Discovery of a Unique Structural Motif in Lanthipeptide Synthetases for Substrate Binding and Interdomain Interactions

Abstract

Class III lanthipeptide synthetases catalyze the formation of lanthionine/methyllanthionine and labionin crosslinks. We present here the 2.40 Å resolution structure of the kinase domain of a class III lanthipeptide synthetase CurKC from the biosynthesis of curvopeptin. A unique structural subunit for leader binding, named leader recognition domain (LRD), was identified. The LRD of CurKC is responsible for the recognition of the leader peptide and for mediating interactions between the lyase and kinase domains. LRDs are highly conserved among the kinase domains of class III and class IV lanthipeptide synthetases. The discovery of LRDs provides insight into the substrate recognition and domain organization in multidomain lanthipeptide synthetases.

A unique structural motif for leader binding and interdomain interactions was identified in the crystal structure of the kinase domain of class III lanthipeptide synthetase CurKC responsible for the biosynthesis of curvopeptin. This motif consists of two N‐terminal antiparallel β‐sheets, a central α‐helix, and three C‐terminal antiparallel β‐sheets.

Introduction

Lanthipeptides belong to the superfamily of ribosomally synthesized and posttranslationally modified peptide (RiPP) natural products and typically contain the β‐thioether crosslinked amino acids lanthionine (Lan) and methyllanthionine (MeLan). ^[1] The Lan/MeLan crosslinks are typically generated by the enzymatic dehydration of serine/threonine residues that yields dehydroalanine (Dha)/dehydrobutyrine (Dhb), respectively, and the subsequent cyclization between cysteine and Dha/Dhb catalyzed by designated cyclases. To date, five classes of lanthipeptides have been reported based on the characteristics of their biosynthetic enzymes.[ ^1b , ² ] Class I lanthipeptides utilize a combination of aminoacyl‐tRNA‐dependent dehydratases (LanBs) and cyclases (LanCs), ^[3] whereas class II lanthipeptides are biosynthesized by ATP‐dependent bifunctional lanthipeptide synthetases (LanMs). ^[4] The biosynthesis of class III and class IV lanthipeptides relies on a group of three‐domain enzymes named LanKCs and LanLs, respectively, which contain an N‐terminal lyase, a central kinase, and a C‐terminal cyclase domain. ^[5] These three domains perform the sequential modification of the core peptides in the order of 1) NTP‐dependent Ser/Thr phosphorylation, 2) phosphate elimination that yields Dha/Dhb and 3) Michael addition between Cys and Dha/Dhb to generate the thioether linkages (Figure 1). LanCs, LanMs, and LanLs utilize a zinc ion in their respective cyclase domains that acts as a Lewis acid for activating the Cys thiol during Lan/MeLan formation. In contrast, the cyclase domains of LanKCs do not contain a conserved Zn binding motif, indicating that LanKC‐catalyzed macrocyclization is Zn‐independent. ^[6] In accordance with the distinct cyclase feature, LanKCs are the only lanthipeptide synthetases observed to catalyze the formation of tri‐amino acid crosslinks, named labionin, through the sequential formation of a C−S and a C−C bond. ^[7] To date, mechanistic investigation of LanBs, LanCs and LanMs is supported by crystallography studies,[ ^3a , ^3b , ^4a ] whereas crystal structures of LanKC/LanL enzymes or their individual domains remain unavailable. Herein, we report the first crystal structure of the kinase domain of CurKC, the class III lanthipeptide synthetase responsible for the biosynthesis of curvopeptin. ^[8] A unique leader recognition domain (LRD) was identified in the kinase domain of CurKC, which consists of two N‐terminal antiparallel β‐sheets, a central α‐helix and three C‐terminal antiparallel β‐sheets. This LRD motif is highly conserved among class III and class IV lanthipeptide synthetases and is responsible for both the leader peptide binding and the interaction between their lyase and kinase domains.

Figure 1.

Biosynthesis of the class III lanthipeptide curvopeptin congeners. The biosynthetic gene cluster contains genes encoding the following proteins: a class III lanthipeptide synthase (CurKC), a precursor peptide (CurA), two ABC transporters (CurT1/T2) and a regulatory protein (CurR). The leader peptide is possibly removed by bifunctional M1 family aminopeptidase(s) encoded in the genome of the producing strain Thermomonospora curvata DSM43183. ^[9] Incomplete removal of the leader peptide by the aminopeptidase(s) leads to the production of congeners with leader overhangs of various lengths.

Results and Discussion

CurKC displays a three‐domain organization:an N‐terminal lyase domain (residues 1–217, denoted as CurKC_lyase), a central kinase domain (residues 218–489, denoted as CurKC_kinase), and a C‐terminal cyclase domain (residues 594–866, denoted as CurKC_cyclase) (Figure S1–S3). Heterologous expression of full‐length CurKC, CurKC_lyase, CurKC_kinase, CurKC_{lyase‐kinase} (residues 1–489) and CurKC_{kinase‐cyclase} (residues 218–866) in E. coli BL21(DE3) all yielded soluble proteins (Figure S4). CurKC_cyclase, however, was not detected in soluble or insoluble fractions after expression in E. coli BL21(DE3), suggesting that CurKC_cyclase is incapable of folding into a stable protein independently. A C‐terminal truncated variant of the CurA peptide, named CurA_S, was employed as a substrate for enzymatic assays as it is significantly more accessible by chemical synthesis than the full length CurA (Figure 2). In the presence of ATP and Mg²⁺ ions, CurKC modified CurA_S by one‐ and two‐fold dehydration along with partial phosphorylation, as determined by matrix‐assisted laser desorption time‐of‐flight mass spectrometry (MALDI‐TOF MS) (Figure S5). The incomplete dehydration of CurA_S by CurKC was in agreement with previous studies showing that the disruption of the C‐terminal Lan formation impairs the modification of Ser/Thr residues at the N‐terminal core peptide. ^[5b] CurKC_kinase modified CurA_S by one‐fold phosphorylation, indicating that the kinase domain of CurKC can fold and function independently. Co‐incubation of CurKC_lyase and CurKC_kinase with CurA_S led to a similar result as the incubation of CurA_s with CurKC_{lyase‐kinase}: a single dehydration of the residue Ser2 (Figure S5 and S6). This result shows that CurKC_lyase retains the function as a lyase to catalyze the phosphate elimination of a phosphorylated serine (pSer) (Figure S5). Similar results were obtained previously when using the artificially separated kinase and lyase domains of the class IV lanthipeptide synthetase VenL with its peptide substrate VenA. ^[5d]

Figure 2.

A) Crystal structure of the CurKC_kinase–adenosine complex. The adenosine molecule is shown in a stick representation. B) Close‐up view of the adenosine binding site. Adenosine and interacting residues are shown as sticks, and the water molecule is shown as a gray sphere. Hydrogen bonds are shown as yellow dashed lines. Oxygen and nitrogen atoms are colored red and blue, respectively. C) MALDI‐TOF MS analysis of the in vitro enzymatic modification of CurA_S peptide by CurKC and CurKC variants. CurA_S (M): calcd: 2714.27 Da, obs.: 2714.34 Da; (M‐2H₂O): calcd: 2678.23 Da, obs.: 2678.67 Da; (M‐H₂O): calcd: 2696.25 Da, obs.: 2696.75 Da; (M+Na): calcd: 2737.26 Da, obs.: 2737.90 Da; (M‐2H₂O+HPO₃): calcd: 2758.21 Da, obs.: 2758.71 Da; (M‐H₂O+HPO₃): calcd: 2776.23 Da, obs.: 2776.82 Da; (M+HPO₃): calcd: 2794.25 Da, obs.: 2794.85 Da. D) Close‐up view of the ATP binding site in the CurKC_kinase–ATP docking structure. ATP and interacting residues are shown as sticks, and the magnesium ions are shown as cyan spheres.

To understand the mechanism of catalysis, we attempted to crystallize CurKC and its soluble variants. Only CurKC_kinase yielded crystals of sufficient quality for diffraction. Crystal structures of CurKC_kinase in its apo form and in complex with adenosine was obtained with resolution limits of 2.40 Å and 2.55 Å, respectively (Table S1, Figure S7A,B and Figure S8). ^[10] In both crystal structures, two identical CurKC_kinase protomers exist in one asymmetric unit with root‐mean‐square deviation (RMSD) values of 0.34 Å, although CurKC_kinase is a monomer in solution (Figure S7A–C). Structural comparison of apo‐CurKC_kinase and the CurKC_kinase–adenosine complex reveals that the binding of adenosine caused no significant alteration of the overall conformation of CurKC_kinase (Figure S7D).

In the CurKC_kinase–adenosine binary complex, CurKC_kinase adopts a typical two‐lobed structure as that observed for protein kinases such as PknB and PknG from Mycobacterium tuberculosis (Figure 2A and Figure S9A). ^[11] The N‐lobe (residues 218–309) comprises five β‐strands and an α‐helix (residues 267–282, denoted as helix C), and the C‐lobe (residues 310–472) is composed of six α‐helices and two β‐sheets. The adenosine resides in a deep cleft formed between the two lobes.

The binding between CurKC and ATP‐γ‐S is saturated at an ≈1 : 1.2 molar ratio, as determined by isothermal titration calorimetry (ITC) analysis (Figure S10), indicating that CurKC contains a single ATP binding site. In the CurKC_kinase–adenosine complex, the adenine motif is bound to the backbone carbonyl oxygen of residue E306 and the backbone amide nitrogen of residue I308 (Figure 2B). A water molecule that coordinates to the amide oxygen of I380 forms additional hydrogen bonds to adenine N6 and N7. Double mutation of E306 and I308 to Ala resulted in CurKC E306 A/I308 A with significantly impaired its phosphorylation and dehydration activity, supporting the roles of E306 and I308 for adenine binding (Figure 2C).

To provide a comprehensive view of ATP binding by CurKC_kinase, we replaced the adenosine with ATP and complemented two Mg²⁺ ions using the tLeaP module of the AMBER 16 software package. ^[12] The system was equilibrated by molecular dynamics (MD) simulations (Figure S11). In the docked CurKC_kinase‐ATP binary structure, the two Mg²⁺ ions were chelated by the D363/N368 and D381/E383 pairs and interacted with the β‐ and γ‐phosphates of ATP (Figures 2D). Ala substitutions of residues D363, N368, D381, and E383 abolished the activity of CurKC (Figure 2C), confirming their roles in ATP and Mg²⁺ binding and catalysis.

Despite their structural similarity, several structural differences are observed between CurKC_kinase and kinases PknB/PknG. The D381–F382–E383 motif in CurKC_kinase is highly conserved among LanKC and LanL enzymes and chelates Mg²⁺ through the D381/E383 pair in the docked CurKC_kinase–ATP model (Figure 2D and S12). Both D381 and E383 are indispensable for the kinase activity of CurKC (Figure 2C). In contrast, the homologous D293‐F294‐G295 motif in PknB lacks a second Mg²⁺ coordinating residue (Figure S9B and S12). The loop region connecting sheets β1 and β2 (spanning residues 230–236, named Loop1, Figure 2A) in CurKC_kinase is highly conserved among LanKC enzymes but shares no sequence similarity to its structural homologs in PknG/PknB, the P‐loop (Figure S12). ^[13]

CurKC and CurKC_kinase are capable of modifying CurA_S but not CurA_{S_CP}, indicating that both enzymes are leader‐dependent (Figure S5 and S16). Indeed, CurKC binds to the CurA leader peptide (CurA_LP) and the CurA_S peptide with similar K _D values of 5.5±0.7 μM and 3.5±0.3 μM, respectively, but showed no binding with CurA_{S_CP} (Figure 3A and Figure S17 and S18), indicating that CurA_LP is essential for enzymatic recognition. CurKC_{lyase‐kinase} and CurKC_kinase bind to CurA_LP with K _D values of (84±12) μM and (801±171) μM, respectively, whereas CurKC_lyase showed no measurable binding affinity to CurA_LP (Figure 3A). These results indicate that the kinase domain of CurKC is responsible for leader binding. The gradually decreased binding affinities of CurKC_{lyase‐kinase} and CurKC_kinase to CurA_LP compared with CurKC might be due to the structural destabilization caused by protein truncation.

Figure 3.

A) Binding affinity between the CurA_LP peptide and the CurKC variants as measured by MST. Data represent the mean±s.d. from three replicates. B) Structural representation of the LRD in CurKC. C) The interior view of CurKC_LRD shows a cleft between the β1‐strand and the helix C as a potential leader binding site. D) A docking model of the CurKC_kinase–ATP–CurA_(‐16)‐4 ternary complex. Important interacting residues are shown as sticks. The leader and core segments of peptide CurA_(‐16)‐4 are shown in red and yellow, respectively.

Leader peptide recognition in many RiPP biosynthetic enzymes is dependent on the so‐called RiPP recognition element (RRE) motif, which typically consists of three N‐terminal β‐strands and a C‐terminal winged helix‐turn‐helix motif containing a total of three alpha‐helices (Figure S19).[ ^3a , ¹⁴ ] These RRE motifs usually accommodate leader peptides in the cleft between the β3‐strand and α3‐helix through both hydrophobic interactions and hydrogen bonding. CurKC_kinase contains an N‐terminal region spanning residues 222–309 that we named leader recognition domain (LRD), consisting of two N‐terminal antiparallel β‐sheets, a central α‐helix, and three C‐terminal antiparallel β‐sheets (Figure 3B). The “3β‐1α‐2β” organization is distinct from typical RRE folds observed in other RiPP enzymes, and this structural unit is referred to as LRD_{3β‐1α‐2β} to highlight its structural features. To understand its function, we expressed and purified CurKC_LRD as a fusion protein conjugated to the C‐terminus of the B1 domain of streptococcal protein G (GB1). Subsequent proteolytic removal of the GB1 tag led to the production of CurKC_LRD as a soluble protein. MST assays showed that CurKC_LRD binds to CurA_LP with a K _D of (1.2±0.3) mM, which is comparable to CurKC_kinase. (Figure 3A). With the LRD motif truncated, CurKC_{kinase‐ΔLRD} completely lost its binding affinity to CurA_LP (Figure S20), indicating the role of the LRD motif for peptide recognition. To further corroborate the interaction between CurKC_LRD and CurA_LP, NMR titration experiments were carried out with ¹⁵N‐labeled GB1–CurKC_LRD. Chemical shift changes of the CurKC_LRD signals in the ¹H‐¹⁵N HSQC spectra were observed when titrated with various concentrations of CurA_LP, indicating a specific interaction between CurKC_LRD and CurA_LP (Figure S21 and S22). Due to the limitation of protein concentration and stability, we were not able to pinpoint specific residues of CurKC_LRD that interact with CurA_LP by three‐dimensional NMR experiments with ¹⁵N and ¹³C doubly labeled GB1–CurKC_LRD. Nevertheless, these results strongly supported the role of CurKC_LRD as the leader recognition site.

Due to the inaccessibility to a co‐crystal structure of CurKC with the CurA_LP peptide, we performed a docking study using CurKC_kinase and the CurA_(‐16)‐4 peptide to understand the substrate binding mode. As the dehydration of residue Ser2 in CurA_S by CurKC was most efficient, the pentapeptide CurA_(‐1)‐4 extracted from the MD representative snapshot of CurA was first docked into the active site of CurKC_kinase. Then, the peptide was gradually extended at the N‐terminus of CurA with residue length intervals of five amino acids to gain a starting conformation for the CurKC_kinase–ATP–CurA_(‐16)‐4 ternary complex (Figure 3D and Figure S23). MD simulation results showed that the binding model was most stable when the region of CurA_LP spanning residues (−16) to (−13) was bound to CurKC_LRD in the cleft between helix C and sheet β5 through hydrophobic interactions between residues V269/L272/L279/L282 in helix C of CurKC_LRD and residues L(−15) and L(−13) of CurA_LP (Figure S23 and S24). Residues V269/L272/L279/L282 of CurKC are highly conserved in LanKC enzymes (Figure S25). Residues L(−15) and L(−13) of CurA_LP are part of the proposed amphipathic “LXXLXXL” motif conserved in class III/IV leader peptides, which is important for enzymatic recognition (Figure S26).[ ⁶ , ^9b ] The “LXXLXXL” motifs in the leader peptides of several class III/V lanthipeptides display tendency to adopt an α‐helical conformation.²⁵ However, secondary‐structure prediction and circular dichroism (CD) spectroscopy analysis indicated that CurA_LP has no tendency for α‐helix formation (Figure S26), which agrees with the extended conformation of CurA_LP in the docking model. In addition to the hydrophobic interactions, hydrogen bonding was also observed in the docking model between residues M(−16)/D(−14)/Q(−12)/M(−10) of CurA_LP and residues K253/A263/R271/E275 of CurKC_LRD (Figure S27).

Mutagenesis studies were then performed to evaluate the key interacting residues between CurKC_LRD and CurA_LP revealed in the docking model. Single and double mutations of residues V269/L272/L279/L282 all resulted in CurKC mutants with significantly impaired leader binding affinity and catalytic activities (Figure 4A, 4 C and Figure S28). Similarly, mutations of K253/R271/E275 decreased the CurKC_LRD–CurA_LP binding affinity by 20–500‐fold and almost abolished their catalytic activities (Figure 4B and Figure S28). Residues E254 and A263 are suggested to interact with CurA_LP through backbone amides, and their Ala mutations had little impact on the binding affinity (Figure 4B and Figure S27). In addition, residue R276 is located on the surface of helix C opposite to the leader binding cleft and has no contact with CurA_LP in the docking model (Figure S29). Accordingly, no significant impact on the leader binding or enzymatic activity towards the peptide CurA_S was observed when R276 was mutated to Ala (Figure S30). Collectively, these data support the predicted CurKC_LRD–CurA_LP binding mode.

Figure 4.

A,B) Binding affinities between CurA_LP and CurKC variants as measured by MST. Assays were performed using fluorophore‐labeled CurKC variants (2.5 nM) and unlabeled CurA_LP. Data represent the mean±s.d. from three replicates. C) Modification of CurA_S by CurKC and CurKC variants as determined by MALDI‐TOF MS analysis. CurA_S (M): calcd: 2714.27 Da, obs.: 2714.34 Da; (M–2H₂O): calcd: 2678.23 Da, obs.: 2678.67 Da; (M–H₂O): calcd: 2696.25 Da, obs.: 2696.75 Da; (M–2H₂O+HPO₃): calcd: 2758.21 Da, obs.: 2758.71 Da; (M–H₂O+HPO₃): calcd: 2776.23 Da, obs.: 2776.82 Da; (M+HPO₃): calcd: 2794.25 Da, obs.: 2794.85 Da. D) Modification of monophosphorylated CurA_S by CurKC_lyase, CurKC_lyase‐LRD or CurKC_lyase‐LRD variants. Assay composition: 20 mM Tricine (pH 8.0), 1 mM EDTA,10 μM protein, and 100 μM monophosphorylated CurA_S peptide. Assay reactions were run at 25 °C for 15 h before MS analysis. CurA_S‐phos was obtained through the modification of CurA_S by CurKC_kinase. CurKC_kinase was inactivated by incubation at 95 °C, and CurA_s‐phos was purified by HPLC before assays. CurA_S (M): calcd: 2714.27 Da, obs.: 2714.10 Da; (M‐H₂O): calcd: 2696.25 Da, obs.: 2696.98 Da; (M+HPO₃): calcd: 2794.25 Da, obs.: 2794.98 Da.

The LRD motif (residues 222–309) is located C‐terminally to the lyase domain of CurKC (residues 1–217) in the amino acid sequence of CurKC, raising the possibility that the LRD motif might also facilitate the function of CurKC_lyase during phosphate elimination. With monophosphorylated CurA_S (CurA_S‐phos) as a substrate, CurKC_lyase‐LRD (residues 1–309) efficiently eliminated the phosphate group by yielding a one‐fold dehydrated product (Figure 4D). In contrast, CurKC_lyase was unable to modify the CurA_S‐phos peptide. Furthermore, the lyase activity of CurKC_lyase‐LRD_V269 A_L272 A and CurKC_lyase‐LRD_L279 A_282 A were significantly impeded (Figure 4D), suggesting that the LRD motif is important for the enzymatic phosphate elimination.

Class III leader peptides often contain a spacer segment as a ruler to determine that Ser/Thr residues of proper positions C‐terminally to the “LXXLXXL” recognition motif can be modified. ^[15] In the docking model of the CurKC_kinase–ATP–CurA_(‐16)‐4 ternary complex, a potential spacer segment spanning Gly(−11) to Asp(−2) of ≈35 Å in length showed no direct interactions with CurKC_kinase (Figure 3D). This spacer is sufficiently long to deliver the core peptide residue Ser2 to the active site of the lyase domain (Figure S31). As a single leader anchoring site, the location of CurKC_LRD at the interface of the kinase and lyase domains appears to be beneficial to mediate the translocation of the CurA_CP between the two active sites for sequential Ser phosphorylation and phosphate elimination.

Although CurKC_lyase was unable to recognize and modify the CurA_S‐phos peptide, co‐incubation of CurKC_lyase and CurKC_kinase fully converted CurA_S‐phos to one‐fold dehydrated CurA_S with similar efficiency as CurKC_{lyase‐kinase} (Figure S5), suggesting that CurKC_lyase and CurKC_kinase might function cooperatively as a complex in the same manner as their covalently linked counterpart CurKC_{lyase‐kinase}. Indeed, CurKC_lyase binds to CurKC_kinase with a K _D of 170±28 μM, supporting that the CurKC_lyase and CurKC_kinase could form a complex in solution (Figure 5A).

Figure 5.

A) MST analysis of the binding affinities between CurKC_lyase and CurKC_kinase variants, as well as CurKC_kinas–CurKC_lyase N103A. Assays were performed using fluorophore‐labeled CurKC_lyase or CurKC_lyase N103A (2.5 nM) and unlabeled CurKC_kinase variants. Data represent the mean±s.d. from three replicates. B) MST analysis of the binding affinities between AciKC_lyase and AciKC_kinase or CurKC_kinase variants, as well as AciKC_kinase–AciKC_{lyase_}S114A. Data represent the mean±s.d. from three replicates. The binding assays were carried out using fluorophore‐labeled AciKC_lyase or AciKC_{lyase_}S114A (2.5 nM) and increasing amounts of unlabeled proteins. C) Interdomain binding affinities between the kinase and lyase domains of CurKC, AciKC, and VenL. D) In vitro modification of CurA_S by VenL_kinase and VenL_kinase/CurKC_lyase. CurA_S (M): calcd: 2714.27 Da, obs.: 2714.93 Da; (M‐H₂O): calcd: 2696.25 Da, obs.: 2696.96 Da; (M+Na): calcd: 2737.26 Da, obs.: 2737.13 Da; (M+HPO₃): calcd: 2794.25 Da, obs.: 2794.98 Da. Assay composition: 20 mM Tricine (pH 8.0), 1 mM DTT, 10 mM MgCl₂, 1 mM ATP, 15 μM protein, and 100 μM CurA_S. The reactions were quenched by the addition of MeOH after incubation at 28 °C for 15 h.

To understand the formation of the CurKC_lyase–CurKC_kinase complex, we performed structural prediction on CurKC_lyase, CurKC_kinase and CurKC_{lyase‐kinase} by AlphaFold2 (Figure S32). ^[16] The predicted structure of CurKC_kinase is highly similar to our crystal structure of CurKC_kinase with a RMSD of only 0.91 Å. In the predicted structure of CurKC_{lyase‐kinase}, the Loop1 and Loop2 of the LRD motif appear to be the major contributors to the CurKC_lyase–CurKC_kinase association (Figure S33A). Indeed, CurKC_lyase binds to CurKC_LRD with a K _D of 139±33 μM, whereas CurKC_{kinase_ΔLRD} showed no binding affinity to CurKC_lyase (Figure 5A and Figure S34). Mutations of residues S233/R234 in the LRD and residue N103 in CurKC_lyase, which are located at the domain interface in the predicted structure of CurKC_{lyase‐kinase}, impaired the binding between CurKC_lyase and CurKC_kinase (Figure 5A), supporting the role of LRD in the domain interactions.

The characteristic LRD_{3β‐1α‐2β} motifs are found in all characterized LanKCs with a similar organization, as indicated by sequence analysis and structural prediction using AlphaFold2 (Figure S12, S25 and S35). Important hydrophobic residues for leader binding are well conserved in these LRD_{3β‐1α‐2β} motifs of LanKC enzymes, suggesting their universal function (Figure S25). To examine the role of the LRD_{3β‐1α‐2β} motifs in LanKC enzymes for lyase–kinase interdomain interaction, we selected AciKC from the biosynthesis of the class III lanthipeptide catenulipeptin as a model enzyme and expressed its lyase (residues 1–219) and kinase (220–495) domains as separate proteins, named AciKC_lyase and AciKC_kinase, respectively. AciKC_lyase binds to AciKC_kinase with a K _D of 43.9±10.9 μM (Figure 5B). Similar to CurKC, structural prediction indicates that Loop1 and Loop2 of AciKC_LRD might be the major contributors to the AciKC_lyase–AciKC_kinase interaction (Figure S36A). Mutations of residues at the AciKC_LRD–AciKC_lyase interface, including S114, N236 and R258, resulted in decreased binding affinity (Figure 5B), supporting the role of AciKC_LRD for interdomain interactions. Interestingly, AciKC_lyase binds to CurKC_kinase with a K _D of (209±71) μM, and CurKC_lyase binds to AciKC_kinase with a K _D of (14.8±8.3) μM (Figure 5C and Figure S37), indicating that the association of the lyase and kinase domains of AciKC and CurKC is interchangeable in vitro. Again, mutations in the Loop 1/2 of CurKC_LRD impaired the AciKC_lyase–CurKC_kinase binding (Figure S34), suggesting that the CurKC_LRD participates in the interdomain association.

The lyase and kinase domains of class III and class IV lanthipeptide synthetases share sequence similarity. ^[6] LRD_{3β‐1α‐2β} motifs are also found at the interface of the lyase and kinase domains in the predicted structures of LanL enzymes (Figure S12, S25, S36 and S37B). To explore the role of the LRD motifs in LanL enzymes, VenL was selected as a model enzyme. ^[5d] MST assays showed that VenL_kinase (residues 213–513) binds to VenL_lyase (residues 1–212) with a K _D value of 32.3±2.9 μM (Figure 5C). In addition, VenL_kinase binds to CurKC_lyase with a K _D of 24.3±3.5 μM, and VenL_lyase binds to CurKC_kinase with a K _D of 177±39 μM (Figure 5A‐5 C and Figure S37). Again, mutations in the Loop 1/2 of CurKC_LRD resulted in an approximately two‐fold decrease in the CurKC_kinase–VenL_lyase binding (Figure S37A). Collectively, these data indicate that the LRDs in class III and IV lanthipeptide synthetases play a role in the cross‐class binding of class III/IV lyase and kinase domains.

Finally, we examined whether the catalytic function of the heterologous complex formed by kinase and lyase domains of distinct enzyme origins can be retained. Using CurA_S as a model peptide substrate, VenL_kinase displayed good substrate tolerance and converted CurA_S to a monophosphorylated product (Figure 5D). Co‐incubation of CurKC_lyase, VenL_kinase, and CurA_S led to the formation of one‐fold dehydrated CurA_S (Figure 5D). As CurKC_lyase is not able to process the phosphorylated peptide substrate independently, these results indicated that CurKC_lyase binds to VenL_kinase and therefore gained access to phosphorylated CurA_S for catalysis. Thus, these results demonstrate that the complex of CurKC_lyase and VenL_kinase retained the catalytic function of each domain when using the CurA_S peptide as substrate. The data also implies that VenL_LRD is both able to recognize CurAs as a substrate and to shuttle the CurA_s core peptide between the VenL_kinase and CurKC_lyase active sites.

Conclusion

Class III and class IV lanthipeptide synthetases are highly modular three‐domain enzymes. Although the heterologous expression of their separate cyclase domains has not yet been successful, previous studies have shown that their kinase and lyase domains can fold and function as individual proteins.[ ^5c , ^5d , ^5e , ^5f ] As the first crystal structure of the kinase domain of a LanKC enzyme, CurKC_kinase displays overall structural similarity to the protein kinases PknB/PknG. However, CurKC_kinase shows almost no structural similarity with the class II synthetase CylM despite their common function of NTP‐dependent phosphorylation of Ser/Thr residues in peptide substrates (Figure S38), ^[4a] suggesting that class II and class III/IV synthetases are of distinct evolutionary origins.

Leader peptide binding is crucial for all three catalysis steps of LanKC enzymes. ^[5c] For multidomain enzymes, the leader peptide of a substrate might anchor at a single binding site, or translocate among different binding sites to meet the requirement of each catalytic domain. In the case of the class I dehydratase NisB, the leader peptide of its substrate NisA binds to a single LP binding site for both glutamylation and elimination. ^[17] The crystal structure of CurKC_kinase reveals an unprecedented LRD_{3β‐1α‐2β} motif that plays essential roles in leader binding. Distinct from the previously discovered RRE motifs found in RiPP biosynthetic enzymes, the LRD_{3β‐1α‐2β} motifs form a hydrophobic cleft between helix C and sheet β5 for leader binding. Our docking model indicated that this cleft likely interacts with the Met/Leu residues in the ML D L QG M motif in CurA_LP mainly through hydrophobic interactions. A conserved (M/L)XX(L/V)XXL motif in class III/IV leader peptides is proposed to be essential for enzymatic recognition, and alterations of the hydrophobic M/V/L residues usually impedes substrate binding and modification. ^[9b] Our structural analysis and mutagenesis studies were in good agreement with previous biochemical results of other LanKC enzymes. It is noteworthy that in the class IV synthetase SgbL, a≈100 amino acid long motif (SgbL_208‐318), which is equivalent to the LRD_{3β‐1α‐2β} motif of CurKC, was already identified as peptide binding site by hydrogen‐deuterium exchange mass spectrometry (HDX‐MS) analysis and binding assays. ^[5f] Similar to CurKC_LRD, the SgbL_208‐318 motif is predicted to adopt a structure comprising of four antiparallel β‐sheets and a central α‐helix. The HDX‐MS experiment strongly indicated the α‐helix and sheet β4 as the major LP‐binding units, which is highly similar to our model of CurA_LP–CurKC_LRD binding. Hence, it makes sense that we also observed how enzymatic activity was retained when using domains from enzymes originating from different classes in trans. In contrast to the previous studies that were based on HDX‐MS data and a mere prediction of the SgbL kinase domain structure, our study provides the first real three‐dimensional structure of a class III kinase domain including its leader peptide binding site. This structure allowed us in turn to not only show the general function of this domain for leader binding, but also what residues in the protein form crucial contacts with the leader peptide. Furthermore, our study for the first time demonstrates the role of the LRD_{3β‐1α‐2β} motif for mediating the interactions between the catalytic centers for peptide phosphorylation and phosphate elimination, which is key for understanding how the substrate can be shuttled back and forth between active sites in a sequential manner. The binding of ATP has little impact on the leader binding and lyase–kinase association (Figure S39). This greatly expands our understanding of how the biosynthesis of both class III and IV lanthipeptides are accomplished on a molecular level.

In conclusion, our study shows that LRD_{3β‐1α‐2β} motifs are conserved among class III and class IV lanthipeptide synthetases and mediate the association between the lyase and kinase domains. The binding modes between LRD_{3β‐1α‐2β} motifs and lyase domains are highly conserved across LanKC and LanL enzymes and facilitate the binding of domains taken from distinct classes. These findings highlight the modularity and the close evolutionary relationship between the lyase–kinase domains of LanKC and LanL enzymes, despite the lack of homology among their cyclase domains.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

S. Huang, Y. Wang, C. Cai, X. Xiao, S. Liu, Y. Ma, X. Xie, Y. Liang, H. Chen, J. Zhu, J. D. Hegemann, H. Yao, W. Wei, H. Wang, Angew. Chem. Int. Ed. 2022, 61, e202211382; Angew. Chem. 2022, 134, e202211382.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.