Functional determinants of a small protein repressor controlling a broadly conserved bacterial sensor kinase

The PhoQ/PhoP two-component system plays a vital role in the regulation of Mg2+ homeostasis, resistance to acid stress, cationic antimicrobial peptides, and virulence in E. coli, Salmonella and related bacteria. Previous studies have shown that MgrB, a 47 amino acid membrane protein that is part of the PhoQ/PhoP regulon, inhibits the histidine kinase PhoQ. Thus, MgrB is part of a negative feedback loop modulating this two-component system, disruption of which is strongly associated with acquired colistin resistance in clinical isolates of Klebsiella pneumoniae. MgrB may also be important for preventing hyperactivation of PhoQ and possibly act as an input for alternate signals in the PhoQ/PhoP pathway. To explore the mechanism of action of MgrB, we have analyzed the effects of point mutations, C-terminal truncations and domain swaps on MgrB activity. In contrast with two other small membrane protein regulators of histidine kinases in E. coli, we find that the MgrB transmembrane (TM) domain is necessary for PhoQ inhibition. Bacterial two-hybrid assays suggest the TM domain mediates interactions with itself as well as with PhoQ. Additionally, alanine scanning mutagenesis of the periplasmic domain of MgrB indicates that, with the exception of a few highly conserved residues, most residues are not essential for MgrB’s function as a PhoQ inhibitor. Our results indicate that the regulatory function of the small protein MgrB is derived via distinct contributions from multiple residues spread across domains. Interestingly, the TM domain also appears to interact with other non-cognate histidine kinases in a bacterial two-hybrid assay, suggesting a potential route for evolving new small protein modulators of histidine kinases.


Introduction
Bacteria thrive in a multitude of niches, often under challenging physicochemical conditions. They have therefore evolved stress response systems to monitor environmental cues and modulate their physiology accordingly. Many of these stress responses are controlled by twocomponent signaling systems, one of the primary modes of signal transduction in bacteria (1). The PhoQ/PhoP system in Escherichia coli, Salmonella, and related Gram-negative bacteria is a well-studied example of a two-component system that is critical for virulence and facilitates adaptation to conditions of low Mg 2+ or Ca 2+ , low pH, and the presence of cationic antimicrobial peptides (2)(3)(4)(5). In response to input signal, the sensor histidine kinase PhoQ autophosphorylates and modulates the phosphorylation state of the response regulator PhoP, which functions as a transcription factor (2,6).
Previous work has shown that a 47-amino acid small inner-membrane protein MgrB is regulated by the PhoQ/PhoP pathway and inhibits the PhoQ sensor kinase, thus constituting a negative feedback loop in this signaling circuit (7). MgrB homologs are found in numerous enterobacteria and PhoQ repression by MgrB appears to be conserved in several Gram-negative genera. MgrB could serve as a point of control for additional input signals modulate PhoQ. For example, MgrB activity is affected by the redox state of the periplasm via the protein's two conserved periplasmic cysteines (8). MgrB is not essential and appears to mainly act in a regulatory role, however it can have a significant impact on bacterial physiology. In E. coli, recent work has shown that strong stimulation of the PhoQ/PhoP system hinders cell division (9). Cells grown in low levels of Mg 2+ in the absence of MgrB form filaments due to strong activation of PhoQ in this condition. MgrB-mediated feedback inhibition thus appears to be important to appropriately limit PhoQ activity and prevent hyperactivation of the system. In at least some contexts, loss of mgrB can also confer a fitness advantage: mutation of mgrB has emerged as one of the primary mechanisms of acquired resistance to the last resort antibiotic colistin in clinical isolates of Klebsiella pneumoniae (10,11).
MgrB is a member of a growing list of small proteins (12,13) that have been shown to be expressed in bacteria. Most of these proteins were initially overlooked due to arbitrary size constraints for defining open reading frames. As a result, they are largely understudied and most have yet to be functionally characterized. However, it is clear that small proteins can play key roles in diverse cellular processes (13). Modulation of histidine kinases, transporters and other integral membrane proteins by small bitopic membrane proteins is one emerging theme. For example, in addition to MgrB, the PhoQ histidine kinase is modulated by another small membrane protein, SafA (14). SafA, which is present exclusively in E. coli, and acts as a connector between the EvgS/EvgA and PhoQ/PhoP two-component systems. In some E. coli isolates, this connection is required for PhoQ activation in response to acid stress (15,16). Another example is the 127-amino acid long MzrA (17), which activates EnvZ and functions as a connector between the CpxA/CpxR and EnvZ/OmpR two component systems.
MgrB acts by inhibiting the kinase activity (18), however, the mechanism by which this small protein inhibits a protein that is roughly ten times larger than itself in size, and the sequence determinants responsible for this action are largely unknown. In this work, we examine the interactions between PhoQ and MgrB, including the role of TM domain. By swapping domains, we show that the TM domain of MgrB is required for the protein to repress PhoQ, in contrast with SafA and MzrA, which do not require their TM domains for their activity (19,20). Using sitedirected mutagenesis, we identified key residues in the MgrB TM and periplasmic domains that are important for PhoQ repression. We find that a large number of amino acids in the periplasmic domain can be replaced with minimal effect on MgrB function. We further show that while the TM domain of MgrB alone is insufficient to confer activity, it interacts with PhoQ and itself in a bacterial two-hybrid assay. Intriguingly, we also uncovered interactions of MgrB with other histidine kinases. Together our findings suggest that PhoQ inhibition by MgrB depends on coordinated interactions from amino acid residues spanning different domains.

Role of MgrB transmembrane domain in PhoQ inhibition
As first step towards understanding how MgrB inhibits PhoQ at a structural level, we looked at the role of the predicted MgrB transmembrane (TM) domain. Based on previous work, MgrB is believed to contain a single 18-amino acid long TM domain oriented so that the N-terminus is in the cytoplasm (7). We performed a domain swapping experiment ( Figure 1A), wherein the putative TM domain of MgrB was replaced with either (i) a 20-amino acid long TM from the E. coli maltose transporter MalF or (ii) a 21-amino acid long human glycophorin A (GpA) TM ( Figure 1B). Note that the GpA TM used in our study is a variant that has a lower propensity to dimerize than the wild-type GpA TM sequence (21). An N-terminally-tagged GFP fusion of MgrB (GFP-MgrB) was previously shown to be functional and localized to the membrane (7). All of the variants (or mutants) of MgrB described henceforth are derivatives of this construct, enabling us to confirm expression and membrane localization by fluorescence microscopy. Both MgrB variants with swapped TM domains localized to the cell membrane ( Figure S1A) and their expression levels were comparable to that of WT MgrB ( Figure S3). Domain swap experiments of two previously studied histidine kinase regulators that also have a single TM domain -SafA and MzrA indicate that the transmembrane sequence is dispensable for their function (19,20). Indeed, our control experiments where we replaced the TM sequences of SafA and MzrA with that of GpA confirmed these results. Thus, for SafA and MzrA, the TM domain appears to primarily act as an anchor to direct the regulatory domain of these proteins to the membrane ( Figure S2). To determine if MgrB behaves in the same way, we monitored the effect of MgrB and its TM swap variants on PhoQ activity using a PhoQ/PhoP-regulated transcriptional reporter that are about 80% of activity of no-MgrB control, indicating that the TM swaps unable to repress PhoQ efficiently. These results indicate that the MgrB TM domain does not simply function as a membrane anchor but is also critical for MgrB's activity as a PhoQ inhibitor.
Given that the MgrB TM sequence is necessary for its function, we wanted to further identify specific residues that might contribute to repression of PhoQ. The predicted TM sequence consists of residues 7-24 (underlined sequence shown in Figure 2), based on results from the TMHMM2 transmembrane domain prediction server (http://www.cbs.dtu.dk/services/TMHMM/). Hydrophobic residues (V, L, A) are often involved in surface interactions facing the hydrophobic interior of the membrane. Prior work suggests that the conserved cysteine within the TM domain (C16) may be less important for MgrB function relative to the two other conserved periplasmic Cys residues (8). Aromatic residues are known to preferentially occur in the lipid-water interface regions (22,23) and could be crucial for membrane anchoring as well as surface interactions. Polar amino acids such as N and Q can facilitate TM domain interactions by forming stable hydrogen bonds (24). Based on these considerations, we chose to focus on residues W20, Q22, F24 of the TM domain and tested the effects of single alanine substitutions. All three mutant proteins localized to the cell membrane ( Figure S1B) and their expression levels were comparable to that of WT ( Figure S4). The W20A mutant greatly affected MgrB function and showed a high b-gal activity (~30-fold higher than WT MgrB) ( Figure 2). Mutants Q22A and F24A displayed modest effects on reporter gene expression (~5-and 8-fold higher b-gal activity compared to WT MgrB, respectively). Recent bioinformatic analyses have found that a 'polar-xxpolar' motif (where polar = S/T/D/E/Q/N) is highly conserved in bacterial TM domains and drives protein-protein interactions (24). We wondered if the motif Q 22 -x-x-N 25 could be relevant, and created a double alanine mutant MgrB Q22A/N25A. This mutant showed a slight increase in reporter gene expression relative to Q22A mutant alone suggesting that this motif could be partly important to MgrB activity.

The MgrB TM domain is necessary but not sufficient for its function
As swapping the TM domain sequence in MgrB severely affected its activity, we tested whether an MgrB mutant lacking the periplasmic domain retained an ability to repress PhoQ. This variant -MgrB-24, in which N25 is replaced with a stop codon-showed proper membrane localization ( Figure S1C), but could not repress reporter gene expression, indicating that this variant composed of MgrB TM domain and the short N-terminal cytoplasmic sequence is not functional. To assess the minimum length of MgrB needed for efficient repression of PhoQ/PhoP-regulated reporter gene expression, we tested other C-terminal truncations of MgrB: MgrB-29, MgrB-34, MgrB-39, MgrB-44, corresponding to replacing the codons for Q30, F35, A40 and I45 with a stop codon, respectively. These mutants displayed similar membrane localization to that of WT MgrB ( Figure S1). Of these four truncations, MgrB-29 and -34 did not show any detectable repression, MgrB-39 showed slight repression, and MgrB-44, which lacks only the last two residues, showed the strongest repression, although last variant was still compromised in activity compared with WT ( Figure S5). Evidently, amino acid interactions in both the TM and the small periplasmic region of MgrB influence the protein's ability to inhibit PhoQ.

Site-directed mutagenesis analysis of MgrB periplasmic domain
To explore the relative importance of different residues in the MgrB periplasmic domain, we performed alanine scanning mutagenesis and analyzed the effects of these mutations on PhoQ inhibition. The majority of the alanine substitutions did not affect PhoQ/PhoP reporter gene expression, but six displayed a notable increase in b-gal activity relative to WT ( Figure 3). Mutants C28A, C39A showed the highest effects (80-90 fold higher b-gal activity than WT MgrB), consistent with previous data indicating that these two cysteine residues are important for MgrB activity (8). In addition, mutants G37A D31A, F34A and W47A displayed a significant increase in reporter gene expression -approximately 30-, 8-, 10-, and 10-fold relative to WT MgrB, respectively. These residues may be important for structural stability of the periplasmic domain or for inter-protein interactions. We confirmed that these mutants localized properly to the cell membrane ( Figure S1D) and that the protein levels were not considerably different from that of the WT ( Figure S6).

MgrB lacking periplasmic domain is capable of physical interaction with PhoQ
Although MgrB-24 (N-terminus+TM domain) does not have the ability to inhibit PhoQ, it could still play a role in establishing physical contact with PhoQ. To explore this possibility, we used a bacterial two-hybrid (BACTH) system based on split adenylyl cyclase (25, 26). To assay interactions of MgrB N-terminus+TM domain with PhoQ, we prepared N-terminal fusions of MgrB-24 (MgrB lacking periplasmic domain) and PhoQ to T18 and T25 fragments, respectively. As controls, we tested interactions between MgrB-PhoQ, MgrB-VcToxR (Vibrio cholerae inner membrane protein ToxR) as well as MgrB-T25 fragment alone. In this spot assay, reconstitution of the T18 and T25 fragments restores the activity of adenylate cyclase (CyaA), which leads to an increase in cAMP levels. This increased cAMP in turn enables efficient induction of the genes required for maltose catabolism, resulting in pink colonies on maltose MacConkey plates. Cells expressing T18-MgrB-24 and T25-PhoQ showed a dark pink colony after 48 h at 30 °C, whereas cells expressing T18-MgrB-24 and T25 fragment alone remained colorless ( Figure 4). In addition, cells expressing T18-MgrB-24 and either T25-MgrB or T25-MgrB-24 were also pink. These results suggest that the MgrB N-terminus+TM domain physically interacts with itself and also with PhoQ. Control cells expressing T18-MgrB and T25-PhoQ or T25-MgrB showed a dark pink spot, consistent with previous work showing an interaction between MgrB with itself and with PhoQ (7) and cells expressing either T18-MgrB or T18-MgrB-24 with either T25-VcToxR or T25 fragment alone remained colorless. Together, these results suggest that the TM domain and N-terminus of MgrB are important for the interaction between MgrB and PhoQ as well as for the interaction between MgrB and itself.

MgrB interaction with non-cognate histidine kinases
To test if MgrB is capable of physically interacting with histidine kinases other than PhoQ, we made constructs of E. coli EnvZ, AtoS, CpxA, and PhoR fused to the T25 fragment of adenylyl cyclase and performed BACTH analysis by spot assays as well as measuring b-galactosidase activities of cells grown in liquid cultures. Intriguingly, cells expressing T18-MgrB and either T25-AtoS, -EnvZ, or -PhoR showed a high level of activity although less than the b-activity of cells expressing T18-MgrB and T25-PhoQ ( Figure 5). Cells expressing T18-MgrB and T25-CpxA displayed a level of b-activity only marginally higher than the negative controls. These results suggest that MgrB is capable of physically interacting with EnvZ, AtoS, and PhoR, at least in the context of the bacterial two-hybrid assay. To test whether MgrB affects the activity of these histidine kinases, we measured gene expression from promoters regulated by these histidine kinases, in the presence or absence of plasmid-driven MgrB expression. However, we did not see any effect of MgrB on gene expression controlled by any of these histidine kinases (data not shown).

Discussion
In this study, we have explored the regions of MgrB that are important for its activity. Using a functional GFP-tagged derivative of MgrB to assay membrane localization and expression level, we determined the amino acid determinants in MgrB TM and periplasmic domains that are important for MgrB regulation of PhoQ. Previous work using a PhoQ chimera indicated that the periplasmic domain of MgrB, and in particular two conserved cysteines in this domain, are critical for PhoQ inhibition (7,8), however little was known about the role of the MgrB TM domain or other residues in the periplasmic domain. Studies of two other bitopic membrane proteins that regulate histidine kinases determined that the transmembrane domains were not required for activity, suggesting the domain may serve primarily as a membrane anchor (19,20). In contrast, our results show a striking loss of function when the MgrB TM is replaced with unrelated TM sequences. Bacterial two-hybrid experiments further indicate that the TM domain alone interacts with PhoQ, as well as with MgrB, although the TM domain alone does not show any inhibitory interaction. We also identified specific residues at the C-terminus of the TM domain that are particularly important for MgrB activity. Taken together, these results indicate that that the specific TM domain sequence is critical for MgrB interactions with PhoQ as well as with itself.
By making a series of C-terminal truncations in MgrB, we found that virtually all of the periplasmic domain is required for full activity. To identify which periplasmic residues in MgrB are required for PhoQ repression activity, we analyzed individual alanine substitutions across the entire periplasmic domain. Remarkably, despite the small size of MgrB (47 amino acids in total with 22 periplasmic residues), substitutions at most positions had little effect if any on MgrB activity. Of the 22 residues in the periplasmic domain, only six were individually essential for full activity: the two conserved cysteines (C28 and C39), as previously noted (8), as well as G37, D31, F34 and W47. Furthermore, substitutions at these latter four residues lowered but did not eliminate MgrB activity. Interestingly, mutations for each of these positions are also associated with colistin-resistant Klebsiella isolates (10,11,(27)(28)(29)(30)(31)(32). Taken together, our findings show that the sequence determinants for MgrB activity are not restricted to either one of the domainsperiplasmic or TM, and that functionality is mediated by collective interactions of key amino acid residues scattered across the polypeptide.
Unexpectedly, we also found that the MgrB lacking its periplasmic domain, as well as full-length MgrB, interact with several other histidine kinases in addition to PhoQ in a bacterial two-hybrid assay. However, we did not observe a significant effect of MgrB on gene expression regulated by these histidine kinases. The TM domain in many histidine kinases such as PhoQ, EnvZ etc. are believed to form a helical bundle consisting of four helices (33). Our results may reflect an interaction between the MgrB TM dimers and a conserved transmembrane conformation shared by multiple histidine kinases. The periplasmic domain of MgrB likely provides additional interactions facilitating specific regulation of PhoQ. It will be interesting to determine if there are other small membrane proteins that have evolved to interact with other histidine kinases using structural features similar to those of the MgrB TM domain. In addition, the results from our study of MgrB, as well as studies of other small membrane proteins, may provide a platform for reengineering these proteins to regulate new targets through rational design and directed evolution.

Materials and Methods
Strains, plasmids and cloning. See Supplementary Materials for tables of strains (Table S1), plasmids (Table S2), and primers (Table S3) used in this study. All strains were derived from E. coli K-12 MG1655. Gene deletions and reporter constructs were transferred between strains using transduction with P1vir. Deletions in the Keio collection (34) that were used for strain construction were confirmed by PCR using primers flanking the targeted gene. Kanamycin resistance markers were excised, when required, using the FLP-recombinase-expressing plasmid pCP20 (35). Strain SAM72 was constructed by first excising the kanamycin resistance cassette from the ompC-mCherry transcriptional reporter strain AFS256 to give SAM71. Kanamycin resistance from the mzrA deletion strain JW3067 was then transduced into SAM71 by P1 transduction. A deletion of safA (SAM73) was constructed as described in (15,35). SAM73 was then used to construct SAM76 by transducing the mgtA-lacZ transcriptional reporter from TIM199. Strain SAM85, which is a lacI q containing version of BTH101 was made by conjugation of the bacterial two-hybrid host strain BTH101 (F−) with XL1-Blue (F+). This strain was used instead of BTH101 as it gave reduced background in our MacConkey plate assays. Plasmids expressing gfpA206K fusions to mgrB mutants (pSY3-9, pSY72-75, pJC1-7, pTG1-15) were created by either QuikChange site-directed mutagenesis (Stratagene) or Inverse PCR (36) using pAL38 as template. Plasmids expressing gfpA206K fusions of mzrA (pSY29) and safA (pSY31) were constructed by Gibson assembly (37). Plasmids pSY32 (MzrA TM swap) and pSY33 (SafA TM swap) were made by Inverse PCR using pSY29 and pSY31 as templates, respectively. Plasmid pSY34 expressing T18-MgrB-24 only was created using pAL33 as template by Inverse PCR. For all the plasmids mentioned above, see Table S3 for specific primer sequences used for each construction. Plasmids pSY61, pSY64 and pSY68 were constructed as follows: atoS, cpxA, and phoR were PCR-amplified off of the MG1655 genome using the following primer pairs -atoS-xbaI-U1/atoS-kpnI-L1, cpxA-xbaI-U1/cpxA-kpnI-L1 and phoR-xbaI-U2/phoR-kpnI-L1, respectively, digested with XbaI/KpnI restriction enzymes and then cloned into pKT25 at XbaI/KpnI sites. All constructs were confirmed by DNA sequencing.
Media and growth conditions. Liquid cultures were grown at 37 °C with aeration, unless otherwise indicated, in either LB Miller medium (Fisher Scientific) or in minimal A medium (38) containing 0.2% glucose, 0.1% casamino acids and 1 mM MgSO4. For routine growth on solid medium, LB or minimal medium containing bacteriological grade agar (Fisher Scientific) was used. The antibiotics ampicillin, kanamycin and chloramphenicol were used at concentrations 50-100, 25 and 20-25 μg mL −1 , respectively. The lac and trc promoters were induced with isopropyl β-D-1-thiogalactopyranoside (IPTG) at a final concentration of 1 mM when indicated. When IPTG was not mentioned in the description of the culture conditions, the basal transcription from the trc promoter was used to drive expression. For bacterial two-hybrid screening, MacConkey/maltose indicator plates were prepared as described (25) using MacConkey agar base powder (Difco™), 100 μg mL −1 ampicillin, 25 μg mL −1 kanamycin, 0.5 mM IPTG and 1% maltose.
Fluorescence microscopy. For each microscopy experiment, strains were grown in minimal medium containing with antibiotics as needed and single-cell measurements were performed essentially as described previously (9). For MgrB localization experiments, saturated overnight cultures of AML67 strains carrying gfpA206K fusions of either wild-type mgrB (pAL38), mgrB mutants (pSY3-9, pSY72-75, pJC1-7, pTG1-15) or gfpA206K-only (pAL39) were diluted 1:1,000 into fresh medium containing 100 µg mL -1 ampicillin and grown at 37 °C with aeration to an OD600 ∼0.2-0.3 (~4-5 h) before visualization. For single-cell fluorescence measurements from transcriptional reporter strains, cultures were grown as outlined above, then rapidly cooled in an ice-water slurry and streptomycin was added to a final concentration of 250 μg mL −1 to stop protein synthesis. Microscope slides were prepared with 1% agarose pads and cell fluorescence was measured by imaging and quantified with in-house software as described previously (9,39).
Bacterial two-hybrid assays. For the colorimetric spot assay, multiple clones from the transformation plate were inoculated in 3 mL of LB containing 100 µg mL -1 ampicillin and 25 µg mL -1 kanamycin. (Several clones were picked in order to reduce heterogeneity (25)) Cultures were grown overnight at 30 °C with shaking. The next day, 3 µL of each culture was spotted on MacConkey/maltose indicator plates and plates were then incubated at 30 °C. For each pair of plasmids, triplicate experiments were performed. Color change on the indicator plates was recorded after 48 h of incubation. Alternatively, β-gal assays were performed on the overnight liquid cultures prepared as described for the colorimetric assay.   putative TM domain expression because we were able to detect PhoQ chim -PhoQ chim interactions at levels comparable to those for (wild-type) PhoQ-PhoQ interactions, consistent with previous reports that both Taken together, these results suggest that th of E. coli PhoQ is important for the int Interestingly, co-expression of T18-MgrB an putative TM domain the N-termini of MgrB and PhoQ, respectively. Based on the known topology of PhoQ, and the topology of MgrB (discussed above), both CyaA fragments should be in the cytoplasm. A strain expressing T18-MgrB and T25-PhoQ showed a significantly higher level of beta-galactosidase activity when compared with strains expressing the T18 and T25 fragments alone ( Figure 5) or expressing the fusions to either MgrB or PhoQ individually (data not shown). These results suggest there is a physical interaction between MgrB and PhoQ. Furthermore, a strain expressing T18-MgrB and the T25-fragment fused to the N-terminus of PhoQ chim showed a minimal increase in beta-galactosidase activity relative to the controls. This is unlikely to be due to a defect in PhoQ chim expression because we were able to detect PhoQ chim -PhoQ chim interactions at levels comparable to those for (wild-type) PhoQ-PhoQ interactions, consistent with previous reports that both PhoQ and PhoQ chim form functional complexes [4 Taken together, these results suggest that the perip of E. coli PhoQ is important for the interaction Interestingly, co-expression of T18-MgrB and a fu   [36]. The arrows labeled I and II denote potential type I ( [34] ) and type II [35] signal sequence cleavage sites, Western blot of the total cell lysate, envelope fraction (pellet), or the soluble protein fraction (sup), of an mgrB 2 strain expressing cytop CFP (AML20) and containing either a control plasmid (pEB52) or an mgrB expression plasmid (pAL8). Both plasmids also express bet enzyme that resides in the periplasm.  Fluorescence of an mgrB reporter, AML21 (mgrB 2 phoQ 2 ), which was transforme plasmid (pEB52) or a plasmid containing mgrB (pAL compatible plasmid expressing wild-type PhoQ (pLPQ2 which the periplasmic domain of E coli PhoQ was rep corresponding domain from P. aeruginosa PhoQ (pLPQ plasmid (pGB2). Fluorescence was measured from cells g minimal glucose medium with 100 mM MgSO 4 as descri and Methods. Cells were grown at 30uC for consistenc two-hybrid experiments- Figure 5 and Figure S3. Simi observed for cells grown at 37uC (data not shown). represent the range of means of two independent cult type, chim: chimera

Figure 3. Effects of site-directed alanine substitutions of MgrB periplasmic residues on
PhoQ activity. b-galactosidase levels were measured in a ∆mgrB strain containing the PhoQ/PhoP-regulated transcriptional reporter PmgtA-lacZ (AML67) and also containing either an empty vector (pAL39), wild-type MgrB (pAL38), or the indicated mutants (pJC1 through pJC7 and pTG1 through pTG15; please see Table S2 for details). Inducer (IPTG) was not added as the leaky expression from the trc promoter resulted in sufficient levels of protein. Data for wildtype MgrB and each mutant was normalized to the no-MgrB control. Averages and standard deviations for three independent experiments are shown. MU = Miller Units.

Fluorescence (AU)
A B Figure S3. GFP-MgrB expression levels for the TM swap variants relative to wildtype. Cells expressing WT MgrB and variants were imaged by fluorescence microscopy and GFP fluorescence levels for quantified (see Methods for details). Data represent averages and ranges for two independent experiments comprised of at least 50 cells each, AU = Arbitrary Units. Figure S4. GFP-MgrB expression levels for the site-directed TM domain alanine mutants relative to wild-type. GFP fluorescence was quantified as described in the legend for Figure S3, AU = Arbitrary Units.