Bacillamides, which have four members of bacillamide A, B, C and D1,2,3,4 (Fig. S1), contain a tryptamide thiazole, a common building block in potential bioactive cyclic peptides such as the protein synthesis inhibitors A21459 A and B from an Actinoplanes sp.5, antibiotic zelkovamycin produced by a Streptomyces sp. K96-06706 and immunosuppressive argyrins from myxobacteria7. Bacillamide A and its derivatives show antibiosis against dinoflagellates, raphidophytes2 and a particular species of cyanobacteria8. In our previous studies, bacillamide C and a new compound of neobacillamide A were isolated from Bacillus atrophaeus C89 associated with the South China Sea sponge Dysidea avara9. Since bacillamide C is obtained in higher yield than that of neobacillamide A9, the fermentation optimization of bacillamide C was carried out and a 26-fold increase in bacillamide C production (71 mg/L) by B. atrophaeus C89 was achieved10. Though chemical synthesis of bacillamide C was carried out recently11,12, the biosynthetic pathway of bacillamide C remains unknown.

Nonribosomal peptides generated by nonribosomal peptide synthetases (NRPS) are extraordinary important for modern medicine because of their diverse biological activities, for example antibiotics tyrocidin A13, vancomycin14, immunosuppressive agents cyclosporine A15, cytostatics bleomycin A216 and toxins thaxtomin A17. Biogenetically, bacillamides including neobacillamide A and bacillamide C belong to nonribosomal peptides. Similarly as neobacillamide A9, bacillamide C could be derived from amino acids alanine, cysteine and tryptophan through NRPS biosynthesis strategy, which was supported by the finding of a NRPS gene cluster (7, 011 bp) by sequencing the genome of B. atrophaeus C89 (Genbank No. JQ 687535)18. Accordingly, the putative biosynthetic pathway of bacillamide C could be proposed based on the NRPSs domain organization (Fig. 1). Among the six domains of the NRPS gene cluster, the first module contains an adenylation domain and a peptidyl carrier protein domain (A-PCP), which selects and activates alanine. The peptide bond formation and heterocyclization is performed with cysteine by the adjacent elongation module containing a condensation domain, an adenylation domain and a peptidyl carrier protein domain (C-A-PCP). However, we do not know whether the decarboxylation of tryptophan to tryptamine is performed before amidation or after.

Figure 1
figure 1

Supposed biosynthetic pathway of bacillamide C. Each square represents a NRPS enzymatic domain: C: condensation domain; A: adenylation domain; P: peptidyl carrier protein domain.

According to the genomic data of B. atrophaeus C89 (Genbank No. JQ 687535), there are fifteen putative decarboxylases in B. atrophaeus C89, so, which one is responsible for the decarboxylation needs to be defined. Here we put forward the hypothesis that the decarboxylase gene (AADC gene), which is at the nearest downstream after the NRPS gene cluster in B. atrophaeus C89, is probably involved in the decarboxylation of L-tryptophan to tryptamine and then tryptamine is combined with the product of NRPS gene cluster by amidation (Fig. 1). In order to clarify the above hypothesis, under the guide of bioinformatics analysis of substrates' similarity and amino acid sequence, the AADC gene of B. atrophaeus C89 was cloned and expressed in Escherichia coli BL21. Subsequently, the isolated AADC enzyme was characterized as a highly efficient catalyst for the decarboxylation of tryptophan to tryptamine, suggesting that tryptamine rather than tryptophan was incorporated into the non-ribosomal peptide bacillamide C. Thus, this study provides an insight to elucidate the biosynthetic mechanism of bacillamide C in B. atrophaeus C89.


Analysis on the potential substrate of AADC

According to the bioinformatic analysis18, there were 15 possible decarboxylases in the genome of B. atrophaeus C89 and all of them were exhibited high similarity (from 72% to 100%) to each other (Table 1). Usually, an enzyme is named based upon its substrate. For example, decarboxylyases are named based upon their substrates19. There are a quite few known substrate molecules in KEGG database. With the substrate-enzyme relationship in KEGG, a few of substrates of decarboxylases in B. atrophaeus C89 were predicted except for phenolic acid decarboxylase (id001001), phenylacrylic acid decarboxylase (id 003690) and pyridoxal-dependent decarboxylyase (id 003505) (Table 1). The alternative name of phenolic acid decarboxylase (id001001) is 4-hydroxybenzoate decarboxylase, thus 4-hydroxybenzoate could be the substrate of phenolic acid decarboxylase (id001001). Amino acid sequence of phenylacrylic acid decarboxylase (YP_004875917.1) showed 94% identity with phenylacrylic acid decarboxylase (id003690) from B. atrophaeus C89. YP_004875917.1 catalyzes the conversion of 4-coumarate (p-coumaric acid) to 4-4-vinylphenol20, so 4-coumarate could be a substrate of phenylacrylic acid decarboxylase (id003690). Similarly, the putative full length amino acid sequence of JQ400024 (id003505) showed 100% and 97% identity to the pyridoxal-dependent decarboxylyase of B. atrophaeus 1942 (YP_003975768.1) and aromatic-L-amino-acid decarboxylase of Muricauda ruestringensis DSM 13 (YP_004788433.1) respectively. Therefore, L-tryptophan and L-phenylalanine are the possible substrates for the functional gene JQ400024/id003505. As shown in Table 1, 14 different substrates corresponding to the 15 decarboxylases were searched, in which JQ608457 (id001001), JQ608467 (id003691) and JQ608468 (id003692) have the same substrate, 4-hydroxybenzoate. Whereas, JQ400024 (id003505) has two possible substrates L-tryptophan and L-phenylalanine.

Table 1 Putative substrates of 15 decarboxylases from B. atrophaeus C89

The amino acid sequence of the catalytic domain of functional gene JQ400024 (id003505) was aligned with certain reported decarboxylases in the GenBank (No: ZP 04126748, YP 004788433, YP 004561607 and ZP 05855305) (Fig. 2). AADC amino acid sequence (No: ZP 04126748) of Bacillus thuringiensis serovar sotto str.T04001 showed 88% similarity with JQ400024 (id003505). Both YP 004561607 and ZP 05855305 were the same type of decarboxylases, which belong to aromatic L-amino acid decarboxylases. Although the AADC sequence has lower similarity (less than 50%) with these aromatic L-amino acid decarboxylases, all of them have identical conserved pyridoxal 5′-phosphate (PLP) binding pockets and catalytic residue (Lys) (Fig. 2). The alignment result suggested that the JQ400024 (id003505) was a type of aromatic L-amino acid decarboxylase and it could catalyze the decarboxylation of aromatic L-amino acids.

Figure 2
figure 2

Amino acid sequence multiple alignments of decarboxylases catalytic domains from B. atrophaeus C89 (AADC), Bacillus thuringiensis serovar sotto str. T04001 (ZP 04126748), Muricauda ruestringensis DSM13258 (YP 004788433), Erysipelothrix rhusiopathiae str. Fujisawa (YP 004561607) and Blautia hansenii DSM20583 (ZP 05855305).

The pyridoxal 5′-phosphate binding pocket was indicated by black arrow. The catalytic residue was indicated by red arrow.

Owing to the relatively low sequence similarity for the same functional decarboxylases and the possible diversity in terms of substrates (as discussed above), a comparison was carried out between the observed substrate molecules. It is straightforward for us to characterize substrate properties based on their volume size, molecular weight, numbers of hydrogen bonding donor and acceptor, oil-water distribution, number of aromatic rings and so on. As shown in Table 2, uroporphyrinogen III, L-tryptophan, L-phenylalanine, 4-hydroxybenzoate and 4-courmarate are aromatic carboxylates and others are not aromatic carboxylates. For these aromatic substrates, molecular size was in order of uroporphyrinogen III > L-tryptophen > L-phenylalanine > 4-courmarate > 4-hydroxybenzoate. Actually, molecular volume of L-tryptophan, L-phenylalanine and 4-courmarate were in a range of 100–130 Å. However, molecular weight (molar mass) and oil-water solubility of L-trytophan and L-phenylalanine were much lower than those of 4-courmarate and 4-hydroxybenzoate. As a result, we concluded that L-tryptophan and L-phenylalanine shared high similarity among the known substrates of decarboxylases (Fig. 3).

Table 2 Molecular characteristics of 14 decarboxylases' substrates
Figure 3
figure 3

3D scatterplot for the molecular properties of fourteen deduced substrates from fifteen decarboxylases.

A: based on molecular properties of volume, H: donor and aromatic bond; B: based on molecular properties of molecular weight, Alogp and solubility. Target substrate was indicated by black arrow.

Function of AADC in the transformation of L-tryptophan into tryptamine

The AADC gene was overexpressed in E.coli BL21 (DE3) as an N-termal His6-tagged fusion protein and purified by an affinity chromatography. The protein purity was judged by SDS-PAGE analysis which revealed AADC with a molecular mass of 58 kDa (Fig. 4). The MW and pI of the enzyme were calculated as 55.899 kDa and 5.96, respectively.

Figure 4
figure 4

SDS-PAGE of purified recombinant decarboxylase.

M: protein Marker; Lane 1: unpurified sample; Lane 2: purified enzyme.

As proposed in Fig. 1, L-tryptophan was tested first as a putative substrate of AADC. Thin layer chromatography (TLC) and high pressure liquid chromatography (HPLC) were employed to monitor the production of tryptamine, the proposed product of AADC. After incubation with 1 mM of AADC and 1 mM of L-tryptophan for 15 min, a small peak occurred at 9.5 min (Fig. 5). The peak was increased with increase in incubation time, with decrease in L-tryptophan (rt = 2.5 min). The peak of 161.1 m/z ([M + H]+) in Fig. 6 proved the product of tryptamine because of its exact mass is 160.22. The fragmentation ion at 144.1 m/z is the product of the loss of ammonia from tryptamine and the fragmentation ion at 146.1 m/z is [M + 2H]2+ (Fig. 6)21. Subsequently, the kinetic parameters were determined based on the tryptamine production. Km and Vmax values were calculated as 0.35 mmol/L and 0.163 mg product/min·mg protein, respectively (Fig. 7). Next, we used two other aromatic amino acids L-phenylalanine and L-tyrosine as substrates to assay the activity of AADC. Thin layer chromatography (TLC) (ninhydrin staining; mobile phase butanol/acetic acid/water: 4/1/1) showed that AADC could not use L-tyrosine as substrate (data not shown). However, AADC could recognize L-phenylalanine as a substrate and catalyze its decarboxylation. The kinetic parameters of AADC toward L-phenylalanine were determined as 0.872 mmol/L and 0.00184 mg product/min·mg protein, respectively (Fig. 7). But, L-tryptophan exhibited lower Km (0.35 mmol/L) and higher Vmax values (0.163 mg product/min mg protein) indicating that L-tryptophan was a better substrate for AADC than L-phenylalanine.

Figure 5
figure 5

HPLC analysis of L-tryptophan and tryptamine in AADC enzyme reaction at different time points.

Red arrow: L-tryptophan, black arrow: tryptamine.

Figure 6
figure 6

LC-MS result of AADC enzyme reaction with L-tryptophan as substrate.

Figure 7
figure 7

Determination of Km and Vmax values for L-tryptophan (A) and L-phenylalanine (B) as substrate, respectively.

Values are means ± standard deviation.


As shown in Fig. 1, bacillamide C was suggested to be derived from the amino acids alanine, cysteine and tryptophan through nonribosomal peptide synthetase (NRPS) biosynthesis strategy. Decarboxylation and amidation are two important steps in the biosynthesis of bacillamide C. Whether the decarboxylation of L-tryptophan to tryptamine is performed before amidation and which decarboxylase is responsible for the decarboxylaiton are two critical issues in the understanding of the biosynthesis mechanism of bacillamide C.

The 15 decarboxylases in B. atrophaeus C89 have 14 different substrates (Table 1). The suggested substrates and their molecular properties analysis might exclude the possibility of decarboxylation after amidation. According to the location of decarboxylases in the genome of B. atrophaeus C89, we deduce that the aromatic L-amino acid decarboxylase (AADC, JQ400024; id003505), which is the closest downstream decarboxylase to NRPS gene cluster, is most likely involved in the biosynthesis of bacillamide C. This was proved by bioinformatics analysis and AADC gene cloning, expression, enzymatic dynamics and activity tests.

The analysis of amino acid sequence alignment between the AADC and some reported decarboxylases showed that the AADC contained the high conserved identical PLP binding pockets and catalytic residue (lys) of aromatic L-amino acid decarboxylase (Fig. 2). Meanwhile, the molecular property analysis of 14 decarboxylases' substrates (Fig. 3; Table 1 and Table 2) suggested that L-tryptophan and L-phenylalanine are probably the substrates based on the substrates' similarity analysis, L-tryptophan and L-phenylalanine are the possible substrates of aromatic L-amino acid decarboxylase (AADC, JQ400024; id003505). This was supported by the production of bacillamide C and neobacillamide A by B. atrophaeus C89 synchronously9.

The analysis of enzymatic dynamics proved that L-tryptophan was a better substrate for AADC than L-phenylalanine because of its lower Km and higher Vmax values (Fig. 7). This result was consistent with our previous study that the yield of bacillamide C was higher than that of neobacillamide A isolated from B. atrophaeus C899. In contrast, if L-phenylalanine is provided as substrate instead of L-tryptophan, the product obtained will be neobacillamide A. Neobacillamide A may share the similar biosynthetic pathway as bacillamide C, only the substrates for AADC are different.

Lots of literatures have reported that aromatic L-amino acid decarboxylases are involved in the biosynthesis of secondary metabolites including terpenoid indole alkaloids, hydroxycinnamic acid amides, serotonin and phenylacetaldehyde22,23,24. AADCs L-tryptophan decarboxylases have been intensively studied because of their importance in the biosynthesis of pharmaceutical secondary metabolites23,24,25,26. When L-tryptophan was used as the substrate, the Km of AADC in B. atrophaeus C89 was 0.35 mM which is four-fold higher than that of the C. roseus (0.075 mM)25. Whereas, the decarboxylases from Ophiorrhiza pumila26 and rice24 showed two-fold higher Km values (0.72 mM, 0.69 mM, respectively) than that of B. atrophaeus C89 AADC. It was suggested that the affinity of L-tryptophan for AADC in B. atrophaeus C89 is different from that of other microorganisms.

On conclusion, the bioinformatics analysis and in vitro dynamics and activity of AADC indicated that AADC is probably involved in the biosynthesis of bacillamide C through decarboxylation of L-tryptophan to tryptamine. In NRPS modular organization (Fig. 1), the last module is only composed of C domain without A domain. This C domain is responsible for peptide bond formation between tryptamine and the growing peptide chain intermediates, which is catalyzed by the other two modules (A-PCP and C-A-PCP). The knockout of AADC gene from B. atrophaeus C89 will provide the direct evidence in future. Even so, the results from this study suggested that the bacillamide C production by B. atrophaeus C89 could be improved with the addition of L-tryptophan. Most importantly, based on the characterization of AADC, manipulation of AADC's specificity by point mutation of the substrate-coordinating amino acid residues mighty be an approach to generate novel bacillamide compounds in further research.



L-tryptophan and L-phenylalanine were purchased from Sigma Chemical Company (St. Louis, MO, USA). Trans Taq™ DNA polymerase and pEASY™-E1 Expression Kit were purchased from TransGen Biotech (Beijing, China). Plasmid Mini Kit I was purchased from Omega (USA). Ni-NTA agarose was purchased from Qiagen (Qiagen Co., Hilden, Germany). All other chemicals used were analytical grade reagents unless stated.

Strains and plasmids

B. atrophaeus C89 isolated from South China Sea sponge Dysidea avara was identified according to the 16S rRNA gene27 (Genbank No. DQ091007). E. coli Trans1-T1 and BL21 (DE3) strains were used for propagation of plasmids and as host for expression of the decarboxylase gene, respectively. B. atrophaeus C89 was incubated in a medium containing 5 g of beef extract and 10 g of peptone in a liter of artificial seawater28. The culture of B. atrophaeus C89 was maintained at 28°C. E. coli Trans1-T1 and BL21 were grown in Luria-Bertani (LB) medium at 37°C.

Analysis of molecular properties of decarboxylases' substrates

The GenBank accession numbers of 15 decarboxylases' sequences from B. atrophaeus C89 were JQ608455 (id 000589), JQ608456 (id 000869), JQ608457 (id 001001), JQ608458 (id 001372), JQ608459 (id 001860), JQ608460 (id 002294), JQ608461 (id 002972), JQ608462 (id 003331), JQ608463 (id 003347), JQ608464 (id 003445), JQ400024 (id 003505), JQ608465 (id 003545), JQ608466 (id 003690), JQ608467 (id 003691) and JQ608468 (id 003692) (Table 1). Amino acid sequences of the 15 decarboxylases and related homolog protein sequences were analyzed based on the Orthologous Genes (COG) and KEGG databases18. Six different characteristics (H Donor, Num_Aromatic bonds, Volume, Molecular Weight, ALogP29 and Solubility, Table 2) of different deduced substrates were selected to construct molecular structure models using Discovery Studio and the energy minimization of molecular structure models through CHARMM (Chemistry at HARvard Macromolecular Mechanics,,31. ALogP indicated the log of the octanol-water partition coefficient using Ghose and Crippen's method. Molecular solubility was expressed as LogS, where S was the solubility in mol/L. Volume is the quantity of three-dimensional space enclosed by some closed boundary. Molecular weight was calculated as the sum of the individual isotopic masses of all the atoms in any molecule. Hydrogen (H) bonds occur when a “donor” atom donates its covalently bonded hydrogen atom to an electronegative “acceptor” atom. An aromatic bond type is used for the ring bonds of benzenoid systems, 6-membered aromatic heterocycles and cyclopentadienyl systems.

Alignment analysis of AADC

The conserved catalytic domain of AADC (JQ400024, id003505) was analyzed in Conserved Domain Database at the NCBI. Database homology searches of amino acid sequences were carried out using BLASTp in NCBI protein database. Amino acid sequence alignment between B. atrophaeus C89 AADC and reference sequences was carried out by DNAMAN.

Genomic DNA extraction

B. atrophaeus C89 was grown at 28°C for 24 h. Genomic DNA was isolated by the modification of Marmur method32; cells were collected by centrifugation at 12,000 × g for 5 min. The pellets were washed twice with 1 mL of TE buffer (10 mM of Tris, 1 mM of EDTA, pH 8.0) and centrifuged at 12,000 × g for 5 min. The pellets were suspended in 500 μL of TE buffer. The suspended cells were treated with lysozyme (2 mg/mL) at 37°C for 1 h followed by 20% sodium dodecyl sulfate (SDS) and proteinase K (20 mg/mL) at 55°C for 1 h. The lysate was extracted with an equal volume of phenol/chloroform/isoamyl alcohol (25:24:1, V:V:V) twice and washed with an equal volume of chloroform/isoamyl alcohol (24:1, V:V). DNA was precipitated with an equal volume of isopropanol at −20°C for 1 h and harvested by centrifugation at 12,000 × g for 15 min. Finally, DNA pellets were washed with 70% ethanol and resuspended in 30 μL of TE Buffer. The extracted DNA was analyzed on 1.0% agarose gel.

PCR amplification of AADC gene

Forward primer decF0403 (5′-GTGAAACAAGTGTC GGAAAACCTG-3′) and reverse primer decR0403 (5′-TTATTCAGCAACGCATGGATACG-3′) used for amplification of AADC genes were designed using Primer Premier 5.0 based on the AADC sequence in the genome of B. atrophaeus C89. In PCR reaction, two hundred nanogram of template DNA was used in 50 μL of reaction system. The PCR mixture (50 μL) contained 0.2 mM of dNTPs, 0.4 μM of each primer, 5 U of Trans Taq™-T DNA Polymerase (Transgen, China) and 1 × PCR buffer (20 mM Tris-HCl , pH 9.0, 10 mM KCl, 10 mM (NH4)2SO4, 2 mM MgSO4). PCR was carried out as follows: initial denaturation at 94°C for 5 min; 30 cycles of 94°C for 30 s, 55°C for 30 s and 72°C for 90 s; and a final extension of 10 min at 72°C. PCR products were purified with Cycle Pure kit and Gel Extraction kit according to the manufacturer's instructions (Axygen, USA). The purified PCR products were ligated into the pEASY™-E1 Expression Vectors (Transgen, China) and the ligated mixture was transformed into competent E. coli Trans1-T1. Positive recombinants were identified by DNA sequencing with T7 promoter and T7 terminator primer. The recombinant plasmids containing AADC gene were purified using the Plasmid Mini Kit I (Omega, China) and transformed into competent cells of E. coli BL21. Transformed cells were plated on LB-agar plates containing 50 μg/mL of ampicillin and incubated at 37°C overnight.

Expression and purification of recombinant AADC

Five milliliter of LB broth containing 100 μg/mL of ampicillin were inoculated with a single transformed E. coli colony and incubated at 37°C overnight. The 5 mL of cells was inoculated into 500 mL of LB broth containing antibiotic and incubated at 37°C. IPTG and PLP (pyridoxal-5′-phosphate) at final concentration of 0.5 mM and 0.1 mM were then added to the culture medium when OD600 of the culture medium was 0.4–0.6, respectively. The incubation was continued at 28°C for 16 h. After centrifugation at 12000 × g, 4°C for 20 min, the cells pellets were suspended in lysis buffer at 2–5 mL per gram wet weight and incubated on ice with lysozyme (1 mg/mL), PMSF (1 mM) and PLP (0.5 mM) and disrupted by sonication. The clear supernatant was collected by centrifugation at 12000 × g and applied to Ni-NTA agarose resin column equilibrated with lysis buffer (pH 8.0) containing 50 mM of NaH2PO4, 300 mM of NaCl and 10 mM of imidazole. The enzyme was eluted with 20, 150 and 250 mM of imidazole in 50 mM of phosphate buffer (pH 8.0) containing 300 mM of NaCl. Recombinant decarboxylse was eluted with the above buffer by gravity. The purified enzyme was monitored by Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) in a 12.0% (w/v) polyacrylamide gel. Protein molecular weight standards in the range of 14.4–116.0 kDa (Fermentas, USA) were used as marker. Protein bands were visualized by Coomassie brilliant blue R-250 staining33.

Activity and dynamics analyses of AADC

AADC activity was assayed according to the slightly modified method of Pennings34. The standard mixture (0.2 mL; final pH 8.0 at 35°C) contained 0.1 M of Tris-HCl buffer, 1 mM of L-tryptophan or 1 mM of L-phenylalanine, 0.4 mM of pyridoxal-5′-phosphate, 3 mM of dithiothreitol and 0.05 mL of the suitably diluted enzyme solution. A blank was prepared by placing the mixture in boiling water for 5 min. The incubation was started by placing the tubes in a water bath at 35°C for 1 h and terminated by adding an equal volume of methanol. The amount of products formed was measured by HPLC and calculated from relative calibration curves. One unit of decarboxylase activity was defined as the amount of enzyme that produced 1 μM of tryptamine or phenylethylamine equivalent per minute34. Protein concentrations were measured by Nano Vue using BSA (bovine serum albumin) as the standard.

The purified decarboxylase was incubated with nine different concentrations (from 0.2 mM to 5.0 mM) of each substrate (tryptophan and phenylalanine) respectively in 0.1 M Tris-Hcl buffer (pH 8.0) at 35°C for 30 min. The Km and Vmax values were calculated by the relative Lineweaver-Burk plot.

For the detection of changes from L-tryptophan to tryptamine during the reaction process, the amount of tryptamine in reaction sample was assayed at different reaction times (0 min, 15 min, 30 min, 60 min and 120 min) by HPLC (Aglient 1100 Series, ZORBAX Eclipse XDB-C18 column (4.6 × 150 mm, 5 μm), mobile phase: methanol/H2O (70/30,V/V, 7 mM SDS,10 mM KH2PO4), 25°C, velocity of flow: 1 mL/min, detection time: 0–11 min, sample size: 20 μL, detection wavelength: 278 nm). The reaction samples were analyzed by LC/MS using an Aglient 1200 Series LC system and 6140 quadrupole MS. Mobile phase A was water with 0.0375% TFA and mobile phase B was ACN with 0.01875% TFA. The detailed mobile phase gradient was shown as follows: 0–0.4 min, 0% B; 0.4–3.4 min, 0–80% B; 3.4–4.0 min, 80%–100% and 4.0–4.5 min, 0% B, while the flow rates were 0.6 mL/min, 0.6 mL/min, 0.8 mL/min and 0.8 mL/min, respectively. The mass spectrometer was operated in a positive ion electrospray ionization mode (capillary potential 3.5 kV, desolvation temperature 350°C, source temperature 100°C).