A new regulatory mechanism for bacterial lipoic acid synthesis

Lipoic acid, an essential enzyme cofactor, is required in three domains of life. In the past 60 years since its discovery, most of the pathway for lipoic acid synthesis and metabolism has been elucidated. However, genetic control of lipoic acid synthesis remains unclear. Here, we report integrative evidence that bacterial cAMP-dependent signaling is linked to lipoic acid synthesis in Shewanella species, the certain of unique marine-borne bacteria with special ability of metal reduction. Physiological requirement of protein lipoylation in γ-proteobacteria including Shewanella oneidensis was detected using Western blotting with rabbit anti-lipoyl protein primary antibody. The two genes (lipB and lipA) encoding lipoic acid synthesis pathway were proved to be organized into an operon lipBA in Shewanella, and the promoter was mapped. Electrophoretic mobility shift assays confirmed that the putative CRP-recognizable site (AAGTGTGATCTATCTTACATTT) binds to cAMP-CRP protein with origins of both Escherichia coli and Shewanella. The native lipBA promoter of Shewanella was fused to a LacZ reporter gene to create a chromosome lipBA-lacZ transcriptional fusion in E. coli and S. oneidensis, allowing us to directly assay its expression level by β-galactosidase activity. As anticipated, the removal of E. coli crp gene gave above fourfold increment of lipBA promoter-driven β-gal expression. The similar scenario was confirmed by both the real-time quantitative PCR and the LacZ transcriptional fusion in the crp mutant of Shewanella. Furthermore, the glucose effect on the lipBA expression of Shewanella was evaluated in the alternative microorganism E. coli. As anticipated, an addition of glucose into media effectively induces the transcriptional level of Shewanella lipBA in that the lowered cAMP level relieves the repression of lipBA by cAMP-CRP complex. Therefore, our finding might represent a first paradigm mechanism for genetic control of bacterial lipoic acid synthesis.


Introduction
Lipoic acid (6,8-dithiooctanoic acid, thioctic acid, or R-5-(1,2-dithiolan-3-yl) pentanoic acid), is a type of two-sulfur inserted eight-carbon fatty acid derivative and acts as a coenzyme widespread in three domains of life (Perham 2000;Cronan et al. 2005). This covalently bound cofactor is required for aerobic metabolism of 2-oxoacids 282 in Escherichia coli and C1 metabolism in plants like Arabidopsis (Perham 2000;Cronan et al. 2005;Engel et al. 2007). In E. coli, the three well-known enzymes whose activities require lipoylation, the post-translational modification, include PDH (pyruvate dehydrogenase), OGDH (2-oxoglutarate dehydrogenase), and GCV (glycine cleavage system) system Hermes and Cronan 2009). All the three enzyme systems possess such subunits (the E2 subunits of both PDH and OGDH, and the H protein of GCV system) that contain no less one lipoyl domains (LD) featuring with a conserved structure of around 80 residues long (Reche 2000;Cronan et al. 2005). Generally, a specific/conserved lysine residue on these LDs is attached by lipoic acid via an amide bond (Perham 2000). Therefore, it seems likely that lipoic acid facilitates shuttle of the activated reaction intermediates amongst the active sites of the lipoate-dependent multienzyme systems (Perham 2000;Cronan et al. 2005).
Most of current knowledge of lipoic acid metabolisms comes from studies with E. coli (Zhao et al. 2003;Cronan et al. 2005). Two alternative strategies have been developed in E. coli to satisfy the trace physiological demand for lipoic acids. It includes de novo biosynthesis pathway and the scavenging route Hermes and Cronan 2009;Rock 2009;Christensen and Cronan 2010). The former pathway is constituted of two consecutive steps: the LipB (octanoyl-ACP: protein N-octanoyl-transferase) transfers the endogenously produced octanoyl moieties from octanoyl-ACP (an intermediate of the fatty acid biosynthesis) to lipoyl domains ( Fig. 1A) (Jordan and Cronan 2003;Zhao et al. 2003Zhao et al. , 2005; in the second step the LipA (lipoyl synthase) uses S-adenosyl-L-methionine (SAM)dependent radical chemistry to insert two sulfur atoms at carbons 6 and ( Fig. 1A) (Zhao et al. 2003;Cronan et al. 2005;Douglas et al. 2006;Christensen and Cronan 2010). The lipoyl protein ligase (LplA) plays a critical role in utilization of exogenous lipoic acids from environments in which the lipoyl-adenylate intermediate is required (Fig. 1A) (Morris et al. , 1995Reed et al. 1994). Although the metabolic mechanism of lipoic aids that was initially discovered in the early of 1940s (Reed 2001) was extensively investigated (Reed 2001;Cronan et al. 2005;Hermes and Cronan 2009;Rock 2009;Christensen and Cronan 2010), its genetic regulation/control is poorly understood (Kaleta et al. 2010;. cAMP receptor protein (CRP, also called catabolic activator protein, CAP) is a type of global regulator, representing a classical model for bacterial gene regulation systems (Green et al. 2014). The paradigm version of CRP is E. coli crp protein product that modulates expression of hundreds of genes involved in a variety of bacterial physiological aspects such as energy metabolism (e.g., galactose catabolism) (Zheng et al. 2004;Green et al. 2014).
Indeed, the activity of CRP requires the presence of its physiological ligand/effector, cyclic AMP (cAMP) (Zheng et al. 2004;Green et al. 2014). Upon the CRP protein is occupied by the cAMP small molecule, it proceeds to an allosteric alteration/structural rearrangement, allowing its acquisition of an ability to specifically bind a collection of specific target DNA sequences (Schultz et al. 1991;Green et al. 2014). As we know, the typical CRP box (cAMP-CRP binding site) is referred to the imperfect palindromic consensus sequence "N 3 TGTGAN 6 TCACAN 3 " (Zheng et al. 2004). In the similarity to the well-studied FadR regulator that has dual functions in fatty acid metabolism (Feng and Cronan 2009a,b, 2010, it appears that the dimeric CRP protein-mediated regulation also can exert two opposite roles, i.e., either activation (Hanamura and Aiba 1992;Ishizuka et al. 1994;Zheng et al. 2004) or repression (Aiba 1983;Hanamura and Aiba 1991;Ishizuka et al. 1994) in response to distinct external and/or internal stimuli/inputs (Green et al. 2014). Recently, comparative genomics-based reconstruction of bacterial regulatory networks RegPrecise (http://regprecise.lbl.gov/ RegPrecise) by Rodionov's research group (Rodionov   Novichkov et al. 2013) predicted that a possible CRP box (AAGTGTGATCTATCTTACATTT) is present in front of lipBA operon (SO1162-S01161) of Shewanella oneidensis MR-1(a marine-borne species of c-proteobacteria family) with considerable potential for the remediation of contaminated environments and application in microbial fuel cells (Fredrickson and Romine 2005;Fredrickson et al. 2008). More importantly, it seemed likely that the predicted site reflects an evolutionally conserved regulatory mechanism in that it is found in nearly all the Shewanella species with known genome sequences and similar scenario were seen even with the two human pathogens Salmonella typhimurium and Klebsiella pneumonia. This might raise a possibility that cAMP signaling is linked to bacterial lipoic acid synthesis in certain species of c-proteobacteria. However, this hypothesis requires further in vitro and in vivo experimental verification.
In this paper, we aimed to resolve this unanswered question. As anticipated, electrophoresis mobility shift assays (EMSA), we conducted and confirmed that the two CRP proteins of E. coli and Shewanella are functionally exchangeable and the predicted CRP sites of Shewanella are functional. Using the chromosome lipBA-lacZ transcriptional fusion in E. coli, we visualize that the removal of E. coli crp gene gave above fourfold increment of lipBA promoter-driven b-gal expression, which is almost identical to the scenario seen with Shewanella. Somewhat it is unexpected, but not without precedent that an addition of glucose into media effectively induces lipBA expression in Shewanella, in that the lowered cAMP level relieves the repression of lipBA by cAMP-CRP complex ( Fig. 1B and C). Therefore, our finding answered the long-term unresolved question in the field of lipoic acid metabolism and might represent a first paradigm illustrating the genetic control of bacterial lipoic acid synthesis by cAMP-dependent CRP signaling in certain species of c-proteobacteria.

Bacterial strains and growth conditions
The bacterial strains used here were derivatives of both E. coli K-12 and S. oneidensis MR-1 (Table 1) and cultivated aerobically at 37°C and 30°C, respectively. For the growth of E. coli, the following three media were utilized, including Luria-Bertani (LB) medium (10 g of tryptone, 5 g of yeast extract and 10 g of NaCl per liter; pH 7.5), rich broth (RB) medium (10 g of tryptone, 1 g of yeast extract, and 5 g of NaCl per liter), and M9 minimal medium with either 5 mmol/L sodium acetate or 0.4% glucose as the sole carbon source Cronan 2009b, 2010). M1-defined medium containing 0.02% (w/v) of vitamin-free casamino acids and 15 mmol/L lactate as electron donor was used to cultivate S. oneidensis (Gao et al. 2008). If required, chemicals or antibiotics were added as follows: 2,6-diaminopimelic acid (DAP), 0.3 mmol/L; sodium ampicillin, 100 lg/mL; kanamycin sulfate, 25 lg/mL; and tetracycline, 15 lg/mL; gentamycin, 15 lg/mL.

Plasmids and DNA manipulations
Using polymersase chain reaction (PCR) with primers crp_she-F plus crp_she-R (Table 2), the S. oneidensis crp gene was amplified and inserted into the BamHI and XhoI sites of pET28a(+) expression vector, giving the recombinant plasmid pET28-crp_she (Table 1). The promoter region of S. oneidensis lipBA (referred to Plip-BA_she) covering the predicted CRP site (Table 3) was amplified with a set of specific primers PlipBA-F plus Pli-pBA-R and cloned into the two cuts SalI and EcoRI of promoter-less plasmid pAH125 to give the recombinant plasmid pAH-PlipBA_she. Consequently, the pAH-Plip-BA_she plasmid was transformed into MC4100 (Dlac), resulting in the LacZ reporter strain FYJ453 with Plip-BA_she-lacZ transcriptional fusion on chromosome ( Table 1). The inserts introduced in the recombinant plasmids we generated were validated by both PCR assays and direct DNA sequencing (Feng and Cronan 2011a,b).
The lipBA promoter activity was assessed using an integrative lacZ reporter system as described recently (Fu et al. 2014). A fragment covering the sequence upstream of the lipB gene from À300 to +1 was amplified and cloned into the reporter vector pHGEI01, verified by sequencing, and the correct plasmid was then transferred into S. oneidensis strains by conjugation. Proper integration of the promoter fusion constructs was confirmed by PCR. To eliminate the antibiotic marker, the helper plasmid pBBR-Cre was transferred into the strains carrying the correct integrated construct. Colonies without the integrated antibiotic marker were screened and verified by PCR, and followed by the loss of pBBR-Cre as described previously ).

In-frame mutant construction and complementation
In-frame deletion strains for S. oneidensis were constructed using the att-based Fusion PCR method as described previously (Jin et al. 2013). In brief, two fragments flanking gene of interest were amplified by PCR, which were linked together by a second round of PCR. The fusion fragments were introduced into plasmid pHGM1.0 by using Gateway BP clonase II enzyme mix (Invitrogen, Grand Island, NY, USA) according to the manufacturer's instruction, resulting in mutagenesis vec-tors in E. coli WM3064, which were subsequently transferred into S. oneidensis MR-1 via conjugation. Integration of the mutational constructs into the chromosome was selected by resistance to gentamycin and confirmed by PCR. The verified transconjugants were grown in LB broth in the absence of NaCl and plated on LB supplemented with 10% sucrose. Gentamycin-sensitive and sucrose-resistant colonies were screened by PCR for deletion of the target gene. Mutants were verified by direct sequencing of the mutated regions.
Plasmids pHG101 and pHG102 were used in genetic complementation of mutants (Wu et al. 2011). For complementation of genes next to their promoter, a fragment containing the gene of interest and its native promoter was generated by PCR and cloned into pHG101. For the remaining genes, the gene of interest was amplified and inserted into MCS of pHG102 under the control of the arcA promoter, which is constitutively active (Gao et al. 2010). The resulting vectors were transferred into its corresponding mutant strain via conjugation and its presence   was confirmed by plasmid purification and restriction enzyme digestion.

RNA isolation and RT-PCR
Mid-log phase cultures of S. oneidensis MR-1 grown in RB media were collected for total bacterial RNA preparations. As we did before, the RNeasy bacterial RNA isolation kit (Qiagen, Hilden, Germany) was adopted (Feng and Cronan 2009b;Feng et al. 2013b). The quality of the acquired RNA samples was visualized using agarose gel The underlined italic letters represent restriction sites, and the bold letters denote the known (and/or predicted) CRP-binding sites. RT-PCR, reverse transcription-polymersase chain reaction; CRP, cAMP-receptor protein. Table 3. CRP binding sites in front of potential lipB/A operons from a variety of species amongst c-proteobacteria.
The position is relative to the translation initiation site. All the information is sampled from the RegPrecise database (http://regprecise.lbl.gov/Reg Precise/search.jsp). electrophoresis. Using the general PCR assay in which the total RNA samples function as templates with primers 16S_she-F plus 16S_she-R (Table 2), the possible contamination of trace genomic DNA in the RNA samples was routinely figured out as we described earlier Cronan 2009b, 2010).
On the basis of above qualified RNA samples, we performed the reverse transcription (RT)-PCR experiments Cronan 2009b, 2010). Briefly, 1 lg of total RNA was mixed with 0.5 lg of random primers (11 lL in total), denatured (70°C for 5 min), and then chilled on ice (5 min). The RT reaction mixture (20 lL total volume) comprised 10 lL of denatured RNA template, 1 lL of random primers, 4 lL of ImProm-II 59 reaction buffer, 2.5 lL of 1 mol/L MgCl2, 1 lL of deoxynucleoside triphosphate mix, 0.5 lL of the recombinant RNasin RNase inhibitor, and1 lL of ImProm-II reverse transcriptase Cronan 2009b, 2011a). The program for RT reaction included the equilibration at 25°C for 5 min, an extension at 42°C for 60 min, and the inactivation of enzyme at 70°C for 15 min. As a result, the cDNA pool (1 lL) was used as the template to PCR-amplify the lipBA operon-related genes/DNA fragments.

Real-time quantitative RT-PCR
On the basis of SYBR Green dye method as we previously mentioned Cronan 2009b, 2010), realtime quantitative RT-PCR (qRT-PCR) experiments were employed to evaluate the altered expression profile of S. oneidensis lipBA operon in the Dcrp mutant. qPCR reaction system (20 lL) contained 12.5 lL of iQ TM SYBR Green Supermix, 1 lL of each primer, 1 lL of the diluted cDNA sample, and 4.5 lL of sterile water. All the data were collected in triplicate on a Mastercycler â eprealplex (Eppendorf, Hauppauge, NY, USA), using the program of a denaturing cycle at 95°C for 15 min, 45 cycles comprising 94°C for 20 sec, 60°C for 20 sec, and 72°C for 20 sec, and a final step featuring with gradient temperature from 60°C to 90°C for dissociating double stranded DNA products. The reference gene was the16S_she rRNA-encoding gene (Table 2) and water acted as blank control to monitor cross-contamination of various cDNA samples. The relative expression levels were calculated with the 2 ÀDDC T method developed by Livak and Schmittgen (2001).  (Feng and Cronan 2011a,b). The nested PCR reactions were established using two sets of combined primers (Outer Primer plus lipBA-GSP and Inner Primer plus lip-BA-Nest primer) ( Table 2). The PCR program was described with a denaturing cycle at 95°C for 5 min followed by 35 cycles comprising 95°C for 30 sec, 55°C for 30 sec, and 72°C for 30 sec. The purified PCR products were sent for direct DNA sequencing. The transcriptional start site was assigned to first nucleotide adjacent to the RLM-RACE adaptor (Feng and Cronan 2009a,b;Feng et al. 2013b).

Enzymatic assays
For PDH assay, cells were grown at 30°C in 25 mL of LB containing the appropriate antibiotics until the beginning of the stationary phase, harvested, and washed twice with a 0.04 mol/L potassium phosphate buffer (pH 7.5). The resulting pellets were frozen rapidly and stored at À80°C. Cell extracts were prepared by resuspending the thawed pellets in 2 mL of the same buffer prior to sonication with a microtip in a Branson model 200 Sonifier (2 min total, with 40-sec pulses at 20-sec intervals). Cell debris was removed by centrifugation (10 min at 12,000g and 4°C), and the supernatants were used for assays at 25°C as described previously (Reed and Cronan 1993). Protein concentrations were determined by the bicinchoninic acid protein assay reagent (Pierce Chemical Co., Rockford, IL, USA).
To measure the b-galactosidase activities in E. coli, bacterial lysates from mid-log phase cultures grown in LB (or M9) media were prepared by treatment with sodium dodecyl sulfate-chloroform (Miller 1972;Feng and Cronan 2009b). Similarly, cells of S. oneidensis (mid-log phase under experimental settings) were pelleted for assaying its b-galactosidase activity with an assay kit as described previously (Wu et al. 2011).

Measurement of intracellular cAMP levels
Cells in mid-log phase cultures (~0.3 of OD 600 ) were collected by centrifugation and washed twice with charcoaltreated phosphate-buffered saline (PBS; pH 7.0). Both supernant and pellet fractions were applied to the cAMP assay using Cyclic AMP EIA kit (Cayman Chemical Co., Ann Arbor, Michigan, USA) according to the manufacturer's instruction.

Expression, purification and identification of two CRP proteins
To prepare the recombinant CRP protein in two versions (CRP_ec and CRP_she), the engineered E. coli strains carrying either pET28-crp_ec or pET28-crp_she (Table 1) were induced with 0.3 mmol/L isopropyl b-D-1-thiogalactopyranoside (IPTG) at 30°C for 5 h (Feng and Cronan 2012). Following bacterial lysis by a French pressure cell, the clarified supernatants by centrifugation (30,966g, 30 min) were loaded onto a nickel chelate column (Qiagen). After removal of the contaminated protein by washing with 19phosphate buffered saline (PBS) with 50 mmol/L imidazole, the interested CRP proteins (CRP_ec or CRP_she) were eluted using elution buffer containing 150 mmol/L imidazole. Finally, the protein was concentrated by ultrafiltration (30 kDa cut-off) and exchanged into 19 PBS (pH 7.4) containing 10% glycerol. The purity of the recombinant CRP proteins was judged by 12% sodiumdodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) Cronan 2009b, 2011b). To verify the identity of the acquired proteins, the de-stained (SDS-PAGE) gel slices were subjected to liquid chromatography quadrupole time-of-flight mass spectrometry using a Waters Q-Tof API-US Quad-ToF mass spectrometer linked to a Waters nanoAcquity UPLC (Feng and Cronan 2011a;Feng et al. 2013a,b).

Electrophoretic mobility shift assays
The function of the predicted CRP-binding site of Shewanella lipBA operon was assessed in vitro using electrophoretic mobility shift assays (EMSA) with little improvements (Feng and Cronan 2011a;Goble et al. 2013;). In the EMSA tests, nine pieces of DNA probes were composed of seven suspected probes (lipBA_she, ybeD_ec, ybeD_es, ybeD_kp, ybeD_st1, ybeD_st2, and ybeD_yp) and the two control probes, the fadD_ec site with known function (the positive control) and the lipA_ec without any function (the negative control) ( Table 3). The digoxigenin (DIG)-labeled DNA probes were prepared in vitro through annealing two complementary oligonucleotides in TEN buffer (10 mmol/L Tris-HCl, 1 mmol/L EDTA, 100 mmol/L NaCl; pH 8.0) and then labeled by the terminal transferase with DIG-ddUTP (Roche, Indianapolis, IN, USA) ).
In the presence/absence of cAMP (20 pmol), the various DIG-labeled DNA probes (0.2 pmol) were incubated with or without CRP protein in the binding buffer (Roche) at room temperature for around 20 min. Following the separation of the DNA-protein complexes with a native 7% PAGE gel, the chemiluminescent signals were further captured by the exposure to ECL film (GE Healthcare, Piscataway, NJ, USA) Cronan 2011b, 2012).

Bioinformatic analyses
The alignments of DNA (and/or protein) sequences were conducted using the ClustalW2 program (http://www.ebi.ac.uk/Tools/clustalw2/index.html) and final output was processed by the ESPript 2.2 server (http://espript.ibcp.fr/ ESPript/cgi-bin/ESPript.cgi). The lipBA regulons and the possible CRP-recognizable sites of c-proteobacteria were collected from the RegPrecise database (Novichkov et al. 2010a) and were analyzed (Feng et al. 2013a) using Reg-Predict software (Novichkov et al. 2010b). The sequence logo for the CRP consensus palindrome was generated by WebLogo (http://weblogo.berkeley.edu/logo.cgi). The software of SPDBV_4.01 (http://spdbv.vital-it.ch/) was used for structure modeling. Figure 2. Gemomic context of the lipB/A operon/genes in the selected c-proteobacteria. Blue arrows represent the lipA genes that encode the lipoic acid synthase catalyzing the last committed reaction of lipoic acid biosynthesis pathway, whereas green arrows indicate the octanoyl-protein ligase-encoding genes (lipB).The gray arrow upstream of the lipB gene refers to the ybeD gene of unknown function. In some cases, the tatE gene (Sec-independent protein translocase) downstream of lipA is shown with yellow arrow. In the four species (Escherichia coli, Enterobacter sp. 638, Klebsiella pneumoniae, and Salmonella typhimurium LT2), the ybeF gene encoding an LysR-type transcription factor (in orange) is located between lipA and lipB. The predicted CRP-binding palindromes are highlighted with dots (red dots represent the experimentally verified sites, whereas the gray ones are not experimentally validated). CRP, cAMP-receptor protein.

Shewanella lipBA is an operon
The paradigm pathway of lipoic acid synthesis is encoded by two genes lipB and lipA of E. coli. The two sequential steps of this pathway included LipB-catalyzed transfer of octanoyl moiety from octanoyl-ACP to lipoyl domains of the cognate enzymes and LipA-mediated insertion of sulfur atoms at C6 and C8 of LD-bound octanoyl moiety to give lipoate (Fig. 1). Therefore, we are interested in examining the genetic context of the lipB and/or lipA in c-proteobacteria using RegPredict software (Novichkov et al. 2010b) (Fig. 2). In addition to the reference strains (e.g., E. coli, Salmonella enterica, Yersinia pestitis, etc.), all the other samples are focused on Shewanella species from the RegPrecise database (Novichkov et al. 2010a). We noted that the ybeD (SO1163) gene is constantly present upstream of the lipB gene (Fig. 2), and YbeD protein of E. coli origin exhibits a striking structural homology to the allosteric regulatory domain of D-3-phosphoglycerate dehydrogenase (Kozlov et al. 2004).
Unlike the scenario seen with E. coli that lipB and lipA are separated by a gene (ybeF) encoding a LysR-family transcription factor of unknown function (Feng and Cronan 2014) (Fig. 2), it seemed likely that lipB and lipA constitutes an operon in most of species of Shewanella (Fig. 1). Although physiological advantages for the cotranscription of these two genes are expected, experimen-tal evidence is lacking. To address this hypothesis, the strain of S. oneidensis MR-1 was selected for our experiments. We established the combined PCR and RT-PCR assays using five pairs of specific primer pairs (Table 2 and Fig. 3A). The positive amplifications (1, 3 and 5) were obtained by both PCR and RT-PCR showed that all three genes (ybeD, lipB and lipA) are transcribed (Fig. 3A). The fact that the primed amplicon (designated to 2) was observed only by PCR, but not by RT-PCR suggested that ybeD is not co-transcribed together with lipB (Fig. 3A). As anticipated, the designed amplicon covering both lipB and lipA was positive in both PCR and RT-PCR assays, validating that lipB and lipA act as an operon (transcriptional unit) (Fig. 3A).

S. oneidensis lipBA promoter
DNA sequences recognized by CRP proteins of E. coli and S. oneidensis are predicted to be similar and the interaction depends on cAMP (Gao et al. 2010;Fu et al. 2013;Zhou et al. 2013). Given the fact that a predicted CRP-binding site (AAGTGTGATCTATCTTACATTT) is located in the intergenic region between the ybeD gene and the lipBA operon of S. oneidensis (Fig. 2), we thus mapped the promoter by employing an improved method of 5 0 -RACE (RLM-RACE). As a result, we acquired the 5 0 -RACE products of approximately 450 bp in length (Fig. 3B). The result of the direct DNA sequencing showed the 5 0 -end of the S. oneidensis lipBA transcript (i.e., transcription start site, A) is located 20 nucleotides upstream its translation initiation codon TTG ( Fig. 4C and D). Apparently, the assumed CRP-recognizable site appears to be 25 bp upstream of the transcription start site (Fig. 4D). Furthermore, the multiple sequence alignment clearly indicated that the CRP binding sites of different origins are extremely conserved, in that 16 of 22 nucleotides are identical at least (if not all) in the examined Shewanella species (Fig.  6A and B). However, the function of these putative sites needs further experimental validation.

Physiological requirement of protein lipoylation
It is reasonable that co-expression of LipB octanoyltransferase and LipA lipoate synthase assures the economical production of lipoic acid (an energy-expansive molecule) to effectively satisfy the metabolic/physiological requirement of protein lipoylation in organisms. Given the fact that both PDH and OGDH are proceeded such kind of post-translational modification, we thereby developed the anti-LA Western blot to detect this metabolic requirement in four Ɣ-protebacteria species (E. coli, S. enterica, V. cholerae, and S. oneidensis). As expected, we did observe 290 that lipoylation occurs in PDH and OGDH of Shewanella, which is in much similarity to the scenario seen with E. coli (Fig. 5A). Because lipoylation is essential for the function of all characterized PDH and OGDH proteins, the result points out the metabolic significance of this common enzyme cofactor in S. oneidensis. Subsequently, we constructed a lipBA null mutant from the S. oneidensis wild-type strain. The mutant was unable to grow on minimal medium unless lipoic acid was supplemented (Fig. 5B), a phenotype observed from E. coli lip mutants (Reed and Cronan 1993). Additionally, the PDH assay revealed that this ΔlipBA strain contained no detectable dehydrogenase activities (Fig. 5C). Importantly, the phenotypes resulting from the lipBA deletion were restored by their expression in trans, indicating that they are due to the intended mutation. These data, collectively, conclude that the lipBA genes are the only enzyme accountable for protein lipoylation in S. oneidensis.

Characterization of S. oneidensis CRP protein
S. oneidensis CRP and its counterpart of E. coli are highly homologous (Fig. S1A), and have been shown to be func- The multiple alignments were conducted using ClustalW2 (http://www.ebi.ac.uk/Tools/clustalw2/index.html), and the resultant output was processed by program ESPript 2.2 (http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi) Cronan 2009b, 2010;Feng et al. 2013a). Identical residues are indicated with white letters on a red background, similar residues are red letters on yellow, varied residues are in black letters, and dots represent missing residues. S, transcription start site; M, translational initiation site. The predicted CRP-recognizable palindrome is underlined. CRP, cAMP-receptor protein. tionally equivalent/exchangeable in vivo (Saffarini et al. 2003). However, whether this is the case in vitro remains undefined. In addition to the E. coli CRP protein, an N-terminal hexahistidine fused S. oneidensis CRP protein was over-expressed, purified to homogeneity and gave a single protein band with an estimated molecular mass (~24 kDa) (Fig. S2B). The tertiary structure of S. oneidensis CRP protein was modeled using SPDBV_4.01 software, which is highly similar to that of E. coli (Fig. S3C). Liquid chromatography mass spectrometry analyses of tryptic peptides of the recombinant CRP protein band excised from an SDS-PAGE gel validated its identity in that the peptides matched S. oneidensis CRP with 70% coverage of the expected peptides (Fig. S4D). The two versions of CRP proteins we prepared were subsequently used for functional analyses of the above predicted CRP-specific palindromic sites.

Shewanella lipBA binds the cAMP-CRP complex
To test the activity of the DNA probe derived from the S. oneidensis lipBA promoter ( Fig. 6A and B), we conducted EMSA assays. First, the positive control fadD_ec probe with a known function (Feng and Cronan 2012) binds well to E. coli CRP protein in the presence of cAMP effector molecule, whereas the negative control lipA_ec probe with a nonfunctional CRP site did not (Fig. 6C). As expected, the lipBA_she probe exhibited the appreciably comparable activity of binding cAMP-CRP complex relative to the positive control. Apparently, our result is much consistent with previous observations with the CRP regulatory protein in the context of other metabolisms (Gao et al. 2010;Fu et al. 2013;Zhou et al. 2013), proving the prediction of Novichkov et al. (2013) is correct. Additionally, the specific binding of lipBA_she to cAMP-CRP complex seemed to be in a protein dosedependent manner (Fig. 6D).Not only does the CRP protein of E. coli origin interacts with E. coli fadD probe (Figs. 6C and 7A) and Shewanella lipBA probe (Figs. 6C and 7C), but also the CRP protein encoded by Shewanella binds to E. coli fadD probe (Fig. 7B) and Shewanella lipBA probe (Fig. 7D). It thus fully demonstrated that the two versions of CRP protein are functionally exchangeable in vitro.
Similarly, we also tested a series of predicted CRPbinding sites located upstream of ybeD-lipB loci (Fig. 2 and Table 3) using EMSA tests with E. coli CRP protein.
Unlike the lipBA_she probe (Fig. S2A), neither the E. coli ybeD probe (ybeD_ec, Fig. S2B) nor the Y. pestis ybeD probe (ybeD_yp, Fig. S2G) are functional for the cAMP-CRP complex. By contrast, the prediction in CRP-recognizable sites (ybeD_es and ybeD_kp) in front of the ybeD gene of both Enterobacter sp. 638 and Klebsiella pneumo- nia are correct in that both bind to the CRP protein ( Fig.  S2C and D). Of particular note, among the two CRP sites (ybeD_st1 and ybeD_st2) proposed for S. enteric ybeD gene, only the ybeD_st1 site is functional (Fig. S2E), while the other one was not (Fig. S2F).  Table 2). The plus sign represents addition of the CRP protein and/or cAMP, whereas the minus sign denotes no addition of the CRP protein and/or cAMP. Designations: ec, E. coli; she, Shewanella. (D) Dose-dependent binding of E. coli CRP binds to Shewanella lipBA promoter. The level of CRP protein in (A) is 2 pmol, and the amount of cAMP is 20 pmol. The protein samples were incubated with 0.2 pmol of DIG-labeled lipBA_she probe (43 bp) in a total volume of 20 lL. A representative result from three independent gel shift assays (7% native PAGE) is given. CRP, cAMP-receptor protein.
in vivo regulatory role of cAMP-CRP complex in expression of S. oneidensis lipBA operon encoding lipoic acid synthesis machinery. First, S. oneidensis lipBA promoter was fused to a LacZ reporter gene to allow direct assaying b-gal activity of lipBA_she-lacZ transcriptional fusion integrated into E. coli chromosome (Table 1). In light of the functional equivalence of Shewanella CRP to E. coli CRP, we firstly compared the alteration of b-gal activity in the model organism E. coli (Dcrp mutant and its parental strain of E. coli). As anticipated, MacConkey plates-based experiments visualized that the lipBA_she promoter-driven b-gal activity is appreciably stronger (illustrated with purple) in the Dcrp mutant than that of the wild type E. coli (low activity denoted by yellow) (Fig. 8A). Direct measurement of LacZ activity revealed that deletion of crp gene gave three-to fourfold increment of lipBA_she transcription level (Fig. 8B). A similar lacZ reporter construct was also integrated into the chromosome of S. oneidensis wild-type and its Dcrp mutant strains (Fig. 8D and E) (Fu et al. 2014). Consequently, the significant alteration/improvement of lip-BA_she-lacZ expression level was detected upon the removal of the crp gene from S. oneidensis ( Fig. 8D and E). Second, the real-time qPCR-based analyses of transcriptional profile showed that no less threefold increment of lipA and/or lipB expression was observed in the Dcrp mutant of S. oneidensis in relative to the wild type strain (Fig. 8C). Of particular note, repression of S. oneidensis lipBA expression by CRP depends on production of cyclic AMP ( Fig. 8E and F). Given the above combined in vitro and in vivo data, we concluded that the cAMP-CRP complex is a repressor for lipBA expression in S. oneidensis.
Glucose improves the expression of S. oneidensis lipBA in the alternative model microorganism E. coli It is well known that an addition of glucose into media can lower the level of cytosolic cAMP in E. coli, which might in turn impair at least partially CRP-mediated regulation.
Somewhat it is unusual that not all the species of Shewanel- la genus can utilize/metabolize glucose in that the S. oneidensis glucose transporter-encoding gene glcP is a pseudogene with a frame-shift Rodionov et al. 2010). Given the above technical problem, we therefore attempted to examine the so-called "glucose effect" with the engineered E. coli strain FYJ457 carrying the lip- BA_she-lacZ transcriptional fusion (Fig. S3). As expected, we observed that the level of lipBA expression was induced by the addition of glucose (5 mmol/L) to about threefold higher than that grown in the M9 minimal media with acetate (5 mmol/L) as the sole carbon source (Fig. S3). Together, we proposed for the first time that the global regulator, the cAMP-CRP complex represses bacterial lipoic acid synthesis in Shewanella, posing the relevance of the cAMP signaling to the production of the sulfurcontaining C8 enzyme cofactor, lipoic acid (Fig. 1A). This regulatory network can respond to the statue of glucose/ cAMP level, i.e., the low glucose/high cAMP level shuts down lipBA expression (Fig. 1B), whereas the high glucose/low cAMP level de-represses lipBA transcription (Fig. 1C).

Discussion
Biotin and lipoic acid both are sulfur-containing fatty acid derivatives and act as enzyme cofactors required for central metabolism in the three domains of life. Unlike the fact that the regulation of bacterial biotin metabolism has been extensively investigated, the knowledge about genetic control of lipoic acid synthesis remains missing or lagged. Although a recent bioinformatics-based proposal was raised, that is, the PDH repressor involved in E. coli lipoic acid synthesis (Kaleta et al. 2010;Gohler et al. 2011), this prediction was not validated by the physiological evidence in vivo . Through revisiting this long-term unanswered issue, the data shown here might establish for the first time the link of cAMP signaling to bacterial lipoic acid metabolism (Fig. 1). Somewhat this finding can resolve the puzzle/discrepancy. As we knew that cAMP-dependent CRP regulatory system is a common global regulator involved in a variety of physiological processes (such as sugar metabolism) in most of the bacterial species, our finding extended this regulatory network into the field of vitamin synthesis.
As an important second messenger, the pool of cAMP molecule is at least determined by the following three factors: First, The activity of cyclic adenylate cyclase (cyaA) is responsible for the formation of cAMP molecules (Of note, 90% of cAMP that is made by intracellular adenylyl cyclases) (Pastan and Perlman 1970;Hantke et al. 2011); Second, the cAMP phosphodiesterase (CpdA) has the opposite enzymatic activity to break a phosphodiester bond of cAMP (Imamura et al. 1996). Not only is the production of CyaA regulated by the CRP regulator at the transcriptional level (Aiba 1985;Qu et al. 2013;You et al. 2013), but was also controlled at the metabolic level by the phosphorylation of EIIA Glc , a core component of PTS system (Crasnier-Mednansky 2008;Deutscher 2008;Gorke and Stulke 2008;Narang 2009). Third, the TolC pump is found to export cAMP outside of E. coli cells in maintaining its sensitivity in the changing metabolic environment (Hantke et al. 2011). Given the fact that all the three homologs of cyaA (SO_4312), cpdA (SO_3901), and tolC (SO_3904) are encoded in Shewanella genomes, it might raise the possibility of unexpected complexity in the linking of the second messenger cAMP signaling to lipoic acid synthesis in the context of Shewanella physiology. More interestingly, we recently discovered a novel enzyme, cAMP deaminase (referred to CadD) from the human pathogen Leptospira interrogans (Goble et al. 2013), and established a new mechanism for quenching the cAMP-dependent signaling. In light that we failed to search a CadD-like homolog, thereby we are not quite sure whether it might be implicated into bacterial lipoic acid synthesis or not yet.
The fact that two critical genes of lipoic acid synthesis lipB and lipA are organized into an operon (Figs. 2 and  3), is somehow what we expected, in that it acts as a physiological advantage for economic and effective production of this enzyme cofactor. It is well-known that the genus Shewanella (belonging to the c-proteobacteria) inhabited in energy-rich, redox-fluctuating environments and in turn evolved to possess diverse metabolic capabilities, e.g., coupling the turnover of organic matter with anaerobic respiration of different electron acceptors (Fredrickson et al. 2008). To our surprise, an appreciably conserved CRP-binding site is constantly present in the promoter regions of lipBA operon from nearly all the Shewanella species with sequenced genomes (Fig. 2). It is of no doubt that a common and novel regulatory mechanism for lipoic acid biosynthesis is present in the genus Shewanella.
What kind of selective pressure or evolution consequence does it reflect in adaption to its unique environmental niche? In the paradigm microorganism E. coli, CRP is a global regulator for catabolite repression (Aiba 1983;Schultz et al. 1991;Chandler 1992). However, the protein in S. oneidensis was initially characterized as a principal regulator controlling anaerobic respiration of many electron acceptors (Saffarini et al. 2003). In recent years, it has been repeatedly shown that the regulator in fact plays a more comprehensive role in the physiology, covering both aerobic and anaerobic respiration (Gao et al. 2010;Dong et al. 2012;Fu et al. 2013Fu et al. , 2014Zhou et al. 2013). In contrast to the pathway-specific regulators BirA (Beckett 2007) and BioR (Feng et al. 2013a,b), both of which negotiate production of the other enzyme cofactor biotin, we believed that Shewanella genus have evolved an unknown strategy to share the cAMP-dependent CRP regulatory architecture with other biological processes to efficiently control lipoic acid synthesis. Given the fact that glucose can induce lipBA expressions (Fig. 1B), together with the above information, we concluded that the logic for this kind of regulation does make sense. The reasons are described as follows: (1) the anaerobic growth environment preferred by Shewanella determines an entry of glucose into the glycolytic pathway, giving two pyruvate molecules each glucose; (2) in the Krebs cycle, the resulting pyruvate is catalyzed by PDH to give acetyl-CoA; (3) the full activity of PDH requires the lipoylation, a posttranslational modification of protein (which is validated by the scenario seen in the Anti-LA Western blot, i.e., PDH is the prevalent protein form relative to OGDH, Figure 5); (4) the protein lipoylation depends on the availability of lipoic acids; (5) de novo LipB-LipA synthesis pathway is necessary to be turned on in addition to the LplA-mediated scavenging route of lipoic acid; (6) derepression of lipBA expression might facilitate meeting the physiological requirement for lipoic acid production in such situation (vice versa, Fig. 1C).
Of particular note, we also detected functional CRPbinding sites ahead of ybeD with unkown function in limited species such as human pathogen S. enterica (Figs. 2 and S2). Although that lipB gene is adjacent to ybeD (of note, we lacked evidence proving if they are co-transcribed or not), it required further experimental evidence for CRP regulate lipB or not in this case. It is of interest to test this hypothesis. In fact, it has already been being our research direction in aiming to answer/pursue this question. To the best of our knowledge, our findings reveal, for the first time, a new molecular mechanism for genetic control of bacterial lipoic acid synthesis.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Figure S1. Characterization of Shewanella CRP protein.
A. Sequence comparison of CRP proteins from three different organisms. As we described in Figures 2 and 4, the multiple alignments of CRP proteins were carried out using ClustalW2 (http://www.ebi.ac.uk/Tools/clustalw2/ index.html). Identical residues are in white letters with red background, similar residues are in black letters with yellow background, varied residues arein black letters, and dots represent gaps. The predicted secondary structure was shown in top. a: a-helix; b: b-sheet; T: b-turns/coils. The three organisms used here are E. coli, V. cholerae, and S. oneidensis, respectively. (B) SDS-PAGE profile of the purified Shewanella CRP protein. The protein sample was separated with 4-20% gradient Mini-PROTEAN@ TGXTM Gel (Bio-Rad).The monomeric CRP protein with the estimated molecular weight of~24 kDa is indicated with an arrow. (C) Modeled structure of Shewanella CRP protein. Structure modeling was proceeded by the software of SPDBV_4.01 using E. coli CRP regulator with known structure (PDB: 2WC2) as structural template. N: N-terminus, C: C-terminus. (D) MS identification the recombinant Shewanella CRP protein. The peptide fragments that match Shewanella CRP protein are highlighted in bold and underlined type (70% coverage in total). The predicted CRP site in front of E. coli ybeD-lipB-ybeF-lipA operon (ybeD_ec) has no ability to bind to the CRP protein. The putative CRP sites of the ybeD-lipB-ybeF-lipA operon from Enterobacter sp. 638 (ybeD_es, C) and Klebsiella pneumonia (ybeD_kp, D) are functional. The predicted CRP site 1 of Salmonella enteric ybeD-lipB-ybeF-lipA operon is functional (E), whereas the site 2 is inactive (F). (G) No binding of the cAMP-CRP complex to the suspected CRP site in front of the ybeD-lipB-lipA operon of Yersinia pestis. All the EMSA experiments (7% native PAGE) were conducted as we described (Feng and Cronan 2012;Feng et al. 2013a) with a minor change. The level of cAMP added is 20 pmol. The E. coli CRP protein samples in various concentrations were incubated with 0.2 pmol of DIG-labeled probe in a total volume of 15 µL. A representative result is given. The sequences of all the DNA probes used here are listed in Tables 2 and 3. The minus sign denotes no addition of the CRP protein and/or cAMP molecule. Designations: she, Shewanella; ec, E. coli; es, Enterobacter sp. 638; kp, Klebsiella pneumonia; st, Salmonella typhimurium LT2, and yp,Yersinia pestis. Figure S3. Induction of Shewanella lipBA expression by glucose in the alternative model E. coli. To test the effect of glucose on Shewanella lipBA expression, the E. coli strain carrying the lipBA_she-lacZ transcriptional fusion (FYJ457) was used here. Mid-log phase cultures in M9 media with acetate and/or glucose (5 mmol/L) as sole carbon source were sampled for assaying b-gal activity. The data from more than three independent experiments is expressed in average AE standard deviation (SD), and error bars indicate SD.