Synthetic Effect between Envelope Stress and Lack of Outer Membrane Vesicle Production in Escherichia coli

Abstract

Outer membrane vesicles (OMVs) are composed of outer membrane and periplasmic components and are ubiquitously secreted by Gram-negative bacteria. OMVs can disseminate virulence factors for pathogenic bacteria as well as serve as an envelope stress response. From a transposon mutant screen for OMV phenotypes, it was discovered that an nlpA mutant of Escherichia coli produces fewer OMVs than the wild type, whereas a degP mutant produces higher levels of OMVs. NlpA is an inner-membrane-anchored lipoprotein that has a minor role in methionine import. DegP is a periplasmic chaperone/protease for misfolded envelope proteins that is critical when cells are heat shocked. To reveal how these proteins contribute to OMV production, the mutations were combined and the double mutant analyzed. The ΔnlpA ΔdegP strain displayed a high-temperature growth defect that corresponded to the production of fewer OMVs than produced by the ΔdegP strain. This phenotype also pertained to other undervesiculation mutations in a ΔdegP background. The hypovesiculation phenotype of ΔnlpA in the wild-type strain as well as in the degP deletion strain was found to be a stationary-phase phenomenon. The periplasm of the ΔnlpA ΔdegP strain was determined to contain significantly more protein in stationary phase than the wild type. Additionally, misfolded DegP substrate outer membrane porins were detected in ΔdegP mutant-derived OMVs. These data suggest that an accumulation of envelope proteins resulting from decreased vesiculation was toxic and contributed to the growth defect. We conclude that OMV production contributes to relieve the envelope of accumulated toxic proteins and that NlpA plays an important role in the production of vesicles in stationary phase.

INTRODUCTION

Outer membrane vesicles (OMVs) are ubiquitously produced by all Gram-negative bacteria studied to date (1–8). They are defined as outer membrane (OM) buds entrapping periplasmic content in the lumen (9). As visualized by electron and atomic force microscopy, OMVs are spherical structures with a diameter ranging from 50 to 250 nm (6, 10). OMVs are a means by which Gram-negative bacteria interact with their environment. They allow for the release of virulence factors by pathogenic Gram-negative bacteria, the transfer of genetic information, and the secretion of degradative factors by bacterial species protecting their ecological niches (11–13). Additionally, our lab has proposed that vesicles mediate envelope stress relief in Gram-negative bacteria (14). Based on the expenditure of energy required for the synthesis of OMV proteins and lipids, it is anticipated that OMV secretion plays a critical role for the bacterial cell and more specifically in envelope stress; however, this has not yet been characterized in molecular detail.

A previous study from our lab identified random transposon mutants of Escherichia coli that exhibited altered vesiculation phenotypes (15). One of the hypovesiculating mutants carried a transposon insertion in nlpA. NlpA is an inner membrane (IM)-anchored lipoprotein (16). The screen also revealed that a transposon insertion immediately following the start of degP resulted in hypervesiculation, a phenotype later confirmed with the full deletion (14, 15). DegP is a periplasmic protease/chaperone that manages envelope stress caused by un- and misfolded proteins (17–19). Transcription of degP is regulated by the σ^E and the Cpx envelope stress pathways (18). DegP is active predominantly as a protease at higher temperatures, whereas its chaperone activity dominates at lower temperatures (17, 20). Severe stress conditions such as heat shock and growth at higher temperatures result in high levels of un- and misfolded proteins in the cell. The deletion of degP becomes lethal at 42°C, presumably because envelope protease activity becomes critical to relieve the envelope from undesired proteinaceous waste products (20).

The hypervesiculation phenotype of ΔdegP can be significantly reduced by growing the cells at a lower temperature (∼30°C), most likely due to the increased time allowed for protein folding, which reduces the amount of proteinaceous waste that accumulates in the envelope (14). This observation led to the hypothesis that strains lacking degP use vesiculation as a survival mechanism, eliminating sublethal amounts of un- and misfolded proteins from the cell envelope. It was not known whether endogenous protein actually accumulated in the periplasm of degP mutants and whether the ability of this strain to hypervesiculate was critical for viability.

In this study, we investigated whether OMV production is critical to relieve envelope stress by combining ΔnlpA and ΔdegP, which are under- and hypervesiculating mutations, respectively. This strain was impaired in OMV production, demonstrated a significant increase in periplasmic protein amount, and exhibited a severe growth defect that began in late log phase. Other undervesiculation mutations cause comparable defects in a degP deletion background. We identified that the role of NlpA in vesicle production is growth phase dependent, occurring in late log/early stationary phase. Furthermore, misfolded outer membrane porins (OMPs), which typically are substrates for DegP, accumulate in ΔdegP mutant-derived OMVs. Together, these data suggest that OMV production is critical for bacteria at a key growth transition and provides an outlet for toxic accumulation of endogenous periplasmic protein during stress.

MATERIALS AND METHODS

Growth conditions and reagents.

The strains and plasmids used in this work are summarized in Table 1. Bacteria were grown in liquid culture in Luria-Bertani (LB) broth (EM Science), unless indicated otherwise, or on plates of solid LB agar supplemented with 50 mg/ml kanamycin or 100 mg/ml ampicillin (Sigma). Isopropyl-β-d-1-thiogalactopyranoside (IPTG) (VWR) was added to induce protein expression if indicated. The single-gene mutants originate from the Keio Collection (21). To create mutants with multiple deletions, the kanamycin resistance marker was removed from the single mutant (22). The additional mutation was then added by transduction of the marked gene deletion using P1 phage (23) from the donor single Keio mutant strain into the unmarked Keio recipient mutant strain. The ΔnlpA ΔdegP strain was constructed at 30°C to minimize the stress on the cells and avoid suppressor mutations. The mutants constructed for this work were either sequenced with primers upstream and downstream of the deleted gene or PCR amplified with primers upstream/downstream of the deleted gene and the kanamycin cassette to confirm the genotypes.

Table 1.

Strains and plasmids

Strain(s) or plasmid	Genotype(s)	Source or reference
Strains
BW25113 (WT of Keio collection)	rrnB3 ΔlacZ4787 hsdR514 Δ(araBAD)567Δ(rhaBAD)568 rph-1	21
Keio collection single mutants	nlpA::kan degP::kan dsbA::kan bolA::kan ompR::kan ompA::kan	21
MK1251	BW25113 nlpA degP::kan	This work
MK1257	BW25113/pCS19	This work
MK1259	BW25113 degP::kan/pCS19	This work
MK1260	BW25113 nlpA degP::kan/pCS19	This work
MK1261	BW25113 nlpA degP::kan/pCS20 (pDegP)	This work
MK1262	BW25113 nlpA degP::kan/pCS21 (pmDegP)	This work
MK1263	BW25113/pTrc99A	This work
MK1264	BW25113 nlpA degP::kan/pTrc99A	This work
MK1265	BW25113 nlpA degP::kan/pNlpA	This work
MK1266	BW25113 nlpA metQ metE::kan/pTrc99A	This work
MK1267	BW25113 nlpA metQ metE::kan/pNlpA	This work
MK1268	BW25113 nlpA metQ metE::kan/pmNlpA	This work
MK1269	BW25113 nlpA degP::kan/pmNlpA	This work
MK1270	BW25113 degP::kan/pCS20 (pDegP)	This work
MK1271	BW25113 degP::kan/pCS21 (pmDegP)	This work
MK1275	BW25113 dsbA degP::kan	This work
MK1276	BW25113 bolA degP::kan	This work
MK1278	BW25113/pNlpA	This work
MK1279	BW25113 degP::kan/pNlpA	This work
Plasmids
pCS19	pQE60-derived vector with lacIq, Ampr, IPTG inducible	17
pCS20 (pDegP)	degP in pCS19, Ampr, IPTG inducible	17
pCS21 (pmDegP)	degP[S210A] in pCS19, Ampr, IPTG inducible	17
pTrc99A	trc promoter, Ampr, pBR322 ori
pNlpA	nlpA in pTrc99a, Ampr, IPTG inducible	This work
pmNlpA	nlpA[N142A, R145A] in pTrc99a, Ampr, IPTG inducible	This work

Construction of expression vectors.

In order to express nlpA under the control of the trc promoter, pNlpA was constructed using the pTrc99A vector. nlpA was amplified from genomic DNA using the primers 5′- GATATCTAGACGCGCTTGCGTGGTCAG-3′, containing the XbaI restriction site, and 5′-GATTAAGCTTTTACCAGCCAGGCACCGC-3′, containing the HindIII restriction site. The nlpA-containing XbaI-HindIII fragment was then ligated with the XbaI-HindIII-cleaved pTrc99A fragment. The restriction enzymes were purchased from New England BioLabs. A pNlpA derivative, pmNlpA, was created to express mNlpA using the inducible trc promoter. The point mutations were introduced by site-directed mutagenesis (24) of pNlpA using the primers 5′-CAACGACCCGACCGCGCTTGGTGCGGCACTTTTACTG-3′ and 5′-CAGTAAAAGTGCCGCACCAAGCGCGGTCGGGTCGTTG-3′. The plasmid insertions were confirmed by sequencing using primers upstream and downstream of the cloned gene.

Growth assays.

Overnight 30°C, 37°C, or 40°C bacterial broth cultures were used to inoculate fresh medium (5 ml) with an amount sufficient to generate an initial optical density at 600 nm (OD₆₀₀) of ∼0.02 to 0.03, and the cultures were grown with shaking at 30°C, 37°C, or 40°C, as indicated. Growth was monitored by measuring the OD₆₀₀.

To assess methionine transport activity, cultures (5 ml) of Vogel Brunner medium E (minimal medium) containing 1% glucose and 20 μg/ml d-methionine were inoculated from overnight 37°C LB cultures, grown with shaking at 37°C or 40°C as indicated, and growth was monitored by OD₆₀₀.

Membrane integrity assay.

A ToxiLight bioassay kit (Lonza) was used to assess membrane integrity. Cultures (5 ml) were grown overnight (40°C, ∼16 h) and diluted 10-fold with LB, and 100 μl was placed in a white 96-well plate (in duplicate). To prepare heat-killed cells for a positive control, 5 ml of bacterial culture was pelleted (10,000 × g, 5 min, room temperature), resuspended in 1 ml sterile deionized water, and boiled for 3 min, followed by sterile filtering (0.45-μm Ultra-free spin column filters; Millipore); lysates were diluted 10- to 1,000-fold as indicated in Fig. 3, and 100 μl was placed in a white 96-well plate (in duplicate). To each well, 100 μl of ToxiLight reagent was added, and the mixture was incubated at room temperature for 30 min. Luminescence was measured with a Molecular Devices SpectraMAX GeminiXS spectrometer. The average of the duplicate sample measurements was multiplied by the dilution factor and divided by the OD₆₀₀ of the original culture to account for differences in culture density.

Fig 3.

Hypervesiculation does not correlate with membrane instability. The luminescence of the indicated strains grown in LB for ∼16 h at 40°C was determined by normalizing to OD₆₀₀. Heat-killed cells were used as a positive control of full lysis; the numbering below the graph indicates the dilution factor to ensure the sample values are within the assay range. Error bars indicate SEM. Heat killed, 0 dilution, n = 4; heat killed, all other dilutions, n = 2; samples, n = 4.

OMV quantitation.

OMVs were isolated from broth cultures as follows. Medium (250 ml) was inoculated (1:250 dilution) from bacterial cultures grown overnight at 40°C. Cultures were grown to an OD₆₀₀ of ∼0.4 to quantify OMV production in log phase. To assess stationary-phase OMV production, the 250-ml overnight cultures were pelleted with a Beckman Avanti J-25 centrifuge (JLA-10.500 rotor; 10,000 × g, 10 min, 4°C), and the cells were resuspended in fresh 250 ml medium and again grown overnight at 40°C. Cells were pelleted with the Beckman Avanti J-25 centrifuge (JLA-10.500 rotor; 10,000 × g, 10 min, 4°C) and the resulting supernatants filtered (low-protein-binding Durapore membrane, 0.45-μm polyvinylidene fluoride; Millipore). Filtrates were centrifuged again with the Beckman Avanti J-25 centrifuge (JLA-16.250 rotor; 38,400 × g, 3 h, 4°C), followed by another centrifugation step with a Beckman Optima TLX ultracentrifuge if the pellets were not visible. In these cases, most of the supernatant was poured off, and the region where pelleted material should be was “resuspended” in the remaining supernatant and repelleted (TLA 100.3 rotor; 41,000 × g, 1 h, 4°C). Pellets were resuspended in Dulbecco's phosphate-buffered saline with added salt (0.2 M NaCl) (DPBSS) and filter sterilized through 0.45-μm Ultra-free spin filters (Millipore). A sample of the filtrate was plated on LB agar overnight to verify that the suspensions were free of bacteria.

To quantitate OMV yield, OMV preparations were boiled for 6 min in 2× Laemmli buffer, separated by 15% SDS-PAGE, and stained with SYPRO Ruby Red (Molecular Probes) overnight in the dark. Prior to and after staining, the gel was fixed for 1 h in a solution of 10% methanol and 7% acetic acid. Ruby-stained proteins were detected under UV light. E. coli OMPs F/C and A were quantified by densitometry (NIH Image J software). The OMP density values were divided by the OD₆₀₀ of the original culture to calculate OMV production, and this value was divided by the OMV production of the wild-type (WT) control strain to determine relative fold OMV production.

Periplasm assays.

Periplasm was isolated using an adapted method (25). Overnight broth cultures or log-phase cultures were grown to an OD₆₀₀ of ∼0.4 (100 ml) and cells pelleted with the Beckman Avanti J-25 centrifuge (JLA-16.500 rotor; 10 min, 10,000 × g, 4°C). The pellets were weighed and resuspended in 20 mM Tris-Cl (pH 8), 20% sucrose buffer (2 ml/1 g of cell pellet), 0.1 M EDTA (100 μl/1 g of cell pellet), and 0.5 mM phenylmethylsulfonyl fluoride (PMSF) protease inhibitor (Sigma) in ethanol. The cell suspension was incubated on ice for 1 h, and then 0.5 M MgCl₂ (160 μl/1 g of cell pellet) was added and mixed well. Lastly, the mixture was centrifuged on a microcentrifuge (8,100 g, 20 min, room temperature) to pellet cell debris, and the supernatant containing the periplasm was collected. To quantitate periplasmic protein, a Bradford assay (Coomassie blue stain; Thermo Scientific) was used with a standard curve generated with known concentrations of bovine serum albumin diluted in water.

OMP localization assay.

To detect surface-exposed OMPs, purified OMVs from stationary-phase cultures were incubated with a membrane-impermeative dye, Texas Red-X succinimidyl ester (Invitrogen) (dissolved in dimethyl sulfoxide [DMSO] at a concentration of 2.5 mg/ml), at a 1:50 ratio (dye to OMVs) for 1 h at room temperature in the dark due to the dye being light sensitive. To quench the unbound dye, 100 μl of 50-mg/ml lysine was added and thoroughly mixed. The OMVs were pelleted with the Beckman Optima TLX ultracentrifuge (TLA 100.3 rotor; 41,000 × g, 1 h, 4°C), washed twice with DPBSS, and stored in DPBSS. To account for the potential loss of OMVs during the wash steps, the absorbance at 280 nm (A₂₈₀) was measured in the original OMV sample and at the end of the surface staining procedure. The surface-stained samples were separated using 15% SDS-PAGE and labeled protein detected with UV light. OMVs from ΔompR and ΔompA strains were used to identify the bands corresponding to OMPs C/F and A that were used for quantification by densitometry. The total amount of OMPs C/F and A (folded and misfolded) was determined by SYPRO Ruby Red staining as mentioned above under “OMV quantitation.” For each strain, the same amount of OMVs based on A₂₈₀ was loaded for surface-labeled and SYPRO Ruby Red detection. Finally, the ratio between surface-stained and total OMPs was normalized to the wild-type strain.

Statistics.

Parameters used for the t test are of equal variance due to the comparison of identical experimental repetitions or unequal variance due to different experimental repetitions and a two-tailed distribution. For direct sample size comparison, the paired t test was used, and for fold comparison, the unpaired t test was used. A t test value of <0.05 was considered statistically significant; if the value was lower, then that was indicated in the significance value under the corresponding data. The number of times that each experiment was repeated (n) is stated in the figure legends.

RESULTS

The ΔnlpA ΔdegP growth defect correlates with reduced OMV production.

We reasoned that if vesiculation is critical for cells to overcome lethal envelope stress, as exists for a ΔdegP mutant at high temperature, a mutation that confers an undervesiculation phenotype should be conditionally synthetically defective for growth in conjunction with the degP deletion. To reveal such a conditional synthetic defect, we examined growth of single and double mutant cultures after they were shifted from 37°C to 40°C. The single mutants did not exhibit significant growth defects: the ΔnlpA mutant grew like the isogenic BW25113 background strain (“wild type” [WT]), whereas the ΔdegP mutant had a slightly decreased level of growth in comparison to WT. However, with the mutations in combination, the ΔnlpA ΔdegP mutant had a severe growth defect upon a shift to high temperature (40°C) (Fig. 1A). As expected, a more modest defect was observed when cultures were maintained at 37°C (see Fig. S1 in the supplemental material).

Fig 1.

Correlation between ΔnlpA ΔdegP growth and vesiculation defects. (A) The indicated strains of bacteria were grown in LB overnight at 37°C, inoculated into fresh LB (t = 0), and then grown at 40°C. The OD₆₀₀ was measured hourly. Error bars indicate standard error of the mean (SEM); n = 2. (B) The relative fold OMV production in cultures of the indicated strains grown in LB overnight at 40°C was determined by quantitating OMVs, normalizing to OD₆₀₀, and dividing by OD₆₀₀-normalized OMV production in a WT culture. Error bars indicate SEM. *, P < 0.05, **, P < 0.01. WT and ΔnlpA ΔdegP, n = 9; ΔnlpA and ΔdegP, n = 10.

NlpA and DegP have been linked to vesiculation levels in multiple E. coli strain backgrounds. Both the under- and hypervesiculation phenotypes have been established previously in a DH5α background (15), used for further studies in an ADA600 background (14), and confirmed here with full deletions in the BW25113 background (Fig. 1B). Furthermore, the growth defect of the ΔnlpA ΔdegP double mutant is also appreciable, although it is diminished in a MC4100 background (data not shown). Therefore, we could conclude that the synthetic growth defect for envelope stress and OMV production is not strain dependent.

We reasoned that if reduced vesiculation was responsible for the growth defect, the double mutant should exhibit reduced OMV production compared to that of the ΔdegP mutant. OMV production was assessed in cultures grown at 40°C using an established quantitative method. This method is based on the quantitation of OMPs in the cell-free culture supernatants, normalized by optical density (OD₆₀₀) to account for differences in bacterial growth (15). As expected from earlier studies, the ΔnlpA mutant produced fewer OMVs and the ΔdegP mutant produced more OMVs than the WT (Fig. 1B). Notably, the ΔnlpA ΔdegP double mutant produced significantly fewer OMVs than the ΔdegP mutant (Fig. 1B). Together with the growth data, these results show that the combination of a mutation that causes hypovesiculation and a temperature-sensitive mutation that increases vesiculation results in a stress-dependent synthetic growth defect.

The ΔnlpA defect depends on growth phase.

We noticed that the growth defect of the double mutant becomes reproducibly more severe after about 3 h of growth, when the cells reach an OD₆₀₀ of ∼0.4 (Fig. 1A). These data suggested that the ΔnlpA defect may be growth phase dependent. To investigate this hypothesis, we first quantified OMVs produced only during log-phase growth. The ΔnlpA and ΔnlpA ΔdegP mutants produced amounts of OMVs during log phase that were comparable to those produced by the WT and the ΔdegP mutant, respectively (Fig. 2A). Thus, there is no observable OMV production defect with respect to the ΔnlpA mutant during log-phase growth. However, when we quantified OMVs produced only in stationary phase, we observed the previously described, significantly reduced vesiculation phenotypes for the ΔnlpA and ΔnlpA ΔdegP mutants compared with the WT and the ΔdegP mutant, respectively (Fig. 2B). These data suggest that the loss of nlpA has growth phase-dependent consequences and that these are most severe during late log phase and stationary phase.

Fig 2.

The ΔnlpA defect is growth phase dependent. (A) The relative fold OMV production in log-phase cultures of the indicated strains, grown in LB to an OD₆₀₀ of ∼0.4 at 40°C, was determined by quantitating OMVs, normalizing to OD₆₀₀, and dividing by OD₆₀₀-normalized OMV production in a WT culture. Error bars indicate SEM. WT, n = 3; ΔnlpA, n = 4; ΔdegP, n = 5; ΔnlpA ΔdegP, n = 6. (B) The relative fold OMV production in stationary-phase cultures of the indicated strains grown in LB overnight at 40°C, after discarding log-phase-produced OMVs at ∼24 h, was determined as for panel A. Error bars indicate SEM. *, P < 0.005. WT, n = 3; ΔnlpA, ΔdegP, and ΔnlpA ΔdegP, n = 5.

Hypervesiculation does not correlate with membrane instability.

Since hypervesiculation has previously been associated with membrane instability (9), we needed to consider whether the so-called “OMVs” we collected were not solely OMVs but also included membrane fragments of compromised cells. If this were the case, the strain with the greatest growth defect should have produced the most apparent “OMVs.” Instead, we noted that the ΔnlpA ΔdegP mutant, the strain with the strongest growth defect, exhibited reduced OMV production in comparison to the single degP deletion strain (compare Fig. 1A and B). Nevertheless, to ensure that the mutations did not compromise membrane integrity, we examined the cytoplasmic content in the culture supernatants and the sensitivity of the mutant strains to actinomycin D and Sytox green, as well as conducting visual inspection by electron microscopy.

Cytoplasmic leakage into the supernatant was assessed by measuring adenylate kinase activity. Adenylate kinase is a cytoplasmic enzyme that is released from lysed cells and can be detected with a luminescence assay (26). Following the protocol of Jacobs et al. (26), we used heat-killed cells as our positive lysis control and to ensure that the content in our samples was not above the maximum detection limit of the assay (Fig. 3). The hypervesiculating ΔdegP mutant exhibited a relatively insubstantial cytoplasmic leakage defect, whereas the highest level of adenylate kinase in the supernatant was observed for the ΔnlpA ΔdegP mutant, which exhibits decreased OMV production with respect to the ΔdegP mutant. These data revealed that increased cytoplasmic leakiness does not correlate with increased OMV production, consistent with previous findings from our laboratory (15).

In other tests for membrane permeability, we used actinomycin D sensitivity and Sytox green incorporation. Actinomycin D sensitivity reveals the presence of compromised membranes, since the antibiotic normally cannot access its cytosolic substrate (27, 28), and similarly, Sytox green is a dye that exhibits fluorescence when bound to DNA but can penetrate only into cells with compromised membranes (29). We first examined strains grown to log phase. The log-phase WT strain was insensitive to actinomycin D and Sytox green, and the ΔnlpA mutant exhibited WT behavior during Sytox green treatment (see Fig. S2A and B in the supplemental material). However, despite the fact that the ΔdegP and ΔnlpA ΔdegP mutants had comparable and significant log-phase vesiculation phenotypes (Fig. 2A), their sensitivities to actinomycin D and Sytox green were not similar (the ΔdegP and ΔnlpA ΔdegP mutants were minimally actinomycin D sensitive, but only the ΔdegP mutant, and not the ΔnlpA ΔdegP mutant, was affected by Sytox green [see Fig. S2A and B in the supplemental material]). The sensitivity of the strains to these two agents was also examined in stationary phase. Whereas the ΔnlpA ΔdegP mutant had the greatest growth defect in the presence of Actinomycin D, the more severe hypervesiculating mutant, the ΔdegP mutant, was insensitive to the drug, similar to the case for the WT (see Fig. S2C in the supplemental material). Furthermore, the hypovesiculating mutant, the ΔnlpA mutant, showed greater Sytox green fluorescence than the WT, and the ΔdegP and ΔnlpA ΔdegP mutants were similarly slightly permeable to Sytox green (see Fig. S2D in the supplemental material). Thin-section transmission electron microscopy of random fields of WT, ΔdegP, and ΔnlpA ΔdegP bacteria further confirmed the membrane integrity of the mutants (see Fig. S3 in the supplemental material). We concluded that although the sensitivities of the strains to these agents differ, importantly, altered membrane integrity and permeability do not correlate with hypervesiculation, either in log or in stationary phase.

Complementation of the ΔnlpA ΔdegP growth defect.

We next tested whether the observed growth defect of the double mutant could be complemented by expression of the deleted genes on IPTG-inducible plasmids. Complementation of ΔnlpA ΔdegP with degP (pDegP) was assessed upon a temperature shift (from 37°C to 40°C). Uninduced, basal expression by pDegP was sufficient to complement the double mutant to WT levels of growth (Fig. 4A). In order to distinguish whether one or both functions of DegP (protease and/or chaperone) were responsible for the complementation, we used a plasmid expressing a protease-deficient point mutant of DegP, pmDegP (17). We found that the basal expression level of pmDegP could not complement the growth defect of the ΔnlpA ΔdegP mutant under the conditions tested (Fig. 4A). Therefore, the chaperone activity of DegP was not sufficient but the protease activity was required to repair the growth defect of the ΔnlpA ΔdegP mutant.

Fig 4.

Complementation of the ΔnlpA ΔdegP temperature shift growth defects by pDegP and pNlpA. (A) The indicated strains of bacteria containing either vector (pCS19) or a DegP- or mDegP-encoding plasmid (pDegP, pmDegP, respectively) were grown in LB overnight at 37°C, inoculated into fresh LB (t = 0), and then grown at 40°C. The OD₆₀₀ was measured hourly. Error bars indicate SEM; n = 4. (B) The indicated strains of bacteria containing either vector (pTrc99A) or an NlpA-encoding plasmid (pNlpA) were grown in LB overnight at 37°C, inoculated into fresh LB (t = 0), and then grown at 40°C. The OD₆₀₀ was measured hourly. Error bars indicate SEM; n = 4. (C) To verify that pNlpA expressed a functional protein, the indicated strains were grown in LB overnight at 37°C, inoculated into d-methionine- and glucose-supplemented minimal medium (t = 0), and grown at 40°C. The OD₆₀₀ was measured hourly. Error bars indicate SEM; n = 3. (D) The indicated strains of bacteria containing either vector (pTrc99A) or an NlpA-encoding plasmid (pNlpA) were grown in LB overnight at 37°C, inoculated into fresh LB (t = 0), and then grown at 40°C. IPTG concentrations used to induce NlpA at 0 h are indicated, if applicable. The OD₆₀₀ was measured hourly. Error bars indicate SEM; n = 3.

Next, we examined complementation of the ΔnlpA ΔdegP temperature-sensitive growth defect by plasmid-expressed NlpA (pNlpA). Basal levels of pNlpA were not sufficient to complement the growth defect of temperature-shifted cultures (Fig. 4B). Induction of NlpA expression above the basal level with a temperature shift was not possible, since IPTG addition to the double mutant containing the empty vector caused a growth defect more severe than the ΔnlpA ΔdegP growth defect, although this was notably not the case for the WT or single-deletion strains (data not shown). Complementation was also unsuccessful for the double mutant induced once the culture reached late log/stationary phase (data not shown). Since we could not successfully complement the ΔnlpA ΔdegP growth defect using pNlpA, we needed to establish whether pNlpA actually expressed functional protein and whether the cells were capable of using the protein under these stressful temperature shift conditions.

We first tested whether the product of pNlpA could functionally complement a methionine transport defect. As shown previously (30), NlpA can substitute for its close homolog MetQ as the periplasmic binding protein for the methionine ABC transporter (MetD) (60). This protein becomes critical when cells lack the ability to synthesize methionine (e.g., in ΔmetE strains) and when MetH is inactive due to the lack of vitamin B₁₂ (31, 32). Thus, to test functional pNlpA expression, we used the ΔnlpA ΔmetQ ΔmetE triple mutant as a background strain. This strain grows poorly in d-methionine-containing minimal medium because it can neither import nor synthesize d-methionine. Growth of the ΔnlpA ΔmetQ ΔmetE mutant carrying the vector was poor, as expected, but the growth defect could be complemented by pNlpA basal level expression, even when the culture was shifted from 37°C to 40°C (Fig. 4C). From these experiments, we concluded that pNlpA does express functional (methionine-binding) NlpA under temperature shift conditions.

We next investigated whether pNlpA toxicity in the ΔnlpA ΔdegP mutant was due to the likely severe defects in the envelope of this temperature-shocked, periplasmic protease-deficient strain. The effect of increasing concentrations of NlpA was studied in a ΔdegP strain under temperature shift conditions and compared to the effect in a temperature-shocked WT strain. The ΔdegP strain was acutely more sensitive to increasing levels of NlpA expression (basal versus IPTG-induced expression) (Fig. 4D). We therefore concluded that strains deficient in envelope protease activity are sensitive to nonnative (presumably too high) levels of NlpA.

In light of these data, we decided to assess pNlpA complementation under somewhat less stressful conditions, by using cultures which had become adapted to high-temperature growth, rather than immediately upon heat shock. First, we needed to ascertain whether the ΔnlpA ΔdegP mutant exhibited a growth defect under these conditions. Compared with the temperature-shifted cultures, a reproducible but somewhat reduced growth defect with somewhat different kinetics was observed for the heat-adapted cultures of the ΔnlpA ΔdegP mutant (Fig. 5A; see Fig. S1C in the supplemental material [compare with Fig. 1A]). To establish if the difference in growth curve shape between the shifted and adapted cultures originates solely from heat shock or the coincident heat shock and recovery from stationary phase, we conducted an experiment in which the cells were grown for 2 h after inoculation into fresh medium at 37°C before shifting to 40°C. The growth curve shape of the double mutant was similar to that when it was shifted directly into fresh 40°C medium but was less severe (see Fig. S4A in the supplemental material). The culture leveled out at an OD₆₀₀ of ∼0.6 instead of 0.4. This is to be expected since the culture was at a higher OD₆₀₀ when it was heat shocked. Furthermore, the growth phenotype of the double mutant became even more severe, leveling out at an OD₆₀₀ of ∼0.3, when the cultures were shifted from overnight growth at 30°C into fresh 40°C medium (see Fig. S4B in the supplemental material). These results suggest that the more extreme the heat shock, the lower the OD₆₀₀ when the double mutant growth arrests, and that the recovery from stationary phase has an additional exacerbating effect. The data support the hypothesis that the temperature-sensitive growth defect was related to acute protein misfolding stress.

Fig 5.

Complementation of the ΔnlpA ΔdegP high-temperature growth defect. (A) The indicated strains containing vector plasmid (pTrc99A) were grown in LB overnight at 40°C, inoculated into fresh medium (t = 0), and grown at 40°C. The OD₆₀₀ was measured hourly. Error bars indicate SEM; n = 2. (B) The indicated strains containing vector (pCS19) or a DegP-encoding plasmid (pDegP) were grown in LB overnight at 40°C, inoculated into fresh medium (t = 0), and grown at 40°C. The OD₆₀₀ was measured hourly. Error bars indicate SEM; n = 3. (C) The indicated strains containing vector (pCS19) or a DegP-encoding plasmid (pDegP or mDegP) were grown in LB overnight at 40°C, inoculated into fresh medium (t = 0), and grown at 40°C. IPTG concentrations used to induce DegP at 0 h are indicated, if applicable. The OD₆₀₀ was measured hourly. Error bars indicate SEM; n = 3. (D) The indicated strains containing vector (pTrc99A) or an NlpA-encoding plasmid (pNlpA or pmNlpA) were grown in LB overnight at 40°C, inoculated into fresh medium (t = 0), and grown at 40°C. IPTG concentrations used to induce NlpA at 0 h are indicated, if applicable. The OD₆₀₀ was measured hourly. Error bars indicate SEM; n = 2. (E) To verify that pmNlpA expressed a protein impaired in methionine binding, the indicated strains were grown in LB overnight at 37°C, inoculated into d-methionine- and glucose-supplemented minimal medium (t = 0), and grown at 37°C. The OD₆₀₀ was measured hourly. Error bars indicate SEM; n = 3. (F) To verify the methionine-independent growth defect of the ΔnlpA ΔdegP mutant, the indicated strains were grown in LB overnight at 37°C, inoculated into glucose-supplemented minimal medium (t = 0), and grown at 40°C. The OD₆₀₀ was measured hourly. Error bars indicate SEM; n = 3. Growth was monitored for 3 days (no nights), and the graph exhibits the monitored growth during these 3 days, eliminating the lines for the time no OD₆₀₀ was measured.

We then assessed complementation of the heat-adapted ΔnlpA ΔdegP cultures. As also seen for the shifted cultures, growth of heat-adapted ΔnlpA ΔdegP cultures could be corrected by complementation with basal expression of pDegP (Fig. 5B). To ascertain if increased amounts of the DegP chaperone (pmDegP) would complement the growth defect, we induced expression with 100 μM IPTG. We found that complementation still required the protease activity of DegP (Fig. 5C). Finally, we examined complementation by pNlpA. Induced expression of NlpA from the plasmid was effective in complementing, rather than exacerbating, the growth defect for the heat-adapted cultures (compare Fig. 5D with Fig. 4B), and IPTG had no toxic side effects. Together these data demonstrated that plasmid-expressed NlpA could complement the ΔnlpA ΔdegP growth defect under less acute denaturing conditions.

With the successful complementation using pNlpA, we could next investigate whether methionine binding by NlpA is necessary for its role in OMV production using a plasmid expressing a mutant variant of NlpA (mNlpA) that cannot bind methionine. The crystal structure of the NlpA homolog, Tp32, from Treponema pallidum was utilized to identify two residues, Asn 116 and Arg 119, which directly hydrogen bond methionine (33). These residues are conserved as Asn 142 and Arg 145 in NlpA. We then designed an expression plasmid for NlpA (pmNlpA) harboring alanine substitutions in these residues. We established that mNlpA could not bind methionine by the fact that basal-level expression of pmNlpA could not complement the methionine transport defect of the ΔnlpA ΔmetQ ΔmetE mutant (Fig. 5E). The ΔnlpA ΔdegP growth defect, however, could be complemented by basal-level expression of pmNlpA (Fig. 5D), suggesting that NlpA's methionine binding ability is not required in OMV production.

We further substantiated the absence of a role for methionine in the nlpA phenotypes by investigating the growth of the strains in medium lacking methionine. If a defect in methionine transport were critical to the growth defect of the ΔnlpA ΔdegP mutant, we would expect the ΔdegP mutant to grow equally poorly as the ΔnlpA ΔdegP mutant in methionine-free medium. Instead, we found that the growth defect of the double mutant persisted in minimal medium, whereas the ΔdegP mutant grew more similarly to the WT (Fig. 5F). In fact, we noted that the growth in minimal medium actually exacerbated NlpA's stationary-phase role: the growth of the double mutant was more comparable to that of WT during early-log-phase growth but then stopped earlier, reaching a maximum OD₆₀₀ of ∼0.5. The modest defect observed previously for this strain in log-phase growth in LB medium is possibly suppressed by the overall lower growth rate in minimal medium (compare log-phase growth in Fig. 5F and 1A), but when NlpA becomes critical (after mid-log phase is reached), slow growth may become insufficient to compensate for the ΔnlpA ΔdegP defect. The results from both the mutant NlpA and methionine-free medium experiments are consistent with the conclusion that NlpA's role in OMV production is methionine independent.

ΔdegP causes accumulation of periplasmic protein and misfolded OMPs which are OMV cargo.

We hypothesized that the growth defect of the ΔnlpA ΔdegP mutant likely originated from an overloaded periplasm that was unable to both effectively degrade proteins due to the degP defect and release accumulated, toxic proteins via OMVs because of the nlpA defect. In order to test this, we first needed to assess whether periplasmic protein accumulates in the ΔdegP strain. DegP is a protease/chaperone that is critical under stressful conditions, presumably to rid the periplasm of toxic product; however, it has never been shown that periplasmic protein actually accumulates in a strain lacking degP. We compared periplasmic protein densities using a Bradford assay to quantitate the amounts of protein in periplasmic preparations from equivalent amounts of cells grown at 40°C. The periplasmic preparation of the ΔdegP strain contained significantly larger amounts of protein than the periplasm from the ΔdegP/pDegP strain expressing basal levels of DegP (Fig. 6A). Complementation with pmDegP, which encodes DegP lacking protease activity, did not reduce the periplasmic protein amount in the ΔdegP strain (Fig. 6A), suggesting that the chaperone activity of DegP was insufficient to prevent periplasmic protein accumulation caused by ΔdegP.

Fig 6.

Periplasmic protein content correlates with altered OMV production, misfolded OMPs accumulate in ΔdegP OMVs, and the defects depend on DegP protease activity. (A) The indicated strains containing vector (pCS19) or a DegP-encoding plasmid (pDegP or pmDegP) were grown for ∼16 to 18 h in LB at 40°C. The periplasmic protein concentrations in the preparations were determined by the Bradford assay. Error bars indicate SEM. *, P < 0.001; n = 4. (B) Periplasm was prepared and quantitated as described for panel A from cultures grown to an OD₆₀₀ of ~0.4 (log phase) or from cultures grown overnight (stationary phase) in LB at 40°C. Error bars indicate SEM. Log phase: ΔdegP, n = 8; ΔnlpA ΔdegP, n = 6. Stationary phase: *, P < 0.05; **, P < 0.001; n = 9. (C) Surface-exposed OMPs of purified stationary-phase OMVs from cells grown at 40°C from WT and ΔdegP strains were labeled with the membrane-impermeable dye Texas Red-X succinimidyl ester and quantitated using UV light. Total OMPs were quantitated by Ruby stain and densitometry of SDS-PAGE. The ratio of surface OMPs/total OMPs was normalized to WT. Error bars indicate SEM. *, P < 0.005; n = 4.

To ensure that the increase in periplasmic protein amounts in the ΔdegP and ΔnlpA ΔdegP mutants was not contaminated with cytoplasmic protein, we compared their protein spectra to that of the WT purified periplasm, and no abundant set of new bands was visible in the mutants (see Fig. S5A in the supplemental material). To further investigate this, we compared the protein profiles of purified WT, ΔnlpA, ΔdegP, and ΔnlpA ΔdegP OMVs (see Fig. S5B in the supplemental material). No substantial differences in the protein species were detected in the OMVs either, suggesting that the periplasm of the ΔdegP and ΔnlpA ΔdegP mutants contain the typical cohort of proteins, but in abnormally large amounts, as would be expected when eliminating a major periplasmic chaperone/protease. However, as one-dimensional SDS-PAGE is a rather crude detection method, a more sensitive assay is required to determine whether more subtle enrichment of particular proteins occurs.

Next, we further examined the growth phase-dependent effect of the ΔnlpA mutation on periplasmic protein amount with the periplasmic protein accumulation assay described above. As expected from their comparable levels of log-phase OMV production (Fig. 2A), no difference in the amount of periplasmic protein was observed for the ΔdegP strain and the double mutant strain for log-phase cultures (Fig. 6B). (The periplasmic protein contents of the WT and the ΔnlpA mutant during log phase could not be evaluated, as these were below the detection level of this assay.) In contrast, for stationary-phase cultures, periplasm isolated from the ΔdegP mutant contained substantially and significantly more protein than WT, and periplasm from the ΔnlpA ΔdegP mutant exhibited a slight, but significant, increase in protein compared to that in the ΔdegP periplasm (Fig. 6B). Periplasm purified from the ΔnlpA mutant in stationary phase did not exhibit an increased protein amount compared to WT (data not shown), consistent with the unaffected growth phenotype of the ΔnlpA mutant (Fig. 1A) but not with its vesiculation defect (Fig. 1B). This could be because there is DegP in this strain and it is able to degrade any accumulated protein. Together, these results support our model that for the ΔdegP mutant, OMVs eliminate some of the toxic protein products that would normally be degraded by DegP's protease activity, and that a consequence of the ΔnlpA mutation in a ΔdegP background is the increased buildup of envelope proteins in the periplasm.

To directly investigate the idea that OMVs are an outlet that can relieve periplasmic overloading, we determined whether native DegP substrates, the OMPs (34), could be detected in the lumen of ΔdegP OMVs. Folded OMPs are integral OM proteins with surface-exposed domains, whereas we reasoned that misfolded OMPs are not surface exposed since they would accumulate in the periplasm (and consequently in the lumen of OMVs). Freire et al. used Texas Red-X succinimidyl ester, a membrane impermeative dye that reacts with primary amines, to label the surface-exposed components of OMPs (35). We anticipated that in a WT strain, nearly all of the OMPs would be folded with surface-exposed domains. In contrast, in a ΔdegP strain, the fraction of surface-exposed OMPs (folded) with respect to total OMPs (folded and misfolded) would be lower in the OMVs. This is indeed what we found: the ratio of surface-stained to total OMPs was reduced in a degP deletion background with respect to the WT value (Fig. 6C). We examined two other hypervesiculation mutants (10, 15) to ensure that this observed ratio decrease is not simply a general hypervesiculation phenomenon but is instead specific to ΔdegP, and this indeed turned out to be the case (see Fig. S6A in the supplemental material). Additionally, we observed a ΔdegP-specific increase in misfolded OMPs in a gel shift assay that allows for the preservation of the native conformation of OMPs (see Fig. S6B in the supplemental material)) (36). It should be noted here that even though we could detect these misfolded OMPs in the vesicle lumen, total OMP expression in this mutant has been previously shown to be similar to that in the WT (15). Therefore, native DegP substrates constitute waste products eliminated in the OMVs. Together, these data support the overall concept that ΔdegP periplasm contains substantially high levels of misfolded protein which cause toxicity that can be alleviated by the production of OMVs.

Other hypovesiculation mutations also have synthetic defects with ΔdegP.

To further investigate the hypothesis that reduced OMV production in a degP deletion background causes growth arrest, we examined the synthetic effects of ΔdegP with two other hypovesiculation mutations, ΔdsbA and ΔbolA, which came out of a high-throughput screen of single-gene mutants with OMV phenotypes (A. Kulp, A. Manning, B. Sun, T. Ai, A. Schmidt, and M. Kuehn, unpublished data). DsbA is a disulfide oxidoreductase which aids in periplasmic protein folding by inducing disulfide bonds (37, 38). BolA is thought to regulate morphology changes as well as increased resistance to antibiotics and detergents in stationary phase and under conditions of stress (35, 39–41). Both double mutants, ΔdsbA ΔdegP and ΔbolA ΔdegP, exhibited a synthetic growth defect comparable with that of the ΔnlpA ΔdegP mutant under temperature shift conditions (Fig. 7A and C; compare with Fig. 1A). Additionally, the growth defect again correlated with reduced amounts of OMV production of the double mutants with respect to that of the ΔdegP mutant (Fig. 7B and D; compare with Fig. 1B). We also investigated these strains for cytoplasmic leakage to exclude cell lysis. As seen previously, hypervesiculation does not correlate with increased luminescence (Fig. 7E). It should be noted that the luminescence data suggest that the ΔnlpA ΔdegP and ΔdsbA ΔdegP mutants have a more permeable envelope than the single mutants (Fig. 3 and 7E), and thus OMV production may be even lower that what our quantitation suggests (Fig. 1B and 7B). This is not a general trend, however, since the ΔbolA ΔdegP mutant had a similarly low luminescence signal as ΔdegP (Fig. 7E). These results further confirm that the combination of a mutation which reduces vesiculation and a mutation which increases vesiculation and periplasmic protein density results in a growth defect. We must stress here that this was not found to be the case for any hypovesiculation mutation combined with a degP deletion. For instance, there was no effect observed when two previously identified hypovesiculation mutations, ΔampG or ΔglnA, were combined with ΔdegP (unpublished data). Whereas this could be explained in the case of ΔampG, which caused production of fewer OMVs than in the WT at 37°C but not at 40°C, the hypovesiculation phenotype of ΔglnA persisted at 40°C (unpublished data). At this point, OMV production by these strains is not characterized well enough to understand why certain hypovesiculation mutations are dominant over the loss of degP to reduce OMV production and others are not.

Fig 7.

Other undervesiculation mutations have a synthetic defect with ΔdegP comparable to that of the ΔnlpA ΔdegP mutant. (A and C) The indicated strains of bacteria were grown in LB overnight at 37°C, inoculated into fresh LB (t = 0), and then grown at 40°C. The OD₆₀₀ was measured hourly. Error bars indicate SEM; n = 6. (B and D) The relative fold OMV production in cultures of the indicated strains grown in LB overnight at 40°C was determined by quantitating OMVs, normalizing to OD₆₀₀, and dividing by OD₆₀₀-normalized OMV production in a WT culture. Error bars indicate SEM. (B) *, P < 0.05; **, P < 0.00005. WT, n = 3; ΔdsbA, n = 3; ΔdegP, n = 4; ΔdsbA ΔdegP, n = 7. (D) *, P < 0.01; **, P < 0.0005. WT, n = 2; ΔbolA, n = 3; ΔdegP, n = 3; ΔbolA ΔdegP, n = 4. (E) The luminescence of the indicated strains grown in LB for ∼16 h at 40°C was determined by normalizing to OD₆₀₀. Error bars indicate SEM. WT, ΔdegP, ΔdsbA, and ΔdsbA ΔdegP, n = 4; ΔbolA and ΔbolA ΔdegP, n = 3.

DISCUSSION

The critical role that OMVs play in pathogenesis has been and continues to be studied quite intensively, but besides their contribution to virulence, roles for OMVs that would benefit the originating organisms, particularly nonpathogens, have been addressed only minimally. Here, we elucidate a conditional growth advantage for bacteria that are able to produce OMVs. We have shown that mutations in both degP, which affects the bacterial response to envelope stress, and either nlpA, dsbA, or bolA, which affect OMV production, lead to a synthetic genetic growth defect under conditions where envelope relief becomes critical. In the case of nlpA, the synthetic defect was shown to be growth phase dependent and to correlate with accumulation of protein in the periplasm. NlpA supports both OMV production and methionine import; however, the methionine binding ability of NlpA was dispensable for its role in OMV production. Evidence that misfolded OMPs accumulate in ΔdegP OMVs and that reduced levels of OMV production correlates with toxicity for the double mutants suggests that OMVs are a critical outlet for accumulated toxic envelope protein products upon exposure to a denaturing stress.

A relationship between the accumulation of protein in the periplasm and OMV production.

It was previously reported that the lethality of ΔdegP at 42°C could be rescued by an additional mutation in lpp, which acts as an envelope “staple,” forming links between the OM and peptidoglycan (20, 42, 43). This mutation caused strong hypervesiculation along with a membrane permeability defect (44), implying that the accumulation of misfolded protein in the periplasm can cause toxicity and that relief could arise by efflux of the toxic material via OMVs as well as leaky membranes. Further, a hypervesiculating mutation was discovered to suppress the toxicity of a heterologously expressed periplasmic protein (14). We add to these findings by demonstrating that in the absence of degP, a substantial amount of protein accumulates in the periplasm upon heat shock, and thus that DegP is responsible for managing a significant amount of envelope protein. That misfolded OMPs accumulate in ΔdegP OMVs and that the protease activity of DegP is sufficient to reduce the proteinaceous accumulation in the degP mutant strain confirm that the material is composed of DegP substrates. Additionally, the accumulation of these misfolded endogenous proteins becomes toxic when OMV production is also reduced. Thus, we show that OMV production acts in concert with DegP to relieve the periplasm of toxic consequences of envelope stress.

Our results also provide a more subtle assessment in comparing periplasmic overcrowding relief by DegP and OMVs. The lack of a substantial growth defect for the ΔdegP strain during constitutive, unstressed conditions suggests that OMV production is sufficient to relieve periplasmic overcrowding. The addition of the ΔnlpA mutation modestly, albeit statistically significantly, increased the periplasmic accumulation of protein in the ΔdegP strain in comparison to the relatively larger decrease in vesiculation (compare Fig. 2B and 6B). The lack of a fully inverse correlation of these phenotypes may be explained if the periplasmic capacity of the ΔnlpA ΔdegP mutant has reached a maximum level due to the lack of the OMV-mediated relief. To the best of our knowledge, the periplasmic capacity has not yet been established, and it may be a technically challenging endeavor to do so. Regardless, the growth defect of this strain (Fig. 1A) supports the hypothesis that upon reaching this maximum periplasmic capacity, the result is toxic, signifying that the OMV stress response is a critical pathway in bacterial viability during stress.

Growth phase-dependent role for NlpA.

It has been well established that vesicle production varies with the bacterial life cycle (11), but to the best of our knowledge, no growth phase-dependent OMV production mutant has yet been identified, which makes ΔnlpA unique. Interestingly, it has been demonstrated for numerous species that the maximum rate of OMV production occurs at the end of log phase (1, 45, 46), which is exactly where we found NlpA's role begins to be critical. Consistent with the growth phase-dependent phenotypes we observed here is NlpA's transcription pattern: NlpA is positively regulated by CsgD (47), a transcriptional regulator important for biofilm formation, a stationary-phase phenomenon, which is induced during mid-log phase (48, 49). Furthermore, it has been shown that CsgD itself is regulated by σ^S (50, 51), the stationary-phase σ factor, and that it can respond to cell density (49). It should be noted that the csg operons are usually induced during growth in minimal medium and lower temperature, but it has been shown that σ^S is induced during heat shock (52), which would consequently lead to CsgD and NlpA expression. This supports the need for NlpA and in turn OMV production, in addition to DegP, as critical responses to heat shock. The observed growth defect of the ΔnlpA mutant at 30°C, compared with 37°C or 40°C (see Fig. S1 in the supplemental material), is consistent with activation of CsgD at lower temperatures and in late log/early stationary phase (OD₆₀₀ of ∼0.4). Further studies are required to determine NlpA's role in low-temperature growth and how it relates to OMV production.

Potential roles for NlpA in OMV production.

NlpA's location in the envelope suggests that it could provide either a mechanistic or signaling role in OMV production. Most E. coli lipoproteins are anchored to the OM (53, 54). NlpA is unusual in that it is one of the minority of lipoproteins anchored to the IM, as well as one of the very few in the family of periplasmic binding proteins that appear to be membrane anchored (55). The atypical IM anchoring of NlpA indicates that location is likely to play a critical biological role.

NlpA's restricted mobility in the membrane could contribute mechanistically to vesiculation. For example, NlpA could initiate the process at a particularly favorable location of the cellular envelope environment later on in the bacterial life cycle. As such, NlpA would serve as a membrane marker where other, as-yet-unidentified, envelope proteins assemble to form a complex that results in OMV budding and release. This model is consistent with the reduced but not ablated levels of OMV production observed in the strains harboring the nlpA null mutation. Without this recruiting factor and/or location marker, the OMV production machinery could still assemble but would produce OMVs less efficiently.

Alternatively, the traditional functions of periplasmic binding proteins, substrate transport and chemotaxis (55), may play a role in vesiculation. As such, NlpA could act as a sensor, transmitting some kind of input signal to the machinery that stimulates OMV production in a growth phase-specific manner. Complementation results using mNlpA suggest that this activity would be independent of its ability to bind methionine, but it could respond to binding of other substrates or polypeptides. In this scenario, residual levels of OMVs produced in the absence of nlpA may originate from a basal level of OMV release that maintains a continuous state of “housecleaning” of the envelope.

Why is NlpA complementation successful at constant high temperature but not upon heat shock?

NlpA complementation has been rather challenging to interpret, since it was not successful under all conditions. Plasmid-expressed NlpA failed to complement under conditions where the cells underwent a rapid heat shock, but it could complement the growth defect of cultures kept at a constant high temperature. From these data, we concluded that there must be differences in the envelopes of heat-shocked cells compared to those of cells acclimated to high temperature. The most obvious culprit is the consequence of the σ^E envelope heat shock response (56–58). In the shift experiment, the bacteria are undergoing a dramatic transition in the state of the envelope. The temperature shift would cause a sudden bolus of misfolded OMPs and periplasmic proteins in the envelope and, as a result of the stress response, also a burst of upregulated proteases and chaperones. In contrast, cells acclimated to growth at high temperature are accustomed to the long-term consequences of σ^E regulon activation, with high levels of active proteases and chaperones along with downregulated OMPs. In this scenario, the periplasm is most likely “cleaner” due to the rescue, degradation, or simply lower number of misfolded membrane and periplasmic proteins.

Consequently, we propose a scenario in which NlpA can complement during acclimation but not heat shock: the extremely crowded periplasm during heat shock may interfere with whatever function the protein fulfills (e.g., localization of budding components). This situation is plausible for several reasons. The absence of DegP caused a substantial increase in misfolded proteinaceous material in the periplasm (Fig. 6; see Fig. S6B in the supplemental material). Additionally, the chaperone activity of DegP was insufficient for complementation under these conditions (Fig. 4A, 5C, and 6A). Finally, the higher the induction of NlpA, the worse the observed growth phenotype (Fig. 4D), suggesting that overexpressed NlpA is adding to the problem (e.g., accumulating proteinaceous garbage) under these conditions. In contrast, the “cleaner” periplasm of the heat-acclimated cells may allow for the observed functional complementation.

Hypovesiculation in the context of ΔdegP.

Our results show that the detrimental effect of a hypovesiculation mutation in a degP deletion background is not solely ΔnlpA specific, further supporting the hypothesis that reduced OMV production is the source of the defect. As yet, we do not understand the role of BolA in OMV production, but it is noteworthy that, like nlpA, bolA is at least partially regulated by σ^S (41), implying that it is expressed at times when the maximum amount of OMV production is observed (1, 45, 46). It is difficult to pinpoint the role of DsbA in OMV production due to its role as a general periplasmic folding factor (59). One of the DsbA downstream targets is likely to be the cause of the reduction in OMVs, since the deletion mutant of the factor responsible for DsbA recycling, DsbB (38), also acts as a hypovesiculator (Kulp et al., unpublished data).

We have to stress here, though, that this is not the case for any hypovesiculation mutation within a degP deletion background. In the case of the ΔampG hypovesiculating mutant, the likely reason that we did not see a substantial synthetic defect with ΔdegP was because vesicle production by the ΔampG mutant was reduced only at 37°C, not at 40°C (unpublished data). However, ΔglnA, identified by McBroom et al. (15), had no effect on ΔdegP mutant growth despite persistence of the hypovesiculation phenotype at 40°C (unpublished data). At this point, these strains are not characterized well enough at least in terms of OMV production to understand why certain hypovesiculation mutations reduce OMV production and others do not in a ΔdegP background. We hypothesize that when a mutation leading to reduced OMV production is dominant over ΔdegP, the gene product may be in one way or another mechanically involved in OMV production. These genes will be of particular interest to fully understand the mechanism of vesiculation.

In sum, these experiments have exposed a critical function for OMV production in general bacterial viability. Further experiments are necessary to identify binding partners of NlpA, as well as to determine how stress-induced periplasmic protein accumulation is sensed, in order to more fully understand how these factors contribute to optimal bacterial growth.