Multisite phosphorylation regulates GpsB function in cephalosporin resistance of Enterococcus faecalis

Abstract

Enterococci are normal human commensals and major causes of hospital-acquired infections. Enterococcal infections can be difficult to treat because enterococci harbor intrinsic and acquired antibiotic resistance, such as resistance to cephalosporins. In Enterococcus faecalis, the transmembrane kinase IreK, a member of the bacterial PASTA kinase family, is essential for cephalosporin resistance. The activity of IreK is boosted by the cytoplasmic protein GpsB, which promotes IreK autophosphorylation and signaling to drive cephalosporin resistance. A previous phosphoproteomics study identified eight putative IreK-dependent phosphorylation sites on GpsB, but the functional importance of GpsB phosphorylation was unknown. Here we used genetic and biochemical approaches to define three sites of phosphorylation on GpsB that functionally impact IreK activity and cephalosporin resistance. Phosphorylation at two sites (S80 and T84) serves to impair the ability of GpsB to activate IreK in vivo, suggesting phosphorylation of these sites acts as a means of negative feedback for IreK. The third site of phosphorylation (T133) occurs in a segment of GpsB termed the C-terminal extension that is unique to enterococcal GpsB homologs. The C-terminal extension is highly mobile in solution, suggesting it is largely unstructured, and phosphorylation of T133 appears to enable efficient phosphorylation at S80 / T84. Overall our results are consistent with a model in which multisite phosphorylation of GpsB impairs its ability to activate IreK, thereby diminishing signal transduction through the IreK-dependent pathway and modulating phenotypic cephalosporin resistance.

Graphical abstract

INTRODUCTION

Enterococci are Gram-positive bacteria that are commensals of the gastrointestinal (GI) tract of many mammals, including humans. However, enterococci are also opportunistic pathogens and a major cause of hospital-acquired infections [1]. These infections can be difficult to treat in part because enterococci possess intrinsic and acquired resistance to various commonly used antibiotics [2, 3]. In particular, enterococci are intrinsically resistant to cephalosporins, a class of β-lactam antibiotics that cause cell wall stress by inhibiting the crosslinking of bacterial peptidoglycan, but the mechanism of cephalosporin resistance is incompletely understood. Prior therapy with 2^nd or 3^rd generation cephalosporins is a wellknown risk factor for developing enterococcal infections [4]: while other bacteria in the GI tract are inhibited by cephalosporin antibiotics, enterococci persist, proliferate to abnormally high numbers, and eventually disseminate to other organs, resulting in life-threatening infections [5, 6]. Therefore, the development of novel therapies that disable cephalosporin resistance have the potential to be useful (when used in combination with cephalosporins) in preventing or treating enterococcal infections in the future.

One important factor in cephalosporin resistance is the transmembrane serine/threonine kinase IreK, a member of the bacterial PASTA kinase family. Deletion of ireK results in a loss of resistance to 2^nd and 3^rd generation cephalosporins [7–9]. IreK activity is stimulated during normal cell growth and division as well as in response to cell wall stress, such as that induced by cephalosporins [8]. Upon detecting growth signals or cell wall stress, IreK autophosphorylates three threonine residues (T163, T166, T168) on a structural element known as the ‘activation loop’ located on its intracellular kinase domain. These autophosphorylation events are critical to boost IreK kinase activity and phosphorylation of downstream substrates (i.e. signaling): mutation of T163, T166, and T168 to alanines (commonly accepted as a phosphoablative mutation to prevent phosphorylation) resulted in a loss of substrate phosphorylation and cephalosporin resistance like that observed for ΔireK mutants [8]. A phosphoproteomics study identified various putative IreK-dependent substrates [10], but only four have been independently validated thus far: IreB, CroS, MltG, and the cytoplasmic protein GpsB [10–13]. To turn off the signaling cascade, the cognate phosphatase IreP dephosphorylates IreK and its substrates [7, 8, 11, 12].

GpsB is important for cephalosporin resistance. Deletion of gpsB in enterococci decreases cephalosporin resistance in otherwise wild-type cells [13], in part because GpsB directly promotes IreK autophosphorylation and downstream signaling in Enterococcus faecalis [13]. Compared to wild-type, IreK autophosphorylation and substrate phosphorylation are dramatically decreased in ΔgpsB mutants and, conversely, are increased when GpsB is overexpressed [13]. GpsB not only promotes IreK autophosphorylation and downstream signaling in growing cells, but is also required for the increase in IreK autophosphorylation and downstream signaling normally observed upon treatment with ceftriaxone, a representative cephalosporin. The exact molecular mechanism of how GpsB promotes IreK autophosphorylation and signaling in enterococci has not been elucidated.

Bacillus subtilis GpsB is the only other GpsB homolog thus far confirmed to be phosphorylated by a PASTA kinase homolog (B. subtilis PrkC). Specifically, B. subtilis GpsB gets phosphorylated solely at residue T75 (highlighted in Figure S1), and this phosphorylation event impairs the ability of GpsB to promote PrkC autophosphorylation and downstream signaling [14]. Our previous phosphoproteomics study reported that E. faecalis GpsB has eight putative phosphorylation sites (Figure S1) [10]. At least one of these sites is important for phosphorylation of GpsB in E. faecalis cells [10, 13]. Two putative sites in E. faecalis GpsB (S80 and T84) are located in the central region of the protein, similar to B. subtilis GpsB T75 (Figure S1). The other six sites in E. faecalis GpsB (T107, S109, T110, T113, T120, and T133) are found in the GpsB “C-terminal extension”, a region of the protein that is unique to enterococcal GpsB homologs and not found in homologs from other bacterial species (Figure S1) [13]. It seems likely that one or more of the six putative sites in the C-terminal extension contribute to GpsB phosphorylation, because a GpsB mutant in which the C-terminal extension was deleted did not get readily phosphorylated by IreK in vitro [13]. However, the functional consequences of phosphorylation on E. faecalis GpsB have not yet been determined.

No crystal structure of full-length GpsB from any bacterial species has been determined. Independently obtained crystal structures of the N- or C-terminal fragments of GpsB homologs revealed dimers [15, 16] or trimers [15] of alpha-helices, respectively. These results, combined with various biophysical data collected with B. subtilis, Streptococcus pneumoniae, and Listeria monocytogenes GpsB homologs suggested that GpsB oligomerizes as a hexamer and adopts a “trimer of dimers” quaternary structure [15–17]. It remains unknown whether E. faecalis GpsB also adopts a “trimer of dimers” arrangement, how the C-terminal extension might impact its quaternary structure, or if phosphorylation influences the structure of GpsB.

Here we experimentally verified the phosphorylation sites of E. faecalis GpsB and investigated the functional role of GpsB phosphorylation in E. faecalis cells. Using in vitro and in vivo methods, we found that GpsB is phosphorylated at multiple sites (S80, T84 and T133) and that phosphorylation at these sites functionally impacts the ability of GpsB to modulate IreK activity and cephalosporin resistance in E. faecalis.

RESULTS

T84 and T133 are major contributors to GpsB phosphorylation in vitro

Eight candidate sites for phosphorylation on E. faecalis GpsB were identified by phosphoproteomics (Figure S1) [10], and we previously found that simultaneous alanine substitutions of all eight sites (phosphoablative mutations, referred to as the “GpsB 8A mutant”) eliminated GpsB phosphorylation [13]. To define the contribution of each site to GpsB phosphorylation, a panel of phosphoablative mutants with substitutions at individual sites (or a subset of sites) was constructed. These GpsB variants were purified and used as substrates in in vitro phosphorylation assays with the cytoplasmic domain of IreK kinase (His₆-IreK-n). Phosphorylated and total GpsB were visualized after SDS-PAGE using Pro-Q Diamond phosphoprotein gel stain and SYPRO Ruby total protein gel stain, respectively. As observed previously [13], wild-type GpsB became phosphorylated in the presence of His₆-IreK-n, but the GpsB 8A mutant did not (Figures 1, S2). Increases in Pro-Q Diamond signal were observed over time in the presence of His₆-IreK-n for all individual GpsB phosphoablative mutants, except for those with T133A mutations (Figure 1A, S2A). The phosphorylation rates of the GpsB S80A and GpsB T120A mutants were not significantly different from wild-type GpsB, indicating that phosphorylation of GpsB is not dependent on either of these residues individually (Figure 1B, S2B). The phosphorylation rate of GpsB T84A was ~50% that of wild-type GpsB and combination of the S80A and T84A mutations (GpsB S80A T84A) had an additive impact, decreasing the rate of phosphorylation to ~30% that of wild-type GpsB (Figure 1B). The phosphorylation rate of the GpsB T107A S109A T110A T113A mutant (these 4 sites were analyzed as a group based on in vivo phenotypic data; see below) was ~65% that of wild-type GpsB, indicating that phosphorylation of GpsB in vitro is partially dependent on these residues collectively (Figure S2B). The phosphorylation rates for the GpsB T133A mutants were the same as the rate of the GpsB 8A mutant (Figure 1B), suggesting that phosphorylation at T133 is required to allow for the subsequent phosphorylation at the other sites, at least in vitro.

Figure 1. Phosphorylation of GpsB S80, T84, and T133 mutants by IreK in vitro.

In vitro phosphorylation assays were performed with purified proteins to assess the phosphorylation of GpsB phosphoablative mutants by IreK. GpsB (wild-type (WT) or phosphoablative mutants) was incubated in the presence or absence of His₆-IreK-n at 37°C. Samples of each reaction were taken at intervals after the addition of excess ATP. (a) Following SDS-PAGE, Pro-Q Diamond Phosphoprotein Gel Stain was used to visualize phosphorylated GpsB (GpsB-P), and SYPRO Ruby protein gel stain was used to visualize total GpsB. Gel images are representative of at least three independent experiments. (b) The phosphorylated and total GpsB signals from the gel images were quantified. Relative GpsB phosphorylation for each timepoint was calculated by determining the ratios of phosphorylated GpsB to total GpsB for each timepoint and graphed. Initial relative GpsB phosphorylation rates were determined by finding the slopes of the best fit lines from timepoints 0 – 7.5 minutes for each reaction. Data are the mean x-fold changes in GpsB phosphorylation rates for each reaction relative to the wild-type GpsB reaction from at least 3 independent experiments. Error bars represent one standard deviation. *, p < 0.05; **, p < 0.01; ns, not significant; student’s T-test compared to WT (heteroscedastic, two-tailed).

To determine which of the eight sites are sufficient for phosphorylation of GpsB in vitro, we adopted a reciprocal approach and constructed a panel of GpsB variants in which only one of the candidate sites remained wild-type (that is, each variant simultaneously carried phosphoablative alanine mutations at seven of the eight candidate sites, leaving one wild-type site; referred to here as “7A mutants”). The eight 7A mutants were purified and in vitro phosphorylation assays were performed as described previously [13]. Despite extended incubation times, only the T133 7A mutant became phosphorylated by His₆-IreK-n (Figure S3), suggesting that phosphorylation at T133 precedes and is required for phosphorylation at other sites on GpsB in vitro. Taken together, the in vitro phosphorylation studies indicate that T84 and T133 are the major contributors to IreK-dependent phosphorylation of GpsB, while S80 and one or more of T107, T109, T110, and T113 appear to play a minor role.

The C-terminal extension of GpsB is highly flexible, regardless of phosphorylation

One of the major sites of GpsB phosphorylation, T133, is located within the C-terminal extension, a region of the protein found in enterococcal GpsB homologs, but not in GpsB homologs of other bacterial species. The structure of E. faecalis GpsB has not been solved, but the C-terminal extension is predicted by MobiDB-Lite [18] to be intrinsically disordered, suggesting that this region may be highly flexible, which might facilitate its phosphorylation by IreK. To investigate the flexibility of the C-terminal extension and the impact of GpsB phosphorylation on that mobility, we used site-directed spin labeling electron paramagnetic resonance (EPR) spectroscopy. In this approach, unique cysteine substitutions are engineered into the protein of interest. The resulting single cysteine variants are purified and labeled with MTSL to create the R1 side chain that can be detected with continuous wave (CW) EPR spectroscopy to provide information about the motion of the label. We created a panel of GpsB variants in which a single cysteine was introduced in the C-terminal extension (V128C or C137, added to the end of GpsB) and in two predicted alpha-helical domains of GpsB that are conserved among GpsB homologs (S49C or S93C). Each single cysteine variant was purified, phosphorylated (or not) by purified His₆-IreK-n, spin labeled at the unique cysteine, and analyzed by CW EPR spectroscopy.

The EPR spectra of the unphosphorylated GpsB variants revealed marked differences in motion of the R1 side chain that depended on its location in GpsB. In particular, R1 in the predicted alpha-helical structured regions of GpsB (S49R1 or S93R1) exhibited intermediate motion, consistent with side chains that are constrained within folded secondary structures (Figure 2A). In contrast, R1 in the C-terminal extension (V128R1 or C137R1) exhibited very fast motion, indicating that the GpsB C-terminal extension is highly mobile in solution and does not adopt any fixed conformation.

Figure 2. The GpsB C-terminal extension is highly flexible, regardless of phosphorylation.

(a) Overlays of X-band CW EPR spectra of apo (black) and phosphorylated (gray) GpsB mutants. The overlaid spectra were recorded over 100 G and are normalized to the same center line height. Arrows indicate motional changes. (b) Q-band DEER distance distributions for apo (black) and phosphorylated (gray) GpsB S93R1. The corresponding fits to the background-corrected dipolar evolution data are shown in the inset.

Phosphorylation of GpsB by His₆-IreK-n led to slight changes in the motion of each of the sites studied, suggestive of some phosphorylation-dependent change in the conformational ensemble of the predicted helical domains and the C-terminal extension (Figure 2A). However, even in the phosphorylated state, V128R1 and C137R1 still exhibited fast motion, so these data do not indicate that phosphorylation causes the C-terminal extension to adopt any structured conformation.

Based on the crystal structures of fragments of GpsB homologs and small angle x-ray scattering and SEC-MALS analyses of full-length GpsB homologs, GpsB has been proposed to adopt a “trimer-of-dimers” arrangement [15–17]. To determine if phosphorylation might alter the quaternary structure of GpsB, we analyzed GpsB S93R1 by double electron-electron resonance (DEER) spectroscopy, an approach that enables the measurement of distances between spin labels in proteins or protein complexes (such as GpsB oligomers). The distance distribution for GpsB S93R1 revealed that the majority of the population exhibited a distance between spin labels of 21.8 Å, and GpsB phosphorylation did not alter the distance distribution (Figure 2B). Although phosphorylation does not appear to influence the quaternary structure of GpsB, it is perhaps worth noting that the observed distance distribution is not consistent with the proposed “trimer-of-dimers” configuration proposed for GpsB homologs in other bacterial species [15–17]. Hence under the conditions used here, E. faecalis GpsB does not appear to exist in a “trimer-of-dimers” state in solution.

T84 and T133 are major contributors to GpsB phosphorylation in vivo

Results from the in vitro phosphorylation assays indicated that T84 and T133 are particularly important for GpsB phosphorylation, while S80 may also play a role. To determine the extent to which these residues contribute to GpsB phosphorylation in vivo, GpsB-His₆ (wild-type or phosphoablative mutants) was enriched from exponentially-growing E. faecalis cells using immobilized metal affinity chromatography. Following SDS-PAGE, phosphorylation of the enriched proteins was determined as described above for the in vitro phosphorylation assays. Because phosphorylation of GpsB may alter its ability to promote IreK autophosphorylation in vivo (which would therefore alter the activity of IreK towards substrates, potentially including GpsB itself) [13], we performed these experiments in a mutant strain of E. faecalis in which the IreK phosphorylation status is “fixed” by phosphomimetic substitutions on the IreK activation loop (ireK T163E T166E T168E mutant). These phosphomimetic substitutions prevent changes in the autophosphorylation status of the IreK activation loop and thereby maintain IreK in an activated state [8], enabling us to decouple the potential effect of GpsB phosphorylation site variants on IreK autophosphorylation from the ability of GpsB to become phosphorylated itself.

Wild-type GpsB and phosphoablative mutants were expressed at comparable levels in E. faecalis and successfully enriched from cell lysates (Figure 3, S4). Staining with Pro-Q Diamond revealed a strong signal for wild-type GpsB that was not present for the GpsB 8A mutant, indicating phosphorylation of GpsB in vivo (Figure 3, S5). Under the conditions of our experiment, we estimated approximately 9.4% of wild-type GpsB was phosphorylated (Fig S5). Some residual Pro-Q Diamond signal was present even for the GpsB 8A mutant, likely a result of nonspecific background interaction of the Pro-Q Diamond stain with GpsB protein (as observed with unphosphorylated, recombinant GpsB, Figure 1, S2). GpsB S80A was phosphorylated as well as wild-type GpsB, whereas the GpsB T84A and GpsB S80A T84A mutants were phosphorylated only ~50% as well as wild-type GpsB, consistent with the model that T84 is a major contributor to GpsB phosphorylation in E. faecalis. In contrast to what was observed in vitro where T133A mutants did not get detectably phosphorylated (Figure 1), the GpsB T133A and GpsB S80A T84A T133A mutants were phosphorylated ~47% and ~30% as well as wild-type GpsB, respectively (Figure 3). Overall, these data indicate that T84 and T133 are both key contributors to GpsB phosphorylation in vivo, but because the GpsB S80A T84A T133A mutant was phosphorylated more than the GpsB 8A mutant, one (or more) of the other sites appears to modestly contribute to phosphorylation in vivo as well.

Figure 3. T84 and T133 are key contributors to GpsB phosphorylation in vivo.

Ectopically-expressed GpsB-His₆ (wild-type (WT) or phosphoablative mutants) were enriched from exponentially-growing E. faecalis ireK T163E T166E T168E cells using immobilized metal affinity chromatography. (a) Immunoblotting was performed on the input fractions (whole cell lysates) using antisera to detect GpsB-His₆. The elution fractions were separated by SDS-PAGE, after which phosphorylated GpsB (GpsB-His₆-P) and total GpsB was analyzed using Pro-Q Diamond Phosphoprotein Gel Stain and SYPRO Ruby Protein Gel Stain, respectively. Images are representative of two independent experiments. (b) Relative GpsB phosphorylation for each GpsB-His₆-enriched sample was determined by calculating the signal intensity ratios of phosphorylated GpsB-His₆ to total GpsB-His₆ and then normalizing the ratios of the phosphoablative mutants to wild-type GpsB. Graphed are the mean relative GpsB-His₆ phosphorylation values from two independent experiments. Error bars represent one standard deviation. *, p < 0.05; **, p < 0.01; ns, not significant; student’s T-test compared to WT (heteroscedastic, two-tailed). Strains were GpsB WT, NV13/pNEM4; GpsB 8A mutant, NV13/pNEM29; GpsB S80A mutant, NV13/pNEM17; GpsB T84A mutant, NV13/pNEM7; GpsB S80A T84A, NV13/pNEM48; GpsB T133A mutant, NV13/pNEM11; GpsB S80A T84A T133A, NV13/pNEM53.

Mutations at GpsB residues S80, T84, and T133 alter cephalosporin resistance

To determine the impact of GpsB phosphorylation at specific sites on the physiologically relevant phenotype of cephalosporin resistance, antibiotic susceptibility assays were performed with E. faecalis strains ectopically expressing phosphoablative or phosphomimetic GpsB variants. As reported previously [13], expression of wild-type GpsB led to a robust increase in resistance to ceftriaxone, a representative cephalosporin, of the ΔgpsB mutant (Table 1). Expression of the GpsB 8A mutant resulted in decreased resistance compared to wild-type GpsB (64-fold decrease, Table 1), suggesting that phosphorylation of GpsB is required for cephalosporin resistance.

Table 1.

GpsB phosphoablative and phosphomimetic mutations alter cephalosporin resistance.

GpsB-His6 gene b	Ceftriaxone MIC (μg/mL) a
None	2
Wild-type	512
	T/S➔A c	T/S➔E d
All 8 phosphorylation sites	8	nd e
S80	256	16
T84	32	4
S80 T84	8	2
T107	512	256
S109	512	512
T110	512	256
T113	512	256
T107 S109 T110 T113	256	nd
T120	512	512
T133	64	64
S80 T84 T133A	8	2

^aMinimal inhibitory concentrations (MIC) for ceftriaxone. Data represent the median MIC from at least 3 independent biological replicates.

^bGpsB-His₆ wild-type or mutant genes were ecmiddleically expressed in Enterococcus faecalis ΔgpsB.

^cPhosphoablative mutations were made by changing the threonine or serine residues to alanine.

^dPhosphomimetic mutations were made by changing the threonine or serine residues to glutamic acid.

^end = not done

To define the importance of specific sites of phosphorylation, we tested GpsB variants with substitutions at each individual site. Expression of GpsB phosphoablative or phosphomimetic T107, S109, T110, T113 (individually or together), or T120 variants resulted in essentially no change in ceftriaxone resistance compared to wild-type GpsB (Table 1). Although T107, S109, T110, or T113 may be phosphorylated in vitro (Figure S2), these data indicate that phosphorylation at these sites does not substantially impact the ability of GpsB to promote cephalosporin resistance, and as a result we did not investigate these residues further. These results are unsurprising given that T107, S109, T110, T113 and T120 are not well conserved among enterococcal GpsB homologs (Figure S1).

In contrast, substitutions at S80, T84, and T133 all resulted in changes to ceftriaxone resistance. Both phosphoablative and phosphomimetic substitutions at S80 and T84 led to reduced ceftriaxone resistance (Table 1), although the magnitude of the effect was not the same at each site or with each substitution. The most pronounced impact was observed with the phosphomimetic T84E substitution, which led to a 128-fold reduction in resistance. Given the proximity of S80 to T84, we considered that substitutions at one site might affect phosphorylation of the other. To determine if this was the case, we constructed double mutants simultaneously carrying substitutions at both sites. The effects of the double mutants were additive with respect to the single mutants, suggesting that each site contributes to ceftriaxone resistance at least in part independently. Expression of the GpsB S80A T84A mutant resulted in the same level of resistance as expression of the GpsB 8A mutant, suggesting that the functional impairment of the GpsB 8A mutant is specifically driven by the alanine mutations at S80 and T84. Despite being expressed as well as, if not better than wild-type GpsB (Figure S4), the GpsB S80E T84E mutant was entirely defective as the ceftriaxone resistance was not different from that of the ΔgpsB mutant. These data suggest that phosphorylation at T84 (possibly in combination with phosphorylation at S80) plays a key role in regulation of cephalosporin resistance of E. faecalis, and moreover that constitutive “phosphorylation” at T84 (and S80) completely prevents GpsB from promoting cephalosporin resistance.

Our in vitro data suggested that phosphorylation of T133 might serve to enable phosphorylation of other sites on GpsB. To test this in E. faecalis, we analyzed ceftriaxone resistance of T133 single variants compared to triple mutants in which S80 and T84 were substituted in combination with T133. In both cases, the phenotype of the triple mutant mirrored that of the corresponding S80 / T84 double mutant (Table 1; for example, compare the S80A T84A mutant vs S80A T84A T133A mutant), indicating that although the T133 mutant had a phenotypic effect when present alone, it was not additive with substitutions at S80 / T84. These results are consistent with a model in which a primary role for phosphorylation at T133 is to enable enhanced phosphorylation at S80 and T84.

Phosphomimetic substitutions at S80 and T84 impair the ability of GpsB to promote IreK signaling

Wild-type GpsB promotes IreK autophosphorylation and downstream signaling in exponentially-growing E. faecalis cells to drive cephalosporin resistance [13], but the extent to which phosphorylation of GpsB at specific sites impacts the ability of GpsB to promote IreK signaling was unknown. Because phosphomimetic substitutions at GpsB S80 and T84 results in a loss of ceftriaxone resistance (Table 1), we hypothesized that phosphorylation at S80 and T84 might inhibit the ability of GpsB to promote IreK autophosphorylation and signaling in E. faecalis cells. To test this, we used Phos-tag SDS-PAGE as described previously [10, 13] to analyze IreK autophosphorylation (which reflects IreK activation) and MltG phosphorylation (which reflects IreK downstream signaling, because MltG is a substrate of IreK [10]) of strains expressing either phosphoablative or phosphomimetic GpsB mutants. Phos-tag enables the separation of phosphorylated proteoforms from nonphosphorylated proteoforms during SDS-PAGE, an approach we have previously used to demonstrate that IreK autophosphorylation and signaling were significantly decreased in the ΔgpsB mutant and that ectopic expression of wildtype GpsB in the ΔgpsB mutant restored IreK autophosphorylation and signaling [13].

Expression of the GpsB S80A mutant resulted in no significant difference in IreK or MltG phosphorylation compared to wild-type GpsB (Figure 4A). Expression of the GpsB S80E mutant led to a modest decrease in IreK autophosphorylation, but no significant change in MltG phosphorylation compared to wild-type GpsB (Figure 4B). Because phosphorylation of MltG is relatively low in wild-type cells, a small decrease in IreK autophosphorylation may not result in a significant decrease in MltG phosphorylation. Expression of the GpsB T84A or GpsB S80A T84A mutants resulted in decreased IreK phosphorylation compared to wild-type GpsB and MltG phosphorylation was nearly absent (Figure 4C and 4E). Expression of the GpsB T84E and GpsB S80E T84E mutants resulted in a greater decrease in both IreK and MltG phosphorylation (Figure 4D and 4F), similar to a ΔgpsB mutant [13]. Overall, these results correlate well with the ceftriaxone resistance phenotypes of these strains (Table 1), indicating that the levels of IreK signaling are driving cephalosporin resistance. These data also suggest that phosphorylation at T84 impairs the ability of GpsB to promote IreK autophosphorylation and downstream signaling in vivo.

Figure 4. Mutations at GpsB S80 and T84 impair IreK phosphorylation and signaling.

Total protein lysates were prepared from exponentially-growing cells, and pairwise comparisons were subjected to Phos-tag SDS-PAGE followed by immunoblot analysis using antiserum to detect IreK or MltG (an IreK substrate). (a) ΔgpsB/p-gpsB WT-His₆ (WT = wild-type) and ΔgpsB/pgpsB S80A-His₆; (b) ΔgpsB/p-gpsB WT-His₆ and ΔgpsB/p-gpsB S80E-His₆; (c) ΔgpsB/p-gpsB WT-His₆ and ΔgpsB/p-gpsB T84A-His₆; (d) ΔgpsB/p-gpsB WT-His₆ and ΔgpsB/p-gpsB T84E-His₆; (e) ΔgpsB/p-gpsB WT-His₆ and ΔgpsB/p-gpsB S80A T84A-His₆; (f) ΔgpsB/p-gpsB WT-His₆ and ΔgpsB/p-gpsB S80E T84E-His₆. Immunoblot images are representative of at least 3 biological replicates per strain or condition. Bar graph data show the average % IreK or MltG phosphorylation (% IreK-P or % MltG-P, respectively) of the 3 replicates. IreK can be phosphorylated at multiple sites and fractionates as multiple bands, all of which were quantified together as indicated. Error bars represent one standard deviation. *, p < 0.05; **, p < 0.01; ns, not significant; student’s T-test (heteroscedastic, two-tailed). Strains were ΔgpsB/p-gpsB-His₆, JL635/pNEM4; ΔgpsB/p-gpsB S80A-His₆, JL635/pNEM17; ΔgpsB/p-gpsB S80E-His₆, JL635/pNEM18; ΔgpsB/p-gpsB T84A-His₆, JL635/pNEM7; ΔgpsB/p-gpsB T84E-His₆, JL635/pNEM8; ΔgpsB/p-gpsB S80A T84A-His₆, JL635/pNEM48; ΔgpsB/p-gpsB S80E T84EHis₆, JL635/pNEM59; ΔireK/vector, JL206/pJRG9; and ΔmltG/vector, JL650/pJRG9.

GpsB T84E is unable to promote IreK phosphorylation and signaling in response to cell wall stress

Wild-type GpsB is required for the increase in IreK autophosphorylation and signaling upon exposure to cell wall stressors, such as ceftriaxone [8, 13]. To determine if phosphorylation at S80 or T84 impairs this function of GpsB, exponentially-growing E. faecalis cells expressing the GpsB phosphoablative or phosphomimetic mutants were treated (or not) with ceftriaxone before harvesting cell lysates. Phos-tag SDS-PAGE was then performed to analyze IreK and MltG phosphorylation as before. The GpsB S80A, S80E, T84A, and S80A T84A mutants exhibited significant increases in both IreK and MltG phosphorylation upon treatment with ceftriaxone (Figure S6). In contrast, the GpsB T84E and S80E T84E mutants exhibited minimal increases in IreK phosphorylation and no significant increases in MltG phosphorylation upon ceftriaxone treatment (Figure S6D and S6F). These results suggest that phosphorylation at GpsB T84 impairs the ability of IreK to respond to cell wall stress induced by cephalosporins.

Role of phosphorylation at GpsB residue T133 on IreK signaling

To determine if phosphorylation at T133 alters the ability of GpsB to promote IreK activation and signaling in vivo, IreK and MltG phosphorylation in strains expressing the GpsB T133A or T133E mutants were analyzed using Phos-tag SDS-PAGE. IreK and MltG phosphorylation were significantly increased in strains expressing either the GpsB T133A or GpsB T133E mutants compared to the strain expressing wild-type GpsB (Figure 5). The mutant strains also exhibited increased IreK and MltG phosphorylation upon ceftriaxone treatment, indicating that IreK was still able to respond to cell wall stress (Figure S7). The observation that the phenotypes resulting from either T133A or T133E substitutions are the same here and with regard to ceftriaxone resistance (Table 1) suggests that the T133E substitution in GpsB is not behaving as a phosphomimetic, but is instead acting as a phosphoablative substitution. Moreover, the observation that IreK phosphorylation and downstream signaling is enhanced in the T133 mutants suggests that T84 is poorly phosphorylated in the mutants. Thus, these results are consistent with a model in which phosphorylation at GpsB T133 of wild-type GpsB impairs IreK phosphorylation and downstream signaling, in part by enabling GpsB phosphorylation at T84 and S80.

Figure 5. GpsB T133 mutants exhibit increased IreK phosphorylation and signaling.

Total protein lysates were prepared from exponentially-growing cells, and pairwise comparisons were subjected to Phos-tag SDS-PAGE followed by immunoblot analysis using antiserum to detect IreK or MltG. (a) ΔgpsB/p-gpsB WT-His₆ (WT = wild-type) and ΔgpsB/p-gpsB T133A-His₆; (b) ΔgpsB/p-gpsB WT-His₆ and ΔgpsB/p-gpsB T133E-His₆. Immunoblot images are representative of at least 3 biological replicates per strain or condition. Bar graph data show the average % IreK or MltG phosphorylation (% IreK-P or % MltG-P, respectively) of the 3 replicates. IreK can be phosphorylated at multiple sites and fractionates as multiple bands, all of which were quantified together as indicated. Error bars represent one standard deviation. *, p < 0.05; ns, not significant; student’s T-test (heteroscedastic, two-tailed). Strains were ΔgpsB/p-gpsB-His₆, JL635/pNEM4; ΔgpsB/p-gpsB T133A-His₆, JL635/pNEM11; ΔgpsB/p-gpsB T133E-His₆, JL635/pNEM12; ΔireK/vector, JL206/pJRG9; and ΔmltG/vector, JL650/pJRG9.

DISCUSSION

In a previous study, we used phosphoproteomics to identify eight putative sites of IreK-dependent phosphorylation on E. faecalis GpsB [10]. Here we comprehensively interrogated GpsB phosphorylation in vitro and in E. faecalis cells using complementary genetic and biochemical methods, revealing that three of the identified phosphorylation sites (S80, T84, and T133) are functionally important for proper regulation of both IreK activity and enterococcal cephalosporin resistance. Our results indicate that T84 and T133 are the major contributors to GpsB phosphorylation and function, and that S80 plays an additive but minor role. Thus, the activity of E. faecalis GpsB is controlled via multisite phosphorylation, a feature that distinguishes E. faecalis GpsB from the other GpsB homolog (B. subtilis GpsB) for which phosphorylation has been studied, in that B. subtilis GpsB is phosphorylated only at a single site [14]. Moreover, the functionally-important phosphorylation of E. faecalis GpsB at T133 occurs in a segment of the protein we termed the “C-terminal extension” because only enterococcal GpsB homologs encode this ~25 amino acid segment at the C-terminal end of the protein. Hence phosphorylation-dependent regulation of the C-terminal extension may represent a novel mechanism for control of GpsB activity among the enterococci.

Sequence and structural analyses of GpsB homologs from other bacterial species revealed the presence of N-terminal and C-terminal alpha helices [15–17]. Although a crystal structure of an intact full-length GpsB homolog has not yet been determined, structural and biophysical studies of GpsB homologs from other bacterial species yielded a model in which GpsB is proposed to oligomerize as a hexamer and form a “trimer of dimers” arrangement, where the N-terminal helices form 3 dimers, and the C-terminal helices form 2 trimers [15–17]. However, none of the GpsB homologs analyzed in those studies were from an enterococcal species; hence it remains unknown if enterococcal GpsB homologs, with their ~25 residue C-terminal extension, also adopt this arrangement. Our DEER analyses (Figure 2B) revealed that the distances between spin labels located in the predicted C-terminal helices (S93R1) are longer than predicted for the trimeric C-terminal helices seen in the existing GpsB crystal structure [15] and without distances that would be expected between the trimer of dimers, suggesting that – at least under the conditions used here – E. faecalis GpsB does not adopt such a “trimer of dimers” quaternary structure. Phosphorylation of GpsB had no effect on the spin label distances observed (Figure 2), indicating that phosphorylation does not result in changes in the GpsB quaternary structure. This is consistent with previous work on B. subtilis GpsB [17] where size exclusion chromatography experiments revealed that B. subtilis GpsB T75D and T75E mutants exhibited identical retention volumes to wild-type GpsB, suggesting that the phosphomimetic substitutions did not alter the oligomerization of GpsB.

The C-terminal extension of E. faecalis GpsB is predicted by MobiDB-Lite [18] to be intrinsically disordered, suggesting that this region may be highly flexible, which might facilitate its phosphorylation by IreK. Consistent with this prediction, we found by CW EPR that the C-terminal extension is highly flexible and does not adopt any fixed conformation (Figure 2A). Moreover, phosphorylation of GpsB did not substantially alter the mobility of the C-terminal extension, indicating that phosphorylation does not induce an intrinsic structural change in the C-terminal extension of recombinant GpsB under the conditions tested. We speculate that phosphorylation of the C-terminal extension might instead alter protein-protein interactions or cellular localization of GpsB in E. faecalis cells, possibilities that will be tested in future work.

By analyzing both IreK and substrate (MltG) phosphorylation in response to physiologically relevant stimuli in live E. faecalis cells, we revealed the functional impact of three GpsB phosphorylation sites on signal transduction by a PASTA kinase and on cephalosporin resistance in E. faecalis. Our data indicate that phosphorylation at sites near the middle of the GpsB protein (primarily T84, but also S80) leads to (i) a substantial reduction in IreK activity in vivo (Figure 4D and 4F); (ii) a nearly complete loss of the ability of IreK to respond to cell wall stress (Figure S6D and S6F); and (iii) a nearly complete loss of cephalosporin resistance (Table 1). In particular, phosphomimetic substitutions at either T84 or S80 led to reductions in IreK phosphorylation, MltG phosphorylation, and ceftriaxone resistance. The magnitude of these effects was greater for the T84E mutant than for S80E, but simultaneous phosphomimetic substitutions at both sites was additive, in that the magnitude of the effect was greater for the double mutant than for either of the individual single mutants, suggesting that phosphorylation of GpsB at both T84 and S80 work cooperatively to influence IreK activity. Overall these results are reminiscent of studies in B. subtilis where phosphomimetic substitutions at the single site of B. subtilis GpsB phosphorylation (GpsB T75E and GpsB T75D) significantly decreased PrkC autophosphorylation in vivo [14]. Thus, phosphorylation of residues at this ‘central’ location in GpsB may be a general strategy for negative feedback on PASTA kinase signaling.

We note that the phosphoablative alanine substitution at GpsB T84 of E. faecalis also led to reduced levels of IreK signaling and ceftriaxone resistance (Figure 4C and 4E, Table 1). Such an effect was not reported for the GpsB T75A mutant in B. subtilis [14]. The magnitude of the reductions for the E. faecalis T84A mutant are smaller than for the phosphomimetic T84E mutant, and importantly the GpsB T84A substitution did not impair the ability of IreK to respond to cell wall stress (Figure S6C and S6E). While it remains unclear why the E. faecalis T84A mutant exhibits these defects, we speculate that it may result from the structural differences between an alanine side chain and an unphosphorylated threonine side chain. For example, perhaps the hydroxyl group of the threonine side chain of residue 84 participates in a hydrogen bonding interaction between the unphosphorylated GpsB protein and IreK (thereby helping unphosphorylated GpsB to stimulate IreK activity), a bond that cannot be formed with the alanine side chain. Future studies to resolve the structure of full-length GpsB and its interaction with IreK will be required to clarify this.

Our data also indicate that E. faecalis GpsB gets phosphorylated at T133, located in the C-terminal extension that is unique to enterococcal GpsB homologs, and suggest that phosphorylation at T133 is a key first step to facilitate efficient phosphorylation at S80 and T84. For example, the T133A mutant does not get phosphorylated by IreK in vitro (Figure 1), despite possessing wild-type sites at S80 and T84. While some phosphorylation of the T133A mutant occurs in E. faecalis cells, the amount of phosphorylation attributable to S80 and T84 on the T133A mutant is reduced compared to otherwise wild-type GpsB (Figure 3; ~50% reduction between wild-type and the S80A T84A mutant, but only ~20% reduction between the T133A mutant and the S80A T84A T133A mutant), suggesting that S80 and T84 are not phosphorylated efficiently in vivo either when T133 cannot be phosphorylated. Note that unlike substitutions at S80 or T84, both alanine and glutamic acid substitutions at T133 result in identical phenotypes (Figure 5, Figure S7, Table 1), indicating that both substitutions at T133 are acting as phosphoablative mutations. The functional consequences of loss of phosphorylation at T133 are consistent with the expected outcome from absence of efficient phosphorylation at S80 and T84 – namely, IreK activity is enhanced by T133A substitutions, presumably because S80 and T84 cannot be efficiently phosphorylated to cause a reduction in IreK activity. This is supported by the phenotype of the S80E T84E T133A triple mutant (Table 1), where introduction of phosphomimetic substitutions at S80 and T84 – effectively bypassing any need for initial T133 phosphorylation - led to a substantial reduction in ceftriaxone resistance of the T133A mutant. Together the data suggest that a key role for phosphorylation at T133 is to modulate the extent of phosphorylation of S80 and T84, thereby influencing the activity of IreK and ultimately, cephalosporin resistance.

Why does the T133A mutant exhibit a defect in ceftriaxone resistance compared to wildtype (Table 1)? The activity of IreK is elevated in the T133A mutant (Figure 5) and the levels of IreK signaling usually strongly correlate with cephalosporin resistance [7, 8]. Although one function of GpsB is to promote enhanced IreK activity, we previously hypothesized that GpsB possesses an as-yet-unidentified additional function that is independent and downstream of IreK signaling [13]. We speculate that phosphorylation at T133 is important for GpsB to carry out this second function. In other words, despite elevated signaling through IreK, GpsB T133A cannot fulfill its additional function in the downstream signaling network to drive cephalosporin resistance. Although it remains unclear what this second function of GpsB is in E. faecalis, GpsB homologs in other Gram-positive bacteria have been shown to interact with penicillin-binding proteins and facilitate their proper localization in the cell [15, 16, 19–21]. Several penicillin-binding proteins are essential for cephalosporin resistance in E. faecalis [22–24], so one possibility is that E. faecalis GpsB promotes cephalosporin resistance by interacting with these penicillin-binding proteins to drive their efficient localization to sites of cell wall synthesis. Future studies will test this possibility.

In conclusion, we propose the following model for the roles of phosphorylation at S80, T84, and T133 on GpsB function in E. faecalis (Figure 6): Unphosphorylated GpsB promotes IreK autophosphorylation and downstream signaling. Activated IreK phosphorylates GpsB at T133, which alters either the conformation or interaction partners of GpsB to expose residues S80 and T84. IreK phosphorylates the S80 and T84 residues, which then releases GpsB and enables it to perform the second function downstream of IreK signaling. Without promotion of IreK autophosphorylation by GpsB, IreP dephosphorylates the kinase to turn off IreK signaling. Overall, the findings described here give new insight into the regulation of GpsB function and IreK signaling and suggest that GpsB could be a target for future therapeutics to disable cephalosporin resistance in enterococci.

Figure 6. Model of how GpsB phosphorylation impacts GpsB function.

(a) In the presence of stimuli that activate IreK, unphosphorylated GpsB promotes IreK autophosphorylation at the activation loop, leading to the activation of IreK to initiate a signaling cascade by phosphorylating substrates. (b) IreK phosphorylates GpsB T133, inducing a change in the protein conformation or interaction partner(s) to expose GpsB residues S80 and T84. (c) IreK phosphorylates GpsB S80 and T84, disabling GpsB from promoting IreK autophosphorylation and allowing GpsB to perform its second function downstream of IreK in the signaling pathway. Without the promotion of IreK autophosphorylation by GpsB, IreP rapidly dephosphorylates IreK to turn off the signaling cascade.

MATERIALS AND METHODS

Bacterial strains, growth, media, and chemicals.

Bacterial strains and plasmids used in this study are listed in Table S1. E. coli strains were grown in Lysogeny broth (LB) and E. faecalis strains were grown in Mueller-Hinton broth (MHB) (prepared according to the manufacturer’s instructions; Difco) for all experiments. Archives of bacterial strains were kept at −80°C in MHB or LB with 30% glycerol. Antibiotics were used at the following concentrations: kanamycin, 50 μg/mL; chloramphenicol, 10 μg/mL.

Construction of plasmids.

All recombinant plasmids were constructed using Gibson assembly [25] and sequencing of the full insert was performed to confirm absence of errors. Ectopic expression of GpsB (wild-type and mutants) in E. faecalis OG1 strains was accomplished using the enterococcal expression plasmid pJRG9 under the control of the constitutive P23s promoter. The native ribosome binding site for GpsB was included. Expression of wild-type and mutant GpsB in E. coli for protein purification was accomplished using the IPTG-inducible expression vector pET28a-His-smt3, which encodes a His₆-SUMO fusion.

Antibiotic susceptibility assays.

Minimal inhibitory concentrations (MICs) were determined using a broth microdilution method. Briefly, bacteria from stationary-phase cultures were inoculated into wells containing 2-fold serial dilutions of antibiotics in MHB supplemented with 10 μg/mL chloramphenicol (for plasmid maintenance) at a normalized optical density at 600 nm (OD₆₀₀) of 4 × 10⁻⁵ (~1 × 10⁵ CFU). Plates were incubated in a Bioscreen C plate reader at 37°C for 24 h with brief shaking before each measurement. The OD₆₀₀ was determined every 15 min, and the lowest concentration of antibiotic that prevented growth was recorded as the MIC.

Enrichment of GpsB-His₆ from E. faecalis cells.

GpsB-His₆ was enriched from exponentially-growing E. faecalis ΔgpsB cells using immobilized metal affinity chromatography as previously described [10]. The input and elution fractions were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Immunoblotting was performed with input samples (whole cell lysates) using antisera to detect GpsB. Phosphorylated and total GpsB in the elution samples were visualized using Pro-Q Diamond Phosphoprotein Gel Stain (Invitrogen) and SYPRO Ruby Protein Gel Stain (Invitrogen), respectively, according to the manufacturer’s instructions.

Whole cell lysate preparation.

Whole cell lysates were collected from ΔgpsB E. faecalis cells harboring the vector alone or plasmid expressing GpsB-His₆ (wild-type or various mutants) as described previously [13]. Cell lysates used for GpsB immunoblot analysis were additionally boiled for 5 minutes, but lysates used for Phos-tag analysis were not.

GpsB immunoblot analysis.

Proteins were separated by SDS-PAGE at 200 V for 45 minutes followed by transfer to polyvinylidene fluoride (PVDF) membranes using a Transblot Turbo system at 25 V for 7 minutes. After transfer, membranes were blocked in 5% milk for 1 hour at room temperature. Membranes were probed with custom primary antiserum to detect GpsB followed by HRP-conjugated secondary antibody before developing with Super-Signal (Thermo Fisher Scientific) using a ChemiDoc touch imaging system (BioRad, Hercules, CA).

Phos-tag SDS-PAGE and IreK/MltG immunoblot analysis.

Phos-tag SDS-PAGE and immunoblot analysis for IreK and MltG were performed as described previously [13].

Protein purification.

Purification of recombinant His₆-SUMO-GpsB (wild-type or various mutants) and wild type His₆-IreK-n (intracellular domain only) were performed as described previously [13].

In vitro GpsB phosphorylation assays.

Purified wild-type or mutant GpsB proteins (14.2 μM) were pre-incubated in the presence or absence of recombinant wild-type His₆-IreK-n (0.33 μM or 1 μM) in a buffer composition of 50 mM Tris [pH = 7.5], 25 mM NaCl, and 5 mM MgCl₂ for 5 minutes at 37°C. 2 mM ATP was added to initiate the reactions after which samples were removed at intervals. The reactions were quenched by the addition of 5X SDS-PAGE Laemmli buffer and boiled for 5 minutes. The samples were subjected to SDS-PAGE at 200 V for 45 minutes. The gels were stained with Pro-Q Diamond Phosphoprotein Gel Stain (Invitrogen) followed by SYPRO Ruby Protein Gel Stain (Invitrogen) according to the manufacturer’s instructions to visualize phosphorylated protein and total protein, respectively. Gels were imaged using an Amersham Typhoon 5 (Cytiva, Marlborough, MA).

EPR spectroscopy.

Single cysteine GpsB mutants were purified as described above. Half of the purified protein preparation was phosphorylated with His₆-IreK-n at a 1:10 His₆-IreK-n:GpsB ratio in 50 mM Tris pH 7.5, 25 mM NaCl, 5 mM MgCl₂, and 2 mM ATP for 2 hours at 37°C. His₆-IreK-n was separated from the GpsB mutants using IMAC and phosphorylation and recovery of GpsB was confirmed using Pro-Q Diamond phosphoprotein gel stain and SYPRO Ruby protein gel stain, respectively. GpsB mutants were treated with 10 mM dithiothreitol (DTT) for 15 minutes at 37°C to reduce disulfide bonds. Rapid dialysis was performed with 50 mM Tris pH 7.5, 25 mM NaCl buffer to remove DTT before spin labeling with S-(1-oxyl-2,2,5,5-tetramethyl-2,5-dihydro-1H-pyrrol-3-yl)methyl methanesulfonothioate (MTSL) at a 10:1 MTSL:GpsB molar ratio overnight at 4°C. Excess spin label was removed by repeated rounds of dialysis with 2 L 50 mM Tris pH 7.5, 25 mM NaCl buffer. CW X-band EPR spectroscopy data were collected on a Bruker ELEXSYS 500 spectrometer (Bruker BioSpin Corporation, Billerica, MA) using an ER4122 cavity at room temperature under nonsaturating conditions over 100 G with a 1.5 G modulation amplitude and a 42 s scan time. Samples were contained in a glass capillary. DEER spectroscopy data were collected on a Bruker Q-band E580 spectrometer using an overcoupled Bruker EN5107D2 resonator. The samples contained 20% deuterated glycerol as a cryoprotectant and were flash frozen in a dry ice and acetone mixture. The data were collected at 50 K and analyzed using the LongDistances software program (http://www.biochemistry.ucla.edu/biochem/Faculty/Hubbell/) [26] written by C. Altenbach (University of California-Los Angeles, CA). The distance distributions with error bars resulted from fitting the background-corrected dipolar evolution data using the model-free algorithms in the LongDistances program.

Image processing and statistical analysis.

a). Phos-tag SDS-PAGE immunoblot analysis.

Quantification of IreK or MltG phosphorylation in whole-cell lysates were performed as described previously [13] with the following alterations. Band intensities of Phos-tag blots were quantified using the lane and band analysis function on ImageLab.

b). In vitro assay analysis.

Quantification of GpsB phosphorylation when incubated with or without His₆-IreK-n in vitro was performed as described previously using Azurespot software [13]. Intensity values of Pro-Q Diamond and SYPRO Ruby signal were exported to Microsoft Excel, where all calculations and graphing was performed. The ratio of phosphorylated protein to total protein of interest was determined and graphed over time. To determine relative initial rates, the time interval exhibiting a linear rate for each reaction type was identified, and the slope was determined from the best fit line. To determine the x-fold change in rates for each reaction compared to the control reaction, the ratio of the rate of each test reaction to the rate of the control reaction was calculated. The mean x-fold change in rate for at least 3 independent experiments was determined and graphed. Standard deviations and statistical analyses (Student’s T-test (two-tailed, heteroscedastic)) were also performed using Microsoft Excel functions.

Quantification of the percentage of phosphorylated GpsB in vivo.

To estimate the percentage of GpsB that was phosphorylated in E. faecalis cells under the conditions of our experiment, we used a standard curve calibration. Purified recombinant wild-type GpsB was phosphorylated to completion by His₆-IreK-n and analyzed by SDS-PAGE and ProQ Diamond / SYPRO Ruby staining as described above with the following modifications. GpsB was incubated in the presence of His₆-IreK-n at a 1:10 His₆-IreK-n:GpsB ratio for 5 hours at 37°C. Serial dilutions of the reaction were made in the presence of 1% bovine serum albumin and subjected to SDS-PAGE followed by staining with Pro-Q Diamond and SYPRO Ruby. The signal intensities for GpsB with each stain were determined using Azurespot software and standard curves of signal intensities versus protein were generated. The Pro-Q Diamond and SYPRO Ruby signal intensities for wild-type GpsB recovered from E. faecalis cells were determined and the amount of phosphorylated GpsB (GpsB-P; ProQ Diamond stained) and total GpsB (SYPRO Ruby stained) were calculated using the standard curves. The percentage of phosphorylated GpsB was calculated by dividing the amount of GpsB-P by total GpsB.

HIGHLIGHTS.

• The functional consequences of GpsB phosphorylation in enterococci are unknown

• Phosphorylation does not impact the flexibility of the C-terminal extension of GpsB

• Phosphorylation at GpsB S80 and T84 impairs IreK signaling and cephalosporin resistance

• Phosphorylation at GpsB T133 enables phosphorylation at other GpsB sites

• Multisite phosphorylation modulates the ability of GpsB to activate IreK signaling