OUP user menu

Effect of rpoS mutations on stress-resistance and invasion of brain microvascular endothelial cells in Escherichia coli K1

Ying Wang , Kwang Sik Kim
DOI: http://dx.doi.org/10.1111/j.1574-6968.2000.tb08902.x 241-247 First published online: 1 January 2000


Escherichia coli K1 strains are predominant in causing neonatal meningitis. We have shown that invasion of brain microvascular endothelial cells (BMEC) is a prerequisite for E. coli K1 crossing of the blood–brain barrier. BMEC invasion by E. coli K1 strain RS218, however, has been shown to be significantly greater with stationary-phase cultures than with exponential-phase cultures. Since RpoS participates in regulating stationary-phase gene expression, the present study examined a possible involvement of RpoS in E. coli K1 invasion of BMEC. We found that the cerebrospinal fluid isolates of E. coli K1 strains RS218 and IHE3034 have a nonsense mutation in their rpoS gene. Complementation with the E. coli K12 rpoS gene significantly increased the BMEC invasion of E. coli K1 strain IHE3034, but failed to significantly increase the invasion of another E. coli K1 strain RS218. Of interest, the recovery of E. coli K1 strains following environmental insults was 10–100-fold greater on Columbia blood agar than on LB agar, indicating that growing medium is important for viability of rpoS mutants after environmental insults. Taken together, our data suggest that the growth-phase-dependent E. coli K1 invasion of BMEC is affected by RpoS and other growth-phase-dependent regulatory mechanisms.

  • Sigma factor
  • RpoS
  • Nonsense mutation
  • Growth medium
  • Escherichia coli K1
  • Brain microvascular endothelial cell invasion

1 Introduction

Escherichia coli K1 is the most common Gram-negative microorganism that causes meningitis during the neonatal period. Most cases of E. coli K1 meningitis develop as a result of hematogenous spread, but the basis of E. coli K1 crossing of the blood–brain barrier is not clear. In studying the mechanisms of E. coli K1 traversal of the blood–brain barrier, we have developed the in vitro model of the blood–brain barrier by isolation and cultivation of brain microvascular endothelial cells (BMEC) [1]. We have found that stationary-phase cultures of E. coli K1 are more invasive than exponential-phase cultures in BMEC ([2] and unpublished results), suggesting that E. coli K1 invasion of BMEC is growth-phase-dependent.

The sigma factor RpoS (also known as σ38, σs or KatF) is the second principal σ subunit after the major σ70 factor. It can reach up to 30% of the level of σ70 in the early stationary phase [3]. RpoS positively and negatively regulates a large set of genes which are expressed when bacteria enter the stationary phase [46]. During such transition, bacteria undergo physiological changes that allow their stationary-phase organisms to survive better in such insults as heat, high osmotic environment, starvation, UV radiation, H2O2 and acid than their exponential counterpart [4,7]. In addition, RpoS has been shown to regulate the expression of the plasmid-encoded spv virulence genes in Salmonella[8,9]. The rpoS mutant of Salmonella typhimurium exhibited the decreased ability to colonize murine Peyer's patches after oral inoculation than its wild-type parent strain in a mouse model. But RpoS was found not to play a role in the invasion of S. typhimurium in several mouse and human cell lines [10]. It has been suggested that RpoS may contribute to the low infective dose associated with shigellosis, since the rpoS mutant of Shigella flexneri lost its ability to survive at pH 2.5 for 2 h in vitro [11]. On the other hand, RpoS has not been shown to be involved in the virulence of Yersinia enterocolitica[12] and Vibrio cholerae[13], although it does contribute to their survival in the environmental stress.

The pathogenesis and pathophysiology of E. coli K1 meningitis have been investigated mostly using two E. coli K1 isolates from the cerebrospinal fluid (CSF) of neonates with meningitis, strains RS218 [1,14,15] and IHE3034 [16], whose serotypes are identical, O18:K1:H7. In the present study, we showed that both of these E. coli K1 strains, which are known to be invasive in BMEC ([2,14,15] and unpublished results), contain a nonsense codon in their rpoS gene. In addition, we showed that the recovery of rpoS mutant bacteria following stress challenge is greatly influenced by growth media.

2 Materials and methods

2.1 Strains, plasmids and culture conditions

E. coli K1 strains used were E44, a spontaneous rifampicin-resistant (RifR) mutant of RS218 [15], strain IHE3034Sm (or 3034Sm), a spontaneous streptomycin (Str)-resistant mutant of IHE3034 [16], strain C5 [17] and strain E412 [1]. Strains RS218, IHE3034 and C5 were isolated from the CSF, and strain E412 was isolated from the blood. E. coli K12 strain MC4100 [18] and its rpoS::Tn10 mutant RH90 [19] were used as RpoS positive and negative controls, respectively. Plasmids used were pACYC184, pACKatF which is the pACYC184 containing the cloned rpoS gene [20], and pCVD442 [21].

E. coli strains were cultured at 37°C in LB (1% tryptone, 0.5% yeast extract, 0.5% NaCl) and stored at −75°C with 20% glycerol. Brain heart infusion (BHI, Difco Laboratories, Detroit, MI, USA) broth and Columbia agar with 5% sheep blood (Remel, Lenexa, KS, USA) were used. When necessary, the medium was supplemented with chloramphenicol (Cm, 24 μg ml−1), rifampicin (Rif, 50 μg ml−1), ampicillin (Ap, 100 μg ml−1) or Str (300 μg ml−1).

2.2 PCR and nucleotide sequencing

DNA amplification was performed as previously described [22] with the Taq DNA polymerase (Promega, Madison, WI, USA). Primers used were: rpoS1, 5′-GAGCTGAACGTTTACCTG which is located at +519 in the rpoS open reading frame (the first nucleotide of the start codon ATG is assigned as +1); rpoS1R, 5′-GTTCAGCTCGAACAGCCA which starts at the +800 site; rpoS2, 5′-TAGCGACCATGGGTAGC, located at the −166 site. The amplified DNA was purified with the PCR purification system (Qiagen, Valencia, CA, USA). Nucleotide sequencing was carried out using the Applied Biosystems automatic sequencer (Foster City, CA, USA).

2.3 Resistance to environmental stress

Bacteria were grown in LB containing Cm at 37°C overnight, and collected by centrifugation. Bacteria were suspended and diluted to 107 colony forming units (CFU) ml−1 in phosphate-buffered saline for the following assays. For acid endurance, 1/10 volume of the bacterial suspension was mixed with LB containing acetic acid (final concentration, 90 mM, ∼pH 2.8) and incubated at 37°C for 20 min. For high osmolarity challenge, bacteria were mixed with an equal volume of 4.8 M NaCl and incubated at 37°C for 6 h. For heat shock, 100 μl of bacteria was heated at 54°C for 5 min. After exposure to these stresses, bacteria were diluted in 0.9% saline and plated in duplicate on LB or Columbia blood agar.

2.4 Catalase assay and Western blot analysis for RpoS

The catalase hydroperoxidases, HP-I and HP-II, were examined in the lysates of stationary-phase E. coli cultures following the method described previously [23]. Briefly, bacteria were grown in BHI broth at 37°C overnight. Cells were washed and resuspended in a phosphate buffer to 1/10 of the original culture volume and sonicated. Protein concentrations in the lysates were determined by using the Bradford reagent (Bio-Rad Laboratories, Hercules, CA, USA) with bovine serum albumin as the standard. The lysates, containing ∼5 mg ml−1 protein, were heated at 55°C for 15 min to inactivate the HP-I. The heat-stable catalase activity was calculated as HP-II as described [23].

For Western blot, bacterial lysates were prepared as above and separated on an 8–16% gradient sodium dodecyl sulfate-polyacrylamide gel (Bio-Rad). Proteins were transferred onto an Immobilon-P Transfer Membrane (Millipore, Bedford, MA, USA) and probed with anti-RpoS antiserum (or anti-sigma S, obtained from Ishihama [3,24]) as previously described [15]. Proteins were visualized with SuperSignal Chemiluminescent Substrate (Pierce, Rockford, IL, USA). The intensity of signals was measured with a GS300 scanning densitometer (Hoefer Scientific Instruments, San Francisco, CA, USA).

2.5 Tissue culture invasion assay

BMEC were prepared from bovine brains [1] and invasion assays were performed as previously described [14]. Briefly, bacteria were cultured in BHI at 37°C overnight without agitation. The bacteria were diluted with the tissue culture medium and added to confluent monolayers of BMEC with a multiplicity of infection of 100. The monolayers were incubated for 1.5 h at 37°C to allow invasion to occur. The number of intracellular bacteria was determined on Columbia blood agar after the extracellular bacteria were killed by incubation of the monolayers with experimental medium containing gentamicin (100 μg ml−1). The bacterial inocula were also prepared by 1:50 dilution of overnight cultures in BHI and incubation at 37°C for 2 h. Results were expressed as percentage invasion ([number of intracellular bacteria/number of total bacteria added]×100). For competition invasion assays, equal numbers of two E. coli strains were mixed and applied on BMEC monolayers. Intracellular bacteria were recovered on LB containing different antibiotics to determine the ratio of the two E. coli strains.

3 Results

3.1 E. coli K1 strains E44 and IHE3034Sm are rpoS mutants

To test whether there is an rpoS gene in these two E. coli K1 strains, primers rpoS1 and rpoS1R, designed from the rpoS sequence of E. coli K12 strain [25], were used to amplify the DNA from E44 and 3034Sm. An E. coli K12 DNA, pACKatF, and pACYC184 were included as positive and negative controls, respectively. An identical PCR product of 281 bp was obtained from all of these templates except for pACYC184, indicating that E. coli K1 strains E44 and 3034Sm have a copy of DNA similar to the rpoS gene of E. coli K12.

We then constructed an rpoS mutant in strain E44 by introducing a suicide plasmid in the middle of rpoS. The 580-bp StuI-AccI fragment in the middle of rpoS (see Fig. 1) was cloned in pCVD442 at the SmaI site. The resulting plasmid was transferred into E44 by conjugation as previously described [14]. The RifRApR transconjugants would have the suicide plasmid inserted into the rpoS locus by homologous recombination. The mutant strain, however, behaved similarly as its parent strain in the stress-surviving tests, including H2O2 sensitivity, high salt, acid endurance and heat shock (data not shown). The invasion frequency of the rpoS mutant in BMEC did not differ significantly from that of the parent strain E44 (data not shown).


Partial nucleotide sequence of DNA containing the E. coli K1 strain E44 rpoS gene (GenBank accession number AF082844). The deduced amino acid sequence is presented. The different nucleotides in E. coli K12 are depicted as small letters on the top of the E44 DNA sequence. Mutations are underlined. The single mutation resulting in a stop codon in strain IHE3034Sm is indicated. *, stop codon.

To examine the rpoS locus in these E. coli K1 chromosomes, the primer rpoS2 was used with the primer rpoS1R to amplify a 966-bp fragment from E44 chromosomal DNA. The PCR DNA was purified and sequenced directly. Within a 900-bp region (Fig. 1), eight nucleotides were found to be different from those in E. coli K12 (99.1% identity). Six of them do not result in coding changes. A C-to-G change at codon #33 site results in a change of CAG glutamine codon to GAG glutamate codon. Of interest, a C-to-A change at codon #221 resulted in a change of TCG, encoding serine, to TAG stop codon, indicating that the strain E44 is an rpoS mutant. A similar piece of DNA was also amplified from the 3034Sm chromosome and sequenced. The sequence was found to be identical to that of E44, except that there is no mutation in the codon #221, instead, a T-to-G mutation was identified at codon #71 resulting in a change of TTA, encoding leucine, to TGA stop codon. These findings indicate that both E. coli K1 strains E44 and 3034Sm have a nonsense mutation in their rpoS gene.

Two similar DNA fragments were sequenced from two other E. coli K1 strains, C5 and E412. The sequenced rpoS DNA (900 bp) from C5 was completely identical to that of E. coli K12, while in E412, only one nucleotide change, a C-to-G change at codon #33, was identified compared to that of E. coli K12.

3.2 The stop codons result in a leaky phenotype in E. coli K1 strains

E. coli synthesizes two catalase hydroperoxidases, HP-I and HP-II. The expression of HP-II, which is more heat-resistant than HP-I, depends on RpoS [23]. The stationary-phase catalase activities were tested in various E. coli strains. Both E44 and 3034Sm had negligible HP-II activities (Table 1), which are supportive of the sequence data indicating that both of these strains are defective in rpoS. The other two E. coli K1 strains, C5 and E412, clearly exhibited HP-II activity compared to that of MC4100, the RpoS+E. coli K12 strain.

View this table:

Stationary-phase catalase activities

StrainCatalase activitya
  • aActivities are given in μmol min−1 mg−1 and are averages of at least two experiments.

The expression of RpoS in the E. coli K1 strains was investigated by Western blot analysis with polyclonal anti-RpoS antiserum. Fig. 2 shows that both E44 and 3034Sm produce a negligible amount of RpoS of the same size as E. coli K12 RpoS (38 kDa, but running at the 42-kDa position as reported in the literature [26]). For quantification, the intensity of the RpoS band in MC4100 was set as 100%, and RH90 0%. Other bands in Fig. 2 were 1.4% for 3034Sm, and 3% for E44, while E44 complemented with rpoS was 78%. The results showed that the expression of RpoS in the E. coli K1 strains, E44 and 3034Sm, was leaky. The truncated RpoS products were not detected in strains RH90, E44 and 3034Sm.


Western analysis of RpoS expression. Lanes: 1, MC4100; 2, RH90 (MC4100 with rpoS::Tn10); 3, IHE3034Sm; 4, E44; 5, E44(pACKatF). Numbers shown on the left are positions and sizes in kDa of prestained molecular mass markers (Bio-Rad).

3.3 Effect of growth media on recovery of stress-challenged rpoS mutants

We tested the E. coli K1 strains and the rpoS-complemented strains for their tolerance to acid, heat and salt. The percentage survival of 3034Sm in the above stress conditions was similar to that of 3034Sm(pACYC184), indicating that the plasmid had little effect on the host in survival following stresses (data not shown). As expected, E. coli strains harboring the K12 rpoS gene were able to survive in the tested stress conditions when plated either on LB or Columbia blood agar (Fig. 3). However, considerable differences in CFU were found when the stress-challenged rpoS mutant bacteria were plated on different media. For example, CFU of RH90(pACYC184) on Columbia blood agar were 10–100-fold higher than those on LB agar following challenge with high osmolarity or heat shock (Fig. 3).


Effect of medium on recovery of stress-challenged bacteria. Different rpoS mutant E. coli strains containing the vector pACYC184 (RpoS) or pACKatF (RpoS+) were treated as described in Section 2. CFUs were determined on either LB or Columbia blood agar. *, CFU below 1%. Error bars indicate S.D. (n=3).

3.4 Effect of RpoS on E. coli invasion in BMEC

We next examined and compared the BMEC invasion frequencies of E44(pACYC184) vs. E44(pACKatF) and 3034Sm(pACYC184) vs. 3034Sm(pACKatF). Fig. 4 shows that overnight cultures of E. coli K1 were 3–4-fold more invasive than exponential-phase cultures, i.e. the 2-h cultures. As shown in Fig. 4, complementation with the rpoS gene significantly increased the BMEC invasion of the overnight cultures of 3034Sm (P=0.03), but did not significantly increase the invasion of strain E44 (P=0.07). Competition invasion assays were also performed and the results were consistent with those of Fig. 4. Using 1:1 ratio of 3034Sm(pACYC184):3034Sm(pACKatF), the ratio of intracellular bacteria recovered was 1:2.1 (average number of three assays), while the ratio of recovered intracellular E44(pACYC184):E44(pACKatF) was 1:1.3.


Effect of RpoS on E. coli K1 invasion of BMEC. The bacterial inocula were 3034Sm or E44 containing the vector pACYC184 (RpoS) or pACKatF (RpoS+) grown for 2 h or overnight. *, P=0.03 compared to 3034Sm(pACYC184). Error bars indicate S.D. (n=3).

4 Discussion

RpoS plays an important role in bacteria surviving in starvation and stress [4,7,19,27]. RpoS+ bacteria, therefore, should have a survival advantage over their rpoS mutants in various environmental stresses. But, mutations in rpoS have been found in apparently wild-type E. coli K12 strains and in strains which have been subjected to long-term starvation [23,24,28,29]. Many of these mutants have a stop codon in the rpoS gene. Since the efficiency of translational stop signals is affected by the neighboring nucleotides of stop codons [30], these mutant stop signals may not function as efficient as the natural ones, which are usually flanked with certain nucleotides [30] or composed of more than one stop codons. Thus the nonsense mutations often result in a leaky phenotype [23,30]. Our results indicated that rpoS mutations occurred in the pathogenic E. coli K1 strains isolated from the CSF of neonates with meningitis. We, however, cannot completely exclude the possibility that both nonsense mutations in E44 and 3034Sm might have arisen from laboratory passage. Nevertheless, it is worth noting that the two widely studied E. coli K1 strains contain nonsense mutations in their rpoS gene.

RpoS+ bacteria have been shown to survive much better after the environmental insults than RpoS strains [19,27]. Interestingly, we found that after environmental stresses, e.g. heat shock or high salt, almost 100-fold more CFU of RpoS strains could be recovered on Columbia blood agar compared to the commonly used LB agar. Our findings suggest that some of RpoS bacteria are viable after exposing to certain stresses, but their recovery requires an enriched medium. These findings show that growth medium is an important factor in examining the phenotypes associated with RpoS.

RpoS has been shown not to play a critical role for the ability of Y. enterocolitica grown at 26°C to survive against diverse environmental stress [12]. As described above, we showed that most rpoSE. coli survived after certain environmental insults when cultured in a rich medium. This suggests that RpoS and other growth-phase-dependent regulatory mechanisms contribute to the survival of E. coli following environmental stress. Other uncharacterized growth-phase-dependent regulatory mechanisms have been indicated to exist. For example, many RpoS-regulated genes have been shown to exhibit growth-phase-dependent expression even in an RpoS background [5].

Stationary-phase cultures of E. coli K1 strain E44 have been shown to be 2–3-fold more invasive in BMEC than exponential-phase cultures [2]. Here, we showed a similar result for another E. coli K1 strain IHE3034Sm (Fig. 4). The invasion experiments with 3034Sm(pACYC184) and 3034Sm(pACKatF) suggested that RpoS contributed to greater invasion of BMEC by stationary-phase culture of E. coli K1 strain 3034Sm. However, the invasion frequency of rpoS-complemented E44 did not differ significantly from that of E44(pACYC184) in their stationary-phase growth (Fig. 4). These findings suggest that growth-phase-dependence of BMEC invasion by E. coli K1 is affected by RpoS and other uncharacterized mechanisms. It is important to note that invasion frequencies of BMEC by E. coli K1 strains E44 and 3034Sm are approximately 0.1–0.2%, which are considerably lower than the invasion frequency of epithelial cells by other Gram-negative bacteria such as Shigella or Salmonella spp. (usually 1–10%). However, we have shown that the BMEC invasion frequency of approximately 0.1% contributes to enhanced bacterial penetration of the blood–brain barrier in vivo and is thus biologically relevant [14,31].

In conclusion, we found that two commonly used CSF isolates of E. coli K1 strains RS218 and IHE3034 are rpoS mutants, while such rpoS mutations were not observed for two other E. coli K1 clinical isolates. We showed that a stop codon mutation in rpoS resulted in 1–3% expression of RpoS. The rpoS mutants were, as expected, impaired in their tolerance to some environmental stresses, but the recovery of the stress-challenged rpoS bacteria was much greater by using an enriched growth medium. The growth-phase-dependent E. coli K1 invasion of BMEC appeared to be regulated by RpoS as well as other growth-phase-dependent regulatory mechanisms.


This study was supported by NIH Grant R01-NS 26310 (to K.S.K.) and a CHLA Research Institute career development fellowship (to Y.W.). We thank J.H. Weiner of University of Alberta for providing pACKatF, MC4100 and RH90, T. Korhonen of University of Helsinki for IHE3034Sm, A. Ishihama of National Institute of Genetics, Japan, for anti-RpoS antiserum, C. Wass for tissue culture, and J. Badger for critical reading.


  1. [1]
  2. [2]
  3. [3]
  4. [4]
  5. [5]
  6. [6]
  7. [7]
  8. [8]
  9. [9]
  10. [10]
  11. [11]
  12. [12]
  13. [13]
  14. [14]
  15. [15]
  16. [16]
  17. [17]
  18. [18]
  19. [19]
  20. [20]
  21. [21]
  22. [22]
  23. [23]
  24. [24]
  25. [25]
  26. [26]
  27. [27]
  28. [28]
  29. [29]
  30. [30]
  31. [31]
View Abstract