Analysis of Strains Lacking Known Osmolyte Accumulation Mechanisms Reveals Contributions of Osmolytes and Transporters to Protection against Abiotic Stress

Abstract

Osmolyte accumulation and release can protect cells from abiotic stresses. In Escherichia coli, known mechanisms mediate osmotic stress-induced accumulation of K⁺ glutamate, trehalose, or zwitterions like glycine betaine. Previous observations suggested that additional osmolyte accumulation mechanisms (OAMs) exist and their impacts may be abiotic stress specific. Derivatives of the uropathogenic strain CFT073 and the laboratory strain MG1655 lacking known OAMs were created. CFT073 grew without osmoprotectants in minimal medium with up to 0.9 M NaCl. CFT073 and its OAM-deficient derivative grew equally well in high- and low-osmolality urine pools. Urine-grown bacteria did not accumulate large amounts of known or novel osmolytes. Thus, CFT073 showed unusual osmotolerance and did not require osmolyte accumulation to grow in urine. Yeast extract and brain heart infusion stimulated growth of the OAM-deficient MG1655 derivative at high salinity. Neither known nor putative osmoprotectants did so. Glutamate and glutamine accumulated after growth with either organic mixture, and no novel osmolytes were detected. MG1655 derivatives retaining individual OAMs were created. Their abilities to mediate osmoprotection were compared at 15°C, 37°C without or with urea, and 42°C. Stress protection was not OAM specific, and variations in osmoprotectant effectiveness were similar under all conditions. Glycine betaine and dimethylsulfoniopropionate (DMSP) were the most effective. Trimethylamine-N-oxide (TMAO) was a weak osmoprotectant and a particularly effective urea protectant. The effectiveness of glycine betaine, TMAO, and proline as osmoprotectants correlated with their preferential exclusion from protein surfaces, not with their propensity to prevent protein denaturation. Thus, their effectiveness as stress protectants correlated with their ability to rehydrate the cytoplasm.

INTRODUCTION

Cellular responses to abiotic stresses include the accumulation and release of organic solutes called osmolytes (1–3). Relevant stressors include noxious chemicals, such as urea, as well as osmotic pressures and temperatures that are sub- or supraoptimal for cell survival and population growth. Osmolyte accumulation clearly promotes bacterial growth in high-osmolality medium, and some osmolytes also confer thermal and/or urea stress tolerance on prokaryotic and eukaryotic cells (1–5). Naturally occurring osmolytes include amino acids, sugars, and related compounds that are highly water soluble and uncharged or zwitterionic. They attain high cytoplasmic concentrations, and many are preferentially excluded from molecular surfaces (6). Thus, they may affect multiple cellular properties, including turgor pressure, cytoplasmic hydration, and the structures, interactions and functions of cytoplasmic macromolecules (1).

Studies of the osmotic stress responses of the enteric bacteria Salmonella enterica and Escherichia coli K-12 provided the first detailed descriptions of osmolyte accumulation mechanisms (OAMs). These mechanisms include transporters that mediate the uptake and accumulation of available osmolytes (e.g., proline and glycine betaine) and enzymes that mediate osmolyte synthesis from cytoplasmic precursors (e.g., glutamate synthesis from α-ketoglutarate, trehalose synthesis from glucose, and glycine betaine synthesis from choline) (7). Diverse prokaryotes and eukaryotes possess arrays of OAMs with overlapping specificities (8).

Organic osmoprotectants are exogenous solutes that stimulate bacterial growth in high- but not low-osmolality medium. Osmoprotectants are transported to the cytoplasm, where they accumulate unaltered or are converted to osmolytes. In E. coli, K⁺ glutamate accumulation confers osmotolerance when osmoprotectants are not available. The pathways of K⁺ and glutamate accumulation are distinct but coordinated (9–11). Synthesis of the glutamate that accumulates under osmotic stress is mediated by glutamate dehydrogenase (GdhA) or glutamate synthase (GltBD), enzymes central to nitrogen metabolism at low or high osmolality. Glutamate accumulation under osmotic stress is believed to result from inhibition of glutamate utilization (9). K⁺ glutamate accumulation is followed and suppressed by trehalose accumulation that is mediated by an osmoregulated trehalose biosynthetic system (OtsBA) (12). Bacterial growth is maximally stimulated if an osmoprotectant like glycine betaine or proline is available and its cytoplasmic accumulation is mediated by osmoregulatory transporters, suppressing the accumulation of K⁺ glutamate or trehalose (12). Osmoprotectant uptake is mediated by the choline transporter BetT and at least two osmoprotectant transporters with broad specificity (ProP and ProU [ProVWX]) (13). Choline, which is an osmoprotectant but not an osmolyte, is transported to the cytoplasm, where it is oxidized to glycine betaine by enzymes BetB and BetA.

There are indications that, as a species, E. coli possesses additional OAMs. Genome, transcriptome, and proteome analyses have revealed putative osmoprotectant transporters (14–18). For example, the core E. coli genome encodes homologues of transporters ProP (YhjE) and ProU (YehZYXW) (19). Approximately one-third of wild-type E. coli genomes also encode the betaine-specific transporter BetU (a BetT homologue) (19). Deletion of proP and proU impaired the growth of the uropathogenic E. coli strain HU734 but not that of the uropathogenic strain CFT073 in high-osmolality human urine. This was surprising, since urine contains glycine betaine (GB) at a level sufficient to provide osmoprotection (approximately 0.1 mM) (20) and the betaine transporter BetU is present in HU734 but not in CFT073 (21). Taken together, these observations led to the hypothesis that additional OAMs may contribute to the osmotic stress tolerance of E. coli.

A variety of factors could account for the redundancy of abiotic stress response mechanisms in each organism. First, OAMs may have evolved as bacteria proliferated in environments that offered different resources. For example, the osmolytes trimethylamine-N-oxide (TMAO) and dimethylsulfoniopropionate (DMSP) occur in marine environments (1, 22). TMAO counteracts the adverse effects of urea when both are present at high levels in the blood of elasmobranch fishes (1). GB is available to uropathogenic bacteria because it is consistently present in urine (20) and it serves as an osmoprotectant for mammalian urinary tract tissues (23). Second, osmolyte selection may reflect metabolic priorities. For example, known regulatory mechanisms determine whether trehalose, GB, and proline serve as osmolytes or as carbon and/or nitrogen sources for bacterial growth (7, 24). In this work, we tested a third hypothesis, that cells use particular osmolytes or osmolyte accumulation mechanisms to mitigate particular stresses. Such specificity might be expected, since osmolytes with different structures have different effects on macromolecular structures and interactions (25).

Most OAMs in E. coli were identified via studies of K-12 strains with extensive histories of genetic manipulation. Here, we report efforts to identify additional osmolytes and OAMs in two wild-type E. coli strains with known genomic sequences: CFT073 (26, 27) and MG1655 (K-12) (28, 29). The former serves as a model for studies of urinary tract infection, while the latter is widely used for fundamental research. We deleted known osmoregulatory loci from these bacteria and sought evidence for the existence and specificity of additional OAMs. In addition, we created variants of E. coli MG1655 retaining each known system in isolation. The single-system variants and the strain lacking known OAMs were used to extend our knowledge of the specificities of transporters ProP, ProU, and BetT. They also supported a comparison of the contributions of diverse solutes and transporters to growth in high-osmolality medium at various temperatures and in the absence and presence of urea. Thus, this work documented the relative abilities of diverse osmolytes and accumulation mechanisms to mitigate specific abiotic stresses.

MATERIALS AND METHODS

Bacteria and genetic manipulations.

The E. coli strains used for this study are listed in Table 1. In-frame deletions of genes encoding known osmolyte accumulation mechanisms (or their components) were introduced to E. coli MG1655 as described by Datsenko and Wanner (30). Existing gene knock-outs (kan element replacements) in Keio collection isolates (31) were obtained for this purpose from E. D. Brown (McMaster University). Deletions were verified with PCR as described by Brown and Wood (32). Oligonucleotides were purchased from Operon Technologies (Eurofins MWG Operon, Huntsville, AL).

TABLE 1.

Escherichia coli strains

Strain	Genotype	Reference(s) and/or source
CFT073	Wild type	H. Mobley; 26
WG696	CFT073 Δ(proP)218 Δ(proV-proX)567	21, 71
WG1248	CFT073 Δ(proP)218 Δ(proV-proX)567 ΔbetT856::FRT	This study
WG1250	CFT073 Δ(proP)218 Δ(proV-proX)567ΔotsA847::FRT	This study
WG1331	CFT073 Δ(proP)218 Δ(proV-proX)567ΔotsA847::FRT ΔbetT856::FRT	This study
MG1655	Wild type	E. coli Genetic Stock Center; 28
WG1228	MG1655 ΔproP837::FRT ΔproW859::FRT ΔotsA847::FRT	This study
WG1230	MG1655 ΔproW859::FRT ΔbetT856::FRT ΔotsA847::FRT	This study
WG1232	MG1655 ΔproP837::FRT ΔbetT856::FRT ΔotsA847::FRT	This study
WG1246	MG1655 ΔproP837::FRT ΔproW859::FRT ΔbetT856::FRT ΔotsA847::FRT	This study
WG1265	MG1655 ΔproP837::FRT ΔproW859::FRT ΔbetT856::FRT	This study
WG1329	MG1655 ΔproP837::FRT ΔproX870::FRT ΔbetT856::FRT ΔotsA847::FRT	This study

The genes betT and otsA were deleted from the CFT073-derived strain WG696 [Δ(proP)218 Δ(proV-proX)567] as described above to create strains WG1248 and WG1250, respectively. The allelic replacement vector pCVD442 was used as previously described (21) to replace the wild-type betT allele in strain WG1250 with ΔbetT856::FRT from strain WG1248, creating strain WG1331. (Efforts to attain this goal by applying the Datsenko-Wanner technique sequentially were not successful.)

Culture media and growth conditions.

Bacteria were cultured in LB broth (33), in modified minimal medium A (MMA), which is comprised of K₂HPO₄ (10.5 g/liter), KH₂PO₄ (4.5 g/liter), (NH₄)₂SO₄ (1.0 g/liter), MgSO₄ (0.5 mM), and d-glucose (5 g/liter), or in MOPS [3-(N-morpholino)propanesulfonic acid] medium (34) supplemented with d-glucose (11 mM, solid medium) or glycerol (0.4%, liquid medium), NH₄Cl (9.5 mM), and vitamin B₁ (0.1 g/liter). NaCl, urea, and/or other organic compounds were added to these media as indicated below. Bacteria were also cultured in high- and low-osmolality urine pools. To prepare the urine pools, first morning urine was collected in sterile vessels by 11 volunteers who had consumed no food or water for at least 14 h. This procedure was approved with REB number 06JA005 by the Research Ethics Board of the University of Guelph. The volumes, pHs, and osmolalities of the urine specimens were measured (pH range, 5.8 to 6.8; osmolality range, 0.37 mol/kg to 1.11 mol/kg), and they were mixed to yield a high-osmolality pool (pH 6.3, 1.00 mol/kg) and a low-osmolality pool (pH 6.4, 0.55 mol/kg). The urea concentrations of the urine pools were measured as described previously (35) and found to be 0.63 M (high-osmolality pool) and 0.29 M (low-osmolality pool). The pools were sterilized by filtration through 0.45-μm nitrocellulose filters (Millipore) and frozen as aliquots at −20°C. The pools were thawed and centrifuged to remove sediment, and the osmolalities were confirmed immediately before use. Osmolalities were measured with a Wescor Vapro vapor pressure osmometer.

To test for stress tolerance during bacterial growth in microtiter plates, bacteria were cultured overnight in LB at 37°C and then subcultured into appropriately supplemented MMA (2%, vol/vol), as specified below, and incubated with rotary shaking (200 rpm) at 37°C to an optical density (OD) at 600 nm of approximately 1 on the Bausch and Lomb Spectronic 88 or 0.5 on the Pharmacia LKB NovaSpec II spectrophotometer. They were then subcultured (at 1% [vol/vol] for growth at 37°C and at 10% [vol/vol] for growth at 15°C) into appropriately supplemented MMA, and 200-μl aliquots were distributed into the wells of microtiter plates containing aqueous solutions (20 μl) of organic solutes (specified below). Each growth condition was represented by 4 to 8 replicate wells per experiment and tested in at least two independent experiments. Each plate was covered with a Breathe-Easy gas-permeable sealing membrane (Sigma-Aldrich, Oakville, ON) to prevent evaporation and shaken at 350 rpm, at 36 or 37°C, in a Nephelostar microplate nephelometer (BMG Labtech, Ortenberg, Germany). Shaking was paused as forward light scattering at 635 nm was monitored at 340-s intervals during 19-h incubations. Alternatively, a Titertek Multiscan plus MKII microtiter plate reader was used to record optical densities (595 nm) over longer intervals during incubation at 15°C, with shaking at 200 rpm.

High-temperature thermotolerance was assessed with a petri plate-based assay. Bacteria were cultivated overnight in LB, harvested, and washed 3 times with an equal volume of saline (0.85% NaCl). The resulting cell suspensions were streaked radially from central filter discs to the edge of petri plates containing MOPS medium supplemented with 0.5 M NaCl (36). Various quantities of solutes, indicated below, were applied to the filter disks, and zones of growth stimulation were measured after 48 h of incubation at the temperatures indicated below.

Cultures were routinely tested by PCR (as described above) to confirm genotypes and by growth on petri plates to confirm the following phenotypes: (i) otsA⁺ bacteria grow on MOPS medium with or without 0.5 M NaCl but otsA bacteria do not grow with 0.5 M NaCl; (ii) growth of strains that are betTBA⁺, proP⁺, and/or proU⁺ on MOPS medium with 0.5 M NaCl is stimulated if GB or choline (2 μmol) is provided in a radial streak test; and (iii) strains WG1246, WG1329, and WG1331 do not grow on MOPS medium with 0.5 M NaCl even if GB or choline is supplied.

Metabolite extraction and analysis.

Bacteria were inoculated into 3 ml of LB in 16- by 150-mm test tubes and incubated at 37°C with rotary shaking at 200 rpm. After approximately 18 h, the cultures were diluted 1/50 into 25 ml of either MMA (strains MG1655 and WG1246) or MOPS medium (strains CFT073 and WG1331) and grown in 125-ml screw-cap flasks for approximately 22 h at 37°C with rotary shaking at 200 rpm. Bacteria were subcultured (1/100 dilution) into test media in 4-liter Fernbach flasks, using MMA supplemented with NaCl (0.5 M) for MG1655 and WG1246 and the high- and low-osmolality urine pools for CFT073 and WG1331. The cultures were incubated at 37°C with rotary shaking at 200 rpm as follows. Strain MG1655 was cultivated with yeast extract (YE; 1 mg/ml) to an OD of 1.7 in 6 h. One day was required to cultivate strain MG1655 with no supplement (final OD, 0.8) or with TMAO (10 mM) (final OD, 0.9) or to cultivate strains CFT073 and WG1331 in the urine pools (final ODs, 0.5 to 0.7). Strain WG1246 was cultivated with YE (1 mg/ml) to an OD of 0.8 in 2 days. ODs were measured at 600 nm with the Pharmacia LKB NovaSpec II spectrophotometer. Before harvesting, a 2-ml aliquot of each culture was set aside for strain tests and a bicinchoninic acid protein assay (37).

Cells were harvested by centrifugation at 14,000 × g for 20 min at 4°C. The pellets were resuspended in 5 ml (MG1655 and WG1246) or 3 ml (CFT073 or WG1331) of 7% (wt/vol) perchloric acid. The suspension was kept on ice for 30 to 60 min, transferred to a 50-ml falcon tube, and centrifuged for 10 min at 4°C. Each supernatant was decanted into a fresh 50-ml falcon tube, and extraction was repeated two more times, pooling the supernatants. Each pooled extract was neutralized with KOH and centrifuged to remove sediment (4,500 rpm for 10 min at 4°C). Supernatants were frozen overnight and lyophilized. Samples were resuspended in 50 mM potassium phosphate buffer (pH 7.4) plus 5% (vol/vol) deuterium oxide (D₂O). The buffer volume was adjusted to normalize the sample volumes to the quantities of cells (as indicated by cell protein) from which each extract was derived. Extracts were then centrifuged (13,000 × g for 5 min), supernatants were decanted into fresh tubes, and the pH was readjusted to 7.4 using H₃PO₄ or NaOH. Extracts were kept frozen at −40°C until analysis. ¹³C 1-dimensional and ¹H-¹³C 2-dimensional heteronuclear single-quantum correlation (HSQC) magnetic resonance (MR) spectra were obtained at the University of Guelph NMR Centre on a Bruker 600 MHz nuclear MR (NMR) spectrometer equipped with a 5-mm TXI cryoprobe. Signals were referenced to trimethylsilyl propanoic acid (TSP), and 3-mm NMR tubes were used to optimize results for high-salt samples.

For metabolomic analyses performed at the National Magnetic Resonance Facility at Madison, samples were dissolved in 0.8 ml of D₂O containing 5 mM 2-(N-morpholino)ethanesulfonic acid (MES), 300 μM TSP, and 0.5 μM NaN₃ (to prevent bacterial contamination). Each sample was titrated with deuterated base and acid to an uncalibrated glass electrode value of 7.400 and placed in a 5-mm tube. Two-dimensional sensitivity-enhanced ¹H-¹³C HSQC spectra were collected on a Varian 600 MHz spectrometer equipped with a cryogenic probe. Spectra were collected, following 16 transients to achieve steady state, as 4 averaged transients with 512 increments in the second dimension. The acquisition time was 300 ms following an initial delay of 1 s. The carbon spectral width was 100 ppm. Time-domain data were Fourier transformed with a shifted exponential sine-bell window function, phased, and chemical shift referenced to 4,4-dimethyl-4-silapentane-1-sulfonic acid (DSS) using custom NMRPipe (38) scripts written in-house. The concentrations of metabolites were estimated from peak intensities on the basis of calibration curves from standards prepared at 2, 5, and 10 mm.

RESULTS

Characteristics of a CFT073 derivative lacking known OAMs.

The enzymes mediating glutamate synthesis under osmotic stress are central to bacterial metabolism, but the other OAMs can be eliminated without affecting bacterial growth in the absence of abiotic stress. For this study, we constructed strain WG1331, a derivative of uropathogen CFT073, which cannot synthesize trehalose (ΔotsA) or take up osmoprotectants via transporters BetT, ProP, or ProU (Table 1 and Materials and Methods). This strain retains BetA (choline dehydrogenase) and BetB (GB aldehyde dehydrogenase) and can therefore oxidize choline to GB.

Previous work revealed that the deletion of proP and proU did not impair the growth of E. coli CFT073 in high-osmolality urine and neither glycine betaine nor proline betaine provided osmoprotection when the mutant strain was grown in high-salinity minimal medium (21). The OAM-deficient strain WG1331 was used to assess the roles of trehalose accumulation, choline uptake-dependent GB accumulation, and other unidentified OAMs in the osmotic stress response of CFT073. Strains CFT073 and WG1331 grew equally well in low-salinity minimal medium. High salinity (0.5 M NaCl) fully inhibited the growth of WG1331 but not that of CFT073, confirming the role of trehalose accumulation in osmotic adaptation (see Fig. S1 in the supplemental material). CFT073 grew slowly in medium supplemented with 0.9 M NaCl but not with 1.0 M NaCl (data not shown), reinforcing the earlier observation that CFT073 is highly tolerant of salinity (21) and supporting the attribution of that tolerance at least in part to trehalose accumulation. Compounds known to serve as osmolytes for E. coli (39–44) and/or other organisms failed to restore the growth of strain WG1331 in minimal medium supplemented with 0.5 M NaCl. They included γ-butrobetaine, choline, DMSP, d- or l-carnitine, ectoine, γ-aminobutyric acid (GABA), GB, hypotaurine, hydroxyectoine, myo-inositol, N,N-dimethlyglycine, pipecolate, proline, sarcosine, sorbitol, taurine, thiaproline, TMAO, and trigonelline (1 mM) (data not shown).

The impacts of the genetic lesions were further assessed by cultivating strains CFT073 and WG1331 in low- and high-osmolality urine pools. They were also cultivated in minimal medium either without supplements or supplemented with NaCl and urea to attain osmolalities (0.55 mol/kg and 1.00 mol/kg, respectively) and urea concentrations (0.55 M and 0.63 M, respectively) equivalent to those of the urine pools (see Fig. S2 in the supplemental material). No difference was observed between the growth of the strains in any of these media, and both grew more rapidly in the urine pools, which are nutrient rich, than in the defined medium (Fig. 1; see also Fig. S2).

FIG 1.

Growth of E. coli strains CFT073 and WG1331 in high-osmolality urine. Representative data are shown; a more extensive data set is shown in Fig. S2 in the supplemental material. Growth of strains CFT073 (wild-type pyelonephritis isolate, solid lines) and WG1331 (ΔotsA ΔbetT ΔproP ΔproU, dashed lines) was monitored for 18 h in a nephelometer (see Materials and Methods). The media included a high-osmolality urine pool (HOU, pH 6.3, 0.63 M urea, 1.00 mol/kg) and MOPS medium adjusted with NaCl and urea to attain the same urea content and osmolality (HOM, see Materials and Methods). Data are means with bars representing standard errors obtained from two separate experiments with nine replicate cultures per experiment (urine pools) or from four separate experiments with nine wells per experiment (MOPS medium). For clarity, an error bar is shown for only every 10th data point.

¹³C-NMR spectroscopy was used to assess the accumulation of solutes by strains CFT073 and WG1331 under these growth conditions. Previous work had revealed high levels of trehalose in extracts from strain CFT073 after growth in a high-osmolality defined medium (MOPS medium supplemented with 0.4 M NaCl, approximately 1 mol/kg) (21). Those levels were comparable to the trehalose accumulation by strain MG1655 after cultivation at high salinity, illustrated in Fig. 2A. In this work, neither trehalose nor glycine betaine was prominent in extracts from strains CFT073 or WG1331 after cultivation in either urine pool (Fig. 2B), even though GB is consistently available in human urine (20). No distinctive difference in prominent metabolites in strains CFT073 and WG1331 was observed (Fig. 2B). All spectra included a prominent peak at 165.6 ppm, attributed to urea. For both strains, the compositions of the cell extracts obtained after growth in high- and low-osmolality urine differed, including apparent increases and decreases in metabolite abundance with increasing osmolality. Peaks with chemical shifts of 33.1 and 59.2 ppm, attributed to creatinine, were much more prominent in the spectra for cell extracts obtained after bacterial growth in the high- than in the low-osmolality urine pool. However, neither creatinine nor its metabolic precursor, creatine phosphate, provided osmoprotection to E. coli CFT073 or WG1331 in radial streak assays (data not shown). Thus, neither low- nor high-osmolality urine imposed sufficient stress to elicit the accumulation of known osmolytes (trehalose and glycine betaine) and no new urinary osmoprotectants were discovered. These results contradict the view that previously undetected osmolyte accumulation mechanisms promote the growth of E. coli CFT073 in urine and reinforce the concept that CFT073 shows unusual osmotic stress tolerance in the absence of osmolyte accumulation (19, 21).

FIG 2.

Compositions of cell extracts. Bacterial cultures were prepared and metabolites were extracted for analysis by ¹³C-NMR spectroscopy as described in Materials and Methods. Peak assignments based on the spectra of standard compounds analyzed under the same conditions by ¹³C 1-dimensional and ¹H-¹³C 2-dimensional NMR spectroscopy were as follows: 1, glutamate; 2, glutamine; 3, trehalose; 4, TMAO; 5, urea; 6, creatinine. (A) Metabolites were extracted from strains WG1246 and MG1655 after cultivation in 1 liter of MMA supplemented with 0.5 M NaCl and yeast extract (1 mg/ml), TMAO (10 mM), or nothing (No Supplement). (B) Metabolites were extracted from strains CFT073 and WG1331 after cultivation in 0.5 liter of the indicated urine pool (except that WG1331 was cultivated in 0.35 liter of the low-osmolality urine pool).

Characteristics of an E. coli MG1655 derivative lacking known osmolyte accumulation mechanisms.

Putative osmolyte accumulation mechanisms identified via genomic analysis are encoded by the core E. coli genome and, hence, are present in E. coli MG1655. Our search for new osmolyte accumulation mechanisms and characterization of the roles of known mechanisms were therefore pursued further with this genetically tractable strain. For this work, we constructed MG1655 derivatives which cannot synthesize trehalose (ΔotsA) or take up osmoprotectants via transporters BetT, ProP, or ProU (strains WG1246 [ΔotsA ΔbetT ΔproP ΔproW] and WG1329 [ΔotsA ΔbetT ΔproP ΔproX]). Each strain retains BetA (choline dehydrogenase) and BetB (GB aldehyde dehydrogenase) and can therefore oxidize choline to GB.

NaCl at a concentration of 0.5 M was necessary and sufficient to fully inhibit the growth of strain WG1246, which is ΔotsA, in minimal medium. Strains MG1655 and WG1265 (which are otsA⁺) grew slowly under these conditions (Fig. 3, No Supplement). Casamino Acids (CAA) strongly stimulated the growth of strain MG1655 (which retains the osmoprotectant transporters), weakly simulated the growth of strain WG1265 (which lacks them), and did not stimulate the growth of strain WG1246 (Fig. 3, CAA). However, yeast extract and brain heart infusion (BHI) stimulated the growth of all three strains (Fig. 3, YE and BHI). In some cases, this growth showed a diauxic appearance, suggesting adaptation to the imposed growth conditions.

FIG 3.

Bacterial growth at high salinity. Growth of strains MG1655 (wild type), WG1265 (ΔbetT ΔproP ΔproU), and WG1246 (ΔotsA ΔbetT ΔproP ΔproU) in MMA supplemented with 0.5 M NaCl was monitored for 18 h in a nephelometer (see Materials and Methods). The medium was unsupplemented (No Supplement) or supplemented with Casamino Acids (CAA), brain heart infusion (BHI), or yeast extract (YE) (1 mg/ml). Data are means with bars representing standard errors for two separate experiments with six replicate cultures per experiment. For clarity, an error bar is shown for only every 10th data point. Brain heart infusion and yeast extract (1 mg/ml) stimulated the growth of strains WG1228 (BetT⁺), WG1230 (ProP⁺), and WG1232 (ProU⁺), but Casamino Acids (1 mg/ml) did not (data not shown).

Analysis of cell extracts by nuclear magnetic resonance (NMR) spectroscopy revealed that glutamate and glutamine were the most prominent osmolytes in the cytoplasm of strain WG1246 after growth in high-salinity medium with YE (Fig. 2A) or BHI (not shown). Full metabolomic analysis of the NMR spectra (detailed in Materials and Methods) also revealed GABA accumulation (data not shown). GABA was not an osmoprotectant for strain WG1246 (Table 2), so the GABA present after cultivation of these bacteria in high-salinity medium with YE was likely metabolic in origin. By comparison, TMAO and/or trehalose was strikingly prominent when wild-type strain MG1655 was cultivated in a high-salinity medium with TMAO, no supplement, or YE (Fig. 2A).

TABLE 2.

Contributions of solutes to abiotic stress tolerance^a

Solute	Acronym	Contribution to stress tolerance
		Mediated by ProP (E. coli WG1230)					Mediated by ProU (E. coli WG1232)
		Osmo stressb	Osmotic stress atc:			Urea stressd	Osmo stress	Osmotic stress at:			Urea stress
			42°C	37°C	15°C			42°C	37°C	15°C
γ-Aminobutyric acid	GABA	+	−	−	+	NT	+	−	−	+	NT
γ-Butyrobetaine		+	NT	+	+	+	+	NT	+	+	+
Choline		+	−	+	+	NT	+	+	+	+	NT
d-Carnitine		+	+	+	+	NT	+	+	+	+	NT
l-Carnitine		+	+	+	+	+	+	+	+	+	+
N,N-Dimethylglycine		+	+	+	+	NT	+	+	+	+	NT
Dimethylsulfoniopropionate	DMSP	+	+	+	+	+	+	+	+	+	+
Ectoine		+	−	+	+	+	+	+	+	−	+
Glycine betaine (N-trimethylglycine)	GB	+	+	+	+	+	+	+	+	+	+
Hydroxyectoine		+	+	+	+	NT	+	+	+	+	NT
Hypotaurine		NT	−	−	−	NT	NT	−	−	−	NT
Myoinositol		NT	−	−	−	NT	NT	−	−	−	NT
Pipecolate		+	−	−	−	NT	+	−	−	+	NT
Proline		+	+	+	+	+	+	+	+	+	+
Sarcosine (N-methylglycine)		−	−	−	−	NT	−	−	−	−	NT
Sorbitol		NT	−	−	−	NT	NT	−	−	−	NT
Taurine		−	−	−	−	NT	−	−	−	−	NT
Thiaproline		−	−	−	−	NT	−	−	−	−	NT
Trimethylamine-N-oxide	TMAO	+	−	−	−	+	+	−	+	−	+
Trigonelline		−	−	−	−	NT	−	−	−	−	NT

^aStress tolerance was assessed as described in Materials and Methods, and solutes were denoted as contributing to stress tolerance (+), not contributing to stress tolerance (−), or not tested (NT).

^bSolutes were classed as osmoprotectants for growth in liquid medium if they stimulated bacterial growth at the optimal temperature (36°C or 37°C) in minimal medium supplemented with 0.5 M NaCl. Representative growth curves are shown in Fig. 4.

^cOsmotic stress tolerance testing was replicated at the optimal (37°C) temperature and extended to a supraoptimal temperature (42°C) with a radial streak test performed on solid minimal medium supplemented with 0.5 M NaCl. Osmoprotection was indicated by a significant increase in the zone of growth stimulation. Representative data are shown in Fig. 6. This test was further extended to growth in liquid minimal medium supplemented with 0.4 M NaCl at a suboptimal temperature (15°C). Osmoprotection was indicated by attainment of a higher optical density in the presence of an osmoprotectant than in its absence. Representative growth curves are shown in Fig. 7.

^dSolutes were classed as urea protectants for growth in liquid medium if they stimulated bacterial growth at the optimal temperature (36 or 37°C) in minimal medium supplemented with 0.3 M NaCl and 0.6 M urea, as shown in Fig. 4.

In some cases, multiple periplasmic binding proteins deliver different substrates to common membrane-integral and cytoplasmic subunits of ABC transporters (e.g., see reference 45). Strain WG1246 lacks the integral membrane subunit ProW of the ProU system, whereas WG1329 lacks the periplasmic binding protein subunit ProX. Each was cultivated in high-salinity minimal medium without or with each of the solutes listed in Table 2 (selected because they accumulate in E. coli and/or other organisms in response to osmotic stress). None of these compounds stimulated the growth of either strain. Thus, ProVW appears to be ProX specific.

These experiments provided no evidence that additional transporters contributed to the accumulation of osmolytes available in the tested media under these conditions. The organic mixtures may facilitate the accumulation of glutamate and glutamine by supplementing nitrogen supplies. For example, Welsh et al. showed that trehalose accumulation was favored over K⁺ glutamate accumulation when E. coli was cultivated at high osmolality under nitrogen limitation, and the converse was true for cultivation under carbon limitation (46). The organic mixtures may also complement auxotrophies that become apparent under osmotic stress. For example, supplementation of minimal medium with arginine, isoleucine, and valine was required to attain strong growth of an E. coli K-12 derivative at very high salinity (0.8 M NaCl) (47), even though these supplements were not required for growth at low salinity. Glutamine and γ-glutamyl peptides were previously detected in osmotically stressed, trehalose-deficient E. coli, but their levels were deemed too low to be osmoprotective (48). The results of this study suggest that glutamine may contribute to osmotolerance under the conditions of our experiments.

Contributions of osmolytes and transporters to osmotolerance.

Strains lacking known OAMs other than the ability to synthesize trehalose (WG1265) or to take up osmoprotectants via transporter ProP (WG1230), ProU (WG1232), or BetT (WG1228) were constructed so that the ability of each mechanism to mitigate various abiotic stresses could be tested in isolation (Table 1). First, we compared the abilities of transporters ProP, ProU, and BetT to mediate osmoprotection by diverse solutes, previously identified as osmoprotectants and/or osmolytes for E. coli (39–44) and/or other organisms. This study reinforced the strict substrate specificity of transporter BetT, which only mediated osmoprotection by choline (data not shown). It also extended our knowledge of the broad and similar substrate specificities of transporters ProP and ProU (Table 2, Osmo stress, and Fig. 4). Solutes were classed as osmoprotectants if they stimulated bacterial growth in minimal medium supplemented with 0.5 M NaCl. In a previous study, choline was found to very weakly inhibit proline uptake via transporter ProP, but no choline uptake via ProP was detected (49). When bacteria retaining the ability to oxidize choline to GB (BetB⁺ BetA⁺) were tested, transporters ProP and ProU clearly mediated osmoprotection by choline (Table 2, Osmo stress). In addition, our data revealed that these transporters mediate osmoprotection by TMAO and GABA (Table 2, Osmo stress, and Fig. 4).

FIG 4.

Contributions of transporters ProP and ProU to osmoprotection and urea protection. Growth of E. coli strains WG1230 (ΔotsA ΔbetT proP⁺ ΔproU) and WG1232 (ΔotsA ΔbetT ΔproP proU⁺) in MMA supplemented with 0.5 M NaCl and the indicated osmoprotectant (top, Osmoprotection) or in MMA supplemented with 0.3 M NaCl, 0.6 M urea, and the indicated osmoprotectant (bottom, Urea Tolerance) was monitored for 18 h in a nephelometer (see Materials and Methods). Osmoprotectants were provided at 1 mM except for TMAO, which was provided at 100 mM. Shown are the means with bars representing standard errors for data obtained from at least 2 separate experiments with 3 replicate cultures per experiment (Osmoprotection) or from 3 separate experiments with 4 replicate cultures per experiment (Urea Tolerance). For clarity, an error bar is shown for only every 10th data point. Dotted midlines are included to facilitate comparisons among plots. The data in the left panels show that strain WG1246 (ΔotsA ΔbetT ΔproP ΔproU) failed to grow in the presence of these compounds (data are global averages ± standard errors for determinations of the growth of strain WG1246 in the presence of each of the listed osmoprotectants).

This work provides the first evidence that TMAO is an osmoprotectant for E. coli and a substrate for transporters ProP and ProU. TMAO stimulated growth only if it was provided at a high exogenous concentration (Fig. 5; the results show maximal stimulation with 100 mM TMAO, whereas other osmolytes were effective at 1 mM), and it did not stimulate growth as strongly as some other osmoprotectants that are less powerful protein stabilizers (e.g., glycine betaine) (50, 51). The stimulatory effects of TMAO could have been limited by its contribution to the osmolality of the medium. However, this was not the case, since TMAO was maximally stimulatory when added at 100 mM, whereas GB, which is effective at exogenous concentrations as low as 10 μM (52), was maximally (and much more strongly) stimulatory at 1, 10, or 100 mM concentrations (Fig. 5). TMAO can decompose to trimethylamine and alkalinize solutions. To determine whether TMAO decomposition affected our osmoprotection assays, the compositions and pHs of our TMAO solutions were examined. No evidence for the presence of trimethylamine was obtained when ¹³C NMR spectroscopy was used to examine a 1 M, sterile aqueous solution of TMAO maintained at 4°C for 17 months, and the pH of that solution was 7.4. The effectiveness of TMAO as an osmolyte could be influenced by its metabolism, so cell extracts were analyzed to determine whether TMAO accumulated unaltered in the cytoplasm of bacteria growing in high-salinity medium. TMAO was present at a high level, as expected (Fig. 2A), and trimethylamine was detected at a very low level relative to the amount of intact TMAO in an extract from bacteria cultured with 100 mM TMAO (data not shown). Thus, the effectiveness of TMAO as an osmoprotectant may be limited by both a low affinity for the osmoregulatory transporters and the nature of its interactions with cellular constituents (discussed further below), but it is probably not limited by decomposition or metabolism. The relative osmoprotective effects of the osmolytes and transporters are summarized in Fig. 4 and discussed further below (see Discussion).

FIG 5.

Glycine betaine is a more powerful osmoprotectant than TMAO. Growth of E. coli strains WG1230 (ΔotsA ΔbetT proP⁺ ΔproU) and WG1232 (ΔotsA ΔbetT ΔproP proU⁺) in MMA supplemented with 0.5 M NaCl and no osmoprotectant or glycine betaine or TMAO was monitored for an 18-h period in a nephelometer (see Materials and Methods). Glycine betaine and TMAO were provided at 1 mM, 10 mM, or 100 mM. Shown are the means with bars representing standard errors for 2 experiments with at least 6 replicate cultures per experiment. For clarity, an error bar is shown for only every 10th data point.

Temperature dependence of the osmotolerance conferred by ProP and ProU.

E. coli cells from stationary-phase cultures survive prolonged exposure to high temperature (55°C), whereas cells from exponential-phase cultures do not (53). This stationary-phase thermotolerance is attributed to trehalose accumulation ensuing from RpoS-dependent transcription of otsAB. Thermotolerance can also be acquired via osmotic induction of otsAB (21, 53).

We examined the impact of high temperature on the abilities of trehalose synthesis and osmoprotectant transport via ProP or ProU to confer osmotolerance. A radial streak test was used to examine bacterial growth in NaCl-supplemented minimal medium (0.5 M NaCl) at 37°C (which is optimal) and 42°C (which is supraoptimal) (Fig. 6). Strain WG1246 (ΔotsA ΔbetT ΔproP ΔproW) did not grow at 37°C or 42°C in the absence of osmoprotectants (there was no growth on the test medium and, hence, no zone of stimulation). In contrast, the wild-type strain MG1655 grew equally well at both temperatures (bacteria grew throughout the streak, yielding maximal zones of growth stimulation [>3 cm]) (Fig. 6). Thus, cytoplasmic trehalose powerfully stimulated growth at optimal and high temperature. None of the solutes listed in Table 2 stimulated the growth of strain WG1246 at 42°C (data not shown). Tests performed with strains WG1230 (ΔotsA ΔbetT proP⁺ ΔproW) and WG1232 (ΔotsA ΔbetT ΔproP proU⁺) showed that osmoprotectants other than DMSP and GB were less effective at 42°C than at 37°C (Fig. 6). The relative effects of the osmolytes and transporters are discussed further below (see Discussion).

FIG 6.

Contributions of transporters ProP and ProU to high-temperature thermotolerance. Radial streak tests were performed as described in Materials and Methods. Petri plates containing MOPS medium supplemented with 0.5 M NaCl were inoculated radially with E. coli strains MG1655 (wild type [WT]), WG1230 (ΔotsA ΔbetT proP⁺ ΔproU), and WG1232 (ΔotsA ΔbetT ΔproP proU⁺). Osmolytes (20 μmol for TMAO, 2 μmol for the other compounds) were applied to the central filter disk. Mean zones of growth stimulation (cm) are shown (± standard error) for 3 replicate tests. Osmolytes γ-aminobutyrate, hypotaurine, myoinositol, sarcosine, sorbitol, taurine, thiaproline, and trigonelline were also tested but did not stimulate growth.

In the course of this work, osmoprotection at optimal growth temperature was assessed during growth in liquid medium (Fig. 4 and Table 2) and on solid medium (Fig. 6 and Table 2). Qualitatively similar patterns of osmoprotection were observed with these two tests except that GABA conferred weak osmoprotection when the test was performed with liquid but not with solid medium (Table 2). The relative effectiveness of osmolytes did vary between growth contexts for each transporter. This may reflect differences in the relative abundance of the two transporters in bacteria cultivated in liquid versus on solid medium.

Osmoprotectant transport can also confer cold tolerance. For example, GB uptake stimulates the growth of Listeria monocytogenes (54) and Corynebacterium glutamicum (55) at low temperature. Each organism possesses at least one transporter that cryoactivates in the absence of osmotic stress (55, 56). The impacts of osmoprotectants on the growth of E. coli at 15°C, which is suboptimal, were examined with a microtiter plate-based assay (see Materials and Methods). No osmoprotectant stimulated the growth of strain WG1246 (ΔotsA ΔbetT ΔproP ΔproW) at 15°C, and tests performed with strains WG1230 (ΔotsA ΔbetT proP⁺ ΔproW) and WG1232 (ΔotsA ΔbetT ΔproP proU⁺) again showed variations in protective activity among osmolytes (Fig. 7). The relative effects of the osmolytes and transporters are discussed further below (see Discussion).

FIG 7.

Contributions of transporters ProP and ProU to low-temperature thermotolerance. E. coli strains WG1230 (ΔotsA ΔbetT proP⁺ ΔproU), WG1232 (ΔotsA ΔbetT ΔproP proU⁺), and WG1246 (ΔotsA ΔbetT ΔproP ΔproU) were cultured at 15°C in microtiter plates containing MMA supplemented with 0.4 M NaCl as described in Materials and Methods. For strains WG1230 and WG1232, data are the means with bars representing standard errors for one representative experiment (of two) and 4 replicate cultures per experiment. For strain WG1246, data are the means with bars representing standard errors for one experiment with 8 replicate cultures per experiment. Nil, no supplement.

Contributions of osmolytes and transporters to urea tolerance.

Urea is a physiologically relevant stressor since high urea levels (e.g., 0.5 M) distinguish the urinary tract from other human tissues commonly infected by E. coli. To establish a system for comparison of the urea-protective activities of diverse osmolytes, we further defined the effects of NaCl, urea, and osmoprotectants on the growth of E. coli strains MG1655 (wild type) and WG1246 (ΔotsA ΔbetT ΔproP ΔproW). Osmoprotectants are known to be urea protective only if E. coli is cultivated under conditions of osmotic stress (57). In accordance with that observation, proline, GB, and ectoine failed to stimulate the growth of strain MG1655 in minimal medium supplemented with only 0.19 M NaCl and urea at concentrations up to 0.7 M (data not shown). However, osmolytes did stimulate the growth of this strain in medium supplemented with 0.3 M NaCl and urea at concentrations up to 1 M (see below). Thus, urea does not activate the osmotic stress response and it is not possible to fully separate the contributions of osmolytes to salt and urea tolerance. Strains MG1655 and WG1246 grew when 0.8 M urea but not 0.9 M urea was present in minimal medium containing 0.3 M NaCl (see Fig. S3 in the supplemental material).

The contributions of diverse osmolytes and of transporters ProP and ProU to urea protection were assessed by cultivating strains WG1230 (ΔotsA ΔbetT proP⁺ ΔproW) and WG1232 (ΔotsA ΔbetT ΔproP proU⁺) in minimal medium supplemented with 0.3 M NaCl and urea at a series of concentrations (0.1 M to 1.1 M in increments of 0.1 M) (data not shown). TMAO (100 mM) elevated the maximum urea concentration at which growth occurred from 0.8 M to 1 M (strain WG1230) or 1.1 M (strain WG1232), and TMAO was not urea protective when added at only 1 mM. No other solute altered the maximum urea concentration for growth. Solutes were further classed as urea protectants if they stimulated growth at a urea concentration of 0.6 M (Table 2 and Fig. 4). The relative effectiveness of the solutes and transporters in urea stress tolerance are discussed further below.

DISCUSSION

This study was designed to test the hypothesis that OAMs other than trehalose synthesis or osmoprotectant uptake via BetT, BetU, ProP, or ProU contribute to the abiotic stress tolerance of E. coli. CFT073 was found to have high tolerance for salinity (growing in minimal medium supplemented with 0.9 M NaCl). Deletion of the genes encoding all known OAMs failed to impair the growth of CFT073 in low- or high-osmolality urine or in defined medium with the same urea concentrations and osmolalities (Fig. 1). No trehalose or GB was evident in the cytoplasm of pyelonephritis isolate CFT073 after cultivation in low- or high-osmolality human urine (Fig. 2B), even though trehalose was found after growth of this strain in a urea-free defined medium with the same high osmolality (21) and GB is available in human urine (20). Urea is a prominent urine constituent, present at concentrations of 0.63 M and 0.29 M, respectively, in the high- and low-osmolality urine pools used for this study. Urea impairs cellular processes by interacting with cellular constituents, but it is membrane permeant and, hence, does not cause cellular dehydration or activate osmoregulatory responses (21; this study). The osmolalities of the urine pools used for this study would have been less than 0.4 mol/kg (high-osmolality urine pool) and less than 0.3 mol/kg (low-osmolality urine pool), respectively, if urea had not been present. Clearly, these conditions did not trigger osmolyte accumulation by CFT073.

These results were unexpected, since deletion of proP impaired the growth of pyelonephritis isolate HU734 in high-osmolality urine and deletion of proU exacerbated that effect (21). The elimination of these osmoprotectant transporters may be particularly deleterious for HU734 because it is intrinsically less osmotolerant than CFT073. Previous work showed that HU734 does not accumulate trehalose in response to osmotic stress and it is less salinity tolerant than CFT073 during growth in osmoprotectant-free MOPS medium (21). This occurred in part because HU734 harbors a defective variant of the stress response sigma factor RpoS, which is required for the expression of otsAB (21). However, this explanation is incomplete, because the growth of strain CFT073 in high-osmolality urine was not impaired by deletion of rpoS, proP, and proU (21). Kunin et al. showed that the salinity tolerance of clinical E. coli isolates varies widely, up to 0.7 M NaCl, regardless of tissue origin (58). The ability to grow in up to 0.9 M NaCl in an osmoprotectant-free medium analogous to that employed by Kunin et al. again shows CFT073 to be a particularly salinity-tolerant isolate.

Since the difference in salinity tolerance between strains CFT073 and HU734 is not based on osmolyte accumulation, it may reside in differential sensitivities of key cellular processes to the consequences of osmotic stress. For example, K⁺ glutamate accumulation impairs transcription by affecting RNA polymerase-DNA interactions in E. coli, and those effects are overcome when GB replaces K⁺ glutamate (59, 60). Perhaps modifications to key enzymes like RNA polymerase increase the osmotolerance of some wild-type E. coli strains. That sort of intrinsic variation has not previously been characterized.

The expectation that additional OAMs would be found in E. coli was reinforced by the identification of ProP and ProU homologues among the putative transporters of E. coli (19) and by evidence for osmotic induction of the encoding genes (14, 61). This study revealed no new osmolyte accumulation mechanisms for wild-type E. coli strain CFT073 or MG1655, though it did suggest that both glutamine and glutamate may contribute to osmotolerance for trehalose-deficient bacteria growing in nitrogen-rich medium (Fig. 2A). Putative osmoregulatory transporters, such as the ProP homologue YhjE and the ProU homologue YehZYXW, may contribute to cellular processes other than abiotic stress tolerance. For example, a ProU homologue in Pseudomonas aeruginosa (CbcXWV) plays a metabolic rather than an osmoregulatory role (62). Furthermore, a PutP homologue (denoted OpuE) plays an osmoregulatory role in Bacillus subtilis (63), but PutP is not osmoregulatory in E. coli. Indeed, other roles were proposed for the Yeh system and its Salmonella homologue, OsmU (otherwise known as OpuC) (64, 65).

This study revealed for the first time that transporters ProP and ProU mediate osmoprotection by TMAO (Table 2 and Fig. 4). This phenomenon is unlikely to be physiologically significant because it was observed only at very high exogenous TMAO concentrations (at or above 10 mM) (Fig. 5). However, osmoprotection by TMAO did correlate with its accumulation in E. coli (Fig. 2B), and the response to TMAO is interesting because TMAO-protein interactions are well characterized (see further discussion below).

This study was also designed to determine whether particular osmolytes or OAMs are particularly effective in mitigating specific abiotic stresses. Such specialized roles might help to explain the redundancy of OAMs in prokaryotic and eukaryotic cells. The abilities of solutes to mitigate osmotic stress were tested at temperatures that were suboptimal (15°C), optimal (36°C or 37°C), or supraoptimal (42°C) for growth and at 37°C in the presence of urea (Fig. 4 to 7). These experiments employed derivatives of E. coli MG1655 that retained only a single osmoprotectant transporter, either ProP or ProU.

The relative effectiveness of each osmolyte in mitigating each abiotic stress was judged qualitatively by comparing the growth curves (liquid cultures) (Fig. 4 and 7) and zones of growth stimulation (radial streak tests) (Fig. 6). Overall, GB and DMSP were the most effective abiotic stress protectants, while TMAO, proline, and ectoine were less effective. This distinction was consistent at each of the temperatures tested, in the absence and presence of urea, and for each transporter (with the notable exception that ProU did not mediate effective urea stress protection by DMSP). Thus, this study did not support the hypothesis that particular osmolytes or osmolyte accumulation mechanisms are particularly effective in mitigating specific abiotic stresses.

In some cases, the relative potencies of various solutes as stress protectants differed for bacteria retaining only ProP or ProU (Fig. 4 to 7). However, these data reinforce the conclusion that these transporters have similar, very broad substrate specificities. The observed variations may in part reflect differences in substrate affinities. For example, the K_m for proline transport via ProP is osmolality dependent, varying from approximately 6 μM to 155 μM in the osmolality range of 0.2 to 0.4 mol/kg, at 25°C (66). At high osmolality, ProP has similar, high affinities for proline, GB, and ectoine (44, 49). In contrast, ProU transports GB with very high affinity (a K_m of 1 μM at high osmolality and 30°C) (67), and it transports ectoine and proline with low affinity (44). The kinetics of DMSP and TMAO transport have not been characterized, but the affinities of ProP and ProU for TMAO are probably low (Fig. 5). The differing relative potencies of solutes as stress protectants for bacteria retaining only one transporter may also reflect direct effects of temperature and urea on transporter structure and function. A particularly low binding affinity (Fig. 5) and inhibitory effects of urea on the periplasmic protein ProX (Fig. 4) may account for the relatively poor performance of TMAO as a urea-protective solute for bacteria expressing only ProU.

The relative protective effects of osmolytes such as glycine betaine, and proline were previously examined in bacteria that could also accumulate trehalose (e.g., see references 60 and 68). This study evaluated the relative effectiveness of diverse osmolytes in trehalose-deficient bacteria. It is particularly informative to compare the impacts of GB, TMAO, and proline on bacterial growth because the interactions of those solutes with macromolecules have been extensively analyzed in vitro. The preferential exclusion of these solutes from the surface of folded bovine serum albumin, a representative globular protein, is ranked in the order GB > proline > TMAO. Trehalose and K⁺ glutamate (which accumulate in bacteria not offered an exogenous solute) are less strongly excluded than TMAO (69, 70). The impacts of GB, proline, and TMAO on protein stability (the transition between folded and denatured states) have also been extensively documented (50, 51), usually by measuring each solute's ability to protect proteins from urea-induced or thermal denaturation. TMAO is generally a more effective protein stabilizer than GB, and proline has little effect. Thus, similar rankings are observed for osmoprotection and exclusion of solutes from the surface of a folded, globular protein, whereas different rankings are observed for osmoprotection and protein stabilization. This difference may reflect the different compositions of the molecular surfaces interacting with the solvent (51). Data suggest that exclusion from the outside surfaces of folded proteins, as would occur in the cytoplasm of a living cell, increases the effectiveness with which solutes rehydrate the cytoplasm (60, 68). In contrast, protein stabilization occurs when solutes are excluded from surfaces that become buried during protein folding. These principles can also be expected to govern the effects of other naturally occurring osmolytes (e.g., DMSP).