All strains, plasmids and primers used in this study were described in Supplementary Tables S1 and S2. Luria-Bertani broth (LB) (per liter, 5 g NaCl, 5 g yeast extract and 10 g tryptone) was used in this work, agar plates contained 1.5% (W/V) agar and supplemented with antibiotics as required. The antibiotics used were 100 μg/ml of ampicillin, 17 μg/ml of chloramphenicol, 50 μg/ml of kanamycin and 50 μg/ml of spectinomycin.

The protocol to construct genome fragment libraries was described previously with some modification (Houot et al., 2012). The genomic DNA of S. Typhimurium strain 14028s was prepared using PureLink Genomic DNA kit (Invitrogen) and partially digested by Sau3AI. The randomly digested DNA was separated on a 0.8% agarose gel, and fragments ranging in size from 2,000 to 4,000 bp were gel-purified using Qiagen gel extraction kit (Qiagen). Three independent two-hybrid libraries were constructed. The first libraries were generated using the pKT25 vectors (Karimova et al., 1998). The other two libraries used pKT25-derivative vectors in which the polylinker site was shifted to +1 and +2 nt, respectively (Handford et al., 2009). The vectors pKT25, pKT25+1 and pKT25+2 were digested by BamH I (New England Biolabs) at 37°C for 2.5 h and dephosphorylated with Shrimp Alkaline Phosphatase (New England Biolabs) at 37°C for 1 h. Several independent pools of DNA fragments were ligated overnight at 16°C into the different pKT25 linearized vectors using T4 ligase (New England Biolabs). The resulting ligation mixture was transformed into E. coli DH5α competent cells, and plated on LB-ampicillin plates. The resulting transformants were collected by scraping the plates with a glass spreader, and pooled before plasmid isolation. The quantity of each final plasmid library was evaluated under UV illumination.

The protein–protein interaction was detected by Bacterial Adenylate Cyclase Two-Hybrid System Kit (Euromedex) according to the product manuals.

Libraries were tested independently against each pUT18 bait (pKT25-SalLib for pat or cobB, pKT25+1-SalLib for pat or cobB and pKT25+2-SalLib for pat or cobB). Basically, 25–50 ng of each p KT25-derivative library was transformed into 100 μl of electrocompetent BTH101 cells carrying the pUT18 bait vector and plated on screening medium containing 5-bromo-4-chloro-3-indolyl- β-D-galactopyranoside (40 μg/ml, X-gal) and isopropyl-β-D-thiogalactopyranoside (0.5 mM, IPTG). Bacteria expressing interacting hybrid proteins will form blue colonies on LB/X-Gal medium, while cells expressing non-interaction proteins will remain white. IPTG was used to increase β-galactosidase expression. A co-transformant containing pKT25-zip and pUT18-zip was used as a positive control for expected growth on the screening medium. A co-transformant containing empty vector pKT25 and pUT18 was used as a negative control. The blue colonies were picked up and recultivated in liquid medium, and plasmids were isolated and further analyzed by DNA sequencing.

To further validate the results from high-throughput two-hybrid assays, the detected genes (rhaR, STM14_1074, lrp) and pat or cobB genes were cloned into pKT25 and pUT18, respectively. The BTH101 strain was co-transformed with the recombinant vectors pKT25-rhaR, pKT25-STM14_1074, pKT25-lrp and pUT18-pat or pUT18-cobB, and then spotted onto screening medium.

Fusion PCR and λ Red recombination were employed to construct S. Typhimurium Lrp K36Q, K36R, K36A and the wild type strain. The chloramphenicol resistance cassette was inserted between lrp and ftsK. PCR products were gel extracted and electroporated into S. Typhimurium containing pKD46 prepared in the presence of 10 mM arabinose. The knock-in strains were verified by PCR and DNA sequencing.

Site-directed mutagenesis of lrp was performed with the corresponding primers (Supplementary Table S2), using KOD-Plus-Mutagenesis Kit according to the manufacturer’s recommendations (Toyobo). The resulting gene mutations were confirmed by DNA sequencing.

For purification of Lrp and its derivative mutants, pET22b-lrp, pET22b-lrp K36Q, pET22b-lrp K36R, and pET22b-lrp K25Q were constructed and verified by sequencing. All the constructed plasmids were transformed individually into E. coli strain BL21, and the resultant strains were grown in LB medium containing 100 μg/ml ampicillin at 37°C. IPTG was added to a final concentration of 0.5 mM at the time point when the absorbance of the culture at OD600 ~0.8. The culture was continuously incubated at 37°C for 2 h. Afterward cells were harvested by centrifugation and stored at -80°C. All subsequent procedures were performed at 4°C. Thawed bacteria were resuspended in lysis buffer [50 mM Tris–HCl, 0.5 M NaCl, 10% (V/V) glycerol, pH 7.6], and were broken by high pressure cracker. Any insoluble material was removed by centrifugation for 60 min at 19,000 g. The soluble extract was applied to a 1 ml column of nickel-nitrilotriacetic acid (Ni-NTA) agarose (Qiagen) that had been equilibrated with lysis buffer containing. The column was subsequently washed with 10 ml of buffer (50 mM Tris–HCl, 0.5 M NaCl, 10% (V/V) glycerol, 20 mM imidazole pH 7.6), and eluted stepwise with 2 ml aliquots of buffer containing 500 mM imidazole. For the protein whose purity was less than 90%, a further purification step was applied in a 1 ml HiTrap Q HP anion-exchange columns (GE Healthcare) as described previously (Bachhawat and Stock, 2007). The polypeptide compositions of the column fractions were monitored in 15% SDS–PAGE, and subsequent staining with Coomassie bright blue.

For expression of site-specifically acetylated Lrp (K36Ac). E. coli strain BL21 was transformed with plasmid pAcKRS-3 and pCDF-PylT-lrp K36 (TAG) and then were grown overnight in LB supplemented with 50 μg/ml kanamycin and 100 μg/ml spectinomycin (LB-KS). One liter of prewarmed LB-KS was inoculated with 20 ml overnight culture and was incubated at 37°C. When OD600 reached 1.5, the culture was supplemented with one liter of fresh LB with 20 mM acetyllysine (AcK). Protein expression was induced by addition of 0.5 mM IPTG for 6 h at 37°C, cells were harvested by centrifugation, and stored at -80°C. The purification method was same as Lrp.

For purification of RhaR, STM14_1074, NrdF and FliT, the plasmids pET22b-rhaR, pET22b-STM14_1074, pET22b-nrdF, pET22b-fliT were constructed and verified by DNA sequencing. Expression and purification of RhaR, STM14_1074, NrdF and FliT were performed as Lrp. The protocols to purify Pat and CobB were described previously (Ren et al., 2016).

The Pat-mediated in vitro acetylation reaction was performed as described previously (Ren et al., 2016). The in vitro acetylation reaction buffer containing 50 mM Tris-HCl (pH 8.0), 0.1 mM EDTA, 10% glycerol, 1 mM dithiothreitol and 10 mM sodium butyrate. The acetylation was carried out by adding 10 μg Pat, 12.5 mM acetyl-CoA, in a volume of 50 μl. Reaction mixtures were completely mixed and incubated at 37°C for 3 h. Specific protein concentrations are indicated in figure legends.

The AcP-mediated in vitro acetylation reaction in the buffer system containing 20 mg/ml bovine serum albumin (BSA), 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA. Lrp was mixed with freshly prepared AcP (20 mM, potassium lithium salt, Sigma) at 37°C for 2 h.

The CobB-mediated in vitro deacetylation reaction was performed as described previously (Ren et al., 2016). Deacetylation reaction by CobB was performed in deacetylation reaction buffer including 50 mM Tris-HCl (pH 8.0), 135 mM NaCl, 2.5 mM KCl, 1 mM MgCl₂ in the presence or absence of 1 mM NAD⁺, and in the presence or absence of 10 mM nicotinamide (NAM), an inhibitor for CobB (Bitterman et al., 2002). Specific protein concentrations and reaction times are indicated in figure legends.

Protein concentration was determined using the Bradford reagent with BSA as a standard (Bradford, 1976).

The standard Western blot procedure was described previously (Ren et al., 2016). For acetylation Western blot, 50 mM Tris (pH 7.5) with 10% (v/v) Tween-20 and 1% peptone was used for blocking and 50 mM Tris (pH 7.5) with 0.1% peptone was used to prepare primary and secondary antibodies.

DNA fragments used for EMSA were amplified by PCR using Salmonella Typhimurium strain 14028s genomic DNA as a template. The promoter region of fimZ was amplified using primers fimZ-F (5′FAM-ACACAGTGGAGCAATAATAA) and fimZ-R (5′FAM-CAGAATATCCTACCCGCTAT). PCR amplification rendered fragments of 202 bp. Approximately 1 nmol of 5′FAM-labeled DNA in a 20 μl volume was incubated at 25°C for 20 min with the indicated amounts of purified the wild type Lrp or mutants. The binding buffer used for protein-DNA incubations contained 100 mM Tris-HCl (pH 8), 375 mM NaCl, 25 mM MgCl₂, 5 mM DTT, 65% glycerol. Samples were separated on a 5% non-denaturing Tris-glycine polyacrylamide gel at 4°C. FAM-labeled fluorescence was detected by the Fujifilm FLA7000.

Bacteria were cultured in LB broth to log phase, harvested and disrupted by Bead beaters. RNA was isolated using Trizol (Invitrogen), and DNase I digestion was conducted as described previously (Sang et al., 2016). Primers for qPCR were listed in Supplementary Table S2. Samples were run in triplicates and amplified using SYBR Premix Ex Taq II (TAKARA). The relative transcriptional level was determined by the methods of 2^-ΔΔCt (Augagneur et al., 2007). 16S rRNA was used as a reference gene.

The yeast agglutination assay was carried out according to the previous protocol with some modification (Roe et al., 2001). A single colony was used to inoculate 5 ml LB broth, cultured at 37°C without shaking, then subcultured twice for overnight static growth. Agglutination was assayed in slide by mixing 5 μl bacterial culture with an equal volume of bakers’ yeast suspension (10 mg/ml). Mannose inhibition of agglutination was confirmed using 3% α-D-mannose in the yeast suspension.

To screen the protein targets of Pat or CobB, the bacterial two-hybrid system (BTH) was employed. At first, the S. Typhimurium strain 14028s genome libraries were constructed. The first one, designated pKT25-SalLib and in order to get a large coverage of the S. Typhimurium genome, we engineered two additional libraries (pKT25+1-SalLib and pKT25+2-SalLib) in which genome libraries were cloned in the pKT25 vector or shifted to +1 and +2 nt in the pKT25+1 and pKT25+2 vectors, respectively. Thus, the final S. Typhimurium prey libraries consisted of 33000 clones, and were estimated to represent 4.39-fold coverage of S. Typhimurium genome.

Next, acetyltransferase Pat and deacetylase CobB proteins were used to test our high-throughput S. Typhimurium two-hybrid library with the aim of identifying unknown targets. At last, in total of 12 DNA fragments encoded proteins were screened and confirmed by DNA sequencing (Supplementary Table S3), and eight proteins interacted with Pat, while four proteins with CobB. Notably, four proteins including STM14_1074, RhaR, FliT and Lrp are transcriptional regulators, suggesting that acetylation is an important regulation mechanism for the activity of regulators in bacteria (Thao et al., 2010; Hu et al., 2013; Ren et al., 2016).

To confirm the above BTH results, we selected several positive clones, which were encoding regulators (nrdF, rhaR, STM14_1074, fliT, lrp) and cloned their full-length genes into pKT25 individually. Since bacteria expressing interacting hybrid proteins formed blue colonies on LB/X-Gal medium, as shown in Figure 1, each blue co-transformant represented expressing Pat and protein interacted with Pat, suggesting these candidate proteins may exist acetylation modification.

To further validate these regulators can be (de)acetylated, the (de)acetylation modification assays in vitro were performed. The purified proteins were incubated with Pat or CobB, respectively, and the acetylation levels of each protein were evaluated by the pan anti-acetyllysine antibody (α-Acetyl). After Pat treatment at 37°C for 3 h, the acetylation levels of STM14_1074 (Figure 2A), RhaR (Figure 2B), NrdF (Figure 2C), and Lrp (Figure 3A) increased 1.24, 1.77, 1.56, and 1.89-fold, respectively, indicating that NrdF, STM14_1074, RhaR and Lrp can be acetylated by Pat in vitro. However, no acetylation signal increase was detected in FliT after incubation with Pat (Supplementary Figure S1).

Then, the CobB-mediated in vitro deacetylation reaction was performed at 25°C for 6 h, afterward, the samples were applied to detect the acetylation levels. The acetylation levels of STM14_1074 (Figure 2D), RhaR (Figure 2E), and Lrp (Figure 3B) decreased after CobB incubation, indicating these three proteins can be deacetylated by CobB. However, there was no decrease in the acetylation level of NrdF with CobB treatment (Figure 2F).

Although the role of acetylation in Lrp activity has not been studied, a whole-proteome analysis uncovers that Lrp can be acetylated in E. coli, and lysine 25 (K25) and lysine 36 (K36) are the acetylated sites (Baeza et al., 2014). To validate whether these lysine residues are acetylated by Pat, K25 and K36 were site-directed mutated to glutamine (Q) individually mimicking acetylation status of lysine. The acetylation level of K36Q did not increase after Pat-mediated acetylation (Figure 3C). However, the acetylation level of K25Q was elevated with Pat treatment (Figure 3D). These results indicate that K36 is a key acetylation site by Pat, rather than K25.

With sequence alignment, we found Lrp K36 is highly conserved in bacteria (Figure 4A), and this residue was also conserved in several other DNA binding proteins (Figure 4B). In addition, K36 is located in the helix-turn-helix (HTH) DNA-binding domain (Hart et al., 2011). Therefore, we propose that acetylation of K36 may be involved in regulating Lrp activity, specifically the DNA-binding ability of Lrp. To test this hypothesis, K36 was mutated to arginine (R) or glutamine (Q) individually. Whereas the K-to-R substitution avoids acetylation but keeps positive charge, thus mimicking the non-acetylated form, the K-to-Q substitution mimics the constitutively acetylated form through neutralization of the positive charge (Luo et al., 2004). The wild type Lrp, K36Q and K36R proteins were overexpressed and purified to apparent homogeneity (Supplementary Figure S2). Electrophoretic mobility shift assay (EMSA) was performed by incubating the above purified wild type Lrp and its derivatives with 6′-FAM-labeled fimZ promoter, which is bound directly by Lrp (McFarland et al., 2008), followed by non-denaturing PAGE analysis. EMSA results showed that both the wild type Lrp and K36R had the similar and strong DNA-binding affinity to the fimZ promoter, while K36Q possessed a significantly weaker ability to bind the fimZ promoter (Figure 4C). These findings indicate that K36 residue is indispensable for DNA-binding ability of Lrp, and suggest that the acetylation of K36 is associated with DNA-binding ability of Lrp. We also determined the DNA-binding ability of K25 by EMSA, since this residue also is acetylated and located in HTH domain. The results showed that mutation of K25 did not affect the DNA-binding ability of Lrp (Supplementary Figure S3).

The above EMSA results demonstrate that Lrp K36 is an essential site for DNA-binding affinity, which may be regulated by acetylation. In order to confirm the role of K36 acetylation on activity of Lrp, the site-specific acetylation incorporated N^𝜀-acetyllysine system was used (Neumann et al., 2008), which has been applied successfully in acetylation of another transcription factor in bacteria (Ren et al., 2016). The lrp gene carrying a TAG stop codon at position 36 was co-expressed with an orthogonal N^𝜀-acetyllysyl-tRNA synthetase/tRNA_CUA pair cultured in LB medium supplemented with N^𝜀-acetyllysine. K36Ac was purified to apparent homogeneity (Supplementary Figure S4), and the incorporation of acetylated lysine of K36Ac was verified by Western blot with pan anti-acetyllysine antibody. The acetylation level of K36Ac increased significantly compared with the wild type Lrp (Figure 5A). Moreover, the acetyl groups in K36 could be effectively removed by CobB in an NAD⁺ dependent manner (Figure 5B).

Next, K36Ac was used to evaluate the effect of acetylation at K36 on DNA-binding affinity of Lrp by EMSA. The wild type Lrp and K36Ac were incubated with 6′-FAM-labeled fimZ promoter followed by EMSA to compare their DNA-binding abilities. Compared with the non-site-specific wild type Lrp, K36Ac possessed a reduced DNA-binding ability (Figure 5C), similar as K36Q mimicking acetylation (Figure 4C). These results indicate that acetylation of K36 impairs the DNA-binding ability of Lrp.

To test whether the acetylation of K36 is involved in the regulation of Lrp activity in vivo, the chromosome K36Q, K36R, and K36A mutations were constructed individually in S. Typhimurium strain 14028s, denoted as eK36Q (engineered K36Q), eK36R (engineered K36R), and eK36A (engineered K36A), respectively. Simultaneously, the similar genetic manipulation was performed in the wild type lrp and resulted in eWT (engineered WT) as a control. The above genetic manipulations did not affect bacterial growth and transcription of its downstream genes (Supplementary Figure S5). The fim operon of S. Typhimurium encodes type 1 fimbriae and can be directly and positively regulated by Lrp (McFarland et al., 2008; Baek et al., 2011). Therefore, fimZ and fimA were selected as representative genes to determine the effect of acetylation of K36 in vivo. The qPCR assay demonstrated that the transcription levels of fimA and fimZ reduced dramatically in eK36Q and eK36A compared with eWT, while eK36R mimicking non-acetylated lysine residue significantly activated the transcription of these genes (Figure 6A). The lrp gene in S. Typhimurium has been found to be negatively autoregulated (McFarland and Dorman, 2008). As expected, the transcription level of lrp increased in eK36Q and eK36A, but relatively low in eWT and eK36R (Figure 6A). These results suggest that acetylation of K36 inhibits the DNA-binding of Lrp and thus alters the transcription of target genes of Lrp in S. Typhimurium in vivo.

The type 1 fimbriae are mannose-sensitive agglutination factors that mediate bacterial adhesion to a broad range of eukaryotic cells by interactions with mannosylated glycoproteins. The expression of the type 1 fimbriae can be evaluated by the yeast agglutination assay (Kukkonen et al., 1993; Baumler et al., 1997). Since Lrp can positively regulate the expression of the type 1 fimbriae (Blomfield et al., 1993; Baek et al., 2011), we used the yeast agglutination assay to confirm the role of acetylation in regulating activity of Lrp. The chromosome K36Q, K36R, K36A mutants along with the wild type strain were subcultured twice for overnight static growth, respectively. Agglutination was assayed in slide by mixing 5 μl bacterial culture with an equal volume of 10 mg/ml bakers’ yeast suspension. The results showed that eWT and eK36R caused mannose sensitive yeast agglutination, however, both eK36Q and eK36A failed to induce yeast agglutination (Figure 6B). These findings demonstrate that acetylation of Lrp K36 inhibits the production of the type 1 fimbriae by decreasing the binding ability of Lrp with the fim operon promoter.