The Phosphocholine-binding Pocket on C-reactive Protein Is Necessary for Initial Protection of Mice against Pneumococcal Infection*

Background: CRP exerts antipneumococcal function in mice if administered during the early stages of pneumococcal infection. Results: CRP loses such antipneumococcal function when its phosphocholine-binding pocket is blocked. Conclusion: The phosphocholine-binding pocket on CRP participates in antipneumococcal function of CRP during the early stages of infection. Significance: Remodeling of CRP may be required for successful treatment of mice during the late stages of infection. Human C-reactive protein (CRP) protects mice from lethal Streptococcus pneumoniae infection when injected into mice within the range of 6 h before to 2 h after the administration of pneumococci. Because CRP binds to phosphocholine-containing substances and subsequently activates the complement system, it has been proposed that the antipneumococcal function of CRP requires the binding of CRP to phosphocholine moieties present in pneumococcal cell wall C-polysaccharide. To test this proposal experimentally, in this study, we utilized a new CRP mutant incapable of binding to phosphocholine. Based on the structure of CRP-phosphocholine complexes, which showed that Phe66, Thr76, and Glu81 formed the phosphocholine-binding pocket, we constructed a CRP mutant F66A/T76Y/E81A in which the pocket was blocked by substituting Tyr for Thr76. When compared with wild-type CRP, mutant CRP bound more avidly to phosphoethanolamine and could be purified by affinity chromatography using phosphoethanolamine-conjugated Sepharose. Mutant CRP did not bind to phosphocholine, C-polysaccharide, or pneumococci. Mutant CRP was free in the mouse serum, and its rate of clearance in vivo was not faster than that of wild-type CRP. When either 25 μg or 150 μg of CRP was administered into mice, unlike wild-type CRP, mutant CRP did not protect mice from lethal pneumococcal infection. Mice injected with mutant CRP had higher mortality rates than mice that received wild-type CRP. Decreased survival was due to the increased bacteremia in mice treated with mutant CRP. We conclude that the phosphocholine-binding pocket on CRP is necessary for CRP-mediated initial protection of mice against lethal pneumococcal infection.

increased several hundredfold in response to pneumococcal infection (2). However, the functions of CRP in pneumococcal infection are not known (3). In mice, CRP is only a trace serum component and is not an acute phase protein (23). Therefore mice are used to explore the in vivo functions of human CRP. In mouse models of infection, passively administered human CRP has been shown to be protective against lethal pneumococcal infection, as determined by increased survival of and decreased bacteremia in the infected mice (24,25). Interestingly, CRP was most effective in protecting mice from infection only when injected within the range of 6 h before to 2 h after administering pneumococci into mice (26). The protective function of CRP was not observed when mice received CRP 24 or 36 h after infection (7,26). Thus, the CRP-mediated protection of mice requires the presence of CRP in the early stages of infection. Mice transgenic for human CRP were also protected from lethal pneumococcal infection and showed both decreased bacteremia and increased survival (27).
Because PnC-complexed CRP activates the complement system in both human and mouse sera (3,28), it has been proposed that CRP is protective through a pathway in which CRP binds to pneumococci through PCh groups present on their surfaces, the pathogen-bound CRP activates the complement system, and bacteremia is then reduced through complement-dependent opsonophagocytosis (29 -31). The aim of this study was to determine whether the binding of CRP to PCh on pneumococci was required for the protection of mice against pneumococcal infection. Employing site-directed mutagenesis, we generated a novel triple mutant of CRP, F66A/T76Y/E81A, incapable of binding to PCh, and used mutant CRP in mouse protection experiments. We hypothesized that if the binding of CRP to PCh was required for the protection of mice against pneumococcal infection, then mutant CRP should not be protective. In a similar previously published study, another CRP mutant, F66A/E81A, incapable of binding to PCh was used (7).

EXPERIMENTAL PROCEDURES
Construction and Expression of the CRP Triple Mutant F66A/T76Y/E81A-The construction of the double mutant F66A/E81A CRP cDNA has been described earlier (13). F66A/ E81A CRP cDNA was used as the template for construction of the triple mutant F66A/T76Y/E81A cDNA (substitution of Phe 66 and Glu 81 with Ala and of Thr 76 with Tyr). Mutagenic oligonucleotides, 5Ј-GGATACAGTTTTTACGTGGGTGG-GTCTG-3Ј and 5Ј-CAGACCCACCCACGTAAAAACTGTA-TCC-3Ј, which substitute Tyr for Thr 76 (codons shown in bold and italicized letters), were designed according to the sequence of F66A/E81A cDNA template and obtained from Integrated DNA Technologies. Mutagenesis was conducted using the QuikChange site-directed mutagenesis kit (Stratagene). Mutations were verified by nucleotide sequencing, utilizing the services of the Molecular Biology Core Facility of our university. Stable transfection of F66A/T76Y/E81A CRP cDNA was carried out in CHO cells as described previously (7). A CHO cell line expressing the F66A/T76Y/E81A mutant CRP was isolated by a series of subcloning steps.
Preparation of PEt-conjugated Sepharose-PEt-conjugated Sepharose was prepared as described previously (32). In brief, 25 ml of packed ECH-Sepharose 4B beads (GE Healthcare) was first washed with water (pH 4.5) and subsequently with 0.5 M NaCl. Then, 180 mg of PEt (catalog number P0503, Sigma-Aldrich) was dissolved in 25 ml water (pH 4.5) and added to the washed beads. 500 mg of N-(3-dimethylaminopropyl)-NЈ-ethylcarbodiimide hydrochloride (catalog number E6383, Sigma-Aldrich) was then added, and the mixture was stirred for 1 h at room temperature. After monitoring the pH for 1 h to ensure that the pH stayed at 4.5, the mixture was left overnight at 4°C with slow stirring. The beads were then washed with 0.1 M acetate buffer, pH 4.0, followed by washing with 100 mM Tris/HCl, pH 8.0, containing 0.5 M NaCl. Washing was repeated three times alternating between the acetate and Tris buffers. Finally, the beads were washed with water and then with TBS containing 2 mM CaCl 2 .
Purification of Native WT CRP-WT CRP was purified from discarded human pleural fluid by affinity chromatography on a PCh-Sepharose column (Pierce) followed by ion-exchange chromatography on a MonoQ column (GE Healthcare) and gel filtration chromatography on a Superose 12 column (GE Healthcare), as described previously (33), and stored frozen. On The PDB coordinates of mutant CRP were generated from the PDB file 1B09 using SYBYL (Tripos, Inc.). The side chains of Phe 66 , Thr 76 and Glu 81 are substituted with Ala, Tyr, and Ala, respectively. One of the five subunits is shown. The figures were rendered using PyMOL (66). the day of the experiments, CRP was repurified by gel filtration chromatography on a Superose 12 column to remove any form of modified CRP that might have been generated due to storage of CRP. Repurified CRP was stored in TBS containing 2 mM CaCl 2 at 4°C and was used within a week. The purity of CRP was confirmed by using denaturing SDS-PAGE.
Purification of Mutant CRP-Purification of mutant CRP from the cell culture supernatants involved two steps: Ca 2ϩ -dependent affinity chromatography on a PEt-conjugated Sepharose column followed by gel filtration chromatography on a Superose 12 column. For affinity chromatography, the culture medium containing CRP was diluted (1:1) in 0.1 M borate buffered saline, pH 8.3, containing 3 mM CaCl 2 and passed through the PEt-conjugated Sepharose column. After collecting the flow-through and washing the column with the same buffer, bound CRP was eluted with 0.1 M borate-buffered saline, pH 8.3, containing 5 mM EDTA. Eluted CRP was concentrated and further purified by gel filtration chromatography on a Superose 12 column. Gel filtration chromatography was carried out as described for WT CRP, except that the column was equilibrated and eluted with TBS containing 5 mM EDTA. It was necessary that the gel filtration chromatography of the mutant CRP be performed in the presence of EDTA because in the presence of Ca 2ϩ , the mutant CRP bound to Superose beads (data not shown). Eluted CRP was immediately dialyzed against TBS containing 2 mM CaCl 2 , stored at 4°C, and used within a week. The purity of CRP was confirmed by using denaturing SDS-PAGE.
The concentration of purified WT and mutant CRP was determined by measuring absorbance at 280 nm. The extinction coefficient value of 19.5 at 280 nm for 1% CRP solution was used for calculations. For mouse protection experiments, both purified WT and mutant CRP were treated with the Detoxi-Gel endotoxin removing gel according to the manufacturer's instructions (Thermo). The concentration of endotoxin in CRP was determined by using the QCL-1000 limulus amebocyte lysate kit according to the manufacturer's instructions (Lonza).
PCh Binding Assay-Binding activity of CRP for PCh was evaluated by using PCh-conjugated BSA and PnC (Statens Serum Institut) as the ligands, as described previously (7,33). Microtiter wells (96-well plates) were coated with 10 g/ml PCh-BSA or PnC in TBS, overnight at 4°C. The unreacted sites in the wells were blocked with TBS containing 0.5% gelatin for 45 min at room temperature. CRP diluted in TBS containing 5 mM CaCl 2 , 0.1% gelatin, and 0.02% Tween 20 (TBS-Ca) was added in duplicate wells. After incubating the plates for 2 h at 37°C, the wells were washed with TBS-Ca. The assays were performed in duplicate plates. In one plate, anti-CRP monoclonal antibody (mAb) HD2.4, diluted in TBS-Ca, was used (1 h at 37°C) to detect bound CRP. In the other plate, rabbit polyclonal anti-CRP antibody (Sigma-Aldrich), diluted in TBS-Ca, was used (1 h at 37°C) to detect bound CRP. HRP-conjugated goat anti-mouse IgG and HRP-conjugated donkey anti-rabbit IgG (Thermo), diluted in TBS-Ca, were used (1 h at 37°C) as secondary antibodies. Color was developed, and the A 405 was read in a microtiter plate reader (Molecular Devices).
Anti-CRP mAb Binding Assay-The anti-CRP mAb HD2.4 and EA4.1 binding assays were performed as described previously (13). Microtiter wells (96-well plates) were coated with 10 g/ml anti-CRP mAb HD2.4 (34,35) or EA4.1 (34) in TBS, overnight at 4°C. The unreacted sites in the wells were blocked with TBS containing 0.5% gelatin for 45 min at room temperature. CRP diluted in TBS-Ca was added in duplicate wells. After incubating the plates for 2 h at 37°C, the wells were washed with TBS-Ca. Rabbit polyclonal anti-CRP antibody (Sigma-Aldrich), diluted in TBS-Ca, was used (1 h at 37°C) to detect bound CRP. HRP-conjugated donkey anti-rabbit IgG (Thermo), diluted in TBS-Ca, was used (1 h at 37°C) as secondary antibody. Color was developed, and the A 405 was read in a microtiter plate reader.
Pneumococci-S. pneumoniae type 3, strain WU2, were made virulent by sequential intravenous passages in mice and were stored in aliquots at Ϫ80°C in Todd-Hewitt broth containing 0.5% yeast extract and 10% glycerol, as described previously (7,28). For each experiment, a separate aliquot of frozen pneumococci was thawed. Pneumococci were then grown in Todd-Hewitt broth containing 0.5% yeast extract and collected from mid-log phase cultures. Pneumococci were washed and resuspended in normal saline (A 600 ϭ 0.35 ϭ 2.5 ϫ 10 8 cfu/ml). The concentration, purity, and viability of pneumococci were confirmed by plating on sheep blood agar plates.
Pneumococci Binding Assay-Microtiter wells (96-well plates) were coated with pneumococci in TBS (10 7 cfu/100 l/well) overnight at 4°C. The unreacted sites in the wells were blocked with TBS containing 0.5% gelatin for 45 min at room temperature. CRP diluted in TBS-Ca was added in duplicate wells. After incubating the plates for 2 h at 37°C, the wells were washed with TBS-Ca. The assays were performed in duplicate plates. The plates were then processed exactly as described for the PCh binding assay.
The binding of CRP to pneumococci in the fluid phase was investigated as follows. Pneumococci (2 ϫ 10 7 cfu; final concentration 10 8 cfu/ml) were incubated with CRP (10 g; final concentration 50 g/ml) in TBS containing 2 mM CaCl 2 and 0.02% Tween 20 at 37°C in a shaking water bath. After 30 min, pneumococci were pelleted, washed three times with the same buffer, resuspended in TBS, and subjected to denaturing SDS-PAGE.
Microtiter wells (96-well plates) were coated with 10 g/ml avidin (catalog number A9275, Sigma-Aldrich) in TBS for 2 h at 37°C. The unreacted sites in the wells were blocked with TBS containing 0.5% gelatin for 45 min at room temperature. After washing the wells with TBS, biotinylated PEt diluted in TBS (10 g/ml) was added to the wells for 2 h at 37°C. After washing the wells with TBS-Ca, CRP diluted in TBS-Ca was added in dupli-cate wells. After incubating the plates overnight at 4°C, the wells were washed with TBS-Ca. The assays were performed in duplicate plates. The plates were then processed exactly as described for the PCh binding assay.
Streptavidin could not be used to capture biotinylated PEt on the wells because in preliminary experiments, CRP was found to bind to streptavidin (data not shown). This was not the case with avidin coating (data not shown).
Mice-Male C57BL/6J mice (Jackson ImmunoResearch Laboratories) were brought up and maintained according to protocols approved by the University Committee on Animal Care. Mice were 8 -10 weeks old when used in experiments.
Mouse Protection Experiments-Two separate mouse protection experiments were performed using two batches of purified WT and mutant CRP. Mice were first injected intravenously with either 25 g or 150 g of WT or mutant CRP in 150 l of TBS containing 2 mM CaCl 2 . The endotoxin content in 25 and 150 g of WT CRP was 0.18 Ϯ 0.09 and 1.08 Ϯ 0.52 endotoxin units, respectively. The endotoxin content in 25 and 150 g of mutant CRP was 0.16 Ϯ 0.09 and 0.93 Ϯ 0.54 endotoxin units, respectively. After 30 min, mice were injected intravenously with 5 ϫ 10 7 cfu (based on A 600 ) of pneumococci in 100 l of saline. The actual number of pneumococci injected, based on the plating results obtained on the next day, was 5.15 Ϯ 0.13 ϫ 10 7 cfu. Survival of mice was recorded three times per day for 7 days. Survival curves were generated using the GraphPad Prism 4 software. To determine p values for the differences in the survival curves among various groups, the survival curves were compared using the software's log rank test.
To determine bacteremia (cfu/ml) in the surviving mice, blood was collected daily for 5 days from the tip of the tail vein, diluted in normal saline, and plated on blood agar plates for colony counting. Bacteremia values for dead mice were taken as Ͼ10 8 cfu/ml because mice died when the bacteremia exceeded 10 8 cfu/ml. The plotting and statistical analyses of the bacteremia data were done using the GraphPad Prism 4 software and Mann-Whitney two-tailed t test.
Clearance of CRP from Mouse Circulation-Mice were injected intravenously with 100 g of CRP in TBS containing 2 mM CaCl 2 through the tail. Blood was collected from the tip of the tail vein at various times up to 30 h. The concentration of CRP in the serum was measured by ELISA (13). The statistical analyses of the data were performed using the Mann-Whitney two-tailed t test.
Repurification of Mutant CRP from Purified Mutant CRPspiked Mouse Serum-Purified F66A/T76Y/E81A mutant CRP (400 g) was added to 2 ml of C57BL/6 mouse serum (Innovative Research), and the final volume was made up to 10 ml by adding 0.1 M borate-buffered saline, pH 8.3, containing 3 mM CaCl 2 . The mixture was incubated for 30 min at 37°C. Mutant CRP was repurified by Ca 2ϩ -dependent affinity chromatography using PEt-conjugated Sepharose beads whose capacity to bind mutant CRP was Ͼ400 g, as described above. After collecting the flow-through and washing the column with the same buffer, bound CRP was eluted with 0.1 M borate-buffered saline, pH 8.3, containing 5 mM EDTA. To control the experiment, mouse serum alone (2 ml), without spiking with purified mutant CRP, was used. The EDTA eluates were subjected to denaturing SDS-PAGE. The concentration of CRP in the EDTA eluates was measured by ELISA to calculate the percentage of recovery.

RESULTS
Characterization of F66A/T76Y/E81A Mutant CRP-Mutant CRP cDNA was successfully expressed in CHO cells and could be purified by PEt affinity chromatography followed by gel filtration chromatography. The elution profiles of WT CRP and mutant CRP from the gel filtration column were almost overlapping; the elution volume of the mutant CRP was only 250 l less than that of WT CRP ( Fig. 2A). Denaturing SDS-PAGE analysis of purified WT CRP and mutant CRP showed single bands, and the molecular mass of the subunits of mutant CRP was the same as that of WT CRP (Fig. 2B). We also evaluated the epitope for anti-CRP mAb HD2.4, which is a pentameric CRP-specific antibody and whose epitope is located on the face opposite to the PCh-binding face of the CRP pentamer (34,36). As shown in Fig. 2C, both WT and mutant CRP recognized mAb HD2.4 equally well. These data demonstrated that the substitution of Phe 66 , Thr 76 , and Glu 81 with Ala, Tyr, and Ala, respectively, did not affect the overall structure of CRP and that mutant CRP was pentameric. The PCh binding activity of mutant CRP was assessed by using two different PCh-containing ligands, PCh-BSA and PnC. WT CRP bound to both ligands in a CRP concentrationdependent manner, but mutant CRP bound neither to PCh-BSA (Fig. 3A) nor to PnC (Fig. 3B). We also characterized the PCh-binding site using anti-CRP mAb EA4.1. The binding of EA4.1 mAb to CRP is Ca 2ϩ -dependent and can be inhibited by PCh, indicating that EA4.1 binds at or near the PCh-binding site (34). As shown in Fig. 3C, the mutations decreased the binding of CRP to EA4.1 by ϳ99%; for equivalent binding to EA4.1, 100 ng/ml mutant CRP was required when compared with 1 ng/ml WT CRP, indicating that the EA4.1-binding epitope, and hence the PCh-binding site, on mutant CRP was lost. We also determined the binding of mutant CRP to whole pneumococci that were used in the mouse protection experiments. In the solid phase pneumococci binding assay, mutant CRP did not bind to whole pneumococci (Fig. 4A). Mutant CRP also did not bind to pneumococci in the fluid phase binding assay (Fig. 4B, compare lanes 4 and 5). Thus, F66A/T76Y/E81A mutant CRP binds neither PCh nor pneumococci.
Because the substitution of Tyr for Thr 76 was based on the structure of SAP, and SAP binds to PEt (8 -10), we next evaluated the effects of the mutations on the binding of CRP to PEt. Both WT CRP and mutant CRP bound to PEt in a CRP concentration-dependent manner (Fig. 5). However, mutant CRP was much more efficient than WT CRP in binding to PEt. For equivalent binding to PEt, 10 g/ml WT CRP was required when compared with 170 ng/ml mutant CRP. By repeating the PEt binding assays four times, we found that ϳ98% less mutant CRP was required when compared with WT CRP for equivalent binding to PEt. Thus, F66A/T76Y/E81A mutant CRP binds to PEt more avidly than WT CRP does.
In the ligand binding assays (Figs. 3, A and B, 4A, and 5), similar results were obtained irrespective of which anti-CRP antibody, the anti-CRP mAb HD2.4 (data shown) or polyclonal anti-CRP antibody (data not shown), was used to detect ligandbound CRP. Because mAb HD2.4 is a pentameric CRP-specific antibody (34,36), the ligand binding data further suggested that the overall structure of mutant CRP was not different from that of WT CRP.
F66A/T76Y/E81A Mutant CRP Does Not Protect Mice from Pneumococcal Infection- Fig. 6A shows the combined results of two separate protection experiments using 25 g of CRP and six mice in each group in each experiment. The median survival time (the time taken for the death of 50% of mice) for mice injected with bacteria alone (control group A) was 40 h. The median survival time for mice injected with bacteria and WT CRP (group B) was 88 h. The median survival time for mice injected with bacteria and mutant CRP (group C) was 44 h. In the WT CRP-treated group, no deaths occurred in 44 h, and 25% of mice survived up to 7 days. By the end of the 3rd day, all mice had died in the mutant CRP-treated group, whereas the survival was 58% in the WT CRP-treated group. Fig. 6B shows the combined results of two separate protection experiments . The unreacted sites in the wells were blocked with gelatin. Purified CRP diluted in TBS-Ca was then added to the wells. Bound CRP was detected by using anti-CRP mAb HD2.4 and HRP-conjugated goat anti-mouse IgG. Color was developed, and the absorbance was read at 405 nm. A representative of four experiments is shown. Similar results were obtained when polyclonal anti-CRP antibody was used to detect ligandbound CRP (data not shown). C, microtiter wells were coated with anti-CRP mAb EA4.1. The unreacted sites in the wells were blocked with gelatin. Purified CRP diluted in TBS-Ca was then added to the wells. Bound CRP was detected by using rabbit polyclonal anti-CRP antibody and HRP-conjugated donkey anti-rabbit IgG. Color was developed, and the absorbance was read at 405 nm. A representative of three experiments is shown.  Microtiter wells were coated with avidin in TBS. The unreacted sites in the wells were blocked with gelatin. Biotinylated PEt diluted in TBS was then added to the wells. After washing the wells with TBS, purified CRP diluted in TBS-Ca was added to the wells. Bound CRP was detected by using anti-CRP mAb HD2.4 and HRP-conjugated goat anti-mouse IgG. Color was developed, and the absorbance was read at 405 nm. A representative of four experiments is shown. Similar results were obtained when polyclonal anti-CRP antibody was used to detect PEt-bound CRP (data not shown).
using 150 g of CRP and six mice in each group in each experiment. The median survival time for mice injected with bacteria and WT CRP (group D) was 92 h. The median survival time for mice injected with bacteria and mutant CRP (group E) was 46 h. In the WT CRP-treated group, no deaths occurred in 46 h, and 17% of mice survived up to 7 days. By the end of the 3rd day, 75% of mice survived in the WT CRP-treated group when compared with only 25% in the mutant CRP-treated group. By the end of the 4th day, all mice died in the mutant CRP-treated group, whereas the survival was 25% in the WT CRP-treated group. Thus, similar results were obtained when mice were given either 25 g of CRP or 150 g of CRP; in contrast to WT CRP, mutant CRP did not decrease mortality and did not prolong survival of infected mice. Because 25 g of WT CRP was protective and mutant CRP was not protective even when used at 150 g, that is, at six times greater than the protective dose, we conclude that CRP-mediated protection of mice from infection is dependent upon the PCh binding activity of CRP. Fig. 7 shows the bacteremia values in mice from the protection experiment shown in Fig. 6. Based on the median bacteremia values, in mice injected with bacteria alone (group A), 1 day after infection, bacteremia was ϳ5.8 ϫ 10 4 cfu/ml blood. In mice injected with bacteria and 25 g of WT CRP (group B), bacteremia was 8.4 ϫ 10 2 cfu/ml 1 day after infection. In mice injected with bacteria and 150 g of WT CRP (group D), bacteremia was 2.5 ϫ 10 3 cfu/ml 1 day after infection. However, in mice injected with bacteria and 25 g of mutant CRP (group C), 1 day after infection, bacteremia was 5.9 ϫ 10 5 cfu/ml. In mice injected with bacteria and 150 g of mutant CRP (group E), 1 day after infection, bacteremia was 2.3 ϫ 10 5 cfu/ml. In groups A, C, and E, bacteremia increased dramatically after day 1, and mice died once bacteremia exceeded 10 8 cfu/ml. In mice administered with WT CRP, there was an increase in bacteremia past day 1, but it took another 2 days to exceed 10 8 cfu/ml when those mice died, when compared with Ͻ1 day for the WT CRP-treated group. Statistically significant differences in bacteremia were observed between the control (group A) and the WT CRP-treated groups (groups B and D) and between the WT CRP-treated and the mutant CRP-treated groups (groups C and E), until day 2. These results indicated that the increased resistance to infection in WT CRP-treated mice was associated with a slower increase in bacteremia and that the PCh binding activity of CRP was critical in this process.

F66A/T76Y/E81A Mutant CRP Stays Free in the Mouse Serum, and Its Clearance Rate in Vivo Is Not Faster than That of
WT CRP-Although mutant CRP did not protect mice from infection even when used at an amount that was six times higher than the protective dose for WT CRP, we determined the rate of clearance of mutant CRP from mouse circulation and compared it with that of WT CRP to confirm that mutant CRP was not protective because of its inability to bind pneumococci and not because of its faster clearance in vivo. The average slope of the four clearance curves for WT CRP was Ϫ0.67 Ϯ 0.02 (Fig. 8A), and the average slope of the four clearance curves for mutant CRP was Ϫ0.33 Ϯ 0.03 (Fig. 8B). Although the rate of clearance of mutant CRP (0.33 g/ml/h) was significantly different (p ϭ 0.03) from that of WT CRP (0.67 g/ml/h), the clearance of mutant CRP was not faster than that of WT CRP. These data suggested that the mutations did not confer instability to mutant CRP in vivo.
Although the clearance data also suggested that mutant CRP was free in the circulation and available for functions because it reacted with anti-CRP mAb HD2.4, we used another approach to confirm that mutant CRP was free in the mouse serum and was not sequestered by any other serum protein. As shown in Fig. 9, mutant CRP present in the mouse serum bound to PEt in a Ca 2ϩ -dependent manner and could be eluted with EDTA (lane 3). The recovery of mutant CRP was 96%. Besides CRP, no additional protein bands were found when compared with the nonspecific bands seen with the serum alone control (compare lanes 3 and 4). Successful repurification of mutant CRP from mutant CRP-spiked mouse serum further suggested that mutant CRP was free in the mouse serum and was not sequestered by any other serum protein.

DISCUSSION
In this study, we tested the hypothesis that the CRP-mediated protection of mice from pneumococcal infection is dependent upon the PCh binding activity of CRP. We mutated three amino acids in the PCh-binding pocket of CRP to generate a CRP mutant incapable of binding to PCh. We then compared the protective ability of WT CRP with that of mutant CRP. Our major findings were as follows. 1) Substitution of Phe 66 , Thr 76 , and Glu 81 with Ala, Tyr, and Ala, respectively, abolished the PCh binding, PnC binding, and pneumococcus binding activity of CRP. 2) F66A/T76Y/E81A mutant CRP was more efficient than WT CRP in binding to PEt. 3) Mutations in the PCh-binding pocket of CRP did not affect the overall pen-tameric structure of CRP. 4) F66A/T76Y/E81A mutant CRP stayed free in the mouse serum, and its clearance rate in vivo was not faster than that of WT CRP. 5) At both 25 g and 150 g, F66A/T76Y/E81A mutant CRP did not protect mice from pneumococcal infection.
Previously, we tested the same hypothesis that the CRP-mediated protection of mice from pneumococcal infection is dependent upon the PCh binding activity of CRP (7). The CRP mutant used in the earlier investigation was a CRP double mutant, F66A/E81A, which also does not bind to PCh, PnC, or pneumococci (7,13). We reported that F66A/E81A mutant CRP was as capable as WT CRP in protecting mice from pneumococcal infection, which was a surprising finding (7). This and the following three points prompted us to make another CRP mutant incapable of binding to PCh and repeat the mouse protection experiments. First, we wanted to use a mutant CRP, incapable of binding to PCh, with more drastic changes in the PCh-binding pocket. We wanted to block the pocket. In the earlier F66A/E81A double mutant, although two residues were mutated, the pocket was not blocked. Second, we wanted to use freshly purified CRP in all experiments. Third, because we were now successful in establishing a mouse model of pneumococcal infection in which 25 g of CRP was protective, as has been shown by others (25,26), we wanted to use two different doses of CRP (25 and 150 g) in protection experiments. Previously (7), we only used one dose of CRP, 150 g.
As shown in Fig. 1A, Phe 66 , Thr 76 , and Glu 81 participate in the CRP-PCh interaction (16). To generate a CRP mutant incapable of binding to PCh, our mutagenesis plan included mutating all three amino acids. Based on the structure of SAP (18), we mutated Thr to Tyr and added the T76Y mutation to our previously published CRP mutant, F66A/E81A. Molecular modeling of F66A/T76Y/E81A CRP triple mutant showed that the large side chain of Tyr could partially block the PCh-binding pocket (Fig. 1B). Loss of binding of mutant CRP to PCh, but dramatically enhanced binding to PEt, further indicated that the PCh-binding pocket was blocked in mutant CRP. In WT CRP, the amino group of PEt makes an electrostatic interaction with Glu 81 (10). In SAP, the amino group of PEt is situated within H-bonding distance of the Tyr 74 hydroxyl group; such H-bonding is not possible for WT CRP as the corresponding   residue is Thr (10). It has also been shown that SAP binds to immobilized PEt much more avidly than CRP does, although the K d values of CRP and SAP for PEt are similar (10). In mutant CRP, the presence of Tyr at position 76 might have facilitated H-bonding between the amino group of PEt and hydroxyl group of Tyr. Mutation of Glu 81 might have further facilitated this interaction and increased the avidity of the binding of mutant CRP to PEt. Also, due to stronger avidity of mutant CRP for PEt, it was easier to purify mutant CRP by PEt affinity chromatography. Interestingly, like SAP (9,10,37,38), mutant CRP exhibited carbohydrate binding property; in the presence of Ca 2ϩ , mutant CRP bound to agarose beads used for gel filtration chromatography (data not shown).
CRP binds to Fc␥R (39 -41), and Fc␥R have been implicated in protection from pneumococcal infection in mice (42). CRP has been shown to enhance uptake and presentation of pneumococcal antigens through Fc␥R on dendritic cells in vitro (43). However, we did not characterize mutant CRP for binding to Fc␥R because Fc␥R do not participate in CRP-mediated protection of mice against pneumococcal infection (44). CRP was found to be equally protective in WT and Fc␥R knock-out mice (44). We did not characterize mutant CRP for activating the complement system because CRP activates complement only when ligand-complexed (30), but mutant CRP was unable to complex with PCh-BSA, PnC, or pneumococci.
We used a high dose, 5 ϫ 10 7 cfu, of pneumococci in mouse protection experiments. This dose of bacteria was necessary because endogenous mouse CRP might have been protective against lower doses of bacteria. It has been reported that endogenous mouse CRP is also protective against pneumococcal infection in mice; the LD 50 of pneumococci was drastically reduced in CRP KO mice (45). We used 25 and 150 g of CRP in mouse protection experiments; however, we also found that even 10 g of WT CRP was protective (data not shown), as has been reported earlier (24 -26). Our finding that WT CRP was protective, whereas mutant CRP was not protective even at a 150-g dose, clearly indicated that the PCh-binding pocket in CRP was critical for the initial protection of mice from pneumococcal infection. Our findings are consistent with other studies that showed that the PCh on pneumococci mediated the function of CRP to block the attachment of pneumococci to platelet-activating factor receptors on human pharyngeal epithelial cells and that the PCh binding activity of CRP was required for the protection of mice from challenge with platelet-activating factor (46,47).
Because CRP protects mice from infection only when injected within the range of 6 h before to 2 h after administering pneumococci, and not when injected 24 or 36 h after infection (7,26), our data suggest that CRP is able to use the PCh bindingbased mechanism for the protection of mice only during the early stages of infection. It has been shown that WT CRP undergoes structural transformations under several different experimental conditions (3, 33, 48 -53). It has also been shown that structurally altered, or slightly to completely denatured, CRP is capable of binding to immobilized factor H (3, 33, 51-56). Factor H is a regulator of complement activation and has been implicated in the resistance of pneumococci to complement attack (3,(57)(58)(59). It remains to be investigated whether such CRP-factor H interaction plays any role in CRP-mediated protection of mice from pneumococcal infection. However, the combined data indicate that the CRP-mediated protection of mice requires the presence of CRP during the early stages of infection.
The outcome of the in vivo experiments obtained with the use of CRP triple mutant in this investigation was different from the results obtained with the use of CRP double mutant reported earlier (7). Both mutants fail to bind to PCh, PnC, or pneumococci. The only difference in the ligand binding properties of the two mutants was their binding avidity for PEt; the binding of CRP double mutant to PEt was not different from that of WT CRP (data not shown). However, there were two major differences in the methods used in the two in vivo experiments. The first difference was in the purity of CRP. Unlike CRP double mutant, CRP triple mutant was freshly purified and never stored frozen. Whether the difference in the results of the two in vivo experiments was due to the difference in the structures of the PCh-binding pocket of the two CRP mutants, or due to the difference in the ages of purified CRP, remains to be explored. The second difference was in the infection models, regarding the protective ability of WT CRP. In the infection model used earlier, Ͻ100 g of WT CRP was not protective and, therefore, we had to use 150 g of both WT and double mutant CRP (7). In the infection model used in the current study, even Ͻ25 g of WT CRP was protective, and thus, we could use two doses, 25 and 150 g, of WT and triple mutant CRP. The data shown in Fig. 6A with the use of 25 g of CRP clearly indicated that the PCh-binding pocket was necessary and required for protection of mice during the initial stage of infection. Statistical analysis of the data shown in Fig. 6B with the use of 150 g of CRP showed that even 150 g of CRP triple mutant was not protective. However, although statistically not significant, there was a slight protective effect of CRP triple mutant when used at 150 g, raising the possibility that Ͼ150 g of CRP triple mutant might have produced results similar to those obtained with 150 g of CRP double mutant reported earlier (7). Alternatively, 25 g of CRP double mutant might not have been protective as opposed to 150 g. Because administering a higher dose of CRP allows CRP to stay longer in the host, we hypothesize that CRP mutants, when given at the beginning of the experiment, undergo remodeling to gain other ligand binding properties, as yet undefined, during their stay in the host, and that may be responsible for the protection of mice from infection independent of the PCh binding activity of CRP. This hypothesis also provides insight into why CRP does not protect mice from infection if administered later and how CRP can be engineered so that it can protect mice from infection even if administered during the late stages of infection (33,51). However, this is just a hypothesis because we have no data.
We conclude that the PCh-binding pocket on CRP is necessary for CRP-mediated initial protection of mice against lethal pneumococcal infection. Development of infection models involving passively administered human WT and mutant CRP in CRP KO mice (45, 60 -62) and SAP KO mice (63)(64)(65), and also the development of mice transgenic for CRP mutants, may provide more information on the mechanisms of antipneumo-coccal function of CRP in protecting mice against early and late stages of infection.