Symbiosis, virulence and natural-product biosynthesis in entomopathogenic bacteria are regulated by a small RNA

Photorhabdus and Xenorhabdus species have mutualistic associations with nematodes and an entomopathogenic stage1,2 in their life cycles. In both stages, numerous specialized metabolites are produced that have roles in symbiosis and virulence3,4. Although regulators have been implicated in the regulation of these specialized metabolites3,4, how small regulatory RNAs (sRNAs) are involved in this process is not clear. Here, we show that the Hfq-dependent sRNA, ArcZ, is required for specialized metabolite production in Photorhabdus and Xenorhabdus. We discovered that ArcZ directly base-pairs with the mRNA encoding HexA, which represses the expression of specialized metabolite gene clusters. In addition to specialized metabolite genes, we show that the ArcZ regulon affects approximately 15% of all transcripts in Photorhabdus and Xenorhabdus. Thus, the ArcZ sRNA is crucial for specialized metabolite production in Photorhabdus and Xenorhabdus species and could become a useful tool for metabolic engineering and identification of commercially relevant natural products. A small RNA regulates the production of metabolites with roles in symbiosis and virulence in Photorhabdus and Xenorhabdus.

stable short form (approximately 50 nucleotides) [9][10][11] . The processed short form of ArcZ activates rpoS translation directly and inhibits the expression of several other genes 11,12 . In Escherichia coli, the expression of arcZ is repressed by the ArcA-ArcB two-component system under anaerobic conditions. In a negative feedback loop, arcZ represses and is repressed by arcB transcription 11 . Although there is a wealth of research on ArcZ in E. coli and Salmonella [9][10][11] , its function in other bacteria remains unclear.
Specialized metabolites (SMs) in bacteria are often responsible for ecologically important activities 13 . In the case of Xenorhabdus and Photorhabdus, SMs play an essential role in cross-kingdom interactions with nematodes, various insects as well as bacterial and fungal species competing for the same food source 14 . Our earlier work on Photorhabdus showed that the deletion of hfq results in severe perturbation of gene networks, including several key regulators 4 . This leads to an overall decrease in SM production and a failure of the bacteria to support their obligate symbiosis with nematodes. Despite SMs playing a central role in the life cycle of the symbiosis, the exact ecological functions of many of these compounds has remained unknown. Notable advances towards the identification of the bioactivities of many of the SMs-with assigned functions including cell-cell communication (photopyrones and dialkylresorcinols 15,16 ), nematode development (isopropylstilbene 17 ), defence against food competitors (isopropylstilbene and rhabdopeptides 17,18 ) or insect pathogenicity (rhabduscin, rhabdopeptides and glidobactin [18][19][20] )-have been made in recent years. However, understanding the full potential of SMs in these bacteria is still hampered by a somewhat limited understanding of when individual SMs are produced and their regulation in general. The regulators Hfq, HexA (also LrhA), LeuO and Lrp have so far been implicated in the regulation of SMs in Photorhabdus and Xenorhabdus 3,4,21,22 . Deletion of hfq in Photorhabdus results in complex regulatory changes, including a strong upregulation of HexA, a known repressor of SM production 4 . Consequently, SM production is completely abolished in this strain and nematode development was severely restricted.
Given the overlapping life cycles and niche occupation, we hypothesized that the deletion of hfq in Xenorhabdus would have  Table 1). We investigated the Hfq binding partners to further elucidate the mechanism of SM regulation. To this end, we sequenced both X. szentirmaii and Photorhabdus laumondii using a sRNA sequencing protocol and combined this with Cappable-seq data to globally annotate the transcriptional start sites belonging to coding sequences or potential previously undescribed sRNAs (Supplementary Note 2 and  Supplementary Table 2). We confirmed the expression of several of these sRNAs by northern blot analysis (Extended Data Figs. 1 and 2). To identify RNA-protein interactions on a global scale, we next employed RNA immunoprecipitation, followed by high-throughput sequencing (RIP-seq) using chromosomally produced Hfq 3×FLAG protein as bait. We performed these experiments at two different cell densities (that is, at an optical density (OD) 600 of 0.5 and 5.0; see Supplementary Table 3 for a full list of the ODs from different experiments). From the corresponding sequencing data, we first identified regions of 5 base pairs (bp) or more that were enriched in our tagged Hfq strain (Methods). We then searched for sRNAs that were specifically enriched in the tagged samples compared with the untagged samples. We identified a total of 37 binding  10  -35   10  20  30  40  50  60  70  80  90  100  110  120  130 -

NATuRE MIcRobIology
sites in annotated sRNAs (35 unique sRNAs) at an OD 600 of 0.5 and 37 binding sites (34 unique) at an OD 600 of 5.0 that were enriched by at least threefold in both replicates (Fig. 1b and Supplementary  Table 4). During early exponential growth, 11 sRNAs (of 35) that have been described to associate with Hfq in other species were identified, whereas 10 (of 34) are known from those that were enriched at an OD 600 of 5.0. As a second step, we examined the mRNAs enriched in the data. At an OD 600 of 5.0, 402 mRNAs and 32 annotated 5′ untranslated regions (UTRs) were identified to associate with Hfq. A total of 1,003 mRNAs and 29 5′ UTRs were detected at an OD 600 of 0.5, (Fig. 1c and Supplementary Table 5). We hypothesized that the Hfq RIP-seq analysis performed would allow us to identify key sRNAs involved in SM repression. However, our analysis identified >50 potential Hfq-binding sRNAs (across both ODs; Fig. 1b). We therefore constructed a transposon-mutant library using pSAM-BT_Kan instead of deleting each sRNA individually (Methods and Supplementary Note 3) and searched for phenotypes consistent with that of the strain from which hfq had been deleted (∆hfq). The red colour afforded to the bacteria by anthraquinone production makes the strain especially suitable for transposon mutagenesis when screening for mutants defective in SM biosynthesis. We screened approximately 60,000 clones for obvious phenotypic alterations. Several mutants were defective in some facets of SM production and showed growth defects (Supplementary Note 3, Extended Data Fig. 3 and Supplementary Table 6); however, only one displayed the desired phenotype. Re-sequencing of this strain, followed by read mapping revealed that the transposon was inserted within an intergenic region associated with the arcZ sRNA gene ( Supplementary Fig. 1).
ArcZ is a well-known Hfq-associated sRNA, which was also present in our list of Hfq-bound sRNAs in P. laumondii ( Fig. 1b and Supplementary Table 4). To verify that the observed phenotype was derived from the transposon insertion, we generated a ∆arcZ mutant by deleting the major part of the sRNA ( Supplementary Fig. 1) and a complemented strain by reintroducing an intact version of arcZ at the original locus. Northern blot analysis was performed to verify the absence of ArcZ in the deletion mutant and the presence of ArcZ in the WT and the complementation mutant (Fig. 2a). Sequencing of the ∆arcZ-mutant RNA showed severe transcriptomic changes compared with the WT and ∆arcZ::arcZ mutant of P. laumondii, reminiscent of that seen in P. laumondii ∆hfq (Supplementary Note 4, Supplementary Tables 7,8 and Supplementary Fig. 2). The SM production titres in the ∆arcZ strain were strongly decreased, similar to that seen in the transposon-insertion mutant, and the complementation strain restored SM production ( Fig. 2c-h).
To corroborate the role of ArcZ in SM production, the ArcZ mRNA targets were predicted using CopraRNA 23 (Supplementary  Tables 9 and 10). One hit warranting further investigation was hexA (lrhA), which was previously identified as a gene that is highly upregulated in our strains and represses SM production in both P. laumondii 24 and Xenorhabdus 3 . CopraRNA predicted a 9-bp-long RNA duplex involving the 5′ UTR of hexA and the processed isoform of ArcZ (Fig. 3a). This base-pairing is reminiscent of previously reported ArcZ targets in other bacteria requiring the RNase E-mediated release of the seed region of the sRNA 12 . We also identified a corresponding enriched RNA sequence upstream of the hexA coding sequence at an OD 600 of 0.5 in the RIP-seq experiments ( Fig. 1c and Supplementary Fig. 3). We hypothesized that, through Hfq, ArcZ might bind to the hexA transcript leading to repression of HexA, and that Hfq and ArcZ prevent HexA production in laboratory cultures in which SMs are produced, thereby allowing the strain to synthesize SM. However, we expected that hexA would no longer be repressed if either hfq or arcZ were deleted, resulting in severely reduced SM production. To test this idea, we altered the predicted site of the ArcZ-hexA interaction to a PacI restriction site (TTAATTAA) and created a knock-in of hexA with the modified sequence in a ∆hexA strain ( Supplementary Fig. 4a,b). We predicted that the knock-in of hexA with an altered 5′ UTR would result in a failure of ArcZ to bind, leading to a reduced SM titre. The SM-production titres of the knock-in mutant with the altered binding site upstream of hexA were indeed greatly reduced ( Fig. 2c-h).
To verify the proposed interaction region, we conducted a compensatory base-mutation study in E. coli. The fifth base pair of the proposed interaction region was exchanged in the arcZ sequence, the hexA 5′ UTR or both using site-directed mutagenesis (Fig. 3a). The hexA 5′ UTR sequence was fused to the green fluorescent protein (gfp) gene. The GFP output was measured to determine the efficiency of inhibition (Fig. 3b,c). For the control, the GFP signal derived from the expression of hexA::gfp was measured and set to one. When p-arcZ was expressed together with hexA::gfp, HexA repression was increased 32-fold compared with the control. In addition, when p-arcZ* (G79C) was expressed, ArcZ* was no longer able to repress HexA. HexA repression was only slightly increased compared with the control for hexA*::gfp (C46G) in combination with the native ArcZ, suggesting that ArcZ can still bind to the 5′ UTR of hexA but with substantially reduced efficiency. When p-arcZ* (G79C) was combined with hexA*::gfp (C46G), HexA::GFP repression was increased 39-fold, which confirms our hypothesis that ArcZ binds to the 5′ UTR of hexA to repress HexA production. Notably, this base-pairing sequence is located approximately 50 nucleotides upstream of the hexA translational start site ( Fig. 3a) and thus ArcZ binding is unlikely to compete with recognition of the mRNA by 30S ribosomes 25 . Instead, alignment of the P. laumondii hexA 5′ UTR revealed that the ArcZ binding site is CA-rich and highly conserved among other SM-producing bacteria ( Supplementary Fig. 5). CA-rich sequences located in proximity to translation initiation sites are well-known translational enhancers and sequestration of these regulatory elements by sRNAs has been reported to downregulate gene expression 26,27 , which might also be relevant for the ArcZ-hexA interaction reported here. In addition, we conducted a proteomic analysis with the wild-type (WT), ∆arcZ, ∆hfq and ∆hexA::hexA_PacI_UTR strains of P. laumondii. We used label-free quantification of quadruplicate samples to determine the HexA abundancy in each strain. The levels of HexA were substantially elevated in all mutant strains (11.8-to 22.7-fold; Supplementary Table 11) compared with the WT, further supporting this mechanism of regulation for SM production.
The arcZ gene and its genomic organization are highly conserved among enterobacterial species 10 (Fig. 1a). As the control of SMs in Photorhabdus relays a fundamental ability for these bacteria to occupy their specific niche, we investigated the possibility that the same mechanism occurs in the closely related Xenorhabdus. Given the SM reduction in X. szentirmaii ∆hfq, we constructed a ∆arcZ mutant in X. szentirmaii in a similar fashion to P. laumondii by deleting 90 bp of the predicted arcZ sequence. Using northern blots, we verified that ArcZ was no longer produced by the deletion mutant and that complementation of the deletion led to resumed production of ArcZ (Fig. 3d). We subsequently investigated the transcriptome and SM profile of the WT as well as the deletion and complementation mutants ( Fig. 3e and Supplementary Table 12). Consistent with P. laumondii, deletion of arcZ resulted in a global effect on the transcriptome as well as severe reductions in the SM titre, both of which were complemented in the ∆arcZ::arcZ complementation mutant (Figs. 3e and 4).
Our results highlight the critical role of ArcZ in regulating specialized metabolism in these strains. In fact, the critical nature of the SMs from Photorhabdus and Xenorhabdus in modulating the insect immune response indicated that ArcZ might be required for niche occupation by these bacteria. We observed an inability of the P. laumondii ∆arcZ strain to support nematode development (Extended Data Fig. 4), consistent with our earlier observations for the ∆hfq mutant 4 . However, the same was not seen with X. szentirmaii. :arcZ, ∆arcZ, ∆arcZ::arcZ, ∆hexA, ∆hexA::hexA_PacI and ∆hexA::hexA. All bars represent the relative production in comparison to the WT. Data are presented as the mean ± s.e.m. The dots represent biologically independent replicates (n = 3). *P < 0.05, **P < 0.005, ***P < 0.0005 and ****P < 0.0001. Statistical significance of the relative production compared with the WT production levels was calculated using a two-sided unpaired t-test.

NATuRE MIcRobIology
We suspect this might be because of the observed increase in protoporphyrin IX (PPIX) production in the X. szentirmaii ∆arcZ strain (Supplementary Note 5). PPIX is a precursor of haem, which is an important cofactor for key biological processes such as oxidative metabolism 28 , protein translation 29 , maintenance of protein stability 30 and many others. However, PPIX cannot be synthesized de novo by Caenorhabditis elegans and other nematodes 31 . The nematodes therefore rely on external PPIX sources (such as from symbiotic bacteria), which positively affects their growth, reproduction and development 32 . It is interesting that, despite P. laumondii also being capable of producing PPIX, reproduction of the Heterorhabditis nematode was not supported in the ∆arcZ mutant or ∆hfq strain. This is possibly indicative of the nematode-specific requirements for reproduction, which may also include isopropylstilbene as an essential factor in Heterorhabditis 17 , where no analogous compound is known to be required for Steinernema. Nearly all of the SM-related genes analysed in both Xenorhabdus and Photorhabdus were found to be downregulated in the ∆arcZ mutant, in accordance with impaired SM production (Fig. 4a,b). This provides a chemical background that is devoid of natural products, which allows for the isolation and identification of a desired compound due to the absence of compounds with similar retention

NATuRE MIcRobIology
times. Therefore, ∆arcZ mutants could offer a powerful tool for the (over-)production and identification of previously undescribed natural products. As a proof of concept, we conducted a promotor exchange in front of gxpS in both X. szentirmaii ∆arcZ and X. szentirmaii ∆hfq, and compared GameXPeptide (GXP)-C production after induction to the WT (Fig. 4c). The production of GXP-C was found to be increased 90.4-fold (±4.7) in X. szentirmaii ∆arcZ::pCEP_GxpS and increased 138.6-fold (±17.1) in X. szentirmaii ∆hfq::pCEP_GxpS compared with the WT (Fig. 4c). The striking increase in production as well as the dramatically reduced chemical background in both strains highlights the potential for exploiting this regulatory cascade for selective SM production in a strain that is well-suited for natural product detection. We recently showed that this strategy could be applied in a high-throughput manner for rapid screening of bioactivities 33 . The same strategy used here in a ∆arcZ strain demonstrates an alternative route to activation, without the complex perturbations associated with deleting the major RNA chaperone in these bacteria. Interestingly, some comparisons between these mechanisms can be drawn in other SM-producing Enterobacteriaceae (Fig. 1a). Erwinia is a genus of plant-pathogenic bacteria that produce SMs, where Hfq and ArcZ have both been implicated in virulence 34 , whereas HexA is a negative regulator of secondary metabolites in these bacteria 35 . Similar parallels can also be seen in Serratia [36][37][38] and Pseudomonas 39 , two other prolific SM producers. Although further investigations will be required to ascertain whether these apparent similarities represent identical mechanisms, the conserved nature of ArcZ in other SM-producing Enterobacteriaceae could suggest that this strategy may yield fresh avenues for rapid investigation into SM biosynthesis in other taxa.

Methods
Bacterial culture conditions. All Photorhabdus and Xenorhabdus strains were cultured in Luria-Bertani (LB) medium for at least 16 h with shaking at 30 °C. The E. coli strains were cultured in LB for at least 16 h with shaking at 37 °C. The medium was supplemented with chloramphenicol (34 µg ml −1 ), ampicillin (100 µg ml −1 ), rifampicin (50 µg ml −1 ) or kanamycin (50 µg ml −1 ) when appropriate. Promotor-exchange mutants were induced by adding l-arabinose (2%, vol/vol) to the cultures. All of the plasmids and strains used in this study are listed in Supplementary Tables 13 and 14.
Nematode bioassays. All nematodes were cultivated in Galleria mellonella and collected on white traps as previously described 40 . The nematode bioassays were also performed as described elsewhere 4 .

Creation of a transposon-mutant library.
For the transposon mutagenesis, the plasmid pSAM_Kan (containing the mariner transposon) was constructed using pSAM_BT 41 as a template. To do this, the plasmid was linearized using the primers NN191 and NN192. The kanamycin-resistance cassette was amplified from the pCOLA_ara_tacI plasmid using the primers NN193 and NN194, introducing complementary overhangs to pSAM_BT at both ends of the PCR fragment.

NATuRE MIcRobIology
The kanamycin-resistance cassette was fused with the linearized pSAM_BT plasmid using Hot Fusion cloning, thereby replacing the erythromycin-resistance cassette with kanamycin resistance. E. coli ST18 was transformed with the plasmid pSAM_Kan and further used for the creation of the transposon-mutant library of P. laumondii TTO1 through conjugation. Transposon-insertion mutants were selected on LB agar containing kanamycin. All primer sequences are listed in Supplementary Table 15.

Construction of mutant strains.
For the deletion of the majority of ArcZ in P. laumondii TTO1, a 1,123-bp upstream and a 1,014-bp downstream product were amplified using the primers NN276 and NN277, and NN278 and NN279, respectively. The PCR products were fused using the complementary overhangs introduced by the primers and cloned into the pEB17 plasmid linearized with PstI and BglII. The resulting plasmid was used for transformation of E. coli s17-1 λpir. Conjugation of the plasmid in the P. laumondii strains and generation of deletion strains by homologous recombination through counter selection was done as previously described 42 . The deletion mutants were verified by PCR using the primers NN281 and NN282, which yielded a 632-bp fragment for mutants genetically equal to the WT and a 502-bp fragment for the desired deletion mutant. Complementation of the ArcZ deletion was achieved by inserting the full and intact version of ArcZ at the original locus. To do this, a 2,207-bp PCR product including the upstream and downstream region required for homologous recombination and the full-length ArcZ was amplified using the primers NN276 and NN279. The fragment was cloned into pEB17 as described earlier. The verified plasmid construct was used to transform E. coli s17-1 λpir cells. The plasmid was transferred into P. laumondii ∆arcZ by conjugation and integrated into the genome of P. laumondii ∆arcZ by homologous recombination. The knock-in mutant was generated by a second homologous recombination through counter selection on LB plates containing 6% sucrose. Knock-in mutants were verified by PCR using the primers NN281 and NN282, yielding a 632-bp fragment. The same strategy was used for the construction of the mutant strains in X. szentirmaii. To generate the promotor-exchange mutants in front of gxpS, the plasmid pCEPKMR_ORF00346 was transferred into X. szentirmaii ∆arcZ and X. szentirmaii ∆hfq by conjugation and integrated into the genome by homologous recombination. DNA extraction. Genomic DNA was extracted using the Gentra Puregene yeast/ bact. kit (Qiagen) following the manufacturer's instructions. For sequencing of transposon-insertion mutants, genomic DNA was extracted using the DNeasy blood and tissue kit (Qiagen).

DNA sequencing and identification of the transposon-insertion site.
DNA isolated from the transposon-insertion mutants was sequenced on an Illumina NextSeq platform. DNA libraries were constructed using the Nextera XT DNA preparation kit (Illumina) and whole-genome sequencing was performed using 2 × 150 bp paired-end chemistry. A sequencing depth of >50× was targeted for each sample. Genomes were assembled using SPAdes (v. 3.10.1) 43 and annotated using Prokka v. 1.12 (ref. 44 ). The completed genome sequences were analysed and viewed using Geneious v. 6 and 9.1 (https://www.geneious.com).

RNA extraction, sequencing and analysis.
Pre-cultures of P. laumondii TTO1, X. szentirmaii DSM16338 and their respective ArcZ deletion and knock-in mutants were cultured overnight in LB broth with shaking at 30 °C. The pre-cultures were used the following day to inoculate fresh LB at an OD 600 of 0.3. The cells were cultured to the mid-exponential phase (the OD values for each experiment can be found in Supplementary Table 3). RNA was extracted using an RNeasy mini kit (Qiagen) following the manufacturer's instructions. To facilitate cell lysis, the cells were pelleted and snap-frozen in liquid nitrogen for 1 min after removing the supernatant. After thawing and resuspending in lysis buffer, the cells were vortexed for 30 s before proceeding with the protocol. The RNA for the sRNA libraries were extracted in duplicate during the mid-exponential phase for P. laumondii TTO1 and X. szentirmaii.
The RNA was sequenced through 150 bp paired-end sequencing by Novogene following rRNA depletion with a RiboZero kit and library preparation following the Illumina protocol for strand-specific libraries. The raw data were trimmed using Trimmomatic 45 and mapped to the reference genome downloaded from NCBI (NC_005126.1 for P. laumondii and NZ_NIBV00000000.1 for X. szentirmaii) using bowtie2 (v2.3.4.3) 46 . The resulting .sam files were converted to .bam files using samtools (v1.8) 47 and featureCounts (a part of the subread package) 48 was used to count reads mapping to annotated genes. The count files were then uploaded to degust (http://degust.erc.monash.edu/) and analysed using the voom/ limma method of normalization. Only genes with an absolute fold change >2 and false discovery rate < 0.01 were considered significantly regulated. Statistical analysis was performed in R (v. 3.6.1) on the degust platform, where the exact code is available to view.

Northern blot analysis.
For the northern blot analyses, total RNA was prepared and analysed as described previously 49 . Briefly, the RNA samples were separated on 6% polyacrylamide and 7 M urea gels and transferred to Hybond-XL membranes (GE Healthcare) by electro-blotting. The membranes were hybridized in Roti-Hybri-Quick buffer (Roth) at 42 °C with gene-specific [ 32 P] end-labelled DNA oligonucleotides, and washed in three subsequent steps with 5×, 1× and 0.5×SSC (in 0.1% SDS) wash buffer. Signals were visualized on a Typhoon FLA 7000 phosphorimager (FUJIFILM). The oligonucleotides used for the northern blot analyses are listed in Supplementary Table 15.
Compensatory base mutation and GFP fluorescence assay. The plasmids pMH078 and pMH079 were generated using Gibson assembly 50 . For plasmid pMH078, the arcZ gene was amplified using P. laumondii TTO1 genomic DNA with the oligonucleotides KPO-6147 and KPO-6148, and fused into a pEVS143 vector backbone 51 , linearized with KPO-0092 and KPO-1397. To construct the plasmid pMH079, the 5′ UTR and the first 20 amino acids of hexA were amplified using P. laumondii TTO1 genomic DNA with KPO-6145 and KPO-6146, and the pXG10-gfp vector 52 was linearized with KPO-1702 and KPO-1703. The plasmids pMH078 and pMH079 served as templates to insert single point mutations in the arcZ gene as well as the hexA 5′ UTR using site-directed mutagenesis and the oligonucleotide combinations KPO-6156 and KPO-6157, and KPO-6164 and KPO-6165, respectively, yielding the plasmids pMH080 and pMH081.
Target regulation using GFP reporter fusions was analysed as described previously 52 . E. coli Top10 cells were cultured overnight in LB medium (37 °C with shaking at 200 r.p.m.). Three independent cultures were used for each strain. The cells were washed in PBS and the GFP fluorescence intensity was determined using a Spark 10M plate reader (TECAN). Samples that did not express fluorescence proteins were used as controls to subtract the background fluorescence.
Cappable-seq analysis. The Cappable-seq was performed as previously described 53 by Vertis Biotechnologies. The raw sequences were trimmed using Trimmomatic 45 and mapped using bowtie2 (ref. 46 ) to NC_005126.1 for P. laumondii and NZ_NIBV00000000.1 for X. szentirmaii. The transcriptional start sites were detected using the built-in TSS detection function read of Xplorer (v2.2.3) 54 with the following settings: use only single perfect matches, minimum number of read starts = 100, minimum percent coverage increase = 750, detect previously undescribed transcripts, minimum transcript extension = 40, maximum distance to feature of leaderless transcripts = 5 and associate neighbouring TSS within 3 bp.

RIP-seq analysis.
Overnight cultures of P. laumondii TTO1 (WT and Hfq 3×FLAG ) were inoculated into fresh LB medium in duplicate and cultured at 30 °C with shaking at 200 r.p.m. The bacteria were harvested by centrifugation at 4,000 r.p.m. for 15 min at 4 °C when the cells reached OD 600 = 0.5 and OD 600 = 5.0. The cell pellets were resuspended in 1 ml lysis buffer (20 mM Tris pH 8.0, 150 mM KCl, 1 mM MgCl 2 and 1 mM dithiothreitol) and pelleted again by centrifugation (5 min, 11,200g, 4 °C). The supernatants were discarded and the pellets were snap-frozen in liquid nitrogen. After thawing on ice, the cells were resuspended in 800 µl lysis buffer and transferred into tubes containing 300 µl glass beads to break the cells using a Bead Ruptor (2 × 150 s, with a 2 min break on ice in between). After a short centrifugation (15,000g, 4 °C), the lysates were transferred into fresh pre-cooled tubes and centrifuged for 30 min at 15,200g at 4 °C. The cleared lysates were transferred into fresh tubes and incubated with 35 µl FLAG-antibody (monoclonal anti-FLAG M2; Sigma, F1804) with rotation for 45 min at 4 °C, followed by the addition of 75 µl Protein G Sepharose (Sigma, P3296) and rotation for 45 min at 4 °C again. After five wash steps with lysis buffer (by inverting the tube gently and centrifuging for 4 min at 4 °C), the samples were subjected to RNA and protein separation using phen ol:chloroform:isoamylalcohol (25:24:1, pH 4.5; Roth) extraction. The upper phase (approximately 500 µl) was transferred into a fresh tube and precipitated overnight at −20 °C with 1.5 ml EtOH:Na(acetate) (30:1) and 1.5 µl GlycoBlue (Ambion, AM9516). After centrifugation at 11,200 r.p.m. for 30 min at 4 °C, the RNA pellets were washed with 500 µl 70% EtOH, dried and resuspended in 15.5 µl nuclease-free H 2 O. The RNA was treated with 2 µl DNase I, 0.5 µl RNase inhibitor and 2 µl 10×DNase buffer at 37 °C for 30 min. The samples were then supplemented with 100 µl H 2 O and again subjected to a phenol:chloroform:isoam ylalcohol extraction. The upper phase (approximately 120 µl) was transferred to a fresh tube with the addition of 2.5-3 volumes (about 350 µl) of EtOH:Na(acetate) (30:1) and stored at −20 °C overnight for RNA precipitation. The RNA pellets were harvested via centrifugation at 13,000 r.p.m. for 30 min at 4 °C, and washed with 500 µl of 70% EtOH, dried and resuspended in nuclease-free H 2 O. Complementary DNA libraries were prepared using a NEBNext small RNA library prep set for Illumina (NEB, E7300S) according to the manufacturer's instructions and sequenced on a HiSeq 1500 system in single-read mode with a read length of 100 nucleotides.
For RIP-seq analysis, the enriched-control sample pairs were normalized to the number of raw reads present after trimming. Depth counts of all samples were obtained using samtools (v1.8) 47 . Only nucleotide positions with a depth of at least 50 reads in the enriched samples were taken for further analysis. The corresponding depth in the unenriched samples was matched for each nucleotide. A region was considered to be enriched if the enrichment factor was at least three and the corresponding 'enriched' nucleotide was present in both sample pairs. Finally, we considered a region to be enriched if more than five consecutive nucleotides were identified as enriched.
ArcZ-binding prediction. ArcZ from E. coli was used to define the boundaries of ArcZ in Xenorhabdus and Photorhabdus. We then took our annotated ArcZ sequence together with several ArcZ homologues from other Enterobacteriaceae (listed in Supplementary Table 9) and used the online CopraRNA tool 23 , a part of the Freiburg RNA tools suite 55 , with default parameters.

Metabolite extraction and HPLC-MS/MS analysis.
Fresh LB medium (10 ml) was inoculated with an overnight culture to an OD 60 of 0.1. After 72 h of cultivation at 30 °C with shaking, 1 ml of the culture was removed from the culture, centrifuged for 20 min at 13,300 r.p.m. and the supernatant was directly subjected for analysis by HPLC-MS/MS using a Dionex Ultimate 3000 system with a Bruker AmaZon X mass spectrometer. The peak areas of the compounds were quantified using TargetAnalysis 1.3 (Bruker). All of the analysed compounds are listed in Supplementary Table 16.
Proteome analysis. The details of the proteomics procedure were previously published 56 . In short, to extract proteins from P. laumondii frozen cell pellets, 300 µl lysis buffer (0.5% Na-desoxycholate in 100 mM NH 4 HCO 3 ) was added to the cell pellet and incubated at 95 °C for 10 min. The protein concentration in the supernatant was determined using a BCA protein assay kit (Thermo Fisher, 23252). Reduction and alkylation was performed at 95 °C using 5 mM TCEP and 10 mM chloroacetamide for 15 min. The protein (50 µg) was transferred to fresh reaction tubes and protein digestion was carried out overnight at 30 °C with 1 µg trypsin (Promega). After the digestion, the peptides were desalted using CHROMABOND spin columns (Macherey-Nagel) that were conditioned with 500 µl acetonitrile and equilibrated with 500 µl and 150 µl 0.1% trifluoroacetic acid (TFA). After loading, the peptides were washed with 500 µl of 0. The HPLC-MS/MS analysis including label-free quantification was carried out as previously described 56 , with minor modifications.
HPLC-MS/MS analysis of protein digests was performed on a Q-Exactive Plus mass spectrometer connected to an electrospray ion source (Thermo Fisher Scientific). Peptide separation was carried out using an Ultimate 3000 nanoLC-system (Thermo Fisher Scientific) equipped with C18 resin column packed in-house (Magic C18 AQ 2.4 µm; Dr. Maisch). The peptides were first loaded onto a C18 pre-column (pre-concentration set-up) and then eluted in backflush mode with a gradient from 98% solvent A (0.15% formic acid) and 2% solvent B (99.85% acetonitrile and 0.15% formic acid) to 35% solvent B over 30 min. Label-free quantification was done using Progenesis QI software (Nonlinear Dynamics, v2.0) and the MS/MS search was performed in MASCOT (v2.5, Matrix Science) against the UniProt P. laumondii protein database. The following search parameters were used: full tryptic search with two missed cleavage sites, 10 ppm MS1 and 0.02 Da fragment ion tolerance. Carbamidomethylation (C) as fixed, and oxidation (M) and deamidation (N,Q) as variable modifications. The progenesis outputs were further processed using SafeQuant 57 .
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
All .mzXML files from the HPLC-MS/MS runs are available at MassIVE (https:// massive.ucsd.edu) under the ID MSV000084163. Raw sequence data are available at the European nucleotide archive (https://www.ebi.ac.uk/ena/) under project accession numbers PRJEB33827 and PRJEB24159. The proteomic data can be accessed at PRIDE (https://www.ebi.ac.uk/pride/) with the project accession number PXD019095. Source data are provided with this paper. Infective juvenile development to hermaphrodites with strains of P. laumondii and X. szentirmaii. Data are presented as the mean ± s.e.m. Dots represent biologically independent replicates (n = 10). Asterisks indicate statistical significance (*P < 0.05, **P < 0.005, ***P < 0.0005, ****P < 0.00005) of relative recovery compared to WT recovery levels. Statistical significances were calculated using a two-sided unpaired t-test. Exact p values (left to right, respectively) for P. laumondii TTO1 correspond to p = <0.0001, 0.0006 and for X. szentirmaii to P = 0. 56

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated Our web collection on statistics for biologists contains articles on many of the points above.

Software and code
Policy information about availability of computer code Data collection During HPLC/MS anaylsis, Chromeleon Chromatography Data System (CDS) was used to collect HPLC data and HyStar (v 3.2) for MS data.
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A list of figures that have associated raw data -A description of any restrictions on data availability All .mzXML files from HPLC-MS runs are available at MassIVE (https://massive.ucsd.edu) under the ID MSV000084163. Raw sequence data is available at the European nucleotide archive (https://www.ebi.ac.uk/ena/) under project accession numbers PRJEB33827 and PRJEB24159. The proteomic data can be accessed at PRIDE (https://www.ebi.ac.uk/pride/) with the project accession number PXD019095. Source Data are provided with this paper.