Functional definition of BirA suggests a biotin utilization pathway in the zoonotic pathogen Streptococcus suis

Biotin protein ligase is universal in three domains of life. The paradigm version of BPL is the Escherichia coli BirA that is also a repressor for the biotin biosynthesis pathway. Streptococcus suis, a leading bacterial agent for swine diseases, seems to be an increasingly-important opportunistic human pathogen. Unlike the scenario in E. coli, S. suis lacks the de novo biotin biosynthesis pathway. In contrast, it retains a bioY, a biotin transporter-encoding gene, indicating an alternative survival strategy for S. suis to scavenge biotin from its inhabiting niche. Here we report functional definition of S. suis birA homologue. The in vivo functions of the birA paralogue with only 23.6% identity to the counterpart of E. coli, was judged by its ability to complement the conditional lethal mutants of E. coli birA. The recombinant BirA protein of S. suis was overexpressed in E. coli, purified to homogeneity and verified with MS. Both cellulose TLC and MALDI-TOFF-MS assays demonstrated that the S. suis BirA protein catalyzed the biotinylation reaction of its acceptor biotin carboxyl carrier protein. EMSA assays confirmed binding of the bioY gene to the S. suis BirA. The data defined the first example of the bifunctional BirA ligase/repressor in Streptococcus.

i.e., the AccB subunit of acetyl-CoA carboxylase (ACC)) that catalyzes the first committed reaction for type II fatty acid synthesis pathway 18 . Biotin protein ligase (BPL) is widespread in three domains of the life in that it transfers/attaches the biotin cofactor to the specific domain of the relevant subunits of key enzymes from the certain central metabolisms 19,20 . Most of bacteria including E. coli 14,18 and Bacillus 21 only encode a single BPL to account for such kind of physiological requirement, while the pathogen Fracisella novicida developed an additional BPL to gain the competitive advantage in the infected host environment 2 . In general, the BPL members are categorized into the following two groups (Group I and Group II) that can be easily distinguished by the presence of N-terminal DNA-binding domain that allow the BirA protein to bind the cognate genes (e.g., bio operon), and thereafter inhibit expression of biotin metabolism 19,21 . Unlike the paradigm Group II BPL proteins, the E. coli birA gene product retaining the DNA-binding activity, the Group I BPL that lacks the N-terminal helix-turn-helix domain solely function as an enzyme responsible for protein biotinylation 6,9 . In particular, the regulatory role of the Group II BPL depends on the participation of the physiological ligand/effector (biotinoyl-5′ -AMP), the product of the first ligase half reaction for biotin utilization/protein biotinylation 12 .
Streptococcus suis (S. suis) is a leading agent of bacterial diseases including meningitis, arthritis and septicemia) in swine industry worldwide 22,23 , and also appears to be an opportunistic zoonotic pathogen responsible for human S. suis infections such as meningitis and even streptococcal toxic shock-like syndrome (STSS) 24 . Given the difference of bacterial capsule, 35 kinds of serotypes (1-34, 1/2) have been attributed to S. suis 25 . Among them, the serotype 2 of S. suis is frequently isolated from diseased piglets and highly-related to the strong virulence 22,23,25 . S. suis 2 (SS2) is a previously-neglected, but newly-emerging human pathogen, claiming a series of occupational/ opportunistic infections. Since the first discovery of human SS2 meningitis was recoded in Denmark, in 1968 25 , SS2 has spread to nearly 30 countries (and/or regions) and caused no less than 1,600 human cases 26 . In particular, two big-scale outbreaks of fatal human SS2 infections had ever emerged in China (one in Jiangsu Province, 1998, and the other in Sichuan Province, 2005), posing a great concern to public health 24,27,28 . In 2007, we also reported three sporadic cases of human SS2 meningitis in China (two cases in Shenzhen City, and one case in Chongqing City) 29 , implying the co-existence of outbreaks and sporadic cases in China 24,26,28,29 . It is unusual that the epidemic strain of Chinese virulent SS2 harbors a pathogenicity island (PAI) referred to 89 K 30,31 . Subsequent functional exploration suggested that this 89 K PAI behaves as a transposon-like genetic element 32 and encodes type IV secretion system 33 and SezAT toxin-antitoxin system 34 . Our epidemiological investigation argued that the 89 K PAI might be undergoing unknown selective pressure in that some variants losing 89 K PAI are emerging 35 . The remodeling of bacterial surface structure significantly was found to attenuate full virulence of the epidemic SS2 strain 36 . Very recently, we observed that regulation of the D-galactosamine (GalN)/N-acetyl-D-galactosamine (GalNAc) catabolism pathway is linked to its infectivity 37 . It seems likely that the regulatory network of bacterial metabolism is complicated into SS2 virulence 26,38 . Given the fact that biotin metabolism and utilization is associated with Francisella pathogenesis 2,39,40 , it is much interest to define the biotin utilization pathway in the human pathogen S. suis 2.
In this paper, the epidemic SS2 strain in China, S. suis 05ZYH33 with the known genome sequence was subjected to the context analyses of the bacterial biotin metabolism and its possible regulation. Unlike the scenarios seen with the paradigm organism E. coli, the possible biotin machinery in the S. suis we detected comprises a single BioY (SSU05_0509) transporter regulated by the BirA bifunctional protein (05SSU_1625) and the biotin-requiring substrate protein AccB (SSU05_1801). By employing integrative approaches that ranged from comparative genomics, bioinformatics, biochemistry/biophysics, metabolomics, to bacterial genetics, we attempted to present a full picture of biotin utilization pathway in the zoonotic pathogen S. suis.

Results and Discussion
S. suis BirA Protein is a Group II BPL Member. It seems likely that S. suis is biotin auxotrophic in that it is deficient in biotin synthesis, and depends on the mechanism of BioY-BirA to scavenge biotin from the inhabiting niche and/or infected host environment (Fig. 1). System biology by Rodionov et al. 41 revealed that bi-functional BPL enzymes (Group II) exemplified with the E. coli BirA, are universal in both Eubacteria and Archaea, implying the group II form might be the ancestor of the BPL. Unlike the group I without DNA-binding domain BPL (e.g., BirA orthologues of Agrobacterium 4 and Brucella 11 ), the multiple sequence alignment analyses suggested that S. suis BirA is generally similar to the paradigm E. coli birA product (Fig. 2). However the situation seemed unusual in the close-relative of S. suis, Lactococcus lactis in that this probiotic bacterium contained two versions of BPL, one of which refers to BirA1 _LL (Group II BPL) and the other is BirA2 _LL lacking the DNAinteracting motif (Group I BPL) (Fig. 2). To address the BPL biochemistry of the S. suis BirA, we applied protein engineering to produce the recombinant protein. As anticipated, we harvested the BirA_ss protein with the mass of around 37 kDa (Fig. 3A). Also, the purity was judged with SDS-PAGE (Fig. 3A). To further assure the identity, the polypeptide fragments digested from the recombinant BirA protein were subjected to the analyses of MALDI-TOFF. The MS result suggested that the recombinant protein matched well the native form of the S. suis BirA in that it exhibited the coverage of 54% (Fig. 3B). Structural modelling assigned the S. suis BirA as a typical version of the group II BPL with the perfect architecture (Fig. 3C). It comprised the following three functional motifs: N-terminal DNA-binding domain, Central domain and C-terminal domain (Fig. 3C). Obviously, the above data defined the S. suis BirA as a member of Group II BPL.

Activity of Biotin Protein Ligase of S. suis
BirA. We employed the in vitro and in vivo approaches to address the BPL activity of S. suis BirA. The two E. coli birA mutants used for functional complementation included the temperature-sensitive mutant of birAts (BM4062) and the birA1 Km mutant (BM4092). As expected, the BM4062 Strain with/without the empty vector pBAD24 cannot grow on the M9 agar plates under the non-permissive temperature of 42 °C (Fig. 4A). In contrast, the arabinose-induced expression (and even basal expression) of the plasmid-borne birA_ss supported the growth of the birAts mutant BM4062 at 42 °C (Fig. 4A).
Scientific RepoRts | 6:26479 | DOI: 10.1038/srep26479 The measurement of bacterial growth curves for BM4062 strains in liquid media also reproduced the similar results to those obtained from the agar plates (Fig. 4B). The presence of pBAD24 plasmid-borne birA_ss allowed the birA1 Km mutant of E. coli, BM4092 to grow on the minimal media supplemented with low level of biotin BioY was illustrated with modeled ribbon structure (in purple) and integrated into the scheme of bacterial membrane. The biotin transported from environment was activated into biotinoyl-5′ -AMP and then transferred to BCCP acceptor protein giving the biotinoyl-5′ -BCCP. Sulphur was labeled in orange, and AMP (and/or ATP/ PPi) was highlighted in red. The biotin acceptor protein BCCP was indicated with a blue rectangle. The multiple sequence alignment of BirA protein was performed using the program of Clustal Omega, an updated version of ClustalW2 (http://www.ebi.ac.uk/Tools/msa/clustalo), and the final output was given with the program ESPript 3.0 (http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi). Identical residues are white letters with red background, similar residues are red letters in white background, the varied residues are in grey letters, and gaps are denoted with dots. The protein secondary structure was shown in cartoon (on top), α : α -helix; β : β -sheet; T: Turn; η : coil.
(25 nM) (Fig. 4C), which agreed well with the scenario seen in the growth curves (Fig. 4D). Obviously, it confirmed that S. suis BirA has a role of being the BPL ligase in the alternative model, E. coli.
To further prove the BPL function of S. suis BirA, we established the assays of enzymatic reaction in vitro. In this system of BirA-catalyzed reaction, the substrate used is biotinylated domain (designated as AccB87 or BCCP87) of the AccB protein ( Fig. 5A,B), which carries a conserved biotinylation site of lysine at the position 122 (K122) (Fig. 5A,C). The method of thin layer chromatography (TLC) was applied to assay conversion of α -32 P-labeled ATP and biotin to biotinoyl-AMP (Fig. 5D). In principle, it represents direct evidence for the first ligase partial reaction (Fig. 5D), and upon an addition of the acceptor protein AccB87 provides an indirect proof of the second ligase partial reaction (i.e., transferring of biotin from biotinoyl-5′ -AMP to the AccB87 acceptor protein) (Fig. 5D). As expected, the S. suis BirA protein was shown to convert biotin and [α -32 P]-ATP to the canonical biotinoyl-5′ -AMP intermediate (Fig. 5D) and transferred the biotin moiety to the AccB-87 acceptor protein (Fig. 5D).
Subsequently, we utilized the matrix-assisted laser desorption/ionization (MALDI) to measure the level of BirA ligase-catalyzed biotinylation of AccB-87 as we recently described 4 . The MS results illustrated that the calculated mass for AccB87 is 10324.2~10327.5 (Fig. 6A), and the expected mass for biotinoyl-AccB87 is 10550.6 ( Fig. 6B). Collectively, the integrative data demonstrated that S. suis BirA acts as a functional member of the BPL family.
Binding of S. suis BirA to the cognate bioY gene. It is unusual that S. suis does not have the ability of de novo biotin synthesis in that this zoonotic pathogen lacks the bio operon found in E. coli 6 and other organisms like Agrobacterium 4 and Paracoccus 1 . However, it seemed likely that the inability of S. suis to make biotin is compensated with the BioY-mediated biotin uptake/scavenging pathway (Figs 1 and 7A). Also, the bioY lous is present in other two closely-relatives of the human pathogen (Enterococcus faecalis and Lactococcus lactis) (Fig. 7A). In particular, the L. lactis encoded two versions of bioY genes as well as two orthologues of BirA (Fig. 7A), implying the complexity of biotin metabolism in the certain species of low-GC contents, gram-positive bacteria.
The transcription start site of the S. suis bioY gene is estimated to be "T" that is 29 bp ahead of the translation initiation site "ATG". Bioinformatics analyses suggested that a putative BirA-binding site (TTT TGT TAA CCA TAA AAT TTT AAG AGG ATA ACA A) covering the transcription start site is present in the S. suis bioY promoter region (Fig. 7B). Given the above observations, we proposed that the bioY might be negatively regulated by the S. suis BirA. While this hypothesis required experimental evidence. We tested the ability of BirA to bind bioY promoter using a 54 bp probe containing the predicted site using the electrophoretic mobility shift (gel shift) assays (Fig. 8A) as we recently performed 2 with minor modifications. Gel shift assays showed that S. suis BirA efficiently bound the bioY probe in a dose-dependent manner (Fig. 8B) in that nearly 100% of bioY probe was transferred into the DNA-protein complex in the presence of 0.5 pmol BirA (Fig. 8B). In light of the appreciable conservation of the BirA sites in the bioY promoter from the related organisms (Fig. 8A), we also examined possible crosstalk of the BirA to bioY of various origins. In fact, the S. suis BirA was found to exhibit comparable binding to the bioY probes of L. lactis (Fig. 8C) and E. faecalitis (Fig. 8D). It demonstrated that physical interaction is present between the BirA bifunctional protein and the biotin transporter-encoding gene bioY.
Physiological Implications for Biotin Utilization Pathway. It is reasonable that S. suis deficient in biotin synthesis evolved the mechanism of BioY-BirA to utilize the biotin scavenged from the inhabiting niche and/or infected host environment (Fig. 1). In the epidemic strain of S. suis serotype 2, 05YH33, three genes with the involvement of biotin metabolism denote bioY (SSU05_0509), birA (05SSU_1625), and accB (05SSU1801), respectively. Following the biotinylation by BirA, the AccB was converted from its apo-form into holo-form, and participated into the initiation of fatty acid biosynthesis. Given the fact that BirA binds the bioY promoter with the help of biotinoyl-5′ -AMP (the intermediate of biotin biotinylation), the regulatory function of the BirA protein is supposed to guarantee that the wasteful production of the BioY transporter is avoided/minimized upon the biotin uptake from the outside environment. Probably, it is a physiological advantage for certain species of Streptococcus/Lactococcus in the context of lipid metabolism. To test above anticipation, we constructed the birA(Δ N) mutant of S. suis of which the DNA-binding domain was in-frame deleted (Fig. 9A,B). The removal of N-terminal DNA-binding motif affect bacterial growth on THB media, but this growth defection can be rescued by supplementation of the 5% defibrinated blood (or blood sera) (Fig. 9C,D). However, the expression of the plasmid-borne bioY promoter-driven lacZ is not altered significantly in the birA(Δ N) mutant in relative to the (D) Growth curves of E. coli BM4092 birA1 Km mutant and its derivatives. E. coli strains were maintained at 30 °C on the defined media M9 with/without 0.2% arabinose for around 36 hours. M9 agar plates were supplemented with 0.5 mM X-gal plus 25 nM biotin. The bacterial growth was measured by optical density at 600 nm, which is automatically recorded using a BioScreen C instrument. Each growth curve assay was carried out in triplicate and the average was used in this plot 50 . Designations: Vec, vector (c); ts: temperature-sensitive; Ara, Arabinose.
wild type (not shown). It might suggest a possibility that the interplay between BirA and bioY represent a developing and/or degenerating system for S. suis.

Conclusions
Our data shown here defined a working model for the route of biotin uptake/utilization in the zoonotic pathogen S. suis 2 (Fig. 1). Unlike the scenarios observed in both Brucella 11 and Paracoccus 1 in that the bioY gene interacts with the BioR regulator, our finding represents a first example for the interplay between the bioY and BirA in the Streptococcus/Lactococcus. Of note, no reaction is present between the bioY gene and BioR in the plant pathogen Agrobacterium 4 , a close relative of the human pathogen Brucella 11 . It suggested the complexity and diversity of bacterial biotin metabolism and regulation. Given the fact that the biotin synthetic genes bioJ 39 and bpl 2 are involved in bacterial virulence of the intracellular pathogen Francisella novicida, it is of much interest to probe  the possible role of biotin metabolism in Streptococcus pathogenesis. While the fact that both bioY and birA are essential for bacterial viability of Streptococcus suis argued the technical feasibility in the genetic removal of the two biotin-related genes. As we knew, biotin and lipoic acid both are sulfur-containing vitamins required for the three domains of the life. Similarly, the scavenging of lipoic acids by LplA was also required for the intracellular growth/survival and virulence 42,43 . Thereby we screened the genome sequence of S. suis 05ZYH33 for the presence of the lplA gene that encodes lipoate-protein ligase, giving the perfect hit (SSU05_1836). We are planning to examine its relevance to bacterial infectivity. Right now, it seemed true that both biotin and lipoic acid are nutritional virulence factors for certain species of bacterial pathogens. Given the fact that S. suis 2 is an emerging/ reemerging infectious agent threatening public health 26 , our finding might be helpful to better understanding biology and even infection of this zoonotic pathogen.   (Table 1), and all the E. coli strains are derived from the wild-type K-12 (Table 1). The two media (Luria Bertani (LB) and rich broth (RB)) were utilized for E. coli, whereas the Todd Hewitt Broth (THB) medium was used for S. suis 37 . Antibiotics were supplemented as follows (in mg/liter): sodium ampicillin, 100; kanamycin sulfate, 50; and Spectinomycin, 100.
Plasmids and genetic manipulations. The birA gene (SSU05_1625) was amplified by PCR with genomic DNA of S. suis 05ZYH33 as the template, and cloned into the expression vector pET28(a), giving the recombinant plasmid pET28-birA_ss (Table 1). To prepare the BirA protein, the expression plasmid pET28-birA_ss was transformed into the strain BL21(DE3), giving the strain FYJ280 (Table 1) 44 .
Also, the birA_ss gene was cloned into the arabinose-inducible expression vector pBAD24 4 , giving the plasmid pBAD24-birA_ss. To evaluate the in vivo activity of BirA, two birA mutants of E. coli were applied, which referred to the birA km mutant strain BM4092, and the temperature-sensitive mutant BM4062, respectively (Table 1). Given the fact the birA is a bifunctional gene and is required for bacterial viability, it is reasonable to delete the partial function of birA at 5′ -end. Therefore we employed an approach of homologous recombination to remove the N-terminal DNA-binding domain from the birA gene of S. suis 05ZYH33, giving the mutant birA(Δ N) ( Table 1). In this case, a thermos-sensitive suicide vector pSET4s 45 was applied. The promoter of S. suis bioY was fused to the promoter-less lacZ gene, creating the plasmid-borne PbioY-lacZ fusion (Table 1). To examine role of birA in vivo, the PbioY-lacZ fusion was separately introduced into the wild-type strain and the birA(Δ N) mutant of S. suis. All the acquired plasmids were verified by the PCR assay and direct DNA sequencing.   48 . The thin-layer chromatograms were dried overnight, exposed to a phosphor-imaging plate and visualized using a Fujifilm FLA-3000 Phosphor Imager.
Electrophoretic mobility shift assays. Gel shift experiments were conducted to test interaction of BirA protein with the bioY promoters of different origins 44,46,49 . Three sets of DNA probes ((bioY_SS, bioY_LL, and bioY_EF) were prepared by annealing two complementary oligonucleotides (Table 2). In the EMSA trials, the digoxigenin-labeled DNA probes (~0.2 pmol) were incubated with the purified BirA_ss protein in the binding buffer (Roche). When necessary, the biotinyl-5′ -AMP ligand was supplemented. The DNA/protein mixtures were separated with the native 7% PAGE and transferred onto nylon membrane by the direct contact gel transfer, giving the chemical-luminescence signals captured via the exposure of the membrane to ECL films (Amersham). β-Galactosidase assays. Overnight cultures of S. suis carrying the lacZ fusion grown in THB medium were subjected to measure direct measurement of β -galactosidase activity 44 . When necessary, the blood sera were added to augment bacterial growth of the mutant S. suis. The bacterial lysates were prepared using French pressure. The data were recorded in triplicate more than three independent assays. org/seq_tools/promoter.html). Structural modelling was proceeded with CPHmodels 3.2 Server (http://www. cbs.dtu.dk/services/CPHmodels).

Primers Sequences
SSbirA-F1 5′ -CG GGATCC ATG AAA ACC TAT CAG AAA ATA T-3'  Table 2. Primers used in this study. The underlined letters in italic are restriction sites, and the bold letters denote the BirA-binding sites.