GASP phenotype: presence in enterobacteria and independence of σ^S in its acquisition

Abstract

The appearance of growth advantage in stationary phase or GASP was originally detected in Escherichia coli. The presence of this phenotype in other enterobacteria such as Enterobacter cloacae, Salmonella typhimurium, Providencia stuartii and Shigella dysenteriae is described in this work. E. cloacae GASP strains presented lower levels of RpoS than the parental strain, although no mutation in the gene or its promoter was detected. This work offers evidence of GASP rpoS-independent pathways as GASP was also acquired in knock-out rpoS E. cloacae and E. coli strains.

Growth advantage in stationary phase, rpoS, Stationary phase, Enterobacteria

1 Introduction

The cycling known as feast and famine is the cell response to the presence and depletion of nutrients that occurs in the bacterial environment. Gene expression changes widely between both states in order to permit the cells to grow and colonise niches or to overcome starvation and stress conditions [1].

Viability of Escherichia coli populations suffers a drop in two orders of magnitude during the first days of nutritional starvation, cell viability diminishing slowly during long periods of time [2]. The cell has to overcome several stress conditions during stationary phase, the resistance being a mechanism induced by the cessation of growth [3]. During this period the mutation rate increases due to different factors, a mechanism that is responsible for the appearance of genetic variability [4].

During both, exponential and stationary phases, some mutants arise that are able to displace parental populations. This phenotype, known as GASP (growth advantage in stationary phase), consists of a continuous cycling of growth and death due to the appearance of new mutants with a better fitness than the parental strain [5]. These mutants can grow scavenging the nutrients released by dead cells with a higher growth rate than the parental strain. Populations during stationary phase are thus highly dynamic undergoing several cycles of displacement [5]. Therefore population switches occur continuously during the course of stationary phase, i.e. 10-day-old cells are genetically different to their parental strain and 20-day-old populations are different to the 10-day-old cells.

The mechanism of acquisition of GASP is not yet fully understood. Up to now, the only described GASP pathways involved rpoS down mutations [5], a gene that encodes an alternative sigma factor that regulates transcription of more than 50 stationary phase genes [6]. Only down mutations but not deletions or knock-out mutations in rpoS are able to confer the GASP phenotype [5]. Mutations in genes such as sgaA, sgaB and sgaC are also involved in subsequent steps [7]. sgaB is an allele of the lrp gene that codifies the leucine-responsive regulatory protein Lrp that regulates amino acid metabolism by increasing anabolism and decreasing catabolism [8]. The release of amino acids is characteristic of dying cells in stationary phase cultures. The presence of certain mutations in these genes (rpoS, sgaA, lrp and sgaC) allows the mutant strains to grow with a higher growth rate than the parental strain in minimal medium with amino acids as the sole carbon source [7]. Abolition of the GASP phenotype can occur due to knock-out mutations in genes such as nuoA and nuoB that encode subunits of NADH dehydrogenase I [9] and, intriguingly, to the presence of a high inhomogeneous magnetic field which has been suggested to increase rpoS expression [10].

GASP has been extensively studied in E. coli, but little is known in other bacteria. The aim of this work was the study of this phenotype in other enterobacteria. Our results indicate that GASP is a generalised behaviour within enterobacteriaceae and also show the presence of GASP pathways independent of rpoS down mutations.

2 Materials and methods

2.1 Culture media, cell growth conditions, bacterial strains and plasmids

Three-ml LB bacterial cultures were grown in 18 mm×150 mm glass tubes with constant aeration in a Cell Mixer Major CM200 at 37°C.

Colony forming units (cfu) were measured by plating different dilutions of the cultures on LB agar containing the appropriate antibiotics at a concentration of 150 µg ml⁻¹ (rifampicin), 20 µg ml⁻¹ (nalidixic acid), 50 µg ml⁻¹ (streptomycin) or 50 µg ml⁻¹ (kanamycin).

The strains used in this work were Enterobacter cloacae CECT960, E. cloacae E1 (CECT960 rpoS::Km) [11], E. cloacae RC1 (clinical isolate), Providencia stuartii CECT866, Salmonella typhimurium CECT443, Shigella dysenteriae CECT853, E. coli ZK126 (W3110 lacU169 tna2) [12] and E. coli ZK918 (ZK126 rpoS::Km^r) [13]. CECT strains correspond to the Colección Española de Cultivos Tipo. Antibiotic resistant derivatives (Rif^r, Nal^r) of E. cloacae and S. typhimurium were isolated by plating aliquots of overnight cultures on LB plates containing the appropriate antibiotic. Rif^r derivatives of P. stuartii and Nal^r derivatives of S. dysenteriae were also isolated.

Eleven-day-old cultures of a particular strain (1D strain) were streaked out on LB plates with the appropriate antibiotic and, after growth, a single colony was isolated (11D strains).

The plasmids used in this work were pMEU16, a pSELEC1 derivative with an EcoRI/BamHI 2.3-kb fragment containing the E. coli rpoS gene [11], pUC19 and pBEcl12, a pBR322 derivative with an EcoRV/BamHI 2.4-kb fragment containing the E. cloacae rpoS gene [11].

2.2 Competition assays

Mixed cultures experiments were performed as previously described by Zambrano et al. [5]. Overnight cultures of strains carrying antibiotic resistance markers were mixed up to a ratio of approximately 1:10 000 or 1:1. Cfu of the different populations of the cocultive were monitored daily by plating serial dilutions in LB agar plates supplemented with the appropriate antibiotic.

As S. dysenteriae and P. stuartii are resistant strains to many different antibiotics, the results of the competition experiments are the cell viability kinetics of antibiotic resistant cells obtained in the following two mixes: antibiotic^s strain (majority) versus antibiotic^r cells (minority) and antibiotic^r strain (majority) versus antibiotic^s cells (minority).

Frequency of acquisition of the GASP phenotype was performed by measuring the number of 10 independent 11-day-old cultures able to displace overnight cultures of the parental strain when mixed up to a ratio of 1:1 (11D strain/wild-type).

2.3 Cloning and sequencing

Chromosomal DNA from strains E. cloacae 11D Rif^r and E. cloacae 11D Nal^r was prepared as described [14]. PCR amplifications were performed in a 50-µl reaction mixture using 1 unit of taq DNA polymerase. The oligonucleotides used were P1 (5′-CTGAATTCCCTGAGTTGCCTACGCCC-3′; rpoS forward primer, located at the end of the upstream gene nlpD), P2 (5′-GGGAATTCGTGCTTAATCAGGAAGGGG-3′; rpoS reverse primer located downstream rpoS), N1 (5′-GGAATTCCTGCACAAAAAACCACCACGG-3′; nlpD forward primer, located in the middle of the gene) and B3 (5′-CGGATCCGCCCTGGACGAGAC-3′; nlpD reverse primer located in the intergenic region between nlpD and rpoS). An EcoRI site was included at the 5′ end of primers P1, P2 and N1 and a BamHI site at the 5′ end of primer B3. Thermal cycling included an initial denaturation step at 95°C for 120 s, followed by 30 cycles of 30 s at 95°C, 30 s at 55°C and 60 s at 72°C.

Plasmid copy number was determined as described [15].

2.4 Western blot

Samples containing 14 µg of total cell protein were electrophoresed on SDS 12.5%–(w/v) polyacrylamide gels and electroblotted onto nitrocellulose membranes (Bio-Rad Mini Protean II system). The membranes, after being probed with a 1:4000 dilution of anti-σ^S polyclonal antibody, were incubated with a 1:5000 dilution of anti-mouse IgG alkaline phosphatase conjugate (Amersham Life Sciences). Detection was made following the ECL method. Densitometry of the bands was performed using the Multi-Analyst 1.1 programme from Bio-Rad.

3 Results and discussion

3.1 GASP in enterobacteria

GASP was initially, and almost exclusively, described in E. coli[5]. Some spontaneous Xanthomonas oryzae[16] and Pseudomonas putida[17] mutants have been shown to be able to outcompete wild-type cells in stationary phase. This phenotype was not found in Caulobacter crescentus[18]. The arise of GASP mutants was studied in several bacteria to determine whether the appearance of this phenotype is a common feature in enterobacteria. The approach followed was similar to that described for E. coli[5]. To monitor growth and death of the subpopulations of cocultives, competition experiments were performed with spontaneous antibiotic resistant mutants of the different strains. The antibiotic resistance markers did not confer an increased or decreased fitness in the conditions assayed as both antibiotic resistant strains remained fully viable in 1:1 overnight mixes (Fig. 1A). On the contrary, the minor subpopulation died in 1:10 000 ratio mixes (Fig. 1B). As GASP mutations occur during cell growth, the probability of appearance of GASP mutants within a population is proportional to its cell number. The probability of presence of GASP mutants within the major population (10⁹ cfu ml⁻¹) is several folds higher than in the minor (10⁴ cfu ml⁻¹). Therefore the GASP mutants present in the major population can take over both, their parental population and the minor one. The presence of GASP mutants was clearly determined in 1:10 000 and 1:1 ratio mixes of a population of overnight 11D strains (minor population) with overnight parental strains (major population) of E. cloacae, S. typhimurium and S. dysenteriae (Fig. 1C,D). As expected for GASP mutants, the 11D strain subpopulations were able to grow and to take over the overnight parental strains that in most of the cases died out. Hence achievement of GASP mutations allows bacteria to overcome starvation. Bacterial Gram-negative cells probably have evolved similar mechanisms and metabolic pathways that enable them to survive under nutrient depletion.

1

Competition assays in E. cloacae, S. typhimurium and S. dysenteriae. Overnight populations with different markers were mixed at a ratio of 1:1 (A and D) and 1:10 000 (B and C). The bacterial strains were either wild-type (1D/1D; A and B) or wild-type and 11D (1D/11D; C and D). For S. dysenteriae, viability of overnight populations was followed in the spontaneous antibiotic-resistant mutant strain (A). The experiments were performed three times and a representative experiment is shown. Similar results were obtained regardless of the antibiotic resistance present in the wild-type or 11D strains.

3.2 GASP frequency

The appearance of the GASP phenotype seems not to be a rare event but rather a common one as it is present in many bacteria. To study also the frequency of appearance within a particular strain, aliquots of 10 different 11-day-old cultures of each, E. coli, E. cloacae, S. dysenteriae and S. typhimurium were mixed up with their respective wild-type cultures to a 1:1 ratio. Population displacement was studied in the different cocultives. The results showed that the GASP phenotype was acquired with a high frequency in E. cloacae (10 out of 10 cultures), E. coli (nine out of 10 cultures), S. dysenteriae (10 out of 10 cultures) and S. typhimurium (eight out of 10 cultures). These results support the fact that GASP arises with high frequency in cell populations, representing a general behaviour in Enterobacteriaceae.

3.3 GASP and rpoS

Several rpoS down mutations have been described to be responsible of acquisition of GASP, these mutations being the first ones to occur during the successive rounds of death and growth that characterise E. coli stationary phase [5]. To examine the relationship between rpoS and the GASP phenotype in bacteria other than E. coli, the E. cloacae rpoS gene of two 11D Rif^r and Nal^r GASP strains was amplified by polymerase chain reaction with rpoS flanking primers P1 and P2. The amplicons, cloned into the EcoRI site of pUC19, were sequenced. No mutations were found in the rpoS gene. E. cloacae CECT960 rpoS expression is driven from a weak promoter present in IS10R, a sequence found in the nlpD gene of this strain that isolates the rpoS gene from its original promoter rpoSp[19]. The insertion sequence IS10R remained the same in size and location in both, GASP mutants and the parental strain as probed by PCR amplification performed with the oligonucleotides N1 and B3 that flank IS10R inserted in the E. cloacae CECT960 nlpD gene (data not shown). The amplicons were cloned into the EcoRI/BamHI sites of pUC19 and sequenced. No mutations were detected in the promoter region of the mutant strains. As rpoS is subject to a rather complex regulation both at the transcriptional and translational levels and at protein stability [20], the RpoS protein level was compared between the parental and GASP E. cloacae strains. The results from a Western blot showed that the RpoS level of the two GASP strains was lower than their parental strain (Fig. 2A). It is highly possible that a mutation in one of the control pathways of RpoS synthesis or stability could have been responsible for this RpoS down phenotype that would lead the 11D strains to acquire the GASP phenotype.

2

RpoS levels in E. cloacae CECT960 and derivatives. Samples containing 14 µg of total cell protein were electrophoresed on SDS–12.5% (w/v) polyacrylamide gels and electroblotted and probed with polyclonal anti-σ^S. A: E. cloacae CECT960 and two GASP derivatives. ZK918 transformed with pBEcl12 (a plasmid that contains E. cloacae rpoS) as a positive control (lane 1). ZK918 as a negative control (lane 2). Prestained marker (lane 3). E. cloacae CECT960 (lane 4). E. cloacae CECT960 Rif^r 11D (lane 5) and CECT960 Nal^r 11D (lane 6). B: E. cloacae CECT960 (lane 1) and E. cloacae RC1 (lane 2).

The results so far obtained and published seemed to indicate that rpoS was the first gene to mutate in the GASP cycling [5]. As shown in Fig. 2B, E. cloacae CECT960 displays much lower levels than the E. cloacae clinical isolate RC1 (a strain in which rpoS transcription is driven by the stronger promoter rpoSp as IS10R is not present in the nlpD gene; [19]). CECT960 was therefore expected to outcompete RC1 in a cocultive, a behaviour that could have important implications in microbial ecology and pathogenesis. Unexpectedly RC1 displaced CETC960 in a mixing experiment (results not shown). This result suggests that there could be GASP pathways independent of RpoS levels. As rpoS down mutations render the mutants more sensitive to environmental stress during stationary phase, it is possible that natural populations could acquire GASP through pathways independent of rpoS down mutations so that they could face both nutrient depletion and environmental stress.

3.4 rpoS-independent GASP

The acquisition of GASP phenotype through a RpoS-independent pathway was studied using E. coli and E. cloacae rpoS knock-out mutants ZK918 (E. coli rpoS::Km) and E1 (E. cloacae rpoS::Km). E. cloacae CECT960 and E. coli ZK126 were cocultured with overnight rpoS strains ZK918 and E1 and with their 11D derivatives. Non-aged strains presented no growth advantage (results not shown) while knock-out rpoS 11D strains outcompeted wild-type strains (Fig. 3). From these results it can be concluded that the GASP phenotype can also be obtained through a pathway independent of rpoS down mutations. Moreover, the frequency of GASP acquisition was similar in wild-type and rpoS knock-out strains (nine out of 10 cultures versus 10 out of 10 cultures in E. coli and 10 out of 10 cultures versus eight out of 10 cultures in E. cloacae).

3

Competition assays between wild-type versus rpoS knock-out strains; competition assay in E. coli harbouring pMEU16 (a plasmid that contains E. coli rpoS). Cell viability in a cocultive of overnight populations of E. cloacae CECT960 Nal^r (wild-type) versus E. cloacae E1 11D (rpoS::Km); E. coli ZK126 Nal^r (wild-type) versus E. coli ZK918 11D (rpoS::Km); ZK126 Nal^r (pMEU16) versus ZK126 (pMEU16) 11D. The experiments were performed three times and a representative experiment is shown. Circles: wild-type strains; squares: 11D strains.

An alternative method to study GASP acquisition in a RpoS-independent pathway was to study its achievement in wild-type E. coli (ZK126) harbouring pMEU16, a recombinant plasmid with the E. coli rpoS gene [11] in which it would be impossible to achieve GASP through a rpoS down mutation. Our results showed that an 11D derivative of ZK126 (pMEU16) displaced 1D ZK126 (pMEU16) in a coculture (Fig. 3). This result was not due to differences in plasmid copy number as it did not vary significantly during stationary phase (39 plasmids per genomic equivalent after 1 day of growth versus 47 after 11 days). As in previous experiments, the non-aged strains did not show an outcompeting behaviour (results not shown). These results point again that a rpoS down mutation is not a necessary step in the achievement of GASP.

As GASP does not require rpoS down mutations to be present, it is possible that bacteria that do not have the rpoS gene could also acquire GASP. This happened to be true for P. stuartii, an enterobacteria that does not have a sequence with a high degree of identity with E. coli rpoS as shown by Southern blot and PCR (11). Overnight wild-type populations in minority are overtaken by the major population. On the contrary a minority population of overnight 11D strain outcompetes the majority (Fig. 4). The results also showed that the GASP phenotype was acquired in P. stuartii with a high frequency (10 out of 10 cultures).

4

Competition assays in P. stuartii. Viability of overnight populations was followed in the spontaneous antibiotic-resistant mutant strain (A) and in mixed populations of antibiotic-resistant and antibiotic-sensitive bacteria at a ratio of 1:10 000 (1D/1D in B and 1D/11D in C) or 1:1 (1D/11D in D). The experiments were performed three times and a representative experiment is shown.

In E. coli multiple pathways in the GASP cycling have been described after achieving a first down mutation in rpoS. The results obtained in this work indicate that the first reported step is not necessary, as there are alternative pathways for GASP achievement. It is interesting to note that if the rpoS gene is present, the most frequent mutations in laboratory conditions that confer a GASP phenotype are rpoS down mutations. Probably the answer to this fact is the phenomena described as cell cheating: expression of the σ^S regulon reduces cell growth in parallel to an increase in stress resistance and therefore, mutations that reduce this expression would permit growth based on scavenging [21]. As it has been pointed out, mutations in rpoS would be the first ones to occur during GASP in order to exit the cell growth arrest caused by the σ^S regulon [21].

Our results show that the GASP phenotype is a generalised behaviour in Enterobacteriaceae. Stationary phase induced by starvation causes a higher rate of mutation, thus increasing the frequency of appearance of new alleles [4] that could enable the cells to face this stress situation by the achievement of a better fitness [7]. GASP could represent a pathway followed by non-sporulating cells to overcome nutrient depletion in an aggressive environment. An important difference between these resistance strategies (spores and GASP) is that the later represents a non-return pathway as cells enter a continuous cycling of mutation events that eventually causes the impossibility of cells to grow in rich media [22].

Acknowledgements

This work was supported by the project BIO097-1246. E.M.G. was a recipient of a scholarship associated with this project. We acknowledge Regine Hengge-Aronis for σ^S antibodies and Maria Marosini (Hospital Ramón y Cajal, Madrid, Spain) for E. cloacae RC1 strain.