OUP user menu

CNE, a collagen-binding protein of Streptococcus equi

Jonas Lannergård , Lars Frykberg , Bengt Guss
DOI: http://dx.doi.org/10.1016/S0378-1097(03)00222-2 69-74 First published online: 1 May 2003


Streptococcus equi subspecies equi is an important horse pathogenic bacterium causing a serious disease called strangles. Using bioinformatics we identified a gene denoted cne (gene encoding collagen-binding protein from S. equi) coding for a novel potential virulence factor of this species called protein CNE. The protein is composed of 657 amino acids and has the typical features found in cell surface-anchored proteins in Gram-positive bacteria. CNE displays amino acid sequence similarities to the previously well-studied collagen-binding protein CNA from Staphylococcus aureus, a proven virulence factor in septic arthritis. Based on similarity to CNA the structure of the mature CNE protein can be divided into an N-terminal A domain and a C-terminal B domain. The highest similarity between CNA and CNE is found in the A domains. The A domain in CNA is known to be the collagen-binding domain. Two parts of cne were amplified using polymerase chain reaction (PCR) and ligated into an expression vector, and recombinant CNE proteins were produced in Escherichia coli. The purified CNE proteins were shown to display collagen-binding activity in a Western ligand blot and to inhibit collagen binding to cells of subsp. equi and to CNE-coated microtitre wells. Furthermore, the A domain of CNE was sufficient for binding collagen, and was shown to compete for the same site on collagen as CNA in inhibition studies. Using PCR, the cne gene was detected in all studied strains of subsp. equi and S. equi subsp. zooepidemicus.

  • Strangles
  • Collagen-binding
  • Cell surface proteins
  • Virulence factors
  • Streptococcus equi
  • Staphylococcus aureus

1 Introduction

Streptococcus equi, belonging to the Lancefield group C, comprises two subspecies: S. equi subsp. equi and subsp. zooepidemicus. Subsp. equi causes a worldwide-distributed and serious respiratory disease in horses called strangles. This subspecies is essentially confined to horses, in contrast to subsp. zooepidemicus, which may also be found in a wide range of other animals. The latter subspecies is considered as an opportunistic commensal, often occurring in the upper respiratory tract of healthy horses, but can also cause disease e.g. in the uterus and wounds. In a recent review on the molecular basis of S. equi infection by Harrington et al. [1] the virulence factors were grouped into three broad categories based on their proposed functions. These were the factors that promote bacterial adherence, those that contribute to immune evasion and those that are involved in nutrient acquisition, although to group individual factors into a specific category is often difficult since some factors have multiple functions, which could contribute both to e.g. adhesion and immune evasion. The M-like proteins [2,3], the fibronectin-binding proteins [46] and the protein G-related proteins [7] are examples of factors in S. equi with multiple functions.

Collagens are a conserved group of proteins found in the extracellular matrix of mammalians, and the ability to adhere to collagen has been shown to be an important feature in some S. aureus infections [8,9]. The present work describes a novel collagen-binding S. equi cell-surface protein called CNE.

2 Materials and methods

2.1 Bacterial strains, plasmids, and growth conditions

S. equi subsp. equi strain 1866 was obtained from NordVacc Läkemedel, Stockholm, Sweden. Other subsp. equi (n=12) and zooepidemicus (n=13) strains used in this study were obtained from the National Veterinary Institute, Uppsala, Sweden. The Escherichia coli strain ER2566 and the plasmid vector pTYB4 were obtained from New England Biolabs, MA, USA (NEB). Streptococcal strains were grown on horse blood agar plates or in Todd–Hewitt broth (Oxoid, Basingstoke, Hampshire, UK) supplemented with 0.5% yeast extract. E. coli was cultured in Luria–Bertani (LB) broth, supplemented with ampicillin (100 µg ml−1) or on LAA-plates [LB broth with ampicillin and agar (15 g l−1)]. Incubations were, if not indicated otherwise, at 37°C.

2.2 Proteins and reagents

Collagen (bovine collagen S, type I) was obtained from Boehringher Mannheim, Germany. The NEB IMPACT™T7 system was used to produce and purify recombinant CNE proteins.

Recombinant CNA of S. aureus was produced from clone pGEX 1.1 [10], an E. coli clone expressing amino acids 30–403 of the A domain of CNA in fusion to the glutathione-S-transferase gene of the vector pGEX-2T (Amersham Pharmacia Biotech). After expression of the fusion protein, in this study called CNA A, the fusion protein was affinity-purified using gluthatione–Sepharose 4B (Amersham).

125I was obtained from Amersham, and used to iodinate 50 µg of each selected protein according to the Iodo-Beads labelling method as described in the manual from the manufacturer (Pierce, Rockford, IL, USA). The specific activity of labelling expressed as counts per minute (cpm) was ~4×106 cpm µg−1 for collagen and ~6×106 cpm µg−1 for CNE L.

2.3 DNA sequencing and homology studies

The nucleotide sequences of the inserts in pCNE L and pCNE S were determined using Dyenamic™ ET Terminator Cycle Sequencing Premix kit, a model 377 Perkin Elmer DNA sequencer and software from the Vector NTI suite (Informax, Bethesda, MD, USA). The NCBI BLAST2 programme was used to analyse sequence similarities (http://www.ncbi.nlm.nih.gov/blast/bl2seq/bl2.html) and the EMBL tool Radar (http://www.ebi.ac.uk/Radar/) was used to find internal repeats. To analyse the structure and properties of CNE, the following web-based tools were used: ProtParam (http://us.expasy.org/tools/protparam.html), DAS (http://www.sbc.su.se/~miklos/DAS/) and SignalP (http://www.cbs.dtu.dk/services/SignalP/).

The nucleotide sequence accession number of the cne gene is AY193773.

2.4 Construction of clones expressing collagen binding

To express and purify CNE, two different constructs were made, pCNE L (large) encoding amino acids 27–615 and pCNE S (small) encoding amino acids 27–328 of CNE. The two constructs were made as follows. Primer OCNE1:5, 5′-CATGCCATGGCAACTAATCTTAGTGACAACAT-3′ combined with primer OCNE3:3, 5′-CCGCTCGAGCTTGTAGCTTGGGTTAAGGGTGT-3′ or OCNE2:3, 5′-CCGCTCGAGAAAGCTGGTATAGCGACTGCCAT-3′ respectively, were used to polymerase chain reaction (PCR)-amplify the corresponding DNA fragments using subsp. equi strain 1866 DNA as template. The underlined nucleotide sequences of the primers hybridise to the cne gene and the introduced restriction endonuclease cleavage sites (NcoI and XhoI) are written in bold. Both PCR amplifications were performed using ReadyToGo™ PCR beads (Amersham) using the following programme: step 1, 95°C 30 s; step 2, 46°C 15 s; and step 3, 72°C 2 min, repeated in 25 cycles. The PCR products were digested with NcoI and XhoI and ligated into the pTYB4 vector (NEB) previously digested with the same restriction endonucleases. After ligation, the plasmids were electrotransformed into E. coli ER2566 and spread on LAA plates. Transformants were transferred to NitroCellulose (NC) membranes and expression induced by incubation on LAA plates with 0.4 mM isopropyl thiogalactose (IPTG). Clones expressing collagen binding were identified by colony screening using 125I-labelled collagen. Of these, two clones (pCNE L and pCNE S), after DNA sequencing of the inserts, were used to produce recombinant proteins according to the manufacturer's (NEB) recommendations.

2.5 SDS–PAGE and Western ligand analysis

The purified recombinant CNE proteins (CNE L and CNE S) were subjected to SDS–PAGE analyses under reducing conditions using the PhastSystem (Amersham) with precasted 8–25% gradient gels. To analyse the collagen-binding activity, the proteins were diffusion-blotted onto an NC membrane for 30 min at 65°C. The NC membrane was subsequently blocked with phosphate-buffered saline (PBS) supplemented with Tween 20 (0.05% v/v) (PBS-T) and casein (0.5% w/v) for 1 h at RT. The NC membrane was incubated with 125I-labelled collagen overnight at 4°C. After extensive washing with PBS-T the NC membrane was exposed to Biomax MS (Kodak) film for 48 h.

2.6 Collagen-binding analysis

Streptococcal cells were grown overnight, harvested and washed with PBS and the OD600 was adjusted to 1, corresponding to 2×108 CFU ml−1. In the collagen-binding assay, 100 µl cell suspension was transferred to an Eppendorf tube, and 100 µl 125I-labelled collagen (~500 000 cpm) with casein (0.1% w/v) were added, and the volume was adjusted to 0.5 ml with PBS-T. After 2 h incubation, the cells were pelleted, washed three times in PBS-T, and the radioactivity bound to the cells measured in a γ-counter.

In a similar assay, unlabelled collagen was added in different concentrations (0.6–192 µg ml−1) to 10 µl cell suspension, together with 100 µl of 125I-labelled collagen and the volume was adjusted to 0.5 ml with PBS-T. After incubation, the radioactivity bound by the cells was measured as described above. All incubations with cells were performed using an end-over-end mixer at RT.

2.7 CNE binding to collagen on bacterial cells

Cell suspensions (OD600=1) of 100 µl were mixed with collagen (0.03–7.5 µg ml−1) in 0.5 ml PBS-T with casein (0.1% w/v). After 1 h of incubation at RT the cells were pelleted, washed twice in PBS-T and resuspended in PBS-T, whereupon 125I-labelled CNE L (~100 000 cpm) and casein (0.1% w/v) were added. Cells were pelleted after 1 h at RT, washed twice in PBS-T, and the radioactivity bound to the cells was measured. In a control experiment, the collagen was replaced by human serum albumin (HSA) at the same concentrations.

2.8 CNE S and CNA A inhibition analysis

The ability of soluble CNE S to inhibit collagen binding to immobilised CNE S was studied. Microtitre wells (Immulon 2, Removawell, Dynatech Laboratories, USA) were coated with 100 µl CNE S (2 µg ml−1) for 1 h at RT. After washing with PBS-T, the wells were blocked for 30 min with casein (0.1%) in PBS-T. CNE S (0.75–50 µg ml−1) was incubated with 125I-labelled collagen (~20 000 cpm) for 30 min at RT. The mixed samples were added to the coated wells and incubated for 1 h at RT. After washing with PBS-T the radioactivity bound to the wells was measured. The same type of experiment was performed with CNA A (0.046–48 µg ml−1) instead of CNE S, and as negative control with HSA (15–500 µg ml−1).

3 Results and discussion

3.1 Identification of the cne gene

The genome of S. equi has been determined by shotgun sequencing and finishing/gap closure is in progress (http://www.sanger.ac.uk/Projects/S_equi/). Using published sequences of virulence factors or potential virulence factors from pathogenic streptococci and staphylococci, it is possible to screen the genome of S. equi for the presence of similar genes. Using the software programme BLAST, we found an open reading frame (ORF) in the S. equi genome encoding a protein similar to the collagen-binding protein CNA of S. aureus [11]. The 1971 nucleotide-long ORF in S. equi is preceded by promoter and a ribosomal binding sequence and can therefore be considered a functional gene, here given the name cne. The cne gene encodes a protein of 657 amino acids termed protein CNE (Fig. 1). Since CNE display typical structures found in other Gram-positive cell surface proteins [12], it is suggested to be a cell surface protein. Computer analysis confirms that CNE contains a typical N-terminal signal sequence (amino acids 1–27) and a hydrophobic C-terminal transmembrane region (amino acids 628–645). Furthermore, the transmembrane region is preceded by an LPXTG motif (LPDTG), a common cell wall-anchoring motif found in surface proteins of Gram-positive bacteria [12]. After export to the cell surface, which results in the removal of the signal sequence and cleavage in the LPXTG motif, the deduced mature cell wall-anchored protein should consist of 592 amino acids with a molecular mass of ~66.7 kDa.


Schematic presentation of CNA and CNE. The signal sequences (S), A domains (A), B subunits (B1.1–B3.2), wall-spanning region (W) and membrane-spanning regions (M) are indicated. Vertical lines indicate regions with high amino acid similarities between the two proteins, and the percentages of identical residues between the regions are presented. The two horizontal bars represent recombinant proteins CNE L and CNE S. Figures in parentheses refer to amino acid positions in CNE.

3.2 Similarity of CNE to CNA of S. aureus

To analyse the similarity between the deduced amino acid sequences of CNE and CNA, various computer programmes were used. The signal sequences and the wall- and membrane-spanning C-terminal parts show no significant similarities. The A domain of CNA, where the collagen-binding site is located [11], shows the most significant similarities (50% identities, 67% similarities) to CNE (Fig. 1). Based on these similarities, the A domain in CNE was defined as ranging from amino acid 28 to 323. The 503 amino acid long A domain of CNA is considerably larger than the corresponding domain of 296 amino acids in CNE. In CNA the amino acid sequence critical for collagen binding has been located between amino acids 209 and 233 [13]. However, an alignment to the corresponding sequence in CNE reveals that these regions of the two proteins are not more conserved than the rest of the A domains. Two amino acids in CNA, Asn232 and Tyr233, appear to be critical for ligand-binding activity [13], yet neither of them is present in CNE.

The A domain in CNE is followed by a region which is similar to the B domain of CNA (Figs. 1 and 2). The B domain in CNA consists of three repetitive units (B1, B2 and B3), each 187 amino acids long [11]. Structure analysis of the B domain suggests that this domain functions as a spacer region, exposing the collagen-binding A domain from the cell [14]. Alignment of the B domain in CNE with CNA suggests that CNE contains one and a half B unit. However, a closer examination of the B units in CNA shows that each unit is composed of two repetitive subunits of alternating 92 and 95 amino acids, and thus the B domain contains six subunits (B1.1, B1.2, B2.1, B2.2, B3.1 and B3.2; Figs. 1 and 2). Accordingly, CNE contains three B subunits (B1.1, B1.2 and B2.1), each ~90 amino acids long (Fig. 2). The last CNE B subunit (B2.1) is the one least similar to its CNA counterpart, probably because it partially spans the cell wall.


Alignment of the six CNA and three CNE B-repeat subunits. Gaps (indicated by dashes) were inserted to obtain optimal alignment. Figures in parentheses refer to amino acid positions in CNA and CNE, respectively.

3.3 Collagen binding of S. equi and binding properties of CNE

Based on similarities to CNA, the CNE collagen-binding site was assumed to be located in the A domain. To study the collagen-binding activity of CNE, two recombinant proteins were expressed. The construct pCNE S encodes the A domain and pCNE L encodes both the A and B domains of CNE (Fig. 1). These proteins were analysed using SDS–PAGE followed by Western ligand blotting with radioactively labelled collagen. The result clearly shows that both recombinant proteins display collagen-binding activity (Fig. 3).


A: SDS–PAGE analysis of recombinant proteins CNE S (lane 1) and CNE L (lane 2). B: Western ligand blot of the same gel, showing binding of 125I-labelled collagen to CNE. Molecular mass markers are indicated.

Strains of subsp. equi were also analysed for their ability to bind collagen in solution and the result showed that strains of this subspecies efficiently bound radioactively labelled collagen (data not shown). The binding was shown to be specific, since it was not affected by the addition of serum albumin but completely inhibited by unlabelled collagen (Fig. 4). Attempts were made using different methods to release the native collagen-binding protein from the cell surface of strain 1866. At present we have no explanation why no collagen-binding activity was detected in the released material. However, similar results have been reported for CNA of S. aureus, where the collagen-binding activity was lost upon the release from the cell surface [11]. In inhibition assays, where increasing amounts of CNE L or CNE S were used, it was possible to reduce the binding of collagen to cells of subsp. equi strain 1866 to 65% (data not shown). The presence of other, yet unidentified collagen-binding structures could explain the incomplete inhibition. Another possible explanation is that there are several binding sites in collagen for the CNE protein. Since collagen molecules have highly repetitive structures, it would be expected that each collagen molecule could bind to more than one CNE protein. To test this hypothesis, an experiment was performed where cells of strain 1866 were incubated with increasing concentrations of unlabelled collagen, or HSA as a negative control (Fig. 5). After binding collagen, the cells were washed and incubated with radioactively labelled CNE L. The results confirmed that CNE L bound to cells which had previously been incubated with collagen, and that the degree of binding was proportional to the amount of collagen used. CNE L did not bind to cells without added collagen or to cells that had been incubated with HSA (Fig. 5).


Inhibition of binding of 125I-labelled collagen to subsp. equi cells using increasing concentrations of unlabelled collagen. Cells (total 2×106 CFU) were mixed with a fixed concentration of labelled collagen and an increasing concentration of unlabelled collagen. After incubation the cells were spun down, washed and the radioactivity bound to the cells was measured. Mean values from triplicate experiments are shown (♦), with bars indicating standard errors.


Binding of 125I-labelled CNE L to cells that have been preincubated with collagen or HSA. Subsp. equi cells (total 2×107 CFU) were incubated with an increasing concentration of unlabelled collagen as indicated (♦). Following binding the cells were spun down, washed and incubated with 125I-labelled CNE L (~100 000 cpm). After incubation the cells were spun down, washed and radioactivity bound to the cells was measured. As a negative control, the same experiment was performed using HSA instead of collagen (□). Mean values from triplicate experiments are shown, with bars indicating standard errors.

To further study the binding properties of CNE, various inhibition assays were performed using CNE S and CNA A. In the first study CNE S and radioactive collagen were mixed prior to incubation in microtitre wells containing immobilised CNE S. The result showed that soluble CNE S inhibited the binding of collagen to immobilised CNE S (Fig. 6). In a separate test CNA A, instead of CNE S, was used as an inhibitor. As shown in Fig. 6, CNA A efficiently inhibits the binding of collagen to CNE S, confirming that the two proteins compete for the same binding sites on collagen.


CNE S and CNA A in solution inhibits collagen binding to immobilised CNE S. A fixed concentration of labelled collagen (~20 000 cpm) was incubated with increasing concentrations of either CNE S (□) or CNA A (♦) in two separately performed experiments. After 30 min the samples were transferred to microtitre wells coated with CNE S and after 1 h the wells were washed and the radioactivity bound to the wells was measured. Mean values from triplicate experiments are shown, with bars indicating standard errors. As a negative control, CNE S and CNA A were replaced by HSA (15–500 µg ml−1), resulting in no inhibition of collagen binding to immobilised CNE S (data not shown).

3.4 The cne gene is present in both subsp. equi and subsp. zooepidemicus

The presence of the cne gene in 12 strains of subsp. equi and 13 strains of subsp. zooepidemicus was determined by PCR, using primer OCNE1:5 combined with OCNE2:3 or OCNE3:3. The obtained fragments were of the same length as the fragments from control strain 1866 for all strains tested (data not shown).

Visai et al. [15] have reported on a collagen-binding cell surface protein from subsp. zooepidemicus. This 57-kDa protein was purified and the amino acid composition determined. A comparison between the amino acid compositions of this protein and CNE suggests that they are not identical (e.g. CNE contains 4.4% Leu, 8.1% Lys, 12.7% Thr, while the protein reported by Visai et al. contains 11.4% Leu, 1.4% Lys, 5.6% Thr). Whether the 57-kDa protein of subsp. zooepidemicus is also present in subsp. equi remains to be investigated.

Adherence of bacteria to host tissue is the first critical step in the pathogenic process of most bacterial infections. CNA has previously been proven a virulence factor of S. aureus [8,9] and in murine vaccination models CNA gives protection upon challenge with S. aureus [16]. Although the importance of CNE in the pathogenesis of S. equi remains to be determined, the high level of sequence similarity between CNE and CNA suggests that these proteins have an evolutionary relationship, and thus CNE probably serves similar functions in S. equi as CNA in S. aureus. Therefore CNE will be an interesting candidate to evaluate in future immunisation experiments, exploring the possibilities for novel vaccines protecting against S. equi infections.


This investigation was supported by the AgriFunGen programme at the Swedish University of Agricultural Sciences, The Swedish Agency for Innovation Systems (2001-00846), NordVacc Läkemedel AB, Stockholm, Carl Tryggers Foundations (CTS 01:113) and FORMAS (22.1/2001-0911). The authors also thank M. Lindberg and K. Jacobsson for critical comments and advice.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
View Abstract