Human C-Reactive Protein Is Protective against Fatal Salmonella enterica Serovar Typhimurium Infection in Transgenic Mice

Abstract

C-reactive protein (CRP) is an acute-phase protein with a well-known association with infection and other inflammatory conditions. We have shown that expression of human CRP by CRP transgenic (CRPtg) mice is protective against lethal infection by Streptococcus pneumoniae, an effect likely mediated by CRP's ability to bind to this gram-positive pathogen. In the present study we tested whether CRPtg mice are resistant to infection with Salmonella enterica serovar Typhimurium, a gram-negative pathogen that causes the murine equivalent of typhoid fever. CRPtg mice experimentally infected with a virulent Typhimurium strain lived longer and had significantly lower mortality than their non-tg littermates. The greater resistance of CRPtg mice could be attributed to significantly increased early (0 to 4 h) blood clearance of salmonellae and significantly decreased numbers of bacteria in the liver and spleen on day 7 postinfection. In addition, 14 days after infection with an avirulent Salmonella strain, the serum titer of anti-Salmonella immunoglobulin G antibodies was higher in CRPtg than non-tg mice. This study provides unequivocal evidence that CRP plays an important role in vivo in host defense against salmonellae during the early stages of infection. In addition, as the beneficial effect of CRP includes enhancement of the host's humoral immune response, CRP may also contribute indirectly to host defense during later stages of infection.

During the acute-phase response, the serum concentration of C-reactive protein (CRP) increases by several hundredfold in humans and rabbits (13), whereas in mice CRP increases only modestly (39). CRP is a pentameric protein exhibiting Ca²⁺-dependent binding specificity for phosphocholine (PCh) (37), phosphoethanolamine (PEt) (31), and certain other ligands (reviewed in reference 1). A variety of activities have been observed in vitro that are consistent with a role of CRP in host defense. For example, CRP binds various pathogens, including bacteria (18) and fungi (28, 29), and promotes their phagocytosis by human leukocytes. CRP is also a potent activator of the classical pathway of complement (17), and therefore it can mediate opsonization of pathogens by complement activation products.

Probably due to the presence of PCh moieties in its cell wall C-polysaccharide, CRP reacts in vivo with the gram-positive bacterium Streptococcus pneumoniae. This reactivity was deduced from bacterial protection studies demonstrating that administration of human CRP increases survival of mice subsequently infected with S. pneumoniae (16, 20, 40). In contrast, protection was not observed after administration of human serum amyloid P-component (SAP) (16), an acute-phase protein in mice but not humans. SAP is structurally similar to CRP and has lectin-like binding specificity for galactose derivatives (14). SAP also binds PEt but not PCh (31). Recently, using CRP transgenic (CRPtg) mice capable of expressing human CRP in an acute-phase manner (9), we confirmed that CRP plays a significant role in vivo in host defense against pneumococcal infections (33). Subsequently we showed that although its protective effect was more pronounced in mice with an intact complement system, CRP offered significant protection even to mice that were decomplemented by cobra venom factor (34).

The gram-negative pathogen Salmonella enterica serovar Typhimurium induces a disease in mice that is a model for human typhoid fever (6, 10, 12, 24). Like all gram-negative bacteria, serovar Typhimurium has phosphatidylethanolamine in its lipid bilayers; however, it does not bind CRP in vitro (21). Thus, based on these in vitro observations, CRP should not be expected to opsonize the bacterium. Nevertheless, in the present study we show that human CRP expressed by transgenic mice is protective against low-dose infection with serovar Typhimurium. These data for the first time extend to gram-negative bacteria previous observations (33, 34) of CRP-mediated protection against pathogens.

MATERIALS AND METHODS

Mice.

We previously described (33) CRPtg C57BL/6J congenic mice. These mice carry a 31-kb ClaI fragment of human genomic DNA comprised of the CRP gene, 17 kb of 5′-flanking sequence, and 11.3 kb of 3′-flanking sequence (9), and they express high levels of human CRP in serum in response to injected endotoxin or after infection with pneumococci (33). C57BL/6J mice are Ity^s, i.e., extremely susceptible to infection with serovar Typhimurium (3, 4, 30), and thus not suitable for infection experiments. In contrast, DBA/2J mice carry the Ity^r allele (3, 15). To generate CRPtg mice resistant to serovar Typhimurium, we crossed female C57BL/6J CRPtg mice with DBA/2J males (Charles River Laboratories, Boston, Mass.) to produce CRPtg and non-tg F₁ hybrids. F₁s were backcrossed to DBA/2J to generate F₂s. Since the Ity^r allele is dominant, all F₂ mice have the Salmonella-resistant phenotype. F₂ mice were screened for the presence of the CRP transgene using a previously described PCR method (23). In addition, since DBA/2J mice are C5 deficient (C5D), F₂ mice were also screened for inheritance of the C5 mutant allele by PCR (38). Finally, due to sexual dimorphism of CRP transgene expression, which results in higher levels of serum CRP in males (33, 35), only male mice were used. Mice were housed in groups of four, fed and watered ad libitum, maintained according to protocols established by the Animal Resources Program at this institution, and 10 to 12 weeks old when used in experiments. Non-tg littermates served as controls.

Bacteria.

Wild-type (virulent) serovar Typhimurium strain LT2L (rpoS⁺) (32) and attenuated (avirulent) recombinant serovar Typhimurium strain BRD-846 (8) were stored as stock cultures at −70°C in Todd-Hewitt broth supplemented with 0.5% yeast extract (Difco Laboratories, Detroit, Mich.). BRD-846 was derived from an aroA aroD live oral vaccine strain (8). The Typhimurium strains used for infecting mice were collected by centrifugation from stationary-phase broth cultures (grown overnight at 37°C) and washed and resuspended in Ringer's lactate solution at 4°C. Concentrations of bacteria were estimated from absorbance at 420 nm (A₄₂₀ of 1 = 2 × 10⁸ bacteria/ml). Inocula were kept on ice, and mice were infected intravenously (i.v.) or orally within 5 min of diluting the bacteria. The density and viability of bacteria were confirmed by plating on trypticase soy agar (TSA) (Difco).

Blood clearance and survival studies.

Groups of CRPtg and non-tg littermates were infected by i.v. injection of 2 × 10¹ to 1 × 10⁶ CFU of strain LT2L suspended in 200 μl of lactate solution. Blood (50 μl) was collected from the retroorbital sinus at 1, 10, 30, and 240 min after infection for analysis of early blood clearance of bacteria. Deaths of mice were recorded at 24-h intervals for a 120-day period after infection.

Quantitation of viable salmonellae in the blood and organs of mice.

To determine bacteremia, 50 μl of blood was serially diluted in Ringer's, and aliquots of each dilution were seeded onto TSA plates. For determination of bacterial load in organs, mice at various stages of infection were killed by cervical dislocation, and the spleen and liver were removed. Each organ was homogenized in 10 ml of ice-cold hypotonic (0.01 M, pH 7.2) phosphate buffer to minimize bacterial multiplication. Aliquots of serially diluted spleen and liver homogenate were seeded onto TSA plates and incubated at 37°C for 24 h before bacterial colonies were counted.

Measurement of human CRP and mouse SAP.

Serum human CRP and mouse SAP were measured by enzyme-linked immunosorbent assay (ELISA) as described (35). The lower limits of detection of human CRP and mouse SAP were 20 ng/ml and 25 μg/ml, respectively.

Immunization and determination of serum antibodies.

The humoral response to salmonellae in mice parallels that in human typhoid fever and is directed against lipopolysaccharide and a number of other undefined antigens (6, 24). Infection of mice with attenuated Typhimurium strains elicits a protective immune response against virulent strains (30), with the advantage that the host survives infection. Therefore, we used the avirulent strain BRD-846 (50% lethal dose [LD₅₀], >10¹⁰) as a live vaccine. For immunizations, mice received by oral gavage 7 × 10⁹ CFU of salmonellae, and blood was collected 14 days later. Serum anti-Salmonella antibody titers were determined by ELISA as previously described (22) using microtiter plates coated with whole-cell lysates of the virulent strain LT2L. Goat anti-mouse immunoglobulin G (IgG)-biotin or goat anti-mouse IgG2a-biotin followed by avidin-peroxidase (all from Bio-Rad Laboratories, Richmond, Calif.) and ABTS [2,2′-azinobis(3-ethylbenzthiazolinesulfonic acid)] were used to develop the plates. The reported titer of each specimen is the reciprocal of the serum dilution giving five times higher absorbance than undiluted (pooled) preimmune serum.

Statistical analyses.

The data presented are from analyses of results pooled from three separate survival experiments, two bacterial clearance experiments, four organ localization experiments, and two immunization experiments. All values are given as mean ± standard error of the mean. Bacteremia data are presented as means of untransformed counts. Data on the numbers of viable salmonellae in the liver and spleen are presented as means of log-transformed counts (log₁₀ CFU). Median survival time (MST) was estimated by interpolation using survival curves. The Mann-Whitney two-sample rank test was used to evaluate differences in length of survival among groups of mice. The χ² test was used to determine if differences between the percentages of tg and control mice surviving infection were significant. Student's t tests were used for comparisons of pre- and postinjection levels of serum CRP and SAP, CRPtg and non-tg antibody titers, and CFU per organ. All procedures used for data analyses were applied using Statview 512+ (Brain Power, Inc., Calabasas, Calif.) software. A P value of less than 0.05 was considered significant in all statistical tests.

RESULTS

Survival of infected mice.

C57BL/6J mice are highly susceptible to infection with S. enterica serovar Typhimurium (15, 20, 21, 30), primarily because they are homozygous for the Ity^s allele at the Ity (immunity to Typhimurium) locus on chromosome 1 (also known as Bcg and Lsh) (26, 27). In contrast, DBA/2J mice carry the dominant Ity^r allele (3, 15). The primary effect of the Ity locus and the Nramp 1 (natural resistance-associated protein 1) gene it contains (36) on resistance to salmonellae is the regulation of growth within relatively nonbactericidal sites in the liver and spleen. To generate CRPtg and non-tg mice with sufficient resistance to salmonellae to allow comparison of responses to graded doses of virulent bacteria, we crossed C57BL/6J-CRPtg mice with DBA/2J mice. Since the Ity^r allele is dominant, all F₂ mice used in the reported experiments had the Salmonella serovar Typhimurium-resistant phenotype. DBA/2J mice are C5D due to a spontaneous mutation in exon 7 of the murine C5 gene (38). CRPtg versus non-tg and C5D versus C5-sufficient F₂ progeny were obtained in the expected Mendelian ratios. All CRPtg and non-tg mice used for infection experiments were male and C5D.

Survival curves for CRPtg and non-tg mice infected with 20 CFU of serovar Typhimurium are shown in Fig. 1. MST for controls was 36 days, and none survived beyond day 72. By comparison, CRPtg mice had increased longevity (MST = 52 days) and their survival rate (50%) was significantly increased (P < 0.05, χ² test). To determine the extent of CRP-mediated protection relative to control animals, we repeated the experiment using mice infected with higher doses. CRP-mediated protection was still observed after injection of 700 CFU per mouse, i.e., two of eight tg mice compared to zero of eight control mice survived infection, and the difference in MST (37 versus 24 days) persisted (Fig. 1, inset). All mice infected with 6 × 10⁵ CFU died, and there was no appreciable difference between the MSTs of control and CRPtg mice (n = 6 each) (Fig. 1, inset).

FIG. 1.

MST and survival rate of mice infected with S. enterica serovar Typhimurium LT2L. CRPtg and littermate non-tg mice (6 mice per group) received 20 CFU of salmonellae i.v. on day 0, and deaths were recorded for 120 days postinfection. Compared to controls, MST of CRPtg mice was more than 2 weeks longer (MST = 52 versus 36 days) and survival rate was significantly increased (P < 0.05, χ² test). (Inset) MST of CRPtg and non-tg mice using increasing doses of salmonellae (six to eight mice per point).

Early blood clearance and dissemination of injected salmonellae.

After i.v. injection, most salmonellae are rapidly killed by serum- and cell-mediated bactericidal mechanisms (10, 12). A few bacteria survive internalization by circulating phagocytic cells, and many of these are transported to the liver and spleen, where they take up residence and multiply (10, 12). Thus, the observed CRP-mediated protection against serovar Typhimurium may have involved enhancement of blood bactericidal activity, reduced transport of intracellular bacteria to liver and spleen, or both.

To determine which of these mechanisms is responsible for CRP-mediated protection against serovar Typhimurium, we compared bacteremia in control and CRPtg mice at different times up to 4 h after i.v. challenge, and quantitated the number of viable organisms delivered to the liver and spleen at 4 h. Serum CRP and SAP levels were monitored. As shown in Fig. 2, blood clearance of salmonellae was more efficient in CRPtg than in non-tg mice, i.e., at 4 h CRPtg mouse blood contained 12-fold fewer viable bacteria than blood from controls (P < 0.05, t test). Also, at 4 h the livers and spleens of CRPtg mice harbored two- to threefold fewer bacteria than controls (Fig. 2, inset). Since fewer bacteria are recovered from the spleen and liver of CRPtg than control mice, the increased clearance of salmonellae from the blood of CRPtg mice cannot be attributed to accelerated transport of bacteria to these organs. Rather, the data are consistent with the notion that CRP enhances killing of the blood-borne pathogens. Within 1 min of bacterial challenge and coincident with severe bacteremia, CRP serum levels were lowered significantly below basal values (27 ± 1 μg/ml) and remained significantly lowered for at least 30 min postinfection (P < 0.05, t tests) (Fig. 3). By 4 h, when the majority of bacteria were cleared from the circulation, serum CRP rebounded to approximately preinfection levels. SAP was also transiently, although not significantly, reduced compared to baseline values (360 ± 90 μg/ml) during the initial 30 min after infection, and its levels at 4 h were significantly elevated (P < 0.05, t test). Although a cause-effect relationship may not exist between clearance of bacteria (Fig. 2) and lowering of serum levels of CRP and SAP (Fig. 3), the correlation is consistent with the hypothesis that the two acute-phase proteins are removed from the fluid phase because they bind to the bacteria and mediate their phagocytosis and clearance from the blood.

FIG. 2.

Blood clearance of salmonellae in infected mice. CRPtg and non-tg mice (three per group) were infected with 10⁶ CFU of strain LT2L. At the indicated times, blood was withdrawn and bacteremia was quantitated. The asterisk indicates a significant difference (P < 0.05, t test) between CRPtg and non-tg mice. (Inset) Numbers of viable salmonellae recovered from the livers and spleens of mice 4 h postinfection.

FIG. 3.

Serum acute-phase proteins in mice injected with 10⁶ CFU of strain LT2L. The concentration of human CRP and mouse SAP was quantitated by ELISA in serum collected during the blood clearance experiment shown in Fig. 2. The asterisks indicate concentrations significantly lower or higher than basal levels (P < 0.05, t tests).

Persistence of salmonellae in spleen and liver.

The difference between CRPtg and non-tg controls in numbers of splenic and hepatic bacteria ascertained at 4 h postinfection (Fig. 2, inset) persisted and in fact was more pronounced on day 7, even when higher doses of bacteria were used for infection (Fig. 4). Depending on the dose of salmonellae injected, the spleens and livers of transgenic mice harbored 6- to 14-fold and 12- to 20-fold fewer bacteria, respectively, than the spleens and livers of controls.

FIG. 4.

Recovery of salmonellae from spleens and livers of infected mice. CRPtg and non-tg mice were infected with 10³ or 10⁵ CFU of strain LT2L (six mice per group per dose). One week later, the liver and spleen were harvested from each mouse and homogenized, and serial dilutions of the homogenates were used to seed culture plates. Asterisks indicate significant differences between the indicated groups (P < 0.05, t tests).

Serum antibody response.

Since humoral immunity is known to play a significant role in protection of mice from serovar Typhimurium (10, 19), we tested if CRP-mediated protection might also involve enhancement of the immune response. As shown in Fig. 5, 14 days after infection with strain BRD-846, CRPtg mice had higher levels of serum antibody, i.e., their anti-Salmonella total IgG titer was on average 11-fold higher than that of non-tg mice and their IgG2a antibody titer was 6-fold higher (P < 0.025, t test).

FIG. 5.

Serum anti-Salmonella responses of CRPtg and control mice. CRPtg and non-tg mice (eight of each) were infected per os with 7 × 10⁹ CFU of avirulent strain BRD-847. Fourteen days later, titers of total IgG and IgG2a anti-Salmonella (whole cell) serum antibodies were determined by ELISA. The asterisk indicates a significant difference between the two groups (P < 0.025, t test).

DISCUSSION

The most significant finding of the present study is that CRPtg mice expressing human CRP exhibit increased resistance to fatal infection with low doses of S. enterica serovar Typhimurium compared to non-tg littermates. The data clearly show that expression of human CRP increases early clearance of i.v. injected bacteria from the blood and reduces dissemination of bacteria to the liver and spleen, allowing CRPtg mice to live longer and survive infection at a higher rate than similarly infected non-tg mice. Furthermore, the specific serum anti-Salmonella antibody response is enhanced in CRPtg mice, which likely also contributes to the observed protective effect. Thus, the host defense function of CRP is not limited to pneumococci, as generally believed, but appears to be wider, extending to at least one gram-negative bacterial pathogen.

In contrast to the present results, demonstrating clearly that endogenously synthesized CRP plays a role in protection of mice from infection with virulent salmonellae, it was reported earlier that passive administration of human CRP failed to protect BALB/c mice from infection with salmonellae (21). We have no direct proof but can offer several explanations for the difference between the results of these two studies. First, BALB/c mice carry the Ity^s genotype and thus are predicted to be much more susceptible to Salmonella infection (3, 15, 30) than the Ity^r hybrids that we employed. Second, in the earlier study, only a single dose of bacteria (2 × 10⁶ CFU) was used (21), and this was twofold higher than the highest dose that we used. At such overwhelming doses, we also failed to observe differences in survival between CRPtg and control mice. Third, in the earlier experiments, mice received only a single injection of human CRP 30 min prior to infection (21). It was ascertained that only 10% of the injected CRP remained in the blood of mice 6 h later and only 2% remained at 24 h, whereas in our experiments the CRP transgene is expressed continuously. The use of a susceptible strain of mice combined with a very high dose of bacteria likely overwhelmed any protective advantage gained by a single injection of CRP.

Despite subsequent reports (2, 25) suggesting that salmonellae express PCh on the surface, binding of CRP to salmonellae could not be detected by using a sensitive radioimmunoassay (21). We also could not detect binding of CRP to salmonellae or to purified Salmonella lipopolysaccharide, using ELISA-based assays (data not shown). Despite these in vitro results, the most likely mechanism for the observed CRPtg-mediated protection is opsonization by human CRP followed by intracellular killing. CRP could opsonize bacteria in vivo directly or indirectly via complement activation products. The observed transient drop in CRP serum concentration immediately after injection of the bacteria (Fig. 3) is certainly consistent with binding of CRP to the bacteria. If that were the case, a likely ligand for CRP would be the PEt polar head group of phosphatidylethanolamine, the main phospholipid of gram-negative bacterial membranes. Questions about access to PEt in the lipid bilayer and about differences between in vitro and in vivo results cannot be answered on the basis of existing information. Nevertheless, it should be pointed out that similar questions can be raised about access of CRP and antiphosphocholine antibodies to PCh residues of the pneumococcal cell wall teichoic acid, particularly since, like serovar Typhimurium, S. pneumoniae does not bind CRP in vitro. Nevertheless, both CRP and anti-PCh antibodies have been shown convincingly to protect mice from experimental pneumococcal infection (4, 5, 16, 40).

A possible role for SAP in the observed protection of mice from Salmonella infection should be entertained. SAP binds PEt with higher avidity than does CRP (31), and similarly to CRP, its serum levels dropped immediately after administration of salmonellae (Fig. 3), suggesting that SAP also bound to the bacteria. Although SAP has not been reported to have opsonic properties, it has been shown to activate complement (13a) and could conceivably mediate opsonization of bacteria by complement activation products. However, there should be no difference between CRPtg and non-tg mice in terms of SAP levels and function, and therefore the observed differences can be attributed solely to CRP.

Our combined data strongly suggest that the protective action of CRP is realized mainly by limiting dissemination of salmonellae during the initial stage of infection. Antibodies directed against salmonellae, which are known to represent a major defense element in mice and humans (10, 19), are probably most effective during later stages of infection, when the antibody concentration in blood reaches protective levels. In this respect, the heightened immune responsiveness of CRPtg mice can be considered a distal CRP protective effect. Like complement (7, 11), CRP might play a role in discriminating among pathogens and enhancing host immune responses to their antigens. This possibility can be further explored by the use of the tg animals used in this study.