The Landscape of Recombination Events That Create Nonribosomal Peptide Diversity

Abstract

Nonribosomal peptides (NRP) are crucial molecular mediators in microbial ecology and provide indispensable drugs. Nevertheless, the evolution of the flexible biosynthetic machineries that correlates with the stunning structural diversity of NRPs is poorly understood. Here, we show that recombination is a key driver in the evolution of bacterial NRP synthetase (NRPS) genes across distant bacterial phyla, which has guided structural diversification in a plethora of NRP families by extensive mixing and matching of biosynthesis genes. The systematic dissection of a large number of individual recombination events did not only unveil a striking plurality in the nature and origin of the exchange units but allowed the deduction of overarching principles that enable the efficient exchange of adenylation (A) domain substrates while keeping the functionality of the dynamic multienzyme complexes. In the majority of cases, recombination events have targeted variable portions of the Acore domains, yet domain interfaces and the flexible Asub domain remained untapped. Our results strongly contradict the widespread assumption that adenylation and condensation (C) domains coevolve and significantly challenge the attributed role of C domains as stringent selectivity filter during NRP synthesis. Moreover, they teach valuable lessons on the choice of natural exchange units in the evolution of NRPS diversity, which may guide future engineering approaches.

Keywords: evolution; microbial ecology; natural products; nonribosomal peptide synthetases; recombination; structural diversity.

Fig. 1.

Evolution of microcystin diversity by recombination, DNA deletion, and point mutations. (a) Compared with microcystin-LR, [Dha⁷] microcystin-LR is produced by strains carrying a mcyA gene in which a segment encoding a N-methyltransferase got deleted (Fewer et al. 2008). (b) Recombination in the segment encoding the first A domain of mcyA likely gave rise to strains producing [Dhb⁷] microcystin-LR (Kurmayer et al. 2005). (c) Recombination in the segment encoding the second A domain of mcyA likely gave rise to strains producing [D-Leu¹] microcystin-LR. Alternatively, point mutations have led to the same chemotype (Shishido et al. 2013). (d) Deletion of two modules encoded by mcyA and mcyB likely was involved in the evolution of microcystin-like nodularins (Rantala et al. 2004). (e) An intragenomic recombination event between mcyB and mcyC likely gave rise to the evolution of strains producing microcystin-RR (Fewer et al. 2007; Tooming-Klunderud et al. 2008; Meyer et al. 2016). Gene segments encoding modules are divided into adenylation (A), condensation (C), thiolation (T), and, if present, methylation (MT) domains. (M)dha, (N-methyl) dehydroalanine; Dhb, dehydrobutyrine.

Fig. 2.

Diversification of cyanobacterial NRPs via recombination in the biosynthesis of (a) microcystins, (b) microginins, (c) anabaenopeptins, (d) spumigins, and (e) anabaenolysins. Structural differences between pairs from compound families (gray squares) correlate with nucleotide sequence divergence of the genes encoding NRPS modules (M). Related sequences have been aligned for pairwise comparison. π values (average number of nucleotide differences per site between two sequences) were computed using the sliding window mode in DnaSP (width, 300 nt; step, 150 nt). The mosaic structure of the genes (Smith 1992) clearly indicates recombination. This notion is also strongly supported by the detection of gene segments that complement divergent sites in a reciprocal fashion (numbered bullet points [BP] 1–6). Notably, the complement sequences stem from modules of the same cluster (BP 1, 4), from different clusters of the same species (BP 3), or from different clusters of different species (BP 2, 5, 6). Amino acid residues in the structures are color-coded to trace back their biosynthetic origin to individual modules. Hty, homotyrosine; Hph, homophenylalanine; mPro, 4-methylproline; mAsp, 3-methylaspartic acid; Te, thioesterase, R, reductive domain. (f) Close-up representation of putative recombination events to evaluate exchange unit boundaries. Gene segments encoding modules are divided into adenylation (A), condensation (C), thiolation (T), and, if present, methylation (MT) domains. Adenylation domain-specific core motifs are indicated by bands and numbers (1–10) (Marahiel et al. 1997). Linkers are indicated as filled squares. Highlighted parts of the graphs represent regions that are more closely related to sequences encoding other modules than to sequence of the respective ortholog.

Fig. 3.

Diversification of Ahp-cyclodepsipeptides via recombination. (a) Structural differences of ahpcyclodepsipeptides (gray squares) correlate with nucleotide sequence divergence of the genes encoding NRPS modules (M). Closely related sequences have been aligned for pairwise comparison. π values (average number of nucleotide differences per site between two sequences) were computed using the sliding window mode in DnaSP (width, 300 nt; step, 150 nt). The mosaic structure of the genes (Smith 1992) clearly indicates recombination. This notion is also strongly supported by the detection of gene segments that complement divergent sites in a reciprocal fashion (BP 7–13). Notably, the complement sequences stem from modules of the same cluster (BP 8, 11, and 13), from related clusters of different species (BP 7 and 10) or from different clusters of different species (BP 9 and 12). Amino acid residues in the structures are color-coded to trace back their biosynthetic origin to individual modules. Ahp, 3-amino-6-hydroxy-2-piperidone; Hty, homotyrosine; Hmp, 3-hydroxy-4-methylproline; Te, thioesterase. (b) Close-up representation of putative recombination events to evaluate exchange unit boundaries. Gene segments encoding modules are divided into adenylation (A), condensation (C), and thiolation (T) domains. Adenylation domain-specific core motifs are indicated by bands and numbers (1–10) (Marahiel et al. 1997). Linkers are indicated as filled squares. Highlighted parts of the graphs represent regions that are more closely related to sequences encoding other modules than to sequence of the respective ortholog.

Fig. 4.

Diversification of noncyanobacterial NRPs via recombination. Putative recombination events in the biosynthesis of (a) iturinic lipopeptides and (b) polymyxins. Structural differences of NRPs (gray squares) correlate with nucleotide sequence divergence of the genes encoding NRPS modules (M). Closely related sequences have been aligned for pairwise comparison. π values (average number of nucleotide differences per site between two sequences) were computed using the sliding window mode in DnaSP (width, 300 nt; step, 150 nt). The mosaic structure of the genes (Smith 1992) clearly indicates recombination. This notion is also strongly supported by the detection of gene segments that complement divergent sites in a reciprocal fashion (numbered bullet points [BP] 14–18). Notably, the complement sequences stem from modules of the same cluster (BP 14 and 18), from related clusters of different species (BP 15), from different clusters of the same species (BP 17), or from different clusters of different species (BP 16). Amino acid residues in the structures are color-coded to trace back their biosynthetic origin to individual modules. Dab, diaminobutyric acid; Te, thioesterase; R, alkyl moiety. (c) Close-up representation of putative recombination events to evaluate exchange unit boundaries. Gene segments encoding modules are divided into adenylation (A), condensation (C), and thiolation (T) domains. Adenylation domain-specific core motifs are indicated by bands and numbers (1–10) (Marahiel et al. 1997). Linkers are indicated as filled squares. Highlighted parts of the graphs represent regions that are more closely related to sequences encoding other modules than to sequence of the respective ortholog.

Fig. 5.

Visualization of exchange unit boundaries in NRPS modules. (a) Schematic visualization of the deduced exchange units (supplementary figs. S18 and S19, Supplementary Material online) that most likely result from a single recombination event (checked pattern). Modules are divided into adenylation (A), condensation (C), thiolation (T) domains, and linkers (L). Adenylation domain-specific core motifs are indicated by numbers 1–10 (Marahiel et al. 1997). Modules that possess an additional methyltransferase (MT) domain between core motif 8 and 9 are marked with an asterisk. The plurality of exchange unit boundaries indicates a pronounced plasticity of the Acore domain, which provides multiple breakpoints for subdomain swaps to be harnessed by evolution. (b) Projection of the deduced exchange units on the structure of SrfA–C (Tanovic et al. 2008) illustrates the obvious trend to keep the native C–A linker, the Asub domain and consequently the Asub–T domain interface intact.

Fig. 6.

Unifying model for the evolution of the present-day variety of NRPs (simplified with amino acids as beads on a string) using the example of cyanobacteria. (a) Intragenomic recombination in last common ancestors that harbored a variety of NRPS gene clusters led to the diversification of multiple compound families. After speciation, many clusters have been lost in individual species due to genome streamlining. This could explain the patchy distribution of NRPS families as well as the unprecedented, network-like mosaic structure of NRPS genes, which is exemplified by the high proportion of putative recombination pairs from different clusters of different species or even different genera (fig. 2, BP2, 5, and 6; fig. 3, BP9 and 12). Similarly, intragenomic recombination in present-day species continuously contributes to generation of structural variants (e.g., fig. 2, BP1 and 4; fig. 3, BP8, 11, and 13). (b) Additionally, horizontal gene transfer (HGT) together with recombination likely drives the diversification of NRPS families, either between related clusters of different species (e.g., fig. 3, BP7 and 10) or between different clusters of different species.