OUP user menu

The presence of genes homologous to the K88 genes faeH and faeI on the virulence plasmid of Salmonella gallinarum

I Rychlik, M.A Lovell, P.A Barrow
DOI: http://dx.doi.org/10.1111/j.1574-6968.1998.tb12869.x 255-260 First published online: 1 February 1998


A Tn3 insertion mutation was produced in the virulence plasmid of a strain of Salmonella gallinarum which conferred avirulence by parenteral and oral routes but which was also less invasive following oral inoculation. The transposon was found to have inserted near an open reading frame (ORF) with no homologies in the data banks. This ORF was adjacent to two additional ORF's with a high degree of homology to Escherichia coli genes encoding the minor structural subunits (FaeH and FaeI) of the K88 fimbria. A similar region of homology was found by DNA-DNA hybridization on the virulence plasmids of S. pullorum, S. dublin and other S. gallinarum strains but not in the plasmids of S. typhimurium, S. enteritidis or S. cholerae-suis.

Key words
  • Salmonella gallinarum
  • Virulence plasmid
  • faeH
  • faeI
  • Invasion

1 Introduction

In most of the Salmonella serotypes that characteristically produce typhoid-like diseases, such as S. typhimurium, S. enteritidis, S. dublin, S. gallinarum-S. pullorum and S. cholerae-suis, the ability to survive and multiply in the reticuloendothelial system is mediated by a large serovar-specific plasmid which is functionally similar in the different serotypes [1]. The plasmids all have a common region of approximately 8 kbp containing several spv genes that are sufficient to restore virulence to plasmid cured derivatives [1]. Their exact role in virulence, however, remains to be determined as regions outside the spv region have been found to be associated with the virulence of S. typhimurium for mice [2, 3]. About 20 kbp from the spv region, fimbrial biosynthesis genes termed pef (plasmid encoded fimbriae) have been identified [4]. The pef operon has been found to contribute to colonization of the ileal mucosa and to fluid secretion in the mouse gut [5]. By hybridization, pefA, encoding the major fimbrial subunit, has been shown to be present in S. enteritidis but not in S. dublin or S. gallinarum[6]. Although invasiveness of S. typhimurium and S. dublin into the Peyer's patches or mesenteric lymph nodes following oral inoculation are thought to be chromosomally mediated [1, 79], with S. gallinarum and its biotype S. pullorum, the virulence plasmid was shown to contribute to in vivo invasiveness [10, 11].

This paper reports the isolation of a transposon mutation in the virulence plasmid 15 kbp from the spv region which conferred reduced invasiveness on a strain of S. gallinarum. Sequence analysis in the region of the insertion indicated the presence of genes with homology to genes encoding minor structural subunits of the Escherichia coli K88 fimbrial adhesin.

2 Materials and methods

2.1 Bacteria

Salmonella gallinarum strain 9 produces typical fowl typhoid in chickens [10]. The strain was used as a spontaneous mutant resistant to nalidixic acid (Nalr) to facilitate enumeration in the alimentary tract. Broth cultures were grown in 10 ml LB broth (Difco) in a shaking incubator (150 revs min−1) at 37°C for 24 h. They contained between 1–3×109 cfu ml−1.

2.2 Transposon mutagenesis

Mutants with Tn3 insertions in the 85-kbp plasmid of S. gallinarum 9 were produced by introducing the temperature-sensitive plasmid pSC101 containing Tn3[10] by transformation [12] into a mutant of S. gallinarum from which its 2.2-kbp plasmid had been cured [10]. After incubation at the restrictive temperature (42°C) transposition was verified by ampicillin resistance and tetracycline sensitivity.

2.3 Animals

Specified-pathogen-free Light Sussex chickens were used at 3 weeks of age. Rearing conditions and diet have been described previously [13].

2.4 Virulence assays

Virulence by the oral route and LD50 estimations by the intramuscular route were carried out as described previously [10]. In vivo invasiveness was measured as follows. Chickens were inoculated orally with 108 cfu in 0.3 ml. At intervals thereafter groups of chickens were killed and organs removed aseptically in the order spleen, ileum and caeca. The spleen was homogenized in phosphate buffered saline (PBS). Ileal and caecal contents were removed by gentle squeezing. These organs were then opened and the residual contents removed from the mucosa with running water. After cursory drying of the mucosa with a clean towel the mucosa was scraped off with a scalpel blade. The caecal tonsil was removed separately. Viable count estimations of inoculated bacteria in tissue homogenates were made by the method of Miles et al. [14] counting Salmonella organisms on brilliant green agar (Oxoid CM263) containing sodium nalidixate (25 μg ml−1) and novobiocin (1 μg ml−1).

2.5 Cloning and sequencing

An ApaI fragment from the virulence plasmid, which included the site of insertion, was cloned using standard procedures [15]. The 3.2-kbp insert was subcloned in four fragments into pBluescript using EcoRI and PstI digestion and sequenced using a Sequenase 2.0 sequencing kit according to the manufacturer's instructions (Amersham). All searches and nucleotide sequence comparisons were made at the internet address http://www.ncbi.nlm.nih.gov. Open reading frame (ORF) searching and other sequence editing was done with the GenCompar software package (Applied Maths, Kortrijk, Belgium).

Southern hybridization was carried out using non-isotopic methods following standard commercially recommended protocols (Amersham).

3 Results

3.1 Isolation of Tn3 mutant SG9 54.4

Of 924 Tn3 mutants examined for virulence for chickens, four were found to be avirulent and to contain single transposon insertions in the virulence plasmid by an increase in restriction fragment size detected by electrophoresis after SalI digestion. Following oral inoculation the parent strain produced 60% mortality with the chickens showing typical signs of fowl typhoid. All four mutants produced no morbidity or mortality. In contrast to the parent strain whose LD50 value by the intramuscular route was less than (log10) 1.0, all four mutants had LD50 values of greater than (log10) 8.0.

The restriction map of the region of the plasmid containing the four insertions is shown in Fig. 1. Three of the Tn3 mutations mapped together on the same 4.6-kbp ClaI-HindIII fragment which also hybridized with the 8-kbp essential virulence region containing the spv genes. Insertion 54.4 was situated 15 kbp away from the spv region.

Figure 1

Restriction map of approximately 40 kbp of the S. gallinarum virulence plasmid showing the site of the four Tn3 mutants which resulted in avirulence. The spv region is indicated as is the orientation of Tn3 in each mutant.

Plasmid preparations from a variety of virulence plasmid-containing and non-containing Salmonella serotypes and other bacteria were electrophoresed, blotted and hybridized with the 9.1-kbp SalI S. gallinarum fragment containing the 54.4 Tn3 insertion. Hybridization to an identical sized fragment occurred with 5/5 S. gallinarum, 5/5 S. pullorum and 4/5 S. dublin strains (the negative S. dublin strain did not contain a virulence plasmid). A total of 18 strains of S. cholerae-suis, S. enteritidis and S. typhimurium, 8 individual strains of other Salmonella serotypes and strains of E. coli and Citrobacter spp. gave negative results.

3.2 In vivo invasive phenotype of mutant 54.4

The results of assessing the mutant for invasiveness in chickens are shown in Table 1. Both strains were isolated from the caecal contents by one day post-infection. The viable numbers of the parent strain were generally higher than those of the mutant. By two days the parent strain was also isolated from the spleen and liver and the viable numbers from the mucosa and lymphoid tissue were consistently high. By contrast the mutant was isolated less frequently from these sites.

View this table:
Table 1

In vivo measurement of invasiveness by Tn3 mutants of S. gallinarum 9 Nalr following oral inoculation

Time (h) after infectionStrain or mutantLog10 median no. of Salmonella organisms g−1 of the following samples (3 birds sampled at each time)
16ParentNN2.5N (3.3)a
7.13NNN (2.3)N (3.2)
5.3NNN (2.3)N (2.0)
7.13N (2.0)N (1.2)NN
5.3NN (2.0)N (3.4)N (4.9)
54.4NNNN (2.0)
64Parent4.14.6NN (2.0) (2.3)
54.4NNN (2.0)N (2.0)
88Parentb (2.3)3.9
54.4NN (1.5)NN (2.6)
  • aN=<1.0. Figures in parentheses indicate highest count if median count=N.

  • bNot done.

3.3 Sequencing analysis of mutant 54.4

The total length of sequence determined was 3252 bp. The site of the Tn3 insertion was at position 466 bp. Three possible open reading frames (ORFs) were found.

The first started at position 1072 with GTG as the start codon or at position 1123 with ATG as the start codon, respectively, and ended at position 1392 with a predicted peptide length of either 107 or 90 amino acids. No homologies were found in the databases either at the nucleotide or amino acid level.

The second ORF started at position 1631 and ended at 2425 bp with a predicted protein length of 265 amino acids. At the nucleotide level this sequence showed 75% homology to the faeH gene encoding a minor structural subunit of the K88 adhesin of E. coli. There was 84% similarity and 76% identity at the amino acid level.

The third ORF started at position 2456 and ended at position 3217 bp with a predicted protein length of 254 amino acids. At the nucleotide level this ORF showed 70% homology to the faeI gene encoding another minor structural subunit of the K88 adhesin. There was 77% similarity and 70% identity at the amino acid level.

Since no ORF was identified at the site of the Tn3 insertion in mutant 54.4 further sequence analysis was carried out. From position 0 to 1666 the average GC content was only 38.42% whereas from position 1666 to the end the content was 63.82%. In addition, two sequences containing direct repeats of different lengths were identified starting at positions 209 and 966, respectively (Fig. 2).

Figure 2

Comparison of sequence of fragments starting at positions 209 and 966 with flanking regions. The direct repeats are overlined. Hatching indicates base identity in identical relative positions in the two regions.

4 Discussion

We have found that the virulence-associated plasmid of S. gallinarum encodes not only the ability to survive and multiply in the cells of the reticuloendothelial system but also aspects of the intestinal phase of infection [10]. A Tn3 insertion which mapped 15 kbp away from the spv region virtually eliminated intestinal invasiveness. In the S. pullorum biotype of S. gallinarum the plasmid has also been found to contribute to the intestinal phase of infection [11]. These results contrast with murine and bovine models of S. typhimurium and S. dublin[1, 79] where the plasmid does not seem to be essential for invasion from the intestine to deeper tissues.

The reduced invasiveness of the Tn3 mutant can also be attributed to reduced adhesion to the enterocytes and/or M cells in the intestine. The reduced virulence after parenteral inoculation suggests reduced spv gene expression. As no ORF was identified at the exact site of insertion and two separate phenotypes were affected, it may be that plasmid gene expression may have been affected by the transposon insertion. This was also suggested by AT richness at the insertion site.

Two ORF's, highly homologous to the faeH and faeI of the K88 fimbria gene cluster of enterotoxigenic E. coli, were identified 1.2 kbp from the Tn3 insertion site in a region which was GC rich. Although the role of faeI in the biosynthesis of K88 is unknown faeH deletants produced fimbriae in reduced number [16]. Genes with homology to fae have already been found in S. typhimurium[6] but none with homology to faeH or faeI. Other homologies to E. coli fimbrial genes were also found, in one case to clpH of the CS31 fimbria (Der Vartanian, unpublished data, GenBank accession number M96152) and in another to ralH in another fimbrial operon (Adams et al., unpublished data, GenBank accession number ECU84144). This suggests a more general role for these genes in fimbrial biosynthesis in E. coli and different Salmonella types.

No hybridization was obtained to plasmid preparations from either S. typhimurium or S. enteritidis using the 9.1-kbp SalI fragment which contained the whole sequenced region. This suggests that the pef genes in these two serotypes are quite different from those present in S. gallinarum and also, in all probability, from those in S. pullorum and S. dublin. This suggests that in disease in the appropriate target host the contribution of these genes to the initial stages of infection produced by S. dublin and S. pullorum would also be worth while investigating.


The authors would like to acknowledge the financial support of the European Community (mobility of researchers scheme) and FEMS.


View Abstract