Pantoea sp. P37 as a novel nonpathogenic host for the heterologous production of rhamnolipids

Abstract Microbially derived surfactants, so‐called biosurfactants, have attracted significant attention as an environmentally friendly alternative to their chemically synthesized counterparts. Particularly, rhamnolipids offer a large potential with their outstanding surfactant properties such as complete biodegradability, low toxicity, and stability. Rhamnolipids are naturally synthesized by the opportunistic human pathogen Pseudomonas aeruginosa under the tight regulation of a highly complex quorum‐sensing system. The heterologous production of mono‐rhamnolipids by a newly isolated nonpathogenic strain of the genus Pantoea was investigated. Analysis of the genome obtained by a chimeric assembly of Nanopore long reads and high‐quality Illumina reads suggested that the strain has evolved to an epiphytic rather than a pathogenic lifestyle. Functional heterologous expression of the mono‐rhamnolipid operon rhlAB derived from a P. aeruginosa strain was established and confirmed by HPLC analysis. Transcriptome analysis indicated destabilizing effects of the produced rhamnolipids on the cell envelope of the host resulting in the induction of molecular stress responses. After integration of the rmlBCDA operon, extracellular rhamnolipids in amounts up to 0.4 g/L could be detected and were identified as a mono‐rhamnolipid Rha‐C10‐C10 by MALDI‐TOF mass spectrometry.

reactions are involved in rhamnolipid biosynthesis: the dimerization of two β-hydroxy fatty acids for the formation of the lipophilic moiety, and the transferase reaction for the glycosidic conjugation of the activated nucleotide carbohydrate dTDP-L-rhamnose to the fatty acid dimer. The dTDP-L-rhamnose moiety is formed by an enzymatic four-step reaction encoded by the rmlBCDA operon. The dimerization step is catalyzed by the 3- (3-hydroxyalkanoyloxy) alkanoate synthetase, encoded by the rhlA gene, and the transferase reaction by the rhamnosyltransferase I, encoded by the rhlB gene. The genes rhlA and rhlB are organized in one operon.
For the synthesis of a di-rhamnolipid, a second transferase step is included, catalyzed by the rhamnosyltransferase II (RhlC) and encoded by the rhlC gene (Abdel-Mawgoud, Hausmann, Lépine, Müller, & Déziel, 2011). Rhamnolipids are naturally produced by Pseudomonas aeruginosa together with various virulence factors during biofilm formation. While rhamnolipids are important structural integrity components in biofilms, they are also responsible for necrosis of host macrophages and leukocytes during acute infections (Lee & Zhang, 2015). The World Health Organisation recently published data underlining the emerging threat of antibiotic-resistant P. aeruginosa strains (Tacconelli, Magrini, Kahlmeter, & Singh, 2017). Rhamnolipid biosynthesis is regulated by multilevel signal transduction networks such as the quorum-sensing system which involves multiple regulatory factors, secondary messengers, and signaling molecules. Since the pathogenicity and complexity of the quorum-sensing system in P. aeruginosa is restricting its use for the industrial-scale production of rhamnolipids, many studies have focused on the discovery of alternative nonpathogenic microbial producers. The majorities of previously described strains were either members of the Pseudomonadaceae family such as P. chlororaphis, P. fluorescens, P. lueola, P. indica, and P. stutzeri or were part of the genus Burkholderia (Abdel-Mawgoud et al., 2010).
A maximum yield of naturally produced rhamnolipids of 9.6 g/L has been reported with a strain of P. indica (Bhardwaj, Cameotra, & Chopra, 2015). Another approach for the safe production of rhamnolipids is by heterologous recombination using a safe alternative host strain instead of P. aeruginosa. Besides the standard model organisms for genetic recombination E. coli and S. cerevisiae, Pseudomonadaceae have been predominantly used for the heterologous production of rhamnolipids with Pseudomonas putida KT2440 being the most prominent strain exhibiting the highest reported rhamnolipid titer of 14.9 g/L (Beuker et al., 2016). Although P. putida KT2440 has been declared as biologically safe by the US Recombinant DNA Advisory Committee, the strain contains a number of potential pathogenicity factors such as phospholipase D (Moore et al., 2006;Nelson et al., 2002).
Pantoea are Gram-negative bacteria belonging to the Enterobacteriaceae family. Within the Pantoea genus, there are currently 21 known type strains (Walterson & Stavrinides, 2015).
While most are assigned to BSL 1, a few species are considered animal or human pathogens (P. agglomerans, P. brenneri, P. eucrina, P. septica) and some are suspected plant pathogens (German Federal Institute for Occupational Safety and Health, 2015).
However, some P. agglomerans BSL 2 strains are used commercially as biocontrol agents (strains C9-1 and E325) and are considered as safe due to the lack of specific virulent determinants (Smits et al., 2010). So far, a total of 21 complete Pantoea genomes and 168 contigs and scaffolds of the 24 identified species have been deposited with the National Center for Biotechnology Information (NCBI). The majority (~80%) of the assemblies belong to only five assigned species providing limited genotypic insights into this diverse bacterial genus (NCBI Resource Coordinators et al., 2014).
The aim of this study was the identification of an alternative host for the heterologous production of rhamnolipids with a low safety risk profile and to uncouple the rhamnolipid synthesis from its complex quorum-sensing system.

| Screening for an alternative host for recombinant rhamnolipid production
Soil samples were mixed with sterile water. After a short sedimentation time, serial dilutions of the liquid phase were spread on selective LB agar plates which were fortified with 1 g/L R90 rhamnolipids in order to only promote the growth of strains resistant to rhamnolipids (AGAE Technologies). After incubation at 37°C overnight, single phenotypically distinguishable colonies were picked and inoculated into sterile LB broth fortified with 1 g/L R90 rhamnolipids, grown overnight and restreaked on the LB-R90-agar plate to assure isolation. Colony PCR was performed with the primer pair rhlI_F/rhlI_R targeting the rhamnolipid biosynthesis controlling acyl-homoserinelactone synthase encoded by rhlI in P. aeruginosa genome. Isolates with a positive result, showing a 236-bp amplicon on the agarose gel, were excluded. All other isolates were further identified with 16S rRNA sequencing using primer pair 27F/1492R.

| Construction of a phylogenetic tree
The phylogenetic tree was constructed with MEGA6 based on the concatenated sequences of the three conserved genes gyrB, rpoB, and 16S rRNA (Kumar, Stecher, & Tamura, 2016). In an initial step, the genome assemblies from various Pantoea strains were downloaded from NCBI and the genes extracted (Table A2). The tree was rooted with the concatenated genes from E. coli K12. It was calculated based on the general time-reversible model with a discrete gamma distribution (+G, parameter = 0.2441) and an allowance for evolutionary invariant sites (+I, 32.53% sites). All positions containing gaps and missing data were excluded from the analysis leading to a final of 9,599 positions in the final dataset. A bootstrap value of 1,000 replications was applied.

| Genome sequencing and data analysis
The total genomic DNA from Pantoea sp. P37 was isolated following the protocol for Gram-negative bacteria of the Promega Wizard Genomic DNA Purification Kit. To obtain high-quality reads, genomic DNA was sequenced on an Illumina HiSeq 4000 platform (Illumina Inc.) yielding 150 bp PE short reads. Furthermore, genomic DNA was sequenced on an Oxford Nanopore MinIon (Oxford Nanopore Technologies) to obtain long reads for genome scaffolding. Short and long reads were assembled by Unicycler in a chimeric approach, and the resulting sequences yielding a complete genome of five circular sequences were annotated by Prokka (Seemann, 2014;Wick, Judd, Gorrie, & Holt, 2017). Geneious (Biomatters) was used for detailed DNA data analysis. Based on the virulence factor database (VFDB) of common bacterial pathogens, the VFanalyzer pipeline was used for the identification of virulence genes (Liu, Zheng, Jin, Chen, & Yang, 2018). IslandViewer 4 was used for the prediction of genomic islands. Prophage sequences were identified by PHASTER (Arndt et al., 2016;Bertelli et al., 2017;Zhou, Liang, Lynch, Dennis, & Wishart, 2011). Further annotations, indicating bacteriophage infections were analyzed manually. The bacterial version of antiSMASH was used to detect secondary metabolite biosynthetic gene clusters (Blin et al., 2019). The results were illustrated in a genome map created by the CGView Server and CIRCOS (Grant & Stothard, 2008;Krzywinski et al., 2009).

| Construction of pETrhlAB
All enzymes used for restriction digestion reactions were purchased from New England Biolabs. The expression vector for heterologous mono-rhamnolipid production was constructed following the Twinprimer Assembly method (TPA) in combination with overlap extension PCR (Liang, Liu, Low, Ang, & Zhao, 2017). All final constructs were sequenced to guarantee the absence of mutations. The rhlAB operon derived from genomic DNA of P. aeruginosa TGR2A. rhlA was amplified with primer pairs rhlA_SF/rhlA_OER (Table A3) using Q5 polymerase (NEB). rhlB was amplified with primer pairs rhlAB_OEF/ rhlB_SR using KOD Xtreme Hotstart (Merck). Both forward primers contained a synthetic RBS sequence designed with the Ribosome Binding Site Calculator tool with translation rates similar to the predicted natural ones of P. aeruginosa PAO1 (Salis, 2011). The two amplicons were subsequently fused via overlap extension PCR and used as a template for amplification with primers rhlA_SF/rhlB_LR and rhlA_LF/rhlB_SR in two separate reactions with KOD Xtreme Hot Start DNA Polymerase (Merck KGaA). The vector backbone part derived from pET28a(+) which was amplified with the primer pairs pET_SF/pET_LR and pET_LF/pET_SR. The final gel-purified PCR products were assembled in one step together with the oligo pair J23108_F/J23108_R containing the constitutive promoter sequence obtained from BioBrick part BBa_J23108 (www.parts regis try.org). For one reaction, each fragment was diluted to reach a final concentration of 20 fmol per 20 µl. Two microlitre Cutsmart Buffer (New England Biolabs) was added to the mixture and topped up with nuclease-free water. All fragments were assembled using the following time/temperature profile: 2 min at 98°C, 2 min at 85°C, 2 min at 75°C, 2 min at 65°C and 2 hr at 56.5°C (T m of overlap regions + 5°C).
Four microlitre of the ligation product mixture was transformed into Mix & Go! (Zymo Research) competent E. coli OmniMAX cells (Thermo Fisher Scientific) and grown overnight. Colonies were picked, plasmids extracted, and restriction digestion with XhoI and NheI-HF performed. The correctly assembled construct, the expression vector pETrhlAB8, was finally transformed via electroporation into Pantoea sp. P37. For the construction of pETrhlAB100, the same fragments were used, only the oligo pairs were changed to J23100_F/J23100 carrying the promoter sequence of BBa_J23100.

| Construction of pET3110 and pET2711
The construction of pET3110 and pET2711 started from the previously built vectors pETrhlAB8 and pETrhlAB100, respectively. The rmlBCDA operon derived from P. aeruginosa TGR2A and was subcloned into the vectors as one fragment by amplifying the region with primer pairs rml_SF/rml_LR and rml_LF/rml_SR. The purified PCR products were assembled together with fragments derived from amplification of the vectors pETrhlAB8 or pETrhlAB100 with primer pairs rhlvb_SF/ rhlvb_LR and rhlvb_LF/ rhlvb_SR. The construct was verified by restriction digestion with NotI-HF.

| Construction of plasmids pACYC_FAS, pACYC_ fabH, pACYC_P450, and pACYC_algC
The backbone fragments of pACYC vectors were amplified with primer pairs pACYC_SF/pACYC_LR and pACYC_LF/pACYC_SR. All subcloned sequences were derived from genomic DNA of P. aeruginosa TGR2A. For pACYC_FAS, the genomic region containing genes fabF1, fabG, and fabD was amplified with fabDFG_SF/fabDFG_LR and fabDFG_LF/fabDFG_SR, the genomic region with accA was amplified with primers accA_SF/accA _LR and accA _LF/accA _SR, and accD with accD_SF/accD _LR and accD _LF/accD _SR. Genes accB and accC could be amplified as one fragment with accBC_SF/ accBC _LR and accBC _LF/accBC _SR. The total amount of 10 fragments was mixed together after purification to a final concentration of 20 fmol of each part. The final hybridization step was performed at two temperatures, 1 hr at 56.2°C and 1 hr at 55.7°C. The extracted plasmids were examined by SmaI digestion.
For the construction of pACYC_fabH, primers fabH_SR/fabH_ LR and fabH_SF/fabH_LF were used. The fragments were assembled at a hybridization temperature of 56°C and digested with BsgI for a final check. As the gene encoding the monooxygenase P450 is close to fabH, only one fragment needed to be subcloned to obtain the final construct pACYC_P450. Primers P450_SF/P450_LF and fabH_SR/fabH_LR were used to amplify the fragment from genomic DNA as a template. The hybridization temperature was 56°C and the assembly was verified by XhoI digestion. Through amplification with alg_LF/alg_SF and alg_SR/alg_LR, pACYC_algC was constructed with hybridization settings of 1 hr at 56.2°C and 1 hr at 55.7°C. The construct was verified by double digestion with SapI and XhoI.

| Rhamnolipid quantification
Rhamnolipids in culture supernatants were quantified by HPLC/ UV according to Smyth, Perfumo, Marchant, and Banat (2010) with some modifications using a Shimadzu LC-20A HPLC with SPD-20A UV/VIS detector (Shimadzu). Briefly, 100 µl of 1:10 water-diluted, cell-free supernatant was mixed with 100 µl derivatization reagent consisting of 0.5 M 2-bromoaceto-phenone: 1 M trimethylamine (8:2% v/v) in acetonitrile. The mixture was heated for 1 hr at 80ºC under constant shaking (1,000 rpm) and filtered with nylon syringe filters (0.22 mm). Gradient HPLC with UV detection set at 244 nm was used with a Phenomenex reversed-phase column (C18 column, 250 mm × 4.6 mm × 5 mm i.d.), acetonitrile (mobile phase A), and phosphoric acid (mobile phase B, 3.3 mM). Gradient conditions were as follows: 50% A and 50% B for 3 min, 100% mobile phase A for 19 min and held for 5 min, 50% A over 3 min and finally held for 10 min. The flow rate was set at 1.0 ml/min, and the injection volume was 20 µl. Standard curves were created using commercial rhamnolipids (Sigma-Aldrich).

| Liquid-liquid extraction of rhamnolipids
The culture supernatant was filtered through a 0.22-µm syringe filter, acidified with HCl to pH 2 and kept overnight at 4°C for rhamnolipid precipitation. The acidified supernatant was centrifuged at 25,000 g and 4°C for 10 min, and the pellet was dissolved in 4 ml water. Subsequently, 8 ml ethyl acetate was added, vortexed vigorously and centrifuged for 5 min for phase separation. The upper phase containing the rhamnolipids was transferred into a new tube, and the lower phase was extracted three more times with 8 ml ethyl acetate, respectively. After extraction, the ethyl acetate fraction was evaporated in a rotary evaporator at 40°C (Behrens, Engelen, Tiso, Blank, & Hayen, 2016).

| MALDI-TOF-MS analysis of rhamnolipids
The ethyl acetate fraction containing the rhamnolipids was dissolved in 100 µl water. The matrix was prepared by mixing 20 mg of 2,5-dihydroxybenzoic acid with 1 ml acetonitrile and agitated until complete dissolution. The matrix was mixed 1:1 with the sample. One microlitre of the obtained solution was spotted on a polished steel MALDI target plate and air-dried before introduction and analysis in an autoflex maX MALDI system (Bruker) with a Smart Beam Laser in linear mode. Lens and reflector voltages were 33 and 3,133 V, respectively. For each spectrum, a number of 300 shots were acquired. The detector voltage was set to 2.5 GS/s. Signals smaller than 480 Da were suppressed (Price, Ray, Vermillion, & Kuo, 2009).

| Transcriptome analysis by RNA sequencing
The recombinant Pantoea sp. P37 carrying the pETrhlAB8 vector and the Pantoea sp. P37 wild type (WT) were grown overnight at 37°C under shaking conditions (200 rpm) in glycerol-fortified LB medium.
The overnight cultures were centrifuged to pellet the cells. RNA sequencing was performed by Novogene (Singapore). For differential expression analysis, the readcount values were normalized by TMM (Trimmed Mean of M-values). The DEGseq package was used for analysis (Wang, Feng, Wang, Wang, & Zhang, 2009). A volcano plot, plotting significance as −log 10 (p-value) versus log2 fold change, was generated showing the overall distribution of differentially expressed genes. No biological replicates were included in the experimental setup; therefore, only outliers with high log2 fold changes and low p-values were examined.

| Isolation of a safe bacterial strain for heterologous production of rhamnolipids
The primary screening of soil samples on rhamnolipid-fortified LB agar yielded 78 rhamnolipid-resistant isolates. The exclusion of undesirable but ubiquitous P. aeruginosa strains by the amplicon-positive/ negative screening narrowed the pool further down to 9 candidates. Subsequent 16S rRNA sequencing revealed that all isolates belong to the Enterobacteriaceae family; however, 8 isolates were identified as members of the genera Klebsiella, Enterobacter, Raoultella, and Cronobacter, common BSL 2 microorganisms. One isolate was identified as a member of the genus Pantoea and designated as Pantoea sp.
P37. Pantoea species are mostly classified as BSL 1. Its 16S rRNA sequence was deposited under accession number MH071150 at NCBI.

| Genome analysis of Pantoea sp. P37
The genome of the isolate Pantoea sp. P37 was sequenced yielding high-quality Illumina short reads and Nanopore long reads.
The sequencing data were submitted to the NCBI Sequence Read Archive under accession numbers SRR7962492 and SRR7962491.
The chimeric genome assembly yielded 5 circular elements with varying characteristics (Table A1). It consisted of one chromosome, two mega-plasmids (>100 kb) and another two plasmids

| Phylogenetic analysis and taxonomic affiliation
For species classification, a maximum likelihood tree illustrating the phylogenetic relationship between Pantoea sp. P37 and 12 selected Pantoea strains was constructed ( Figure A1). The isolated strain Pantoea sp. P37 showed only a loose evolutionary relationship with the virulent strains P. agglomerans, P. brenneri, P. septica, and P. eucrina within the displayed Pantoea entity. It exhibited a close relatedness to P. dispersa and P. wallisii which are both classified as BSL 1 strains.

| Genome Mapping
The genome of Pantoea sp. P37 was sequenced and mapped for a deeper analysis of virulence factors. The genome maps highlighting particular features found on the main chromosome of Pantoea sp. P37 and its four plasmids are shown in Figures 1, 2,  The outer ring displays specific biosynthetic clusters (blue), MFS transporter, multidrug efflux pumps and transporter (red), hypothetical proteins (purple), and phage-associated genes (yellow). The second outer ring displays genomic islands (red). The third outer ring represents the GC content. The fourth ring represents the GC skew a p-hydroxybenzoic acid efflux pump, and general stress proteins.
Multiple antibiotic resistances including β-lactam, virginiamycin, and bicyclomycin resistance were also detected. Carbohydrate metabolism genes occurring in a predicted GI were comprised of three of the four homolog genes required for dTDP-L-rhamnose biosynthesis, rmlA (rffH_2), rmlC (rfbC), and rmlD.
Plasmid 1 carried three BGC which could be identified (Figure 2).
One of the clusters contained genetic elements encoding for enzymes participating in bacteriocin synthesis. A siderophore cluster found on plasmid 1 consisted of genes encoding siderophore biosynthesis proteins, an anion permease as well as iron and a major facilitator superfamily (MFS) transporter. An isoprenoid BGC for the production of zeaxanthin was also detected. One GI was predicted in plasmid 1. It was flanked by a phage-associated gene encoding a transposase and by genes of unknown function. Within the GI, 6 of the 17 CDS were genes of unknown function.
One of the main features of plasmid 2 was a cluster-like segment composed of various transfer (tra) genes located in one of the two predicted GI ( Figure A2). Besides the tra genes on plasmid 2, several genomic sequences were identified which encoded for enzymes involved in the biosynthesis of the endotoxic cell membrane component lipid A. Almost 20% of the plasmid 2 CDS were genes with unknown function.
A range of genes encoding for VirB proteins was also found on plasmid 3 while 43% of all genes had an unknown function. The smallest plasmid 4 with a total size of 5,593 bp contained 5 known and one unknown CDS. No specific BGC or virulence factors could be identified ( Figure A2). F I G U R E 2 Genome map of Pantoea sp. P37 plasmid 1. The outer ring displays specific biosynthetic clusters (blue), MFS transporter, multidrug efflux pumps and transporter (red), hypothetical proteins (purple), and phage-associated genes (yellow). The second outer ring displays genomic islands (red). The third outer ring shows the GC content. The fourth ring displays the GC skew F I G U R E 3 MALDI-TOF-MS spectra of extracts from culture supernatants of recombinant Pantoea sp. P37 isolate (pET3110), Pantoea sp. P37 wild type (WT) and P. aeruginosa isolate TGR2A  (Table 1). The pET3110 transformants produced 409.4 mg/L of the rhamnolipid which was the highest concentration detected. The other transformants produced much lower amounts, and no product was detected in the culture supernatants of the transformants pETrhlAB100 and pET2711.

| Impact of heterologous rhamnolipid production on the transcriptome of Pantoea sp. P37
An RNA-seq analysis was performed to assess the impact of the recombinant production of mono-rhamnolipids on the Pantoea sp. P37 transcriptome (Figure 4). The majority of upregulated genes in the recombinant strain are genes encoding proteins involved in genetic information processing pathways including chaperones (groL, groS), ribosomal proteins (rpsE, rplF, rplO, rplV), or polymerases (rpoB, rpoC).
A more than twofold expression was also observed of the genes cyoB and cydB2, both encoding cytochromes. Additionally, with a log2 fold change value of ~1.6, enhanced expression of the lipoprotein (lpp) gene in the recombinant strain encoding the enterobacterial major outer membrane lipoprotein was found. A significant downregulation of the cysG2 gene which is involved in heme biosynthesis was observed.
However, the majority of down-and upregulated genes in the transcriptome of the recombinant strain were unidentified genes.

| D ISCUSS I ON
Screening of soil samples on rhamnolipid-fortified agar was performed to select microbial strains with resistance to rhamnolipids.
As mainly pathogenic members of the ubiquitous P. aeruginosa were expected to grow, a colony PCR-mediated exclusion based on rhlI, coding for the acyl-homoserine-lactone synthase responsible for activating the expression of the rhlAB operon and other virulence-associated genes was employed (Raychaudhuri, Jerga, & Tipton, 2005).
The remaining strains were predominantly identified as pathogenic members of the Enterobacteriaceae family suggesting a genetic preset of resistance to rhamnolipids. A possible mechanism of this bacterial resistance is the production of specific cleaving enzymes or multidrug efflux pumps and transporter proteins of the MFS which provides nonspecific resistance to a number of compounds (Partridge, 2015). Pantoea sp. P37 was identified as the only rhamnolipid-resistant strain with a potential BSL 1.
A prediction of the pathogenicity of novel Pantoea strains is problematic and the misidentification of clinical samples has often resulted in the wrong attribution of Pantoea species to a variety of infections (Walterson & Stavrinides, 2015). Besides a phylogenetic analysis based on the housekeeping genes 16S rRNA, gyrB and rpoB ( Figure A1), a de novo whole-genome sequencing was performed to both clarify the identity of the isolate and also to allow an in-depth pathogenomic evaluation of the strain (Figures 1, 2 and A2; Table   A1). The phylogenetic analysis indicated a close relatedness of the isolate to the nonpathogenic strains P. dispersa and P. wallisii. Also, the structure of its genome, the presence of a range of plant-associated genes, and the absence of major virulence factors suggest an epiphytic lifestyle of Pantoea sp. P37 rather than that of a human pathogen.
A total of four plasmids were present in Pantoea sp. P37 which is a common feature of the Pantoea genus. The F+-mediated conjugation system, responsible for horizontal gene transfer, is a major feature of the Pantoea genus providing biological robustness by the integration of GI (De Maayer et al., 2012). The GI found in Pantoea sp. P37 consisted of genes indicating environmental adaption processes. One was tehB encoding a tellurite methyltransferase and tehA encoding a resistance protein against the toxic tellurium dioxide (Chasteen, Fuentes, Tantaleán, & Vásquez, 2009). Other clusters were the carotenoid BGS for protection against UV irradiation and genes supporting the utilization of sucrose (Reid & Abratt, 2005). No genes of the cluster encoding for components of the type III secretion systems (T3SS), an essential prokaryotic virulence determinant of invasive Gram-negative pathogenic bacteria for injection of effector proteins into plant, animal, or human host cells were found (Deng et al., 2017). Pantoea sp. P37 also lacked other virulent determinants including proteases and lipases which target structural components of host cells leading to the dissemination of infection (Titball, 1993).
Pantoea sp. P37 also did not harbor any major toxin-encoding sequences similar to Pantoea vagans C9-1, a commercially available biocontrol agent considered as safe (Smits et al., 2010). The only toxin found was the endotoxic surface lipopolysaccharide (LPS) which is part of the outer membrane of all Gram-negative bacteria, including the safe E. coli strain K12 (Brade, 1999). Identified critical clusters like the type VI secretion system (T6SS), chemotactic flagella, or bacteriocin biosynthetic cluster are predominately survival mechanisms used for the ecological adaption to changing environmental processes and interbacterial defense and are considered as safe (Cotter, Ross, & Hill, 2013;Russell, Peterson, & Mougous, 2014;Wadhams & Armitage, 2004). However, the presence of multiple multidrug efflux pumps, MFS domains and the resistance against β-lactam, virginiamycin, and bicyclomycin needs to be considered.
The heterologous expression of the rhlAB operon in Pantoea sp.
P37 resulted in the production of the mono-rhamnolipid Rha-C 10 -C 10 with a titer of approximately 0.4 g/L. In contrast, in the early research stage, the recombinant strain of Pseudomonas putida KT2440 produced a total rhamnolipid concentration of 0.22 g/L (Wittgens et al., 2011). After fermentation process design experiments for optimized rhamnolipid production, the same strain was able to achieve a titer of 14.9 g/L (Beuker et al., 2016). The highest concentration of mono-rhamnolipids was obtained by integration of the rmlBCDA operon derived from P. aeruginosa into the rhamnolipid rhlAB expression vector. This is in accordance with a previous study where the availability of dTDP-L-rhamnose has been considered as a limiting factor for in the recombinant production of mono-rhamnolipids in E. coli (Cabrera-Valladares et al., 2006). The dTDP-L-rhamnose limitation may be due to the presence of several pathways competing for carbohydrates which are common in Enterobacteriaceae but not in Pseudomonads.
The relatively low production of mono-rhamnolipids following the integration of the rmlBCDA operon derived from P. aeruginosa  (Kumar, Mande, & Mahajan, 2015). These results suggest that the heterologous production of rhamnolipids induced stress to the cells which reduced the fitness of Pantoea sp. P37 explaining its inability to produce mono-rhamnolipids when rhlAB expression was driven by a strong promoter.  Grant Number: 1526004161).

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
Margarete Nawrath conceived the study, curated the data, performed formal analysis, involved in the investigation, contributed to the methodology and visualization, and wrote the original draft.
Christoph Ottenheim conceived the study, curated the data, performed formal analysis, involved in supervision and validation, and wrote, reviewed, and edited the manuscript. Jinchuan Wu conceived the study; acquired fund; involved in the investigation, project administration, resources, and supervision; and wrote, reviewed, and edited manuscript. Wolfgang Zimmermann conceived the study, contributed to the methodology, supporting, and supervision; wrote, reviewed, and edited the manuscript.

E TH I C S S TATEM ENT
None required.  F I G U R E A 2 Genome map of the Pantoea sp. P37 plasmids 2, 3 and 4. The outer ring displays all known CDS (blue), MFS transporter, multidrug efflux pumps and transporter (red), hypothetical proteins (purple), and phage-associated genes (yellow). The second outer ring displays genomic islands (red). The third outer ring shows the GC content. The fourth ring displays the GC skew