Department of Clinical Microbiology, University Hospital, Uppsala, Sweden.
Interaction of vitronectin with Haemophilus influenzae
Article first published online: 9 JAN 2006
FEMS Immunology & Medical Microbiology
Volume 34, Issue 3, pages 215–219, November 2002
How to Cite
Eberhard, T. and Ullberg, M. (2002), Interaction of vitronectin with Haemophilus influenzae. FEMS Immunology & Medical Microbiology, 34: 215–219. doi: 10.1111/j.1574-695X.2002.tb00627.x
- Issue published online: 9 JAN 2006
- Article first published online: 9 JAN 2006
- Received 10 May 2002, Revised 8 August 2002, Accepted 5 September 2002
- Haemophilus influenzae;
- Extracellular matrix;
Eight strains of Haemophilus influenzae were tested for binding to human vitronectin. All strains adhered to vitronectin-coated glass slides but no binding was detected using soluble vitronectin, suggesting that surface association of vitronectin is a prerequisite. Vitronectin binding was not likely to be mediated by fimbriae as non-fimbriated and fimbriated isogenic strains adhered equally. Adhesion could be blocked by heparin, which is also known to block vitronectin binding to Staphylococcus aureus. However, no blocking was achieved with sialic acid-rich glycoproteins such as fetuin and mucin contrasting with Helicobacter pylori for which sialic acid seems to play an important role. With Streptococcus pneumoniae binding was detected both with soluble and surface-associated vitronectin and could not be blocked by heparin. Our results suggest that H. influenzae, Streptococcus pneumoniae and Helicobacter pylori all use distinct modes to interact with vitronectin.
Haemophilus influenzae is a common coloniser of the human nasopharynx and is associated with both localised and invasive disease. The vast majority of isolates from children are non-encapsulated (non-typable) except a minor fraction which are encapsulated with type b as the predominant serotype. The non-encapsulated isolates are associated with milder disease such as otitis media, sinusitis and bronchitis. Type b isolates on the other hand are associated with severe diseases as sepsis, meningitis and arthritis . Both encapsulated and non-encapsulated strains adhere to epithelial cells in the nasopharynx. This process is mediated both by fimbria [1–4] and by non-fimbrial high-molecular-mass adhesins  as well as lipid-binding proteins . Protein D has also been shown to mediate cell adhesion . The colonisation is often associated with epithelial damage and breakdown of tight junctions whereby the bacteria come in contact with the underlying basement membrane and extracellular matrix (ECM) [2,3,8,9]. We have earlier been able to demonstrate that H. influenzae can interact with basement membrane as well as with distinct matrix components such as laminin, fibronectin and collagen. By acquisition of plasmin(ogen) on the surface the bacterium can also provide itself with a tool for matrix penetration and thereby gain access to deeper tissue layers . A study of non-encapsulated strains isolated from patients with chronic obstructive pulmonary disease showed only high binding of plasminogen in those isolates that had low binding of ECM proteins . However, other studies have found strains that demonstrate a high binding to both plasminogen and ECM proteins [10,12].
Vitronectin is a 75-kDa glycoprotein which is present in many tissues but also in plasma. Vitronectin is, in addition to fibronectin, the main mediator of cell anchoring to ECM. Receptors for vitronectin have been described on many cells, including platelets, endothelial cells and melanoma cells, and have in many cases been defined as integrins reacting with the RGD motif of the vitronectin molecule . Vitronectin also interacts with many bacterial species, for example Staphylococcus aureus, Neisseria gonorrhoeae, hemolytic streptococci groups A, C and G as well as Escherichia coli[14–23]. In this investigation we extend previous knowledge of bacterial interaction with ECM by describing vitronectin binding to H. influenzae.
2Materials and methods
2.1Bacterial strains and culture condition
Eight H. influenzae strains (six nontypable and two of type b) were isolated from clinical laboratory specimens. Strains 770235fim+ (type b fimbriated), 770253fim− (type b nonfimbriated), E1a and R2625 (E1a derivative with deletion of the 28-kDa invasion factor)  were kindly provided by Dr A.L. Smith, Missouri, USA. Eight S. pneumoniae strains were all isolated from clinical laboratory specimens. Strains were kept frozen in broth supplemented with 20% glycerol at −70°C and recultivated for 2–3 days prior to testing. Strains were cultured overnight at 37°C on haematin agar plates in a humid atmosphere containing 5% CO2. For testing, bacteria from the agar plates were inoculated into brain heart infusion (BHI) broth supplemented with 1% IsoVitalex (BBL Microbiology Systems, Cockeyville, MD, USA) and 40 mg/l of haemin (Sigma, St. Louis, MO, USA). After an overnight cultivation, bacteria were collected by centrifugation (3000 rpm 10 min, Tamro-Adapta 90–54401, Labsystems, Helsinki, Finland) and washed twice in phosphate-buffered saline (PBS), pH 7.1.
Bacterial adhesion to vitronectin coated on microscope glass slides, Diagnostica K107 (Danlab, Helsinki, Finland) was tested using bacterial concentrations ranging from 5×107 to 1×109. Vitronectin prepared according to the method of Yatohgo et al.  was kindly provided by Dr Åsa Ljungh, Lund, Sweden. Wells were coated using a procedure earlier described [26,27]. Bovine serum albumin and fetuin (Sigma) were used as negative controls representing unglycosylated and glycosylated proteins, respectively. The adhering bacteria were stained with metylene blue and then visualised in a Nikon Diaphot TDM microscope (Nikon, Tokyo, Japan). Images were captured using a Kappa CF15MC video camera connected to a Kappa MCU control unit (Kappa Messtechnik, Gleichen, Germany). Images were digitised using a Scion LG-3 Frame Grabber card (Scion, Frederick, MD, USA) connected to a Power Macintosh 6100/60 (Apple Computer, Cupertino, CA, USA), using a variant of the public domain software NIH Image version 1.59 as supplied by Scion. The number of bacteria in 10 microscopic fields was determined by particle counting and density slicing, using a macro developed by Mirko Brummer at Department of General Microbiology, Helsinki University, Finland.
In the case of lactoferrin, mannose, fetuin, asialofetuin or BSA bacteria were preincubated for 60 min using a concentration of 0.1 mg ml−1. The RGD containing peptides G-1269 (GRGDSPK)  at 2 mM; A-6677 (RGDSKASSKP) [29,30] at 0.5 mM; as a control peptide HIV gag fragment S-5151 (AQNYPIV)  at 2.0 mM. When blocking with glycosaminoglycans, vitronectin-coated glass slides were preincubated with heparin (5000 IU ml−1) or hyaluronic acid (0.1 mg ml−1) for 60 min. All reagents were purchased from Sigma. The adhesion of bacteria to vitronectin was then evaluated as described above.
2.4Binding of soluble vitronectin
Ligand binding assays were performed essentially as described earlier . After culturing, the bacterial concentration was measured as turbidity at 600 nm using a Pharmacia LKB Novaspec II spectrophotometer (LKB Biochrom, Cambridge, UK), using a standard curve based upon colony forming units. 2.5×109 bacteria suspended in 1 ml PBST were used in each test. 20 ng of 125I-labelled vitronectin was added to each test tube. After 30 min incubation the bacteria were collected by centrifugation at 3600×g for 10 min and the supernatant was removed. The radioactivity in the pellet was measured using a gamma counter (LKB-Wallac Clini Gamma 1272, Turku, Finland). Vitronectin binding was expressed as percent of total radioactivity added.
3.1Bacterial adhesion to immobilised vitronectin
Eight selected H. influenzae strains, including two type b isolates, were tested for their ability to adhere to vitronectin immobilised on glass slides. All strains tested demonstrated a dose-dependent adhesion when using bacterial concentrations in the range of 5×107 to 109 ml−1 (Fig. 1). Parallel control experiments using glass slides coated with bovine serum albumin (BSA) were also performed. Negligible binding to control slides was observed with all strains.
3.2Binding of soluble vitronectin
For comparative reasons the binding of 125I-labelled vitronectin to eight H. influenzae strains was also tested. No significant uptake of vitronectin (<6%) was scored with any strain, suggesting that the immobilisation of vitronectin is a prerequisite for the binding to occur. The S. pneumoniae strains used as controls resulted in uptake values in the range of 10–20% (Fig. 2).
3.3Role for known virulence factors
In order to test the role of fimbriae in vitronectin binding, isogenic strains with and without fimbriae were compared. No significant difference in adhesion values was demonstrated (Fig. 3), suggesting that fimbriae are unlikely to be involved in vitronectin binding. Also the importance of a 28-kDa invasion factor was tested. The knockout mutant strain adhered similarly to the native strain, suggesting that also this factor is of no importance for vitronectin binding.
3.4Blocking experiments with glycoproteins and glycosaminoglycans
To further investigate the nature of vitronectin binding to H. influenzae, blocking experiments were performed (Fig. 4). Blocking agents tested included heparin, lactoferrin and mannose, which are known to block vitronectin binding to S. aureus, as well as two sialic acid-rich glycoproteins known to interfere with the vitronectin binding to Helicobacter pylori. More than 90% blocking was achieved using heparin, whereas only a minor effect was scored with other agents. The inhibitory effect of different concentrations of heparin is presented in Fig. 5. For comparative reasons we also tested the capacity of heparin to block pneumococcal adhesion to immobilised vitronectin. Using a concentration of 5000 IU ml−1 no inhibitory effect could be seen, suggesting a different mechanism for this bacterium (data not shown).
Our results show that a majority of H. influenzae strains have the potential to react with vitronectin in a dose-dependent manner and that this reaction is likely to be of a specific nature. However, the conformation of the vitronectin molecule seems to be important as only surface-associated but not free vitronectin interacts with the bacterial surface. This finding is in contrast to what has been demonstrated for several Gram-positive bacteria. Thus S. aureus as well as several coagulase negative staphylococci are all efficient binders of free vitronectin [18,23,34,35]. Similar findings have been made for streptococci groups A, C and G  and for S. pneumoniae. Also certain Gram-negative bacteria are capable of binding free vitronectin, as demonstrated for E. coli. The nature of vitronectin-binding structures of H. influenzae is at present unclear. Fimbriae can be ruled out, as a nonfimbriated mutant strain binds equally well as the fimbriated parental strain. Neither can the binding be attributed to the 28-kDa invasion factor , as the deletion mutant and parental strains bind equally well. The part of the vitronectin molecule responsible for the binding is most likely the C-terminal heparin-binding region, as heparin is an efficient blocking agent. No blocking was observed with sialic acid-rich proteins such as fetuin and mucin. This is in contrast to what has been reported for H. pylori, where sialic acid plays an important role.
Early in the infection process H. influenzae will bind to the epithelial cell layer of the nasopharynx. This adhesion is mediated both by fimbriae and by nonfimbrial high-molecular-mass adhesion molecules [5,36]. However, the epithelial layer in the nasopharynx will often be disrupted as a consequence of the infection process, giving the bacterium access to the matrix layer below. We have earlier demonstrated that H. influenzae can interact with reconstituted basement membrane as well as with ECM from cultures of human endothelial cells , showing that H. influenzae has the potential to bind the deeper layers of the mucosa. This interaction with ECM and basement membrane includes binding of fibronectin, laminin and various collagens. Thus the capacity to bind vitronectin described here seems logical, as this molecule is also a part of the submucosal layer in many tissues. From our earlier investigations we also know that H. influenzae can gain proteolytic activity from the host by the acquisition of plasmin(ogen) on the bacterial surface and thereby penetrate the matrix layer . It is likely that this property is of importance for generalised infection to occur.
Another interesting aspect of the bacterial vitronectin binding is the effect on cellular invasion as demonstrated for N. gonorrhoeae. Certain serotypes of N. gonorrhoeae expressing the Opa50 protein can specifically bind vitronectin. This binding will trigger bacterial internalisation into epithelial cell lines. Blocking of integrin functions by RGDS peptides or by antibodies specific to αvβ5 or αvβ3 results in abolition of vitronectin-triggered bacterial internalisation. Neisseria meningitidis can also use vitronectin as a bridge for cellular attachment , but no information is available telling us whether this will also lead to internalisation.
Finally bacterial binding of vitronectin could play a protective role due to the interactions with the complement C5b–9 complex . Binding of vitronectin would enable protection from bacterial lysis following opsonisation. However, no clear-cut evidence for such a protective role of vitronectin exists at present. Further work, including purification and cloning of the receptor structure, will be needed to address different functional aspects.