Identification and distribution of accessory genome DNA sequences from an invasive African isolate of Salmonella Heidelberg


  • Editor: Stephen Smith

Correspondence: Craig Winstanley, Division of Medical Microbiology, University of Liverpool, Duncan Building, Daulby Street, Liverpool L69 3GA, UK. Tel.: +44 151 706 4388; fax: +44 151 706 5805; e-mail:


Nontyphoidal salmonellae (NTS) are a leading cause of invasive disease in young children in sub-Saharan Africa. We used suppression subtractive hybridization (SSH) to identify 41 sequences within the accessory genome of an invasive strain of Salmonella Heidelberg from Malawi. PCR assays and database searches, used to determine the distribution of 14 SSH sequences among a panel of African and UK NTS isolates and published genomes, indicated that two were specific for S. Heidelberg. However, we found no evidence for major differences in the accessory genome content between African invasive and gastrointestinal isolates of S. Heidelberg. Six of the SSH sequences were within fimbrial operons. The tcf operon, associated with the host specificity of Salmonella Typhi, and the stk operon, reported previously in Salmonella Paratyphi, were both present in either all (tcf) or most (stk) isolates of S. Heidelberg, but had restricted distributions among the other serovars tested. Reverse transcription PCR analysis of seven SSH sequences indicated variable expression of the stk operon among isolates of S. Heidelberg. Three of the seven targeted genes were not expressed in a UK veterinary isolate of S. Heidelberg, suggesting that although genome content per se may not explain the different pathogenicity of the invasive isolates, it is possible that variations in gene expression may play a role.


Nontyphoidal salmonellae (NTS) are a common cause of meningitis and septic arthritis in children (Graham, 2002; Lavy et al., 2005), and of bacteraemia in both children and HIV-infected adults (Bahwere et al., 2001; Gordon et al., 2001; Berkley et al., 2005; Gordon, 2008) in sub-Saharan Africa. However, the factors determining this invasive pathogenicity are not well understood. There is some evidence that clinical NTS isolates from Kenya differ from isolates taken from livestock and the environment of patients in both their genetic profile, as determined by molecular typing, and their antibiotic susceptibility patterns (Kariuki et al., 2002, 2006). This implies an alternative route of transmission of African NTS, compared with North America and Europe, where domestic animals represent the major reservoir of NTS and foodstuffs of animal origin are the vehicle of human infection (Threlfall et al., 2000).

Salmonella Heidelberg is one of the more virulent and invasive Salmonella serovars, causing extraintestinal infections associated with severe disease symptoms, such as myocarditis and bacteraemia (Rice et al., 1976; Burt et al., 1990; Wilmshurst & Sutcliffe, 1995). As well as being common among gastrointestinal isolates, S. Heidelberg has been reported as the most common Salmonella serovar among blood culture isolates in Canada (Demczuk et al., 2003), and the third most common serovar causing invasive NTS infection, after Salmonella Typhimurium and Salmonella Enteritidis, in the United States (Vugia et al., 2004). The WHO Global Salmonella Survey (GSS) showed that between 2000 and 2004 S. Heidelberg was the fourth most common serotype among human isolates, and the second most common serotype among non-human isolates in the world. However, because African countries participating in the WHO GSS are predominantly located in West Africa (Galanis et al., 2006), data on NTS from sub-Saharan Africa are limited.

Suppression subtractive hybridization (SSH) is a method for the identification of DNA sequences present within the genome of one strain but absent from the genome of another (Winstanley, 2002), and it has been used previously on several Salmonella serovars, including Typhimurium and Enteritidis (Agron et al., 2001; Kang et al., 2006). The method is ideal for identifying genomic regions that vary between strains of the same species (comprising the ‘accessory genome’). In this study, we report the use of SSH to screen for sequences in the accessory genome of an African paediatric bacteraemia isolate of S. Heidelberg (strain D23734), and report the distribution of such sequences among a collection of Salmonella isolates including African NTS, S. Heidelberg isolates from Europe, and previously sequenced strains of Salmonella enterica.

Materials and methods

Bacterial strains used in this study

African NTS isolates were obtained from existing collections maintained in our University from Uganda, Malawi, Kenya and the Democratic Republic of Congo (Zaire). UK isolates were obtained from human and animal faeces (Table 1). Other isolates were obtained from John Wain (Health Protection Agency, Colindale, London). For UK isolates, serovar designations were confirmed by the Salmonella Reference Laboratory, Colindale, London. For the Malawi isolates, serovars other than S. Typhimurium and S. Enteritidis were confirmed by the National Salmonella Reference Laboratory, Galway, Republic of Ireland.

Table 1.   Summary of African and UK Salmonella strain sets
Country of originSerovarNumberDetails and sources
UgandaTyphimurium9Adult diarrhoea isolates (M. Okong, unpublished data)
MalawiTyphimurium17Children bacteraemia isolates 1998–2004 (Gordon, 2008)
KenyaTyphimurium4Adult bacteraemia isolates, 1994–2003 (Kariuki et al., 2005)
DRC (Zaire)Typhimurium7Children bacteraemia isolates (Green & Cheesbrough, 1993)
UKTyphimurium6Adult diarrhoea isolates (C. Parry, unpublished data)
UK VeterinaryHeidelberg3Veterinary isolates (P. Wigley, unpublished data)
UnknownHeidelberg3Gut (Sanger Institute, unpublished data)
ZanzibarHeidelberg1Gut (Sanger Institute, unpublished data)
KenyaHeidelberg1Blood (Sanger Institute, unpublished data)
TanzaniaHeidelberg1Faeces (Sanger Institute, unpublished data)
NigeriaHeidelberg1Gut (Sanger Institute, unpublished data)
PeruHeidelberg1Gut (Sanger Institute, (unpublished data)
ThailandHeidelberg1Gut (Sanger Institute, unpublished data)
MalaysiaHeidelberg1Gut (Sanger Institute, unpublished data)

Construction and screening of subtraction libraries

The SSH tester strain was the bacteraemia isolate D23734, from a Malawian child. The serotype of the tester strain was confirmed by both serology and flagellin gene sequencing. The driver strain used in the SSH was the laboratory strain S. Typhimurium LT2 for which the complete genome sequence is known (McClelland et al., 2001).

Genomic DNA was isolated from S. enterica strains using the Wizard Genomic DNA Purification kit and following the manufacturer's protocol (Promega). SSH was carried out using the Clontech PCR-Select Bacterial Genome Subtraction kit, as recommended by the supplier (Clontech). Tester and driver strain DNAs were digested with RsaI and the PCR amplicons obtained following SSH were cloned into pGEM-T (Invitrogen). The subtraction libraries of RsaI fragments thus constructed were screened by sequencing of plasmid DNA extracted from individual clones using forward (M13-F: 5′-ACGTTGCACAATCCGGAT-3′) and reverse (M13-R: 5′-CCACCGAAGAAGGAGCAA-3′) vector primers (Cogenics Lark). In order to identify genuinely subtracted sequences, blastn searches targeting the genome of S. Typhimurium LT2 were conducted. Sequences sharing >90% identity with the driver genome were omitted from further study. Sequences sharing <90% identity with the genome of the driver strain were further analysed using blastn and blastx searches of the general database. Similar blastn searches were used to determine the presence or the absence of SSH sequences from the genomes of published Salmonella genomes, including S. Heidelberg SL476 and SL486. All searches were performed using the NCBI website (

PCR amplification screening of strains

Oligonucleotide primers (Sigma-Genosys) for PCR screening using amplifications are listed in the Supporting Information, Table S1, along with the annealing temperatures used. DNA for PCR amplification was prepared by boiling a suspension of a few colonies in 5% 200 μL Chelex-100 (Bio-Rad) for 5 min. After centrifugation, 150 μL was removed and stored at −20 °C. For PCR amplification, typically 1 μL DNA was used directly in 25-μL volumes containing 1.25 U GoTaq DNA polymerase (Promega), 300 nM each primer, GoTaq buffer, 2.5 mM MgCl2 and 100 mM nucleotides (dATP, dCTP, dGTP, and dTTP). Amplifications were carried out in an Eppendorf MasterCycler thermal cycler for 30 cycles consisting of 95 °C for 1 min, annealing temperature for 1 min and 72 °C for 2 min, with an additional extension time at 72 °C for 10 min following completion of the 30 cycles.

Preparation of cDNA for reverse transcription (RT) PCR amplification

A subset of six S. Heidelberg strains was used to determine the expression of SSH sequences. Cells were gown overnight in Luria broth at 37 °C with shaking at 200 r.p.m. Cells were collected by centrifugation of 1 mL of overnight culture at 7000 g for 2 min. RNA extraction was performed using the RiboPure Bacteria whole RNA isolation kit (Ambion, Applied Biosystems) following the manufacturer's instructions, except that two rounds of the recommended DNAse I treatment of the RNA samples were performed. The RNA was converted to cDNA using the SuperScript II RT kit (Invitrogen). For each RNA preparation, a control reaction lacking the reverse transcriptase was also prepared. We confirmed that expression of the flagellin gene (fliC) was detectable in all cDNA samples, but not in any of the controls.


Identification of genetic sequences present in the Malawian S. Heidelberg isolate D23734 but absent from the S. Typhimurium strain LT2

In order to identify novel sequences that might be associated with the invasive phenotype of S. Heidelberg D23734, we carried out SSH using S. Typhimurium LT2 as the reference (driver) strain. We chose this strategy with the aim of identifying novel genes common to both invasive and gastrointestinal isolates of S. Heidelberg, as well as genes specific to the African invasive isolates. Following SSH, a total of 114 clones were sequenced, resulting in the identification of 52 (46%) genuinely subtracted sequences, 11 of which were repeated more than once, leaving an output of 41 different S. Heidelberg D23734 subtracted sequences. A summary of all the SSH sequences is given in Table S2, organized according to their putative function as determined by the blastx search. Although many of the SSH sequences matched hypothetical proteins, among a number of matches of potential relevance to virulence were six related to fimbriae, one related to a putative autotransporter, one related to a lipoprotein, two related to transcriptional regulators and one related to a ferrichrome-iron receptor.

Of the 41 SSH sequences, 37 (88%) matched at least partly one or both of the genome-sequenced S. Heidelberg strains SL476 or SL486, and the other four matched a Salmonella serovar other than S. Heidelberg. Using blastn, the genome of Salmonella Typhi CT18 was screened for the presence of the SSH sequences. Nineteen (46%) of the 41 SSH sequences were found to be at least partly present in S. Typhi CT18.

Distribution of SSH sequences among a panel of African and UK NTS isolates

In order to determine the distribution of sequences identified in the genome of S. Heidelberg D23734 using SSH, we designed oligonucleotide primers to 14 of the subtracted sequences and screened a panel of African, Tropical and UK human and veterinary NTS isolates using PCR assays (Table 2). The SSH sequences chosen for further study were selected on the basis of (1) potential roles in virulence according to the best blastx matches (fimbrial, autotransporter, iron receptor, transcriptional regulator, and lipoprotein) and (2) matches to enzymes or hypothetical proteins with limited distributions among Salmonella serovars according to blastx searches.

Table 2.   Distribution of selected SSH sequences according to PCR assays
SSH sequence putative function based on blastx
(and sequence identifier)
Sub-Saharan African and Tropical IsolatesUK isolates
S. TyphimuriumS. EnteritidisS. HeidelbergS. Bovis-morbificansOther Salmonella serovarsS. TyphimuriumS. EnteritidisS. HeidelbergS. Bovis-morbificans
  1. The table shows the number of strains that tested positive for a given sequence by PCR assay; percentages are given in parentheses. The group of sub-Saharan African and Tropical S. Heidelberg isolates includes the tester strain D23734. +,All of the strains (100%) were PCR positive; −, all of the strains tested were PCR negative.

2-Nitropropane dioxygenase NPD (D23734-4)+1 (33)+
Short-chain dehydrogenase/reductase (D23734-10)++
Putative autotransporter/pertactin (D23734-14)11 (92)+1 (33)+
Putative fimbrial protein stkD (D23734-A1)8 (67)2 (66)
Putative fimbrial protein tcfA (D23734-A5)+2 (66)+
Fimbrial subunit (D23734-G3)+10 (83)++++
Probable lipoprotein (D23734-B10)7 (58)+2 (66)1 (17)++
Ferrichrome-iron receptor (D23734-D2)+10 (83)1 (33)++
Transcriptional regulator tinR (D23734-D4)4 (15)++1 (17)+11 (25)
Conserved hypothetical protein (D23734-B9)++
Restriction enzyme (D23734-B2)++
Hypothetical protein (D23734-5)++
Hypothetical protein (D23734-2)++
Hypothetical protein (D23734-1)11 (92)2 (66)++

Our collection of invasive NTS isolates from Africa is dominated by S. Typhimurium and S. Enteritidis. In order to screen for common sequences among NTS invasive isolates in this collection regardless of serotype, we included greater numbers of these more common serovars in the panel, alongside S. Heidelberg and other less common serotypes. However, none of the invasive S. Typhimurium isolates tested positive for any of the subtracted sequences. SSH sequence D23734-G3 represents a putative fimbrial subunit present in a restricted number of serovars. All S. Enteritidis and Salmonella Bovismorbificans, and the majority of S. Heidelberg isolates, were PCR positive for this sequence (Table 2). Sequence D23734-G3 shares 94% identity with a putative gene identified in the S. Enteritidis phage type 4 strain P125109 (SEN2799; putative fimbrial subunit protein) (Thomson et al., 2008).

According to the PCR assays, two other fimbriae-related sequences (representing tcf and stk fimbriae, respectively) were either present only in S. Heidelberg isolates (stk) or found in all S. Heidelberg isolates, but also in some other serovars (tcf). Either S. Enteritidis or S. Bovismorbificans was PCR positive for three other SSH sequences, but all S. Enteritidis and S. Bovismorbificans isolates were PCR negative for 10 of the SSH sequences (Table 2). For eight of the 14 SSH sequences, all S. Heidelberg isolates were PCR positive. We found no evidence among the African isolates for distribution according to geographical source (data not shown).

Distribution of SSH sequences among genome-sequenced Salmonella serovars

We searched Salmonella genomes in the database for the presence of the 14 SSH sequences. Thirteen of the SSH sequences were found to be present in the genomes of both S. Heidelberg strains SL476 and SL486. However, D23734-B10, part of a gene encoding a putative lipoprotein, was absent from both genomes. After S. Heidelberg, D23734 SSH sequences were most commonly found in Salmonella Paratyphi A (Table 3).

Table 3. blastx screening results for D23734 SSH sequences
SSH sequence putative function (and sequence identifier)S. Heidelberg (SL476)S. Heidelberg (SL486)S. Choleraesuis (SC-B67)S. Typhi (Ty2)S. Typhi (CT18)S. Agona (SL483)S. Javiana (GAMM 040433)S. Kentucky (CDC191)S. Kentucky (CVM29188)S. Saintpaul (SARA23)S. Saintpaul (SARA29)S. Schwarzengrund (CVM19633)S. Schwarzengrund (SL480)S. Newport (SL254)S. Newport (SL317)S. Paratyphi A (ATCC9150)S. Dublin (CT_020211853)
2-Nitropropane dioxygenase NPD (D23734-4)+++++++
Short-chain dehydrogenase/reductase (D23734-10)++
Putative autotransporter/pertactin (D23734-14)+++++
Putative fimbrial protein stkD (D23734-A1)+++++
Putative fimbrial protein tcfA (D23734-A5)++++++++
Fimbrial subunit (D23734-G3)++++
Probable lipoprotein (D23734-B10)++++++
Ferrichrome-iron receptor (D23734-D2)++++++
Transcriptional regulator tinR (D23734-D4)+++++++++
Conserved hypothetical protein (D23734-B9)++
Restriction enzyme (D23734-B2)++
Hypothetical protein (D23734-5)+++++
Hypothetical protein (D23734-2)+++++
Hypothetical protein (D23734-1)+++++++++++

Expression of genes represented by SSH sequences

A subset of six S. Heidelberg strains was chosen for RT-PCR screening to detect the expression profile of seven of the SSH sequences. According to PCR assay results, each of the six strains contained each of the seven SSH sequences tested. The results are shown in Table 4. The majority of SSH sequences lie within genes expressed in all of the African and Tropical S. Heidelberg isolates. The exception was SSH sequence D23734-A1 (putative stk fimbrial gene), for which expression could not be detected in three of the isolates. For three of the SSH sequences, expression could not be detected in the UK veterinary isolate KMS1977 (Table 4).

Table 4.   Expression profile of seven SSH sequences among a panel of six Salmonella Heidelberg strains
SSH sequence putative function (and sequence
(UK Veterinary)
Putative autotransporter/pertactin (D23734-14)++++++
Putative fimbrial protein stkD (D23734-A1)+++
Putative fimbrial protein tcfA (D23734-A5)++++++
Fimbrial subunit (D23734-G3)+++++
Probable lipoprotein (D23734-B10)++++++
Ferrichrome-iron receptor (D23734-D2)+++++
Transcriptional regulator tinR (D23734-D4)+++++


We used SSH to identify 41 sequences that were present in the accessory genome of African bacteraemia isolate D23734, but absent from S. Typhimurium LT2. Our results suggest that there is little variation between the genomes of African invasive and faecal isolates of S. Heidelberg. We did identify one SSH sequence (D23734-B10) lacking from the two genome-sequenced S. Heidelberg strains but present in the majority of African isolates. The presence of this sequence in some other invasive serovars, such as Typhi, suggests that the genomic region represented by this sequence may merit further study. However, we found the SSH sequence to be present also in UK veterinary isolates of S. Heidelberg, and demonstrated the expression of the putative lipoprotein-encoding gene in all six isolates tested, including one of the veterinary isolates. Hence, there was no clear association of either the presence of the gene or the expression of the gene with the invasive isolates. Likewise, the potentially virulence-related putative autotransporter/pertactin-related SSH sequence D23734-14 was found in 92% of the African/Tropical S. Heidelberg isolates and all of the UK veterinary isolates, and was expressed in all the S. Heidelberg isolates tested. Interestingly, based on PCR assays, this sequence was present in the five African isolates of S. Bovismorbificans, but absent from UK gastrointestinal isolates of this serovar. The sequence was also present in some other invasive serovars. However, overall, there was no clear correlation between SSH sequence distributions and those serovars generally regarded as more virulent/invasive, including serovars Cholaerasuis, Dublin, Schwarzengrund and Newport (Threlfall et al., 1992; Chiu et al., 2006).

Salmonella Heidelberg is in the same serogroup as S. Typhimurium (serogroup B, antigen profile 1,4,5,12 : r : 1,2). In a study of the clonal diversity of eight NTS serovars, using enzyme electrophoresis to detect allelic polymorphisms, it was reported that S. Heidelberg, when compared with serovars Choleraesuis, Dublin, Derby, Enteritidis, Typhimurium, Infantis, and Newport, showed the least diversity among strains from Europe and the Americas (Beltran et al., 1988). However, we detected variations between African/Tropical isolates of S. Heidelberg with respect to six of the SSH sequences used for PCR assays.

Two of the SSH sequences (D23734-B2 and D23734-B9) were present in all S. Heidelberg isolates and in none of the other Salmonella serovars tested in this study, including all of those in the database. Thus, these sequences may represent S. Heidelberg-specific markers. Genome sequences are available in the database for two American isolates, SL476 (GenBank accession number CP001120), a multidrug-resistant strain, and SL486 (ABEL01000001–ABEL1000048) a drug-susceptible strain. With respect to the distribution of SSH sequences, we found no difference between these strains. Indeed, with the exception of Salmonella Saintpaul and Salmonella Newport, all of the same-serovar genomes showed no difference in terms of SSH sequence distributions. Salmonella Saintpaul strain SARA23 was the only strain negative for all 14 SSH sequences, and is described as falling within the main clade of the Saintpaul serovar, whereas its partner strain SARA29, which we found to be positive for three SSH sequences (D23734-5, -14, and -B10), has been described as an outlier ( Salmonella Newport SL254 and SL317 are strains from the two distinct lineages that exist within the S. Newport serovar ( We found that both strains carry the lipoprotein-associated SSH sequence D23734-B10, but strain SL254 also carries the fimbrial-associated SSH sequence D23734-G3.

The genomes of S. enterica possess numerous fimbrial gene clusters implicated in host colonization and adaptation. Indeed, the genome of S. Typhimurium LT2 alone carries 11 fimbrial operons, some of which have been implicated directly in virulence (van der Velden et al., 1998; Humphries et al., 2001). The repertoire of fimbrial operons varies between serovars, with some widely distributed, but the others restricted to a limited number of serovars (Townsend et al., 2001; Porwollik & McClelland, 2003). Our observations indicate that fimbrial gene clusters make a major contribution to the accessory genome of S. Heidelberg. The stk gene cluster has been reported to be specific for S. Paratyphi A (Edwards et al., 2002). Based on analysis of the distribution of SSH sequence D23734-A1, we can add S. Heidelberg and Salmonella Kentucky to this, although our PCR assay data suggest that the operon may not be carried by all S. Heidelberg. We also demonstrate that the expression of the stk operon gene could be detected from some, but not all, of the S. Heidelberg isolates tested. Given the proven role of fimbriae in pathogenicity, the variable carriage of this fimbrial operon gene (stkD), and the variable expression even among those strains carrying the gene, leads us to the conclusion that the role of this operon merits further investigation.

Salmonella Typhi carries the fimbriae designated tcf, for Typhi colonizing factor (Folkesson et al., 1999). It has been reported previously that the tcf operon is present in S. Heidelberg and other invasive serovars such as Paratyphi A, Sendai and Choleraesuis (Townsend et al., 2001). We identified two SSH sequences matching genes within this operon, namely tcfA (D23734-A5) and tcfD (D23734-E10). In addition, we identified an SSH sequence matching tinR (D23734-D4), which lies downstream of the tcf genes in S. Typhi, and encodes a transcriptional regulator (Folkesson et al., 1999). However, based on the S. Heidelberg SL476 genome, it appears that this proximity does not occur in S. Heidelberg, and our observations suggest that the distributions of tinR and tcfA differ among both our strain collections and the serovars represented in the database. Our PCR assays indicated that the SSH sequence D23734-A5 (representing tcfA) was present in 100% of African, Tropical and UK isolates of S. Heidelberg, as well as in the two genome-sequenced isolates SL476 and SL486. Expression of tcfA was detected in all six isolates of S. Heidelberg tested.

A third fimbrial operon, represented by the SSH sequence D23734-G3, was present in serovars Enteritidis, Bovismorbificans, Saintpaul, Newport and Dublin as well as all but two S. Heidelberg isolates, and was expressed in all S. Heidelberg isolates tested, except for the UK veterinary isolate KMS1977. Expression of genes associated with a further two of the seven SSH sequences tested, encoding a putative ferrichrome-iron receptor and transcriptional regulator, respectively, were also expressed in all S. Heidelberg isolates except KMS1977. These observations suggest that although we found little evidence for genome content variations between invasive and gastrointestinal isolates of S. Heidelberg, we did find variations in gene expression. It is possible that differences in gene expression play a role in the different pathogenic abilities exhibited by isolates of this serovar.


We would like to acknowledge the guidance and support of Professor Tony Hart, who sadly died in September 2007. We would also like to thank Martin Okong, Chris Parry, John Cheesbrough, Melita Gordon, John Wain, Paul Wigley and Sam Kariuki for providing strains. This work was funded by the Medical Research Council in the form of a capacity building studentship for C.B.