The evolutionary origin of naturally occurring intermolecular Diels-Alderases from Morus alba

Biosynthetic enzymes evolutionarily gain novel functions, thereby expanding the structural diversity of natural products to the benefit of host organisms. Diels-Alderases (DAs), functionally unique enzymes catalysing [4 + 2] cycloaddition reactions, have received considerable research interest. However, their evolutionary mechanisms remain obscure. Here, we investigate the evolutionary origins of the intermolecular DAs in the biosynthesis of Moraceae plant-derived Diels-Alder-type secondary metabolites. Our findings suggest that these DAs have evolved from an ancestor functioning as a flavin adenine dinucleotide (FAD)-dependent oxidocyclase (OC), which catalyses the oxidative cyclisation reactions of isoprenoid-substituted phenolic compounds. Through crystal structure determination, computational calculations, and site-directed mutagenesis experiments, we identified several critical substitutions, including S348L, A357L, D389E and H418R that alter the substrate-binding mode and enable the OCs to gain intermolecular DA activity during evolution. This work provides mechanistic insights into the evolutionary rationale of DAs and paves the way for mining and engineering new DAs from other protein families.


REVIEWER COMMENTS
Reviewer #1 (Remarks to the Author): The manuscript (Manuscript ID: NCOMMS-23-56002-T) by Xiaoguang Lei, et al. entitled "The evolutionary origin of naturally occurring intermolecular Diels-Alderases from Morus alba" explores evolutionary origins of the intermolecular Diels-Alderases in the biosynthesis of plant-derived Diels-Alder type secondary metabolites.The authors claim the findings suggest that these Diels-Alderases have evolved from an ancestor functioning as a flavin adenine dinucleotide dependent oxidocyclase, which catalyses the oxidative cyclisation reactions of isoprenoid substituted phenolic compounds.Through crystal structure determination, computational calculations, and site-directed mutagenesis experiments, they identified several critical mutations, including S348L, A357L, D389E, and H418R, that alter the substrate-binding mode and enable the flavin adenine dinucleotide dependent oxidocyclases to gain intermolecular Diels-Alderase activity during evolution.The results provide mechanistic insights into the evolutionary rationale of Diels-Alderases and pave the way for mining and engineering new Diels-Alderases from other protein families.However, MaDA could not catalyze stereoselective Diels-Alder reaction during the formation of chalcomoracin (5) as an endo form and 7 as an exo form catalyzed by wild type MaDA in Figure 4.The authors should show the productions of both compounds by using a knock-out strain of the gene, MaDA, in Morus alba.Regarding the nonstereoselective Diels-Alder reaction, the authors should clearly show the transition state energies for both transformations instead of Figure 5c, which is hard to understand what you show.They have to show the computational results such as Figure 4b in reference number 17 in this manuscript or Figure 4b in Sato, M., et al. Nature Catalysis, 2021, 4, 223-232.The section of antifungal and antibacterial activity of Diels-Alder-type adducts should be deleted in this manuscript because it does not matter with Diels-Alderase.The authors should focus on demonstrating the intermolecular Diels-Alder reaction by MaDA.
Reviewer #2 (Remarks to the Author): This study seeks to learn how diels-alderases (DAs) in Moraceae plants evolved their mechanisms.Evolutionary mechanisms showed these enzymes emerged from an ancestor with oxidocyclase activity.The authors use phylogenetics, enzymatic assays, structural studies, MD simulations and mutagenesis to study ancestral and modern variants to identify amino acids that drove the acquisition of diels-Alderase activity and substrate binding.These studies are rigorous and provide mechanistic insight into how this family of enzymes evolved, while also giving insight into how proteins can be rationally engineered to achieve new catalytic functions.
A few comments are provided below to help to improve the manuscript.After these comments are addressed, acceptance is recommended.This is a semantic matter: authors should consider replacing the word mutations/mutagenesis in connection to evolutionary changes with 'substitution'.
Authors state that the ionic hydrogen bond involving E389 and R418 promotes the D-A reaction by lowering the LUMO energy.It is not clear how this determination is made, as no energy calculation is reported in the manuscript.The authors should provide an explanation of this result or a reference explaining how this conclusion was arrived at.
Authors have used MD simulations to determine that substrates or transitions do or do not bind to enzymes.However, the only data they show to support this claim is distances, which is not enough.Distances can be used to support a more sustained interaction between components of the protein-TS complex, but they cannot be used to descirbe tight binding (or even binding),unless binding affinities have been calculated, which can be done from MD simulations.Authors should either perform binding calculations or remove the quantitative language about binding, replacing it with something like 'transition state maintains an interaction with the active site over the course of our simulation' .This comment refers to data shown in Figures 5C and 5D.
Because models were prepared by Alphafold, authors should consider providing the confidence values (per-residue) of their structures.This will provide a sense of how reliable the models are.
Critical details of the MD simulation are missing.Was solvent added, and how much?Were ions added?Were minimization and equilibration performed?What type of ensemble, thermostat etc were used.
Reviewer #3 (Remarks to the Author): Lei et al. report an interesting study to trace the evolutionary origin of the intermolecular Diels-Alderases and their functionally coupled dehydrogenases in Morus alba (named MaDAs and MaDSs, respectively).Based on a thorough phylogenetic analysis with an association of enzymatic activity assays, the authors raised the hypotheses that both MaDAs and MaDSs share with the oxidative cyclases in Moraceae plants (named MaOCs) an ancestor enzyme possessing a berberine bridge enzyme (BBE)-like fold and that OC activity appears to be more ancestral and can be developed to individual dedicated OCs, Das and DSs through gene duplication and mutagenesis.To validate these hypotheses, they constructed the ancestral Diesl-Alderase (ancDA) and the last common ancestor of DAs and DSs (ancDADS) for comparative analysis.Consequently, the ancDADS, which basically is an OC and has no DA and DS activities, was developed to an enzyme with DA activity via a minimum of the mutations of four key residues.Eventually, computational studies rationalized the catalytic processes for ancDA, ancDADS and ancDADS-derived enzymes with DA activity, leading to a model for the evolution of OC to DA.
Overall, this is a very interesting and comprehensive study that showcases how enzymes evolve for functional differentiation to process structurally related substrates in different ways by combining genetic, biochemical, structural biology and computational approaches together.The manuscript is well-written.The experiments are clearly and comprehensibly presented, and figures are wellarranged.Understanding the evolutionary origin and mechanism of MaDAs will ultimately facilitate design and develop new enzyme tools for synthetic biology utilization.This reviewer is supportive of publication (although minor revision appears to be necessary) because the results should be of great interest to the interdisciplinary readership of the journal.
Enzymatic activity assays of selected MaOCs, MaDAs and MaDSs revealed that a few are functionally promiscuous, i.e., two DAs and two DSs have OC activity for moracin D formation.Did the authors determine whether DAs have DS activity or DSs have DA activity?If not, could you provide an explanation, particularly based on the evolutionary model shown Fig. 2? Page 7. The single mutation of I417L in ancDA resulted in the complete loss of DA activity.This results appears to be a bit contradictory to the weak DA activity of ancDADS-mut4 in which the I417L mutation is not involved.
Extended Data Fig. 7, the lane for the incubation of ancDADS-mut5 with substrates 3 and 4. The product profile is complicated and have some peaks in addition to those indicated.What do these peaks represent?It seems additional activities to OC and DA activities were observed?Chemically, the evolution of an OC from a DS is possible, given the notion that cyclization activity can be developed for 1,4-addition by following dehydrogenation immediately.Thus, the mechanism by which MaOCs function is interesting: during the catalysis, does cyclization occur prior to dehydrogenation?
Others: 1. Page 2, abstract, line 35: "plant-derived Diels-Alder-type secondary metabolites".It could be better to specialize to "moraceae plant-derived".2. Page 2, abstract, line 40: please remove the comma in "D389E, and H418R" and in "and H418R, that alter".3. Page 3, line 83: please remove the first "typically" in "Unlike most BBE-like enzymes typically, which typically catalyse".4. Page 4, line 100: "FAD-dependent DAs in Moraceae plants evolve from the BBE-like enzymes" could be better.5. Page 6, line 177: please leave a blank place between "MaDA1(Extendend Data Fig. 5a".6. Page 7, line 207: please leave a blank place between "mongolicin F(exo-7)".7. Page 10, Line 318: the sentence "" needs to be rephrased.It could be better "This evolutionary strategy represents a rare case for metabolic pathway development, by evolving a same ancestor to two new enzymes catalysing two consecutive reactions".
Reviewer #4 (Remarks to the Author): Although many enzyme-catalyzed Diels-Alder reactions have been identified, the evolutionary mechanisms of natural Diels-Alder enzymes remain unclear.Lei and co-authors use phylogenetic analysis, ancestral sequence reconstruction, and other methods to elucidate the evolutionary origin of Diels-Alder enzymes in Moraceae plants.They found that these Diels-Alder enzymes originated from gene duplication followed by neofunctionalization of BBE-like enzymes with oxidocyclase activity.The results are interesting and novel to the natural product biosynthesis research.

Questions and suggestions:
1.This work involved docking and Molecular dynamics simulation analyses to investigate the substrate binding pattern change in the evolutionary process from OCs to DAs.It might be better to perform quasi-classical direct-dynamics simulations as Ken Houk and co-authors did (J Am Chem Soc, 2016, 138, 3631-3634).2. In line 70 of page 3, the authors wrote that "the monofunctional DAs PyrE3, SdnG and SpnF".However, the work of Hung-wen Liu and Ken Houk showed that SpnF is a multifunctional enzyme with [4+2], [6+4], and cope rearrangement activities.(J Am Chem Soc, 2016, 138, 3631-3634;PNAS, 2017, 114, 10408-10413.)3.In line 360 of page 11, the authors wrote, "Protein sequences with fewer than 450 or more than 650 amino acids were removed."Would this prevent the identification of new Diels-Alder enzymes? 4. In line 430 of page, "0. 0.97915 Å" should be replaced by "0.97915 Å".We would like to thank this reviewer for his/her thorough review of the manuscript and for providing valuable feedback.
1.However, MaDA could not catalyze stereoselective Diels-Alder reaction during chalcomoracin (5) formation as an endo form and 7 as an exo form catalyzed by wildtype MaDA in Figure 4.
Response #1-1: As Like the exo-selective MaDAs (MaDA7 and MaDA8), ancDA was found to harbor both R294 and R443 (Fig. R1b).We believe that this feature of ancDA supports its capacity to catalyze the formation of exo-configurated D-A products as we observed in Figure 4. Unlike ancDA, the evolutionally more distant ancDADS only contains the conserved R294 in exo-seletive D-Aases but not the conserved R443 across both endo-and exo-selective D-Aases.When we introduce the conserved R443 as well as the other three residue into ancDADS, the generated ancDADS-mut4 was proved to catalyse both endo-and exo-selective Diels-Alder reactions without obvious bias (Figure 4).These results suggested that the evolution of Diels-Alderases (DAases) underwent a process of functional specialization.This transition involved a shift from the ancestral state of enzymes with a broad spectrum of catalytic activities, encompassing both endo-type and exo-type products, to enzymes exhibiting a tendency toward either endo or exo products.This aligns with findings reported by some researchers (Fig. R2 reviewer's insightful suggestions.We have deleted this section and revised it accordingly. Reviewer #2 (Remarks to the Author): Comments: This study seeks to learn how Diels-Alderases (DAs) in Moraceae plants evolved their mechanisms.Evolutionary mechanisms showed these enzymes emerged from an ancestor with oxidocyclase activity.The authors use phylogenetics, enzymatic assays, structural studies, MD simulations and mutagenesis to study ancestral and modern variants to identify amino acids that drove the acquisition of Diels-Alderase activity and substrate binding.These studies are rigorous and provide mechanistic insight into how this family of enzymes evolved, while also giving insight into how proteins can be rationally engineered to achieve new catalytic functions.A few comments are provided below to help to improve the manuscript.After these comments are addressed, acceptance is recommended.
We thank this reviewer for dedicating time to assess our manuscript and providing positive feedback on our work.We have successfully addressed all the suggestions and comments raised by this reviewer.
1.This is a semantic matter: authors should consider replacing the word mutations/mutagenesis in connection to evolutionary changes with substitution.

Response #2-1:
We sincerely appreciate the excellent suggestion from this reviewer and have revised it accordingly.
2. Authors state that the ionic hydrogen bond involving E389 and R418 promotes the D-A reaction by lowering the LUMO energy.It is not clear how this determination is made, as no energy calculation is reported in the manuscript.The authors should provide an explanation of this result or a reference explaining how this conclusion was arrived at.

Response #2-2:
Many thanks to this reviewer for highlighting this aspect.In our prior investigations (Gao, L. et al. Nat. Catal. 2021, 4, 1059), utilizing DFT theozyme calculations, we have discerned that R443 plays a pivotal role in activating the dienophile through an ionic          1 .This interaction significantly diminishes the gap between the highest occupied molecular orbital (HOMO) of the diene and the lowest unoccupied molecular orbital (LUMO) of     Figure R5).
The binding model analysis of MaDATS(endo) reveals a hydrogen bond triad involving Y412, E414, and R443, with R443 establishing contact with the carbonyl oxygen of the dienophile (Figure R6).Site-directed mutation studies suggest that, among these residues, Y412 holds relatively less significance than E414 and R443 (Figure R6).Consequently, E414 and R443 emerge as crucial contributors to facilitating the Diels-Alder reaction by reducing the LUMO energy in MaDA. to describe tight binding (or even binding), unless binding affinities have been calculated, which can be done from MD simulations.Authors should either perform binding calculations or remove the quantitative language about binding, replacing it with something like transition state maintains an interaction with the active site over the course of our simulation.This comment refers to data shown in Figures 5C and

5D
Response #2-3: We greatly appreciate the insightful comments provided by the reviewer.As the reviewer suggested, we have removed the quantitative language about binding and rephrased the related sentences as follows:  our MD simulations showed that the transition state for the intermolecular D-A reaction between diene 3 and morachalcone A (4) could not bind to the active site of ancDADS (Fig. 5c)  our MD simulations showed that the transition state for the intermolecular D-A reaction between diene 3 and morachalcone A (4) could not maintain a stable interaction with the active site of ancDADS during the simulations (Fig. 5c).  We found that the transition state could also tightly bind to the active site of ancDADS-mut4 like ancDA and MaDA1 (Fig. 5c).  In contrast, the transition state maintains an interaction with the active site of ancDADS-mut4 as well as ancDA and MaDA1 over the course of our simulation (Fig. 5c). 4. Because models were prepared by Alphafold, authors should consider providing the confidence values (per-residue) of their structures.This will provide a sense of how reliable the models are.
Response #2-4: We appreciate the suggestions provided by the reviewer.Following your advice, we have generated structural diagrams of proteins predicted by AlphaFold, employing colorization based on the numerical values of per-residue confidence score (pLDDT) (Figure R7).The results reveal that the pLDDT values for over 90% of the residues 14 predicted structure (light purple) of MaDA and the crystal structure of MaDA (cyan, PDB ID 6JQH).
5. Critical details of the MD simulation are missing.Was solvent added, and how much?
Were ions added?Were minimization and equilibration performed?What type of ensemble, thermostat etc were used.
Response #2-5: We thank the reviewer for raising important queries regarding the critical details of our Molecular Dynamics (MD) simulations.In response, we have provided the required details in the method section as follows: For the molecular dynamics (MD) simulations, each docking complex featuring the transition state (TS) with ancDA, ancDADS, and ancDADS-mut4 served as the foundational structure.Protonation states were determined using the Protein Preparation Wizard in Maestro.We employed the OPLS4 force field parameter set for both the protein and ligand, accompanied by the TIP3P model for water.
Each protein complex was immersed in a pre-equilibrated orthorhombic box with a 10 Å buffer of TIP3P water molecules, and the volume was minimized.Explicit counterions (Na or Cl ) were added to neutralize the systems, and 0.15 M NaCl was introduced to simulate ionic effects in a realistic environment.
Molecular dynamics simulations were conducted using the Desmond software package, employing a standard NPT (isothermal-isobaric ensemble) relaxation protocol.The simulation protocol unfolded in multiple stages to explore the system's behavior under diverse conditions.All time units are reported in picoseconds (ps); energy is expressed in kilocalories per mole (kcal/mol).
Utilizing default parameters, the simulation was performed with a single CPU, employing a cutoff radius of 9.0 Å, a time step of [0.002 0.002 0.006] ps, and a total simulation time of 500,000 ps.For temperature control, the Berendsen algorithm was employed with a target temperature of 343.0 K and a coupling time of 1.0 ps.Pressure control was achieved using the NPT ensemble with a pressure setting of 1.01325 bar.
The pressure coupling time was 2.0 ps, and the temperature coupling time was 1.0 ps.
The restraint protocol employed a force constant of 50 kcal/mol/Å² on heavy atoms.To define the transition state geometry, a harmonic constraint with a force constant of 500 kcal mol ¹ Å ² was applied.Torsional constraints were applied to the planes of diene and dienophile carbons with a force constant of 500 kcal mol ¹ Å ².These restraint parameters were selected to ensure the preservation of the desired structural features during the simulation while allowing for the exploration of dynamic behavior.
We trust that these clarifications provide a more detailed understanding of our MD simulation methodology.
from the literature and our experiments suggests that the first one is probably the catalytic mechanism of MaOCs.They are as follows: 1) There have been some studies on the catalytic mechanism of BBE-like enzymes, among which THCAS (Tetrahydrocannabinolic acid synthase) catalyzes a very similar reaction as MaOCs (Figure R13c).Minor issues: 5. Page 2, abstract, line 35: "plant-derived Diels-Alder-type secondary metabolites".
It could be better to specialize to "moraceae plant-derived".
Response #3-7: Thank you for your correction.We have eliminated the redundant "typically" and revised the phrase to "Unlike most BBE-like enzymes, which typically catalyze." 8. Page 4, line 100: "FAD-dependent DAs in Moraceae plants evolve from the BBElike enzymes" could be better.
Response #3-10: Understood.We have added a space, and it now reads "mongolicin F (exo-7)." 11. Page 10, Line 318: the sentence "" needs to be rephrased.It could be better " This evolutionary strategy represents a rare case for metabolic pathway development, by evolving a same ancestor to two new enzymes catalysing two consecutive reactions".
Response #3-11: Thank you for your valuable input.We have rephrased the sentence  This evolutionary strategy represents a rare case for metabolic pathway development, by evolving a same ancestor to two new enzymes catalysing two consecutive reactions.
Since all the newly characterized BBE-like enzymes including MaDAs from Morus alba have a protein sequence length ranging from 450 to 650 amino acids, this empirical feature, rooted in our experience, guided the genome mining of novel Diels-Alderases in other species and the construction of the updated phylogenetic tree for the DAs and oxidases (Extended Data Fig. 2).The chosen length criteria resulted in a phylogenetic tree with high Bootstrap values, minimizing inaccuracies that may arise from the extended length span of the initial phylogenetic tree.In this case, as the reviewer pointed out, novel Diels-Alderases from other species might be missed if their complete protein sequences are unavailable in the database.
To present this filtering process more comprehensively and clearly, we have deleted   rotein sequences with fewer than 450 or more than 650 amino acids were removed         description in the phylogenetic analysis section as follows: ' For candidate genes from other species, the protein sequences with fewer than 450 or more than 650 amino acids were removed to enhance the accuracy of the phylogenetic tree.'

1
Point-by-Point ResponsesContents Response to Re   Page 28 Response to Re  .Page 915 Response to Reviewer #3 .. Page 1624 Response to Reviewer #4 .. Page 2529 Reviewer #1 (Remarks to the Author): Comments: The manuscript (Manuscript ID: NCOMMS-23-56002-T) by Xiaoguang Lei           -      evolutionary origins of the intermolecular Diels-Alderases in the biosynthesis of plant-derived Diels-Alder type secondary metabolites.The authors claim the findings suggest that these Diels-Alderases have evolved from an ancestor functioning as a flavin adenine dinucleotide dependent oxidocyclase, which catalyses the oxidative cyclisation reactions of isoprenoid substituted phenolic compounds.Through crystal structure determination, computational calculations, and site-directed mutagenesis experiments, they identified several critical mutations, including S348L, A357L, D389E, and H418R, that alter the substrate-binding mode and enable the flavin adenine dinucleotide-dependent oxidocyclases to gain intermolecular Diels-Alderase activity during evolution.The results provide mechanistic insights into the evolutionary rationale of Diels-Alderases and pave the way for mining and engineering new Diels-Alderases from other protein families.
Figure R1.Statistical analysis of R443 and R294 in MaDAs, MaDSs and MaOCs.a, Statistical analysis was conducted on R443 within MaDAs, MaDSs and MaOCs.The figure illustrates a phylogenetic tree encompassing MaDAs (Diels-Alderases), MaDSs (diene-synthases), and MaOCs (oxidocyclases), wherein residues corresponding to the catalytic residue R443 are annotated within parentheses.The corresponding residue at positions aligning with R443 in each enzyme was indicated by the single letter codes of amino acids.b, The statistical analysis of R294 within MaDAs, MaDSs and MaOCs.

Figure R5 .
Figure R5.Orbital analysis.a, Orbital analysis for the LUMOs of the dienophile, b, Orbital analysis for the HOMO of the diene, c, Orbital analysis for the LUMO of the dienophile bound to R443 (MaDA), d, Orbital analysis for the dienophile bound to R294(MaDA-3).Our previous paper cited this data (Gao, L. et al.Nat.Catal.2021, 4, 1059).

Figure R13 ,
Figure R13, Proposed mechanisms of MaOCs and their homologous protein THCAS.a, The first proposed mechanism of MaOCs.b, The second proposed