Correspondence: Abraham Eisenstark, Cancer Research Center, 3501 Berrywood Drive, Columbia, MO 65201, USA. Tel.: +1 573 875 2255; fax: +1 573 443 1202; e-mail: email@example.com
Previously, we reported extensive diversity among survivors of Salmonella enterica ssp. enterica serovar Typhimurium that were stored for four decades in sealed agar stabs. Thus raising the question: was there selection for greater fitness among eventual survivors? To address this, we cocultured archived LT2 survivors with nonarchived (parental) LT2 strains in competition experiments. Selected archived strains outgrew a nonarchived LT2 sequenced strain. Although we initially assumed this was the result of mutations empowering greater nutritional utilization, we found phage selection was also involved. Phage fels-1 and fels-2 in supernatants were identified by primer/PCR as a putative selective force following single plaque isolations on a prophage-free strain and testing on appropriate hosts. In confirmatory experiments, instead of coculture in Luria–Bertani requiring antibiotic marker insertions, competing strains without markers were inoculated at opposite edges of motility plates. Not only did the archived LT2 population overgrow the nonarchived LT2 population, but also clear zones appeared at edges of encounters from which phage fels-1 and fels-2 (but not gifsy-1 nor gifsy-2) were recovered. However, in competitions of an archived strain with S. Typhimurium ATCC 14028, phage emerged that had a DNA base sequence segment of prophage ST64B but the sequence differed from the reported homologous segment in ST64B.
Five decades ago, Miloslav Demerec had the foresight to collect, analyze and curate several thousand Salmonella enterica ssp. enterica serovar Typhimurium LT2 auxotrophic mutants (Demerec et al., 1958; Demerec & Hartmen, 1959). These mutants were used to (1) develop the S. Typhimurium genetic map (Sanderson & Roth, 1988), (2) the XbaI–BlnI–CeuI genomic restriction map of S. Typhimurium LT2 determined by pulsed-field gel electrophoresis (Liu et al., 1993), (3) unravel biochemical pathways, (4) construct the Ames test to detect mutagens, (5) advance phage typing schemes for epidemiological tracking, as well as (6) establish basic genetic procedures and concepts, including horizontal gene transfer.
A large collection of over 10 000 of these auxotrophic mutants from the Demerec collection in sealed vials is now curated in our laboratory. Upon opening of over 200 of these sealed vials, all yielded colonies, despite storage of over 40–50 years at ambient temperatures and depletion of favorable nutritional and physical conditions. The survivors are extensively diverse in phenotype (Sutton et al., 2000; Edwards et al., 2001; Tracy et al., 2002; Liu et al., 2003; Porwollik et al., 2004; Rabsch et al., 2004), suggesting a continual, albeit slow, mutation–selection process. This raised the question of whether the survivors would have a competitive advantage over nonarchived strains. To test this, in Luria–Bertani broth (LB), archived isolates were cocultured with a LT2 strain considered to be the parent of the Demerec collection. This required insertion of antibiotic markers in each competing strain, a laborious procedure with other potential genetic consequences. Additional matches were observed by inoculations on each side of soft agar LB plates. This simpler matching procedure (Bettelheim & Carlile, 1976; Bettelheim, 1982), not only avoided laborious marker insertion but also allows characterization of progeny directly from archived vials (thus avoiding numerous laboratory regrowths with the danger of further genetic changes). It also expedites the testing of increased numbers of S. Typhimurium samples.
Subsequent to our competition matches of the archived vs. nonarchived cells described in Fig. 1 (Figueroa-Bossi & Bossi, 2004), reported that when selected nonarchived laboratory strains of S. Typhimurium ATCC 14028 and SL1344 were cocultured and as resources were depleted, one strain destroyed the other by phage lysis. Based on this model, we examined whether the same phenomenon of ‘sudden death’ might have occurred among cells in the archived cultures of the Demerec collection. Although each vial was initially inoculated from an individual colony, we now know that extensive mutation occurred during extended storage (Sutton et al., 2000; Edwards et al., 2001; Tracy et al., 2002; Liu et al., 2003; Porwollik et al., 2004; Rabsch et al., 2004). We hypothesize that S. Typhimurium subpopulations with diverse characteristics evolved in each sealed vial, and as resources were depleted, survival-prone progeny were selected. These bacteria contain prophages in their genomes, thus survival strategy could include prophage induction, transduction, chromosomal rearrangements, and/or the killing of cells that were less survivor prone.
In the present study, we found that in soft agar motility plate competitions, when archived LT2 strains met nonarchived S. Typhimurium LT2, a clear zone at the edges of their encounter was observed. This zone consisted of increased PFU that emerged from the encounter of the two hosts. Killing was only observed when the two strains encountered each other. Characterization of the phage DNA revealed that the recovered phage have DNA sequences of fels-1 and fels-2 when S. Typhimurium pairs were LT2 derivatives, which confirmed our previous results of coculturing in broth (data not shown). We also matched archived LT2 with nonarchived S. Typhimurium ATCC 14028 with the expectation that fels phage would emerge. However, to our surprise, a different phage (lab stock no. P153) was recovered from a zone of clearing where archive LT2 confronted nonarchival ATCC 14028. As described below, after sequence analysis, recovery of P153 phage was the result of the induction of prophage ST64B in the ATCC 14028 competitor, but with an interesting base pair difference.
Materials and methods
Salmonella Typhimurium strains utilized are described in Table 1. The experimental procedures were as follows: (1) coculturing in LB liquid medium of selected LT2 archived strains with isolates of nonarchived LT2 strains and (2) coculturing S. Typhimurium LT2 pairs and S. Typhimurium LT2/ATCC 14028 pairs on soft motility agar plates, allowing each to grow into competitor territory.
S. Typhimurium LT2 archived thyA68. Two colony types were isolated when vial was opened. In microarray analysis, one contained 180-kb amplification and given the new lab number 2375. The other without amplification is 2374
S. Typhimurium LT2, wild-type sequenced strain
ATCC 14028 with tetR inserted upstream of purA+
ATCC 14028 with camR inserted upstream of purA+
2239 with tetR insertion from 2277
2239 with camR insertion from 2278
2376 with tetR insertion from 2277
S. Typhimurium LT2 archived 1656. Colony isolated that did not contain 180 kb amplification
S. Typhimurium. Cured of fels-1, fels-2, gifsy-1, and gifsy-2 by L. Bossi
Fels-1 removed by red swap, transduced into MA8507 by phage P22. Resistance marker removed with recombinase
Bacterial strain construction
J. Slauch (University of Illinois, Urbana-Champaign) generously provided strains 2277 and 2278 (Table 1) in which tetracycline (tet) and chloramphenicol (cam) resistance markers, respectively, were introduced into virulent S. Typhimurium strain ATCC 14028 upstream of purA+ via the Wanner red-swap strategy. The antibiotic markers from these ATCC 14028 derived strains were transduced into LT2 and secondarily transduced (backcrossed) from wild-type LT2 (2239) into archival and nonarchival strains for competitions (Table 1). Transduction involved 44 kb of donor DNA into the new background. Because the cam and tet markers were much <44 kb, backcrossing eliminated the excess DNA from the LT2 competition strains.
Competitions in broth
Competitions were performed in LB liquid media at 37 °C using distinct ratios of an archived and nonarchived S. Typhimurium strains, with antibiotic markers, tetracycline tet (12 μg mL−1) or chloramphenicol cam (20 μg mL−1), for census of each. Each competition set consisted of two experiments, one with the archival strain in a 100 : 1 majority (vs. nonarchival), the second with the archival strain in a 1 : 100 minority. Additionally, each competitor was also grown and scored separately as a control. Periodically, samples were diluted and plated on LB agar plates containing either tet or cam antibiotic. Population composition was determined over a 4-week period for each competition by plating on LB agar plates supplemented with tet and cam antibiotics (Fig. 2).
Competitions on soft agar plates
Competitor cells were inoculated using a toothpick at opposite sides on soft motility peptone agar plates, incubated at room temperature to minimize nonbiological movement in the soft agar. Motile offspring grew out to form rings as cells found their chemical attractants in the medium (chemotaxis) (Fig. 1). One of the added advantages of observing competition on soft agar plates is that there would be no need to insert antibiotic markers into strains, as well as the ability to perform numerous matches expediently. Thus, in addition to matching LT2 strains, we matched archived LT2 with S. Typhimurium ATCC 14028.
Characterization of phage/prophage
Collection of phage
When cultures were started upon opening archived vials, we recovered PFUs from over 90% of the first 228 culture supernatants. For this competition study, we focused on those phages recovered from each of the competitors and phage recovered after confrontation between competitors. We initially compared host range and plaque morphology, and then examined phage DNA for sequence differences between them. Phage experiments were performed by standard procedures (Eisenstark, 1967).
Primer/PCR examination of phage from archived cultures
Primers (Table 2) were obtained for sequences that straddled the four known prophages of LT2 (fels-1, fels-2, gifsy-1, and gifsy-2) to determine which of these was the lethal agent. Sequencing primers were also obtained for ST64B phage after our discovery of its release from strain ATCC 14028 soft agar competition (Table 3).
Table 2. Primer sequences, target genes, and expected PCR product sizes
atg acg ctc ttg gac gca gga gga
gcg gcc tgg cct ggt aat atg atg
tgg cgc tgg gct gga aag
tgc tcc ggt tgg cgt agg
ttc ccg ctc tcc tta cac
ccg atc gtc atc ttg tcc
atg gct aat gca tca tac ccg aca
tcg tga gtg gat ctg gtt ccc ctt
Table 3. Primer sequences used to amplify and partially sequence phage P153 DNA
Primer sequences (5′–3′)
Source of plaque-forming phage
To determine the source of the phage that lysed nonarchived competitors, we collected supernatants from our matches and tested them on appropriate hosts. In the case of competitions between the LT2 strains, it was obvious that the archived strain supernatants plaqued on nonarchived strains, but supernatants from nonarchived LT2 yielded only a rare plaque. We observed no indication of an induction resulting from coincubation. However, in our soft agar match with an ATCC 14028 strain, there appeared to be an induction of P153 prophage, subsequently identified as ST64B. This phenomenon was observed previously by Figueroa-Bossi & Bossi (2004).
Comparison of phage DNA on agarose gels
Phage DNA was isolated from the lysates using the Qiagen Lambda Mini kit (Qiagen, Valencia, CA) and digested with EcoRl and BamH1. The resulting phage DNA fragments were separated via electrophoresis in a 1.0% agarose gel and stained with ethidium bromide for visualization/comparison using a Kodak EDAS Gel Electrophoresis Imaging System (Kodak, Rochester, NY). Growth from single plaques were used in the identification of fels-1 and fels-2 obtained from competitions of archival and nonarchival LT2 strains.
Cloning, PCR analysis, and DNA sequencing
Phage P153 (isolated from archival vs. nonarchival ATCC 14028 competition) DNA and the pZerO-2 cloning vector were digested with EcoRI, ligated with T4 DNA ligase, and transformed into competent cells of Escherichia coli Top 10 according to the Zero Background/Kan Cloning kit protocol (Invitrogen, Carlsbad, CA). Cloned phage DNA was then obtained from overnight (LB) broth cultures of the resulting kanamycin-resistant transformants via the QIAprep Miniprep kit protocol (Qiagen). Two plasmid clones, one containing a 1.1-kb insert and another containing a 3.5-kb insert, were subjected to nucleotide sequencing with vector primers and BigDye terminator chemistry on a PE Biosystems 3730 capillary DNA sequencer at the University of Missouri Molecular Biology Program DNA Core Facility. DNA sequences and deduced peptide sequences were manipulated and analyzed using the genetics computer group software package through the Pittsburgh Supercomputing Center (http://psc.edu/) and the blast program (and linked databases) at the National Center for Biotechnology Information (http://ncbi.nlm.nih.gov).
Once the possible coidentity of phage P153 and the recently reported phage ST64B had been established, long-range PCR was used to amplify and paired end sequence seven regions of the phage genome, each of which was c. 4 kb in length. Briefly, appropriately positioned oligonucleotides corresponding to sequences with the published ST64B genome were utilized in PCR with phage P153 DNA as template, using the Expand Long Template PCR system (Roche Molecular Biochemicals, Indianapolis, IN) as described previously (Calcutt et al., 1999). The resulting amplicons were analyzed by direct sequencing with the amplification primers, following purification of the PCR products through QIAquick spin columns (Qiagen). Oligonucleotides were all written 5′–3′ and purchased from Integrated DNA Technologies Inc. (Coralville, IA) their sequences were as listed in Table 3.
Competition in broth
In the competition examples (Fig. 2), the archived cells started as a majority of 100 : 1 in one set and in the other set, the archived strain started as minority of 1 : 100. As seen in Figs 1a and b, the archived LT2 strain 2381 outgrew the sequenced nonarchived LT2 2285 strain. Figure 1c–f show results of coculture with the additional archived strains, 2286 and 2379, with similar results. Figure 1g and h represent the same archived strains as in Fig. 1a and b, but in which the tet and cam markers were reversed. These strains were constructed to evaluate whether the introduction of the tet and cam antibiotic-resistance genes would have an influence in the selection process. When samples were assayed from these competitions, the competition data were essentially the same, regardless of which antibiotic marker was used (Fig. 2). It is apparent that when the archived cells started as 1 : 100 minority, the population shifted abruptly within 24 h. and ended with almost complete reversal of ratios. However, when the archived strains started 100-fold excess, they remained in the majority.
These experiments were performed to test whether the selected archived strains had greater fitness than nonarchived strains. The striking results indicated that the archived strains, rather than having a mere growth advantage, were actually toxic to the nonarchived LT2. Examination of supernatants from the archived strains revealed that fels-1 and fels-2 phage were being released, but there was no evidence of gifsy-1 nor gifsy-2. Further testing after growth from single plaques grown on prophage-free host 2754 indicated that fels-1 was the major lethal agent, with little evidence of lysis by fel-2. Further support of this conclusion is explained in Discussion.
Competition tests on soft agar plates
Subsequent to broth experiments, competition experiments were performed on soft agar plates, which in addition to being less labor intensive reduced the need for antibiotic markers and allowed us to score results of numerous matches at the same time. Also, bacterial samples directly from archived vials, without regrowth, could be tested (currently under study). Although the patterns of migration varied, in general, archived cells moved faster across the plate than the nonarchived cells. However, in every match of an archived LT2 with a nonarchived S. Typhimurium, there was a clear zone at the zone of confrontation (Fig. 1), although the migration patterns differed in the various matches. Although, in the case of competitions of cells from the same colony matched as controls, no clear zones were observed as noted by Bettelheim (Bettelheim & Carlile, 1976; Bettelheim, 1982). This is an interesting observation of self-recognition. However, Bettelheim did not refer to phage as a factor in this recognition phenomenon.
Primer/PCR examination of phage from archived cultures
Primers (Table 2) were obtained for sequences that straddled the four known prophages of LT2 (fels-1, fels-2, gifsy-1, and gifsy-2) to determine which of these was the lethal agent. Extraction of purified DNA from supernatants and electrophoresed on agarose gels revealed bands for fels-1 and fels-2 only. By adding virions from single plaques of archived LT2 to S. Typhimurium hosts that lack either fels-1 (strain 2144) or fels-2 (strain 2273), or lacking all four prophages (strain 2754), supernatants were obtained that putatively had either fels-1 or fels-2 virions only. When tested on the nonarchived strains used in Table 1, the putative fels-1 resulted in positive lysis in most cases but also fels-2 in only a few cases. Thus, fels-1 appears to be the major virion that destroys the nonarchived cells. This is further supported by the microarray study that describes the multitranscription of fels-1 (Frye et al., 2005).
When phage DNA was recovered from the virions that emerged from the soft agar competition, we expected that it would identify with one or more of the four prophages. However, while this was true in cases of LT2 matches, when archived strain LT2 was matched with nonarchived S. Typhimurium ATCC 14028, the isolated DNA differed from any of the prophage of the archived LT2. DNA from the putative phage (P153) was analyzed to determine its identity. Initially, two plasmid clones were generated, each of which contained a differently sized EcoRI fragment derived from the purified phage DNA. Sequence analysis using vector primers revealed that these two cloned fragments were most likely derived from phage ST64B because for each sequence read of c. 700 nucleotides, there was 100% nucleotide sequence identity with the published ST64B sequence. The two clones encompassed nucleotides 37 466–923 (spanning the terminal sequences, as submitted to GenBank, accession no. 055382) and 24 922–AY26 055. Further analysis of the phage DNA was conducted by nucleotide sequencing of the ends of seven amplicons of c. 4 kb, generated by long-range PCR with primers derived from the ST64B sequence. End sequencing of these amplicons yielded c. 11.2 kb of unambiguous sequence interspersed over the ST64B genome. Each sequence was 100% identical to the published sequence, a result that was not anticipated in light of polymorphisms among ST64B-borne alleles that have been used to discriminate between closely related definitive types of S. Typhimurium (Ross & Heuzenroeder, 2005). Although the complete genome sequence of phage P153 was not determined, the finding that more than one-third of the phage genome is 100% identical to ST64B strongly supports the notion that phage P153 and ST64B are very closely related, if not identical.
The isolation of a ST64B-like phage was somewhat surprising because previous reports have indicated that the prototypic ST64B isolate that was sequenced is defective for transfer, most likely due to a defect in the gene encoding the tail protein SB21 (Bossi et al., 2003; Mmolawa et al., 2003). The defect was identified as a mutation in a homopolymeric G tract: the presence of an 8-bp G tract (as found in the prototype) results in a premature stop codon whereas a 7-bp G tract fuses the annotated SB21 and SB22 ORFs to create an uninterrupted reading frame equivalent in length and with extended sequence homology to putative tail proteins of other phages. Because phage P153 could be readily propagated, it was of interest to determine the length of the homopolymeric G tract. Direct sequencing of amplicons encompassing the SB21–SB22 junction unambiguously revealed the presence of an 8-bp poly-G tract, which was previously reported to correlate with a defective phage phenotype. It should be noted, however, that although compelling, a direct causal relationship between poly-G tract length and phage phenotypes has not yet been demonstrated. Furthermore, it is formally possible that multiple mechanisms may exist to overcome the apparent +1 expansion of the G tract in ‘defective’ ST64B-like phages.
These experiments were initiated to explore the question, after reduction of nutritional resources, whether there would be a selection of fittest survivors. We observed that phage/prophage relationships may be important factors in selecting survivors in populations with limited resources. This conclusion is based on coculturing archived LT2 and nonarchived LT2 isolates in LB broth. Rather than a gradual outgrowth, we observed a sharp population drop of the nonarchived LT2; this indicated a toxic effect. After agarose gel electrophoresis via primer/PCR of supernatants from archived strains, it was evident that either fels-1 and/or fels-2 lysed the nonarchived strain. Distinction of the two was determined by growing on appropriate permissive and nonpermissive strains; these results indicated that fels-1 was the major lytic agent. Supporting the view of high titer release of fels-1 is the microarray observation by Frye et al. (2005) of a ninefold amplification fels-1 DNA upon H2O2 or mitomycin induction. During long periods of starvation it is obvious that chromosomal breakages would occur resulting in phage induction.
Another puzzle is that fels-1 is the phage that lysed the nonarchived LT2. But this LT2 also has fels-1 prophage; thus, it should have been immune. Whether a bypass such as this is a survival strategy is currently under study in our laboratory, including the possibility that the released fels-1 from archived LT2 has a mutation that bypasses the fels-1 immunity of the nonarchived LT2.
We have observed that when cultures have been started from vials that had been sealed and stored for over four decades, PFU (virions) were recovered from supernatants of most archived cultures to over 103 mL−1 in almost every case. Phage samples have been collected from over 200 of such vials. In contrast, when 22 other LT2 cultures (nonarchived) were sampled in a similar manner, only in a few cases were plaques observed from supernatants when plated on hosts that were lysed by supernatant from archived 2382, without induction manipulation. The nonarchived LT2 cultures were obtained from Ken Sanderson, Salmonella Genetic Stock Center. These strains were lyophilized several years ago and are considered closest to the original strain used to obtain auxotrophic mutants in the collection. It is also the strain used for sequencing (McClelland et al., 2001).
Comparison of competition studies by others
Initially, we focused on the concept that mutations in rpoS in E. coli would be a factor in fitness as reported by others (Finkel & Kolter, 1999; Vulic & Kolter, 2001). We assumed that the rpoSatt might be a key ‘champion’ gene in S. Typhimurium also, as indicated for E. coli. These rpoSatt were designated as growth advantage in stationary phase. Although we observed rpoS mutations in many of our archived cultures (Sutton et al., 2000), we recognized that there were a number of complexities in determining survival. Thus, subsequent experiments were designed to score additional diversities in search of ‘champion’ survivor genes, including catalase content, losses in ability to utilize 190 different carbon and nitrogen compounds (Tracy et al., 2002), susceptibility to lysis by a large battery of typing phages (Rabsch et al., 2004), a role for mutator gene (Liu et al., 2003), and significant genomic changes (Porwollik et al., 2004).
Other long-term growth studies under starvation include those of (Lenski & Riley, 2002). Naas et al. (1995) described insertion sequence transpositions as a cause of genetic diversity, utilizing comparative gene hybridization in microarrays (Faure et al., 2004), speculated that the loss of crl gene is responsible for selective advantage of deletions that occurred in the aged cells. Upon examining Vibrio cholerae 01 populations from aged agar stabs, Paul et al. (2004) noted that mutations involving the toxR reading frame might be involved in fitness. Because the environmental conditions differed from those in our experiments, especially in cases where there was replenishment of nutrients, it is difficult to compare results.
Figueroa-Bossi & Bossi (1999) observed that S. Typhimurium strains ATCC 14028s and SL1344 harbor defective ST64B prophage that did not yield plaques until cocultured with an isogenic sibling lacking the prophage. Phage ST64B is considered to be a defective prophage until methods for induction revealed that it could easily develop to a PFU virion (Figueroa-Bossi & Bossi, 2004). As described by Mmolawa et al. (2003), ST64B is a genetic mosaic whose genome has been acquired from numerous portions from sources outside of the genus Salmonella. However, it should be noted that only Bossi et al. (2003) have described a phage role in bacterial population dynamics.
These experiments surfaced some anomalies that are under current study. The main unexpected result was that archived LT2 yielded fels-1 and fels-2 phage, which lysed nonarchived LT2. The nonarchived LT2 contained prophages fels-1 and fels-2; thus, they should have been immune. Because they were not, we are examining the basis of this phage/prophage/host relationship. Another surprise was the phage that emerged from archived LT2 nonarchived ATCC 14028.
We thank J. Slauch for ATCC 14028 derived strains containing the tet (2277) and cam (2278) markers, A. Segall for LT2 strain with deletion of all prophages (2754) and K. Sanderson for the nonarchival LT2 strains used in this study. This research was supported by private donations to the Cancer Research Center. R.A.H. and W.F. were recipients of Raymond Freese Postdoctoral Fellowships (Cancer Research Center).