Abstract
As a ubiquitous bacterial secondary messenger, c-di-GMP plays key regulatory roles in processes such as bacterial motility and transcription regulation. CobB is the Sir2 family protein deacetylase that controls energy metabolism, chemotaxis and DNA supercoiling in many bacteria. Using an E.coli proteome microarray, we found that c-di-GMP strongly binds to CobB. Protein deacetylation assays showed that c-di-GMP inhibits CobB activity and thereby modulates the biogenesis of acetyl-CoA. Through mutagenesis studies, residues R8, R17 and E21 of CobB were shown to be required for c-di-GMP binding. Next, we found that CobB is an effective deacetylase of YdeH, a major diguanylate cyclase (DGC) of E.coli that is endogenously acetylated. Mass spectrometry analysis identified YdeH K4 as the major site of acetylation, and it could be deacetylated by CobB. Interestingly, deacetylation of YdeH enhances its stability and cyclase activity in c-di-GMP production. Thus, our work establishes a novel negative feedback loop linking c-di-GMP biogenesis and CobB-mediated protein deacetylation.
Introduction
Cyclic diguanosine monophosphate (c-di-GMP) was first identified in Gluconacetobacter xylinus, where it was found to regulate cellulose synthesis1. Subsequently, c-di-GMP was shown to be involved in a wide range of bacterial biological processes such as bacterial motility, biofilm formation, virulence and transcription regulation2-4. However, these processes likely represent only a portion of the full scope of its diverse functions in the cell, owing to the typical challenges of unambiguously ascribing functionality directly to the activity of a specific small molecule second messenger5. A first step toward this understanding has often emerged from an identification of the proteins to which it strongly interacts, as exemplified in studies of c-di-GMP binding to YcgR6, CckA7, BldD8 and CheY-like (Cle) proteins9. Thus, we speculated that a better understanding of the full repertoire of c-di-GMP functions in bacteria would emerge from a comprehensive knowledge of the complete range of proteins to which it binds.
c-di-GMP is synthesized by diguanylate cyclases (DGCs)10 and degraded by specific phosphodiesterases (PDEs)11-13. In E.coli, the dominant DGCs are YdeH (also known as DgcZ)14 and DosC15, and the major PDEs are YhjH16 and DosP15. Each of these proteins has been shown to be modulated by “first” messengers such as light, oxygen and temperature2. Yet, owing to the involvement of c-di-GMP in the aforementioned fundamental biological processes that are all well-regulated by many basic metabolic mechanisms17, 18, there is also the possibility that c-di-GMP biosynthesis is also regulated via metabolism-related intracellular signals.
The Sir2 family protein CobB is a NAD+-dependent deacetylase that is highly conserved in prokaryotes19, 20. In E.coli, CobB is the sole Sir2 homolog, although there are two forms of CobB, namely, CobB and CobBS, the former of which has an additional 37 aa N-terminal tail21. CobB exhibits protein deacetylation activity and it regulates a variety of physiological functions. For example, CobB deacetylates lysine-609 of acetyl-coenzyme A synthetase (Acs) to activate its activity22, resulting in an increased cellular concentration of acetyl-coenzyme A (acetyl-CoA), which is a central component of energy metabolism. In addition, CobB regulates E.coli chemotaxis by deacetylating the chemotaxis response regulator protein (CheY)23, as well as the activity of N-hydroxyarylamine O-acetyltransferase (NhoA)24 and topoisomerase I (TopA)25. Yet, despite the critical role that CobB plays in many biological processes, its inherent regulation remains poorly understood26-28.
To globally identify c-di-GMP effectors and explore new functions of c-di-GMP, we employed an E.coli proteome microarray29 for proteome-wide identification of c-di-GMP binding proteins. Surprisingly, we found that c-di-GMP strongly binds to CobB. Subsequent biochemical analysis confirmed that c-di-GMP binds to CobB and inhibits its deacetylation activity both in vitro and in vivo. Furthermore, we found that the major DGC in E.coli, YdeH, is endogenously acetylated and CobB promotes the stability and activity of YdeH through deacetylation of lysine-4. Altogether, we have established an evidence-based regulation loop underlying the cytoplasmic concentration of c-di-GMP that involves its direct binding to, and thereby inhibition of CobB.
Results
Protein deacetylase CobB is a novel c-di-GMP effector
To identify novel effectors of c-di-GMP, an E.coli proteomic microarray was probed with biotin-c-di-GMP. In this way, CobB (specifically, the version of CobB with the extra N-terminal tail) was identified as a strong binder of c-di-GMP (Fig. 1a). To validate this interaction, we developed a simple in vitro assay in where purified CobB was incubated with biotin-c-di-GMP, UV crosslinked, and then probed with fluorescent streptavidin.30, 31 We found that c-di-GMP indeed exhibits strong binding to CobB, although trace bindings were also observed for cGMP and c-di-AMP (Fig. 1b). Mixing c-di-GMP with biotin-c-di-GMP before the two were added to CobB resulted in a significant reduction in biotin-c-di-GMP binding to CobB (Fig. 1b). To quantify the binding strength of this interaction, we performed Isothermal Titration Calorimetry (ITC) and found that the c-di-GMP/CobB affinity constant, Kd, is 21.6 μM with a binding stoichiometry of 0.97. We also performed ITC assays with cGMP and c-di-AMP, and no binding was detected with these two molecules (Fig. 1c and supplementary Fig. 1a-c). To confirm these results, we also determined the affinity constant using Microscale Thermophoresis (MST), which yielded a Kd of 22.7 ± 1.63 μM for the interaction between c-di-GMP and CobB (Supplementary Fig. 2). We note however that while the affinity constants determined from ITC and MST are consistent with each other, we observed some polymerization of CobB in the presence of c-di-GMP (Supplementary Fig. 3), which may somewhat affect the measurement of the affinity for c-di-GMP. Nonetheless, taken together, these results clearly show that c-di-GMP specifically binds to CobB.
c-di-GMP inhibits the deacetylase activity of CobB and down-regulates the biogenesis of acetyl-CoA
With the demonstration of binding between c-di-GMP and CobB, we speculated that c-di-GMP may affect the deacetylase activity of CobB. To test this, we examined the activity of CobB in the presence of c-di-GMP by monitoring the deacetylation of the well-known CobB substrates, Acs22, CheY23, NhoA24. Before incubation with CobB, we found that purified Acs from a CobB-deficient E.coli (ΔcobB) is highly acetylated (Fig. 2a). Incubation with CobB resulted in significant reduction in acetylation, consistent with previous work22 (Fig. 2a). However, in the presence of c-di-GMP, the deacetylation of Acs was significantly inhibited in a dose dependent manner (Fig. 2a and supplementary Figure 4a). By contrast, neither the presence of cGMP nor c-di-AMP affected the CobB deactylation of Acs. Similar results were obtained for CheY (Supplementary Figure 4b) and NhoA (Supplementary Figure 4c).
To examine the inhibition of c-di-GMP on CobB in more detail, we measured the kinetics of CobB deacetylation of an acetylated peptide32at different concentrations of c-di-GMP. We found that c-di-GMP significantly reduces the maximal catalytic rate of CobB, with no changes to the Km values, yielding a Ki of c-di-GMP for CobB of 32.27 μM (Fig. 2b). Thus, c-di-GMP noncompetitively inhibits the deacetylation activity of CobB, suggesting that the binding region of c-di-GMP on CobB is not in the catalytic pocket.
To further confirm the inhibition of c-di-GMP on CobB activity, we sought to alter the endogenous levels of c-di-GMP and test the deacetylation activity of CobB in vivo. In E.coli, YdeH14 is the major DGC that produces c-di-GMP. Thus, we examined an E.coli strain with overexpressed YdeH (ydeH+) and a strain expressing a dysfunctional mutant YdeH (ydeHG206A,G207A)14 to produce strains with high and low levels of c-di-GMP, respectively. To monitor the activity of CobB, we examined the acetylation level of endogenous Acs to which a chromosomal C-terminal FLAG-tag was attached. We also examined a series of CobB-depleted strains, namely ΔcobB, ΔcobB::cobB and ΔcobB::cobB ydeH+, to further investigate this interaction. As expected, high levels of c-di-GMP were clearly observed for strains with YdeH+ (Fig.2c). In these strains, we found a significant reduction of deacetylation of Acs (Fig.2d). We found that this deacetylation of Acs is abolished by the deletion of CobB (ΔcobB), and that this could be recovered by putting back CobB (ΔcobB::cobB). Thus, these results indicate that c-di-GMP increases the acetylation levels of Acs in a CobB-dependent manner.
It is known that CobB activates Acs through deacetylation of K609, and Acs is responsible for the synthesis of acetyl-CoA, which is essential for cell growth22. It is also known that both acetate and propionate can serve as a donor for acetyl-CoA for cell growth33, 34 (Fig.2e). To further confirm the regulatory role of c-di-GMP on acetyl-CoA synthesis, all of the aforementioned strains (Fig.2f) were cultured using acetate or propionate as the sole carbon source. For the CobB deficient cells (ΔcobB), we observed an obvious growth defect, consistent with a lack of activated Acs. Similarly, overexpression of YdeH significantly inhibited cell growth in the ydeH+ and ΔcobB::cobB ydeH+ strains. These results are consistent with an inhibition of CobB by high levels of c-di-GMP (Fig.2f, Supplementary Table 1). We also found that the ydeHG206A,G207A strain exhibited a similar growth as that of the WT strain. The inhibition of growth in the strains with ydeH+ was similar at concentrations of acetate or propionate of 10 mM or 30 mM (Supplementary Fig. 5, Supplementary Table 1). Thus, in addition to confirming the c-di-GMP inhibition of CobB in vivo, these results strongly suggest that c-di-GMP is a physiologically relevant effector of the regulation of acetyl-CoA biogenesis through the inhibition of the deacetylation activity of CobB.
c-di-GMP globally affects CobB-dependent deacetylation in vivo
To determine whether c-di-GMP could affect CobB-dependent deacetylation in a global setting, we applied Stable Isotope Labeling with Amino acids in Cell culture (SILAC) coupled with MS to quantitatively compare the levels of protein acetylation in WT and ydeH+ cells. Previously, Weinert et al.34 identified 366 CobB regulated acetylation sites when E.coli was cultured in M9 media supplemented with 0.2% glucose. In addition, Cerezo et al.35 identified 283 acetylation sites in E.coli under carbon-limited conditions. Since the acetylome of E.coli varies significantly under different culture conditions, to facilitate comparisons with previous work, we adopted the procedure developed by Weinert et al.34 with slight modifications (see Methods). A total of 802 acetylation sites (Data set S1) were identified, of which 107 (Data set S2) exhibited enhanced acetylation upon overexpression of YdeH (Fig. 3a, b, supplementary Fig. 6). Of the CobB regulated acetylation sites identified by Weinert et al.34 (Data set S3), 43 were discovered in our study, 28 of which were among those that exhibited greater acetylation in ydeH+ (Fig. 3c, data Set S4). Hence, CobB regulated sites are enriched in ydeH+ cells. Furthermore, we mapped the 107 c-di-GMP upregulated sites to 87 proteins and found that 42 of these proteins overlap with the 271 CobB regulated proteins identified previously34 (Fig. 3d, data Set S5). We also compared the functional categories of the c-di-GMP upregulated proteins with the CobB regulated proteins and found that these two sets of proteins are highly similar in several classifications (Supplementary Fig. 7). We note that slight differences in our procedure from that described by Weinert et al. (particularly the absence of fractionation before MS analysis) may explain the lower numbers of acetylation sites identified here compared to the earlier work (802 vs. 3,680). Nevertheless, our data strongly indicate that c-di-GMP regulated acetylation is closely related to CobB dependent deacetylation. Thus, these results suggest that c-di-GMP globally affects CobB-dependent protein deacetylation in vivo.
Mutagenesis and binding studies of the CobB and c-di-GMP interaction
There is an additional 37 aa N-terminal tail in CobB compared to that of CobBS21 (Fig. 4a). Streptavidin blotting assays showed that truncation of the N-terminal fragment of CobB disrupts c-di-GMP binding, suggesting that residues 1-37 of CobB are essential for c-di-GMP interaction (Fig. 4b). It is known that c-di-GMP binds to many of its effectors by Arg, Leu, Asp and Glu residues36. To determine the exact binding sites on CobB, we mutated all Arg, Leu, Asp and Glu residues to Ala within the CobB N-terminal domain. We found that only CobB mutants with R8A, R17A and E21A exhibited a weakened interaction with c-di-GMP (Fig. 4c). We next performed ITC titration with these mutants and found that the Kdvalues are 1.8 mM, 1.6 mM and 0.32 mM for CobBR8A, CobBR17A and CobBE21A, respectively (Fig. 4d), which is 15 to 83-fold lower than that of the wild type CobB (Fig. 1c and supplementary Fig. 1d-f). Despite a loss of c-di-GMP binding, CobBR8A, CobBR17A and CobBE21A displayed similar deacetylase activity as CobB. Importantly, addition of c-di-GMP did not inhibit the activity of these mutants (Fig. 4e). Collectively, these data suggest that R8, R17 and E21 are important for c-di-GMP binding but do not directly participate in the catalytic activity of CobB.
To further validate the specific binding of c-di-GMP to CobB, we constructed strains of these three CobB mutants, i.e., ΔcobB::cobBR8AydeH+, ΔcobB::cobBR17A ydeH+ and ΔcobB::cobBE21A ydeH+ based on an E.coli strain that includes the chromosomal Flag-tagged Acs. These strains were cultured with acetate or propionate and the c-di-GMP concentrations were measured (Supplementary Fig. 8a). To monitor the in vivo activity of CobB, we examined the acetylation of endogenous Acs. As expected, the acetylation levels of Acs were significantly lower in the strains with these CobB mutants, compared with that in the ΔcobB::cobB ydeH+ strain (Fig. 4f). Additionally, these CobB mutant strains showed similar growth rates as ΔcobB::cobB, but higher growth rates than ΔcobB::cobB ydeH+ (Fig. 4g, supplementary Fig. 8b-d). Hence, these results provide in vivo evidence for the role of these residues in CobB in the binding of c-di-GMP.
The binding and inhibition of the deacetylation activity of CobB by c-di-GMP is conserved among prokaryotes
CobB is a Sir2 homolog that is highly conserved in prokaryotes. We thus hypothesized that the binding and inhibition of c-di-GMP observed with CobB from E.coli is the same for the CobB homologues in other prokaryotes. To test this hypothesis, CobB protein sequences from a series of highly diverse bacteria were aligned (Supplementary Fig. 9). We found that the c-di-GMP binding region is reasonably well conserved in these bacteria (Fig. 5a). Examining S. typhimurium CobB as an exemplary member of this conserved set, we found that CobBS. typhimurium binds to c-di-GMP, but not to c-di-AMP and cGMP (Fig. 5b). Furthermore, we determined the affinity of S. typhimurium CobB and c-di-GMP to be 15.5 μM (Supplementary Fig. 10), and that the CobBS.typhimurium deacetylation of Acs could be clearly inhibited by c-di-GMP, but not cGMP and c-di-AMP (Fig. 5c).
CobB deacetylates YdeH on K4 to activate its DGC activity
Protein acetylation is one of the most abundant post-translational modifications in bacteria (and eukaryotes), with hundreds of acetylated proteins already identified in E.coli37, 38. We speculated that some of the proteins involved in c-di-GMP metabolism are endogenously acetylated and could be deacetylated by CobB. We selected 4 DGCs (YdeH, YaiC, YegE and YeaJ), 2 PEDs (YahA, DosP), and 3 GTPases (Era, YihK and YihI) from E.coli to test this possibility. We found that the 2 DGCs (YdeH, YegE) and 2 GTPases (Era, YihK) were acetylated (Fig. 6a). Further, treating these proteins with CobB, we found that YdeH (DGC) and Era (GTPase) could be effectively deacetylated by CobB (Fig. 6a). To determine the residue(s) of YdeH targeted for deacetylation by CobB, both acetylated YdeH and CobB-treated deacetylated YdeH were subjected to mass-spectrometry analysis. In this way, YdeH K4, K170, K277 were determined as the sites that could be deacetylated by CobB (Fig. 6b, supplementary Fig. 11). To validate these sites, we mutated all 8 Lys residues of YdeH to Ala individually, and found that only the YdeH K4 mutant exhibited a significantly decreased acetylation level (Fig. 6c). These results indicate that K4 is the dominant acetylated site on YdeH. In addition, two other acetylation sites were also identified on Era (K171) (Supplementary Fig. 12a) and YegE (K936) (Supplementary Fig. 12b).
To verify the functional role of YdeH K4 acetylation, we mutated the Lys to Arg (YdeHK4R), Gln (YdeHK4Q) or Ala (YdeHK4A). Western blotting showed that these mutations significantly reduced the level of acetylation of YdeH compared to that of WT YdeH (Fig. 6d). Since YdeH is the major DGC in E.coli, we then set to test whether CobB deacetylation could affect the DGC activity of YdeH using UPLC-IM MS (Fig. 6e). Upon CobB treatment, WT YdeH showed a 2-fold increased production of c-di-GMP, while the activities of YdeHK4R, YdeHK4Q and YdeHK4A were 69.5%, 29.5%, and 27.1% of that of the WT, respectively (Fig. 6f). Unexpectedly, the mutant YdeHK4R that structurally resembles the deacetylation form of YdeH also exhibited a lower DGC activity. Evidently, the “K” to “R” mutation at residue 4 fails to fully mimic the deacetylation status of this site. Nonetheless, these results suggest that K4 is critical for the DGC activity of YdeH.
To test whether CobB can regulate the c-di-GMP levels in vivo, we constructed several ydeH+ and ΔcobB ydeH+ strains, including those with CobB mutants or with YdeH mutants, and then determined the YdeH acetylation and c-di-GMP levels. We found that the acetylation level of YdeH in ydeH+ is lower than that of ΔcobB ydeH+ and the c-di-GMP level of ydeH+ is significantly higher than that of ΔcobB ydeH+ (Fig. 6g). These results confirm that CobB can regulate the DGC activity of YdeH through deacetylation. The strains with the YdeH mutants exhibited the lowest DGC activity, confirming that K4 is critical to the DGC activity of YdeH (Fig. 6g). Interestingly, the strains with the CobB mutants exhibit lower acetylation levels than WT CobB (Fig. 6g). These results indicate that c-di-GMP does not affect the activity of the CobB R8A, R17A and E21A mutants in vivo. Still, the CobB mutants exhibited higher c-di-GMP levels, further confirming that CobB promotes the DGC activity of YdeH through deacetylation.
CobB enhances the solubility/stability of YdeH through deacetylation
We found that the acetylation level of endogenous YdeH is significantly decreased in WT cells compared to that of CobB-deficient cells (Fig. 7a). In order to better characterize the endogenous stoichiometry of YdeH K4 acetylation, we used the absolute quantification (AQUA) method39, 40 to quantify the level of the K4 acetylated peptide. As shown in Fig. 7b, the YdeH K4 acetylation stoichiometry was 1.3 ± 0.2 % for the CobB deficient cells, while it is undetectable for the WT cells.
Since it is known that acetylation can lead to protein degradation in vivo41, we speculated that acetylation of YdeH may affect its stability. Indeed, we found that the level of endogenous soluble YdeH deceased 40% in CobB deficient cells as compared to that of the WT cells (and also the CobB recovered cells) in exponential growth (Fig. 7c), and more significantly, a decrease of 70% soluble YdeH was observed for stationary growth (Supplementary Fig. 16c). Furthermore, for stationary growth, a significant amount of precipitated YdeH was observed for CobB deficient cells, but not for WT and CobB recovered cells (Supplementary Fig. 16c). We then examined the protein stability of YdeH mutants in vitro and found that YdeHWT and YdeHK4Q are less stable than YdeHK4A and YdeHK4R (Supplementary figure 13). In addition, we also noticed that the level of soluble YdeH following the addition of CobB is 2-fold of that in the absence of CobB (Fig. 7e).
Additionally, our data show that the c-di-GMP increase by 36% in WT cells than CobB deficient cells in vivo under exponential growth when YdeH was overexpressed (Fig. 6g). To understand the underlying mechanism, we then determined the level of YdeH when it was overexpressed in WT cells and CobB deficient cells. The results showed that the soluble YdeH level decreased ~45% in CobB deficient cells as compared to that of the WT cells (Fig. 7f), which is similar to that of the endogenous YdeH under exponential growth (Fig. 7c). We also measured the YdeH level in the sediment and observed significant amount of precipitated YdeH in CobB deficient cells as compared to that of WT and CobB recovered cells (Fig. 7f).
Thus, it appears as though YdeH K4 acetylation lead to protein aggregation/precipitation by reducing its solubility/stability (Fig. 8a). In WT cells, the acetylation of YdeH is removed by CobB and retained in soluble state, with little or undetectable protein sediment (Supplementary Fig. 16c). In CobB deficient cells, more acetylated YdeH will precipitate with both acetylated and/or deacetylated YdeH because of the lack of CobB deacetylation. Additionally, this model also explains why YdeH K4 may be frequently acetylated but cannot accumulate to high level in vivo. It is known that acetylation on specific sites promote protein aggregation42, 43. For example, K280 acetylation causes the aggregation of tau. And tau promotes neuronal survival, thus the acetylation induced tau aggregation is pathologically significant. Interestingly, tau K280 belongs to a double lysine motif, i.e., 275VQIINKK281, and YdeH K4 belongs to a similar double lysine motif, i.e., 1MIKK4. Thus, it is possible that the underlying mechanism of lysine acetylation induced protein aggregation of YdeH may be similar to that of tau. However, because of the possible complexity, the detailed mechanism is yet to be discovered.
DISCUSSION
c-di-GMP is a key secondary messenger in prokaryotes. CobB is the first, and major, protein deacetylase identified in prokaryotes. Herein, we discovered that c-di-GMP specifically binds to CobB and inhibits its deacetylase activity, and down-regulates the cellular concentration of acetyl-CoA through modulation of the acetylation levels of Acs. Interestingly, we also found the major E.coli DGC, YdeH, is endogenously acetylated, and CobB enhances the solubility/stability of YdeH, and activates the DGC activity of YdeH, through deacetylation. Thus, we established a negative feedback regulatory loop between c-di-GMP biogenesis and CobB dependent protein deacetylation.
The binding region and kinetics of CobB and c-di-GMP
The most well-known motif of c-di-GMP binding is the EXLXR motif44. Interestingly, the c-di-GMP binding site on CobB contains RXLXE, which is precisely EXLXR in reverse. As expected in the EXLXR motif, both R17 and E21 are critical for c-di-GMP binding, while the L19 residue does not directly bind to c-di-GMP in the RXLXE motif. This motif is located in the N-terminal tail of this protein that we found is responsible for its dimerization (Supplementary Figure 3). Since many of the c-di-GMP binding proteins are dimers or tetramers4, 31, 45, a plausible explanation for this effect of c-di-GMP described here is that c-di-GMP interferes with the dimerization of CobB.
A concern with our model though is that the Kd and Ki of c-di-GMP for CobB measured in vitro are larger than the sub- to micro-molar concentrations of c-di-GMP generally observed in vivo46. While this may be owing to presently unidentified molecular factors that reduce this affinity in vivo, we believe that this discrepancy could be explained by the spatially and temporally uneven distribution of c-di-GMP in the bacteria. Indeed, direct measurement of the concentration of c-di-GMP in bacteria using a FRET biosensor has revealed a wide range of concentrations within individual cells47, including some resolution-limited locations (~200 x 200 x 1000 nm3) that exhibit a local concentration significantly greater than 1 μM. We note that this concentration corresponds to only 25 molecules within this volume. Further, it is well known that the bacterial cytoplasm is extremely crowded48, and recent work has shown that, as a result, effectively, there are caging effects on the free diffusion of particles49. A single molecule of c-di-GMP within a “cage” of only 20 x 20 x 100 nm3 is at a concentration of 40 μM. Thus, we speculate that it is indeed physically possible that the local concentration of c-di-GMP could exceed the measured affinity constants for CobB in the WT bacteria. In fact, owing to the mechanism described here, the negative regulatory loop involving CobB and YdeH may be principally responsible for keeping the local (and thus global) cytosolic concentration to the observed maximal levels.
c-di-GMP regulates physiological functions through inhibiting of CobB deacetylation
As the most important substrate of CobB, Acs is responsible for the synthesis of acetyl-CoA, which controls cell energy metabolism and global protein acetylation, and affects cell growth and proliferation50-52. Our results indicate that c-di-GMP can lower the concentration of acetyl-CoA in vivo through inhibiting CobB and increasing the acetylation level of Acs. Previous studies have identified several possible overlapping effects of c-di-GMP and acetyl-CoA. For example, c-di-GMP affects the expression of acetate kinase (AckA) through the binding of the transcription factor that regulates AckA transcription in B. burgdorferi53. In addition, acyl-CoA dehydrogenase, a key enzyme in acetyl-CoA metabolism pathways, is a genuine c-di-GMP effector in B. bacteriovorus54. Here, we strengthened the link between c-di-GMP and acetyl-CoA by showing, for the first time, that c-di-GMP directly modulates the biogenesis of acetyl-CoA.
c-di-GMP has emerged as a key regulator in the decision between motile and sedentary forms of bacteria2. Elevated c-di-GMP levels inhibit bacterial motility via effects on the flagella-associated protein6. Most recently, Nesper et al. found that c-di-GMP directly binds to the CheY-like regulators and tunes the bacterial flagellar motor9. According to our data, c-di-GMP can regulate the acetylation level of CheY via inhibition of the deacetylase activity of CobB. Thus, it is highly possible that c-di-GMP regulates bacterial motility through CobB-mediated regulation of CheY activity.
CobB is a member of the sirtuin protein family, which are highly conserved across prokaryotes to eukaryotes. Thus, it is possible that c-di-GMP may also bind human sirtuins and play critical functional roles in a variety of biological processes, such as host-pathogen interactions.
CobB regulates the YdeH DGC activity in vitro
The overexpressed YdeH was used for DGC activity assay in vitro. We determined the K4 acetylation stoichiometry of the overexpressed YdeH following the same AQUA procedure that we used for the endogenous YdeH. The results showed that the overexpressed YdeH K4 acetylation stoichiometry is 28.7 ± 4.9% (Supplementary Fig. 14). Assuming that all the K4 acetylation could be deacetylated by CobB, we could thus expect a ~1.4-fold activity increase upon CobB deacetylation. But it is still could not fully explain the 2-fold activity increase upon CobB deacetylation (Fig. 6f). To this end, we sought to determine, with a simple model, whether the loss of stable YdeH could completely account for the 2-fold difference in c-di-GMP. In particular, based on the stability data (Fig. 7e), we considered the case in which the difference in c-di-GMP is completely owing to the decrease in soluble protein. Assuming that the substrate is not limiting throughout the assay and that steady state is achieved after a very short time, the lower amount of c-di-GMP produced would simply be a result of the lower amount of soluble protein at any time point. We found that the decrease in soluble protein is well-described by a single exponential (Supplementary Fig. 15) where C is the concentration of YdeH and k is the rate of the protein loss, determined to be 0.49 hr-1 in the fit. Thus, the fold-increase, F, expected for the fully stable (CobB treated) YdeH relative to the more unstable version can be calculated (for the 2 hrs assay) by with the aforementioned value of k. Thus, simple loss of the active protein would result in a 1.57-fold difference in c-di-GMP. Since we observed a 2-fold difference, we conclude that the loss of soluble protein can account for ~78% (1.57/2) of the difference measured with the acetylated and deacetylated YdeH. Hence, a difference in the inherent activity between acetylated-YdeH and deacetylated-YdeH is needed to fully account for observed difference in c-di-GMP produced.
We noticed that the K4 acetylation stoichiometry of overexpressed YdeH is much high that of endogenous YdeH. The endogenous acetylation stoichiometry of YdeH K4 was measured using the exponential growth of E. coli. While as a common practice, when overexpress a protein in E. coli, the strain is usually induced at exponential growth, and then cultured for another ~2 hours for fast protein producing. At the time of protein purification, the strain is already at stationary phase. Thus, it is possible that the big difference of acetylation stoichiometry between the overexpressed YdeH and the endogenous one is at least partially due to the difference of the growth phases.
To test this possibility, we measured the YdeH K4 acetylation stoichiometry using E. coli of stationary growth. The results showed that the YdeH K4 acetylation stoichiometry is 1.1 ± 0.3% in WT cells and 4.2 ±0.5% in CobB deficient cells (Supplementary Fig. 16a-b). For CobB deficient cells, the YdeH K4 acetylation stoichiometry is 3-fold(4.2%/1.3%) in stationary growth than that of exponential growth. At least, this data could partially explain the stoichiometry gap between the endogenous and overexpressed YdeH.
CobB regulates the YdeH DGC activity in vivo
It is known that the expression of YdeH is tightly regulated by the RNA binding protein CsrA and the the csrA mutant can increase the transcription of YdeH in 15-fold and another DGC YcdT in 45-fold55. Thus, to artificially generate a condition of YdeH high expression for functional analysis, the csrA mutant has often been applied56. However, the csrA mutation can also activate the expression of another DGC, i.e., YcdT, which could cause high c-di-GMP background when studying YdeH. So, we chose to overexpress YdeH in E. coli to study the regulation role of CobB to YdeH inside the cell. Our data show that the c-di-GMP increase by 36% in WT cells than CobB deficient cells in vivo under exponential growth when YdeH was overexpressed (Fig. 6g), which is consistent with the observed 45% decrease of soluble YdeH level in CobB deficient cells than that of WT cells (Fig. 7f). Hence, the 36% lower c-di-GMP in CobB deficient cells is mainly due to the instability/precipitation of YdeH.
Though it is clear now that the major functional consequence of K4 acetylation is the decreasing of YdeH stability, the direct activity change upon acetylation and deacetylation by CobB may still play important role. Our results showed that YdeHK4A, which has comparable stability with WT YdeH, also exhibited a loss ~50 % of its DGC activity (Fig. 6f). It is possible that the YdeH K4 acetylation may significant affect the structure of YdeH. To understand the mechanism, we proposed a working model that illustrates how CobB could regulate the structure of YdeH, based on an analysis of the atomic structure of YdeH56. It has been reported that dimerization and proper conformational rearrangement of the active center are required for optimal YdeH DGC activity (Supplementary Fig. 17a). Reside K4 is located within a “hinge” region of YdeH and may potentially regulate the conformational rearrangement of the active center upon dimer formation (Supplementary Fig. 17b-d). In the acetylated form of K4, the two GGDEF catalytic domains of YdeH are misaligned that may cause the aggregation and then lead to precipitate. (Supplementary Fig. 17b) Besides, this misaligned region could prevent productive ligation of the two GTP molecules, each captured by a GGDEF domain. According to our enzymatic studies, it is very likely that deacetylation of YdeH K4ac triggers the proper realignment of the GGDEF dimer so that the active center is rearranged to allow the formation of intermolecular phosphoester bonds between the two GTP substrates (Supplementary Fig. 17c). Interestingly, the proposed model is close to the Zn-regulation mechanism in YdeH, i.e., inhibits YdeH’s DGC activity by hindering its dimerization56. Thus, at least in part, CobB may regulate the active/inactive switch of DGC activity of YdeH through deacetylation of YdeH on K4 (Supplementary Fig. 17d).
The interplay between CobB and c-di-GMP
Our data show that c-di-GMP can significantly reduce of the decetylation activity of CobB through direct binding, and that the deacetylation of overexpressed YdeH by CobB leads to an increase of c-di-GMP by 36% in vivo. In addition, c-di-GMP is synthesized from GTP by DGC. CobB regulates the levels of acetyl-CoA through deacetylating Acs and then modulates the generation of GTP in the tricarboxylic acid (TCA) cycle. Thus, to a certain extent, the biogenesis of c-di-GMP could be activated by CobB through its effects on the TCA cycle. We also found that a GTPase Era and the DGC YegE are also endogenously acetylated. These results further demonstrate that the biogenesis of c-di-GMP is regulated by (de)acetylation.
The interplay between c-di-GMP and CobB may play important roles in bacteria. It is known that c-di-GMP participates in motility, biofilm formation and virulence, all of which are regulated by energy metabolism57, 58. We found that c-di-GMP modulates the bacterial energy metabolism through inhibiting the deacetylation activity of CobB, thus a connection between important bacterial processes and energy metabolism could be established. In addition, CobB is prominently implicated in the regulation of metabolism59 and c-di-GMP is involved in GTP metabolism and purine metabolism, suggesting a possible two-way regulation between c-di-GMP and CobB at the metabolic level.
Taken together, we established a novel feedback regulation loop between c-di-GMP and the deacetylation activity of CobB (Fig. 8b). The findings of both directions (c-di-GMP inhibits CobB and CobB promotes c-di-GMP biogenesis) are novel. From this loop, we can envision a tightly regulated balance between the levels of c-di-GMP and the protein deacetylase activity of CobB. We strongly believe that our findings will facilitate future functional studies of both c-di-GMP and CobB-based regulation of protein acetylation.
Methods
Bacterial Strains, Plasmids, and CobB Mutant Construction
In this study, we used E.coli BW25113 as the wild type strain and the plasmid pSUMO10 for the overexpression of CobB, pET32a for the rescue experiment, and pCA24N for the overexpression of YdeH, Acs, CheY, NhoA, Era, YeaJ, YaiC, YahA, DosP, YihI, YegE and YihK. We performed the CobB mutations using the QuikChange® Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, USA).
E.coli Proteome Microarray Screening and Data Processing
The E.coli proteome microarrays were prepared as described previously29. The proteome microarrays were first blocked with blocking buffer (3% BSA in 0.1% Tween 20; TBST) for 1 h at room temperature. Bio-c-di-GMP was then diluted to 1 μM incubated on the microarray at room temperature for 1 h, and the same concentration of biotin was included as a negative control. The microarrays were next washed with TBST three times for 5 min each and then were incubated with Cy3-Streptavidin at a 1:1000 dilution (Sigma-Aldrich, Darmstadt, Germany) for 1 h at room temperature, followed by three washes with TBST for 5 min each. The microarrays were spun dry at 250 × g for 3 min and were scanned with a GenePix pro 6.1 microarray scanner to visualize and quantify the results. The raw data have been published in the Protein Microarray Database (www.proteinarray.cn) with the accession number, PMDE271. The auto-analysis tool in Protein Microarray Database (http://www.proteinarray.cn/index.php/analysis-toolkits) was used to process the microarray data.
Preparation of c-di-GMP
It is known that c-di-GMP (BIOLOG Life Science Institute, Bremen, German) may form dimer or polymer, affects the binding with its effectors. Thus, we first determined the c-di-GMP monomer population. The OD276 and OD289 of c-di-GMP were measured using the cuvette mode of Nanodrop 2000c (Thermo Fisher Scientific, MA, USA) with a volume of 100 μL. To calculate the ratio of monomer of c-di-GMP (Pmono)at various concentrations, we applied the equation (pmono = 1.15 (A276/A289) − 1.64) developed by Gentner et al.60 To eliminate the possible effect of c-di-GMP polymer, we heated the c-di-GMP solution in a water bath at 60°C for 1 h to depolymerize ~80% c-di-GMP to monomer before testing.
Streptavidin blotting assay
In this assay, CobB (0.5 mg/mL, 16.13 μM) and 10 μM bio-c-di-GMP were incubated in the deacetylation buffer (50 mM Tris-HCl, 4 mM MgCl2, 50 mM NaCl, 50 mM KCl, 1 mM NAD+, pH 8.0) at 37°C for 1 h and the same amount of biotin, cGMP, c-di-AMP were included as negative controls. The samples were then UV-cross linked on ice with 10 min for 1.2 mJ/cm2 energy to further link the c-di-GMP to CobB. These samples were then analyzed by Western blotting. After incubation with IRDye 800-CW Conjugated Streptavidin (LI-COR Biosciences, Nebraska, USA) for 2 h, the membranes were washed with TBST three times and visualized with an Odyssey Infrared Imaging System (LI-COR Biosciences, Nebraska, USA). And the protein level was determined by Ponceau S staining is the (Sangon Biotech, Shanghai, China).
ITC assay
In the ITC assay, c-di-GMP and the wild-type CobB and mutant CobB proteins were prepared in the titration buffer (20 mM Tris, 250 mM NaCl, 2 mM DTT, and pH 7.5). Protein concentrations were measured based on the UV 280nm absorption. The ITC titrations were performed using a MicroCal iTC200 system (GE Healthcare, Pittsburgh, USA) at 25°C. Each titration consisted of 17 successive injections (the first at 0.4 μL and the remaining 16 at 2.4 μL). The stock c-di-GMP, cGMP and c-di-AMP at 1.5 mM were titrated into wild-type or mutant CobB (0.1 mM) in the sample cells of 200 μL volume individually. c-di-GMP, cGMP and c-di-AMP of 1.5 mM were titrated into 200 μL titration buffer as controls for data processing. The resultant titration curves were processed using the Origin 7.0 software program (OriginLab) according to the “one set of sites” fitting model.
CobB deacetylase activity assay
CobB (50 μg/mL, 1.6 μM) and c-di-GMP (at either 0.25 or 0.5 mM) were incubated in 20 μL deacetylation buffer (50 mM Tris-HCl, 4 mM MgCl2,50 mM NaCl, 50 mM KCl, 1 mM NAD+, pH 8.0) at 37°C for 0.5 h. For negative controls, we used 0.5 mM cGMP and c-di-AMP. The CobB substrates (4.16 μM Acs, 1.41 μM CheY, 0.62 μM NhoA) were then added and incubated at 37°C for 1.5 h. These proteins were analyzed by both silver staining and Western blotting. Membranes were incubated with a pan anti-acetyl antibody (Cell Signaling Technology, MA, USA, with a 1:1000 dilution) for at 4°C for 12 h and then incubated with an IRDye 800 secondary antibody at room temperature for 1 h. The membranes were visualized with an Odyssey Infrared Imaging System.
Measuring the catalytic kinetics of CobB
CobB (9.3 μg/mL, 0.3 μM) was incubated with 0, 10, 20 and 80 μM c-di-GMP at 37°C for 20 min in 80 μL deacetylation buffer, and then 20 μL gradient concentrations acylated peptide (LEQIAELAGVSKacTNLLYYFPSK)32 (1, 2, 4, 8, 16, 32, 50, 75, 100, 150 and 200 μM) were added and co-incubated at 37°C for 30 min. 100 mM HCl and 160 mM acetic acid were added to stop the reactions and spun for 10 min at 18,000 x g to separate the enzyme from the reactions. These samples were analyzed by HPLC. Briefly, samples were injected onto a C-18 column (AlltimaTM C18 4.6 x 250 mm) and analyzed by reversed-phase HPLC (Shimadzu, Japan). Solution A (0.065% trifluoroacetic acid in 100% water (v/v)) and solution B (0.05% trifluoroacetic acid in 100% acetonitrile (v/v)) were used in a gradient program (0.01 min with 5% Solution B, 25 min with 65% Solution B, 25.01 min with 95% Solution B, 31 min with 95% Solution B, 31.01 min with 5% Solution B, 40 min with 5% Solution B and stop in 40.01 min) with a flow rate of 1 mL/min. Peptides were detected at 220 nm wavelength. This assay was performed three preparations and Vmax, Kivalues were calculated by curve-fitting the plot using GraphPad Prism 6.
Construction of an E.coli strain (BW25113) harboring chromosomal 3xFLAG-tagged Acs and 3xFLAG-tagged YdeH
E.coli strain (BW25113) harboring chromosomal 3xFLAG-tagged Acs and 3xFLAG-tagged YdeH were constructed using the Red recombination system61, as described previously32.
Isolation and Quantification of c-di-GMP in E.coli
The c-di-GMP isolation methods were described previously62. Briefly, the total amount of E.coli cells with a 50 OD were harvested and re-suspended in 2 mL ddH2O. To extract intracellular c-di-GMP, 8 mL extract mixture of 50% methanol and 50% acetonitrile were added. We also added 1 μM cGMP as the reference. The mixture was incubated in boiling water for 10 min and then centrifuged at 10,000 rpm for 10 min. The supernatant was transferred to a new tube for freeze-drying. The freeze-dried pellet was re-suspended in 100 μL water containing 50% methanol. Samples were analyzed by UPLC-IM-MS (Ultra high-performance liquid chromatography coupled ion mobility mass spectrometry). UPLC-IM-MS was performed using a Waters UPLC I-class system equipped with a binary solvent delivery manager and a sample manager, coupled with a Waters VION IMS Q-TOF Mass Spectrometer equipped with an electrospray interface (Waters Corporation, Milford, USA) at the Instrumental Analysis Center of Shanghai Jiao Tong University. UPLC was performed on a ZIC-HILIC column (100 mm × 2.1 mm i.d., 3.5 µm; Merck). The column was eluted with gradient solvent from A: B (5: 95) to A: B (40: 60) at a flow rate of 0.40 mL/min, where A is 50 mM ammonium formate and B is acetonitrile. We employed the following MS experimental parameters: a negative polarity with 2.0 kV capillary voltage, 20 V sampling cone and 6 eV collision energy. c-di-GMP was detected in M-1 ion of m/z 689.086 with a fragment ion of m/z 344.040, and cGMP as an input in M-1 ion of m/z 344.039.
The SILAC MS assay of quantitative acetylation proteomics
E.coli BW25113 (△lysA) and BW25113 with YdeH overexpression (△lysA ydeH+) were subjected for SILAC MS assay. The two strains were activated in LB medium, and then were cultured in 2% glucose M9 minimal media supplemented with heavy isotopes of lysine (13C614N2-lysine) or light isotopes of lysine (12C614N2-lysine) (Silantes, Munich, Germany) for △lysA and △lysA ydeH+, respectively. Strains were induced by 0.1 mM IPTG during exponential growth (OD 600nm = ~0.4) and the cells were harvested after inducing for 4 h (OD 600nm = ~1.0). These cells (~30 OD) were added with 1 mL lysis buffer (8 M Urea, 100 mM NH4HCO3, 2 mM sodium butyrate, 5 mM nicotinamide, 1x protease inhibitor (Roche, Basel, Switzerland), pH 8.0) and lysed for 2 min at 4°C by an Ultrasonic Cell Disruptor (Cheng-cheng Weiye Science and Technology, Beijing, China). Protein concentration was determined by BCA kit (Pierce, MA, USA). The labeling efficiency of E.coli cultured in “heavy” medium was checked before sequential proteomic experiments.
Light-labeled and heavy-labeled lysate were equally mixed. Cysteine bonds were reduced by 5 mM dithiothreitol (DTT) at 56°C for 30 min and followed by alkylation reaction with 15 mM iodoacetamide at room temperature in darkness for 30 min. The alkylation reaction was quenched by 30 mM cysteine. The protein solution was diluted to less than 2 M Urea by addition of 100 mM NH4HCO3 (pH 8.0) and then digested with sequencing grade trypsin at a trypsin-to-protein ratio of 1: 50 (w/w) at 37°C for 16 h. For complete digestion, additional trypsin was added at trypsin-to-protein ratio of 1: 100 (w/w) at 37°C for another four hours. The tryptic peptides were desalted through SepPak C18 cartridges (Waters, MA, USA) and vacuum dried.
To enrich the lysine acetylated peptides, 2 mg desalted peptides were dissolved in NETN buffer (600 mM NaCl, 1 mM EDTA, 50 mM Tris-HCl, 0.5% NP-40, pH 8.0) and incubated with 10 μL drained pre-washed anti-acetyl beads (Immunechem, Burnaby, Canada) at 4°C overnight with gentle shaking. The beads were gently washed for four times with NETN buffer and twice with deionized water. The bound peptides were eluted with 0.1% TFA and vacuum dried. The eluted peptides were desalted with C18 ZipTips (Millipore, MA, USA) according to the manufacturer’s instructions.
Enriched acetylated peptides were analyzed by nano flow LC-MS/MS using an EASY-nLC 1000 system connected to Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific, MA, USA). Peptides were dissolved in solvent A (0.1% FA in 2% ACN) and separated using a homemade reverse-phase C18 analytical column (75 μm ID × 18 cm length, 3 μm particle size) with a 90-min gradient from 5% to 80% solvent B (0.1% FA in 90% ACN) at a constant flow rate of 300 nL/min. Intact peptides with m/z 300-1400 were detected at a resolution of 120,000 at m/z 200. Ions with intensity above 5,000 were isolated and sequentially fragmentized by Higher Collision Dissociation (HCD) with normalized collision energy of 32% in top speed mode. The automatic gain control (AGC) targets were set at 5.0e5 for full scan and 7.0e3 for MS/MS scan, respectively. The dynamic exclusion duration was set as 30s.
MS/MS data files were processed with MaxQuant software (version 1.5.3.8) against Escherichia coli (strain K12) database from Uniprot (proteome ID: UP000000625, 4309 sequences, last modified on May 13th, 2017) with a reversed decoy database. SILAC was selected as “doublets” and “Heavy labels” panel was selected as heavy lysine (Lys6). Trypsin/P was chosen as the digestion enzyme and two maximum missing cleavages was allowed. Carbamidomethyl (C) was specified as the fixed modification and variable modifications were oxidation (M), acetylation (Protein N-term) and acetylation (K). False discovery rates (FDR) at protein, peptide and modification level were all set as 1%. For quantitative analysis, the normalized H/L ratio of each acetylated peptide exported by MaxQuant software was corrected at the protein level to eliminate the protein abundance difference. The SILAC mass spectrometry proteomics raw data have been deposited to the ProteomeXchange63 Consortium via the PRIDE64 partner repository with the dataset identifier PXD007616 (Username: reviewer16485{at}ebi.ac.uk, Password: dD4exhSO).
Identification of deacetylation sites by Q Exactive plus MS
Five proteins of the DGC pathway, YdeH, Era, YegE, YeaJ and YihK were constructed to Pca24N and purified from E.coli BL21. After SDS-PAGE separation and tryptic digestion, these proteins were mixed and analyzed by mass spectrometry. Briefly, nanoLC−MS/MS-experiments were performed on an EASY-nLC system (Thermo Scientific, Odense, Denmark) connected to a Q Exactive Plus (Thermo Scientific, Bremen, Germany) through a nanoelectrospray ion source. Samples (1 µL) were loaded by an autosampler onto a 2-cm packed pre-column (75 µm ID x 360 µm OD) in 0.1% HCOOH/water (buffer A) at a flow rate of 1 µL/min for 5 min. Analytical separation was performed over a 15-cm packed column (75 µm ID x 360 µm OD) at 300 nL/min with a 60 mins gradient of increasing CH3CN (buffer B, 0.1% HCOOH/CH3CN). Both pre-column (5 µm diameter, 200 Å pore size) and analytical column (3 µm diameter, 100 Å pore size) were packed with C18-reversed phase silica (DIKMA-inspire TM, CA, USA) using a pressure bomb. Following sample loading, buffer B was increased rapidly from 3% to 6% over 5min and then shallowly to 22% over 36 min, and then to 35% over 9 min followed by a quick increase to 95% over 3min, and hold at 95% for 7 min. The total acquisition duration lasted for 60 min. The Q Exactive Plus mass spectrometer was operated in the data dependent mode to automatically switch between full scan MS and MS/MS acquisition. Survey full scan MS spectra (m/z 350−1800) were acquired in the Orbitrap with 70 000 resolution (m/z 200) after accumulation of ions to a 3×10 6 target value based on predictive AGC from the previous full scan. Dynamic exclusion was set to 60 s. The 15 most intense multiply charged ions (z≥2) were sequentially isolated and fragmented in the octupole collision cell by higher-energy collisional dissociation (HCD) with affixed injection time of 55 ms and 17500 resolutions for the fast scanning method. Typical mass spectrometric conditions were as follows: spray voltage, 1.7 kV; heated capillary temperature, 320°C; normalized HCD collision energy 27%. The MS/MS ion selection threshold was set to 9×103 counts. A 1.6 Da isolation width for the samp7les was chosen. The mass spectrometry raw data have been submitted to PRIDE63, 64 with project accession PXD007651 (User: reviewer60424{at}ebi.ac.uk, password: 6rxEMkjm). And the raw data file was named “20161202_ZHN_AC.raw” with the search file “20161202_ZHN_AC-01.msf”.
YdeH was chosen for in-depth quantitative MS analysis using Q Exactive plus mass spectrometer. YdeH was overexpressed in E. coli and affinity purified. YdeH was treated with CobB for deacetylation and untreated WT YdeH as the control. After SDS-PAGE separation and trypsin digestion, the samples were analyzed by Q Exactive plus mass spectrometer under the same experiment condition. The MS raw data was processed using Protein Discovery software (ThermoFisher) to identify the lysine acetylation sites. The acetylation sites, which were identified in untreated WT YdeH sample but not in CobB treated sample, were considered to be deacetylated by CobB. The mass spectrometry raw data have been submitted to PRIDE with project accession PXD008113 (Username: reviewer00527{at}ebi.ac.uk, password: jKBngFLH).
Determination of the deacetylation activity of CobB in vivo
The strains, △ cobB, WT, ydeH+, ydeHG206A,G207A, △ cobB::cobB, △cobB::cobB ydeH+ were used for to assay the deacetylation activity of CobB. These strains were grown in Vogel-Bonner medium (0.81 mM MgSO4·7H2O, 43.8 mM K2HPO4, 10 mM C6H8O7·H2O, 16.7 mM NaNH4HPO4·4H2O) with 10 mM acetate at 25°C for 12 h and induced by 0.2 mM IPTG at 25°C for 12 h. 20 OD cells were harvested and repeatedly freeze-thawed three times. The cells were then treated with 4mL lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole (Sigma-Aldrich), 1mg/mL lysozyme (Sangon Biotech, Shanghai, China), 1x CelLytic B (Sigma-Aldrich), 50 units/mL of Benzonase (Sigma-Aldrich) and 1 mM PMSF (Sigma-Aldrich), pH 8.0) at 4°C for 20 min with vigorous shaking. After lysis and centrifugation at 10,000 rpm for 5 min, 20 μL anti-FLAG antibody (Sigma-Aldrich) and 50 μL protein G conjugated agarose beads (Roche) were added at 4°C for 2 h with gentle agitation to enrich the 3xFLAG-tagged Acs. Protein G conjugated agarose was harvested and washed three times by buffer A (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole, pH 8.0). The 3xFLAG-tagged Acs was then eluted by heating at 95°C for 10 min. The amount of Acs protein and deacetylation level of these samples were analyzed by western blotting. Membranes were incubated with the pan anti-acetyl antibody (Cell Signaling Technology with a 1:1000 dilution) for at 4°C for 16 h and an anti-FLAG antibody (Sigma-Aldrich with a 1:2000 dilution) at 4°C for 16 h. The IRDye 800 antibody was used as a secondary antibody, and membranes were incubated for 1 h at room temperature. The membranes were washed three times in TBST between each antibody incubation step. Final visualization was performed using an Odyssey Infrared Imaging System.
Determination of the strain growth curve in Vogel-Bonner medium
The strains described above were grown in Vogel-Bonner medium with 10 or 30 mM acetate or propionate at 25°C for 12 h and then induced by 0.2 mM IPTG at 25°C for 20 h. During the entire 32 h growth period, the cell concentrations were measured at OD600 using Nanodrop 2000s at 8, 12, 16, 24 and 32 h. The growth curves were then drawn using GraphPad Prism 6.
Homology analysis, phylogenetic tree construction, and sequence alignment
The 25 bacteria that we selected for homology analysis were Escherichia coli, Salmonella typhimurium, Yersinia pestis, Shigella dysenteriae, Citrobacter amalonaticus, Erwinia tracheiphila, Vibrio parahaemolyticus, Klebsiella oxytoca, Cedecea neteri, Achromobacter sp.ATCC35328, Siccibacter colletis, Lelliottia amnigena, Buttiauxella brennerae, Kluyvera ascorbata, Mangrovibacter sp.MFB070, Cronobacter dublinensis, Hafnia alvei, Chania multitudinisentens, Serratia symbiotica, Yokenella regensburgei, Trabulsiella odontotermitis, Leclercia adecarboxylata, Pantoea dispersa, Shimwellia blattae, Kosakonia radicincitans. The CobB sequences were examined at the NCBI website and the phylogenetic tree was constructed by EMBL-EBI Clustalw2 online tool. Sequence alignment was performed using DNAMAN 2.0.
Determination of the YdeH DGC activity
The purified WT YdeH 5 uM (0.14mg/ml) protein was incubated with CobB 1.6 uM (0.05mg/ml) for deacetylating at 37℃ for 30 min. For DGC activity determination, the WT YdeH, deacetylated YdeH and YdeH mutants were incubated with 1 mM GTP, 5 mM MgCl2 in 100 ul at 30℃ for 2 h. The reaction was terminated with heating at 95℃ for 10 min and spun for 10 min at 18,000 g to separate the enzyme from the reactions. Before analyzed, we add 2 uM cGMP to these reactions as the control. The c-di-GMP concentrations were determined by UPLC-IM-MS with the same parameters as mentioned earlier.
Determination of the YdeH K4 acetylation stoichiometry using AQUA quantification
We used the absolute quantification (AQUA) method39, 40 to quantify the peptide levels. We synthesized the AQUA peptides for deacetylated (KTTEIDAIL(13C6,15N)LNLNK and TTEIDAIL(13C6,15N)LNLNK) and acetylated peptides (KacTTEIDAIL(13C6,15N)LNLNK) of YdeH K4. To acquire the endogenous YdeH protein, we inserted a chromosomal C-terminal Flag-tag to YdeH and immuno-precipitated by an anti-FLAG antibody. The strains, △cobB and WT, were used for to assay the YdeH K4 acetylation stoichiometry. These strains were grown in LB medium overnight and transfer to 1 L VB-E medium with 1:500 dilution and grown to OD600 = ~0.3 at 25℃ for 16 h as the exponential growth cells and to OD600 = ~0.6 at 25℃ for 42 h as the stationary phase cells. These cells were harvested and lyzed by high pressure with 50 mL lysis buffer (50 mM NaH2PO4, 300 mM NaCl, and 1 mM PMSF (Sigma-Aldrich), pH 8.0). After lysis and centrifugation at 10,000 rpm for 5 min, 100 μL anti-FLAG antibody (Sigma-Aldrich) was added and incubated at 4°C for 20 h with gentle agitation to enrich the 3xFLAG-tagged YdeH. 300 μL protein G conjugated agarose beads (Roche) were added and incubated at 4°C for 4 h with gentle agitation to enrich the anti-Flag antibody. Protein G conjugated agarose was harvested and washed for three times by buffer A (50 mM NaH2PO4, 300 mM NaCl, pH 8.0). The 3xFLAG-tagged YdeH was eluted by 2 mL 0.2 mg/mL Flag peptide at 4°C for 4 h. The free Flag peptide was separated through dialysis.
To acquire the overexpressed YdeH, we constructed the YdeH to pCA24N in △cobB strain (△cobB ydeH+). This strain was grown in LB medium overnight and transfer to 100 mL LB medium with 1:100 dilution and grown to OD600 = ~0.6 at 37℃ for 2 h, 0.1 mM IPTG was added to induce at 37℃ for 2 h. The cell was harvested and lyzed with 35 mL lysis buffer. After lysis and centrifugation at 10,000 rpm for 5 min, 1 mL Ni-IDA beads (Senhuimicrosphere, Suzhou, China) was added and incubated at 4°C for 1 h with gentle agitation to enrich the His-tagged YdeH. The His-tagged YdeH was eluted by 250 mM imidazole at 4°C for 20 min. The free imidazole was separated through dialysis.
For YdeH acetylation stoichiometry, affinity purified YdeH proteins were digested by trypsin overnight after cysteine reduction and alkylation reaction. Then tryptic peptides were desalted. The synthetic AQUA standard peptides were spiked into the digested YdeH samples with the close MS intensity to the native peptides. Peptide mixture containing AQUA peptides were analyzed by Q-Exactive mass spectrometer with three independent measurements. Full MS scan mode was used and the scan range was set as 700 to 900 m/z, containing all three targeted peptides. Extracted ion chromatography (XIC) peak areas of native peptide and corresponding heavy labeled peptide were used for stoichiometry calculation. The mass spectrometry proteomics raw data have been deposited to the ProteomeXchange63 Consortium via the PRIDE64 partner repository with the dataset identifier PXD007616 (Username: reviewer16485{at}ebi.ac.uk, Password: dD4exhSO).
Determination of stability of YdeH in vitro
After overexpressed in CobB deficient cells, YdeH was purified and diluted to 0.14 mg/mL, followed with or without CobB treatment. These samples were incubated at 37℃ for deacetylation for 0.5 h. Then 1 mM GTP was added, the samples were incubated at 30℃ for another 1 h and 2h. These samples were centrifuged at 12,000 g for 10 min to separate the soluble protein from the precipitation.
For YdeH mutants, these were diluted to 0.14 mg/mL incubated at 37℃ for 0.5 h. Then 1 mM GTP was added, the samples were incubated at 30℃ for another 1 h and 2h. These samples were centrifuged at 12,000 g for 10 min to separate the soluble protein from the precipitation.
Statistical analysis
Graphs were plotted using GraphPad Prism 6 and the statistical analyses were performed using Excel. Pairwise comparisons were performed using two-tailed Student’s t-test, and statistical significance was set at P <0.05. Error bars represent the mean ± standard errors of mean (S.E.M.).
Author Contributions Statement
S.C.T. conceived the idea. L.J.B. provided key reagents. Z.W.X. and C.X.L. performed interaction assay and functional analysis. Z.W.X. and C.X.L. performed enzyme activity assay. Z.W.X., H.N.Z., X.R.Z. and H.T.L. performed structure analysis and protein mutation. H.N.Z. and X.R.Z. performed the ITC and MST assay. Z.W.X., H.N.Z., X.R.Z., H.W.J., S.J.G and F.L.W. prepared the figures with the help of S.H.W. L.L.Q. and M.J.T. performed the SILAC-MS. F.L. performed the UPLC-IM-MS analysis. J.L.H. performed Nano LC maXis impact UHR-TOF MS analysis. Z.W.X., H.N.Z., X.R.Z., D.M.C. and S.C.T. wrote the manuscript.
Additional Information
Competing financial interests
The authors declare that we have no conflict of interest.
Acknowledgements
We thank Prof. Yufeng Yao of Shanghai Jiao Tong University for providing S. typhimurium CobB overexpression plasmid. We thank Prof. Yan Wei and Liyun Ji of Shanghai Jiao Tong University for performing the Q Exactive plus assay. We thank Ademi Zhakyp of Nazarbayev University for proof-reading the manuscript. This study was supported in part by The National Key Research and Development Program of China Grant 2016YFA0500600, National Natural Science Foundation of China Grants 31670831, 31370813, to S.C.T., 31370750, 31670722 to D.M.C., and 31670066 to M.J.T., and National Basic Research Program of China (973 Program) (No. 2014CBA02004) to M.J.T..