Caerulomycin and collismycin antibiotics share a trans-acting flavoprotein-dependent assembly line for 2,2'-bipyridine formation

Abstract

Linear nonribosomal peptide synthetases (NRPSs) and polyketide synthases (PKSs) template the modular biosynthesis of numerous nonribosomal peptides, polyketides and their hybrids through assembly line chemistry. This chemistry can be complex and highly varied, and thus challenges our understanding in NRPS and PKS-programmed, diverse biosynthetic processes using amino acid and carboxylate building blocks. Here, we report that caerulomycin and collismycin peptide-polyketide hybrid antibiotics share an assembly line that involves unusual NRPS activity to engage a trans-acting flavoprotein in C-C bond formation and heterocyclization during 2,2'-bipyridine formation. Simultaneously, this assembly line provides dethiolated and thiolated 2,2'-bipyridine intermediates through differential treatment of the sulfhydryl group arising from L-cysteine incorporation. Subsequent L-leucine extension, which does not contribute any atoms to either caerulomycins or collismycins, plays a key role in sulfur fate determination by selectively advancing one of the two 2,2'-bipyridine intermediates down a path to the final products with or without sulfur decoration. These findings further the appreciation of assembly line chemistry and will facilitate the development of related molecules using synthetic biology approaches.

Fig. 1. Biogenesis of the polyketide-peptide hybrid, 2,2’-bipyridine antibiotics CAE-A and COL-A.

a Organization of related genes in the CAE and COL biosynthetic gene clusters that codes for the hybrid NRPS/PKS assembly lines, the associated trans-acting flavoproteins and other proteins. The didomain encoding gene caeA1 in the cae cluster corresponds to the two discrete genes, colA1a and colA1b, in the col cluster. Gene functions are annotated by color on the left. ID, sequence identity. b Hybrid NRPS/PKS assembly line for 2,2’-bipyridine formation. PKS and NRPS modules are indicated. The functional domains (i.e., AT and A domains) for substrate specificity are indicated in different colors with their associated building blocks: blue for picolinyl, yellow for malonyl, red for l-cysteinyl, and light gray for l-leucinyl.

Fig. 2. In vitro reconstitution of the 2,2’-bipyridine assembly line.

2,2’-Bipyridine products were examined by HPLC (λ 315 nm). Each assay was performed at least three times, and every time at least two parallel samples were used. a Determination of the trans-acting partner. In the reaction mixture where the proteins CaeA1, CaeA2, and CaeA3 were combined with the substrates picolinic acid, malonyl-CoA, l-cysteine, and l-leucine as well as ATP (i), CaeB1 (ii), CaeA4 (iii), or both CaeB1, and CaeA4 (iv) were added individually. Synthetic 1 was used as a standard (v). b Validation of the necessity of the Ct and Cy domains in the atypical NRPS module for 2,2’-bipyridine formation. In the above CaeB1-containing, 1 producing reaction mixture, wild-type CaeA2 was replaced with truncated CaeA2ΔCt (i) and mutated CaeA2_D1165A (ii), respectively. c Determination of the similarity and difference in the formation of dethiolated and thiolated 2,2’-bipyridines by protein/domain exchange. In the above CaeB1-containing, 1 producing reaction mixture, CaeB1 was replaced with ColB1 (i), CaeA3 was replaced with ColA3 (ii), CaeA3 was replaced with ColA3 and additional ColG2 (iii), ColG2 was added (iv), and CaeA3 was replaced with CaeA3_CColA3_A-PCP (v). Synthetic 16 was used as a standard (vi).

Fig. 3. Examination of PCP-tethered intermediates during 2,2’-bipyridine formation (top) by nanoLC-MS/MS (below).

The sequence SLGGDSIMGIQL₂₀₄₂VSR (in the rectangle) arises from the complete digestion of CaeA2^F2042L with trypsin and chymotrypsin. For details of the HR-MS/MS analyses, see Supplementary Figs. 5~7, 10, and 13. The two IAA-treated sequences (ESI m/z [M + 3H]³⁺ for the left, calcd. 678.3143; for the middle, calcd. 677.9704) come from l-cysteinyl-S-CaeA2^F2042L (3) and dehydrocysteinyl-S-CaeA2^F2042L (7), respectively, which were examined by incubating CaeA2^F2042L and l-cysteine in the absence (i, negative control) and presence (ii) of ATP, with CaeB1 (iii), CaeB1 and CaeA2^F2042LΔCt (replacing CaeA2^F2042L) (iv), or CaeB1, CaeA1, picolinic acid and malonyl-CoA (v). The right sequence (ESI m/z [M + 3H]³⁺, calcd. 690.9854) comes from 2,2’-bipyridinyl-S-CaeA2^F2042L (9). The positive control was prepared by incubating CaeA2^F2042L (with PCP in apo form) and 2,2’-bipyridinyl-S-CoA (i). 9 was examined by incubating CaeA2^F2042L and l-cysteine (ii), with CaeB1 (iii), CaeB1, CaeA1, picolinic acid, and malonyl-CoA (iv), or CaeB1, CaeA1, CaeA3, picolinic acid, malonyl-CoA and l-leucine (v). All examinations were performed at least in triplicate, and each had at least two parallel samples.

Fig. 4. Measurement of the interactions of the Ct domain of CaeA2 with related flavoproteins by ITC.

Raw data were shown on top, and the integrated curves containing experimental points and the best fitting line obtained from the single binding site model were shown on the bottom. This measurement was conducted in triplicate. For negative controls, see Supplementary Fig. 11. a Titrating MBP-fused CaeB1 to Trx-tagged Ct. b Titrating ColB1 to Trx-tagged Ct.

Fig. 5. Proposed mechanisms for 2,2’-bipyridine formation.

The PCP domain of CaeA2 is labeled in red. The C domains of CaeA3 and ColA3 are highlighted in green and blue, respectively. In contrast to favored routes b and c, unfavored route a is shown by gray color. Fl_ox oxidized flavin, Fl_red two-electron reduced flavin.

Fig. 6. Domain organization of the NRPSs CaeA2, ColA2, and ColA2_PKS-CaeA2_NRPS (Top) and HPLC analysis of COL-related products in S. roseosporus strains (below).

i, the wild-type COL-producing strain (positive control); ii, the ΔcolA2 mutant in which a chimeric gene colA2_PKS-caeA2_NRPS is expressed in trans; and iii, the ΔcolA2 mutant (negative control).