Increased Osmolarity in Biofilm Triggers RcsB-Dependent Lipid A Palmitoylation in Escherichia coli

ABSTRACT Biofilms are often described as protective shelters that preserve bacteria from hostile surroundings. However, biofilm bacteria are also exposed to various stresses and need to adjust to the heterogeneous physicochemical conditions prevailing within biofilms. In Gram-negative bacteria, such adaptations can result in modifications of the lipopolysaccharide, a major component of the outer membrane characterized by a highly dynamic structure responding to environmental changes. We previously showed that Gram-negative biofilm bacteria undergo an increase in lipid A palmitoylation mediated by the PagP enzyme, contributing to increased resistance to host defenses. Here we describe a regulatory pathway leading to transcriptional induction of pagP in response to specific conditions created in the biofilm environment. We show that pagP expression is induced via the Rcs envelope stress system independently of the Rcs phosphorelay cascade and that it requires the GadE auxiliary regulator. Moreover, we identify an increase in osmolarity (i.e., ionic stress) as a signal able to induce pagP expression in an RcsB-dependent manner. Consistently, we show that the biofilm is a hyperosmolar environment and that RcsB-dependent pagP induction can be dampened in the presence of an osmoprotectant. These results provide new insights into the adaptive mechanisms of bacterial differentiation in biofilm.

and host immune defenses (2,3) at the origin of chronic and device-related biofilm infections (4).
Several studies showed that modifications in lipopolysaccharide (LPS) structure in Gram-negative bacteria also contribute to biofilm tolerance of antimicrobial compounds (5,6). Consistently, we had previously reported that biofilm formation in Escherichia coli and other Gram-negative bacterial species is associated with increased incorporation of a palmitate acyl chain into the LPS lipid A domain. Lipid A palmitoylation attenuates the Toll-like receptor 4 (TLR4)-mediated inflammatory host response and increases in vivo survival of biofilm bacteria (7). Lipid A palmitoylation is mediated by the palmitoyl transferase PagP (8)(9)(10) and is increased in both biofilm and prolonged planktonic cultures in late stationary phase (7), suggesting that these two environments could share common conditions triggering this modification. However, while E. coli pagP expression is repressed by H-NS and positively regulated by the anti-H-NS factor SlyA, the exact mechanism underlying biofilm-associated increased lipid A modification remains unclear.
In planktonically grown E. coli, the pagP gene has been demonstrated to depend on the PhoPQ and EvgAS transcriptional regulatory systems (11,12). The PagP enzyme itself can also be activated posttranslationally by a redistribution of phospholipids within the outer membrane (11). Here we showed that increased lipid A palmitoylation in biofilm results from two distinct molecular mechanisms: a posttranslational activation of PagP observed in biofilm and under late-stationary-phase conditions (12)(13)(14) and a previously undescribed transcriptional induction of pagP expression in biofilm only. We determined that induction of pagP expression in mature biofilm is dependent on RcsB, a regulator of the Rcs envelope stress response phosphorelay (15,16). This Rcs-dependent induction of pagP expression can be recapitulated by exposing planktonic bacteria to high saline stress, suggesting that pagP expression is triggered in response to specific physicochemical cues developing within mature biofilms. This report therefore provides new insights into how physiological adaptations to the local biofilm microenvironment can contribute to increased stress resistance in biofilm bacteria.

RESULTS
pagP transcription is specifically induced in E. coli biofilms. We previously reported that E. coli biofilm bacteria show a higher level of lipid A palmitoylation on LPS gels than planktonic bacteria. However, this difference between planktonic and biofilm LPS profiles is attenuated with time ( Fig. 1A). Since accumulation of lipid A palmitoylated species can result either from posttranslational activation of PagP (12,13) or from transcriptional induction of the pagP gene (17), we decided to monitor pagP expression in E. coli using a pagP-lacZ transcriptional operon fusion, in which lacZ is inserted downstream of an intact pagP gene. In contrast to planktonic and biofilm lipid A profiles on LPS gels, we observed a strictly biofilm-associated increase of the pagP expression level, whether tested at 24 or 96 h (Fig. 1B). To further confirm the correlation between biofilm formation and transcriptional induction of pagP expression in biofilm, we compared the levels of pagP expression in BW25113 and MG1655 E. coli K-12 backgrounds carrying or not the biofilm-promoting F conjugative plasmid (18). In both cases, F ϩ biofilm-forming strains displayed earlier maximum induction of pagP ( Fig. 1C and D; see also Fig. S1 in the supplemental material), confirming the biofilmassociated induction of pagP expression. In the rest of this report, pagP expression levels are compared at 48 h using F-carrying strains.
Transcriptional regulator RcsB regulates pagP expression in biofilms. pagP expression is controlled by histone-like protein repressor H-NS and the SlyA regulator (7). However, while inactivation of slyA reduced pagP expression in both planktonic culture and biofilms, it did not abolish pagP induction in biofilms, suggesting that, in addition to SlyA, other regulators play a role in biofilm-associated pagP expression (7). We therefore tested the contribution of various envelope stress and general response regulators ( Fig. 2A; see also Fig. S2) and found that a mutation in rcsB was the only mutation abolishing all differences between planktonic and biofilm pagP expression in a biofilm-forming E. coli strain carrying the pagP-lacZ fusion ( Fig. 2A). RcsB is the response regulator of the Rcs system, a phosphorelay transduction system that responds to various environmental stresses (19)(20)(21). This phenotype was fully complemented upon plasmid-based expression of rcsB in medium-copy-number plasmid pCA24N-rcsB (Fig. 2B), confirming the contribution of RcsB to pagP biofilm regulation. Finally, we investigated the contribution of RcsB transcriptional induction to total accumulation of palmitoylated lipid A species. Using matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS), we compared the relative abundances of hexa-acylated (nonpalmitoylated) and hepta-acylated (palmitoylated) lipid A species in E. coli wild-type (WT) and ΔrcsB biofilms. The MALDI-TOF MS analysis shows two major peaks, at m/z ϭ1,797.4, corresponding to the hexa-acylated nonpalmitoylated lipid A species, and at m/z ϭ2,035.6, corresponding to the heptaacylated palmitoylated lipid A species. These results show that inactivation of rcsB leads to a 2-fold decrease in the ratio of palmitoylated and nonpalmitoylated lipid A species [i.e., lipid A(2035.8)/lipid A(1797.4)] ( Fig. 2C and D). To further confirm that the decreased level of palmitoylated lipid A species in the RcsB mutant was due to the absence of pagP transcriptional induction, we tested whether deletion of rcsB could also impact PagP at the posttranslational level. To address this issue, we used a Pcl-pagP strain, in which pagP is placed under the control of a constitutive promoter, and monitored the lipid A palmitoylation profile in the presence or absence of RcsB. We did not observe any decrease of lipid A palmitoylation in the rcsB mutant (Fig. S3). Taken together, these results indicate that lipid A palmitoylation occurs both through posttranslational activation of the basal PagP enzyme inserted in the outer membrane and through a RcB-dependent transcriptional induction of the pagP gene.
RcsB-dependent biofilm regulation does not require RcsACDF or phosphorylation. The Rcs pathway is composed of several proteins (RcsA, RcsC, RcsD, and RcsF) involved in phosphorylation and dephosphorylation of response regulator RcsB, mod- Biofilm-Dependent Lipid A Palmitoylation in E. coli ® ulating its ability to bind DNA and regulate target gene expression. To study the contribution of Rcs components to RcsB-dependent induction of pagP, we monitored planktonic and biofilm pagP expression in several rcs mutants and showed that none of the tested rcs mutants affected pagP expression as much as rcsB, suggesting that, in addition to signaling from the outer membrane, pagP induction might also be partially independent of the classical Rcs phosphorelay cascade (Fig. 3A).
RcsB is known to be alternatively phosphorylated via an independent pathway that uses acetyl-phosphate (AcP), an intermediate metabolite of the Pta-AckA pathway, as a phosphoryl group donor (22). Although deletion of ackA leads to accumulation of AcP in the cell and therefore favors RcsB activation, deletion of both pta and ackA results in the absence of AcP production. To test whether the acetyl-phosphate level drives pagP induction via alternative and Rcs-independent RcsB phosphorylation, we monitored pagP induction in a pta-ackA mutant background, in which AcP production is abolished FIG 2 RcsB-dependent regulation of pagP increases lipid A palmitoylation in biofilm bacteria. (A) E. coli K-12 BW25113 F ϩ pagP-lacZ strains, including the wild-type (WT) strain or strains with the pspF, degP, baeR, rcsB, phoP, ompR, recA, fnr, or rpoS gene deleted, were grown in planktonic cultures (Pk) and biofilm (Bf) for 48 h, and ␤-galactosidase activity was measured. (B) E. coli K-12 BW25113 F ϩ pagP-lacZ pCA24N, E. coli K-12 BW25113 F ϩ ΔrcsB pagP-lacZ pCA24N, and E. coli K 12 BW25116 F ϩ ΔrcsB pagP-lacZ pCA24N-lacZ were grown in planktonic (Pk) cultures and biofilm (Bf) for 48 h in M63B1 minimal medium supplemented with 0.4% glucose and 0.05 mM isopropyl-␤-D-thiogalactopyranoside (IPTG), and ␤-galactosidase levels were measured. Statistical significance was assessed using one-way analysis of variance (ANOVA) followed by Bonferroni's post hoc comparison tests (***, P Ͻ 0.001; N.S., not significant). (C and D) MALDI-TOF MS analysis of lipid A extracted from E. coli K-12 MG1655 F ϩ , E. coli K-12 MG1655 F ϩ ΔpagP, and E. coli K-12 MG1655 F ϩ ΔrcsB 48h biofilm bacteria was performed using a DHB matrix (C) or a CMBT matrix (D). Analysis of lipid A by MALDI-TOF MS in biofilm bacteria shows two major peaks, at m/z ϭ1,797.4, corresponding to a hexa-acylated nonpalmitoylated lipid A species, and at m/z ϭ2,035.6, corresponding to a hepta-acylated palmitoylated lipid A species. Results are presented as the ratio of palmitoylated and nonpalmitoylated lipid A species: A(2035.8)/A(1797.4). (9). We did not observe a significant difference in pagP expression in the double pta/ackA mutant compared to the wild-type strain (Fig. 3B). Moreover, we found that pagP expression was still induced in biofilms in a strain combining both rcsC and ackA/pta mutations (Fig. 3C).
These results support the hypothesis that regulation of pagP expression does not require RcsB phosphorylation via the Rcs phosphorelay or acetyl phosphate pathway.
Consistently, whereas expression of the rprA gene, whose transcription requires the phosphorylated form of RcsB (14,23), was induced in a strain harboring a constitutively phosphorylated variant of RcsB, RcsB D56E , and was decreased in the nonphosphorylatable variant RcsB D56S (Fig. S4), biofilm-specific induction of pagP was not impaired in any of the rcsB alleles (Fig. 3D).
Contribution of GadE, SlyA, and H-NS in RcsB-dependent biofilm regulation of pagP. We then tested the potential contribution of several auxiliary transcription regulatory proteins known to interact with nonphosphorylated RcsB, including BglJ, GadE, MatA, and DctR (15,(24)(25)(26). We showed that only a mutation in gadE abolished biofilm-specific induction of pagP (Fig. 4). GadE was previously shown to regulate transcription of the glutamate decarboxylase genes gadA and gadB in complex with RcsB but independently of RcsB phosphorylation (26). To study the regulatory interplay between H-NS-, SlyA-, and RcsB/GadE-dependent regulation of pagP under biofilm conditions, we first tested whether slyA transcription is regulated by RcsB. We therefore monitored slyA expression in the wild-type strain carrying the pCA24N or pCA24N-rcsB coli K-12 MG1655 F ϩ pagP-lacZ strains, including the wild-type (WT) strain and strains with the rcsB, rcsD, rcsC, rcsA, or rcsF gene deleted, were grown in planktonic (Pk) cultures and biofilm (Bf) for 48 h, and ␤-galactosidase activity was measured. (B) E. coli K-12 MG1655 F ϩ pagP-lacZ strains, including the WT or strains with the pta gene or the ackA gene or both the pta gene and the ackA gene deleted, were grown in planktonic (Pk) cultures and biofilm (Bf) for 48 h, and ␤-galactosidase activity was measured. (C) The E. coli K-12 MG1655 F ϩ pagP-lacZ WT strain or strains with the rcsC or the pta/ackA genes or both the rcsC and ptA/ackA genes deleted were grown in planktonic (Pk) cultures and biofilm (Bf) for 48 h. ␤-Galactosidase activity was measured. (D) The E. coli K-12 MG1655 F pagP-lacZ WT strain and the ΔrcsB strain as well as strains rcsBD56E and rcsBD56S (with a constitutively activated variant and a nonphosphorylatable variant of the rcsB gene, respectively), were grown in planktonic (Pk) cultures and biofilm (Bf) for 48 h, and ␤-galactosidase levels were measured. Statistical significance was assessed using one-way analysis of variance (ANOVA), followed by Bonferroni's post hoc comparison tests (***, P Ͻ 0.001; N.S., not significant).
Biofilm-Dependent Lipid A Palmitoylation in E. coli ® inducible plasmid, using a slyA-lacZ transcriptional fusion. We showed that overexpression of rcsB led to increased expression of slyA ( Fig. 5A), suggesting that RcsB could positively regulate slyA expression, which would in turn positively regulate pagP.
However, overexpression of rcsB in a ΔslyA pagP-lacZ reporter strain still led to increased expression of pagP (Fig. 5B). We also tested the possibility that H-NS controls pagP expression in an indirect way that involves repression of the positive regulator rcsB. However, deletion of rcsB in an hns-deficient strain, in which pagP expression is derepressed, did not restore the wild-type level of pagP expression (Fig. 5C). These results therefore suggest that, while RcsB acts upstream of SlyA, other SlyAindependent inputs can contribute to alleviate H-NS-mediated repression of pagP (see model in Fig. 7).
Increased ionic strength triggers RcsB-and biofilm-dependent pagP expression. The implication of stress response regulators RcsB and GadE in regulation of pagP expression suggested that biofilm-associated induction of pagP could be triggered in response to physicochemical stress conditions created locally within the biofilm environment (27). The Rcs envelope stress system is known to respond to several physicochemical cues, including osmotic stress and exposure to highly ionic environments (20). To test whether RcsB/GadE-dependent expression of pagP is activated in response to osmotic stress in biofilm, we measured the osmolarity of planktonic and biofilm supernatants of E. coli after 24, 48, and 72 h of growth and showed that the osmolarity was consistently higher in biofilm supernatants than in planktonic samples (Fig. 6A). We further studied the contribution of osmotic stress to pagP gene expression by briefly exposing exponentially growing bacteria to salts such as NaCl or KCl and to polymers (sucrose or dextran) commonly used for studying bacterial response to osmotic stress, and we then monitored pagP expression. A significant increase in pagP gene expression was observed only when bacteria were exposed to NaCl or KCl (0.5 to 0.8 M) (Fig. 6B), and we found that this induction was dependent on rcsB (Fig. 6C). Consistently, monitoring the lipid A palmitoylation profile on LPS gels in the presence or absence of NaCl showed that the additional band corresponding to palmitoylated lipid A was observed when bacteria were exposed to NaCl (0.6 and 0.8 M) (Fig. 6B) and that the impact of NaCl was strongly reduced in the rcsB mutant strain (Fig. S5). Moreover, NaCl-dependent induction of pagP was strongly reduced in planktonic bacteria upon addition of the osmoprotectant osmolyte betaine (Fig. 6D), suggesting that ionic stress triggers pagP expression.
In E. coli, sensory proteins such as the KdpD sensor of the KdpDE two-component system can detect changes in osmolarity (28). Moreover, mechanosensitive channels such as that corresponding to the MscL membrane protein also participate in regulation of turgor pressure by modulating ion fluxes (29). To test whether changes in ion fluxes could activate RcsB-GadE-dependent regulation of lipid A palmitoylation, we monitored pagP induction in biofilms formed by trkA, kdpD, mscL, or mscS mutants but could not detect any impact on pagP transcriptional induction in biofilm (Fig. S6).
The GadE stress regulator is involved in maintenance of pH homeostasis and is induced under low-pH conditions (30). We therefore measured the pH of planktonic and biofilm supernatants of E. coli after 24, 48, and 72 h of growth to test whether expression of pagP is activated in response to pH changes and showed that the pH of biofilm supernatants is acidified compared to planktonic samples (Fig. S7A). However, growing planktonic bacteria at different pH levels had only a limited impact on pagP gene expression, indicating that pagP transcription is not induced by acid stress (Fig. S7B). Moreover, whereas gadE is required to induce pagP expression in biofilms, NaCl-dependent induction of pagP in planktonic bacteria was not abolished in a ΔgadE mutant (Fig. S8), suggesting that exposure to NaCl partially mimics conditions created in the biofilm environment.
These results demonstrate that increased ionic stress can trigger pagP expression in a RcsB-dependent and GadE-independent manner and suggest that increased ionic stress associated with increased osmolarity could be part of the signal activating RcsB and inducing pagP expression and lipid A palmitoylation in E. coli biofilms.

DISCUSSION
LPS is a major component of the Gram-negative bacterial outer membrane and possesses a highly dynamic structure responding to environmental changes (31). We previously showed that formation of Gram-negative bacterial biofilms is associated with transient palmitoylation of lipid A mediated by palmitoyl transferase enzyme PagP, contributing to protection of biofilm bacteria from the host immune system (7). Here we investigated mechanisms underlying induction of this biofilm-associated LPS modification.
Increased lipid A palmitoylation is a general feature of Gram-negative bacterial biofilms which, in E. coli, can result either from transcriptional induction of pagP or from posttranslational activation of the PagP enzyme in response to perturbations of outer membrane lipid asymmetry (12). In prolonged planktonic cultures, absence of transcriptional induction of pagP while palmitoylation occurs suggests that, under these growth conditions, increased lipid A modification could result solely from posttranslational activation of PagP. Stress created under these extreme stationary-phase conditions could alter the integrity of the outer membrane, for instance, by causing phospholipid substrate migration to the , and ␤-galactosidase activity was measured. Inset: analysis of LPS extracted from control and NaCl-exposed (0.6 M or 0.8 M) planktonic bacteria by Tricine SDS-PAGE. Arrowheads indicate a modified LPS band. (C) E. coli K-12 BW25113 F ϩ pagP-lacZ pCA24N, E. coli K-12 BW25113 F ϩ ΔrcsB pagP-lacZ pCA24N, and E. coli K-12 BW25113 F ϩ ΔrcsB pagP-lacZ pCA24N-rcsB were grown in planktonic cultures in M63B1 supplemented with 0.4% glucose and 0.1 mM isopropyl-␤-D-thiogalactopyranoside (IPTG) to an OD of 0.4 to 0.5. NaCl (0.6 or 0.8 M) was added to the cultures, and ␤-galactosidase activity was measured after 1 h of incubation. (D) E. coli K-12 MG1655 F ϩ pagP-lacZ was grown in planktonic cultures to an OD of 0.4 to 0.5 and exposed to 0.6 M of NaCl in the presence or absence of betaine (an osmoprotective osmolyte). ␤-Galactosidase activity was measured after 1 h of incubation. Statistical significance was assessed against the corresponding untreated control (CT) using one-way analysis of variance (ANOVA) followed by Bonferroni's post hoc comparison tests (***, P Ͻ 0.001; N.S., not significant).

Szczesny et al.
® outer leaflet, where they can activate the basal level of resident PagP molecules (32,33). Given the well-established stationary-phase character of the biofilm environment (34), it is likely that a significant proportion of the lipid A palmitoylation observed in biofilm bacteria also originates from posttranslational activation of PagP.
In addition to potential posttranslational activation of PagP, we showed that the biofilm environment triggers RcsB-dependent transcriptional induction of pagP, also contributing to the global increase in lipid A palmitoylation. This effect was not detected in our previous study (7), probably due to the parallel confounding increase in posttranslational activation of PagP under these conditions. The Rcs phosphorelay system regulates colanic acid synthesis, cell division, growth on surfaces, and motility, all important functions involved in biofilm formation and bacterial physiological adaptation to the biofilm lifestyle (21,35). Several major and accessory components (RcsF, RcsC, RcsD, and RcsA) are involved in RcsB phosphorylation and subsequent regulation of a subset of Rcs-regulated genes (36). Here we show that RcsB regulates pagP expression independently of major contributions of other Rcs components and of its phosphorylation. Consistently, we discovered that the auxiliary transcriptional regulator GadE, one of the 4 transcriptional regulators known to interact with the nonphosphorylated form of RcsB (MatA, BglJ, DtcR, and GadE) (24,26), contributes to RcsBdependent regulation of pagP induction. GadE is expressed under low-pH conditions (30,37), whereas RcsB can be activated by several stress conditions, including membrane perturbations (38,39), osmotic stress (20), and growth on solid surfaces (40). Since the RcsB-BglJ complex specifically regulates at least 10 genes, 8 of which correspond to unknown or predicted proteins (41), this suggests that, in addition to pagP, other genes regulated by the RcsB-GadE complex remain to be discovered, and their discovery could lead to identification of novel biofilm properties. Although the RcsB-GadE complex was reported to regulate only gadA/BC genes (26), it is worth noting that the glutamate decarboxylase genes gadA and gadB are, like pagP, under the negative control of H-NS (Fig. 7) and were shown to be induced in biofilm (34,42,43).
Our results suggest that, in addition to pagP, conditions prevailing within biofilm could specifically trigger the expression of other RcsB-regulated and H-NS-regulated genes. Consistently, the expression of RcsB-regulated genes cpsB and rprA, or of RcsB/GadE-regulated genes gadA and gadB, showed significant induction under biofilm conditions (see Fig. S9 in the supplemental material) (44,45). Hence, RcsB could play a central role in H-NS antisilencing and could enable the expression of rcs genes regulated both through activation of the phosphorelay cascade and through activation of RcsB complexes formed with auxiliary regulators. Consistently, among genes positively regulated by the RcsB-BglJ complex, leuO is an anti-H-NS factor (46,47) that could, when expressed, contribute to global antagonism of H-NS-mediated repression. However, it has been shown that all RcsB auxiliary regulators, like some of their targets, are repressed by H-NS (48)(49)(50)(51), raising the question of how their expression can be activated. Since H-NS preferentially binds to curved and A-T-rich DNA regions (52,53), alteration of DNA structure under environmental conditions reducing the degree of DNA curvature, such as high temperature or increased osmolarity (54), could lead to H-NS destabilization and derepression of H-NS-repressed genes.
The Rcs system was previously described as being activated by various cues such as LPS or peptidoglycan defects, temperature, or osmotic shock (16). We showed that RcsB-dependent pagP expression and increased lipid A palmitoylation can be induced by ionic solutes but not by nonionic compounds such as sucrose or dextran. The mechanisms behind this specificity could involve an osmotic stress response triggered to minimize loss of water and maintain positive turgor pressure (55). In bacteria subjected to salt stress, one key mechanism of bacterial osmoadaptation lies in rapid K ϩ uptake through different K ϩ sensory and transport systems (56). However, deletion of major components of the E. coli potassium transporters or mechanosensitive channels had no impact on pagP transcriptional induction in biofilm. Further investigations will determine the contribution of additional ion transporters to biofilm-associated pagP induction.
Consistently with NaCl-dependent and KCl-dependent induction of pagP, we detected higher osmolarity in the biofilm environment, suggesting that increased osmolarity could be the signal triggering RcsB-dependent induction within biofilms. In Dickeya dadantii, the level of osmoregulated periplasmic glucans (OPGs), a periplasmic component of the cell envelope, was shown to modulate the activity of the Rcs system. While accumulation of OPGs in the periplasm inhibits the Rcs pathway under conditions of low osmolarity, in biofilms, which are characterized by higher osmolarity, absence of OPGs could serve as a signal activating the Rcs pathway and leading to increased expression of pagP (57). While the origin of this increased osmolarity is not known, it might be explained by accumulation of metabolic wastes, cell lysis, or production of matrix components that characterize the biofilm mode of growth. Increased osmolarity could therefore be a general feature regulating other biofilm-specific responses beyond lipid A modifications. Interestingly, several examples of membrane modifications in response to osmotic stress have been described, including modification of the LPS structure in Rhizobium meliloti (58) and the increase in the amount of saturated fatty acids in Pseudomonas putida colonies, contributing to increased membrane rigidity (59). While PagP-dependent palmitoylation was also shown to contribute to membrane rigidity (60), we speculate that this phenotype is part of a more global mechanism for adaptation to life under conditions of increased osmolarity and contributes to maintaining cell homeostasis by limiting loss of water.
In Fig. 7 we present a model for the pagP transcriptional regulation in biofilm described in this study. We propose that the biofilm environment, characterized by increased osmolarity, triggers activation of RcsB and its interaction with the auxiliary regulator GadE. Formation of a RcsB-GadE complex, occurring independently of RcsB phosphorylation, leads to positive regulation of pagP expression. Our results further suggest that this new regulatory pathway could be at least partially independent of the previously described anti-H-NS SlyA factor.
The analyses presented in this study were performed on a biofilm population. The observed increased lipid A palmitoylation could be interpreted as a general feature of all biofilm cells. However, biofilm environments are highly heterogeneous, and, depending on their localization within the biofilm, bacteria encounter different stress conditions (27). It is therefore possible that the level of lipid A palmitoylation that we observe in biofilm reflects an average level between those of highly palmitoylated and nonpalmitoylated cells which, in addition, could result either from increased pagP expression or from activation of the basal level of the dormant enzyme. The resulting heterogeneity among biofilm cells could represent an adaptive advantage ensuring population survival in hostile environments. Consistently, our previous in vivo analyses showed that lipid A palmitoylation increased biofilm bacterial survival by reducing the host inflammatory response (7). However, while the presence of highly palmitoylated lipid A at the surface of the biofilm could represent a defensive barrier preventing biofilm eradication by the host immune system, our results suggest instead that the conditions required for pagP induction are more closely correlated with deeper biofilm layers in which accumulation of metabolites and other wastes could increase external osmolarity. Further localization studies with oxygen-insensitive reporters or imaging mass spectrometry will help clarify the correlation between induction of pagP and the physicochemical cues present in the biofilm environment.

MATERIALS AND METHODS
Bacterial strains and growth conditions. Bacterial strains and plasmids used in this study are described in Table S1 in the supplemental material. Antibiotics (concentrations) were used as follows: kanamycin (50 g/ml); chloramphenicol (25 g/ml); ampicillin (100 g/ml); zeocin (50 g/ml); and tetracycline (7.5 g/ml). When specified, isopropyl-␤-D-thiogalactopyranoside (IPTG) was used to induce gene expression at the following concentrations: for planktonic cultures, 0.1 mM; for biofilms, 0.05 mM.
Biofilm growth in continuous-flow microfermentors. Biofilm formation on an internal glass slide stand in continuous-flow microfermentors was performed as previously described (18). Inoculation was performed by dipping the glass spatula for 5 min in a culture adjusted to an optical density at 600 nm (OD 600 ) of 2 from overnight bacterial cultures grown in M63B1 minimal medium supplemented with 0.4% glucose and the appropriate antibiotics. The spatula was then reintroduced into the microfermentor, and biofilm culture was performed at 37°C in M63B1 with glucose. Biofilm biomass produced at different time points was rapidly resuspended in 10 ml of fresh M63B1 medium (without glucose), and biofilm bacteria were used for LPS analysis of measures of ␤-galactosidase activity.
Planktonic growth. Planktonic cultures were also grown in M63B1 medium supplemented with 0.4% glucose at 37°C.
Strain construction. The E. coli deletion mutants used in these studies either originated from E. coli K-12 BW25113 wild-type and isogenic mutant strains from the Keio collection (61) or were derived by -red linear DNA gene inactivation using pKOBEG plasmids or derivates (62,63). To construct pagP-lacZ derivatives, mutations were transferred by P1vir phage transduction to BW25113 pagP-lacZ, BW25113 F ϩ pagP-lacZ, MG1655 pagP-lacZ, or MG1655 F ϩ pagP-lacZ strains. When required, kanamycin resistance markers flanked by two FLP recombination target (FRT) sites were removed using Flp recombinase (64).
LPS analysis by Tricine SDS-PAGE. A total of 10 8 bacteria were pelleted and resuspended in 100 l of lysing buffer (Bio-Rad) containing 1% SDS, 20% glycerol, 100 mM Tris (pH 6.8), and Coomassie blue G-250. Lysates were heated at 100°C for 10 min; proteinase K was then added at 1 mg/ml, and the reaction mixture was incubated at 37°C for 1 h. These samples (4 l) were electrophoresed in a Tricine SDS-PAGE system, which improves resolution of the low-molecular-weight LPS band, using 4% stacking gel and 20% separating gel (65). LPS was then visualized by a periodate-silver staining method adapted from a previous study (66). Gels were immersed in fixing solution (30% ethanol, 10% acetic acid) for 1 h, washed in water three times for 5 min each time, oxidized in 0.7% periodic acid for 10 min, and washed in water three times for 5 min each time. Gels were then immersed for 1 min in 0.02% thiosulfate pentahydrate, rinsed quickly in water, and stained in 25 mM silver nitrate for 10 min. After a 15 s wash in water, gels were developed in 3.5% potassium carbonate-0.01% formaldehyde. Development was stopped for 30 min in 4% Tris base-2% acetic acid.
Direct lipid A isolation from bacterial cells. Lipid A was isolated directly by hydrolysis of bacterial cells as described previously (67,68). Briefly, 5 mg of lyophilized bacterial cells was suspended in 100 l of a mixture of isobutyric acid-1 M ammonium hydroxide (5:3 [vol/vol]) and kept for 1.5 h at 100°C in a screw-cap test tube in a Thermomixer system. The suspension was cooled in ice water and centrifuged (2,000 ϫ g, 5 min). The recovered supernatant was diluted with 2 volumes of water and lyophilized. The Biofilm-Dependent Lipid A Palmitoylation in E. coli ® sample was then washed once with 100 l of methanol (by centrifugation at 2,000 ϫ g for 5 min). Finally, lipid A was extracted from the pellet in 50 l of a mixture of chloroform, methanol, and water (3:1.5:0.25 [vol/vol/vol]).
MALDI-TOF MS analysis. LPS samples were dispersed in water at 1 g/l. Lipid A extracts in chloroform-methanol-water were used directly in this mixture of solvents. In both cases, a volume of a few microliters of sample solution was desalted with a few grains of ion exchange resin (Dowex 50W-8 [H ϩ ]). Aliquots of 0.5 to 1 l of the solution were deposited on the target, and the spot was then overlaid with matrix solution and left to dry. Dihydroxybenzoic acid (DHB) (Sigma-Aldrich) or 5-chloro-2mercaptobenzothiazole (CMBT) was used as the matrix. It was dissolved at 10 mg/ml in 0.1 M citric acid solution in the same solvents as were used for the analytes (69). Different analyte/matrix ratios (1:2, 1:1, and 2:1 [vol/vol]) were tested to obtain the best spectra. Negative-and positive-ion mass spectra were recorded on a PerSeptive Voyager-DE short tandem repeat (STR) time of flight mass spectrometer (Applied Biosystems) in the linear mode with delayed extraction. The ion-accelerating voltage was set at Ϫ20 kV, and the extraction delay time was adjusted to obtain the best resolution and signal-to-noise ratio. LPS analyses were performed by LPS-BioSciences (IGM, Orsay, France).
␤-Galactosidase activity assay. To determine the ␤-galactosidase enzyme activity of strains carrying a transcriptional lacZ fusion, bacteria were grown in M63B1 minimal medium with glucose at 37°C for the time indicated for each experiment. The enzyme activity was assayed at least in triplicate for each strain under each condition as previously described (70), and the results are expressed in Miller units.
Screening for conditions impacting lipid A palmitoylation and pagP expression. To evaluate the roles of different physicochemical stresses in pagP expression and lipid A palmitoylation, we proceeded as follows.
(i) Test of osmotic agents. Planktonic cultures in M63B1 minimal medium supplemented with 0.4% glucose and the appropriate antibiotics were started at an optical density at 600 nm (OD 600 ) of 0.05 from overnight precultures performed in LB. Bacteria were incubated at 37°C until they reached an optical density at 600 nm (OD 600 ) approaching 0.5. The tested chemical was then added to the cultures, and ␤-galactosidase activity was measured after 1 h if incubation. NaCl, KCl, sucrose, and dextran were used to induce osmotic stress. The listed compounds (concentrations) were used as follows: NaCl (0.6 M and 0.8 M); KCl (0.5 M); 17.5-kDa and 70-kDa dextran (10%); and sucrose (20%).
Osmolarity and pH measurements. To determine the osmolarity level and the pH in planktonic and biofilm cultures, we proceeded as follows.
(i) Sample preparations. For biofilm measurements, the biofilm biomass was recovered from the glass spatula using a cell scraper to minimize dilution of the biofilm supernatant. The biofilm sample was centrifuged and filtered to eliminate residual bacteria. For planktonic cultures, an aliquot of the bacterial suspension was centrifuged and filtered.
Samples were recovered from at least 3 independent biofilm and planktonic cultures and 3 different time points (24 h, 48 h, and 72 h).
(ii) Measures. Osmolarity measurements were performed on 100-l samples using a freezing-point osmometer and repeated 3 times. pH was assessed using pH strips (Sigma P-4536) with a unit resolution of 0.5 pH.
Statistical analysis. Statistical analyses were performed using one-way analysis of variance (ANOVA) and the Bonferroni multiple-comparison test with Prism 5.0 for Mac OS X (GraphPad Software, Inc.). Each experiment was performed at least 3 times. Statistical significance is indicated as follows: *, P ϭ Ͻ0.05; **, P ϭ Ͻ0.01; ***, P ϭ Ͻ0.001.

ACKNOWLEDGMENTS
We thank Nadim Majdalani, Emmanuelle Bouveret, Daniela De Biase, and David Clarke for helpful discussions and critical reading of the manuscript. We are also grateful to Emmanuelle Bouveret, Nadim Majdalani, and Inigo Lasa for generously providing some of the strains and plasmids used in this study and to Nicolas Desprat for his help in osmolarity measurements. We also thank Alexei Novikov and Martine Caroff (LPS-Biosciences, IGM, Orsay, France) for their help in LPS analysis.