Annex O3.6:

MOLECULAR CHARACTERISATION OF PERSISTENT AND

NON-PERSISTENT SE STRAINS USING MICROARRAY

Methods

Bacterial strains and genomic DNA isolation

Five S. Enteritidis PT4 strains were chosen from farms on which Salmonella had been persistently found (7914/03, 1851/03, 5090/06, from Farm 11 and 6714/02, 1890/06 from Farm 5) and 4 S. Enteritidis PT4 strains were chosen from farms without a persistence problem (8718/99 from Farm 4, 6469/02 from Farm 8, 6027/00 from Farm 3, and 219/02 from Farm 7). Strains were grown overnight in LB broth and genomic DNA isolated using a Dneasy tissue kit (catalog no. 69504; QIAGEN).

Comparative genomic hybridization (CGH) Microarray

The microarray used in this study comprises specific unique PCR products (100–500 bases) of 4097 CT18 genes, plus ~1900 specific gene segments representing the S.bongori,S.Gallinarum, S. Enteritidis PT4, S. Typhimurium SL1344, DT104, LT2 strains. Appropriate positive and negative controls, such as Cy3 and Arabidopsis oligos were also printed on the arrays. Two micrograms of test and control DNA was labeled using BioPrime DNA labeling System kit (Invitrogen) following manufacturer’ instructions. Cy3/Cy5- dCTP was incorporated into the DNA. Previous protocols used in our laboratory for hybridization and scanning of microarrays were followed (see Anjum et al, 2005 for details).

CGH microarray Data Analysis

Spots with a value for median signal intensity minus the background below 50 (in both the reference and test channel) were discarded due to low signal intensity leading to ambiguous results. Per spot per chip intensity-dependent Lowess normalization was performed to compensate for unequal dye incorporation. The ln(Cy5/Cy3) ratio intensity of all spots from the hybridization was calculated and the mean ln(Cy5/Cy3) from up to eight data points per gene (i.e. two slides with four repeats per slide), was used in all subsequent data analyses. The relationship between strains was calculated by average-linkage hierarchical clustering using the Pearson Correlation Coefficient as part of the GeneSpring microarray analysis software V7.0 (Silicon Genetics).

Multilocus sequence typing (MLST)

Three house keeping genes (hisD, aroC, pduF) and two virulence genes (fliC, hilA), were chosen for MLST. Primers previously described () were chosen to PCR amplify products from all 9 strains. PCR was performed once at 94°C for 4 min, 30 cycles at 94°C for 30 sec, 30 sec annealing at temperatures between 50 to 55°C and 72°C for 1 min. Sequencing was performed using a deoxy terminator cycle sequencing kit (Perkin-Elmer, Applied Biosystems) with both forward and reverse primers for each gene.

Results

CGH Microarray studies

The genetic profile of four persistent (7914/03, 1851/03, 6714/02, 1890/06), two non-persistent (6469/02, 219/02) and a positive control strain (PT4 genome sequenced) was compared by CGH microarray for ~6000 genes. Hierarchical clustering of the CGH microarray data using the Pearson Correlation Coefficient is shown in Figure 1. It shows that strains 1851/03, 6714/02 and 1890/06, which fall within the persistent group, are genetically more similar to each other; whilst strains 6469/02, 2169/02 and PT4 show greater similarity. However, to interrogate the data further for variably absent and present genes, all genes commonly present or absent in the 7 strains were taken out. This resulted in 79 variable genes to be left in the data set (Fig. 2a).

A more detailed analysis of the genes within the variable data set showed a group of contiguous genes (STY1534 – 39) to be absent only from strains 6714/02 and 1890/06 (Fig. 2b). These strains were collected from the same farm but at different times (2002 and 2006). The functions of the genes absent in both strains are: putative membrane protein (STY1534); conserved hypothetical protein (STY1535); putative aldehyde-dehydrogenase (STY1536); putative regulatory protein (STY1537); putative membrande transport protein (STY1538); and putative membrane protein (STY1539). The absence of these genes may infer an insertion/deletion event whereby another gene fragment has been gained in this group of strains enabling them to persist longer in farm environments.

We also identified a single gene (STY1592) that was absent in all 4 persistent strains included in our study but present in the 2 non-persistent strains (Fig. 2b). This gene has been annotated as a conserved hypothetical protein of unknown function. Its absence in the persistent strains may again be an indicator of a gene exchange or a deletion event, as these strains adapted to specific niche.

A cluster of 10 genes (SG0266, SG0272-75, SG1185, SGP0052-3, SGP0095, SLP2_0017, SLP2_0043) was identified to be absent in all 6 Enteritidis isolates included in our study, but present in the sequenced Enteritidis PT4 strain (Fig. 2c). Several of these genes were S. Gallinarum chromosomal and plasmid associated genes; whilst two were plasmid encoded genes from a Typhimurium sequenced strain (SL1344). Both Enteritidis and Gallinarum are Group D Salmonella and therefore it is interesting to note that several Gallinarum associated genes that were present in the sequenced PT4 strain are absent from more recent farm isolates collected between 1999 and 2006.

The data from this work was compared to CGH microarray data compiled from 24 Salmonella Enteritidis strains of phage types 4, 6, 8, 13a, 9 and 11 from previous work (Pan et al, unpublished data). In particular the variable presence of the genes described above were studied. Figure 3 shows that two strains (6, and 50) also had the region STY1534 – 39 missing; strain 6 was phage type 4, whilst strain 50 was phage type 9. It was interesting to note that the cluster of 10 genes previously found to be present only in the PT4 sequenced strain was mainly absent from this group of 24 strains. However, one gene STM0923 was present in all six PT11 strains included in our study.

MLST analysis

MLST analysis of all nine strains included in this study showed no variability within their sequence profile except in strain 2169/02 which showed a single base substitution (G to T transition) in the fliC amplicon. This did not change the MLST sequence type of the strain but did result in change of the codon from UUC to GUC i.e. a phenylalanine to valine substitution.

Conclusion

In this study genomic and phylogenetic characterisation was undertaken for several persistent and non-persistent strains collected from farms in England from 1999 to 2006. Several genes have been identified that may be implicated in presence of persistent Salmonella Enteritidis strains on farm. However this study has been performed on small subset of strains and future work on a larger number of strains is required to verify these findings. If these findings are substantiated then it will indicate that genetic changes associated with on farm persistence of Enteritidis PT4 strains may occur. These genes may be monitored in future by a simple PCR or high through-put microarray as part of routine surveillance to determine the likely significance of new and emerging strains.

Reference

Anjum, M.F., Marooney C., Fookes, M., Baker, S., Dougan G., Ivens, A., Woodward M.J. (2005). Identification of core and variable components of the Salmonella enterica subspecies I genome by microarray. Infect Immun.73:7894-78905.

1

1