Journal of the American Chemical Society

Developing enzymatic mechanisms for C-F bond formation remains a long-standing challenge. Here, we repurposed the biosynthetic nonheme Fe enzyme EgtB, which features a three-histidine facial triad, to catalyze C(sp3)-H fluorination reactions. Directed evolution of EgtB afforded two new-to-nature fluorine atom transferases with opposite enantiopreference, EgtBCHF1 and EgtBCHF2, with up to 28-fold improved total activity. In contrast to our previously evolved nonheme Fe fluorine atom transfer biocatalyst ACCOCHF, which contains a two-histidine-one-carboxylate facial triad, the evolved EgtBCHF variants displayed unexpected hydroxylation activity. 18O-labeling experiments showed that the hydroxy group originated from water rather than residual O2. Computational studies suggested that the three-histidine-supported Fe(III) center exhibits enhanced Lewis acidity compared to the two-histidine-one-carboxylate system, allowing deprotonation of Fe(III)-bound water to form a Fe(III)-OH species to catalyze radical hydroxylation. Primary coordination-sphere mutagenesis in EgtB and ACCO further supported the critical role of Fe coordination chemistry in controlling radical rebound reactivity and selectivity. Computational studies revealed that Fe coordination chemistry strongly influences both fluorine atom abstraction and radical rebound, with the intrinsic C-X (X = F, OH, and N3) bond forming radical rebound preference following the order N3 > OH > F. Furthermore, multivariate linear regression analysis revealed that fluorine atom abstraction is primarily governed by the intrinsic Fe-F bond strength, whereas fluorine rebound is predominantly controlled by the electronic structure of the Fe(III) intermediate. Together, these findings provide mechanistic insights into nonheme Fe enzymology and reprogramming toward selective radical rebound reactions, including challenging C-H fluorination. Table of Contents (TOC) O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=106 SRC="FIGDIR/small/737789v1_ufig1.gif" ALT="Figure 1"> View larger version (30K): org.highwire.dtl.DTLVardef@1ad85b2org.highwire.dtl.DTLVardef@1248bd4org.highwire.dtl.DTLVardef@58268dorg.highwire.dtl.DTLVardef@14b2da0_HPS_FORMAT_FIGEXP M_FIG C_FIG

Kaitocephalin (KCP) is a fungal neuroactive natural product bearing a peptide-like yet nonpeptidic amino acid-derived scaffold in which amino acid-like units are connected by C-C bonds rather than peptide bonds. The enzymatic construction of this unusual scaffold has remained unresolved. Here, we identify KpbH as a PLP-dependent enzyme that couples pyrroline-5-carboxylate, generated from L-ornithine, with L-aspartate to form (2S,5R)-5-((S)-2-amino-2-carboxyethyl)pyrrolidine-2-carboxylic acid (ACPCA), which corresponds to the nonpeptidic Ala-Pro substructure of KCP. D2O-labeling experiments showed enzyme-controlled, solvent-derived deuterium incorporation at C7 of ACPCA, supporting a decarboxylative Mannich-type mechanism. Feeding of a deuterium-enriched ACPCA-containing reaction mixture to the KCP-producing fungus Eupenicillium shearii resulted in deuterium incorporation into KCP, linking ACPCA to KCP biosynthesis. These results identify KpbH as the first native PLP-dependent enzyme that catalyzes an L-aspartate-dependent decarboxylative Mannich-type C-C bond-forming reaction and reveal a biosynthetic strategy for constructing a noncanonical amino acid-like C-C bond scaffold. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=106 SRC="FIGDIR/small/733665v1_ufig1.gif" ALT="Figure 1"> View larger version (29K): org.highwire.dtl.DTLVardef@a27fb7org.highwire.dtl.DTLVardef@6eea95org.highwire.dtl.DTLVardef@1eae086org.highwire.dtl.DTLVardef@13a92e9_HPS_FORMAT_FIGEXP M_FIG C_FIG

The lysosome is a major catabolic organelle responsible for the breakdown of both intra- and extracellular substrates1,2. Lysosomal membrane damage mediated by pathologic amyloid fibrils is an area of recent focus3-6. The dipeptide ester LLOMe is typically employed to model lysosomal membrane damage7-11; however its mechanism of membranolysis was previously incompletely understood. Here, in vitro and cell-based analyses, and cryo-electron microscopy and tomography studies reveal LLOMe-derived oligopeptides generated by the lysosomal protease Cathepsin C assemble into cross-{beta}-sheet amyloid fibrils within the lysosome. Additionally, we report lysosome membrane damage triggers the broadly nonspecific dipeptidyl ligase activity of Cathepsin C, facilitating the tagging of proximal proteins within the damaged lysosome lumen with a click chemistry handle: to our knowledge, the first reported localized proximity labeling approach exploiting a fully endogenous, non-engineered enzyme. While Cathepsin C ligase activity has been demonstrated in vitro12,13, our observations of dipeptidyl ligation onto proximal proteins in cells suggests an unexplored role of Cathepsin C in lysosomal biology and broadly exemplifies how other endogenous enzymes might be similarly exploited for proximity labeling. Altogether our results unveil two mechanisms by which dipeptide esters perturb lysosomal homeostasis and provide a roadmap for their utilization toward targeted studies of the lysosome.

The cysteine to serine mutation at residue 481 of Bruton's tyrosine kinase (BTK) is the most common mechanism of clinical resistance against ibrutinib for the treatment of mantle cell lymphoma and chronic lymphocytic leukemia. We report small molecule ligands containing chiral {beta}-lactone electrophiles to address this challenge. The asymmetric warhead enabled stereoselective covalent modification of wild-type and ibrutinib-resistant mutant BTK(C481S) through distinct sites of reactivity. Building on these findings, we developed kinase-directed {beta}-lactone probes and demonstrated that individual enantiomers preferentially engage distinct subsets of the kinome. These studies establish {beta}-lactones as stereochemically encodable covalent warheads whose stereochemistry can serve as a selectivity filter in covalent drug discovery.

Engineered bacterial routes for oxidation of non-native alcohols face three challenges: Nicotinamide-dependent enzymes are coupled to cellular redox metabolism, nicotinamide-independent aryl-alcohol oxidases (AAOs) usually express poorly in bacteria, and aldehyde products are rapidly modified by host enzymes. Here, we address these limitations by engineering aldehyde-retaining Escherichia coli for discovery and application of soluble bacterial AAOs. Screening 51 candidates revealed a high-expression sequence cluster containing enzymes that are active on diverse aromatic and furan-based alcohols. Pairing the top-performing AAO with designer pathways in aldehyde-retaining cells enabled modular C-N and C-C bond forming cascades starting from supplied alcohols. By making both the oxidase and its product compatible with the host, this work advances air-driven oxidation of diverse alcohols as a programmable entry point to aldehyde-derived chemistry in engineered bacteria.

Proteins are dynamic molecular machines that change shape in response to physical and chemical perturbations. Although single-molecule force spectroscopy provides precise information about the stretching of proteins in response to tunable forces, it does so without structural detail. Circular protein-DNA chimeras, with DNA attached to pairs of surface sites, have been introduced as an alternative way to tunably apply forces to proteins. Intriguingly, these chimeras should be tractable for atomic-level study by nuclear magnetic resonance (NMR) spectroscopy and other structural methods. Here, we describe the NMR-scale synthesis of circular chimeras of single- and double-stranded DNA with ubiquitin, an essential component of many cellular pathways. We designed these chimeras to probe a two-residue retraction of ubiquitin's C-terminal {beta}5 strand, normally triggered by phosphorylation of serine 65 during initiation of mitophagy. We probed the resulting conformational changes by NMR and found that the attachment of a single strand of DNA suffices to alter this conformational equilibrium. A control bearing two separate short single DNA strands recapitulated much of the circular chimera's NMR properties, supporting a dominant role for local protein-DNA interactions rather than spring-like action by single- or double-stranded DNA. These results provide a necessary benchmark for future studies using DNA springs to probe the functional dynamics of proteins.

Precisely modifying proteins at multiple sites in their native, folded structures offers unique opportunities to answer molecular and cellular-level biological questions. Here, we present a genetic code expansion strategy for site-specific integration of a fluorophore-quencher pair comprising two non-canonical amino acids--acridonylalanine (Acd) and methyltetrazinyl phenylalanine (Tet) -- into a protein expressed in E. coli. The Acd and Tet pair requires no post-translational labeling, and quenching can be switched off by biorthogonal or photochemical reactions of Tet for convenient internal control experiments. Mechanistic studies based on Stern-Volmer quenching, fluorescence lifetime measurements, and "proline ruler" peptides established the distance dependence of quenching. As proof-of-concept, we applied this strategy to study: 1) calmodulin, a calcium-sensing protein, 2) RecA, a DNA damage sensor in bacteria, and 3) LexA, a transcriptional repressor whose activation by RecA governs acquired antibiotic resistance in bacteria. Using these proteins, we demonstrate that dual Acd/Tet labeling provides molecular-level insights into protein dynamics, enables high-throughput drug screening, and advances tools for studying protein structure-function relationships.

The function of a protein depends not only on its sequence but on post-translational modifications and folding that produce functionally distinct proteoforms. Single-molecule methods for protein identification, such as nanopore sequencing, typically require denaturation or proteolysis, sacrificing conformational information that contributes to proteoform diversity. Here, we identify intact, folded proteins by recording an optical fingerprint of their local surface chemistry using a single covalent label. The signal is produced by a spontaneously blinking fluorophore attached to the protein through established bioconjugation reactions. The thermodynamics and kinetics of its switching between a fluorescent and a dark state are influenced by the immediate protein environment in a chemically interpretable manner. Further discriminative information can be extracted using deep learning to achieve excellent identification accuracy. Using this approach, we distinguish different proteins, different pockets of the same protein, and the presence of a single post-translational modification, in each case tracing the classification back to a distinct physicochemical mechanism. These results establish single-molecule fluorescence blinking as both a protein fingerprinting method and a probe of local chemistry on the surface of folded proteins.

Programmable molecular biology increasingly requires strategies for converting engineered recognition or proximity modules into measurable outputs, particularly within transcriptional regulation, RNA imaging, and CRISPR-associated systems. Synthetic chemically induced dimerization (CID) systems provide a class of programmable recognition modules for such applications, yet generalized strategies for coupling structurally diverse CIDs to functional readouts remain limited. Here, we introduce a CID-to-output conversion strategy based on engineering of the linker-mediated coupling interface. Using single-fluorescent-protein sensors as an experimentally tractable optical model readout, we systematically varied paired N- and C-terminal linkers flanking circularly permuted green fluorescent protein (cpGFP) to map coupling landscapes across synthetic CID systems derived from combinatorial selection and computational protein design. The results revealed strong non-additive interactions across paired linkers and suggest that linker length is a first-order determinant of CID-to-output coupling. Across nanobody-, monobody-, and de novo-designed CID architectures, this framework yielded functional sensors with dynamic ranges up to 1270% and robust responses in mammalian cells. Together, this work demonstrates that effective CID-to-output conversion can be achieved by empirically mapping the linker-mediated coupling interface, providing a practical route for adapting synthetic CID to diverse programmable molecular readouts and nucleic-acid-associated synthetic biology systems O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=94 SRC="FIGDIR/small/735888v1_ufig1.gif" ALT="Figure 1"> View larger version (25K): org.highwire.dtl.DTLVardef@1111094org.highwire.dtl.DTLVardef@1579e8aorg.highwire.dtl.DTLVardef@16981feorg.highwire.dtl.DTLVardef@1d588f7_HPS_FORMAT_FIGEXP M_FIG C_FIG

I.Rapid developments in magic-angle-spinning (MAS) hardware over the past two decades have made possible the acquisition of high-resolution spectra of protons in solids, fuelling studies of small and large molecules alike. Nevertheless, proton resolution, limited by the strong dipole-dipole coupling network, remains a bottleneck even at MAS frequencies exceeding 100 kHz. We present here techniques based on phase-modulated homonuclear decoupling that dramatically improve proton coherence times and resolution compared to 60-95 kHz MAS alone using low average radio-frequency amplitudes (< 100 kHz). A relatively high sensitivity (40- 70%) and a straightforward optimization procedure directly on the sample being studied allows these gains to be realised in large biomolecules, as demonstrated here on a 326-residue cytoskeletal protein in its filamentous state. These techniques enable experiments with improved resolution on biomolecules while simultaneously taking advantage of the higher sensitivity available on probes with relatively large rotor volumes that cannot reach higher MAS frequencies.

The incorporation of non-canonical amino acids (ncAAs) into proteins, developed in the past 20 years, has opened new avenues with respect to protein structure, protein modification, protein-protein interaction or enzyme catalysis beyond what is possible with the 20 proteinogenic AAs. Although >300 unusual building blocks including several ncAAs have been described in nonribosomal peptides (NRPs) naturally, we aimed to further expand the scope of the underlying nonribosomal peptide synthetases (NRPS) to incorporate ncAAs beyond the naturally available ones. We have therefore systematically screened for ncAA accepting NRPS systems, applied NRPS engineering to transfer the respective ncAA-accepting parts into other NRPSs and thereby created novel peptides that were further derivatized in post-enzymatic chemical synthesis reactions directly in bacterial culture extracts. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=177 SRC="FIGDIR/small/738027v1_ufig1.gif" ALT="Figure 1"> View larger version (38K): org.highwire.dtl.DTLVardef@90552forg.highwire.dtl.DTLVardef@1c8a5e0org.highwire.dtl.DTLVardef@2549dorg.highwire.dtl.DTLVardef@1012911_HPS_FORMAT_FIGEXP M_FIG C_FIG

Biomolecular condensates exhibit spontaneous electrochemical microenvironments characterized by asymmetric ion distributions and pH gradients that emerge from protein-sequence-dependent charge regulation. Despite their biological importance, mechanistic understanding of these microenvironments has been constrained by the absence of computationally tractable frameworks capable of treating proton exchange, counterion partitioning, and buffer equilibria on consistent thermodynamic footing. Here, we introduce the buffered Charge-Regulation Monte Carlo (b-CR-MC) framework, which couples grand-canonical exchange of ions and buffer species with explicit charge regulation of titratable residues. By extending the CR-MC ion-merging strategy to multicomponent reservoirs and employing the Restricted Primitive Model, b-CR-MC achieves computational efficiency while maintaining thermodynamic rigor, with quantitative agreement to the more expensive generalized G-RxMC approach. Applied to full-length FUS (net positive) and PGL-3 (net negative) under physiological conditions, the framework reveals sequence-dependent pH gradients: the dense phase of FUS exhibits an alkaline shift, while PGL-3 exhibits an acidic shift, in both cases driving the condensate interior toward the protein's isoelectric point. Slab-geometry simulations further resolve the Donnan potential and continuous ion profiles across the condensate interface, confirming the direction and magnitude of these electrochemical shifts. Additionally, we identify spatially resolved buffer depletion within dense phases, establishing that dynamic charge regulation is a primary determinant rather than a secondary correction to condensate electrochemistry. By establishing a sequence-resolved, thermodynamically consistent computational platform, b-CR-MC enables quantitative prediction of how mutations and post-translational modifications reprogram condensate microenvironments across biological and pathophysiological contexts.

The emergence of respiratory, photosynthetic, and assimilatory complexes in evolution required proteins capable of binding multiple catalytic and electron-transfer cofactors while exerting fine control over their spatial arrangement. Across natural systems these cofactors are preferentially positioned in loop regions. In contrast, most protein design strategies have focused on installing cofactor-binding sites within helical elements. Here we show that introducing only a pair of appropriately placed histidine ligands into the interhelical loop regions of a canonical single-chain four-helix bundle is sufficient to create new well-defined high affinity heterocofactor binding sites. This simple modification enables the self-assembly of complexes containing up to three distinct cofactors in a single designed domain with positional specificity. Using this strategy, we creat-ed constructs containing one or two hemes in combination with Zn(II) phthalocyanine monosulfonate, Zn-heme, and the light-harvesting Zn(II) tetraphenylporphyrin tetrasulfonate. Fluorescence measurements of constructs containing the latter show efficient energy transfer between photoactive donor cofactors. By demonstrating that loop-embedded ligands support robust, modular, and evolutionarily plausible cofactor recruitment, this work provides a mechanistic explanation for the widespread placement of redox and catalytic cofactors in loops in natural proteins: only limited packing complementarity is needed, meaning that just a few mutations can introduce a functional cofactor binding site, after which additional mutations can tune affinity, reactivity, and specificity. More importantly, it establishes a straightforward path toward constructing func-tional protein domains that mirror the complexity of biological energy-conversion architectures.

Dynamic biomolecular condensates play crucial roles in intracellular compartmentalization and physiological functions. While engineering tools for compartmentalization have expanded add-on functionalities, directly amplifying the inherent catalytic machinery within biological phase-separated droplets has remained elusive. Herein, we developed a phase-separated oxidative folding reaction chamber based on protein disulfide isomerase A6 (PDIA6) by chemically targeting its active site CxxC motif to enhance enzymatic activity within PDIA6 droplets. A para-substituted N-methylated pyridinylmethanethiol (pMePySH) enhanced the catalytic oxidative folding of bovine pancreatic trypsin inhibitor, proinsulin, and antibody up to 12-fold within in vitro PDIA6 droplets. Furthermore, pMePySH targeted PDIA6 foci within the endoplasmic reticulum, significantly promoting insulin secretion. These findings offer a powerful platform for the spatiotemporal manipulation of protein folding, with profound implications for the scalable manufacturing of therapeutic antibodies and other complex biopharmaceuticals.

Serotonin is a key neurotransmitter, and aptamer-based tools using the 44 nt Apt44 have been successfully developed for its in vitro and in vivo detection. Nevertheless, the structural basis of recognition by this aptamer remains unclear. Here we report high resolution NMR structures of Apt38, a 6-nt truncated variant in the third loop of Apt44, in free and serotonin-bound states. Both structures reveal a two layered antiparallel chair type G quadruplex core with three edgewise loops and a terminal duplex, forming a G quadruplex duplex hybrid structure. Serotonin binds at the G quadruplex duplex junction, stabilized by stacking, electrostatic attraction, hydrogen bonding, and hydrophobic contacts. Apt38 is preorganized for binding, whereas the longer third loop of Apt44 introduces conformational dynamics into the G quadruplex scaffold, which enables a pronounced binding triggered conformational switch in PBS buffer, explaining its sensing mechanism. Our work reveals the recognition and sensing mechanism of the serotonin aptamer and provides a framework for aptamer design in serotonin biosensing.

The non-amyloid {beta} component (NAC) domain of alpha synuclein (Syn) drives Syn aggregation in Parkinson disease, yet as an intrinsically disordered segment it lacks a stable epitope for conventional ligand discovery. Using deep learning-based protein design, we generated compact single-chain binders that take advantage of the propensity of the NAC domain to adopt an extended {beta}-strand conformation, which they engage and stabilise. From 21 expressed designs, 3 engaged the target both in vitro and within live cells. A single round of partial diffusion improved their affinity, maturing the strongest binder to a dissociation constant of 1.94 nM with no detectable cross-reactivity to tau, amylin or amyloid {beta}. Solution NMR spectroscopy confirmed Syn peptide association with all three binders and showed that peptide binding induces conformational changes consistent with the intended design architecture. Each binder engaged a distinct point on the aggregation pathway. Kinetic analysis combining seeded aggregation with the measured affinities for monomer, oligomer and fibril resolved the specific microscopic step that each binder inhibits. Notably, the most potent binder acted by selectively capturing on-pathway oligomers, the species most closely linked to toxicity, and suppressed fibril formation at substoichiometric ratios without engaging the bulk monomer. Together, these findings establish de novo-designed Syn binders that selectively target distinct aggregation intermediates to mechanistically reshape Syn assembly, providing a framework for the rational design of aggregation-modifying proteins.

Achieving precise control of CRISPR/Cas trans-cleavage depends on understanding how nucleic acid activators engage Cas effectors, yet the fundamental principles of split-trigger activation of Cas12a remain unclear. Here, we uncover the mechanistic determinants that enable fragmented nucleic acids to collectively initiate Cas12a activity. We show that split triggers bearing external extensions fully support the R-loop formation, whereas internal extensions which disrupt the spacer complementarity abolish Csa12a activation. We further demonstrate that covalent linkage of split-trigger fragments prevents R-loop propagation, revealing that Cas12as activation strictly requires two physically independent split fragments. Together, these findings establish a synergistic split-trigger activation mechanism in which cooperative hybridization of two individually fragments nucleates and extends the Cas12a R-loop with high efficiency. Conceptually, this mechanism enables a cascade architecture that transforms CRISPR diagnostics from a one-target one-Cas ribonucleoprotein (RNP) paradigm into a highly amplifying process in which a single target molecule activates numerous downstream Cas RNPs. Building on this principle, we show that the cleavage of a rationally designed linear DNA-RNA-DNA mediator by LbuCas13a generates optimally configured split triggers for Cas12a activation, thereby coupling RNA recognition to large-scale Cas12a activation without enzymatic preamplification. The resulting Split Trigger Activated Cas13-Cas12 Cascade System (STACS) achieves amplification-free detection down to 1 copy/{micro}L within 15 minutes and maintains robust performance in complex biological (serum, saliva) and environmental (mud) matrices. This work establishes a generalizable strategy for engineering programmable CRISPR cascades with high Cas RNP activation multiplicity for ultrasensitive molecular diagnostics. Graphic abstract.Mechanism and detection workflow of the Split Trigger Activated Cas13-12 Cascade System (STACS). O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=101 SRC="FIGDIR/small/734747v1_ufig1.gif" ALT="Figure 1"> View larger version (19K): org.highwire.dtl.DTLVardef@1407364org.highwire.dtl.DTLVardef@57d5f5org.highwire.dtl.DTLVardef@a00693org.highwire.dtl.DTLVardef@fab93_HPS_FORMAT_FIGEXP M_FIG C_FIG

Messenger RNA (mRNA) is a prerequisite for programmable protein expression, but its therapeutic and synthetic-biology applications are limited by instability and susceptibility to degradation. Hybridizing mRNA to short DNA strands can fold it into a compact origami nanostructure, protecting it from degradation but impeding ribosome access. However, how such a folded mRNA is translated, and which parts must be left unpaired, remain unclear. Here we fold an EGFP-encoding mRNA into a six-helix bundle and leave defined regions of the coding sequence unpaired to examine what the ribosome requires. We find that the start of the coding sequence must be accessible for translation, whereas leaving the far end unpaired makes no difference. Counterintuitively, leaving more of the coding sequence unpaired does not help: translation first falls and then partially recovers as the unpaired region lengthens, a reproducible pattern set by how that region folds rather than by its length. Modified mRNAs carrying 5-methoxyuridine or N1-methylpseudouridine still fold correctly into the six-helix bundle and show the non-monotonic translation pattern; the modification only shifts the overall level of protein produced, with N1-methylpseudouridine giving the most. Together these results begin to define how a folded mRNA can be made both stable and efficiently translated. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=82 SRC="FIGDIR/small/734245v1_ufig1.gif" ALT="Figure 1"> View larger version (36K): org.highwire.dtl.DTLVardef@e26126org.highwire.dtl.DTLVardef@580c65org.highwire.dtl.DTLVardef@95ed54org.highwire.dtl.DTLVardef@110271f_HPS_FORMAT_FIGEXP M_FIG C_FIG

-Synuclein (S), a protein that plays a central role in Parkinsons disease and related synucleinopathies, is an intrinsically disordered protein (IDP) whose functional interactions and aggregation behavior can be strongly influenced by post-translational modifications (PTMs). Phosphorylation, acetylation, and other PTMs regulate Ss interactions with lipid membranes and binding partners, whereas their dysregulation is associated with aggregation and neuronal toxicity. Despite significant progress through chemical and semi-synthetic approaches, investigating the combinatorial effects of PTMs has remained challenging due to the lack of accessible, site-specific methods. Here, we present an integrated strategy combining genetic code expansion, enzymatic modification, and intein-mediated click chemistry to generate S variants bearing multiple defined PTMs and a C-terminal fluorescent label. The resulting constructs enable direct evaluation of how individual and combined PTMs influence S structure, lipid binding, and cellular internalization. Our approach expands the molecular toolkit for dissecting PTM crosstalk in S and other aggregation-prone IDPs, advancing mechanistic understanding and supporting the development of therapeutic strategies for neurodegenerative disease.

Camptothecin derivatives are first-line anticancer drugs used worldwide for the treatment of diverse malignant tumors. However, the biosynthetic pathway of camptothecin has remained elusive for five decades. Here, we fully map its entire biosynthetic route. We discovered five key missing enzymes (OpCAR, OpSDR11, OpCS, OpGH1, and OpSTR) via the combination of MALDI mass spectrometry imaging, single-cell RNA sequencing and co-expression analysis. Meanwhile, we demonstrated a free flavin mononucleotide triggered the non-enzymatic 6-5-6 to 6-6-5 fused-ring skeleton rearrangement, filling the last gap in camptothecin biosynthesis. Finally, we validated this identified pathway and achieved the de novo biosynthesis of camptothecin in Saccharomyces cerevisiae. These discoveries uncover the long-standing mystery underlying camptothecin and pave the way for manufacturing camptothecin and its derivatives through synthetic biology approaches.