INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

SYSTEMIC COMPLEMENT ACTIVATION has been noted in a variety of shock states, including sepsis and severe trauma. In human studies, the degree of activation correlates with severity of injury, development of multiple organ failure, and death (5-9, 14, 23, 26). Similar observations have been made in animal studies, and there is growing evidence that, in addition to being important proinflammatory effectors, products of complement activation may contribute directly to more generalized manifestations of shock, such as acidosis, hypotension, and O₂ consumption (12, 15, 25). Complement activation may also participate adversely in other major systemic stresses, such as cardiopulmonary bypass, where recent clinical work has shown that treatment with a humanized monoclonal antibody to C5 reduces requirements for the postoperative transfusion of blood products and may improve neuropsychiatric function postoperatively (4).

The influence of complement activation products on hemodynamics is complex and varies with the species studied and the complement product generated. In 1967, it was noted that hypo- and hypertensive responses in guinea pigs infused with activated rat or porcine serum were blocked by cyclooxygenase (COX) inhibitors and antihistamines (3). Similar hemodynamic responses have been seen in dogs intravenously injected with yeast-activated porcine serum (16). Mediators implicated in these responses include histamine, prostanoids, and catecholamines (3, 10, 16). Much of this work must be interpreted with care, inasmuch as the infusion of activated serum and, in particular, xenogeneic serum suggests that these adverse responses may, in part, reflect complement-dependent acute vascular damage induced by natural IgM antibody reacting with epitopes on endothelial cells.

One means of accomplishing acute endogenous complement activation without the use of activators such as inulin or yeast is with cobra venom factor (CVF). CVF is a C3 convertase that initiates the formation of C3a, C5a, and the terminal complement complex independently of the classical, alternative, or mannose-binding lectin complement pathways. Intravascular administration of CVF produces widespread neutrophil activation and has been used as a model of acute lung injury (Ref. 22; S. Sawada, K. Matsuda, K. Johnson, J. Younger, R. Bartlett, and R. Hirschl, unpublished observations). CVF acts at a single step in the complement cascade (i.e., conversion of C3 to C3a and C3b), making it a potentially useful tool for examining changes related to systemic complement activation. This model also allows for the direct evaluation of complement activation products in the absence of other confounding factors such as infectious agents, blood loss, or mechanical trauma.

In the present studies we examined a number of responses to systemic activation of complement with CVF, including blood pressure, metabolic acidosis, changes in vascular permeability, and lung function. High doses of CVF produced circulatory collapse, extravasation of radiolabeled protein into the lung, and modest decreases in arterial PO₂. Prior depletion of complement protected against these abnormalities. Other interventions, including neutrophil depletion and COX inhibition, prevented lung injury but had much less effect on systemic hemodynamics or gut permeability. Inhibiting carboxypeptidase N (CPN), an enzyme that converts C3a and C5a to their less-active des-Arg forms, significantly worsened injury.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

All reagents were purchased through Sigma Chemical (St. Louis, MO) unless otherwise noted. Specific-pathogen-free male Sprague-Dawley rats (300-350 g) were obtained from Harlan Industries (Indianapolis, IN).

Purification of CVF. CVF was prepared from the whole venom of Naja naja kaouthia first by dialysis against 40 mM PBS for 24 h. Anion-exchange chromatography using a DEAE-cellulose column (DE52, Whatman, Kent, UK) was used to isolate the CVF-containing protein fraction (1). One unit of CVF activity was defined as the amount of CVF necessary to reduce the hemolytic activity in 1 ml of human serum by 50%. Enzymatic activity of secretory phospholipase A₂ (sPLA₂), which is known to coelute with CVF during exchange chromatography, was measured in the elution fractions and in the final pooled CVF preparation with the use of a commercially available kit (Cayman Chemical, Ann Arbor, MI). PLA₂ activity was expressed as the quantity (in mmol) of a chromophore hydrolyzed from the sn-2 position of 1,3-diphosphatidylglycerol per minute per milliliter of solution.

Rat model of complement activation. Before experimentation, rats were allowed free access to food and water. After induction of anesthesia and analgesia with subcutaneous ketamine hydrochloride (50 mg/kg) and xylazine hydrochloride (10 mg/kg), carotid arterial and jugular venous catheters were placed and a tracheostomy was performed. To assess intracardiac pressures, in some animals the carotid catheter was advanced into the left ventricle. Mechanical ventilation (inspiratory O₂ fraction = 1.0, tidal volume = 8 ml/kg, respiratory frequency = 72 breaths/min, positive end-expiratory pressure = 2 cmH₂O) was instituted using a piston ventilator (model 643, Harvard Apparatus, South Natick, MA). Core temperature was monitored with a rectal thermometer and maintained at 37-39°C. Animals were allowed 30 min to recover after instrumentation. To activate circulating complement, CVF (12.5 U/kg in 0.5 ml of lactated Ringer solution; Abbott Critical Care, North Chicago, IL) was injected intravenously. This dose was found in preliminary experiments to produce profound but nonlethal hemodynamic compromise 30 min after injection. At the conclusion of the experiment, animals were euthanized by exsanguination from the abdominal aorta. All surgical protocols were approved by the local animal use committee and were in compliance with local and federal regulations.

Physiological measurements. Mean arterial blood pressure (MAP) was measured from the carotid artery with the use of a transducer (Transpac IV, Abbott Critical Care, North Chicago, IL) interfaced to a digital data acquisition system (model MP100A, Biopac, Santa Barbara, CA). Heart function was evaluated by heart rate, left ventricular developed pressure, rate-pressure product, and positive and negative change in pressure over time (dP/dt). Metabolic acidosis was gauged by serum bicarbonate concentration calculated from arterial blood-gas samples (model ABL 606, Radiometer, Copenhagen, Denmark). The same clinical blood-gas analyzer was used to measure arterial PO₂. Static lung compliance (C_s) was determined immediately after euthanasia and was expressed as the change in volume divided by the change in pressure from 1 to 10 ml of instilled gas.

Vascular permeability to protein. Extravasation of intravascular protein was measured with ¹²⁵I-labeled BSA. At the beginning of the experimental protocol, animals received 2 × 10⁶ counts/min intravenously. At the conclusion of the protocol, 1 ml of blood was reserved. After euthanasia, the proximal aorta was ligated, the main pulmonary artery was incised, and 10 ml of saline were gently injected into the left ventricle, flushing the cardiac and, in a retrograde fashion, the pulmonary circulation. These organs were placed in scintillation vials. The small intestine, from the duodenojejunal flexure to the cecum, the left kidney, and a portion of the liver and the hemidiaphragm were flushed of intravascular blood by instillation of 30 ml of lactated Ringer solution into the thoracic aorta at the diaphragmatic crux over 30 min by means of a syringe pump. The organs were harvested and placed in scintillation vials. Radioactivity of blood and organs was measured using a gamma counter (Beckman Instruments, Irvine, CA). An injury index for each organ was defined as the organ radioactivity per gram of wet tissue divided by the radioactivity in 1 g of blood.

Extravascular tissue water. Erythrocytes from a donor rat, after incubation for 1 h with Na⁵¹Cr, were washed and resuspended in buffered saline until a labeling efficiency of >98% was achieved. Animals being studied for extravascular water content received 2 × 10⁶ counts/min of labeled cells in a volume of 0.5 ml 5 min before the conclusion of the experiment. Immediately before euthanasia, 1 ml of arterial blood was recovered. Radioactivity in the blood reference and in the heart, lung, and gut was measured, and wet organ weights were determined. Dry organ weight was measured after 10 days of dehydration at 80°C. Extravascular tissue water was expressed as a wet-to-dry weight ratio corrected for the weight of intravascular blood.

Inhibition or augmentation of complement activation. To confirm whether the abnormalities observed after CVF administration were specifically a result of generation of complement-derived anaphylatoxins, some animals were complement depleted before study. These rats received CVF (30 U/kg ip) 36, 24, and 12 h before the experimental protocol. This treatment produced slow continuous complement consumption, such that, on the day of the experiment, complement-mediated hemolytic activity of the animal's serum was undetectable using the total hemolytic complement technique (CH₅₀ < 20). In other animals, complement activation was augmented by inhibition of CPN. This enzyme cleaves the terminal arginine from C3a, C4a, and C5a (the anaphylatoxins), yielding the less active des-Arg form of each; inhibition of CPN thus may potentiate anaphylatoxin activity (17). CPN activity was blocked by administration of DL-2-mercaptomethyl-3-guanidinoethylthiopropanoic acid (100 mg/kg ip; Calbiochem, La Jolla, CA) 30 min before CVF injection. This agent has been shown to reduce CPN activity >90% for 3 h (11). Preliminary experiments in our laboratory revealed that CPN inhibition followed by 12.5 U/kg CVF was frequently lethal within minutes of CVF injection. Therefore, a reduced CVF dose of 6.25 U/kg was used in CPN-inhibited animals.

Neutrophil depletion. In other animals, neutropenia was established before experimentation by injection of 0.5 ml of rabbit anti-rat neutrophil antiserum intraperitoneally 18 h before experimentation. Neutropenia was confirmed at the time of instrumentation by complete absence of neutrophils on a hematoxylin- and eosin-stained peripheral smear.

COX inhibition. Nonspecific COX and COX-2-specific inhibitors were studied. COX-1 and COX-2 were nonspecifically inhibited with 9 mg/kg of ketorolac administered orally as the tromethamine salt (Cayman Chemical) dissolved in PBS. The COX-2 inhibitor NS-398 (Cayman Chemical) was prepared by suspending the powdered drug into warm soybean lecithin, which was then sonicated in PBS for 10 min to form a suspension that was administered orally in a dose of 5 mg/kg. Ketorolac and NS-398 were given 60 min before CVF administration, a period previously demonstrated to yield COX-inhibitory serum concentrations (21).

Effect of contaminating sPLA₂. Potential confounding effects of cobra sPLA₂ that had coeluted during anion-exchange chromatography were also studied. In some animals, commercially available sPLA₂ isolated from the venom of N. naja was administered intravenously in a dose equipotent to the measured contaminating sPLA₂ activity in 12.5 U/kg of CVF.

Preparation of tissue for histology. To correlate quantitative measures of injury with histological appearance, lung, heart, and small intestinal tissue from some animals was fixed in a glutaraldehyde-cacodylate buffer. Representative images from 1-µm slices of the organs of three control and three CVF-treated animals were photographed.

Statistical analysis. Values are means ± SD. In all groups, n  5. Initial comparisons between groups were performed using ANOVA. Blood pressure responses between control and CVF-treated animals were examined using repeated-measures techniques. Otherwise, post hoc comparisons between groups were limited to 1) a comparison of the 12.5 U/kg CVF group (serving as the positive control) with sham animals and with the interventions that might lessen or intensify the injury (i.e., complement and neutrophil depletion and COX and CPN inhibition) and 2) a comparison of the sham group (serving as the negative control) with the sPLA₂-treated group. Pressure-volume curves obtained during C_s measurements were analyzed with repeated-measures ANOVA. All analyses were carried out using SAS 6.12 software (SAS Institute, Cary, NC).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Response to CVF. Intravenous injection of CVF produced a characteristic blood pressure response, consisting of a brief (<5 min) and modest period of hypertension and widened pulse pressure followed by progressive, severe, and sustained hypotension (Fig. 1). There was a corresponding deterioration in heart performance, as demonstrated by decreased developed pressure, contractility, and diastolic relaxation (Table 1). A compensatory tachycardia, which would be predicted given the degree of depressed contractility, was not seen in the CVF-treated animals. This finding suggests that the impaired cardiac function may be mediated in part vagally. These hemodynamic changes were associated with metabolic acidosis and decreased arterial PO₂ (Table 2). When measured at the 30-min time point, C_s did not decrease after CVF infusion (Fig. 2). CVF-treated animals also exhibited increased accumulation of ¹²⁵I-BSA in the lung and in the intestine but not in the heart (Fig. 3, Table 3). An increase in extravascular tissue water was noted only in the small intestine (Table 4). Permeability measures did not appreciably change in the kidney, liver, or diaphragm after infusion of CVF (data not shown). These data suggest that, after systemic activation of complement, the lung and gut are susceptible to albumin leak but not the heart, liver, or kidney. Reasons for these discrepant responses are not clear.

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Mean arterial pressure (MAP) response to cobra venom factor (CVF, 12.5 U/kg). Intravascular activation of complement consistently produced progressive and severe hypotension. Values are means ± SD; n  5 in each group. *P < 0.05 by repeated-measures ANOVA.

View larger version (14K): [in this window] [in a new window]   Fig. 2.   Static lung compliance after systemic activation of complement. There was no change in static lung compliance after CVF injection. Values are means ± SD; n = 5 in each group. Values are statistically not significant by repeated-measures ANOVA.

View larger version (170K): [in this window] [in a new window]   Fig. 3.   Representative histological findings. All images represent 1-µm sections stained with toluidine blue. A and B: lung tissue from sham and CVF-treated animals, respectively. Original magnification: ×100. CVF was associated with an intravascular influx of neutrophils with preservation of the alveoli. C and D: myocardium from sham and CVF-treated animals, respectively. Original magnification: ×100. No morphological differences were noted. E and F: distal ileum from sham and CVF-treated animals, respectively. Original magnification: ×40. CVF was associated with submucosal edema and patchy loss of intestinal villi.

Inhibition or augmentation of complement activation. Prior complement depletion provided significant protection from all abnormalities seen with CVF injection, supporting the hypothesis that CVF produces injury through a complement-dependent pathway (Tables 2 and 3). Prior inhibition of CPN greatly potentiated CVF-associated hemodynamic collapse and loss of vascular integrity.

Neutrophil requirements for CVF-induced abnormalities. Neutrophil depletion improved oxygenation and reduced extravasation of protein into the lung (Tables 2 and 3). This treatment, however, had much more modest effects on mean arterial pressure, metabolic acidosis, or vascular permeability (albumin leak) in the gut. These results indicate that neutrophil-dependent and -independent pathways mediate the tissue-adverse effects of complement activation in the lung and gut, respectively.

Effect of COX inhibition. Nonspecific inhibition of COX-1 and COX-2 with ketorolac was similar to neutrophil depletion, in that the most protective effects were limited to the lung. Treatment of rats with ketoralac greatly reduced the albumin leak into the lung and had much less protective effect against vascular leak into the gut (Table 3). The COX-2-specific inhibitor NS-398 did not ameliorate any measured parameter, suggesting that COX-1 is the enzyme isoform responsible for the pathophysiological abnormalities.

PLA₂ activity in purified CVF and the effect of intravenously administered PLA₂. Analysis of effluent from the chromatography column used to purify the CVF demonstrated coelution of PLA₂ with CVF activity. The PLA₂ activity of the final CVF preparation was 60 mmol · min¹ · ml¹, or 6 mmol · min¹ · 12.5 U/kg CVF¹. Therefore, 6 mmol/min of sPLA₂ were injected into rats. Administration of this enzyme in the absence of CVF produced no detectable changes in blood pressure, serum bicarbonate concentration, arterial oxygenation, or lung or intestinal injury index.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this rat model of acute systemic activation of complement, CVF injection (12.5 U/kg) produced rapid, persistent, near-lethal hypotension and accompanying metabolic acidosis. These derangements are different from those reported in the hemodynamic response of guinea pigs to activated porcine serum, characterized by a brief episode (~45 s) of hypotension followed by elevated blood pressure for another several minutes (3). The hypotension that occurred in our experiments was accompanied by a large decrease in left ventricular performance, suggesting that hypovolemia or intrinsic myocardial dysfunction was present. In recently reported isolated perfused heart experiments, our laboratory found that, within 10 min of ex vivo perfusion, hearts harvested from CVF-intoxicated rats were functionally no different from hearts retrieved from normal animals (25). Additionally, in the present studies, no difference in myocardial permeability to ¹²⁵I-BSA was found after CVF injection, nor was there an increase in cardiac extravascular water, suggesting that hypovolemia and underfilling of the left ventricle are the most likely explanation for the perturbation in cardiac performance after CVF injection. Our findings therefore are more consistent with those described in dogs, where MAP and cardiac output fell by 50% after administration of activated porcine serum (16). In the present experiments, these hemodynamic changes were largely independent of neutrophils and products of COX and, of the interventions tested, were ameliorated only by prior depletion of complement.

The nature of the injury in the lung appeared to be distinct from the abnormalities occurring systemically and in the intestine. Administration of antineutrophil serum before experimentation provided a 70% reduction in the vascular leak of protein. Nonspecific blockade of COX with ketorolac was similarly protective. The COX-2-specific inhibitor NS-398 failed to provide protection, suggesting that COX-1, rather than the COX-2 isoform, is responsible for the early changes in lung function seen in this model. This pattern is in striking contrast to changes in the gut, in which prior complement depletion was highly protective, but neither neutrophil depletion nor treatment with ketoralac provided substantial benefit. These results suggest that complement-mediated changes in the lung proceed along pathways that are quite different from those that simultaneously affect the gut after systemic infusion of CVF. This pattern of difference, at least involving the role of neutrophils, is similar to other reports in which ischemic injury of gut seems to develop in a manner that, although classical pathway and Ig dependent, is neutrophil independent (24).

Although ¹²⁵I-BSA accumulated in the lung during complement activation, there was no detectable change in extravascular lung water as measured by the wet-to-dry weight ratio. Normal extravascular lung water content after CVF administration correlates with our observations of unchanged C_s, only modest decreases in arterial PO₂, and alveolar sparing noted histologically. Further studies would be useful in examining any protective mechanisms, such as enhanced clearance of alveolar or interstitial water, which might be engaged by the complement system. Given the overall preservation of alveoli histologically and the maintenance of normal C_s, it is most likely the diminished arterial PO₂ observed in this model is a reflection of severely depressed mixed venous oxygenation in the presence of right-to-left shunting, rather than intrinsic failure of gas exchange within the lung.

Assessment of the integrity of the endothelium using labeled albumin has certain limitations that may be relevant to the present studies. Protein may accumulate in the interstitium as a result of increased transcapillary flow, deterioration of the reflection coefficient of the endothelial boundary, or a decrease in lymphatic drainage of the organ (2). In neither the lung nor the intestine are our methods capable of clearly distinguishing which mechanism (or combination thereof) is responsible for the increased injury index observed. In the intestine, the presence of low perfusion pressure, as indicated by severely reduced MAP, suggests that transcapillary flow was likely reduced and that, therefore, accumulation of protein was a result of failure of the endothelial barrier or a significant decrease in lymphatic outflow.

Without a direct measurement of pulmonary arterial pressure, the role of intravascular hydrostatic forces present in the lung during complement activation is not known, and to what extent the elevated albumin leak into the lung is due to impaired endothelial barrier function or increased hydraulic forces cannot be determined. Acute pulmonary vasoconstriction would promote transvascular water and protein flux, could potentially cause dysregulation of ventilation-perfusion matching, resulting in decreased arterial oxygenation, and might impair left ventricular filling, resulting in decreased myocardial performance and systemic hypotension. Our observations that COX inhibition and neutrophil depletion significantly improved lung permeability and oxygenation but had little effect on systemic hemodynamics suggest that a process more complex than isolated pulmonary vasoconstriction is responsible for the abnormalities noted.

As an inducer of C3 convertase, CVF produces anaphylatoxins and the terminal complement complex (C5b-9), both of which may be participating in our model. Additional study will be required to determine the relative contribution of these two mediators. C5a, the most likely anaphylatoxin contributing to the abnormalities seen in this model, is an 8-kDa fragment (if glycosylation is not taken into account) of the much larger C5 protein. An antibody with sufficient affinity for C5a and without cross-reactivity to the parent C5 (the normal function of which is required for the development of the terminal complement complex) is under development. Evidence for the importance of C5a remains indirect in the absence of blocking experiments. The lethality of CPN inhibition seen in our experiments suggests that the anaphylatoxins are the relevant mediators of organ damage after systemic activation of complement.

Phospholipase activity was noted to coelute with CVF during anion-exchange chromatography, but administration of an equipotent dose of PLA₂ isolated from Naja venom produced no detectable effect in the animals studied, indicating that any PLA₂ in the CVF preparation does not account for the observed organ injury. Although the physiological derangements developing after CVF administration did not occur in complement-depleted animals, complement depletion was not entirely protective of changes in MAP (82% preservation of MAP) or serum bicarbonate (89% preservation, Table 1). Thus it is possible that some other activity within the purified CVF may be directly linked to biochemical changes found in this model.

CVF is a useful agent for rapidly activating C3, yet it is unlikely that this intensity occurs in human conditions encountered clinically. However, our findings are consistent with recent animal studies using a variety of anticomplement strategies, including sCR-1 for hemorrhagic shock (20) and C1 esterase inhibitor for traumatic shock (13). A single intravenous injection of CVF has been shown to reduce complement-dependent serum hemolytic activity (as measured by CH₅₀) by >95% (22). Our laboratory recently showed in a rat model of hemorrhagic shock producing a degree of hypotension and acidosis comparable to that in the present studies that CH₅₀ was reduced by 35% (25). Reductions of CH₅₀ by roughly one-third have also been seen in models of mesenteric ischemia (12) and in human acute respiratory distress syndrome (14) and burn patients (19). Although the degree of complement activation is severe, the present experiments indicate that complement activation, independent of other insults, is sufficient to cause profound alterations in systemic perfusion, acid-base status, and vascular integrity.

Conclusion. Systemic activation of complement results in circulatory collapse, metabolic acidosis, impaired oxygenation, and extravasation of intravascular protein. The lung was protected in large part by neutrophil depletion or COX-1 (but not COX-2) inhibition. These treatments were much less effective in providing protection against the systemic hemodynamic changes and intestinal injuries produced by CVF. These studies define the pathophysiological changes in the lung, heart, and gut after systemic activation of complement. They also suggest that mediator pathways leading to injury in the lung and gut operate by fundamentally different mechanisms.