Decoding and reprogramming fungal iterativenonribosomal peptide synthetases

ARTICLE

Received2Sep2016|Accepted22Mar2017|Published23May2017

DOI:10.1038/ncomms15349OPEN Decoding and reprogramming fungal iterative nonribosomal peptide synthetases

Dayu Yu1,2,*,Fuchao Xu2,*,Shuwei Zhang2&Jixun Zhan2

Nonribosomal peptide synthetases(NRPSs)assemble a large group of structurally and

functionally diverse natural products.While the iterative catalytic mechanism of bacterial

NRPSs is known,it remains unclear how fungal NRPSs create products of desired length.Here

we show that fungal iterative NRPSs adopt an alternate incorporation strategy.Beauvericin

and bassianolide synthetases have the same C1-A1-T1-C2-A2-MT-T2a-T2b-C3domain

organization.During catalysis,C3and C2take turns to incorporate the two biosynthetic

precursors into the growing depsipeptide chain that swings between T1and T2a/T2b with

C3cyclizing the chain when it reaches the full length.We reconstruct the total biosynthesis of

beauvericin in vitro by reacting C2and C3with two SNAC-linked precursors and present

a domain swapping approach to reprogramming these enzymes for peptides with altered

lengths.These?ndings highlight the difference between bacterial and fungal NRPS

mechanisms and provide a framework for the enzymatic synthesis of non-natural

nonribosomal peptides.

1Department of Applied Chemistry and Biological Engineering,College of Chemical Engineering,Northeast Electric Power University,Jilin,Jilin132012,China. 2Department of Biological Engineering,Utah State University,4105Old Main Hill,Logan,Utah84322,USA.*These authors contributed equally to this work. Correspondence and requests for materials should be addressed to J.Z.(email:jixun.zhan@https://www.sodocs.net/doc/3b3328066.html,).

N onribosomal peptides(NRPs)represent an important family of bioactive natural products,such as daptomycin,

penicillin and vancomycin.They are assembled by nonribosomal peptide synthetases(NRPSs)that consist of a series of catalytic domains1.Adenylation(A),thiolation (T)and condensation(C)are essential domains for the formation and elongation of the peptide chain.Other tailoring domains, such as methyltransferase(MT),oxidation(Ox)and epimeri-zation(E),contribute to the large structural diversity in natural NRPs2.Iterative NRPSs generate NRP products with repeated moieties/monomers3,such as bacterial metabolites enterobactin, echinomycin and gramicidin,as well as fungal natural products beauvericin(1),beauvericins A–C(2–4),bassianolide(5),ennia-tins A–C(6–8)and verticilide(9)(Fig.1a).These molecules possess a wide range of biological activities,including antimi-crobial,insecticidal,anthelmintic,herbicidal,anti-haptotactic, anti-cholesterol and anticancer activities3–8.1–9are cyclic oligomers of monomer synthesized from a hydroxycarboxylic acid and an N-methylated amino acid.For instance,1is a cyclic trimer of a dipeptidol monomer synthesized from N-methyl-L-phenylalanine(N-Me-L-Phe)and D-hydroxyisovaleric acid (D-Hiv),and5is an octadepsipeptide containing four units of D-Hiv-N-Me-L-Leu(leucine:Leu).However,how fungal iterative NRPSs are programmed to assemble these compounds is not well understood.

Beauveria bassiana is a?lamentous fungus known to produce 1–5,and the enzymes responsible for the synthesis of these compounds have been reported9,10.The beauvericin and bassianolide synthetases(BbBEAS and BbBSLS)share signi?cant sequence homology(66%identity and79%similarity)and the same domain organization(Fig.1b).In spite of the high sequence similarity,BbBEAS catalyses recursive head-to-tail condensation of three dipeptidol monomers,while BbBSLS condenses four monomers.Thus,these enzymes constitute a promising model to understand the product elongation and length control mechanism by fungal iterative NRPSs.The gene sequences of several other fungal NRPSs with the domain organization of C1-A1-T1-C2-A2-MT-T2a-T2b-C3(Fig.1b)have also been reported11.The common architecture of these NRPSs suggests that they share a general assembly rule.

Previous studies on bacterial iterative NRPSs,such as those involved in the biosynthesis of tyrocidine and gramicidin S, revealed that these enzymes use an oligomerization approach for chain elongation and length control through the C-terminal thioesterase(TE)domain.They oligomerize a monomer inter-mediate formed by upstream condensation domains and catalyse the subsequent head-to-tail cyclization for product release12,13. However,TE or a homologous domain is absent in fungal iterative NRPSs such as BbBEAS and BbBSLS.Based on the conserved signature regions,it was generally proposed that A1of BbBEAS and BbBSLS activates D-Hiv,while A2recognizes L-Phe and L-Leu,respectively.The activated D-Hiv is transferred to T1, and MT methylates the A2-activated amino acid4.However,the roles of several key domains in these fungal NRPSs remain unknown,including the twin T2domains and three C domains. More importantly,the product assembly process by fungal NRPSs is poorly understood.

In this work,we report that BbBEAS and BbBSLS adopt an alternate incorporation strategy to assemble the products.The two precursors,D-Hiv and N-Me-L-Phe/N-Me-L-Leu,are alter-nately incorporated into the extending depsipeptide chain by C2and C3.The C3domain controls the product length and cyclizes the full-length depsipeptide chain to form the?nal products.Products with altered chain lengths are obtained by swapping the C3domain,providing a domain swapping approach to reprogramming these enzymes for new molecules.Results

Puri?cation and reconstitution of BbBEAS and BbBSLS.We have recently expressed BbBEAS and BbBSLS and achieved high-yield production of1–5(33.8mg là1for1–4and21.7mg là1 for5)in Saccharomyces cerevisiae BJ5464-NpgA14.To decode the functions of the catalytic domains in BbBEAS(352kDa)and BbBSLS(348kDa),we?rst expressed and puri?ed these giant enzymes from S.cerevisiae BJ5464-NpgA.As shown in Supplementary Fig.1,both enzymes were expressed in the yeast and obtained in pure form(0.8mg là1for both BbBEAS and BbBSLS).Reaction of C-His6-tagged BbBEAS with adenosine triphosphate(ATP),S-adenosylmethionine(SAM),L-Phe and D-Hiv yielded1(trace i,Fig.1c),which was con?rmed by electrospray ionisation mass spectrometry(ESI-MS) (Supplementary Fig.2),high-resolution electrospray ionisation mass spectrometry(HR-ESI-MS)(Supplementary Fig.3)and a comparison with the authentic sample prepared and fully characterized in our previous work(trace ii,Fig.1c)14.Similarly, BbBSLS was found to synthesize5(traces iii and iv,Fig.1c, Supplementary Fig.2and Supplementary Fig.4)from L-Leu and D-Hiv14.Formation of1and5allowed direct functional characterization of BbBEAS and BbBSLS.These results indicated that both BbBEAS and BbBSLS were functionally expressed in S.cerevisiae and their catalytic activity can be reconstituted in vitro,providing a great platform to further study the catalytic domains in these modular NRPSs.

Roles of the twin T2domains and the biosynthetic model.Two possible biosynthetic models were proposed for BbBEAS and other similar NRPSs:linear(alternate incorporation of the precursors)and parallel(oligomerization of a monomer synthesized from the precursors)(Fig.2a)4.In the linear model, T1and T2a/T2b are alternately used for tethering the growing depsipeptide chain for addition of the D-hydroxycarboxylic acid and N-Me-L-amino acid units.T2a and T2b have the same function and only one of them is required.In contrast,in the parallel model,one of the twin T2domains is proposed to hold the synthesized dipeptidol monomer after it is synthesized at the other T2domain4.Additional monomers are added to the T2b-or T2a-linked monomer through oligomerization to form the dimer and subsequently the trimer(1)or tetramer(5).In this parallel model,both twin T2domains are required.Bacterial iterative NRPSs adopt a product assembly process that is more like the parallel model in Fig.2a,except that the growing product chain is linked to the TE instead of a T domain.To understand what model is used by fungal iterative NRPSs,it is necessary to reveal the roles of the twin T2domains in BbBEAS and BbBSLS. Sequence analysis revealed that the twin T2domains in both BbBEAS and BbBSLS contain the conserved motif (I/L)GG(D/H)SL,in which the key serine(Ser or S)residue serves as the phosphopantetheine attachment site(Supplementary Fig.5).We?rst examined how inactivation of T2a and/or T2b affects the biosynthesis.Mutation of S2591in T2a to alanine(A) did not abolish the production of beauvericins(trace i,Fig.2b), but the titre(0.7±0.1mg là1)calculated from three replicates was signi?cantly lower than that(33.8±1.4mg là1)by wild-type BbBEAS.In contrast,the T2b mutant BbBEAS-S2688A produced beauvericins(trace ii,Fig.2b)with a titre of23.0±1.0mg là1. When both key Ser residues were mutated,no products were synthesized by BbBEAS-S2591A/S2688A(trace iii,Fig.2b).

We next tested how removal of the twin T2domains from BbBEAS affects beauvericin biosynthesis using a dissection and co-expression approach.BbBEAS was dissected at different positions and the fragments were functionally expressed in S.cerevisiae BJ5464-NpgA(traces i–iv,Fig.2c).Using this system,

we removed T 2a from BbBEAS by co-expressing C 1-A 1-T 1-C 2-A 2-MT and T 2b -C 3in the yeast.Liquid chromatography–mass spectrometry (LC–MS)analysis revealed that beauvericins were produced at 5.0±1.2mg l à1(trace v,Fig.2c)in the absence of T 2a .This was validated by the construction of an intact enzyme BbBEAS-D T 2a ,which produced beauvericins at 0.5±0.1mg l à1(trace vi,Fig.2c),con?rming that the dissection and co-expression approach is effective and reliable.To remove T 2b ,C 1-A 1-T 1-C 2-A 2-MT-T 2a and standalone C 3were co-expressed in the yeast,yielding 1–4at 13.0±3.9mg l à1(trace vii,Fig.2c).When both T 2domains were removed,no products were detected (trace viii,Fig.2c).These results were consistent with those from Ser mutations,which con?rmed that only one of the twin T 2domains is required for beauvericin biosynthesis.

Similarly,we also dissected BbBSLS before and after T 2a ,and con?rmed that the fragments can be co-expressed in the yeast to reconstitute bassianolide biosynthesis (traces i and ii,Fig.2d).Removal of T 2a (trace iii,Fig.2d)or T 2b (trace iv,Fig.2d)did not abolish the production of 5,although the titre by the former was much lower (2.9mg l à1versus 10.6mg l à1).This further con?rmed that T 2a plays a major role in the biosynthetic process as inactivation or removal of this domain signi?cantly lowered the ef?ciency of the product assembly line.By contrast,T 2b is less important in the biosynthetic process and may serve as an auxiliary T domain to contribute to the overall ef?ciency.Because the presence of both twin T 2domains is not necessary for the biosynthesis,the possibility of oligomerization (parallel model)in chain elongation can be ruled out.Thus,it is concluded that 1–5are assembled through an alternate incorporation approach (linear model,Fig.2a)by the fungal NRPSs.

The condensation activity of the C 1and C 2domains .Another mystery of BbBEAS and BbBSLS is the unknown roles of the three C domains in the biosynthetic process.We analysed the C domains in four reported fungal iterative NRPSs.As shown in Fig.3a,the C 2domains of these fungal NRPSs have the conserved HHxxxDG motif 15.It was generally believed that the second histidine (H)in this motif serves as a general base for the condensation in most NRPSs,while aspartic acid (D)plays an important role in the structure of C domains 16,17.The ?rst H residue in this motif is not highly conserved,as some C domains such as the C 3domains shown in Fig.3a have an S at this position.The exact function of the G residue in this motif is not https://www.sodocs.net/doc/3b3328066.html,pared to C 2and C 3,the C 1domains of these fungal NRPSs are more divergent and C 1(BbBEAS)does not have the essential H residue.As the C-terminal condensation domain,C 3is likely involved in the ?nal cyclization and concomitant release of the peptide chain from the end of the NRPS assembly lines.The remaining two condensation domains,C 1and C 2,are possibly responsible for forming the ester and amide bonds in the product assembly line,respectively.We ?rst tested the condensation activity of C 2.C 2(BbBEAS)was overexpressed and puri?ed from E.coli BL21(DE3)(Supplementary Fig.6),and reacted with D -Hiv-SNAC (S1)(SNAC:N -acetylcysteamine)18and N -Me-L -Phe-SNAC (S2).As shown in Fig.3b,D -Hiv-N -Me-L -Phe-SNAC (S3)was synthesized by C 2(BbBEAS),and then spontaneously cyclized to form cyclo-D -Hiv-N -Me-L -Phe (10,Fig.3b).In contrast,this domain could not form an ester bond as the reaction of S1and (D -Hiv-N -Me-L -Phe)2-SNAC (S4)with C 2(BbBEAS)did not yield any products (Supplementary Fig.7).Consequently,the role of C 2was determined to be forming the amide bond in the assembly process of NRPs.

We next attempted to understand whether C 1forms the ester bond.C 1(BbBEAS)was overexpressed and puri?ed from E.coli BL21(DE3)(Supplementary Fig.6).Incubation of C 1(BbBEAS)with S1tS2or S1tS4did not yield any products (Fig.3c,d).

Because

b

Bassianolide (5)

Module 1

BbBEAS

BbBSLS

C 1

C 1

C 2

C 2

A 1

A 1

A 2

A 2

T 2a T 2a

T 2b

T 2b

C 3

C 3

MT

MT T 1

T 1

Beauvericin (1)Beauvericin A (2)Beauvericin B (3)Beauvericin C (4)R 1=R 2=R 3=CH 3

R 1=R 2=CH 3, R 3=CH 2CH 3R 1=CH 3, R 2=R 3=CH 2CH 3R 1=R 2=R 3=CH 2CH 3

Enniatin A (6)Enniatin B (7)Enniatin C (8)R=CH(CH 3)CH 2CH 3R=CH 2CH(CH 3)2

R=CH(CH 3)2Figure 1|Fungal iterative NRPSs and their products.(a )Representative NRPs (1–9)synthesized by fungal iterative NRPSs.The monomer unit is green shaded.(b )Domain organization of BbBEAS and BbBSLS.Both modular enzymes have the same architecture of C 1-A 1-T 1-C 2-A 2-MT-T 2a -T 2b -C 3.T o

differentiate these two enzymes,BbBEAS is shown in blue and BbBSLS in red.(c )HPLC analysis of the in vitro reaction of BbBEAS and BbBSLS with ATP,SAM,D -Hiv and L -amino acid (L -Phe for BbBEAS and L -Leu for BbBSLS).

C1(BbBEAS)lacks the active site H,we mutated D179to A and expressed this mutant enzyme BbBEAS-D179A in S.cerevisiae BJ5464-NpgA.LC–MS analysis revealed the synthesis of beauvericins by BbBEAS-D179A(trace i,Fig.3e)at 24.2±1.0mg là1,indicating that the mutation did not affect the biosynthetic process.Similarly,mutation of H170and D174 in the conserved motif of C1of BbBSLS did not interfere with bassianolide biosynthesis(traces ii and iii,Fig.3e),indicating that mutation of these functionally or structurally important residues does not in?uence the biosynthesis of5.These results clearly indicate that C1has no condensation activity.To further probe the role of C1in beauvericin biosynthesis,we constructed a truncated version of BbBEAS by removing C1.No beauvericins were produced by BbBEAS-D C1.Addition of standalone C1(BbBEAS)into the yeast did not recover beauvericin biosynthesis. SDS–PAGE analysis revealed that BbBEAS-D C1(308kDa)was not expressed and the attempt to purify His6-tagged BbBEAS-D C1using Ni-NTA chromatography from the yeast host failed. Similarly,BbBSLS-D C1could not be expressed in the yeast either. Thus,removal of C1resulted in unsuccessful expression of the truncated enzymes,suggesting that the presence of C1is critical for the expression of BbBEAS and

BbBSLS.

a

1 Linear model

Parallel model

Macrocyclization

Figure2|Roles of the twin T2domains in BbBEAS and BbBSLS.(a)Two possible models for the assembly process of1:linear model(top,with?lled arrows)and parallel model(bottom,with un?lled arrows).The two precursors and the growing depsipeptide chain are tethered to the T domains through phosphopantetheinyl prosthetic groups.In the linear model,T2a and T2b have the same function of holding the growing depsipeptide chain.(b)HPLC analysis of the products of mutant enzymes BbBEAS-S2591A,BbBEAS-S2688A and BbBEAS-S2591A/S2688A in S.cerevisiae BJ5464-NpgA.(c)HPLC analysis of the effect of removal of T2a and/or T2b on beauvericin biosynthesis.(d)HPLC analysis of the effect of removal of T2a and/or T2b on bassianolide biosynthesis.

The condensation activity of the C 3domain .To understand how the ester bond is formed,we next investigated the condensation activity of C 3.We ?rst removed this domain to yield a truncated variant of BbBEAS.Expression of BbBEAS-D C 3yielded no products (trace i,Fig.4a).In trans addition of standalone C 3(BbBEAS)into S.cerevisiae BJ5464-NpgA reconstituted the synthesis of beauvericins (26.5±1.3mg l à1,trace ii,Fig.4a).This suggests that C 3plays an essential role in beauvericin biosynthesis.We next examined the effect of mutation of the conserved H on beauvericin biosynthesis.BbBEAS-H2901A failed to produce beauvericins (traces iii,Fig.4a),indicating that the condensation activity of C 3is required for the biosynthetic process.

To gain more insight into the role of C 3,C-His 6-tagged BbBEAS-D C 3was puri?ed from S.cerevisiae BJ5464-NpgA and reacted with ATP,SAM,D -Hiv and L -Phe.No beauvericins were detected by LC–MS.Instead,small amounts of 10and 11(trace i,Fig.4b)were produced,which were respectively identi?ed as the cyclic and linear D -Hiv-N -Me-L -Phe based on the ESI-MS (Supplementary Fig.2)and NMR (Supplementary Table 1)as well as a comparison with the chemically prepared standards (traces ii and iii,Fig.4b).10can be hydrolysed in 0.1N NaOH to yield 11.Production of 10and 11indicated that the biosynthetic process stopped after the formation of the monomer.The same products were observed from the reaction of puri?ed C-His 6-tagged BbBEAS-H2901A (trace iv,Fig.4b),con?rming that mutation of H2901did cause the loss of the condensation ability of C 3.

We also overexpressed and puri?ed C 3(BbBEAS)from E.coli BL21(DE3)(Supplementary Fig.6).Co-reaction of C 3(BbBEAS)and BbBEAS-D C 3with necessary components yielded 1(trace i,Fig.4c).(D -Hiv-N -Me-L -Phe)2-D -Hiv-SNAC (S5)was synthesized from S1and S4by standalone C 3(BbBEAS),then spontaneously hydrolysed to form (D -Hiv-N -Me-L -Phe)2-D -Hiv (14,Fig.4d)that

has a molecular mass of 640(Supplementary Fig.2).Addition of S2into the reaction system did not produce any extra products including 1(Supplementary Fig.8).Thus,C 3is responsible for the formation of the ester bond between T 1-linked D -Hiv and the monomer D -Hiv-N -Me-L -Phe or the dimer (D -Hiv-N -Me-L -Phe)2that is tethered to T 2a or T 2b .

The same experiments were applied to BbBSLS.BbBSLS-D C 3and BbBSLS-H2861A failed to synthesize 5(traces i and iii,Fig.4e).In trans addition of standalone C 3(BbBSLS)recovered bassianolide biosynthesis (traces ii and iv,Fig.4e).In vitro reactions of ATP,SAM,D -Hiv and L -Leu with puri?ed BbBSLS-D C 3(trace i,Fig.4f)or BbBSLS-H2861A (trace ii,Fig.4f)did not yield 5,but the linear (12)and cyclic D -Hiv-N -Me-L -Leu (13)(traces i–iv,Fig.4f),respectively.Addition of C 3(BbBSLS)into the reaction system reconstituted bassianolide biosynthesis (trace ii,Fig.4c).These results further con?rmed that C 3plays an essential role in forming the ester bond during the biosynthesis of 1–5.Removal or inactivation of C 3did not affect the functions of the upstream domains,but stopped the biosynthetic process at the monomer stage.

Identi?cation and reprogramming of chain length control .With the understanding of the roles of the twin T 2and three condensation domains,we next attempted to identify the product-length-controlling domain(s)by constructing a series of chimeric enzymes.Since module 1of BbBEAS and BbBSLS are exchangeable without affecting the product pro?les 19,we tested the effects of swapping the C-terminal domains on the product formation.We ?rst constructed an enzyme C 1-A 1-T 1-C 2-A 2-MT (BbBEAS)T 2a -T 2b -C 3(BbBSLS).Unlike wild-type BbBEAS (trace i,Fig.5a),C 1-A 1-T 1-C 2-A 2-MT (BbBEAS)T 2a -T 2b -C 3(BbBSLS)did not synthesize beauvericins,but a new product FX1(15)(trace ii,

S2

S1

S3

10

(i) S1+S2+inactivated C 2(BbBEAS)

(ii) S1+S2+C 2(BbBEAS)(iii)a

c

b

S2

S1+S2+inactivated C 1(BbBEAS)

S1+S2+C 1(BbBEAS)

S1

d

S1

S4

(i) S1+S4+inactivated C 1(BbBEAS)(ii) S1+S4+C 1(BbBEAS)6BbBEAS-D179A BbBSLS-H170A BbBSLS-D174A

5

1

2

3

4

(i)

(ii)(iii)

NRPS

BbBEAS FvBEAS BbBSLS XsBSLS

Standards of S3 and10151020253015102025

30

81012141618Min

15

10

20

2530

Min

e

SHALYDG SHALYDG SHALYDG

HHIVSDG HHIISDG HHIISDG HHIISDG

HLALVDS SHALVDS SHSFVDS SHALVDN

O O O

O

O

O O O N

Spontaneous cyclization

D-Hiv-SNAC (S1)N-Me-L-Phe-SNAC (S2)

Condensation

D-Hiv-N-Me-L-Phe-SNAC (S3)Cyclo-D-Hiv-N-Me-L-Phe (10)

O

OH

N S +

OH

HN

S

S

H N H N

H N (i)(ii)(i)

(ii)

C 1

C 2

C 3

Min Min

SHALYDG O

C 2

Figure 3|Investigation of the condensation activity of C 1and C 2.(a )Comparison of the conserved HHxxxDG region in the C domains of four reported

fungal iterative NRPSs.The active site H residues are highlighted.(b )HPLC analysis of the in vitro reaction of C 2(BbBEAS)with D -Hiv-SNAC (S1)and N -Me-L -Phe-SNAC (S2).(c )HPLC analysis of the in vitro reaction of C 1(BbBEAS)with S1and S2.(d )HPLC analysis of the in vitro reaction of C 1(BbBEAS)with S1and S4.(e )HPLC analysis of the products of C 1-mutated BbBEAS and BbBSLS in S.cerevisiae BJ5464-NpgA.

Fig.5a)with a molecular mass of 1,044(Supplementary Fig.2)at 3.2±0.6mg l à1.The molecular formula of 15was determined to be C 60H 76N 4O 12by HR-ESI-MS (Supplementary Fig.9).The structure was established as a cyclic tetramer (Fig.5b)based on the one-dimensional (Supplementary Table 1)and two-dimensional NMR data (Supplementary Fig.10).This was further con?rmed by chemical hydrolysis of 15in 0.1N NaOH,which yielded the monomer 11(Supplementary Fig.11).We then made C 1-A 1-T 1-C 2-A 2-MT-T 2a(BbBEAS)-T 2b -C 3(BbBSLS)and C 1-A 1-T 1-C 2-A 2-MT-T 2a -T 2b(BbBEAS)-C 3(BbBSLS).Both enzymes produced 15as the sole product (traces iii and iv,Fig.5a).These results revealed that replacement of C 3(BbBEAS)with C 3(BbBSLS)is suf?cient to reprogramme chain length control in BbBEAS.The same results were obtained for BbBSLS (traces i–iv,

Fig.5c).C 1-A 1-T 1-C 2-A 2-MT-T 2a -T 2b(BbBSLS)-C 3(BbBEAS)gene-rated 8(Supplementary Fig.12)as the only product.Therefore,C 3was unambiguously identi?ed as the chain-length-controlling domain.

C-terminal TE,R,Cy or C is often involved in the cyclization and concomitant release of the peptide chain from the end of the NRPS assembly lines 20.Since no TE,Cy or R domains are present in BbBEAS and BbBSLS,we considered that C 1and C 3might be involved in the macrocyclization.To probe the cyclization activity and speci?city of C 3,three depsipeptidyl-SNACs that mimic the linear monomer,dimer and trimer intermediates tethered to T 2a or T 2b were reacted with C 3(BbBEAS).S3is unstable and can undergo quick spontaneous cyclization to yield 10in the reaction buffer (traces i and ii,Fig.5d).Addition of C 3(BbBEAS)into the

a

b

)

Figure 4|Investigation of the ether bond-forming activity of C 3.(a )HPLC analysis of the in vivo products of C 3-less or C 3-mutated BbBEAS in S.cerevisiae BJ5464-NpgA.(b )HPLC analysis of the in vitro reactions of C 3-less or C 3-mutated BbBEAS with ATP,SAM,D -Hiv and L -Phe.(c )HPLC analysis of the in vitro reconstitution of the biosynthesis of 1and 5by the co-reactions of C 3-less NRPS and standalone C 3with ATP,SAM,D -Hiv and L -amino acid (L -Phe for BbBEAS and L -Leu for BbBSLS).(d )HPLC analysis of the in vitro reaction of C 3(BbBEAS)with D -Hiv-SNAC (S1)and (D -Hiv-N -Me-L -Phe)2-SNAC (S4).(e )HPLC analysis of the in vivo products of C 3-less or C 3-mutated BbBSLS in S.cerevisiae BJ5464-NpgA.(f )HPLC analysis of the in vitro reactions of C 3-less or C 3-mutated BbBSLS with ATP,SAM,D -Hiv and L -Leu.

system did not cause observable changes in the formation rate of 10(trace iii,Fig.5d).The dimer substrate S4is stable in the buffer (trace iv,Fig.5d),but was not cyclized by C 3(BbBEAS)to yield 16(trace v,Fig.5d).The trimer substrate (D -Hiv-N -Me-L -Phe)3-SNAC (S6)mimics the full-length intermediate.This substrate is stable in the reaction buffer (trace vi,Fig.5d).Incubation of S6with C 3(BbBEAS)yielded 1(trace vii,Fig.5d),while C 3(BbBEAS-H2901A)failed to cyclize it (trace viii,Fig.5d).The same reactions were applied to C 1(BbBEAS)but no cyclization products 16and 1were detected.These in vitro reactions further con?rmed that C 3catalysed the macrocyclization in beauvericin biosynthesis.Mutation of the active site residue H in C 3(BbBEAS)to A abolished its macrocyclization activity,indicating that this domain uses the same active site for the condensation and macrocyclization,both leading to the formation of an ester bond (intermolecular or intramolecular)between the carboxyl of L -Phe and hydroxyl of D -Hiv.Moreover,C 3only cyclizes the mature depsipeptide intermediate.This property is critical for chain length control,which also explains why 10and 16are not produced by the beauvericin-producing strain B.bassiana and the yeast strain that expresses BbBEAS.

In vitro total biosynthesis of 1using the C 2and C 3domains .Our results revealed that C 2and C 3work collaboratively to generate a linear full-length intermediate that is subsequently cyclized and released by C 3.We next attempted to use these two domains for in vitro total biosynthesis of 1from the beginning precursors.To this end,we reacted C 2(BbBEAS)and C 3(BbBEAS)with the two mimicking substrates S1and S2,while the control contains the same components except the two domains were inactivated.The reactions were subjected to LC–MS analysis.As shown in Fig.6a,there is a small product peak at 25.8min that had a [M tH]tpeak at m/z 642(Supplementary Fig.2).Extraction of the ion chromatogram at m/z 642(i,Fig.6a)from the negative control (left panel)and the reaction (right panel)revealed that this product was only formed in the reaction,which has the same molecular mass and retention time (Fig.4d)as S4.The extracted ion chromatogram (EIC)at m/z 784clearly showed the synthesis of 1(ii,Fig.6a).Similarly,EICs at m/z 262,641and 903indicated the formation of 10(iii),14(iv)and S6(v)in the reaction mixture.We further enlarged the high-performance liquid chromatography (HPLC)traces of the negative control (i,Fig.6b),the reaction (ii,Fig.6b)and the standard of 1(iii,Fig.6b),which clearly revealed that 1was synthesized from the two beginning precursors S1and S2by the two isolated domains.Furthermore,the peak of S6was also observed,which showed the same retention time as the extracted ion peak shown in trace v of Fig.6a.These results indicated the success of in vitro total biosynthesis of 1by C 2and C 3and showed a series of biosynthetic intermediates or their hydrolysed products.

a

c

d

b

FX1 (15)

108

5

(i)

(ii)(iii)(iv)

A 1A 1A 1A 1A 2A 2A 2A 2MT

MT

MT

MT

T 2a T 2a T 2b T 2a T 2b T 1T 1T 1T 1C 1C 1C 1C 1C 2C 2C 2C 2C 3

12141618Min

Figure 5|Identi?cation and reprogramming of chain length control.(a )HPLC analysis of the products of wild-type BbBEAS and its engineered versions in S.cerevisiae BJ5464-NpgA.(b )Structure of the new tetrameric product FX1(15)generated by reprogramming BbBEAS.(c )HPLC analysis of the products of wild-type BbBSLS and its engineered versions in S.cerevisiae BJ5464-NpgA.(d )HPLC analysis of the in vitro reactions of depsipeptidyl-SNACs with C 3(BbBEAS).Three SNAC derivatives including D -Hiv-N -Me-L -Phe-SNAC (S3),(D -Hiv-N -Me-L -Phe)2-SNAC (S4)and (D -Hiv-N -Me-L -Phe)3-SNAC (S6)were used to mimic the intermediates in beauvericin biosynthesis.

Discussion

Fungal NRPs represent an important group of bioactive natural products.BbBEAS and BbBSLS are two fungal iterative NRPSs that are involved in the biosynthesis of the anticancer NRPs 1–5.However,the assembly process of these cyclooligomer depsipeptides is unclear,largely due to the poorly understood domains including T 2a ,T 2b ,C 1,C 2and C 3in these modular enzymes.We used a combination of enzyme dissection,domain swapping,site-directed mutagenesis and in vitro enzymatic reactions to study BbBEAS and BbBSLS.This work elucidates fungal NRPSs’chain elongation and length control strategy.Highly consistent results were obtained from both BbBEAS and BbBSLS,suggesting that this family of fungal iterative NRPSs follows a general biosynthetic mechanism,which is different from the approach used by bacterial iterative NRPSs.

A typical NRPS module contains a C domain,an A domain and a T domain for chain elongation.Co-existence of two T domains in module 2of fungal iterative NRPSs not only represents an interesting structural feature,but also raises questions about the roles of these T domains in fungal NRP biosynthesis.There are several examples of twin carrier proteins in a megasynthase.For example,it was previously reported that module 6of the leinamycin polyketide synthase from Streptomyces atroolivaceus S-140contains twin acyl carrier protein (ACP)domains that are separated by a MT domain.Either is suf?cient for the biosynthesis,and ACP 6–2is preferred 21.In our work,we found that T 2a plays

major role in the formation of the NRPs.Two possible models have been previously proposed to explain biosynthetic process of fungal NRPs based on how the chain is elongated and how the twin T 2domains involved in the assembly process.We speci?cally mutated key Ser residue in these T domains that is required for the 0-phosphopantetheinylation.Inactivation of either T 2domain did abolish the formation of the corresponding NRPs,

but affected yield to different extents.Similar results were observed one of the T 2domains was removed,providing solid against the parallel model (Fig.2a),which requires the of both T 2a and T 2b .Thus,it is hypothesized that the of 1–5proceeds through the linear model (Fig.2a).was con?rmed by the in vitro reactions of C 2(BbBEAS)and 3(BbBEAS),which demonstrated that C 2forms the amide bond and 3catalyses the synthesis of the ester bond.Co-reactions C 2(BbBEAS)and C 3(BbBEAS)with D -Hiv-SNAC and N -Me-L -Phe-reconstituted the in vitro biosynthesis of 1with only two condensation domains.Furthermore,some of the intermediates or hydrolysis products were observed,including cyclo-D -Hiv-N -Me-L -Phe,(D -Hiv-N -Me-L -Phe)2-SNAC,(D -Hiv-N -Me-L -Phe)2-D -Hiv (hydrolysis product)and (D -Hiv-N -Me-L -Phe)3-SNAC,which provides direct evidence for the alternate incorporation approach described in the linear biosynthetic model (Fig.2a).

C 3is a crucial domain that integrates three functions,including condensation,chain length control and macrocyclization.While it acts like normal condensation domains to form the ester bond in the depsipeptide chain,this domain is actively involved in chain elongation and catalyses the macrocyclization of the mature intermediate for product release.It is different from C T domains in fungal noniterative NRPSs that only perform cyclization 22and Cy domains that form oxazoline or thiazoline rings using the

D residues in the conserved motif DXXXXD 23,24.Bacterial iterative NRPSs typically use a C-terminal T

E domain to cyclize and release the ?nal products.The enterobactin synthetase is such an enzyme whose TE plays a key role in the oligomerization 25.The monomer intermediate is transferred from the T domain of Ent

F (C-A-T-TE)to the active site Ser residue of the TE,which allows the T domain to be used for the formation of the next monomer.The TE catalyses the oligomerization and holds the growing peptide chain until it reaches the full length for macrocyclization.A similar TE was observed in the tyrocidine 12and gramicidin NRPSs 13.It was also reported that cyclooligomerization can be catalysed through ATP-dependent condensation reactions by NRPS-independent siderophore synthetases,such as DesD 26,PubC 27and BibC 28that participate in the biosynthesis of desferrioxamine,putrebactin and bisucaberin,respectively.

Here we show that fungal iterative NRPSs use a different strategy for chain elongation and product length control.Bacterial NRPSs use an oligomerization approach through their C-terminal TE for product chain elongation.The cyclooligomer-ization by bacterial iterative NRPSs involves acylation and deacylation of the active site serine residue of the C-terminal TE.The monomer intermediate is formed by upstream domains and transferred from a T domain to the Ser of TE.Additional units of the monomer are then added to the monomer-O-TE intermediate to elongate the peptide chain to the full length for product cyclization and release.Therefore,bacterial iterative NRPSs oligomerize a monomer unit that is synthesized from the precursors for chain elongation,and the growing oligomer intermediate is formed at and always tethered to the TE.In contrast,fungal iterative NRPSs use an alternate incorporation approach to extend the peptide chain.The growing depsipeptide chain in the biosynthesis of 1–5swings between T 1and the twin T 2domains.The monomer is not used as a unit for chain

+C 3(BbBEAS)

2(BbBEAS)a

b

Figure 6|In vitro total biosynthesis of 1using C 2and C 3domains.

(a )LC–MS analysis of the in vitro reaction of C 2(BbBEAS)and C 3(BbBEAS)with S1and S2.The left panel is the negative control with inactivated C 2and C 3,while the right is the actual reaction.The HPLC traces are shown on the top,followed by the EICs of the intermediates and ?nal product.(i)–(v)are EICs at m/z 642(i),784(ii),262(iii),641(iv)and 903(v),respectively.

(b )Enlarged view of the region of 30.1–34.4min of the HPLC traces of the negative control (i),reaction (ii)and the standard of 1(iii).

elongation at all.C3needs to work with the other condensation domain(C2)to alternately incorporate the two precursors (D-Hiv and N-Me-L-Phe or N-Me-L-Leu).Therefore,bacterial iterative NRPSs adopt a product assembly process that is more like the parallel model in Fig.2a,while fungal iterative NRPSs use a linear biosynthetic route.

In the linear model(Fig.2a),C3(BbBEAS)forms an ester bond between T1-tethered D-Hiv and T2a-or T2b-tethered D-Hiv-N-Me-L-Phe and(D-Hiv-N-Me-L-Phe)2in beauvericin biosynthesis, while C2catalyses the formation of an amide bond between T2a-or T2b-tethered N-Me-L-Phe and T1-tethered D-Hiv, D-Hiv-N-Me-L-Phe-D-Hiv and(D-Hiv-N-Me-L-Phe)2-D-Hiv. C3(BbBSLS)has the same function except that N-Me-L-Leu is used. Both C2and C3work collaboratively to elongate the depsipeptide chain.Once C2catalyses the last condensation step to yield the full-length depsipeptide chain,C3acts as a decision maker to determine and inform the biosynthetic machinery of whether further elongation is required.For instance,when a hexadepsi-peptide intermediate is synthesized,C3(BbBEAS)will shut down the elongation line and perform the macrocyclization.Instead, C3(BbBSLS)will continue the elongation to get an octadepsipeptide intermediate and then conduct the macrocyclization and product release.When C3(BbBEAS)was substituted for the C3in BbBSLS, it continued to condense D-Hiv and the hexadepsipeptide intermediate(D-Hiv-N-Me-L-Phe)3,and C2(BbBEAS)will continue to form an additional amide bond to get an octadepsipeptide. Cyclization of this intermediate yielded the new product15. Similarly,substitution of C3(BbBSLS)with its counterpart in BbBEAS switched BbBSLS from an octadepsipeptide synthetase to a hexadepsipeptide synthetase and yielded8.Thus,C2 is?exible and can catalyse extra or fewer condensations according to the‘command’of C3.While C3is strict with the number of condensation,it is?exible enough to condense‘unnatural’precursors.Consequently,reprogramming of BbBEAS and BbBSLS can be conveniently and readily achieved by swapping the C3domain.Our work thus provides an unprecedented tool for engineering fungal iterative NRPSs to yield‘unnatural’cyclooligomer depsipeptides with varied chain lengths.This work also presents in vitro total biosynthesis of a fungal NRP using only two condensation domains.

Methods

Analysis and puri?cation of https://www.sodocs.net/doc/3b3328066.html,pounds were analysed at210nm using an Agilent Ecilpse XDB-C18column(5m m,4.6mm?250mm)on an Agilent1200HPLC coupled with an Agilent6130single quadrupole mass spectrometer.High-resolution mass spectrum of FX1was collected on an Agilent 6210LCMS.Four HPLC programmes were used.Programme1(for Figs1c and 3c,d and Figs4b,d,f and6):5–90%acetonitrile–water with0.1%formic acid from0 to30min,90–100%acetonitrile–water with0.1%formic acid from30to35min and100%acetonitrile–water with0.1%formic acid from35to50min at

1ml minà1.Programme2(for Figs2b–d and3e and Figs4a,c,e and5a,c):

80–100%acetonitrile–water with0.1%formic acid over20min at a?ow rate of 1ml minà1.Programme3(for Fig.5d):50–100%acetonitrile–water with

0.1%formic acid over30min at1ml minà1.Programme4(for Fig.3b):5–70% acetonitrile–water with0.1%formic acid over40min at1ml minàhttps://www.sodocs.net/doc/3b3328066.html,pound puri?cation was performed on the same HPLC.

Strains and plasmids.E.coli XL1-Blue(Agilent Technologies)was used for routine cloning and pJET1.2(Fermentas)was used as the cloning vector.E.coli cells were grown in Luria-Bertani(LB)medium at37°C.When necessary,

50m g mlà1ampicillin was added.E.coli BL21(DE3)(Agilent Technologies)cells were used for expression of C1(BbBEAS),C2(BbBEAS),C3(BbBEAS),C3(BbBEAS-H2901A), C3(BbBSLS)and MT(BbBEAS).S.cerevisiae BJ5464-NpgA(MAT a ura3-52his3-D200 leu2-D1trp1pep4::HIS3prb1D1.6R can1GAL)was obtained from Dr.Nancy Da Silva at the University of California,Irvine.The strain was maintained on YPD (10mg là1yeast extract;20mg là1peptone;20mg là1dextrose)agar plates at 30°C and used for expression of BbBEAS,BbBSLS,their mutants or truncated variants,and co-expression of the dissected fragments.The E.coli/S.cerevisiae shuttle vectors YEpADH2p-URA3and YEpADH2p-TRP1were gifts from

Dr.Yi Tang at the University of California,Los Angeles.Gene ampli?cation and plasmid construction.The gene fragments C3(bbBeas), bbBeas-D C3,C1(bbBeas),C1(bbBsls),T2a T2b C3(bbBeas),T2b C3(bbBeas),C3(bbBsls),

C1A1T1C2A2MT(bbBeas),bbBeas-D C1,C1A1T1C2A2MTT2a(bbBeas),bbBsls-D C3,bbBsls-D C1,C1A1T1C2A2MT(bbBsls),T2a T2b C3(bbBsls),T2b C3(bbBsls)and C1A1T1C2A2MT-

T2a(bbBsls)were ampli?ed by PCR from pDY37or pDY42(ref.14)using Phusion High-Fidelity DNA Polymerase(New England Biolabs)with speci?c primers (Supplementary Table2).These PCR products were ligated into the cloning vector pJET1.2to yield16plasmids including pDY83,pDY85,pDY92,pDY93,pDY104, pDY105,pDY106,pDY108,pDY109,pDY111,pDY112,pDY113,pDY135,

pDY136,pDY137and pDY138.These plasmids were con?rmed by digestion checks and gene sequencing.

The C3(bbBeas),T2a T2b C3(bbBeas)and T2b C3(bbBeas)inserts were excised from pDY83,pDY104and pDY105with Nhe I and Pml I and ligated into YEpADH2p-URA3between the same sites to generate pDY87,pDY114and pDY115, respectively.The bbBeas-D C1,bbBsls-D C3,bbBsls-D C1,bbBsls-D T2a T2b C3and bbBsls-D T2b C3inserts were excised from pDY109,pDY112,pDY113,pDY135and pDY138with Spe I and Pml I and ligated into YEpADH2p-URA3between the same sites to generate pDY117,pDY119,pDY121,pDY150and pDY141.The bbBeas-D C3,bbBsls-C3,bbBeas-D T2a T2b C3,bbBeas-D T2b C3,T2a T2b C3(bbBsls)and T2b C3(bbBsls) inserts were excised from pDY85,pDY92,pDY93,pDY106,pDY108,pDY111, pDY136and pDY137with Nde I and Pme I and ligated into YEpADH2p-TRP1 between the same sites to yield pDY88,pDY100,pDY101,pDY116,pDY122, pDY118,pDY140and pDY139,respectively.

Site-directed mutagenesis in pDY37and pDY42that harbour the original bbBeas and bbBsls genes was carried out to construct the mutant plasmids.The PCR conditions were as follows:initial activation of the Phusion High-Fidelity DNA Polymerase for5min at95°C,followed by36cycles of40s denaturation at 95°C,annealing for40s at63°C and extension for15.5min at65°C.PCR products were treated with Dpn I(New England Biolabs)for24h at37°C to remove the templates,ligated and introduced into E.coli XL1-Blue.Colonies were subjected to digestion checks and sequencing to con?rm the correct mutation. pDY145(bbBeas-H2901A),pDY149(bbBeas-D179A),pDY158(bbBeas-S2591A) and pDY162(bbBeas-S2688A)were made from pDY37,while pDY151 (bbBsls-H170A),pDY152(bbBsls-H2861A)and pDY161(bbBsls-D174A)were made from pDY42.For double mutations,pDY183(bbBeas-S2591A)was made from pDY162.

Based on the Asc I site in bbBeas,the gene fragments(AscI)bbBeasàD T2a, (AscI)bbBeas with T2a T2b C3(bbBsls),(AscI)bbBeas with T2b C3(bbBsls)and(AscI)bbBeas with C3(bbBsls)were ampli?ed by splicing by overlap extension PCR using pDY37 and pDY42as the templates.Similarly,(BsrGI)bbBsls with T2a T2b C3(bbBeas), (BsrGI)bbBsls with T2b C3(bbBeas)and(BsrGI)bbBsls with C3(bbBeas)were ampli?ed. These gene fragments were ligated into the cloning vector pJET1.2to yield seven plasmids including pDY165,pDY188,pDY189,pDY190,pDY191,pDY192and pDY222.They were con?rmed by digestion checks and gene sequencing.The (AscI)bbBeas-D T2a,(AscI)bbBeas with T2a T2b C3(bbBsls),(AscI)bbBeas with

T2b C3(bbBsls)and(AscI)bbBeas with C3(bbBsls)inserts were excised with Asc I and Pml I from the corresponding pJET1.2-derived plasmids and ligated into pDY37between the same sites to generate pDY173,pDY201,pDY204and pDY203,respectively. The(BsrGI)bbBsls with T2a T2b C3(bbBeas),(BsrGI)bbBsls with T2b C3(bbBeas)and (BsrGI)bbBsls with C3(bbBeas)inserts were excised with BsrG I and Pml I from the corresponding pJET1.2-derived plasmids and ligated into pDY42between the same sites to generate pDY215,pDY205and pDY224,respectively.

The gene fragments C1(bbBeas),C2(bbBeas),C3(bbBeas),C3(bbBsls)and MT(bbBeas)were ampli?ed by PCR from the genomic DNA of B.bassiana ATCC7,159with Phusion High-Fidelity DNA Polymerase with speci?c primers(Supplementary Table2). These gene fragments were ligated into the cloning vector pJET1.2to yield?ve plasmids including pFC1,pFC62,pFC2,pFC9and pZJ134.The gene fragment C3(bbBeas-H2901A)was ampli?ed from pDY145and ligated into pJET1.2to yield pFC44.These plasmids were con?rmed by digestion checks and gene sequencing. The C1(bbBeas),C3(bbBsls)and MT(bbBeas)inserts were excised from pFC1,pFC9and pZJ134with Nde I and BamH I and ligated into pET28a between the same sites to generate pFC3,pFC11and pJCZ21,respectively.The C2(bbBeas)insert was excised from pFC62with Nde I and Hind III and ligated into pET28a between the same sites to generate pFC63.The C3(bbBeas)and C3(BbBEAS-H2901A)inserts were excised from pFC2and pFC44with Nhe I and BamH I and ligated into pET28a between the same sites to generate pFC4and pFC46,respectively.

The plasmids constructed in this work are shown in Supplementary Table3.

Analysis of the products of the engineered yeast strains.The NRPSs were expressed in S.cerevisiae BJ5464-NpgA.The correct transformants were selected by autotrophy of uracil and/or tryptophan.Yeast strains harbouring one plasmid were cultured in50ml of SC-Ura(or-Trp)dropout medium(6.7mg là1yeast nitrogen base;20mg là1glucose;0.77mg là1-Ura or0.74mg là1-Trp dropout supplement)at30°C with shaking at250rpm.For co-expression experiments, -Trp/-Ura dropout was used.After the OD600value reached0.6,an equal volume of YP medium(10mg là1yeast extract;20mg là1peptone)was added,and the cultures were maintained under the same conditions for an additional3days. The cultures were then extracted three times with100ml of ethyl acetate and subjected to analysis on an Agilent1200HPLC(at210nm)coupled with an Agilent6130single quadrupole mass spectrometer.The product titres were

calculated from three independent experiments based on the standard curves of puri?ed compounds.

Protein expression and puri?cation.C-terminal His6-tagged BbBEAS,BbBSLS and their mutants and truncated variants were expressed in S.cerevisiae

BJ5464-NpgA for protein puri?cation.For1l of yeast culture,the cells were grown at25°C in modi?ed YPD medium(with appropriate dropout supplement)that contains1%dextrose for72h.The cells were collected by centrifugation(4,000rpm for20min),resuspended in20ml of lysis buffer(50mM NaH2PO4,pH8.0,0.15M NaCl,10mM imidazole)and lysed with sonication on ice.Cell debris was removed by centrifugation(35,000g for1h at4°C).Two mililitres of Ni-NTA agarose resin was incubated with the supernatant at4°C for7h.The mixture was loaded into a gravity?ow column.Buffer A(50mM Tris-HCl,pH7.9,2mM EDTA,2mM DTT) with increasing concentrations of imidazole was used as washing buffers.Puri?ed proteins were concentrated and exchanged into the reaction buffer(50mM Tris-HCl,pH8.0).The yields of these modular enzymes were B0.8mg là1.N-terminal His6-tagged C1(BbBEAS)(22.5mg là1),C2(BbBEAS)(35.1mg là1),C3(BbBEAS) (29.2mg là1),C3(BbBEAS-H2901A)(22.1mg là1),C3(BbBSLS)(23.6mg là1)and

MT(BbBEAS)(23.0mg là1)were expressed with the induction of200m M isopropyl b-D-1-thiogalactopyranoside and puri?ed from E.coli BL21(DE3)strains harbouring pFC3,pFC63,pFC4,pFC46,pFC11and pJCZ21,respectively,using a similar Ni-NTA chromatography method.

Puri?cation and structural characterization of8and15.To isolate the two products from reprogrammed BbBEAS and BbBSLS,the corresponding cultures were scaled up to2l and grown at30°C with shaking at250rpm for72h.The ethyl acetate extracts were dried under reduced pressure and subjected to MCI column chromatography,successively eluted with a gradient of methanol-water

(0:100,20:80,50:50and100:0,v/v).The fractions containing the target compounds were further separated by HPLC,eluted with a gradient of acetonitrile–water (80–100%over20min)at a?ow rate of1ml minà1,to yield8(10.1mg)and 15(2.5mg).The puri?ed compounds were identi?ed based on the spectral data.

Chemical preparation of substrates and intermediate products.Beauvericins (1–4)and bassianolide(5)were extracted with ethyl acetate from5l of fermen-tation broths of S.cerevisiae BJ5464-NpgA/pDY37and S.cerevisiae BJ5464-NpgA/ pDY42,respectively.After evaporation of the solvent,the crude extracts were hydrolysed in20ml of acetonitrile–water(1:1)containing0.1N NaOH for48h at 40°C with stirring.The resultant hydrolytic products were extracted by ethyl acetate,loaded on a silica gel60open column,and eluted with a gradient of acetone-hexanes(50,70and100%,v/v)and methanol.The monomer(11or12)-containing fractions were puri?ed by HPLC,washed with40%acetonitrile–water at a?ow rate of1ml minà1to yield71.2mg of11and51.6mg of12.

A mixture of1-ethyl-3-(3-dimethyllaminopropyl)carbodiimide hydrochloride

(0.1mmol)and hydroxybenzotriazole(0.1mmol)was added into15ml of tetrahydrofuran containing0.1mmol11or12.After stirring the resulting mixture for1h at40°C,K2CO3(0.05mmol)was added,and the reaction was stirred for overnight at40°C.The reaction was then stopped and concentrated by rotary evaporation.The resultant cyclized products10and13were puri?ed by HPLC.

A gradient of acetonitrile–water(5–60%over30min)was programmed at a?ow rate of1ml minà1and17.6mg of10and12.2mg of13were puri?ed from the respective reactions.

To prepare D-Hiv-N-Me-L-Phe-SNAC(S3),11(0.1mmol)and SNAC

(0.12mmol)were added to1ml of dimethylformamide in a10-ml round?ask and cooled down to0°C.The solution was then treated with diphenylphosphorylazide (0.15mmol)and triethylamine(0.2mmol).The reaction mixture was stirred for3h before the reaction was stopped and concentrated by rotary evaporation.The resultant monomer-SNAC was puri?ed by HPLC.A gradient of acetonitrile–water (5–90%over30min)was programmed at a?ow rate of1ml minà1to yield

7.7mg of S3.D-Hiv was purchased from Sigma-Aldrich(St Louis,USA),and (D-Hiv-N-Me-L-Phe)2and(D-Hiv-N-Me-L-Phe)3were purchased from the Chinese Peptide Company(Hangzhou,China),which were used to synthesize D-Hiv-SNAC (S1),(D-Hiv-N-Me-L-Phe)2-SNAC(S4)and(D-Hiv-N-Me-L-Phe)3-SNAC(S6) using the same method for S3.

To prepare N-Methyl-L-Phe-SNAC(S2),L-Phe-SNAC was synthesized?rst. t-Butyloxy carbonyl(Boc)-L-Phe(1mmol),N,N-dicyclohexylcarbodiimide

(1mmol)and1-hydroxybenzotriazole(1mmol)were dissolved in15ml of

tri?uoroacetic acid(TFA),followed by adding N-acetylcysteamine(1mmol).After stirring the resulting solution for45min at25°C,K2CO3(0.5mmol)was added and the reaction mixture was stirred for an additional3h at25°C.After?ltration, the solvent was removed in vacuo.The resulting residue was dissolved in ethyl acetate and washed once with an equal volume of10%aqueous NaHCO3.The organic layer was dried by MgSO4,?ltered and concentrated in vacuo.The crude product was subjected to silica gel column chromatography,eluted with4%(v/v) MeOH-CHCl3,to afford Boc-L-Phe-SNACs.Boc group was removed by dissolving Boc-L-Phe-SNAC in50%TFA/CH2Cl2and stirring at25°C for1h.After removing solvent,the residue was taken up in a minimal volume of CH2Cl2and precipitated with ether.The resulting solid was washed twice with ether and dried to afford L-Phe-SNAC.

A typical methylation reaction of L-Phe-SNAC(100m l)consisted of6.4m M MT(BbBEAS),0.8mM L-Phe-SNAC and2.4mM SAM in100mM Tris-HCl buffer (pH7.5).The reaction mixtures were incubated at25°C for30min and then quenched with MeOH(50m l).The mixtures were brie?y vortexed and centrifuged at15,000rpm for5min to remove the precipitated protein before the samples were injected into LC–MS for analysis.The supernatants were analysed by HPLC under 235nm,eluted with an increasing gradient of acetonitrile(10–15%)in H2O containing0.1%TFA with a?ow rate of1ml minà1.Puri?cation of S2was carried out using the same HPLC method.

In vitro enzymatic studies.For a typical in vitro NRPS activity assay,a400m l reaction contained50mM Tris-HCl buffer(pH8.0),2m M NRPS(or dissected fragments),5mM ATP,25mM MgCl2,7.5mM L-Leu or L-Phe,2.25mM D-Hiv and 3mM SAM.After3h incubation at25°C,the reactions were quenched with methanol for LC–MS analysis.

To test the condensation activity of C2,S1(0.3mM)and S2(0.3mM)or S4 (84m M)were incubated with30m M C2domain in1ml of50mM Tris-HCl buffer (pH7.8)at25°C for12h.

To test the condensation activity of C1,S1(0.3mM)and S2(0.3mM)or S4 (84m M)were incubated with30m M C1domain in1ml of50mM Tris-HCl buffer (pH7.8)at25°C for12h.

To test the condensation activity of C3,S4(84m M),S1(0.3mM)and S2 (0.3mM,when needed)were incubated with30m M C3domain in1ml of50mM Tris-HCl buffer(pH7.8)at25°C for12h.

To test the macrocyclization activity of C1and C3,the SNAC substrates S3,S4 or S6(75m M)were,respectively,incubated with30m M C1or C3domain or their mutants in100m l of50mM Tris-HCl buffer(pH7.8)at25°C for4h.

For in vitro total biosynthesis,S1(1mM)and S2(1mM)were incubated with C2domain(50m M)and C3domain(50m M)in2ml of50mM Tris-HCl buffer (pH7.8)at25°C for12h.

Data availability.The authors declare that the data supporting the?ndings of this study are available within the article and its Supplementary Information Files.The GenBank accession numbers of BbBEAS and BbBSLS are ACI30655and

ACR78148,respectively.All other data supporting the?ndings of this study are available from the corresponding author on reasonable request.

References

1.Fischbach,M.A.&Walsh,C.T.Assembly-line enzymology for polyketide and

nonribosomal peptide antibiotics:logic,machinery,and mechanisms.Chem.

Rev.106,3468–3496(2006).

2.Keating,T.A.,Marshall,C.G.,Walsh,C.T.&Keating,A.E.The structure of

VibH represents nonribosomal peptide synthetase condensation,cyclization and epimerization domains.Nat.Struct.Biol.9,522–526(2002).

3.Mootz,H.D.,Schwarzer,D.&Marahiel,M.A.Ways of assembling complex

natural products on modular nonribosomal peptide synthetases.ChemBioChem 3,490–504(2002).

4.Sussmuth,R.,Muller,J.,von Dohren,H.&Molnar,I.Fungal cyclooligomer

depsipeptides:from classical biochemistry to combinatorial biosynthesis.Nat.

Prod.Rep.28,99–124(2011).

5.Zhan,J.,Burns,A.M.,Liu,M.X.,Faeth,S.H.&Gunatilaka,A.A.L.Search for

cell motility and angiogenesis inhibitors with potential anticancer activity: beauvericin and other constituents of two endophytic strains of Fusarium oxysporum.J.Nat.Prod.70,227–232(2007).

6.Waetjen,W.et al.Enniatins A1,B and B1from an endophytic strain of

Fusarium tricinctum induce apoptotic cell death in H4IIE hepatoma cells

accompanied by inhibition of ERK phosphorylation.Mol.Nutr.Food Res.53, 431–440(2009).

7.Chen,B.F.,Tsai,M.C.&Jow,G.M.Induction of calcium in?ux from

extracellular?uid by beauvericin in human leukemia cells.Biochem.Biophys.

https://www.sodocs.net/doc/3b3328066.html,mun.340,134–139(2006).

8.Shiomi,K.et al.Verticilide,a new ryanodine-binding inhibitor,produced by

Verticillium sp.FKI-1033.J.Antibiot.63,77–82(2010).

9.Xu,Y.et al.Biosynthesis of the cyclooligomer depsipeptide beauvericin,a

virulence factor of the entomopathogenic fungus Beauveria bassiana.Chem.

Biol.15,898–907(2008).

10.Xu,Y.Q.et al.Biosynthesis of the cyclooligomer depsipeptide bassianolide,

an insecticidal virulence factor of Beauveria bassiana.Fungal Genet.Biol.46, 353–364(2009).

11.Jirakkakul,J.et al.Identi?cation of the nonribosomal peptide synthetase gene

responsible for bassianolide synthesis in wood-decaying fungus Xylaria sp.

BCC1067.Microbiology154,995–1006(2008).

12.Trauger,J.W.,Kohli,R.M.,Mootz,H.D.,Marahiel,M.A.&Walsh,C.T.

Peptide cyclization catalysed by the thioesterase domain of tyrocidine

synthetase.Nature407,215–218(2000).

13.Hoyer,K.M.,Mahlert,C.&Marahiel,M.A.The iterative gramicidin S

thioesterase catalyzes peptide ligation and cyclization.Chem.Biol.14,13–22 (2007).

14.Yu,D.et al.Engineered production of fungal anticancer cyclooligomer

depsipeptides in Saccharomyces cerevisiae.Metabol.Eng.18,60–68(2013).

15.Gaudelli,N.M.,Long,D.H.&Townsend,C.A.b-Lactam formation by

a non-ribosomal peptide synthetase during antibiotic biosynthesis.Nature520,

383–387(2015).

16.Bergendahl,V.,Linne,U.&Marahiel,M.A.Mutational analysis of the C-domain

in nonribosomal peptide synthesis.Eur.J.Biochem.269,620–629(2002). 17.Samel,S.A.,Schoena?nger,G.,Knappe,T.A.,Marahiel,M.A.&Essen,L.O.

Structural and functional insights into a peptide bond-forming bidomain from

a nonribosomal peptide synthetase.Structure15,781–792(2007).

18.Ehmann,D.E.,Trauger,J.W.,Stachelhaus,T.&Walsh,C.T.Aminoacyl-

SNACs as small-molecule substrates for the condensation domains of

nonribosomal peptide synthetases.Chem.Biol.7,765–772(2000).

19.Yu,D.,Xu,F.,Gage,D.&Zhan,J.Functional dissection and module swapping

of fungal cyclooligomer depsipeptide https://www.sodocs.net/doc/3b3328066.html,mun.49,

6176–6178(2013).

20.Keating,T.A.et al.Chain termination steps in nonribosomal peptide

synthetase assembly lines:directed acyl-S-enzyme breakdown in antibiotic and siderophore biosynthesis.ChemBioChem2,99–107(2001).

21.Tang,G.L.,Cheng,Y.Q.&Shen,B.Polyketide chain skipping mechanism in

the biosynthesis of the hybrid nonribosomal peptide-polyketide antitumor antibiotic leinamycin in Streptomyces atroolivaceus S-140.J.Nat.Prod.69, 387–393(2006).

22.Gao,X.et al.Cyclization of fungal nonribosomal peptides by a terminal

condensation-like domain.Nat.Chem.Biol.8,823–830(2012).

23.Marshall,C.G.,Hillson,N.J.&Walsh,C.T.Catalytic mapping of the

vibriobactin biosynthetic enzyme VibF.Biochemistry41,244–250(2002). 24.Keating,T.A.,Miller,D.A.&Walsh,C.T.Expression,puri?cation,and

characterization of HMWP2,a229kDa,six domain protein subunit of

yersiniabactin synthetase.Biochemistry39,4729–4739(2000).

25.Shaw-Reid,C.A.et al.Assembly line enzymology by multimodular nonribosomal

peptide synthetases:the thioesterase domain of E.coli EntF catalyzes both

elongation and cyclolactonization.Chem.Biol.6,385–400(1999).

26.Kadi,N.,Oves-Costales,D.,Barona-Gomez,F.&Challis,G.L.A new family of

ATP-dependent oligomerization-macrocyclization biocatalysts.Nat.Chem.

Biol.3,652–656(2007).

27.Kadi,N.,Arbache,S.,Song,L.,Oves-Costales,D.&Challis,G.L.Identi?cation

of a gene cluster that directs putrebactin biosynthesis in Shewanella species: PubC catalyzes cyclodimerization of N-hydroxy-N-succinylputrescine.J.Am.

Chem.Soc.130,10458–10459(2008).

28.Kadi,N.,Song,L.&Challis,G.L.Bisucaberin biosynthesis:an adenylating

domain of the BibC multi-enzyme catalyzes cyclodimerization of https://www.sodocs.net/doc/3b3328066.html,mun.41,5119–5121(2008).Acknowledgements

This work was supported by a Utah State University Research Catalyst grant and grants from the National Natural Science Foundation of China(31170763,31470787).We are grateful to Dr.Yi Tang,the University of California at Los Angeles,for the YEpADH2p vectors,and Dr.Nancy Da Silva,the University of California at Irvine,for the S.cere-visiae BJ5464-NpgA strain used in this research.

Author contributions

J.Z.conceived of the overall idea and designed the experiments.D.Y.conducted the plasmid construction,protein expression and puri?cation in S.cerevisiae and enzymatic studies on the intact and dissected enzymes.F.X.carried out the fermentation of engineered strains,protein expression and puri?cation of C1,C3and MT in E.coli,as well as LC–MS analysis and structural characterization of the products.S.Z.prepared and characterized the monomers from the hydrolysis of beauvericin and bassianolide.

D.Y.and F.X.synthesized the SNAC derivatives.All authors analysed and discussed the results.J.Z.,D.Y.and F.X.prepared this manuscript.

Additional information

Supplementary Information accompanies this paper at https://www.sodocs.net/doc/3b3328066.html,/ naturecommunications

Competing interests:The authors declare no competing?nancial interests.

Reprints and permission information is available online at https://www.sodocs.net/doc/3b3328066.html,/ reprintsandpermissions/

How to cite this article:Yu,D.et al.Decoding and reprogramming fungal iterative nonribosomal peptide https://www.sodocs.net/doc/3b3328066.html,mun.8,15349doi:10.1038/ncomms15349 (2017).

Publisher’s note:Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional af?liations.

This work is licensed under a Creative Commons Attribution4.0

International License.The images or other third party material in this article are included in the article’s Creative Commons license,unless indicated otherwise in the credit line;if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license,visit https://www.sodocs.net/doc/3b3328066.html,/licenses/by/4.0/

r The Author(s)2017