Preparation of screening antigens
To screen for autoreactive antibodies, blood from healthy pigs was prepared as previously described for M. suis antigen preparations [28, 29] with slight modifications. Briefly, blood cells from M. suis-negative pigs were sedimented by centrifugation at 300 × g for 15 min at room temperature (RT). Plasma and buffy coat were discarded. The erythrocytes were suspended in phosphate-buffered saline (1 × PBS; Biochrom Ag, Berlin, Germany) containing 0.15% Tween20 and 3% EDTA and incubated for 20 min at RT with shaking. Debris and erythrocytes were removed by low-speed centrifugation (500 × g, 20 min, RT). The supernatant was centrifuged at 25,000 × g for 120 min at 4°C. The resulting pellet was resuspended (1 mg/ml) in sterile 1 × PBS. The preparation was depleted of IgG and albumin using a ProteoExtract® Albumin/IgG Removal Kit, Maxi (Calbiochem, Merck, Geneva, Switzerland) according to the manufacturer's recommendations and stored in 100 μl aliquots at -80°C until further use. The preparations were referred to as negative antigen (neg Ag).
Sera reactive with neg Ag were further tested with IgG-depleted erythrocyte lysate (ECL) and with several actin preparations. ECL was prepared by hypotonic lysis of erythrocytes from a healthy pig. Red blood cells were washed three times with 1 × PBS (500 × g, 15 min, RT) and resuspended in ice cold lysis buffer (5 mM Na2PO4, 1 mM EDTA, pH 7.6, Sigma) at a 1:1 (v/v) ratio. After incubation on ice for 5 min, the ghosts (remnants of lysed cells) were pelleted (30,000 × g, 15 min, 4°C). This procedure was repeated until the pellet turned white. The white pellet was washed in 1 × PBS, homogenised by ultrasonication on ice and IgG depleted. The concentration was set to 1 mg/ml with 1 × PBS and the suspension was stored in 1 ml aliquots at -80°C until further use.
Pig and rabbit muscle actin (α-actin) were obtained from Sigma, Buchs, Switzerland. Porcine cytoskeletal actin (β-actin) was not commercially available and was therefore produced recombinantly (see below).
Screening of sera by ELISA
Sera were obtained in previous studies using a splenectomized pig model for experimental M. suis infections . They were divided into two groups. Group 1 consisted of sera from animals that had been immunised with a recombinant immunogenic surface protein of M. suis (MSG1)  prior to splenectomy and M. suis infection. Group 2 sera were from splenectomised, M. suis-infected pigs that had not been immunised . The studies were approved by the government of upper Bavaria under the registration number 211-2531-77/98 and were performed in compliance with animal care legal prescriptions.
The porcine sera were screened for autoreactivity by ELISA as described previously . Briefly, 100 ng of IgG-depleted neg Ag were coated on individual wells of microtiter plates. Sera were diluted 1:100 in 1 × PBS containing 0.05% (v/v) Tween20 (PBST) and tested in duplicate. Sera were drawn prior to M. suis infection, after immunisation in group 1 (I), during the first clinical attack (II), between first and second clinical attacks (III), and during the second clinical attack (IV). Sera reactive with neg Ag were further tested for reactivity with ECL, pig and rabbit α-actin, and porcine β-actin. The protein concentration was 100 ng/well. OD values (λ = 405 nm) were determined and normalised to a negative control (pool of 10 sera from healthy pigs). Cut-off values were set at three times the standard deviation of the negative controls.
To determine the subtypes of autoreactive antibodies, monoclonal antibodies targeting porcine IgG1 and IgG2 (Prionics, Schlieren, Switzerland) were diluted 1:1000 in PBST. Horse-radish-peroxidase-labelled antibodies against mouse IgG (Sigma) diluted 1:5000 in PBST were used for detection.
Cloning and expression of porcine cytoskeletal actin
Porcine cytoskeletal actin (β-actin) was not commercially available. Using the encoding mRNA sequence (Genbank:AY550069), the gene was de novo synthesised and optimised for E. coli codon usage by Eurofins (Martinsried, Germany). For cloning, specific recognition sites for endonucleases were introduced (XhoI at the 5' end and HindIII at the 3' end). For expression, the gene was ligated into the expression vector pBadMycHisA (Invitrogen) using T4 ligase from Roche, Basel, Switzerland (overnight, 14°C). The insert-containing vector was transformed into E. coli LMG194 (Invitrogen). The E. coli transformants were grown at 37°C, and expression of the recombinant porcine β-actin was induced by adding 0.02% arabinose at OD600 = 0.4. The cells were incubated for a further 4 h at 37°C and the 6×His-tagged protein was purified by nickel affinity chromatography (GE Healthcare, Glattbrugg, Switzerland) according to the manufacturer's recommendations. To extract the protein from inclusion bodies, 8 M urea was used for purification. The recombinant protein was further purified by electro-elution from 10% polyacrylamide gels using an Electro-Eluter Model 422 (Biorad, Reinach, Switzerland). The eluted protein was precipitated by adding three volumes of ice cold acetone and incubating overnight at -20°C. The protein was harvested by centrifugation (10,000 × g, 90 min, 4°C), allowed to dry, resuspended in 1 × PBS, and stored in aliquots of 1 mg/ml at -80°C.
Production of hyperimmune sera
One milligram of protein (recombinant MSG1 , recombinant porcine β-actin or porcine α-actin (Sigma, Buchs, Switzerland)) was mixed with complete Freund's adjuvant (Sigma) at a 1:1 (v/v) ratio and injected subcutaneously into rabbits. Two booster injections (the proteins were mixed 1:1 (v/v) with incomplete Freund's adjuvant) were given two and four weeks later. Six weeks after the first injection the rabbits were bled; the serum was harvested and stored in aliquots at -20°C. Immunisations were approved by the Veterinary Office of Zurich and conducted under the registration number 144/2008 in accordance with legal prescriptions.
The cross-reactivity of rabbit sera against MSG1 and porcine α-actin between α-actin and MSG1 was tested by western blotting as described previously . Briefly, α-actin and MSG1 were loaded on 10% polyacrylamide gels containing 1% SDS (w/v) and transferred to a nitrocellulose membrane (Whatman Protran®, GE Healthcare). Free binding positions on the membrane were blocked with skim milk (Sigma, 2% w/v in Tris buffered saline, TBS) and probed with rabbit hyperimmune sera (diluted 1:100 in TBS containing 2% skim milk powder) against recombinant MSG1 and α-actin, respectively. A secondary antibody reactive with pig IgG (Sigma) and coupled to horseradish peroxidase (HRP) was used, diluted 1:5000 in TBS containing 2% (w/v) skim milk powder. The substrate for the HRP was H2O2 (Sigma); 4-Chloro-1-naphthol (Sigma) was used as chromogen. The reaction was stopped by adding MQ water (Millipore, Zug, Switzerland).
Search for cross-reactive epitopes
To identify potential cross-reactive epitopes, an epitope finder program referred to as the NetMHCpan algorithm was used . Published protein sequences of MSG1 (UniProtKB/TrEMBL:Q05G10), porcine α-actin (Swiss-Prot:P68137.1) and porcine β-actin (Swiss-Prot:Q6QAQ1.2) were used as input.
The ELISA OD values measured at 405 nm were compared by linear regression models using SigmaPlot software, Version 10.0 (SPSS Inc., Chicago, IL, U.S.A). To show independence of values between the two groups, the unpaired Student's t-test was applied (significant difference if P ≤ 0.05).