Vmri.hu

ANALYSIS OF VIRULENCE PLASMIDS OF ENTEROTOXIC ESCHERICHIA COLI
(ETEC) STRAINS 2173 ISOLATED FROM PORCINE POSTWEANING DIARRHOEA
Ph.D. thesis
Eötvös Lóránd University of Sciences, Faculty of Science head: Prof. Dr. Erdei Anna, corr. member of Hung. Acad. Sci. Classical and molecular genetics Ph.D. program head: Prof. Dr. Orosz László, corr. member of Hung. Acad. Sci. leader of the Ph.D. work: Prof. Dr. Nagy Béla, ord. member of Hung. Acad. Sci. Veterinary Medical Research Institute of Hung. Acad. Sci. INTRODUCTION
The most frequent enteral colibacterial infections of pigs and cattle are the diseases caused by enterotoxigenic Escherichia coli (ETEC) strains which also often cause diarrhoea in humans. Specific prevention of ETEC induced diarrhoea is still an unsolved problem in weaned pigs. Live, oral vaccines seem to be a new way of prevention. For an effective oral vaccine, the candidate vaccine strains should be able to colonize the intestine (as well as the pathogenic strains) but must be free from toxins or other factors destructive for the microvilli. To produce this kind of vaccine strain detailed informations are required about the virulence genes of the wild type (virulent) strains and about the mobile elements corresponding for the virulence gene-transfer. The ETEC bacteria causing diarrhoea of weaned pigs are able to adhere to the microvilli of the epithelial cells of small intestine through their fimbrial adhesins (F4, F18), meanwhile they produce enterotoxins (STa, STb and/or LT) stimulating the fluid secretion of the absorptive intestinal The production of ETEC virulence factors (adhezins and toxins) determined by plasmid encoded genes. The earlier described plasmid encoded virulence factor genes were the virulence genes of ETEC strains originated from pigs and cattle. There are a number of mobile virulence elements: pathogenicity islands (PAI), bacteriophages, plasmids and transposons taking part in the horizontal evolution of pathogenic E. coli bacteria. In the evolution of ETEC strains transposons carrying heat stable enterotoxin genes are playing an important role. STa as carried by Tn1681 and STb has been described as part of Tn4521. Besides there are data about IS mediated transfer mechanism of genes encoding LT (heat labile) toxin. So far there were no reports about 10-30kb sized DNA regions in ETEC plasmids, which would carry more virulence genes in contrast plasmids of Shigella and Yersinia strains. The survival of plasmids is determined by two main factors: by their copy number and by their ability to co-exist with other plasmids (plasmid incompatibility). Some of them are able to exist only in one copy, but the high copy number plasmids can exists in several hundred identical copies The definition of pathogenicity islands (PAI) is based on the observation that the virulence genes of pathogenic bacteria localized and form a distinct region on special parts of the chromosome together with genes corresponding for their horizontal spreading. The pathogenicity islands were first described in extraintestinal (uropathogenic) E. coli bacteria, but later examinations of a number of other enteral E. coli pathotypes and other bacteria showed that their virulence genes are forming Pathogenicity islands are large (10 kb to 200 kb) DNA fragments encoding for virulence proteins, which are absent from non-pathogenic members of the same or closely related bacterial species. Their G+C content and codon usage often differs in the flanking regions. PAIs were described on the chromosome, but some data suggest that special parts of virulence plasmids can also be defined as pathogenicity islands. The direct repeat (DR) ends of the pathogenicity islands are generated during the integration into the host genome and are able to promote excision further transfer. PAIs typically integrate into tRNA genes, and often encode cryptic or functional genes of mobilization factors: integrases, transposases and IS elements. As a consequence PAIs are instabile regions: their transfer and/or deletion is often mediated by the DR ends or by other mobile elements. General goal of my work was the analysis of the virulence plasmids of the ETEC strain 2173 (O147:NM, Hly, F18ac, STa, STb) (Nagy, B et al. 1990. J. Clin. Microbiol. 28, 651-653) and other ETEC strains causing postweaning diarrhoea-, or of verotoxic Escherichia coli (VTEC) strains Characterization and genotyping of F18 fimbria encoding plasmids by replicontyping Genetic analysis of the plasmid pTC bearing the sta and stb enterotoxin genes and a tetB tetracycline resistance gene of the strain 2173: primarily the sequencing and analysis of the origin Characterization of the linkage between the sta and stb enterotoxin genes of the pTC plasmid: in order to respond the question wether the pTC bears a larger mobile virulence element which encoding for more virulence genes similar to the pathogenicity islands in chromosomal MATERIALS AND METHODS
The general molcular biological methods were performed according to Maniatis, T., Fritsch, E. F., Sambrook, J., 1989. Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory Press, Second Edition, Cold Spring Harbor, New York. Characterization and replicon typing of F18 fimbria encoding plasmids
The 15 Hungarian E. coli strains tested were isolated from fatal cases of porcine postweaning diarrhoea or oedema disease in Hungary, and their phenotype and genotype were described earlier (Nagy et al. 1997) as ETEC (10 strains), as VTEC (4 strains), or ETEC/VTEC (one strain). Additional VTEC isolates were included as follows: 107/86 (F107 prototype strain from Switzerland, H. Bertschinger) and 2228 (F107+ strain from A. D. O’Brien, Bethesda, Md.). As negative control, we used the F18- VTEC strain (2206) and the TG1 K12 strain. The pathotype, O antigen, entero- and vero- (Shiga-) toxin genes and F18 fimbrial type of the strains were determined Plasmid DNA was isolated from the E. coli strains by a modified Kado and Liu extraction method. The plasmids were electrophoresed in 0.5% TAE agarose gels. The gels were dried for prehybridization and for the hybridization with the P32 labelled F18 and the Rep probes (using PHARMACIA Ready to Go DNA Labelling Kit). Hybridization was performed directly in the gel overnight on 60 Cº. After hybridization the we photographed the gels on X-ray films (AGFA) for 24 (sometimes 48 or 72 hours) developed at -80 Cº. Conjugational transfer of the F18 fimbria plasmid from the ETEC strain 2173
The donor NBI 2/11 (derivative of the 2173) and the recipient (E. coli K12 XL1Blue) bacteria were grown in 3 ml Luria-Bertani (LB) broth at 37Cº overnight. 100 µl of the donor and 100 µl of the recipient bacterial cultures were mixed on non-selective LB agar plates and incubated at 37Cº overnight. The bacterial lawn of the plate was suspended in 5 ml saline, and 100 µl were inoculated onto 5% sheep blood containing agar with 10 µg/ml final concentration of tetracycline added. After overnight incubation at 37Cº, the haemolytic colonies were collected, and after the confirmation of the TetR, Hly+ phenotype, one isolate proved to be suitable for further experiments. Conjugational transfer of the pTC plasmid from the ETEC strain 2173
To test the conjugation ability of pTC, the plasmid was introduced into the strains S17-1, TG1 and HB101 (HB101 does not harbour mobilisation or transfer functions, TG1 carries an F’ plasmid without the ability to be conjugated and S17-1 has a mobilisation function integrated in the chromosome). Subsequently we used the pTC cured Ec2173 derivatives NB-1/36, 2/2 and 2/11 as recipients. Transconjugants were selected on tetracycline containing GTS minimal agar plates. Cloning the origin of replivation of the pTC plasmid
To determine the origin of replication of the pTC it was digested with BamHI and the fragments were cloned randomly to a CmR cassette (cat) gene of the plasmid pAW302. As a result, the deletional pTC derivatives were transformed to the TG1 K12 strain. The transformants were able to multiply in chloramphenicol containing LB broth, in case the cat gene was ligated to a fragment containing the origin of replication of pTC. After physical mapping performed using restriction enzymes the resulting chloramphenicol resistance encoding pPFE1 plasmid was sequenced (EMBL Nucleotide Sequence Database; accession No: AJ627566). Examinations on the stability of the pTC plasmid
In the plasmid segregation experiments two or three plasmids with ColE1 type replicons were introduced into the E. coli K12 strain HB101 (see below). pEMBL19 (AmpR), pFOL5 (CmR)
pTC (TcR), pEMBL19 (AmpR), pFOL5 (CmR)
pPFE2 (KmR), pEMBL19 (AmpR), pFOL5 (CmR)
pPFE2 (KmR), pTC (TcR), pEMBL19 (AmpR)
Single colonies of the plasmid containing bacteria were inoculated into LB broth and were passed 5 times at 37°C in non-selective LB after overnight culture (representing about 50 generations). The number of plasmid containing bacteria were determined after every passage on antibiotics containg LB plates. The elimination of antibiotic resistance pattern was interpreted as a segregation event. Determination of toxin specific locus (TSL) of the pTC plasmid
DNA sequences were determined directly on the intact pTC or pAKR2 plasmid (purified by QIAGEN Plasmid Maxi kit) using the ABI PRISM® 310 Genetic Analyzer (Applied Biosystems) and the LI-COR DNA Sequencer model 400 automated sequencer (MWG Biotech) using primer walking. For comparison with NCBI GenBank database records blastx (BLAST Nucleotide query - Protein db), blastn (standard nucleotide-nucleotide BLAST) and BLAST 2 Sequences similarity search tools were used. We used the following elements of the BioEdit 5.0.9 Software package: ClustalW Multiple alignment for analysing raw nucleotide data and the CAP Contig assembly program for assembly of consensus sequences. ORF search and physical map construction of the 9872bp long pTC fragment was performed by the Vector NTI6 (InforMax Inc. 2000) package. The complete nucleotide sequence of the TSL of plasmid pTC has been submitted to EMBL Nucleotide Sequence Database and was given the accession number AJ555214. Transposition experiments using conjugation
In these experiments first we transformed TG1 K12 with the pTC (colE1 origin of replication) and afterwards with the pACYC177 (p15 origin of replication) plasmid. The resulting donor strain TG1/pTC+ pACYC177+ and the recipient (E. coli K-12 J5-3) bacteria were grown in 3 ml Luria-Bertani (LB) broth at 37 °C overnight. 100 µl of both of the donor and the recipient bacterial cultures were mixed on LB agar plates, which contained no antibiotics, and incubated at 37 °C overnight. The bacterial layer of the plate was suspended in 5 ml saline. This suspension was diluted (10-1-10-6) and 100 µl was inoculated onto LB agar plates with ampicillin, kanamycin and rifampicin added. After overnight incubation at 37 °C the AmpR, KmR, RifR colonies were collected, from which the TetS phenotype was searched. Three isolates (designated as AKR1, AKR2, AKR3) proved to be suitable for further analysis. In the partial sequencing of the pAKR2 transposition product we identified IS10 elements on both ends of the inserted fragment in the Examination on the distribution of the TSLpTC in F4(K88+) and F18+ ETEC strains
To determine the stb gene and its 5’ flanking regions of the TSLpTC by PCR we used primers is1rev- stbrev and 4521fw-4521rev primerpairs. The strains were origated from Hungary and Austria (12) RESULTS AND DISCUSSION
Number of virulence genes of pathogenic E. coli bacteria were found on uni- or bireplicon plasmids which are the members of the F incompatibility complex. It has been known that the adhesion factor genes of F18 fimbria expressing ETEC and VTEC strains can be found on plasmids, but the replication characteristics of these plasmids were not investigated so far. In our experiments we verified the plasmid localization of the F18 genes, and first time we determined the type of its origin of replication. Most of the examined strain we found 3 high molecular weight plasmid. In every cases a big (>36 kb) plasmid band showed hybridization signal with radioactive (P32) labelled F18 gene probe (originated from the 107/86 strain), showing that in all the ETEC (F18ac) and VTEC (F18ab) strains the adhezin gene was located on plasmids (including the reference strain 107/86). There was a difference in the size of the F18ab and F18ac plasmids. The F18ac plasmids of ETEC strains showed similar size (about 150 kb), which similarity raised the possibility of clonal origin and horizontal spreading of these plasmids. In contrast the size of the adhezin plasmids of VTEC strains (even from the same serogroup) differed between 63-150 kb, indicating that during the horizontal spreading not the intact plasmids but only the DNA fragments bearing the F18ab operon could move onto the plasmids of these strains. The results of replicon typing experiments confirmed the common origin of the F18ac plasmids: the adhezin (and haemolysin) plasmids of the ETEC strains belonged to the RepFI incompatibility group or to its subgroups: FIIa or I1. From this point of view there is only a partial similarity between F18 plasmids and the ETEC virulence plasmids described earlier. The size of the F18ac plasmid of the 2173 strain (which is a typical The results of the conjugation experiments done on the NB2/11 strain revealed that the α- haemolisin gene (which can be found mostly on the chromosome) was linked to the F18ac operon and was located on the same high molecular weight and (by our results) conjugative plasmid. Whereas the pF18 plasmid determines the adhesive phenomena of the ETEC strain 2173, the enterotoxin genes (sta and stb) corresponding for the postweaning diarrhoea can be found on its other plasmid with the approx. size of 120 kb (called pTC). To characterize the pTC it was necessary to get information about its stability, and about its plasmid incompatibility mechanisms in the strain 2173. Therefore we constructed two plasmid derivatives: pPFE1 and pPFE2. We sequenced the origin of replication of the pTC on the pPFE1, and found that it was a modified colE1, which had a deletion in the Rom (regulator) protein gene. To examine whether the deletion plays a role in the plasmid stability mechanisms of the 2173, we performed plasmid segregation experiments to compare the „survival” of the pTC, pPFE2 and other plasmids bearing an intact colE1 origin of replication (pEMBL19, pFOL5) in the HB101 K12 strain. In comparison with other plasmids (containing an colE1 origin of replication with a full Rom gene) based on their antibiotic resistance pattern after 50 division -we found the pTC or the pPFE2 (its kanamycin resistant derivative) in the majority of colonies. The pTC showed a stronger incompatibility effect: 97,55% of the colonies were TetR (characteristic to the pTC). In the case of the pPFE2 the incompatibility was weaker, however the plasmids bearing the „colE1pTC” origin prevailed altough but in decreased number. When we introduced both plasmids with „colE1pTC” origin aside the pEMBL19 we observed the survival both of pTC and pPFE2, and segregation of the pEMBL19. One interpretation of this could be that the gene product regulating the replication of the pTC recognize the replication origin of the pPFE2 to be the same that of pTC, and they act as a trans-regulators. In order to confirm, that the deletion of the rom gene contributes to the stability of the pTC further experiments will be required. The gene encoding STa enterotoxin was described earlier as part of the Tn1681 transposon, and the STb encoding gene is a part of the Tn4521. Based our earlier observations, that the same ETEC strain can produce both heat-stable enterotoxins (STa and STb) we assumed, there is may be a genetic linkage between the toxin. This linkage was indicated by the 10 kb TSLpTC what we found by the sequencing of the enterotoxin genes of the pTC plasmid. The sequence analysis of the 9872 nucleotide long TSL fragment revealed that sequence of both enterotoxin genes are equal to the sequences described for STa and STb earlier. In the case of sta gene we found DNA fragments specific for the Tn1681 on both flanking region, but in the case of the stb gene of the TSLpTC its flanking sequences are completely different from the Tn4521: we did not find the defective IS2 specifc sequences characteristic for the Tn4521. Upstream from the stb there was a full IS1 element, and their close linkage can indicate that the stb gene of the TSLpTC is a part of a new (non-Tn4521) virulence transposon. We didn’t find connection between the tetB tetracycline resistance gene and the TSLpTC. Downstream from the stb we identified a deleted pheV gene. The partial deletion of this tRNA gene can refer a recombinational event during the integration of a pathogenicity island. In the TSLpTC the toxin genes (in both direction) flanked by numerous (complete or deleted) mobile genetic element, which can suggest interbacterial gene- transfer events. Based on this data the structure of the TSLpTC seemed comparable to pathogenicity In the TSLpTC we also found DNA fragments homologous to genes originating from other E. coli patothypes or other Enterobacteriaceae species ( i. e: some parts of the PAI I536 of the UPEC strain 536, fragments of the pO157 virulence plasmid of EHEC strains, and the downstream sequences of the senA gene of Shigella flexneri and the pWR100 virulence plasmid). The sequence analysis of the stb gene of pTC plasmid revealed two point mutations: based on the nucleotid data: (performing a theoretical translation) there can be a histidin→asparagin mutation in amino acid position 12; and a lysin→isoleucin mutation in amino acid position 23 of the active toxin molecule. Altough the strain 2173 shows the characteristic toxic effect, these two mutation may alter the activity of the toxin molecule. To clarify this question we started a cooperation with a Canadian research group (which had earlier characterised the STb toxin receptor). As a result of the mobilization experiments we have transposed succesfully an approx. 40 kb fragment of the pTC plasmid into pACYC177. The transposition product (pAKR2) contained the TSLpTC. The succesful transposition confirmed our hypothesis, that the enterotoxin genes of the pTC are located on a mobile genetic element. This 40 kb fragment contained not only the sta, stb enterotoxin genes, and the pheV gene but also the colE1 type origin of replication and mob region which corresponds for the plasmid mobilization functions. The mobility of the 40kb fragment follows from the IS10 elements found on both ends. This confirms our hypothesis that the genetic element bearing the TSLpTC can be defined as a new plasmid-encoded enterotoxigenic pathogenicity island (PAI I2173). Sequencing the 40 kb PAI I2173 –what will be the next step in our investigations- will provide further data about its characters and its possible role in horizontal spreading of sta and Since our data suggested the existence of a new ETEC plasmid borne pathogenicity island including a toxin specific locus (TSL), we raised the question about the frequency of that TSL among porcine postweaning ETEC strains from different geographical origin.For that purpose we designed primers specific for the originally described Tn4521 (known to be absent from TSL) and for the detection of the stb gene and its 5’ flanking regions (known to be present in TSL). For the latter purpose, the is1rev-stbrev primer pair were used. As expected, the majority of F18+ strains (11/13) proved to be positive for the presence of the 5’ flanking region of stb and –with one exception- negative for the transposon Tn4521. On the other hand the Tn4521 was present in more than half of F4 (K88)+ strains (6/10). Some exceptions occurred in both directions. In consideration of our knowledge that the F18 and K88 plasmids can be found on plasmids differing in many characteristics our results not only point at a new, plasmid encoded pathogenicity island, but also suggesting the existence of the interaction between enteroxin and fimbrial gene encoding plasmids. Focusing on the virulence plasmids of ETEC strains and the function of the virulence genes we gained valuable informations that could be of further use in vaccine development studies. As a result of the plazmid analysis and replicon typing experiments Hungarian ETEC and VTEC strains isolated from porcine postweaning diarrhoea and oedema disease were found to harbour 2-3 high molecular size (>30 kb) plasmids. The F18 fimbria (and haemolysin) encoding plasmids belonged to the FIIa or I1 subgroups of the FIc incompatibility group. From this viewpoint there was only partial similarity between F18 plasmids and the porcine ETEC virulence plasmids, described earlier. The F18 plasmid of the ETEC 2173 strain proved to have a molecular Both of the enteroxin genes (sta, stb) of the ETEC 2173 strain were found on a 120 kb plasmid encoding for tetracycline (pTC). Sequencing of the origin of replication of pTC revealed a colE1 type origin (characteristic for the enterotoxin plasmids), however this plasmid contained a 463 bp long deletion in the mob region. Analyzing the genetic localization and the linkage between the sta and stb enterotoxin genes by primer walking sequencing we described and characterized a 40 kb sized new plasmid encoded pathogenicity island (PAI I2173), which was successfully mobilized together with its its toxin specific locus (TSLpTC), of approx. 10 kb. We found that the TSL seems to be a new common vector of the heat-stable enterotoxin genes (sta and stb) of ETEC from weaned pigs. In contrast to the sta gene (that we found as a part of the Tn1681 transposon), the genetic localization of the stb gene was completely different from the Tn4521 transposon described to carry stb earlier. Based on our experiments the TSL could be a new tool for the horizontal virulence gene transfer, characteristic for the F18+ ETEC strains, from different geographical origin. The atoxic variant of the strain 2173 (based on the features of the F18 and pTC plasmids) supplemented with a K88 fimbria encoding plasmid can serve as a starting point for a vaccine candidate strain in the prevention of porcine postweaning diarrhoea. THE THESIS IS BASED ON THE FOLLOWING PUBLICATIONS
Fekete, P. Z., Gerardin, J., Jacquemin, E., Mainil, J. G., Nagy, B., 2002. Replicon typing of F18 fimbriae encoding plasmids of enterotoxigenic and verotoxigenic Escherichia coli strains from porcine postweaning diarrhoea and oedema disease. Vet. Microbiol. 22, 275-284. (IF: 1,65)
Fekete, P. Z., Schneider, G., Olasz, F., Blum-Oehler, G., Hacker, J. H., Nagy, B. 2003. Detection of a plasmid-encoded pathogenicity island in F18+ enterotoxigenic and verotoxigenic Escherichia coli from weaned pigs. Int. J. Med. Microbiol. 293(4), 287-298. (IF: 1,362)
ADDITIONAL PUBLICATIONS IN THE SUBJECT OF THE DISSERTATION
Nagy, B., Fekete, P. Zs., 1999. Enterotoxigenic Escherichia coli (ETEC) in farm animals. Vet. Res. (IF: 1.49)
BOOK CHAPTERS
Nagy, B., Tóth, I., Fekete, P. Zs. 2004. Adhesins and receptors for colonisation by different pathotypes of Escherichia coli in calves and young pigs. In: Holzapfel, W. and Naughton P. eds.: Microbial ecology of the growing animal. 2003 Elsevier Science, Amsterdam, The Netherlands, in Élő, szájon át adható Escherichia coli vakcina készítésére alkalmas törzs a sertések választási hasmenésének megelőzésére és a törzs készítésére alkalmas eljárás. List number: 221 664 (based ont he statement Nr. P 99) Day of the statement and beginning of the protection: 31st of March, 1999. Patent entitled to: Veterinary Medical Research Institute of Hungarian Academy of Sciences, Budapest. Inventors: dr. Nagy Béla, Budapest 50%; Olasz Ferenc, Gödöllő, 30%; Fekete Péter Zsolt, Dunaharaszti, 15%; Tóthné Szekrényi Márta, Budapest 5% A strain suitable for producing a live, orally applicable Escherichia coli vaccine for the prevention of post-weaning diarrhoea in pigs, and the procedure suitable for producing that strain. International Patent submitted on 29th of March, 2000, Nr. PCT/HU00/00026 CONFERENCE ABSTRACTS
Fekete, P. Zs., Olasz, F., Szeverényi, I., Nagy, B., 1997. Characterization of the pTC plasmid from the enterotoxigenic E. coli strain EC2173, presentation, Abstracts of presentations and posters of the Annual Conference of Hungarian Society of Microbiology 1997., Szekszárd, p23 Fekete, P. Zs., Olasz, F., Szeverényi, I., Blum-Oehler, G., Nagy, B., 1998. Examinations ont he deletinal derivatives of pTC plasmid (EC2173), poster, Abstracts of presentations and posters of the Annual Conference of Hungarian Society of Microbiology 1998., Miskolc, p32 Fekete, P. Zs., Gerardin, J., Mainil. J. G., Nagy, B., 2000. Replicon typing and characterization of plasmids originated from Hungarian enterotoxigenic and verotoxigenic E. coli strains expressing F18 fimbriae. Abstracts of presentations and posters of the Annual Conference of Hungarian Society of Microbiology 2000., Szekszárd, p34 Fekete, P. Zs., Bluhm-Oehler, G., Hacker, J., Nagy, B., 2001. Partial characterization of a plasmid encoded „enterotoxigenic pathogenicity island” Abstracts of presentations and posters of the Annual Conference of Hungarian Society of Microbiology 2001., Balatonfüred, p49 Fekete, P. Z., Schneider, G., Olasz, F., Blum-Oehler, G., Hacker, J. H., Nagy, B. 2003. Detection of a plasmid-encoded pathogenicity island in F18+ enterotoxigenic Escherichia coli (ETEC) and verotoxigenic E. coli (VTEC) from weaned pigs. Scientific program & abstracts of the 11th European Conference on Bacterial Protein Toxins (ETOX11), Čelakovice, Czech Republic,Poster FEMS 2003 Fekete, P. Z., Schneider, G., Olasz, F., Blum-Oehler, G., Hacker, J. H., Nagy, B. 2003. Plasmid borne pathogenicity island of F18+ enterotoxic Escherichia coli. Abstract book of the 1st FEMS Congress of European Microbiologists, Ljubljana, Slovenia, P13-5 Nagy, B., Olasz, F., Fekete, P. Zs., Blum-Oehler, G., Hacker, J. 2004. Peculiarities of virulence plasmids of F18+ enterotoxigenic Escherichia coli causing postweaning diarrhoea. Proceedings of the 18th IPVS Congress, Hamburg, Germany, 2004 – Volume 1. p.259 CITATION (30 of June, 2004.)
Independent citations (in SCI journals) of the scientific publication of Fekete Péter Zsolt: 30

Source: http://www.vmri.hu/NB_pdfs/FPZs_PhD_thesis_2005_en.pdf