Secondary Logo

Journal Logo

Basic Science Aspects

Complement Activation During Hemorrhagic Shock and Resuscitation in Swine

Szebeni, Janos*; Baranyi, Lajos; Savay, Sandor; Götze, Otto; Alving, Carl R.*; Bünger, Rolf§; Mongan, Paul D.

Author Information
doi: 10.1097/01.shk.0000082444.66379.17
  • Free



Previous studies have shown that traumatic blood loss leads to complement (C) activation, which in turn, plays a key role in the pathogenesis of subsequent shock, respiratory distress syndrome, and multiorgan failure (1–3). It has also been established that a major contributor to C activation under these conditions is tissue damage caused by the trauma and/or reperfusion, upon restoration of the circulation in ischemic organs (4,5). However, although there is substantial information on C activation after trauma and ischemia/reperfusion (I/R) injury, other possible C activation mechanisms in hemorrhagic shock have attracted relatively little attention to date. In particular, it is not clear whether C is activated in hemorrhagic shock without trauma or I/R damage. This is a crucial question in understanding the complex pathomechanism of C activation in civilian or battlefield hemorrhagic trauma, whereupon various degrees of tissue injury, blood loss, shock, and reperfusion injuries may be simultaneously present.

Here, we studied C activation in a porcine severe hemorrhagic shock/resuscitation model, which mimicked real-life situations in that the animals were subjected to resuscitation from shock first with different plasma expander fluids and then with whole blood. The controlled, rapid bleeding protocol simulated severe hemorrhage without tissue injury. The data obtained suggest that hemorrhagic shock per se can cause C activation, and that severe metabolic changes along with progressive endotoxemia could be major contributors to this process. We also report unexpected early biphasic course of plasma hemolytic C and C5a changes after hemorrhage, which may reflect until now unrecognized systemic responses to acute hemorrhage and/or C activation.



Veronal-buffered saline containing 0.15 mM Ca2+, 0.5 mM Mg2+, and 0.2% gelatin (VBS2+-gel), sheep red blood cell (SRBC) hemolysin (rabbit anti-sheep erythrocyte antiserum), and gelatin were from Difco Laboratories (Detroit, MI). SRBC was obtained from a local farm. All other chemicals and reagents, including antibodies whose sources are not specified, were from Sigma Chemical (St. Louis, MO). Lactated Ringer's solution (LR) containing 28 mM sodium (d,l)-lactate was obtained from McGaw Inc. (Irvine, CA), whereas the rest of resuscitation fluids were prepared using cell culture grade reagents, sterile water for injection, sterile disposable plastic containers and aseptic filtration methods.

Animal preparation

The Institutional Animal Care and Use Committee approved the procedures described, which were in compliance with the Animal Welfare Act and adhered to the principles enunciated in the Guide for the Care and Use of Laboratory Animals. Adolescent Yorkshire swine weighing 39.6 ± 0.4 kg were fasted overnight with free access to water. In the morning, they were sedated with an intramuscular injection of ketamine (10 mg/kg), and were anesthetized with halothane by nose cone to facilitate tracheal intubation. During surgical preparation, anesthesia was maintained with halothane (1.0%–1.5% end tidal concentration) while the animals spontaneously ventilated oxygen/nitrogen (21%/79%) through a semiclosed circle system (Narkomed 2B; North American Drager, Telford, PA). Expired oxygen, CO2, and halothane concentrations were continuously monitored (M1026A Gas Analyzer and 68 Clinical Monitor; Hewlett-Packard, Andover, MA). The right external jugular vein was percutaneously cannulated with an 8.5-FR introducer sheath, and a continuous thermodilution cardiac output pulmonary artery catheter was advanced through the sheath to measure pulmonary artery pressure and cardiac output (Qvue; Abbott Critical Care, North Chicago, IL). The left femoral artery was isolated at the groin and was cannulated with 8.5-FR introducer sheath. After isolation of the right common femoral artery, a micromanometer (MPC-500; Millar Instruments, Houston, TX) was advanced to the mid-thoracic aorta for the measurement of mean arterial pressure (MAP). Physiological data were displayed on an eight-channel clinical monitor (model 68; Hewlett-Packard). Other details of the model and calculation of hemodynamic parameters were described earlier (6).

Protocol of hemorrhagic shock and resuscitation

One hour before initiation of hemorrhage, the expired halothane concentration was reduced to 0.8% with the animals spontaneously ventilating. At time 0, controlled arterial hemorrhage was started using a roller pump (MasterFlex Digital Console Drive; Cole-Parmer Instruments, Chicago, IL) run by Labview 5 (National Instruments, Austin, TX), a program that determined the speed and direction of blood flow to maintain a constant, preset pressure through a proportional control feedback algorithm. At the start of hemorrhage, the program initiated blood withdrawal to decrease the MAP to 35 mmHg over 5 min. During hemorrhage, the blood was stored in a closed reservoir primed with sodium citrate (3.22 g) and porcine heparin (5000 units) to inhibit clot formation. After the initial 5-min rapid hemorrhage, the program maintained the MAP at 35 mmHg either by withdrawal or by reinfusion of the blood, as necessary. The volume of shed blood was determined by gravimetric analysis of the reservoir weight on a balance. At 90 min, around the time of decompensation, hypotensive volume-limited resuscitation was initiated with one of six resuscitation fluids listed in Table 1. Finally, at 180 min, resuscitation was continued by reinfusion of the total shed blood. The animals were observed until 270 min and were then sacrificed. A scheme of the above protocol is presented in Figure 1.

Table 1
Table 1:
Complement consumption in pigs during hemorrhage and resuscitation with different fluids
Fig. 1
Fig. 1:
Time line of pressure controlled hemorrhagic shock with hypotensive resuscitation in pigs. The specifics of resuscitation fluids are in Table 1. The blood pressure during fluid and blood resuscitations were raised to ∼50 to 60 mmHg and 80 to 90 mmHg, respectively. Further experimental details are described in the text and in Reference 6.

Arterial blood was sampled in standard heparin-tubes every 30 min for measurement of pH, base excess, blood gases, ions (Na+, K+, Cl, and HCO3), osmolality, lactate, and pyruvate (6). Samples were immediately centrifuged and stored at 80°C until further C analysis, as described below.

Measurement of complement activation in pig blood in vitro

Blood was taken from a pig before the experimental procedure described above, and was anticoagulated with 20 IU/mL heparin. Aliquots were supplemented with 20 mM free l(+) lactic acid or sodium lactate with or without 0.25 mg/mL zymosan, and the pH was adjusted to values indicated in the text, using dilute NaOH. Volume adjustment was done with physiological saline (0.9% NaCl). Blood was then incubated at 37°C in closed Eppendorf tubes with constant shaking, and serial samples were taken at the times indicated. Cells were pelleted within 3 min and the plasma was frozen and stored at 20°C until running the C5a or hemolytic C assays, as described below. In the case of C5a determinations, 20 mM EDTA was added to the samples before centrifugation.

Measurement of complement consumption in vivo and in vitro

Complement consumption was assessed by Mayer's CH50 assay adapted to 96-well plates. In brief, plasma samples (obtained at the indicated times during shock, or taken during the in vitro experiments as described below) were diluted 5-fold in VBS2+-gel, and aliquots in the 2 to 20 μL range were added to quadruplicate wells of 96-well Costar plates. The volumes were adjusted with VBS2+-gel to match the highest added volume (e.g., 20 μL). Positive (maximum hemolysis) and negative (spontaneous hemolysis) controls were wells to which 20 μL of 5% Triton X-100 and 20 μL VBS2+-gel were added, respectively. SRBC was suspended at 109 cell/mL in VBS2+-gel and was sensitized by incubation with 1/1000 volume hemolysin at room temperature for 30 min with occasional shaking. Sensitized SRBC (200 μL) were then added to the microplate wells and the plate was incubated at 37°C for 1 h with shaking (80 rpm). Finally, the plates were centrifuged at 4000 rpm for 10 min in a plate centrifuge (Beckman Instruments, Fullerton, CA), the supernatant was transferred to an empty plate, and A540 was read in a plate reader. Hemolysis was expressed as A540, the percentage of baseline, using the mean of quadruplicate wells.

Porcine C5a enzyme-linked immunoabsorbant assay (ELISA)

Plasma C5a levels were determined by an ELISA described earlier (7,8). In brief, wells of 96-well Costar plates were coated with rabbit polyclonal IgG directed against mouse IgG (1:500 diluted in Na2CO3/NaHCO3 coating buffer, pH > 10), 100 μL/well, overnight at 4°C. After blocking with 1% gelatin in coating buffer and washing in phosphate-buffered saline (PBS)/Tween, 100 μL from a 10 μg/mL anti hog-C5a mab, T13/9, was incubated in the wells for 2 h at room temperature. This was followed by three washings and addition of 100-μL aliquots from 2-fold-diluted plasma samples (in PBS/Tween/20 mM EDTA, pH 7.2–7.4). After a 1-h incubation at room temperature, wells were washed three times with PBS/Tween and 100 μL from the 2 μg/mL detection antibody, biotinylated rabbit anti-hog-C5a IgG, was added and incubated for 1 h at room temperature. In the final steps, wells were washed and incubated with streptavidin-peroxidase (1:4000 in PBS/Tween/EDTA, 100 μL/well) for 1 h at room temperature, followed by staining with ABTS/H2O2 substrate and reading of A410 with a plate reader. The C5a standard was affinity purified porcine C5a (7) and the log(dose) versus A410 curve was linear in the 0.625 to 10 ng/mL range (R2 = 0.999).

Measurement of plasma thromboxane B2 and LPS levels

Plasma TXB2 was measured using an ELISA kit from Amersham Life Sciences (Buckinghamshire, UK). LPS was measured by a Limulus amebocyte lysate-based chromogenic (Pyrochrome) assay obtained from Associates of Cape Cod (Falmouth, MA).

Statistical methods

Comparisons of the hemodynamic, C, and other laboratory changes at different time points were done by analysis of variance (ANOVA), followed by Student-Neumann-Keuls' or Bonferroni's post tests, as indicated. The time course of C5a production in whole blood in the absence and presence of lactic acid at different adjusted pH were analyzed by repeated-measures two-way ANOVA (matched by time), using pairwise comparisons among replicate means. The difference in C hemolytic activity among the different resuscitation groups was analyzed by the Kruskal-Wallis nonparametric test. All statistical computations were performed using GraphPad Prism, version 4.00 for Windows (GraphPad Software, San Diego, CA).


Complement consumption during shock and resuscitation

The induction (in 5 min) and conditions of isobaric shock (systemic arterial pressure maintained at 35 mmHg for 90 min) were identical in each group (Fig. 1), therefore, the 0-, 30-, 60-, and 90-min hemolytic C values were pooled from all animals. These data showed a trend for initial increase in hemolytic C at 30 min, followed by a steep decline, reaching statistically significant, 40% C consumption at 90 min (Fig. 2). Considering that C consumption is a relatively insensitive measure of C activation, the latter figure suggests massive C activation in hemorrhagic shock before resuscitation. However, the exact start of activation could not be established as a consequence of the initial rise of hemolytic C activity. Although this rise was statistically not significant, it obscured the initial kinetics of C activation.

Fig. 2
Fig. 2:
Complement consumption during hemorrhagic shock and resuscitation in pigs. Data represent A540, percentage of baseline, and mean ± SEM for n number of pigs (shown above the circles) from whom samples were available at the indicated times. In each pig, C consumption was calculated from quadruplicate OD readings, as described in “Materials and Methods.” ANOVA indicated significant differences among the groups (P < 0.0001), and the SNK post hoc test showed significant differences (*) between 90 vs. 0, 30; 180 vs. 0, 30, 60; and 270 vs. 0, 30 at P < 0.001, and between 60 vs. 30 and 180 vs. 90, 270 min at P < 0.05. The exact P of difference from baseline at 30 and 60 min was also calculated by Student's paired t test, two-tailed, giving 0.16 and 0.12, respectively (n = 6).

Plasma samples taken at 180 and 270 min, corresponding to the end of resuscitation with different fluids and shed blood, respectively, were grouped for a preliminary assessment of the impact of these fluids on C activation. The data obtained in two to three animals in each group (Table 1) were subjected to nonparametric one-way ANOVA by ranks (Kruskal-Wallis test), and were used for sample size calculations for a more conclusive study that would give 80% chance (power) to detect a 20% difference in mean CH50/mL as significant. The Kruskal-Wallis test indicated no significant differences among the groups (P = 0.42), whereas the sample size calculations predicted a minimum of 15 pigs to be included in each group in order see significant differences even with the simplest assumptions (see legend to Table 1). Based on these findings we did not analyze further the effect of resuscitation fluids on C activation, and we pooled the postresuscitation (180 and 270 min) CH50/mL values from all six groups.

As shown in Figure 2, the pooled hemolytic C activity at 180 min indicated 20% further C consumption during resuscitation, whereas at 270 min, the hemolytic C activity started to rise, indicating recovery of hemolytic C during resuscitation with shed blood. These differences were significant at P < 0.05 (see legend to Fig. 2).

Taken together, the above data demonstrate considerable C activation during shock possibly followed by slight extra activation during resuscitation with fluids. However, the exact start of C activation remained unresolved, and the minor C consumption during fluid resuscitation could also arise from dilution of blood by the resuscitation fluids. These ambiguities prompted us to perform further, alternative C measurements focusing on the initial (30- and 60-min) and late (180-min) changes.

C5a and TXB2 changes during shock and resuscitation

The alternative methods that we used for quantifying C activation were a sandwich ELISA of porcine C5a (7) and the measurement of TXB2, the stable metabolite of TXA2. The latter analyte was found earlier to be a sensitive indirect marker of C activation in pigs (9) and rats (10).

As shown in Figure 3A, plasma C5a slightly declined at 30 and 60 min relative to baseline, then it increased over baseline at 180 min, the difference becoming significant between 60 and 180 min. This net production of C5a between 60 and 180 min is consistent with the significant decline of hemolytic C activity during the same period (Fig. 2), and provides direct evidence for C activation. Although the reason for the initial decline of C5a is not clear, it should be emphasized that it is not necessarily inconsistent with, and actually may reflect, significant early C activation, as pointed out in the Discussion.

Fig. 3
Fig. 3:
Time course of plasma C5a (A) and thromboxane B2 (B) changes during hemorrhagic shock and resuscitation in pigs. C5a and TXB2 were measured by ELISA, as described in the text. Bars are mean ± SEM for n = 22 (C5a) and n = 10 (TXB2) pigs, from whom samples were available for these assays at the indicated times. Data were analyzed by ANOVA followed by the SNK post hoc test. (A) ANOVA:P < 0.0001, SNK:P < 0.01 for 0 vs. 60 min, 30 vs. 180 min and 60 vs. 180 min. (B) ANOVA:P < 0.0001, SNK:P < 0.001 for 0 vs. 30, 60 and 180 min.

The most convincing data suggesting that C could have been activated earlier than 60 min came from the TXB2 measurements, which indicated a significant rise of TXB2 over baseline already at 30 min, with continued elevation at 60 and 180 min (Fig. 3B).

Hemodynamic and metabolic changes during compensated shock

In an attempt to understand the mechanism of C activation in shock, we analyzed the traffic of blood between the reservoir and the circulation, as well as the cardiovascular and metabolic changes during the compensated (preresuscitation) phase of shock. As shown in Figure 4, from 5 to 90 min, the mean arterial pressure was maintained at 35 mmHg (Fig. 4A), requiring increasing amounts of blood withdrawal until the shed blood volume reached its peak at 60 ± 2 min (at 42.0 ± 0.6 mL/kg, Fig. 4B). Figure 4B also shows that during this first hour, no blood was reinfused into the circulation, and only a small amount (2 mL/kg, average 58 ± 5 mL, i.e., ∼2% of the total blood volume) was reinfused between 60 and 90 min. Figure 4, C through F, documents the cardiac hemodynamic responses during shock: a 60% increase in heart rate at 30 min with further 10% increase later at 60 and 90 min (Fig. 4C); maximal, >80% drops in cardiac index and stroke volume at already 30 min (Fig. 4, D and E), and gradual, 60% to 260% increases in systemic vascular resistance over 30 to 90 min (Fig. 4F). The latter changes reflect severe derangement of heart pump function with compensatory tachycardia and systemic vasoconstriction.

Fig. 4
Fig. 4:
Hemodynamic and cardiac changes in pigs during hemorrhagic shock, before resuscitation. Data are mean ± SEM for 12 pigs. Further details of the analysis are described in Materials and Methods. An asterisk indicates significant difference (P < 0.05) relative to baseline, as calculated by ANOVA followed by the SNK test.

Figure 5 demonstrates that the hemodynamic changes during the initial 90 min of shock were associated with progressive drop of pH from 7.43 to 7.28 (Fig. 5A), declines of base excess (from +6 to –10 mEq/L, Fig. 5B) and serum bicarbonate (from 31 to 16 mEq/L, Fig. 5C), and a considerable increase of plasma lactate (from 1 to 12 mM, Fig. 5D). These values imply lactic acidosis partially compensated by base consumption and bicarbonate excretion (hyperventilation with consequent decrease of PaCO2; data not shown) and they reflect severe oxygen debt (11).

Fig. 5
Fig. 5:
Metabolic changes in pigs during hemorrhagic shock, before resuscitation. Values are mean ± SEM for 12 pigs. All other details are similar to those in Figure 4.

Plasma endotoxin changes during shock

Figure 6 shows that plasma LPS levels displayed a slight but significant increase over baseline at 90 and 180 min. Taken together with the circulatory and metabolic evidence of organ hypoperfusion described above, this endotoxemia could arise from ischemic intestinal damage during shock.

Fig. 6
Fig. 6:
Plasma LPS changes during hemorrhagic shock and resuscitation in pigs. Values are mean ± SEM for 24 pigs from whom samples were available at the indicated times. An asterisk indicates significant difference (P < 0.05) relative to baseline, as calculated by ANOVA followed by the SNK post hoc test.

The effect of lactic acidosis on C5a formation in pig blood in vitro

Considering that acidosis has been known to promote C activation, our further efforts to identify the mechanism of C activation in shock used in vitro studies modeling the impact of lactic acidosis on C activation. Pig whole blood was incubated at 37°C with shaking, and C5a formation was measured at pH and/or lactate levels corresponding to the values observed at 90 min of shock. Figure 7A shows that blood C5a level did not change at neutral pH, whereas at pH 7.28 C5a formation was significantly (P < 0.05) increased at 120 min. Stronger acidification to pH 6.8 (i.e., the pH of blood after addition of 20 mM free lactic acid) induced a more expressed (∼100%) increase of C5a at 120 min. Acidification of blood to identical pH values by free l-lactic acid led to similar, or even faster generation of C5a (P < 0.01 at 90 min).

Fig. 7
Fig. 7:
Effects of lactic acid, pH, and zymosan on C5a formation and complement consumption in pig bloodin vitro. (A) The pH of freshly drawn pig blood (7.46, ○) was adjusted to 6.8 (□) or 7.28 (▵) by either HCl (open symbols) or by adding 20 mM l-lactic acid with or without NaOH (filled symbols). Filled diamonds represent samples in which blood was first acidified with 20 mM l-lactic acid and was then neutralized, whereas filled circles are those that were supplemented with 20 mM l-lactic acid that had previously been neutralized with NaOH (i.e., contained 20 mM Na lactate, pH 7.6.) These notations are also shown in the key, with “L” meaning l-lactic acid. Blood was than incubated in closed tubes at 37°C with shaking, and C5a was measured by ELISA at the indicated times. Points represent mean ± SD for triplicate measurements. Significant differences, whenever indicated in the text, were obtained by two-way repeated-measures ANOVA followed by pairwise comparisons according to Bonferroni. (B) Similar experiment, except that the effects of 0.25 mg/mL zymosan (Zym) ± 20 mM sodium lactate (Lac) was tested, using the hemolytic assay to measure C consumption. The graph shows the time course of C consumption, expressed as a percentage of baseline, mean ± SD for duplicate measurements. Other details are described in “Materials and Methods.”

These results are consistent with a significant influence of lactic acidosis on C activation in hemorrhaged pigs, however, the observed C5a changes did not duplicate the time course of C consumption in vivo, nor did they reflect the massive C consumption seen at 90 min of shock. Hence, we examined the possibility that rather than being the sole cause, lactic acidosis might have contributed to C activation in shock by potentiating the efficacy of (an)other C activation trigger(s).

The effect of lactate on zymosan-induced complement consumption in vitro

As shown in Figure 7B, incubation of pig blood at 37°C for 120 min caused no changes in blood hemolytic C levels regardless of the presence of 20 mM lactate. Incubation with 0.25 mg/mL zymosan led to significant C consumption at 90 min, whereas, together with 20 mM lactate, the same amount of zymosan caused significant C consumption at already 10 min, with maximal effect at 60 min. Thus, lactate accelerated the C-activating effect of zymosan to such degree that the kinetics of C consumption were very similar to that observed in shock.


While providing first line defense against microbial infections, activation of the C system is also known for its pathogenic role in some 30 human illnesses or adverse conditions (12). Trauma with or without shock, respiratory distress syndrome, multiorgan failure, and I/R injury are known examples for these conditions where the cause and pathogenic role of abnormal C activation have been well established. However, surprisingly little is known about the presence and possible pathogenic role of C activation in hemorrhagic shock without resuscitation, under conditions when trauma is not playing a major role. Most recently, Younger et al. (13) have shown significant C consumption in a rat model of hemorrhagic shock that was not related to tissue injury or resuscitation. They also demonstrated that amplification of anaphylatoxin production with a carboxypeptidase N inhibitor worsened the shock, whereas decomplementation with cobra venom factor improved the hemodynamic and acid-base recovery after resuscitation (13). However, these or other authors have not clarified the mechanism by which C is activated during shock in the absence of trauma and/or resuscitation.

The main goals of the present study were to explore C activation and its correlation with other physiological changes in a porcine hemorrhagic shock/resuscitation model, and to conduct relevant in vitro experiments addressing the mechanism of activation. The model was used recently for addressing other questions (6) and resembled the rat model of Younger et al. (13) in that hemorrhagic hypotension was maintained at 30 to 35 mmHg. However, in our pigs, this value was reached faster (in 5 min vs. 30–40 min) and it was maintained for longer (for 90 min instead of 50 min). Furthermore, unlike Younger et al. (13), we resuscitated the pigs first with different plasma expanders and then with shed blood, and we performed serial analysis of hemolytic C activity during shock. Our studies reproduced the C consumption described by Younger et al. (13) and provided additional, new information on the phenomenon, as discussed below.

Complement consumption

We measured a 39% decline of blood hemolytic C activity at the end of the hemorrhage, which is in reasonable agreement with the 33% average posthemorrhage C consumption in rats reported by Younger et al. (13). As for the underlying mechanism of this C consumption, the authors did not reinfuse fluid into the rats during this period, ruling out dilution (of hemolytic C) or C activation by exogenous triggers. Although in our experiments, 5% of shed blood was reinfused between 60 and 90 min, this could not account for the 39% drop in hemolytic C even if all C had been consumed in the reservoir. Considering that the blood was anticoagulated in the reservoir with citrate and heparin, both acting as inhibitors of C activation within the dose range achieved in shed blood (i.e., ≥10 mM citrate and ≥3.0 IU/mL heparin) (14), C consumption could be excluded in the reservoir in any way. Infusion of heparin or C activating substances released from the reservoir wall were also unlikely contributors to C consumption within the animal because the decline of CH50/mL preceded the exposure of blood to these agents, and, after fluid resuscitation, we reinfused 100% of shed blood (∼1.6 L, or 57% of total blood volume) into the pigs, yet the CH50/mL levels rose instead of dropping. With regard to heparin, it should be noted that depending on dose, heparin can promote or inhibit C activation. For example, Mollnes et al. (14) have found that 0.2 to 2 IU/mL heparin slightly increased, whereas 20 IU/mL significantly decreased spontaneous C activation in human serum. Thus, the infusion of ≤170 IU heparin into the pigs between 60 and 90 min, raising the blood level to ≤0.13 IU/mL, would be expected to have no significant influence on C activation. On the other hand, the heparin dose that reached the blood during the final reinfusion of shed blood (≤1.8 IU/mL) would be expected to promote C consumption, rather than increase the CH50/mL. Based on this information, the observed C consumption in our study was unlikely to be due to the extracorporeal circuit or anticoagulants used.

As for the tendency for initial rise of hemolytic C after hemorrhage, this paradoxical observation can most easily be rationalized by acute release of C or C-binding proteins (e.g., C-reactive protein or other acute-phase proteins of the pentraxin family) into the blood (15), or by transient depletion of some C control proteins (e.g., C1INH, C4bBP, and factors H and I). Although further studies are needed to understand the phenomenon, it should be mentioned that similar initial rise in hemolytic C was observed in rats subjected to controlled arterial hemorrhage and in other pig studies whereupon pigs were subjected to uncontrolled arterial hemorrhage (C.R. Alving, L. Baranyi, M. Bodo, M. Dubick, J. Savay, J. Sondeen, J. Szebeni, unpublished observations).

Plasma C5a changes

In an attempt to provide direct evidence for C activation during hemorrhagic shock, particularly at times when the hemolytic C data were inconclusive, we extended the C measurements with ELISA of plasma C5a. Although the significant rise of this anaphylatoxin at 180 min was consistent with C activation, the initial (30 and 60 min) decline relative to baseline was unexpected and, at first glance, contradicting to C activation. Nevertheless, C activation is not necessarily incompatible with the absence of substantial elevation, or even decline, of plasma C5a. It is known that C5a rapidly (within seconds) and irreversibly binds to specific high-affinity receptors (C5aR) on leukocytes and macrophages, followed by rapid interiorization of the receptor-ligand complex (8,16). Thus, C5aR+ cells might act as C5a scavengers and delay the net buildup of this anaphylatoxin in the blood after C activation, or even decrease its plasma level provided their C5aR expression is upregulated by non-C5a triggers (e.g., by C3a, C5b-9, and other secondary mediators). As proposed by Oppermann et al. (8) earlier, this C activation-related C5a decrease paradox may represent a control mechanism protecting the body from systemic effects of a potent phlogistic mediator.

Plasma TXB2 changes

TXB2 is a stable metabolite of the short-lived TXA2, a major secondary mediator of leukocyte, mast cell, and macrophage activation. TXB2 was shown to be the immediate, rate-limiting mediator of liposome-induced and C-mediated pulmonary vasoconstriction in pigs (9) and a sensitive measure of liposome-induced C activation in rats (10). Its rise is consistent with the upregulation of hepatic cyclooxygenase-2 mRNA in a rat model of hemorrhagic shock (17). TXB2 was also shown to rise parallel with C5a in a dog model of intestinal I/R damage (18), attesting to its use as a surrogate marker of slow, gradual C activation as well. Its significant rise at already 30 and 60 min provides the most straightforward (albeit indirect) support for early C activation in our model.

The influence of resuscitation fluids on complement parameters during resuscitation

Apart from the C analysis reported here, the present study was conceived to compare the resuscitating efficacies of hypertonic, monocarboxylate-augmented plasma expanders with those of control LR and hypertonic saline. This analysis will be described elsewhere. Although the high within-group variation of CH50/mL values and low number of animals in the different resuscitation groups did not allow definitive conclusions regarding the individual effects of resuscitation fluids on C activation, the lack of significant differences among the groups at 180 and 270 min by the nonparametric Kruskal-Wallis test is consistent with a study by Svennevig et al. (19) showing that C activation during open-heart surgery was only marginally affected by the choice of fluid for volume replacement.

Possible mechanisms of complement activation

Based on our previous pig studies (9,20), anesthesia and vascular instrumentation could be excluded as the underlying cause of C activation in the present study. Our model did not involve trauma or tissue injury, and a major effect of the extracorporeal circuit and anticoagulants could also be ruled out. Furthermore, we observed massive C activation before infusing into the animals significant amounts of blood or fluid, which argues against a role of reperfusion injury. Although we cannot entirely rule out a potentiating influence of very low doses of heparin on C activation, the explanation most probably lies in one ore more physiological or chemical abnormalities associated with hypoperfusion-related tissue ischemia.

The triggering events of such I/R-related C activation include the deposition of natural antibodies (21,22), C-reactive protein (23), or mannose-binding lectin (24) to the surface of injured cell membranes. However, in our experiments, neither the perfusion nor the oxygenation was restored during the time of massive C activation. Considering that the production of oxygen radicals is critical to reperfusion injury (25,26), it is unclear whether hypoperfusion alone can lead to tissue damage typical, at least in part, of I/R. Hence, the involvement of oxidative injury, or prolonged ischemia in the C activation described here, remains to be established. In the present study, we focused on two other possible mechanisms that are known to cause systemic C activation independent of tissue injury: metabolic acidosis and bacteremia/endotoxemia.

It is common knowledge that acidification of plasma (to pH 6.5) results in increased C-mediated hemolysis of red blood cells from paroxysmal nocturnal hemoglobinuria (PNH) patients; a phenomenon underlying the laboratory diagnosis of this disease (Ham's test). Key elements of this abnormal alternative pathway C activation include a deficiency of PNH red cells in C-inhibitory membrane proteins (CD59 and DAF), and increased association of factor B and C5 to membrane-bound C3b at pH 6.5 (27,28). The relevance of the above facts to hemorrhagic shock was highlighted by Emeis, Sonntag, and associates (29–31) by demonstrating increased C activation in human blood after acidification with 5.5 to 22 mM lactate. It was also shown that supplementation of heparinized adult blood with 5.5 mM lactate reduced the pH from 7.34 to 7.17 and increased the levels of C3a, C5a, and SC5b-9 after 1 h incubation by 5%, 16%, and 62%, respectively (31). The authors concluded that lactic acidosis occurring in birth asphyxia, septic shock, and many other conditions associated with tissue hypoxia leads to the activation of the entire C system. Considering that the lactacidemia seen in our pigs reached the above-defined threshold for C activation, it seems likely that lactic acidosis could play an important role in C activation in our study as well. To bear out this hypothesis, we carried out in vitro studies as described below.

As for the possible role of endotoxin in shock-related C activation, bacterial translocation through the intestinal wall with subsequent endotoxemia has been previously demonstrated in numerous models of intestinal I/R and hemorrhagic shock (4,5,18,32,33). LPS is known to activate C by various routes, including the classical and alternative pathways, and also indirectly, via stimulation of C-reactive protein release into the blood (15). In our experiments, the quantitatively small, yet significant rise of plasma LPS at 90 and 120 min is consistent with the CH50/mL curve reaching its nadir during this period. As for the low degree of changes, there are several processes contributing to the rapid clearance of the LPS in plasma, such as the binding to plasma proteins, high-density lipoproteins, and CD14 receptors on different (monocytic) cells, and enzymatic degradation by esterases (33,34). Thus, our analysis might have underestimated the endotoxin exposure of the animals under shock. It should also be considered that hemorrhagic shock could sensitize the pigs to LPS in a process similar to that described in rabbits (35). A recent study by Zhi-Yong et al. (18) is particularly relevant in this regard, as the authors demonstrated close parallelism between bacteremia, hypotension, and the elevation of plasma C5a and TXB2 after restoration of bowel circulation in dogs after partial occlusion of the superior mesenteric artery.

In vitro studies on the mechanisms of complement activation in hemorrhagic shock

To test the above theoretical considerations relating to the role of lactic acidosis and bacteremia/endotoxemia in C activation during shock, we measured C5a formation in vitro under conditions that mimicked hemorrhagic shock in terms of lactic acidosis and endotoxemia/endotoxemia. We used whole blood (instead of serum or plasma) to preserve the buffering capacity of red blood cells (which is known to account for some 50% of the total buffer capacity of blood), and zymosan to simulate bacterial C activation. The data showing that acidification of blood to pH 7.28 with l-lactic acid led to a 2-fold increase in C5a formation after 90 min incubation was in keeping with the mentioned human data of Hecke et al. (31); however, this change was not associated with C consumption. To reproduce the decline of CH50/mL in vivo, we had to combine zymosan with lactic acidosis, suggesting that the simultaneous presence of these conditions might have been critical in the massive C consumption in pigs during shock.

It should be mentioned that the dose of heparin used to effectively prevent coagulation during in vitro shaking of pig blood at 37°C (20 IU/mL) was shown to suppress spontaneous C activation in human serum (14). Therefore, it is possible that the effects of lactic acidosis and/or zymosan on C5a formation were underestimated in our in vitro studies.

In conclusion, C activation during hemorrhagic shock may be multifactorial and complex even without trauma or reperfusion injury. Our data suggest that metabolic acidosis and endotoxemia play important roles in this activation, along with the possible contribution of other adverse processes at the level of hypoxic cell membranes (Fig. 8A). Considering that C activation is a key step in a vicious cycle of pathological changes ultimately leading to circulatory collapse (Fig. 8B), studies on its mechanism and inhibition may help in alleviating the outcome of hemorrhagic shock.

Fig. 8
Fig. 8:
Hypothetical mechanism (A) and role (B) of C activation in hemorrhagic shock. (A) The scheme illustrates that activation of C3, i.e., the central step in the C cascade, might occur in hemorrhagic shock as a consequence of rises in LPS and lactic acid (LAC) levels in blood. In addition, the binding of natural antibodies (nAb), C3b, mannose binding lectin (MBL), and C-reactive protein (CRP) to hypoxic cell membranes might provide additional C activating triggers. Bold characters and arrows show the changes described in the present study, and thin characters and arrows show potential mechanisms reported in the I/R literature. (B) The scheme illustrates the interrelationship among the different pathological processes in hemorrhagic shock acting in a vicious cycle in inducing circulatory collapse. It is seen that C activation is a critical step in the cycle via numerous adverse effects.


The skillful technical assistance of Ms. Eva Fleischmann is gratefully acknowledged.


1. Yao YM, Redl H, Bahrami S, Schlag G: The inflammatory basis of trauma/shock-associated multiple organ failure. Inflamm Res 47:201–210, 1998.
2. Baue AE, Durham R, Faist E: Systemic inflammatory response syndrome (SIRS), multiple organ dysfunction syndrome (MODS), multiple organ failure (MOF): are we winning the battle? Shock 10:79–89, 1998.
3. Goris RJ: Pathophysiology of shock in trauma. Eur J Surg 166:100–111, 2000.
4. Spain DA, Fruchterman TM, Matheson PJ, Wilson MA, Martin AW, Garrison RN: Complement activation mediates intestinal injury after resuscitation from hemorrhagic shock. J Trauma 46:224–233, 1999.
5. Rehrig S, Fleming SD, Anderson J, Guthridge JM, Rakstang J, McQueen CE, Holers VM, Tsokos GC, Shea-Donohue T: Complement inhibitor, complement receptor 1-related gene/protein y-Ig attenuates intestinal damage after the onset of mesenteric ischemia/reperfusion injury in mice. J Immunol 167:5921–5927, 2001.
6. Mongan PD, Capacchione J, West S, Karaian J, Dubois D, Keneally R, Sharma P: Pyruvate improves redox status and decreases indicators of hepatic apoptosis during hemorrhagic shock in swine. Am J Physiol 283:H1634–H1644, 2002.
7. Hopken U, Mohr M, Struber A, Montz H, Burchardi H, Gotze O, Oppermann M: Inhibition of interleukin-6 synthesis in an animal model of septic shock by anti-C5a monoclonal antibodies. Eur J Immunol 26:1103–1109, 1996.
8. Oppermann M, Gotze O: Plasma clearance of the human C5a anaphylatoxin by binding to leukocyte C5a receptors. Immunology 82:516–521, 1994.
9. Szebeni J, Fontana JL, Wassef NM, Mongan PD, Morse DS, Dobbins DE, Stahl GL, Bünger R, Alving CR: Hemodynamic changes induced by liposomes and liposome-encapsulated hemoglobin in pigs: a model for pseudo-allergic cardiopulmonary reactions to liposomes. Role of complement and inhibition by soluble CR1 and anti-C5a antibody. Role of complement and inhibition by soluble CR1 and anti-C5a antibody. Circulation 99:2302–2309, 1999.
10. Szebeni J, Wassef NM, Spielberg H, Rudolph AS, Alving CR: Complement activation in rats by liposomes and liposome-encapsulated hemoglobin: evidence for anti-lipid antibodies and alternative pathway activation. Biochem Biophys Res Commun 205:255–263, 1994.
11. Rixen D, Raum M, Holzgraefe B, Sauerland S, Nagelschmidt M, Neugebauer EA: A pig hemorrhagic shock model: oxygen debt and metabolic acidemia as indicators of severity. Shock 16:239–244, 2001.
12. Sahu A, Lambris JD: Complement inhibitors: a resurgent concept in anti-inflammatory therapeutics. Immunopharmacology 49:133–148, 2000.
13. Younger JG, Sasaki N, Waite MD, Murray HN, Saleh EF, Ravage ZA, Hirschl RB, Ward PA, Till GO: Detrimental effects of complement activation in hemorrhagic shock. J Appl Physiol 90:441–446, 2001.
14. Mollnes TE, Brekke OL, Fung M, Fure H, Christiansen D: Essential role of the C5a receptor in E. coli-induced oxidative burst and phagocytosis revealed by a novel lepirudin-based human whole blood model of inflammation. Blood 100:1869–1877, 2002.
15. Szalai AJ, van Ginkel FW, Wang Y, McGhee JR, Volanakis JE: Complement-dependent acute-phase expression of C-reactive protein and serum amyloid P-component. J Immunol 165:1030–1035, 2000.
16. Hetland G, Moen O, Bergh K, Hogasen K, Hack CE, Mollnes TE, Fosse E: Both plasma- and leukocyte-associated C5a are essential for assessment of C5a generation in vivo. Ann Thorac Surg 63:1076–1080, 1997.
17. Tsukada K, Hasegawa T, Tsutsumi S, Kuwano H. Roles of cyclooxygenase-2 in tissue injury during hemorrhagic shock. Shock 13:392–396, 2000.
18. Zhi-Yong S, Dong YL, Wang XH: Bacterial translocation and multiple system organ failure in bowel ischemia and reperfusion. J Trauma 32:148–153, 1992.
19. Svennevig JL, Tollofsrud S, Kongsgaard U, Noddeland H, Mohr B, Ozer M, Mollnes TE: Complement activation during and after open-heart surgery is only marginally affected by the choice of fluid for volume replacement. Perfusion 11:326–332, 1996.
20. Szebeni J, Baranyi B, Savay S, Bodo M, Morse DS, Basta M, Stahl GL, Bunger R, Alving CR: Liposome-induced pulmonary hypertension: properties and mechanism of a complement-mediated pseudoallergic reaction. Am J Physiol 279:H1319–H1328, 2000.
21. Weiser MR, Williams JP, Moore FDJ, Kobzik L, Ma M, Hechtman HB, Carroll MC: Reperfusion injury of ischemic skeletal muscle is mediated by natural antibody and complement. J Exp Med 183:2343–2348, 1996.
22. Fleming SD, Shea-Donohue T, Guthridge JM, Kulik L, Waldschmidt TJ, Gipson MG, Tsokos GC, Holers VM: Mice deficient in complement receptors 1 and 2 lack a tissue injury-inducing subset of the natural antibody repertoire. J Immunol 169:2126–2133, 2002.
23. Barrett TD, Hennan JK, Marks RM, Lucchesi BR: C-reactive-protein-associated increase in myocardial infarct size after ischemia/reperfusion. J Pharmacol Exp Ther 303:1007–1013, 2002.
24. Jordan JE, Montalto MC, Stahl GL: Inhibition of mannose-binding lectin reduces postischemic myocardial reperfusion injury. Circulation 104:1413–1418, 2001.
25. Vega VL, Maldonado M, Mardones L, Manriquez V, Vivaldi E, Roa J, Ward PH: Inhibition of nitric oxide synthesis aggravates hepatic oxidative stress and enhances superoxide dismutase inactivation in rats subjected to tourniquet shock. Shock 9:320–328, 1998.
26. Mota-Filipe H, McDonald MC, Cuzzocrea S, Thiemermann C: A membrane-permeable radical scavenger reduces the organ injury in hemorrhagic shock. Shock 12:255–261, 1999.
27. Gotze O, Muller-Eberhard HJ: Lysis of erythrocytes in the absence of antibody. J Exp Med 132:898–915, 1970.
28. Fishelson Z, Horstmann RD, Muller-Eberhard HJ: Regulation of the alternative pathway of complement by pH. J Immunol 138:3392–3395, 1987.
29. Emeis M, Sonntag J, Willam C, Strauss E, Walka MM, Obladen M: Acidosis activates complement system in vitro. Mediators Inflamm 7:417–420, 1998.
30. Sonntag J, Emeis M, Strauss E, Obladen M: In vitro activation of complement and contact system by lactic acidosis. Mediators Inflamm 7:49–51, 1998.
31. Hecke F, Hoehn T, Srauss E, Obladen M, Sonntag J: In vitro activation of complement system by lactic acidosis in newborn adults. Mediators Inflamm 10:27–31, 2001.
32. Redl H, Bahrami S, Furst W, Schiesser A, Schlag G, Mollnes TE: Evidence for early bacteremia/endotoxemia and inflammatory responses after trauma-hemorrhage and complement activation in a nonhuman primate model. Eur J Trauma 6:300–307, 2000.
33. Guo W, Magnotti LJ, Ding J, Huan Q, Xu D, Deitc EA: Influence of gut microflora on mesenteric lymph cytokine production in rats with hemorrhagic shock. J Trauma 52:1178–1185, 2002.
34. Erwin PJ, Lewis H, Dolan S, Tobias PS, Schumann RR, Lamping N, Wisdom GB, Rowlands BJ, Halliday MI. Lipopolysaccharide binding protein in acute pancreatitis. Crit Care Med 28:104–109, 2000.
35. Mileski WJ, Winn RK, Harlan JM, Rice CL: Sensitivity to endotoxin in rabbits is increased after hemorrhagic shock. J Appl Physiol 73:1146–1149, 1992.

Anaphylatoxins; endotoxin; thromboxane; trauma; Ringer's lactate; plasma expanders; ischemia/reperfusion injury; lactic acidosis

©2003The Shock Society