Insecticide resistance by a host-symbiont reciprocal detoxification

Insecticide resistance is one of the most serious problems in contemporary agriculture and public health. Although recent studies revealed that insect gut symbionts contribute to resistance, the symbiont-mediated detoxification process remains unclear. Here we report the in vivo detoxification process of an organophosphorus insecticide, fenitrothion, in the bean bug Riptortus pedestris. Using transcriptomics and reverse genetics, we reveal that gut symbiotic bacteria degrade this insecticide through a horizontally acquired insecticide-degrading enzyme into the non-insecticidal but bactericidal compound 3-methyl-4-nitrophenol, which is subsequently excreted by the host insect. This integrated “host-symbiont reciprocal detoxification relay” enables the simultaneous maintenance of symbiosis and efficient insecticide degradation. We also find that the symbiont-mediated detoxification process is analogous to the insect genome-encoded fenitrothion detoxification system present in other insects. Our findings highlight the capacity of symbiosis, combined with horizontal gene transfer in the environment, as a powerful strategy for an insect to instantly eliminate a toxic chemical compound, which could play a critical role in the human-pest arms race.

The experiments with the dissected guts were also ingenious. Some of the physiological mechanisms are not really clear-how does the 3M4N exit the gut, and then how is it excreted? However, these are details that are not really needed for the main story.
Altogether, I found this to be a very rich paper with abundant experimental data to support the proposed route of detoxification. It was also easy to read. The figures are very nice. It is too bad so much good work is buried in the SI, but I can see that it would be a very long paper if included in the main text.
Very minor points: l. 41 "prosperity" is not the right word to use here.
Extended data Fig 4a: the little dotted lines on the illustratio of the MEP degradation assay seem to be shifte to the left, and not pointing to the correct spots on the plate. Extended data Fig 6: please define M2, M3, M4, H in the legend. Some people might think H is head, but it seems tobe the other end! Reviewer #3: Remarks to the Author: In this study Sato and co-authors study the molecular and physiological basis of insecticide detoxification in an insect by its associated bacterial symbiont. The study builds upon several previous studies by this same group that have shown how this symbiont that is acquired from the soil at every generation is responsible for insecticide resistance in the insect. The authors perform a series of genomic analyses that identify the potential genes responsible for such degradation. The functionality of these genes is verified through gene silencing coupled with biochemistry assays. The authors also demonstrate that while the symbiotic bacteria degrades the insecticide, which is not bactericidal, the product of this degradation is bactericidal so that in return, the host insect immediately excretes it. The authors confirm such excretion by detecting the degradation product in insect faeces. The authors perform these experiments with bacteria both in vitro and in vivo. In vivo experiments are performed by reinoculating the symbiont (and created mutants) in the insect host. By doing so the authors can test symbiont effects on insect fitness. A surprising result is, however, that the gene expression levels found in vitro and in vivo are quite different, which reveals the important role played by the host in regulating the metabolism of the microbe. The authors also demonstrate that insecticide degradation, which is located in a gene inside a plasmid, can be horizontally transmitted among bacteria. This is important as it shows how insecticide resistance can be transmitted among bacteria and potentially among insect hosts. This part of the study is maybe less clear to me as the authors demonstrate such horizontal transmission between bacteria in vitro, but not in vivo. Also, even if horizontal transmission of this trait is demonstrated among bacteria, whether the same occurs in nature among insect hosts is not demonstrated. As far as I am aware this study is one of the few reported examples where an interaction of mutual benefits mediated by an insect toxic compound has been reported. The manuscript is well presented, clear and suits the general audience of Nature Communications.
Enric Frago CBGP -Centre for Biology and Management of Populations -Montpellier -France ( I sign all my reviews ) We appreciate the constructive and favorable comments made by the reviewers regarding our manuscript entitled "Insecticide resistance by a host-symbiont reciprocal detoxification" (NCOMMS-21-21341-T). We have carefully revised the manuscript and addressed in detail each of the comments/suggestions made by the reviewers. The revised parts and sentences have been highlighted in the marked-up manuscript. Below, we reply point-by-point to all comments (copied in blue).

Response to Reviewer #1:
In this manuscript, Sato and colleagues elucidate the molecular basis of insecticide resistance in the Riptortus-Burkholderia symbiosis. Following up on earlier reports of the same group on insecticide-degrading symbionts in the bean bug, they show that the symbionts encode two different pathways for fenitrothion (MEP) degradation, only one of which is highly upregulated in vitro upon exposure to MEP. Surprisingly, in vitro RNAseq revealed that only the first enzyme of the degradation pathway is constitutively highly expressed, whereas the subsequent ones are effectively silent even when the host is fed a MEP-containing diet. Deletion mutants of the first and second step in the MEP degradation pathway reveal that the first step (conversion of MEP to 3M4N) is necessary and sufficient to confer MEP resistance to the host, but both steps are required for in vitro growth of the symbionts when grown on MEP as the sole carbon source. Interestingly, as 3M4N has bactericidal activity on the symbionts, the authors conclude that the insect effectively remove 3M4N from the midgut crypts, and they provide in vivo evidence that this is indeed the case. This manuscript is a tour -de-force on the fascinating interplay between host and symbiont for the detoxification of an insecticide and reveals the molecular basis for the rapid acquisition of a novel ecological trait through symbiosis in an elegant series of experiments. As such, it will make an outstanding contribution to the scientific literature, and I am convinced that it will receive tremendous attention from a broad scientific readership. While I do have a few comments for the authors to consider in a revision, I am very enthusiastic about the quality and novelty of the science presented in this manuscript.
---We thank the reviewer for the helpful and useful comments, which have significantly improved this manuscript.
General comment: 1. The authors present convincing data that the symbiont's mpd expression is necessary and sufficient for conferring insecticide resistance to the host. However, the dynamics of 3M4N clearance still remain mysterious to me. I urge the authors to consider and address the following aspects: a. The authors claim that pnp is upregulated upon accumulation of 3M4N in vitro, but they do not explicitly show this. If this is the case, it would indeed make sense that pnp and the following steps in the pathway are not upregulated in vivo, if 3M4N is in fact quickly cleared from the midgut crypts by as yet unknown host mechanisms. Given that this is an important aspect, it would be worth testing whether pnp upregulation indeed responds to the presence of 3M4N. This should be easy to do in vitro, i.e. by assessing pnp expression upon exposure to 3M4N via qPCR or RNAseq. (i) The cells were cultured in minimal medium containing citrate, and at the log phase, MEP was added. One hour after the addition of MEP, the cells were collected and subjected to RNA-seq.
(ii) Cells were transferred from the citrate medium to the MEP medium, and RNA-seq was conducted after yellow color of the MEP medium started to disappear.
(iii) Cells were transferred from the spent MEP medium to freshly prepared MEP medium.
The RNA-seq data clearly confirmed that pnp is not constitutively expressed but is induced by b. I am somewhat puzzled by the fact that the symbionts convert MEP into 3M4N, and then export the latter to then be cleared by the insect, even though they could convert it to non-bactericidal products. Or am I misunderstanding this, and MEP degradation actually occurs via extracellular enzymes (Mpd)? Since localization is crucial here, it would be good to explain in a bit more detail what is known about the localization of MEP-degrading enzymes.
---The cellular localization of Mpd has been thoroughly examined in Pseudomonas strains, wherein Mpd is localized in the inner membrane and the degradation from MEP to 3M4N occurs in the periplasmic space (Liu et al. 2005, Biochem Biophys Res Commun;Zhongli et al. 2001 Appl Environ Microbiol;Yang et al. 2008, Biotechnol Bioeng). By heterologous expression of mpd in E. coli, Yang et al. (J Agric Food Chem 2009) confirmed that Mpd is localized and functions in the inner membrane (and is not secreted into the culture supernatant). The localization of Mpd is now mentioned in the main text with reference to these studies (L114 -115 in the revised manuscript). Although the localization was shown in the summarizing illustration ( Figure   3d), we apologize for not including this important information in the main text. Thus, under in vivo conditions, we expect that 3M4N accumulates in the luminal region of the gut symbiotic organ (M4), which is immediately eliminated by the gut epidermal cells of the host insect without damaging symbiont cells.
c. The authors nicely show the presence of 3M4N in the bugs' feces after oral exposure to this compound. I was a bit surprised to not see data on the excretion of 3M4N in MEP-fed bugs, since this would directly demonstrate 3M4N as the end product of MEP degradation. This experiment would be a good support of the presented scenario on host-symbiont interplay to deal with the insecticide. If this could be done quantitatively (i.e. quantifying the amount of ingested MEP by assessing the amount of consumed liquid and then quantifying 3M4N), the authors could even draw conclusions on the efficiency of degradation, but semi-quantitative measurements would also be sufficient to make the point (i.e. quantifying MEP vs. 3M4N in the feces of MEP-fed bugs).
---Thank you for your useful comment. We agree with the reviewer's suggestion. Although we made multiple attempts to detect MEP and 3M4N in the feces of MEP-fed insects, the detection was unsuccessful. Perhaps, the failure occurred due to foreign substances present in the feces and/or the detection limit of the chemical compounds in feces. When the insects were fed a high dose of MEP for easier detection, the insects did not feed well and produced less feces. Moreover, MEP is a photodegradative compound, which can also hamper quantitative detection. Despite these efforts, we concluded that it is difficult to obtain quantitative data in MEP-fed insects, and to experimentally confirm the excretion of 3M4N by the host. We conducted in vivo detoxification  figure 7). Since 3M4N was not toxic to the insects, these experiments were successful and we managed to accumulate supportive evidence for the 3M4N efflux by the host insect.
d. What is the point (from the bacterial perspective) of having a degradation pathway with the first step constitutively expressed that converts a non-toxic substrate into a toxic intermediate, and then have a lag time to upregulate the following enzymes? I have a hard time understanding how this could evolve, unless the substrate is a really common carbon source that makes it worthwhile and compensates for the costs of possible self-intoxication. This is a general question on the degrading ability of the bacteria, regardless of their symbiotic association with the host, so maybe I am missing something important here.
---Although the driver of such unfavorable metabolic system is not well understood, this is a very interesting point. In the soil environment where these bacteria live, the complex aggregated structure of the soil could not allow sufficient accumulation of 3M4N to harm the bacterium itself.
Alternatively, it is reasonable to suggest that the production and accumulation of 3M4N could contribute to the inhibition and sterilization of other surrounding bacteria. In turn, it can benefit the eventual growth of the bacterium and lead to the dominance of the bacterium in the soil habitat.
Although this is not the focus of the present study, it can be an intriguing topic for investigating microbial evolution in future studies.
Specific comments: l. 47: Since this paper is not about an insect feeding on plant sap or vertebrate blood, please keep this more general on insect symbiosis.
---Thank you for your useful comment. We agree, and have revised the sentence in the revised manuscript: "Many insects possess symbiotic bacteria in the bacteriocytes or the gut, wherein symbionts play pivotal metabolic roles such as the provision of essential amino acids, supplementation of vitamins, and digestion of indigestible food materials, such as plant cell walls." (L47 -50 in the revised manuscript) l. 61: delete "environomental" ---The term has been deleted. (L60 in the revised manuscript) l. 64: delete "to" ---As suggested "to" was deleted. (L63 in the revised manuscript) l. 83-85: While the phylogenies do indicate quite a bit of horizontal transfer, the symbiont genes are usually embedded in a clade of Burkholderia/Caballeronia, so I am not completely convinced that this provides strong evidence for horizontal acquisition of the plasmid by the symbionts. Am I missing anything here? This is a small point, though, since the authors nicely and convincingly demonstrate that the plasmid can be transferred horizontally.
---In the revised version, a phylogeny of 16S rRNA gene sequences was included to show the relationships between the bacteria carrying the MEP-degrading genes ( Supplementary Fig. 3e).
This phylogeny also includes a number of alliances that have neither MEP-degradation abilities nor mpd, wherein MEP-degraders carrying mpd do not form a monophyletic group. This provides a strong argument for the presence of frequent horizontal transmission of the degradation trait.
Although it remains unclear why bacteria of the same genus tend to be clustered together, conjugation compatibility between bacterial species may play a pivotal role. To avoid confusion, the sentence has been revised as follows: "The phylogeny of bacterial mpd genes as well as other genes of the Mhq pathway ( l. 166: Please rephrase, as the effect of trehalose was not explicitly assessed, since the assays were only done in the presence (but not the absence) of trehalose. What you mean to say is that the effect is observed in the presence of trehalose.  Figure 6b). Trehalose was added because it acts as an energy source for insect cells and a higher effect on 3M4N efflux was expected.
However, there was no statistically significant difference between the treatments. Even though the data appears to be unimportant, we would like to retain it as some readers may be concerned about the effect of trehalose on the in vivo assay.
l. 206: Consider replacing "Analogous" with "Thus", and removing "similar" ---All the words have been revised accordingly. (L215 -216 in the revised manuscript). Ext. Data Fig. 3: Please highlight the placement of the symbiont. It is not stated anywhere that "BSFA1" is the symbiont. Although it can be deduced from the text, please make the reader's life easier by stating it here. Also, it would be good to add a phylogenomic analysis of the symbiont to show its affiliation to the genus Burkholderia (or Caballeronia?) and to put the displayed phylogenies into perspective.
---In the phylogenetic trees, "BSFA1" has been replaced by "Burkholderia symbiont strain SFA1" The phylogenetic tree of the 16S rRNA gene was included to show the phylogenetic placement of strain SFA1 (it is placed in the stinkbug-associated Burkholderia [Caballeronia] clade). ---As the feces are a little greenish-to-brownish in color, and, as we put in about the same amount of feces, we believe that the difference in color is due to the 3M4N excreted by the insects. This color difference is clear and easy to understand, thus, we would like to retain this picture in this figure.
---The mass and other spectral data of the 3M4N standard are available online from the Spectral Database for Organic Compounds (https://sdbs.db.aist.go.jp/sdbs/cgi-bin/direct_frame_top.cgi).
The link is shown in the legend of Supplementary fig. 7. In addition, the UV-Vis absorbance spectrum of 1 mM 3M4N has been added to Supplementary Fig. 7e.

Response to Reviewer #2:
This study examined the molecular basis for insecticide detoxification by a gut symbiont of the bean bug. In 2012, some of the same researchers published evidence that an environmentally acquired Burkholderia bacterium confered insecticide resistance upon the insect host; this was a very fascinating result (which has been cited >500 times). This paper provides a very significant extension, showing that the symbiont provides a first enzymatic step, that produces a noninsecticidal product (3M4N). Interestingly, the product is bacteriocidal, which would potentially be self-defeating, since killing the bacteria would remove their beneficial effects for the insect.
Other experiments show that the insect excretes the 3M4N, removing it from the gut lumen where the bacteria reside.
I read all of this paper with much interest, including the (long) SI. In general, the claims seem very interesting and also quite well supported. There is quite a lot of data here, including the full bacterial genome, which turns out to be very complex, as it is larger than most bacterial genomes, and contains several chromosomes and numerous plasmids. The genome contained candidate loci for the detoxification activity, as characterized in other bacteria. As it turns out, these are indeed the loci responsible. One core finding of the paper was the use of knockout strains lacking the initial Mpd detoxification enzyme and complemented strains, and the role of this locus is confirmed with classic molecular microbiology methods. Similar knockouts and complements were used to confirm a role of another locus (pnpA2) in further transformation of the 3M4N.
Interestingly these loci are on the same bacterial plasmid, though in different locations, and are inducible by MEP.
Phylogenetic trees for these genes show frequent movement among bacterial strains and species, and this ability to transfer among genomes is confirmed by experiments in which a plasmidharboring strains transmits to another strain.
The experiments with the dissected guts were also ingenious. Some of the physiological mechanisms are not really clear-how does the 3M4N exit the gut, and then how is it excreted?
However, these are details that are not really needed for the main story.
Altogether, I found this to be a very rich paper with abundant experimental data to support the proposed route of detoxification. It was also easy to read. The figures are very nice. It is too bad so much good work is buried in the SI, but I can see that it would be a very long paper if included in the main text.
Thank you for your review and evaluation of our manuscript. The question pointed out by the reviewer "how does the 3M4N exit from the gut" is indeed one of the challenges in future studies.
We expect that some types of transporters may be involved in 3M4N excretion, which would be elucidated by RNA-seq of the gut tissue in conjunction with RNAi screening.
Very minor points: l. 41 "prosperity" is not the right word to use here.
---The term has been replaced by "success" (L41 in the revised manuscript)

Response to Reviewer #3:
In this study Sato and co-authors study the molecular and physiological basis of insecticide detoxification in an insect by its associated bacterial symbiont. The study builds upon several previous studies by this same group that have shown how this symbiont that is acquired from the soil at every generation is responsible for insecticide resistance in the insect. The authors perform a series of genomic analyses that identify the potential genes responsible for such degradation.
The functionality of these genes is verified through gene silencing coupled with biochemistry assays. The authors also demonstrate that while the symbiotic bacteria degrades the insecticide, which is not bactericidal, the product of this degradation is bactericidal so that in return, the host insect immediately excretes it. The authors confirm such excretion by detecting the degradation product in insect faeces. The authors perform these experiments with bacteria both in vitro and in vivo. In vivo experiments are performed by reinoculating the symbiont (and created mutants) in the insect host.
By doing so the authors can test symbiont effects on insect fitness. A surprising result is, however, that the gene expression levels found in vitro and in vivo are quite different, which reveals the important role played by the host in regulating the metabolism of the microbe. The authors also demonstrate that insecticide degradation, which is located in a gene inside a plasmid, can be horizontally transmitted among bacteria. This is important as it shows how insecticide resistance can be transmitted among bacteria and potentially among insect hosts. This part of the study is Insecticide resistance is one of the most serious problems in contemporary agriculture 21 and public health 1,2 . Although recent studies revealed that insect gut symbionts contribute 22 to resistance 3,4 , the symbiont-mediated detoxification process remains unclear. Here we 23 report the in vivo detoxification process of an organophosphorus insecticide, fenitrothion, 24 in the bean bug Riptortus pedestris. Using transcriptomics and reverse genetics, we reveal 25 that gut symbiotic bacteria degrade this insecticide through a horizontally acquired 26 insecticide-degrading enzyme into the non-insecticidal but bactericidal compound 3-27 methyl-4-nitrophenol, which is subsequently excreted by the host insect. This integrated 28 "host-symbiont reciprocal detoxification relay" enables the simultaneous maintenance of 29 symbiosis and efficient insecticide degradation. We also find that the symbiont-mediated 30 detoxification process is analogous to the insect genome-encoded fenitrothion 31 detoxification system present in other insects. Our findings highlight the capacity of 32 symbiosis, combined with horizontal gene transfer in the environment, as a powerful 33 strategy for an insect to instantly eliminate a toxic chemical compound, which could play 34 a critical role in the human-pest arms race. 35 36 37

Introduction 38
Insects live in a world abounding with toxic compounds such as plant toxins and man-39 made pesticides. To overcome these toxins, herbivorous insects have evolved elaborate 40 mechanisms for their detoxification 1,2,[5][6][7] . Toxin resistance has brought insects to success 41 in the terrestrial ecosystem, while at the same time, insecticide resistance is one of the 42 most serious problems in contemporary agriculture and public health 2,8,9 . Although the 43 resistance mechanisms are often encoded by the insects' own genomes, recent studies 44 revealed that in many insects specific gut microorganisms also contribute to toxin 45 resistance by degrading the chemical compounds 3,4 . 46 Many insects possess symbiotic bacteria in the bacteriocytes or the gut, wherein 47 symbionts play pivotal metabolic roles such as the provision of essential amino acids, 48 supplementation of vitamins, and digestion of indigestible food materials, such as plant 49 cell walls 10-12 . In addition to the nutritional contribution, recent studies have revealed that 50 symbiotic bacteria confer also other functions, including heat tolerance, parasite or 51 pathogen resistance, body coloration, as well as toxin degradation 13,14 . However, how 52 host-symbiont metabolic interactions play a role in these different symbiont-mediated 53 ecological traits is poorly investigated. Here we report the in vivo detoxification process 54 of the insecticide fenitrothion by gut symbionts in the bean bug Riptortus pedestris, 55 revealing that a reciprocal host-symbiont detoxification of the insecticide and its 56 bactericidal degradation-product is pivotal to maintain the stable association and thus the 57 efficient detoxification. 58 The bean bug R. pedestris, a serious pest of leguminous crops in Eastern Asia 15 , 59 acquires a specific bacterial symbiont of the genus Burkholderia from the soil every 60 generation and harbors 10 7 -10 8 cells of these bacteria in midgut crypts (Fig. 1a) 16,17 . The 61 Burkholderia symbiont contributes to the recycling of the host's metabolic wastes, which 62 benefits growth and reproduction of the bean bug host 18  The symbiont-mediated resistance is effective for oral and cuticle treatments with MEP 19 . 72 Here, using transcriptomics and reverse genetics, we report the in vivo 73 detoxification process of the insecticide fenitrothion by gut symbionts in the bean bug, 74 revealing that a reciprocal host-symbiont detoxification of the insecticide and its 75 bactericidal degradation-product is pivotal to maintain the stable association thereby 76 enabling efficient detoxification. 77 78 Results and discussion 79

MEP-degradation pathway in the Burkholderia symbiont 80
To clarify the genetic bases of the symbiont-mediated MEP resistance, we first 81  The genomic data suggested that MEP can be assimilated by strain SFA1 via two 97 metabolic pathways branching at 2-methylhydroquinone: the para-nitrophenol reductase 98 (Pnp) pathway and the methylhydroquinone metabolizing enzyme (Mhq) pathway 99 ( Supplementary Fig. 2a). When the symbiont was cultured in a minimal medium 100 containing MEP as the sole carbon source, the Mhq pathway genes, as well as the 101 upstream methyl parathion degrading enzyme (mpd) and pnpAB genes, were highly 102 expressed (Fig. 1d, Supplementary Fig. 2b and c). This indicated that SFA1 assimilates 103 MEP mainly via the Mhq pathway. 104 105 First catabolic step of MEP-degradation is critical and sufficient for the symbiont-106 mediated insecticide resistance 107 We then investigated the expression levels of the MEP-degrading genes in the SFA1 108 bacteria colonizing the midgut crypts of the insect. Unexpectedly, only the first gene, mpd, 109 was highly expressed while all the downstream genes were nearly silent (Fig. 1d,  110 Supplementary Fig. 2b). Moreover, the expression level of the genes was not affected by 111 MEP treatment of the insects. The mpd gene was also highly expressed in bacterial culture 112 grown without MEP (Fig. 1d), indicating that it is constitutively expressed in SFA1. 113 Previous studies on Pseudomonas strains reported that Mpd is a membrane protein and 114 the first step of the MEP degradation occurs in the periplasmic space 24-27 . While MEP is 115 highly toxic to insects, the degradation-product, 3-methyl-4-nitrophenol (3M4N; a 116 yellow-colored phenolic compound) (Fig. 2a), is non-toxic ( Supplementary Fig. 5a). 117 These results strongly suggested that the MEP resistance is conferred exclusively by the 118 expression of mpd in the gut and by the Mpd-mediated conversion of MEP to 3M4N. 119 To confirm this point, the mpd and pnpA genes were deleted and MEP-exposure 120 tests were performed with insects infected with the mutants. PnpA, degrading 3M4N to 121 2-methyl-1,4-benzoquinone, is encoded in strain SFA1 by two homologous genes, pnpA1 122 and pnpA2 (Fig. 1b, d, Supplementary Fig. 2), which were both deleted. These gene 123 deletion mutants, as well as wild type SFA1, grew well in a minimal medium containing 124 glucose as the sole carbon source (Fig. 2b). However, the Δmpd and ΔpnpA2 single 125 mutants, and the ΔpnpA1/ΔpnpA2 double mutant, but not the ΔpnpA1 single mutant, were 126 not able to grow in a minimal medium containing MEP as the sole carbon source (Fig.  127 2c), confirming the involvement of mpd and pnpA2 in the MEP degradation by SFA1. All 128 the deletion mutants colonized well the midgut crypts of the bean bug when insects were 129 reared in the absence of MEP (Fig. 2d). However, when the infected insects were treated 130 with MEP, those harboring the Δmpd mutant showed a significant reduction of the 131 survival rate but not those harboring the ΔpnpA1, ΔpnpA2 or ΔpnpA1/ΔpnpA2 mutants 132 (Fig. 2e). A genetically complemented mutant of mpd, Δmpd/mpd + , restored the growth 133 ability in MEP medium (Fig. 2c) and the ability for conferring MEP resistance in the bean 134 bug host (Fig. 2e). Together, these results demonstrated that (1)

Insecticide-degradation product 3M4N is highly bactericidal 140
Notably, the MEP medium of ΔpnpA2 and ΔpnpA1/ΔpnpA2 cultures became yellow in 141 color due to the accumulation of 3M4N (Fig. 2c). When transferred to MEP medium after 142 culturing in nutrient medium containing citrate as a carbon source, the growth of wild 143 type SFA1 is halting by 10 to 20 hours before resuming ( Supplementary Fig. 5b-d). Since 144 pnpA expression depends on its substrate 3M4N 28,29 , this long growth lag is probably due 145 to the accumulation of 3M4N in the cultures, as suggested by their transient yellowish 146 color (Fig. 2a, Supplementary Fig. 5c), until induction of the pnpA expression 147 ( Supplementary Fig. 5e). Moreover, it could also indicate that this compound is toxic to 148 the symbiotic bacteria. Indeed, when cultured SFA1 cells were incubated either with MEP 149 or 3M4N and subsequently spotted on agar plates, 3M4N but not MEP showed 150 bactericidal activity (Fig. 3a). Our previous study reported that the symbionts in the 151 midgut crypts show a thinner cell wall compared with the cultured bacteria and that they 152 are highly sensitive to various stresses such as detergents, proteases, and cell surface-153 attacking antimicrobial peptides 18 . The spot test revealed that, while MEP is still nontoxic, 154 3M4N shows a drastically higher toxicity to SFA1 cells isolated from midgut crypts than 155 to SFA1 grown in culture ( Fig. 3a and b). 156 Bactericidal product 3M4N is efficiently eliminated by the host from the gut 158 symbiotic organ 159 The imbalanced expression pattern of mpd and pnpA in the midgut crypts, i.e. constitutive 160 mpd expression and little pnpA expression, is expected a priori to cause an accumulation 161 of 3M4N in the midgut, which could lead to a severe effect on the symbiont population. 162 However, the symbiont titer in the insects was not affected by 3M4N feeding (Fig. 3b). 163 Furthermore, even the ΔpnpA1/ΔpnpA2 mutant, which lacked the 3M4N-degrading 164 enzyme and was thus susceptible to the compound (Fig. 2c), was also not affected by 165 3M4N feeding in the midgut crypts (Fig. 3b), strongly suggesting non-accumulation of 166 3M4N in the symbiotic organ. When dissected midguts infected with SFA1 were knotted 167 with a nylon wire at the anterior and posterior portions to avoid bacterial leakage 168 ( Supplementary Fig. 6a) and incubated in a phosphate buffer containing 2 mM MEP, 169 3M4N accumulated in the solution (Fig. 3c). This finding indicates that MEP permeates 170 into the crypt lumen and is degraded by the symbiont and that subsequently its 171 degradation product 3M4N is actively excreted from the midgut to the outside (i.e. the 172 hemolymph in the living insect), even though the transportation system of 3M4N through 173 crypt epithelia remains unclear. The low expression of the 3M4N-inducible pnpA2 gene 174 in the midgut (Fig. 1d) supports this conclusion. The exclusion ability of 3M4N by the 175 midgut was not affected by the presence of trehalose that is known as the major sugar in 176 insect hemolymph (Supplementary Fig. 6b). Although MEP degradation was 177 accomplished in midgut crypts (Fig. 3d), the degradation activity per symbiont cell was 178 approximately ten times lower in the midgut cells than that in cells in culture 179 ( Supplementary Fig. 6c). This could be because (1) the expression activity of mpd or its 180 protein product is low in the gut; (2) the transport systems of MEP and 3M4N in the gut 181 epithelial cells limit the degradation efficiency; or (3) symbiont cells are compacted in 182 the gut crypts, which reduces the available surface area of the symbiont for uptake and 183 also the degradation efficiency. When feces of bean bugs fed with water containing 3M4N 184 were analyzed by LC-MS, the compound was detected in the feces ( Supplementary Fig.  185 7), suggesting that 3M4N is not further metabolized in the bean bug host but excreted in 186 unchanged form. 187 188

Host-symbiont reciprocal detoxification 189
Taken together, we conclude that the symbiont-mediated insecticide detoxification is 190 accomplished by an elaborate host-symbiont collaboration. Symbiotic bacteria actively 191 degrade the insecticidal (but not bactericidal) MEP, and in return, the host insect 192 immediately excretes its bactericidal (but not insecticidal) degradation product 3M4N 193 (Fig. 3d). Hence, this coordinated detoxification enables the maintenance of the 194 symbiosis, and thereby sustains high degradation activity, even under insecticide stress. 195 Host-symbiont metabolic integration, wherein both the host and the bacterial symbiont 196 contribute to metabolic pathways, is a common feature in many symbiotic interactions of 197 plant-sucking and blood-feeding insects with some biosynthetic steps performed by 198 symbiont enzymes and others by host-encoded enzymes 10-12 . This study demonstrates 199 that, in the detoxification symbiosis, the host-symbiont metabolic integration is also 200

pivotal. 201
It should be noted that the symbiont-mediated detoxification process of MEP is 202 remarkably analogous to another known MEP-detoxification mechanism in insects that 203 is mediated by an insect-encoded glutathione S transferase (GST) ( Supplementary Fig.  204 8). Insect GSTs detoxify MEP by conjugating glutathione while the symbiont Mpd 205 hydrolyzes MEP and their respective products, 5-S-glutathionyl-1-methyl-2-nitrobenzene 206 and 3M4N, are subsequently eliminated from the insect cells and tissues. Constitutive 207 high expression of GSTs, reported in MEP-resistant insects 30 , can thus be mimicked 208 functionally and mechanistically in insects lacking such a GST by a MEP-degrading 209 symbiont that is constitutively expressing mpd, without the need of any mutation in the 210 insects' own genomes to evolve a MEP-modifying GST. Another notable point is that gut 211 symbionts can rapidly gain the detoxification ability through horizontal gene transfer. 212 Diverse bacteria encode insecticide-degrading genes on plasmids 3 , and in Burkholderia 213 species, MEP-degrading plasmids can be acquired or lost dynamically, depending on the 214 presence of the insecticide in the environment 21 ( Supplementary Fig. 4). Thus, highly 215 flexible symbiont-mediated detoxification mechanisms may play a critical role in the 216 evolutionary arms race and integrated co-evolution of toxin-mediated relationships such 217 as plant-herbivore and human-pest antagonisms. 218

Insects and bacteria 221
Bean bugs were reared in petri dishes (90 mm in diameter and 20 mm high) at 25°C under 222 a long-day regimen (16 h light, 8 h dark) and fed with soybean seeds and distilled water 223 containing 0.05% ascorbic acid (DWA). Burkholderia symbiont strain SFA1 19 , a MEP-224 degrading strain conferring MEP resistant in the bean bug, and its GFP (green fluorescent 225 protein) labeled derivative, strain SJ586, were used in this study. The symbiont was 226 cultured at 30°C on YG medium (0.5% yeast extract, 0.4% glucose, 0.1% NaCl). The 227 GFP-labelled strain was constructed by the Tn7 mini-transposon system, as previously 228

Preparation of midgut symbiont cells for RNA-seq 278
The oral administration of the symbiont strain SFA1 was performed as described 19,44 . The 279 symbiont was inoculated to 2nd instar nymphs, and three days after molting to the 3rd 280 instar, nymphs were transdermally administrated with 1 µl of 0.2 µM or 20 µM of MEP 281 (dissolved in acetone). One-or three-day after the treatment, insects were dissected and 282 the crypt-bearing symbiotic gut region was subjected to the RNA extraction and RNA-283 seq analysis. As a control, untreated insects were analyzed. 284

RNA-seq analysis 286
Total RNA was extracted from triplicate samples from cultures by the hot phenol method 287 as previously described 45 or from the midgut symbiont cells by using RNAiso Plus 288 (Takara Bi, Kusatsu, Shiga, Japan) and the RNeasy mini kit (Qiagen). The extracted total In total, 21 cDNA libraries were constructed and sequenced by MiSeq (Illumina, Inc., 300 San Diego, CA, USA). To ensure high sequence quality, the remaining sequencing 301 adaptors and the reads with a cutoff Phred score of 15 (for leading and tailing sequences, 302 Phred score of >20) and a length of less than 80 bp in the obtained RNA-seq data were 303 removed by the program Trimmomatic v0.30 using Illumina TruSeq3 adapter sequences 304 for the clipping 47 . The remaining paired reads were analyzed by FastQC 305 were pre-cultured in minimal medium containing 1.0 mM MEP on a gyratory shaker (210 325 rpm) at 30ºC overnight, and then cultured in newly-prepared MEP-containing minimal 326 medium under the same condition. The growth of cultures was estimated by OD600 327 measurements. To confirm the basic growth abilities of the mutants, these bacterial strains 328 were pre-and sub-cultured in minimal medium containing 0.1% glucose under the same 329 conditions. These symbiont strains and mutants were inoculated to the bean bug as

LC-ESI-MS detection of 3M4N in feces from 3M4N-fed insects 375
An insect rearing system for feeding 3M4N and collecting feces is shown in 376 Supplementary Fig. 7. Insects were fed with DW or DW containing 10 mM 3M4N in a 377 plastic container, in which the solution supplier was covered by 0.5 mm mesh so that 378 insects were able to drink the solution by probing with their proboscis but did not directly

Competing interests 587
The authors declare no competing interests.   After 10-20 h delay, bacterial growth was recovered because of the pnpA2 expression was induced by 3M4N. The growth curves of three independent cultures are shown. d, Growth of SFA1 in MEP medium after pre-culture in MEP medium. No growth delay was observed. e, Gene expression of pnpA2 during cultivation with MEP investigated by RNA-seq. Culture conditions: (i) SFA1 was cultured in minimal medium containing citrate and analyzed one hour after MEP addition; (ii) cells were transferred from the citrate medium to MEP medium, wherein growth delay was observed, and analyzed after yellow color of MEP medium started to disappear; (iii) cells were transferred from MEP medium to newly-prepared MEP medium, wherein no growth delay was observed. The values are mean from 2 replicates. The bars denote standard error.    Supplementary Fig. 7 | Detection of 3M4N from feces. a, An insect rearing system for feeding 3M4N and collecting feces.
Insects were fed with DW or DW containing 10 mM 3M4N in a plastic container, where the solution supplier was covered by 0.5 mm mesh so that insects were able to drink the solution by probing with their proboscis but did not directly touch the solution by their legs or body. b, Color of isolated feces. Feces isolated from the 3M4N-fed group are more yellowish in color.
Feces were dissolved in DW adjusted to pH 8.0. c, Chromatogram of the feces extracts analyzed by HPLC. Monitoring of absorbance at 400 nm, which corresponds to 3M4N specific absorbance, detected a clear peak at 14.2 min only in the 3M4N-fed group. Triplicate chromatograms of the DW-and 3M4N-fed groups are stacked. d, Mass spectrum of the HPLC 14.2 min fraction in negative mode. The m/z of 151.57 is identical to that of 3M4N standard (m/z = 151.53). The mass and other spectral data of 3M4N standard are available from Spectral Database for Organic Compounds (https://sdbs.db.aist.go.jp/sdbs/cgibin/direct_frame_top.cgi). e, UV-V is absorbance spectrum of 1 mM 3M4N. A large peak was observed near 390 nm.