents into the extracellular space. We made no attempt to identify the molecular composition of these reducing equivalents but they likely included both reactive oxygen intermediates and nitric oxide derivatives (22, 28, 29). Activated leukocytes did not appear to migrate from the peritoneal space into the diaphragm. We detected no increase in tissue activity of MPO, a marker enzyme for neutrophils (24, 25); diaphragm fiber bundles did not release reducing equivalents at exaggerated rates; nor did iNOS immunostaining identify interstitial macrophages within the muscle. This differs from the effects of endotoxemia, which stimulates infiltration of inflammatory cells into the perivascular space of rat diaphragm (14).

Inflammation was not restricted to the peritoneal cavity, however. iNOS staining was increased in the microvasculature, a systemic response that was evident in limb skeletal muscles as well as diaphragm. In contrast, we did not detect iNOS staining within the fibers of any muscle examined. Our findings closely mimic the pattern of iNOS staining observed by Thompson and colleagues (30) in diaphragm and limb muscle of mice injected with Escherichia coli endotoxin: microvessels and macrophages stained for iNOS, muscle fibers did not. Other investigators reported more widespread upregulation of iNOS following systemic endotoxemia. Boczkowski and coworkers (14) examined the diaphragms of rats 12-24 h after lipopolysaccharide (LPS) inoculation and found positive iNOS staining in 40-45% of total fiber cross-sectional area. Gath and coworkers (15) found that guinea pigs pretreated with LPS exhibited iNOS upregulation in diaphragm and that iNOS expression was restricted to slow muscle fibers. Thus, it appears that individual cell types within the diaphragm differ in their sensitivity to inflammatory mediators, in their regulation of iNOS expression, or both. Interestingly, peritonitis did not appear to upregulate iNOS in capillary endothelium or other cells of the heart and spleen of our animals. This suggests that humoral mediators of peritoneal inflammation evoke tissue-specific responses and that effects on the vascular endothelium of skeletal muscle can differ from those observed in other tissues.

Diaphragm Dysfunction

Peritonitis produced decrements in force production of the unfatigued diaphragm at most stimulus frequencies. Such losses attest to weakening of the muscle and document loss of contractile function, confirming our original hypothesis. In contrast, peritonitis had little effect on acute fatigue of the excised muscle. These findings closely resemble the pattern of diaphragmatic dysfunction observed by Boczkowski and coworkers (14) following 12-24 h of systemic endotoxemia.

It was not surprising that peritonitis caused diaphragm weakness. The muscle was exposed to a localized inflammatory process that included upregulation of iNOS in the diaphragm microvasculature, accumulation of activated neutrophils in the adjacent peritoneal space, and direct adhesion of macrophages to the diaphragm surface. These processes implicate free radicals and their redox derivatives as potential mediators of the resulting pathology. iNOS was clearly identifiable within the vasculature of diaphragm isolated from animals with peritonitis. Unlike constitutive isoforms of NO synthase, NO formation by iNOS is not calcium-regulated; synthesis is limited only by substrate availability when cofactors of the enzyme are available (31). High rates of unregulated NO production would favor formation of highly reactive NO derivatives (e.g., peroxynitrite, peroxynitrous acid) in and around diaphragm microvessels and would predispose adjacent muscle fibers to nitrosative injury (32). In addition, the adjacent peritoneum was flooded with activated leukocytes. Macrophages were directly attached to the abdominal surface of the muscle; these cells stained heavily for iNOS and represented a second source of unregulated NO synthesis. Activated neutrophils in the peritoneal space represented an additional site of free radical synthesis. These cells contain NADPH oxidase, myeloperoxidase, and other membrane-bound oxidoreductase systems that generate reactive oxygen intermediates as part of their integrated inflammatory response (29). It is likely that parallel increases in NO derivatives and reactive oxygen intermediates contributed to diaphragm dysfunction in peritonitis, a mechanism similar to that which injures the diaphragm in endotoxemia (5, 7, 8, 14, 15).

Other inflammatory mediators also may have contributed. In the human disease, activated leukocytes invade the peritoneal cavity and release tumor necrosis factor (TNF), interleukin-1 (IL-1), arachidonic acid metabolites, and other cytokines (3, 4). These mediators can cause muscle wasting and impair contractile function of the diaphragm (10, 16, 17) although such effects may not be direct (13, 17, 33). It is likely that inflammatory mediators were absorbed by the mesenteric microvasculature and distributed systemically. The supposition that our rats developed a low level of systemic inflammation is supported by the observation that iNOS was upregulated in the microvessels of limb skeletal muscles. The limb muscles were not directly exposed to peritoneal inflammation and must have responded to humoral signals in the vascular compartment.

Clinical Relevance

Peritonitis is a frequent cause of respiratory failure in humans, resulting in ventilator dependence and prolonged stays in the intensive care unit. The combination of peritonitis and respiratory failure result in mortality rates of 80% (34) to 89% (2). A number of factors contribute to the respiratory complications seen in peritonitis. Lung function may be compromised by pulmonary edema, pulmonary embolization, and bronchopneumonia (2), processes that both perturb gas exchange and increase the work of breathing. The present study suggests that respiratory muscle dysfunction may aggravate this clinical picture. Direct exposure to peritoneal inflammation is likely to weaken the diaphragm, limiting the ventilatory reserve of patients with peritonitis and predisposing them to respiratory failure. Further, losses in contractile function of diaphragm myocytes may be compounded by reflex inhibition of neural drive to the muscle as occurs following laparotomy (35). Respiratory muscle function has not been evaluated in humans with peritonitis but the present data provide a rationale for such research.

	CONCLUSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

Experimental peritonitis causes generalized diaphragm weakness in rats. The most likely mediators of muscle dysfunction are free radicals and inflammatory cytokines produced within the peritoneal cavity. A similar process may compromise the diaphragm of humans with peritonitis thereby contributing to the frequent occurrence of respiratory failure in affected individuals.