Stress and Survival of Aging Escherichia coli rpoS Colonies

Abstract

In Escherichia coli, the expression of the RpoS regulon is known to be crucial for survival in liquid cultures during stationary phase. By measuring cell viability and by transcriptome analysis, here we show that rpoS cells as well as wild-type cells survive when they form colonies on solid media.

ESCHERICHIA coli possess seven RNA polymerase σ-factors, which compete for association with the core polymerase subunit, each coordinating the transcription of a set of genes, which allows fine control of adaptation to different physiological conditions (Record et al. 1996). In cells that approach stationary phase, the σ-factor σ³⁸ (RpoS) competes with the primary σ-factor σ⁷⁰ (RpoD) for association with the core polymerase subunit. To date ∼100 genes have been reported to belong to the RpoS regulon whose induction results in physiological and morphological modifications that increase resistance to various stresses such as heat shock, cold shock, acid shock, oxidative stress, osmotic stress, and UV light (Storz and Hengge-Aronis 2000).

Paradoxically, mutations and even inactivation of the rpoS gene might be tolerated or advantageous in nature, despite the partial or complete loss of RpoS-dependent resistance to environmental challenges, as suggested by the high degree of allelic variation of the rpoS genes from E. coli natural isolates (Herbelin et al. 2000; Jorgensen et al. 2000). Several studies have shown that the fitness of rpoS mutants depends on environmental conditions (Ferenci 2001; Farrell and Finkel 2003). However, the influence of the structured environment on fitness of the rpoS mutant is unknown. In nature, E. coli is often found in structured environments, e.g., microcolonies, aggregates, or biofilms. Hence, in this study we have tested how a structured environment influences the survival of an E. coli rpoS mutant to determine whether the selection of rpoS alleles found in nature can occur in such environments.

While the decrease in the number of colony forming units (CFUs) obtained from planktonic cells between 1 day (D1) and 7 days (D7) in culture was 200-fold for rpoS mutant and 10-fold for wild-type strain (wt), there was little difference in colonies between D1 and D7 for each strain as well as between the two strains (Table 1a). Furthermore, the measurements of the cell viability in aging colonies, as well as of the colony area, show that the rpoS mutant had difficulties relative to wt during the first 24 hr (D1) but there was no difference between the two strains after 7 days (D7; Table 1b), indicating better survival of rpoS cells at D7 than at D1, and/or increased cell turnover. To identify the functions that might be responsible for fitness modifications of rpoS mutants in structured environments, we established genome-wide transcription profiles of cells in aging colonies of wt and rpoS strains using macroarray technology.

TABLE 1.

Growth and death within wild-type andrpoS colonies

a	CFU per milliliter of liquid LB	SD	CFU per colony	SD
wt D1	2.74 × 109	4.1 × 108	2.19 × 109	1.14 × 109
wt D7	2.98 × 108	1.4 × 108	4.05 × 109	2.15 × 109
rpoS D1	3.11 × 109	6.3 × 108	1.9 × 109	0.7 × 109
rpoS D7	2.96 × 107	3.2 × 107	4.32 × 109	1.7 × 109
Colony area (mm2)		% of dead cells per colony
b	D1	D7	D1	D7
wt	1.76	136	2.94	58
rpoS	1.56	147	7.2	59
P-value	0.003	0.36	5.1 × 10−5	0.56

We used the E. coli K12 MG1655 strain (wt) and the derived rpoS mutant strain (the rpoS359::Tn10 allele was kindly provided by M. E. Winkler, Indiana University, Bloomington, IN). Bacteria were grown in agitated Luria-Bertani broth (LB) liquid medium at 37° for 1 day or 7 days. For colony analysis, 10²–10³ cells from an overnight liquid culture were inoculated onto nitrocellulose filters (NC 45 from Schleicher & Schuell, Keene, NH), laid on fresh LB plates (agar 15 g/liter), and incubated at 37° for 1 day (D1) or 7 days (D7). Afterward, the filter with the growing colony was put in 1 ml of LB and vortexed, which enabled the recovery of all the cells. To calculate the number of CFUs per milliliter of liquid culture or per colony, aliquots of diluted cells were plated on LB and the CFUs were counted 1 day after plating. The average number of CFUs and the standard deviation (SD) from six experiments are indicated in a. The areas of 22 wt colonies and 22 rpoS colonies at D1 and D7 were measured using the software Ellix (Microvision, Evry, France). Percentages of dead cells within 7 wt colonies and 7 rpoS colonies at D1 and D7 were estimated by flow cytometry analysis (FASCscan, Becton Dickinson) using the two-color fluorescence (syto 9 green/propidium iodide) assay from Molecular Probes (Eugene, OR; live/dead Baclight bacterial viability kits). The null hypothesis of equal mean surface and equal mean percentage of dead cells (H1) between the two strains at each time was tested by calculating the t-test and the P-value (significant when P < 0.05). The area averages, the percentage of dead cell averages, and P-value are reported in b.

The transcriptome analysis supports the observation that the rpoS colonies have more difficulties at D1 compared to wt colonies (Tables 1b and 2). For example, at D1, we observed a lower expression of genes coding for translation apparatus [ribosomal proteins (50S and 30S subunits)] in rpoS than in wt colonies, which may indicate that the rpoS cells have a lower growth rate than wt cells (Gausing 1977). Furthermore, at D1, we observed a higher expression of the oxyR and soxR genes (coding for transcriptional regulators involved in the response to hydrogen peroxide and superoxide, respectively; Storz and Imlay 1999) in rpoS colonies than in wt colonies, indicating that oxidative stress might be the major problem for rpoS cells (Table 2). The induction of these two stress responses in rpoS mutant may be the consequence of the deficiency in RpoS-mediated protection against oxidants (Dps, KatE; Hengge-Aronis 2002) and/or an increase in production of reactive oxygen species. Moreover, some genes that are not regulated by OxyR but are induced by H₂O₂ treatment (Zheng et al. 2001)—such as the ydcH (unknown function) and the yfiA gene encoding a protein associated with 70S ribosomes in stationary phase (Maki et al. 2000)—were also upregulated in rpoS compared to wt at D1. The decrease in viability of rpoS cells at D1 could also result from the absence of downregulation of the ompF gene (coding for major outer membrane porin; Liu and Ferenci 2001), which is observed in wt. The downregulation of the ompF gene reduces membrane permeability and consequently the entry of toxic compounds into cells (Martinez-Martinez et al. 2000).

TABLE 2.

Genes with altered expression inrpoS compared to wt colonies

Functional categories and gene products	Genes	Fold wt D7 vs. wt D1	Fold rpoS D1 vs. wt D1	Fold rpoS D7 vs. wt D7	Fold rpoS D7 vs. rpoS D1
Adaptations: atypical conditions
Heat-shock protein	htgA	−1.06	−2.55	1.27	3.04
Heat-shock protein	htpX	1.25	−1.05	1.43	1.88
Heat-shock protein	ibpB	1.26	−4.69	−1.09	5.41
Heat-shock protein	ibpA	1.31	−1.96	−1.19	2.16
Heat-shock protein	groES	−1.83	−1.37	2.06	1.54
Heat-shock protein	hslV	1.13	−1.95	1.33	2.93
H2O2-inducible genes activator	oxyR	1.14	2.38	−1.01	−2.11
Manganese superoxide dismutase	sodA	−1.15	−2.64	1.35	3.11
Regulatory protein of SoxRS regulon	soxR	−1.12	2.00	1.08	−2.09
Ribosome binding protein, H2O2 inducible	yfiA	1.54	4.74	−1.37	−4.23
RNA polymerase σ-32 subunit	rpoH	−1.01	−1.64	1.23	1.99
Unknown, H2O2 inducible	ydcH	−1.01	2.7	1.64	−1.66
DNA repair
Damage-inducible protein J	dinJ	1.43	−2.14	−1.01	3.02
DNA-damage-inducible protein G	dinG	1.68	−1.14	1.05	2.00
Endonuclease IV	nfo	1.52	−2.01	1.13	2.50
Exodeoxyribonuclease I	sbcB	−1.04	−1.90	1.15	2.11
Exodeoxyribonuclease small subunit	xseB	1.01	−1.78	1.63	2.92
Exonuclease V subunit	recB	−1.15	−1.55	1.42	1.92
Uracil-DNA glycosylase	ung	−1.03	−1.26	1.56	1.90
Replication-transcription
DNA polymerase III, χ-subunit	holC	−1.60	−2.78	1.10	1.90
Factor for inversion stimulation	fis	−1.89	−1.55	1.74	1.43
RNA polymerase, α-subunit	rpoA	−1.89	−1.58	1.82	1.52
RNA polymerase, β-subunit	rpoB	−2.01	−2.74	1.63	2.23
RNA polymerase, β′-subunit	rpoC	−1.51	−3.13	1.25	2.59
TCA cycle
2-Oxoglutarate dehydrogenase	sucA	1.19	1.03	2.31	2.65
Aconitate hydratase 2	acnB	−1.19	−2.61	−1.11	1.97
Citrate hydro-lyase 1	acnA	1.01	−2.04	1.04	2.15
Citrate synthase	gltA	−1.15	1.06	2.08	1.71
Dihydrolipoamide succinyltransferase	sucB	1.02	1.02	1.81	1.82
Fumarate hydratase class II	fumC	2.45	−1.09	1.43	3.85
Malate dehydrogenase	mdh	−1.07	−2.08	1.20	2.32
Succinate dehydrogenase	sdhC	1.67	−1.08	1.18	2.13
Succinate dehydrogenase	sdhD	1.23	−1.02	2.78	3.51
Succinate dehydrogenase flavoprotein	sdhA	1.03	−1.05	1.69	1.83
Succinate dehydrogenase iron-sulfur protein	sdhB	−1.09	1.02	1.93	1.73
Succinyl-coA synthetase α-chain	sucD	−1.02	1.96	3.40	1.70
Succinyl-coA synthetase β-chain	sucC	1.09	−1.19	1.90	2.47
Outer membrane constituents
Outer membrane porin F	ompF	−1.11	3.21	−9.32	−33.19
Phage shock proteins
Phage shock protein A	pspA	3.18	2.06	1.18	1.82
Phage shock protein B	pspB	1.62	−1.16	2.26	4.25
Phage shock protein C	pspC	−1.13	1.07	2.56	2.11
Phage shock protein D	pspD	2.17	−1.06	1.71	3.94
Ribosomal proteins
30S subunit protein S3	rpsC	−1.81	−2.02	1.64	1.83
30S subunit protein S5	rpsE	−1.70	−1.75	2.16	2.21
30S subunit protein S7	rpsG	−2.57	−3.26	1.51	1.92
30S subunit protein S8	rpsH	−1.92	−2.50	1.43	1.86
30S subunit protein S14	rpsN	−1.45	−1.97	1.92	2.61
30S subunit protein S16	rpsP	−1.26	−1.30	1.66	1.71
30S subunit protein S19	rpsS	−1.31	−1.98	1.37	2.07
50S subunit protein L1	rplA	−2.07	−3.15	1.44	2.19
50S subunit protein L2	rplB	−1.49	−2.48	1.43	2.38
50S subunit protein L3	rplC	−1.83	−2.53	1.19	1.64
50S subunit protein L6	rplF	−1.76	−1.81	2.00	2.06
50S subunit protein L10	rplJ	−1.79	−2.55	1.24	1.77
50S subunit protein L11	rplK	−1.51	−2.98	1.13	2.23
50S subunit protein L7/L12	rplL	−1.42	−2.36	1.22	2.03
50S subunit protein L15	rplO	−1.44	−2.24	1.81	2.81
50S subunit protein L16	rplP	−1.55	−2.01	1.56	2.01
50S subunit protein L19	rplS	−1.61	−1.33	1.81	1.50
50S subunit protein L23	rplW	−1.53	−1.80	1.37	1.61
50S subunit protein L24	rplX	−1.18	−1.58	1.51	2.02
50s subunit protein L25	rplY	−1.63	−2.21	1.28	1.74
50S subunit protein L29	rpmC	−1.54	−1.76	1.78	2.03
Ribosomal protein L11	prmA	−1.60	−1.88	1.78	2.08

After 1 day (D1) and 7 days (D7), total RNA was extracted from several colonies (∼10⁹ cells) and 2 μg was used for production of [α-³³P]dCTP-labeled cDNA probe. Panorama arrays (Sigma-Genosis) containing the PCR-amplified 4290 ORFs of E. coli genes printed in duplicate were hybridized with the entire labeled cDNA probe following the recommendation of the furnisher. The software XdotsReader (Cose, Dugny, France) was used to calculate the pixel density for each spot. Each hybridization experiment was repeated four times with independent RNA preparation. The eight values for each gene have been used to test the null hypothesis of equal mean expression between the two strains and two different times by calculating the t-test and the Wilcoxon test. To analyze the fold change in transcript abundance, the (wt D7/wt D1), (rpoS D1/wt D1), (rpoS D7/wt D7), and (rpoS D7/rpoS D1) mRNA ratios were calculated. When the ratio was <1, the inverse ratio was reported with a negative value. This allows a symmetric view of the data (hence the direct value shift from +1 to −1). Genes with the (rpoS D1/wt D1) or (rpoS D7/rposS D1) ratios >1.5 and with a P-value ≤0.01 for both the t-test and the Wilcoxon test were considered to have significant difference of expression and are reported in this table. Genes upregulated in the rpoS strain compared to the wt strain at D1 are underlined. For the other genes listed, an increase in mRNA level was observed in the rpoS strain between D1 and D7. (For gene annotation, see http://www.genome.wisc.edu)

Between D1 and D7, the fraction of dead cells within wt and rpoS colonies reached 60%. The difficulties faced by both strains were reflected by the increase in expression of RpoS-independent genes involved in stress resistance and DNA repair: e.g., the σ³² (RpoH)/RpoN-regulated heat-shock genes ibpA, ibpB (Kitagawa et al. 2002), the psp genes (phage shock proteins; Model et al. 1997), some SOS genes (dinJ, dinG; Courcelle et al. 2001), and the nfo gene encoding endonuclease IV (Hosfield et al. 1999). We also observed, but only in rpoS, an increase in transcripts of RpoH-regulated genes (groES and htgA; Missiakas and Raina 1997), as well as the sbcB, recB (respectively, the exodeoxyribonuclease I and the exonuclease V subunit; Seigneur et al. 1999), xseB (exonuclease subunit; Viswanathan et al. 2001), sodA (manganese superoxide dismutase; Compan and Touati 1993), and ung (uracil-N-glycosylase; D'souza and Harrison 2003) genes. Therefore, rpoS cells in aging colonies appear to face various stresses that damage DNA and proteins. However, the fact that the rpoS strain does not show an increase in mortality relative to wt at D7 suggests that it succeeds in coping with the stresses encountered in aging colonies.

Remarkably, the expression of genes coding for ribosomal proteins globally decreases between D1 and D7 in the wt strain, while it increases in rpoS colonies (Figure 1). A similar tendency was observed for genes coding for RNA (rpoB, rpoC, and rpoH) and DNA (holC) polymerase subunits (Record et al. 1996). Finally, the expression of fis (Appleman et al. 1998), a gene coding for a protein required for adaptation of cells to rapid growth conditions, increases in rpoS D7 colonies relative to rpoS D1, as well as compared to wt D7. The resulting increase in transcription, protein, and DNA synthesis in aging rpoS colonies may simply be a side effect due to the absence of RpoS regulation. However, this global change may be the result of rpoS cell turnover, as some cells grow and divide while others die.

Figure 1.—

Fold change in transcript abundance of genes coding for ribosomal proteins. The two graphs compare expression of the 60 genes encoding ribosomal proteins in the rpoS strain to wt strain: top after 1 day (D1; blue) and bottom after 7 days (D7; green). The mRNA ratios for each gene between D1 and D7 in the wt strain [(wtD7/wtD1) or if wtD1 > wtD7 (−wtD1/wtD7)] are reported on the x-axis. mRNA ratios of the same genes for D1 and D7 between the wt strain and the rpoS strain [(rpoS/wt) or if wt > rpoS (−wt/rpoS)] are reported on the y-axis. At D1, most of the values are in the bottom left quadrant, corresponding to wt D7 < wt D1 and rpoS < wt. At D7, most of the values are in the top left quadrant, corresponding to wt D7 < wtD1 and rpoS > wt.

In conclusion, our study shows that RpoS-independent stress responses may be sufficient to ensure the survival of an rpoS population in aging colonies, which may not be the case in planktonic cells where the level of stress may be higher, particularly in oxidative stress (Dukan and Nystrom 1998). Furthermore, besides providing a collective defense against antagonists (Ma and Eaton 1992), the cellular aggregation may facilitate access to resources that cannot be efficiently utilized by isolated cells in suspension. This may be the case for reutilization of nutrients from the dead cells concentrated inside the colony (Shapiro 1992). Under such conditions, rpoS cells might use the available nutrients more efficiently than wt as suggested by the significant differences in transcriptional modifications of energy-producing pathways between the two strains. A compensatory transcription in rpoS colonies is suggested by the expression of genes involved in energy salvage: transcription of the psp operon and genes belonging to the tricarboxilic acid and the glyoxylate cycles (Cronan and LaPorte 1996), which are, respectively, RpoN and RpoD dependent, was higher in rpoS than in wt D7 colonies (Table 2). Therefore, it may be that the structured environments are favoring the observed enrichment for modified or null rpoS alleles in nature by providing an environmental niche in which the handicap resulting from the lack of various RpoS-dependent functions is attenuated through compensation.