Main

Complement component C9 is the principal cytolytic protein of the complement system. After activation of the alternative or classical complement pathway, the complement proteins C5b, C6, C7, and C8 are sequentially assembled on the surface of a target cell to form C5b-8 complexes that insert into the cell membrane(1–3). A complete membrane attack complex is created when a C5b-8 complex binds several molecules of C9(4). The C9 molecules unfold and self-polymerize to form a pore that penetrates the lipid bilayer of the cell membrane(5–8). If a sufficient number of membrane attack complexes are deposited, cell injury or death occurs(9–11).

The Gram-negative bacteria that cause sepsis in adult humans are characteristically resistant to the complement-mediated cytolytic activity of normal serum(12–16). In contrast, the strains of E. coli that cause sepsis in human neonates are usually killed during incubation in normal adult serum(“serum-sensitive”)(17). Even strains of Escherichia coli that express the pathogenic K1 capsule are usually serum-sensitive if they are isolated from neonatal blood(17). Similarly, about one-half of the K1-positive strains isolated from the cerebrospinal fluid of neonates are serum-sensitive(18). These conflicting observations in septic neonatal and adult subjects could be reconciled if the sera of neonates were deficient in the bactericidal activity required to kill serum-sensitive bacteria. Consistent with this theory, the sera of human neonates, compared with adults, contain diminished concentrations of C9 which restrict the capacity of the sera to kill pathogenic strains of Escherichia coli(19–21). The clinical significance of this observation is not known. Further investigation has been impeded by the lack of a suitable animal model of C9 deficiency in neonates.

Therefore, this study was undertaken to determine whether a diminished concentration of C9 predisposes neonatal rats to bacterial invasion and to determine whether, by ameliorating the cytolytic defect, the administration of C9 improves neonatal survival during experimental E. coli sepsis. Experiments were designed to 1) quantitate C9 serum concentrations in neonatal rats during development, 2) determine whether diminished C9 concentrations restrict the bactericidal and hemolytic capacities of neonatal rat serum, 3) assess whether the C9 concentration, bactericidal activity, and hemolytic activity of serum are correlated with the capacity of developing rats to survive experimental infection with a pathogenic strain of E. coli, and 4) determine the effect of the administration of C9 on the survival of neonatal rats infected with E. coli.

METHODS

Animals. Pregnant Sprague-Dawley rats were obtained from Harlan Sprague Dawley, Inc., Indianapolis, IN. After delivery, the pups were housed with their mothers under standardized conditions at the vivarium of the Research Resource Center at the University of Louisville School of Medicine. Use of the animals as described in this report was approved by the Animal Care and Use Committee of the University of Louisville.

Buffers and reagents. The following buffers and reagents were used: Dulbecco's PBS (Media Tech, Washington, DC), pH 7.4 (PBS); PBS containing 1 mM MgCl2 (PBS+); PBS+ containing 0.5% BSA(BPBS+; Sigma Chemical Co., St. Louis, MO); tryptic soy broth (Difco Laboratories, Detroit, MI); veronal-buffered saline (VBS), pH 8.6, containing 4.18 g/L sodium barbital (Sigma Chemical Co.) and 0.80 g/L barbital (Fisher Scientific, Pittsburgh, PA); agarose containing 10 g of reagent grade agarose(Bio-Rad, Hercules, CA) and 20 g of dextran 10 (Pharmacia, Piscataway, NJ) per liter of VBS; 100 mM MgEGTA (Sigma Chemical Co.). Buffers used during rocket immunoelectrophoresis and in the hemolytic assays contained 0.02% sodium azide. Buffers used in the bactericidal assays and in the in vivo experiments contained no bacteriostatic agents.

Serotyping. The serotypes of the two E. coli strains were determined by standard methods at the E. coli Reference Center of Pennsylvania State University, University Park, PA. Briefly, the strains were propagated on veal infusion yeast extract agar, harvested into formalin saline, heated at 100 °C for 2 h, and then tested for “O” serotype using 178 standard O antisera, or subcultured onto veal infusion yeast extract agar and tested for plaque formation by K-1 phages A-E. The strains were also evaluated for motility after inoculation and maintenance in motility media for 10 d at 37 °C.

Antibody titration. Agglutinating antibody was quantified in 96-well microtiter plates. Test sera were serially diluted 2-fold (1:2 to 1:256) in PBS, mixed with an equal volume of antigen O1:K1:NM in 0.6% formalin-saline, and incubated at 37 °C. Agglutination titers were determined after 18 h of incubation and compared with standard rabbit anti-O1:K1 sera and nonimmunized rabbit sera.

Bacteria. E. coli serotypes O1:K1:NM and O7:K1w:NM were used in the bactericidal assays and were administered to neonatal rats to induce experimental infection. These isolates of E. coli were characteristic of previously reported pathogenic strains obtained from human neonates. The isolates expressed the K1 antigen, were isolated from the blood of two septic human neonates, and were killed during incubation with normal serum pooled from adult humans(17). Before an experiment, the bacteria were grown overnight in 50 mL of tryptic soy broth. The organisms were washed three times in BPBS+ or PBS+ and adjusted by absorbance at 600 nm to a concentration of 5 × 105 cfu per mL of BPBS+ for use in the bactericidal assays. The concentrations of the bacteria were adjusted to 1.0 or 0.85 × 109(E. coli O1:K1:NM) or 2.5 × 106 (E. coli O7:K1w:NM) cfu/mL of PBS+ for use in the experiments in which neonatal rats were injected with E. coli. The bacterial concentrations were confirmed by quantitative culture.

Serum. Separate pools of serum were prepared from adult rats and from neonatal rats ranging in age from 1 to 21 d of life. Blood was obtained from the adults by needle puncture of the inferior vena cava and from the neonates by incision of the external jugular vein. The blood from each animal was allowed to clot in plastic test tubes for 1 h at room temperature. Next, the tubes were centrifuged at 12,000 × g for 5 min at room temperature. The sera were then aspirated, pooled according to the age of the animals, passed through sterile 0.2-μm filters, and stored in aliquots at -70 °C.

Antisera. Rabbit anti-rat C9 antiserum was produced by intradermal immunization of New Zealand White rabbits with purified rat C9 by standard methods(22). Rabbit antisera specific for human C3, C5, and C9 were purchased from Harlan Bioproducts, Indianapolis, IN. The antisera were stored at -20 °C.

Rat complement component C9. Complement component C9 was isolated from rat plasma by the method described by Jones et al.(23). The concentration of rat C9 was measured by the BCA protein assay (Pierce, Rockford, IL)(24) and adjusted to 1 mg/mL in PBS. A purity of >95% was determined by SDS-PAGE and by Western blot, which used rabbit anti-rat C9 antiserum. Electrophoretic transfer was accomplished using the method of Towbin et al.(25) as modified by Matsuidaira(26). The primary antibody was rabbit anti-rat C9 antiserum, and the secondary antibody was goat anti-rabbit IgG coupled to alkaline phosphatase and adsorbed with human plasma proteins (Sigma Chemical Co.).

Human complement component C9. Purified human C9 was purchased from Quidel Corporation, San Diego, CA, as a solution containing 1 mg/mL of PBS without bacteriostatic agents. The human C9 was isolated by the method of Biesecker and Müller-Eberhard(27) and was subjected by the manufacturer to an anti-impurities solid-phase immunoaffinity column containing the IgG fraction isolated from monospecific goat anti-human C3, -C5, -C6, -C7, and -C8 to ensure the complete removal of these complement proteins. SDS-PAGE and hemolytic assays by the manufacturer demonstrated that>95% of the protein in the solution was pure C9 and that the solution was completely free of C3, C5, C6, C7, and C8 functional activities.

To confirm the purity of the human C9 preparation used in this study, various assays were also performed in our laboratory. By nephelometry (Beckman Array Protein System, Beckman Instruments, Brea, CA), the solution did not contain detectable IgA, IgM, or IgG. Because neonatal serum normally contains diminished concentrations of various complement components, such as C3 and C5, assays were performed to determine the amount of these two proteins in the C9 preparation used in experiments to supplement neonatal serum. The C9 preparation was subjected to Western immunoblot analysis(26) using rabbit antisera specific for human C3, C5, and C9 and a secondary antibody as described above. Trace amounts of C3 were detected based upon a series of standards but represented <0.1% of the protein in the C9 preparation. C5 was not detected in the C9 preparation. SDS-PAGE analysis revealed that greater than 90% of the protein in the solution was represented by a single band with an approximate molecular mass of 71 kD. Each of the minor bands detected by SDS-PAGE, including a band with a molecular mass >180 kD, reacted with the anti-C9 antiserum in the Western blot assay, suggesting trace amounts of aggregated C9 or C9 degradation products that may have developed during storage. The C9 was stored in 10-μL aliquots at -70 °C.

Purification of C9 from human plasma. For use in vivo, a large quantity of C9 was required, and commercially available C9 was prohibitively expensive. Therefore, C9 was purified from human plasma using a modification of previously described methods(27, 28). Briefly, Cohn fraction III (Armour Pharmaceuticals, Kankakee, IL) was solubilized in PBS containing protease inhibitors, fractionated with PEG-4000 (BDH, Liverpool, England), and purified by chromatography on lysine-Sepharose, Q-Sepharose, S-Sepharose, and Sephacryl HR-200 (all from Pharmacia Biotech Inc., Piscataway, NJ). The purity and biologic activity of the C9 were confirmed in SDS-PAGE analyses, in hemolytic and bactericidal assays, and in the in vivo experiments described below.

IgG for in vitro and in vivo use. IgG antibodies pooled from adult human plasma and modified for i.v. infusion (IVIG; Iveegam) was purchased from Immuno-U.S., Inc. The IVIG was dialyzed against PBS without bacteriostatic agents, adjusted to a concentration of 50 mg/mL, and stored in 1-mL aliquots at -70 °C.

Cobra venom factor. Lyophilized cobra (Naja naja kaouthia) venom factor was purchased from Quidel (San Diego, CA) and dissolved in sterile water. The solution contained 200 μg (105 U) of cobra venom factor per mL of PBS. The cobra venom factor was purified by the manufacturer by sequential column chromatography using DEAE-Sephacel then Sephacryl-300. Phospholipase contaminants were inactivated with 2 mM p- bromophenacyl bromide. SDS-PAGE performed by the manufacturer revealed a purity of at least 95%. In our laboratory, the composition and a purity of at least 95% were confirmed by SDS-PAGE in the presence and absence of a reducing agent. Also, the i.p. injection of 10 μg of cobra venom factor abolished the hemolytic activity of serum subsequently obtained from 2-d-old rats for at least 4 d (see below for description of the hemolytic assay used). The cobra venom factor was stored in 20-μL aliquots at -70°C.

Quantitation of rat C9 in serum by rocket immunoelectrophoresis. Five-microliter samples of serum diluted with VBS were placed into wells created in agarose gels [1% agarose, 2% dextran 10 in VBS (wt/vol)] containing 7 μL of anti-rat C9 antiserum per mL. After electrophoresis at 10 V/cm for 4 h, the gels were rinsed, dried, and then stained with Coomassie Blue R-250 (0.2% dye in 50% methanol/10% acetic acid) and destained with 30% methanol/10% acetic acid. The heights of the rocket-shaped bands of precipitation were measured, and the results were compared with a standard curve derived from rockets produced by samples containing various quantities of purified rat C9. All gels containing the samples were run concurrently with a VBS negative control and rat C9 standards.

Quantitation of human C9 in rat serum. A radial immunodiffusion kit (Human C9 NL RID Kit) was purchased from The Binding Site (San Diego, CA). In duplicate, 5 μL of serum were transferred into the agar well, and the diameter of the immunoprecipitation ring was measured after 72 h of incubation at room temperature. The concentration of C9 was calculated from the regression equation derived from concurrent assays of control sera, provided by the vendor, which contained known concentrations of C9. The antibody provided in the agar did not cross-react with rat C9.

Hemolytic assay. The capacity of neonatal rat serum to lyse erythrocytes was determined by using a commercially available kit, the EZ Complement CH50 Test (Diamedix, Miami, FL). Five microliters of serum were mixed with 3 mL of buffer containing 6.7 × 106 sheep erythrocytes/mL coated with rabbit anti-sheep erythrocyte IgG antibodies. The buffer, provided by the vendor, was Veronal-buffered saline containing gelatin, glucose, calcium, and magnesium. After incubation for 1 h at room temperature, the mixture was centrifuged, and the supernate was separated. The absorption at 415 nM was used to quantitate the proportion of erythrocytes lysed. Control samples included adult rat serum, heat-treated (56 °C, 30 min to inactivate the complement system) serum from adult rats and from neonatal rats, and buffer without erythrocytes. The A415 resulting from 100% lysis was determined by analyzing the supernate of erythrocytes that had been incubated with 3 mL of distilled water rather than buffer.

Bactericidal assay. For E. coli O1:K1:NM, 10 μl of BPBS+ containing 5 × 103 cfu were incubated at 37 °C for 120 min with 30 μL of neonatal rat serum in a final reaction volume of 100 μL. In some experiments, the serum was supplemented with human or rat C9. In each case, the volume of the mixture was adjusted to 90 μL with BPBS+ before the addition of the bacteria. Controls included mixtures containing adult rat serum, adult rat serum treated with heat (56 °C, 30 min) or 10 mM MgEGTA, heat-treated neonatal serum, buffer without serum and, in some cases, human or rat C9 without serum. Immediately after adding the bacteria and at 15 min intervals, 10 μL of the mixture were removed for quantitative culture in triplicate on tryptic soy agar by the pour-plate technique.

For E. coli O7:K1w:NM, E. coli suspended in BPBS+ were incubated with an equal volume of PBS or PBS containing 5 mg IVIG/mL for 10 min at 37 °C just before each assay. Each mixture of 100μL contained 50 μL of neonatal rat serum and 5 × 103 cfu of the preincubated E. coli O7:K1w:NM. Some mixtures also contained IVIG (500 μg/mL) and/or supplemental human C9 (100 μg/mL). The bacteria were also incubated with control mixtures containing adult rat serum, adult rat serum treated with heat or 10 mM MgEGTA, and heat-treated neonatal rat serum. Serum-free control mixtures contained IVIG or human C9 in buffer or buffer alone without additives. Just before the incubation and after 60 min of incubation, 10-μL aliquots were removed for quantitative culture.

The results of the quantitative cultures were used to calculate a bactericidal index to indicate the capacity of serum to kill the bacteria. The bactericidal index was calculated by using the followingformula:

Experimental infection of neonatal rats with E. coli. First, studies were conducted to determine whether transthoracic injections of bacteria resulted in bacteremia. At various intervals after injecting 1.7× 106 cfu of E. coli O1:K1:NM or 5 × 103 cfu of E. coli O7:K1w:NM per g body weight into the right lung, blood was obtained by incision of the external jugular vein for quantitative culture.

Next, groups of 9-12 neonatal rats, ranging in age from 2 to 21 d, were injected with sterile buffer or E. coli O1:K1:NM. The litters were mixed on the 2nd d of life. Each neonate was injected in the right lung by the transthorax route with 2 μL per g body weight of PBS+ or PBS+ containing 1.0 × 106 cfu of E. coli O1:K1:NM per μL. Each litter injected with bacteria was matched for age with a litter of 10 neonates injected with sterile buffer. After the injections, the number of surviving animals was recorded daily for 1 wk.

Other experiments were designed to determine the effect of the administration of human C9 on the survival of neonatal rats. Litters of 2-d-old rats were mixed and divided into groups of 9-12 animals. Each animal was injected i.p. with 200 μL of PBS, PBS containing 10 μg of cobra venom factor, or PBS containing 500 μg of human C9. Four hours later, each animal received a transthoracic injection into the right lung of 2 μL of PBS+ per g body weight or PBS+ containing 0.85 × 106 cfu E. coli O1:K1:NM per μL. Other controls included groups of neonates injected with placebo both i.p. (PBS) and transthoracically(PBS+) and groups of neonates injected i.p. with cobra venom factor and transthoracically with PBS+. The number of surviving animals was recorded at intervals for 1 wk.

Experiments were also conducted to determine the protective effect of human C9 administered to neonatal rats infected with E. coli O7:K1w:NM and to determine whether the C9 would potentiate the protective effect of IVIG. Groups of 9-12 2-d-old rats were used. Each animal was injected i.p. with 30μL/g body weight of PBS or PBS containing 2.5 μg of C9/μL. Three hours later, each animal also received an i.p. injection of 20 μL of PBS/g body weight or PBS containing 5.75 μg of IVIG/μL. One hour later, each neonate was injected transthoracically with 2 μL/g body weight of PBS+ containing 5 × 103 cfu of E. coli.

Statistical analysis. Correlations between continuous variables were determined by multiple linear regression analysis. The means of continuous variables were compared by the t test for independent groups. Categoric variables were compared by using the χ2 test for homogeneity. In the experiments designed to compare the bactericidal effect of human C9 with rat C9, the slopes of the dose-response curves were compared using the sequential regression F test. The effects of the administration of human C9, cobra venom factor, or IVIG on the survival of experimentally infected neonates were determined using the log rank test.

RESULTS

C9 serum concentrations. The concentration of C9 in the pooled adult rat serum was 224 ± 7.2 μg/mL (mean ± SEM). In contrast, the concentration of C9 in the serum pooled from 1-d-old rats was only 43 ± 3.8 μg/mL. During the first 3 wk of life, the C9 serum concentration increased to 170 ± 20 μg/mL. In the neonatal animals, the C9 serum concentration was positively correlated with age in days(r = 0.95, p < 0.01; Fig. 1A).

Figure 1
figure 1

Serum C9 concentrations, serum bactericidal activity, survival after E. coli O1:K1:NM infection, and serum hemolytic activity in neonatal rats during the first 3 wk of life. (A) Serum C9 concentrations were measured by rocket immunoelectrophoresis. (B) The bactericidal indices were derived from bactericidal assays. (C) The mean percentages of surviving neonates (9-12 per group) were determined 1 wk after intrapulmonary injections with E. coli. (D) The percentages of sensitized sheep erythrocytes lysed by the neonatal serum were measured by hemolytic assay. In panels A, B, and D, at each point the data are expressed as the mean ± SEM of at least three replicate experiments.

Full size image

Serum bactericidal activity. E. coli O1:K1:NM were killed during 120 min of incubation with adult serum and with serum from neonates of various ages ranging from 2 to 21 d (bactericidal index > 90). However, during the first 30 min of incubation, the capacity of neonatal serum to kill E. coli was reduced compared with the adult serum. The bactericidal index of the adult serum during 30 min of incubation was 72± 3.2. In contrast, the bactericidal index was 25 ± 4.4 in the serum obtained from neonates on the 2nd d of life. The bactericidal index increased during the first 3 wk of life to 53 ± 2.6 (r = 0.97, p < 0.01; Fig. 1B).

The bacteria proliferated during incubation with serum from adult or neonatal rats treated with heat to inactivate both the classical and alternative pathways of complement. Also, the bacteria proliferated in adult serum treated with 10 mM MgEGTA, which blocks activation of the classical, but not the alternative complement pathway(29). The number of surviving bacteria did not change significantly during incubation with buffer in the absence of serum.

Survival of neonatal rats injected at various ages with E. coli. All animals injected transthoracically with sterile PBS+ survived. During 1 wk of observation, the survival of the animals was positively correlated with age at the time of injection with E. coli(r = 0.79, p < 0.04) (Fig. 1C). Similarly, neonatal survival after experimental infection with E. coli was positively correlated with the serum C9 concentration (r= 0.76, p < 0.05). By χ2 analysis, the survival was positively correlated with a serum C9 concentration ≥90 μg/mL(χ2 = 50.8, p < 0.01) and with a bactericidal index> 30 (χ2 = 39.1, p < 0.01).

Agglutinating antibody titers to E. coli O1:K1:NM. The antibody titers of serum obtained from rats of various ages were: adult, 1:4w (weak reaction); 2 d, 1:8w; 7 d, <1:4; 14 d, 1:4; 21 d, 1:4w. The titers were not correlated with the C9 concentrations, serum bactericidal activity, or survival after E. coli O1:K1:NM infection in neonatal rats during the first 3 wk of life.

Serum hemolytic activity. The adult serum lysed 100% of the sensitized sheep erythrocytes used in the hemolytic assay. In contrast, serum obtained from 2-d-old rats lysed only 10 ± 0.12% of the cells. During the first 3 wk of life, the hemolytic capacity of the neonatal serum increased such that 88 ± 3.6% of the erythrocytes were lysed during incubation with the serum from 21-d-old rats (r = 0.82, p < 0.02versus age; Fig. 1D). The hemolytic capacity of the neonatal serum was positively correlated with the C9 serum concentration(r = 0.78, p < 0.03) and with the survival of neonates experimentally infected with E. coli O1:K1:NM (r = 0.97,p < 0.01). Survival was more likely if the hemolytic capacity was high (hemolysis >80%) and was less likely if the hemolytic capacity was low(hemolysis <15%; χ2 = 50.1, p < 0.01).

Effect of in vitro supplementation with C9 on the hemolytic activity of neonatal serum. In the serum of 2-d-old rats, supplemental human C9 enhanced the capacity of neonatal rat serum to lyse sensitized sheep erythrocytes. The proportion of cells lysed was positively correlated with the dose of supplemental C9 (r = 0.87, p < 0.02;Fig. 2).

Figure 2
figure 2

Effect of supplemental C9 on the hemolytic activity of neonatal rat serum. A hemolytic assay was used to determine the percentage of sensitized sheep erythrocytes lysed during incubation with serum from 2-d-old rats. The serum was supplemented in vitro with various doses of human C9. At each point, the data are expressed as the mean ± SEM of three replicate experiments.

Full size image

Effect of in vitro supplementation with C9 on the bactericidal capacity of neonatal serum. The bactericidal activity of serum from 2-d-old rats against E. coli O1:K1:NM was enhanced after in vitro supplementation with either human or rat C9 in doses ranging from 0 to 200μg per mL. The slope of the human C9 dose-response curve was not significantly different from the slope of the rat C9 dose-response curve (Fig. 3).

Figure 3
figure 3

Effect of supplemental C9 on the bactericidal activity of neonatal rat serum. E. coli O1:K1:NM were incubated 45 min with serum obtained from 2-d-old rats and supplemented in vitro with increasing doses of either human or rat C9. The bactericidal index was calculated as described in the text. *p < 0.05 vs unsupplemented neonatal serum. At each point, the data are expressed as the mean ± SEM of three replicate experiments.

Full size image

The number of surviving bacteria did not change significantly during incubation with human or rat C9 in the absence of serum.

Effect of i.p. human C9 on the concentration of human C9 and the bactericidal capacity of neonatal serum against E. coli O1:K1:NM. The concentration of human C9 was determined in serum pooled from 2-d-old neonates 18 h after the i.p. administration of human C9 in doses ranging from 0 to 1000μg per animal. The serum concentration of human C9 was positively correlated with the dose (r = 0.99, p < 0.01). The bactericidal index of the serum was also increased after i.p. human C9(Fig. 4). A dose of 500 μg resulted in a serum concentration of 117 μg/mL and an increased bactericidal index of 70.

Figure 4
figure 4

Effect of i.p. human C9 in neonatal rat serum. The concentration of human C9 and the bactericidal index were quantitated in serum pooled from groups of neonatal rats 18 h after increasing i.p. doses of human C9. E. coli O1:K1:NM was used in the bactericidal assays. At each point, the data are expressed as the mean of at least three replicate experiments.

Full size image

In a different set of experiments, the concentration of human C9 was measured in serum obtained from 2-d-old rats at various intervals after the i.p. injection of 75 μg of human C9 per g of body weight. After a distribution phase of 8 h, the calculated elimination t1/2 of human C9 in the neonatal rats was 8.2 h (Fig. 5).

Figure 5
figure 5

Kinetics of i.p. human C9 in neonatal rats. Human C9 was quantitated in serum obtained from 2-d-old rats at various intervals after the i.p. administration of 75 μg of human C9/g of body weight. At each point, the data are expressed as the mean of at least three replicate experiments.

Full size image

Effect of human C9 on bacteremia and survival in neonatal rats infected with E. coli O1:K1:NM. Two-day-old rats received i.p. injections of PBS or PBS containing 500 μg of human C9. Four hours later, 1.7 × 106 cfu E. coli per g of body weight were injected into the right lung of each animal. Eight hours after the transthoracic injections,E. coli was isolated from the blood of all 13 animals that had received i.p. placebo. In each case, the concentration of bacteria exceeded 300 cfu/mL of blood (range: 356 to >1000 cfu/mL). Similarly, E. coli was isolated from the blood of all 10 of the C9 recipients (range: 70 to >1000 cfu/mL). However, the concentration of bacteria exceeded 300 cfu/mL in only 4 of the 10 neonates treated with C9 (p < 0.01versus placebo).

Compared with 45 2-d-old rats treated with i.p. placebo, the survival of 44 neonates treated with 500 μg of human C9 was enhanced after the intrapulmonary injection of E. coli O1:K1:NM (Fig. 6). In the neonatal animals that received placebo, the median survival time was 48 ± 6 h, and the mean survival time was 78 ± 8 h. In contrast, in the neonates treated with C9, the median and mean survival times were, respectively, 120 ± 30 h (p < 0.05 versus placebo) and 119 ± 8 h (p < 0.05 versus placebo). No deaths occurred in 30 neonates that were not infected but instead were injected with placebo buffer in both the right lung and the peritoneum. Similarly, no deaths occurred in 21 neonates injected with transthoracic PBS+ and i.p. cobra venom factor. However, of 23 neonates injected i.p. with 10 μg of cobra venom factor before transthoracic injections with E. coli, none survived more than 42 h.

Figure 6
figure 6

Effect of i.p. human C9 on the survival of neonatal rats infected with E. coli. Two-day-old rats received either i.p. placebo (PBS; n = 45) or 500 μg of human C9 (n = 44). Four hours later, E. coli O1:K1:NM were injected into the right lung(log rank p < 0.01, C9 vs placebo). At each point, the data are derived from four replicate experiments using 9-12 animals per group.

Full size image

Effect of supplemental C9 on the bactericidal activity of IVIG against E. coli O7:K1w:NM in neonatal rat serum. E. coli O7:K1w:NM proliferated during incubation with unsupplemented serum from 2-d-old rats. The bacteria also proliferated during incubation with neonatal rat serum supplemented with a sublethal dose of IVIG (500 μg/mL; a supplemental dose of 5 mg/mL resulted in a bactericidal index of 92). The titer of agglutinating antibody to E. coli O7:K1w:NM was 1:16 in the undiluted IVIG, and was 1:4w in the neonatal serum. The bactericidal index of neonatal serum supplemented with 100 μg C9/mL was 21. In contrast, the bactericidal index of neonatal rat serum containing both IVIG and supplemental C9 was 70. Other control experiments revealed that the bacteria proliferated in heat-treated serum (to inactivate both the classical and alternative pathways of complement) from adult or neonatal rats. Similarly, the bacteria proliferated in adult rat serum treated with 10 mM MgEGTA to selectively block activation of the classical complement pathway(29). The number of surviving bacteria did not change significantly during incubation with buffer alone or with buffer containing IVIG or C9 in the absence of serum.

To determine whether the IVIG might contain significant contaminants which possess intrinsic bactericidal activity or which (in addition to antibodies) enhance the complement-mediated cytolytic activity of neonatal serum, other control experiments were performed. The survival of E. coli O7:K1w:NM was unaffected during incubation with buffer containing IVIG (500μg/mL) in the absence of serum. The bacteria proliferated during incubation with MgEGTA-treated neonatal serum which had been supplemented with both C9 and IVIG, indicating a requirement for activity of the classical complement pathway. Also, supplemental IVIG (5 mg/mL) did not enhance the capacity of serum from 2-d-old rats to lyse sensitized sheep erythrocytes.

Effect of transthoracic injections of E. coli O7:K1w:NM on the development of bacteremia in neonatal rats. Nine 2-d-old rats were injected transthoracically with 2 μL of PBS+ containing 5 × 103 cfu E. coli per g body weight. At 1, 2, or 24 h after the injections, animals were killed, and blood was obtained for quantitative culture. One hour after the injections, the blood of both animals that were cultured contained >3.5 × 103 cfu E. coli/mL. Two hours after the injections, the blood concentration of bacteria exceeded 6× 103 cfu/mL in the blood of all three animals that was cultured. Blood cultures were performed in the remaining animals 24 h after the injections. In each case, the concentration of bacteria exceeded 1 × 104 cfu/mL.

Effect of intraperitoneal C9 on the capacity of IVIG to protect neonatal rats infected with E. coli. Before the transthoracic injection of E. coli O7:K1w:NM, 37 animals received only i.p. PBS (buffer placebo only), 40 received i.p. IVIG, 38 received i.p. C9, and 40 received both IVIG and C9. None of the placebo controls survived. The mean duration of survival in the IVIG-treated group was 109.8 ± 9.73 h and was 116.63 ± 9.35 h in the group treated with C9. In contrast, the mean duration of survival in the group treated with both IVIG and C9 was 146.2 ± 7.54 h(p < 00.01 versus IVIG or C9). At the end of the week of observation, survival was significantly increased in the animals that received either IVIG or C9 (p < 0.01 versus placebo). Moreover, the proportion of animals that survived was significantly greater in the group which received the combination of C9 and IVIG compared with the other three groups (p < 0.01; Fig. 7).

Figure 7
figure 7

Effect of human C9 on the capacity of IVIG to protect neonatal rats infected with E. coli. Two-day-old rats received i.p. placebo (PBS) or 75 μg of C9/g of body weight. Three hours later, they also received either i.p. PBS or 115 μg of IVIG/g of body weight followed 1 h later by the transthoracic injection of E. coli O7:K1w:NM. The number of subjects in each group was: placebo only, 37; C9 alone, 38; IVIG alone, 40; both C9 and IVIG, 40 (log rank p < 0.01, C9vs placebo, IVIG vs placebo, and C9 + IVIG vs each of the other three groups). At each point, the data are derived from four replicate experiments using 9-12 animals per group.

Full size image

DISCUSSION

The serum of neonatal rats, like neonatal humans(19, 20), contained a diminished concentration of complement component C9 compared with adult serum. During the first 3 wk of life, the C9 serum concentration increased to a level which approximated that observed in adult rats. Similarly, the bactericidal activity of the neonatal serum was diminished but increased during the first 3 wk of life. Whether added in vitro or administered in vivo, supplemental C9 enhanced the capacity of the neonatal serum to kill pathogenic E. coli. This observation indicated that the diminished C9 concentration restricted the bactericidal capacity of neonatal serum and that the serum contained sufficient quantities of the other components of the classical complement pathway required to kill the bacteria. Furthermore, the survival of neonates experimentally infected at various ages with E. coli was positively correlated with the C9 concentration and with the bactericidal activity of their serum. Intraperitoneal human C9 diminished the intensity of bacteremia and enhanced survival in neonatal rats infected with E. coli. The data indicate that C9 deficiency is one of the defects in antibacterial host immunity which predisposed the neonatal rats to invasion and septic death caused by E. coli.

These observations in the neonatal rat parallel previous findings in human newborn infants. The serum C9 concentration of full-term human neonates is approximately 15% of that observed in adults(20). Supplemental C9 enhances the capacity of serum from human neonates to kill pathogenic strains of E. coli, suggesting that diminished C9 concentrations restrict the bactericidal activity of their serum(21). Whether C9 deficiency predisposes human neonates to invasion by E. coli is not known. This report describes, for the first time, an animal model of neonatal C9 deficiency that allows the in vivo assessment of the effects of C9 administration and blockade of C9 deposition in the development of E. coli infection, experimentation which is not ethically justified in humans.

In vitro, the bactericidal activity of rat serum was abolished with heat inactivation to prevent complement activation or by treatment with MgEGTA to selectively block activation of the classical, but not the alternative, complement pathway(29). The i.p. administration of cobra venom factor to induce the consumption of complement components in vivo(30) abolished hemolytic activity in the neonatal serum and resulted in rapid death after E. coli infection. These observations are consistent with the theory that complement activation is a critical determinant of antibacterial host defense in neonates. In contrast, herein reported for the first time, the prophylactic administration of a purified complement component enhanced antibacterial host defense.

The protection afforded by the administration of C9 was incomplete. During the week of observation, approximately 50% of the treated animals died. Whether larger or repetitive doses of C9 would have altered the outcome is not known. Possibly, the unexpectedly short elimination t1/2(about 8 h) may have limited the protective effect of the protein. Four days after the i.p. injection of 75 μg/g body weight, human C9 was no longer detectable in serum from the neonates. The elimination t1/2 of native C9 in the neonatal rat is not known. Consequently, it is possible that human C9 is catabolized or excreted rapidly compared with native C9 in the neonatal rat. Alternatively, rapid catabolism and/or elimination may be a characteristic of C9 in neonates, whether the protein is derived from the same or a different species. It is unlikely that the observed clearance resulted from the development of anti-human C9 antibodies because the onset and rate of clearance was so rapid. The factors which determine the concentration of C9 in neonatal serum will require continued investigation into the genetic regulation of C9 synthesis as well as the mechanism(s) of catabolism and excretion during development.

Supplemental C9 provided only partial restoration of hemolytic activity, and C9 administration prolonged survival during E. coli sepsis without completely preventing death. These observations suggest that C9 deficiency is only one of several developmental defects in host immunity that predisposed the neonates to bacterial invasion. The neonatal immune system is characterized by reduced numbers and function of phagocytic leukocytes, reduced concentrations of several complement components other than C9, and variable concentrations of anti-E. coli antibodies(19, 31–34). In adults and older children, inherited C9 deficiency impairs the bactericidal activity of serum against E. coli(35) and may predispose affected subjects to the acquisition of neisserial infections(36). Nonetheless, the incidence of invasive E. coli disease does not appear to be increased in adults with isolated C9 deficiency who, unlike neonates, are otherwise immunologically intact(37, 38). Therefore, the enhancement of serum bactericidal activity and resistance to E. coli infection observed in neonatal rats during the first 3 wk of life probably resulted from the maturation or interaction of several factors participating in antibacterial immunity, in addition to rising C9 serum concentrations.

In sufficient dosage, IVIG enhanced serum bactericidal activity and improved survival after E. coli sepsis. Supplemental C9 enhanced the bactericidal and protective effects of IVIG. These findings suggest that deficiencies of both anti-E. coli antibodies and C9 contributed to the inefficient serum bactericidal activity and diminished host defense observed in the neonatal animals. However, the agglutinating antibody titers to E. coli O1:K1:NM did not change significantly during the first 3 wk of life. This observation suggests that an increasing antibody concentration in serum is an unlikely explanation for the enhancement of serum bactericidal activity and the capacity to survive E. coli infection seen in the developing animals. The data did reveal that the serum C9 concentration and bactericidal activity were positively correlated with resistance to E. coli infection in the neonates, and that the administration of C9 enhanced antibacterial host resistance against two pathogenic serotypes of E. coli. The results consequently indicate that C9 deficiency was one of the defects in antibacterial host resistance that predisposed the animals to bacterial invasion. As previously noted, C9 deficiency has been shown to restrict the bactericidal activity of serum from neonatal humans. Therefore, it is possible that C9 deficiency is a defect in host resistance that predisposes neonatal humans to the acquisition of invasive disease and sepsis caused by E. coli. Defining the relative importance of C9 deficiency, compared with other immune defects, will require further investigation.

Similar to the serum bactericidal activity, the hemolytic activity of serum obtained from neonatal rats during the first few days of life was diminished compared with the hemolytic activity of adult serum. Serum obtained during the first 4 d of life contained a serum concentration of C9 that was less than 60μg/mL and lysed less than 20% of the sensitized sheep erythrocytes in the hemolytic assay. In contrast, serum obtained on or after the 7th d of life contained a serum C9 concentration equal to or greater than 90 μg/mL and lysed more than 80% of the erythrocytes. Possibly, the serum C9 concentration increased during the 1st wk of life above a threshold concentration that is required for efficient hemolytic activity. Alternatively, the concentration of several complement components, in addition to C9, may have increased during the 1st wk and thereby enhanced the serum hemolytic activity.

Supplemental C9 enhanced the hemolytic activity of serum obtained from 2-d-old rats in a dose-dependent fashion. This observation suggests that the diminished concentration of C9 restricted the hemolytic capacity of the neonatal serum. However, even after supplementation with saturating quantities of supplemental C9, the neonatal serum lysed only 50% of the sheep erythrocytes used in the hemolytic assay. In contrast, the adult rat serum lysed 100% of the erythrocytes. Possibly, the neonatal serum was deficient in one or more components of the classical complement pathway, other than C9, which are required to lyse antibody-sensitized sheep erythrocytes(39). The concept that developmental deficiencies of complement components other than C9 also limit cytolytic activity is further supported by the observation that unsupplemented serum from 7-d-old animals lysed erythrocytes more efficiently than C9-supplemented serum from 2-d-old animals. Alternatively, C5b-8 complexes assembled from rat complement components on the erythrocyte surface might bind more avidly to rat C9 than to the human C9 used as a supplement in these experiments. An explanation for the limited effect of supplemental human C9 on the hemolytic activity of neonatal rat serum will require further investigation.

In summary, a diminished concentration of C9 restricted the bactericidal and hemolytic activity of serum from neonatal rats. The serum C9 concentration, bactericidal activity, and hemolytic activity increased during the first 3 wk of life and were correlated with the capacity of neonatal rats to survive an experimental infection with E. coli. The prophylactic administration of human C9 enhanced neonatal survival after septic infection with two pathogenic strains of E. coli.

We conclude that the diminished C9 concentration of neonatal serum is one of the defects in antibacterial host immunity that predispose neonatal rats, and possibly neonatal humans, to the acquisition of invasive E. coli disease.