M protein of a Streptococcus dysgalactiae human wound isolate shows multiple binding to different plasma proteins and shares epitopes with keratin and human cartilage


*Corresponding author. Tel.: +49 (3641) 934223; Fax: +49 (3641) 933474, E-mail address: khschmidt@bach.med.uni-jena.de


Besides group A (GAS), Lancefield group C β-haemolytic streptococci (GCS) have been implicated as a causative agent in outbreaks of purulent pharyngitis. In this study we have investigated a class CI M protein of a Streptococcus dysgalactiae human wound isolate designated MC. MC shares similar properties with M proteins of GAS. It contributes to the virulence of the investigated GCS strain as revealed by in vivo phagocytosis in chicken embryos. Further, MC showed multiple binding to the human plasma proteins fibrinogen, albumin, plasminogen, IgA and all subclasses of IgG. Until now, an M protein, especially from a group C strain, with such a multiple binding behaviour has not been described. Immunoblot experiments with 150 patient sera, having a rheumatoid factor titre >1:256, revealed that 26% of these sera showed serological cross-reactivity between a 68-kDa cartilage protein and the N-terminal part of MC. Only 8% of the sera of healthy patients showed this property. In additional, MC also cross-reacted with antibodies recognising epidermal keratins. The cross-reacting 68-kDa protein from cartilage was different from human serum albumin, but was recognised with anti-vimentin immune serum. The MC was cloned and the gene sequenced. By using PCR, recombinant gene fragments encoding characteristic peptide fragments of MC were expressed in Escherichia coli. The peptides were used to map the binding sites for plasma proteins and to locate the cross-reacting epitopes on the MC molecule. In consequence, sequence alignments revealed that MC shared homologous regions with vimentin and different keratins. Our data, obtained with MC, suggest that not only infections with GAS but also infections with GCS and possibly GGS (the latter species can also produce class CI M-like proteins) may be responsible for the formation of streptococcal-associated sequel diseases.


colony forming units


group A streptococci


group C streptococci


group G streptococci


horseradish peroxidase


human serum albumin


M protein of GCS


phosphate-buffered saline


PBS plus 0.02% Tween 20


recombinant MC.


Besides group A (GAS), Lancefield group C β-haemolytic streptococci (GCS) have been implicated as a causative agent in outbreaks of purulent pharyngitis [5,8,10,39]. GCS were also isolated from wound infections and some case reports were published recently, describing the connection of GCS and group G streptococci (GGS) with streptococcal toxic shock syndrome [15,20,21,26,41]. Both GCS and GGS isolated from humans were designated Streptococcus dysgalactiae group C or group G because of characteristic taxonomic similarities [27].

It has been shown that GCS and GGS isolated from human sources can express M proteins with high sequence homology in the conserved C-terminal region to class CI M proteins of group A streptococci [3,35]. This implies that GCS and GGS have similar tools to group A streptococci to resist host defence mechanisms.

In this study, from a S. dysgalactiae group C strain, an M protein with similarity to the class CI M protein family of GAS [14] has been identified and was designated MC (M protein of group C streptococci). This protein was found to contribute to the virulence of the GCS strain as revealed by i.v. infection of chicken embryos. MC further showed multiple binding to fibrinogen, human serum albumin, IgG, IgA and plasminogen. In addition, antibodies induced against this protein showed cross-reactivity to keratin and a 68-kDa protein extracted from joint cartilage.

2Materials and methods

2.1Bacteria and phages

The S. dysgalactiae (former designation S. equisimilis) group C, T-type 4 strain 25287 of the strain collection of the Streptococcal Laboratory of the University of Jena, Germany, was used in this study. The strain had been isolated from a wound infection. It was stored at −70°C in Todd-Hewitt broth supplemented with 20% calf serum and 20% glycerol or in lyophilised form.

Escherichia coli strains TG1, JM109 and Epicurian coli XL1-Blue MRF’Kan (Stratagene) were used for subcloning experiments. E. coli M15(pREP4) (Qiagen, Hilden, Germany) was used for expression of recombinant MC protein (rMC) and MC protein subfragments. Host for M13 phage was E. coli strain JM103 Y. The gene encoding rMC was cloned in M13mp18 to isolate single-stranded DNA for nucleotide sequence determination. The phage library was established with phage EMBL3 in strain E. coli NM539.

2.2Patient sera and biopsies

One hundred and fifty sera from patients with the presumable diagnosis rheumatoid arthritis and with a titre of rheumatoid factor >1:256 were selected from routinely performed rheumatoid factor investigations in our lab. The age of the patients was not considered. We thank Dr Bartha from the hospital of skin diseases, Jena, for the serum of a psoriasis patient. As healthy controls 46 human sera were selected from sportsmen during routine checks.

Biopsy samples from knee joint cartilage after implantation of an artificial knee joint as well as skin pieces were homogenised and washed 10 times in 0.05 M Tris-HCl buffer, pH 7.5. The insoluble precipitates were boiled in five volumes of 0.05 M Tris-HCl buffer, pH 7.5 containing 2% SDS for 30 min. After cooling and centrifugation the supernatant was used for further investigations.

2.3Proteins, antisera and antibodies

The human plasma proteins fibrinogen, serum albumin, IgG and IgA were purchased from Sigma Chemical Co. (St. Louis, MO, USA). Vitronectin was purified according to the method of Yatohgo et al. [44] and fibronectin as described recently [40]. For binding assays the plasma proteins were directly labelled with horseradish peroxidase (HRPO) as described recently [34,38]. Vimentin (bovine) was obtained from Sigma Chemical Co. The N-terminal half of M5 protein, pepM5, was produced as described recently [30]. Antiserum against MC was raised in rabbits with aliquots of 100 μg of the entire purified protein for each immunisation step by the protocol of Louvard et al. [17]. In the first subcutaneous injection complete Freund's adjuvant was added, in the following i.m. and i.v. applications the protein was dissolved in phosphate-buffered saline (PBS). Anti-human vimentin serum from goat came from Chemicon International Inc. and anti-keratin polyclonal rabbit serum was purchased from Sigma Chemical Co. HRPO-conjugated anti-human IgG, anti-goat IgG and anti-rabbit IgG were ordered from DAKO Diagnostika GmbH (Hamburg, Germany).

MC-specific antibodies cross-reacting with human cartilage and skin proteins were isolated from anti-MC rabbit immune sera or human patient sera by immune absorption. The SDS protein extracts from cartilage or skin were separated by SDS-PAGE and the protein bands transferred onto nitrocellulose membranes. Reactive bands were identified with anti-MC rabbit serum after Western blotting of gels run in parallel. The identified bands were cut out and the small nitrocellulose strips were blocked with skimmed milk in PBS containing 0.02% Tween 20 (PBST) as described recently [34]. After washing with PBST the strips were transferred into 1.5-ml reaction tubes and 250 μl of anti-MC serum was added and incubated by shaking at room temperature for 3 h. The supernatant was then removed and the strips were washed three times with PBST. Absorbed antibodies were eluted with 200 μl 0.1 M glycine-HCl buffer, pH 2.5. After separation the antibody-containing supernatant was immediately neutralised with a drop of 2 M Tris buffer, pH 9.0. The antibodies were stabilised by addition of 100 μl rabbit preimmune serum.

Vice versa, patient sera which reacted with the cartilage 68-kDa band were absorbed on the recombinant MC fragment AC1 (see Section 3) and the antibodies were isolated as described above.

2.4Purification of MC protein and rMC protein

MC was found at the cell surface as well as in the culture supernatant of strain 25287. Native MC was isolated from mutanolysin extract of harvested cells from a 5-l streptococcal culture [30] and from the supernatant of this culture [34]. The proteins were precipitated with ammonium sulfate at 70% saturation in the culture supernatant. The resulting precipitate was resuspended in distilled water and dialysed against 0.02 M ammonium hydrogen carbonate. Dialysed material was centrifuged and the supernatant containing the crude protein was purified by affinity chromatography on fibrinogen Sepharose [30,32]. The column was equilibrated and washed with 0.05 M Tris, 0.15 M NaCl, pH 7.5. Proteins bound were eluted with 2 M potassium isothiocyanate, 0.1 M sodium acetate buffer, pH 5.0. Positive fractions were identified by SDS-PAGE and dot binding assays on nitrocellulose with peroxidase-labelled human serum albumin (HSA) and fibrinogen [32,34]. After neutralisation and dialysis of MC against 0.02 M ammonium hydrogen carbonate a rechromatography on fibrinogen Sepharose was performed with the same steps described above. After dialysis against PBS the purified protein was stored in aliquots at −20°C.

The recombinant MC (rMC) and its large subfragments AD, AC4, and AC1 were isolated from a 100 ml expression culture in E. coli strain M15(pREP4) (see below) according to the protocols of the manufacturer (Qiagen). After sonication of bacterial cells in PBS, the cell debris was sedimented by centrifugation. rMC and rMC peptide fragments were purified from the supernatant by affinity chromatography on fibrinogen Sepharose as described above.

2.5SDS-PAGE, blotting, and binding of plasma proteins

SDS-PAGE and electrotransfer of proteins to nitrocellulose membranes was performed with the Mini-Protean equipment of Bio-Rad according to the procedures of the manufacturer. Before serological and binding reactions the nitrocellulose membranes were blocked for 10 min with 5% defatted skimmed milk in PBST. Different rabbit or human sera were used as primary antibodies. For detection HRPO-labelled anti-rabbit IgG or anti-human IgG were used. The binding of human plasma proteins to MC or peptide fragments of MC was investigated after immobilisation of the streptococcal-derived proteins onto nitrocellulose membranes by Western blotting. The membranes were blocked with defatted skimmed milk and then directly incubated with HRPO-labelled human proteins for 4 h. The bacterial dot blot assay to examine binding of human proteins to streptococcal cells was performed as described recently [29,33]. Blots were visualised with 3-amino-9-ethylcarbazole [32].

2.6N-terminal amino acid sequencing

For amino acid sequencing purified MC and rMC protein transferred onto Immobilon membranes (Millipore) were used. Sequences were determined by Edman degradation on Applied Biosystems protein sequencer 473A.

Recombinant MC and recombinant MC peptide fragments were expressed as fusion proteins which start at the N-terminal site with the heterologous amino acids M-R-G-S. This sequence is derived from vector pQE50 after ligation of foreign DNA in the BamHI restriction site (see below).

2.7Cloning techniques, PCR and DNA sequencing

To establish a gene library in E. coli NM539, the partially Sau3A-digested genomic DNA of strain 25287 was ligated into the predigested lambda replacement vector EMBL3 [28]. Recombinant phages were produced with the help of the packaging extracts from Promega. The phage library was screened for expression of a fibrinogen binding protein as described recently [32]. One clone, EMBL3 k5, could be identified and was used as template for further investigations.

To determine whether M-like genes are present in strain 25287, we performed PCR by using the so-called all emm primers MPf 5′-ggg ggg gga tcc ATA AGG AGC ATA AAA ATG GCT (BamHI site) as forward primer and MPr 3′-aag atc taa gct TGT GAT AGC TTA GTT TTC TTC TTT (HindIII and BglII sites) as reverse primer [23] and the phage clone EMBL3 k5 as well as genomic DNA of strain 25287 as templates. The PCR was run in 30 cycles with 100-μl volumes reaction mixture using the GeneAmp PCR System 2400 and the AmpliTaq Gold Kit of Perkin Elmer. The following amplification protocol was used: first step, DNA was denatured at 95°C for 30 s; second step, annealing was performed between 50 and 60°C (the temperature was primer-dependent) for 30 s; third step, the elongation was run at 72°C for 90 s. The PCR fragments were purified with the Qiagen Gel Extraction Kit according to the instructions of the manufacturer. For DNA sequencing the PCR fragments were subcloned in M13mp18 phage. The single-stranded DNA was sequenced with the Sequenase Version 2.0 DNA Sequencing Kit (USB) according to the manufacturer's instructions. The DNA sequence was analysed using the program HUSAR (DKFZ Heidelberg). It encodes an M-like protein and was submitted to the NCBI Database, accession number X93464.

From the DNA sequence oligonucleotide primers were designed to produce DNA subfragments which allow an in-frame cloning in vector pQE50. Forward primers with a BamHI site in the 5′ end were used. Reverse primers with flanking HindIII or PstI sites had a stop codon before their restriction site. The following primers were used:

  • 1(A) 5′-ggt agg atc cGG GCA GAC AGT TAA GGC AGG C-forward
  • 2(B) 5′-gtg ctg gat ccG ATC AAA TTA AAC AAC TAG AA-forward
  • 3(C) 5′-aag cag gat ccG CTG CTC TCA AAC AAC TTG AA-forward
  • 4(D) 5′-ttt gta agc ttc aTT CTA GTT GTT TAA TTT GAT C-reverse
  • 5(E) 5′-ttt tga agc ttc aTT CAA GTT GTT TGA GAG CAG C-reverse
  • 6(F) 5′-gga aga tct gca gtc gac tta TGG GTT AGT TGC TTC ACC TGT TGA-reverse

PCR fragments were subcloned in vectors pCR-Script™ SK(+) (Stratagene) or pGEM®-T vector (Promega) before ligation into the expression vector pQE50. For expression of the recombinant peptides the appropriate ligation product in vector pQE50 was transformed in E. coli M15(pREP4). Fragment AC4 was produced from the PCR fragment encoding AD (see Section 3) by digestion with BamHI and HindIII before ligation into the PQE50 vector. The HindIII restriction site in the DNA sequence encoding MC is located at position 989 at the end of the encoding C4 repeat. Expression clones were screened by small-scale expression cultures according to the manufacturers’ instructions. The harvested cells were lysed with 0.5 M SDS, 0.025 M Tris and 0.02 M glycine-HCl buffer and the recombinant protein identified with anti-MC rabbit serum by Western blot analysis.

2.8In vivo phagocytosis assay

Fertile eggs, 11 days post fertilisation, from white leghorn hens were used. The chicken embryos were infected by i.v. inoculation of 100 μl of different numbers of streptococci as described recently [29,33]. Before infection, with the help of a counting chamber the bacteria from a 5-ml culture, incubated for 3 h at 37°C in Todd-Hewitt broth, were adjusted to 4×103 colony forming units (CFU) per ml with Todd-Hewitt broth, which was diluted with the same volume of PBS [29]. For the protective assay with antiserum, two 3-ml aliquots of the adjusted streptococcal culture were sedimented by centrifugation. The bacterial pellets were suspended either in 0.75 ml anti-MC rabbit serum or in the same amount preimmune serum. After incubation for 15 min at room temperature, 0.75 ml PBS and 1.5 ml Todd-Hewitt broth were added. The bacteria were carefully suspended and immediately used for infection. The bacterial number injected was confirmed by plating 100 μl of each dilution on blood agar. Eggs were candled daily for the next 6 days. Samples of the chorioallantoic fluid from the embryos that died were spread on blood agar plates to ensure that the death was caused by streptococci.


3.1Binding properties of group C streptococcal isolates

In our laboratory we routinely check streptococcal isolates for expression of plasma protein binding proteins at the cell surface using a bacterial dot blot assay. Recent experiments revealed that such strains of GAS which bind fibrinogen and/or albumin at their cell surface mostly express proteins of the M family [29,31–33]. Fibrinogen binding seems to be a property of most M proteins [1,7,25,30–32,34] and contributes to the antiphagocytic properties of these cell surface proteins [29,33,42,43]. In our binding experiments with human isolates of GCS and GGS we found that such strains showed complex multiple binding to a test set of plasma proteins comparable to that of GAS [29]. For further investigations we selected the group C strain 25287, a wound isolate, which expressed at the cell surface multiple binding to the plasma proteins fibrinogen, human serum albumin, plasminogen, IgG, IgA, and to vitronectin (not shown).

3.2Identification of a multiple binding M protein (MC) from GCS

For isolation of the binding component(s), the mutanolysin cell extract and the culture filtrate of strain 25287 were fractionated on fibrinogen Sepharose. From both crude extracts a protein could be isolated which formed a band triplet with a main middle band at 55 kDa after separation by SDS-PAGE. When tested by Western blotting, all three bands of the protein showed binding to the human plasma proteins fibrinogen, albumin, plasminogen, IgA, polyclonal IgG and all four subclasses of IgG (Fig. 1). The same pattern of plasma proteins also bound to the whole streptococcal cell surface. The multiple binding protein, designated MC, did not show binding to vitronectin in this approach. The binding to vitronectin was found only at the whole streptococcal cell surface, suggesting that this affinity was caused by a protein different from MC. The purified MC was N-terminal sequenced. The obtained sequence of 31 amino acids did not show any similarity to known streptococcal M protein sequences in the protein and DNA databases. Because we assumed that this multiple binding protein belongs to the M protein family, a PCR was performed with the ‘all emm primers’ published recently [23] and the DNA of EMBL phage clone k5 as well as the chromosomal DNA of strain 25287 as templates. In both cases a 1.5-kb fragment was obtained with an identical DNA sequence. A computer search for sequence homology was made in the EMBL database. The DNA sequence (accession number X93464) encoded an M-like protein with 475 amino acids (Fig. 2). It showed homology in the C domain to class CI M proteins of group A streptococci, but in the N-terminal part no similarity to other M proteins was found. The 31 amino acids obtained by N-terminal sequencing of the native MC could be identified in the translated sequence of the cloned MC starting at position 42 (Fig. 2, bold letters). The sequence before position 42 is characteristic for a signal peptide described for M proteins of GAS [1] (Fig. 2). The found N-terminal amino acid sequence of the native expressed MC (Fig. 2, bold letters) starts at amino acid 42 which points out that the signal peptide was removed from the mature MC protein when it entered the streptococcal membrane. In Fig. 3 the localisation of the different domains of MC is schematically illustrated. In different M proteins of GAS the domain between the end of signal peptides and the beginning of the conservative C repeat region is mostly variable. This region defines the M type of GAS strains. The sequence of MC is new in the part between the signal sequence and the C domain. It contains two completely identical B repeats unique to MC and not found in other M and M-like protein sequences of GAS, GCS, or GGS accessible so far in the database. The C repeats and the anchor region are conserved between different M proteins. But single C domains can show a small sequence variability inside one and between other M proteins. Thus, the four C repeats of MC are similar but not identical (Fig. 2). However, the whole C domain of MC shows high homology to class CI M protein sequences of human isolates of group A (M11338, X80168), C (X60097), and G (X62467, Z32677, Z32678, X60098, M95774) streptococci available in the database (not shown). The genes encoding MC and MC fragment AD were ligated into the expression vector pQE50 and expressed in E. coli, strain M15 (pREP4). rMC and the large recombinant fragment AD showed the same binding pattern to human proteins as the native MC from streptococci in Fig. 1. For analysis of the binding epitopes in the MC molecule, additional smaller gene fragments of MC were amplified by PCR and expressed in E. coli using vector pQE50. The MC peptides were tested for binding to HRPO-labelled plasma proteins by Western blotting (not shown). The results are illustrated in the table of Fig. 3.

Figure 1.

Separation of MC by SDS-PAGE and detection of its binding properties by Western blotting with HRPO-labelled human proteins fibrinogen (FBG), serum albumin (HSA), plasminogen (PLMG), IgA, polyclonal IgG, and monoclonal subclasses of IgG (types 1–4) as well as vitronectin (VN). MWST, molecular mass standard.

Figure 2.

A: Peptide sequence of MC translated from its nucleotide sequence, accession number X93464. Bold characters indicate the amino acid sequence revealed for native processed MC in streptococci. Segments with homology to signal peptides of other M proteins1, to vimentin2 and to different keratins 3,4,5 are aligned. Arrows indicate the start of the domains: signal peptide (SS), B repeats B1, B2, C repeats C1, C2, C3, C4 and domain D which includes the anchor region. The angular arrow indicates the end of SS. B: Analysis for heptad periodicity of the regions of MC and human vimentin carrying the motif DIMRLREKLQ (bold letters). The numbers indicate the location of amino acids in the appropriate molecule.

Figure 3.

Schematic representation of complete MC and related MC peptide fragments produced by cloning and expression of appropriate PCR fragments derived from the nucleotide sequence AC X93464. The numbers indicate the first amino acids of the individual regions (MC complete numbering). The localisation of PCR primers used for synthesis of the DNA fragments encoding the cloned peptides is shown (A–F). All fragments carry the heterologous amino acids MRGS at the N-terminus, derived from vector pQE50. The amino acid composition of MC is characteristic for class CI M proteins of group A streptococci. The translated amino acid sequence of complete MC consists of a signal sequence (SS), a region A without repeats, two repeating regions B, four class I C repeats and a characteristic D region including an anchor sequence. The left column of the table summarises the results of mapping experiments to localise the binding regions in the MC molecule for the human proteins fibrinogen (FBG), albumin (HSA), plasminogen (PLMG), IgA, and IgG, as evaluated by Western blotting. The right column illustrates the cross-reactivity of MC domains to antibodies which also react with a 68-kDa protein in cartilage extract of knee joint (CAR) or with keratin bands (KER) in human skin extracts (see also Figs. 5 and 6).

The N-terminal half of MC, including the regions A and B, was found to bind fibrinogen. This confirms the variable N-terminal part of M proteins as fibrinogen binding epitope [1,25,30]. However, by database sequence comparisons with the N-terminal parts of known M protein amino acid sequences a characteristic sequence motif for fibrinogen binding could not be identified. On the other hand, the C domain of MC carries four C repeats characteristic for this domain in other class CI M proteins. This region of MC was recognised as the albumin binding epitope which confirms earlier data obtained with M proteins of GAS. In M proteins of GAS the C repeats were found to be essential for the interaction with albumin [1,12,25]. However, for the binding neither of plasminogen, IgG nor of IgA to MC could we detect a distinct binding motif on the MC molecule. In MC only the large peptides of AD and AC4, but no smaller fragments, were reactive (Fig. 3). Different consensus sequences in M proteins were published for the interaction with IgG [1,4], with IgA [4] and with plasminogen [6] but none of these motifs was present in the amino acid sequence of MC.

3.3In vivo phagocytosis experiments in the chicken embryo model

In the previously described chicken embryo model it was demonstrated that M proteins contribute to the virulence of GAS for chicken embryos [29,33]. We used this in vivo phagocytosis model to test the GCS strain 25287 with the assumption that the MC may act as virulence antigen like M proteins of GAS. In a first experiment we determined the effect of different doses of strain 25287 6 days post i.v. infection. From these experiments the LD90 was calculated to be 300–500 CFU. This bacterial concentration was used in the following experiments. Here we tested whether the rabbit immune serum raised against purified MC had a protective effect for chicken embryos if they were infected with GCS strain 25287. As demonstrated in Fig. 4, the animal group treated with bacteria which were incubated with anti-MC showed a delayed and significantly decreased killing over the observation period of 6 days. In the control groups which were treated with bacteria alone or with bacteria plus preimmune rabbit serum a significantly higher death rate was found. This result revealed MC to be a virulence factor for GCS which could partially be neutralised by opsonisation with anti-MC at the cell surface of strain 25287. Thus, MC can be considered an M protein of GCS comparable to the M proteins for GAS. Samples of 50 μl of chorioallantoic fluid from each egg, regardless of whether the embryos survived or had been killed, were plated out on blood agar. From all samples of killed embryos streptococci could be isolated. But no bacteria were found in samples from eggs where the embryos survived. This indicates that in surviving animals the streptococci were killed by the phagocytic system of the chicken embryo.

Figure 4.

Virulence of GCS strain 25287 injected i.v. with 100 μl bacterial suspension, containing 400 CFU, in chicken embryos. Time-dependent killing was observed over 6 days. □, Bacteria alone; ▪, bacteria plus rabbit preimmune serum; ◯, bacteria plus anti-MC rabbit immune serum. The figure represents the evaluation of three independent experiments. In the first two experiments, 15 eggs in each group were used, in the third test series, 20 eggs per group were infected, altogether 50 eggs per group. Treatment with anti-MC rabbit immune serum resulted in a significant delay and decrease of the death rate.

3.4Cross-reaction to joint cartilage and keratin

By testing different anti-M protein immune sera with human proteins, we found that the anti-MC rabbit immune serum recognised different human proteins in extracts from cartilage and human skin. A strong reaction was found against two bands of 68 kDa and 54 kDa present in an SDS extract of human knee joint cartilage (Fig. 5A). The 68-kDa and 54-kDa bands together with smaller bands of the cartilage extract were also recognised by commercial anti-vimentin antibodies (Fig. 5A). However, the 54-kDa band was also, though weakly, recognised by the anti-human IgG used as secondary antibody. Thus we assumed that despite extensive washing of the cartilage sample, traces of human IG could be still present in the SDS cartilage extract. The secondary antibody consequently identified the heavy chain of human IgG which arises from sample preparation under reductive conditions.

Figure 5.

Reaction of the rabbit anti-MC immune serum as well as different human patient sera against SDS extracts of human knee joint cartilage (A) and human skin (C) after SDS-PAGE and Western blotting. B: Identification of vimentin (bovine) after SDS-PAGE and Western blot by antibodies reactive with MC fragment AC1. Abbreviations: MWST, molecular mass standard; Anti-MC, rabbit immune serum against purified MC; Anti-VIM, anti-vimentin serum from goat; RA-P (1,2), sera from different patients with rheumatoid arthritis; PS-P, serum from patient with psoriasis; Healthy(1,2), sera from two healthy sportsmen; Rabbit-NS, rabbit preimmune serum; Anti-human IgG from goat, Anti-rabbit IgG from goat, Anti-goat IgG from rabbit.

Thus, we focused the further examinations on the 68-kDa band. This band was different from human serum albumin because the cross-reactive antibodies directed against this protein, isolated from anti-MC, did not react with purified HSA in the Western blot (not illustrated). Separation of commercial bovine vimentin by SDS-PAGE and following Western blotting revealed a double band at 68 kDa. This band was recognised with antiserum against human vimentin but also with anti-MC antibodies which were isolated from rabbit immune serum by immunoabsorption on the MC fragment AC1 (Fig. 5B). Database sequence alignment revealed bovine vimentin (accession number P48616) to have 96.8% identity with human vimentin (accession number X56134). A recently described 58-kDa protein from cartilage cross-reacting with antibodies directed against the pepM5 fragment of M5 protein and with anti-vimentin antibodies [2] could not be identified with our anti-MC. This is supported by the result that anti-MC rabbit serum did not cross-react with pepM5, the N-terminal half of M5 protein. Vice versa, anti-pepM5 rabbit serum did not react with MC (not shown). To determine which antigenic regions in the MC molecule could induce the production of cross-reactive antibodies, we performed immunoabsorption of anti-MC rabbit serum on the 68-kDa band from cartilage as described in Section 2. The isolated antibodies were then tested against different fragments of MC as illustrated in Fig. 6. They bound to MC fragments carrying the region AB2 in the N-terminal part (Fig. 6 and table in Fig. 3). Sequence comparison of MC with human vimentin revealed that three related segments in the AB2 segment of MC showed 55–75% homology to the vimentin motif DIMRLREKLQ (Fig. 2). A computer search with the program ‘Matcher’[9] showed that the motif is located in the α-helical coiled-coil domains of MC as well as of vimentin. Such coiled-coil domains are characterised by the periodically repeating heptad form (a.b.c.d.e.f.g)n[19]. The hydrophobic residues I and L of the motif DIMRLREKLQ were found to be located in MC and in vimentin at the same positions a and d (Fig. 3B). The charged amino acids K/R of vimentin and K/K of MC occupy positions e and g. Positions a and d are responsible for hydrophobic interactions and positions e and g for ionic pair interactions between two α-helical protein chains to form a coiled-coil dimer. Such regions were assumed to be exposed in coiled-coil structures [9]. Because of these similarities we tested human sera against the MC fragment AC1 by Western blotting. Among 150 human sera of patients with a rheumatoid factor titre >1:256 we found that 39 (26%) of these sera recognised this fragment. All of the AC1-reactive sera also stained the cartilage 68-kDa band. In 46 control sera of healthy donors only four (8%) of the sera stained the AC1 fragment. With some of the patient sera we performed immunoabsorption on the MC fragment AC1. The antibodies retained by AC1 cross-reacted with the 68-kDa cartilage band (Fig. 5A) and with bovine vimentin (Fig. 5B). In bovine vimentin the motif DIMRLREKLQ is also completely present and forms the same heptad structure as in human vimentin if analysed with the program ‘Matcher’. Thus, it seems likely that the vimentin motif DIMRLREKLQ is one target for cross-reacting antibodies between MC and vimentin.

Figure 6.

Separation by SDS-PAGE of different MC peptide fragments as illustrated in Fig. 3. After electrotransfer of peptides from parallel gels onto nitrocellulose membranes, the blots were primary probed with anti-MC rabbit immune serum (left), anti-MC antibodies purified by absorption to the 68-kDa cartilage protein (middle) or to the keratin band from human skin (right). As secondary antibody HRPO-labelled anti-rabbit IgG was used.

Besides reacting with the cartilage protein, anti-MC rabbit serum also recognised keratin bands between 40 and 60 kDa in human skin extracts after separation by SDS-PAGE (Fig. 5C). But in this pattern of bands staining with anti-vimentin and anti-keratin serum was observed, suggesting that in the SDS skin extract, besides other proteins, keratin and vimentin were present together. Sera of rheumatoid patient 1 and a psoriasis patient recognised the same pattern of bands as found with anti-vimentin serum and not with anti-keratin serum (Fig. 5A,C).

We also performed immunoabsorption of anti-MC rabbit serum on the keratin bands from skin extract between 60 kDa and 40 kDa (Fig. 5C) as described above. The isolated antibodies were tested against different fragments of MC (Fig. 6, right). The antibody fraction which bound to human skin keratins also recognised MC fragments carrying the C repeat region. In the C repeat region of MC sequence motifs were recognised which showed 54–65% homology to the human cytokeratins 1, 10 and 18 (Fig. 3A). Comparable cross-reacting peptide segments were recently identified in M6 protein of GAS [36]. Also the fragment AC1 which contains only one C repeat was stained. In this region we did not find an alignment with a human cytokeratin. This suggests on the one hand that the database is not complete for human keratin sequences. But also induction of cross-reactive antibodies by other structures as homologous sequence epitopes could be possible. None of the keratin motifs formed the characteristic heptad structure in the analysis with ‘Matcher’ as found for the vimentin motif DIMRLREKLQ.


In this study we investigated an M protein of class CI of a S. dysgalactiae human wound isolate, designated MC, which showed multiple binding to the human plasma proteins fibrinogen, albumin, plasminogen, IgA and all subclasses of IgG. MC was found to contribute to the virulence of the GCS strain 25287 as tested in the chicken embryo model. This in vivo phagocytosis assay was recently found to be suitable for the virulence determination of M protein-positive GAS [29,33]. Further, MC showed serological cross-reactivity to a 68-kDa protein extracted from cartilage, and to epidermal keratins of different molecular masses. The cross-reacting 68-kDa protein from cartilage was different from albumin, but was recognised with anti-vimentin immune serum. In consequence, sequence alignments revealed that MC shared homologous regions with different types of keratin and with vimentin. Until now, an M protein, especially from a group C strain, with such a multiple binding behaviour has not been described. We mapped the binding regions of MC for different plasma proteins by Western blotting. Recombinant peptide fragments which represent characteristic domains of MC were used as target. Fibrinogen and albumin bound to distinct separate regions of MC. Fibrinogen bound to the N-terminal part which includes domains A and B. The C repeat region was not necessary for this interaction. The majority of M proteins bind to fibrinogen [1,25,30,31] and the N-terminal half, which represents the variable region in these molecules, seems to be the fibrinogen binding site. However, by sequence comparisons of the N-terminal variable segments of M protein sequences accessible in the EMBL database a consensus sequence for fibrinogen binding could so far not be identified.

Albumin binds to the C repeats of MC. This region was found to be essential for albumin binding of class C M and M-like proteins [1,12,25].

For the interactions with plasminogen, IgG and IgA, a defined binding region in MC could not be identified. Only the large MC fragments which include the main regions A, B, C, and D showed affinity to IgG, IgA and plasminogen. Although MC expressed plasminogen binding activity, as was also found for a series of M family proteins in group A, C, and G streptococci, the sequence of MC did not contain the plasminogen binding motif EKLTADAELQRLKNERHEEAELERLKSE as reported recently [6]. The same was true for IgG and IgA binding. The consensus sequences described so far for binding of these immunoglobulins to M and M-like proteins [1,4] were not found in MC. Thus it seems that variable sequences of different M proteins can form structures with a unique property, for example, to bind to a distinct plasma protein. One important property of the different M protein molecules is the formation of coiled-coil structures shown to be essential to develop binding affinity to different human proteins [7,12,22]. Consequently, the consensus sequences reported so far for a special interaction are only correct for a few of the existing M protein molecules.

The name M protein was defined for those cell wall proteins of GAS which can impede phagocytosis. Because MC shares characteristic properties with M proteins of GAS, for example fibrinogen binding and sequence homology in the characteristic C domain, we tested the contribution of MC to the virulence of the GCS strain 25287. Compared to GAS, the virulence for chicken embryos of strain 25287 was about 2 log less (LD90 400 CFU) in comparison to highly virulent GAS strains (LD50 1 CFU) [29,33]. The virulence of the GCS strain 25287 for chicken embryos was significantly decreased when the bacteria before infection were opsonised with rabbit antiserum raised against purified MC. From this result we propose that MC is an M protein.

With MC we firstly describe that M proteins of GCS can induce antibodies which cross-react with human proteins of joint cartilage and skin. Such an effect has so far been described for M proteins of GAS. We found that in extracts from knee joint and hip joint cartilage a 68-kDa protein was present which showed cross-reaction with anti-MC immune serum from rabbit as well as with 26% of sera from patients with a rheumatoid factor titre >1:256. An anti-vimentin immune serum also recognised the 68-kDa band, suggesting that vimentin may participate in this reaction. The final proof, N-terminal sequencing of the 68-kDa band to verify vimentin as the counterpart cross-reacting with MC, was not successful. However peptide sequence alignment of vimentin (accession number X56134) with MC revealed that three repeating segments in the N-terminus of MC showed high homology with the vimentin-derived motif DIMRLREKLQ. Cross-reactivity between cartilage proteins and another M protein, pepM5, was described recently [2] and vimentin was shown to be one of the cartilage cross-reacting proteins. However, sequence alignment did not reveal a common sequence motif between pepM5 and vimentin such as we found in this report between MC and vimentin. Moreover, rabbit immune sera, against MC or pepM5, did not show cross-reactivity between both M protein molecules. Thus, we conclude that epitopes of different M proteins, cross-reacting with vimentin, can share different regions and possibly different quaternary structures with the vimentin molecule. The increased presence of autoreactive anti-vimentin antibodies in human sera with increased rheumatoid factor titre need not be the consequence of a preceding streptococcal infection. It may also be a result of other autoimmune reactions without streptococcal contribution. However, a new streptococcal infection of such patients could trigger a booster effect via cross-reactive epitopes resulting in an increase of autoreactive antibodies. Experiments are under way to prove the existence of cross-reactive T cells for MC fragments and vimentin.

Keratins of human skin are another group of human proteins which share homologous epitopes with MC. Anti-MC antibodies, which recognise keratin, bound to the C-terminal half of MC. This regions is conserved between different class C M proteins of GAS, GCS and GGS. Similar keratin motifs, published recently, which showed cross-reactivity with M 6 protein of GAS [18,36] could also be found in the C repeat domain of MC. Because many other M proteins carry those conserved class C domains, such cross-reaction can be expected with M proteins of different types.

After infection with GAS some patients develop sequel diseases with an autoimmune character such as rheumatic fever, often complicated with endocarditis, acute glomerulonephritis, arthritis or psoriasis. Tissue components of the organs which were attacked by streptococcal sequel diseases show in all cases immunological cross-reactivity with components of GAS, mainly the M family proteins. For example, heart muscle myosin shares sequence homology with different M proteins [24]. Other cross-reactions have been described between vimentin, extracted from kidney, and M protein [16], skin keratin and M6 protein [36] and between cartilage vimentin and M5 protein [2]. Group C streptococci have so far not been described to cause such chronic development. Only one article was published 10 years ago [13] describing acute psoriasis associated with group C and G streptococci. In recent years the numbers of cases have increased where GCS and GGS showed an association with streptococcal infections normally caused by GAS, for instance with streptococcal toxic shock syndrome. These results suggest a development to an increased participation of GCS in such human diseases. This may also be true for human isolates of GGS where the sequences of emm genes show a large similarity to those of GCS [3,35]. Moreover, the genomic regions adjacent to the emm gene are nearly identical between GCS and GGS and rather different from GAS [11]. Horizontal gene transfer, concerning the emm genes, between GAS and GCS or GGS is an important process discussed in this context [37]. Our data obtained with MC suggest that not only infections with GAS but also infections with GCS or GGS may be involved in streptococcal-associated sequel diseases.


We thank Ms Roswitha John for excellent technical assistance. The work was supported by Grant 01 ZZ 9602 from the ‘Verbund für Klinische Forschung’ of the Friedrich-Schiller-University, Jena, which is sponsored by the Bundesministerium für Forschung und Technologie, Germany.


  • 1

    Signal sequence of M1 protein of GAS as one example [1].

  • 2

    Human vimentin, accession number X56134, aa 181–190.

  • 3

    Motif in human keratin type 10, accession number M77663, aa 51–66, additional in keratins 1, 12, 13, 15, 16.

  • 4

    Motif in human epidermal keratin type I, accession number J00124, aa 925–941.

  • 5

    Motif in human keratin 18 mRNA, accession number M26326, aa 318–396.

  • 1

    Antibodies from different sera isolated by immunoabsorption on MC fragment AC1.

  • 2

    Antibodies labelled with horseradish peroxidase.