Lysis Cassette-Mediated Exoprotein Release in Yersinia entomophaga Is Controlled by a PhoB-Like Regulator

While theoretical models exist, there is not yet any empirical data that links ALC phage-like lysis cassettes with the release of large macro-molecular toxin complexes, such as Yen-Tc in Gram-negative bacteria. In this study, we demonstrate that the novel Y. entomophaga RoeA activates the production of exoproteins (including Yen-Tc) and the ALC at the transcriptional level. ABSTRACT Secretion of exoproteins is a key component of bacterial virulence, and is tightly regulated in response to environmental stimuli and host-dependent signals. The entomopathogenic bacterium Yersinia entomophaga MH96 produces a wide range of exoproteins including its main virulence factor, the 2.46 MDa insecticidal Yen-Tc toxin complex. Previously, a high-throughput transposon-based screening assay identified the region of exoprotein release (YeRER) as essential to exoprotein release in MH96. This study defines the role of the YeRER associated ambiguous holin/endolysin-based lysis cluster (ALC) and the novel RoeA regulator in the regulation and release of exoproteins in MH96. A mutation in the ambiguous lysis cassette (ALC) region abolished exoprotein release and caused cell elongation, a phenotype able to be restored through trans-complementation with an intact ALC region. Endogenous ALC did not impact cell growth of the wild type, while artificial expression of an optimized ALC caused cell lysis. Using HolA-sfGFP and Rz1-sfGFP reporters, Rz1 expression was observed in all cells while HolA expression was limited to a small proportion of cells, which increased over time. Transcriptomic assessments found expression of the genes encoding the prominent exoproteins, including the Yen-Tc, was reduced in the roeA mutant and identified a 220 ncRNA of the YeRER intergenic region that, when trans complemented in the wildtype, abolished exoprotein release. A model for Y. entomophaga mediated exoprotein regulation and release is proposed. IMPORTANCE While theoretical models exist, there is not yet any empirical data that links ALC phage-like lysis cassettes with the release of large macro-molecular toxin complexes, such as Yen-Tc in Gram-negative bacteria. In this study, we demonstrate that the novel Y. entomophaga RoeA activates the production of exoproteins (including Yen-Tc) and the ALC at the transcriptional level. The translation of the ALC holin is confined to a subpopulation of cells that then lyse over time, indicative of a complex hierarchical regulatory network. The presence of an orthologous RoeA and a HolA like holin 5′ of an eCIS Afp element in Pseudomonas chlororaphis, combined with the presented data, suggests a shared mechanism is required for the release of some large macromolecular protein assemblies, such as the Yen-Tc, and further supports classification of phage-like lysis clusters as type 10 secretion systems.

and, therefore, do not form an anti-holin to regulate cell lysis, necessitating an alternate regulatory mechanism for cell lysis. Uniquely to strains in which the lysis cassette is not co-located with the Tc, the RoeA-like regulator is found adjacent to the lysis cassette (Fig. 1A). Phylogenetic assessment of HolA, PepB, and RoeA identified a high conservation of HolA and PepB within the Yersinia genus (Fig. S1), but RoeA orthologs rarely found in the Yersiniaceae, but more prevalent within the Photorhabdus genus (Fig. S2).
Mutation of the ALC in H4 and the constructed DALC (Table S1), as well as mutations of roeA in K18 and H12, altered cell morphology with shortened cells in the H12 roeA mutant but elongated cells in the ALC mutants (Fig. S3) and abolished exoprotein release in each of these mutants (Fig. 1B) The deletion of the ALC in DALC (Table S1) did not alter cell viability as determined through CFU counts and LIVE/DEAD stain (Fig. S5), indicating that the ALC may not cause cell lysis or, alternatively, that ALC-mediated cell lysis and death is compensated for by growth of the non-lysed cell population and is therefore not detectable through CFU counts.
Induction of the arabinose inducible vector pAY-ALCDrz1 (Table S2) harboring the ALC with its native gene arrangement of holA, pepB, and rz but devoid of rz1 resulted in no change in optical cell density (OD 600 ) in either MH96 ( Fig. 2A) or Escherichia coli DH10B (Fig. 2B). This likely reflects the requirement for back-translational coupling (termination-reinitiation), which is typical for l phage-like lysis (29,30). However, the use of the uncoupled ALC-encoding ORF in the pAY-ALC-opt vector series (Table S3) resulted in rapid cell lysis and a concurrent decrease of OD 600 from 1 to 0.2 within 15 min post induction (mpi) (Fig. 2C and Movie S1). Live-cell imaging of pAY-ALCDRz1-opt revealed cell elongation with brightening at cell poles; some elongated cells showed 1 to 3 small localized dark areas, which disappeared within 15 min ( Fig. 2C and Movie S2). Prior to cell lysis (20 mpi) localized bulging of cells leading to membrane blebbing (15 mpi) was observed ( Fig. 2C and Movie S2). In MH96, the expression of the ALC missing either i-spanin Rz or o-spanin Rz1 in pAY-ALCDrz-opt and pAY-ALCDrz1-opt caused rapid cell lysis as seen in ALC-opt expression ( Fig. 2A). However, lysis activity of the ALC-opt in E. coli increased in the absence of either spanin (Fig. 2B), which may reflect cell wall composition differences between these species. In the absence of pepB, slow cell lysis was observed when inducing pAY-ALCDpepB-opt in E. coli (Fig. 2B).
To correlate ALC activity with MH96 cell lysis, the translational HolA-sfGFP and Rz1- sfGFP constructs were assessed in LB broth cultures at 25°C. During early exponential growth (6 to 8 h), 15% of HolA-sfGFP cells were fluorescing, increasing to 44% at midlate exponential phase (10 to 12 h postinoculation [hpi]) and decreasing slightly to 41% at stationary phase (.12 hpi). The HolA-sfGFP signal was observed over the entire cell membrane (Fig. 3) and in elongated cells through these growth phases. Parallel assessments of Rz1-sfGFP found that all stationary phase cells were fluorescing with the entire cell membrane fluorescent (Fig. 3).
The PhoB-like regulator RoeA alters ALC, exoprotein, and global gene expression. Through amino acid alignments and correlation of RoeA to the resolved structures of the TCRs, CadC (31), OmpR (20,32), and PhoB (33), RoeA shares a PhoB-like helix-turn-helix (HTH) motif, comprising 3 a-helices, of which a2 and a3 are connected by a DNA loop forming the DNA binding structure (33,34), in which the DNA binding motif resides (Fig. 4). Of interest, the C termini of RoeA and RoeA-like proteins extend up to 40 amino acids compared to PhoB-like proteins, with the exception of CadC, which harbors a N-terminal HTH motif. Unlike PhoB, RoeA has no cognate phosphorylation domain.
Assessing the non-redundant National Center for Biotechnology Information's (nrNCBI) protein database, RoeA orthologues were identified in a limited number of bacteria, mainly within Yersiniaceae (sharing 30 to 100% amino acid identity) and Enterobacteriaceae (,30% amino acid identity) (Fig. S3). Of note, a RoeA orthologue (40% amino acid similarity to RoeA) is located 59 of the Afp/PVC-like eCIS protein complexes of Pseudomonas chlororaphis, and is in juxtaposition to a ALC HolA ortholog (Fig. 1A). Interestingly, a second MH96 RoeA-homolog Yen7 is located 59 of Yen-Tc, which is a similar position to the LysR-like regulator tcaR of other Tc-encoding Yersinia strains (12,25).
Attempts to delete roeA in its entirety were unsuccessful, while SDS-PAGE assessment of MH96DroeA151::Spec containing a spectinomycin cassette 151 bp 39 of the roeA initiation codon (Table S1) showed an exoproteome profile similar to MH96 (data not shown). This contrasts to the highly reduced exoprotein profile of the transposon insertion H12. Through the positioning of these mutations on the RoeA amino acid sequence (Fig. 4), the H12 insertion prevents translation of the entire RoeA HTH-motif, while the MH96DroeA151::Spec insertion retained 90% of the a2 helix and a partial HTH motif, which may enable its functionality. Based on the different exoproteome profile of H12 and MH96DroeA151::Spec the genome of H12 was sequenced from where a single transposon insertion within roeA was validated and the H12 roeA-strain was used in subsequent assessments.
To further define the role of RoeA in the transcriptional regulation of exoproteins and their release mechanism, the roeA H12 mutant and the wild type MH96 were subjected to transcriptomic assessments, targeting early stationary growth phase at approximately 9.6 log 10 CFU mL 21 . Through DEseq2 analysis of the transcriptome, the H12 roeA mutant resulted in the differential expression of 2235 (53%) genes relative to MH96. For analysis purposes we concentrated on genes with transcription levels increased or decreased by log 2 fold .j1j, and identified 406 genes that were significantly overexpressed and 500 that were significantly downregulated (Fig. 5).
While most differentially regulated genes in roeA mutants have no assigned COGclassification, the H12 roeA mutation affected genes with a wide range of functions, with greater effects on genes involved in: (i) post-translational modification, such as heat shock proteins, DnaJK and GroEL molecular chaperones; (ii) translation, such as ribosomal proteins; and (iii) intracellular trafficking and secretion, such as genes of the T2SS (File S1). Other genes function in energy production, and carbohydrate, amino acid and inorganic ion transport, and metabolism (File S1). Importantly, the transcription of genes encoding for the ALC, its associated roeA, Yen-Tc, the roeA homologue yen7, and the genes encoding the predominant wildtype MH96 exoproteins PL78_18785, PL78_05495, PL78_11910, and PL78_05310 (validated through LC-ESI-MS/MS) were significantly reduced in the roeA H12 mutant (File S1).
Temperature dependent regulation of RoeA and its effect on exoprotein production. Based on the role of RoeA in the transcription of exoproteins including the Yen-Tc and the effect of temperature on virulence regulation in vitro (35) and in vivo in G. mellonella at 25°C and 37°C (15), we investigated the effect of temperature on the translation of RoeA and the Yen-Tc associated Chi1, as an exoprotein proxy.
Using a P roeA ::lacZ cis-merodiploid strain cultured at 25°C in LB broth, the b-galactosidase (b-gal) levels increased during exponential growth phase but stabilized at stationary growth phase (Fig. 6), which correlates with the increased exoprotein levels produced through the exponential growth phase (17). In parallel assessments, at 37°C P roeA ::lacZ and P chi1 ::lacZ b-gal levels were significantly reduced compared to 25°C (Fig. 6).
The YeRER intergenic region affects the regulation of the ALC and RoeA. As previously demonstrated in the HESA, 3 mutations (H4, H31, and H45, 59 of the ACL) and the spontaneous K18 mutation (59 roeA) were identified in the YeRER intergenic region (17), suggesting an important role for the intergenic region in ALC regulation, and, FIG 4 Amino acid alignment and secondary structure prediction of RoeA and its Yen7 orthologue and selected PhoB type regulators. Black filled circles denote PhoB residues linked to DNA binding (33), and area marked by a star denotes HTH-motif (34). Red and blue arrows denote respective MH96DroeA151::Spec and H12 mutation sites.
Regulation of Exoprotein Release in Y. entomophaga Microbiology Spectrum therefore, exoprotein release. Through transcriptome assessments, the predicted transcriptional start site of roeA and holA was located 3 bp and 350 bp 59 of the respective gene (Fig. S6A). A saliant finding through the assessments of the RNAseq mRNA reads was the identification of a low number of mRNA reads spanning a 220 bp region of the YeRER intergenic region in both WT MH96 and the roeA-H12 mutant, and the predicted ncRNA designated ncALC220 (Fig. S6A). Based on this information we further interrogated the YeRER nucleotide sequence for potential regulatory signatures (Fig. S6B). The YeRER intergenic region is AT rich with 33.7% G1C relative to 48.6% G1C of the MH96 genome, and harbors several protein binding motifs including those for PhoB and Furlike proteins, identified using the Prodoric software (Table S4). The Vibrio cholerae cyclic AMP receptor protein (CRP) DNA binding motif was found in the core sequence of the degenerate repeats (Fig. S6B). Further investigation of the nucleotide sequences 59 of roeA  Based on the bioinformatic analysis of the YeRER associated intergenic region and the potential role of ncALC220 in YeRER regulation, a series of pACYC184 (p184) based vectors ( Fig. 7A and Table S2) were constructed enabling their effect on exoprotein production in either a MH96, DALC, H4, or K18 background to be determined. The vectors comprise various sections of the intergenic region, ncALC220, and the intergenic region where the ncALC220 sequence was deleted (p184INTDncALC220) (Fig. 7).

Regulation of Exoprotein Release in Y. entomophaga
Microbiology Spectrum restoration was observed in H4 (transposon insertion 59 of the ALC). Importantly, trans complementation with p184INT_ALC in either DALC and H4 resulted in partial restoration of exoprotein but not the Yen-Tc (Fig. S7B, C), and cell morphology with cells of similar size to MH96 (Fig. S3). While trans complementation of DALC with either p184INT, p184 5'holA, and to a lesser extent with p184INT(K18) or p184ncALC220, resulted in cells of a similar length as observed for MH96 (Fig. S3). Further to this, trans complementation with the intergenic region, in full or in part, including the ncRNA220, did not significantly alter cell morphology in H4 being similar in appearance to vector only control (Fig. S3). In K18, none of the trans complementation p184 vectors altered either the observed exoprotein ( Fig. S7D) or cell morphology (Fig. S3).
To determine if cell elongation observed in H4 and DALC and cell shortening observed in K18 reflects an accumulation/absence of pre-exoprotein within the cell, the culture (;log 10 9.6 CFU/mL) cell pellets and supernatant of MH96, K18, H4, and DALC were assessed by SDS-PAGE (Fig. 1B). Pre-exoproteins corresponding to the Yen-Tc (as proxy for exoproteins) were seen in the H4, DALC and MH96 but were absent in K18 and the roeA H12 mutant (Fig. 1B). To further define the role of RoeA in exoprotein release, the vector pAY-RoeA was induced in the roeA K18 mutant where, with the exception of the Yen-Tc, exoprotein was released (Fig. S7D).

DISCUSSION
The presented data supports the role of the YeRER associated RoeA regulator and the ALC in production of exoprotein by Y. entomophaga-a model which is presented in Fig. 8. Differing from MH96, SDS-PAGE assessments of supernatant of the roeA mutants H12 and K18, and the ALC mutants DALC and H4 revealed an absence of exoprotein (Fig. 1B) and changes in cell morphology (Fig. S3). The altered phenotype of reduced exoprotein release and changes in cell morphology in H4 and DALC could be restored by trans complementation of p184INT_ALC, revealing that its putative RoeA regulator can act in trans. The presence of pre-exoprotein precursors, such as the Yen-Tc in the cell pellet of DALC and H4, but no exoprotein (Fig. 1B) supports the mechanistic role of the ALC in exoprotein release. Under the same conditions no pre-exoprotein was observed in the cell pellets of either the H12 roeA mutant or K18 (Fig. 1B), further validating the regulatory role of RoeA in the induction of pre-exoprotein including the Yen-Tc.
The induction of the native ALC cassette pAY-ALCDrz1 devoid of the second rz1 spanin did not cause cell lysis. In Y. enterocolitica W22703 induction of its lysis cassette deficient in spanins caused cell lysis (37). For W22703, holA and pepB are adjacent genes while in MH96 holA and pepB are overlapping genes on different reading frames. Through uncoupling the ALC system, the induction of pAY-ALC-opt, pAY-ALCDrz-opt (devoid of Rz spanin), and pAY-ALCDrz1-opt (devoid of the Rz1 spanin) caused rapid cell lysis in MH96, indicating the necessity of holin and endopeptidase for lysis. Similar effects were also noted on the induction of these constructs in E. coli. Of interest, although the induction pAY-ALCDpepB-opt devoid of pepB had no effect on MH96, its induction in E. coli resulted in a low level of cell lysis (Fig. 2B). This may reflect the nonspecific pore of the holin allowing other peptidoglycan degrading enzymes to enter the periplasmic space in E. coli (38, 39) (Fig. 8).
Fluorescent microscopy of the HolA-sfGFP fusion found fluorescence was confined to a subset of cells of various sizes, wherein the fluorescence was observed throughout the cell. This may reflect the integration of HolA in the inner membrane and subsequent formation of the holin holomer required for release of the endopeptidase as documented in E. coli phage l system and the mycobacteriophage D29 system expressed in E. coli prior to cell lysis (40,41). In contrast, the fluorescence of the spanin Rz1-sfGFP was observed throughout the population and across the entire cell, which was also demonstrated in the spanin complex of the phage l using a GFPURz fusion by Berry et al. (2012) (42). The post-transcriptional regulation of ALC is likely through the need for back-translational coupling to uncouple the holA and pepB ORFs, the process of which was confined to a subset of cells as evidenced using the HolA-sfGFP reporter. Similar to phage systems, such as the l phage, the translation of HolA is likely a rate limiting step of cell lysis (43).
Through RNAseq, the expression 2235 genes including those encoding components of the ALC cluster were significantly reduced in the H12 roeA mutant. This included the reduced transcription of several key exoproteins (PL78_18785, PL78_05495, PL78_11910, PL78_05310 [as validated through LC-ESI-MS/MS of wild type MH96 culture supernatant]) and the Yen-Tc. In addition, through the use of P roeA ::lacZ and the Yen-Tc P chi1 ::lacZ reporters, the b-gal activities of both strains proportionally increased through exponential growth in LB broth at 25°C, and were both significantly reduced at 37°C, linking the translation of both RoeA and the Yen-Tc.
Based on the high similarity of RoeA to PhoB and PhoB-like regulators, such as CadC and OmpR (Fig. 4), it is plausible that altered expression of many genes of the H12 roeA mutant may reflect the ability of RoeA to bind other yet to be determined genomic sites. Other genes significantly reduced in expression in the H12 roeA mutant included ribosomal genes, and genes involved in metabolism. Based on the absence of a ribosomal depletion step in our methodology (refer to materials and methods), these transcriptional changes may cause the cells to enter a state of reduced metabolic activity. The reduced metabolism in H12 and K18 may account for the observed homogenous, mid is likely activated as part of a larger regulatory system responding to environmental stimuli, such as temperature or stress response. The presence of an H-NS binding consensus sequence at similar nucleotide distances from roeA and its yen7 orthologue suggests that H-NS regulates the expression of these genes. The RoeA-DNA binding motif enables its binding to PhoB-like DNA binding sites in the genome, including within the YeRER intergenic region. RoeA then activates the transcriptional expression of the ALC and exoproteins, including the Yen-Tc, as evidenced through their reduced transcription and the absence of pre-exoprotein in the H18 and H12 roeA-cell pellet. The ability of the ncALC220 to reduce exoprotein release when placed in trans in MH96 supports the hypothesis that the YeRER intergenic region may interact with mRNA or proteins e.g., Hfq, wherein Hfq binding sites reside within the YeRER intergenic region and within the ncALC220. (B) The ALC is tightly regulated at post-transcriptional level by the termination-reinitiation complex of the ALC operon, as well as yet to be determined posttranscriptional factors linked to ncALC220. Based on its observed expression in a subset of cells, HolA is under tight post-translational regulation. Prior to holin pore formation, exoproteins accumulate in the cell causing cell elongation. Upon holin pore formation, the endolysin PepB and the spanin complex enter the periplasm, which subsequently (C) causes cell lysis allowing the release of the exoproteins including the Yen-Tc. Due to inactivation of the ALC, no cell lysis occurs, which inhibits the release of proteins and leads to accumulation of exoproteins in the cell, as observed by SDS-PAGE of the H4 and DALC cell pellets and supernatants, which then causes cell elongation. In the H12 roeA mutant, neither the ALC nor exoproteins are produced resulting in uniform sized cells, which do not elongate.

Regulation of Exoprotein Release in Y. entomophaga
Microbiology Spectrum -exponential cell culture comprising cells of a shorter length relative to the heterogenous population with a small number of elongated cells observed in wildtype MH96 (Fig. S3).
In contrast, a high number of atypically elongated cells were observed in the DALC mutant (Fig. S3).
Interrogation of the roeA/holA intergenic nucleotide sequence revealed the presence of several nucleotide repeats, PhoB-and H-NS binding motifs, and secondary structures ( Fig. S6B and C, and Table S4). The prevalence of a range of different DNA binding motifs and the ncALC220 suggests that the YeRER intergenic region is the substrate for a complex of transcriptional, translational, and post-translational regulation (44) (Fig. S6B and Table S4). The positioning of a H-NS binding consensus sequence at similar nucleotide distances from roeA and its yen7 homologue (Fig. S6C and Table S4) suggests that RoeA is likely under the control of the H-NS regulatory cascade, from where RoeA in turn acts on exoprotein and the ALC expression (Fig. 8). This, in part, parallels the thermoregulation of the Y. enterocolitica W22703 LysR-like transcriptional regulator TcaR1/2 and H-NS (25,26). Of interest, H-NS, a global transcriptional regulator, reacts to environmental cues (45,46), and may indirectly take the place of a TCR response sensor that is absent in RoeA. Similar H-NS regulatory cascades have been described for H-NS in acid stress resistance in E. coli (45,47). Adding to this, Schoof et al. (16) found transposon mutants H23 and H45 for quorum sensing (QS) N-acyl-homoserine lactone synthetase resulted in reduced exoprotein through the exponential growth phase (17), indicating growth phase dependent regulation of RoeA resulting in exoprotein expressions.
A saliant finding was the ability of the intergenic ncALC220 region to reduce exoprotein release in MH96 when complemented. This reduction was not observed through complementation of the YeRER intergenic region devoid of the ncALC220 region (Fig. S7A). This data revealed the ncALC220 region as a key component of a complex regulatory network. Based on this, we can speculate that the ncALC220 may inhibit transcription or translation of ALC or RoeA (48,49). Trans complementation of p184INT, p1845'holA, p184INT(K18), and p184ncALC220 in DALC did not restore exoprotein release in DALC but did reduce the proportion of elongated cells, with cell sizes similar to those observed for MH96 (Fig. S3). Based on these findings, we hypothesize that the complemented regions are diluting out an activator of RoeA, decreasing RoeA activity, which reduces the production of intracellular pre-exoprotein (Fig. 1B), and prevents cell elongation. The inability to complement within the intergenic region, in full or in part, to alter the cell morphology (Fig. S3) or exoprotein profile of the K18 strain, revealed that the missing nucleotides of the K18 3 bp deletion, 131 bp 59 of roeA are required for the expression of RoeA.
The absence of elongated cells in either the K18 or the roeA H12 mutants (Fig. S3) further supports the notion that RoeA is required to induce intracellular pre-exoprotein expression, which causes cell elongation. Of interest, the induction pAY-RoeA restored exoprotein release to K18 and H12 (similar to the trans complementation of p184INT_ALC in DALC) but no Yen-Tc associated bands were observed. We assume this may relate to a combination of factors, such as the proximity of nc220 to its yet defined target, and/or most likely, the regulation of the Yen7-RoeA homologue.
The heterogeneity of the observed cell shapes and HolA expression reflect observations in other systems. In S. marcescens and E. coli, the expression of the exoproteins chitinase AB and colicin, respectively, caused cell elongation which was confined to a low proportion of cells (13,50). In this respect, 0.6%, 6%, and 2% of E. coli cells that expressed colicin A, E2, and E7, respectively, were elongated (50). Using mKate fluorescence peptidase ChiX reporter, 1% of S. marcescens cells fluoresced which correlated to the co-expression of the ChiWXYZ lysis cluster and the ChiAB chitinases (13). The heterogeneous colicin expression is caused by bet-hedging, a risk spreading strategy in which stochastically occurring phenotypes of an isogenic population may adapt to changing environmental conditions (50,51). Under in vitro conditions, exoprotein, including Yen-Tc, release is restricted to temperatures #25°C (35), which reflects a responsive switch rather than a stochastic switch (bet-hedging). In contrast to these in vitro findings, through the use of a YenA1 GFP reporter, elongated MH96 cells were observed in vivo in G. mellonella during early infection at 25°C but were absent at 37°C, where only a limited number of cells fluoresced (16,18), indicating an additional layer of complexity in the regulatory system. Based on the results above, a model for the regulation of exoprotein production in Y. entomophaga is proposed in Fig. 8.
Based on the phylogenetic data, RoeA-like regulators identified in some bacteria of the Enterobacteriaceae and Yersiniaceae may also enable the release of other large macromolecular toxin-transporting assemblies, such as AFP/PVC complex of P. chlororaphis, through the activation of an ALC-like lysis cassette. Further research to define the role of RoeA and the ncRNA, ncALC220, in exoprotein regulation is required. The proposed model of Y. entomophaga MH96 mediated exoprotein release (Fig. 8) provides phenotypic evidence of the crucial role of a holin/endolysin-based system in the programmed release of proteins. It provides further support to the roles of these lysis systems as T10SS, as initially proposed by Palmer et al. (12) based on in silico analysis.
Molecular cloning. Standard DNA techniques were performed as described by Sambrook (52). Chromosomal DNA was isolated using PrepMan Ultra Sample Preparation Reagent (Thermo Fisher), and plasmid DNA was isolated using a High Pure Plasmid isolation kit (Roche). For amplification of genetic elements, Platinum Taq DNA polymerase (Invitrogen) was used according to manufacturer's guidelines. Amplicons were purified using High Pure PCR Product purification kit (Roche). Primers are listed in Table S5.
When required, PCR purified amplicons were ligated into pGEM-T Easy (Promega) following the manufacturer's instructions. The resultant construct sequences were validated using M13_F and M13_R universal primers, or in other cloned constructs using construct specific primers (Table S5). Sequencing was performed using Macrogen Sequencing Services (Macrogen Inc.). DNA was electroporated into Y. entomophaga and its derivatives using the method of Dower et al. (53).
Construction of ALC mutant. For the deletion of the ALC, 2 kb 59 and 2 kb 39 of the ALC operon were PCR amplified using the primer pairs MS82/83 and MS84/85, respectively. The primers MS83 and MS84 harbor a complementary sequence to the pKD4 encoded kanamycin cassette that was amplified using the primers MS01 and MS02. The resultant amplicons were then assembled using fusion-PCR. The purified fusion-PCR amplicon DALC was cloned into pGEM to form pGEM-DALC, from where it was cloned into the suicide vector pJP5608 using SacI restriction sites (pJP5608-DALC). E. coli ST18 was used to conjugate pJP5608-DALC into MH96. Trans-conjugants were selected on LB-agar with kanamycin and tested for loss of pJP5608 tetracycline resistance. The final mutant MH96-DALC was subjected to PCR and sequence was validated using the peripheral validation primers MS29 and MS30.
Construction of p184 trans complementation vectors. To construct the p184 vector series (Fig. 7), the various regions of the MH96 wild YeRER region were PCR amplified using PCR primers (Table S5). The resultant PCR amplicons were cloned into pGEM Teasy, from where they were cloned into pACYC184 using the pGEMTeasy derived EcoRI restriction site. The K18 intergenic region was amplified (INT[K18]) from K18. Using GeneScript DNA of INT with a deletion of ncYLC220 (INTDnvYLC220) and ncYLC220, truncations ncYLC80 and ncYLC110 flanked by EcoRV restriction were synthesized and used to clone into pACYC184 EcoRV to form the respective plasmids p184 INTDnvYLC220, p184ncYLC80, and p184ncYLC110. The pACYC184 constructs were sequence validated with primer set MS/MS and transformed into MH96, K18, DYLC, and H4.

Regulation of Exoprotein Release in Y. entomophaga
Microbiology Spectrum Protein visualization and characterization. Standard sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed as described (54). Proteins were visualized by silver staining according to Blum et al. (55).
LC-ESI-MS/MS. For LC-ESI-MS/MS, the appropriate band was excised from 0.1% (wt/vol) SDS polyacrylamide gel stained with Coomassie brilliant blue, and prepared for LC-ESI-MS/MS spectrometry. Each gel band was analyzed by mass spectrometry after de-staining, reduction with 0.1 M tris (2-carboxyethyl) phosphine (Fluka Chemie GmbH), alkylation with 20 mL of 0.15 M iodoacetamide (Sigma), and digestion for 18 h with 1 mg of TPCK-trypsin (Promega) in the presence of 10% acetonitrile (ACN). After digestion, the peptides were dried and resuspended in 50 mL of 0.1% FA prior to injection on the mass spectrometer.
LC-ESI-MS/MS was performed on a nanoflow Ultimate 3000 UPLC (Dionex) coupled to maXis impact HD mass spectrometer equipped with a CaptiveSpray source (Bruker Daltonik). For each sample, 1 mL of the sample was loaded on a C18 PepMap100 nano-Trap column (300 mm ID Â 5 mm, 5-micron 100 Å) at a flow rate of 3000 nl/min. The trap column was then switched in line with the analytical column ProntoSIL C18AQ (100 mm ID Â 150 mm 3-micron 200 Å). The reverse phase elution gradient was from 2% to 20% to 45% over 60 min, total 84 min at a flow rate of 600 nL/min. Solvent A was LCMS-grade water with 0.1% formic acid (FA); solvent B was LCMS-grade ACN with 0.1% FA.
The Q-TOF Impact HD (Bruker Daltonics) mass spectrometer was set up in a data-dependent automatic MS/MS mode where a full scan spectrum (50-2000 m/z, 2 Hz) followed by 10 MS/MS (350 to 1500 m/z, 1-20Hz) of the most intense ions with charge states 2 to 3 selected.
Genome sequencing of K18. Genomic DNA for genome sequencing was isolated using the ISOLATE II Genomic DNA Kit (Bioline). For identification of DNA alterations in strain K18, Illumina HiSeq 2500 System by Macrogen Sequencing Services was used. DNA sequences were trimmed using Trim_Galore (http://www .bioinformatics.babraham.ac.uk/projects/trim_galore/). Nucleotide differences were identified by alignment of Illumina reads against the MH96 genome sequence using the conda package for breseq version 0.33.0 with default parameters http://barricklab.org/twiki/bin/view/Lab/ToolsBacterialGenomeResequencing (56).
Bioinformatic analysis. DNA sequences were trimmed and aligned against the genome of strain MH96 (GenBank accession number NZ_CP010029.1) using the Map to Reference function of Geneious Prime (57). Protein sequences were assessed using Phyre2 (http://www.sbg.bio.ic.ac.uk/phyre2/html/ page.cgi?id=index) and BLASTP (https://blast.ncbi.nlm.nih.gov/Blast.cgi) with default settings. For gene synteny, the highest BLASTP hits to the query protein were used to pull genome data from the respective organism, covering 10 kb over the homologous genes. Multiple nucleotide alignments and Neighbor-Joining tree were then performed using standard settings in Geneious 10.0.9.
Amino acid alignments were performed using Geneious 10.0.9 using ClustalW and BLOSUM Matrix with a Gap open cost of 10 and a Gap extension cost of 0.1. Amino acid alignments were visualized using GeneDoc 2.7.000 (58).
Cloning of ALC constructs and RoeA in pAY2-4, and their arabinose-based induction. Variations of the ALC operon were cloned into pAY2-4 NdeI and XhoI site. The PCR amplicon of holA/pepB/rz using primer pair MS101 and MS104 was cloned into pAY2-4 to form pAY-ALCDrz1 (hoLA, pepB, and rz). Optimized amplicons of the lysis cassette were synthesized at GeneArt (Thermo Fisher Scientific) and designed to encode: (i) holA, pepB, rz, and rz1 as non-overlapping ORFs (pAY-ALC-opt), while maintaining ribosomal binding sites (rbs) and amino acid identity; (ii) optimized regions encoding holA, pepB, and rz (pAY-ALCDrz1-opt); (iii) optimized region encoding holA, rz, and rz1 (pAY-ALCDpepB-opt); and (iv) holA, pepB, and rz1 (pAY-ALCDrz-opt), refer to Table S3 for synthesized nucleotide sequence. Cells harboring pAY-ALCDrz1, pAY-ALCopt, pAY-ALCDpepB-opt, pAY-ALCDrz-opt, and pAY-ALCDrz1-opt were grown in LB (40%) broth at 25°C and 200 rpm until an OD 600 of 1 was reached. The cultures were induced with arabinose (0.2% final concentration), or the same volume of dH 2 O was added as control and placed at ambient temperature ;22°C on a rotating platform at 40 rpm. The OD 600 was measured every 15 min until 3 h and at 24 hpi.
For construction of pAY-RoeA, the primers set MS65/MS66 were used to PCR amplify the amplicon RoeA, the purified product then cloned into pGEM T-easy, and sequence validated with M13F/R primer. Using NdeI and XhoI cloning sites, the RoeA amplicon was cloned into the analogous sites of pAY-2 to form pAY-RoeA. Prospective pAY-RoeA clones were sequence validated using the AY_F and AY_R primers. pAY-RoeA was then electroporated into K18 cells. For pAY-RoeA induction, 50-mL cultures were grown in 40% LB to which 0.02% arabinose was added. The cultures were incubated for 16 h at 25°C under 250 rpm shaking from where samples were centrifuged at 8.000 Â g for 5 min to collect cell pellets and culture supernatant to assess using SDS-PAGE.
Light and fluorescence microscopy. For light microscopy, 3 mL of a MH96 cell culture at 16 hpi were observed under phase contrast. For Live/Dead staining the Syto9/PI stain (LIVE/DEAD BacLight kit; Invitrogen) was used at a 1:1 ratio and incubated for 5 min in a 1.5 mL UV-safe tube. Cells were observed under an Olympus BX50 light microscope at Â400 magnification for both light and fluorescence microscopy. The SYTO9 stain was visualized using a FITC filter with excision of 460/515 nm, and the PI stain using Texas Red 545/610 nm filter. Cell counts were measured using the software ImageJ 1.47v (59).
Live-cell imaging. All cultures were grown in 100% LB broth (200 mg/mL ampicillin) until OD = 1. For live-cell imaging, 10 mL of the culture were induced with 0.6% arabinose and immediately pipetted onto agarose covered glass slides, and assessed by light microscopy within the first minute post induction. Agarose pads were used to eliminate cell movement during the imaging process. Live-cell imaging was undertaken using the LSM710 microscope operated with Axiovision System (Carl Zeiss).
Three culture flasks per strain (H12 and MH96) were incubated to reach log 10 CFU mL 21 of 9.5. From each culture flask, 1 mL of sample culture was immediately transferred into 2 mL RNA to protect bacteria (Qiagen) and vortexed. After 5 min incubation at 25°C, the samples were pelleted at 5,000 Â g for 10 min. The supernatant was decanted, and pellets left to air dry at 37°C before freezing at 220°C.
RNA was isolated using the RNA minikit (Qiagen) following the manufacturer's instructions. Following the on-column DNA digest with RNase-free DNase, a second, off-column DNA digest was performed. To 40 mL RNA, RDD buffer (40 mL) and DNase stock I (2.5 mL) was added, and the volume adjusted to 100 mL with DNA-free water. After 10 min incubation time at 25°C, the RNA cleanup protocol (provided in the Qiagen RNA minikit) was followed. The RNA was eluted in 40 mL RNase-free water and isopropanol precipitated. The total volume was adjusted to 180 mL and 1% sodium acetate (3M) was added. Three times, ice cold 100% ethanol (600 mL) were added to the solution and vortexed. The microcentrifuge tube was then placed at 220°C overnight, after which the suspension was centrifuged (10,000 g for 30 min at 4°C) and supernatant discarded. The pellet was washed twice with ice cold 75% ethanol (500 mL) and pelleted at 10,000 Â g for 5 min at 4°C, and the supernatant discarded. After the final wash step, the samples were pulse spun to remove residual ethanol by pipetting out residual supernatant. The pellets were air dried at 37°C for 30 min, and then resuspended in RNase-free water. The resuspended sample was quantified by nanodrop. Of the sample RNA, 6 mg/mL were placed into a reaction tube, and liquid was evaporated in a SpeedVac. RNAseq was quality controlled and performed by Macrogen.
RNAseq. The Illumina short reads were inspected for quality using FASTQC (https://www.bioinformatics .babraham.ac.uk/projects/fastqc/). Bases with low quality PHRED scores (PHRED , 15 using a sliding window of 4 bases) were trimmed (using TRIMMOMATIC) from the short-read library, as well as any Illumina adapter sequences. Paired reads that were longer than 36 bp were kept for further analysis. The Y. entomophaga MH96 genome (Aug2018.NCBI.gb) was converted from GenBank format to fasta format using a custom BIOPYTHON script prior to indexing. Gene annotations were converted to gff format using the BIOPERL program bp_genbank2gff3.pl (https://manpages.debian.org/testing/bioperl/bp_genbank2gff3.1p.en.html). The genome fasta file was indexed, and the short-read libraries aligned to the reference genome using HISAT2 with the default parameters. STRINGTIE was used for novel transcript assemblies, and BALLGOWN calculated the transcript count matrix for each sample.
The transcript count matrix was read into R, and differential gene expression calculated using the DESeq2 package (60). Genes were considered differentially expressed when the adjusted P value (p adj ) was less than 0.05. For analysis purposes, differently expressed genes of a log2Fold change , 21 and . 1 were considered of interest, and further assessed.
b-gal assays. For b-gal assays, an over-night culture was used to inoculate (1%) 50-mL LB flasks that were incubated at either 25 or 37°C with 200 rpm shaking. For the assay, at each time point, 2 Â 200 mL of each culture were collected into a sterile 96-well plate (F-bottom) (Greiner Bio-One Cellstar) and then frozen at 280°C. Using a 96-well microplate reader SPECTROstarNano (BMG Labtech) and the MARS Data Analysis software (BMG Labtech), the rate of b-gal production was measured at OD 420 following the methods of Schaefer et al. (61,62) with a custom b-gal mix: 60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl,1mM MgSO 4 ; 1.8 mL mL 21 b-mercaptoethanol, 0.2 mg mL 21 Lysozyme from chicken egg (Sigma), 1:150 diluted Bacterial Protein Extraction Reagent (Thermo Fisher); and 1 mg mL 21 of 2-nitrophenyl-b-galactopyranoside (Sigma). To control for temperature-dependent differences in the MH96 calibration curves measured at OD 600 for 25 and 37°C, an appropriate calibrating factor was applied to the Miller Unit Equivalent calculation.