{"paper_id":"1d129196-8a45-4e86-b5fa-983eb383dc7a","body_text":"1 \n \nEngineering auxin degradation into r oot-associated bacteria promotes plant \ngrowth \nTing Jiang1, 10, Yihui Shen2, 3, 9, 10,*, Xi Li2, 3, Michal J. Kozlowski1, Philip D. Jeffrey4, John \nT. Groves2, Joshua D. Rabinowitz2, 3, 5 and Jonathan M. Conway1, 4, 6, 7, 8,* \n \n1Department of Chemical and Biological Engineering, Princeton University, Princeton, \nNJ, USA. \n2Department of Chemistry, Princeton University, Princeton, NJ, USA. \n3Lewis-Sigler Institute for Integrative Genomics, Princeton University, Princeton, NJ, \nUSA. \n4Department of Molecular Biology, Princeton University, Princeton, NJ, USA. \n5Ludwig Institute for Cancer Research, Princeton Branch, Princeton, NJ, USA. \n6Omenn-Darling Bioengineering Institute, Princeton University, Princeton, NJ, USA. \n7High Meadows Environmental Institute, Princeton University, Princeton, NJ, USA. \n8Andlinger Center for Energy and the Environment, Princeton University, Princeton, NJ, \nUSA. \n9Present address: Department of Bioengineering, University of Pennsylvania, \nPhiladelphia, PA, USA. \n10These authors contributed equally: Ting Jiang, Yihui Shen \n* Correspondence to: yihuis@seas.upenn.edu; jmconway@princeton.edu \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n2 \n \nAbstract \nOverproduction of indole-3-acetic acid (IAA) by rhizosphere bacteria disrupts auxin \nhomeostasis and induces root growth inhibition (RGI) in plants. Bacteria from the genus \nVariovorax mitigate this effect by degrading IAA, and in our prior work we identified the \niad locus as being required for this activity. Here, we refine our understanding of the iad \npathway using bacterial genetics, metabolomics, and isotope tracing to assign roles to \nindividual Iad pathway enzymes and show that IadDE, though resembling a Rieske \ndioxygenase, functions instead as a monooxygenase that initiates catabolism via a \nnovel intermediate. Guided by these insights, we installed chromosomal iad  cassettes \ninto root-associated commensals ( Polaromonas MF047 and Paraburkholderia MF376), \ncreating the first engineered bacteria that reprogram rhizosphere auxin homeostasis in \nmicrobially complex environments to benefit the plant. In natural soil, engineered \nParaburkholderia enhanced plant biomass, and co mmunity profiling revealed no \nsignificant differences in microbiome composition between engineered and wild type \ntreatments, supporting that auxin degradation conferred plant benefit without broader \ndisruption of the rhizosphere community. Together, this work refines the pathway logic \nof microbial auxin degradation and demonstrates that commensals can be rationally \nengineered to deliver auxin-balancing functions in complex rhizosphere microbiomes. \nMore broadly, it provides a framework for leveraging mechanistic insight to engineer \nplant-associated commensals that enhance plant growth, laying the foundation for \ndeployment in agricultural settings. \n \nIntroduction \nPlant roots develop within a complex and densely populated microbial environment, \nwhere interactions with soil and rhizosphere microbes strongly influence root \narchitecture and function\n1,2. A central regulator of root development is auxin, a group of \nsignaling molecules that function as plant hormones but are also synthesized by \nbacteria, fungi, and animals\n3. Among them, indole-3-acetic acid (IAA) is the \npredominant and most indispensable auxin 4. While plants maintain auxin homeostasis \nthrough biosynthesis, transport, conjugation, and degradation, many rhizosphere \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n3 \n \nmicrobes also synthesize IAA. In microbial contexts, IAA contributes to bacterial \nphysiology, stress adaptation, and microbe–microbe communication 3. However, in \nplants, microbial IAA production can disrupt auxin homeostasis and lead to root growth \ninhibition (RGI)\n5. Genomic surveys indicate that over 80% of soil- and plant-associated \nbacterial genomes harbor complete or partial IAA biosynthetic pathways 6,7, \nunderscoring the widespread microbial influence on plant hormone dynamics. \nLike a natural counterbalance, certain rhizosphere bacteria can mitigate RGI by \ndegrading IAA. Two major types of auxin-degrading pathways have been identified in \nsoil- or plant-associated bacteria: the iad-like and iac-like pathways\n8,9. The iad-like (IAA \ndegradation) pathway, found in genera such as Variovorax, Alcaligenes, Achromobacter \nand Bradyrhizobium , converts IAA into anthranilic acid 5,9. In contrast, the iac-like \n(indole-3-acetic acid catabolism) pathway, originally described in Pseudomonas putida \n1290, degrades IAA into catechol 10,11. Compared to the iac-like pathway, the iad-like \npathway is more effective in reversing RGI induced by root-associated bacteria 9. For \nexample, Variovorax strains carrying the  iad  pathway fully restored root growth in \nArabidopsis thaliana  seedlings exposed to a 175-member RGI-inducing synthetic \ncommunity, whereas iac-containing strains failed to do so9. \nDespite its critical role, the molecular mechanisms underlying iad-mediated IAA \ndegradation remain incomplete. The iad locus is known to be regulated by two MarR-\nfamily transcription factors, MarR73 and MarR50, with MarR73 serving as the primary \nrepressor\n9. The first catalytic module, comprising IadDE (annotated as a Rieske non-\nheme dioxygenase) and the associated reductase IadC, is postulated to form a two-\ncomponent dioxygenase system that facilitates electron transfer and enhances catalytic \nefficiency\n9,12.  \nPrevious studies have demonstrated the feasibility of engineering IAA-degrading activity \nby introducing iad genes into heterologous hosts 5,9,12. Plasmid-based expression of the \niad locus in the root-associated isolate Acidovorax  Root219 conferred IAA-degrading \nactivity in vitro and promoted primary root elongation under exogenous IAA treatment 5. \nHowever, this Acidovorax Root219 strain failed to fully reverse RGI caused by the \nauxin-producing Arthrobacter CL028, let alone more complex auxin-producing bacterial \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n4 \n \ncommunities5. Additional knowledge of the contribution of iad-mediated auxin \ndegradation to microbiome structure and mechanistic understanding of the genetics and \nbiochemistry of the iad pathway would facilitate rational engineering of the iad locus into \nother strains. These past findings further suggest the best chassis strains would \npossess traits such as plant colonization capacity, competitiveness in the rhizosphere, \nand ecological stability as well as be genetically tractable for chromosomal genetic \nmanipulation. Investigating new strains engineered with the iad pathway will lay the \ngroundwork for future applications of the iad pathway and delivery of robust auxin \nhomeostasis phenotypes in agricultural and ecological settings. \nHere, we systematically dissect the iad-mediated IAA degradation pathway in V. \nparadoxus CL014 using a combination of genetic, metabolomics, and isotope tracing. \nWe identify a nine-gene region ( iadCDEFGHIJK2) within the iad locus that is \nresponsible for IAA catabolism. V. paradoxus CL014 initiates IAA degradation through \nan unreported two-step oxidative mechanism. In the first step, the IadCDE complex \nfunctions as a monooxygenase that incorporates one oxygen atom from molecular \noxygen into IAA. Although IadCDE is structurally homologous to canonical dioxygenase \ncomplexes, our metabolomic and isotopic analyses demonstrate that it functions as a \nmonooxygenase, revealing a striking divergence between structure and catalytic \nmechanism. The second oxidative step is mediated by the dehydrogenase IadJ \nfollowing hydrolysis, leading to oxygen incorporated from water. Notably, we identify the \nproduct of IadCDE to be an uncharacterized compound (C\n₁₀ H₉ NO₃ ), rather than the \ncommonly known intermediate, 2-oxindole-3-acetic acid (oxIAA, C ₁₀ H₉ NO₃ ), which is \nsupportive of an epoxide-forming mechanism by IadCDE. To explore the broader \napplicability of this pathway, we genomically integrated the iad  pathway genes, under \ncontrol of the MarR73 regulator, into two additional root-associated bacteria, \nPolaromonas MF047 and Paraburkholderia MF376, and evaluated their IAA-degrading \nactivity and ability to mitigate RGI in both simplified and complex bacterial community \ntreatments, using Arabidopsis thaliana and Medicago truncatula as host plants. \nEngineered Paraburkholderia MF376 emerged as a promising chassis for agricultural \napplications, as it effectively reversed RGI caused by strains that V. paradoxus CL014 \ndid not fully counteract, enhanced Arabidopsis growth in natural soil over long-term \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n5 \n \nincubation, and exhibited robust colonization when introduced into new microbial \ncommunities. These findings advance our understanding of microbial auxin metabolism \nand its relevance to restoring rhizosphere auxin balance and enhancing plant growth. \n \nResults \nIdentification of a nine-gene region within the iad locus responsible for IAA \ndegradation in Variovorax paradoxus CL014 \nPrevious studies indicated that V. paradoxus CL014 degrades IAA through a pathway \nsimilar to that of Bradyrhizobium japonicum, producing intermediates such as dioxindole, \nisatin, isatinic acid, and anthranilic acid—an intermediate in tryptophan \nbiosynthesis\n9,13,14. However, the specific genes responsible for each step remained \nunidentified. The iad locus (Hot Spot 33, HS33), consisting of 25 genes (Fig. 1a), was \nidentified as responsible for IAA degradation via genomic deletion 5. To pinpoint key \ngenes, we cloned different segments of the iad locus into the broad-host vector pBBR1 \nand introduced them into an iad locus deletion mutant (Δ HS33) (Extended Data Fig. 1a). \nMetabolomic analysis revealed that mutants carrying the iadC-K2 region restored the \nlevel of intermediates and the final product, anthranilic acid, indicating that this nine-\ngene region encodes the complete IAA degradation pathway (Fig.1a, Extended Data \nFig. 1b).  \nTo identify specific substrates and products of iad genes, we performed liquid \nchromatography in tandem with high-resolution mass spectrometry (LC-MS), focusing \non metabolites previously associated with IAA degradation in V. paradoxus CL014\n9. We \nreasoned that knocking out a gene downstream of a metabolite or overexpressing a \ngene upstream would lead to metabolite accumulation or depletion, respectively. \nSpecifically, we created knockout mutants in the wild-type background and \noverexpression mutants in the \nΔ HS33 background (Fig. 1b, Extended Data Fig. 2). \nBased on gene annotations and structural data, iadD and iadE encode the large and \nsmall subunits of a Rieske non-heme dioxygenase, while iadC encodes a reductase, \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n6 \n \nforming a two-component dioxygenase system (Fig. 1k) 9,12. Accordingly, iadCDE was \ntreated as a single functional unit for mutant construction.  \nOur analysis revealed that deletion of iadCDE, iadF, iadG, iadH, iadI, or iadJ  \nsignificantly reduced anthranilic acid (C ₇ H₇ NO₂ ) production (Fig. 1c), confirming their \nessential roles in IAA degradation. Complementation of the Δ HS33 mutant with the full \niadC–K2 gene region fully restored anthranilic acid levels (Fig. 1c). Notably, the \npresence of iadCDE in these segments was able to rapidly degrade IAA. This confirms \nthat IadCDE directly uses IAA as its substrate and catalyzes the initial step of IAA \ndegradation (Fig. 1d, Extended Data Fig. 1c). We next examined metabolite \naccumulation in the \nΔ HS33::iadCDE strain and identified two compounds, C ₁₀ H₉ NO₃  \nand likely its hydration product C ₁₀ H₁₁ NO₄ , suggesting C ₁₀ H₉ NO₃  to be the product of \nIadCDE (Fig. 1e, f). Interestingly, iadJ deficiency also led to the accumulation of \nC₁₀ H₉ NO₃  and C ₁₀ H₁₁ NO₄ , indicating that IadJ likely consumes one of these \nintermediates (Fig. 1e). Similarly, we found that C₁₀ H₉ NO₄  is depleted with iadJ deletion \nwhile it accumulates in Δ HS33::iadCDEJK2, suggesting C ₁₀ H₉ NO₄  to be the product of \nIadJ (Fig. 1g, h). Deletion of iadF, iadG, iadH and iadI all lead to C ₁₀ H₉ NO₄  \naccumulation, among which Δ iadF shows the strongest accumulation suggesting \nfunctional proximity of IadF to IadCDEJ  (Fig. 1g). Other projected intermediates were \nnot detected, thus IadFGHI likely carry out a chain of reactions, with most intermediates \nquickly channeled through the enzymes. Therefore, we resorted to annotated enzyme \nfunction to infer reactions carried out by these genes. Annotated as an acyl-CoA \nsynthase (AMP-forming), IadF likely activates the carboxylic group in the acetyl side \nchain, enabling its subsequent removal by IadG, an acetyl-CoA acetyltransferase \n(ketothiolase, EC 2.3.1.16) (Fig. 1k). This reaction yields CoA-bound intermediates that \nare membrane-impermeable, ensuring their retention within bacterial cells and enabling \nefficient recognition and processing\n14. Meanwhile, IadH, annotated as an alcohol \ndehydrogenase, likely facilitates further processing (Fig. 1k). Together, IadF, IadG, and \nIadH mediate the reductive removal of the acetyl side chain (-C ₂ H₂ O₂ ), producing \nC₈ H₇ NO₂  (Fig. 1i). Indeed, C ₈ H₇ NO₂  accumulates in Δ HS33::iadCDEFGH_JK2 \nexpression strain as well as the iadI  deletion strain (Fig. 1i), suggesting that IadI acts \ndownstream of C ₈ H₇ NO₂ . We also detected C ₇ H₇ NO and C ₈ H₇ NO₃ , which strongly \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n7 \n \naccumulate only in strains overexpressing the whole iadC-K2 gene locus, suggesting \nthat they are likely intermediates between the IadI product and the final product \nanthranilic acid (C ₇ H₇ NO₂ ) (Fig. 1b, j). Although iadI is annotated as a kynurenine \nformamidase, our MS/MS analysis confirmed that C ₈ H₇ NO₃  is not N-formylanthranilic \nacid—the expected formylated derivative of anthranilic acid—indicating that IadI may \ninstead function as a broad-specificity amidase. Lastly, IadK2 was previously identified \nas a highly IAA-specific ATP-binding cassette (ABC) transporter solute-binding protein \ninvolved in IAA uptake\n12. However, it is not essential, as Δ HS33::iadCDE was still \ncapable of taking up and utilizing IAA, suggesting alternative uptake mechanisms9. \nIn summary, we identified a nine-gene region (iadC-K2) within the iad locus responsible \nfor IAA degradation in V. paradoxus  CL014 (Fig. 1a, Extended Data Fig. 1a, b). By \nintegrating metabolite abundance changes with gene functional annotations, we \nassigned putative reaction steps to each gene and reconstructed the degradation \npathway (Fig. 1k). This pathway links specific functional genes to corresponding \nintermediates, providing a framework for understanding IAA metabolism in this \nmicroorganism.  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n8 \n \nFig. 1 | Metabolomic profiling of iad pathway mutants reveals substrates and \nproducts of genes involved in IAA degradation in V. paradoxus CL014. a, Genomic \norganization of Hot Spot 33 in V. paradoxus CL014. Gene annotations are shown above \nas the final two digits of the IMG gene ID (26436136##) and below with assigned gene \nnames. The MarR-family transcriptional regulator marR73 is shown in red; the core \nnine-gene IAA degradation locus is highlighted in yellow. b, Heatmap of Log ₂  fold \nchanges in metabolite abundance in iad  pathway mutants relative to the wild type. Full \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n9 \n \nLC–MS profiles are provided in Extended Data Fig. 2. Data represent the mean of \nn/i2 =/i2 3 biological replicates. c–j, Bar plots showing the relative abundance of anthranilic \nacid ( c), IAA ( d), and key metabolic intermediates ( e–j) across wild-type and mutant \nstrains. Data represent mean ± s.e.m. (standard error of the mean) of n = 3 biological \nreplicates. k, Schematic representation of the iad-mediated IAA degradation pathway in \nV. paradoxus CL014. \n \nIsotope tracing reveals new intermediates and an unreported two-step oxidative \nmechanism revising the iad-mediated IAA degradation pathway \nPreviously, IAA degradation via the iad pathway was proposed to proceed through \nmonooxygenation to generate 2-oxindole-3-acetic acid (oxIAA, C ₁₀ H₉ NO₃ ), followed by \nconversion to dioxindole 9,12. Our metabolomic analysis, however, revealed two distinct \nC₁₀ H₉ NO₃  peaks in wild-type extracts: a major peak at 4 min and a minor peak at 7 min \n(Extended Data Fig. 3a). Using an oxIAA standard, we confirmed that the minor peak \ncorresponds to oxIAA (Extended Data Fig. 3a); however, this compound is not depleted \nin the \nΔ HS33 mutant, making it unlikely to be the product of IadCDE (Extended Data Fig. \n3b-c). By contrast, the major C ₁₀ H₉ NO₃  peak is depleted in the Δ HS33 mutant and \naccumulates in iadCDE overexpressing strains, indicating that it represents the product \nof IadCDE (Extended Data Fig. 3b-d). These findings indicate that the observed \nC₁₀ H₉ NO₃  is not oxIAA but rather a previously uncharacterized intermediate, suggesting \nthat IAA degradation in V. paradoxus CL014 proceeds via a mechanism distinct from \nthe canonical oxIAA pathway. \nTo elucidate the mechanism of IAA degradation, we performed isotope tracing and LC-\nMS/MS analysis of both labeled and unlabeled pathway intermediates (Fig. 2, Extended \nData Fig. 3 and 4). Specifically, we employed uniformly deuterium-labeled IAA ([²H₇ ]IAA) \nto monitor hydrogen rearrangements during catabolism, providing greater mechanistic \nresolution than the previously used [¹³C\n₆ ]IAA9. Tracing with [²H ₇ ]IAA revealed distinct \nmass shifts in pathway intermediates relative to unlabeled controls, indicating the \nnumber of retained deuterium atoms. The first intermediate, C\n₁₀ H₉ NO₃ , retained all \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n10 \n \nseven deuterium atoms, suggesting no isotope loss during the initial oxidation step (Fig. \n2a, Extended Data Fig. 4a, c). This observation is inconsistent with mechanisms \ninvolving oxIAA or its enol form, 2-hydroxy-IAA\n9,12, or a radical intermediate from \nhydrogen atom abstraction15. Indeed, the minor C₁₀ H₉ NO₃  peak corresponding to oxIAA \nretained only six deuterium atoms (Extended Data Fig. 3d). The subsequent \nintermediate, C ₁₀ H₉ NO₄ , showed the loss of one deuterium, whereas all downstream \nintermediates consistently retained four deuterium atoms on the aromatic ring (Fig. 2a, \nExtended Data Fig. 4a, c). \nTo identify the source of oxygen in the formation of C ₁₀ H₉ NO₃ , C ₁₀ H₁₁ NO₄ , and \nC₁₀ H₉ NO₄ , respectively, we performed ¹ /i2 O-labeled water (H ₂ ¹/i2 O) tracing in bacterial \nculture (Fig. 2b, Extended Data Fig. 4b). The IadCDE monooxygenase complex uses \nmolecular oxygen (O₂ ) and therefore will not generate labeled product (C₁₀ H₉ NO₃ ) from \nH₂ ¹/i2 O. On the contrary, the IadJ dehydrogenase mechanism leads to oxidized product \nthat can be labeled by H ₂ ¹/i2 O. The minor C ₁₀ H₉ NO₃  peak corresponding to oxIAA is \nsignificantly labeled by H ₂ ¹/i2 O, further confirming oxIAA is not formed by IadCDE \n(Extended Data Fig. 3e). In contrast, we detected no labeling in the major C ₁₀ H₉ NO₃  \npeak in iadCDE  overexpressing strain, consistent with it being the product of IadCDE \n(Fig. 2b). Meanwhile, both C ₁₀ H₁₁ NO₄  and C ₁₀ H₉ NO₄  incorporated ¹ /i2 O in the same \nstrains, supporting that C ₁₀ H₁₁ NO₄   is formed via hydration of C ₁₀ H₉ NO₃  in the cells, \nand is further processed by IadJ  to C ₁₀ H₉ NO₄  through a dehydrogenase mechanism \n(Fig. 2b). Further MS/MS analysis of C ₁₀ H₉ NO₄  revealed that the incorporated ¹ /i2 O \nresides either on the carboxylate or the ketone moiety (Extended Data Fig. 4b-d). \nInterestingly, C\n₁₀ H₉ NO₄  remained almost unlabeled in the wild type, which may be due \nto the rapid hydrolysis of C ₁₀ H₉ NO₃  within the IadCDE catalytic center, where H ₂ O is \nderived from catalytic reduction of molecular oxygen possibly by IadC. \nCollectively, our data support a revised IAA degradation pathway in V. paradoxus  \nCL014 (Fig. 2c). Specifically, the initial oxidation of IAA proceeds through two sequential \nsteps (C ₁₀ H₉ NO₂  →  C ₁₀ H₉ NO₃  →  C ₁₀ H₉ NO₄ ) (Fig. 2c). IadCDE functions as a \nmonooxygenase complex that carries out the first oxidation, incorporating oxygen \nderived from molecular oxygen, whereas IadJ acts as a dehydrogenase that completes \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n11 \n \nthe second oxidation, incorporating the oxygen atom from H₂ O. Our data reveal that the \nintermediate C₁₀ H₉ NO₃  is distinct from commonly believed pathway intermediate, oxIAA, \nalthough they share the same molecular formula. Following the observed deuterium \nlabeling and mechanisms of known heme-dependent tryptophan or indole \ndioxygenases\n16–19, we propose that oxidation by IadCDE proceeds via a 2,3-epoxide \nintermediate. This is followed by spontaneous ring opening via hydrolysis and IadJ \ncatalyzed dehydrogenation to produce dioxindole-3-acetic acid (C ₁₀ H₉ NO₄ ). This \nrevised pathway provides a more accurate framework for understanding IAA \ndegradation in V. paradoxus CL014.  By homology, it suggests that the iad-like pathway \nrepresents a conserved and unique auxin-catabolic strategy among certain plant-\nassociated commensals\n9,20. This work also lays the foundation for further exploration of \niad-like pathways in these plant-associated bacteria in genera such as Bradyrhizobium, \nAlcaligenes, and Achromobacter9. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n12 \n \nFig. 2 | Isotope tracing with [²H ₇ ]IAA and H2\n18O reveals a revised IAA degradation \npathway in V. paradoxus CL014. a , LC-MS analysis of [²H ₇ ]IAA catabolism identifies \nkey pathway intermediates and their deuterium labeling profiles. Left, LC–MS peak \nintensities of major metabolites. Right, deuterium isotope distributions following \nincubation with unlabeled IAA or [²H₇ ]IAA. Data represent mean ± s.e.m. (standard error \nof the mean) of n  = 3 biological replicates.  b, H₂ ¹/i2 O (20% v/v) tracing confirms the \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n13 \n \nsource of oxygen atoms incorporated during the two-step oxidation reactions. Data are \nshown for two biological replicates. c , A revised mechanistic model of IAA degradation \nconsistent with isotope tracing. \n \nThe structurally and functionally divergent oxygenase complex IadCDE reverses \nbacteria-induced root growth inhibition \nOxygenases play a crucial role in aerobic bacteria by incorporating oxygen into \nchemically stable aromatic compounds, enabling their degradation into metabolically \naccessible substrates\n14,21. To date, the most extensively studied indole oxygenases fall \ninto two major families: the tryptophan dioxygenase (TDO) superfamily, which catalyzes \ndioxygenation or monooxygenation via an Fe/heme-dependent mechanism\n18,19,22; and \nthe flavin-dependent monooxygenases (FMOs), which use a flavin cofactor to mediate \noxygen transfer\n23–25. Rieske-type dioxygenases may function divergently as \ndioxygenase, monooxygenase, or hydroxylase 15–17. Our metabolomics and isotope \ntracing reveal a monooxygenation-driven mechanism, identifying IadCDE in V. \nparadoxus CL014 as a two-component indole monooxygenase, essential for initiating \nIAA degradation, rather than a dioxygenase as previously annotated 9,12. The proposed \nreaction mechanism closely resembles that of MarE, a heme-dependent aromatic \nmonooxygenase involved in the maremycin biosynthetic pathway of Streptomyces sp. \nB9173\n17–19,22. While MarE belongs to the TDO superfamily and relies on a heme \ncofactor17,18, IadDE is structurally similar to Rieske non-heme dioxygenases (Fig. 3a, b) \n12,15.  \nTo investigate the molecular basis of IadCDE function, we determined the crystal \nstructure of the IadDE complex (PDB code: 9O71) from V. paradoxus CL014 at 1.28\n/i2 Å \nresolution, providing a high-resolution view of the complete holoenzyme. While a \nhomologous structure was previously resolved by cryo-EM at 1.8\n/i2 Å resolution (with \n98.4% and 100% protein sequence identity for IadD and IadE, respectively) 12 the \nimproved resolution of our X-ray structure allows for more precise visualization of \noverall structural features. Structurally, IadDE adopts the canonical fold of Rieske-type \nnon-heme dioxygenases\n26. IadD and IadE assemble into a heterodimeric α₃ β₃  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n14 \n \nquaternary complex with threefold symmetry and a characteristic mushroom-shaped \nmorphology (Fig. 3b)12. In the IadCDE complex, IadD contains both a Rieske-type [2Fe–\n2S] cluster and a mononuclear iron center and serves as the catalytic subunit, while \nIadE provides structural support. Trimerization of IadDE heterodimers brings the Rieske \niron-sulfur cluster of IadD close to the iron center of an adjacent IadD, potentially \nfacilitating electron transfer during catalysis (Fig. 3b). IadC is annotated as a reductase \nand is proposed to mediate electron transfer from NADH to IadD to activate molecular \noxygen\n12. \nPhylogenetic analysis revealed that IadD forms a distinct subclade within the Rieske \ndioxygenase and indole oxygenase family, showing close evolutionary relationships to \nthe phthalate family of Rieske non-heme dioxygenases\n27, as well as to members of the \nTDO and FMO superfamilies. This positioning may reflect shared functional features \namong distinct oxygenase lineages (Extended Data Fig. 5). \nTo determine whether iadCDE alone is sufficient to reverse RGI in the context of \ncomplex microbial communities, we genomically integrated the native repressor \nmarR739 alongside either the minimal 9 (iadCDE) or full ( iadC-K2) degradation pathway \ninto the Variovorax Δ HS33 strain. This enabled direct functional comparison of the \noxygenase module with the complete pathway under IAA-rich conditions. In M9 minimal \nmedium supplemented with glucose and 0.1 mg/mL IAA, both engineered strains \nshowed significantly enhanced IAA degradation relative to the wild type, independent of \nbacterial growth (Fig. 3c, Extended Data Fig. 6a). In a gnotobiotic system, engineered \nstrains fully rescued RGI in Arabidopsis seedlings challenged with 100 nM exogenous \nIAA, the auxin-producing strain Arthrobacter  CL028, or a 32-member synthetic \ncommunity (SynCom32, Supplementary Table 1), with comparable effects observed in \nMedicago (Fig. 3d, e). These results demonstrate that iadCDE alone is sufficient to \nreverse RGI caused by both exogenous IAA and auxin-producing microbes. However, \nunder elevated IAA concentrations (1 \nμ M or 10 μ M), the strain harboring the full \ndegradation pathway outperformed the minimal mutant, suggesting that downstream \nenzymatic steps confer an advantage in detoxifying excess IAA (Fig. 3f).   \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n15 \n \n \nFig. 3 | Structural and genetic evidence show that IadCDE enables auxin \ndegradation and alleviates root growth inhibition. a , Crystal structure of the IadDE \ncomplex from V. paradoxus  CL014 resolved at 1.28 Å (PDB code: 9O71). b, A \nheterohexameric IadDE complex is formed by trimerization of IadDE heterodimers. The \nRieske-type [2Fe–2S] cluster and mononuclear iron within each IadD subunit are shown \nin matching colors. Trimer assembly reduces the spatial distance between the electron \ndonor (Rieske cluster) and acceptor (iron center), potentially enhancing catalytic \nelectron transfer efficiency. c, in vitro  IAA degradation by engineered strains. IAA \nconcentrations were measured from culture supernatants of strains grown in M9 \nminimal medium supplemented with glucose and 0.1 mg/mL IAA. The Salkowski \nreagent reacts with IAA to generate a pink-to-red chromophore with an absorption \nmaximum at 530 nm (OD\n₅₃₀ ). Data represent the mean of two biological replicates per \nsample. d–f, Primary root length of Arabidopsis seedlings exposed to 100 nM IAA or the \nauxin-producing strain Arthrobacter CL028 ( d), A 32-member synthetic community \n(SynCom32) in Arabidopsis  and Medicago  seedlings ( e), and elevated IAA \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n16 \n \nconcentrations (1 μ M and 10 μ M) in Arabidopsis ( f). Letters above boxplots indicate \nstatistically significant differences as determined by one-way ANOVA with Tukey’s post \nhoc test (P < 0.05). Groups not sharing the same letter differ significantly. “NB” indicates \nthe no-bacteria control. Sample sizes (left to right): d, n = 23, 29, 31, 31, 27, 34, 26, 34, \n26, 25, 31, 16, 21, 21, 25; e, n = 64, 81, 56, 52, 53, 43, 52, 78, 52, 71, 47, 48; f, n = 19, \n15, 18, 14, 9, 19, 17, 19, 19, 14. Box plots represent the median (center line), \ninterquartile range (box), and 1.5× interquartile range (whiskers).  \n \nEngineering of iad genes into two distinct root- associated bacteria confers IAA-\ndegrading activity and promotes auxin homeostasis  \nTo evaluate whether root-associated bacteria can acquire IAA-degrading activity \nthrough genetic engineering, we introduced iad genes into two root-associated strains: \nPolaromonas MF047 and Paraburkholderia MF37628. Both strains, like Variovorax, are \nBetaproteobacteria and do not impair host growth when co-inoculated with Arabidopsis  \nseedlings (Fig. 4b). Neither strain produced IAA when cultured in M9 minimal medium \nsupplemented with tryptophan and glucose, nor were they able to reverse RGI triggered \nby exogenous IAA or by the auxin-producing strain Arthrobacter CL028 (Fig. 4b, \nExtended Data Fig. 6d). Additionally, as an amino acid auxotroph, Polaromonas MF047 \nexhibited no growth in M9 minimal medium unless supplemented with amino acids \n(Extended Data Fig.\n/i2 6e). \nTo confer IAA-degrading capacity, we genomically integrated marR73 together with \neither the minimal (iadCDE) or complete (iadC-K2) pathway into three intergenic sites in \nPolaromonas MF047 and one site in Paraburkholderia MF376 (Extended Data Fig. 6b). \nTo minimize potential disruption of native gene expression, insertion sites were located \nwithin non-coding intergenic regions ranging from 400 to 900 bp in length. All \nengineered strains degraded IAA within 24 hours, and Paraburkholderia MF376 knock-\nins achieved complete IAA removal within four hours (Fig. 4a). This rapid degradation \nalleviated RGI in Arabidopsis seedlings exposed to 100 nM exogenous IAA (Fig. 4b). \nGrowth curves in 50% TSB medium revealed no significant differences between wild-\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n17 \n \ntype and engineered strains, indicating that chromosomal integration did not impair \nbacterial fitness (Extended Data Fig. 6a). \nTo assess their function in plant-microbe interactions, we co-inoculated engineered \nstrains with Arthrobacter CL028 or SynCom32 on 7-day-old Arabidopsis seedlings in a \ngnotobiotic system. After 7 days, all engineered strains fully rescued RGI triggered by \nArthrobacter CL028, restoring primary root length to control levels (Fig. 4b). In parallel, \nthey significantly alleviated RGI induced by SynCom32, with Paraburkholderia MF376 \nconsistently showing greater efficacy than Polaromonas MF047 (Fig. 4c). To test cross-\nspecies efficacy, we repeated the SynCom32 experiment using Medicago seedlings. In \nthis context, strains carrying the full iad pathway promoted stronger root growth than \nthose expressing only the minimal iadCDE module, highlighting the importance of a \ncomplete IAA degradation pathway for effective function in complex microbial \nenvironments (Fig. 4c). These results demonstrate for the first time that stable genomic \nintegration and expression of iad genes in root-associated bacteria is able to restore \nroot growth under microbial auxin stress. The engineered strains are effective across \nmicrobial contexts and plant hosts, underscoring their potential as tools for microbiome-\nbased modulation of plant development. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n18 \n \n \nFig. 4 | Engineering root-associated bacteria with iad genes enables IAA \ndegradation and alleviates root growth inhibition (RGI). a, Engineered Polaromonas \nMF047 strains cultured in 50% TSB medium and Paraburkholderia MF376 strains \ncultured in M9 medium efficiently degraded IAA. IAA levels were quantified from culture \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n19 \n \nsupernatants using the Salkowski reagent, which forms a pink-to-red chromophore upon \nreaction with IAA and is measured at OD ₅₃₀ . Data represent the mean of two biological \nreplicates per sample. b–c, Engineered chromosomal knock-in strains reversed RGI, as \nindicated by enhanced primary root elongation in Arabidopsis seedlings treated with 100 \nnM IAA or the auxin-producing strain Arthrobacter CL028 ( b), and in Arabidopsis  and \nMedicago seedlings exposed to a 32-member synthetic bacterial community \n(SynCom32) (c). Letters above boxplots indicate statistically significant differences as \ndetermined by one-way ANOVA with Tukey’s post hoc test ( P < 0.05). Groups not \nsharing the same letter differ significantly. “NB” denotes the no-bacteria control. Sample \nsizes (left to right): b, n = 23, 42, 37, 31, 34, 26, 30, 27, 31, 26, 25, 51, 42, 44, 33, 39, \n52, 49, 42, 26, 28, 35, 31, 21, 15, 21, 21, 23, 21, 17, 22, 22, 29; c, n = 64, 81, 45, 41, 33, \n28, 61, 36, 17, 56, 75, 89, 52, 78, 76, 58, 67, 74, 57, 68. Box plots show the median \n(horizontal line), interquartile range (boxes), and whiskers extending to 1.5× the \ninterquartile range. \n \nEngineered Paraburkholderia MF376 reverses RGI that V. paradoxus  CL014 \ncannot fully rescue \nTo assess the contribution of the iad pathway to RGI mitigation, we selected 23 \npreviously identified RGI-inducing strains from nine bacterial genera (Supplementary \nTable 1)5. Each strain was co-inoculated with either wild-type V. paradoxus CL014 or \nthe iad deletion mutant Δ HS33 on 7-day-old Arabidopsis  seedlings. After 7 days, root \nlength measurements revealed that 22 strains induced RGI (Fig. 5a, b; Extended Data \nFig. 7). Among these, RGI caused by 16 strains was reversed by wild-type V. \nparadoxus CL014, and in 12 of these cases (12/16), the wild type restored root growth \nmore effectively than the iad-deficient mutant \nΔ HS33, indicating an iad-dependent \nmitigation mechanism (Extended Data Fig. 7). In contrast, for Agrobacterium MF224, \nArthrobacter MF135, and Pseudomonas MF051, both wild-type and Δ HS33 strains \nrestored root growth to similar levels, suggesting an iad-independent mechanism (Fig. \n5b).  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n20 \n \nGenome analysis of V. paradoxus CL014 revealed a gene encoding 1-\naminocyclopropane-1-carboxylic acid (ACC) deaminase, an enzyme that degrades ACC, \nthe ethylene precursor29. Ethylene, a plant hormone, can inhibit primary root elongation \nat high levels 30. Prior studies show that RGI induced by Arthrobacter CL028 and a \nsynthetic microbial community requires both auxin and ethylene signaling in the host 5.  \nWhile plants produce ethylene endogenously, certain soil and rhizosphere microbes \nalso contribute to ethylene levels\n29. Notably, some Agrobacterium, Arthrobacter, and \nPseudomonas strains have been reported to produce ethylene, which can suppress root \nelongation and alter root architecture 31–33. Based on these observations, we \nhypothesized that the three iad-independent strains may produce both auxin and \nethylene, with IAA degradation mediated by the iad pathway and ethylene detoxification \npotentially supported by ACC deaminase activity in V. paradoxus  CL014. However, \ndeletion of the ACC deaminase gene in both wild-type and Δ HS33 backgrounds did not \nimpair their ability to reverse RGI, ruling out ethylene degradation through this ACC \ndeaminase as the primary mechanism (Fig. 5b). These findings suggest that while the \niad pathway is a central mechanism for suppressing RGI caused by many root-\nassociated strains, V. paradoxus  CL014 also employs iad -independent strategies to \nmitigate RGI from certain microbes. \nDespite its broad effectiveness, V. paradoxus CL014 was unable to fully rescue RGI \ninduced by Pseudomonas MF048. To test whether iad-engineered strains could provide \nenhanced reversion, we evaluated Polaromonas  MF047 and Paraburkholderia MF376 \nengineered with the iad pathway. These strains were co-inoculated with Pseudomonas \nMF048 alone or in combination with Arthrobacter CL028 on Arabidopsis seedlings. Root \nlength analysis showed that Paraburkholderia MF376 knock-in strains significantly \noutperformed both V. paradoxus  CL014 and engineered Polaromonas  MF047 under \nboth conditions (Fig. 5c). Collectively, these results demonstrate that engineering \nParaburkholderia MF376 with the iad  pathway enhances its ability to mitigate complex \nRGI interactions, highlighting its potential as a robust and versatile chassis for \npromoting plant growth in diverse microbial environments. \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n21 \n \nFig. 5 | Root growth inhibition by certain strains is mitigated independently of the \niad locus, while engineered Paraburkholderia MF376 enhances mitigation of iad-\ndependent inhibition. a , Heatmap showing average primary root lengths of \nArabidopsis seedlings inoculated with 23 previously identified RGI-inducing strains 5, \neither alone (self) or co-inoculated with V. paradoxus  CL014 wild type or the iad-\ndeficient mutant Δ HS33, to assess iad-dependent reversion. Blue squares indicate \nsignificant differences compared to the no-bacteria (NB) control demonstrating RGI \nphenotypes; black squares indicate significant differences compared to the strain \ninoculated alone (self). Corresponding box plots are shown in Extended Data Fig. 7.  b, \nPrimary root length of Arabidopsis seedlings co-inoculated with iad -independent RGI-\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n22 \n \ninducing strains and treated with V. paradoxus CL014 wild type, Δ HS33, Δ acc (ACC \ndeaminase gene deleted), or Δ HS33 Δ acc (both iad  and ACC deaminase genes \ndeleted). c, Primary root lengths of Arabidopsis seedlings inoculated with Pseudomonas \nMF048 alone or with Arthrobacter CL028, and co-inoculated with V. paradoxus CL014, \nor iad-engineered strains of Polaromonas MF047 or Paraburkholderia MF376. Letters \nabove boxplots indicate statistically significant differences as determined by one-way \nANOVA with Tukey’s post hoc test ( P < 0.05). Groups not sharing the same letter differ \nsignificantly. “NB” denotes the no-bacteria control. Sample sizes (left to right): b, n = 40, \n56, 48, 41, 48, 49, 14, 17, 25, 25, 21, 16, 28, 17, 15, 14, 15, 15; c, n = 42, 41, 41, 17, 25, \n17, 29, 27, 29, 25, 26, 28, 22, 19, 30, 21, 30, 24, 25. Box plots show the median (center \nline), interquartile range (boxes), and whiskers extending to 1.5× the interquartile range.  \n \nImpact of V. paradoxus CL014 and engineered strains on the root microbiome \nand plant growth \nV. paradoxus CL014 can degrade IAA without harming plant growth at normal treatment \nlevels (OD ₆₀₀  = 0.05). To test whether this effect holds under high bacterial load, we \napplied V. paradoxus CL014 at OD ₆₀₀  = 2 to Arabidopsis  seedlings in a gnotobiotic \nsystem. No significant changes in primary root length were observed, indicating that \neven at high densities, V. paradoxus CL014 does not adversely affect plant growth or \ndisturb auxin balance (Fig. 6a). These findings suggest that its IAA-degrading function \nmay primarily influence microbe–microbe interactions rather than directly altering plant \ndevelopment. \nTo assess the impact of V. paradoxus CL014, Polaromonas MF047, Paraburkholderia \nMF376, and their iad-gene-containing engineered strains on the root microbiome, we \ntreated Arabidopsis and Medicago seedlings with SynCom32 (Supplementary Table 1) \nin the presence or absence of each strain. After 9 days, root-associated communities \nwere profiled by 16S rRNA amplicon sequencing. Despite receiving the same \nSynCom32 inoculum, Arabidopsis and Medicago developed distinct root microbiomes, \nhighlighting the influence of host genotype (Fig. 6b). Among the three introduced genera, \nParaburkholderia exhibited the highest relative abundance in roots (50.84%), followed \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n23 \n \nby Variovorax (3.40%), whereas Polaromonas (0.01%) showed minimal colonization \n(Fig. 6b, and Supplementary Table 6). Due to its robust root colonization, \nParaburkholderia MF376—both wild-type and engineered—substantially altered root \nmicrobiome composition, explaining 52% and 33.8% of the community variation in \nArabidopsis and Medicago, respectively, as shown by unconstrained principal \ncoordinate analysis (PCoA) using Bray–Curtis distances (P = 0.001; Fig. 6c). In contrast, \nV. paradoxus CL014 and Polaromonas MF047 had more modest effects (17.1-28.3%, P \n= 0.05), consistent with their lower colonization (Fig. 6b, Extended Data Fig. 8 and \nSupplementary Table 6). \nNotably, no major differences were observed between wild-type and knock-in strains of \nV. paradoxus  CL014, Paraburkholderia  MF376 and Polaromonas MF047, suggesting \nthat IAA degradation pathway did not markedly alter their interactions within the \nmicrobiome (Fig. 6c, Extended Data Fig. 8). Log\n₂  fold-change analysis revealed no \nsignificant differences in the relative abundance of any genera between wild-type and \nengineered strain inoculated conditions, further suggesting that microbiome shifts were \nprimarily driven by colonization capacity rather than IAA degradation activity (Extended \nData Fig. 9). \nTo evaluate potential benefits for plant performance, we inoculated untreated natural \nsoil with each strain and measured fresh shoot weight after 33 days. Arabidopsis grown \nin soil supplemented with Paraburkholderia  MF376 carrying the complete iad pathway \n(marR73 iadC-K2) exhibited a significant increase in shoot biomass compared to \nuninoculated controls and those treated with the wild-type strain (Fig. 6d, Extended \nData Fig. 10, and Supplementary Table 7). These findings indicate that combining a \nstrong plant colonizer chassis strain with the full IAA degradation pathway can enhance \nplant growth in natural soil, offering a promising strategy for microbiome-based \nagricultural interventions. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n24 \n \nFig. 6 | Engineered strains do not alter root microbiome composition but promote \nArabidopsis growth. a , Primary root length of Arabidopsis seedlings treated with V. \nparadoxus CL014 at OD ₆₀₀  = 2. Statistical significance was assessed using Welch’s \ntwo-tailed t-test ( n = 19, 16).  b , Relative abundance of bacterial genera in the root \nmicrobiomes of Arabidopsis and Medicago seedlings treated with SynCom32 in \ncombination with wild-type or iad -engineered strains of V. paradoxus  CL014, \nPolaromonas MF047, and Paraburkholderia MF376. Sample sizes are indicated above \neach bar.  c , Unconstrained principal coordinate analysis (PCoA) of Bray–Curtis \ndissimilarity showing root microbiome profiles in Arabidopsis and Medicago seedlings \ntreated with SynCom32 alone or with engineered Paraburkholderia MF376 strains. \nEllipses represent 68% confidence intervals. Statistical significance was determined by \nPERMANOVA (Adonis2). d, Shoot fresh weight of 33-day-old Arabidopsis plants grown \nin untreated natural soil inoculated individually with Paraburkholderia MF376 wild type \nand marR73 iadC-K2 knock-in strains. Statistical significance was determined using \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n25 \n \none-way ANOVA with Tukey’s post hoc test (n = 36, 32, 31). Data shown in panels a–d \nare derived from two independent experiments. \n \nDiscussion \nMembers of the genus Variovorax play a critical role in plant–microbe interactions by \nmitigating RGI induced by auxin-producing bacteria, establishing them as key plant-\nbeneficial microbes 5,8,34,35. Building on this foundation, we used bacterial genetics, \nmetabolomics, and isotope tracing to dissect the iad-mediated IAA degradation pathway \nin V. paradoxus CL014. We illuminated the iad pathway biochemistry and defined the \niad genes associated with each pathway step. We then used this knowledge to evaluate \nthe functional transferability of the iad  pathway to other root-associated bacteria and \nassess the effects of these engineered strains on microbiome composition and plant \nphenotype. \nA previous study using [¹³C ₆ ]IAA (benzene ring-labeled) proposed that V. paradoxus  \nCL014 degrades IAA via a pathway similar to that of B. japonicum , in which IAA is \nsequentially processed through 2-hydroxyindole-3-acetic acid, dioxindole-3-acetic acid \n(DOAA), isatin, isatinic acid, and anthranilic acid 9,13. However, since key structural \nmodifications occur at the pyrrole ring and acetic acid side chain, we used [²H ₇ ]IAA to \ntrack these changes with greater resolution. Surprisingly, our isotope tracing revealed a \ndistinct degradation route (Fig. 2c). Rather than proceeding through the 2-\nhydroxyindole-3-acetic acid or 2-oxindole-3-acetic acid pathway, IAA was metabolized \nvia a likely epoxide mechanism followed by hydration, and a subsequent \ndehydrogenation, ultimately leading to the production of the final product, anthranilic \nacid (Fig. 2c). This finding redefines the function of IadDE: although annotated as, and \nmost structurally similar to Rieske non-heme dioxygenases (Fig. 3a, b), it functions as a \nmonooxygenase (Fig. 2). While classical Rieske dioxygenases typically catalyze \ndioxygenation reactions, exceptions such as naphthalene dioxygenase have been \nshown to perform monooxygenation via radical intermediates\n15. However, indole \noxidation has rarely been attributed to Rieske-type enzymes and is more commonly \nassociated with the tryptophan 2,3-dioxygenase (TDO) superfamily, which features a \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n26 \n \nheme cofactor coordinated by a conserved histidine and a distinct core architecture 22,36. \nDespite their structural divergence, the catalytic mechanism of IadCDE parallels that of \nMarE, a heme-dependent monooxygenase in the TDO superfamily that catalyzes 2-\noxoindole formation during maremycin biosynthesis in Streptomyces (Extended Data \nFig. 5)\n17,18. While some TDO enzymes catalyze dioxygenation via two consecutive \nmonooxygenation steps, others—including MarE, SfmD, and TyrH—mediate single \noxygen-atom transfer, with the second oxygen likely reduced to water in the presence of \nan electron donor such as ascorbate\n17–19,22,37. Given the central role of oxygenases in \nbacterial aromatic compound degradation 14, our findings provide new mechanistic \ninsights into O ₂ -dependent ring cleavage strategies that overcome the inherent \nchemical stability of aromatic substrates. This revised pathway highlights the functional \ndiversity of oxygenases and suggests that IadDE-mediated degradation of indole \ncompounds more closely resembles the activity of heme-dependent monooxygenases \nthan classical Rieske-type dioxygenases, representing a distinct biochemical route for \naromatic catabolism (Extended Data Fig. 5). Notably, pathways for direct conversion of \nindoles to oxindoles have garnered increasing attention due to their relevance to various \npathogenic processes in humans and the multipotent therapeutic value of oxindole \npharmacophores\n16. Our study expands the understanding of microbial oxygenase \ndiversity and provides a valuable link between bacterial aromatic catabolism and \nbroader biological processes involving indole monooxygenation. \nIadC and IadDE form a two-component oxygenase system, with IadC transferring \nelectrons from NADH to IadDE to activate molecular oxygen 12,26. Although IadC \nenhances catalytic efficiency 9, it is dispensable in V. paradoxus CL014, likely due to \nfunctional redundancy with endogenous reductases. This is supported by the \nobservation that overexpression of iadDE alone in the \nΔ HS33 background restores IAA \ndegradation9. Moreover, IadDE activity is strain-dependent when expressed \nheterologously; it functions in E. coli 10 Beta but not in BL21(DE3), likely due to host-\nspecific differences in redox environments 9,12. Analogous redundancy has been \nreported in Sphingopyxis granuli , where ThnA4, a ferredoxin reductase, is dispensable \nfor tetralin degradation38. While iadC is not required for IAA degradation in V. paradoxus \nCL014, it is essential for RGI mitigation. In Arabidopsis, only Δ HS33 strains expressing \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n27 \n \niadCDE, but not iadDE  alone, reversed RGI induced by the auxin-producing \nArthrobacter CL0289, indicating that iadC is essential for functional rescue.  \nTo test broader applicability, we engineered Polaromonas MF047 and Paraburkholderia \nMF376 with the same two versions of the iad pathway, selecting these chassis strains \nfor their root-association and potential plant-beneficial traits. All engineered strains \ndegraded IAA and alleviated RGI induced by exogenous IAA, Arthrobacter CL028, and \nSynCom32 (Fig. 4). Notably, Paraburkholderia MF376 consistently outperformed \nPolaromonas MF047 across all tested conditions and reversed RGI caused by \nPseudomonas MF048, where V. paradoxus  CL014 is ineffective (Fig. 5c). 16S rRNA \namplicon sequencing revealed that integration of the iad pathway into any of these \nstrains (Δ HS33, MF047, or MF376) did not substantially alter root community structure \ncompared to the respective wild type (Fig. 6b). Instead, colonization capacity emerged \nas the major driver of microbiome shifts (Fig. 6b, c; Extended Data Fig. 8). Among all \ntested strains, Paraburkholderia MF376 strains exhibited the highest colonization \nefficiency whether wild type or engineered (Fig. 6b). And, in natural soil with a native \nmicrobiome, the engineered Paraburkholderia MF376 strain significantly enhanced plant \ngrowth, increasing shoot biomass relative to uninoculated controls and the wild type (Fig. \n6d, Extended Data Fig. 10). This beneficial effect likely reflects its superior root \ncolonization ability and the efficient IAA-degrading activity introduced through our \nengineering. Together, our findings underscore the importance of chassis strain \nselection when engineering IAA-degradation into new bacteria and demonstrate that \ncombining heterologous expression of the iad genes with strong root colonization \nappears best for promoting rhizosphere auxin homeostasis and plant growth. This study \nadvances our mechanistic understanding of microbial auxin metabolism and provides a \nplatform for designing next-generation microbial bioinoculants to modulate plant-\nmicrobe interactions and enhance crop productivity in diverse soil environments. \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n28 \n \nMethods \nEngineering bacterial strains \nKnock-out mutant construction. Gene deletions in V. paradoxus  CL014 were \ngenerated using the suicide vector pMo130, following previously established \nmethods\n5,9,39.  The vector backbone was PCR amplified and treated with DpnI to \nremove the template DNA. Primers used for mutant construction are listed in \nSupplementary Table 2. Upstream and downstream flanking regions of target genes \nwere amplified from V. paradoxus CL014 genomic DNA using Platinum SuperFi II PCR \nMaster Mix (Thermo Fisher Scientific). PCR products were purified using the DNA \nClean & Concentrator Kit (Zymo Research) and assembled into pMo130 using HiFi \nGibson Assembly Master Mix (New England Biolabs). Assembled plasmids were \ntransformed into E. coli  NEB 5-alpha (New England Biolabs), selected on Lysogeny \nBroth (LB, Thermo Fisher Scientific, BP1427) agar (2% w/v) supplemented with \nkanamycin (50 µg/mL), and verified by Sanger sequencing (Genewiz). Sequence-\nconfirmed plasmids were transferred into the diaminopimelic acid (DAP) auxotrophic E. \ncoli WM3064 for conjugation. Transformants were selected on LB agar containing \nkanamycin (50 µg/mL) and DAP (0.3 mM), and grown at 37\n/i2 °C for 24 h. For biparental \nmating, donor E. coli WM3064 and recipient V. paradoxus CL014 (pre-grown in 50% \nTSB (Tryptic Soy Broth, Thermo Fisher Scientific, CM0129) agar (2% w/v) with \nampicillin (100 µg/mL)) were washed twice with 50% TSB, mixed at a 1:1 volume ratio, \npelleted (5,000 × g, 5 min), resuspended in 1/10 volume of 50% TSB, and spotted onto \n50% TSB agar containing DAP (0.3 mM). After overnight incubation at 28\n/i2 °C, \nexconjugants were selected on 50% TSB agar containing ampi cillin and k anamycin \nwithout DAP, and incubated for 3–4 days. Resulting colonies were re-streaked onto \nfresh antibiotic 50% TSB agar to ensure clonality and remove residual donor cells. \nSingle colonies were screened by colony PCR to confirm single-crossover integration. \nVerified integrants were cultured in 50% TSB with ampicillin and Isopropyl \nβ -D-1-\nthiogalactopyranoside (IPTG, 1 mM). Cultures were plated on sucrose counter-selection \nagar (10 g/L tryptone, 5 g/L yeast extract, 10% w/v sucrose, 2% w/v agar, 100 µg/mL \nampicillin, 1 mM IPTG) and incubated at 28\n/i2 °C for 3-4 days. Resulting colonies were \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n29 \n \npassaged in the same liquid medium, and deletion events were verified by PCR. Final \ndeletion strains were re-streaked on 50%  TSB with ampicillin and confirmed by \ndiagnostic PCR using one primer located outside the deletion region and one within the \ndeleted gene to ensure complete excision and strain purity. DNA templates for all PCRs \nwere prepared using the following rapid lysis protocol\n40. A single colony or 6 µl of \nbacterial culture was mixed with 10 µl of alkaline lysis buffer (25 mM NaOH, 0.2 mM \nNa₂ -EDTA, pH 12), incubated at 95 /i2 °C for 30 min, and neutralized with 10 µl Tris-HCl \n(40 mM, pH 7.5). The resulting material was used directly as PCR template (1:10, \ntemplate:total PCR reaction volume).  \nOverexpression mutant construction. V. paradoxus  CL014 marR73  iadCDE and \nmarR73 iadC-K2 were cloned into the broad-host-range vector pBBR1MCS-2 as \npreviously described5,9,41, genomic fragments were amplified using Platinum SuperFi II \nPCR Master Mix and assembled into pBBR1MCS-2 via Gibson assembly using HiFi \nGibson Assembly Master Mix (New England Biolabs). Primers used for mutant \nconstruction are listed in Supplementary Table 2. Circular template DNA was digested \nwith DpnI, and assembled plasmids were transformed into E. coli  NEB 10-beta (New \nEngland Biolabs). Transformants were selected on LB agar containing kanamycin (50 \nµg/mL), and plasmids were extracted (ZR Plasmid Miniprep Kit, Zymo Research) and \nverified by Sanger sequencing.  Verified constructs were introduced into V. paradoxus  \nCL014 \nΔ HS33 by tri-parental mating, using E. coli pRK2013 as a helper strain. Donor \nand helper strains were grown in LB with kanamycin (50 µg/mL) at 37 /i2 °C, and the \nrecipient strain was cultured in 50% TSB with ampicillin (100 µg/mL) at 28 /i2 °C. All \nstrains were pelleted (5,000 × g, 5 min), washed in 50% TSB twice, mixed at equal \nvolumes, and spotted onto 50% TSB agar for overnight conjugation at 28 /i2 °C. \nExconjugants were selected on 50% TSB agar supplemented with kanamycin (50 \nµg/mL) and ampicillin (100 µg/mL), confirming successful plasmid transfer into V. \nparadoxus CL014 \nΔ HS33. \nKnock-in mutant construction. Two gene combinations, marR73 iadCDE and marR73 \niadC-K2, were integrated into the iad locus of V. paradoxus CL014 Δ HS33, as well as \nthree intergenic genomic sites in Polaromonas MF047 (IMG genome ID 2636416056 \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n30 \n \nwith insertion positions between Gene IDs: 2639079279–80, 2639079819–20, and \n2639080354–55) and one site in Paraburkholderia MF376 (IMG genome ID  \n2521172625 with insertion position between Gene IDs 2521671121–22), using the \npMo130 suicide vector (Extended Data Fig. 6b). Gene amplification, vector assembly, \nand verification followed the same procedure as knockout mutant construction, except \nthat plasmids were initially propagated in E. coli NEB 10-beta. Primers used for mutant \nconstruction are listed in Supplementary Table 2. Sequence-verified plasmids were \nelectroporated into V. paradoxus  CL014 \nΔ HS33, Polaromonas MF047, and \nParaburkholderia MF376, as described below. Strains were cultured in 50% TSB at \n28\n/i2 °C with shaking (250 rpm) for 2 days, followed by 24 h incubation at 4 /i2 °C. Cells \nwere harvested (5,000 × g, 10 min, 4/i2 °C), washed twice with ice-cold sterile water, and \nresuspended in 10% sterile glycerol for electroporation. For each reaction, 100 ng of \nplasmid DNA was electroporated into 100 µl of competent cells using a 0.1 cm gap \ncuvette using the following conditions: 1,800 V, 25 µF, 200 \nΩ  for marR73 iadCDE, and \n2,500 V, 25 µF, 200 Ω  for marR73 iadC-K2. After electroporation, cells were recovered \nin SOC medium (New England Biolabs, B9020) at 28 /i2 °C (250 rpm) for 3 h and plated \non 50% TSB agar with selective antibiotics: ampicillin (100 µg/mL) + kanamycin (200 \nµg/mL) for Polaromonas MF047 and kanamycin (50 µg/mL) for Paraburkholderia \nMF376. After 4–5 days of incubation, single colonies were screened by colony PCR \nusing crude DNA extracted with alkaline lysis buffer. PCR-confirmed integrants were \ngrown overnight in 50% TSB with 1 mM IPTG, supplemented with ampicillin (100 µg/mL) \nfor Polaromonas MF047 and without antibiotics for Paraburkholderia MF376. Cultures \nwere then diluted 1,000-fold and plated on sucrose counter-selection agar to induce \nsecond recombination (5% w/v sucrose for Polaromonas MF047; 20% w/v sucrose for \nParaburkholderia MF376). Double-crossover mutants were confirmed by PCR and \nSanger sequencing.  \nLiquid chromatography -mass spectroscopy (LC–MS) metabolomics  \nSample preparation. V. paradoxus strains were streaked from glycerol stocks onto 50% \nTSB agar plates supplemented with appropriate antibiotics: ampicillin (100 µg/mL) for \nknockout mutants and wildtype, and ampicillin (100 µg/mL) plus kanamycin (50 µg/mL) \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n31 \n \nfor overexpression mutants. Plates were incubated at 28/i2 °C for 3 days. Single colonies \nwere inoculated into 5 mL of 50% TSB containing the appropriate antibiotics and \ncultured at 28 /i2 °C with shaking at 250 rpm for 48 h. Bacterial cultures were harvested \nand washed following the root growth inhibition assay protocol, then resuspended in 5 \nmL of modified M9 medium supplemented with 15 mM succinic acid to a final OD ₆₀₀  of \n0.05. Cultures were incubated at 28 /i2 °C, 250 rpm for 15 h, after which IAA or \ndeuterium-labeled IAA ([ 2H7]IAA, DLM-8040-0.1, Cambridge Isotope Laboratories) was \nadded to a final concentration of 0.1 mg/mL. Cultures were incubated for an additional 4 \nh. Cells were collected at a total biomass of OD\n₆₀₀  × volume (mL) = 2, centrifuged at \n5,000 × g for 10 min, and pellets were resuspended in 400 µl of cold quenching solvent \n(acetonitrile : methanol : water, 40:40:20, v/v/v). Samples were stored at −80 /i2 °C prior \nto metabolite extraction and LC–MS/MS analysis. \nLC-MS analysis. LC-MS analysis was performed on a Vanquish UHPLC system \n(Thermo Fisher Scientific) coupled to a quadrupole Orbitrap Exploris 480 mass \nspectrometer.  LC separation of polar metabolites was achieved using a Waters \nXBridge BEH Amide column (2.1 mm × 150 mm, 2.5-µm particle size, 130-Å pore size). \nThe LC method has a 25-min solvent gradient at a flow rate of 150 µL/min, with the \nfollowing gradient parameters: 0 min, 90% B; 2 min, 90% B; 3 min, 75%; 7 min, 75% B; \n8 min, 70%, 9 min, 70% B; 10 min, 50% B; 12 min, 50% B; 13 min, 25% B; 14 min, 25% \nB; 16 min, 0% B, 20.5 min, 0% B; 21 min, 90% B; 25 min, 90% B, where Solvent A was \n95:5 water : acetonitrile with 20 mM ammonium hydroxide and 20 mM ammonium \nacetate (pH 9.4) and solvent B was acetonitrile. The autosampler temperature was 4 °C, \nthe column temperature was 25 °C, and the injection volume was 10 \nμ l. The Exploris \n480 mass spectrometer was operated in full scan mode in negative polarity on MS1 \nlevel, which allows the relative quantitation of the metabolite across by ion count.  \nFollowing parameters are used for the full scan: resolution, 120,000; scan range, m/z \n70-1000 (negative mode); AGC target, 1e6; ITmax, 500 ms. Other instrument \nparameters are spray voltage 3000 V, sheath gas 35 (Arb), aux gas 10 (Arb), sweep \ngas 0.5 (Arb), ion transfer tube temperature 300 °C, vaporizer temperature 35\n◦ C, \ninternal mass calibration on, RF lens 50. The MS2 spectra were collected in targeted \nmode using the parallel reaction monitoring (PRM) function at higher energy C-trap \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n32 \n \ndissociation (HCD) energy of 20eV, and other instrument settings as following: \nresolution 30,000, AGC target 1e6, maximum injection time 250 /i2 ms, and isolation \nwindow 1.0 /i2 m/z. For the MS1 data analysis, raw LC–MS data were converted to \nmzXML format using ProteoWizard42 Peak picking was performed with EL-Maven \n(v0.12.1-beta; Elucidata) for unlabeled and 2H-labeled compounds, and MAVEN \n(v2.10.14c) for 18O-labeled compounds. Relative abundance changes of each \nmetabolite were quantified using relative peak area tops in the chromatogram. For 2H-\nlabeled data analysis, natural isotope abundance was corrected using the AccuCor R \npackage\n43 (https://github.com/lparsons/accucor). For 18O-labeled data analysis, isotope \ncorrection was performed using the Iso-Autocorr package \n(https://github.com/xxing9703/Iso-Autocorr\n). For the MS2 data, raw LC–MS files were \nprocessed and peaks were extracted with the built-in Xcalibur Qual Browser (Thermo \nScientific, v4.4). \n \nIadDE protein expression, purification and crystallization \nProtein expression. The iadD and iadE genes (IMG gene IDs: 2643613669, \n2643613668) from V. paradoxus CL014 (IMG genome ID: 2643221508) were cloned \ninto the pET28b expression vector with an N-terminal His-tag and transformed into E. \ncoli BL21(DE3) for recombinant protein expression. Transformants were plated on LB \nagar supplemented with kanamycin (50 µg/mL) and chloramphenicol (33 µg/mL) and \nincubated at 37/i2 °C for 24 h. Individual colonies were picked and grown overnight in LB \nmedium at 37/i2 °C with shaking at 250 rpm.  Overnight cultures were diluted 1:1000 into \nautoinduction (AI) medium (ZYM-5052)44 containing the same antibiotics, and incubated \nat 37/i2 °C, 250 rpm for 22 h. Cells were harvested by centrifugation at 5,000 × g for 20 \nmin at 4 /i2 °C, and pellets were collected and stored at –80 /i2 °C for subsequent protein \npurification. \nProtein purification. Cell pellets were resuspended in IMAC Buffer A (20 /i2 mM \nNaH₂ PO₄ , 500/i2 mM NaCl, pH 7.4) at a ratio of 1 /i2 g pellet per 10/i2 mL buffer. To reduce \nviscosity, 0.1/i2 µL benzonase nuclease (Sigma-Aldrich, 70746) was added per gram of \ncell pellet. Cells were lysed using an EmulsiFlex-C5 high-pressure homogenizer (three \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n33 \n \npasses), and the lysate was clarified by centrifugation at 25,000 × g for 30 min at 4/i2 °C. \nThe supernatant was filtered through a 0.22 µm syringe filter and loaded onto a 5 mL \nNi-charged Nuvia IMAC column (Bio-Rad) using a fast protein liquid chromatography \n(FPLC) system (Bio-Rad). The column was washed with IMAC Buffer A containing 15 \nmM imidazole, and bound proteins were eluted with IMAC Buffer B (20 mM NaH\n₂ PO₄ , \n500 mM NaCl, 500 mM imidazole, pH 7.4). Eluted fractions were analyzed by SDS–\nPAGE (Bio-Rad Stain-Free gels), pooled, concentrated, and buffer-exchanged into \n50\n/i2 mM Tris-HCl (pH/i2 7.0), 150/i2 mM NaCl using 10/i2 kDa MWCO concentrators (Pierce, \nThermo Fisher). Purified protein was stored at 4/i2 °C, and concentration was determined \nusing the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). \nProtein crystallography. Purified IadDE was concentrated to 10 mg/mL in buffer \nsupplemented with 1 mM dithiothreitol (DTT) and crystallized using the sitting-drop \nvapor diffusion method at 20 /i2 °C. Crystallization drops were prepared by mixing 1.5 µl \nof protein solution with 1.5 µl of reservoir solution in 24-well sitting-drop plates. The \nreservoir solution contained 9.6–10.6% (w/v) PEG 3350 and 0.1 M sodium citrate \ntribasic dihydrate (pH 5.5). Crystals were cryoprotected by brief soaking in the reservoir \nsolution supplemented with 30% (v/v) ethylene glycol and subsequently flash-cooled in \nliquid nitrogen for data collection.  Diffraction data were collected at beamlines 17-ID1 \n(AMX) and 17-ID2 (FMX) at Brookhaven National Laboratory to a maximum resolution \nof 1.28 Å.  Crystals grew in space group H3 (hexagonal setting of R3) with typical cell \ndimensions a=b=130.5 Å c=100.7 Å \nα =β =90° γ =120° with one complex per asymmetric \nunit.  Data were processed with XDS 45 and scaled with AIMLESS 46.  The structure was \ndetermined by the method of molecular replacement using the program PHASER 47 \nutilizing sequential placement of the two subunits with the models derived from \nAlphaFold models. The structure was rebuilt in COOT 48, incorporating a [2Fe–2S] \ncluster and a mononuclear iron-binding site, and subsequently refined using \nPHENIX.REFINE\n49. The final model showed good agreement with the experimental \ndata and displayed excellent geometry. The final model and associated X-ray data have \nbeen deposited with the Protein Data Bank with code 9O71. Relevant refinement \nstatistics are summarized in Supplementary Table 5. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n34 \n \n \nMeasurement of IAA degradation \nIAA degradation by bacterial strains was assessed using a spike-in approach. Bacterial \npreparation followed the same procedure as the root growth inhibition assay, including \nstreaking from glycerol stocks, cultivation, washing, and OD ₆₀₀  measurement. Washed \ncells were inoculated into M9 medium 5,9 (3 g/L KH ₂ PO₄ , 0.5 g/L NaCl, 6.78 g/L \nNa₂ HPO₄ , and 1 g/L NH ₄ Cl), supplemented with 2 mM MgSO ₄ , 0.1 mM CaCl ₂ , 10 µM \nFeSO₄ , and 5 g/L glucose, to a final OD ₆₀₀  of 0.05. Cultures were incubated at 28 /i2 °C \nwith shaking (250 rpm) for 15 h before IAA was spiked-in to a final concentration of 0.1 \nmg/mL. Aliquots (300 µl) were collected at 2 h or 4 h intervals, centrifuged at 5,000 × g \nfor 10 min, and 50 µl of the supernatant was mixed with 100 µl of freshly prepared \nSalkowski reagent (10 mM FeCl ₃  and 35% perchloric acid). After incubation for 40 min \nat room temperature, absorbance was measured at 530 nm using a BioTek Synergy H1 \nmicroplate reader. As Polaromonas MF047 is an amino acid auxotroph and cannot grow \nin minimal medium, this strain and its engineered strains were assayed in 50% TSB \nmedium instead of M9. All strains were tested in three biological replicates to ensure \nreproducibility. \nBacterial growth curve \nBacterial preparation followed the same protocol as the root growth inhibition assay. \nWashed cells were inoculated into 200 µl of either M9 minimal medium or 50% TSB \nmedium at a final OD\n₆₀₀  of 0.05. Each strain was tested in three biological replicates to \nensure reproducibility. Cultures were grown in sterile 96-well cell culture plates, sealed \nwith Breathe-Easy gas-permeable film (Diversified Biotech), and incubated at 28 /i2 °C \nwith continuous linear shaking. Optical density at 600 /i2 nm (OD₆₀₀ ) was recorded every \n2/i2 h over a 24-hour period using a BioTek Synergy H1 microplate reader to monitor \nbacterial growth dynamics. \nRoot growth inhibition assay \nSeedling and seed preparation. Arabidopsis thaliana  Col-0 seeds were surface-\nsterilized by vortexing in 70% bleach containing 0.2% Tween-20 for 10 min, followed by \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n35 \n \nfive washes with sterile distilled water. Seeds were sown on half-strength Murashige \nand Skoog (MS) agar medium (2.22 g/L MS basal medium with Gamborg vitamins \n(PhytoTech Labs, M-404), 0.5 g/L MES, 5 g/L sucrose, 10 g/L agar, pH 5.7 adjusted \nwith 3 M NaOH) in 12 × 12 cm square plates and grown vertically under short-day \nconditions (21\n/i2 °C day / 18 /i2 °C night, 10 h light / 14 h dark, 70% relative humidity, 170 \nμ mol m ⁻ ² s ⁻ ¹ light intensity) for 7 days. Medicago seeds were sterilized following \npublished protocols50,51. Briefly, seeds were treated with concentrated sulfuric acid for \n10 min with agitation, rinsed once with sterile water, then treated with 70% bleach for 3 \nmin. After five additional washes, seeds were imbibed in sterile water for 2–6 h at room \ntemperature prior to use. \nIAA-containing plate preparation. Half-strength MS agar medium was autoclaved and \ncooled until warm to the touch. IAA stock solutions (1 mM or 100 mM in 100% ethanol) \nwere added to final concentrations of 100 nM, 1 µM, or 10 µM. Stocks were stored at –\n20\n/i2 °C for up to two months. \nBacterial and SynCom32 preparation. Bacterial strains were revived from 20% \nglycerol stocks by streaking onto 50% TSB agar plates (15 g/L tryptic soy broth, 20 g/L \nagar) and incubated at 28 /i2 °C for 3–4 days. Single colonies were inoculated into 5 mL \n50% TSB and grown for 2 days at 28 /i2 °C with shaking (250 rpm). Cultures were \npelleted at 5,000 × g for 10/i2 min, washed twice with 3/i2 mL of sterile 10/i2 mM MgCl₂ , and \nresuspended in 750/i2 µl of the same buffer. The optical density at 600 /i2 nm (OD₆₀₀ ) was \nmeasured using a NanoDrop One C with semi-micro cuvettes and adjusted to 0.05. For \nSynCom32 assembly, each strain’s OD ₆₀₀  was measured individually. To ensure equal \nbiomass contribution, volumes were calculated using the equation:  OD₆₀₀  × volume (µl) \n= 300, and combined accordingly. The pooled culture was washed and resuspended as \ndescribed above. \nBacterial inoculation. For plant-microbe co-inoculation assays, 100 µl of strain (OD ₆₀₀  \n= 0.05) was evenly spread onto the surface of half-strength MS agar plates. For \nSynCom32, 100 µl of the pooled culture (OD\n₆₀₀  = 0.05) was applied, followed by 100 µl \nof V. paradoxus  CL014, Polaromonas MF047, Paraburkholderia MF376, or their \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n36 \n \nengineered strains at OD ₆₀₀  = 0.005. Plates were incubated overnight at room \ntemperature prior to seedling transfer. \nPlant growth and measurement. Seven-day-old Arabidopsis seedlings or sterilized \nMedicago seeds were transferred onto pre-inoculated plates. Plates were sealed with \n3M Micropore tape and incubated vertically in a growth chamber under short-day \nconditions for 7 days. For 16S rRNA amplicon sequencing, plants were grown for 9 \ndays to allow sufficient biomass for DNA extraction.  Plates were imaged using a digital \ncamera, and primary root elongation was measured as the distance from the initial to \nfinal root tip position using the freehand line tool in ImageJ.  All primary root elongation \ndata are provided in Supplementary Table 3. \n \nMicrobiome analysis \nSample collection. After nine days of co-incubation with bacteria, roots from \nArabidopsis and Medicago were harvested for microbiome profiling. Each sample \nconsisted of 5–10 roots pooled per plate, with 3–5 biological replicates per treatment. \nRoots were transferred to 15 mL Falcon tubes containing 7 mL sterile water and \nwashed by vigorous vortexing. Excess water was removed using sterile filter paper, and \ndried roots were transferred to Lysing Matrix E tubes (MP Biomedicals) and stored at –\n80\n/i2 °C until DNA extraction. For SynCom32 input controls, bacterial cultures were \npelleted at 5,000 × g for 10 min, the supernatant was discarded, and pellets were stored \nat –80/i2 °C. \nDNA extraction. Samples were homogenized using a FastPrep-24™ 5G instrument \n(MP Biomedicals) with two 40 s cycles at 6.0 m/s for root samples, and one cycle for \nSynCom32 input controls, with 2 min on ice between runs to prevent overheating. DNA \nwas extracted using the FastDNA SPIN Kit for Soil (MP Biomedicals), eluted in 55 µl \nDES buffer (provided in the kit), quantified using the Quant-iT PicoGreen dsDNA Assay \nKit (Thermo Fisher), and diluted to 3.5 ng/µl with DES buffer. \n16S rRNA library preparation and sequencing. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n37 \n \nThe V5–V7 region52 of the 16S rRNA gene was amplified using a two-step dual-indexed \nPCR approach. The first PCR was performed using Platinum SuperFi II PCR Master \nMix (Thermo Fisher) in a 25 µl reaction containing: 12.5 µl Master Mix, 7 µl nuclease-\nfree water (Qiagen), 2 µl of 5 µM forward primer 799F (5\n′ -\nAACMGGATTAGATACCCKG-3′ , with 10-bp sample barcode), 1 µl of 10 µM reverse \nprimer 1192R (5 ′ -ACGTCATCCCCACCTTCC-3′ , with 6-bp library barcode), and 2.5 µl \nof 3.5 ng/µl DNA template. Thermal cycling conditions were: 98/i2 °C for 30 s, followed by \n30 cycles of 98/i2 °C for 10 s, 55 /i2 °C for 10 s, and 72 /i2 °C for 15 s, with a final extension \nat 72/i2 °C for 5 min. Three technical PCR replicates were pooled per sample to minimize \namplification bias. Products were verified by 1.2% agarose gel electrophoresis. For gel \npurification, 25 µl of each PCR product (two samples per lane) were pooled with 10 µl \n6× loading dye, run on a 1.2% agarose gel, and ~400 bp bands were excised and \npurified using the Wizard SV Gel and PCR Clean-Up System (Promega). DNA \nconcentration was assessed by PicoGreen, and 100 ng of each sample was pooled for \nlibrary construction and cleaned with 0.9× AMPure XP beads (Beckman Coulter). A \nsecond PCR was performed to add Illumina sequencing adapters. Final libraries were \nsequenced on an Illumina MiSeq platform (2 × 300 bp paired-end reads)\n52,53 at the \nPrinceton Genomics Core Facility. After quality filtering, a total of 14,907,735 high-\nquality sequences were obtained from 175 samples, with an average of 85,187 reads \nper sample. \nAmplicon data processing. Raw reads were demultiplexed in QIIME2 (v2024.10)\n54 \nusing qiime cutadapt demux-paired 55, and primer sequences were trimmed with qiime \ncutadapt trim-paired. Denoising, quality filtering, and chimera removal were performed \nwith DADA2 (qiime dada2 denoise-paired) 56. After testing multiple truncation lengths, \nforward reads were truncated at 220 bp and reverse reads at 180 bp to optimize read \nretention and ASV diversity. ASVs were taxonomically classified using a naïve bayes \nclassifier trained on a custom database of root-associated bacterial sequences\n57 via \nqiime feature-classifier classify-sklearn. Assignments were filtered based on confidence \nscores, prevalence, and relative abundance. Relative abundance tables were generated \nusing qiime feature-table relative-frequency and merged with taxonomy and metadata \nfor visualization in R using ggplot2. Beta diversity was calculated using Bray–Curtis \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n38 \n \ndissimilarity (R package vegan, vegdist) 58, followed by principal coordinate analysis \n(PCoA; cmdscale). Differences in community composition were assessed by \nPERMANOVA using adonis2 58. Differential abundance analysis was conducted using \nthe Mann–Whitney U test with FDR correction. Log ₂  fold changes were calculated, and \nheatmaps were visualized using pheatmap (R v4.4.2)59.  \n \nPlant growth promotion assay in natural soil \nSoil was collected from the Stony Ford Research Station (Princeton, NJ, USA), where \nno chemical fertilizers, pesticides, or plants had been applied or grown in recent years. \nThe soil was sieved twice to remove rocks and plant debris, then distributed into pots \nplaced within 9 × 13-inch aluminum foil trays. Pots were saturated overnight with 1.2 L \nof sterile water containing bacterial inoculum. Bacterial preparation followed the same \nprotocol as the root growth inhibition assay, including streaking, cultivation, washing, \nand OD\n₆₀₀  measurement. Individual bacterial strains were suspended in 1.2 L of sterile \nwater to a final OD ₆₀₀  of 0.03 prior to soil application. Each treatment included four \nbiological replicates (pots). Arabidopsis  Col-0 seeds were surface sterilized as \ndescribed above and placed directly onto the surface of inoculated soil. Pots were \nmaintained under short-day conditions in a growth chamber and watered with 600–800 \nmL of distilled water every 4 days. Afte r 33 days, above-ground tissues were harvested \nand shoot fresh weight was measured using an analytical balance. \n \nAcknowledgements \nWe thank Prof. Jeffery L. Dangl (University of North Carolina at Chapel Hill, USA) for \nproviding bacterial strains used in this study and for valuable suggestions on the \nmanuscript. We also acknowledge the Princeton University Genomics Core Facility for \nperforming 16S rRNA amplicon sequencing. This research used resources of the AMX \n(17-ID-1) and FMX (17-ID-2) beamlines at the National Synchrotron Light Source II \n(NSLS-II), a U.S. Department of Energy (DOE) Office of Science User Facility operated \nby Brookhaven National Laboratory under Contract No. DE-SC0012704. The Center for \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n39 \n \nBio Molecular Structure (CBMS) is supported by the NIH National Institute of General \nMedical Sciences (P30GM133893) and the DOE Office of Biological and Environmental \nResearch (KP1605010). This work was supported by the Lidow Independent \nWork/Senior Thesis Fund to M.J.K.; the National Science Foundation grant CHE-\n2246289  to J.T.G.; the Department of Energy (DOE) DE-SC0018260 to J.D.R; the DOE \nCenter for Advanced Bioenergy and Bioproducts Innovation (U.S. Department of Energy, \nOffice of Science, Biological and Environmental Research Program under Award \nNumber DE-SC0018420) to J.D.R., Y.S., and X.L.; the Project X Fund administered by \nthe School of Engineering and Applied Science at Princeton University to J.M.C.; \nstartup funds from the Department of Chemical and Biological Engineering to J.M.C. \n \nAuthor contributions \nJ.M.C. and J.D.R. supervised the project. T.J., Y.S., and J.M.C. conceived the study \nand designed the experiments. T.J. constructed mutant bacterial strains and prepared \nsamples for LC-MS analysis. Y.S. and X.L. conducted metabolomics experiments, and \ntogether with J.T.G., analyzed the resulting data. T.J. performed plant-microbe \ninteraction experiments and data analysis. T.J. and M.J.K. expressed and purified \nproteins, while T.J. and P.D.J. carried out protein crystallization and structural data \nanalysis. T.J. also prepared samples, constructed libraries, and analyzed 16S rRNA \nsequencing data, and conducted the plant growth promotion assay in natural soil. T.J., \nY.S., X.L., and J.M.C. wrote the manuscript, with input and feedback from all co-authors. \n \nCompeting interests \nPrinceton University has filed pending patent applications covering aspects of the auxin \ndegradation pathway engineering described in this work, listing T.J. and J.M.C. as \ninventors. J.D.R. is a co-founder, director and stockholder in Raze Therapeutics and \nFarber Partners; a co-founder and stockholder in Fargo Biotechnologies; and an advisor \nand stockholder in Empress Therapeutics, Bantam Pharmaceuticals, Faeth \nTherapeutics, Colorado Research Partners and Rafael Pharmaceuticals. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n40 \n \n \nData availability \nAll data supporting the findings of this study are available within the paper and its \nSupplementary Information. Source data underlying the figures are also provided with \nthis paper. The IadDE protein structure has been deposited in the Protein Data Bank \nunder accession code 9O71. \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n41 \n \n \nReferences \n1. Acharya, M. et al. Soil microbial diversity in organic and non-organic pasture systems. \nPeerJ 9, e11184 (2021). \n2. Banerjee, S. & Van Der Heijden, M. G. A. Soil microbiomes and one health. Nat. Rev. \nMicrobiol. 21, 6–20 (2023). \n3. Duca, D. R. & Glick, B. R. Indole-3-acetic acid biosynthesis and its regulation in plant-\nassociated bacteria. Appl Microbiol Biotechnol 104, 8607–8619 (2020). \n4. Bartel, B. Auxin biosynthesis. Annu. Rev. Plant Biol. 48, 51–66 (1997). \n5. Finkel, O. M. et al. A single bacterial genus maintains root growth in a complex \nmicrobiome. Nature 587, 103–108 (2020). \n6. Spaepen, S. & Vanderleyden, J. Auxin and plant-microbe interactions. Cold Spring \nHarbor Perspectives in Biology 3, a001438–a001438 (2011). \n7. Zhang, P. et al. The Distribution of Tryptophan-Dependent Indole-3-Acetic Acid \nSynthesis Pathways in Bacteria Unraveled by Large-Scale Genomic Analysis. \nMolecules 24, 1411 (2019). \n8. Wang, L. et al. Systematic characterization of plant-associated bacteria that can \ndegrade indole-3-acetic acid. PLoS Biol 22, e3002921 (2024). \n9. Conway, J. M. et al. Diverse MarR bacterial regulators of auxin catabolism in the \nplant microbiome. Nat. Microbiol. 7, 1817–1833 (2022). \n10. Leveau, J. H. J. & Gerards, S. Discovery of a bacterial gene cluster for \ncatabolism of the plant hormone indole 3-acetic acid: Genes in pseudomonas putida \n1290 for the degradation of IAA. FEMS Microbiology Ecology 65, 238–250 (2008). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n42 \n \n11. Scott, J. C., Greenhut, I. V. & Leveau, J. H. J. Functional characterization of the \nbacterial iac genes for degradation of the plant hormone indole-3-acetic acid. J Chem \nEcol 39, 942–951 (2013). \n12. Ma, Y. et al. Structural and biochemical characterization of the key components \nof an auxin degradation operon from the rhizosphere bacterium Variovorax. PLoS \nBiol. 21, e3002189 (2023). \n13. Jensen, J. B., Egsgaard, H., Van Onckelen, H. & Jochimsen, B. U. Catabolism of \nindole-3-acetic acid and 4- and 5-chloroindole-3-acetic acid in bradyrhizobium \njaponicum. J Bacteriol 177, 5762–5766 (1995). \n14. Fuchs, G., Boll, M. & Heider, J. Microbial degradation of aromatic compounds — \nfrom one strategy to four. Nat. Rev. Microbiol. 9, 803–816 (2011). \n15. Chakrabarty, S., Austin, R. N., Deng, D., Groves, J. T. & Lipscomb, J. D. Radical \nintermediates in monooxygenase reactions of rieske dioxygenases. J. Am. Chem. \nSoc. 129, 3514–3515 (2007). \n16. Mondal, P., Rajapakse, S. & Wijeratne, G. B. Following nature’s footprint: \nMimicking the high-valent heme-oxo mediated indole monooxygenation reaction \nlandscape of heme enzymes. J. Am. Chem. Soc. 144, 3843–3854 (2022). \n17. Shin, I., Nguyen, R. C., Montoya, S. R. & Liu, A. Structural insights into 2-\noxindole-forming monooxygenase MarE: Divergent architecture and substrate \npositioning versus tryptophan dioxygenases. Journal of Biological Chemistry 301, \n108241 (2025). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n43 \n \n18. Zhang, Y. et al. Characterization of 2-Oxindole Forming Heme Enzyme MarE, \nExpanding the Functional Diversity of the Tryptophan Dioxygenase Superfamily. J. \nAm. Chem. Soc. 139, 11887–11894 (2017). \n19. Basran, J. et al. The mechanism of formation of N -formylkynurenine by heme \ndioxygenases. J. Am. Chem. Soc. 133, 16251–16257 (2011). \n20. Bai, Y. et al. Functional overlap of the Arabidopsis leaf and root microbiota. \nNature 528, 364–369 (2015). \n21. Kovaleva, E. G. & Lipscomb, J. D. Versatility of biological non-heme fe(II) centers \nin oxygen activation reactions. Nat Chem Biol 4, 186–193 (2008). \n22. Shin, I., Wang, Y. & Liu, A. A new regime of heme-dependent aromatic \noxygenase superfamily. Proc. Natl. Acad. Sci. U.S.A. 118, e2106561118 (2021). \n23. Thodberg, S. et al. A flavin-dependent monooxygenase catalyzes the initial step \nin cyanogenic glycoside synthesis in ferns. Commun Biol 3, 507 (2020). \n24. Wu, M. et al. The promiscuous flavin-dependent monooxygenase PboD from \naspergillus ustus increases the structural diversity of hydroxylated pyrroloindoline \ndiketopiperazines. J. Nat. Prod. 87, 1171–1178 (2024). \n25. Li, S. et al. Biochemical characterization of NotB as an FAD-dependent oxidase \nin the biosynthesis of notoamide indole alkaloids. J. Am. Chem. Soc. 134, 788–791 \n(2012). \n26. Ferraro, D. J., Gakhar, L. & Ramaswamy, S. Rieske business: Structure–function \nof Rieske non-heme oxygenases. Biochemical and Biophysical Research \nCommunications 338, 175–190 (2005). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n44 \n \n27. Duarte, M., Jauregui, R., Vilchez-Vargas, R., Junca, H. & Pieper, D. H. \nAromaDeg, a novel database for phylogenomics of aerobic bacterial degradation of \naromatics. Database 2014, (2014). \n28. Levy, A. et al. Genomic features of bacterial adaptation to plants. Nat. Genet. 50, \n138–150 (2018). \n29. Nagahama, K., Ogawa, T., Fujii, T. & Fukuda, H. Classification of ethylene-\nproducing bacteria in terms of biosynthetic pathways to ethylene. Journal of \nFermentation and Bioengineering 73, 1–5 (1992). \n30. Arshad, M., Saleem, M. & Hussain, S. Perspectives of bacterial ACC deaminase \nin phytoremediation. Trends in Biotechnology 25, 356–362 (2007). \n31. Eckert, C. et al. Ethylene-forming enzyme and bioethylene production. Biotechnol \nBiofuels 7, 33 (2014). \n32. KVpczynska, E., Zielinska, S. & KVpczynski, J. Ethylene production by \nagrobacterium rhizogenes strains in vitro and in vivo. Plant Growth Regulation 39, \n13–17 (2003). \n33. Primrose, S. B. Ethylene-forming bacteria from soil and water. Journal of General \nMicrobiology 97, 343–346 (1976). \n34. Durán, P. et al. Microbial Interkingdom Interactions in Roots Promote Arabidopsis \nSurvival. Cell 175, 973-983.e14 (2018). \n35. Qi, M. et al. Identification of beneficial and detrimental bacteria impacting \nsorghum responses to drought using multi-scale and multi-system microbiome \ncomparisons. The ISME Journal 16, 1957–1969 (2022). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n45 \n \n36. Nolan, K. & Wang, Y. Combined spectroscopic and structural approaches to \nexplore the mechanism of histidine-ligated heme-dependent aromatic oxygenases. in \nMethods in Enzymology vol. 685 405–432 (Elsevier, 2023). \n37. Nelp, M. T., Zheng, V., Davis, K. M., Stiefel, W. K. J. & Groves, J. T. Potent \nactivation of indoleamine 2,3-dioxygenase by polysulfides. J. Am. Chem. Soc. 141, \n15288–15300 (2019). \n38. Floriano, B., Santero, E. & Reyes-Ramírez, F. Biodegradation of tetralin: \nGenomics, gene function and regulation. Genes 10, 339 (2019). \n39. Hamad, M. A., Zajdowicz, S. L., Holmes, R. K. & Voskuil, M. I. An allelic \nexchange system for compliant genetic manipulation of the select agents \nburkholderia pseudomallei and burkholderia mallei. Gene 430, 123–131 (2009). \n40. Zhang, J. et al. High-throughput cultivation and identification of bacteria from the \nplant root microbiota. Nat. Protoc. 16, 988–1012 (2021). \n41. Kovach, M. E. et al. Four new derivatives of the broad-host-range cloning vector \npBBR1MCS, carrying different antibiotic-resistance cassettes. Gene 166, 175–176 \n(1995). \n42. Chambers, M. C. et al. A cross-platform toolkit for mass spectrometry and \nproteomics. Nat Biotechnol 30, 918–920 (2012). \n43. Su, X., Lu, W. & Rabinowitz, J. D. Metabolite spectral accuracy on orbitraps. \nAnal. Chem. 89, 5940–5948 (2017). \n44. Studier, F. W. Protein production by auto-induction in high-density shaking \ncultures. Protein Expression and Purification 41, 207–234 (2005). \n45. Kabsch, W. XDS. Acta Crystallogr D Biol Crystallogr 66, 125–132 (2010). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n46 \n \n46. Evans, P. R. & Murshudov, G. N. How good are my data and what is the \nresolution? Acta Crystallogr D Biol Crystallogr 69, 1204–1214 (2013). \n47. McCoy, A. J. et al. Phaser crystallographic software. J Appl Crystallogr 40, 658–\n674 (2007). \n48. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development \nof coot. Acta Crystallogr D Biol Crystallogr 66, 486–501 (2010). \n49. Adams, P. D. et al.  PHENIX/i2 : A comprehensive python-based system for \nmacromolecular structure solution. Acta Crystallogr D Biol Crystallogr 66, 213–221 \n(2010). \n50. Liu, H. et al. Constitutive activation of a nuclear-localized calcium channel \ncomplex in medicago truncatula. Proc. Natl. Acad. Sci. U.S.A. 119, e2205920119 \n(2022). \n51. Feng, F. et al. A combination of chitooligosaccharide and \nlipochitooligosaccharide recognition promotes arbuscular mycorrhizal associations in \nmedicago truncatula. Nat Commun 10, 5047 (2019). \n52. Bulgarelli, D. et al. Revealing structure and assembly cues for Arabidopsis root-\ninhabiting bacterial microbiota. Nature 488, 91–95 (2012). \n53. Huang, A. C. et al.  A specialized metabolic network selectively modulates \nArabidopsis root microbiota. Science 364, eaau6389 (2019). \n54. Bolyen, E. et al. Author correction: Reproducible, interactive, scalable and \nextensible microbiome data science using QIIME 2. Nat Biotechnol 37, 1091–1091 \n(2019). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint \n\n  \n47 \n \n55. Martin, M. Cutadapt removes adapter sequences from high-throughput \nsequencing reads. EMBnet j. 17, 10 (2011). \n56. Callahan, B. J. et al. DADA2: High-resolution sample inference from illumina \namplicon data. Nat Methods 13, 581–583 (2016). \n57. Pedregosa, F. et al. Scikit-learn: Machine learning in python. MACHINE \nLEARNING IN PYTHON. \n58. Oksanen, J. et al. vegan: Community ecology package. 2.6-10 \nhttps://doi.org/10.32614/CRAN.package.vegan (2001). \n59. Raivo Kolde. pheatmap: Pretty heatmaps. 1.0.12 \nhttps://doi.org/10.32614/CRAN.package.pheatmap (2010). \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted October 26, 2025. ; https://doi.org/10.1101/2025.10.25.684584doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}