Identification and Solution Structures of a Single Domain Biotin/Lipoyl Attachment Protein from Bacillus subtilis*

Protein biotinylation and lipoylation are post-translational modifications, in which biotin or lipoic acid is covalently attached to specific proteins containing biotin/lipoyl attachment domains. All the currently reported natural proteins containing biotin/lipoyl attachment domains are multidomain proteins and can only be modified by either biotin or lipoic acid in vivo. We have identified a single domain protein with 73 amino acid residues from Bacillus subtilis strain 168, and it can be both biotinylated and lipoylated in Escherichia coli. The protein is therefore named as biotin/lipoyl attachment protein (BLAP). This is the first report that a natural single domain protein exists as both a biotin and lipoic acid receptor. The solution structure of apo-BLAP showed that it adopts a typical fold of biotin/lipoyl attachment domain. The structure of biotinylated BLAP revealed that the biotin moiety is covalently attached to the side chain of Lys35, and the bicyclic ring of biotin is folded back and immobilized on the protein surface. The biotin moiety immobilization is mainly due to an interaction between the biotin ureido ring and the indole ring of Trp12. NMR study also indicated that the lipoyl group of the lipoylated BLAP is also immobilized on the protein surface in a similar fashion as the biotin moiety in the biotinylated protein.

The biotin/lipoyl attachment domain (IPR000089) is a signature structural motif, which has a conserved lysine residue that can bind biotin or lipoic acid. This domain can be found in enzymes that require biotin or lipoic acid as cofactor (1). Biotin plays a catalytic role in carboxyl transfer reactions. It is covalently attached to a lysine residue, via an ⑀-amino group, in enzymes such as pyruvate carboxylase, acetyl-CoA carboxylase, and propionyl-CoA carboxylase, etc. (2). This process, called biotinylation, is catalyzed by biotin-protein ligase (BPL) 2 (1,3).
Lipoic acid can serve as cofactor in a variety of proteins, such as E2 acyltransferases (E2p) in the pyruvate dehydrogenase complex and H-protein of the glycine cleavage system (4,5). It is also covalently bound via an amide linkage to a lysine group. This process, named lipoylation, is catalyzed by lipoate-protein ligase (LPL) (1,6). It is known that BPLs have broad substrate ranges. Mammalian biotinyl domain can be biotinylated by bacterial BPL and vice versa. The situation for LPLs is also similar (1).
The structures of several biotin/lipoyl attachment domains have been determined. Two of them are biotinyl domains, the C-terminal domain of Escherichia coli biotin carboxyl carrier protein (BCCP) and the biotinyl domain of 1.3 S subunit of transcarboxylase from Propionibacterium shermanii (7)(8)(9)(10). Several structures of lipoyl domains have also been reported (11)(12)(13)(14)(15). All biotinyl domains and lipoyl domains share a very similar overall fold, which is a flattened ␤-barrel formed by two ␤-sheets. The conserved lysine residue is located at the tip of a tight ␤-turn.
Although biotinyl and lipoyl domains are quite similar in their three-dimensional structures, there is no natural protein currently reported to be recognized by both BPL and LPL. However, it is found that the biotinyl domain of E. coli BCCP can be lipoylated after its protruding thumb is removed, and the removal does not affect its biotinylation property (16).
All the proteins containing a biotin/lipoyl attachment domain, reported so far, are multidomain proteins. The other domains are required for them to be functional (10,(17)(18)(19).
Here we report the identification and the structural characterization of a single domain protein (GenBank TM accession number NP_570905) with 73 amino acid residues from Bacillus subtilis strain 168. Because the size of this protein only corresponds to the sizes of the biotin/lipoyl attachment domains and can be both biotinylated and lipoylated in E. coli, we named it as biotin/lipoyl attachment protein (BLAP).

MATERIALS AND METHODS
Cloning, Expression, and Purification of BLAP-The gene of BLAP was amplified from B. subtilis strain 168 genome by PCR and cloned into protein expression vector pET-21a(ϩ). The recombinant plasmid was transformed into E. coli BL21(DE3)/ pLysS for protein expression. A single colony of the bacteria was cultured overnight in 2 ml of LB medium at 35°C, with 100 mg/liter ampicillin and 34 mg/liter chloramphenicol. The cells were collected by centrifugation and transferred into 250 ml of M9 minimal medium to continue incubation. When the cell density reached A 600 of 1.0, isopropyl 1-thio-␤-D-galactopyranoside was added to a final concentration of 0.5 mM to induce protein expression. D-Biotin was added into M9 medium to express the biotinylated protein, and ␣-lipoic acid was added for the lipoylated protein production. 15 NH 4 Cl and [ 13 C]glucose were used in M9 medium for preparing 15 N/ 13 C-labeled protein samples. BLAP was purified with a Q-Sepharose anion exchange column followed by gel filtration using ACTA fast protein liquid chromatography (Amersham Biosciences).
Reverse Transcription-PCR-The total mRNA was extracted from cells of B. subtilis strain 168 grown in LB media with 600 mg/liter streptomycin and harvested at A 600 of 0.6. Reverse transcription was carried out with 2.5 g of total RNA using superscript II reverse transcriptase and random hexamers as primers. Real time PCR was performed using the Qiagen SYBR Green TM PCR master mix kit according to the manufacturer's instruction. Each reaction was repeated three times.
Western Blot-The primary antibody was from the rabbit serum after 45 days of immunization by four times intramuscular injection of purified recombinant BLAP protein.
Anti-Rb IgG (H ϩ L) horseradish peroxidase was used as secondary antibody. Western blot analysis was performed according to the standard method (20) with purified BLAP as positive control.
NMR Spectroscopy-All the NMR samples contained about 1.2 mM protein in 50 mM phosphate buffer, pH 7.0, with 1 mM EDTA and 0.01% NaN 3 , in 90:10% H 2 O/D 2 O. 2,2-Dimethyl-2silapentanesulfonic acid was added as the chemical shift reference. All the NMR spectra were collected at 298 K on a Bruker AVANCE 600 MHz NMR spectrometer. The backbone chemical shift assignments were obtained based on two-dimensional 1 H-15 N HSQC, three-dimensional HNCACB, three-dimensional CBCA(CO)NH, three-dimensional HNCA, and threedimensional HNCO experiment data, whereas the side chain resonance assignments were obtained from two-dimensional 1 H-13 C HSQC, three-dimensional HCCH-COSY, three-dimensional HBHA(CO)NH, three-dimensional 15 N total correlation spectroscopy-HSQC, three-dimensional HCC(CO)NH, and three-dimensional CC(CO)NH experiment data (21). Three-dimensional 13 C-NOESY-HSQC and three-dimensional 15 N-NOESY-HSQC spectra (22) were collected with mixing times of 60 and 120 ms, respectively. The side chain assignments of aromatic residues were obtained based on three-dimensional 13 C-NOESY-HSQC experiment data for the aromatic carbons. Two-dimensional 13 C/ 15 N-filtered NOESY spectra (23, 24) (mixing time of 200 ms) were collected for the assignments of proton resonances from the attached biotin or lipoate moiety. One-dimensional 1 H NMR spectrum, two-dimensional NOESY, and double quantum-filtered correlation spectroscopy data were used for obtaining the proton chemical shifts of free lipoic acid. The steady state heteronuclear { 1 H}- 15 N NOE experiments (25) were performed in the presence and absence of a 3-s proton pre-saturation period (26,27). All NMR spectra were processed using NMRPipe (28) and analyzed using NMRView (29).
Structure Calculations-The solution structures were calculated using CYANA (30) and refined with AMBER (31). The distance restraints were obtained from analysis of NOESY data using SANE (32). Dihedral angle restraints ( and ) were determined based on analysis results from TALOS (33) and CSI (34). 1 angles and stereo-specific assignments of H ␤ were determined based on NOE intensity information from threedimensional 15 N-NOESY-HSQC and three-dimensional 13 C-NOESY-HSQC data. Hydrogen bond constraints were used based on NOE analysis and the secondary structure predictions by CSI.
A total of 200 structures was calculated with CYANA, and 100 of them with the lowest target functions were selected for further refinement with AMBER. The final 20 structures with the lowest AMBER energies were selected for structure representation. The final structures were analyzed using MOLMOL (35) and PROCHECK-NMR (36). The root mean square deviation (r.m.s.d.) data were calculated using SUPPOSE (37). The mean structures were generated using SUPPOSE and energyminimized with AMBER.

Cloning and Recombinant Expression of BLAP in E. coli-The
DNA fragment encoding BLAP was successfully amplified from the genome DNA of B. subtilis strain 168 and was ligated into pET expression vector. DNA sequencing results showed that the coding sequence of the cloned BLAP gene is identical to the published sequence in the NCBI data base (GenBank TM accession number NC_000964).
BLAP was expressed in E. coli with a typical protein yield of 60 -80 mg/liter. Mass spectrometry (MS) analysis showed that the molecular mass of the purified protein is 7913 Da, which is 128 Da less than the theoretical value. This mass difference is because of the missing of the first methionine, which was confirmed by N-terminal protein sequencing analysis. The MS data indicated that the expressed BLAP is without any post-translational modification. When the bacteria were grown in M9 minimal medium supplied with 40 mg/liter biotin, both apo and biotinylated (btl) BLAP were obtained. Apo-BLAP and btl-BLAP can be separated on an anion exchange column because btl-BLAP has one less positive charge than apo-BLAP. At pH 9.0, btl-BLAP was eluted from the Q-Sepharose column at 80 mM NaCl, and apo-BLAP was eluted at 130 mM NaCl. Interestingly, lipoylated (lpl) BLAP was produced when the cells were cultured in M9 minimal medium with 40 mg/liter lipoic acid. The molecular masses of both btl-BLAP and lpl-BLAP were verified by MS. The molecular weight difference between btl-BLAP and apo-BALP is 225.9 Da and that between lpl-BLAP and apo-BLAP is 188.5 Da. These values agreed with the corresponding theoretical values.
The amount of btl-BLAP or lpl-BLAP produced in E. coli is dependent on the incubation time after isopropyl 1-thio-␤-D-galactopyranoside induction. When the bacteria were incubated for 6.5 h after induction, about 10% BLAP was biotinylated (ϳ50% lipoylated). When it was elongated to 16 h, about 50% BLAP was biotinylated (ϳ75% lipoylated). The rates of biotinylation and lipoylation for BLAP in E. coli are comparable with the biotinylation rate of the biotinyl domain of E. coli BCCP and the lipoylation rate of the lipoyl domain of E. coli E2p, respectively (38).
Verification of BLAP Expression in B. subtilis-To determine whether BLAP is expressed in B. subtilis, we first tried to detect the mRNA of BLAP in B. subtilis total mRNA extraction. After reverse transcription, real time PCR was carried out, and the results showed that the mRNA of BLAP has a transcription level close to that of B. subtilis 16 S rRNA gene rrnO (data not shown). The fluorescence threshold was set at 0.02, and the threshold cycle (CT) for BLAP coding sequence and rrnO is 28.8 and 28.2, respectively. Melting curve analysis indicated that the PCR product is a single fraction, which means no nonspecific PCR occurred.
Western blot was used to verify the protein expression of BLAP in B. subtilis. Cells of B. subtilis strain 168 were cultured in 500 ml of LB media until the A 600 reached 1.5. The bacteria were harvested, and the cells were broken by using a French press. The supernatant was then loaded onto a Q-Sepharose column at pH 9.0, and the fraction eluted with 0.2 M NaCl was collected and concentrated to 0.5 ml. 7.5 l of this concentrated fraction was used for Western blot analysis. A single band was detected from this sample by immunoblot using multiclonal antibody generated against the recombinant BLAP. This single band appeared at exactly the same position as the purified recombinant BLAP.
Solution Structure of Apo-BLAP-Nearly all the chemical shift assignments were obtained for apo-BLAP except for Asn 70 , whose NH signal was missing in the two-dimensional 1 H-15 N HSQC spectrum. The solution structure of apo-BLAP was calculated based on NOE, dihedral angle, and hydrogen bond restraints ( Table 1). The superimposition of the final 20 structures with the lowest AMBER energies for apo-BLAP, together with a ribbon diagram of the energy minimized mean structure, are shown in Fig. 1. Residues 2-69 of apo-BLAP form a well defined structure, whereas the last four residues are flexible. Structural statistics indicated the solution structure is in very good quality (Table 1).
Solution Structure of btl-BLAP-We obtained nearly complete chemical shift assignments for btl-BLAP except Val 3 and the NH of Asn 70 . An extra signal appeared in the two-dimensional 1 H-15 N HSQC spectrum of btl-BLAP, which was assigned to the ⑀-NH of the biotin-attached Lys 35 .
Comparing the two-dimensional 1 H-15 N HSQC spectrum of btl-BLAP with that of apo-BLAP, it is found that the backbone NH signal of Lys 35 has the largest chemical shift change (Fig. 2). Meanwhile, NH signals from residues near Lys 35 (Ile 30 -Ile 38 ) have significant chemical shift changes because of the biotinylation. Chemical shift changes are also observed for NH signals of residues 7, 9 -15, 28, 56, and 60, as well as the NH 2 signals of Asn 10 (Fig. 2).
The 1 H chemical shifts of the attached biotin were all assigned and are listed in Table 2 and Structure 1. The attached biotin moiety showed significant 1 H chemical shift changes as compared with free biotin. Most of the 1 H chemical shifts of the attached biotin are decreased, and generally, those further to the carboxyl group have larger decrease in chemical shift. The chemical shifts for HN3 and one proton of H10 are increased.
The superimposition of the 20 btl-BLAP structures with the lowest AMBER energies is shown in Fig. 3A. The structural statistics for btl-BLAP are listed in Table 1. The structures of apo-BLAP and btl-BLAP are very similar (Fig. 3B). The r.m.s.d.
(residues 2-69) for backbone heavy atoms between the mean structures of the two forms is 0.83 Å. The major structural displacement occurs at the end of ␤4, the biotin-attached turn, and the first half of ␤5. The ␤-turn with the attached biotin is converted into a type IЈ ␤-turn from a type IIIЈ ␤-turn in apo-BLAP. As a result, the end of ␤4 moves closer to the beginning of ␤2. The biotinylation of BLAP stabilizes the side chain of Lys 35 , which leads to the side chain heavy atom r.m.s.d. of Lys 35 (not including biotin moiety) for the ensemble of structures reduced from 1.72 Å in apo-BLAP to 0.69 Å in btl-BLAP.
Conformation of the Attached Biotin-The biotin moiety is well defined in the btl-BLAP structure. The heavy atom r.m.s.d. from mean for the biotin group is 0.61 Å in the ensemble of structures. The bicyclic ring of the attached biotin is folded back and immobilized on the protein surface, mainly on top of ␤4. The side chains of Glu 32 (on ␤4) and Glu 37 (on ␤5) are pushed away from the biotin for its bicyclic ring to fit (Fig. 4, A  and B). The NH 2 signals of Asn 10 in the two-dimensional   H-15 N HSQC spectrum of btl-BLAP show a very large shift compared with those of apo-BLAP. In the btl-BLAP structure, the side chain of Asn 10 moves toward the side chain of Glu 32 , and the NH 2 group of Asn 10 is in a position to form a hydrogen bond with the carboxyl group of Glu 32 in all 20 structures of btl-BLAP. Most interestingly, the indole ring of Trp 12 turns almost 90°toward the biotin to face its ureido ring. The conformation changes for the side chains of Asn 10 and Trp 12 on ␤2 are probably responsible for the NH chemical shift changes for residues 9 -15. The small chemical shift changes observed for NH signals of Val 28 , Gly 56 , and Asn 60 could be attributed to a secondary effect from the conformation changes of Asn 10 and Trp 12 side chains, because of the hydrogen bond network of the S2 ␤-sheet. The binding site of the biotin bicyclic ring is mainly negatively charged (Fig. 3C), and no specific interaction (hydrogen bond, hydrophobic interaction, etc.) between the biotin bicyclic ring and the protein can be identified. Analysis of NOEs between the biotin and the protein showed that all the protons on the biotin bicyclic ring have NOEs with the indole ring of Trp 12 . The distance between the center of the Trp 12 hexagon ring and the center of the biotin ureido ring is about 3.8 Å, and the angle between the planes of the two rings is 29.5°in the mean structure.
The protons on the biotin bicyclic ring and those close to this ring, whose chemical shifts decrease significantly, must be within the shielding cone of the indole ring of Trp 12 . Their chemical shift decreases should be a result of aromatic ring current effect, although the HN3 of biotin is probably located outside of the shielding cone and therefore has an increased chemical shift. The distances between H10 protons and the centers of either the pentagon or the hexagon rings of the indole group are over 10 Å. Therefore, the chemical shift changes for H10 are probably related to local conformation change because of the immobilization of biotin.
The steady state heteronuclear { 1 H}-15 N NOE value of the ⑀-N H of Trp 12 changes from 0.50 Ϯ 0.02 in apo-BLAP to 0.71 Ϯ 0.02 in btl-BLAP. This indicates that the motion of the indole ring is somewhat restricted upon the attachment of biotin. As only this indole ring moved toward biotin and all other nearby side chains moved away, we suspected that the immobilization of the biotin is mainly due to an interaction between the Trp 12 indole ring and the biotin ureido ring.
To test this hypothesis, we mutated Trp 12 to a methionine (W12M) and studied the W12M BLAP mutant with NMR. Comparing the two-dimensional 1 H-15 N HSQC spectrum of wild-type (WT) apo-BLAP with that of W12M apo-BLAP, it was found that all residues with significant NH chemical shift changes are located near the mutated residue 12 (Fig. 5A). Comparison of 1 H ␣ chemical shifts between W12M apo-BLAP and WT apo-BLAP showed that the 1 H ␣ chemical shift differences for most residues are smaller than 0.02 ppm, except that the 1 H ␣ chemical shift differences for Lys 13  The W12M mutation did not cause any change to the protein biotinylation in E. coli. Fig. 5B shows the overlay of the two-   biotin ureido ring is responsible for the immobilization of the biotin moiety in WT BLAP. The nature of this interaction is probably electrostatic interaction.
NMR Study of lpl-BLAP-Backbone resonance assignments were carried out for lpl-BLAP. Except that Asn 70 lacked an NH signal, all the possible backbone NH and side chain NH 2 resonances in the two-dimensional 1 H-15 N HSQC spectrum were assigned. The ⑀-NH signal of the lipoate attached Lys 35 was also identified (Fig. 6).
Similar to btl-BLAP, the residues with obvious NH chemical shift changes because of lipoylation mainly exist in two regions as follows: one is near Lys 35 (Ile 30 , Glu 32 -Ile 38 ) and the other is near Trp 12 (Asn 10 -Lys 13 ). Val 14 , Gly 56 , and Asn 60 also show small chemical shift changes. The residues with obvious chem-ical shift variations because of biotinylation or lipoylation are almost identical. Actually, the NH chemical shift differences between lpl-BLAP and btl-BLAP are much smaller than those between them and the apo-BLAP for residues with very large NH chemical shift changes (residues 10 ␦ , 35 and 36) (Fig. 6).
The 1 H chemical shifts of the attached and free lipoic acid were also assigned (Table 3 and Structure 2). All the 1 H chemical shifts are decreased for the attached lipoic acid compared with those of the free lipoic acid except for H9. Those further from the carboxyl group have larger chemical shift changes. This is similar to the case of the attached biotin in btl-BLAP. The significant chemical shift decreases are probably also because of the ring current effect of Trp 12 .
The steady state heteronuclear { 1 H}-15 N NOE value of the ⑀-N H of Trp 12 increases from 0.50 Ϯ 0.02 in apo-BLAP to 0.61 Ϯ 0.02 in lpl-BLAP. Therefore, we believe that the attached lipoate moiety is also immobilized on the protein surface, and it is in a similar conformation as the attached biotin in btl-BLAP. We suspected that the immobilization of lipoate moiety is also mainly due to an electrostatic interaction between the rings of lipoic acid and Trp 12 .

DISCUSSION
The Existence of BLAP-The B. subtilis BLAP (GenBank TM accession number NP_570905) was initially annotated as a BCCP. Unlike other bacterial BCCPs that normally have 150 -    The entry for GenBank TM accession number NP_570905 was later discontinued in the NCBI protein data base after we have cloned the gene of BLAP. In this study, we have proved the existence of BLAP in B. subtilis at both the mRNA level and the protein level.
About 30 homologous protein sequences were found when we searched the nonredundant protein data base with the B. subtilis BLAP sequence on Biology Workbench. These protein sequences all have over 60% positive matches with the B. subtilis BLAP, and their sequence lengths are all less than 90 amino acid residues (Fig. 7). The MKM motif and the glutamic acid at the Ϫ3 position with respect to the target lysine, which is very important for biotinylation, are highly conserved (Fig. 7). All of them are from prokaryotes, except one is from Danio rerio (GenBank TM accession number XP_701405). Therefore, BLAP is not only restricted to B. subtilis but also exists in other species.
Immobilization of the Biotin-The bicyclic ring of biotin is also attached on the protein surface near the protruding thumb of the biotinyl domain of E. coli BCCP (7). It was first proposed that a hydrogen bond between the HN 2 of biotin and the O ␥ of Thr 94 was the reason for the immobilization of biotin based on the crystal structure (7), although NMR study results of the biotinyl domain of E. coli BCCP did not support this hydrogen bond in solution structure (9). However, the immobilization of the attached biotin in E. coli BCCP was generally attributed to the interaction between biotin and the protruding thumb (8 -10).
By Examining the structure of the biotinylated biotinyl domain of E. coli BCCP, we found that Tyr 92 of E. coli BCCP is in a similar position to Trp 12 of btl-BLAP. The phenyl ring of Tyr 92 in E. coli BCCP is also facing the ureido ring of biotin, and the distance between the centers of the two rings is 3.7 Å (Fig. 4C). In addition, it is also reported that extensive NOEs have been observed between the phenyl ring of Tyr 92 and every proton of the biotin bicyclic ring (9). Based on this information, we propose that the immobilization of the biotin moiety in E. coli BCCP is also because of an interaction between the phenyl ring of Tyr 92 and the biotin ureido ring, similar to the case for btl-BLAP described above.
On the other hand, it has been reported that the attached biotin is flexible in the biotinyl domain of the 1.3 S subunit of transcarboxylase from P. shermanii (39). The structure of the biotinyl domain of the 1.3 S subunit shows that there is no aromatic residue in similar positions as Trp 12 of BLAP or Tyr 92 of E. coli BCCP (Fig. 4D). This should explain the flexibility of the attached biotin in the biotinyl domain of the 1.3 S subunit.
Substrate Recognition by BPL and LPL-Although biotinyl domains and lipoyl domains share a very similar overall fold, it is a general belief that BPL and LPL enzymes can distinguish biotinyl and lipoyl domains as substrates (1). Several studies have been carried out to reveal the mechanisms for the enzyme specificities of BPL and LPL (16,38,40,41).
Mutagenesis study has shown that removing the protruding thumb, which lipoyl domains do not have, from the biotinyl domain of E. coli BCCP made it an efficient substrate for LPL, and it still retained the ability to be recognized by BPL (16). So the protruding thumb of E. coli BCCP prevents it from being lipoylated. Also, based on sequence and structure comparisons, it has been proposed that loop 1 between strands 1 and 2 is responsible for BPL not recognizing lipoyl domains (42). This loop in lipoyl domains is generally a few residues longer than that in biotinyl domains.
BLAP can be both biotinylated and lipoylated, which indicates that it has the structural characteristics recognized by both BPL and LPL. In Fig. 8, the structure of BLAP is compared with two biotinyl domain structures and three lipoyl domain structures. The fact that BLAP lacks the protruding thumb and has shorter loop 1 makes it a good substrate for both BPL and LPL.
The protruding thumb of the biotinyl domain of E. coli BCCP is not a common feature for all biotinyl domains. It only exists in the biotinyl domains of bacterial acetyl-CoA carboxylase (1). Other biotinyl domains, such as that of the 1.3 S subunit from P. shermanii, have much shorter loops at the corresponding places, as does BLAP. It is possible that this kind of biotinyl domain can also be an effective substrate for LPL.
Biological Implications-Being both biotinylated and lipoylated efficiently does not mean that BLAP is involved in carboxylase or dehydrogenase complex, as the biotin/lipoate attachment domain cannot function independently. The N domain and the linker fragment of E. coli BCCP are indispensable for the assembly and function of the acetyl-CoA carboxylase complex in vivo (10,18). Similarly, an active dehydrogenase can never assemble if it lacks the binding domain and the catalytic domain on the E2 component (19). Because BLAP is a single domain protein, its biological function should be different from currently known biotinyl/lipoyl domain-containing proteins.
The fact that BLAP can be both biotinylated and lipoylated in E. coli suggests that it can compete with the biotinyl/lipoyl domain-containing proteins for biotin or lipoic acid. Thus, it is possible for BLAP to play some kind of regulatory role.
In summary, we have identified a new protein (BLAP) from B. subtilis that can be both biotinylated and lipoylated. Unlike all other biotin/lipoyl attachment domain-containing proteins, it is a single domain protein. However, its biological function is still not clear, and further studies are needed to reveal the biological role of BLAP.