rpoS Mutants in Archival Cultures of Salmonella enterica Serovar Typhimurium

Abstract

Long-term survival under limited growth conditions presents bacterial populations with unique environmental challenges. The existence of Salmonella enterica serovar Typhimurium cultures undisturbed in sealed nutrient agar stab vials for 34 to 45 years offered a unique opportunity to examine genetic variability under natural conditions. We have initiated a study of genetic changes in these archival cultures. We chose to start with examination of the rpoS gene since, among gram-negative bacteria, many genes needed for survival are regulated by RpoS, the stationary-phase sigma factor. In each of 27 vials examined, cells had the rpoS start codon UUG instead of the expected AUG of Salmonella and Escherichia coli strains recorded in GenBank. Ten of the 27 had additional mutations in the rpoS gene compared with the X77752 wild-type strain currently recorded in GenBank. The rpoS mutations in the 10 strains included two deletions as well as point mutations that altered amino acid sequences substantially. Since these stored strains were derived from ancestral cells inoculated decades ago and remained undisturbed, it is assumed that the 10 rpoS mutations occurred during storage. Since the remaining 17 sequences were wild type (other than in the start codon), it is obvious that rpoS remained relatively stable during decades of sealed storage.

Several investigators have examined “aged” bacterial cultures, leading to consensus that mutations occur under stressed conditions of minimal metabolism and minimal chromosomal activity (7, 10–12, 26). Since we have a collection of more than 2,000 Salmonella enterica serovar Typhimurium cultures that have been stored for 34 to 45 years (2–5, 24), there was an opportunity to examine questions of mutations in aged cultures from an additional viewpoint. Miloslav Demerec and associates stocked and curated the extensive collection of these auxotrophic mutants (2–5). These isolates stem from single strain LT2 and LT7 cultures supplied by Norton Zinder and Joshua Lederberg (24). This collection consists of auxotrophic isolates used primarily for inter- and intragenic mapping of the Salmonella chromosome. At the time of isolation, they were assumed to be isogenic except for their nutritional requirements. We are currently examining samples from this collection for specific variations that may have occurred during this storage period. Since the number of survivors is only a small fraction of the earlier population, it is assumed that components from dead cells provided sufficient nutrients to allow occasional cell division. In this report we focus on alterations in the rpoS gene, which are assumed to have occurred during this “hibernation.” This gene was chosen for special attention in keeping with a model presented by Kolter and associates (11, 12, 28–30) that certain non-null rpoS alleles have a selective advantage over wild-type alleles after emergence from stationary phase back to growth phase. In particular, they observed that when cells from an old culture of an Escherichia coli ZK226 rpoS mutant with a 46-base redundant region were mixed with fresh wild-type cells, the mutant outgrew the wild type (29). Under more natural conditions, we searched for rpoS mutations that might offer further insight into evolutionary changes in bacterial populations.

The RpoS sigma factor in gram-negative bacteria regulates an array of genes in response to environmental stress and DNA damage, as well as to prepare the cell for entrance into stationary phase and for survival during minimal nutrition and cell division (10). Microarray assays of 4,290 E. coli genes that are transcribed under limited nutrition reveal that RpoS is pivotal in the complex regulation of a large fraction of total genes (25).

We now report that, upon sequencing the rpoS region of isolates from 27 of the stored vials (Table 1), 10 had mutations in the coding region (Table 2), but the remaining 17 had the same sequence as recorded for the wild type in GenBank.

TABLE 1.

Strains used in this study

Strain no.	GenBank accession no.a	Mutationb	rpoS allele	Date of inoculation (mo/yr)
1594	AF184095	his-2555 DIId	Mutant	4/1967
1595	AF184103	his-2555 DII	Mutant	4/1967
1596	AF184102	his-2555 DII	Mutant	4/1967
1669	AF184101	his-2550 DIIR42A	Mutant	4/1967
1670	AF184100	his-2550 DIIR47A	Mutant	4/1967
1674	AF184099	his-2550 DIIR49B	Mutant	4/1967
1704	AF184104	thy-316 2265A	Mutant	8/1964
1706	AF184098	thy-317 2266	Mutant	8/1964
1747	AF184097	his-2555 DII	Mutant	4/1958
1748	AF184096	his-2555 DII	Mutant	4/1958
1602c	NA	χ3000	Wild type	10/1998
1597	NA	his-140	Wild type	6/1954
1599	NA	his-142	Wild type	6/1954
1600	NA	his-143	Wild type	6/1954
1601	NA	his-144	Wild type	6/1954
1657	NA	his-2122 R5	Wild type	4/1958
1683	NA	his-214	Wild type	4/1958
1684	NA	thy-273 2222	Wild type	8/1964
1685	NA	thy-269	Wild type	8/1964
1686	NA	thy-273 2222	Wild type	8/1964
1687	NA	thy-273 2222	Wild type	8/1964
1695	NA	thy-272 2211	Wild type	8/1964
1698	NA	thy-313 2262B	Wild type	8/1964
1699	NA	thy-314 2263A	Wild type	8/1964
1700	NA	thy-314 2263B	Wild type	8/1964
1701	NA	thy-314 2263C	Wild type	8/1964
1702	NA	thy-315 2264A	Wild type	8/1964
1703	NA	thy-314 2264B	Wild type	8/1964

^aNA, no accession number. 

^bStrains with the same allele number (i.e., his-2550 and his-2555) are replicates from the original auxotroph isolation. 

^cWild-type strain χ3000 was generously provided by Cheryl Nickerson, Washington University. 

^dDII indicates the position of the mutation within the histidine operon (5). 

TABLE 2.

Mutant rpoS sequences^a

Strain	DNA lesion	Change in nucleotide sequence	Change in aa sequence	HPII activity
1594	Frameshift	C deletion at 743	Truncation: retains first 223 native aa + 30 nonsense aa added	Very low
1595	Base change	C to T at 964	Arginine to cysteine	Normal
1596	Frameshift	CG insertion at 415	Truncation: retains first 114 native aa + 4 nonsense aa added	Normal
1669	Base change	G to T at 796	Truncation: retains first 241 native aa	None
1670	Frameshift	Δ290–701	Truncation: retains first 72 native aa + 11 nonsense aa added	Very low
1674	Base change	G to T at 168	Glutamic acid to aspartic acid at aa 32	Normal
1704	Base change	G to A at 591	Silent mutation	Very low
1706	Base change	G to C at 463	Glutamic acid to glutamine at aa 131	Normal
1747	Deletion	Δ565–576	In-frame deletion of 4 aa, RLPI, Δ165–168	Very low
1748	Frameshift	C insertion at 743	Truncation: retains first 224 native aa + AGR replaces LGG	Normal

^aNumbers describing DNA lesions are based on sequences submitted to GenBank (see Table 1), with 1 corresponding to the wild-type reference allele X77752. aa, amino acid. 

MATERIALS AND METHODS

The nutrient agar stab cultures were stored in small vials sealed with paraffin at ambient room temperatures. Even after 34 to 45 years of storage, viable cells were recoverable from these cultures, yielding 10³ to 10⁵ CFU per vial. It is assumed that a very low level of metabolism and cell division may have occurred in these cultures, perhaps as a result of dead cells providing meager nutrients to viable cells (11, 26). All isolates were confirmed as S. enterica serovar Typhimurium by lysis with P22 phage and clumping with Difco Salmonella O antiserum group B.

Culture selection.

Although we have over 2,000 auxotrophic mutants in storage, the 27 used in this study were not selected at random. This study was restricted to his mutants because of extensive records and publications used in fine-structure mapping and thy mutants because these were isolated by one of the authors (A.E.) and original notebooks and records are available (8). Also, since thy mutants have higher mutation rates (13), these could be an interesting group.

In our selection, we also wanted samples of cultures that initially came from the same colony but were stored in replicate vials. These include his-2550 and his-2555 sets. Strains 1747 and 1748, both rpoS mutants, were included in the collection because they had previously been assayed as low in hydroperoxidase II (HPII).

Sample preparation.

Table 1 lists the aged Salmonella strains that were examined for rpoS sequence variability using an ABI-377 automated sequencer. Strains with the same auxotrophic number (i.e., his-2555 and his-2550) were derived from the same colony in April 1958.

The agar plug from each vial was removed into 2 ml of phosphate-buffered saline (PBS), vortexed vigorously, and allowed to rehydrate for an hour at room temperature. An aliquot of the freshly vortexed buffer suspension was plated on solid Luria-Bertani (LB) agar and on minimal medium containing the appropriate supplements for each strain and allowed to grow for 1 or 2 days at 37°C. A single colony from each culture was transferred to 25 μl of PCR-grade water and stored at −20°C.

Amplification of rpoS.

Primers flanking the rpoS gene were designed using the web-based Primer program (http://www-genome.wi.mit.edu/cgi-bin/primer/primer) and the Salmonella rpoS sequence from GenBank (accession no. X77752). The primers were as follows: forward primer beginning at base 198, 5′-CAAGGGGAAATCCGTAAACCC-3′, and reverse primer beginning at base 1391, 5′-GCCAATGGTGCCGAGTATC-3′. The entire rpoS sequence from base 198 to base 1391 was amplified from each sample using 1 μl of the frozen bacterial suspension made above as a DNA template in the following PCR: initial denaturation for 5 min at 95°C, 38 cycles at 95°C for 30 s, 58°C for 30 s, and 72°C for 1 min, followed by a 10-min extension at 72°C using HotStarTaq enzyme and buffers from Qiagen. The resulting 1,193-bp fragments were analyzed by polyacrylamide gel electrophoresis (PAGE) and purified using Qiagen's Qiaquick purification columns.

DNA sequencing.

PCR products for the rpoS region of 27 aged Salmonella strains were sequenced in both directions and compared to the published sequence (GenBank no. X77752) as well as a concurrently sequenced wild-type strain, χ3000. The amplified 1,193-bp fragment was used as a sequencing template. To control for PCR artifacts, separate amplification reactions were used for forward- and reverse-strand sequencing. Sequencing samples were prepared using plus-strand primers 198 and 797 (5′-GATTCGCTTGCCGATTCAC-3′) and minus-strand primers 822 (5′-TAACAATGTGAATCGGCAAG-3′) and 1391 along with BigDye chemistry (ABI) and run on an ABI-377 automated sequencer (DNA Core Facility, University of Missouri, Columbia). The resulting sequences were visually reconciled to confirm base-calling and compared to each other and the wild-type X77752 and U01150 rpoS sequences using ALIGN (17, 20; http://vega.igh.cnrs .fr/bin/align).

Catalase assay.

Activity gels were prepared from protein extracts from overnight cultures by the method described previously (23).

RESULTS

Variation in rpoS sequence.

The region containing rpoS was PCR amplified and analyzed by 7.5% PAGE to confirm the presence of a product and any gross size differences from the expected size. These 27 strains included histidine and thymidine auxotrophic isolates (Table 1). Every sample analyzed, including the wild-type LT2 laboratory strain and the four aged LT7 strains, had a UUG start codon. Ten of the 27 samples contained mutant rpoS sequences (see Table 2). The coding region of these mutants is shown in Table 2. In five of these cases, a single base pair change was found. These changes were distributed randomly throughout the coding sequence. These base changes created a single amino acid substitution in strains 1595, 1674, and 1706 and a silent mutation in 1704. In 1669, the base change created a stop codon resulting in a truncated protein of 241 amino acids (Table 2). Three frameshift mutations were found. In two strains, 1594 and 1748, a deletion and insertion, respectively, of a cytosine 673 bp downstream of the translation start site resulted in a frameshift that truncates the protein. The mutation removed 102 amino acid residues in 1748 and added 30 residues in 1594 at the point of mutation. The third frameshift mutant, 1596, had a 2-bp insertion (GC) 345 bp into the coding region that resulted in a truncated protein 119 amino acids long.

The two deletion mutants, 1670 and 1747 (Table 2), had direct repeat sequences, one exterior to the deletion and the other just within the deleted sequence. The 411-bp deletion in 1670 resulted in a frameshift that truncated the protein after adding 11 nonsense amino acids, retaining only the first 72 amino acids of the native protein. The 12-bp deletion in 1747 removed 4 amino acids (165 to 168) in frame (Table 2).

Variation in rpoS among replicates of initial isolates.

When these mutants were collected decades ago, quintuplicate copies were made. Within our current collection, we have several replicates. Note that several strains from the Demerec collection (Table 1) have the same allele number. These are from different vials but stem from the same auxotrophic mutant colony isolated decades ago. We chose these for this study because any differences between them would certainly represent mutations that had occurred during storage. Laboratory strains 1669, 1670, and 1674 were initially inoculated from the same colony, yet all three differ in rpoS sequence. Strains 1594, 1595, and 1596 also differ in rpoS sequences. In checking on the history of these mutants, we learned that all of these were spontaneous his mutants isolated by J. Yourno in 1967 (personal communication from Phil Hartman). However, in other cases of cultures stemming from “siblings,” we detected no difference in sequence from that recorded in GenBank for wild-type LT2. This was the case for strains 1684, 1686, and 1687 and again for the four strains 1699, 1700, 1701, and 1703.

Variation of rpoS among isolates from same vial.

When the vials were opened, cultures were started from single colonies that grew on LB plates. An obvious question is whether survivors were allelic with regard to rpoS, or whether the population was heterogeneous with regard to rpoS. To address this, the DNA from 46 colonies grown from the original vial of culture number 1670 were PCR amplified and examined by PAGE for variation. While this would not resolve small size differences, larger deletions would be detected by this method. Two of the isolates did not contain a template for the rpoS primers 198 and 1391, while the other 44 produced fragments indistinguishable from wild-type lengths.

Variation in catalase.

RpoS directly regulates HPII (katE gene product [23]), and it also has an indirect role in transcription of HPI (katG gene product) (9, 14). Table 2 compares the catalase HPII activity of the strains listed in Table 1. As may be noted, the correlation between rpoS mutation and catalase activity is not absolute (see Discussion). We have assayed catalase activity several times as well as checking for rpoS mutation in these strains. When catalase activity differed from expectations, perhaps another sigma factor has substituted in this regulatory role.

Changes other than in RpoS.

It should be noted that we chose to focus on differences in rpoS sequences since a number of other investigators have postulated the importance of rpoS to survival in old cultures (10–12, 27, 28). We have also undertaken examination of other genetic changes in these aged cultures (unpublished). One pertinent observation was that when the strains listed in Table 1 were examined for mutations in rpoE, none were found (data not shown).

Broth cultures of the strains listed in Table 1 were incubated for 80 days to observe whether the rpoS mutants would lose survival capacity. Every few days, CFU were determined. Although the population in all 27 cultures dropped approximately 100-fold in the first few days, the final populations among each of the 27 did not vary by more than 10-fold.

DISCUSSION

It is of particular interest that these cultures have survived decades of confinement under very limited growth conditions. We have opened hundreds of these archival cultures, and all yielded CFU regardless of the auxotrophic mutation. As described by Koch (18), mutations would be expected in genes that are no longer needed in this environment. Whatever these mutations might be, cells manage to survive.

We chose to initiate our study of survival of cells in these archival cultures with observations of RpoS, especially since it is the regulator of a large number of stationary-phase genes (9, 25).

Distribution of rpoS mutations.

Based on previous observations of survival advantages of rpoS^att mutants (29), we expected to find mutants in our archival cultures. The notation rpoS^att refers to mutants that are attenuated in activity but have a growth advantage over the parent strains (29). Of the 27 strains that were tested, only 10 were rpoS mutants. Since the other 17 were wild type, it is obvious that rpoS mutation is not a condition of survival. It is puzzling that, of the 10 strains with altered rpoS sequence, 8 were derived from two strains, his-2550 and his-2555, spontaneous his mutants (see Results). This suggests that there may have been something unusual about these two strains at the time of isolation.

Without knowing the rpoS sequence of the ancestral S. enterica serovar Typhimurium isolate, we can only speculate as to genetic events that may have occurred. An obvious speculation is that the UUG rpoS start codon existed in the original Zinder-Lederberg strains, since these were assumed to be nonpathogenic laboratory cultures in the Yale collection. Also, nonpathogenic Typhimurium strains in current use also have UUG as the start signal (19). In E. coli, UUG is used as an alternative start codon in ∼1% of genes (6). In general, this alternative codon lowers the efficiency of translation initiation and therefore lowers the protein level (6). There is evidence to suggest that UUG may make translation more responsive to environmental stress (6). Perhaps the lowered level of RpoS contributed to the survival of these strains in a nonanimal environment.

As for changes other than AUG in the rpoS sequence, assuming that almost all, if not all of the cells inoculated decades ago had no rpoS base sequence differences, we can only speculate as to the time of occurrence. According to one scenario, the changes may have occurred very early, perhaps within the first few weeks of storage, since cells may have needed this change for survival superiority. This is essentially the view of Zambrano and colleagues (11, 12, 28), that rpoS^att GASP (growth advantage in stationary phase) mutants occur early after reaching stationary phase and that they have an early selective advantage over the wild type.

It is difficult to understand how strain 1670 manages to survive, especially since no RpoS could be detected by Western blot (data not shown). Population takeovers similar to those observed in the laboratory may be occurring in nature. Although we observed mutations other than in rpoS within these aged strains (data not shown), we have focused on rpoS mutation in this report because it may be pivotal to understanding how a Salmonella population faces depletion of nutrients. Tao et al. (25), in a comprehensive analysis of 4,290 putative open reading frames of E. coli, noted that over half of the known RpoS-dependent genes are expressed under nutrient-deprived conditions. Their hypothesis invokes RpoS as a universal stress protein that coordinates various aspects of metabolic shifts. They also point out that production of RpoS is subject to complex posttranscriptional and translational regulation, and therefore it cannot be presumed that rpoS transcript levels correlate with RpoS activity.

A plausible prediction for such long-term survival is that a mutant form of RpoS, the pivotal stress sigma factor protein, would allow transcription and translation to favor “survival” proteins, as well as to shut down products that would be unfavorable to survival. Since we found only 10 out of 27 sequences to have mutations (other than the UUG start site), perhaps a more plausible explanation is that there is more than one type of RpoS among individual survivors within the same vial. Other than the 46 colonies isolated from the 1670 culture (as described under Results), we picked only two to four colonies from the other vials. Obviously, there could be additional rpoS mutants in the same vial. Perhaps selection of only the fittest within a population is not ideal for long-term survival; rather, diversity may be a preferred way of coping with the stresses of long-term hibernation, as suggested by Finkel and Kolter (11). We know that there is diversity among survivors from the same vial, since we see differences in such characteristics as colony size, but this does not tell us whether this diversity relates to survival. This is testable by examining a number of colonies for rpoS mutations that stem from a single vial and doing competition experiments similar to those done by others (11, 12, 27).

From examination of the mutant rpoS sequences by other investigators, there is no reported single change that would indicate a critical survival advantage. Most natural isolates of serovar Typhimurium show little variation in rpoS sequence. Jordan (17) examined the variability of rpoS sequence in environmental isolates of salmonellae of various species and found that the differences were at the single-base level. Waterman and Small (27) found that 11 of 58 clinical isolates carried rpoS^att alleles but only relatively small changes in sequence.

Among E. coli strains, there have been several observations with regard to rpoS variability. One strain had a tandem duplication of 46 bases in rpoS that resulted in a striking alteration of phenotype (15, 28). Jishage and Ishihama (16) reported variations in rpoS composition among 11 stocks of E. coli W3110. The differences stemmed from an E. coli W3110 strain sent to Japan four decades ago and subsequently subcultured in various laboratories in Japan (16). In one of these subcultures, the RpoS that normally has 330 residues is now truncated to only 269.

Although bacterial cells remain viable for extended periods in stationary-phase cultures, after reaching the maximum growth population in a broth culture (∼2 × 10⁹/ml for serovar Typhimurium), the population is reduced in the broth culture to ∼10⁷/ml, or a loss of over 99%. However, survivors, upon subculturing, grow back up to ∼2 × 10⁹/ml. Whether the selection process is purely random or whether there is a basis for selection of survivors is not completely understood (1). One hypothesis is that population shifts are an important general strategy for survival in the natural environment. When nutrients are plentiful, bacteria in liquid culture grow exponentially. When essential nutrients are depleted, however, the culture enters the so-called stationary phase, in which there is no net increase in cell mass.

Table 2 shows the relationship between rpoS sequence and expression of HPII activity in the strains tested. RpoS, as a positive regulator, controls synthesis of such enzymes as HPII (shield against oxidative damage [10]) and exonuclease III (repair of oxidatively damaged DNA [10, 23]). Since RpoS has a complex regulatory role, interacting with many other gene products (25), it is not surprising that the correlation is not exact. Sigma factors other than RpoS can take over transcription functions. For example, HPII can be upregulated when strains are defective in oxyR, the normal regulator of HPI (14) that is under sigma 70 control.

The negative regulatory power of RpoS could also assist in determining survival capacity (25). These could protect starved cells for some time, with some cells in the population dying but others remaining mortal. Perhaps, after the log phase of growth, this advantage is due to cells storing glycogen as a carbon source and trehalose as a membrane protector. Thus, RpoS manages a large array of gene products with varied functions to survive the adversities of minimal metabolism and cell division. Decades ago, Atwood and Ryan noted that under nutrient deprivation, mutants were selected with survival capabilities greater than that of the parental strain (1, 21, 22). Thus, we might expect survivor mutants under the confined environment in our vials, and we will continue to search for what these might be, starting with mutations in mutator genes, other regulator genes, and deletions that may be suspected after chromosomal analysis.