Deciphering a unique biotin scavenging pathway with redundant genes in the probiotic bacterium Lactococcus lactis

Biotin protein ligase (BPL) is widespread in the three domains of the life. The paradigm BPL is the Escherichia coli BirA protein, which also functions as a repressor for the biotin biosynthesis pathway. Here we report that Lactococcus lactis possesses two different orthologues of birA (birA1_LL and birA2_LL). Unlike the scenario in E. coli, L. lactis appears to be auxotrophic for biotin in that it lacks a full biotin biosynthesis pathway. In contrast, it retains two biotin transporter-encoding genes (bioY1_LL and bioY2_LL), suggesting the use of a scavenging strategy to obtain biotin from the environment. The in vivo function of the two L. lactis birA genes was judged by their abilities to complement the conditional lethal E. coli birA mutant. Thin-layer chromatography and mass spectroscopy assays demonstrated that these two recombinant BirA proteins catalyze the biotinylation reaction of the acceptor biotin carboxyl carrier protein (BCCP), through the expected biotinoyl-AMP intermediate. Gel shift assays were used to characterize bioY1_LL and BirA1_LL. We also determined the ability to uptake 3H-biotin by L. lactis. Taken together, our results deciphered a unique biotin scavenging pathway with redundant genes present in the probiotic bacterium L. lactis.

as within the biotin utilization pathway. While unexpected, such an arrangement is not without precedent as a similar scenario was observed in Paracoccus 12 . This rare situation where redundant genes of biotin metabolism are present within the minimal genome of L. lactis might render some unknown physiological advantage, raising a possibility that it represents a relic of bacterial evolution and selection by its environmental niche 13,14 .
In this paper, we performed systematic functional analyses of the redundant genes in the context of biotin metabolism. We demonstrate in vivo evidence for biotin transport by L. lactis and also determined distinct functional assignments for the two BirA orthologues. Finally, we formulated a working model for biotin scavenging by L. lactis (Fig. 1). The atypical occurrence of two different biotin protein ligases might suggest its unique evolution history in adaption to its growing environment and/or ecological niche.

Results and Discussion
Redundancy of biotin metabolism genes in L. lactis. Similar to its closely-related cousin, the zoonotic pathogen Streptococcus suis 15 , L. lactis is also a low GC Gram-positive bacterium with a reduced genome (35.3% GC percentage, 2.37 Mb) 9 . However, it seemed unusual that the genetic organization of L. lactis features gene  (Table S1). Unlike S. suis that encodes only one birA gene and one bioY gene, L. lactis has two birA orthologues (called birA1 [L0191] and birA2 [L0192], Table S1) and two bioY orthologues (bioY1 [L24031] and bioY2 [L1011], respectively (Table S1). The birA2 gene encodes a putative "simple" biotin protein ligase (BPL) that lacks the N-terminal DNA binding motif found in the putative dual-functional birA1 (Figs 1 and 2). While the presence of multiple copies of birA and bioY homologues is atypical, two copies of birA have been found in Francisella 16 and two different bioY orthologues are present in Paracoccus 12 . Apart from redundancies in birA and bioY, redundancy is also found within the fatty acid synthesis loci in L. lactis (Table S1). Three loci of fabG (3-oxoacyl-ACP reductase) exist in L. lactis, namely fabG1 (L0185), fabG2 (L27694), and fabG3 (L1530) (Table S1). Of note, Wang and Cronan reported that the fabG2 is not an active 3-oxoacyl-ACP reductase 17 , indicating a possible role of these fabG loci in sugar metabolism. Moreover, L. lactis has two loci (fabZ [L0188] and fabZ1 [L160425]) that encode a 3-hydroxyacyl-ACP dehydratase (Table S1). Two fatty acid degradation (fad) genes (fadA [L25946] fadD [L54546]) also are present in L. lactis, whereas the fad system is cryptic for Streptococcus (Table S1). These gene arrangements suggest an unexpected complexity of lipid metabolism (including biotin utilization) in L. lactis compared with its close relative, S. suis (Table S1).
Characterization of the Two BirA homologues. BPL has been classified into two groups: the Group I BPLs lack an N-terminal DNA-binding motif whereas the Group II BPLs (BirAs) have an N-terminal DNA-binding domain 4 . The paradigm Group II BPL is the E. coli BirA and such enzymes also are found in the Archaea, suggesting that the group II version may be the ancestor of the BPL enzyme family. Based upon The multiple alignment was performed using ClustalW2 (http://www.ebi.ac.uk/Tools/clustalw2/index.html), and the final output was processed using the program ESPript 2.2 (http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi). The well-studied BirA_ec protein was used from E. coli K-12 MG1655 (Accession no. AAC43075), and the two BirA homologues were derived from L. lactis IL1403. BirA1 _LL (Accession no. NP_267937) is shown in red and BirA2 _LL (Accession no. NP_268060) is indicated in blue. Identical residues are in white letters with red background, similar residues are in red letters with white background, varied residues are in black letters, and dots represent gaps. The predicted secondary structure of BirA is shown at the top of the alignment. α : α -helix; β : β -sheet; T: β -turns/coils. Designations: EC, E. coli; LL, L. lactis. annotation, L. lactis IL1403 likely possesses two types of BPLs, which are referred to here on as BirA1 _LL (323 residues, a putative Group II BPL) and BirA2 _LL (250 aa, a putative Group I BPL). Multiple sequence alignments of the two BirA proteins showed that they exhibit 20.7% identity and 34.6% similarity compared to the E. coli BirA (Fig. 2).
To test the BPL activity of the two L. lactis IL1403 BirA proteins, we purified the hexahistidine-tagged proteins to homogeneity. SDS-PAGE profiles of the purified recombinant proteins suggested that BirA1 _LL is around 40 kDa, whereas the BirA2 _LL is estimated to be about 30 kDa (Fig. 3A). To further verify their identity, MALDI-TOF was used to analyze digested polypeptide fragments. The MS results supported that the two purified, recombinant proteins are BirA1 _LL (36.8% coverage) and BirA2 _LL (45.6% coverage) (Fig. 3B,C).
To better visualize the architectures of the two distinct forms of L. lactis BPLs, structural modelling was carried out. In similarity to that of the E. coli BirA (Fig. 4A) 18 , BirA1 _LL also possesses the N-terminal DNA-binding motif and the C-terminal enzymatic domain (Fig. 4B,C), fitting the criteria for the Group II BPLs. The structure of the BirA2 _LL was almost identical to that of the Aquifex aeolicus BirA, a member of Group I BPLs 19 (Fig. 4D-F), supporting its classification as a Group I BPL. Comparing the BPL activity of the two L. lactis BirA proteins. BPL activity was measured using the well-studied E. coli model together with in vitro enzymatic assays. The model E. coli strain for functional complementation of birA homologues is the birA1 mutant BM4092, which lacks the ability to synthesize biotin or to transport biotin with high affinity 20 . As expected, the BM4092 strain either with or without the empty vector pBAD24 did not grow on M9 agar plates supplemented with 25 nM of biotin (Fig. 5A). In contrast, the arabinose-induced expression (and even basal expression) of both birA1 _LL and birA2 _LL supported the growth of BM4092 under the non-permissive biotin levels (25 nM) (Fig. 5A). The bacterial growth curves of the BM4092 derivatives in liquid media also gave similar results to those obtained from the agar plates ( Fig. 5B,C). Of note, overexpression of birA1 _LL seemed slightly toxic for E. coli in that bacterial growth of the BM4092 strain supplemented with 0.2% arabinose was significantly inhibited when compared with growth conditions lacking the inducer, arabinose (Fig. 5C). Together, these data provided in vivo evidence that the two BirA orthologues of L. lactis (BirA1 _LL and BirA2 _LL ) have BPL activity.
Subsequently, BPL function of the BirA orthologues was further compared using in vitro assays to assess biotinylation activity. In this reaction, three versions of BirA were included (BirA_ec, BirA1 _LL and BirA2 _LL ), and the domain of the AccB protein targeted for biotinylation by BPLs (designated as AccB87 or BCCP87) was used as the substrate (Fig. 6A). In principle, the conserved lysine at position 122 (K122) of AccB87 should be biotinylated upon the presence of a functional BPL (Fig. 1A). The conversion of α -32 P-labeled ATP and biotin to the intermediate product, biotinoyl-AMP, by the BPLs (Fig. 1A) was directly visualized using thin layer chromatography (TLC) (Fig. 6B). The TLC approach also provided indirect proof of the second ligase partial reaction that occurs upon introduction of the acceptor protein AccB87 (i.e., transferring of biotin from biotinoyl-5′ -AMP to the AccB87 lysine) resulting in loss of the biotinoyl-5′ -AMP intermediate and formation of AMP (Figs 1A and 6B). The increased consumption of ATP in the presence of AccB87 (most notably for BirA-ec) was due to the fact that in absence of the acceptor protein, biotinoyl-AMP remains tightly bound within the ligase active site so that only one molecule of biotinoyl-AMP was formed per molecule of ligase (i.e., biotinoyl-AMP synthesis is not catalytic). Biotin transfer to an acceptor protein allows for catalysis and continued conversion of ATP to biotinoyl-AMP. These results further confirmed BPL activity.
Our results demonstrate that both BirA proteins from L. lactis possess the ability to convert biotin and [α -32 P]-ATP to the canonical biotinoyl-5′ -AMP intermediate (Fig. 6B) and indirectly support transfer of the biotin moiety to the AccB87 acceptor protein (Fig. 6B). The most striking difference between BirA1 _LL and BirA2 _LL was that upon addition of the acceptor protein, the ratio of AMP to ATP in the BirA2 _LL assay is appreciable higher than that seen in the BirA1 _LL assay and BirA2 _LL appears to form more of the biotinoyl-5′ -AMP intermediate (Fig. 6B). This indicates that BirA2 _LL catalyzes biotin attachment more quickly than BirA1 _LL . The E. coli BirA consumed the majority of the α -32 P-labeled ATP whereas no appreciable increase in ATP consumption was seen in the L. lactis BirA assays (Fig. 6B). This implies that the BirA_ec is the most active version amongst the three BPL enzymes under the conditions tested. Mass alteration of the AccB87 caused by biotinylation was directly measured using MALDI-TOF as previously described. In our assays, the mass for AccB87 was calculated to be 9335.5~9336.6 ( Fig. 7A), and the mass for the biotinoyl-AccB87 was determined to be 9562.3~9562.9 ( Fig. 7B-D). Collectively, the data illustrated that: 1) the two versions of L. lactis BirA (BirA1 _LL and BirA2 _LL ) functioned as BPLs with weak activity; 2) BirA2 _LL was more catalytically active relative to BirA1 _LL . The inoculated plates were grown at 30 °C for ~20 hours. Representative plate results are shown. Designations: Ara, arabinose; Vec, the pBAD24 empty vector, LL, L. lactis. Growth curves of the BM4092 birA Km mutant complemented with either plasmid-borne L. lactis birA genes (either birA1 _LL or birA2 _LL ) in liquid minimal media with (B) or without (C) the addition of 0.2% arabinose. 25 nM of biotin was supplemented into the M9 liquid media. 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 is shown in this plot 32 .
Scientific RepoRts | 6:25680 | DOI: 10.1038/srep25680 Binding of L. lactis BirA to the bioY genes. Relative to the scenario seen in the closely-relative cousin, S. suis, the genomic context of the bio loci is complex due to the presence of duplicated genes for birA and bioY (Fig. 8A). The putative BirA1 DNA binding site (ACA GTT AAC CTA AAT TTG ATT TTA GGG TTA CTG T) was detected in front of the bioY1 promoter region (Fig. 8B). According to the position of the transcriptional start site "T" predicted with the Neutral Network Program of Promoter Prediction (http://www.fruitfly.org/seq_tools/ promoter.html), the BirA1-binding site appears to overlap the "− 10" to "− 35" promoter regions (Fig. 8B), suggesting the transcription of bioY1 might be negatively regulated by BirA1 in L. lactis.  To test the function of the predicted BirA site (Figs 8A and 9A), electrophoresis mobility shift assays (EMSA) were conducted using the purified BirA1 _LL and BirA2 _LL (Fig. 9B-D). EMSAs confirmed that BirA1 _LL effectively binds the bioY1 probe in a dose-dependent manner (Fig. 9B), whereas BirA2 _LL , which lacks the putative N-terminal DNA binding motif, did not bind to the probe (Figs 2 and 9C). Interestingly, the S. suis bioY (bioY _SS ) promoter has the ability to interact with the L. lactis BirA1 _LL (Fig. 9D). A similar scenario was found in the case of S. suis BirA in that it also binds the L. lactis bioY promoter. Therefore, we anticipated that crosstalk would be present between the bioY and BirA of both S. suis and L. lactis. Given the fact that interaction occurs between the BirA1 and bioY1 promoter, while bioY2 lacks a predicted BirA1 binding site, it might be of interest to probe the physiological relevance of this regulation mechanism to biotin assimilation.
The ability of L. lactis to transport 3 H-biotin. Structural modelling against a solved BioY structure 8 suggested that both BioY1 _LL (Fig. 10A) and BioY2 _LL (Fig. 10B) consists of seven α -helices and exhibits appreciable similarity in their overall configuration. Given the fact that the L. lactis encoded two BioY transporters, we hypothesized that it would be capable of robust biotin transport. L. lactis cultures were grown in the presence of either limiting biotin or excess biotin (either 1 nM or 1 μ M) for 4.5 hours, pelleted and washed with PBS to remove free biotin, and used for 3 H-biotin transport assays (Fig. 10C,D). Relative transport of 3 H-biotin by biotin depleted cultures (predicted to produce both BioY transporters) and biotin replete cultures (predicted to require single transporter) was similar over the intervals measured. Minor differences between the samples were only observed at the initial transport time point (Fig. 10D). In contrast, at one minute and later, uptake/accumulation of biotin is roughly equal (Fig. 10C). It seems likely that the increment of exogenous biotin tested did not significantly augment activity of biotin uptake.

Conclusions
The data reported here suggests a specific pathway for biotin utilization in the probiotic bacterium L. lactis (Fig. 1) differing with the closely related animal pathogen S. suis in the redundancy (and/or duplication events) of the birA and bioY loci (Fig. 8). L. lactis is anticipated to be a biotin auxotroph in that it lacks a full biotin biosynthesis pathway. Unlike the regulatory machinery for biotin uptake in S. suis in which BirA modulates the transcription of the bioY gene, L. lactis has two BioY transporters responsible for biotin uptake, and two BPL enzymes (BirA1 _LL and BirA2 _LL ) catalyzing protein biotinylation. Among the two bioY genes (bioY1 _LL and bioY 2 _LL ), it seems likely that transcription of bioY1 _LL is regulated by BirA1 _LL , whereas the expression of bioY2 _LL is constitutive (Fig. 1). This unique mechanism might guarantee L. lactis possesses the ability to respond to fluctuating levels of biotin in the environment and/or the gastrointestinal tract.
To the best of our knowledge, regulation of bacterial biotin metabolism occurs by at least three diverse mechanisms represented by the E. coli BirA 4 , Agrobacterium BioR 20 , and Mycobacterium smegmatis BioQ 21 . The regulated transport of biotin is exemplified by BioR of Brucella 22 plus Paracoccus 12 and S. suis via BirA. The atypical  The minus sign denotes no addition of either BirA1 _LL or BirA2 _LL protein, and the DIG-labeled probe shifted by BirA1 _LL protein is indicated with an arrow. The concentrations of BirA1 _LL (BirA2 _LL ) in the right four lanes of each panel were (from left to right) 3, 6, 9 and 12 pmol, respectively. The protein samples were incubated with 0.2 pmol of DIG-labeled probe in a total volume of 15 μ l. To make biotinoyl-AMP, the ligand of BirA protein, 100 uM biotin plus 100 uM ATP was added into the gel shift system and incubated for ~1 hour. Designations LL and SS denote L. lactis and S. suis, respectively. BirA1 _LL and BirA2 _LL denote the two putative versions of L. lactis BirA protein, whereas bioY _LL and bioY _SS denote the bioY gene from L. lactis and S. suis, respectively. A representative graph from three independent gel shift assays is presented here. 7.5% native PAGE gels were used for the assays.
occurrence of two distinct, functional biotin protein ligases in L. lactis adds to the breadth of unique examples for bacterial biotin utilization (Fig. 1). In the organism Francisella, we also noted the atypical presence of two different BirA orthologues 16 . However, we failed to detect any orthologue of the BioY transporter 16 , ruling out the possibility of regulated transport of biotin. Unlike the scenario with Francisella where the BirA bifunctional protein can cross-talk with the E. coli bio operon in vitro and in vivo, the L. lactis BirA1 cannot regulate expression of the E. coli bio operon (note: blue color is due to appreciable expression of the bio-lacZ fusion, Fig. 5). Of note, the L. lactis BirA can crosstalk with the S. suis bioY gene (Fig. 9D), and vice versa. Consistent with the scenario in E. coli, the accB gene (L0187) encoding the biotin-acceptor protein was detected in the genome of L. lactis (Table S1). The reduced BPL activities of the two individual BirA homologues from L. lactis compared to the BPL activity of the single E. coli BirA (Fig. 6) might argue that redundancy of BirA LL is required to fulfill the need for protein biotinylation in L. lactis. The inability of L. lactis to synthesize biotin and the unique ecosystem of the gastrointestinal tract which is filled with a variety of nutritional elements and competing bacteria could select for the redundancy in biotin scavenging pathways or conversely, the gene organization might simply be a relic of bacterial evolution. Future work could explore the regulation of transporter production under different biotin conditions as well as differences in transporter efficiency.

Methods
Bacterial strains and growth conditions. The E. coli strains used included MG1655, BM4092 (Km mutant of birA) 20 , DH5α , and BL21 (DE3). The strains of L. lactis IL1403 9 and S. suis 2 15 were used for functional analyses. Luria Bertani (LB) and rich broth (RB) were used for the growth of E. coli and both Todd Hewitt Broth (THB) and minimal medium were used for the maintenance of L. lactis IL1403 and S. suis 2 23 . When necessary, antibiotics were added as follows (in mg/liter): sodium ampicillin, 100; and kanamycin sulfate, 50.
Plasmids and genetic manipulations. The two birA genes (birA1 [L0191] and birA2 [L0192]) were amplified by PCR and cloned into the expression vector pET28(a) to create histidine-tagged proteins for affinity purification and the arabinose-inducible vector pBAD24 for gene complementation experiments as previously described 24 . The derivatives of pET28 were introduced into BL21 (DE3) for protein production, whereas the pBAD24 derivatives expressing birA were transformed into the Km mutant of birA (BM4092) for functional assay of birA 25,26 . All the acquired plasmids were validated by PCR and DNA sequencing.
Protein Purification. Expression, purification, and quantification of the truncated AccB (AccB87) apo acceptor protein was performed as previously described 27 . The recombinant BirA1 _LL and BirA2 _LL proteins were purified from 1-L LB cultures grown at 37 °C to an OD 600 of 0.8 and protein production was initiated by the addition of 0.5 mM IPTG for 4 h at 30 °C. Cell lysis and protein purification were performed using the previously described protocols 28 . The purified proteins were dialyzed overnight in storage buffer containing 50 mM Tris-HCl [pH 8.0], 150 mM KCl, 10% glycerol and 0.1 mM DTT, concentrated using Millipore concentrators, flash frozen, and stored at − 80 °C. The purity of all the protein samples was judged by separation on 12% SDS-PAGE gels and staining with Coomassie brilliant blue.

Liquid chromatography quadrupole time-of-flight mass spectrometry. A Waters Q-Tof API-US
Quad-ToF mass spectrometer was used for the determination of the two L. lactis BirA orthologues (BirA1 _LL & BirA2 _LL ). The protein band of interest was removed from the gel and digested with Trypsin (G-Biosciences St. Louis, MO). Finally, the resultant peptides were loaded on a Waters Atlantis C-18 column (0.03 mm particle, 0.075 mm × 150 mm) and the acquired data were subjected for further analyses by the ms/ms. Bio-5ʹ-AMP Synthesis Reactions. The assay for BirA-catalyzed in vitro protein biotinylation activity was performed as described previously 28 with some modifications. Protein concentrations were determined using the extinction coefficients calculated from the protein sequence using the ExPASY Tools website. The assays contained 50 mM Tris-HCl (pH 8), 5 mM Tris-(2-carboxyethyl) phosphine, 5 mM MgCl 2 , 20 μ M biotin, 5 μ M ATP plus 16.5 nM [α -32 P] ATP, 100 mM KCl and 2 μ M BirA protein. Each of the reaction mixtures were incubated at 37 °C for 30 min. For each BirA protein tested, two identical tubes were used and at the end of the 30 min reaction AccB87 (50 μ M) was added to one of each pair of tubes while the other tube was left untreated. The tubes were incubated for an additional 15 min at 37 °C. 1 μ l of each reaction mixture was applied to a cellulose thin-layer chromatography plate of microcrystalline cellulose and the plates were developed in isobutyric acid-NH 4 OH-water (66:1:33) 29 . The thin-layer chromatograms were dried for 10 h, exposed to a phosphor-imaging plate and visualized using a Fujifilm FLA-3000 Phosphor Imager and Fujifilm Image Gauge software (version 3.4 for Mac OS).
Mass spectrometry. MALDI TOF/TOF mass spectrometer was used to measure the level of BirA-catalyzed biotinylation of AccB87. Reactions containing 100 μ M AccB-87, 3 μ M BirA, and 100 μ M biotin, 1 mM ATP, 10 mM MgCl 2 , 100 mM KCl, 5 mM tris-(2-carboxyethyl) phosphine in 50 mM Tris-HCl, (pH 8), at 37 °C, were incubated for 16 hours. Prior to the low-resolution matrix-assisted laser desorption/ionization analyses, the mixtures were dialyzed in 25 mM ammonium acetate and lyophilized to dryness. Data processing was performed using the FlexAnalysis 3.3 software package (Bruker Daltonics). Spectra were smoothed and a baseline correction was applied using the built-in features of the software package Electrophoretic mobility shift assays. Gel shift assays were performed to probe binding of BirA1_LL (and BirA2_LL) protein to the bioY promoters of L. lactis and S. suis 2 as previously described 26,30,31 . Two sets of DNA probes (bioY_LL and bioY_SS) were prepared by annealing two complementary oligonucleotides (bioY _LL -F: 5′ -CAA ATA ATA AAA TTA ACA GTT AAC CTA AAT TTG ATT TTA GGG TTA CTG TTT GAT ATG-3′ ; bioY _LL -R: 5′ -5′ -CAT ATC AAA CAG TAA CCC TAA AAT CAA ATT TAG GTT AAC TGT TAA TTT TAT TAT TTG-3′ ). In the binding buffer (Roche), the purified BirA1_LL (and BirA2_LL) protein in a series of dilutions was mixed with the digoxigenin-labeled DNA probes (~0.2 pmol). If required, the biotinyl-5′ -AMP ligand was added. The DNA/protein mixtures were separated on native 7% PAGE.
Transport of 3 H-biotin. L. lactis was grown overnight in THB. A one ml sample was pelleted, washed three times in PBS, and used to inoculate minimal medium supplemented with either 1 nM or 1 μ M of biotin. Samples were grown for 4.5 h, pelleted, and washed three times with PBS to remove external biotin. Pellets were then suspended in PBS plus 0.5% glucose for transport assays 6 and quantification of total protein in each sample using the Bio-Rad Protein assay. For transport assays, samples were incubated with 0.5 μ M 3 H-biotin at 30 °C for 0.5, 1, 2, 5, 30, or 60 minutes. Reactions were halted via dilution in ice cold PBS and bacteria were collected on 0.45 μ m filters. Filters were then mixed with scintillation fluid and DPM were counted using a Packard Scintillation counter.