1. Biochemistry and Chemical Biology
Download icon

Comment on ‘YcgC represents a new protein deacetylase family in prokaryotes’

  1. Magdalena Kremer
  2. Nora Kuhlmann
  3. Marius Lechner
  4. Linda Baldus
  5. Michael Lammers  Is a corresponding author
  1. University of Cologne, Germany
  2. Institute of Biochemistry, Synthetic and Structural Biochemistry, University of Greifswald, Germany
  • Cited 6
  • Views 1,146
  • Annotations
Cite this article as: eLife 2018;7:e37798 doi: 10.7554/eLife.37798

Abstract

Lysine acetylation is a post-translational modification that is conserved from bacteria to humans. It is catalysed by the activities of lysine acetyltransferases, which use acetyl-CoA as the acetyl-donor molecule, and lysine deacetylases, which remove the acetyl moiety. Recently, it was reported that YcgC represents a new prokaryotic deacetylase family with no apparent homologies to existing deacetylases (Tu et al., 2015). Here we report the results of experiments which demonstrate that YcgC is not a deacetylase.

https://doi.org/10.7554/eLife.37798.001

Introduction

The progress in quantitative mass spectrometry in the last decade enabled the identification of thousands of lysine acetylation sites in all kingdoms of life (Choudhary et al., 2009; Lundby et al., 2012; Weinert et al., 2014, 2017, 2013b; Zhang et al., 2009). While histones were known since the 1960 s to be modified by lysine acetylation, these studies revealed that proteins of all cellular compartments covering essential cellular functions are lysine acetylated in eukaryotes, archaea and prokaryotes (Finkemeier et al., 2011; König et al., 2014). Lysine acetylation is a dynamic post-translational modification that is catalysed by lysine acetyltransferases, which add the acetyl moiety and lysine deacetylases (KDACs) which remove it. Site-specific non-enzymatic acetylation has also been reported (Baeza et al., 2015). While two classes of deacetylases exist in eukaryotes, the classical Zn2+-dependent KDACs and the NAD+-dependent sirtuins, prokaryotes only encode for sirtuins. Escherichia coli encodes for only one single KDAC, the sirtuin deacetylase CobB, which is structurally highly similar to eukaryotic sirtuins showing only minor structural differences in the small Zn2+-binding domain (Zhao et al., 2004). In contrast, in mammals there are seven sirtuin deacylases (Sirt1-7), which show distinct subcellular localisation. While the nuclear Sirt1, the cytosolic Sirt2 and the mitochondrial Sirt3 possess a robust deacetylase activity, the remaining deacetylases show preferences for longer acylations or act as ADP-ribosyltransferases (Du et al., 2011; Feldman et al., 2013; Liszt et al., 2005; Smith et al., 2008). Mammals encode eleven classical KDACs and seven sirtuins enabling to use some activities with rather low specificity, while other enzymes show a remarkable specificity for some substrates and some lysine acylation sites (Knyphausen et al., 2016). This allows mammals to use post-translational lysine acylation as a molecular switch with consequences on signal transduction and regulation of protein function. In contrast, the prokaryotic CobB has a broad substrate range and it is quite promiscuous with respect to its substrate proteins. This suggests CobB’s major role to be the detoxification of lysine acylations occuring upon metabolic fuel switching or metabolic stress under conditions of accumulation of acetyl-CoA or acetyl-phosphate, which can both serve as acetyl group donor molecules in prokaryotes (Weinert et al., 2013a, 2017).

Recently, Tu et al. reported the presence of another deacetylase class in prokaryotes with YcgC as a representative (Tu et al., 2015). YcgC (also termed DhaM) together with YcgT (DhaK) and YcgS (DhaL) was originally reported to be part of the active dihydroxyacetone kinase (DhaK) complex. It is a component of a phosphorelay system using phosphoenolpyruvate as phosphoryl donor in which the phosphoryl group is finally transferred by DhaK to dihydroxyacetone yielding dihydroxyacetone phosphate (Gutknecht et al., 2001). YcgC has no sequence or structural homology to either eukaryotic classical KDACs or to sirtuin deacylases and it does not depend on Zn2+ or NAD+ for catalysis. Tu et al. reported that YcgC acts as serine hydrolase using a distinct catalytic mechanism involving a catalytic serine residue (S200); they also reported that YcgC catalyses deacetylation of the transcriptional regulator RutR and supports subsequent autoproteolytic cleavage at the RutR N-terminus (Tu et al., 2015). Here we report the results of our efforts to build on this work. In summary, we were unable to detect any deacetylase activity of YcgC, which means that CobB would seem to be the only deacetylase present in E. coli, and that YcgC should not be annotated as a lysine deacetylase in UniProt and other databases.

Results

Expression and purification of YcgC, CobB and RutR

We recombinantly expressed YcgC, the supposed deacetylase-dead mutant YcgC S200A, CobB and the catalytically inactive mutant CobB H110Y in E. coli BL21 (DE3) cells and purified them to a high level of purity using a two-step purification strategy composed of a glutathione (GSH)-affinity purification step followed by size exclusion-chromatography (SEC). For YcgC and YcgC S200A Tobacco etch virus (TEV) cleavage was performed to remove the glutathione-S-transferase (GST)-tag. In contrast, CobB and CobB H110Y were purified as GST-fusion proteins to allow for a better discrimination of RutR and CobB in subsequent SDS-PAGE analyses. All proteins could be successfully produced and showed final purities of more than 95% (Figure 1A). YcgC and the mutant YcgC S200A behaved similar on analytical SEC further underlining the high quality of proteins (Figure 1B, upper panel). Tu et al. reported that YcgC deacetylates the transcriptional regulator RutR acetylated at K52 (RutR AcK52) and at K62 (RutR AcK62). For expression of acetylated RutR, we constructed an E. coli BL21 (DE3) strain carrying a genomic deletion of ycgC and cobB (E. coli BL21 (DE3) ΔycgCΔcobB) to exclude the possibility that RutR is deacetylated endogenously during the expression (Figure 1—figure supplement 1).

Figure 1 with 1 supplement see all
Preparation of YcgC, CobB and non-acetylated/acetylated RutR.

(A) SDS-PAGE analysis of proteins used in this study. All proteins were expressed and purified as GST-fusion proteins. In terms of YcgC and YcgC S200A the GST-tag was removed by TEV protease during the purification steps. Staining of the gel was done by coomassie brilliant blue (CBB). For molecular masses see figure legend for Figure 2A. (B) Analytical size exclusion chromatography on a S200 10/300 GL column shows that YcgC WT and YcgC S200A as well as CobB and the corresponding catalytically inactive variant CobB H110Y display an almost identical elution profile. Moreover, RutR proteins show a nearly identical elution profile in analytical SEC runs indicating that RutR acetylation at K52 and K62 does not interfere with protein folding or its oligomeric state. (C) RutR-His6 AcK52 and AcK62 are quantitatively acetylated. Shown are SDS-PAGE and immunoblot analyses of all RutR-His6 proteins used in this study. Staining for AcK using an anti-acetyl-L-lysine antibody revealed a strong signal for RutR AcK52 and AcK62, whereas no signal was obtained for RutR WT. As loading control anti-His6 staining was performed. (D) ESI-MS data show the quantitative and homogenous incorporation of acetyl-L-lysine into RutR. Shown is the deconvoluted spectrum on the true mass scale after software transformation yielding one single peak and the corresponding molecular mass as indicated. Expected mass non-acetylated RutR: 24567.5 Da; acetylated RutR: 24609.5 Da.

https://doi.org/10.7554/eLife.37798.002

We used the genetic code expansion concept to site-specifically introduce acetyl-L-lysine at positions K52 and K62 into RutR. To this end, we constitutively expressed the synthetically evolved acetyl-lysyl-tRNA synthetase MbtRNACUA pair from Methanosarcina barkeri and co-expressed it with RutR carrying an amber stop codon at the positions encoding for K52 and K65 to allow the site-specific incorporation of acetyl-L-lysine. To exclude the possibility that the His6-tag used for purification via Ni2+-NTA chromatography interferes with YcgC catalysed deacetylation and subsequent N-terminal autoproteolytic cleavage of RutR, we purified RutR proteins placing the His6-tag at the RutR C-terminus. We were able to produce non-acetylated RutR wildtype (WT), RutR AcK52 and RutR AcK62 purified to high degree and obtained a yield sufficient to perform biochemical studies. Immunoblotting, electrospray ionisation mass spectrometry (ESI-MS) as well as tryptic digest and subsequent ultra-performance liquid chromatography tandem mass spectrometry (UPLC-MS/MS) confirmed that RutR is quantitatively, homogenously and site-specifically lysine acetylated at the positions K52 and K62 (Figure 1C,D, Figure 1—figure supplement 1). Notably, RutR WT was not acetylated on Lys upon expression in either E. coli BL21 (DE3) or in E. coli BL21 (DE3) ΔycgCΔcobB. This is in contrast to the report by Tu et al., according to which RutR expressed in E. coli W3110 is almost quantitatively lysine acetylated (Figure 1C,D, Figure 1—figure supplement 1). As shown later, our LC-MS/MS data revealed that also the RutR protein from Tu et al. is not quantitatively lysine acetylated. In analytical SEC runs, all RutR proteins eluted not in the exclusion volume, showed a highly symmetric elution peak and eluted at an almost identical elution volume. This shows that all proteins are in a similar oligomeric state and it furthermore suggests that lysine acetylation of RutR does not interfere with protein folding or the oligomeric state (Figure 1B, lower panel).

YcgC does not deacetylate RutR AcK52 and AcK62

To check for deacetylase activity of YcgC we used the site-specifically lysine acetylated RutR AcK52 and AcK62 proteins as substrates and performed an in vitro deacetylation assay. YcgC was used in a twofold molar excess to RutR substrate to ensure that the reaction proceeds to completion. As a readout we conducted immunoblotting using a specific anti-acetyl-L-lysine antibody (anti-AcK AB) that we have shown earlier to be well suited for following deacetylation reactions. We did not observe any YcgC catalysed deacetylation of RutR AcK52 or RutR AcK62 (Figure 2A). As a control, we also analysed the deacetylation of RutR AcK52 and AcK62 by CobB. As shown by immunoblotting, RutR AcK52 was strongly deacetylated by CobB whereas RutR AcK62 was only slightly deacetylated (Figure 2A). To confirm the immunoblotting results and to exclude the possibility that this assay is not sensitive enough to detect a low deacetylase activity of YcgC on acetylated RutR, we also analysed the reaction products by ESI-MS. Again, no YcgC deacetylase activity towards either RutR AcK52 or RutR AcK62 was measurable and only one single peak with a mass of 24608 Da exactly corresponding to mono-lysine acetylated RutR was detectable (Figure 2B, Figure 2—figure supplement 1). In contrast, RutR AcK52 was deacetylated by CobB to more than 60% resulting in a mass shift of 42 Da, which agrees with removal of an acetyl-group, while RutR AcK62 was only marginally deacetylated by CobB (Figure 2B, Figure 2—figure supplement 1). These data support the immunoblotting experiments and clearly show that YcgC does not have any deacetylase activity towards RutR AcK52 and AcK62. Moreover, we did not observe any (auto-)proteolytic cleavage of RutR WT, RutR AcK52 or RutR AcK62, independently of presence or absence of YcgC. Except for the deacetylation of RutR AcK52, and to a much lesser extent of RutR AcK62, by CobB the molecular weight of the RutR proteins is not affected as shown by SDS-PAGE, immunoblotting and ESI-MS (Figure 2, Figure 2—figure supplement 1).

Figure 2 with 1 supplement see all
YcgC does not show any deacetylase activity towards RutR AcK52 or RutR AcK62 and it does not stimulate RutR (auto-) proteolytic cleavage.

(A) RutR AcK52 and RutR AcK62 were treated with a two-fold molar excess of YcgC, YcgC S200A, CobB or CobB H110Y in the presence of 5 mM NAD+. As visible from the immunoblotting using an anti-AcK AB CobB was able to completely deacetylate RutR AcK52 while RutR AcK62 is only faintly deacetylated under the conditions used. YcgC, YcgC S200A and the catalytically dead CobB H110Y did neither deacetylate RutR AcK52 nor RutR AcK62 suggesting that YcgC is no deacetylase for RutR. We used the GST-fusion proteins of CobB and CobB H110Y here to get a better separation of CobB and RutR proteins. Probing of the His6-tag using an anti-His6 antibody was done to show equal loading with RutR proteins. SDS-PAGE analysis and CBB staining show that there is only one band of the same size for RutR for all the conditions tested suggesting that YcgC did also not stimulate (auto-) proteolytic cleavage of RutR. The control lane (ctrl) shows the respective RutR protein, AcK52 or AcK62, without addition of CobB or YcgC. Molecular weights of proteins used (all masses calculated without N-terminal methionine): RutR, 24567.5 Da; acetylated RutR, 24609.5 Da; GST-CobB, 53271.09 Da; GST-CobB H110Y, 53297.12 Da.; YcgC, 51649.80 Da; YcgC S200A, 51633.80 Da. We used the GST-fusion proteins for CobB to obtain a better separation from RutR as CobB (MW: 26314.86 Da) and CobB H110Y (26340.90 Da) without GST-tag have similar molecular weights compared to RutR. (B) Quantification of ESI-MS spectra of YcgC and CobB reaction products shown as immunoblots in (A). As a support for the data obtained by immunoblotting, neither YcgC nor YcgC S200A nor catalytically inactive CobB H110Y did alter the molecular mass of RutR WT, AcK52 and AcK62. However, active CobB led to more than 60% deacetylation of RutR AcK52 while RutR AcK62 was only marginally deacetylated (app. 30% deacetylated). Again, these data confirm that YcgC does neither deacetylate or proteolytically cleave RutR nor does it stimulate autoproteolytic cleavage of RutR.

https://doi.org/10.7554/eLife.37798.004

Analysis of YcgC deacetylation of RutR expressed in E. coli W3110 (by Tu et al.) and site-specifically lysine-acetylated RutR AcK52 and AcK62 expressed in BL21 (DE3) (this work)

To resolve discrepancies observed between data obtained by Tu and co-workers and by our group we initiated experiments to compare activities of the proteins of both groups (RutR and YcgC; Figure 3, Figure 3—figure supplement 1, Supplementary file 5). To this end, we exchanged all proteins between both laboratories. We performed YcgC catalysed deacetylation reactions under conditions described by Tu et al. (reaction buffer: 50 mM Tris, 4 mM MgCl2, 50 mM NaCl, 50 mM KCl, 5% (v/v) Glycerol, pH 8.0). We also compared the antibodies used for detection in both studies (our lab: ab21263 (abcam); Tu et al.: CST9441 (Cell Signaling Technology)). Using the antibody from abcam we confirmed the results obtained before. We only detected site-specifically lysine-acetylated RutR AcK52 and AcK62, while none of the other proteins were stained with the anti-acetyl-lysine antibody from abcam (ab21263) (Figure 3A). Notably, the antibody CST9441 used by Tu et al. detected all RutR proteins used in this study, i.e. non-acetylated RutR prepared by our group, RutR AcK52 and AcK62, as well as RutR from Tu et al. Moreover, while we observed a CobB catalysed decrease in acetylation-level using the antibody from abcam for RutR Ack52 (and more weakly also for AcK62) as expected, the antibody CST9441 did not detect a decrease in acetylation signal intensity upon CobB treatment (Figure 3B). This suggests that it detects trace amounts of acetylation occurring at very low stoichiometry most likely non-enzymatically during expression in E. coli. These trace amounts must be present even on non-acetylated RutR protein prepared by us. Alternatively, the CST9441 antibody shows cross-reactivity with other antigenic components, for example the RutR protein sequence. In both cases, the antibody CST9441 might not be suitable for these assays. To underline these conclusions, we observed for our RutR proteins no additional lysine acetylations by ESI-MS showing that if further acetylations happen on the RutR protein, these are present at very low, highly sub-stoichiometric levels (Figure 1D).

Figure 3 with 1 supplement see all
Comparison of YcgC and RutR proteins as well as antibody detection from Tu et al.

(T) and from our lab (L). (A) RutR from Tu et al. (RutR T) and our non-acetylated RutR (RutR L) were treated with YcgC prepared by our lab (L), YcgC and YcgC S200A from Tu et al. (T). RutR T and non-acetylated RutR L are not detected by the antibody ab21623 from abcam (left panels). Site-specifically lysine-acetylated RutR AcK52 and AcK62 prepared by us (RutR AcK52 L and RutR AcK62 L) are detected by the antibody ab21623. Treatment with CobB reduces acetylation level for RutR AcK52 L and AcK62 L as expected (right panels). Treatment of RutR AcK52 L and RutR AcK62 L with YcgC T, but not with YcgC S200A T and YcgC L, results in reduction of molecular size, while acetylation level is not decreased. (B) Same as in A but staining was done with anti acetyl-lysine antibody CST9441 from Cell Signaling Technology. CST9441 antibody detects site-specifically acetylated and non-acetylated RutR (or RutR that contains highly sub-stoichiometric, trace amounts of lysine acetylation). Acetylation signal is completely removed for RutR T upon treatment with YcgC T but neither with YcgC S200A nor with YcgC L. His6-staining shows that YcgC T treatment also removes N-terminal His6-tag suggesting proteolytic cleavage from N-terminus (Figure 3—figure supplement 1B). RutR AcK52 L and AcK62 L also show reduction in molecular size upon treatment with YcgC T (right panels). However, acetylation signal using CST9441 antibody of the degradation band is not removed, showing that deacetylation at neither RutR AcK52 L nor RutR AcK62 L takes place. CobB treatment of no RutR protein results in visible reduction in acetylation signal using CST9441 antibody, although it is shown that CobB deacetylates AcK52 and AcK62 from RutR.

https://doi.org/10.7554/eLife.37798.006

We observe a reduction of the molecular size upon treatment with the YcgC prepared by Tu et al., but neither for YcgC prepared by us, nor by YcgC S200A prepared by Tu et al., suggesting that the YcgC preparation from Tu et al. contains an activity explaining this observation (Figure 3A, Figure 3—figure supplement 1). This is true for RutR from Tu et al. as well as for all RutR proteins prepared by us, that is non-acetylated RutR and RutR AcK52 and AcK62. This proteolytic modification must occur from the N-terminus as our C-terminally His6-tagged protein still gives a signal in His6 staining, while N-terminally His6-tagged RutR from Tu et al. completely looses His6-staining. Importantly, for both RutR AcK52 and AcK62 the lower molecular weight RutR protein is still lysine acetylated as shown by immunoblotting using the anti-acetyl lysine antibody from abcam suggesting that both proteins are not deacetylated on either AcK52 or AcK62 prior to proteolytic cleavage. This clearly shows that the autoproteolytic cleavage of RutR is not preceded by the deacetylation of RutR AcK52 and/or AcK62, as was suggested by Tu et al. As additional support, we are able to express and purify non-acetylated RutR in full-length without observing truncation at the N-terminus.

Mass-spectrometry of YcgC and RutR proteins

As the results above clearly show that YcgC does not have an intrinsic lysine deacetylase activity, it is still an unresolved question what leads to the reduction of the molecular size of RutR proteins upon incubation with YcgC prepared by Tu et al. We concluded that a unspecific or a RutR specific protease activity might be present in the YcgC preparation of Tu et al., which could explain the observed reduction in molecular size upon treatment with YcgC from Tu et al. To analyse protein preparations for presence of proteases, we performed LC-MS/MS analyses (Supplementary file 1). We observed, that the YcgC preparation by Tu et al. contained several contaminants that were neither present in the YcgC S200A preparation by Tu et al. nor in our YcgC protein preparation (Supplementary file 1, Supplementary file 2). Amongst these proteins, we found several proteases and peptidases (metallo- and serine proteases) and several proteins with unknown function (Supplementary file 1, Supplementary file 2). Notably, we even detected Lon protease in YcgC by Tu et al., which is genomically deleted in most commercially available bacterial expression strains to avoid proteolytic cleavage of expressed proteins. One of Lon protease’s best-characterised substrates is the helix-turn-helix (HTH) LuxR-type transcriptional regulator RscA. This fuels the idea that maybe the HTH TetR-type transcriptional repressor RutR might be a specific substrate as well. This needs further investigation.

We also used LC-MS/MS to analyse the RutR preparation of Tu et al., and can draw several conclusions from the results of this analysis. First, the RutR protein is not quantitatively lysine acetylated, as was reported by Tu et al. In fact, from our mass spectrometrical analyses we can conclude that overall the amount of acetylated RutR is 1.09% (total intensity of 5.74*1011 vs. intensity of acetylated peptides: 6.26*109). These analyses can be used to estimate lysine acetylation occupancy. We obtained an overall sequence coverage of 90% in the LC-MS/MS experiments showing that almost all lysines present in RutR are detectable (Supplementary file 3). Moreover, although we cannot determine precise RutR-acetylation stoichiometries by LC-MS/MS without performing a standard curve, these data clearly show that the RutR acetylation is of very low occupancy. Acetylated and non-acetylated peptides is reported to show similar ionization efficiencies and as a consequence the MS intensities can be used for quantification of the occupancy (Cho et al., 2016). From the identified acetylated lysines, RutR K150 is the acetyl acceptor site with highest occupancy of about 0.7% of total RutR. Lysine 21 is the second best acetyl acceptor site with an occupancy of 0.3% followed by AcK95 with 0.06%. Second, we found AcK52 in the RutR preparation from Tu et al. but again in a very low occupancy of only 0.02% in relation to all peptides identified. Third, while Tu and co-workers identified K62 in RutR as the acetylation site of highest functional importance, our LC-MS/MS data revealed that K62 is not lysine-acetylated at all, although this lysine has been identified in the non-acetylated state and in a peptide with one missed cleavage suggesting that the peptide would be analysable by LC-MS/MS if present. These data clearly show that the model presented by Tu et al. needs to be revised.

Discussion

In summary, we have shown that YcgC does not show any deacetylase activity towards site-specifically lysine acetylated RutR. Additionally, YcgC prepared in our lab did neither directly nor indirectly affect RutR proteolytic cleavage or autoproteolysis. LC-MS/MS analyses revealed that RutR from Tu et al. is only marginally lysine-acetylated with an overall occupancy of 1%. Of the identified lysine acetylation sites, AcK52 has an occupancy of only 0.02% and AcK62 was not identified at all. For our site-specifically lysine-acetylated RutR proteins, we also observe proteolytic cleavage by treatment with the YcgC preparation from Tu and co-workers. However, also the lower molecular weight RutR proteins were lysine-acetylated.

Taken together, these data call into question the molecular model presented by Tu et al.: in other words, the data show that the (auto-)proteolysis of RutR is not preceded by the YcgC-catalysed deacetylation of RutR at AcK52 and/or AcK62. Our LC-MS/MS analyses revealed the presence of a number of contaminants, including some that have proteolytic activity, in the YcgC preparation provided by Tu et al. Finally, threading analyses of YcgC using the programme iTASSER revealed no obvious homologies to enzymes with deacetylase or protease activity (Supplementary file 4) (Yang et al., 2015).

Materials and methods

Generation of E. coli BL21 (DE3)ΔycgCΔcobB

Request a detailed protocol

The double knockout strain E. coli BL21 (DE3) ΔycgCΔcobB was created by applying the Quick and Easy E. coli Gene Deletion Kit (Gene Bridges). Consecutive deletion of ycgC and cobB was done by homologous recombination according to the manufacturer’s constructions. Finally, successful gene deletion was verified by PCR and sequencing.

Expression and purification of proteins

Request a detailed protocol

E. coli YcgC, YcgC S200A, CobB and CobB H110Y were expressed as GST-fusion proteins using the vector pGEX-4T5/TEV derived from pGEX-4T1 (GE Healthcare). RutR proteins were expressed as fusion proteins carrying a C-terminal His6-tag using the vector pRSF-Duet-1. YcgC, YcgC S200A, CobB and CobB H110Y were expressed in E. coli BL21 (DE3), whereas RutR proteins were expressed in E. coli BL21 (DE3) ΔycgCΔcobB. Cells were grown to an optical density at 600 nm (OD600) of 0.6 (37°C, 160 rpm) before protein expression was induced by addition of 300 µM isopropyl-β-D-thiogalactopyranoside (IPTG). Protein expression was performed over night at 18–20°C at 160 rpm. Subsequently, the bacterial cultures were harvested (4000 g, 10 min) and resuspended in either (YcgC and CobB) buffer A (50 mM Tris/HCl pH 7.4, 100 mM NaCl, 5 mM MgCl2, 2 mM β-mercaptoethanol) or (RutR) buffer B (100 mM K2HPO4/KH2PO4 pH 6.4, 150 mM NaCl, 20 mM imidazole, 2 mM β-mercaptoethanol) containing 200 µM Pefabloc protease inhibitor cocktail. Cells were lysed by sonication and after centrifugation (20000 g, 45 min) the soluble fraction was applied to an equilibrated GSH- or Ni2+-NTA affinity chromatography column. Washing was performed with either buffer A (YcgC and CobB) or buffer B (RutR) supplemented with 500 mM NaCl. YcgC GST-fusion proteins were treated with TEV protease on the GSH-column to remove the GST-tag (4°C, 0.5 ml/min, over night). CobB was either eluted as GST-fusion protein with buffer A containing 30 mM glutathione or it was treated with TEV protease to remove the GST-tag. RutR proteins were eluted as GST-fusion proteins with buffer C (buffer B with imidazole gradient from 50 to 300 mM imidazole). Eluates were concentrated using 10 kDa MWCO amicon ultrafiltration devices and subsequently applied to a size exclusion chromatography (SEC) column (GE healthcare). SEC was performed in buffer A (YcgC and CobB) or buffer B (RutR), respectively. Peak fractions containing the protein of interest were pooled, shock frozen in liquid nitrogen and stored at −80°C. Protein concentrations were determined by measuring the absorption at 280 nm using the protein’s extinction coefficient (http://web.expasy.org/protparam/).

Incorporation of N-(ε)-acetyl-lysine into RutR at K52 and K62 using the genetic code expansion concept

Request a detailed protocol

The site-specific incorporation of N-(ε)-acetyl-lysine into RutR at position K52 and K62 was conducted as described previously (de Boor et al., 2015; Knyphausen et al., 2016; Kuhlmann et al., 2016; Lammers et al., 2010; Neumann et al., 2008). In brief, E. coli BL21 (DE3) ΔycgCΔcobB cells bearing the vector pRSF-Duet-1 encoding for RutR K52amber or RutR K62amber, the synthetically evolved acetyl-lysyl-tRNA-synthetase AcKRS3 and the amber suppressor tRNACUA, MbtRNACUA derived from Methanosarcina barkeri were cultivated to an OD600 of 0.6 (37°C, 160 rpm). After addition of 10 mM N-(ε)-acetyl-L-lysine (Chem-Impex International Inc.) and 20 mM nicotinamide (NAM) cells were grown for further 30 min before protein expression was induced by adding 300 µM IPTG. The quantitative incorporation of N-(ε)-acetyl-L-lysine into RutR at position K52 and K62 was verified by immunoblotting using an anti-acetyl-L-lysine antibody and mass spectrometry as described earlier (de Boor et al., 2015; Lammers et al., 2010).

Immunoblotting

Request a detailed protocol

Proteins were separated by SDS-PAGE and analysed by immunoblotting using a standard protocol. Bound antibodies were visualised by using enhanced chemiluminescence (Roth). Rabbit polyclonal anti-acetyl-L-lysine antibody (ab21623, 1:1500), mouse monoclonal anti-His6 antibody (ab18184, 1:2000), rabbit monoclonal anti-acetyl-L-lysine antibody (CST9441, 1:1000) as well as HRP-coupled secondary antibodies against rabbit (ab6721, 1:10000) and mouse (ab6728, 1:10000) were purchased from Abcam.

Deacetylase assay

Request a detailed protocol

To test for deacetylation of RutR-His6 AcK52 and RutR-His6 AcK62 by YcgC and GST-CobB 12.5 µM of recombinant RutR-His6 were incubated with 25 µM YcgC, YcgC S200A, GST-CobB or GST-CobB H110Y. All reactions were done in buffer A supplemented with 5 mM NAD+. To this end, all proteins were transferred into buffer A by using 10 kDa MWCO Viaspin 500 microcentrifugal units. After incubation for one hour at 37°C, either samples were boiled at 95°C for five minutes and subjected to SDS-PAGE and immunoblotting or reaction products were analysed by ESI-MS.

For comparison of proteins from Tu et al. and our group (Tu et al.: YcgC T, YcgC S200A T, RutR T; Kremer and Kuhlmann et al.: YcgC L, RutR L, RutR AcK52 L, RutR AcK62 L) to resolve discrepancies between results, we performed deacetylation assays in the exact buffer as described in Tu et al. (50 mM Tris, 4 mM MgCl2, 50 mM NaCl, 50 mM KCl, 5% (v/v) Glycerol, pH 8.0). We incubated 6.3 µg (≈ 25 µM) of the respective RutR protein with 6.3 µg (≈ 12 µM) YcgC T, YcgC S200A T, YcgC L or 3.15 µg (12 µM) CobB filled up with buffer (and NAD+ for CobB) to a total volume of 10 µL for 2 hr at 37°C. Analysis of the lysine acetylation status, the molecular size of RutR and protein loading was done by immunoblotting using an anti-acetyl-lysine antibody from abcam (ab21623) or Cell Signaling Technology (CST9441), anti-His6-antibody (ab18184, 1:2000) and Coomassie-brilliant blue (CBB) staining.

Electrospray-ionisation (ESI)-mass spectrometry

Request a detailed protocol

To verify successful incorporation of acetyllysine, intact proteins were analysed by ESI-MS on a LTQ Orbitrap Discovery mass spectrometer (Thermo Scientific). To validate specificity of the acetylation site, RutR was trypsin-digested and desalted prior to analysis by UPLC-MS/MS. In brief, 4 µg protein were denatured in urea buffer (50 mM TEAB, 8 M urea) and reduced by adding dithiothreitol (DTT) to a final concentration of 5 mM and incubation at 37°C for one hour. Subsequently, proteins were alkylated with chloroacetamide (CAA) (40 mM, 30 min in the dark) followed by LysC digestion for 4 hr at 37°C. Afterwards, tryptic digest was done over night at 37°C. For both enzymes an enzyme to substrate ratio of 1:75 was used. The next day, obtained peptides were acidified by adding formic acid to a final concentration of 1%, desalted and analysed with a Q Exactive Plus Orbitrap LC-MS/MS system (Thermo Scientific) as described earlier (Kuhlmann et al., 2016). MaxQuant and the implemented Andromeda search engine were used to analyse raw data. MS/MS spectra were correlated with the Uniprot Escherichia coli database containing the target protein (Cox and Mann, 2008; Cox et al., 2011).

Liquid chromatography (LC)-mass spectrometry and data analysis

Request a detailed protocol

Peptides were analysed on a Q Exactive Plus (Thermo Scientific) mass spectrometer that was coupled to an EASY nLC 1000 UPLC (Thermo Scientific). Samples were loaded with solvent A (0.1% formic acid in water) onto an in-house packed analytical column (50 cm × 75 µm I.D., filled with 2.7 µm Poroshell EC120 C18, Agilent). Peptides were chromatographically separated at a constant flow rate of 250 nL/min using 60 min methods: 5–30% solvent B (0.1% formic acid in 80% acetonitrile) within 40 min, 30–50% solvent B within 8 min, followed by washing and column equilibration. The mass spectrometer was operated in data-dependent acquisition mode. The MS1 survey scan was acquired from 300 to 1750 m/z at a resolution of 70,000. The top 10 most abundant peptides were subjected to HCD fragmentation at a normalised collision energy of 27%. The AGC target was set to 5e5 charges. Product ions were detected in the Orbitrap at a resolution of 17,500. All mass spectrometric raw data were processed with Maxquant (version 1.5.3.8) using default parameters [Tyanova et al., 2016]). Briefly, MS2 spectra were searched against the E. coli UniProt database, including a list of common contaminants and sequences of the specific purified proteins in the samples. False discovery rates on protein and PSM level were estimated by the target-decoy approach to 0.01% (Protein FDR) and 0.01% (PSM FDR) respectively. The minimal peptide length was set to seven amino acids and carbamidomethylation at cysteine residues was considered as a fixed modification. Oxidation (M), Acetyl (K) and Acetyl (Protein N-term) were included as variable modifications. The resulting output was processed using Perseus as follows: Protein groups flagged as ‘reverse’, ‘potential contaminant’ or ‘only identified by site’ were removed from the proteinGroups.txt. Intensity or iBAQ values were log transformed.

Data availability

All data generated or analyses during this study are included in the manuscript and supporting files.

References

Decision letter

  1. Wilfred A van der Donk
    Reviewing Editor; University of Illinois at Urbana-Champaign, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your work entitled "Comment on "YcgC represents a new protein deacetylase family in prokaryotes"" for consideration by eLife. Your article has been reviewed by three peer reviewers including Wilfred van der Donk as the Reviewing Editor and Reviewer #2, and the evaluation has been overseen by Michael Marletta as the Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. All three reviewers recognize the importance to publish your work if indeed the previous assignment of YcgC as a deacetylase turns out to be incorrect. However, all three reviewers also believe that at present, your experiments have not used the same conditions and enzymes as those used in the original work. Hence, the discrepancies could well lie in those differences in conditions and enzymes. Based on these discussions and the individual reviews below, we regret to inform you that your work as submitted will not be considered further for publication in eLife. However, if you were to investigate the deacetylase activity of YcgC on acetylated RutR under the same conditions as the previous study and find that no activity is observed, we would be willing to reconsider. The journal would also like to help in exchange of materials with the initial authors to find the source of the discrepancies.

Reviewer #1:

In this manuscript, Kuhlmann et al. reported the function characterization of YcgC, which conflicts with what was reported by Tu and co-workers before. The negative results of the authors led to their claim that "YcgC is not a protein deacetylase". After comparing the two manuscripts, the reviewer found out the following key differences. On the basis of the limited data shown here, the reviewer could not decide whether the listed difference led to the conflicting results presented in the two manuscripts. However, given that the current manuscript built up its argument on the basis of the negative results and reported the finding after the first report, it should be the burden of the current authors to rule out that any of the following differences or their combination is not the cause of the conflicting observation. If possible, the authors should contact the lab for identical reagents unless it is inaccessible. It is not sufficient to simply justify why the new conditions were used without conducting the same experiments under the identical conditions as reported previously.

a) Buffer conditions for deacetylase assay:

Current manuscript: 50 mM Tris-HCl pH 7.4, 100 mM NaCl, 5 mM MgCl2, 5 mM NAD+, 2 mM β-mercaptoethanol, 12.5 μm substrate and 25 μm YcgC at 37°C for 1 hr. The manuscript of Tu: 50 mM Tris-HCl pH 8.0, 50 mM NaCl, 4 mM MgCl2, 50 mM KCl, 1 mM NAD+, 0.15 μg/μl substrate, and 2.25 μg/μl enzyme at 37 °C for 1 hr.

b) RutR Substrates:

Current manuscript: C-terminal His6-tag containing either K52Ac or K62Ac expressed in E. coli BL21 (DE3) ΔycgCΔcobB strain through amber-suppressor-mediated non-natural amino acid incorporation.

The manuscript of Tu: C-terminal 3xFLAG-tagged expressed in E. coli strain (W3110) with K52Ac and K62Ac through posttranslational modifications.

c) YcgC:

Current manuscript: GST-TEV-tagged-YcgC in combination with TEV cleavage.

The manuscript of Tu: GST-YcgC

d) Antibody readouts:

Current manuscript: a specific anti-acetyl-L-lysine antibody (anti-AcK AB) that we have shown earlier to be well suited for following deacetylation reactions (de Boor, Knyphausen et al., 2015, Knyphausen et al., 2016, Kuhlmann, Wroblowski et al., 2016). The manuscript of Tu: an α-AcK antibody (Cell Signaling Technology, Shanghai, China)

Reviewer #2:

In this study, Lammers and coworkers report experiments that potentially call into question the conclusions of a previous study reported in eLife by Tu et al. In that previous paper, the authors reported that YcgC is a new type of deacetylase and that YcgC regulates transcription by catalyzing deacetylation of Lys52 and Lys62 of a transcriptional repressor RutR. The work was exciting because YcgC is not related to sirtuins or classical Lys deacetylases, and would add a new type of bacterial deacetylase (E. coli has only a single validated deacetylase, CobB). The proposed YcgC activity was discovered by using a clip-chip strategy, and activity was supported with His-tagged YcgC expressed in E. coli and purified and acetylated RutR obtained from an E. coli strain that apparently results in strongly acetylated protein. I read the previous paper and its initial reviewer comments and the response and I must say that the previous study appears solid.

In the current work, YcgC was expressed as a GST-tagged protein in E. coli, purified, and the tag removed by TEV protease, and C-terminally His-tagged RutR was obtained acetylated at the previously reported Lys52 and Lys62 by stop-codon suppression methods and purified. Unfortunately, in vitro incubation of the putative deacetylase and acetylated substrate did not lead to any activity.

If indeed YcgC is not a deacetylase, it will be very important for the community and hence I believe this work is important. However, there could be explanations for the discrepancies other than that the previous study was flawed, and these need to be explored. Both the YcgC enzyme and its substrate appeared to have been prepared differently in this study than in the previous study and apparently with different affinity tags. To truly test whether the previous work is reproducible, the same constructs should be used. The observations in the current study that the protein "behaved similarly on analytical SEC" is not really a measure of "the high quality of proteins". It shows that the proteins behave the same, but says nothing about the "quality". Similarly, I don't think that attaching the His-tag at the C-terminus of YcgC ensures it retains its activity as is written in the second paragraph of the subsection “Expression and purification of YcgC, CobB and RutR”, and identical behavior on SEC says nothing about proper folding, just that they are folded the same. I do not think the authors can rule out that either YcgC or RutR may be improperly folded in the current study, or that the hyper acetylation of RutR in the previous work may have been important.

I think at minimum, the authors of the current paper should prepare the same form of YcgC and the same form of RutR as in the previous study (same affinity tags and at the same positions). eLife should assist in requesting materials from Tu et al. In the end it is in the interest of both research groups and of the research community that the conclusions of the previous study are tested, but it is not in the interest of anyone if it turns out that the differences in conditions/proteins caused the apparent discrepancy and that YcgC does in fact have deacetylase activity.

When I read the previous study by Tu et al. I could not find if their proteins for their in vitro data were His-tagged at the N- or C-terminus. This is critical information and if it indeed was not in the paper, eLife should request that information from the original authors.

I did not think that the section: "YcgC does not show any deacetylase or proteolytic activity" added much if any value. Tu et al. never suggested YcgC was a general protease like trypsin. Instead they clearly proposed that upon deacetylation (by either YcgC or by CobB) RutR undergoes autoproteolysis. They also never suggested YcgC to act on a p53 based peptide. Hence, I did not think that these studies had much merit. The focus should be on YcgC-catalyzed deacetylation of RutR with proteins that are the same as those in the Tu et al. study. If that indeed leads to no activity, then I think it is important to publish this study.

Reviewer #3:

This report by Lammers et al. rebutted the previous discovery by Tu et al. that Ycgc is a unique protein deacetylase in bacteria.

Subsection “Expression and purification of YcgC, CobB and RutR”, first paragraph: GSH should be GST.

Figure 2A: in this reviewer's visual judgement, YcgC had some deacetylase activity but YcgC S200A did not.

Figure 2B was poorly presented and easy to cause confusion. The authors should include data of RutR WT, AcK52 and AcK62 for each cluster (i.e. Control, YcgC, and CobB). Error bars should be included. The figure legend does not seem to match the figure graph.

YcgC is a bacterial protein but p53 is a mammalian protein; therefore it is not a surprise that p53 peptide is not a substrate of YcgC. This point is particularly important as the previous study pointed out that YcgC has very narrow substrate specificity.

Tu et al. showed that RutR expressed in E. coli W3110 was highly acetylated, but this is not the case in this present study where BL21 (DE3) was used. Such drastic difference means physiologies of the two cellular systems are distinct and conclusions on RutR acetylation cannot be compared. This deserves further investigation.

This reviewer is worried about whether the differences in the protein constructs, acetyl-lysine antibody, and experimental assay protocols used between Tu's study and this one, may be the main reasons accounting for the divergent conclusions on whether YcgC has deacetylase activity. YcgC was assumed to have different mechanism from CobB, but why did the authors used the same protocols for deacetylase activity test with NAD present?

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for resubmitting your work entitled "Comment on "YcgC represents a new protein deacetylase family in prokaryotes"" for further consideration at eLife. Your revised article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Michael Marletta as the Senior Editor.

In this revision, you have carried out experiments to further test whether YcgC is not a deacetylase. You provide data suggesting that the observed activity in the previous eLife paper might arise from the trace contamination of protease(s). The evidence presented in the manuscript using both reagents, enzyme, and substrate preparations from the original authors and the current authors is strong. The methods used are rigorous using both MS and western blotting, You also show data suggesting that the original findings were highly dependent on blots using an antibody that appears non-selective for the desired antigen.

The manuscript has been improved and all three reviewers are strongly supportive of publication in eLife but you will need to address some editorial points (the article is much longer than our guidelines recommend, and the tone also needs attention in a number of places).

Revisions needed:

In the third paragraph of the Discussion, the authors provide quantification of the amount of acetylation of RutR in the preparation of Tu et al. They use the fact that they achieve 90% sequence coverage as argument that the mass spectrometry intensities can be used for quantification. The reviewers do not think this is valid. If the acetylated peptides ionize much more poorly than the nonacetylated peptides, then peak intensities can only be used if the authors have standard curves of mixtures of different ratios. Hence, they should be more qualitative in their discussions.

In Figure 2A, what is in the ctrl lane? Also, do YcgC and CobB happen to have the same mass? If so, it may be good to provide the calculated molecular weights of RutR, YcgG and CobB in the legend to avoid confusion for readers.

https://doi.org/10.7554/eLife.37798.015

Author response

Our decision has been reached after consultation between the reviewers. All three reviewers recognize the importance to publish your work if indeed the previous assignment of YcgC as a deacetylase turns out to be incorrect. However, all three reviewers also believe that at present, your experiments have not used the same conditions and enzymes as those used in the original work. Hence, the discrepancies could well lie in those differences in conditions and enzymes. Based on these discussions and the individual reviews below, we regret to inform you that your work as submitted will not be considered further for publication in eLife. However, if you were to investigate the deacetylase activity of YcgC on acetylated RutR under the same conditions as the previous study and find that no activity is observed, we would be willing to reconsider. The journal would also like to help in exchange of materials with the initial authors to find the source of the discrepancies.

As we were really inspired by the great findings by Tu and co-workers which reported the identification of a novel lysine-deacetylase class in prokaryotes, we initiated studies to elucidate structure and function to characterise the catalytic mechanism employed by YcgC to deacetylate lysine side-chains of substrates such as RutR suggested to result in its autoproteolytic cleavage.

We used a synthetic biological approach to produce quantitatively and site-specifically lysine acetylated RutR K52 (RutR AcK52) and K62 (RutR AcK62). These are the two by Tu and colleagues described acetyl-acceptor lysines in RutR. Moreover, we recombinantly expressed YcgC and purified it to a high level of purity. However, we were quite surprised that our data robustly show that YcgC was not capable to deacetylate neither RutR AcK52 nor RutR AcK62. In contrast, CobB was a potent deacetylase for RutR AcK52 and less efficiently also for RutR AcK62. We confirmed our data by electrospray-ionization mass-spectrometry next to immunoblotting experiments.

Reviewer #1:

In this manuscript, Kuhlmann et al. reported the function characterization of YcgC, which conflicts with what was reported by Tu and co-workers before. The negative results of the authors led to their claim that "YcgC is not a protein deacetylase". After comparing the two manuscripts, the reviewer found out the following key differences. On the basis of the limited data shown here, the reviewer could not decide whether the listed difference led to the conflicting results presented in the two manuscripts. However, given that the current manuscript built up its argument on the basis of the negative results and reported the finding after the first report, it should be the burden of the current authors to rule out that any of the following differences or their combination is not the cause of the conflicting observation. If possible, the authors should contact the lab for identical reagents unless it is inaccessible. It is not sufficient to simply justify why the new conditions were used without conducting the same experiments under the identical conditions as reported previously.

a) Buffer conditions for deacetylase assay:

Current manuscript: 50 mM Tris-HCl pH 7.4, 100 mM NaCl, 5 mM MgCl2, 5 mM NAD+, 2 mM β-mercaptoethanol, 12.5 μm substrate and 25 μm YcgC at 37°C for 1 hr. The manuscript of Tu: 50 mM Tris-HCl pH 8.0, 50 mM NaCl, 4 mM MgCl2, 50 mM KCl, 1 mM NAD+, 0.15 μg/μl substrate, and 2.25 μg/μl enzyme at 37 °C for 1 hr.

We repeated the experiments in exactly the same buffer and used the original proteins from Tu et al., which we got from Tu and co-workers. However, apart from reduction in molecular size, maybe due to proteolytic cleavage, we could not detect any deacetylase activity. We observed cleavage for all RutR preparations from Tu et al. and from our lab, including our site-specifically acetylated RutR AcK52 and AcK62 proteins. However, the acetylation signal was not removed showing that deacetylation and (auto-)proteolytic cleavage are not coupled. Furthermore, treatment with catalytically active CobB did not show any proteolytic degradation of no RutR protein analysed, including RutR from Tu et al. We also performed LC-MS/MS analyses on the original proteins from Tu et al. and observed that RutR is not quantitatively lysine acetylated with an overall acetylation occupancy of app 1%. AcK62 was not identified as being lysine acetylated. In the preparation of YcgC of Tu and co-workers, we identified several contaminants with protease/peptidase activity not in the YcgC S200A preparation from Tu et al., and also not in our YcgC preparation.

b) RutR Substrates:

Current manuscript: C-terminal His6-tag containing either K52Ac or K62Ac expressed in E. coli BL21 (DE3) ΔycgCΔcobB strain through amber-suppressor-mediated non-natural amino acid incorporation. The manuscript of Tu: C-terminal 3xFLAG-tagged expressed in E. coli strain (W3110) with K52Ac and K62Ac through posttranslational modifications.

Tu and co-workers used recombinantly expressed and purified RutR protein carrying an His6tag (see sequences in supplemental section of the revised manuscript, which we got from Tu eat al.). We observed the same results independently from where the His6-tag is placed, either at RutR N-terminus or at C-terminus.

We examined the RutR preparation of Tu et. Al by LC-MS/MS and found, firstly, that their RutR protein is not quantitatively lysine acetylated but total acetylation stoichiometry is approximately 1%. Secondly, AcK21 and AcK150 are the major acetylation sites, although both also occur at very low occupancy with 0.3% and 0.7%. K52 is found to be lysineacetylated but only with an overall stoichiometry of 0.02% and K62 is not acetylated at all. The working model presented by Tu and co-workers cannot be correct if acetylation is not quantitative. Considering these occupancies, almost quantitative proteolytic degradation of RutR as observed by treatment with YcgC from Tu et al. cannot be explained.

We also examined YcgC and YcgC S200A preparations from Tu and co-workers and found that in their YcgC preparation there are many contaminants not present in YcgC S200A or our YcgC preparation. Amongst those we identified several proteases including Lon protease, which is genetically deleted in most bacterial expression strains.

c) YcgC:

Current manuscript: GST-TEV-tagged-YcgC in combination with TEV cleavage.

The manuscript of Tu: GST-YcgC

According to communication with Tu et al., they used YcgC/YcgC S200A with N-terminal His6-tag as shown in supplemental material in the revised manuscript.

d) Antibody readouts:

Current manuscript: a specific anti-acetyl-L-lysine antibody (anti-AcK AB) that we have shown earlier to be well suited for following deacetylation reactions (de Boor, Knyphausen et al., 2015, Knyphausen et al., 2016, Kuhlmann, Wroblowski et al., 2016). The manuscript of Tu: an α-AcK antibody (Cell Signaling Technology, Shanghai, China)

Tu et al. used an anti-acetyl-lysine antibody from Cell Signaling Technology (CST9441). We applied anti-acetyl-lysine antibody (ab21623) from abcam, which we used in several studies in our lab. We compared both antibodies for detection of our protein preparations and protein preparations from Tu and co-workers. We found that the abcam antibody ab21623 did, in contrast to the antibody CST9441, not stain non-acetylated RutR proteins, neither from our lab nor the RutR protein from Tu et al.. This supports the notion that RutR from Tu et al. is not quantitatively lysine-acetylated and CST9441 might be unsuitable for this study. Using the abcam antibody ab21623 we see downshifted site-specifically acetylated RutR AcK52 and AcK62 proteins upon treatment with YcgC from Tu et al.. However, the lower molecular weight band is still lysine-acetylated. This shows that deacetylation by YcgC is not the molecular event preceding RutR (auto-)proteolytic cleavage as postulated by Tu et al.

Importantly, only the abcam antibody ab21623 detects deacetylation catalysed by CobB as the signal decreases for RutR AcK52 and to lesser extent also for AcK62 (which is supported by our ESI-MS data). According to the CST9441 antibody, treatment with CobB does not affect the overall signal of our quantitatively lysine-acetylated proteins RutR AcK52 (nor for AcK62). The CST9441 antibody detects all RutR proteins, from our lab and from Tu et al. Our mass-spec data show that RutR from Tu et al. is acetylated but acetylation occurs highly sub-stoichiometrically. Our hypothesis is that the CST9441 either detects trace amounts of acetyl-lysine and is therefore much more sensitive than the abcam antibody (or it detects something else such as the RutR protein sequence or N-terminal acetylation or a combination of all three). Apparently, if the CST9441 antibody detects lysine acetylation, it detects trace amounts of lysine-acetylation on RutR that could occur non-enzymatically during recombinant expression in E. coli. However the total occupancy of acetylation is so low, that it is most likely not of any physiological significance and it can for sure not explain the observed almost quantitative reduction in molecular size upon treatment with (only) YcgC from Tu et al. of all RutR proteins used in these studies (prepared by Tu et al. and by our lab). In sum, we can conclude that for these purposes to draw conclusions on lysine-acetylation and deacetylation the CST9441 antibody is not suitable.

From our point of view, the data can be explained by a protease contamination present in the YcgC preparation. We found many possible candidates selectively only in the YcgC preparation from Tu et al. by LC-MS/MS (but not in YcgC S200A from Tu et al. and not in our YcgC preparation). Proteolytic degradation of RutR might take place, which removes parts of the proteins, that are major sub-stoichiometric lysine-acceptor sites, K21 and K150, in RutR so that they are not even detectable by the antibody CST9441. One candidate might be the protease Lon. One of the Lon protease’s best-characterised substrates is the helix-turn-helix (HTH) LuxR-type transcriptional regulator RscA. This fuels the idea that maybe the HTH TetR-type transcriptional repressor RutR might be a specific substrate as well.

YcgC is clearly no deacetylase and the model must be revised. AcK52 is of very low stoichiometry in preparation of RutR from Tu et al., AcK62 is not present at all and total acetylation occupancy is very low.

Reviewer #2:

In this study, Lammers and coworkers report experiments that potentially call into question the conclusions of a previous study reported in eLife by Tu et al. In that previous paper, the authors reported that YcgC is a new type of deacetylase and that YcgC regulates transcription by catalyzing deacetylation of Lys52 and Lys62 of a transcriptional repressor RutR. The work was exciting because YcgC is not related to sirtuins or classical Lys deacetylases, and would add a new type of bacterial deacetylase (E. coli has only a single validated deacetylase, CobB). The proposed YcgC activity was discovered by using a clip-chip strategy, and activity was supported with His-tagged YcgC expressed in E. coli and purified and acetylated RutR obtained from an E. coli strain that apparently results in strongly acetylated protein. I read the previous paper and its initial reviewer comments and the response and I must say that the previous study appears solid.

We performed LC-MS/MS experiments that showed that RutR from Tu et al. is not quantitatively lysine acetylated. Instead overall acetylation occupancy is very low, app. 1%. We found K21 and K150 to be the major acetyl-acceptor lysines from low-occupancy sites. We found K52 in RutR from Tu et al. to be acetylated with occupancy of 0.02% and K62 was not at all found to be acetylated. This shows that the model from Tu et al. cannot be valid. For such a quantitative (auto-proteolytic) degradation of RutR as suggested by Tu and coworkers, if elicited by YcgC mediated deacetylation of RutR AcK52 and AcK62, the lysines have to be also almost quantitatively acetylated.

In the current work, YcgC was expressed as a GST-tagged protein in E. coli, purified, and the tag removed by TEV protease, and C-terminally His-tagged RutR was obtained acetylated at the previously reported Lys52 and Lys62 by stop-codon suppression methods and purified. Unfortunately, in vitro incubation of the putative deacetylase and acetylated substrate did not lead to any activity.

If indeed YcgC is not a deacetylase, it will be very important for the community and hence I believe this work is important. However, there could be explanations for the discrepancies other than that the previous study was flawed, and these need to be explored. Both the YcgC enzyme and its substrate appeared to have been prepared differently in this study than in the previous study and apparently with different affinity tags. To truly test whether the previous work is reproducible, the same constructs should be used.

We exchanged proteins with both groups and performed assays with the original protein preparations from Tu et al.. We clearly see that YcgC is no deacetylase. We only observe reduction of molecular size of all RutR proteins upon treatment with YcgC from Tu et al., but not with YcgC S200A from Tu et al. and not with our YcgC. Furthermore, in contrast to the publication of Tu et al., treatment with catalytically active CobB (prepared by us) did not show any proteolytic degradation of no RutR protein analysed, also not of RutR from Tu et al.

We analysed YcgC and YcgC S200A preparations from Tu et al. (and our YcgC preparation) and found several proteases present only in the YcgC from Tu et al. Using our site-specifically acetylated RutR AcK52 and AcK62 proteins we see reduction in molecular size upon treatment with YcgC from Tu et al. However, the degradation band does still show acetylation signal in immunoblotting showing that deacetylation is not the molecular event preceding (auto-)proteolysis as suggested by Tu et al.

The observations in the current study that the protein "behaved similarly on analytical SEC" is not really a measure of "the high quality of proteins". It shows that the proteins behave the same, but says nothing about the "quality". Similarly, I don't think that attaching the His-tag at the C-terminus of YcgC ensures it retains its activity as is written in the second paragraph of the subsection “Expression and purification of YcgC, CobB and RutR”, and identical behavior on SEC says nothing about proper folding, just that they are folded the same. I do not think the authors can rule out that either YcgC or RutR may be improperly folded in the current study, or that the hyper acetylation of RutR in the previous work may have been important.

We rephrased some of the statements as suggested by reviewer 2. For our RutR proteins (RutR, RutR AcK52 and RutR AcK62) we can exclude a hyperacetylation as we see the correct molecular mass by ESI-MS. We performed LC-MS/MS also with the original RutR preparation from Tu et al. and found that the overall acetylation occupancy is low, only app. 1%. Size-exclusion chromatography can give a hint if the oligomeric state is comparable. If acetylation of RutR would mass up the structure, we would see this also by SEC as most likely we would observe an increase of protein elution very early as aggregates in the exclusion volume. Therefore, we think analytical SEC runs can be used for quality control.

I think at minimum, the authors of the current paper should prepare the same form of YcgC and the same form of RutR as in the previous study (same affinity tags and at the same positions). eLife should assist in requesting materials from Tu et al. In the end it is in the interest of both research groups and of the research community that the conclusions of the previous study are tested, but it is not in the interest of anyone if it turns out that the differences in conditions/proteins caused the apparent discrepancy and that YcgC does in fact have deacetylase activity.

As written above, we exchanged all proteins between both labs and we included data with proteins from Tu et al. in this revised manuscript. These data include immunoblottings using all proteins and both antibodies used for deletion. Additionally, we performed LC-MS/MS with YcgC and RutR preparations Tu et al. These data clearly show that YcgC is no deacetylase. RutR from Tu et al. is not quantitatively lysine-acetylated and this per se shows that the molecular model cannot be correct as quantitative autoproteolysis would also request that acetylation is quantitative. Moreover, although AcK52 was found on RutR from Tu et al. the occupancy is very low (app. 0.1%) and AcK62 was not found at all. Our LC-MS/MS data on ycgC show clearly contaminations with several proteases only present in YcgC from Tu et al. but neither in our YcgC preparation nor in YcgC S200A prepared by Tu et al.

When I read the previous study by Tu et al. I could not find if their proteins for their in vitro data were His-tagged at the N- or C-terminus. This is critical information and if it indeed was not in the paper, eLife should request that information from the original authors.

I did not think that the section: "YcgC does not show any deacetylase or proteolytic activity" added much if any value. Tu et al. never suggested YcgC was a general protease like trypsin. Instead they clearly proposed that upon deacetylation (by either YcgC or by CobB) RutR undergoes autoproteolysis. They also never suggested YcgC to act on a p53 based peptide. Hence, I did not think that these studies had much merit. The focus should be on YcgC-catalyzed deacetylation of RutR with proteins that are the same as those in the Tu et al. study. If that indeed leads to no activity, then I think it is important to publish this study.

We removed this section from the revised manuscript.

Reviewer #3:

This report by Lammers et al. rebutted the previous discovery by Tu et al. that Ycgc is a unique protein deacetylase in bacteria.

Subsection “Expression and purification of YcgC, CobB and RutR”, first paragraph: GSH should be GST.

Figure 2A: in this reviewer's visual judgement, YcgC had some deacetylase activity but YcgC S200A did not.

Figure 2B was poorly presented and easy to cause confusion. The authors should include data of RutR WT, AcK52 and AcK62 for each cluster (i.e. Control, YcgC, and CobB). Error bars should be included. The figure legend does not seem to match the figure graph.

We corrected the points suggested by reviewer 3, for YcgC, we are sure that it does not have deacetylase activity. If it acted like an enzyme the signal should be completely gone regarding the huge enzyme:substrate ration we used in this assay. These assays are meant only to make qualitative statements. In Figure 2B we show the quantification of ESI-MS spectra of the reactions as described. These reactions are shown in A using immunoblotting as a readout. We corrected it accordingly to make this clearer.

YcgC is a bacterial protein but p53 is a mammalian protein; therefore it is not a surprise that p53 peptide is not a substrate of YcgC. This point is particularly important as the previous study pointed out that YcgC has very narrow substrate specificity.

Reviewer 3 is correct in saying that p52 might not be a substrate as also already mentioned by reviewer 2. Therefore, we deleted this section from the revised manuscript.

Tu et al. showed that RutR expressed in E. coli W3110 was highly acetylated, but this is not the case in this present study where BL21 (DE3) was used. Such drastic difference means physiologies of the two cellular systems are distinct and conclusions on RutR acetylation cannot be compared. This deserves further investigation.

We performed LC-MS/MS analyses on the original RutR protein preparation from Tu et al. We found that lysine acetylation occupancy of the RutR of Tu et al. is very low (app. 1%). We found that K21 and K150 are the major acetyl-acceptor sites, although of course occurring at very low occupancy (0.3% and 0.7%, respectively). We found K52 to be acetylated with about 0.02% and K62 was not identified to be acetylated at all. These data show that the working model presented by Tu et al. cannot be correct.

This reviewer is worried about whether the differences in the protein constructs, acetyl-lysine antibody, and experimental assay protocols used between Tu's study and this one, may be the main reasons accounting for the divergent conclusions on whether YcgC has deacetylase activity. YcgC was assumed to have different mechanism from CobB, but why did the authors used the same protocols for deacetylase activity test with NAD present?

We exchanged proteins with both groups and performed assays with the original protein preparations from Tu et al. under the same conditions. We see clearly that YcgC is no deacetylase. We only observe reduction of molecular size of all RutR proteins upon treatment with YcgC from Tu et al., but not with YcgC S200A from Tu et al. and not with our YcgC. We analysed YcgC and YcgC S200A preparations from Tu et al. (and our YcgC preparation) by LC-MS/MS and found several proteases present only in YcgC preparation by Tu et al. Amongst these proteases we found Lon protease in YcgC by Tu et al. Using our site specifically acetylated RutR acK52 and AcK62 proteins we see reduction in molecular size upon treatment with YcgC preparation from Tu et al. However, the degradation band does still show acetylation signal in immunoblotting showing that deacetylation is not the molecular event preceding (auto-)proteolysis as suggested by Tu et al.. Furthermore, treatment with catalytically active CobB did not show any proteolytic degradation of no RutR protein analysed, also not of RutR from Tu et al. This is also in contrast to the data shown by Tu et al. in the original publication.

[Editors' note: the author responses to the re-review follow.]

In the third paragraph of the Discussion, the authors provide quantification of the amount of acetylation of RutR in the preparation of Tu et al. They use the fact that they achieve 90% sequence coverage as argument that the mass spectrometry intensities can be used for quantification. The reviewers do not think this is valid. If the acetylated peptides ionize much more poorly than the nonacetylated peptides, then peak intensities can only be used if the authors have standard curves of mixtures of different ratios. Hence, they should be more qualitative in their discussions.

In Figure 2A, what is in the ctrl lane? Also, do YcgC and CobB happen to have the same mass? If so, it may be good to provide the calculated molecular weights of RutR, YcgG and CobB in the legend to avoid confusion for readers.

We are delighted that the reviewers are now convinced that YcgC has not the desired deacetylase activity. We included all of the reviewers’ and the eLife editors’ suggestions in the revised version of the manuscript. From my point of view this strongly improved the manuscript in terms of its conciseness and in its tone.

https://doi.org/10.7554/eLife.37798.016

Article and author information

Author details

  1. Magdalena Kremer

    Institute for Genetics and Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), University of Cologne, Cologne, Germany
    Contribution
    Conceptualization, Formal analysis, Methodology
    Contributed equally with
    Nora Kuhlmann
    Competing interests
    No competing interests declared
  2. Nora Kuhlmann

    Institute for Genetics and Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), University of Cologne, Cologne, Germany
    Contribution
    Investigation, Methodology
    Contributed equally with
    Magdalena Kremer
    Competing interests
    No competing interests declared
  3. Marius Lechner

    Institute for Genetics and Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), University of Cologne, Cologne, Germany
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared
  4. Linda Baldus

    Institute for Genetics and Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), University of Cologne, Cologne, Germany
    Contribution
    Investigation
    Competing interests
    No competing interests declared
  5. Michael Lammers

    1. Institute for Genetics and Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), University of Cologne, Cologne, Germany
    2. Institute of Biochemistry, Synthetic and Structural Biochemistry, University of Greifswald, Greifswald, Germany
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    michael.lammers@uni-greifswald.de
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4168-4640

Funding

Deutsche Forschungsgemeinschaft (LA 2984/3-1)

  • Michael Lammers

Deutsche Forschungsgemeinschaft (CECAD)

  • Magdalena Kremer
  • Kuhlmann Nora
  • Marius Lechner
  • Linda Baldus
  • Michael Lammers

Boehringer Ingelheim Fonds (Exploration Grant)

  • Kuhlmann Nora

Deutsche Forschungsgemeinschaft (Cologne Graduate School of Ageing Research)

  • Magdalena Kremer

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

Acknowledgements

We thank the CECAD proteomics facility headed by Dr. Christian Frese and Dr. Stefan Müller for the helpful discussions setting up the MS experiments and conducting the experiments. This work was funded by Cologne Excellence Cluster on Cellular Stress Responses in Aging-Associated Diseases (CECAD), the Heisenberg Programm (grant: LA2984/3-1) of the German Research Foundation (Deutsche Forschungsgemeinschaft; DFG) and the Boehringer Ingelheim Foundation (Exploration Grant Programme).

Reviewing Editor

  1. Wilfred A van der Donk, University of Illinois at Urbana-Champaign, United States

Publication history

  1. Received: April 23, 2018
  2. Accepted: June 18, 2018
  3. Accepted Manuscript published: June 25, 2018 (version 1)
  4. Version of Record published: June 28, 2018 (version 2)

Copyright

© 2018, Kremer 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

  • 1,146
    Page views
  • 113
    Downloads
  • 6
    Citations

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

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)

  1. Further reading

Further reading

    1. Biochemistry and Chemical Biology
    2. Genetics and Genomics
    Krishna S Ghanta et al.
    Research Article

    Nuclease-directed genome editing is a powerful tool for investigating physiology and has great promise as a therapeutic approach to correct mutations that cause disease. In its most precise form, genome editing can use cellular homology-directed repair (HDR) pathways to insert information from an exogenously supplied DNA repair template (donor) directly into a targeted genomic location. Unfortunately, particularly for long insertions, toxicity and delivery considerations associated with repair template DNA can limit HDR efficacy. Here, we explore chemical modifications to both double-stranded and single-stranded DNA-repair templates. We describe 5′-terminal modifications, including in its simplest form the incorporation of triethylene glycol (TEG) moieties, that consistently increase the frequency of precision editing in the germlines of three animal models (Caenorhabditis elegans, zebrafish, mice) and in cultured human cells.

    1. Biochemistry and Chemical Biology
    2. Structural Biology and Molecular Biophysics
    Paul Fischer et al.
    Research Article

    Enzymerhodopsins represent a recently discovered class of rhodopsins which includes histidine kinase rhodopsin, rhodopsin phosphodiesterases and rhodopsin guanylyl cyclases (RGCs). The regulatory influence of the rhodopsin domain on the enzyme activity is only partially understood and holds the key for a deeper understanding of intra-molecular signaling pathways. Here we present a UV-Vis and FTIR study about the light-induced dynamics of a RGC from the fungus Catenaria anguillulae, which provides insights into the catalytic process. After the spectroscopic characterization of the late rhodopsin photoproducts, we analyzed truncated variants and revealed the involvement of the cytosolic N-terminus in the structural rearrangements upon photo-activation of the protein. We tracked the catalytic reaction of RGC and the free GC domain independently by UV-light induced release of GTP from the photolabile NPE-GTP substrate. Our results show substrate binding to the dark-adapted RGC and GC alike and reveal differences between the constructs attributable to the regulatory influence of the rhodopsin on the conformation of the binding pocket. By monitoring the phosphate rearrangement during cGMP and pyrophosphate formation in light-activated RGC, we were able to confirm the M state as the active state of the protein. The described setup and experimental design enable real-time monitoring of substrate turnover in light-activated enzymes on a molecular scale, thus opening the pathway to a deeper understanding of enzyme activity and protein-protein interactions.