Bacteria employ lysine acetylation of transcriptional regulators to adapt gene expression to cellular metabolism

The Escherichia coli TetR-related transcriptional regulator RutR is involved in the coordination of pyrimidine and purine metabolism. Here we report that lysine acetylation modulates RutR function. Applying the genetic code expansion concept, we produced site-specifically lysine-acetylated RutR proteins. The crystal structure of lysine-acetylated RutR reveals how acetylation switches off RutR-DNA-binding. We apply the genetic code expansion concept in E. coli in vivo revealing the consequences of RutR acetylation on the transcriptional level. We propose a model in which RutR acetylation follows different kinetic profiles either reacting non-enzymatically with acetyl-phosphate or enzymatically catalysed by the lysine acetyltransferases PatZ/YfiQ and YiaC. The NAD+-dependent sirtuin deacetylase CobB reverses enzymatic and non-enzymatic acetylation of RutR playing a dual regulatory and detoxifying role. By detecting cellular acetyl-CoA, NAD+ and acetyl-phosphate, bacteria apply lysine acetylation of transcriptional regulators to sense the cellular metabolic state directly adjusting gene expression to changing environmental conditions.

M a rk e r R u tR -H

Supplementary Fig 2: Quantitative incorporation of N-(e)-acetyl-L-lysine into RutR at distinct sites shown by electrospray-ionization mass-spectrometry (ESI-MS).
Electrospray-ionization mass-spectrometry (ESI-MS) of RutR proteins.The determined molecular masses do correspond exactly (≤ 1.5 Da) to the protein masses (RutR: 24567.5 Da; acetylated RutR: 24609.5 Da; acetyl group: 42 Da).All RutR proteins, i.e. lysine-acetylated and non-acetylated RutR, carry a C-terminal hexahistidine-tag (His6-tag).The table summarizes the expected and determined molecular weights of the proteins.Supplementary Figure 3: LC-MS/MS fragmentation spectra of lysine-acetylated peptides obtained for recombinantly expressed and purified site-specifically lysine-acetylated RutR.Except for RutR AcK11 all acetylation sites were confirmed.For RutR Ack21, Ack95 and AcK150 two peptides were identified one of which containing a missed cleavage.RutR AcK11 was not detectable the peptide resulting after proteolytic cleavage was too small (8-TTGKR-12).However, as the total mass of RutR AcK11 was correct, sequencing of the DNA-construct revealed presence of an amber stop codon at the correct position and the immunoblotting by anti-AcK antibody showed that it is lysineacetylated, the site-specific acetylation of RutR at K11 was confirmed.
Supplementary Figure 4: Quantification of the electrophoretic-mobility shift assays (EMSAs).a-l.Protein-DNA interaction between RutR variants and promoter (boxrutA: 50 bp segment of PromrutA, boxcarA:46 bp dsDNA segment of PromcarA) as well as control (ctrl) oligonucleotides were studied in EMSAs.The signal intensities were quantified using ImageJ and plotted as a function of the RutR protein concentration.The values are given as means ± standard deviations calculated from at least three replicates (n≥3).The binding curves were analysed by applying a quadratic model to fit the data.The    For ectopic expression cells of E. coli U65 DrutR with a genomic PcarA-lacZ fusion were transformed with pRSFDuet-1/rutR/ackRS3/MbpylT or pRSFDuet-1/rutRK52amber/ackRS3/MbpylT.Expression of rutR-FLAG at endogenous level was achieved using E. coli U65 rutR-FLAG-kanR.As indicated, using varying IPTG concentrations (0, 5, 10 and/or 1000 µM) were applied to assess the impact of IPTG on the expression.Cells were harvested at OD600 = 0.6.RutR-FLAG protein levels were detected by immunoblotting with an anti-FLAG-antibody.TCE staining served as loading control.For endogenous expressions (g: genomically inserted rutR-FLAG) double the volume of the cell lysate was loaded compared to the ectopic expressions.a.u.arbitrary units.The protein level was quantified by ImageJ software.Statistical analyses was performed using students t-test (unpaired, two-tailed).Experiments were performed in three biologically independent experiments, two of which contain two technical replicates (n=3).Source data including statistical analyses are provided as Source Data file.b.The samples described in A for IPTG-dependent ectopic expressions of rutR-FLAG (encoding for non-acetylated and K52-acetylated RutR-FLAG) were analysed for b-galactosisase activity as an indicator for RutR-FLAG transcriptional regulator activity.All RutR-FLAG AcK52 samples show a statistically significant reduction in b-galactosidase activity compared to the samples of nonacetylated RutR, reaching a level similar to the empty vector controls.Notably, these effects are independent of the IPTG concentration used.Statistical analyses was performed using students ttest (unpaired, two-tailed; Source Data file).Experiments were performed in four biologically independent experiments, each with two technical replicates (n=4).Source data including statistical analyses are provided as Source Data file.Other reported sites reported earlier include K62 in a4 3,4 .b. RutR full-length was treated for 18 h (B) with 10 mM acetyl-phosphate in 100 mM K2PO4/KH2PO4 pH7.5 at room temperature.Afterwards the proteins were digested with trypsin and analyzed by mass-spectrometry.The spectra obtained for acetylated peptides identified after treatment for 18 h are shown.For full-length RutR we could identify AcK7, AcK21, AcK52 and AcK95.In a study performed earlier in our group we also identified K7, K11, K19, K20, K21, K95 and K150 in fulllength RutR.c. RutR D1-12 was treated for 18 h (B) with 10 mM acetyl-phosphate in 100 mM K2PO4/KH2PO4 pH7.5 at room temperature.Afterwards the proteins were digested with trypsin and analyzed by massspectrometry.The spectra obtained for acetylated peptides identified after treatment for 18 h are shown.For RutR D1-12 we could identify AcK21, AcK52 and AcK95.7: Single values for the kinetic characterization of the interaction of boxrutA and boxcarA dsDNA towards non-acetylated RutR (WT), RutR AcK7, RutR AcK11 and double acetylated RutR AcK7/11 by surface plasmon resonance studies.The values obtained correspond to the association rate constant, kass, the dissociation rate constant, kdiss, and the calculated dissociation equilibrium constant, KD.Rmax shows the maximum of obtained resonance units (RU).For kass, kdiss and KD, the mean values are calculated.Experiments were performed in three biologically independent experiments (n=3).

Supplementary Figure 6 :
Uracil and RutR acetylation are additively impairing DNA-binding.Electrophoretic-mobility shift assays were performed to assess the impact of uracil on binding of acetylated RutR variants to boxcarA and boxrutA DNA.Presence of 200 µM uracil (+) impairs RutR DNAbinding to both DNA fragments.For RutR AcK7, AcK11, AcK52 and AcK62 addition of uracil results in an additional decrease of RutR DNA-binding suggesting that acetylation and uracil exert their impact using different molecular mechanisms.Experiment was conducted once (n=1).Source data are provided as Source Data file.

Supplementary Figure 11 : 2 .Supplementary Figure 12 :Supplementary Figure 13 :
Primary sequence and structural alignment of selected TetR-related transcriptional regulators of various Gram-positive and Gram-negative bacterial strains (UniProt accession numbers are shown).a. Structural alignment of AlphaFold2 models of selected TetR-related transcriptional regulators of various bacterial strains (Supplementary Data 3) were aligned onto the HTH motif of RutR.Shown are only the N-terminal tails and the HTH-motifs.Many transcription factors carry an overrepresented number of positively-charged residues in the N-terminal tail preceding the HTHmotif and within the HTH-motif suggesting a potential to be regulated by lysine acetylation.The area, in which K52 of RutR is located (at N-terminus of a3) and in which K62 is located (N-terminal region of a4) is shown.This resembles histone-like N-terminal tails of eukaryotes and suggests a similar regulation of transcription factor DNA-binding by lysine acetylation as observed for histones in eukaryotes.Lysine and arginine side chains are shown in stick representation (Supplementary Data 3).b.Sequences of all proteins listed in Supplementary Data 3 were aligned by ClustalW multiple sequence alignment (Bioedit).Several proteins contain a positively-charged N-terminal tail preceding the HTH-motif.The secondary structure elements and numbering is shown for RutR above the sequence alignment.RutR K52 and K62 are conserved in several family members.The 19-KKK-21 basic motif of RutR is also conserved in several family members.The alignment was created by ESPript version 3.0 The KATs PhnO, YjaB and RimI are inactive in acetylating RutR and PatZ/YfiQ and YiaC are not able to acetylate RutR K52.a. PhnO, YjaB and RimI are not active in acetylating RutR.As described before, RutR was incubated with the respective KAT enzymes (inactive: PhnO Y128A; RimI Y115A; YjaB Y117A) in presence/absence of acetyl-coenzyme A (Ac-CoA) as indicated.The samples were subsequently analyzed in immunoblots (IB) probed with anti-AcK AB and total protein staining using 2,2,2trichloroethanol (TCE) was used as loading control.For PhnO, YjaB and RimI an autoacetyltransferase activity is observed.One exemplary result is shown and the results were confirmed in at least three replicates (n≥3).Long exposure was conducted to visualize weak signals.Red color indicates oversaturation of signal.Source data are provided as Source Data file.b.PatZ/YfiQ and YiaC are not active in acetylating RutR at K52.In vitro KAT activity assays with RutR and RutR K52Q show that the signal is not impaired in RutR K52Q.The samples were subsequently analyzed in immunoblots (IB) stained with anti-AcK AB. 2,2,2-trichloroethanol (TCE) staining was used as loading control.One exemplary result is shown and the results were confirmed in three replicates (n=3).Red indicates oversaturation of signal.Source data are provided as Source Data file.MS/MS fragmentation spectra of lysine-acetylated peptides obtained for RutR acetylated in vitro upon treatment with 10 mM acetyl-phosphate for 18 h.a.We show here that RutR is acetylated non-enzymatically by acetyl-phosphate at lysines in the Nterminal histone-like tail (K7, K11 and K7/K11), in the helix a1 preceding the HTH-motif (triple Kmotif, 19-KKK-21), in the HTH-motif (K52) and in the LBD (K95 and K150).The acetylation sites are shown in stick representation highlighted in green using the AlphaFold2 structure of RutR.

Supplementary Figure 15 :
CobB-mediated deacetylation of RutR AcK7, AcK11 and AcK7/11.a. CobB-catalysed deacetylation of RutR AcK7, AcK11 and AcK7/11 is not inhibited using 10 mM nicotinamide (NA) under the reaction conditions used resulting in complete RutR deacetylation.This shows that these sites are more efficiently deacetylated by CobB compared to RutR AcK52 or AcK62.Source data are provided as Source Data file.b.CobB can completely deacetylate RutR protein acetylated enzymatically by PatZ/YfiQ or YiaC.Red indicates oversaturation of the signal.

Supplementary Figure 5: Glutamine is a poor molecular mimic for RutR K52-acetylation as shown by EMSAs and by isothermal titration calorimetry (ITC).
table summarizes the KD values as indicated.Source data are provided as Source Data file.Interaction between RutR K52R and K52Q mutants towards boxrutA and boxcarA DNA analyzed by ITC.Shown are exemplary ITC traces (DP: differential power).A one-site binding model was used to fit the data.All interactions were determined at least in three biologically independent experiments and the values are given as means ± standard deviations (Supplementary Table3; Source Data file) (n≥3).
and mutation of K to Q to mimic a lysine acetylation.However, apart from the neutral charge a Q does sterically not resemble an AcK.Sterically a mutation of K to R might mimic an AcK more than a Q.One exemplary result of at least three replicates is shown (n=3).Source data are provided as Source Data file.b. c.Summary of the thermodynamic characterization of the interaction of acetylated RutR AcK52 and AcK62 and RutR K52R/K52Q mutants towards boxrutA and boxcarA DNA analyzed by ITC.Shown are the results of all measurements.A one-site binding model was used to fit the data.All interactions were determined at least in three biologically independent experiments and the values are given as means ± standard deviations (Supplementary Table3; Source Data file) (n≥3).Source data are provided as Source Data file.