C-Reactive Protein and Complement Are Important Mediators of Tissue Damage in Acute Myocardial Infarction

Human CRP 26, human serum amyloid P component (SAP; reference 27), rat CRP 28, rat C3 28, and cobra (Naja naja) venom factor 29 were isolated and purified and/or assayed precisely as previously described. The preparations of human CRP and SAP and of cobra venom factor used in vivo were all >99% pure.

Female Wistar rats bred at Imperial College School of Medicine, Hammersmith campus were used at age 9–10 wk and weighed 230–280 g. Within each experiment, the whole body and heart weights were the same in control and experimental groups. All animals were carefully examined and were healthy before fasting overnight before surgery. General anesthesia was induced with inhaled isoflurane, and supplemental oxygen was provided at 1.0 liter/min. The chest was opened through the sixth intercostal space, anesthesia was discontinued but supplemental oxygen continued at 0.5 liter/min, and the pericardium was opened to allow the heart to be lifted out of the thorax. The left anterior descending coronary artery was immediately ligated at a constant distance just below the atrium with a 5/0 silk suture, the heart was replaced, and the chest was closed while active ventilation with 100% oxygen was continued. The duration of pneumothorax was ∼30 s. Resuscitation was completed by gentle chest massage with the rat lying supine, and there was usually rapid recovery of consciousness and the righting reflex. Buprenorphine (0.05 mg/kg) was given immediately for postoperative analgesia and repeated 12-hourly. Development of myocardial infarction was confirmed by early postoperative four lead electrocardiogram showing marked ST segment elevation in leads I and aVL as well as various arrhythmias.

Intraoperative and immediate postoperative mortality was up to 20%, but thereafter there were few deaths before day 5, when the rats were killed for measurement of infarct size. Deep anesthesia was induced with isoflurane, and the hearts were excised while still beating in sinus rhythm and immediately arrested in diastole by immersion in 10 ml of 30 mM KCl. They were then cleaned of any extraneous adherent tissue and weighed before being briefly chilled at −20°C to produce sufficient rigidity to facilitate cutting into defined sections. Each heart was cut into four slices of equal thickness perpendicular to the course of the left coronary artery, starting from the apex of the heart and ending at the position of the ligature around that artery. These slices were designated A, B, C, and D starting from the apex, and each was washed with pure water and then immersed in 0.02 M phosphate buffer, pH 7.4, containing 0.5 mg/ml of nitroblue tetrazolium (NBT; Sigma Chemical Co.) at 37°C for 30 min. NBT stains viable but not infarcted myocardium. After staining with NBT, the slices were washed briefly in cold water before fixation in 10% buffered formalin for 48 h, and the proximal cut surface of each slice was then imaged under standard conditions with a high resolution digital camera. The captured images were coded and analyzed ‘blind,’ without knowledge by the operators of the treatment received by the rat in question. Lines were drawn around the total area of the slice, the NBT-negative infarct zone, and any parts of the image to be excluded from the estimation, including valves, chordae tendinae, shadows, and edges, and the area of the infarct was determined as a percentage of the total area. The infarcts were confined to slices B, C, and D, and the mean percentage in these three was taken as the infarct size in each heart. Statistical significance of differences between infarct sizes in different treatment groups within each experiment was sought by one-way analysis of variance and by Bonferroni t tests.

Complement depletion was induced in vivo by a single intraperitoneal injection of cobra venom factor at 250 U/kg. Human CRP or SAP was injected intraperitoneally at 40 mg/kg 1 h after coronary artery ligation and then at 24-h intervals until the rats were killed on day 5. The proteins were dissolved in 10 mM Tris and 140 mM NaCl, pH 8.0, and also 2 mM CaCl₂ in the case of human CRP, and the control rats received injections of the buffer alone at the same intervals. In one experiment, rats were killed 24, 48, and 72 h after coronary artery ligation. For analysis of the effect of cobra venom factor on plasma C3, the usual dose was given and normal, unoperated rats were bled from the tail at intervals up to 6 d. For measurement of plasma clearance of human CRP and SAP, normal unoperated rats received a single 40 mg/kg dose of each protein and were bled at intervals up to 72 h later. Plasma clearance of human CRP was also studied in the same protocol in normal unoperated rats that had received cobra venom factor 24 h previously. These experiments could not be performed in rats that had undergone coronary artery ligation because they do not survive the anesthesia required for bleeding, but there is no reason to suppose that the results in those rats would be significantly different.

The formalin-fixed heart slices were embedded in wax and sectioned at 2-μm thickness for routine hematoxylin and eosin staining and for immunohistochemical staining. Optimal dilutions of the IgG fractions of monospecific goat anti–human CRP, rabbit anti–rat CRP, and sheep anti–rat C3 antisera, raised by immunization with the respective isolated pure antigens, were used as primary antibodies after blocking the sections by incubation in 20% vol/vol normal serum of the species of origin of the secondary antibody. In all cases, endogenous peroxidase activity was blocked by incubation with H₂O₂ before testing. Secondary antibodies were optimal dilutions of goat anti–rabbit IgG or rabbit anti–goat IgG (both from DAKO Corp.), which cross-reacts with sheep IgG. Finally, rabbit or goat peroxidase–antiperoxidase complexes (DAKO Corp.) were used, followed by visualization with diaminobenzidine. All positive staining was completely abolished when the primary antibodies had been absorbed before use with an excess of the respective specific antigen. In the case of rat C3, which was not available in pure form, parallel absorptions of the sheep anti–rat C3 antibody were performed with normal rat serum and with C3-depleted serum from a cobra venom–treated rat. The latter did not affect anti-C3 staining, whereas C3-sufficient rat serum abolished it completely. To confirm that the antigens under investigation were not affected by formalin fixation and tissue processing, sample hearts were snap frozen without fixation, and 5-μm cryostat sections were cut for immunoperoxidase staining. For all antigens, appearances were identical to those of the fixed sections, except for poorer preservation of histological detail as expected. Frozen sections of a heart from a human CRP–treated rat killed on day 3 were also stained for human CRP by indirect immunofluorescence and the nuclei counterstained with ethidium bromide to visualize binding of CRP to cell surfaces.

We have previously shown that rat CRP does not activate rat complement, whereas human CRP does so very efficiently 28. The rat model therefore permits specific analysis of the complement-dependent effect of human CRP on infarct size in vivo. Rats that received 40 mg/kg of isolated pure human CRP by intraperitoneal injection 1 h after coronary artery ligation rapidly became clinically less well than buffer-treated controls, and some died during the next 3 d while receiving further daily injections of the same dose of human CRP. In contrast, rats that had not been operated on but received the same doses of human CRP showed absolutely no ill effects. Injections of human SAP, the pentraxin protein very closely related to CRP 30, had no adverse clinical effects in either normal or coronary artery–ligated rats. When all surviving animals were killed on day 5 after coronary artery ligation, the infarcts in those receiving human CRP were ∼40% larger than in control rats treated with either buffer or human SAP (Table).

The plasma clearance of human CRP and SAP after single intraperitoneal injections in control, nonoperated rats and the clearance of human CRP in decomplemented rats is shown in Fig. 1. The clearance of human CRP was not affected by complement depletion. The peak values for human CRP were comparable to massive acute phase responses in humans 31, and those for human SAP were much higher than ever seen in humans 32. Rats that have undergone coronary artery ligation do not survive the anesthesia required for bleeding and were therefore not bled before they were killed on day 5. However, when these animals were killed 24 h after the last of five daily injections, the serum concentrations of human CRP were still in the range of 13–53 mg/liter, typical of a moderate clinical acute phase response in humans. In animals that had received human SAP, the serum concentrations of this protein were between 45 and 175 mg/liter. Values for rat CRP in serum at the time of exsanguination 5 d after coronary artery ligation were 215–525 mg/liter, which is within the normal range for this species 28.

Administration of cobra venom factor in vivo rapidly produces profound and sustained depletion of C3 29, with no active C3 remaining in the circulation at 6 h. Traces of C3 antigen detectable thereafter are inactive cleavage fragments 33. With the dose of cobra venom factor used here, 250 U/kg 29, active C3 starts to reappear in the circulation after ∼4 d and is within the normal range by day 5 or 6. When rats had been decomplemented by in vivo administration of cobra venom factor 24 h before coronary artery ligation, their infarcts at 5 d were ∼60% smaller than those in control untreated animals (Table). Complement-sufficient rats injected daily with human CRP developed, as before, infarcts ∼40% larger than those in control untreated animals. However, injection of human CRP had no effect at all on the reduced infarct size in decomplemented rats (Table). The damaging effect of human CRP in this model is thus absolutely complement dependent.

Human CRP and rat C3 (Fig. 2) and rat CRP (not shown) were all deposited in the infarcted myocardium. On day 5, human and rat CRP were present homogeneously throughout the infarcted muscle and also in a more intense, speckled pattern in multiple foci that occasionally coincided with hematoxyphil nuclear remnants. These foci may be nuclear ghosts from which chromatin has been cleared but that retain the small nuclear ribonucleoprotein particles to which human CRP binds avidly 34,35 and possibly other CRP ligands. Rat C3 was present predominantly in the same speckled foci, with no diffuse immunoreactivity on the infarcted muscle cells, in contrast to the distribution of CRP. Immunofluorescence staining of sections of unfixed, snap-frozen myocardial tissue taken 72 h after coronary ligation clearly demonstrated the presence of CRP on the surfaces of damaged myocardial cells in and around the infarct, as well as the same distribution seen in fixed sections stained by the immunoperoxidase method. Staining for human CRP was always more intense than for rat CRP, although it is not clear whether this represents greater abundance of the human protein in the tissue sections or just greater sensitivity of the respective immunostaining procedure. Numerous mononuclear cells in the dense periinfarct infiltrate also stained strongly for human CRP (Fig. 2) but not for rat CRP (not shown) or rat C3 (Fig. 2).

Marked reductions in infarct size and in ischemia/reperfusion injury have previously been demonstrated in animals in which complement activation had been blocked by treatment, either with cobra venom factor or with recombinant soluble complement receptor type 1, before or at the time of induction of ischemia 4,5,6,7,8. The therapeutic implications of this observation are obvious, but for clinical implementation, it is critical to know whether complement depletion initiated after the onset of coronary occlusion can still have a protective effect. We therefore administered cobra venom factor at various times between 24 h before and 24 h after coronary artery ligation. There was, as shown before, a marked, ∼60% reduction in infarct size in rats that had received treatment 24 h beforehand and that were maximally complement depleted at the time of operation (Table). Although injection of cobra venom factor 6 or 24 h after ligation had no effect, the infarcts in animals that received cobra venom factor either 0.5 or 2 h after ligation were almost 50% smaller (Table). Maximal complement depletion is achieved only from 6 h after intraperitoneal injection of cobra venom factor, and in this rat model there is therefore a window of therapeutic opportunity of perhaps up to 8 h after acute coronary artery occlusion during which immediate inhibition of complement activation could potentially reduce infarct size.

CRP has been very stably conserved in evolution 30,36,37, and no structural polymorphism or deficiency of CRP has yet been reported in humans, suggesting that this protein has important normal functions that contribute to survival. In experimental models, CRP is protective against pneumococcal infection 38,39 and may contribute to innate immunity to other microorganisms to which it binds 40. By analogy with this role of SAP in relation to chromatin that we have recently demonstrated 41, CRP also probably plays an important role in scavenging autologous ligands and preventing development of autoimmunity. This does not mean, however, that CRP may not also contribute to pathogenesis of disease, especially conditions developing in postreproductive later life. Natural selection is ‘blind’ to phenomena occurring after reproduction, provided they do not affect viability of the species as a whole 42.

The results presented here unequivocally demonstrate, in a robust experimental model, that human CRP markedly enhances the extent of myocardial damage produced by ischemic injury. Although the time–concentration profile of human CRP produced by daily CRP injections was not the same as the monophasic acute phase response that follows uncomplicated naturally occurring myocardial infarction in humans, it was comparable to the persistent, high, and fluctuating CRP pattern typically found in patients with postinfarct complications 11 and was therefore not ‘unphysiological.’ Endogenous rat CRP, as well as the injected human CRP, was deposited in the infarct in vivo, but rat CRP does not activate rat complement 28. In contrast, human CRP is a potent activator of rat complement 28, and the enhancement of infarct size caused by administration of human CRP is completely abrogated by in vivo complement depletion. Human CRP production is always greatly increased after acute myocardial infarction, CRP is always deposited in human myocardial infarcts, and early and late clinical outcomes are significantly associated with peak postinfarction plasma levels of CRP. It is therefore very likely that CRP contributes significantly to the extent of damage in human acute myocardial infarction, and, based on our results here, it probably does so via complement activation. Although complement activation by CRP is not efficient in generating the terminal lytic complement complex, it very effectively cleaves C3, the critical step for opsonization by C3b and liberation of the C3a anaphylatoxin.

These findings have important therapeutic implications, suggesting that a drug capable of inhibiting the binding of human CRP to its target ligands in vivo and thereby preventing it from activating complement should reduce infarct size, with corresponding clinical benefit. Furthermore, increased CRP production is a feature of the nonspecific acute phase response to a very wide range of traumatic, infectious, inflammatory, and neoplastic tissue-damaging conditions 31. In all of these, including disorders as diverse as burns, surgical trauma, rheumatoid arthritis, sepsis, and invasive neoplasia, there are nonirremediably damaged cells that, by analogy with myocardial infarction, are likely to be targeted by the opsonic and proinflammatory actions of CRP and complement. Specific inhibition of CRP binding in vivo might thus be expected to be of wide clinical benefit, and suitable compounds are currently being sought, supported by our recent description of the high resolution 3-dimensional structure of the physiological CRP–ligand complex 43.

The cardioprotective effect of total complement depletion at the time of experimental coronary artery occlusion is well established 4,5,6,7,8. However, in the clinical context of patients presenting with acute myocardial infarction, complement depletion could only be initiated after the onset of symptoms. We show here that injection of cobra venom factor up to 2 h after ligation of the coronary artery significantly reduces infarct size, despite the fact that maximal depletion of plasma C3 is only achieved 6 h after cobra factor administration. This encouragingly suggests that, after acute ischemia, the major pathogenic action of complement occurs after some hours, and there is therefore a window of therapeutic opportunity during which inhibition of complement activation could still reduce infarct size, with corresponding reduction in morbidity and mortality. Although no anticomplement drugs are yet in clinical use, our results support their development and testing.

We thank Mr. John Arnold for assistance with image analysis, Ms. Michelle Leppo for advice on surgical technique, and Ms. Janet Gilbertson for technical assistance.

This work was supported by UK Medical Research Council Programme grant G97900510 to M.B. Pepys.