| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Microbiology, Swedish University of Agricultural Sciences, SE-750 07 Uppsala, Sweden
Correspondence Lars Frykberg lars.frykberg{at}mikrob.slu.se
Received February 18, 2004
Accepted June 25, 2004
A phage display library made from Staphylococcus aureus DNA was sorted against a central venous catheter (CVC) that had been removed from a patient 2 days after insertion. After the first panning, approximately 50 % of the clones encoded proteins known to interact with mammalian proteins. After the second and third pannings, fibrinogen-binding and ß2-glycoprotein I (ß2-GPI)-binding phage particles were clearly dominating. Proteins adsorbed to different CVCs were investigated using specific antibodies. Among the proteins probed for, fibrinogen was most abundant, but, interestingly, ß2-GPI was also detected on all tested CVCs.
 |
INTRODUCTION |
|---|
 TOP INTRODUCTION METHODSRESULTS AND DISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
S. aureus encodes a large number of proteins that interact with host proteins, so-called receptins (Kronvall & Jönsson, 1999). Many are well characterized, but additional receptins have been predicted from the genome sequence (Kuroda et al., 2001). Several plasma proteins promote S. aureus adherence to polymeric surfaces, i.e. fibrinogen (Fg), fibronectin (Fn), laminin, vitronectin, thrombospondin and von Willebrand factor (vWf) (Herrmann et al., 1988, 1991, 1997; Vaudaux et al., 1984; Fuquay et al., 1986). Studies of S. aureus adherence to biomaterial coated with various plasma proteins have given valuable information on the host proteins with which S. aureus interacts. However, information on which receptin(s) mediates the adherence is limited since S. aureus encodes several receptins with similar binding specificity. In addition, such studies depend on the bacterial growth conditions and growth phase and may not totally reflect the in vivo situation.
Since its introduction by Smith in 1985 (Smith, 1985), phage display has been a very useful method for identifying protein–protein interactions. A foreign protein is displayed at the phage surface as a fusion to one of the phage coat proteins. Usually the minor (pIII) or major (pVIII) coat protein is used for this purpose. To minimize disturbance of phage functions, systems that allow fusion to only some of the coat proteins, e.g. phagemid vectors, are frequently used. Insertion of foreign DNA into gene-VIII-based phagemid vectors results in multivalent display, but the number of displayed polypeptides will vary considerably for different polypeptides (Malik et al., 1996, and references therein). Phage libraries made from bacterial DNA consist of phage particles that together theoretically display all proteins encoded by the bacterial genome. Simultaneously, the genetic information for the displayed polypeptide is contained in the phagemid vector inside the phage particle. Such libraries can be used for identification of receptins by panning against various ligands (Jacobsson & Frykberg, 1995, 1996, 1998). Here, we have sorted an S. aureus library against a central venous catheter (CVC) removed from a patient 2 days after insertion.
|
METHODS |
|---|
|
TOPINTRODUCTIONMETHODSRESULTS AND DISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Phage library.
The library used is described by Bjerketorp et al. (2002). It was constructed from genomic DNA of S. aureus strain Newman, consisted of 1 x 107 clones and had a titre of 1.5 x 109 c.f.u. ml–1. The library was made in the gene-VIII-based phagemid vector pG8SAET (Jacobsson & Frykberg, 2001), which should result in multivalent display of the foreign polypeptides on the phage surface. The vector contains a screening tag (E-tag) in reading frame with gene VIII but out-of-frame with the signal peptide. The cloning site is located just in front of the E-tag and hence the expression can be restored if an inserted DNA fragment, with an open reading frame, corrects the frameshift. Expression of the E-tag is used to follow enrichment for putative correct clones.
Escherichia coli strain TG1 was grown in Luria–Bertani (LB) broth or on LA plates (LB broth supplemented with 1.5 % agar). Ampicillin (50 µg ml–1) was added when appropriate (LA-Amp). All bacterial incubations were done at 37 °C.
Pannings.
A 400 µl portion of the phagemid library was mixed with casein (Sigma) at a final concentration of 100 µg ml–1 and incubated with a 
5-mm-long piece of CVC removed from a patient 2 days after insertion. After panning at room temperature (RT) with shaking for 4 h, the CVC was washed thoroughly with PBST and bound phages were eluted with 400 µl elution buffer (0.05 M sodium citrate, 0.15 M NaCl, pH 2.0). The eluate was neutralized with 50 µl 2 M Tris-buffer, pH 8.7, and 0.1–50 µl was added to 25 µl stationary phase E. coli TG1, together with LB broth to a final volume of 200 µl. The infection proceeded for 20–30 min at RT before the suspension was spread on LA-Amp plates. The resulting colonies were screened (see below) and infected with helper phage R408 (Promega) for production of enriched phage stocks for repanning, as described by Jacobsson et al. (2003). Repannings were carried out as described above, but for only 2 h.
Panning against in vitro coated microwells (Maxisorp; Nunc) was done as described by Jacobsson & Frykberg (1998). Wells were coated for 1 h with human plasma (Uppsala University Hospital, Sweden) diluted 20 times in coating buffer (50 mM NaHCO3, pH 9).
Individual phage stocks were prepared by R408 infection of bacterial cultures of clones Nbm21, Nbm29 and Nbm69 (Table 1). The titre of each phagemid stock was adjusted to 108 c.f.u. ml–1. Pannings with these phage stocks against 5 mm pieces of CVCs (either unused, or removed from patients after 6, 9, 13 or 18 days) or against pure ligands in microwells were performed essentially as described above. Human Fg was obtained from IMCO and human IgG was purchased from Kabi Vitrum. Recombinant human vWf (rvWf) was a kind gift from Dr Å. Ljungh, Lund, Sweden (originally from Professor F. Dorner, Vienna). The rvWf has been shown to have an activity comparable to human plasma-derived vWf (Fischer et al., 1997). ß2-Glycoprotein I (GPI) was acquired from ICN (sold under the alternative name apolipoprotein H). Fg, IgG, rvWf or ß2-GPI were coated in microwells, at a concentration of 5 µg ml–1, in order to establish the binding capacity of the individual phage stocks.
|
Screening and sequencing of phagemid clones.
After each round of panning, colonies were transferred to nitrocellulose membranes (Schleicher & Schuell) and subsequently screened for expression of the E-tag with anti-E-tag antibodies (Amersham Biosciences) as described by Jacobsson et al. (2003). Phagemid DNA from E-tag-positive clones was prepared with the Qiagen Miniprep kit according to the manufacturer's instructions. The inserted DNA fragments were sequenced using the DYEnamic ET Terminator Cycle Sequencing Premix kit (Amersham Biosciences) and the samples were analysed using the ABI 377 DNA Sequencer (Perkin Elmer) according to the manufacturer's instructions. The Vector NTI computer software (Informax) was used for handling of the sequences obtained.
Analysis of proteins adsorbed to the ex vivo biomaterials.
Pieces of CVCs (5 mm) were boiled in 40 µl sample buffer (62.5 mM Tris/HCl, pH 6.8; 10 % glycerol; 2 % SDS; 5 % ß-mercaptoethanol; and 0.01 % bromophenol blue) for 5 min and the released proteins were subjected to SDS-PAGE using the PhastSystem with 4–15 % gels (Amersham Biosciences). For analysis of ß2-GPI, pieces of the same CVCs were boiled in a non-reducing buffer (composition as above, but without the ß-mercaptoethanol), since this protein contains 11 disulfide bridges (Kato & Enjyoji, 1991), and the monoclonal antibodies only reacted weakly with the reduced form (unpublished observation). Proteins were diffusion blotted (PhastSystem at 65 °C for 40 min) onto nitrocellulose membranes (ECL; Amersham Biosciences), which were treated with the Quentix Western Blot Signal Enhancer kit (Pierce). Thereafter the membranes were blocked for 30 min in PBST containing casein at a concentration of 1 mg ml–1. Proteins were detected using horseradish peroxidase (HRP)-labelled antibodies against Fn (DakoCytomation), Fg (DakoCytomation) or IgG (ICN). ß2-GPI was detected using an anti-ß2-GPI monoclonal antibody (Chemicon) followed by HRP-labelled anti-mouse antibodies (Amersham Biosciences). Bound HRP-labelled antibodies were visualized using BM Blue POD substrate (Roche).
|
RESULTS AND DISCUSSION |
|---|
|
TOPINTRODUCTIONMETHODSRESULTS AND DISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
From each panning, approximately 50 E-tag-positive clones were isolated and the DNA sequences of the inserts were determined. Clones were considered to represent an authentic interaction if overlapping inserts of a gene were found in two or more clones, or if a clone encoded a known receptin. After the first panning, approximately 50 % of the clones were correct. In the second and third pannings, only one background clone was found. However, despite the large number of clones investigated, no novel receptin was identified.
Receptins identified
The receptins identified were: coagulase (Coa), an Fg- and prothrombin (PT)-binding protein with the ability to coagulate plasma; protein A, an IgG- and vWf-binding protein; Efb, a Fg-binding protein; FnbpA/B, two similar Fn- and Fg-binding proteins; and Sbi, an IgG- and ß2-GPI-binding protein (Table 1). The binding domains in these proteins have been mapped, and therefore it is possible to deduce which binding property is displayed by a selected phage particle. However, it is possible that additional, so far unidentified, binding domains may exist or that a domain might bind more than one protein. The majority of isolated phage encoded either Fg-binding or ß2-GPI-binding polypeptides. Part of the coa gene was found in 35 clones, all encoding the Fg-binding repeat region, and only in one clone was the PT-binding domain also present. Altogether, 74 clones contained parts of the sbi gene. Interestingly, 51 of these clones encoded only the part known to bind ß2-GPI, while the others encoded polypeptides with the ability to bind both IgG and ß2-GPI.
Protein Sbi contains a signal peptide and has been found associated with the surface of S. aureus. However, the protein does not contain the typical LPXTG motif and the transmembrane domain found in most cell-wall-bound proteins, suggesting a different type of association to the surface. Still, similar to cell-wall-spanning domains, Sbi contains a repetitive domain with proline residues repeated every fifth amino acid (Zhang et al., 1998, 1999).
Panning against microwells coated in vitro with plasma
The results from the panning experiments against ex vivo biomaterials suggest that ß2-GPI is easily accessible for Sbi displayed on the phage surface. One possible explanation for this could be that ß2-GPI is deposited later than Fg and Fn. As a comparison, plasma proteins were coated onto microwells for 1 h, and the phage display library was panned in two cycles against the wells. Following the first and second pannings, 24 E-tag-positive clones were analysed. After the first panning, nine clones contained the coa gene, five the fnbB gene, three the spa gene (encoding protein A) and one the efb gene, while six clones were considered as background. After the second panning, only clones encoding Fg-binding proteins were found. These results suggest that Fg is the main protein recognized by S. aureus proteins after a short in vitro coating.
Proteins released from ex vivo biomaterials
The high number of clones with ß2-GPI-binding capacity suggests that either ß2-GPI is a major component on ex vivo biomaterial, or it is more accessible than other adsorbed proteins. Proteins were released from CVCs removed from patients after different implantation periods and analysed by SDS-PAGE and Western blotting. Due to the limited amount of ex vivo CVC material available, the CVC removed after 2 days was not included in this experiment. To detect relevant proteins, antibodies specific for Fg, Fn, IgG and ß2-GPI were used. Fg was the major component released from all CVCs (Fig. 1b), but Fn was also detected in all samples (data not shown). IgG was detected in some of the samples, and relatively weak signals were obtained (Fig. 1c). Interestingly, ß2-GPI was detected in all samples, although the signal was weaker than for Fg (Fig. 1e). However, the signal strength in a Western blot is semi-quantitative and depends on a number of different factors besides the amount of protein, e.g. the quality of the antibodies used, the ability of the antibodies to recognize the target protein in a denatured state, as well as the efficiency with which the protein is transferred to the nitrocellulose membrane.
|
To our knowledge it has not previously been shown that ß2-GPI is deposited on biomaterial. ß2-GPI is a single-chain 50 kDa protein, and although present in relatively high amounts in serum (0.2 mg ml–1), the biological function of ß2-GPI is not completely clear. It binds to negatively charged substances such as heparin and phospholipids, and is structurally related to regulators of the complement system (Kato & Enjyoji, 1991; Steinkasserer et al., 1991). ß2-GPI has been reported to have several biological functions, such as promoting clearance of liposomes and maybe foreign particles from the circulation, but also effects on the coagulation system have been reported (Chonn et al., 1995; Nimpf et al., 1986). The mechanism for the anticoagulating activity of ß2-GPI was suggested in a recent publication, which showed that ß2-GPI binds to factor XI and inhibits its activation (Shi et al., 2004).
Panning with phage stocks against ex vivo biomaterials and pure proteins
Considering bacterial adhesion to biomaterial, the total amount of protein deposited is less important than its accessibility on the surface. To confirm the accessibility of Fg, IgG and ß2-GPI on different CVCs, phage stocks were made from three relevant clones. After adjusting their titres, the binding to CVCs as well as to pure proteins immobilized in microwells was investigated. Although CVCs were derived from different individuals and the insertion time varied, this did not seem to greatly affect the ability of the phage clones to bind to the CVCs. As expected from the Western blot data, the Fg-binding clone Nbm69 demonstrated good binding to all CVCs, but 100 times more phage particles bound to pure Fg (Fig. 2a). The binding capacity of Nbm29, encoding part of protein A, varied to some extent between CVCs, but it was much higher than to an uncoated CVC (Fig. 2b). The phage particles also bound to pure IgG, but no binding to rvWf was observed (Fig. 2b). The lack of binding to rvWf may be caused by differences between rvWf and vWf purified from serum; alternatively, the conditions in our experiments might not be ideal to detect this interaction. The sbi-derived Nbm21 phage bound well to ß2-GPI but, as expected, not to pure IgG (Fig. 2c). Interestingly, these phage particles also bound well to all tested CVCs (Fig. 2c), showing that ß2-GPI is accessible on all tested CVC surfaces, further strengthening our findings regarding ß2-GPI in the Western blot (Fig. 1e). A negative control phage stock containing the vector pG8SAET showed no binding to either of the CVCs (data not shown).
|
Even though the signal strengths in the Western blots of proteins released from the CVCs may not show the exact relation between protein amounts, the signals imply that Fg is present in higher amounts than ß2-GPI (Fig. 1). However, the Fg-binding and ß2-GPI-binding phage particles bound to the CVCs in comparable numbers (Fig. 2), thus demonstrating that it is the surface accessibility rather than the total amount of protein that is of importance for the phage binding. The binding of phage particles to pure Fg was approximately 25 times higher than to pure ß2-GPI, and since similar numbers of Nbm69 and Nbm21 phage particles bound to the CVCs, this suggests that ß2-GPI is actually more accessible than Fg on the CVCs.
Concluding remarks
When Dickinson et al. (1995) investigated the adherence of S. aureus to in vitro coated glass surfaces under flow, it was concluded that adherence to Fg or plasma almost exclusively depended on the Fg-binding protein clumping factor (CF). Francois et al. (2000) also demonstrated the importance of CF for S. aureus binding to Fg-coated polyvinyl chloride (PVC) tubing. However, it was also demonstrated that Fg deposited on the PVC tubing in vitro was much more efficient in promoting S. aureus adherence than the Fg deposited during haemodialysis. In addition, an S. aureus mutant lacking the CF bound in 10-fold higher numbers to the tubing coated during haemodialysis than to control tubing coated with albumin. Similarly, our result from panning against a well coated with plasma in vitro yielded mainly Fg-binding phage. A possible explanation for these differences is the over layering of initially adhered Fg by other plasma proteins on the haemodialysis tubing. When the tubing coated during dialysis was subjected to analysis of the protein content, it was suggested that an unidentified protein(s) different from Fg, vitronectin, thrombospondin and vWf was responsible for the additional binding of S. aureus to the tubing (Francois et al., 2000). Our panning against the CVCs removed from different patients shows that ß2-GPI is an easily accessible component, and may be the unidentified target on the haemodialysis tubing above. Thus the ability to bind ß2-GPI could be one mechanism by which S. aureus adheres to biomaterial.
ß2-GPI has attracted a lot of attention in recent years due to its involvement in autoimmune disease. In antiphospholipid syndrome (APS), autoantibodies are often not directed against phospholipids alone, but against phospholipids bound to ß2-GPI or to ß2-GPI only (McNeil et al., 1990; Galli et al., 1990). Although the exact specificity of antiphospholipids (aPL) antibodies is under debate, it has been suggested that the concept ‘aPL antibodies’ needs to be revaluated since the major target is presumably not phospholipids but ß2-GPI (Li & Krilis, 2003). The cause for the appearance of these autoantibodies is not known, but aPL antibodies are frequently found in conjunction with various infections (McNeil et al., 1991). It has been suggested that a change in conformation of ß2-GPI upon contact with negatively charged phospholipids or surfaces may be responsible for exposure of autoepitopes (Borchman et al., 1995; Chamley et al., 1999; Wang et al., 2000). In this context, the binding of ß2-GPI to biomaterial is interesting. It is known that binding of plasma proteins to polymer surfaces can result in a conformational change of the protein (Barbucci & Magnani, 1994). Perhaps as a consequence, haemodialysis patients were found to have a significantly higher incidence of aPL antibodies compared to control groups (Fabrizi et al., 1999; McIntyre & Wagenknecht, 2003). The finding of ß2-GPI on the surface of CVCs is intriguing, but whether adherence of ß2-GPI to biomaterials is a general phenomenon needs to be investigated.
Sorting an S. aureus phage display library against ex vivo biomaterials revealed several possible interactions. However, this type of experiment identifies proteins that bind to the target regardless of the normal localization and function of the protein. Thus it may be possible to find interactions without biological meaning. Still, all receptins identified in this study contain a signal peptide and are extracellular proteins, thus they could be of biological significance. Two of them, Efb and Coa, are secreted, although a small fraction of Coa seems to be surface associated (Bodén & Flock, 1989; McDevitt et al., 1992). Their primary function probably is not adherence; still the binding of these proteins to biomaterial might serve other biological functions. After the initial adherence of S. aureus to biomaterial, it is conceivable that the Fg binding of Coa could cause a localized coagulation, which could promote biofilm formation on the biomaterial. Interestingly, Efb and Coa were also found when an S. aureus library was sorted against platelets, suggesting that these proteins bind platelets via Fg as a bridging molecule and possibly could aid S. aureus adherence to platelets (Heilmann et al., 2002).
The importance of CF for S. aureus adherence to Fg has been demonstrated in several publications (e.g. Dickinson et al., 1995; Francois et al., 2000), but CF was not found in this study. Most likely, phages that display CF are out-competed by other binding phages, perhaps due to its large ligand-binding domain. Although the major selection should be for binding to the immobilized ligands, other factors may also influence which phage dominate after panning. For example, it is possible that different phagemid clones replicate to different numbers, or that different polypeptides, with the same binding specificity, might be displayed with different efficiency. Ten different clones encoding parts of the coa gene were found after the first and second pannings. Among these, Nbm42 contains almost five complete Fg-binding repeats, but this clone is not found after the second or third pannings. Instead, Nbm69, encoding only two Fg-binding repeats, dominates after the third panning. Clearly factors other than Fg-binding alone are responsible for the dominance of this clone.
Although no new receptin was identified in this study, it was shown that ß2-GPI is deposited on the CVCs. The results also showed that with this kind of phage library, only one panning cycle is needed to obtain a high frequency of correct clones. Further pannings resulted in an even higher frequency of correct clones, but much fewer unique clones. Thus in order to find new interactions, it might be better to make only one panning and investigate a larger number of clones.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPINTRODUCTIONMETHODSRESULTS AND DISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPINTRODUCTIONMETHODSRESULTS AND DISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| INT J SYST EVOL MICROBIOL | J MED MICROBIOL | MICROBIOLOGY | J GEN VIROL | ALL SGM JOURNALS |
