1. Biochemistry and Chemical Biology
  2. Structural Biology and Molecular Biophysics
Download icon

Serine phosphorylation regulates the P-type potassium pump KdpFABC

  1. Marie E Sweet
  2. Xihui Zhang
  3. Hediye Erdjument-Bromage
  4. Vikas Dubey
  5. Himanshu Khandelia
  6. Thomas A Neubert
  7. Bjørn P Pedersen
  8. David L Stokes  Is a corresponding author
  1. Skirball Institute, Dept. of Cell Biology, New York University School of Medicine, United States
  2. PHYLIFE, Physical Life Science, Department of Physics, Chemistry and Pharmacy, University of Southern Denmark, Denmark
  3. Department of Molecular Biology and Genetics, Aarhus University, Denmark
Research Article
  • Cited 0
  • Views 597
  • Annotations
Cite this article as: eLife 2020;9:e55480 doi: 10.7554/eLife.55480

Abstract

KdpFABC is an ATP-dependent K+ pump that ensures bacterial survival in K+-deficient environments. Whereas transcriptional activation of kdpFABC expression is well studied, a mechanism for down-regulation when K+ levels are restored has not been described. Here, we show that KdpFABC is inhibited when cells return to a K+-rich environment. The mechanism of inhibition involves phosphorylation of Ser162 on KdpB, which can be reversed in vitro by treatment with serine phosphatase. Mutating Ser162 to Alanine produces constitutive activity, whereas the phosphomimetic Ser162Asp mutation inactivates the pump. Analyses of the transport cycle show that serine phosphorylation abolishes the K+-dependence of ATP hydrolysis and blocks the catalytic cycle after formation of the aspartyl phosphate intermediate (E1~P). This regulatory mechanism is unique amongst P-type pumps and this study furthers our understanding of how bacteria control potassium homeostasis to maintain cell volume and osmotic potential.

Introduction

Potassium is the primary osmolyte used by cells to maintain homeostasis and facilitate cell growth. A sizeable difference in K+ concentration across the plasma membrane is largely responsible for setting the membrane potential, is essential for regulating intracellular pH, for generating turgor pressure required for cell growth and division and as an energy source for diverse transport processes (Altendorf et al., 2009; Epstein, 2003). When external K+ concentrations are in the millimolar range, Trk and Kup are two constitutively expressed K+ uptake systems that are sufficient to maintain K+ homeostasis in bacteria. However, when extracellular K+ concentrations fall into the micromolar range, a high-affinity, active transport system called Kdp takes over. The kdp genes, which underlie the only inducible K+ transport system in bacteria, are organized into two tandem operons (kdpFABC and kdpDE). The latter is responsible for sensing the environment and controlling the expression of the former (Jung and Altendorf, 2002). The resulting KdpFABC protein complex is an ATP-dependent K+ pump that has micromolar affinity for K+ and is capable of generating a gradient up to six orders of magnitude, thus maintaining intracellular concentrations between 0.1 and 1M even when extracellular K+ is present in trace amounts (Ballal et al., 2007).

Although the mechanisms for activating KdpFABC have been studied in considerable detail, the field has largely overlooked the need to suppress the pump once external K+ levels return to the millimolar range. On the one hand, it is clear that activation of transport is done at the transcriptional level via KdpD/E, a two-component system in which KdpD serves as the sensor kinase and KdpE as the response regulator (Walderhaug et al., 1992). When the cell's need for K+ is not being met, KdpD phosphorylates and thus activates KdpE, which then induces expression of the KdpFABC pump (Jung et al., 2012). The signal acting on KdpD is still controversial, with recent studies providing evidence for external K+ concentration (Laermann et al., 2013), internal Na+ or NH4+ concentrations (Epstein, 2016) or dual binding sites for K+ that sense the magnitude of the gradient across the membrane (Schramke et al., 2016). On the other hand, inhibition of transport by KdpFABC has been observed when cells are returned to K+-rich media (Rhoads et al., 1978; Roe et al., 2000). Inhibition of transport under these conditions is a logical necessity to prevent wasteful usage of ATP and a skyrocketing of the intracellular K+ concentration. Although elevated K+ levels do result in transcriptional repression, this is an unlikely explanation for the observed inhibition because the effect is far more rapid (<2 min in K+ rich media) than the documented rate of protein turnover in bacteria (Trötschel et al., 2013). Indeed, the authors reporting this effect speculated that existing KdpFABC molecules may be directly inhibited by some unreported mechanism (Roe et al., 2000).

Recent structural analysis revealed phosphorylation of a serine residue in a cytoplasmic domain of the KdpB subunit within a highly conserved sequence motif (Huang et al., 2017). This observation was surprising, given the prominence of this motif (TGES162) in the catalytic cycle of related P-type ATPases and the lack of precedent for an analogous post-translational modification in this well-studied superfamily of ATP-dependent cation pumps (Palmgren and Nissen, 2011). P-type ATPases including KdpB share a reaction cycle known as the Post-Albers scheme (Figure 1) that begins with formation of an aspartyl phosphate on a conserved sequence (D307KTGT) in the centrally located cytoplasmic domain (P-domain). This phosphoenzyme intermediate (E1~P) transiently harnesses the energy of ATP, which is then used to drive conformational changes that lead to ion transport. The highly conserved TGES motif is found in a second cytoplasmic domain (A-domain) with Glu161 playing a crucial role in hydrolysis of the aspartyl phosphate in the second half of the catalytic cycle (E2-P → E2) (Møller et al., 2010). Biochemically, The E1~P and E2-P states are distinguished by their reactivity to ADP, with the former, high-energy form rapidly dephosphorylating to produce ATP, whereas the latter, low-energy is unreactive (Toustrup-Jensen et al., 2001). An X-ray structure of KdpFABC showed that Ser162, which is immediately adjacent to this catalytic glutamate residue, carried a phosphate moiety and the ability of a protein phosphatase to increase ATPase activity suggested that phosphorylation of Ser162 was inhibitory (Huang et al., 2017). This crystal structure also revealed an interaction between the phosphate moiety and positively charged residues on the third cytoplasmic domain (N-domain), suggesting a possible mechanism for this inhibition. However, subsequent cryo-EM structures from another group which also showed phosphorylation of Ser162 (Stock et al., 2018), revealed different conformations indicating that the interaction between A- and N-domains was not stable and therefore probably not the basis for inhibition.

The Post-Albers scheme for the reaction cycle of P-type ATPases.

The cycle alternates between two major conformations denoted E1 and E2. The E1 state binds ATP and produces the high-energy aspartyl phosphoenzyme intermediate (E1~P), which remains reactive with ADP. A spontaneous transition produces the low-energy E2-P intermediate, which no longer reacts with ADP but instead undergoes hydrolysis to produce the E2 species.

To further explore the inhibitory effects of serine phosphorylation, we have established growth conditions that lead to phosphorylation of Ser162 on KdpB and have characterized its effect on individual steps in the reaction cycle. Consistent with previous observations (Roe et al., 2000), when wild-type (WT) KdpFABC was expressed using the native promoter at low K+ concentrations, we found that moving cells to K+-rich media resulted in an inhibition of ATPase activity. We found that KdpFABC inhibition was time-dependent, was proportional to the extent of serine phosphorylation, and was restored by removing the phosphate with lambda protein phosphatase (LPP). To elucidate the mechanism of inhibition, we expressed a variety of mutants using the pBAD promoter in cells grown in K+-rich media and measured steady-state levels of the aspartyl phosphate intermediates (EP). We introduced the KdpB-S162D mutation to mimic the effect of serine phosphorylation and the KdpB-S162A mutation to prevent serine phosphorylation. Despite a lack of inhibition with KdpB-S162A, we found that EP levels for this mutation were much lower than for serine-phosphorylated, WT KdpFABC or for the KdpB-S162D mutation. Pulse chase studies with ADP indicated that both serine phosphorylation and the KdpB-S162D mutation trapped the pump in the E1~P state. Furthermore, although EP formation with the KdpB-S162A mutant was K+-dependent, serine phosphorylation as well as the phosphomimetic S162D mutation eliminated this K+-dependence. These observations suggest that serine phosphorylation disrupts allosteric coupling of the pump and provides secondary level of control over K+ homeostasis.

Results

Expression of KdpFABC

We used three different expression strains and a variety of different growth conditions to study the physiological stimulus for serine phosphorylation and to produce KdpFABC for in vitro experiments (Figure 2, Table 1). In native E. coli cells, the promoter that controls transcription of kdpFABC genes is regulated by KdpE in response to ‘potassium need’ of the cell. In K+ replete conditions, KdpFABC is not expressed, but when K+ levels in the media fall into the micromolar range, the constitutive transport systems (Trk, Kup) fail to maintain the necessary chemo-osmotic gradient. KdpD senses this deficiency and activates KdpE such that transcription of kdpFABC is initiated. WT KdpFABC is a powerful pump, so the deficit is normally overcome by modest levels of expression. In order to obtain over expression, as required for structural or biochemical studies, a multicopy plasmid with kdpFABC driven by the native promoter is introduced into a cell line (TK2499) with Trk and Kup knocked out and cells are cultured with limiting concentrations of K+, which must remain in the low micromolar range for expression of WT kdpFABC (Epstein and Davies, 1970). Because KdpFABC activity is required to rescue these cells, mutations that generate lower activity cannot survive in these conditions and, in those cases, precise K+ concentrations that allow cell growth and also produce robust expression depend on the K+ affinity and/or turnover rate of the particular mutant (e.g. 0.2 mM for the KdpA-Q116R mutation used for the X-ray structure). Inactive mutants cannot be expressed using this strategy, because they are incapable of rescuing cell growth in a K+-deficient environment. In those cases, the pBAD promoter can be used for expression with Top10 cells in K+ replete conditions (e.g. for previous cryo-EM structures), or a chromosomal copy of the kdpFABC gene can be used for rescue (TK2498 cells) and the mutant protein can be expressed from a plasmid carrying the native promoter. In this case, a His-tag is used to separate the mutant from the chromosomal copy by affinity chromotagraphy. In the following studies, we have used all three expression strategies. As we learned over the course of this work, the K+ concentration in the culture media is key to serine phosphorylation, making the expression protocol an important parameter to keep in mind.

Bacterial strains used to express various KdpFABC mutants.

The TK2498 strain contains a chromosomal copy of kdpFABC that carries the KdpA-Q116R mutation behind the native promotor. This gene rescues the cells in K+-deficient media (0.2 mM), which also serves to induce expression not only of the chromosomal copy, but also of His-tagged mutants on plasmids derived from pSD107, which carries the same native promoter. The genotype is shown below together with mutants expressed from this strain. The TK2499 strain has the same genotype as TK2498, except that it lacks the chromosomal copy of kdpFABC. It was used for expression of the His-tagged WT protein, which is induced at μM K+ concentrations. This strain was used for K+ shock experiments. The Top10 strain was used for expression of several His-tagged mutants behind the pBAD promoter grown in standard LB media and induced with arabinose. This strain was used for expression of proteins for biochemical assays of aspartyl phosphate formation. The mutants and their expression conditions are summarized in Table 1.

Table 1
KdpFABC mutants.
AnnotationMutationsStrainPromoterConditionFigures
WT-TK2499KdpEμM K+3,4,4-s1,6,6-s1
KdpA-E370AKdpA-E370ATK2498KdpE0.2 mM K+3
KdpA-R400A/T401AKdpA-R400A
KdpA-T401A
TK2498KdpE0.2 mM K+3
KdpA-S517AKdpA-S517ATK2498KdpE0.2 mM K+3
KdpA-R400AKdpA-R400ATK2498KdpE0.2 mM K+3
KdpA-R493AKdpA-R493ATK2498KdpE0.2 mM K+3
KdpB-D300AKdpB-D300ATK2498KdpE0.2 mM K+3
KdpB-D302AKdpB-D302ATK2498KdpE0.2 mM K+3
KdpB-D300A/D302AKdpB-D300A/
KdpB-D302A
TK2498KdpE0.2 mM K+3
KdpB-D307AKdpB-D307ATK2498KdpE0.2 mM K+3,6-s1
KdpB-D583AKdpB-D583ATK2498KdpE0.2 mM K+3
KdpB-K586AKdpB-K586ATK2498KdpE0.2 mM K+3
Q116R/K357A/R363AKdpA-Q116R/
KdpB-K357A/
KdpB-R363A
TK2498KdpE0.2 mM K+6-s2
Q116RKdpA-Q116RTop10pBADLB media*5,5-s1,s2,6
S162A/Q116RKdpA-Q116R/
KdpB-S162A
Top10pBADLB media*5,5-s1,s2
S162D/Q116RKdpA-Q116R/
KdpB-S162D
Top10pBADLB media*5,5-s1
D307A/Q116RKdpA-Q116R/
KdpB-D307A
Top10pBADLB media*5,5-s1,s2
S162A/E161Q/
Q116R
KdpA-Q116R/
KdpB-S162A/
KdpB-E161Q
Top10pBADLB media*5,5-s1,s2
  1. *LB media has been reported to contain 8 mM K+ (Su et al., 2009).

Survey of KdpFABC mutants reveals unexpected inhibition

An overarching goal of our work is to understand the allosteric coupling between KdpA and KdpB that underlies ATP-dependent K+ transport. To this end, we setup a survey of alanine substitution mutants to assess the role of various residues implicated by the X-ray crystal structure. For expression of these mutants, we used E. coli strain TK2498 grown at 0.2 mM K+ (Figure 2, Table 1). After purification, the ATPase activities of all of the mutants were very low (Figure 3a), causing us to explore parameters that might have influenced the protein. Based on our earlier evidence that phosphorylation of Ser162 in KdpB is inhibitory (Huang et al., 2017), we used mass spectrometry to determine levels of this post-translational modification (Figure 3—figure supplement 1), which were all very high (Figure 3b). We then used LPP to remove the phosphate and observed significant stimulation in ATPase activity for several of these mutants (Figure 3c), thus supporting an inhibitory role for Ser162 phosphorylation. WT protein and the D307A mutation in KdpB served as positive and negative controls, respectively. As expected, the D307A mutant was catalytically inactive because this residue is the site of aspartyl phosphate (E1~P) formation. In contrast to the mutants, the WT construct had virtually no serine phosphorylation. This is because the WT protein has much higher K+ affinity and was therefore expressed at much lower levels of K+ (<10 μM) in the TK2499 strain. These results motivated us to examine whether phosphorylation of Ser162, and consequent inhibition of KdpFABC, was a cellular response to elevated K+ in the media.

Figure 3 with 1 supplement see all
Survey of alanine substitution mutants reveals inhibition and serine phosphorylation of KdpB.

(A) Very low levels of ATPase activity were obtained from a series of Ala substitution mutants. As indicated in Table 1, WT protein has been expressed in TK2499 at very low levels of K+, whereas the other constructs were expressed in TK2498 at 200 μM K+. (B) Mass spectrometry of these Ala mutants revealed high levels of phosphorylation of Ser162 on KdpB. (C) Treatment of several of these mutants with LPP to remove serine phosphorylation resulted in significant stimulation of ATPase activity, supporting the notion that serine phosphorylation is inhibitory. Data in panels A and B are single measurements, whereas data in panel C were collected in duplicate.

Increased K+ is a physiological stimulus for serine phosphorylation

To test whether elevated K+ concentrations lead to serine phosphorylation, we first grew TK2499 cells with WT KdpFABC under K+ limiting (micromolar) conditions and then split the culture into two batches immediately prior to harvest. To one batch, we added 20 mM K+ - so called K+ shock - and growth was continued for 20 min. KdpFABC was then purified from both batches. ATPase assays documented a 40% inhibition for protein derived from the cells subjected to the K+ shock, which was fully restored after treatment with LPP (Figure 4a). Mass spectrometry confirmed that K+ shock produced a higher level of serine phosphorylation: ion currents for the phosphorylated peptide relative to the unphosphorylated peptide were 27% for K+ shock vs 1% for the control (Figure 4b). This quantification is nominal because phosphorylated and nonphosphorylated versions of the same peptide are not ionized and detected with equal efficiency and, as a result, positive ion mass spectrometry typically underestimates the level of phosphorylation (Xu et al., 2005). However, relative phosphorylation levels of different samples are reliable (Huang et al., 2016) and these data show that the K+ shock resulted in much higher levels of Ser162 phosphorylation as well as inhibition of ATPase activity.

Figure 4 with 1 supplement see all
K+ shock leads to serine phosphorylation of KdpB and inhibition of ATPase activity.

For these experiments, TK2499 cells were cultured at low K+ concentrations to induce expression of WT kdpFABC; thereafter, the culture was split and one half was subjected to 20 mM K+ for varying periods of time. (A) A 20 min K+ shock resulted in loss of ~40% of ATPase activity, which was fully recovered by LPP treatment. (B) Serine phosphorylation levels for samples from panel A determined by mass spectrometry were much higher after K+ shock. (C) ATPase rates from additional individual experiments with varying periods of K+ shock. The time periods are nominal because harvest of cells required a 20-min centrifugation in their respective media. (D) Serine phosphorylation levels for samples in panel C determined by Phos-tag staining of SDS gels show an increase in serine phosphorylation levels that is proportional to the observed inhibition of ATPase activity. (E) ATPase activities from a single culture split into four batches and treated with K+ for variable times show a graded inhibitory response. (F) Levels of serine phosphorylation for the cultures in panel E were determined by Phos-tag staining. (G) Data from all experiments were combined and show a reasonable correlation (R = 0.56) between ATPase activity and levels of serine phosphorylation determined by mass spectrometry. (H) Correlation of ATPase activities from all the experiments with levels of serine phosphorylation determined by Phos-tag staining produces a much better correlation (R = 0.97). (I) Correlation of serine phosphorylation levels from mass spectrometry and Phos-tag staining. ATPase measurements were performed in triplicate and shown as mean plus SEM; phosphoserine levels represent single measurements.

Time dependence of serine phosphorylation

To extend these studies, we varied the length of time that cells were subjected to K+ shock in order to characterize the time dependence of the response. For each experiment, the culture was split, 20 mM added to one half and incubated for either 0, 10, or 90 min before harvesting the cells. The incubation time is nominal given that cells had continued exposure to the media during the ensuing 20-min centrifugation step. ATPase inhibition increased as incubation time increased (15%, 25%, and 74%) and LPP treatment largely relieved this inhibition (Figure 4c). In addition to mass spectrometry, we used SDS-PAGE together with Phos-tag stain (Figure 4—figure supplement 1) to quantify levels of serine phosphorylation. Results from mass spectrometric phosphorylation analysis were consistent with this graded response to the K+ shock (increased phosphorylation levels of 2.07-, 4.02-, and 5.23-fold, respectively) and confirmed Ser162 as the site of phosphorylation (Figure 3—figure supplement 1). However, data from Phos-tag staining was more consistent across independent experiments and these data clearly illustrate a time-dependent increase in serine phosphorylation and the inverse correlation with ATPase activity (Figure 4d).

In addition to these isolated experiments, we conducted a time course in which a single culture was split into four batches that were shocked for varying times. In addition to a control, untreated batch, the cultures were exposed to 20 mM K+ for 0, 20 and 60 min prior to harvesting the cells. ATPase activities of the corresponding preparations of KdpFABC show an increasing amount of inhibition, which was largely reversed by treatment with LPP (Figure 4e), and increasing levels of serine phosphorylation (Figure 4f). Aggregation of data from all of these experiments showed a clear correlation between ATPase activity and serine phosphorylation for both mass spectrometry (Figure 4g) and Phos-tag staining (Figure 4h). Although the mass spectrometry data showed a lot of scatter, it was well correlated with measurements from Phos tag staining (Figure 4i). Combining the site specificity of mass spectrometry with the quantification accuracy of Phos-tag staining, provides strong evidence that a return to K+-rich media causes phosphorylation of Ser162 with a concomitant and proportional inhibition of ATPase activity.

Effect of serine phosphorylation on partial reactions

Transient formation of an aspartyl phosphate at the catalytic site in the P-domain of KdpB is a key step in the reaction cycle. To evaluate how serine phosphorylation affects individual steps in the cycle, we used a conventional assay employing [γ32P]-ATP to initiate the cycle followed by acid quench at defined time intervals. This assay reports on steady-state levels of aspartyl phosphate (EP), which includes molecules in both E1~P or E2-P states (Figure 1). These two states were then distinguished by their reactivity to ADP, based on the fact that the high-energy form (E1~P) is readily dephosphorylated by ADP, whereas the low-energy form (E2-P) is not (Toustrup-Jensen et al., 2001). For these studies, we introduced a number of mutations (Table 1) to elucidate the effect of serine phosphorylation, a stable post-translational modification that is distinct from the transient phosphorylation of the catalytic Asp307. In particular, the KdpB-S162A mutation prevented serine phosphorylation and the KdpB-S162D mutation served as a surrogate for the fully phosphorylated protein. The KdpB-E161Q mutation is expected to block hydrolysis of the E2-P species (step four in Figure 1), based on work with other P-type ATPases (Clausen et al., 2004), and KdpB-D307A serves as a negative control that prevents EP formation altogether. All these mutations in KdpB were combined with the KdpA-Q116R mutation, denoted kdp-42 or strain TK2242-42 in previous work (Epstein et al., 1978; Siebers and Altendorf, 1989), which lowers K+ affinity to the millimolar range and facilitates study of K+-dependence by eliminating the background activity generally observed with the WT construct (Siebers and Altendorf, 1988; Siebers and Altendorf, 1989). Although it may be surprising that a substitution in the signature TGES sequence is tolerated, an extensive phylogenetic analysis of the P-type ATPase superfamily indicates that the serine has the greatest variability in this motif and appears as Thr, Ala, Arg, or Pro in other family members (Chan et al., 2010). All relevant constructs were expressed behind a pBAD promoter in K+ rich media (Figure 2); these conditions activated the pathway for serine phosphorylation, which is evident from the stimulation in ATPase activities that was observed after treatment of the KdpA-Q116R construct with LPP (Figure 5—figure supplement 1). No inhibition was seen when Ser162 was mutated to Ala (KdpB-S162A) and mutations of KdpB-D307A, KdpB-E161Q and KdpB-S162D all resulted in complete inhibition of ATPase activity even after LPP treatment.

Typical experiments are shown in Figure 5a and b, which represent the time course of EP formation at room temperature in the absence and presence of 10 mM K+, respectively. In the presence of K+, EP levels of active constructs (e.g. S162A/Q116R) decay over time as the ATP in the solution is exhausted (Figure 5b). In contrast, EP levels were stable in the absence of K+, because there is no enzymatic turnover (Figure 5a), and at 4°C, because of a much slower turnover rate (Figure 5—figure supplement 2). EP levels were also stable for inactive constructs such as S162D/Q116R and E161Q/S162A/Q116R. Data from 15 individual experiments have been normalized and pooled in Figure 5c. These data included all points for experiments with low turnover rates (i.e. at 4°C or lacking K+) and initial points for experiments with high turnover rates (i.e. at room temperature in the presence of K+) or pulse chase (see below). Normalization of data from each experiment was based on data from E161Q/S162A/Q116R, which consistently produced the highest EP levels and which were stable under all the experimental conditions that were tested. Thus, data in Figure 5c indicate that Q116R and S162D constructs, which both reflect the effects of Ser162 phosphorylation, produced high levels of EP both in the presence and absence of K+. In contrast, EP levels for the Ser162A construct, which cannot be serine phosphorylated, are much lower and showed a significant dependence on K+. As expected, the D307A construct produced negligible levels of EP. These data suggest that formation of the high-energy aspartyl phosphate (step two in Figure 1) is K+-dependent and that phosphorylation of Ser162 not only uncouples this step but also prevents the hydrolysis of the phosphoenzyme intermediate.

Figure 5 with 2 supplements see all
Steady-state levels of aspartyl phosphate (EP).

(A–B) Individual experiments comparing time-dependence of EP formation at room temperature of several different constructs in the absence (A) or presence (B) of 10 mM K+. EP formation was initiated by addition of [γ32-P]-ATP and aliquots were taken at various time intervals and quenched with TCA. The constructs are indicated in the legend. (C) Aggregated data from 15 individual experiments that have been normalized relative to levels for the E161Q construct. (D) Pulse-chase experiment in which samples have been incubated for 15 s with [γ32-P]-ATP at 4°C to achieve maximal levels of EP (plotted at time zero) followed by addition of 10 mM ADP and 10 mM EDTA for the indicated time periods. Data was collected in triplicate and lines are drawn through mean values at each time point. Error bars in panel C correspond to SEM.

The data from S162D/Q116R and E161Q/S162A/Q116R suggest that the KdpB-S162D mutation and the KdpB-E161Q mutation cause the cycle to be blocked at some point after aspartyl phosphate formation. In order to determine more precisely where the cycle was blocked, we used a pulse-chase strategy to distinguish E1~P from E2-P intermediates. For this assay, we initiated phosphoenzyme formation in the presence of K+ with [γ32P]-ATP at 4°C where steady-state levels are stable. After 15 s, an aliquot was removed (plotted as time zero in Figure 5d) and a solution containing 10 mM ADP and 10 mM EDTA was added to determine the proportion of E1~P present. Figure 5d shows that EP levels for S162A/Q116R and S162D/Q116R are immediately reduced to zero (within the time resolution of this assay). This result indicates that E1~P is the prevalent species at steady state and that serine phosphorylation prevents the molecule from progressing to the E2-P state (step three in Figure 1). In contrast, the E161Q/S162A/Q116R construct was completely unreactive with ADP, indicating the the E2-P state was trapped in accordance with the effects of the Glu to Gln mutation on other P-type pumps (Clausen et al., 2004).

Although steady levels of EP from S162D/Q116R are consistent with blockage of the E1~P to E2-P transition, data from the Q116R construct reflect a more complex situation. Unlike the S162D/Q116R construct, the Q116R construct generates a mixture of active (no serine phosphorylation) and inactive (serine phosphorylated) molecules. At first glance, one might expect that the fraction of inactive molecules with serine phosphorylation would produce stable EP levels while unmodified, active molecules would turnover and very slowly exhaust the ATP in the solution. However, ADP is produced as the active molecules turnover which will then react with and dephosphorylate the molecules arrested in the E1 ~P state, as seen in the pulse chase experiments. This behavior would then account for the observed fall in overall EP levels of partially phosphorylated Q116R mutants over time.

Effect of serine phosphorylation on ligand binding and transport

An alternative explanation for the observed inhibition could be that serine phosphorylation simply slows down one of the steps of the reaction cycle or changes the affinity for one of the substrates. To evaluate this alternative, we used ATPase activity to estimate affinities for K+ and ATP as well as inhibition by orthovanadate. Orthovanadate acts as a transition state analogue for inorganic phosphate and typically binds to the E2 state of P-type ATPases to form an inhibitory complex (Clausen et al., 2016). In this way, the IC50 for inhibition by orthovanadate reflects not only an intrinsic binding affinity but also the conformational equilibrium between E2 and E1 states of P-type ATPases (Toustrup-Jensen et al., 2001). For the K+ titration, we used the KdpA-Q116R mutant and for the ATP and vanadate titrations we used WT protein, both with high levels of serine phosphorylation, evident from a threefold to fivefold stimulation after LPP treatment (e.g. Vmax = 1.9 and 9.3 μmoles/mg/min in Figure 6b). Fitting of data from untreated and treated samples (Figure 6a & b) indicated that differences in KM for K+ (21 mM and 7 mM, respectively) and ATP (60 μM and 72 μM) were not statistically significant. Similarly, inhibition by orthovanadate under standard ATPase conditions (150 mM K+ and 2.4 mM ATP) indicate no significant change in IC50 (9.2 μM vs. 8.4 μM). Although not a comprehensive analysis of reaction cycle kinetics, these results reveal no significant changes in the affinity of these substrates and are therefore consistent with the interpretation that the KdpA-Q116R preparation consists of a fraction of molecules with normal binding affinities and a fraction of inactive molecules with serine phosphorylation. Although we consider it unlikely, it is possible that the Q116R mutation is masking an effect of serine phosphorylation on K+ affinity of the WT protein, which is expected to be in the micromolar range (Buurman et al., 1995; Dorus et al., 2001).

Figure 6 with 2 supplements see all
Substrate affinities are not affected by serine phosphorylation.

(A) For K+ titrations, the KdpA-Q116R mutant was expressed with the pBAD system, which produced high levels of serine phosphorylation (see Figure 5—figure supplement 1). K+-dependence curves show that TPP treatment increased Vmax significantly from 2.4 to 5.8 μmoles/mg/min, but that differences in KM (21.5 mM vs. 7.0 mM) where not statistically significant at the 95% confidence level. (B) ATP dependence of WT protein was fitted with Vmax = 1.9 and 9.3 μmoles/mg/min for untreated and treated samples respectively with insignificant change in KM (60 vs 73 μM). (C) Titrations of WT protein with orthovanadate produced maximal activities of 1.7 and 8.7 μmoles/mg/min for untreated and treated samples and IC50 values of 9.2 and 8.4 μM, respectively. Data were collected in triplicate and fitted with the Michaelis Menton equation (A,B) and inhibitor-response equation (C) in Prism8.

Membrane transport assays indicate that transfer of K+ across the membrane remain coupled to ATPase activity of the complex. For these assays, KdpFABC was reconstituted into proteoliposomes and a voltage-sensitive dye (DiSC3) was used to report electrogenic transport of K+ into the vesicles (Fendler et al., 1996). Upon addition of ATP, a robust decrease in fluorescence indicated a buildup of a positive membrane potential (Figure 6—figure supplement 1). The initial part of the curve was fitted with an exponential to quantify the rate (Damnjanovic and Apell, 2014a) and to compare the preparation before and after LPP treatment. This analysis shows an excellent correlation between ATPase activity and transport. If an uncoupling of serine-phosphorylated molecules allowed passive transport of K+ in the absence of ATPase activity, the membrane voltage generated by unphosphorylated molecules would be rapidly dissipated, as is seen by the addition of valinomycin in Figure 6—figure supplement 1a (red arrow). Thus, these data suggest that the fraction of active molecules are fully capable of coupled transport whereas the fraction of inactive molecules with serine phosphorylation do not allow passive transport of K+ across the membrane.

Finally, we tested our previous hypothesis for inhibition, involving salt bridges between the phosphorylated Ser162 in the A-domain and basic resides in the N-domain. Specifically, Lys357 and Arg363 were mutated to Ala to prevent this interaction. The corresponding construct, Q116R/K357A/R363A was expressed under K+-replete conditions, resulting in substantial levels of serine phosphorylation as shown by Phos-Tag staining. (Figure 6—figure supplement 2). Comparison of ATPase activity before and after LPP treatment document a robust inhibition by serine phosphorylation despite the inability to form these salt bridges, indicating that a different inhibitory mechanism is in play.

Discussion

In this study, our goal was to explore the effects of serine phosphorylation on KdpFABC that was first described by Huang et al., 2017. In a physiological context, it is well established that kdpFABC expression is induced by K+-deficient media in order to maintain sufficient levels of K+ inside the cell. Our data show that phosphorylation of Ser162 on KdpB occurs when these cells are returned to K+-rich media. This posttranslational modification results in inhibition of ATPase activity as well as K+ transport, thus preventing unnecessary ATP consumption and uncontrolled increases in intracellular K+. This novel inhibitory mechanism explains earlier studies in which cell-based K+ uptake was inactivated under similar conditions (Rhoads et al., 1978; Roe et al., 2000). The work by Roe et al., 2000 also showed that inactivation was irreversible in situ, that the effect was specific to K+ and was not due to osmotic upshock, which in any case would be very modest with the addition of only 20 mM KCl. Although these authors showed complete inhibition in cell-based assays of transport, we achieved only ~75% maximal inhibition of purified protein after much longer incubation times in K+-rich media (90 min vs. 2 min). This difference might reflect the much higher expression levels used for the current study. Specifically, we employed a multicopy plasmid in order to produce high levels of protein expression suitable for biochemical purification compared to the single chromosomal copy used in previous work. The higher expression level might affect the extent of the phosphorylation either by exceeding the capacity of the (still unknown) kinase system or by changing the ionic balance of cells and thus influencing the sensing mechanism. In any case, the inability to completely inhibit KdpFABC is likely to be detrimental to the cells, either by exhausting ATP levels or by creating an ionic imbalance. Either effect could interfere with progressive inhibition during long K+ shock experiments (e.g. the 90 min time point in Figure 4c) and affect the overall viability of the culture.

P-type ATPases are subject to a variety of regulatory strategies, but the inhibition of KdpFABC by phosphorylation of the highly conserved TGES motif represents a novel approach for this superfamily. In the case of plasma-membrane Ca2+-ATPase (PMCA) (Calì et al., 2017), yeast and plant H+-ATPases (Haruta et al., 2015), and the yeast lipid flippase (Drs2p) (Jacquot et al., 2012), the pumps are held in autoinhibited states by C-terminal domains. Similarly, phospholamban (PLB) and sarcolipin (SLN) are regulatory subunits that inhibit Ca2+-ATPase from sarcoplasmic reticulum (SERCA) (Primeau et al., 2018) and FxYD proteins regulate Na/K-ATPase (Garty and Karlish, 2006). In most cases, these regulatory domains/subunits interact with the conserved catalytic domains to prevent turnover; distinct physiological factors bind to or modify these regulatory domains/subunits to disrupt these interactions and allow catalytic domains to function. These stimulatory factors are diverse and include Ca2+/calmodulin for PMCA, phosphatidylinositol-4-phosphate for Drs2p and serine phosphorylation for the others. The regulatory mechanism for KdpFABC is novel in that the modification inhibits rather than stimulates the transport cycle and that the modification is applied directly to one of the catalytic domains rather than to an accessory structural element.

Although serine phosphorylation is well known in eukaryotes, there are relatively few examples in bacteria. More commonly, signaling cascades in bacteria involve phosphorylation of histidine or aspartate residues, typically involving two-component systems such as KdpD/E. Correspondingly, few Ser/Thr kinases have been identified in bacteria. In fact, a catalogue of bacterial signal transduction systems lists only two eukaryotic-like Ser/Thr kinases in E. coli: ubiB and yegI (Galperin et al., 2010). The former is involved in the ubiquinone biosynthesis pathway (Poon et al., 2000), whereas the latter has been characterized in vitro but its function in vivo remains unknown (Rajagopalan et al., 2018). These Ser/Thr kinases have been proposed to originate in eukaryotes and their presence in bacteria to be the result of horizontal gene transfer (Lai et al., 2016). If so, it would seem unlikely that a process as fundamental as osmotic homeostasis would be acquired by this mechanism. Atypical Ser/Thr kinases have also been reported in prokaryotes, such as the Hpr kinase that controls metabolism of carbon sources in gram positive bacteria, SpoIIAB, RsbT and RsbW that act on sigma factors to control sporulation and stress response in B. subtilis (Pereira et al., 2011), and AceK in E. coli that phosphorylates isocitrate dehydrogenase to control metabolism through the TCA cycle (Cortay et al., 1988). Their involvement in regulation of KdpFABC is worth considering. An intriguing alternative is that KdpB engages in autophosphorylation; if so, it must be very inefficient or require special conditions, because self-inhibition did not occur over the time course of our ATPase assays, nor has it been reported in the literature. High levels of phosphorylation observed on the D307A/Q116R mutant (Figure 3b) preclude transfer of phosphate from Asp307 to Ser162. The alternative is that phosphate transfers directly from ATP to Ser162, but the complete lack of signal in EP formation experiments using the D307A mutation (Figure 5) is inconsistent with this idea. Another intriguing possibility is that KdpD, which is already designed to sense the K+ need of the cell (Schramke et al., 2016), would regulate activity of a Ser/Thr kinase that targets KdpFABC, as has been seen in several other systems (Pereira et al., 2011).

The dependence of serine phosphorylation on cell growth conditions and its inhibitory effects on KdpFABC has important ramifications for our understanding of the reaction cycle. Earlier biochemical studies concluded that formation of E1~P was K+-independent, that K+ bound to the E2-P state and was transported across the membrane during hydrolysis of the aspartyl phosphate (Damnjanovic and Apell, 2014b; Siebers and Altendorf, 1989). According to this scheme, K+ transport by KdpFABC is analogous to K+ transport by Na+/K+-ATPase (Goldshleger et al., 2001) or lipid by lipid flippases (Jacquot et al., 2012). However, this conclusion is at odds with recent structural studies in which KdpB adopted an E1-like structural state when K+ was bound in the selectivity filter of KdpA (Huang et al., 2017; Stock et al., 2018). In the current biochemical experiments, we also observed K+-independence of EP formation, but only when KdpB was serine phosphorylated or carried the S162D mutation to mimic serine phosphorylation. In contrast, when the S162A mutation was used to prevent this phosphorylation, we observed much lower levels of EP formation and a clear dependence on the presence of K+. The lower levels of EP for the S162A construct in the presence of K+ likely reflects catalytic turnover of these molecules which reduces the fraction of time spent in one of the phosphoenzyme states (E1~P or E2-P). In contrast, inactive constructs such as S162D, E161Q, and the serine phosphorylated species are essentially stuck in one of these phosphoenzyme states. ADP-sensitivity seen in the pulse chase experiments show that E1~P is the prevalent species and that serine phosphorylation prevents, or at least greatly slows down, the E1~~P to E2-P transition. In contrast, the E161Q mutation typically interferes with hydrolysis of the aspartyl phosphate (Clausen et al., 2004 #2555), thus preventing the E2-P to E2 transition and, as expected, the phosphoenzyme formed by this construct does not react with ADP. Therefore, we conclude that formation of E1~P by functional KdpFABC molecules is dependent on K+ binding by KdpA from the periplasm and that K+ is likely imported across the membrane during the transition from E1~P to the E2-P state.

In addition to this inhibition of catalytic activity, serine phosphorylation of KdpFABC also eliminates the K+-dependence of EP formation. As described above, the inhibited pump is quite capable of reacting with ATP to form aspartyl phosphate intermediates, but has lost the K+ dependence of this step. Interestingly, the E161Q and S162D mutations also resulted in uncoupling, which suggests that this crucial TGES162 loop has long-range allosteric effects over K+ sensing within the transmembrane domain. Although the path taken by K+ across the membrane is still uncertain (Pedersen et al., 2019), the canonical substrate-binding site on M4 is connected directly to the P-domain and is likely to play an important role in sensing K+ and initiating EP formation. In addition, work on other P-type ATPases has shown how large movements of the A-domain propagate to the transmembrane domain via its linkages to M1 and M2 helices. These helices control occlusion of Ca2+ in the M4 site by SERCA (Sørensen et al., 2004) and access of phosphatidyl serine to this site by lipid flippases (Hiraizumi et al., 2019). The unusual pose of the A-domain in the crystal structure of serine-phosphorylated KdpFABC indicates that it has a larger range of motion relative to these other pumps. Thus, the loss of K+ dependence by serine phosphorylated KdpFABC could reflect enhanced mobility of the A-domain and associated structural elements that allow cytosolic ions access to the canonical binding site, thereby triggering the conformational change leading to E1~P in an unnatural and non-selective way. Although not observed in other P-type ATPases, the KdpB-E161Q mutation appears to have an analogous effect on the K+-dependence of this first step. Once E1~P is formed, our biochemical data suggest that serine phosphorylation prevents KdpFABC from progressing to the E2-P state. E2-P is characterized by the loss of ADP sensitivity, which results from an interaction between the TGES162 loop and the aspartyl phosphate (Asp307). It seems plausible that a phosphate moiety on Ser162 would sterically interfere with this interaction, thus preventing formation of E2-P. For the Q116R/S162A/E161Q, the lack of serine phosphorylation allows this construct to achieve the E2-P state, but the isosteric E161Q mutation prevents hydrolysis of the aspartyl phosphate that is normally orchestrated by this residue. Finally, although serine phosphorylation eliminates the K+-dependence of EP formation, the complex is not truly uncoupled since transportof K+ across the membrane, both active and passive, is also inhibited.

This mechanism of inhibition differs from our earlier speculation that salt bridges between the serine phosphate and the N-domain simply held the protein in the inactive conformation seen in the X-ray structure (Huang et al., 2017). Indeed, this interaction was not seen in the more recent cryo-EM structures (Stock et al., 2018) and we found that mutations designed to break up this interaction (K357A/R363A) had no effect on the inhibitory properties of serine phosphorylation. This finding is not surprising given the lack of conservation in this region of the N-domain and specifically for the two basic residues (K357 and R363) that mediated the interaction in the crystal structure. Interestingly, the cryo-EM structures show that both E1 and E2-P like conformations can be adopted by serine phosphorylated KdpFABC, which might appear to contradict our biochemical data. However, our studies of EP formation were designed to study the cycle initiated by ATP addition (clockwise in Figure 1), whereas E2-P can also be formed directly from phosphate by so-called back-door phosphorylation (reversal of steps 4 and 5 in Figure 1). Although this step is energetically unfavorable, it may have been facilitated by the presence of AlF4 in samples used for the cryo-EM studies.

In conclusion, our work has characterized a novel mechanism for regulating P-type ATPases and maintaining K+ homeostasis in bacteria. Fundamental questions remain regarding the molecular mechanism of KdpFABC, including the role of a unique tunnel that runs between the subunits and the precise path taken by K+ as it crosses the membrane. Given the effects that we have documented on the enzymatic cycle, efforts should be made to ensure that serine phosphorylation is not a factor in future structural and biochemical work on this intriguing membrane pump.

Materials and methods

Expression of KdpFABC complex

Request a detailed protocol

For studies of serine phosphorylation, the pSD107 plasmid was transformed into E. coli strain TK2499 (both from W. Epstein, Univ. of Chicago), which lacks all other K+ transport systems with genotype F, thi, rha, lacZ, nagA, trkA405, trkD1, Δ(kdpFABC)5, Δ(ompT) (Figure 2). This plasmid was originally derived from pBR322 and contained the kdpFABC operon along with its endogenous promoter and an 8x histidine tag at the C-terminus of kdpC. Cells were grown in 4 L flasks at 31°C in K0-medium (46 mM Na2PO4, 23 mM NaH2PO4, 25 mM (NH4)2SO4, 0.4 mM MgSO4, 6 μM FeSO4, 1 mM sodium citrate, 0.2% glucose, 1 μg/mL thiamine, 50 μg/mL carbenicillin) supplemented with additions of 10 μM KCl every 90 min after culture density reached an OD600 of ~0.1. Total culture volumes ranged between 4 and 9 L. Cells were harvested at an OD600 of ~0.3–0.4 by centrifugation at 3500 g for 20 min; the resulting cell pellets were frozen at −80°C for storage. For K+ shock experiments, the culture was divided into equal portions and 20 mM KCl was added from a 2M stock solution followed by further incubation for defined periods of time prior to harvest.

To generate site-specific Ala mutants, site-directed PCR mutagenesis was used to alter codons in the WT operon on the plasmid described above. These plasmids were transformed into E. coli strain TK2498 with genotype F, thi, rha, lacZ, nagA, trkA405, trkD1, kdpA-Q116R, Δ(ompT), also from W. Epstein (Figure 2). Unlike TK2499, this strain carries a chromosomal copy of kdpFABC with the Q116R mutation on kdpA, which lowers potassium affinity to ~6 mM and has been identified as kdp-42 or kdpA42 in previous publications (Epstein et al., 1978; Siebers and Altendorf, 1989). This tag-less copy served to ensure cell survival in the low K+ conditions required to induce pump expression in cases where the alanine mutant complexes were inactive. For expression, cells were incubated overnight at 37°C in 10 mL K5-medium (K0-medium supplemented with 5 mM KCl). This culture was transferred to 500 mL K1-medium (K0-medium supplemented with 1 mM KCl) and incubated at 37°C for 8 hr. The 500 mL cell culture was transferred again to 18 L K0.2-medium (K0-medium supplemented with 0.2 mM KCl) and incubated at 31°C to induce expression of kdpFABC. Cells were harvested at a culture density of OD600 = ~0.8 by centrifugation at 3500 g for 20 min at 4°C and cell pellets were either frozen or used immediately for purification.

For studies of EP formation, the kdpFABC operon with the 8x histidine tag was cloned into the pBAD vector (Invitrogen, Carlsbad, CA) and introduced into E. coli strain Top10 (Invitrogen). Cells were grown in 4L flasks at 37°C in LB media supplemented with 100 mg/mL ampicillin to an OD600 of 0.5. Expression was then induced by addition of 0.02% arabinose and the culture continued to grow at 37°C for 4 hr. Cells were then harvested by centrifugation as above and frozen for storage. Total culture volumes were typically 2 L.

Purification of KdpFABC

Request a detailed protocol

For purification of the KdpFABC complex, cells were resuspended in 50 mM Tris pH 7.5, 1.2 M NaCl, 10 mM MgCl2, 10% glycerol, protease inhibitor tablets (Roche, Basel Switzerland) and 1 mM DTT, and lysed with an Emulsiflex C3 high-pressure homogenizer (Avestin, Ottawa Canada). Whole cells and debris were removed by centrifugation for 15 min at 12,000 g, and membranes were pelleted by centrifugation for 2.5 hr at 90,140 g. Membranes were solubilized by overnight incubation in 50 mM Tris pH 7.5, 600 mM NaCl, 10 mM MgCl2, 10% glycerol, 1 mM TCEP and 1.2% n-decyl-β-maltoside (DM) at 4°C using 20 mL per gram of membrane. After centrifugation for 30 min at 90,140 g, the solution was loaded on a 5 mL Ni-NTA HiTrap chelating column (GE Healthcare, Chicago, IL) that was equilibrated in Ni-NTA base buffer (50 mM Tris pH 7.5, 600 mM NaCl, 10 mM MgCl2, 10% glycerol, 1 mM TCEP, and 0.15% DM) supplemented with 20 mM imidazole. The column was washed with 20 mL of this same buffer, followed by 20 mL of Ni-NTA base buffer supplemented with 50 mM imidazole. Stepwise 5 mL elutions (collected in 1.5 mL fractions) were then performed as the imidazole concentration was incremented in steps of 50 mM up to a final concentration of 300 mM imidazole. Alternatively, elution was done with a continuous gradient running from 50 to 500 mM imidazole. The fractions containing KdpFABC were pooled, concentrated to ~0.7 mL using a 100 kDa cut-off Amicon centrifugal filter (Sigma Aldrich, St. Louis, MO) and introduced onto a Superdex 200 size exclusion column (GE Healthcare) pre-equilibrated with 25 mM Tris pH 7.5, 100 mM KCl, 10% glycerol, 1 mM TCEP, and 0.15% DM. For measurements of EP formation, the NaCl was substituted for KCl during this purification. Fractions were either stored for short periods (days) at 4°C or frozen in liquid nitrogen and stored at −80°C.

Quantification of serine phosphorylation

Request a detailed protocol

The Phospho-Tag phosphoprotein gel stain kit (ABP Biosciences, Beltsville, MD) was used to evaluate levels of serine phosphorylation. Specifically, 20 μg of precipitated protein was resuspended in sodium dodecylsulfate (SDS) buffer and run on a 10% SDS-PAGE gel. After staining according to manufacturer’s instructions, gels were imaged in a Chemidoc imaging system (BioRad, Hercules, CA) using the epi-green illumination for Cy3 fluorescence. KdpB-S162A was used as a negative control and as background in quantifying band intensities.

Localization of the site of serine phosphorylation and estimation of stoichiometry were done by nano-liquid chromatography-tandem mass spectrometry (nano-LC-MS/MS). Slices from SDS gels were digested with 500 ng trypsin solution (10 ng/μL) (Promega, Madison, WI) in 50 mM NH4HCO3 and incubated for 4 hr at 37°C, followed by 500 ng GluC or AspN protease (Promega) overnight. Digestion was stopped with 10% formic acid and peptides extracted with 5% formic acid in 50% acetonitrile, followed by final extraction with acetonitrile. Combined peptide pools were concentrated to a small droplet by centrifugation under vacuum. The resulting peptides were desalted using a Stage Tip manually packed with Empora C18 High-Performance Extraction Disks (3M) (Rappsilber et al., 2007). Desalted peptide pools were dried under vacuum to a small droplet, and finally reconstituted to 20 μL with 0.1% formic acid.

LC-MS/MS was performed with a ThermoEasy nLC 1000 ultra-high-pressure UPLC system (ThermoFisher Scientific, Waltham MA) coupled online to a Q Exactive HF mass spectrometer with NanoFlex source (ThermoFisher Scientific). Analytical columns (25–30 cm long and 75 μm inner diameter with 8 μm spray needle) were packed in-house with ReproSil-Pur C18 AQ 3 μm reversed-phase resin (Dr. Maisch GmbH, Ammerbuch-Entringen, Germany). The analytical column was placed in a column heater (Sonation GmbH, Biberach, Germany) regulated to a temperature of 45°C. The peptide pool was loaded onto the analytical column with buffer A (0.1% formic acid) at a maximum back-pressure of 300 bar; peptides were eluted with a 2-step gradient of 3% to 40% buffer B (100% ACN and 0.1% formic acid) over 40 min and 40% to 90% B over 5 min, at a flow rate of 250 nL/min over 60 min before direct infusion into the mass spectrometer. Mass spectrometric data were acquired using a data-dependent (DDA) top-15 method, dynamically choosing the most abundant, not-yet-sequenced precursor ions from the survey scans (300–1750 Th). Peptide fragmentation was performed via higher energy collisional dissociation with a target value of 106 ions. Precursors were isolated with a window of 1.6 Th, survey scans were acquired with a resolution of 120,000 at m/z 200 and HCD spectra with 15,000 at m/z 200.

Mass spectra were first processed with Proteome Discoverer 1.4 and then compared with a SwissProt Database with a Taxonomy filter for E coli (22,982 sequences) from 072014 using Mascot (Matrix Science, London, UK). Decoy protein sequences with a reversed order were included to estimate false discovery rates, which averaged ~1% at protein and peptide levels. Peptide spectral matches were further analyzed with Thermo Scientific Xcalibur software. Phosphoserine peptide ion intensities (within 5 ppm of calculated mass) were manually extracted, and intensity ratios of singly phosphorylated peptides relative to total (phosphorylated and nonphosphorylated) peptide intensities were calculated for each sample.

Activity assays

Request a detailed protocol

Rates for ATP hydrolysis were measured with a coupled enzyme assay (Warren et al., 1974) with 5 μg of purified KdpFABC in a total volume of 0.5 mL at 25°C. The assay buffer contained of 75 mM TES pH 7, 150 mM KCl, 7.5 mM MgCl2, 9.6 U/mL lactate dehydrogenase, 9.6 U/mL pyruvate kinase, 2.4 mM ATP, 0.5 mM phosphoenol-pyruvate, 0.36 mM NADH, and 0.15% DM. Concentrations of KCl and ATP were varied for determination of KM. Orthovanadate was prepared by adjusting the pH of a 200 mM solution of sodium othovanadate to 10 using NaOH, boiling until the yellow color cleared, cooling to room temperature, and repeating this cycle until the pH stabilized. Assays were run in triplicate and data from titrations with K+, ATP and VO4 were fit using Prism8 software (GraphPad Software, San Diego, CA).

For measurement of K+ transport, KdpFABC was reconstituted into proteoliposomes using the method of Lévy et al., 1992. Specifically, a thin film of lipid (3:1 weight ratio of E. coli polar extract and 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) (Avanti Polar Lipids, Alabaster, AL) was prepared by evaporating a chloroform stock solution (25 mg/ml) with Ar gas followed by 2 hr in a vacuum chamber. This dry film was resuspended in transport buffer (20 mM HEPES pH 7.2, 5 mM MgSO4, 140 mM potassium sulfamate, 1 mM N-methyl-D-glucamine sulfamate) to make a 10 mg/mL lipid stock solution, which was subjected to five cycles of freezing and thawing using liquid N2. The stock solution was then extruded 13 times through a 400 nm nucleopore filter (Whatman plc, Maidstone, UK) to create homogenous, unilamellar liposomes. Proteoliposomes were made at a lipid:protein weight ratio of 20:1 by combining 1.25 mg liposomes, 62.5 μg of KdpFABC, and 375 μg Triton X-100 (Sigma Aldrich) in a total volume of 250 μL. This solution was stirred at room temperature for 30 min followed by addition of 7.5 mg BioBeads (BioRad) and a further 90 min incubation. Finally, 15 mg BioBeads were added before overnight incubation with stirring at 4°C. Reconstituted proteoliposomes were harvested and bath-sonicated three times for 10 s each before use in the transport assay.

The transport assay was performed as described by Damnjanovic et al., 2013 using the voltage-sensing dye DiSC3 (Anaspec, Fremont, CA). The fluorescent signal was measured with a Fluoromax-4 spectrofluorimeter (Horiba Scientific, Piscataway, NJ) with excitation at 650 nm (5 nm slit), and emission at 675 nm (5 nm slit). Each assay was done at 4°C in 2 mL reaction volumes comprising 25 μL proteoliposomes diluted in a transport buffer consisting of 25 mM HEPES pH 7.2, 5 mM magnesium sulfamate, 140 mM potassium sulfamate and 1 mM N-methyl-D-glucamine sulfamate, followed by the addition of 1 μM DiSC3 from a 1 mM stock in DMSO. Electrogenic transport by inside-out KdpFABC molecules was initiated by addition of 2 mM ATP, and the subsequent drop in fluorescence was fit to an exponential curve in order to calculate initial rate of transport. Assays were performed in triplicate.

For measurements of EP formation, aliquots of purified KdpFABC were diluted to 0.25 mg/ml in 50 mM HEPES pH 7.8, 2 mM MgCl2, 0.15% DM and 10 mM of either KCl or NaCl. The reaction was started by addition of 100 μM [γ32P]-ATP and 200 μL aliquots containing 25 μg of protein and 2.7 μCi were taken at intervals of 0, 15, 30, 60, 120 s and quenched with 300 μl of 35% ice cold trichloroacetic acid (TCA). For the pulse-chase experiments, samples were incubated for 15 s at 4°C, an aliquot was recovered followed by addition of 10 mM ADP and 10 mM EDTA with additional aliquots taken various time intervals. After completing the assay, aliquots were centrifuged for 10 min at 14,000 g and washed twice with either 5% ice-cold TCA (for filtration) or water (for SDS-PAGE). For filtration, pellets were resuspended in 500 μL 5% TCA, filtered with 0.22 mm MCE filters (MilliporeSigma, Burlington, MA) and washed three times with 5% TCA. Radioactivity was then quantified with a scintillation counter. For SDS-PAGE, the pellets were resuspended in 30 μL of 5 mM TRIS-phosphate pH 5.8, 6.7 M urea, 0.4 M DTT, 5% SDS, 0.014% bromophenol blue. The stacking gel contained 4% polyacrylamide, 65 mM TRIS-phosphate pH 5.5, 0.1% SDS. The resolving gel contained 7.5% polyacrylamide, 65 mM TRIS-phosphate pH 6.5, 0.1% SDS. The running buffer was 0.17 M MOPS pH 6 and 0.1% SDS. Gels were dried and exposed to a storage phosphor screen overnight which was then imaged with a Typhoon Trio+ (GE Healthcare). For each construct, data was obtained from three to four independent cell expression/purification procedures and individual samples were measured in triplicate.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
    A K+ transport ATPase in Escherichia coli
    1. W Epstein
    2. V Whitelaw
    3. J Hesse
    (1978)
    The Journal of Biological Chemistry 253:6666–6668.
  14. 14
    The roles and regulation of potassium in Bacteria
    1. W Epstein
    (2003)
    Progress in Nucleic Acid Research and Molecular Biology 75:293–320.
    https://doi.org/10.1016/s0079-6603(03)75008-9
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
    Towards an understanding of the molecular mechanisms of stimulus perception and signal transduction by the KdpD/KdpE system of Escherichia coli
    1. K Jung
    2. K Altendorf
    (2002)
    Journal of Molecular Microbiology and Biotechnology 4:223–228.
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
    Characterization of the phosphorylated intermediate of the K+-translocating Kdp-ATPase from Escherichia coli
    1. A Siebers
    2. K Altendorf
    (1989)
    The Journal of Biological Chemistry 264:5831–5838.
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51

Decision letter

  1. Lucy R Forrest
    Reviewing Editor; National Institute of Neurological Disorders and Stroke, National Institutes of Health, United States
  2. Olga Boudker
    Senior Editor; Weill Cornell Medicine, United States
  3. Howard Young
    Reviewer

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This paper provides convincing evidence that post-translational modification by phosphorylation is a key feature of the bacterial potassium uptake system KdpFABC, an unusual complex of a P-type ATPase (subunit KdpB) and a potassium channel. Specifically, the results show that phosphorylation serves to inhibit the complex once cellular potassium levels have been restored, a novel regulatory strategy that may present another unique feature of KdpFABC systems, but that also provides a new lens through which other P-type ATPases should be viewed.

Decision letter after peer review:

Thank you for submitting your article "Serine Phosphorylation Regulates the P-type Potassium pump KdpFABC" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Olga Boudker as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Howard Young (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission. Please aim to submit the revised version within two months.

Summary:

The bacterial potassium uptake system KdpFABC is an unusual complex of a P-type ATPase (subunit KdpB) and a potassium channel. Its expression levels are increased in response to low-potassium stress conditions, at which point it utilizes ATP to pump K+ into the cell thereby restoring the intracellular potassium concentration. Surprisingly, recent structural data for KdpFABC complexes revealed the phosphorylation of a conserved TGES motif in the A domain of KdpB, and subsequent experiments demonstrated that phosphorylation reduces the ATPase activity of KdgFABC. Using activity and phosphorylation assays, and mass spectrometry data, Sweet et al. demonstrate that this post-translation regulation occurs in response to increased K+ levels. Phosphorylation therefore serves to inhibit the complex once the cellular potassium levels have been restored. This novel regulatory strategy may present another unique feature of KdpFABC systems, but also provides a new lens through which other P-type ATPase should be viewed.

Essential revisions:

While the observations regarding the role of Ser162 in phosphorylation are clear and robust, more conclusive data must be provided regarding the molecular basis of inhibition, according to the following:

1) The results describing the experiments identifying the step in the Post-Alber cycle affected by phosphorylation are confusing, in part because of the various mutants and growth conditions. This section requires clarification. It may also help to rename E161Q to reflect the additional mutations that accompany it.

2) Based on Figure 4 it is concluded that KdpFABC is inhibited in the E1~P state. There are two issues with the experiment.

- First, all variants were produced at high K, and thus in the Q116R mutant, Ser162 would be mostly phosphorylated. The authors claim that the phosphorylation of Asp307 is uncoupled from the presence of potassium, which is why high phosphorylation is found for the inhibited variants Q116R and S162D (Figure 4A). But why does the EP level of Q116R decrease over time, if it is inhibited? The "rate" of dephosphorylation is at least as fast as for S162A, which is not inhibited and the dephosphorylation is nearly 100%. Considering that only a fraction of the protein (e.g. 75%) is inactive, the dephosphorylation should only apply to a small portion of the phosphorylated protein. How does this experiment look for Q116R expressed at low K? Is the initial EP level much lower due to fast turnover as suggested for S162A? And is the dequenching even faster?

- Second, from the pulsed-chase experiment the authors claim that the protein is trapped in an E1~P state. However, from the inhibition with orthovanadate the authors concluded that the equilibrium between E1/E2 is not affected by the phosphorylation. In the best case this statement is only a bit misleading as it only refers to the non-phosphorylated fraction. If something else was meant, this statement should be clarified. Further, the pulsed-chase experiment can be explained by two scenarios: the inhibition of the protein in one state and the E1-E2 transition as rate-limiting step. For S162A the authors claim the experiment reflects the latter, while S162D is inhibited in E1-~P.

3) The plots in 2G and 2H are misleading as they are based on the assumption that upon 100% phosphorylation the ATPase activity is completely abolished. However, there is no formal proof of this. In fact, the maximal inhibition seen is approx. 60% after 90 min K shock when compared to the dephosphorylated sample. On the other hand, the level of phosphorylation cannot be reliably estimated, neither from mass spec nor from the phos-tag signal. One could thus speculate that the rate of phosphorylation is even higher, particularly because no significant increase was detected upon K shock after 20 minutes (Figure 2F), and that the phosphorylation does not completely inhibit KdpFABC but instead slows down the state transitions (e.g. E1 to E2) such that 100% inhibition would never be seen. This scenario may also explain the experiments performed in Figure 4.

4) The molecular dynamics simulations presented rely on the deletion of the phosphate group from Ser162 as a proxy for the dephosphorylated state. This assumption is flawed as it involves artifactual, high-energy states, and does not consider the metastable conformations that are likely to be involved immediately before or after kinase/phosphatase activity. Moreover, the simulations of the phosphorylated structure were initiated using a system in which the cavities in the structure have not been hydrated (e.g., using Dowser), with the exception of one highly-ordered water molecule observed in the structure. Such cavities are unlikely to become hydrated on the timescale of the equilibration, but could have a significant effect on the dynamics of the protein system. Therefore, these simulation data should be removed. In case the authors chose to provide new computational data in its place, substantially more detail must be provided to describe the analysis, e.g. which atoms are included in RMSD calculations. Please also note that radial distribution functions are not typically appropriate for describing residue-residue interactions; distance distributions are preferable.

5) A central theme of the proposed mechanism is that Ser162-Pi interacts with K357+R363, as observed in the available crystal structure (PDB code 5MRW) and that dephosphorylation would abolish this interaction leading to an allosteric effect on the transmembrane domains. However, other structures of KdpFABC (PDB codes 6HRA and 6HRB; from Stock et al.,) contain Ser162-Pi and yet do not show this interaction, despite representing, at least in one case, a similar state in the transport cycle. Moreover, one of those structures is an E2 state of Ser-phosphorylated KdpB, which questions the hypothesis that the phosphorylated state is truly inhibited in an E1 conformation. Thus, any discussion of a molecular mechanism must carefully and convincingly consider those structures and their implications. [We note that the lower resolution of the cytoplasmic domains in the cryo-EM structure obtained for the E1 state point towards a higher flexibility of this region (supporting the conclusion drawn here that phosphorylation increases the mobility). Yet, despite the lower resolution the overall position of the A, N and P domains is unambiguous enough to show that these contacts are disrupted.]

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "Serine Phosphorylation Regulates the P-type Potassium pump KdpFABC" for consideration by eLife. Your article has been reviewed by Olga Boudker as the Senior Editor, a Reviewing Editor, and two reviewers. The reviewers have opted to remain anonymous. The Reviewing Editor has also reviewed the article, and has drafted this decision to help you prepare a revised submission.

Summary:

The bacterial potassium uptake system KdpFABC is an unusual complex of a P-type ATPase (subunit KdpB) and a potassium channel. Its expression levels are increased in response to low-potassium stress conditions, at which point it utilizes ATP to pump K+ into the cell thereby restoring the intracellular potassium concentration. Recently, structural data for KdpFABC complexes revealed a surprising observation, namely phosphorylation of a serine in a conserved TGES motif in the A domain of KdpB, based on which it was shown that phosphorylation reduces the ATPase activity of KdgFABC. In the present work, Sweet et al., demonstrate that this post-translation regulation occurs in response to increased K+ levels, using activity and phosphorylation assays, and mass spectrometry data. Phosphorylation therefore serves to inhibit the complex once the cellular potassium levels have been restored. This novel regulatory strategy may present another unique feature of KdpFABC systems, but also provides a new lens through which other P-type ATPases should be viewed.

The authors satisfactorily addressed the major concerns regarding the biochemical aspects of the work, and the experimental systems and major conclusions of the work are more clearly presented. They also provided additional experimentation that test a foundational assumption regarding the simulation data. Unfortunately, the new experiments illustrate that key residues interacting by the phosphorylserine group in the starting structure of the MD simulations are not required for inhibition. Therefore, the simulations of this structure are unlikely to be relevant for the (very challenging) question at hand. Moreover, the authors use, as their control, a modified (dephosphorylated) version of that structure, which is bound to have unfavorable interactions that most probably do not reflect the true dephosphorylated protein. [On this latter point, based on the authors responses it is apparent that one of the comments in the first round of reviews was insufficiently clear, and therefore, we provide a clarification below.] Based on these concerns, if the authors wish to resubmit a revised version of the manuscript, the MD simulations must be removed.

Clarification:

In referring to a high-energy, artifactual state, we referred not to the crystal structures themselves, but rather to the molecular system created by the authors when the phosphate group was deleted from the serine in the crystal structure prior to running the MD simulations, i.e., the SER system. The concern is that introducing this dramatic artificial change into a structure (in this case, one in which the phosphoryl group interacts with two positively-charged residues) creates an system with a set of new high-energy interactions; these could be the cause of the "uncoupling of the A- and N-domains" (subsection “Effects of serine phosphorylation on structural dynamics”). In other words, the SER system does not reflect a true dephosphorylated protein. The true dephosphorylated protein may or may not ever visit the exact same conformation that is present in this starting structure. In an ideal world, one would initiate the simulations with an experimental structure of a dephosphorylated protein in the same state of the transport cycle. Obviously, such a structure is not available, but the alternative adopted here is insufficiently robust to address the question at hand.

https://doi.org/10.7554/eLife.55480.sa1

Author response

Essential revisions:

While the observations regarding the role of Ser162 in phosphorylation are clear and robust, more conclusive data must be provided regarding the molecular basis of inhibition, according to the following:

1) The results describing the experiments identifying the step in the Post-Alber cycle affected by phosphorylation are confusing, in part because of the various mutants and growth conditions. This section requires clarification. It may also help to rename E161Q to reflect the additional mutations that accompany it.

We agree that the diversity of samples used for this work could cause confusion. Keeping track of the wide array of mutations and varying growth conditions is critical for understanding the implications of individual experiments and for the overall functioning of KdpFABC. To assist in this task, we have added a new paragraph at the beginning of the Results section to enumerate the mutations and to explain differences in the growth conditions used to produce them. This explanation is accompanied by a newly added Figure 2, which diagrammatically illustrates the bacterial strains as well as K+ concentrations used for growth, as well as Table 1, which summarizes information for each construct. We have also changed the nomenclature for the constructs in the text to explicitly acknowledge double and triple mutations.

2) Based on Figure 4 it is concluded that KdpFABC is inhibited in the E1~P state. There are two issues with the experiment.

Note that data for these experiments are now shown in Figure 5 of the revised manuscript.

- First, all variants were produced at high K, and thus in the Q116R mutant, Ser162 would be mostly phosphorylated. The authors claim that the phosphorylation of Asp307 is uncoupled from the presence of potassium, which is why high phosphorylation is found for the inhibited variants Q116R and S162D (Figure 4A). But why does the EP level of Q116R decrease over time, if it is inhibited? The "rate" of dephosphorylation is at least as fast as for S162A, which is not inhibited and the dephosphorylation is nearly 100%. Considering that only a fraction of the protein (e.g. 75%) is inactive, the dephosphorylation should only apply to a small portion of the phosphorylated protein.

We do have an explanation for the decrease in EP levels for the Q116R mutant. As the reviewers acknowledge, this preparation has a mixture of serine phosphorylated (inhibited) and unmodified (fully active) protein. We believe that the fraction that is serine phosphorylated is blocked in the E1~P conformation, whereas the active fraction will turnover. As a result of this enzymatic turnover, ADP will be produced and, as demonstrated by the pulse-chase experiments (Figure 5D), ADP rapidly reacts with the E1~P species to reverse the phosphorylation of Asp307 (step 2 in Figure 1). Thus, as time goes on and more ADP is produced, the total steady-state levels of EP will fall to zero. This effect only occurs in the presence of K, because otherwise there is no turnover and no ADP is produced. We previously thought that this detail might be confusing to readers, but we are pleased that the reviewers brought it up and we have included an explanation in subsection “Effect of serine phosphorylation on partial reactions” of the revised manuscript.

How does this experiment look for Q116R expressed at low K? Is the initial EP level much lower due to fast turnover as suggested for S162A? And is the dequenching even faster?

With our current expression strategies, we are not able to produce the Q116R mutant at low K. Because of the lower K affinity produced by this mutation, it cannot rescue cell growth at μM levels of K, but requires at least 0.2 mM at which concentration there is a already a substantial amount of serine phosphorylation. In principle, one could try to express a His-tagged Q116R construct in a background of WT KdpFABC. Although we do not have the strain necessary to attempt this experiment, it would be interesting to pursue this approach in follow-up studies.

- Second, from the pulsed-chase experiment the authors claim that the protein is trapped in an E1~P state. However, from the inhibition with orthovanadate the authors concluded that the equilibrium between E1/E2 is not affected by the phosphorylation. In the best case this statement is only a bit misleading as it only refers to the non-phosphorylated fraction. If something else was meant, this statement should be clarified. Further, the pulsed-chase experiment can be explained by two scenarios: the inhibition of the protein in one state and the E1-E2 transition as rate-limiting step. For S162A the authors claim the experiment reflects the latter, while S162D is inhibited in E1-~P.

The titrations with orthovandate, K and ATP address the possibility that serine phosphorylation alters the rate of a step in the reaction cycle rather than blocking it altogether. If rates were affected, the fraction of unphosphorylated molecules would behave normally, but the serine-phosphorylated molecules – comprising the majority of the population – would behave differently and thus shift one of these titration curves. The fact that all three titration curves, now shown in Figure 6, were not significantly affected by levels of serine phosphorylation is consistent with (though not necessarily proof of) our hypothesis that serine phosphorylation completely blocks turnover and that the observed activity comes only from unphosphorylated molecules. In order to make this reasoning clearer, we have moved this section to a later point in the manuscript so that there is better context for these experiments. We have also rewritten the relevant paragraphs in subsection “Effect of serine phosphorylation on ligand binding”.

3) The plots in 2G and 2H are misleading as they are based on the assumption that upon 100% phosphorylation the ATPase activity is completely abolished. However, there is no formal proof of this. In fact, the maximal inhibition seen is approx. 60% after 90 min K shock when compared to the dephosphorylated sample. On the other hand, the level of phosphorylation cannot be reliably estimated, neither from mass spec nor from the phos-tag signal. One could thus speculate that the rate of phosphorylation is even higher, particularly because no significant increase was detected upon K shock after 20 minutes (Figure 2F), and that the phosphorylation does not completely inhibit KdpFABC but instead slows down the state transitions (e.g. E1 to E2) such that 100% inhibition would never be seen. This scenario may also explain the experiments performed in Figure 4.

As discussed above for point 2, the data are consistent with the idea that serine phosphorylation completely inhibits turnover. This conclusion is further supported by results from the S162D mutation, which is completely inactive and produces a steady level of EP in both the presence and absence of K. However, the reviewers are correct that we have no formal proof that serine phosphorylation has this same effect and perhaps S162D is a misleading – though commonly used – mimetic of phosphorylation. The right-hand axes are not essential to our story so we have removed them from the revised figures and will simply report phosphorylation in arbitrary units, since this will convey the point without unnecessary confusion or controversy.

The data after 90 min of K shock is somewhat puzzling, as one would logically expect continued increases in inhibition over time. After reconsidering this matter, however, we believe that cell metabolism is likely to be compromised under extended conditions of K shock. Unlike the complete inhibition observed in the previous publication by Roe et al., (2000), the inhibitory system (as yet uncharacterized) is not 100% effective in shutting down KdpFABC activity in our expression system. In the original manuscript, we speculated that this outcome is due to forced overexpression of KdpFABC from a multicopy plasmid that has overwhelmed this regulatory system. As a result of this deficiency, we expect that K shock will be detrimental to ionic homeostasis and potentially compromise survival of cells after they are returned to K replete conditions. If so, K shock would cause unexpected results during long incubations This effect may be reflected in the lower activity seen in LPP treated samples from the longer periods of K shock. This issue is discussed in more detail in the Discussion section in the revised manuscript.

4) The molecular dynamics simulations presented rely on the deletion of the phosphate group from Ser162 as a proxy for the dephosphorylated state. This assumption is flawed as it involves artifactual, high-energy states, and does not consider the metastable conformations that are likely to be involved immediately before or after kinase/phosphatase activity.

The reviewers raise important points about our analysis of MD simulations that have caused us to reconsider their contribution. The bottom line is that simulations show a dramatic difference in the behavior of serine phosphorylated and unphosphorylated KdpFABC and, although they do not offer definitive answers, we believe that they make a valuable contribution to our story. For the revised manuscript, we have refocused the analysis on the dynamic properties of cytoplasmic domains of KdpB and on the linkers connecting the A-domain to the membrane helices. We have downplayed the issue of hydration and have presented a testable hypothesis to explain the functional effects seen in our biochemical experiments.

We are not certain what is meant by "artifactual high-energy states" and the "metastable conformations […] immediately before or after kinase/phosphatase activity". In our minds, this statement could refer to either (1) states related to the conformational cycle of the P-type pump and phosphorylation/hydrolysis of Asp307, (2) the regulatory process involving phosphorylation/dephosphorylation of the Serine162 or (3) the influence of crystal environment that could force flexible regions to adopt a high-energy conformation. For our response, we have assumed that the "artifactual high-energy state" refers to the packed environment in the crystals that produced the starting model for our MD simulations (#3 above) and that kinase/phosphatase activity refers to the unknown agent that is acting on Ser162 (#2 above). We sincerely hope that this is as intended.

We agree that the X-ray structure is possibly in a high-energy conformation, stabilized by the chemical environment of the crystallization experiment (crystal packing, chemical composition of reservoir solution etc.). We nevertheless used the X-ray structure because it is based on data of considerable higher quality compared to the EM models. This is exemplified by the higher resolution of the X-ray data (2.9 Å) compared to the cryo-EM structures (3.7 and 4.0 Å), and is especially evident in the definition of the cytoplasmic domains (cryo-EM structures have local resolution of 5-6 Å in this region). We note that crystal packing is likely not the primary determinant of the observed conformation, given that the asymmetric unit contained three copies of KdpFABC, each having unique interactions with neighboring molecules but all adopting the same conformation (suitable for non-crystallographic symmetry averaging). In general, the use of crystal structures as a starting point of Molecular Dynamics is standard practice, and any "high-energy" strain imposed by the crystallographic experiment is normally relieved either by the initial equilibration procedure or during the restraint-free production simulations. It is the undeniable that the choice of the starting conformation can potentially influence the average thermodynamic quantities calculated from the simulations, but we feel that our choice is well justified and was effective not only in revealing distinct character for the two systems but also in highlighting a region of the molecule that may be responsible for allosteric coupling. For the revised manuscript, we added a brief explanation for the basis of selecting the X-ray structure over the cryo-EM structures in subsection “Effects of serine phosphorylation on structural dynamics”.

With regard to the metastable intermediate states associated with phosphorylation, it was not our goal to study this process, but rather to compare the properties of the protein in the presence and absence of serine phosphorylation.

Moreover, the simulations of the phosphorylated structure were initiated using a system in which the cavities in the structure have not been hydrated (e.g., using Dowser), with the exception of one highly-ordered water molecule observed in the structure. Such cavities are unlikely to become hydrated on the timescale of the equilibration, but could have a significant effect on the dynamics of the protein system. Therefore, these simulation data should be removed. In case the authors chose to provide new computational data in its place, substantially more detail must be provided to describe the analysis, e.g. which atoms are included in RMSD calculations.

There is ample evidence in literature that the free water movement in MD simulations is able to accurately hydrate protein cavities. For example, Tajkhorshid and colleages showed that cavities of amino acid transporters were hydrated on a timescale comparable to our simulations (~100 to 200 ns) (Table 1 in Li et al., 2013). In another example, water molecules spontaneously hydrate or dehydrate a key enzymatic active site in the cytochrome c oxidase (Goyal et al., 2013). With regard to ion pumps, we have not been able to find any report on MD simulations in which a hydrating procedure was used (Rui et al., 2016; Ratheal et al., 2010; Sugita et al., 2010; Espinoza-fonseca et al., 2015; Sonntag et al., 2011, and several of our own investigations: Yamamoto et al., 2019; Dubey et al., 2018; Mahmmoud et al., 2015; Poulsen et al., 2010). These examples illustrate the broad acceptance of MD simulations of transmembrane protein systems without using cavity-hydrating programs such as Dowser. As a result, we believe that the distribution of water seen during our simulations has good precedent in the field as a measure of hydration for KdpFABC.

We did nevertheless use Dowser++ as proposed on the KdpFABC as well as on the Na/K-ATPase system that we have previously examined in a number of publications. During this process, we communicated with the authors of Dowser++ to ensure that our protocol was optimal (specifically that a draining process was applied after the initial hydration). Despite our best effort, we obtained gross over-hydration of the membrane domain for both protein systems. As a result of this over-hydration, the ensuing MD runs produced anomalous results. In reviewing the use of Dowser++ in other publications (Farahvash et al., 2018, Morozenko et al., 2016), we noted that the number of water molecules inserted by Dowser++ often increases significantly as the resolution of the structure decreases and that very high-resolution (sub-Å) crystal structures are typically used for benchmarking Dowser++ performance.

Given the uncertainties of using Dowser++ with low resolution structures, and the poor behavior of the resulting hydrated complexes of KdpFABC and Na/K-ATPase during simulations, we believe that this issue requires further experimentation which is outside of the scope of the current manuscript. We have therefore refocused the revised manuscript on the dynamics of the cytoplasmic domains and the identification of a potential allosteric control site in the linker between the A-domain and the membrane domain. Although our observation of increased hydration is noted as a potential explanation for the K+ independence of EP formation caused by serine phosphorylation, we have de-emphasized this point by moving it to the Discussion section with the relevant data moved to the supplemental material (Figure 9—figure supplement 1). We believe that this observation provides a potentially important lead for future studies into the allosteric effects of serine phosphorylation and therefore makes a valuable contribution to the manuscript.

Please also note that radial distribution functions are not typically appropriate for describing residue-residue interactions; distance distributions are preferable.

We agree that distance distributions may be more suited to represent residue-residue interactions. We have therefore included RMSD distance distributions alongside the radial distribution functions. In general, RMSD was calculated from the last 100 ns of the simulations after correcting individual structures for bulk translational motion; the calculations include all atoms from the relevant residues. The radial distribution functions and RMSD values both show comparable effects and support the striking difference in behavior after phosphorylation of Ser162. The details of the RMSD calculation have been included in legend for Figure 7—figure supplement 1.

5) A central theme of the proposed mechanism is that Ser162-Pi interacts with K357+R363, as observed in the available crystal structure (PDB code 5MRW) and that dephosphorylation would abolish this interaction leading to an allosteric effect on the transmembrane domains. However, other structures of KdpFABC (PDB codes 6HRA and 6HRB; from Stock et al) contain Ser162-Pi and yet do not show this interaction, despite representing, at least in one case, a similar state in the transport cycle. Moreover, one of those structures is an E2 state of Ser-phosphorylated KdpB, which questions the hypothesis that the phosphorylated state is truly inhibited in an E1 conformation. Thus, any discussion of a molecular mechanism must carefully and convincingly consider those structures and their implications. [We note that the lower resolution of the cytoplasmic domains in the cryo-EM structure obtained for the E1 state point towards a higher flexibility of this region (supporting the conclusion drawn here that phosphorylation increases the mobility). Yet, despite the lower resolution the overall position of the A, N and P domains is unambiguous enough to show that these contacts are disrupted.]

These are very valid observations that indicate our failure to express clearly our point of view about the interaction between Ser162-Pi and K357-R363. Although this interaction was highlighted in the X-ray structure and proposed as a potential mechanism for inhibition, the paper also noted lack of conservation for K357 and R363 in KdpB from other species. Given the cryo-EM structures from the Hänelt group and the MD studies presented here, our view on this matter has shifted. We now believe that the observed interaction with K357 and R363 in the crystal structure is not essential for the inhibition to occur. We have brought this issue front-and-center in the Introduction of the revised manuscript (Introduction) and return to the lack of sequence conservation in the Discussion section. In addition, we have added a key experiment to show that mutation of Lys357 and Arg363 does not prevent inhibition (Figure 7—figure supplement 1G and H). In particular, we expressed the K357A/R363A/Q116R mutant in the TK2498 strain in 0.2 mM KCl (same conditions as for Figure 3) and used the Phos-tag stain to show that serine phosphorylation had occurred. We then compared ATPase activity of this preparation before and after LPP treatment. The ~4-fold stimulation produced by the phosphatase is analogous to the effect on WT protein, providing clear evidence that Lys357 and Arg363 are not essential for the inhibitory mechanism. These new data are referenced together with the MD results (subsection “Effects of serine phosphorylation on structural dynamics”).

With regard to the cryo-EM structures from the Hänelt group, these two structures were derived from a single sample prepared in the presence of AMP-PCP, K, and AlF4. Given our evidence that serine-phosphorylation eliminates K-dependence of the reaction cycle, this combination of ligands could plausibly stabilize E1, E1-ATP, E1~P, or E2-P states. Although we conclude that serine-phosphorylation blocks the reaction cycle in the E1~P state, the E2-P state would still be accessible via back-door phosphorylation, especially given the uncoupling of the Post-Albers reaction cycle with respect to K. These thoughts about the cryo-EM structures are included in the penultimate paragraph of the Discussion section, which also dismisses roles for Lys357 and Arg363 and offers an alternative explanation for stabilization of the E1~P state.

https://doi.org/10.7554/eLife.55480.sa2

Article and author information

Author details

  1. Marie E Sweet

    Skirball Institute, Dept. of Cell Biology, New York University School of Medicine, New York, United States
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
  2. Xihui Zhang

    Skirball Institute, Dept. of Cell Biology, New York University School of Medicine, New York, United States
    Contribution
    Formal analysis, Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
  3. Hediye Erdjument-Bromage

    Skirball Institute, Dept. of Cell Biology, New York University School of Medicine, New York, United States
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  4. Vikas Dubey

    PHYLIFE, Physical Life Science, Department of Physics, Chemistry and Pharmacy, University of Southern Denmark, Odense, Denmark
    Contribution
    Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  5. Himanshu Khandelia

    PHYLIFE, Physical Life Science, Department of Physics, Chemistry and Pharmacy, University of Southern Denmark, Odense, Denmark
    Contribution
    Conceptualization, Supervision, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  6. Thomas A Neubert

    Skirball Institute, Dept. of Cell Biology, New York University School of Medicine, New York, United States
    Contribution
    Formal analysis, Supervision, Validation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  7. Bjørn P Pedersen

    Department of Molecular Biology and Genetics, Aarhus University, Aarhus, Denmark
    Contribution
    Conceptualization, Validation, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7860-7230
  8. David L Stokes

    Skirball Institute, Dept. of Cell Biology, New York University School of Medicine, New York, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Methodology, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    stokes@nyu.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5455-8163

Funding

National Institutes of Health (R01 GM108043)

  • David L Stokes

National Institutes of Health (S10 RR027990)

  • Thomas A Neubert

European Research Council (637372)

  • Bjorn P Pedersen

Independent Research Fund Denmark (DFF-8021-00161)

  • Bjorn P Pedersen

Lundbeckfonden (R82-2011-7280)

  • Himanshu Khandelia

Novo Nordisk Fonden (NNF18OC0034936)

  • Himanshu Khandelia

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Jingjing Deng for initial mass spectrometry analyses. Mass spectra data has been deposited in the MassIVE repository (MSV000084906). The work was supported by NIH grants R01 GM108043 to DLS and S10 RR027990 to TAN, by funding from the European Research Council (grant agreement No. 637372) and the Independent Research Fund Denmark (grant agreement No. DFF-8021-00161) to BPP, and by Lundbeckfonden (grant No. R82-2011-7280) and Novo Nordisk Fonden (grant No. NNF18OC0034936) to HK.

Senior Editor

  1. Olga Boudker, Weill Cornell Medicine, United States

Reviewing Editor

  1. Lucy R Forrest, National Institute of Neurological Disorders and Stroke, National Institutes of Health, United States

Reviewer

  1. Howard Young

Publication history

  1. Received: January 25, 2020
  2. Accepted: September 19, 2020
  3. Accepted Manuscript published: September 21, 2020 (version 1)
  4. Version of Record published: October 5, 2020 (version 2)

Copyright

© 2020, Sweet et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 597
    Page views
  • 90
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Biochemistry and Chemical Biology
    2. Cell Biology
    İbrahim Avşar Ilik et al.
    Research Article

    The nucleus of higher eukaryotes is a highly compartmentalized and dynamic organelle consisting of several biomolecular condensates that regulate gene expression at multiple levels (Banani et al., 2017; Shin and Brangwynne, 2017). First reported more than 100 years ago by Ramón y Cajal, nuclear speckles (NS) are among the most prominent of such condensates (Spector and Lamond, 2011). Despite their prevalence, research on the function of NS is virtually restricted to colocalization analyses, since an organizing core, without which NS cannot form, remains unidentified (Chen and Belmont, 2019; Galganski et al., 2017). The monoclonal antibody SC35, which was raised against a spliceosomal extract, is a frequently used reagent to mark NS since its debut in 1990 (Fu and Maniatis, 1990). Unexpectedly, we found that this antibody has been misidentified and the main target of SC35 mAb is SRRM2, a large (~300 kDa), spliceosome-associated (Jia and Sun, 2018) protein with prominent intrinsically disordered regions (IDRs) that sharply localizes to NS (Blencowe et al., 1994). Here we show that, the core of NS is likely formed by SON and SRRM2, since depletion of SON leads only to a partial disassembly of NS as reported previously (Ahn et al., 2011; Fei et al., 2017; Sharma et al., 2010), in contrast, combined depletion of SON together with SRRM2, but not other NS associated factors, or depletion of SON in a cell line where IDRs of SRRM2 are genetically deleted, leads to a near-complete dissolution of NS. This work, therefore, paves the way to study the role of NS under diverse physiological and stress conditions.

    1. Cell Biology
    2. Biochemistry and Chemical Biology
    Melissa V Gammons et al.
    Research Article Updated

    Feedback control is a universal feature of cell signaling pathways. Naked/NKD is a widely conserved feedback regulator of Wnt signaling which controls animal development and tissue homeostasis. Naked/NKD destabilizes Dishevelled, which assembles Wnt signalosomes to inhibit the β-catenin destruction complex via recruitment of Axin. Here, we discover that the molecular mechanism underlying Naked/NKD function relies on its assembly into ultra-stable decameric core aggregates via its conserved C-terminal histidine cluster (HisC). HisC aggregation is facilitated by Dishevelled and depends on accumulation of Naked/NKD during prolonged Wnt stimulation. Naked/NKD HisC cores co-aggregate with a conserved histidine cluster within Axin, to destabilize it along with Dishevelled, possibly via the autophagy receptor p62, which binds to HisC aggregates. Consistent with this, attenuated Wnt responses are observed in CRISPR-engineered flies and human epithelial cells whose Naked/NKD HisC has been deleted. Thus, HisC aggregation by Naked/NKD provides context-dependent feedback control of prolonged Wnt responses.