Learn more: PMC Disclaimer | PMC Copyright Notice
Characterization of Binding of Human Lactoferrin to Pneumococcal Surface Protein A
Abstract
Human lactoferrin is an iron-binding glycoprotein that is particularly prominent in exocrine secretions and leukocytes and is also found in serum, especially during inflammation. It is able to sequester iron from microbes and has immunomodulatory functions, including inhibition of both complement activation and cytokine production. This study used mutants lacking pneumococcal surface protein A (PspA) and PspC to demonstrate that the binding of human lactoferrin to the surface of Streptococcus pneumoniae was entirely dependent on PspA. Lactoferrin bound both family 1 and family 2 PspAs. Binding of lactoferrin to PspA was shown by surface colocalization with PspA and was verified by the lack of binding to PspA-negative mutants. Lactoferrin was expressed on the body of the cells but was largely absent from the poles. PspC showed exactly the same distribution on the pneumococcal surface as PspA but did not bind lactoferrin. PspA's binding site for lactoferrin was mapped using recombinant fragments of PspA of families 1 and 2. Binding of human lactoferrin was detected primarily in the C-terminal half of the α-helical domain of PspA (amino acids 167 to 288 of PspA/Rx1), with no binding to the N-terminal 115 amino acids in either strain. The interaction was highly specific. As observed previously, bovine lactoferrin bound poorly to PspA. Human transferrin did not bind PspA at all. The binding of lactoferrin to S. pneumoniae might provide a way for the bacteria to interfere with host immune functions or to aid in the acquisition of iron at the site of infection.
Lactoferrin is an iron-binding glycoprotein present in milk and mucosal secretions. It is also released by specific granules of polymorphonuclear leukocytes during inflammation (35, 36). It is a member of the siderophilin family and is structurally related to the more abundant serum protein transferrin (40). Lactoferrin has been ascribed many diverse biological functions, most of which are immunomodulatory or antibacterial (5, 7, 20–22, 37, 57, 58). It can inhibit cytokine activation, myelopoiesis, and complement activation (21, 34, 37, 54). It also plays a role in host resistance by sequestering from bacteria the free iron necessary for bacterial growth and by the bactericidal activity of an N-terminal fragment released after pepsin digestion in the gut (22, 56, 58).
Streptococcus pneumoniae is an important cause of respiratory tract infections, bacteremia, and meningitis. These infections are especially common in young children and in the elderly (3, 26). Infection usually starts with asymptomatic carriage in the nasopharynx. Bacteria can then, in some cases, spread to other locations such as the lungs, middle ear, and blood (4, 26, 55). To effectively infect the host, pneumococci have to survive and evade the immune system in the nasopharynx as well as at other sites within the host. This may be accomplished by binding immunomodulatory molecules, such as lactoferrin, at the site of infection.
S. pneumoniae has been reported to bind lactoferrin (28). Using radiolabeled, milk-purified lactoferrin, Hammerschmidt et al. observed interaction of lactoferrin with 88% of the clinical S. pneumoniae isolates tested. The bacterial receptor was purified by affinity chromatography and identified as pneumococcal surface protein A (PspA). This interaction with purified lactoferrin was further verified using purified PspA.
In the present study, we have more completely characterized the binding of lactoferrin to S. pneumoniae. To avoid the potential presence of other copurified proteins frequently associated with the purification of lactoferrin and other proteins from milk, we used recombinant human lactoferrin in our binding studies. Also, we used fluorescence methodology to quantitate and microscopically visualize binding of lactoferrin to the bacterial surface, something that was not attempted in the original study (28). By using recombinant lactoferrin and an isogenic pneumococcal strain lacking expression of PspA, it has been shown for the first time that the binding of lactoferrin to the pneumococcal surface is dependent on PspA and that PspC is not involved in lactoferrin binding. These studies have also revealed localized surface distribution of PspA and identified the region of PspA that binds to lactoferrin.
MATERIALS AND METHODS
Reagents.
Protein markers were from Amersham Pharmacia Biotech (Piscataway, N.J.). NBT (nitroblue tetrazolium) and BCIP (5-bromo-4-chloro-3-indolylphosphate) were from Fisher Scientific (Atlanta, Ga.). Alkaline phosphatase (AP)-conjugated streptavidin, biotin-conjugated goat anti-mouse immunoglobulin (Ig), and biotin-conjugated goat anti-rabbit Ig antibodies were from Southern Biotechnology Associates (Birmingham, Ala.). Fluorescein isothiocyanate (FITC)-conjugated streptavidin, R-phycoerythrin (RPE)-conjugated streptavidin, FITC-conjugated rabbit anti-mouse Ig antibodies, and FITC-conjugated goat anti-rabbit Ig antibodies were from Dako A/S (Rothskild, Denmark). Bacto-Todd Hewitt medium and yeast extract were from Difco Laboratories (Detroit, Mich.). Human and bovine milk lactoferrin and human transferrin were from Sigma Chemical Co (St. Louis, Mo.).
Recombinant human lactoferrin was kindly provided by Arne Forsgren, Department of Medical Microbiology, Lund University, University Hospital Malmö, Malmö, Sweden.
Monoclonal anti-PspA antibodies Xi126 and XiR278 have been previously described (38). Polyclonal anti-PspC serum was raised by immunization with a truncated PspC molecule from S. pneumoniae L81905 expressed in Escherichia coli (19). Polyclonal anti-PspA family 1 antiserum was pooled from two rabbits immunized with recombinant PspA/L82016 (clade 1) or recombinant PspA/Rx1 (clade 2) expressed in E. coli. Anti-PspA family 2 antiserum was produced by pooling serum from two rabbits immunized with either recombinant PspA/V-024 (clade 3) or recombinant PspA/V-032 (clade 4) expressed in E. coli. To this pool, recombinant PspA/Rx1 was added to reduce its cross-reactivity with family 1 (clade 1 and 2) PspAs. Approximately 10 μg of the purified PspA proteins was injected subcutaneously into a rabbit twice on consecutive weeks, and blood was collected 10 days after the last injection. The primary immunization was given with Freund's complete adjuvant, and the booster immunization was given in saline.
Bacteria.
The strains and plasmids used in this study are described in Table Table1.1. The pneumococcal strains were stored at −80°C in fetal calf serum, transferred to blood agar plates, and incubated at 37°C in a 5% CO2 atmosphere overnight. Colonies grown on blood agar were used to inoculate liquid growth medium (Todd-Hewitt medium containing 0.5% yeast extract [THY]). Upon reaching late log phase, the bacteria were harvested by centrifugation at 1,500 × g for 15 min and suspended in 60 mM phosphate-buffered saline (PBS, pH 7.2). The bacterial concentration was estimated by interference contrast microscopy (TE Leitz Ortolux II microscope with interference contrast equipment; Leitz, Wetzlar, Germany) using a Bürker chamber and confirmed by counting viable cells. Appropriate dilutions of the bacteria were suspended in PBS.
TABLE 1
Bacterial strains and plasmids
| Strain or plasmid | Characteristics | Source or reference |
|---|---|---|
| S. pneumoniae | ||
| D39 | Type 2 encapsulated | 6, 59 |
| Rx1 | Nonencapsulated mutant of D39 | 48 |
| JY53 | pspA mutant of D39 | 59 |
| TRE118 | pspC mutant of D39 | This study |
| TRE121 | pspA pspC mutant of D39 | This study |
| WU2 | Type 3 encapsulated | 17 |
| EF3296 | Type 4 encapsulated | 2, 59 |
| L81905 | Type 4 encapsulated | 19 |
| EF3030 | Type 19F encapsulated | 1 |
| CCUG10175 | Type 19F encapsulated | 2 |
| M. catarrhalis | Clinical isolate | Clinical bacteriology lab, Lund, Sweden |
| E. coli M15 | Qiagen | |
| Plasmids | ||
| pHR101 | pQE40::BglII-HindIII fragment, encodes PspA/EF3296 (1–115) | This study |
| pHR102 | pQE40::SalI-SphI fragment, encodes PspA/EF3296 (75–305) | This study |
| pHR105 | pQE40::SalI-SphI fragment, encodes PspA/EF3296 (1–411) | This study |
| pHR107 | pQE40::SalI-SphI fragment, encodes PspA/EF3296 (75–490) | This study |
| JAS218 | PQE40::BamHI-SalI fragment, encodes PspA/Rx1 (167–288) | This study |
| PBAR4285 | Encodes PspA/Rx1 (1–115) | 38, 44, 59 |
| pBAR4310 | Encodes PspA/Rx1 (1–192) | 38, 44, 59 |
| pBAR501 | Encodes PspA/Rx1 (288–563) | 38, 44, 59 |
| pUAB55 | Encodes PspA/Rx1 (1–370) | 12, 19 |
| pUAB103 | Encodes PspA/Rx1 (1–303) | 19 |
The pspC gene was insertionally inactivated in S. pneumoniae D39. An internal fragment of pspC was amplified using PCR and cloned into pSF143 (51). The ligated plasmid was electroporated into E. coli DH5α, and clones were selected for tetracycline resistance. The plasmid, which contained the internal fragment of pspC, was isolated from recombinant E. coli using standard procedures and transformed into S. pneumoniae D39 (30). Tetracycline-resistant recombinants were screened by both Southern hybridization and Western blotting to confirm inactivation of pspC. Lysate from the strain containing the insertionally inactivated pspC (TRE118) was transformed into JY53 (erythromycin resistant, pspA negative) to create a mutant that lacked both pspA and pspC (TRE121) (23, 59).
Binding of lactoferrin and transferrin to bacterial cells.
Purified lactoferrin from human or bovine milk, recombinant human lactoferrin, and human transferrin were biotinylated using the Roche biotin labeling kit according to the manufacturer's instructions (Roche Molecular Biochemicals, Indianapolis, Ind.).
Bacteria were grown in THY medium (S. pneumoniae) or on chocolate agar plates (Moraxella catarrhalis) and suspended in PBS at a concentration of approximately 5 × 108 bacteria/ml. The bacterial suspension (100 μl) was mixed with 0.5 to 10 μl of biotinylated protein (2-mg/ml stock solution in PBS) for 30 min at room temperature and washed by centrifugation at 1,500 × g for 5 min in PBS. FITC-conjugated streptavidin (1:100 dilution in PBS) was added for an additional 30 min at room temperature, and after a final wash in PBS, the cells were inspected by epifluorescence and laser scanning confocal microscopy using MRC-1024 confocal equipment (Bio-Rad Laboratories, Hemel-Hampstead, United Kingdom) attached to a Nikon Eclipse E800 upright microscope (Nikon, Tokyo, Japan). The binding was quantitated by flow cytometry using a FACSCalibur flow cytometer (Becton Dickinson Biosciences, Rutherford, N.J.).
Antibody staining of S. pneumoniae.
Studies of the colocalization of PspA and lactoferrin were performed using the monoclonal anti-PspA antibodies Xi126 and XiR278, recognizing the N-terminal and the more distal part of the α-helical region of PspA, respectively (38). Bacteria were first incubated with 5 μl of biotinylated lactoferrin and washed in PBS, and the bacteria were then fixed for 5 min in 4% formaldehyde in PBS. After the bacteria had been washed in PBS, monoclonal antibodies (undiluted hybridoma supernatant) were added for an additional 30 min. After a third wash in PBS, the bacteria were incubated with RPE-conjugated streptavidin (1:100 in PBS) and FITC-conjugated rabbit anti-mouse Ig antibodies (1:100 in PBS) for 30 min at room temperature, and binding was inspected by epifluorescence and confocal microscopy. Controls treated without the monoclonal anti-PspA antibody showed no staining with FITC.
Staining for PspC on the bacterial surface was done using anti-PspC antibodies as described above. As the anti-PspC antibodies also recognize PspA (19), staining for PspC was performed using the PspA-negative mutant JY53.
Staining for the cell wall was achieved using anti-phosphoryl choline monoclonal antibody 59.6C5 (13) using the technique described above.
PspA and PspA fragments.
Full-length PspA was purified from S. pneumoniae Rx1 and EF3296 as described (16, 60). PspA fragments BAR4285, BAR4310, and BAR501 from strain Rx1 were produced as described (38, 44). BAR4285 and BAR4310 are derived from pJY4285 and pJY4310, respectively, originally cloned into the pUC18 vector, as described (59). The inserts were moved into pMal-p2 and expressed as described (44). The predicted sizes of the expression constructs were 55.2 kDa (BAR4285), 63.6 kDa (BAR4310), and 72.8 kDa (BAR501). PspA fragments UAB055 and UAB103 from strain Rx1 were produced as described (12, 19). PspA fragments HR101 (primer pair ABW23 and LSM12), HR102 (primer pair HR10 and HR11), and HR107 (primer pair HR10 and HR14) from S. pneumoniae EF3296 pspA and JAS218 (primer pair LSM150 and LSM16) from S. pneumoniae Rx1 pspA were expressed in E. coli strain M15 using the expression vector pQE40 (Qiagen Inc, Chatsworth, Calif.). The primers used for PCR are described in Table Table2.2. PCR-amplified fragments of pspA were cloned into SphI-SalI-, BglI-HindIII-, or BamHI-SalI-digested pQE40 vector (Table (Table1)1) and transformed into M15(pREP4), a K-12-derived E. coli strain containing a plasmid which carries a lac repressor, allowing control over expression. Clones containing the different pspA inserts were identified by Southern blot analyses using digoxigenin-labeled pspA probes (38). Expression of PspA fragments from positive clones was induced with 1 mM IPTG (isopropylthiogalactopyranoside) during growth at room temperature. The overexpressed protein fragments were purified by affinity chromatography using a nickel resin according to the manufacturer's instructions. The different constructs encoded predicted 38.6-kDa (HR101), 52-kDa (HR102), 71.8-kDa (HR107), and 39.0-kDa (JAS218) PspA fragments, which were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and quantified using the Bio-Rad DC protein assay (Bio-Rad Laboratories, Hercules, Calif.). PspC was purified as described (19).
TABLE 2
Primers used for expression of recombinant PspA fragmentsa
| Primer | Position in sequence (bp) | Sequence |
|---|---|---|
| ABW23 | 400–417 | 5′-TCTGATATTCAGCGTCAG-3′ |
| HR10 | 428–447 | 5′-GCATATAAAGAGTACCGAGA-3′ |
| HR11 | 1103–1121 | 5′-GATACTGCTGCTCTTCCAAA-3′ |
| HR14 | 1656–1675 | 5′-CTACTCAACCAGAAAAACCA-3′ |
| LSM12 | 149–173 | 5′-AGCGTCGCTATCTTAGGGGCTGGTT-3′ |
| LSM150 | 577–591 | 5′-CCAAGGATCCGTGGATGCTGAAGAA-3′ |
| LSM16 | 928–942 | 5′-CGCGTCGACACTCTCATTAACTGCTTT-3′ |
Dot blot.
PspA, fragments of PspA, and PspC (10 μg/ml and 1:3 serial dilutions of these stock solutions) were applied to a 0.45-μm nitrocellulose membrane (Millipore, Bedford, Mass.), and the membrane was allowed to dry. The membrane was blocked with 1% bovine serum albumin (BSA) in PBS for 45 min at room temperature and washed three times with PBS containing 0.1% Tween 20 (PBS-T). The membrane was overlaid with biotinylated recombinant and milk-purified human lactoferrin, bovine lactoferrin, or biotinylated human transferrin (1:500 dilution in PBS-T of the 2.0-mg/ml stock solution) for 45 min at room temperature and washed three times in PBS-T. After an additional incubation with AP-conjugated streptavidin (1:500 dilution in PBS-T) for 45 min at room temperature, the membrane was developed using 1 mg of NBT and 5 mg of BCIP per 10 ml of 0.15 M Tris-HCl (pH 8.8).
Western blot.
PspA, fragments of PspA, and PspC (0.5 μg) were run on 10% polyacrylamide gels (Bio-Rad Ready gels; Bio-Rad Laboratories); and the gels were electroblotted to a 0.45-μm nitrocellulose membrane (Bio-Rad) in Tris-glycine buffer (20% methanol, 25 mM Tris, 192 mM glycine [pH 8.1 to 8.4]) at 100 V for 1 h at 4°C. The blotted membrane was incubated with 1% BSA in PBS-T for 45 min at room temperature and washed three times (5 min each) with PBS-T. The membranes were overlaid with biotinylated human recombinant or milk-purified lactoferrin (1:500 dilution in PBS-T of 2.0-mg/ml stock solution), with anti-PspA antibodies (monoclonal Xi126 or polyclonal anti-PspA family 1 antiserum), or with polyclonal anti-PspA family 2 antiserum for 30 min at 37°C and washed three times in PBS-T. The anti-PspA-exposed membrane was further incubated with a mix of biotinylated goat anti-mouse or goat anti-rabbit Ig antibodies (1:1,000 in PBS-T) and AP-conjugated streptavidin (1:500 dilution in PBS-T) for 30 min at 37°C, and the lactoferrin-exposed membrane was incubated with streptavidin only. After washing, the membrane was developed using 1 mg of NBT and 5 mg of BCIP per 10 ml of 0.15 M Tris-HCl (pH 8.8).
RESULTS
Binding of human lactoferrin to S. pneumoniae.
S. pneumoniae D39 was incubated with human recombinant lactoferrin (Fig. (Fig.1A)1A) or lactoferrin purified from human milk (data not shown), counterstained with FITC-conjugated streptavidin, and inspected by confocal or epifluorescence microscopy. Recombinant lactoferrin and lactoferrin from human milk showed identical binding patterns of strong binding to D39. This observation made it clear that lactoferrin, and not a contaminating milk protein, was responsible for the binding to the pneumococcal surface. Similar binding was observed for S. pneumoniae EF3296 (data not shown).
Binding of lactoferrin to S. pneumoniae D39. Bacteria were incubated with biotinylated recombinant lactoferrin, costained with FITC-conjugated streptavidin, inspected by confocal or epifluorescence microscopy, and quantitated by flow cytometry. (A) Binding of recombinant human lactoferrin (hLF) to S. pneumoniae D39 (left), a transmission light detection image (middle) of the same bacteria, and a merged picture of the two (right). Binding was detected by confocal microscopy, and the staining showed a pattern with localized accumulation of binding to the sides of the bacteria, with less binding to the poles and the interbacterial zones. (B) Staining of the cell wall of S. pneumoniae Rx1 with anti-phosphoryl choline (anti-PC) antibodies (left), a transmission light detection image (middle) of the same bacteria, and a merged picture of the two (right). Binding was more or less homogenous around the bacterial surface. (C and D) Flow cytometry analysis of recombinant lactoferrin binding to S. pneumoniae D39 (C) and EF3296 (D). Fluorescence detected from bacteria treated with biotinylated lactoferrin. With lactoferrin at 10 μg/ml (arrow 10), binding was 12.9 and 5.5 times above that of the controls for D39 and EF3296, respectively; at 40 μg/ml (arrow 40), binding was 27.6 and 9.9 times above that of the controls for D39 and EF3296, respectively; at 100 μg/ml (arrow 100), binding was 31.1 and 22.2 times above that of the controls for D39 and EF3296, respectively; and at 150 μg/ml (arrow 150), binding was 32.1 and 22.1 times above that of the controls for D39 and EF3296, respectively. Binding was compared to that with streptavidin-alone-treated (SA) bacteria.
The surface binding of recombinant lactoferrin was further quantitated using flow cytometry, resulting in a clear shift of fluorescence intensity with increased concentrations of lactoferrin (Fig. (Fig.1C1C and D). At binding saturation (50 to 100 μg of lactoferrin per ml), 31.1 (range, 17.3 to 65.4) times higher fluorescence intensity was observed for S. pneumoniae D39 compared with the streptavidin-treated control cells (Fig. (Fig.1C1C and D). A similar fluorescence intensity shift was seen for the binding of human lactoferrin to S. pneumoniae EF3296 (22.2 times above the control [range, 15.3 to 27.6]).
Binding of human lactoferrin to PspA.
The binding of lactoferrin to S. pneumoniae has been reported to involve PspA. Recombinant PspA has been shown to inhibit the association of radiolabeled lactoferrin with whole bacteria (28). However, the functional interaction of lactoferrin and PspA and the dependence on PspA for lactoferrin binding on the bacterial surface were not previously addressed. We have investigated these questions in two ways: colocalization experiments of lactoferrin and monoclonal anti-PspA antibodies on the bacterial surface and experiments using mutants of S. pneumoniae lacking PspA surface expression. Binding was analyzed by confocal microscopy and flow cytometry.
Staining of the bacteria with lactoferrin and monoclonal anti-PspA antibodies showed a pattern of colocalization consistent with the binding of lactoferrin to PspA (Fig. (Fig.2A).2A). The pattern for lactoferrin binding and anti-PspA antibody staining over the bacterial surface was not uniform but displayed localized binding with areas of higher and lower intensity. In most cases PspA was only poorly expressed at the poles of the cells. Most of the localized staining was thus present along the body of the cell, although some examples of other staining patterns could be observed. On close examination, the most common pattern of staining could be discerned based on diplococcal units. Within a four-coccus chain, there was faint staining near the poles of the most recent cell division but none near the poles of the preceding cell division (Fig. (Fig.1A1A and and2A).2A). This pattern of staining was seen consistently for all wild-type strains used in the study regardless of capsular type (D39, type 2; WU2, type 3; EF3296, type 4; L81905, type 4; EF3030, type 19F; and CCUG10175, type 19) and was also present in pneumococci lacking a capsule (Rx1). The staining pattern of PspA was different from the homogenous staining of the cell wall of S. pneumoniae Rx1 using anti-phosphoryl choline antibodies (Fig. (Fig.1B).1B). This suggests that the localized expression of PspA away from the ends of the cells was not an artifact of the experimental conditions.
Colocalization of lactoferrin and PspA on isogenic mutants of D39. S. pneumoniae D39 and isogenic mutants lacking surface expression of PspA, PspC, or both PspA and PspC were incubated with biotinylated recombinant lactoferrin, followed by incubation with anti-PspA antibodies (monoclonal Xi126), and counterstained with RPE-conjugated streptavidin and FITC-conjugated anti-mouse Ig antibodies. Binding was inspected by confocal microscopy. Green fluorescence indicates staining with anti-PspA antibodies, red fluorescence indicates staining with lactoferrin, and yellow staining indicates colocalization of anti-PspA antibodies and lactoferrin. (A) D39 wild-type bacteria; (B) JY53 bacteria lacking PspA on the surface; (C) TRE118 bacteria lacking PspC on the surface; (D) TRE121 bacteria lacking PspA and PspC on the surface. JY53, TRE118, and TRE121 are all specific mutants of strain D39. Bar, 2 μm.
The specificity of the binding of lactoferrin to PspA was demonstrated using PspA-negative mutants of S. pneumoniae D39. The PspA-negative mutant did not express PspA on the surface, as seen from the lack of staining with monoclonal anti-PspA antibodies, and did not bind milk-purified or recombinant lactoferrin (Fig. (Fig.2B).2B). The intensity of binding in flow cytometry was similar to that of the streptavidin-treated bacteria, verifying the confocal microscopy results (Fig. (Fig.3B).3B).
Flow cytometry analysis of lactoferrin binding to isogenic mutants of D39. S. pneumoniae D39 and isogenic mutants lacking surface expression of PspA, PspC, or both PspA and PspC were incubated with biotinylated recombinant lactoferrin and counterstained with FITC-conjugated streptavidin, and binding was quantitated by flow cytometry. The white area indicates bacteria treated with streptavidin alone, and the gray area indicates binding of lactoferrin. (A) D39 wild-type (wt) bacteria (binding 44.3 times the streptavidin control value); (B) JY53 lacking PspA on the surface (binding 0.96 times the streptavidin control value); (C) TRE118 lacking PspC on the surface (binding 40.4 times the streptavidin control value); (D) TRE121 lacking PspA and PspC on the surface (binding 0.96 times the streptavidin control value). The fluorescence intensity of the lactoferrin-treated and streptavidin-treated PspA-negative cells revealed no residual binding of lactoferrin to other components on the bacterial surface.
Lactoferrin does not bind to PspC.
S. pneumoniae expresses a second choline-binding protein with high homology to PspA, designated PspC, SpsA, or CbpA (15, 29, 55). PspC has been suggested to be involved in adherence to epithelial cells in the respiratory tract as well as in binding to the secretory component of IgA and complement factor C3 (19, 29, 45, 49). In the original report, PspC was shown not to bind lactoferrin in Western blot (28). This did not, however, exclude the involvement of PspC as a cofactor in lactoferrin binding. A complex of two proteins constituting the lactoferrin receptor is a common feature in gram-negative bacteria (9–11, 27, 46). To investigate the potential role of PspC expressed on the surface of the bacteria for lactoferrin binding, we used a mutant of S. pneumoniae D39 lacking PspC. This mutant still expressed PspA, as shown by staining with anti-PspA antibodies, and had an identical lactoferrin binding pattern in confocal microscopy and the same intensity of binding by flow cytometry as the wild-type control (Fig. (Fig.2C2C and and33C).
Similarly, PspA-negative bacteria, which by fluorescence staining still expressed PspC on the surface (data not shown), showed no residual binding of lactoferrin (Fig. (Fig.2B2B and and3B).3B). Neither lactoferrin nor anti-PspA antibodies bound a double mutant strain lacking both PspA and PspC (Fig. (Fig.2D2D and and3D).3D). These results demonstrate that PspC is not able to bind lactoferrin to the pneumococcal surface and is not a necessary cofactor for PspA-dependent binding. Interestingly, the cell surface distribution of PspC was identical to that seen for PspA, with areas of lower and higher intensity of expression (data not shown).
Finally, recombinant as well as milk-purified human lactoferrin was capable of detecting PspA but not PspC in dot blot and Western blot analyses (Fig. (Fig.4B4B and C). The total lack of binding to the PspA-negative bacteria further suggests that lactoferrin does not bind to PspC or any other component expressed on the bacterial surface under the conditions used in these experiments.
Binding of human lactoferrin to PspA and recombinant PspA fragments in dot blot and Western blot analyses. (A) Map of the recombinant PspA/Rx1 (family 1) and PspA/EF3296 (family 2) fragments used in the study. (B) Dot blot. Full-length PspA and PspC and recombinant fragments of PspA were dot blotted to a nitrocellulose membrane. The membrane was blocked, overlaid with human recombinant lactoferrin, and developed with NBT after incubation with AP-conjugated streptavidin. Binding of lactoferrin to both full-length PspA/Rx1 and PspA/EF3296 was detected but not to PspC. When the PspA fragments were investigated, human lactoferrin was shown to bind to all Rx1 fragments except BAR4285, consisting of amino acids 1 to 115, and BAR4310, consisting of amino acids 1 to 192. The same was seen for fragments of PspA/EF3296, where lactoferrin was shown to bind to all fragments except HR101, consisting of amino acids 1 to 115. (C) Western blot. Full-length PspA, recombinant fragments of PspA, and PspC were run on 10% gels and electroblotted to a nitrocellulose membrane. The membrane was blocked, overlaid with human recombinant lactoferrin, and stained with NBT after incubation with AP-conjugated streptavidin. No binding was detected for BAR4285, BAR4310, HR101, or PspC. These results confirmed the dot blot results. The positions of size markers are shown on the left (in kilodaltons).
Mapping of the binding of lactoferrin to PspA.
To identify the general location of the binding site for lactoferrin on PspA, we compared the binding of lactoferrin to full-length PspA with that of recombinant E. coli-expressed fragments consisting of various amino acids from the family 1 and family 2 PspA sequences (Fig. (Fig.4A).4A). The PspA molecule is genetically variable and has been classified into six clades by sequence homology (31). Five of these clades make up families 1 and 2 of PspA sequences and comprise over 90% of PspA in pneumococcal isolates. PspAs of families 1 and 2 are only partially cross-reactive in enzyme-linked immunosorbent assays with immune sera and display over 40% divergence at the level of nucleotide sequence. Full-length PspA from Rx1 (family 1) and EF3296 (family 2) bound milk-purified and recombinant lactoferrin in dot blot and Western blot analyses, suggesting that the binding site for lactoferrin was conserved between the two families despite the variability in their amino acid sequences (Fig. (Fig.4B4B and C).
Lactoferrin failed to bind the most N-terminal region of PspA, shown by the inability to bind Rx1 fragments BAR4285 (amino acids 1 to 115) and BAR4310 (1 to 192) and EF3296 fragment HR101 (1 to 115). On the other hand, lactoferrin bound significantly to the Rx1 fragments UAB055 (1 to 303), UAB103 (1 to 370), and JAS218 (167 to 288) and the EF3296 fragments HR102 (75 to 305) and HR107 (75 to 490) (Fig. (Fig.4B4B and C). Lactoferrin appeared to show weak binding to fragment BAR501 (288 to 563), consisting of the proline-rich and choline-binding domains. The binding results with these fragments suggest that the primary binding site of lactoferrin resides in the C-terminal part of the α-helical domain, with a possible secondary interaction with some of the more C-terminal elements of the molecule.
S. pneumoniae is bound more strongly by human lactoferrin than by bovine lactoferrin.
Among Moraxella spp., it has been observed that the strength of binding to lactoferrin differs for lactoferrins of different origins, with the strongest binding being to lactoferrin of the host that it infects (10, 27). In addition to binding lactoferrin, most gram-negative bacteria can also bind and utilize transferrin as an iron source, using a system very similar to that for acquisition of iron from lactoferrin (27). Hammerschmidt et al. demonstrated that bovine lactoferrin could not block binding of radiolabeled human lactoferrin to pneumococcal cells (28). The direct binding of bovine lactoferrin to pneumococci was not attempted.
As S. pneumoniae is exclusively a human pathogen, we investigated the difference in binding between lactoferrins of human and bovine origin. Binding of lactoferrin to the surface of M. catarrhalis was used as a control, and binding of both the human and bovine forms of the protein was detected (data not shown), indicating that both mammalian proteins exhibit functional binding under the conditions used. Human lactoferrin bound much more strongly to S. pneumoniae D39 and EF3296 than did bovine lactoferrin (Fig. (Fig.5A5A and B). The human protein bound to the bacterial surface with an intensity of 32 (D39) and 16 (EF3296) times that of the respective controls treated with streptavidin alone. In contrast, bovine lactoferrin showed only 2.1-fold (D39) or 1.3-fold (EF3296) greater binding compared to the respective controls. The interactions of human and bovine lactoferrins with PspA were verified by dot blot analyses (Fig. (Fig.5C).5C). These results indicate species specificity towards components present in the natural host.
Binding of human and bovine lactoferrin. S. pneumoniae D39 and EF3296 were incubated with biotinylated recombinant human (hLF) or bovine (bLF) lactoferrin or with biotinylated human transferrin (hTF) and counterstained with FITC-conjugated streptavidin. Binding was quantitated by cell sorting (A and B) or visualized by dot blotting (C). (A and B) Binding of human and bovine lactoferrin and human transferrin to S. pneumoniae D39 and EF3296. Human recombinant lactoferrin (hLF) had a binding intensity 35.4 (D39) and 22.2 (EF3296) times that of the streptavidin control. Bovine lactoferrin (bLF) only bound with intensities of 2.1 (D39) and 1.3 (EF3296) times that of their streptavidin-treated controls. Human transferrin (hTF) showed no greater binding than the streptavidin-only controls. (C) Dot blot. Full-length PspA was dot blotted in a 1:3 serial dilution series to a nitrocellulose membrane. The membrane was blocked, overlaid with recombinant human or bovine lactoferrin (LF) or human transferrin (TF), and developed with NBT after incubation with AP-conjugated streptavidin. Only human lactoferrin showed significant binding to PspA.
Although human transferrin bound well to the surface of M. catarrhalis (binding intensity 30 times above the control; data not shown), it failed to bind to the surface of S. pneumoniae, having a fluorescent signal intensity identical to that of the bacteria treated with streptavidin alone (Fig. (Fig.5A5A and B). This supported earlier studies showing no association of radiolabeled transferrin with the bacterial cells (28). The lack of association with PspA was confirmed by dot blot analysis (Fig. (Fig.5C).5C). Thus, in contrast to strains of Neisseria and Moraxella, S. pneumoniae lacks the ability to bind human transferrin.
DISCUSSION
In the original demonstration of Hammerschmidt et al. that lactoferrin binds to PspA (28), the lactoferrin used was purified from human milk. Because of the well-known difficulty of purifying proteins from complex fluids such as milk, our results with recombinant human lactoferrin are an important confirmation of the earlier report that PspA binds lactoferrin. In our fluorescence studies, lactoferrin binding was detected at concentrations down to 1 μg/ml, which is the concentration present in normal serum (8), but 50% saturation of binding to the bacterial surface required approximately 40 μg/ml. Lactoferrin is a known acute-phase protein released in large amounts during inflammation by polymorphonuclear cells. Thus, the concentrations used for half-saturation may very well be within the concentration range seen in an infected individual or locally at a site of inflammation.
The binding of lactoferrin to the bacterial surface was totally dependent on the expression of PspA. Hammerschmidt et al. suggested that lactoferrin did not bind to PspC in Western blots (28). The reactivity of lactoferrin with PspA but not PspC was confirmed in this study using both Western and dot blot analyses with recombinant PspA and PspC fragments purified from E. coli. To assess binding as it occurs under native conditions on the bacterial surface, we used mutants lacking expression of PspA and/or PspC. Using these mutants, we found that lactoferrin binding to PspA and PspC mutants of pneumococci was dependent on the expression of PspA, but the presence of PspC was neither sufficient nor necessary for lactoferrin binding. Hammerschmidt et al. also proposed a second low-affinity interaction of lactoferrin with pneumococci (28): evidence for this was not observed in our study.
The binding of lactoferrin and of the antibodies to PspA and PspC indicated that the expression of both proteins was, in contrast to phosphoryl choline, not uniform over the pneumococcal surface. The most common staining pattern showed that PspA was localized mainly along the lateral body of the bacteria, with little binding at the poles or between bacterial cells. Faint staining was observed near the poles of the most recent cell division, but not near the poles of the preceding cell division. A minority of cells displayed staining between cells or at the poles. On close examination, these cells were irregular in shape and appeared to be in an early stage of cell division. The localized staining may indicate either a loss of surface-expressed PspA over time or the fact that proteins like PspA and PspC may be secreted to the surface in a localized manner. It is interesting that production of the cell wall of pneumococci was limited to a thin growing zone in the lateral body of the bacterial cell (18). Whether PspA and PspC are secreted in conjunction with the production of cell wall remains to be determined.
Sequence homology data (31) and serologic cross-reactivity (41; M. C. V. Coral, N. Fonseca, E. Castaneda, J. L. Di Fabio, S. Hollingshead, and D. E. Briles, submitted for publication) have led to the classification of PspAs into different families and clades. Although PspA is one of the more variable gene products in pneumococci, genetically variable PspAs from different strains can still display some cross-reactivity and can elicit broadly cross-protective antibodies (14, 39, 41, 52). In this report, we show that lactoferrin binds to PspAs from both of the major PspA families. Thus, although PspA is highly variable between strains, there are apparently conformationally conserved regions of the molecule that are responsible for lactoferrin binding. One interpretation of the conservation of lactoferrin binding among strains expressing very variable PspAs sequences is that lactoferrin binding is important and beneficial for the bacteria.
The PspA molecule has been divided into three distinct regions based on its sequence. It has an N-terminal α-helix-rich domain, which is suggested to form a coiled-coil structure similar to that of many gram-positive fibrillar surface proteins. This is the most variable domain of the protein and is exposed on the surface of the cell (31, 38). C-terminal to the α-helical domain is the proline-rich domain, which is known to span the cell wall of pneumococci (32). C-terminal to the proline-rich region is the repeat region that forms a choline-binding site that anchors PspA to the cell wall. Using recombinant fragments of family 1 and family 2 PspAs, we were able to show that lactoferrin binds to the carboxy end of PspA's α-helical region. Lactoferrin bound to full-length PspA from both strains in dot blot and Western blot analyses, consistent with the results using whole bacteria. When investigating binding to the different fragments, we observed that no binding could be detected to fragments constituting the first 115 amino acids of the N-terminal region. Thus, lactoferrin binds to the same general region of PspA that has been found to be most important in eliciting cross-protective immune responses (38, 52).
S. pneumoniae was shown to bind human lactoferrin with higher intensity than the bovine protein. This confirmed the results by Hammerschmidt et al. (28). Although this earlier study did not investigate direct binding of bovine lactoferrin to the pneumomcoccal surface or to purfied PspA, it did show that bovine lactoferrin could not inhibit binding of human lactoferrin to whole S. pneumoniae. A similar situation exists with M. catarrhalis and Moraxella bovis, which have the highest affinity for the lactoferrin of their natural host (10, 27). The fact that each of these three bacterial species recognizes the lactoferrin of its host more strongly than that of an unrelated mammal argues that the ability of each species to bind lactoferrin is important to its ability to colonize or infect its hosts.
Most infections start at the mucosal surface and require that the infecting pathogens have the ability to assimilate nutrients for survival and growth at the site of infection and that they also have ways to effectively evade the host immune system. Binding of lactoferrin may serve both these purposes. Lactoferrin binding has been documented for numerous bacterial species as a way to acquire iron at the site of infection (9, 10, 24, 25, 42, 47). Although iron utilization by S. pneumoniae has not been extensively studied, it is known that pneumococci do not produce siderophores during invasive infection but can utilize hemin and hemoglobin as iron sources in the circulation (50). The means by which the pneumococcus acquires iron at the mucosal surface is less well understood, but there is evidence that it cannot use either lactoferrin or transferrin as an iron source (50). If pneumococci do not use lactoferrin to acquire iron, it must play some other role in human infections.
Lactoferrin has also been shown to inhibit complement activation and to depress immune activity (21, 33, 34, 37, 54). Human tear lactoferrin was shown to block the assembly of the C3 convertase of the classical pathway, probably through interactions with complement factor C2 (34, 54). The results of complement inhibition are, however, conflicting, as there are also reports claiming that lactoferrin binding to bacterial surfaces will enhance complement activation. The modulatory effects of lactoferrin may thus depend on the bacterial surface, the way it is bound, and the environment where activation occurs (43). PspA has been shown to inhibit complement activation in vivo (53). Infection with PspA-negative S. pneumoniae caused higher levels of complement activation in the serum of mice than infection with bacteria carrying PspA on the surface. Moreover, PspA-negative pneumococci are cleared more rapidly from the circulation of mice than those expressing PspA. Binding of lactoferrin by PspA may be a mechanism for pneumococci to inhibit complement activation. It may also be a way to subdue the immune system through the immune-suppressive effects inherent in the lactoferrin molecule (21, 37).
Finally, lactoferrin receptors are known to exist on host cells and may play a role in pneumococcal adherence by allowing lactoferrin to form a bridge between the bacteria and host cells. A similar situation has recently been described for complement protein C3 and its binding to PspC on the pneumococcal surface. This interaction caused increased binding to host epithelial cells (49). Further studies aim at understanding the significance of lactoferrin binding to PspA and its overall role in S. pneumoniae infections.
ACKNOWLEDGMENTS
We acknowledge Catharina Svanborg for her input, interest, and support of these studies; Susan Hollingshead for input in the study and generously providing us with some of the cloned PspA fragments; Beth Ralph and Xinping Wu for help with producing some of the cloned PspA fragments; and Jason Caldwell for help with producing fragment JAS218. We also acknowledge William Benjamin for input in the study, Janet Yother for sharing her earlier experiences looking at PspA with fluorescent techniques, and Flora Gathof, whose handling of the administrative details greatly facilitated this study.
This study was supported by the Swedish Cancer Society (A.H.) and grants AI21548 and Hl54818 (D.E.B.) and AI43653 (L.S.M.) from the National Institutes of Health.