Role of serum amyloid P component in bacterial infection: Protection of the host or protection of the pathogen (original) (raw)

Abstract

Serum amyloid P component (SAP) binds to Streptococcus pyogenes, and we show here that it also binds to_Neisseria meningitidis_, including a lipopolysaccharide (LPS)-negative mutant, and to rough variants of Escherichia coli. Surprisingly, this binding had a powerful antiopsonic effect both in vitro and in vivo, reducing phagocytosis and killing of bacteria. Furthermore, SAP knockout mice survived lethal infection with S. pyogenes and rough E. coli J5, organisms to which SAP binds. The susceptibility of SAP−/− mice was fully restored by injection of isolated human SAP. However, SAP−/− mice were more susceptible than wild-type animals to lethal infection with_E. coli_ O111:B4, a smooth strain to which SAP does not bind, suggesting that SAP also has some host defense function. Although SAP binds to LPS in vitro, SAP−/− mice were only marginally more susceptible to lethal LPS challenge, and injection of large amounts of human SAP into wild-type mice did not affect sensitivity to LPS, indicating that SAP is not a significant modulator of LPS toxicity in vivo. In contrast, the binding of SAP to pathogenic bacteria enabled them to evade neutrophil phagocytosis and display enhanced virulence. Abrogation of this molecular camouflage is thus potentially a novel therapeutic approach, and we show here that administration to wild-type mice of (R)-1-[6-(R)-2-carboxy-pyrrolidin-1-yl]-6-oxo-hexanoyl]pyrrolidine-2-carboxylic acid, a drug that inhibits SAP binding, significantly prolonged survival during lethal infection with E. coli J5.

The pentraxin plasma proteins, C-reactive protein (CRP), the classical acute phase protein, and serum amyloid P component (SAP) (1), belong to the highly conserved lectin fold superfamily (2) with functions related to specific calcium-dependent ligand binding (3, 4). SAP binds to DNA in chromatin (5) exposed to or released in the extracellular environment by apoptosis or necrosis (6, 7), stabilizes it, and protects it from degradation (8, 9). SAP-deficient mice, created by gene targeting (10), degrade chromatin more rapidly than normal, have enhanced antibody responses to exogenous chromatin, and spontaneously develop marked anti-chromatin autoimmunity (9). A major normal function of SAP is thus to modulate handling of chromatin autoantigens and prevent loss of tolerance. On the other hand, SAP also binds to all types of amyloid fibrils, the abnormal protein aggregates responsible for the amyloidosis diseases (1). Binding of SAP stabilizes amyloid fibrils_in vitro_ (11) and contributes to pathogenesis of amyloidosis_in vivo_, as shown by delayed and reduced amyloid deposition in SAP-deficient mice (10). However, in addition to these normal and pathological endogenous ligands, SAP also binds to some bacteria, including the important human pathogen Streptococcus pyogenes, suggesting a possible role in host defense (12). Here we report studies of bacterial infection in SAP knockout mice and the surprising finding that although SAP does contribute to survival during some infections, it also inhibits phagocytosis of those bacteria to which it binds and thus enhances their virulence.

Materials and Methods

In Vivo Infection Experiments.

Pure line 129SV mice with targeted deletion of the gene for SAP and total deficiency of SAP (SAP−/−) were created as previously reported (10). The SAP gene deletion was also introduced into pure line C57BL/6 mice by back-crossing for at least 10 generations before the animals were used in the present experiments. Groups of 15–24 SAP−/− and wild-type 129SV mice and of 10–13 SAP−/− and wild-type C57BL/6 mice, all 10–14 weeks old, weighing 20–25 g, and age weight and sex matched within each experiment, were infected on day 0; morbidity and mortality were then recorded until no more deaths occurred. Bacteria were grown in broth as described below and harvested in log phase, and the concentrations used for infection were checked by colony counting. Pure human SAP was isolated as described (13).

Treatment of Infection by Inhibition of SAP Binding.

A potent, nontoxic inhibitor of SAP binding, (R)-1-[6-(R)-2-carboxy-pyrrolidin-1-yl]-6-oxo-hexanoyl]pyrrolidine-2-carboxylic acid (designated here as compound R), which inhibits and reverses binding of SAP to amyloid deposits in vivo (14), has been developed as a possible therapeutic agent for amyloidosis (refs. 11 and14; patent EP-A-915088). This drug in solution in water or neutral pH buffer was administered by repeated i.p. injection to groups of 10 age-, weight-, and sex-matched adult wild-type C57BL/6 mice receiving an LD90 dose of Escherichia coli J5, and the response to infection was observed as above. Controls received the solvent alone.

Bacteria.

E. coli O111:B4 (from B. J. Applemelk, Vrije Universiteit, Amsterdam, The Netherlands), E. coli J5, and_E. coli_ F515 (from U Mamat, Forschungzentrum, Borstel, Germany) were grown overnight to stationary phase in Luria broth (Oxoid, Basingstoke, UK) and then reinoculated and grown to log phase with an optical density of 1.0, corresponding to about 109 cells per ml, and checked by colony counting.S. pyogenes H305, a scarlet fever serotype M1T1 clinical isolate from a positive blood culture of a patient with invasive streptococcal infection (rederived from strain H250 provided by the Public Health Laboratory Service, Colindale, England), used extensively in murine models of infection (15), was grown overnight to stationary phase in Todd Hewitt broth containing 0.2% yeast extract (Oxoid) and then reinoculated and grown to log phase with an optical density of 1.0, and the number of cells per milliliter was checked by colony counting. In some experiments organisms were killed by heating for 30 min at 80°C before testing, but this had no effect on SAP binding. Neisseria meningitidis strains (from P. Brandtzaeg, Ullevål University Hospital, Oslo, Norway) were isolated from typical cases of meningococcal disease, apart from the lipolpolysaccharide (LPS)-negative strain provided by P. van der Ley (Bilthoven, The Netherlands) (16). All were grown overnight in brain heart infusion broth containing 5% FCS, then centrifuged, heat killed (80°C for 30 min), and resuspended to standard optical density corresponding to 2 × 108 cells per milliliter. Calcium-dependent binding of SAP from 1-ml aliquots of pooled whole normal human serum, containing ≈30 mg/liter SAP, to aliquots of 109 to 1010 bacterial cells, was tested precisely as described in ref. 12 by using both immunoassay for SAP eluted with EDTA and direct counting of bound125I-SAP spiked into the serum. Identical results were obtained with the two methods, and the values shown are representative of reproducible replicate experiments with each organism. The number of SAP molecules bound was calculated from the number of bacteria used and the molecular mass, 127,310 Da, of the physiological pentameric form of human SAP (17, 18).

Phagocytosis by Neutrophils.

Smooth E. coli O111:B4 and rough (Rc) mutant E. coli J5 were transformed, by using plasmid pFPV25.1 containing the_gfp-mut 3a_ gene under the constitutive S13 ribosomal protein promoter (from D. Holden, Imperial College School of Medicine, London), to express green fluorescent protein (19), and were grown as described above. After washing in sterile PBS, pellets of 109 organisms were resuspended and incubated for 1 h at 4°C in 1-ml volumes of the following reagents: 10 mM Tris-buffered 140 mM NaCl containing 2 mM CaCl2 and 4% wt/vol BSA at pH 8.0 (TCB buffer) [this concentration of albumin is essential for preventing the autoaggregation of isolated human SAP (18)]; TCB buffer containing isolated pure human SAP (13) at 30 mg/liter; whole heat-inactivated (56°C, 30 min) pooled normal human serum containing 30 mg/liter SAP; the same serum from which all SAP had been removed by absorption with sheep anti-SAP antibodies immobilized on CNBr-activated Sepharose (20); whole serum, heat inactivated as above, absorbed with immobilized sheep anti-C-reactive protein antibodies instead of anti-SAP (21); and SAP-depleted heat-inactivated serum reconstituted with 30 mg/liter of isolated SAP. Leukocytes from healthy normal individual donors were isolated by centrifugation and washing after lysis of erythrocytes with ammonium chloride, resuspended in Hanks' balanced salt solution, and incubated for 30 min at 37°C with 0.1 μg/ml dihydroethidine. Volumes of 10 μl of bacteria suspended in the various reagents and 100 μl of leukocyte suspension in Hanks' were then mixed to provide 10 bacteria per polymorph and were rotated at 37°C for 15 min to allow phagocytosis to proceed before fixing in 1% paraformaldehyde in PBS. Ingestion of bacteria containing green fluorescent protein (absorption 480 nm, emission 540 nm; ref. 22) and triggering of the respiratory burst, indicated by oxidation of dihydroethidine to red fluorescent ethidium bromide (absorption 480 nm, emission 593 nm; ref.23), were quantified by flow cytometry (Coulter Epics XL-MCL with System II software) with the forward scatter threshold set to 140 to exclude free bacteria from the analysis. At least 10,000 neutrophils, identified by well-established light scatter and autofluorescence parameters, were analyzed in each sample. Within each experiment, the relative phagocytosis of bacteria in the different reagents was quantified by comparison of mean cellular green fluorescence in cells that showed red fluorescence, demonstrating that they had mounted an oxidative burst response; this quantification excluded neutrophils that had not actively phagocytosed bacteria. For investigation of neutrophil phagocytosis of N. meningitidis, the serogroup B strain H44/76 was grown in broth and then heat killed, as described above, before labeling by incubation at 109 organisms per milliliter with 50 μg/ml fluorescein isothiocyanate in 0.1 M NaHCO3 at pH 9.6. In two separate sets of experiments, labeled bacteria were mixed with either undiluted whole serum immunodepleted of SAP, as described above, or control serum treated with immobilized normal sheep IgG or with 10% vol/vol SAP-depleted whole serum in TCB buffer, containing 1% wt/vol BSA, to which was added a range of concentrations of isolated pure SAP. After preincubation for 1 h at 37°C the mixtures were added to leukocyte preparations, and the phagocytosis assay and flow cytometry were performed as described above for E. coli. Three different individual leukocyte donors were used in each experiment with each bacterium. For display purposes, results within each experiment have been normalized to a value of 100% assigned to the mean of the maximal phagocytosis in that experiment.

Bacterial Killing in Vivo.

An acute phase response was induced in both wild-type SAP+/+ and SAP−/− C57BL/6 mice by s.c. injection of 0.5 ml of aqueous 2% wt/vol AgNO3, and pooled serum collected 24 h later contained 210 mg/liter SAP from the wild-type mice and 0 mg/liter SAP from the SAP−/− animals. Isolated pure mouse SAP (24) was added, to 100 mg/liter, to some of the SAP−/− serum, and all of the reagents were then heat inactivated at 56°C for 30 min. E. coli J5 were washed in PBS, and separate pellets containing 108 organisms were resuspended and then rotated for 1 h at 4°C with 100-μl volumes of each of the serum reagents described above. The bacterial suspensions were then diluted to 107 organisms per milliliter in Hanks' balanced salt solution, and 1-ml volumes were injected intraperitoneally into individual adult female C57BL/6 mice weighing 20–25 g each. Exactly 15 min later the mice were killed by CO2 narcosis, the peritoneal cavity of each was washed out with 10 ml of PBS, and the number of surviving bacteria was determined by colony counting. The results shown are pooled from two identical experiments, each containing five mice per group.

Lipopolysaccharide Toxicity in Vivo.

Groups of 6–12 adult BALB/c mice, age, weight, and sex matched in each experiment, received i.p. injections of purified human SAP (100 mg/kg) or equimolar amounts of purified human serum albumin or human CRP (25), all in solution in Tris-buffered physiological saline (pH 8.0). Twenty minutes later all animals received 200 μg per mouse of_Salmonella typhimurium_ LPS by i.p. injection at time 0. In some experiments, a second injection of each protein was given after 4 h. The toxicity of E. coli O111:B4 LPS was also compared in groups of 10 age-, weight-, and sex-matched SAP−/− and wild-type SAP+/+ C57BL/6 mice, after i.p. injection of 10 mg/kg or 1 mg/kg of LPS. In all experiments, morbidity and mortality were observed until no more deaths occurred.

Results

Binding of SAP to Bacteria in Vitro and the Outcome of Bacterial Infection in SAP-Deficient Mice.

Pure line 129SV SAP−/− mice infected with_E. coli_ O111:B4, a mouse pathogenic smooth strain to which SAP shows no appreciable binding in vitro (ref. 12; Table1), had significantly greater morbidity and mortality than wild-type SAP+/+ controls (Fig.1a). SAP evidently contributes to survival during infection with this organism. However, surprisingly, and in marked contrast, infection with S. pyogenes, a pathogen to which SAP binds in substantial amounts (ref. 12; Table 1), produced significantly lower mortality in the SAP knockouts than in wild-type controls (Fig. 1b).

Table 1.

Binding of human SAP to bacteria in vitro

Bacterial species SAP binding (molecules of SAP bound per bacterial cell)
E. coli O111:B4 (smooth) <5 × 102
E. coli J5 (Rc) 2.1–6.0 × 104
E. coli F515 (Re) 2.8 × 104
S. pyogenes H305 2.8–6.0 × 103
N. meningitidis clinical isolates, types A, B, and C 4.3 × 103 to 1.7 × 104
N. meningitidis H44/76 (LPS-sufficient wild type) 4.9 × 104
N. meningitidis H44/76[pHBK30] (LPS-deficient mutant) 2.8–4.0 × 104

Figure 1.

Figure 1

Effect of SAP deficiency on survival of 129SV mice during bacterial infection. (a) E. coli O111:B4, 4.4 × 108 organisms injected i.v. (b) S. pyogenes H350, 2.1 × 105 organisms injected i.p. P values (log rank test) represent significant differences in survival between SAP−/− and SAP+/+ mice. Each result shown is typical of two to four identical experiments with each organism.

SAP specifically recognizes cyclic pyruvate acetals (1, 12, 26), but the ligand for SAP expressed by S. pyogenes is not known. However, binding of SAP to Gram-negative bacterial LPS has recently been reported (27), and we therefore investigated the interaction of SAP with smooth and rough E. coli strains expressing different forms of LPS. No binding to the smooth strain was detected, but substantial amounts of SAP bound to rough variants on which the central and core structures of LPS were exposed (Table 1). Nevertheless LPS is not the only SAP ligand on Gram-negative bacteria, as there was marked binding of SAP to an LPS-deficient mutant of N. meningitidis (Table 1).

The difference between binding of SAP to rough and smooth E. coli in vitro was reflected in a dramatic difference in survival between SAP−/− and SAP+/+ mice after lethal infection in vivo. Thus whereas pure line C57BL/6 SAP−/− mice were again more susceptible than wild-type animals to infection with E. coli O111:B4 (Fig.2a), SAP-deficient animals survived remarkably better than wild-type mice when infected with the rough E. coli J5 strain (Fig. 2b). The critical specific role of SAP itself was confirmed by the administration of isolated pure human SAP to SAP−/− mice during infection with E. coli J5. This administration of human SAP completely restored the greater mortality typical of wild-type SAP+/+ animals (Fig. 2c).

Figure 2.

Figure 2

Effect of SAP deficiency on the survival of C57BL/6 mice during bacterial infection. (a) E. coli O111:B4, 108 organisms injected i.v. (b) E. coli J5, 109 organisms injected i.v. Each result shown in a and b is typical of at least two identical experiments with each organism. (c)E. coli J5, 109 organisms injected i.v. at time 0; isolated human SAP, 10 mg per mouse, was injected i.p. at −2 h and 16 h. P values (log rank test) represent significant differences in survival between SAP−/− and SAP+/+ or SAP-reconstituted SAP−/− mice.

Treatment of Bacterial Infection by Inhibition of SAP Binding.

The binding of SAP to amyloid fibrils makes a significant contribution to the pathogenesis of amyloidosis (10, 11). We have therefore lately developed compounds that inhibit ligand binding by SAP in vitro and in vivo, for use as drugs to treat amyloidosis (14). The addition of one of these inhibitors, compound R, to human serum at 50 μM completely inhibited binding of SAP to_E. coli_ J5. For example, in a typical experiment, 4 × 104 molecules of SAP bound to each bacterial cell in the presence of Tris-buffered saline containing calcium, whereas no SAP binding was detectable in the presence of 10 mM EDTA or any concentration of compound R between 5 mM and 50 μM. Lower concentrations of compound R were not tested. To investigate the effect of compound R on the course of infection in vivo, animals received the drug i.p. at 10 mg per mouse immediately before and 4 h after infection with E. coli J5. They showed significantly prolonged survival compared with mice receiving solvent only (Fig.3a). In a second experiment the survival of treated animals was similarly prolonged significantly by doses of 5 mg per mouse immediately before infection followed by 2 mg per mouse at hourly intervals up to 12 h (Fig. 3b).

Figure 3.

Figure 3

Effect of inhibition of SAP binding on the survival of C57BL/6 mice during bacterial infection. All mice received an LD90 of E. coli J5, 109 organisms injected i.v. (a) The treated group received compound R at 10 mg per mouse by i.p. injection immediately before infection and at 4 h. (b) The treated group received compound R in doses of 5 mg per mouse by i.p. injection immediately before infection and 2 mg per mouse at hourly intervals up to 12 h. In both experiments the control group received injections of solvent alone at the same times. Note that the repeated manipulation and injections involved in_b_ were associated with accelerated morbidity and mortality, but the treated group still showed significantly prolonged survival. P values (Wilcoxon test based on Kaplan–Meier analysis) represent significant differences in survival between treated and control groups.

SAP Reduces Killing of Bacteria in Vivo.

When E. coli J5 organisms were incubated with acute phase serum, either from SAP−/− or SAP+/+ mice, or with SAP−/− serum reconstituted with isolated pure mouse SAP, and then introduced into the peritoneal cavity of SAP−/− animals, there was significantly more immediate killing of the bacteria in the absence of SAP (Fig.4). Binding of SAP evidently enables the organism to evade the host's essential early defenses.

Figure 4.

Figure 4

Effect of SAP on immediate killing of E. coli J5 in SAP−/− mice in vivo. E. coli J5 was preincubated with whole acute phase serum from wild-type SAP+/+ mice, or from SAP−/− animals, or with SAP−/− acute phase serum reconstituted by the addition of isolated mouse SAP, and 107 organisms were then injected i.p. into individual C57BL/6 SAP−/− mice. After 15 min the peritoneal cavity was lavaged, and the numbers of bacteria surviving in each animal were determined. Horizontal lines indicate medians, and the P value (Wilcoxon rank sum test) indicates the significant difference between bacterial killing in the presence or absence of SAP.

SAP Binding Inhibits Phagocytosis of Bacteria by Neutrophils_in Vitro_.

Comparison of in vitro phagocytosis of E. coli J5 by human neutrophils in the presence of whole heat-inactivated serum, in SAP-depleted serum, or in SAP-depleted serum replenished with isolated pure human SAP, showed robustly reproducible inhibition by SAP (Table 2). Indeed, the presence of isolated human SAP alone notably reduced phagocytosis. The presence or absence of SAP had no effect on the phagocytosis of E. coli O111:B4, to which SAP does not bind appreciably (Table 2). These_in vitro_ observations with human cells, serum, and SAP indicate that SAP binding is likely to promote bacterial virulence in humans as it does in the mouse. However, rough E. coli strains are not human pathogens, and E. coli J5 was used only as a convenient model system. In contrast, the binding of SAP to_N. meningitidis_ (Table 1), the pathogen responsible for the serious human infections meningococcal meningitis and septicemia, is clearly of considerable clinical interest. N. meningitidis does not infect mice, and virulence studies of the type used here for_E. coli_ and S. pyogenes are therefore not possible. However, binding of SAP markedly inhibited neutrophil phagocytosis of N. meningitidis in vitro (Table 2). The effect was dose dependent and demonstrable below and within the physiological concentration range of circulating human SAP, 15–40 mg/liter (28).

Table 2.

Effect of human SAP binding on phagocytosis of bacteria by human neutrophils

Bacterium Medium Relative GFP or FITC fluorescence of ethidium positive neutrophils Mean (SD), n = 3
E. coli J5 TCB buffer 100.0 (10.3)
SAP in TCB buffer 22.5 (4.6)
E. coli J5 Whole serum 32.9 (4.2)
CRP-depleted serum 33.1 (0.9)
SAP-depleted serum 100.0 (9.4)
SAP-repleted serum 44.5 (7.1)
E. coli O111:B4 TCB buffer 100.0 (4.6)
SAP in TCB buffer 97.7 (3.3)
E. coli O111:B4 Whole serum 94.4 (8.6)
CRP-depleted serum 98.4 (18.0)
SAP-depleted serum 100.0 (16.2)
SAP-repleted serum 105.2 (17.2)
N. meningitidis H44/76 Whole serum 40.3 (7.7)
SAP-depleted serum 100.0 (6.0)
N. meningitidis H44/76 SAP-depleted serum 100.0 (10.4)
+ SAP 0.1 mg/l 91.9 (19.0)
+ SAP 1.0 mg/l 89.3 (16.0)
+ SAP 10.0 mg/l 62.6 (12.2)
+ SAP 100 mg/l 47.3 (6.5)
+ SAP 1000 mg/l 41.7 (7.4)

Effect of SAP on Lipopolysaccharide Toxicity in Vivo.

Administration to wild-type mice of even a very large amount of isolated human SAP (100 mg/kg) shortly before challenge with a lethal dose of 10 mg/kg of LPS did not confer any protection at all, with the same observed mortality rate as that observed after administration of human serum albumin as a control. In contrast, an equimolar quantity of human CRP significantly reduced morbidity and mortality (Fig.5a), as previously reported in mice transgenic for rabbit CRP. In further experiments in which a second equal dose of human SAP or CRP was given 4 h after injection of LPS, exactly the same result was obtained, with a final percentage of survivors in one experiment of 62.5% with CRP treatment, 25.0% with SAP, and 28.6% with albumin, and in a second, separate experiment, 25% with CRP and 0% with SAP. Furthermore, although SAP−/− mice showed accelerated morbidity after a lethal challenge with 10 mg/kg of LPS, the eventual mortality was the same as in SAP+/+ wild-type animals (Fig. 5b). Importantly, challenge with 1 mg/kg of LPS, which is not clinically toxic for wild-type C57BL/6 mice, similarly caused no morbidity or mortality in SAP−/− mice. Thus, in contrast to the in vitro observations of de Haas et al. (27, 29), there is no evidence that SAP plays a significant role in protection from, or modulation of, LPS toxicity in vivo.

Figure 5.

Figure 5

Effect of SAP deficiency on survival after LPS challenge. (a) BALB/c mice received i.p. injections of 100 mg/kg of isolated pure human SAP, or equimolar amounts of isolated pure human CRP or serum albumin, followed after 20 min by 200 μg per mouse of S. typhimurium LPS, injected i.p. at time 0. (b) SAP−/− or wild-type SAP+/+ C57BL/6 mice received 10 mg/kg of whole LPS from E. coli O111:B4, about 200 μg per mouse, injected i.p. at time 0.

Discussion

The antiopsonic property of SAP is presumably critical for its physiological function of retarding degradation of autologous chromatin in the extracellular environment (9). However, bacteria that express ligands to which SAP binds evidently use this mechanism to their advantage, and this is an example of a normal host protein acting like an antiopsonin, promoting bacterial survival, pathogenicity, and virulence.

In mice the only cellular site of significant in vivo catabolism of SAP is the hepatocyte (30), and the present findings confirm the apparent absence of receptors for SAP on human neutrophils that presumably underlies the antiopsonic effect of bound SAP. In addition to the direct, complement-independent effect of SAP that we show here using isolated SAP in serum-free medium or heat-inactivated sera, ligand binding by SAP may also interfere with opsonization by complement. Artificially aggregated SAP activates complement (8, 31), as do some of the macromolecular and particulate ligands to which SAP binds, including DNA and certain bacteria. However, the complex of SAP with chromatin, its physiological ligand, does not activate complement (8). This absence of activation is presumably important for the normal role of SAP in retarding and modulating chromatin clearance (9), but it has been subverted by bacteria that have evolved ligands for SAP, and, for example, binding of SAP to the rough S. typhimurium Re strain inhibits complement activation by these organisms (32).

As well as binding to bacteria, SAP also binds to influenza virus (33), and we have lately demonstrated that, in comparison with wild-type mice, SAP−/− mice show markedly enhanced survival during lethal influenza virus infection and mount accelerated and increased anti-viral antibody responses (M. Kopf, B. Ecabert, G. Köhler, M.B.P., and M.F. Bachmann, unpublished observations). Thus in addition to its antiopsonic effects, binding of SAP can suppress the immunogenicity of microbial antigens as well as chromatin autoantigens (9), although the duration of the present bacterial infection models was too brief for this mechanism to have been significantly involved.

Regardless of the underlying mechanisms, the enhancement of virulence conferred by bound SAP suggests a possible new therapeutic target in infections caused by SAP-binding pathogens, and we have shown here, in an experimental model system, that in vivo administration of a drug that inhibits SAP binding prolongs survival during a lethal bacterial infection. Compound R is a dicarboxylic acid that must be given by injection, and it is very rapidly excreted. When given by continuous infusion from an osmotic minipump, doses as low as 30 μg per mouse per day are sufficient to completely inhibit and reverse all SAP binding in vivo (M.B.P., unpublished observations). However, implantation of such pumps was not compatible with the established infection models used here. Although the very short course of drug R administered by repeated injection in these experiments did not affect the final number of survivors, the present results (Fig. 3) and the reconstitution of SAP−/− mice with pure SAP (Fig. 2c) clearly confirm the role of SAP in enhancing bacterial virulence. The results with S. pyogenes infection and our in vitro observations with N. meningitidis, human SAP, and human neutrophils (Tables 1 and 2) suggest that this mechanism may well be operative in clinically significant human infections.

In contrast to the effects of SAP binding that are harmful to the host, SAP evidently also contributes to survival during infections with organisms to which it does not bind. We show this contribution here for_E. coli_ O111:B4 and have previously found it with smooth_S. typhimurium_ 12023, an intracellular Gram-negative bacterium to which little or no SAP binds (ref. 14; M. Botto, J. Warren, M. J. Walport, and M.B.P., unpublished observations). We have also observed increased parasitemia in SAP−/− animals compared with wild-type mice after infection with the malarial parasite Plasmodium chabaudii (M. Botto, P. Taylor, M. J. Walport, and M.B.P., unpublished observations). Several mechanisms may underlie the role of SAP in host resistance in these different infections, although our present results do not support the idea that binding to LPS is important. It is not yet clear whether SAP has any specific anti-infective property or whether it may also be beneficial during other inflammatory or tissue-damaging processes. The toxicity and proinflammatory properties of materials released by cell death of both autologous and microbial cells contribute to morbidity and mortality during infection. The critical role of SAP in appropriate handling of chromatin, and perhaps other ligands, exposed or released by apoptosis and necrosis (9), may thus be important. For example, prokaryotic DNA is toxic in mice (34), and the major DNA-binding property of SAP (5) may be protective.

The contrasting roles of SAP, which we report here, in host resistance to infections are a novel example of the interplay between evolution in the host and in the pathogen. The protean and important protective functions of the evolutionarily ancient SAP molecule, with respect to antinuclear autoimmunity and perhaps some pathogens, probably accounts for its high degree of conservation. In contrast, the evolution by some bacteria of ligands to which SAP binds permits evasion of host defenses, causing enhanced virulence. Inhibition of this binding may offer a new approach to therapy.

Acknowledgments

We thank Beth Jones for expert secretarial assistance. This work was supported by the UK Medical Research Council, Program Grant G7900510 (to M.B.P.) and Training Fellowship (to M.N.). M.C.M.B. was supported by a Training Fellowship from the Arthritis Research Council.

Abbreviations

CRP

C-reactive protein

SAP

serum amyloid P component

compound R

(R)-1-[6-(R)-2-carboxy-pyrrolidin-1-yl]-6-oxo-hexanoyl]pyrrolidine-2-carboxylic acid

TCB buffer

10 mM Tris-buffered 140 mM NaCl containing 2 mM CaCl2 and 4% wt/vol BSA at pH 8.0

LPS

lipopolysaccharide

Footnotes

Xia, D., Lin, C., Yun, J., Wagner, T., Magnuson, T. & Samols, D. (1991) FASEB J. 5, 1628 (abstr.).

References