RESEARCH ARTICLE SUMMARY◥

BIOSYNTHETIC ENZYMES

Structures of a dimodular nonribosomal peptidesynthetase reveal conformational flexibilityJanice M. Reimer*, Maximilian Eivaskhani*, Ingrid Harb, Alba Guarné,Martin Weigt, T. Martin Schmeing†

INTRODUCTION:Nonribosomal peptide synthe-tases (NRPSs) are microbial megaenzymesthat make a wide variety of small-moleculeproducts, including many that are clinicallyused as antitumors, antibiotics, or immuno-suppressants. Nonribosomal peptide synthesisproceeds with assembly-line logic, where eachstation on the NRPS assembly line is a multi-domain unit called a module. An excellentunderstanding of the structures and activitiesof isolated modules has been established, butmuch less is known about howmodules workwith each other in the context of the largerNRPS. Structural investigation of multimod-ular NRPSs is needed to understand NRPSarchitecture, organization, and intramodularfunction during the synthetic cycle of anNRPSand to facilitate the longstanding goal of bio-engineering for production of new-to-naturebioactive small molecules.

RATIONALE: To gain insight into outstandingtrans- and supermodular questions in NRPSfunction, we performed x-ray crystallographywith a series of constructs of the dimodularNRPS protein linear gramicidin synthetasesubunit A (LgrA). We performed complemen-

tary small-angle x-ray scattering experimentsto analyze the behavior of the NRPS in so-lution. We also performed direct couplinganalysis to confirm the biological relevanceand evolutionary conservation of observedinterdomain interfaces. Both the structuresand direct coupling analyses were used toguide mutagenesis studies designed to en-hance the activity of a chimeric NRPS.

RESULTS: We have determined five indepen-dent crystal structures of constructs of LgrA,bound with a series of ligands and interme-diate analogs, to resolutions between 2.2 and6 Å. The crystallized constructs include thecomplete initiation module and from one toall three canonical domains from the elonga-tionmodule. Some structures are inmarkedlydifferent conformations, inferring large move-ments, and each structure seems to be in acatalytically relevant state. Small-angle x-rayscattering indicates that LgrA is also very flex-ible in solution, confirming that markedly diff-erent conformations are a bona fide feature ofNRPS biology. The structures reveal previouslyunobserved states, including a full condensa-tion conformation, where the thiolation (T)

domains from both the initiation and elon-gation modules are simultaneously bound atthe condensation (C) domain. Similar confor-mations in high-resolution structures allowanalyses of the productive T:C domain-domain interface, which mediates the only

known functional linkbetween modules. Directcoupling analysis appliedto large collections ofNRPSsequences provides strongsupport for the biologicalrelevance and evolution-

ary conservation of observed interdomain in-terfaces. Furthermore, both the structures andcoupling scores for mutational effects wereused to guide bioengineering, and we wereable to double the activity of amodule-swappedchimeric NRPS by introducing two point mu-tations at the unnatural T:C domain-domaininterface.

CONCLUSION: The structures and small-anglex-ray scattering show NRPSs undergo verylarge conformational changes and challengethe general assumption that NRPSs have regu-lar higher-order architecture. Theydemonstratethat there is no strict coupling between the cat-alytic state of a particular module and the over-all conformation of themultimodularNRPS andsuggest that the T:C interaction for condensa-tion is the only point where adjacentmodulesmust coordinate. This feature can be exploitedinmodule-swappingbioengineering to producenew useful nonribosomal peptides.▪

RESEARCH

Reimer et al., Science 366, 706 (2019) 8 November 2019 1 of 1

The list of author affiliations is available in the full article online.*These authors contributed equally to this work.†Corresponding author. Email: martin.schmeing@mcgill.caCite this article as J. M. Reimer et al., Science 366,eaaw4388 (2019). DOI: 10.1126/science.aaw4388

Module 1 Module 4’ Activity Activity

Module 1

Module 2

Module 2

Module 1 Module 1Module 1

Module 2

Module 2

Structures of a dimodular NRPS protein reveal the central condensationstate and infer very large conformational changes. A series of crystalstructures of the dimodular nonribosomal peptide synthetase protein LgrAincludes a structure of the condensation state (left). Condensation is the centralevent in synthesis, elongating the peptide intermediate and passing it to thedownstream module. Additional structures in condensation and thiolation states

show large conformational differences (indicated by arrows), which aresupported by solution small-angle x-ray scattering data. These structuresshow decoupling of the catalytic state and overall conformation and implythat coordination of adjacent modules’ catalytic states is only required atcondensation. The structures and coevolution analyses enable improvementof activity of a module-swapped chimeric enzyme (bottom left).

ON OUR WEBSITE◥

Read the full articleat http://dx.doi.org/10.1126/science.aaw4388..................................................

on April 17, 2021

http://science.sciencem

ag.org/D

ownloaded from

RESEARCH ARTICLE◥

BIOSYNTHETIC ENZYMES

Structures of a dimodular nonribosomal peptidesynthetase reveal conformational flexibilityJanice M. Reimer1*, Maximilian Eivaskhani1*, Ingrid Harb1, Alba Guarné1,Martin Weigt2, T. Martin Schmeing1†

Nonribosomal peptide synthetases (NRPSs) are biosynthetic enzymes that synthesize natural producttherapeutics using a modular synthetic logic, whereby each module adds one aminoacyl substrateto the nascent peptide. We have determined five x-ray crystal structures of large constructs of the NRPSlinear gramicidin synthetase, including a structure of a full core dimodule in conformations organizedfor the condensation reaction and intermodular peptidyl substrate delivery. The structures revealdifferences in the relative positions of adjacent modules, which are not strictly coupled to the catalyticcycle and are consistent with small-angle x-ray scattering data. The structures and covariationanalysis of homologs allowed us to create mutants that improve the yield of a peptide from a module-swapped dimodular NRPS.

Nonribosomalpeptide synthetases (NRPSs)are intricate macromolecular machinesthatmake small-molecule productswithvery high chemical diversity and activity(1). Compounds made by NRPSs have

found widespread clinical use and are on listsof United Nations–designated essential medi-cines (2) and top-selling pharmaceuticals (3).Nonribosomal peptide synthesis usesmodular,assembly-line logic where each multidomainmodule adds one amino acid substrate to thegrowing peptide (Fig. 1) (4). A module’s ade-nylation (A) domain selects and activates theamino acid and then covalently attaches itas a thioester to the thiolation (T) domain’sphosphopantetheine (ppant) arm. The con-densation (C) domain catalyzes peptide bondformation between that aminoacyl-T domainand the donor peptidyl-T domain from theupstreammodule (5, 6). The newly elongatedpeptidyl-T domain is then the donor substratefor condensation in the downstreammodule,passing off and further elongating the peptide.Many modules also have tailoring domainsintegrated within them, which cosyntheticallymodify the nonribosomal peptide, such as thetailoring formylation (F) domain (7) found inthe initiationmodule of theNRPS studied here,linear gramicidin synthetase (Fig. 1).An excellent structural understanding of the

synthetic cycle of isolated modules has beengained from structures of domains, didomains,and individual modules [reviewed in (4, 8)].However, modules typically function within

the context of the full NRPS. They are phys-ically attached to their neighbors by flexiblepeptide linkers or through small docking do-mains (9). Adjacentmodulesmust functionallycoordinate at least once during the syntheticcycle, when the C domain catalyzes peptidebond formation between aminoacyl and pep-tidyl moieties attached to T domains of adja-cent modules. Little else is known about howmodules work with each other in the con-text of the larger NRPS. Two previous high-resolution structures contain domains fromadjacent modules: The T5C6 didomain of tyro-cidine synthetase is in an unproductive con-formation (10), and A1T1C2 of bacillibactin

synthetase (11) showed that the sole observedintermodule contactmust break in the courseof peptide synthesis. Structural data for mul-timodular NRPS are limited to 26- to 29-Ånegative-stain electron microscopy recon-structions of two modules of bacillibactinsynthetase (CATCAT), which showed hetero-geneity in the module:module conformation(11). Hypothetical models of multimodularNRPSs can be constructed by consecutivelyoverlappingmultidomain structures from dif-ferent synthetases, and the models often takethe form of rigid superhelices (4, 12), but thereis no evidence that any of these conformationsoccur in vivo. More data are needed to under-stand NRPS architecture, organization, andintramodular function during the syntheticcycle of an NRPS and to facilitate their use tomake new-to-nature compounds.

ResultsCrystallography of five large NRPS constructs

Linear gramicidin synthetase is a 16-module,4-protein NRPS which makes the clinicallyused eponymous antibiotic (Fig. 1) (13). Theantibiotic acts by forming dimeric b-helicalpores in Gram-positive bacterial membranes,which kills the bacteria by allowing free pas-sage of monovalent cations across the mem-brane (14). To gain insight into outstandingtrans- and supermodular questions in NRPSfunction, we undertook more than 100,000crystallography screening trials with con-structs of linear gramicidin synthetase sub-unit A (LgrA, with domains FATCATE0; Fig. 1)complexed with substrates, substrate ana-logs, and dead-end inhibitors. This yielded fivestructures: two structures of the four-domain

RESEARCH

Reimer et al., Science 366, eaaw4388 (2019) 8 November 2019 1 of 7

1Department of Biochemistry and Center de Recherche enBiologie Structurale, McGill University, Montréal, QC H3G0B1, Canada. 2Sorbonne Université, CNRS, Institut deBiologie Paris-Seine, Laboratory of Computational andQuantitative Biology, F-75005 Paris, France.*These authors contributed equally to this work.†Corresponding author. Email: martin.schmeing@mcgill.ca

SH SH

Val-AMP Gly-AMP Gly-SVal-S

Adenylation2 ATP

Val, Gly2 PPi

A

Thiolation2 AMP

A

FormylationF1THF

fTHF

Gly-SfVal-S

Peptide Transfer(Condensation)

C2

Peptide Transferto LgrB

O

HN

O

NH

O

HN

O

NH

O

HN

O

NH

NH

O

HN

O

NH

NH

O

HN

O

NH

NH

O

HN

O

NH

NH

O

HN

OHNH

OHN

ONH

O

Gramicidin A

SH

A1 T1

SH

A2 T2C2 Eo2F1

Module 1 Module 2A

B

C3fVal-Gly-Ala-S

3

3

Ala-SfVal-Gly-S

SH

Fig. 1. Overview of the biosynthetic steps performed by LgrA. (A) LgrA is a dimodular protein withinitiation (F1A1T1) and elongation modules (C2A2T2E

o2). Valine and glycine are selected and adenylated by

A1 and A2 and then transferred to T1 and T2. Val-T1 is formylated by F1, and then peptide bond formationbetween fVal-T1 and Gly-T2 by C2 produces fVal-Gly-T2. This is the donor substrate for peptide bondformation in the C3 domain of the next NRPS subunit, LgrB. (This schematic is not intended to indicatethe timing of rebinding of substrates.) Eo, inactive epimerization domain; ATP, adenosine triphosphate;AMP, adenosine monophosphate; PPi, inorganic pyrophosphate; THF, tetrahydrofolate. (B) Chemical structureof linear gramicidin A, with a box highlighting the fVal-Gly portion assembled by LgrA.

on April 17, 2021

http://science.sciencem

ag.org/D

ownloaded from

construct FATC in peptide donation confor-mation (FATfValC and FATfValC*; Fig. 2A andfig. S1A); one of FATCA in peptide dona-tion conformation (FATCA; Fig. 2B); oneof FATCA in two thiolation conformations(FATVadCA, where Vad is valinyl-adenosine-vinylsulfamonamide; Fig. 2, C and D); andone of the full dimodule FATCAT in overallcondensation conformation (FATCAT; Fig. 2E)(figs. S1 and S2 and table S1). In every struc-ture, each domain assumes its canonical forms(4, 8): The F domain bears the formyltrans-ferase catalytic domain fold; the C domain isa V-shaped pseudodimer of chloramphenicolacetyltransferase-like lobes with a tunnel toits active site; each A domain has its majorportion, which includes the amino acid bind-ing site (Acore) and its mobile C-terminal sub-domain (Asub); and the T domains are smallfour-helix bundles with prosthetic ppant arms.As examined below, together these structuresof LgrA demonstrate three general features ofNRPS architecture: (i) The didomain structuralunit (FAcore or CAcore) of an NRPS modulelargely maintains its overall conformation(15–19), (ii) the small domains (T and Asub)move according to the catalytic state (15–20),and (iii) observed in detail here, the relative

orientations of adjacent modules in an NRPScan vary markedly.

Module conformation varies between structures

The main structural units of the modules arethe FAcore or CAcore didomains. The currentstructures include 11 crystallographically inde-pendent FAcore or CAcore didomains,more thandoubling the number available (15–19) (Figs. 2and 3A). These show the didomains in eachmodule as “catalytic platforms” (15) that pres-ent the binding site for each module’s T do-mains (the F1 and A1 active sites for T1 and theC2 acceptor site and A2 active site for T2) onthe same face to facilitate substrate delivery.The didomains are fairly rigid, because theF:A or C:A configurations shift by only ~1° to12°, propagating to ~10 Å (Fig. 3 and figs. S3and S4). In FATCA, there are few crystal con-tacts at the distal end of A2, and variation inthe C:Acore orientation from unit cell to unitcell is evident from progressively increasingB-factors and weaker electron density at thedistal end of A2. Notably, there is substan-tiallymore variation in C:Acore conformationsbetween different NRPSs than in a singleNRPS: A2core superimposition withmodules ofenterobactin, AB3403, and surfactin synthe-

tases places some equivalent Cdomain residues>20 Å apart, because of variations of the C:Ainterface and “openness” (21) of the V shape ofthe C domain (figs. S4 and S5).The positions of the Asub and T domains do

vary depending on the catalytic state (Fig. 3A)(15–20). As further explored below, T1 is boundat the donor site of C2 in four structures, andin three of these (FATfValC, FATfValC*, andFATCAT), A1sub is bound to A1core in the ade-nylation conformation (Fig. 2, A and E, and fig.S1A). The simultaneous positioning of T1 forcondensation andA1 for adenylation reiteratesthat NRPSs can start a second synthetic cyclebefore finishing the first (Fig. 1) (18). In bothFATVadCA molecules, T1 and A1sub are boundto A1core in the thiolation conformation (Figs.1 and 2, C and D) (17, 22, 23). This means thatFATfValC and FATVadCA represent consecutivesteps in synthesis (Fig. 1). To move betweencatalytic conformations observed here, Asub

rotates up to ~151° and translates up to ~17 Å,and T1 rotates up to ~153° and translates upto ~47 Å (fig. S6) (22, 24). These transitions areas large as those that Asub and T1 require tomove between their positions in the rest ofthe synthetic cycle, for example to and fromformylation conformation (17).

Reimer et al., Science 366, eaaw4388 (2019) 8 November 2019 2 of 7

FATVadCAMolecule 1

A2core

F1

A1core

C2

A1sub

A2sub

T1

FATfValCfTHF, AMPcPP, valine

FATVadCAMolecule 2

A1core

A2sub

T1

FATV dCAC

A2core

F1

A1core

C2A1sub

T1

A2sub

A

C D

A1

T1C2

F1

FATCA

T1

F1

A1sub

A1core

C2

A2core

A1 T1 A2 T2C2F1

A1subA1core A2subA2core

BT1

F1

A1sub

A1core

C2

A1

T1

C2

F1

A2

A1

T1

C2

F1

A2

A1

T1C2

F1

A2

FATCATMolecule 1

T1

T2

F1

C2

A1sub

A1core

A2sub A2coreE

A1

T1C2

F1

A2T2

Fig. 2. Crystal structures of dimodular LgrA. (A to E) Structures of LgrA constructs. FATfValC, which hasfVal-amino-ppant on T1 and is bound with valine, AMPcPP, and fTHF (space group P212121; 2.5-Å resolution)(A); FATCA (P212121; 2.5-Å resolution) (B); two crystallographically independent molecules of FATVadCA,for which dead-end Vad were used to stall T1 at A1 during thiolation (P212121; 6-Å resolution) [(C) and (D)]; andFATCAT, which has ppant on both T domains (C2221; 6-Å resolution) (E). See fig. S1, A and B, for additionalstructures. Cartoon insets show schematics depicting protein constructs that were crystallized. Domains whichare grayed in the labels are disordered in the crystal structure.

A

B

Module 2

Module 1

~2 - 12°

~1-8°

A1 T1 A2 T2C2F1

A1 T1 A2 T2C2F1

F1

A1sub

T1

A1core

T2 A2sub

A2core

C2

Fig. 3. Comparison of intramodular conformations.(A and B) The structural unit for the initiationmodule (F1A1core) (A) and elongation module (C2A2core)(B) are in similar conformations in all structures.The F1:A1core interface buries 773 to 860 Å2 and the C2:A2core interface buries 565 to 770 Å2 of surface area.Superimposing each structure by their Acores showsa ~1° to 12° shift of F1 or C2. Previously, EntF C1:A1 wasseen to shift ~15° (18) between structures.

RESEARCH | RESEARCH ARTICLEon A

pril 17, 2021

http://science.sciencemag.org/

Dow

nloaded from

Multiple structures show condensation andsubstrate donation conformationsCondensation is the central chemical eventof peptide synthesis. It requires that donorT domain (here T1) and acceptor T domain(here T2) bind simultaneously to the C domain(Fig. 4). The structure of FATCAT featuresthe full condensation state, with both T1 andT2 bound to C2, and represents a detailedthree-dimensional view of a multimodularNRPS (Fig. 4A). The resolution of this struc-ture is 6 Å, but the high-resolution structuresof F1, A1, T1, C2, A2core, and homology modelsof the ~100-residue A2sub and ~90-residue T2enabled the building of a high-quality struc-ture for the full, 1800-residue dimodule (fig.S1). T2 occupies the acceptor binding site onC2, located near helices a1 and a10 (15, 18)(Fig. 4B) and positions the phosphate of itsppant arm at the entrance of the C domainactive site tunnel (fig. S2H). This T2 positionagrees with our direct coupling analysis (DCA)(25) and that in AB3404 (18) but is rotated by

~55° from that in surfactin synthetase (15) (fig.S7, A to C, and table S2).Four structures (FATfValC, FATfValC*, FATCA,

and FATCAT) showT1 binding to the donor siteof C2 (Fig. 4, B and C). This canonical Tn:Cn+1interaction is the functional link between mo-dules 1 and 2, which allows the nascent peptideto be elongated and passed downstream inthe condensation reaction. The donor site isa shallow depression between helices a4 anda9 on the opposite side of the C domain tunnelfrom the acceptor site (Fig. 4, B and C). EachLgrA donor structure has slightly differentresidue-level T1:C2 contacts, all dominated byvan der Waals interactions, which shift distalT1 residues up to ~3.5 Å (Fig. 4B and fig. S7, Dand E). DCA of the Tn:Cn+1 interaction showeda strong coevolution signal between the areas ofT1 and C2 that we observe in direct contact (Fig.4C and table S3). We established a multiple-turnover peptide synthesis assay by fusingFATCAT to the terminal C (CT) domain ofbacillamide synthetase, which catalyzes peptide

release by condensation with free tryptamine(Tpm) (26, 27). This FATCAT-CT construct pro-duces fVal-Gly-Tpm tripeptide (Fig. 4D). Wethen usedDCA, the capacity of which to predictmutational effects in proteins has recently beenestablished (28), to guide mutational analysisof the T1:C2 interface. Of four mutations in C2predicted to be deleterious for the T1:C2 inter-action but not for C2 folding, three showedmoderate, but significant, decrease in tripeptideproduction (Fig. 4E and fig. S7, J to M). Theobserved binding is thus likely a faithful repre-sentation of an important T1:C2 interaction.T domains have previously been observed

bound to sites analogous to the donor site intwo specialized C domain homologs. E domains,found downstream of T domains in somemod-ules, catalyze chirality inversion in the peptideintermediate (29). Fungal NRPSs often endwitha terminal condensation-like (Ct) domain, whichcatalyzes peptide release by macrocyclizationwith an internal nucleophile in the peptideintermediate (30). Both E and Ct domains have

Reimer et al., Science 366, eaaw4388 (2019) 8 November 2019 3 of 7

Donor Tbinding site

Acceptor Tbinding site

T1

C2

1008

1088

1086

1051

T2

A2core

C2

A2sub

T1

F1

A1sub

A1core

Condensation ConformationAcceptor T binding site

Donor Tbinding site

Donor Tbinding site Acceptor T

binding site

T1 T2

C2

A B C

D F

Active site

T1C2

Ser729

G

fValHis

1134

Tyr810

carbonyl

approach of acceptor

amino- ppant

His1134

fVal

E

wild ty

pe

A1008K

R1051N

P1086G

L1088Y

0

3x108

6x108

9x108

12x108***

********

ns

Inte

rgra

ted

EIC

pea

k (c

ount

s)

amino- ppant

fTHF2 ATP

Val, Gly, Tpm

THF2 AMP2 PPi

O

NH

O

OHN

NH

NH

A1 T1 A2 T2C2BmdB

CT3F1

SH SH

Fig. 4. Condensation in LgrA. (A) FATCAT shows a full condensation state, withT1 and T2 docked at the donor and acceptor binding sites, respectively.FATfValC and FATfValC* are overlayed to show that the first four domains arein analogous positions regardless of space group or resolution of the structures.(B) Overlay of LgrA structures with T1 at C2. (C) DCA between Tn and Cn+1displayed on LgrA and coevolution signal (red lines) between residues in closeproximity in the LgrA structures. Residues selected for mutation are indicatedwith brown a-carbon spheres. (D) Schematic of the LgrA-BmdB chimera proteinFATCAT-CT3 and its product fVal-Gly-Tpm. (E) Liquid chromatography–massspectrometry (LC-MS) peptide synthesis assay of mutations near the T1 binding

site of C2. All reactions were performed in triplicate (n = 3) and are shown asmean values from integration of extracted ion chromatogram (EIC) peaks,normalized against the average wild-type value. Statistical significance wasdetermined by two-sided Student’s t test [not significant (ns), p > 0.05;***p ≤ 0.001; ****p ≤ 0.0001]. Error bars indicate plus or minus the standarderror of the mean. A1008K, Ala1008→Lys; R1051N, Arg1051→Asn; P1086G,Pro1086→Gly; L1088Y, Leu1008→Tyr. (F) Unbiased FO-FC (3s) simulated annealingPolder omit electron density map of the C2 active site of FATfValC, calculated withphases from a model that never included fVal-ppant ligand. (G) Magnified view ofthe FATfValC C2 active site, which shows that the Val rotates for condensation.

RESEARCH | RESEARCH ARTICLEon A

pril 17, 2021

http://science.sciencemag.org/

Dow

nloaded from

evolutionarily diverged from canonical C do-mains (5, 31), and each has one only T domainbinding site, which is analogous to the donorsite. The structure of TqaA didomain T3Ct (30)shows contacts clustered on the a9 side ofthe donor site depression, similar to our T1:C2interaction, with T3 shifted by maximally ~7 Å(fig. S7G). However, the position of T1 in thestructure of the GrsA didomain T1E1 (29) isquite different, rotated by ~37°, and translated~13 Å to the a4 side of the donor site (fig. S7F).Correspondingly, DCA between Tn and En do-mains (within CnAnTnEn modules) shows sig-nal clustered on that a4 side (fig. S7I and tableS4). T domains thus bind C and E domains indistinct ways, with our structures and TqaATCt representing Tn:Cn+1 binding, and the GrsAdidomain representing Tn:En binding.T domain binding to the donor site places

the ppant arm into that side of the active sitetunnel. The tunnel leads to the C domain’sconserved HHXXXDG catalytic motif, whereX is any residue, with the second histidine(His908) most important for activity (32–34).FATfValC, FATfValC*, and FATCA all show elec-tron density for the (amino-)ppant (35) arm inthe C2 tunnel (Fig. 4F and fig. S2, C, E, and F).FATfValC contains extra density attached to theamino-ppant, which fits fVal (fig. S2, E and F),placing the formyl group within hydrogenbonding distance of Tyr810 (Fig. 4G). The donorppant-fVal would require a small shift of theVal to expose the reactive carbonyl carbon tothe acceptor site (Fig. 4G), which may only oc-cur when acceptor substrate binds to the activesite. A transient “opening” or “closing” of the Vshape formed by the C domain’s N- and C-lobes(5, 34) could also be involved, though C2 is in

very similar conformations in all the structures,not greatly influenced by whether the T do-mains are interacting with C2 or what is at-tached to the ppants (fig. S6). The small shift ofa donor substrate to achieve a fully reactive con-formation is reminiscent of the large ribosomalsubunit, which maintains peptidyl-tRNA in anonreactive conformation until the aminoacyl-tRNA binds (36).

Large conformational changes indimodule structures

FATCAT, FATCA, and both molecules ofFATVadCA provide insight into questions ofsupermodular architecture. In FATCAT andFATCA, T1 is similarly bound at the donor siteof C2, but C2A2 or C2A2T2 is folded back towardthe initiation module in two distinctive ways(Fig. 5A): In FATCAT (and also in FATfValC andFATfValC*), C2 makes contact with F1 near itsactive site. By contrast, the entire secondmodulein FATCA is rotated ~114° around the F1A1

didomain, theC2A2core didomain center ofmassis translocated by ~80Å (Fig. 5A), andC2makesextensive contacts with A1. Notably, the C2

acceptor site is not obstructed by any of theseinteractions, meaning that this conformationwould allow a full condensation state in solu-tion. Thus, the overall conformations of FATCATand FATCA are very different, but both seemcapable of peptide bond formation.To obtain the crystals of FATVadCA, we used

a Vad dead-end inhibitor of the thiolation re-action (Fig. 1A) (37, 38). Vad binds A1 and allowsthe nucleophilic attack of ppant-T1 on Vad butis not cleaved by the reaction, tethering T1 toVad and stalling T1 and A1 in the thiolationstate. Fortuitously, the crystals of FATVadCA

contain twomolecules in the asymmetric unit,revealing two different views of the complex.In both molecules, Vad is indeed at the A1

active site and T1 and A1sub are in thiolationconformation, but the elongation module hasnotably different orientations. In onemolecule,C2A2 extends away from F1A1T1 at ~45° (Figs.2C and 5B), and, in the other, it makes a ~135°angle on a different axis (Figs. 2D and 5B). Thetransitionbetween the twoconformationswouldrequire a ~82-Å translation and ~140° rotationof the C2A2 (Fig. 5B). The initiation and elonga-tion modules do not form substantial inter-actions with each other in either conformation,with the T1-C2 linker acting as a flexible tetherbetween the two modules.Comparing the fourdimodular conformations

observed in our structures dramatically showsthe scale of the conformational changes pos-sible in a dimodular NRPS. Residues on thedistal side of A2core wouldmove by between 85and 216 Å to transition between observed con-formations. This is similar to the length of thefull dimodular NRPS, because the longest dis-tance within any structure is 220 Å. The kindsof conformational changes required for thesetransitions is presented in movie S1.To assess whether the conformational vari-

ability seen in the crystal structures reflectedflexibility in solution, we analyzed the behaviorof FATCA, the LgrA construct in three of thefour different crystallographic-observed con-formations, using small-angle x-ray scattering(SAXS). FATVadCA, FATfValCA, and apo-FATCAsamples behaved well in solution, as judged bythe initial characterization of their scatteringcurves (fig. S8). Notably, their pair-distance dis-tribution functions had limited features andlarge Dmax values (fig. S8, K to M), which is acharacteristic of molecules that either adoptextended conformations or are flexible (39).Comparison of the experimental scatteringcurves to theoretical scattering curves calculatedfrom the crystal structures resulted in very poorfits (fig. S9, A to F, and table S6), indicating thatnone of the individual conformations observedin the crystal structures fully described the con-formation of FATCA in solution. Usingweightedcombinations of the crystal structures improvedthe chi-square values somewhat (table S7) (40).Modeling of the scattering curves by using theensemble optimization method (41) resultedin ensembles with excellent fits (fig. S9, G to L).The ensembles were reminiscent of the seriesof conformations observed in the crystal struc-tures (fig. S9, M to P) and retained a level offlexibility similar to the original pool of 1000independent models (table S5), consistent withthe interpretation that LgrA is highly flexible.

Structures and sequence statistics enablemodule-swapping bioengineering

The structures and SAXS suggest that there islittle constraint onpositions of adjacentmodules

Reimer et al., Science 366, eaaw4388 (2019) 8 November 2019 4 of 7

Overlay

F1A1core

T1C2

A2core

A1sub

A2sub

FATVadCA Molecule 1 FATVadCA Molecule 2 Overlay

151 Å 216 Å

85 Å171 Å

164 Å

162 Å1

1 Å

B

CA

FF1F1

A1core

T1C2

A2core

A1subA2sub

F1A1core

T1 C2

A2core

A1subA2sub

F1 A1

A2

T1

C2

T2

A2sub

A2core

A1sub

A1coreF1

T1C2

A2core

C2

T1

A1core

A1sub

F1

F1

T2

A2core

T1

C2

FATCAFATCAT

Fig. 5. Different dimodular conformations for the same catalytic states. (A) FATCAT and FATCA bothshow T1 binding to the donor site of C2 but have very different overall conformations. (B) The twocrystallographically independent molecules of FATVadCA both show module 1 in thiolation conformationbut have very different positions of module 2. (C) The distances between positions of residue Asp1236 in thefour dimodular conformations. The structures are superimposed by their A1core.

RESEARCH | RESEARCH ARTICLEon A

pril 17, 2021

http://science.sciencemag.org/

Dow

nloaded from

in NRPSs. Other than the intramodular linker,only the condensation reaction and the T1:C2interaction during this reaction strictly coupleneighboring modules. Thus, high-resolutionviews of the T1:C2 interaction should enablemodule swapping experiments. We fused mod-ule 1 of LgrA (LgrAM1) with the terminationmodule 4 of cereulide synthetase (CesBM4) toproduce a chimera that synthesizes fVal-Val(Fig. 6 and fig. S10). We then used DCA toguide mutagenesis of the T1:CCesB-4 interfaceaimed at increasing activity of the chimera.Mutations in CCesB-4 rather than T1 were tar-geted because T1 must interact with F1, A1, andCCesB-4 for synthesis, but the donor site ofCCesB-4 must interact only with T1. Saturationmutagenesis in silico of 75 residues around thedonor site of CCesB-4 was performed, and fivemutations predicted to improve T:C inter-action without substantially affecting C do-main folding were constructed and tested invitro (Fig. 6, B and C, and table S8). Three ofthe mutations decreased activity, but E925A(Glu925→Ala) and H1008Q (His1008→Gln)showed a significant increase in fVal-Val for-mation (Fig. 6D). DCA scores predicted thateffects of C domain mutations should be ad-ditive; LgrAM1-CesBM4(E925A, H1008Q) didindeed show an additive effect, doubling the pep-tide production of the LgrAM1-CesBM4 chimera.

Discussion

The series of structures and SAXS data pre-sented portray multimodular NRPSs as veryflexible. Previous results showed that withina single module, the catalytic state can definethe positions of all domains in themodule (e.g.,the thiolation state has specific positions of alldomains in an initiation or elongationmodule)(15, 17, 18). By contrast, our current structuressuggest that for multimodular NRPSs, there isno strict coupling between the catalytic stateof a particular module and the overall confor-mation of themultimodular NRPS. Rather, itappears that many overall conformations canallow the various catalytic states: FATVadCAmolecule 1, FATVadCA molecule 2, and a con-tinuum of unobserved other conformationsshould allow thiolation; FATCA, FATCAT, andacontinuumofunobservedconformationsshouldallow condensation. The continuum of con-formations need not be equally populated: Theobservation of the same condensation confor-mation in multiple crystal forms hints that itmay be more common than others. Each con-formation we observedwas fortuitously selectedthrough crystallization in the very differentpacking environments of the five unrelatedcrystal forms. Inspection of the crystal packingreveals a myriad of different crystal contactsand few trends, though perhaps predictably,many more contacts are mediated by thelarger domains and subdomains (F, Acores,and C) than the small, mobile ones (Asub and

T), and the lower-diffracting crystal formshave more porous packing and spacious sol-vent channels.The four conformations we observed are all

markedly different from each other, and none

of the observed positions of module 2 wouldallow module 1 to perform each step of itscatalytic cycle. For example, module 2 has tomove from any observed position to allow T1to reach the F1 active site. To test if there is asingle overall dimodule conformation that iscompatible with the full synthetic cycle, wevisualized possible positions of module 2 bydrawing spheres with radii of the length ofthe T1-C2 linker, at the observed positions ofthe last residue of T1. In FATCAT, as well asin an extrapolated CATCAT dimodule, there islittle or no available position within the over-lapping spheres (fig. S11). In addition, if NRPSspossess supermodular architecture, it wouldbe mediated by domain:domain interactionsinvolving CAcore structural units of adjacent mod-ules, but we are unable to detect CnAn:Cn+1An+1

coevolution expected for such interactions. Thevolume constraints and lack of detectableCnAn:Cn+1An+1 coevolution make it likely thatno static module:module conformation existsthat accommodates the full NRPS syntheticcycle. Although it is possible that some NRPSshave a different nature from LgrA, we suggestthat NRPSs, in general, do not possess con-stant and rigid supermodular architecture.The high flexibility of NRPSs could facilitate

synthetic cycles that include cosynthetic mod-ification (tailoring), are noncanonical, or arenonlinear. Flexibility could be important forNRPSs with tailoring domains inserted at dif-ferent positions in their architecture (17, 42) orNRPSs with an abnormal order of domains,like heterobactin synthetase (which includesa C-T-A module) (43), obafluorin synthetase(which has an A domain C-terminal to its TEdomain) (44, 45), or vibriobactin synthetase(which includes VibF: Cy-Cy-A-C-T-C) (46).Equally unusual systems include beauvericinsynthetase, which uses tandem T domains toperform iterative condensation (47); myxo-chromide synthetase, which perform skippingof module 4, with T3 donating the peptidedirectly to module 5 (48); and thalassospira-mide B synthetase, where A2 is proposed tothiolate T1, T2, and T5 (49). Presumably, some ofthese specialized systems have additionalinteractions that bring domains that are farapart in sequence close together in space.Because of their straightforward synthetic

logic and important bioactive products, NRPSshave long been the subject of bioengineeringattempts to create new-to-nature peptides, withmixed success (50, 51). Strategies include muta-tion of the A domain substrate-binding pocket(52, 53), domain swaps (54), module deletion orinsertion (55, 56), module swaps (54), swaps ofmodule-sized segments (57–59), andmultimod-ular swaps via docking domains (60, 61). Thisincludes interesting recent results using swappedAnTnCn+1 segments (57, 58), a strategy thatconserves the native donor Tn:Cn+1 interactionbut not the acceptor Cn:Tn interaction (57, 58),

Reimer et al., Science 366, eaaw4388 (2019) 8 November 2019 5 of 7

A1 T1 A4 T4C4F1

SH SH

TE4

(Val) (Val)

LgrA M1 CesB M4 fTHF 2 ATP2 Val H2O

THF 2 AMP 2 PPi

fVal-Val

A

B

C

T1 A1sub

C2

1078

1089

1008

1052

925

D****

*****

*

wild ty

peE925A

H1008Q

D1052R

K1078Q

N1089A

E925A + H10

08Q

Inte

grat

ed E

IC p

eak

(cou

nts)

*******

2.5x108

2.0x108

1.5x108

1.0x108

5.0x107

0.0

O

O

HN

O

NH

HO

-2.5

0

2.5

5

7.5

-0.8 -0.6 -0.4 -0.2 0.80.60.40.20Difference in Tn : Cn+1 interaction score

Diff

eren

ce in

C d

omai

n sc

ore

N1089A00 AAA88NN 9999808 AA010 989AA888800

E925A

H1008Q

D1052R

K1078Q88QQ101K1 88Q88Q88KK Q88QQQQQQ

Fig. 6. Module-swapping bioengineering usingstructural and direct coevolution analysis.(A) Schematic of the LgrAM1-CesBM4 chimera andits product fVal-Val. (B) Scatter diagram of thedifference in C domain score against the differencein Tn:Cn+1 interaction score for the DCA analysisof 75 residues around the donor T domain bindingsite of CesB C4 mutated to each of the other19 amino acids. CesB C4 mutations in red wereanalyzed biochemically. E925A, Glu925→Ala;K1078Q, Lys1078→Gln; D1052R, Asp1052→Arg;N1089A, Asn1089→Ala; H1008Q, His1008→Gln.(C) The corresponding positions of these five muta-tions in LgrA C2. (D) Wild-type LgrAM1-CesBM4 andmutant proteins were assayed for peptide productionby LC-MS. All reactions were performed in triplicate(n = 3) and are shown as mean values from integrationof EIC peaks, normalized against the average wild-type value. Statistical significance was determined bytwo-sided Student’s t test (*p ≤ 0.05; ***p ≤ 0.001;****p ≤ 0.0001). Error bars indicate plus or minus thestandard error of the mean.

RESEARCH | RESEARCH ARTICLEon A

pril 17, 2021

http://science.sciencemag.org/

Dow

nloaded from

and using the junction of the N and C lobe ofthe C domain as the split point (58, 59). Notably,both the N and C lobes contribute to both thedonor and the acceptor T domain binding sites.The lack of strong interactions between the

didomain structural units of adjacent modulesobserved here should facilitate module swap-ping experiments, but unnatural Tn:Cn+1 in-teractions could inhibit synthesis. That weincreased the activity of a chimera by im-proving the unnatural T1:C2 interaction indi-cates that this interaction can be rate-limitingin module-swapped NRPSs. Although it didnot produce orders of magnitude higher prod-uct quantity, our approach for improving un-natural Tn:Cn+1 interactions could be combinedwith other bioengineering strategies and mayhelp NRPSs fulfill their long-held promiseas sources of new designer bioactive smallmolecules.

Materials and methods summary

Constructs were cloned with cleavable octa-histidine and calmodulin binding peptide tagsandmodified by site-directedmutagenesis. Pro-teins were expressed in Escherichia coli andpurified for crystallization by calmodulin affin-ity, nickel affinity, tag removal, anion exchange,and gel filtration chromatography. FATCAT-CTandLgrAM1-CesBM4werepurifiedby calmodulinaffinity, nickel affinity, and gel filtration. Ppantswere added to apo T domains using Sfp andfVal-NH-CoA or coenzyme A. Valinyl-adenosine-vinylsulfamonamide was complexed to FATCAby including it in the Sfp reaction.Initial nanovolume crystallization conditions

were optimized in large format to the condi-tions listed in the supplementary materials.FATC was phased by molecular replacementin Phenix (62) with FAcore (17) and the N-lobeof TycC [Protein Data Bank (PDB) 2JGP] (10),followed by (re)building in Coot (63) and re-finement in Phenix. Models of F, A1core, and Cand homologymodels A2 and T2 were used formolecular replacement phasing or as a start-ing point for (re)building and refinement forother structures.Small-angle scattering data were collected at

three concentrations andprocessedusingATSAS(40). Because of evidence of high flexibility,EOM2 (41) was used to generate ensembles ofFATCA conformations whose theoretical com-bined scattering matches the experimentallymeasured scattering well.For peptide synthesis by FATCAT-CT3, 3.7 mM

protein was incubated with 0.2 mMN10-fTHF,5 mM valine, 2 mM glycine, 1 mM tryptamine,and 5 mM adenosine triphosphate (ATP) at23°C for 5 hours before quenching and liquidchromatography–mass spectrometry (LC-MS).For peptide synthesis by LgrAM1-CesBM4, 5 mMproteinwas incubatedwith 5mMvaline, 5mMATP, and 0.5 mM N10-fTHF for 6 hours beforequenching and LC-MS.

For DCA (64), we extracted 45,015 Tn:Cn+1pairs, 14,506 Tn:En pairs, and 29,700 Cn:Tn

pairs and calculated interdomain contact scoresand domain-domain interaction scores. To com-putationally suggest mutations to improve theTn:Cn+1 interaction, we altered all single aminoacids in Cn+1 interface positions to all 19 otheramino acids and evaluated the interactionscores, looking for those that improved theTn:Cn+1 interaction score and did not substan-tially affect the C domain score.

REFERENCES AND NOTES

1. C. T. Walsh, Polyketide and nonribosomal peptide antibiotics:Modularity and versatility. Science 303, 1805–1810 (2004).doi: 10.1126/science.1094318; pmid: 15031493

2. E. A. Felnagle et al., Nonribosomal peptide synthetasesinvolved in the production of medically relevant naturalproducts. Mol. Pharm. 5, 191–211 (2008). doi: 10.1021/mp700137g; pmid: 18217713

3. N. A. McGrath, M. Brichacek, J. T. Njardarson, A graphicaljourney of innovative organic architectures that have improvedour lives. J. Chem. Educ. 87, 1348–1349 (2010).doi: 10.1021/ed1003806

4. J. M. Reimer, A. S. Haque, M. J. Tarry, T. M. Schmeing, Piecingtogether nonribosomal peptide synthesis. Curr. Opin. Struct.Biol. 49, 104–113 (2018). doi: 10.1016/j.sbi.2018.01.011;pmid: 29444491

5. K. Bloudoff, T. M. Schmeing, Structural and functional aspectsof the nonribosomal peptide synthetase condensation domainsuperfamily: Discovery, dissection and diversity. Biochim.Biophys. Acta 1865, 1587–1604 (2017). doi: 10.1016/j.bbapap.2017.05.010; pmid: 28526268

6. G. H. Hur, C. R. Vickery, M. D. Burkart, Explorations of catalyticdomains in non-ribosomal peptide synthetase enzymology.Nat. Prod. Rep. 29, 1074–1098 (2012). doi: 10.1039/c2np20025b; pmid: 22802156

7. C. T. Walsh et al., Tailoring enzymes that modify nonribosomalpeptides during and after chain elongation on NRPS assemblylines. Curr. Opin. Chem. Biol. 5, 525–534 (2001).doi: 10.1016/S1367-5931(00)00235-0; pmid: 11578925

8. K. J. Weissman, The structural biology of biosyntheticmegaenzymes. Nat. Chem. Biol. 11, 660–670 (2015).doi: 10.1038/nchembio.1883; pmid: 26284673

9. C. Chiocchini, U. Linne, T. Stachelhaus, In vivo biocombinatorialsynthesis of lipopeptides by COM domain-mediatedreprogramming of the surfactin biosynthetic complex.Chem. Biol. 13, 899–908 (2006). doi: 10.1016/j.chembiol.2006.06.015; pmid: 16931339

10. S. A. Samel, G. Schoenafinger, T. A. Knappe, M. A. Marahiel,L. O. Essen, Structural and functional insights into a peptidebond-forming bidomain from a nonribosomal peptidesynthetase. Structure 15, 781–792 (2007). doi: 10.1016/j.str.2007.05.008; pmid: 17637339

11. M. J. Tarry, A. S. Haque, K. H. Bui, T. M. Schmeing, X-raycrystallography and electron microscopy of cross- andmulti-module nonribosomal peptide synthetase proteins reveala flexible architecture. Structure 25, 783–793.e4 (2017).doi: 10.1016/j.str.2017.03.014; pmid: 28434915

12. M. A. Marahiel, A structural model for multimodular NRPSassembly lines. Nat. Prod. Rep. 33, 136–140 (2016).doi: 10.1039/C5NP00082C; pmid: 26429504

13. G. Schoenafinger, N. Schracke, U. Linne, M. A. Marahiel,Formylation domain: An essential modifying enzyme for thenonribosomal biosynthesis of linear gramicidin. J. Am. Chem.Soc. 128, 7406–7407 (2006). doi: 10.1021/ja0611240;pmid: 16756271

14. B. A. Wallace, Common structural features in gramicidinand other ion channels. BioEssays 22, 227–234 (2000).doi: 10.1002/(SICI)1521-1878(200003)22:3<227::AID-BIES4>3.0.CO;2-6; pmid: 10684582

15. A. Tanovic, S. A. Samel, L. O. Essen, M. A. Marahiel, Crystalstructure of the termination module of a nonribosomal peptidesynthetase. Science 321, 659–663 (2008). doi: 10.1126/science.1159850; pmid: 18583577

16. J. M. Reimer, M. N. Aloise, H. R. Powell, T. M. Schmeing,Manipulation of an existing crystal form unexpectedly results ininterwoven packing networks with pseudo-translationalsymmetry. Acta Crystallogr. D Struct. Biol. 72, 1130–1136(2016). doi: 10.1107/S2059798316013504; pmid: 27710934

17. J. M. Reimer, M. N. Aloise, P. M. Harrison, T. M. Schmeing,Synthetic cycle of the initiation module of a formylatingnonribosomal peptide synthetase. Nature 529, 239–242(2016). doi: 10.1038/nature16503; pmid: 26762462

18. E. J. Drake et al., Structures of two distinct conformationsof holo-non-ribosomal peptide synthetases. Nature529, 235–238 (2016). doi: 10.1038/nature16163;pmid: 26762461

19. B. R. Miller, E. J. Drake, C. Shi, C. C. Aldrich, A. M. Gulick,Structures of a nonribosomal peptide synthetase modulebound to MbtH-like proteins support a highly dynamic domainarchitecture. J. Biol. Chem. 291, 22559–22571 (2016).doi: 10.1074/jbc.M116.746297; pmid: 27597544

20. J. Alfermann et al., FRET monitoring of a nonribosomal peptidesynthetase. Nat. Chem. Biol. 13, 1009–1015 (2017).doi: 10.1038/nchembio.2435; pmid: 28759017

21. K. Bloudoff, D. Rodionov, T. M. Schmeing, Crystal structures ofthe first condensation domain of CDA synthetase suggestconformational changes during the synthetic cycle ofnonribosomal peptide synthetases. J. Mol. Biol. 425,3137–3150 (2013). doi: 10.1016/j.jmb.2013.06.003;pmid: 23756159

22. A. S. Reger, R. Wu, D. Dunaway-Mariano, A. M. Gulick,Structural characterization of a 140° domain movement in thetwo-step reaction catalyzed by 4-chlorobenzoate:CoA ligase.Biochemistry 47, 8016–8025 (2008). doi: 10.1021/bi800696y;pmid: 18620418

23. C. A. Mitchell, C. Shi, C. C. Aldrich, A. M. Gulick, Structureof PA1221, a nonribosomal peptide synthetase containingadenylation and peptidyl carrier protein domains.Biochemistry 51, 3252–3263 (2012). doi: 10.1021/bi300112e;pmid: 22452656

24. A. M. Gulick, Structural insight into the necessaryconformational changes of modular nonribosomal peptidesynthetases. Curr. Opin. Chem. Biol. 35, 89–96 (2016).doi: 10.1016/j.cbpa.2016.09.005; pmid: 27676239

25. F. Morcos et al., Direct-coupling analysis of residue coevolutioncaptures native contacts across many protein families.Proc. Natl. Acad. Sci. U.S.A. 108, E1293–E1301 (2011).doi: 10.1073/pnas.1111471108; pmid: 22106262

26. L. Yuwen et al., The role of aromatic L-amino aciddecarboxylase in bacillamide C biosynthesis by Bacillusatrophaeus C89. Sci. Rep. 3, 1753 (2013). doi: 10.1038/srep01753; pmid: 23628927

27. K. Bloudoff, C. D. Fage, M. A. Marahiel, T. M. Schmeing,Structural and mutational analysis of the nonribosomal peptidesynthetase heterocyclization domain provides insight intocatalysis. Proc. Natl. Acad. Sci. U.S.A. 114, 95–100 (2017).doi: 10.1073/pnas.1614191114; pmid: 27994138

28. M. Figliuzzi, H. Jacquier, A. Schug, O. Tenaillon, M. Weigt,Coevolutionary landscape inference and the context-dependence of mutations in beta-lactamase TEM-1. Mol. Biol. Evol.33, 268–280 (2016). doi: 10.1093/molbev/msv211;pmid: 26446903

29. W.-H. Chen, K. Li, N. S. Guntaka, S. D. Bruner, Interdomain andintermodule organization in epimerization domain containingnonribosomal peptide synthetases. ACS Chem. Biol. 11,2293–2303 (2016). doi: 10.1021/acschembio.6b00332;pmid: 27294598

30. J. Zhang et al., Structural basis of nonribosomal peptidemacrocyclization in fungi. Nat. Chem. Biol. 12, 1001–1003(2016). doi: 10.1038/nchembio.2202; pmid: 27748753

31. C. Rausch, I. Hoof, T. Weber, W. Wohlleben, D. H. Huson,Phylogenetic analysis of condensation domains in NRPS shedslight on their functional evolution. BMC Evol. Biol. 7, 78 (2007).doi: 10.1186/1471-2148-7-78; pmid: 17506888

32. C. Y. Chang et al., Structural insights into the free-standingcondensation enzyme SgcC5 catalyzing ester-bond formationin the biosynthesis of the enediyne antitumor antibioticC-1027. Biochemistry 57, 3278–3288 (2018). doi: 10.1021/acs.biochem.8b00174; pmid: 29533601

33. V. Bergendahl, U. Linne, M. A. Marahiel, Mutational analysis ofthe C-domain in nonribosomal peptide synthesis. Eur. J. Biochem.269, 620–629 (2002). doi: 10.1046/j.0014-2956.2001.02691.x;pmid: 11856321

34. K. Bloudoff, D. A. Alonzo, T. M. Schmeing, Chemical probesallow structural insight into the condensation reaction ofnonribosomal peptide synthetases. Cell Chem. Biol. 23,331–339 (2016). doi: 10.1016/j.chembiol.2016.02.012;pmid: 26991102

35. Y. Liu, S. D. Bruner, Rational manipulation of carrier-domaingeometry in nonribosomal peptide synthetases. ChemBioChem8, 617–621 (2007). doi: 10.1002/cbic.200700010;pmid: 17335097

Reimer et al., Science 366, eaaw4388 (2019) 8 November 2019 6 of 7

RESEARCH | RESEARCH ARTICLEon A

pril 17, 2021

http://science.sciencemag.org/

Dow

nloaded from

36. T. M. Schmeing, K. S. Huang, S. A. Strobel, T. A. Steitz, Aninduced-fit mechanism to promote peptide bond formation andexclude hydrolysis of peptidyl-tRNA. Nature 438, 520–524(2005). doi: 10.1038/nature04152; pmid: 16306996

37. C. Qiao, D. J. Wilson, E. M. Bennett, C. C. Aldrich, Amechanism-based aryl carrier protein/thiolation domainaffinity probe. J. Am. Chem. Soc. 129, 6350–6351 (2007).doi: 10.1021/ja069201e; pmid: 17469819

38. J. A. Sundlov, C. Shi, D. J. Wilson, C. C. Aldrich, A. M. Gulick,Structural and functional investigation of the intermolecularinteraction between NRPS adenylation and carrier proteindomains. Chem. Biol. 19, 188–198 (2012). doi: 10.1016/j.chembiol.2011.11.013; pmid: 22365602

39. P. Bernadó, Effect of interdomain dynamics on the structuredetermination of modular proteins by small-angle scattering.Eur. Biophys. J. 39, 769–780 (2010). doi: 10.1007/s00249-009-0549-3; pmid: 19844700

40. D. Franke et al., ATSAS 2.8: A comprehensive data analysissuite for small-angle scattering from macromolecularsolutions. J. Appl. Crystallogr. 50, 1212–1225 (2017).doi: 10.1107/S1600576717007786; pmid: 28808438

41. G. Tria, H. D. Mertens, M. Kachala, D. I. Svergun, Advancedensemble modelling of flexible macromolecules using x-raysolution scattering. IUCrJ 2, 207–217 (2015). doi: 10.1107/S205225251500202X; pmid: 25866658

42. S. Mori et al., Structural basis for backbone N-methylation byan interrupted adenylation domain. Nat. Chem. Biol. 14,428–430 (2018). doi: 10.1038/s41589-018-0014-7;pmid: 29556104

43. M. Bosello et al., Structural characterization of theheterobactin siderophores from Rhodococcus erythropolis PR4and elucidation of their biosynthetic machinery. J. Nat. Prod.76, 2282–2290 (2013). doi: 10.1021/np4006579;pmid: 24274668

44. J. E. Schaffer, M. R. Reck, N. K. Prasad, T. A. Wencewicz,b-Lactone formation during product release from anonribosomal peptide synthetase. Nat. Chem. Biol. 13,737–744 (2017). doi: 10.1038/nchembio.2374;pmid: 28504677

45. D. F. Kreitler, E. M. Gemmell, J. E. Schaffer, T. A. Wencewicz,A. M. Gulick, The structural basis of N-acyl-a-amino-b-lactoneformation catalyzed by a nonribosomal peptide synthetase.Nat. Commun. 10, 3432 (2019). doi: 10.1038/s41467-019-11383-7; pmid: 31366889

46. N. J. Hillson, C. T. Walsh, Dimeric structure of the six-domainVibF subunit of vibriobactin synthetase: Mutant domain activityregain and ultracentrifugation studies. Biochemistry 42,766–775 (2003). doi: 10.1021/bi026903h; pmid: 12534289

47. Y. Xu et al., Biosynthesis of the cyclooligomer depsipeptidebeauvericin, a virulence factor of the entomopathogenic fungusBeauveria bassiana. Chem. Biol. 15, 898–907 (2008).doi: 10.1016/j.chembiol.2008.07.011; pmid: 18804027

48. S. C. Wenzel et al., Structure and biosynthesis ofmyxochromides S1-3 in Stigmatella aurantiaca: Evidence for aniterative bacterial type I polyketide synthase and for moduleskipping in nonribosomal peptide biosynthesis. ChemBioChem6, 375–385 (2005). doi: 10.1002/cbic.200400282;pmid: 15651040

49. A. C. Ross et al., Biosynthetic multitasking facilitatesthalassospiramide structural diversity in marine bacteria.J. Am. Chem. Soc. 135, 1155–1162 (2013). doi: 10.1021/ja3119674; pmid: 23270364

50. M. Winn, J. K. Fyans, Y. Zhuo, J. Micklefield, Recent advances inengineering nonribosomal peptide assembly lines. Nat. Prod. Rep.33, 317–347 (2016). doi: 10.1039/C5NP00099H; pmid: 26699732

51. A. S. Brown, M. J. Calcott, J. G. Owen, D. F. Ackerley,Structural, functional and evolutionary perspectives oneffective re-engineering of non-ribosomal peptide synthetaseassembly lines. Nat. Prod. Rep. 35, 1210–1228 (2018).doi: 10.1039/C8NP00036K; pmid: 30069573

52. K. Eppelmann, T. Stachelhaus, M. A. Marahiel, Exploitation ofthe selectivity-conferring code of nonribosomal peptidesynthetases for the rational design of novel peptide antibiotics.Biochemistry 41, 9718–9726 (2002). doi: 10.1021/bi0259406;pmid: 12135394

53. J. Thirlway et al., Introduction of a non-natural amino acid intoa nonribosomal peptide antibiotic by modification ofadenylation domain specificity. Angew. Chem. Int. Ed. Engl. 51,7181–7184 (2012). doi: 10.1002/anie.201202043;pmid: 22711659

54. K. T. Nguyen et al., Combinatorial biosynthesis of novelantibiotics related to daptomycin. Proc. Natl. Acad. Sci. U.S.A.103, 17462–17467 (2006). doi: 10.1073/pnas.0608589103;pmid: 17090667

55. H. D. Mootz et al., Decreasing the ring size of a cyclicnonribosomal peptide antibiotic by in-frame module deletion inthe biosynthetic genes. J. Am. Chem. Soc. 124, 10980–10981(2002). doi: 10.1021/ja027276m; pmid: 12224936

56. D. Butz et al., Module extension of a non-ribosomal peptidesynthetase of the glycopeptide antibiotic balhimycin producedby Amycolatopsis balhimycina. ChemBioChem 9, 1195–1200(2008). doi: 10.1002/cbic.200800068; pmid: 18399534

57. K. A. J. Bozhüyük et al., De novo design and engineering ofnon-ribosomal peptide synthetases. Nat. Chem. 10, 275–281(2018). doi: 10.1038/nchem.2890; pmid: 29461518

58. C. Steiniger et al., Harnessing fungal nonribosomalcyclodepsipeptide synthetases for mechanistic insights andtailored engineering. Chem. Sci. 8, 7834–7843 (2017).doi: 10.1039/C7SC03093B; pmid: 29163920

59. K. A. J. Bozhüyük, A. Linck, A. Tietze, F. Wesche, S. Nowak,F. Fleischhacker, H. B. Bode, Modification and de novodesign of non-ribosomal peptide synthetases (NRPS) usingspecific assembly points within condensation domains. bioRxiv354670 [Preprint]. 24 June 2018. https://www.biorxiv.org/content/10.1101/354670v1.

60. M. Hahn, T. Stachelhaus, Harnessing the potential ofcommunication-mediating domains for the biocombinatorialsynthesis of nonribosomal peptides. Proc. Natl. Acad. Sci. U.S.A.103, 275–280 (2006). doi: 10.1073/pnas.0508409103;pmid: 16407157

61. C. Hacker et al., Structure-based redesign of docking domaininteractions modulates the product spectrum of arhabdopeptide-synthesizing NRPS. Nat. Commun. 9, 4366(2018). doi: 10.1038/s41467-018-06712-1; pmid: 30341296

62. P. D. Adams et al., PHENIX: Building new software forautomated crystallographic structure determination. Acta

Crystallogr. D Biol. Crystallogr. 58, 1948–1954 (2002).doi: 10.1107/S0907444902016657; pmid: 12393927

63. P. Emsley, B. Lohkamp, W. G. Scott, K. Cowtan, Features anddevelopment of Coot. Acta Crystallogr. D Biol. Crystallogr. 66,486–501 (2010). doi: 10.1107/S0907444910007493;pmid: 20383002

64. M. Weigt, R. A. White, H. Szurmant, J. A. Hoch, T. Hwa,Identification of direct residue contacts in protein-proteininteraction by message passing. Proc. Natl. Acad. Sci. U.S.A.106, 67–72 (2009). doi: 10.1073/pnas.0805923106;pmid: 19116270

ACKNOWLEDGMENTS

We thank D. Alonzo (MS and ligand building); A. Pistofidis andY. Ripstein (purification help); K. Bloudoff and C. Fortinez (bmdBplasmid); J. Jiang (cloning); Schmeing lab members (discussions);N. Rogerson (editing); G. Hura (SAXS advice and analyses);K. Burnett (SAXS data collection); A. Berghuis (discussions); andB. Nagar (diffraction data discussions). We thank C. Chalut forBL21(DE3)entD– cells; J. Colucci, M. Guerard, and R. Zamboni(ZCS); staff at CLS 08ID-1 (S. Labiuk, J. Gorin, M. Fodje, K. Janzen,D. Spasyuk, and P. Grochulski); APS 24-ID-C (grants GM124165,RR029205, and DE-AC02-06CH11357; F. Murphy); and SAXS dataAdvanced Light Source (ALS) SIBYLS beamline (US-DOE-BERIntegrated Diffraction Analysis Technologies, NIGMS ALS-ENABLE-P30-GM124169 and S10OD018483). Funding: This work wasfunded by CIHR (FDN-148472) and a Canada Research Chair toT.M.S., a European Union H2020 research and innovation programMSCA-RISE-2016 (#734439 INFERNET) grant to M.W., andstudentships from NSERC (J.M.R.), Boehringer Ingelheim Fonds(M.E.), and CIHR (I.H.) Author contributions: J.M.R. performedthe crystallography with assistance of I.H. J.M.R performed activityassays for LgrA-CesB chimeric proteins. M.E. performed activityassays for LgrA-BmdB proteins and structure refinement. J.M.R.and M.E. prepared samples for SAXS, and A.G. performed analysesof SAXS data. M.W. performed coevolution and bioinformaticanalyses. T.M.S. directed the project. T.M.S., J.M.R., and M.W.wrote the manuscript. Competing interests: The authorsdeclare no competing interests. Data and materialsavailability: Coordinates and structure factors are available in theRCSB Protein Data Bank under the following PDB IDs: FATfValC,6MFW; FATfValC*, 6MFX; FATVadCA, 6MFY; FATCAT, 6MFZ; andFATCA, 6MG0. All other data are available in the main text or thesupplementary materials.

SUPPLEMENTARY MATERIALS

science.sciencemag.org/content/366/6466/eaaw4388/suppl/DC1Materials and MethodsFigs. S1 to S11Tables S1 to S10References (65–91)Movie S1

View/request a protocol for this paper from Bio-protocol.

30 December 2018; resubmitted 4 June 2019Accepted 10 October 201910.1126/science.aaw4388

Reimer et al., Science 366, eaaw4388 (2019) 8 November 2019 7 of 7

RESEARCH | RESEARCH ARTICLEon A

pril 17, 2021

http://science.sciencemag.org/

Dow

nloaded from

Structures of a dimodular nonribosomal peptide synthetase reveal conformational flexibilityJanice M. Reimer, Maximilian Eivaskhani, Ingrid Harb, Alba Guarné, Martin Weigt and T. Martin Schmeing

DOI: 10.1126/science.aaw4388 (6466), eaaw4388.366Science 

, this issue p. eaaw4388Scienceand handoff between modules.authors used small-angle x-ray scattering to confirm that large conformational changes are possible during biosynthesis positioning differed between these structures even when the same intermediate was attached to the enzyme. Thedetermined crystal structures of portions of a nonribosomal peptide synthetase, including a full dimodule. Module

et al.nature. Part of the challenge is in understanding how modules interact and hand off intermediates. Reimer is a distant goal in the lab despite a huge diversity of modular systems in−−complexes to produce desired products

where enzyme units can be swapped in and out of assembly line−−Modular biosynthesis of small moleculesMoving modules drive biosynthesis

ARTICLE TOOLS http://science.sciencemag.org/content/366/6466/eaaw4388

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2019/11/06/366.6466.eaaw4388.DC1

REFERENCES

http://science.sciencemag.org/content/366/6466/eaaw4388#BIBLThis article cites 89 articles, 11 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Science. No claim to original U.S. Government WorksCopyright © 2019 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of

on April 17, 2021

http://science.sciencem

ag.org/D

ownloaded from