Substrate Specificity of Acyltransferase Domains for Efficient Transfer of Acyl Groups

Acyltransferase domains (ATs) of polyketide synthases (PKSs) are critical for loading of acyl groups on acyl carrier protein domains (A) via self- and trans-acylation reactions, to produce structurally diverse polyketides. However, the interaction specificity between ATs and unusual acyl units is rarely documented. In Streptomyces tsukubaensis YN06, we found that AT4FkbB [an AT in the fourth module of tacrolimus (FK506) PKS] transferred both allylmalonyl (allmal) and emthylmalonyl (ethmal) units to ACPs, which was supposed responsible for the production of both FK506 and its analog FK520, respectively. Mutations of five residues in AT4FkbB (Q119A, L185I-V186D-V187T, and F203L) caused decreased efficiency of allmal transfer, but a higher ratio of ethmal transfer, supposedly due to less nucleophilic attacks between Ser599 in the active site of AT4FkbB and the carbonyl carbon in the allmal unit, as observed from molecular dynamics simulations. Furthermore, reverse mutations of these five residues in ethmal-specific ATs to the corresponding residues of AT4FkbB increased its binding affinity to allmal-CoA. Among these residues, Val187 of AT4FkbB mainly contributed to allmal recognition, and V187K mutant produced less FK520 than wild type. Our findings thus suggested that five critical residues within AT4FkbB were important for AT functionality in polyketide extension and potentially for targeting biosynthesis by generating desirable products and eliminating undesirable analogs.


INTRODUCTION
Polyketide natural products function as a wide range of therapeutic agents, such as immunosuppressants (Mo et al., 2011), anti-cancer agents , antibiotics (Urem et al., 2016), which are invaluable natural resources for pharmaceutical development (Fegan et al., 2010;Mo and Suh, 2016). Many polyketides are produced by type I assembly line PKSs containing several catalytic modules, each of which is responsible for one round of polyketide chain elongation. The essential catalytic module is basically comprised of β-ketoacyl synthase domain (KS), AT, and ACP domain (Fischbach and Walsh, 2006;Khosla et al., 2007;Fegan et al., 2010;. The KS within each module accepts a polyketide chain from the upstream ACP and condenses acyl units bound to KS and ACP. AT selects the appropriate acyl unit and load it onto the ACP via two-step reactions: a self -acylation reaction to recognize acyl units from acyl donors to the active site Ser to form acyl-O-AT intermediates; and a trans-acylation reaction to transfer the acyl units from AT intermediates to the phosphopantetheine arm of ACP to form acyl-S-ACPs Ye et al., 2014).
Loading several different acyl units to produce structurally diverse polyketide backbones depends on natural ATs in modular PKSs specifically (Musiol and Weber, 2012;. Studies on ATs mainly focus on understanding the catalytic mechanism and substrate specificity at present. The catalytic processes of several ATs have been interpreted for their appropriate transfer of acyl groups (Marsden et al., 1994;Musiol et al., 2013;Jiang et al., 2015;Wang Y.Y. et al., 2015). AT3 in 6-Deoxyerythronolide B Synthase (DEBS) recognizes only the electrophilic malonyl (M) component but not methylmalonyl (MM) component, which is strictly specific in the self -acylation reaction . The discrete AT KirCI of kirromycin PKS transfers only M not MM or ethylmalonyl (ethmal) unit to ACP, which is specific in the tranacylation reaction (Musiol et al., 2013). Structural studies of ATs have elucidated the catalytic mechanism and the key residues contributing to substrate specificity (Wong et al., 2011;Liew et al., 2012;Sundermann et al., 2013;. Cocrystallization of AT DYN10 with acyl groups showed that AT DYN10 protects the M-enzyme intermediate to specifically facilitate the transfer of M to ACP and the bulky residues Phe752 and Met680 are primarily specific for M binding (Liew et al., 2012). Gln198 of AT6 in DEBS promotes the activation of the thioester group and helps to keep MM-CoA in a conformation suitable for the nucleophilic attack from the active site Ser197 (Sundermann et al., 2013). However, the mechanism of substrate specificity between ATs and acyl units still remains unclear.
FK506 and its analog FK520 as secondary metabolites are coproduced by Streptomyces tsukubaensis Huang et al., 2013). In our industrial strain S. tsukubaensis YN06, the yield of FK506 and FK520 in shake-flask fermentation is 103.3 and 17.0 mg/L respectively, showing a ratio of FK506 : FK520 of 6.1:1 ( Figure 4F). This process requires the transfer of a unique allylmalonyl (allmal) unit, synthesized by TcsA-D encoded within FK506 gene cluster (Goranovič et al., 2010;Mo et al., 2011;Barreiro and Martínez-Castro, 2014), and ethmal unit by AT4 FkbB (AT in the fourth module of FK506 PKS) . So here we used AT4 FkbB as a model to study the molecular basis for substrate specificity of acyl units. We found AT4 FkbB transfers both allmal-and ethmal-CoA onto diverse ACPs in self -and trans-acylation reactions simultaneously. Five residues were identified to be potentially important for AT4 FkbB in allmal group recognition by protein alignment between AT4 FkbB and ethmal-specific ATs. The essentiality of these key AAs of AT4 FkbB were tested by mutations and it was found that those five AAs controlled allmal specificity. This information allowed us to enhance the transferred efficiency of allmal by AT4 FkbB via a single AA mutation. Our study thus provides new insights into the molecular basis of substrate specificity of ATs, and will contribute to targeting biosynthesis of structure-specific polyketides.
For biochemical analyses of trans-acylation reactions, a reaction mixture in 50 µl contains 1.6 µM ATs, 16 µM allmal-CoA (ethmal-CoA or allmal-CoA:ethmal-CoA = 1:1, M-CoA, MM-CoA), 16 µM holo-ACPs obtained above and 100 mM Tris-HCl, pH 8.0. The reaction mixture was incubated at 25 • C for 2 h. ATs include AT4 FkbB from FK506 and FK520 PKS, AT4 TiA2 from fidaxomicin PKS, AT3 lsd12 from lasalocid PKS, AT5 monAIV from monensin PKS and their mutants. However, the concentration of allmal-or ethmal-CoA was changed to 2 mM and other conditions were unchanged in the reactions of AT4 TiA2 and its mutants in fidaxomicin PKS. holo-ACPs include holo-ACP2/4 FkbB and ACP10 FkbA . Allylmalonylation and ethylmalonylation conversion yields in trans-acylation reactions, comparing ion peak heights of FIGURE 2 | Amino acid sequence alignments and phylogenetic tree of AT4 FkbB of FK506 PKS and ethmal-specific ATs. (A) Amino acid sequence alignments between AT4 FkbB and ethmal-specific ATs. Sequences colored in yellow and blue or green were corresponded to the results of identify positions and consensus positions. The residues in AT4 FkbB labeled by red stars were mutated to the correspondingly conserved residues in ethmal-specific ATs. The residues labeled by black line marks were the catalytic centers of ATs. (B) Phylogenetic tree of AT4 FkbB and ethmal-specific ATs. A phylogenetic tree was generated using Mega 5.05 neighbor-joining method. ATs are shown as the locations in PKSs and names of corresponding polyketides in parenthesis. The selection of 18 ethmal-specific ATs were from PKSs of FK520, fidaxomicin, lasalocid, monensin, kirromycin, phoslactomycin, meridamycin, salinomycin, indanomycin, oligomycin, tautomycetin, concanamycin A, niddamycin, and phoslactomycin. acylated and intact protein in MS, were selected the same to HPLC.

HPLC-MS Analysis for Acyl-Transfer
The reaction mixtures were analyzed on HPLC-MS equipped with an Agilent 1200 HPLC system (Agilent, Santa Clara, CA, United States) and a Thermo Finnigan LCQDeca XP Max LC/MS system (Thermo Finnigan, Waltham, MA, United States). HPLC separation was performed on an Agilent SB-C18 column (3.5 um particle size, 80 Å, 2.1 mm × 150 nm) at 35 • C. Solvent A was water containing 0.1% formic acid. Solvent B was acetonitrile. HPLC condition for self -acylation reaction: a linear gradient from 90 to 70% solvent A from 0 to 5 min, a linear gradient from 70 to 60% solvent A from 5 to 40 min, a linear gradient from 60 to 50% solvent A from 40 to 60 min, a linear gradient from 50 to 30% solvent A from 60 to 70 min. While, HPLC condition for trans-acylation reaction: a linear gradient from 90 to 70% solvent A from 0 to 5 min, a linear gradient from 70 to 50% solvent A from 5 to 55 min, a linear gradient from 50 to 30% solvent A from 55 to 60 min. The equilibration to initial conditions was for 13 min at a flow rate of 0.2 mL/min. UV detection was performed at both 220 and 280 nm. MS with an electrospray ionization (ESI) source was performed as follows: positive mode, source voltage of 2.5 kV, capillary voltage of 41 V, sheath gas flow of 45 arbitrary units, auxiliary/sweep gas flow of 5 arbitrary units, capillary temperature 330 • C.

Docking Models and Molecular Dynamics Simulations (MDs)
The homology models of [KS4][AT4] FkbB and ACP4 FkbB of FK506 PKS were constructed by SWISS-MODEL 1 . Before 1 | The activities of AT4 FkbB and its mutants of Fk506 PKS with allmal-or ethmal-CoA as the substrate in self-and trans-acylation reactions.

Mutant
Self-acylation reaction (%) Trans-acylation reaction (%) The process that PAC-B18 containing AT4 FkbB in the FK506 gene cluster was mutated from amino acid 187 Val (GTC) to Lys (AAG), is depicted in Supplementary Figure S1. Briefly, the PAC-B18 plasmid was electroporated into E. coli DH10β harboring pSC101-ccdA-gbaA. CcdA is as antidote for the counterselectable agent CcdB, which expressed to be toxic to E. coli (Wang et al., 2014). The linear targeting molecule containing ccdB-amp was amplified from p15A-ccdB-amp by PCR using primers 67-68. The first recombinants were cultured on LB plates containing ampicillin and analyzed by sequence. In the second round of recombineering, the linear targeting AT4 FkbB fragment was amplified from pET28a-V187K using primers 69-70 for electroporation. The second recombinants were incubated on LB plates at 37 • C overnight and analyzed by sequence using primers 69-70. The resulting recombinant was named as PAC-B18-V187K in following experiments.

Introduction of PACs Into S. tsukubaensis YN06 and Fermentation
The E. coli DH10β containing PAC-B18 or PAC-B18-V187K (Apra R ) was used in a dipparental mating with the other E. coli strain ET12567 that contained the RP4 derivative pUZ8002 (Kan R ) respectively (Supplementary Figure S2). The method of dipparental mating was accoding to Jones et al. (2013) to obtain the ET12567 strains containing PAC-B18 or PAC-B18-V187K (Apra R ) and pUZ8002 (Kan R ). Then the obtained ET12567 strains were conjugated with the mycelia on ISP4 medium with 20 mM MgCl 2 and 50 mM CaCl 2 after 20 h with 20 µg/ml apramycin and 25 µg/ml nalidixic acid for 6-9 days at 28 • C. Exconjuants were streaked on ISP4 plates with 20 µg/ml apramycin and further confirmed by PCR amplification, T-vector construction, and sequenced using the primers 69-70. The resulting S. tsukubaensis strains YN06-01 (PAC-B18), YN06-02 (PAC-B18-V187K) and wild type were grown on ISP4 plates for 5-9 days at 28 • C to allow colonies to sporulate for fermentation. For fermentation, an agar piece of approximately 1 cm 2 was inoculated into a 250-ml flask containing 50 ml of the seed medium, consisting of 2% TSB and 5% PEG6000, and was maintained at 28 • C and 220 rpm for 24 h. The seed culture to final concentration of OD 600 = 0.4 was inoculated into a 250-ml flask containing 30 ml of the fermentation medium (5% maltodextrin, 1% yeast extract, 3% cotton seed meal, 0.2% K 2 HPO 4 , 0.1% CaCO 3 , pH 6.8). The culture condition was maintained at 28 • C and 220 rpm for 5 days (Wang et al., 2016).

Detection of FK506 and FK520
Culture samples (2 ml) from each S. tsukubaensis strain were withdrawn and ultrasonic extracted with 500 µl of methanol. Methanol layer was recovered by centrifugation at 12000 rpm for 15 min. The concentration of FK506 and FK520 was determined using an HPLC system (Agilent Series 1100, Agilent) equipped with a SB-C18 column (150 mm × 2.1 mm, Agilent). The column temperature was maintained at 60 • C and UV detector was set at 215 nm. The mobile phase, which had a flow rate of 1.0 mL/min, contained 0.1% H 3 PO 4 solution (A) and acetonitrile (B). HPLC condition: a linear gradient from 40 to 70% solvent B from 0 to 10 min, a linear gradient at 70% solvent B from 10 to 35 min. The FK506 and FK520 titers were calculated with standard curves.

Preference of AT4 FkbB on Allmal-CoA Transfer
We have reported that AT4 F kbB in FK506 PKS can transfer both allmal and ethmal units to ACP4 FkbB , accounting for simultaneous production of FK506 and FK520 in S. tsukubaensis YN06 . To further understand the possible competition between allmal and ethmal units to AT4 F kbB in self -acylation reactions, equal molar ratio of allmal-and ethmal-CoA was added to AT4 F kbB in vitro assays. HPLC data showed that a new large peak appeared along with AT4 FkbB (Figures 1A-D), and this HPLC peak contained two products, which were identical to the biochemically synthesized allmal-and ethmal-AT4 FkbB based on mass spectrum (MS) data, indicating self -acylation has occurred on AT4 FkbB . Allylmalonylation and ethylmalonylation conversion yields were estimated to be 46.4 and 38.6%, respectively, by comparing the ion peak strength of both acylated and un-acylated proteins on MS which selected was the same to the retention time of HPLC Wang Y.Y. et al., 2015) (Figures 1M-P). These data suggested that allmal-CoA is the preferred substrate for AT4 FkbB in formation of intermediates in the self -acylation reactions.
To further discover the competition of allmal and ethmal units onto the acyl receptor ACPs in trans-acylation reactions, equal molar of allmal and ethmal units were incubated with AT4 F kbB and different holo-ACPs, including holo-ACP2 FkbB , holo-ACP4 FkbB , and holo-ACP10 FkbA (ACP in the 2nd, 4th, and 10th module of FK506 PKS to accept MM units, both allmal and ethmal units, M unit, respectively). We observed appearance of one or two new peaks with all holo-ACPs (Figures 1E-G,I,K), compared to control reactions without AT4 FkbB (Figures 1H,J,L). Both the retention time and the MS data confirmed that these peaks were allmal-and ethmal-ACPs (Figures 1E-L,Q-X). The acylation ratios of allmal and ethmal onto ACP2 FkbB , ACP4 FkbB , and ACP10 FkbA were 44.3 and 37.9% (Figure 1Q), 42.1 and 35.2% (Figure 1U), 44.4 and 37.6% (Figure 1W), respectively. These TABLE 2 | The activities of AT4 FkbB of FK520 PKS, AT4 TiA2 of fidaxomicin PKS, AT3 lsd12 of lasalocid PKS and AT5 monAIV of monensin PKS and their mutants with allmalor ethmal-CoA as the substrate in self-acylation reactions and with allmal-or ethmal-, M-, MM-CoA as the substrate in trans-acylation reactions.

Mutant
Self-acylation reaction (%) Trans-acylation reaction (%) results suggested that allmal-CoA is preferred over ethmal-CoA as a substrate for AT4 FkbB in transferring the acyl group to diverse ACPs in the trans-acylation reactions, and that the specificity of acyl groups is determined by AT4 FkbB rather than ACPs. And ACP10 FkbA was chosen as an acyl receptor in the biochemical assays below.
Gln119, Leu185-Val186-Val187, and Phe203 Residues Are Critical for Allmal Transfer by AT4 FkbB Our above results suggested that allmal-CoA was preferred over ethmal-CoA by AT4 F kbB in both self -and trans-acylation reactions for FK506 biosynthesis. Next, protein alignment of AT4 FkbB with 18 ethmal-specific ATs from other diverse PKSs was demonstrated that 14 residues (Lys54-Ser55, Met69, Gln119, Thr156, Ile160, Thr183, Leu185-Val186-Val187, Cys189-Pro190-Thr191, and Phe203) of AT4 FkbB might be involved in the discriminative recognition of allmal-and ethmal groups, since residues in these positions from ethmal-specific ATs are highly conserved (Figure 2A). Then all these residues in AT4 FkbB were mutated to the corresponding AAs in ethmal-specific ATs individually, and all the mutant proteins were expressed and purified from E. coli for in vitro biochemical assays. Notably, both Q119A and L185I-V186D-V187T mutants showed less than half self -acylation activities of wild type AT4 FkbB with allmal-CoA, and dramatically decreased trans-acylation activities by 28 and 36%, respectively. However, these mutants had no significant difference in enzymatic activities with ethmal-CoA as the substrate in either self -or trans-acylation reactions ( Table 1). Additionally, F203L mutant of AT4 FkbB led to dramatic decrease in trans-acylation activity with allmal-CoA as the substrate. However, it had a higher trans-acylation activity with ethmal-CoA as the substrate (Table 1). These results suggested that mutations of Q119A and L185I-V186D-V187T might have negative effects on allmal-AT4 FkbB intermediate formation during self -acylation reactions, leading to less allylmolonylation of ACP10 FkbA in trans-acylation reactions, while F203L has an important role on allmal transfer to ACP10 FkbA in transacylation reaction. The other 11 mutants displayed similar acylation activities to the wild type ( Table 1), suggesting that the residues at these positions had no significant impacts on AT4 FkbB in recognizing and transferring the allmal unit to ACP10 FkbA . A Q119A-L185I-V186D-V187T-F203L mutant was further constructed and displayed a lower (<16%) trans-acylation activity with allmal-CoA but a higher activity with ethmal-CoA (Table 1). Furthermore, all 15 mutants had very similar circular dichroism (CD) spectra to the wild type AT4 FkbB , suggesting that they had similar secondary structures ( Supplementary  Table S1). Cumulatively, these data suggested that residues Gln119, Leu185-Val186-Val187, and Phe203 of AT4 FkbB are crucial for determining the substrate specificity.  ± 0.1 Å), indicating the roles of His702 in activating the catalytic residue Ser599 on nucleophilic attack on the allmal or ethmal thioester (Ser599 and Ser89 are the same site in [KS4][AT4] FkbB and AT4 FkbB , respectively). And the distance between the nucleophilic hydroxyl oxygen atom of Ser599 (O) and the electrophilic carbonyl carbon atoms along with side chain carboxyl of allmal (C) and ethmal (C) were 3.3 ± 0.2 and 3.2 ± 0.2 Å, respectively (Figures 3A,C,E,G) (Figures 3F,H). Our results suggested that mutations (Q119A, L185I, V186D, V187T, and F203L) reduced allmal specificity due to less nucleophilic attacks between Ser599 and allmal for formation of the allmal-enzyme intermediate.
(F) Production of FK506 and FK520 in WT, YN06-01 and YN06-02. Analyses of variance were conducted to determine the difference of WT and mutants using SPSS 20. The LSD multiple range tests were evaluated for significant differences among WT and mutants (P < 0.01). The data are expressed as mean ± SE (n = 3). * P < 0.01. but no altered self -and trans-acylation activities with ethmal-CoA as a substrate compared with wild type AT4 FkbB of FK520 PKS ( Table 2). The results suggested that mutations of L145Q, I211L-A212V, and L229F in AT4 FkbB of FK520 PKS enhanced acyl unit specificity on allmal unit. Similar results were observed in mutants of AT4 TiA2 , AT3 lsd12 , and AT5 monAIV ( Table 2), further confirming the essentiality of these five residues for allmal specificity.
Moreover, other commonly used acyl units, such as Mand MM-CoA, were used for specificity tests. Both acyl units were incubated with all above mutants and their corresponding wild type versions of ATs (AT4 FkbB , AT4 TiA2 , AT3 lsd12 , and AT5 monAIV ) in trans-acylation reactions. LC-MS data showed that none of all mutants and their wild type counterparts had detectable trans-acylation activities with M-or MM-CoA, except that AT5 monAIV had activities with MM-CoA as previously reported (Pospisil et al., 1994;Leadlay et al., 2001), while the activities of its mutants to MM-CoA also remained ( Table 2). Furthermore, the secondary structures of all mutants were similar to those of their wild type proteins as determined by CD spectra (Supplementary Table S1). These data further suggested that these five residues are specific for controlling allmal uploading by ATs.

V187K of AT4 FkbB Enhancing Allmal Specificity in FK506 PKS
We showed the roles of residues Gln119, Leu185-Val186-Val187, and Phe203 in controlling allmal specificity of ATs. Next we attempted to enhance the substrate specificity of FK506 PKS AT4 FkbB on the allmal group. We found that three residues Leu185-Val186-Val187 had the most important roles in controlling allmal recognition by AT4 FkbB based on the allymalonylation conversion data (Tables 1, 2). From the position of Leu695-Val696-Val697 with allmal, Val697 constrained the size of the allmal carbon chain to prevent more spacious polyketide building blocks such as ethmal, MM and M from incorporating to generate the diversely undesirable analogs (Supplementary Figure S2A).
To test the important role of Val187 (Val697 and Val187 are the same site in [KS4][AT4] FkbB and AT4 FkbB respectively) on allmal group recognition, V187A mutant was constructed and showed decreased trans-acylation activity to ∼60% of WT AT4 FkbB with allmal-or ethmal-CoA as the substrate as expected ( Table 1) Figure S2B). Thereafter, mutants of Val187 to the six individual AAs in top scores were constructed for in vitro self -and trans-acylation reactions. We found that V187K mutant increased the ratio of allmal/ethmal from 1:1 to 2:1 (Figure 4A), while the overall activities of V187K were similar to WT in transacylation reactions ( Table 1). The ratio of allmal/ethmal in other five mutants also increased ( Figure 4A), but the overall activities were lower than WT (Table 1). To further test that V187K mutant had enhanced activities of allmal transfer, equal molar ratio of allmal and ethmal units were incubated with V187K mutant in trans-acylation reactions. HPLC data showed a new peak allmal-ACP10 FkbA , larger than that of ethmal-ACP10 FkbA in size (Figures 4B,C), as confirmed by MS (Figures 4D,E). Allylmalonylation and ethylmalonylation conversion yields were estimated 57.2 and 28.3%, respectively, by comparing ion peak heights of acylated and intact proteins in MS (Figures 4D,E). These experiments suggested that targeted point mutation V187K in AT4 FkbB of FK506 PKS could enhance the substrate specificity for allmal unit.
Based on the biochemistry results of V187K, we introduced PAC-B18-V187K or PAC-B18 into S. tsukubaensis YN06 for the production of FK506 and FK520 (Supplementary Figure S2). The FK506 and FK520 production (172:29.1 mg/L) of strain YN06-01 (PAC-B18) in shake-flask fermentation had obvious change compared to that of S. tsukubaensis YN06 (103.3:17.0 mg/L). But the ratio of them in YN06-01 was the same to wild type about ∼6:1 (Figure 4F and Supplementary Figure S3), suggesting that the exogenous FK506 gene cluster PAC works in wild type. In strain YN06-02 (PAC-B18-V187K), the yield of FK506 and FK520 was 175.0 and 22.7 mg/L and the ratio of them was about ∼7.7:1 ( Figure 4F and Supplementary Figure S3), suggesting V187K in AT4 FkbB of FK506 PKS could decrease the production of FK520.

DISCUSSION
Selection of the suitable acyl units in self -and trans-acylation reactions to load the starter or extending units for PKSs to produce diverse and biologically active natural products is dependent on ATs Bravo-Rodriguez et al., 2014). Full understanding of interaction between ATs and the starter or extending units is required to overcome limitations related to the substrate specificity of ATs. Although several structural studies of ATs that utilize malonyl-CoA have been demonstrated (Liew et al., 2012), little is known regarding the specificity of ATs on unusual acyl-CoA. In this study, AT4 FkbB was selected as a model to analyze the recognition specificity on the extending units allmal and ethmal, since AT4 FkbB can transfer both allmal-and ethmal-CoA onto diverse ACPs in selfand trans-acylation reactions and allmal-CoA is preferred over ethmal-CoA as a substrate (Figures 1E-L,Q-X).
There are very few reports about the key AAs in ATs responsible for the intrinsic substrate specificity. Most studies focused on some sequence motifs from M-and MM-specific ATs (Haydock et al., 1995;Reeves et al., 2001;Vecchio et al., 2003). It was reported that replacement of Gln with Leu in the active site GHSQGE motif produced a mixture of 6dEB and the corresponding 6-desmethyl-6-dEB (Reeves et al., 2001). However, our results showed that allmal-specific motif GHSQGE in AT4 FkbB is the same to the most of ethmal-specific ATs. The M unit was incorporated after mutation of YASH to YAFH or HASH in AT6 of DEBS, where YASH and HAFH are the conserved sequence motifs for M and MM specificity, respectively (Vecchio et al., 2003). The motif CPTH in AT4 FkbB were mutated to ethmal-specific motif VASH, resulting in no altered enzymatic activities with allmal-and ethmal-CoA as substrates compared to WT (Table 1). Therefore, uploading of allmal or ethmal group into polyketide backbone is not dependent on the conserved motifs GHSQGE or CPTH in AT4 FkbB. Gln119, Leu185-Val186-Val187, and Phe203 residues are newly identified residues in controlling allmal specificity, and confirmed by reverse mutations and functional analysis (Tables 1,  2). Furthermore, Mutant Q119A-L185I-V186D-V187T-F203L transferred less allmal because of possible less nucleophilic attacks between the active site Ser599 and allmal unit as observed from MDs (Figures 3B,D). Here, we for the first time found that Gln119, Leu185-Val186-Val187, and Phe203 residues in AT4 FkbB were critical in determining the specificity toward allmal unit.
On Tab. 2, F203L mutant had a dramatically decreased transacylation activity with allmal unit, whereas the self -acylation activity was similar to WT, implying a potential role of F203L on the interaction between AT4 FkbB and ACP. But Phe203 did not interfere with the interaction between Ser89 of AT4 FkbB and phosphopantetheine attachment site Ser33 of ACP4 FkbB according to the homology model of ACP4 FkbB bound to the homology model [KS4][AT4] FkbB and the cross-linking results (Supplementary Figure S4). How does F203L alter the substrate specificity? We speculated that F203L helped to allow ethmal group as a substrate to the active site Ser599, which was supported by the structural study of the M-AT DYN covalent complex (Liew et al., 2012). F203L recognized sufficient allmal unit so that the self -acylation activity with allmal was similar to that of WT, if adequate reaction time was provided in the self -acylation reaction. When holo-ACP10 FkbA , allmal-CoA, and F203L mutant were incubated together for 12 and 24 h, respectively, the activities of trans-acylation turned out similar to that of WT (data not shown), which partly confirmed our speculation. Therefore, further analysis of co-crystallizing F203L with allmal-or ethmal-CoA is still needed to discover the role of F203L in substrate specificity of AT4 FkbB .
Reverse mutations of these five residues in ethmal-specific ATs to the corresponding AAs of AT4 FkbB resulted in enhanced transfer of allmal onto ACP ( Table 2). These data indicated that Gln119, Leu185-Val186-Val187, and Phe203 residues alter the substrate specificity of ethmal-specific ATs. The experimental data open up the possibility to control the substrate specificity in ATs by substitution of the critical residues, which will lead to the incorporation of non-native extending units for the production of 'non-native' natural products. Despite the protein engineering in ethmal-specific ATs for allmal specific transfer, the role of other residues on allmal or ethmal-specific recognition of ATs is not yet thoroughly elucidated. For this reason, determination of the three-dimensional structures of AT4 FkbB with allmal and ethmal units would substantially improve our understanding of the mechanism of other residues in allmal specificity of AT for efficiently governing FK506 biosynthesis.