Use of Single-Strand Conformation Polymorphism Analysis To Examine the Variability of the rpoS Sequence in Environmental Isolates of Salmonellae

Abstract

The natural environment places its resident microflora under stress, which may often result in adaptation by the microflora in order to increase the probability of survival. One such mechanism that has been postulated involves rpoS, which encodes a sigma factor that is known to enhance survival upon exposure to stress. The present work aimed to examine the genetic variability of rpoS in a selection of Salmonella enterica subspecies environmental isolates with an automated single-strand conformation polymorphism analysis technique. The results indicated that sequence variation does occur and that these changes are mainly located in two areas: at the center and near the end of the coding region. The variability was generally at the single-base level, although one strain (S. arizonae) did demonstrate significant differences in nucleotide sequence.

RpoS (ς^S, ς³⁸) is a key element for the survival of several gram-negative bacteria (including the salmonellae, Escherichia coli, pseudomonads, and Vibrio spp.) in adverse situations, e.g., entry into stationary phase (11, 21) and exposure to sublethal stresses such as osmotic shock (12) and low pH (23). The protein aids transcription of at least 30 genes and operons that form part of the bacterial defense (5) and virulence (8) mechanisms. Alignment of RpoS with RpoD (ς⁷⁰) (25) suggests that functional regions present are conserved in core binding (to the core polymerase), in the “RpoD box” (implicated in DNA strand opening), and in the −10 and −35 binding sites (for promoter recognition).

Over the last few years, isolates responsible for outbreaks of food-borne salmonellosis, e.g., Salmonella typhimurium DT104 (38), have been reported to be increasing their ranges of antibiotic resistance, and further work has indicated that some E. coli and Salmonella sp. isolates are capable of entering a hypermutatable state (observed as a gain in resistance to several antibiotics [22, 37]). Most food products create a relatively harsh microenvironment, resulting in the resident flora being in stationary phase (34) due to the levels of stress experienced in this situation. Recent reviews suggest that the preservative techniques utilized by the food industry to control and/or reduce the microbial load within these products may actually facilitate the evolution of more-resistant food pathogens (1, 20). One of the problems in improving food safety is the potential for rpoS variability between different isolates of the same strain, which occurs both in the laboratory and in natural environments (17, 40, 41). In certain circumstances, this variation has been thought to confer a selective advantage on individual bacteria within a population experiencing nutrient deprivation (41).

In previous work, screening for mutations of rpoS has been approached by identifying strains demonstrating phenotypic differences and then sequencing the gene (39, 41); however, this technique does not detect silent mutations. The use of genetic screening allows all nucleotide differences to be located; one technique that is capable of performing this task is single-strand conformation polymorphism analysis (SSCPA) (19). This method relies on electrophoretic separation of fragments according to their sizes and secondary DNA structures, which are sequence dependent (26, 33). Multiple fluorescence-based PCR-SSCPA (6, 15) has provided the capacity for simultaneous labelling of several fragments as well as the potential for an automated data collection system (4).

The present work aimed to evaluate the level of nucleotide variation in rpoS in the salmonellae. A selection of 18 environmental and human isolates of predominantly Salmonella enterica subspecies were screened along the entire length of the rpoS gene by SSCPA, with potential changes being confirmed by sequence analysis.

MATERIALS AND METHODS

Strains.

A list of strains utilized is shown in Table 1.

TABLE 1.

Salmonella strains utilized in this study

Strain	Source or referencea
S. senftenberg 775W	NCTC 9959
S. bovis morbificans	Laboratory strain, University of Nottingham Division of Food Sciences W03
S. amsterdam	Laboratory strain, University of Nottingham Division of Food Sciences W04
S. infantis	Laboratory strain, University of Nottingham Division of Food Sciences W06
S. arizonae	Laboratory strain, University of Nottingham Division of Food Sciences W07
S. amina	Laboratory strain, University of Nottingham Division of Food Sciences W08
S. kubacha	Laboratory strain, University of Nottingham Division of Food Sciences W09
S. montevideo	Laboratory strain, University of Nottingham Division of Food Sciences W10
S. kedouyou	Laboratory strain, University of Nottingham Division of Food Sciences W11
S. stanley	Laboratory strain, University of Nottingham Division of Food Sciences W12
S. derby	Laboratory strain, University of Nottingham Division of Food Sciences W13
S. virchow	Laboratory strain, University of Nottingham Division of Food Sciences W14
S. java	Laboratory strain, University of Nottingham Division of Food Sciences W15
S. bareilly	Laboratory strain, University of Nottingham Division of Food Sciences W17
S. braenderup	Laboratory strain, University of Nottingham Division of Food Sciences W18
S. enteritidis PT4 “E”	14
S. typhimurium (Ratten)	Hospital isolate, QMC, Nottingham, United Kingdom
S. typhimurium DT104 “30”	Bovine isolate, T. Humphrey, PHLS, Exeter, United Kingdom
S. togba	Laboratory strain, University of Nottingham Division of Food Sciences (W24)

^aPHLS, Public Health Laboratory Service.  

Primer design for PCR fluorescent labelling.

Primer sequences together with their fluorescent labels are shown in Table 2. In designing the primers, two major points were considered: (i) the optimal length for detection of single-base-pair changes is 200 bp (10), and (ii) changes at the end of the fragment are often more difficult to detect than those at the center (36). The six regions (A to F) covered the whole coding region of the gene, together with a small upstream region to ensure that the transcriptional start site was included. The primers were designed by utilizing Primer Express software, version 1.0 (Perkin-Elmer Applied Biosystems, Foster City, Calif.), with the four available sequences (GenBank) for rpoS in salmonellae (accession numbers are given in parentheses): S. typhi (X81641), S. enterica (X82129), and S. typhimurium (U05011 and X77752). The numbering system employed used the base pair numbering of S. typhi, which, as the longest sequence (1,707 bp [35]), avoided confusion over base pair locations within rpoS. The area to be screened (Fig. 1) starts just upstream of the transcriptional start site, which, according to published data (35), is at bp 423.

TABLE 2.

Primer details for fluorescent labelling of rpoS fragments

Primera	Labelb	Sequence	Region of rpoS amplified (bp)	Length of product (bp)
Afwd	6-FAM	CGTAAACCCGCTGCGTTAT	326–526	200
Arev	NA	GGTTCCTCTTCACTCAAGGCTT	326–526
Bfwd	6-FAM	AGTGATAACGACCTGGCTGAAGAA	527–731	204
Brev	NA	GTTACTCTCAATCATGCGACGG	527–731
Cfwd	6-FAM	GAGAGTAACCTGCTGGTGGTA	691–985	294
Crev	NA	TATGCGACAACTCACGTGCG	691–985
Dfwd	HEX	CGCACGTGAGTTGTCGCATA	966–1168	202
Drev	NA	TGTCTTCCGGACCGTTCTCTT	966–1168
Efwd	TET	AGAGAACGGTCCGGAAGACA	1149–1350	201
Erev	NA	GCCTTCAACCTGAATCTGACG	1149–1350
Ffwd	HEX	AGATTCAGCTTGAAGGCCTGC	1334–1698	364
Frev	NA	CCTTGCCCGGGCTGTGCCGATGCAC	1334–1698

^afwd, forward; rev, reverse. 

^bAbbreviations: NA, not applicable; 6-FAM, 6-carboxyfluorescein; HEX, 6-carboxy-2′,4′,7′,4,7-hexachlorofluorescein; TET, 4,7,2′,7′-tetrachloro-6-carboxyfluorescein. 

FIG. 1.

rpoS primers utilized for SSCPA of rpoS and predicted functional regions within RpoS (taken from reference 25). In the upper section, regions of function within RpoS are highlighted on the protein along with their locations. The lower section shows the rpoS gene and the locations of the primers utilized for the screening process. aa, amino acids.

PCR amplification of rpoS regions.

Cultures were streaked onto Luria (Lennox) agar (Difco, Surrey, United Kingdom) and incubated overnight at 37°C. The template was prepared for PCR by emulsifying a single colony from the overnight culture in 100 μl of sterile reverse-osmosis (RO) water. Reaction mixtures of 50 μl, containing 16 pmol of primers (Perkin-Elmer Applied Biosystems), 25 mM (each) deoxynucleoside triphosphates (Pharmacia Biotech, Herts, United Kingdom), 1 mM MgCl₂ (Advanced Biotechnologies, Leatherhead, United Kingdom), and 4.6 μl of buffer IV (200 mM [NH₄]₂SO₄, 750 mM Tris-HCl [pH 9.0] at 25°C, and 0.1% [wt/vol] Tween) supplied with Taq DNA polymerase AB0289 (Advanced Biotechnologies), were prepared and made up to volume with sterile RO water. Following a hot start of 10 min at 95°C and subsequent addition of 1 U of Taq DNA polymerase (Advanced Biotechnologies), PCR was performed with 35 cycles of 95°C for 15 s, 55°C for 30 s, and 72°C for 30 s and a final cycle of 72°C for 10 min (Progene [FPROG050]; Techne [Cambridge] Ltd., Cambs, United Kingdom). Correct amplification was verified by electrophoresing the products together with 1.5 μg of a 100-bp ladder (Pharmacia Biotech) on a 1.5% (wt/vol) agarose gel (NuSieve 3:1; FMC, Kent, United Kingdom) containing 1 μl of a solution containing 100 mg of ethidium bromide (Sigma, Dorset, United Kingdom) ml⁻¹.

SSCPA of rpoS.

Optimization of all conditions for SSCPA was carried out by using paired strains with known single-base-pair changes in a particular region.

The labelled PCR products were purified with a PCR purification kit (Qiagen, West Sussex, United Kingdom) according to recommended instructions (32), and the DNA was eluted in 50 μl of sterile RO water. A sample dilution (ranging from 1:2 to 1:30) was performed as previously described (16). From the resulting dilution, 5 μl was added to 0.5 μl of 0.1 M NaOH, which was heated at 95°C for 5 min and immediately snap-cooled on ice. After denaturation, 0.5 μl of Genescan-500 (TAMRA) (Perkin-Elmer Applied Biosystems) was added to each sample to act as the internal lane standard. The samples (for regions A to E) were loaded onto a 0.4-mm prechilled 10% (wt/vol) polyacrylamide gel (12.5 ml of Ultra Pure electrophoresis-grade 40% [wt/vol] acrylamide-bis acrylamide [37.5:1] [Sigma], 5 ml of 10× TBE buffer [890 mM Tris-borate, 890 mM boric acid, and 20 mM EDTA {Sigma Ultra Pure} at a pH of 8.3 at ambient temperature], and 32.5 ml of distilled water prepared as previously described [29]). Region F samples required a lower-percentage gel (8% [wt/vol]).

Electrophoresis was performed in prechilled 1× TBE buffer at 500 V for 20 h in a 373 DNA Sequencer (Perkin-Elmer Applied Biosystems); buffer was kept chilled for the first 30 min of the run time by the periodic addition of miniature sealed ice blocks. Data were collected and analyzed with Genescan 672 software (version 1.2.1) (Perkin-Elmer Applied Biosystems). Relative mobility calibration was constructed by utilizing the second-order least squares curve (i.e., linear regression) to provide the best interlane comparison. Internal lane standard peaks were allotted relative mobility values (scan number divided by 10) in order to provide a numerical comparison of fragment mobility between lanes.

Sequence analysis.

Sequencing was performed with a 373 DNA sequencer (Perkin-Elmer Applied Biosystems) with the Taq Dye Deoxy terminator cycle sequencing kit (Perkin-Elmer Applied Biosystems) and the relevant reverse SSCPA primers (A to F) as required.

RESULTS

Calculating gel variation of a “standard sequence.”

SSCPA relies on detection of differences in relative mobility during electrophoresis between PCR products of the same region. Because of gel variation, the same sequence run on different occasions may give differences in relative mobility. To allow changes in nucleotide sequence to be determined, this intrinsic variation in the relative mobility of a given sequence must first be estimated. For each region, a known sequence was taken as standard (“standard sequence”). PCR products of this standard sequence were run in five separate wells of duplicate gels, and a mean and standard deviation of relative mobility for each gel for the standard sequence of each region was calculated. Table 3 shows the duplicate means and standard deviations [ς_(n−1)] of relative mobility for each standard sequence from all of the regions within rpoS (A to F). The standard deviations of the five samples of identical sequence varied slightly between duplicate gels, with region D showing the largest discrepancy. This suggests that comparison between gels is possible, although it is not advisable for complete accuracy in the prediction of sequence variability. The variations in standard deviation between regions ranged from 0.3 for region E to 1.54 for region D. Therefore, to ensure that a change in relative mobility reflected true sequence variation from the standard sequence, a value of more than 2 standard deviations from the mean for a region was taken as a threshold value with defined confidence limits of 95%.

TABLE 3.

Calculated standard deviations and means of relative mobility of a single sample for the six regions of rpoS^a

Region	Strain	SD		Mean
		Gel 1	Gel 2	Gel 1	Gel 2
A	S. typhimurium DT104 ‘30’	0.48	0.39	574.29	659.12
B	S. kedouyou	1.35	1.09	652.13	635.62
C	S. virchow	1.01	1.22	694.60	587.20
D	S. typhimurium DT104 ‘30’	1.04	1.54	646.27	568.71
E	S. typhimurium DT104 ‘30’	0.3	0.32	581.39	561.38
F	S. enteritidis PT4 ‘E’	0.63	0.68	553.66	540.27

^aStandard deviations and means of relative mobility for the standard sequence on each gel were calculated from five samples of the same fragment. Relative mobility values were assigned following construction of the calibration curve (scan number divided by 10 against time) with Genescan 672 software (Perkin-Elmer Applied Biosystems). 

Detection of nucleotide differences in rpoS.

PCR products for the six regions of the 18 strains of Salmonella were analyzed by SSCPA, and sequence variations were determined by comparison of their relative mobilities with those of the standard sequences. These 18 strains included some isolates where single-base-pair changes were known from previous sequencing work to demonstrate single-base-pair changes in one of the regions. Those strains for which the relative mobilities for a region were outside the threshold value of 2 standard deviations are shown in Table 4. Three of the rpoS regions (A, D, and E) showed no variation in sequence, as indicated by the lack of shifts in fragment relative mobilities, and these were not analyzed further. Of the regions where strains demonstrated potential nucleotide differences (B, C, and F), C had the greatest level of variation. No strains showed variation in more than one region. The degree of shift was variable, suggesting that the type and extent of discrepancy may be strain dependent. In particular, the shifts with S. arizonae in region F seemed to be significantly larger than in other strains. Although S. amina was already known to contain a single-base-pair change (C to T) in region B (from prior sequence analysis) (Table 5), the shift in relative mobility fell inside the 2-standard-deviation threshold and therefore was not detectable by this method.

TABLE 4.

Relative mobilities for the six regions of rpoS from Salmonella environmental isolates demonstrating relative mobility shifts

Region of rpoS	Relative mobility of standard sequencea		Strain(s) outside threshold valueb
	Mean	SD
A	574.29	0.48
B	652.13	1.35	S. bovis morbificans (649.46)
C	694.60	1.22	S. amsterdam (686.49), S. stanley (698.71), S. typhimurium (690.66)
D	646.27	1.54
E	581.39	0.32
F	553.66	0.68	S. arizonae (546.90), S. senftenberg 775W (565.62)

^aData are from a single gel for each region of rpoS (A to F). 

^bValues in parentheses are relative mobilities of strains demonstrating values greater than the 2-standard-deviation threshold from the mean of the standard sequence. 

TABLE 5.

Sequence discrepancies in S. enterica subspecies as detected by SSCPA

Region of rpoS	Strain	Location of change (bp)	Nucleotide change
B	S. amina	687	C to T
B	S. bovis morbificans	619	To be determined
C	S. amsterdam	897	T to C
C	S. stanley	750	A insertion
		915	G to A
C	S. typhimurium	897	T to C
F	S. senftenberg 775W	1466	C to T

For S. arizonae, primer pair C failed to amplify a fragment with the cycle parameters described above; therefore, each region C primer was used with a primer from another region. The primer combination A forward and C reverse was successful in producing a PCR product, but the combination C forward and E reverse was not. These products were analyzed for sequence variation as described below.

Sequence analysis of variant strains.

Confirmation of potential nucleotide changes was verified with sequence analysis by using duplicate PCR amplifications containing the region of interest. In most cases, only a single-base-pair change was responsible for the shift in fragment migration (Table 5). As expected, greater shifts in mobility were seen with region C. Of the three sites where changes were evident, one was consistent: a T-to-C change at bp 897 seen in S. amsterdam and S. typhimurium.

While the location of the change in region B of S. bovis morbificans has been identified, the exact sequence change has yet to be determined: sequence data suggest that the population is a mixture of G and T at this site, the latter representing a sequence change.

The increased fragment migration time in S. arizonae region F samples was a result of 14 nucleotide changes (Fig. 2) comprising 12 single-base-pair changes and two insertions. Sequence data for the amplified product for region C revealed that a sequence complementary to the C forward primer binding region was present in S. arizonae. Failure of PCR with primer set C was postulated to be due to interference resulting from a more stable DNA secondary structure; PCR with increased binding and denaturation times resulted in weak amplification of this region. Addition of dimethyl sulfoxide (5%) to the reaction mixture gave a result equivalent to the PCR results for the other strains. These results support the hypothesis that secondary structure interfered with amplification of region C in this strain.

FIG. 2.

Alignment of S. arizonae rpoS with the published sequence of S. typhi rpoS to show differences in region F. Nucleotide changes are highlighted in black.

Region C of S. arizonae also showed significant variation, with 6 base changes evident (one A-to-T change, three C-to-T changes, and two T-to-C changes, as shown in Fig. 3). None of these corresponded to the changes seen in other strains.

FIG. 3.

Alignment of S. arizonae rpoS with the published sequence of S. typhi rpoS to show differences in region C. Nucleotide changes are highlighted in black.

DISCUSSION

Reproducibility of SSCPA.

In this study a threshold value for relative mobility of 2 standard deviations from the mean of five replicates of a standard sequence was set as the limit for determining changes in base sequence. In the majority of cases (five of six), this arbitrarily chosen value was able to discriminate between fragments with known single-base-pair differences. Variation in relative mobilities produced by identical sequences has been attributed to a number of factors. Previous work has shown that some single-stranded DNA is capable of producing two different conformations (15), and this would result in differences in mobility for identical sequences; in some instances both conformations can be present and run as separate bands in the same lane. Reproducibility of results has also been linked to constant run conditions, including power and temperature (7). The automated analysis system used here allows conditions to be recorded, enabling comparison between electrophoresis runs for voltage, temperature, and laser current. Temperature variation of the system here was minimized through the use of prechilled gels and buffer for electrophoresis. These changes limited the degree of reannealing following denaturation, especially during the period prior to DNA strand separation in the gel after the samples had been loaded onto the wells. The stability of single-stranded DNA has been verified as being temperature dependent (15, 28), with ambient temperature facilitating reannealing. Without prechilling, the level of single-stranded DNA visualized by the laser detection system employed by the 373 DNA sequencer was undetectable when small fragments (i.e., approximately 100 bp) were run (results not shown).

However, one base pair variation in region B of the S. amina strain (C-to-T change at bp 686) was not always detected by this technique. One possible explanation for this is the incorporation of errors during PCR (2, 13, 27, 31). Published literature suggests that the fidelity of a proofreading DNA polymerase is able to reduce the level of nucleotide misincorporation by a factor of 10 (30); in this instance, however, use of a polymerase with proofreading capacity (Ultma DNA polymerase; Perkin-Elmer Applied Biosystems) amplified a fragment of identical sequence. The likely reason for this observation is the relatively short length of DNA amplified in each instance and optimization of the cycling parameters and reagents for PCR.

Previous work with SSCPA has resulted in a wide range of detection rates, ranging from 50 to nearly 100% (4), which can be enhanced by the utilization of more than one set of electrophoretic conditions (9). The sensitivity of the technique relies greatly on the sequence change to influence fragment mobility (33), and in order to do this, it must cause a disruption in the folding of the DNA fragment. In certain instances, C-to-T changes have not been readily detected (36), which suggests that they are not in regions of single-stranded DNA that directly influence its secondary structure. Both cytidine and thymidine are pyramidine bases consisting of a single six-membered ring and therefore will not exert as much steric hindrance to single-stranded DNA folding as the purines, which comprise a five- and a six-membered ring.

Analysis of strains.

Of the 18 isolates analyzed by SSCPA, 6 of them contained discrepancies in the nucleotide sequence, as confirmed by sequence analysis. The level of variation ranged from a single base pair change to several substitutions, in the case of S. arizonae. The latter was to be expected, as this group of strains is reported to be a distant relative of S. enterica subspecies in terms of biochemical (24) and genetic (3) classification strategies. The lack of binding of the C forward primer to the totally complementary binding site in this strain indicates that the secondary structure of this region may be interfering with primer binding. This was confirmed by a weak amplification when the denaturing time was doubled and by a level of amplification similar to other S. enterica isolates in this region upon addition of dimethyl sulfoxide to the PCR mixture.

The locations of the nucleotide changes found indicate that certain areas of the gene (A, D, and E) remain conserved within the subspecies of S. enterica. These areas may be of specific importance to the cell, such as regions coding for protein functionality or turnover. RpoS is an important cell constituent which facilitates survival under adverse conditions; therefore, it is to be expected that some sections of the gene may be conserved. Alignment with other sigma factors with known functional regions (25) indicates that regions C and F, which showed the greatest variation, may be involved in binding to the core polymerase and the −35 binding site, respectively (Fig. 1). The predominant type of nucleotide variation in this work is a single substitution or insertion along the sequence, which is similar to previous work published for rpoS from both laboratory and environmental isolates of E. coli (40, 41). Not all mutations are beneficial to the bacterial population, and in most instances they lead to a reduced ability to survive (18, 40). However, there have been reported incidences of mutations in rpoS which enhance survival in long-term storage, as defined by the GASPing phenotype (41). These mutations are not necessarily imperative for survival but have been postulated to provide individuals with a competitive edge. The role of the mutations detected in the present work is yet to be discovered but will hopefully provide better insight into the effects of genetic variation in rpoS and the reasons for its persistence within the natural population.