Pantoea ananatis carotenoid production confers toxoflavin tolerance and is regulated by Hfq‐controlled quorum sensing

Abstract Carotenoids are widely used in functional foods, cosmetics, and health supplements, and their importance and scope of use are continuously expanding. Here, we characterized carotenoid biosynthetic genes of the plant‐pathogenic bacterium Pantoea ananatis, which carries a carotenoid biosynthetic gene cluster (including crtE, X, Y, I, B, and Z) on a plasmid. Reverse transcription–polymerase chain reaction (RT‐PCR) analysis revealed that the crtEXYIB gene cluster is transcribed as a single transcript and crtZ is independently transcribed in the opposite direction. Using splicing by overlap extension with polymerase chain reaction (SOE by PCR) based on asymmetric amplification, we reassembled crtE–B, crtE–B–I, and crtE–B–I–Y. High‐performance liquid chromatography confirmed that Escherichia coli expressing the reassembled crtE–B, crtE–B–I, and crtE–B–I–Y operons produced phytoene, lycopene, and β‐carotene, respectively. We found that the carotenoids conferred tolerance to UV radiation and toxoflavin. Pantoea ananatis shares rice environments with the toxoflavin producer Burkholderia glumae and is considered to be the first reported example of producing and using carotenoids to withstand toxoflavin. We confirmed that carotenoid production by P. ananatis depends on RpoS, which is positively regulated by Hfq/ArcZ and negatively regulated by ClpP, similar to an important regulatory network of E. coli (HfqArcZ →RpoS Ͱ ClpXP). We also demonstrated that Hfq‐controlled quorum signaling de‐represses EanR to activate RpoS, thereby initiating carotenoid production. Survival genes such as those responsible for the production of carotenoids of the plant‐pathogenic P. ananatis must be expressed promptly to overcome stressful environments and compete with other microorganisms. This mechanism is likely maintained by a brake with excellent performance, such as EanR.


| INTRODUC TI ON
Carotenoids are widely used in functional foods, cosmetics, and health supplements, and their importance and scope of use are continuously expanding (Ram et al., 2020;Song et al., 2013).
Carotenoids are produced by plants and microorganisms, including algae, fungi, yeast, and bacteria, but animals must obtain carotenoids from dietary sources. Interestingly, aphids, which are capable of synthesizing carotenoids, are reported by later gene transfer from fungi (Moran & Jarvik, 2010). Several carotenoid-producing bacteria have been identified (Dufossé et al., 2005;Fasano et al., 2014;Fidan & Zhan, 2019;Lorquin et al., 1997;Lu et al., 2017;Ram et al., 2020;Sajilata et al., 2008;Sedkova et al., 2005;Virtamo et al., 2014). Carotenoids are highly hydrophobic, restricted to essential parts of the complex membrane and cell wall in bacteria, and mainly responsible for enhancing various functions related to the cell membrane and walls, including physical strength, fluidity, cell wall rigidity, and lipid peroxidation (Kirti et al., 2014;Lutnaes et al., 2004;Vila et al., 2019). Several functions are closely related to the habitats of bacteria; in particular, the carotenoids of bacterial species living in low-or high-temperature environments are used to control the membrane fluid, while those of bacteria continuously exposed to UV radiation increase tolerance to UV (Dundas & Larsen, 1963;Kunisawa & Stanier, 1958;Mathews & Sistrom, 1959, 1960Mostofian et al., 2020;Stanier, 1959). Also, carotenoids aid bacteria in combating stresses related to oxidation, salt, and desiccation (Oren, 2009;Tian & Hua, 2010). When bacteria are placed in a stressful environment, carotenoid production increases to protect against particular stressors, such as temperature, salt, light, and acidity (Paliwal et al., 2017;Ram et al., 2019). This is consistent with the fact that bacterial carotenoid production is closely related to habitat characteristics.
Bacteria are surprisingly rich producers of carotenoids. However, bacteria with low carotenoid content are unsuitable for commercial use. The production of plant-based carotenoids in bacteria is easier than in eukaryotic organisms such as yeasts, fungi, and plants (Ram et al., 2020). Previously, the biosynthesis of carotenoids has relied on bacterial carotenoid genes and DNA recombination techniques.
Because these methods depend on restriction sites, generating recombinant DNA fragments and rearranging multiple carotenoid genes is problematic. The technique of splicing by overlap extension by polymerase chain reaction (SOE by PCR) using asymmetric amplification was first developed for introducing mutations into the center of a PCR fragment (Higuchi et al., 1988;Mullis et al., 1986), making site-directed mutagenesis more flexible. Horton et al. (1989) modified SOE by PCR to allow DNA segments from two different genes to be spliced together by overlap extension. SOE has been used to enhance site-directed mutagenesis (Duan et al., 2013;Hussain & Chong, 2016;Xiao et al., 2007), generation of non-polar, markerless deletions in bacteria Merritt et al., 2007;Xu et al., 2013), multiple-site fragment deletion (Zeng et al., 2017), and generation of hybrid proteins of immunological interest (Warrens et al., 1997).
Pantoea ananatis is considered an emerging pathogen based on the increasing number of reports of diseases occurring in various economically important crops worldwide. This pathogen can also infect humans and numerous insects (Coutinho & Venter, 2009;Dutta et al., 2016;Weller-Stuart et al., 2017) and cause bacteremia infection (De Baere et al., 2004). P. ananatis PA13 causes plant diseases such as rice grain rot, sheath rot, and onion center rot disease in Korea . This pathogen is a potential threat to stable rice production, in particular during the growing season, when the weather is hot and humid. The pathogenicity of this bacterium is controlled by bacterial quorum sensing (QS), which is bacterial cell-to-cell communication with extracellular signaling molecules called autoinducers that are present in the environment in proportion to cell density (Lee, 2015;Morohoshi et al., 2007;Platt & Fuqua, 2010). The QS system facilitates community coordination of gene expression and benefits group behaviors. QS of P. ananatis, which uses EanRI homologous to P. stewartii subsp. stewartii EsaRI, has revealed that EanR negatively regulates self-expression and EPS production, but not eanI expression (Beck von Bodman & Farrand, 1995;Lee, 2015;Minogue et al., 2005;Morohoshi et al., 2007). In P. ananatis, 3-oxo-hexanoyl homoserine lactone (3-oxo-C6AHL) and hexanoyl homoserine lactone (C6AHL) signals are generated by EanI and secreted extracellularly. AHL signals bind EanR, an AHL receptor; this interaction de-represses the EanR negative regulator (Morohoshi et al., 2007). We revealed that EPS production, the hypersensitive response in tobacco, and virulence in rice are regulated by AHL-mediated QS in P. ananatis PA13 (Lee, 2015). In a previous study by Morohoshi et al. (2007), the QS system of onion pathogenic P. ananatis regulated EPS biosynthesis, biofilm formation, and infection of onion leaves.
Hfq, a ring-shaped hexameric RNA binding protein, has many important physiological roles that are mediated by interaction with Hfq-dependent small RNAs (sRNAs) in bacteria (Brennan & Link, 2007). Hfq was first reported in Escherichia coli as a host factor important in the replication of bacteriophage Qβ (Muffler et al., 1996). Hfq regulates the stress response protein RpoS, which controls many stress response genes (Brown & Elliott, 1996;Hwang et al., 2011;; it also regulates virulence in several pathogenic bacteria (Chao & Vogel, 2010;Sittka et al., 2007;Zeng et al., 2013). Also, it modulates a wide range of physiological responses in bacteria. The hfq deletion mutant exhibits several different phenotypes (Figueroa-Bossi et al., 2006).
It is established that the Hfq ArcZ →RpoS ʵ ClpXP regulatory networks are based on E. coli. RpoS is positively regulated by Hfq and its cognate sRNA ArcZ. RpoS levels are kept low by constitutive degradation of the ClpXP protease until the stationary phase (Raju et al., 2012). RpoS-dependent carotenoid production in P. agglomerans (formerly Erwinia herbicola) has been previously reported (Becker-Hapaka et al., 1997). In this study, we therefore investigated how QS, Hfq, and RpoS are involved in and regulate the carotenoid production in P. ananatis. Here, we also found that carotenoids were responsible for toxoflavin tolerance in P. ananatis.

| Bacterial strains and plasmids
Bacterial strains and plasmids used in this study are listed in Table   A1. E. coli strains were cultured on the lysogeny broth (LB) medium at 37°C. The P. ananatis PA13 was cultivated at 28°C on LB medium.

| DNA manipulation and data analyses
Manipulation of genomic DNA and plasmids and DNA cloning were performed as previously described (Sambrook & Russell, 2001).
Restriction enzymes (TaKaRa) were used for DNA digestion and modification. DNA sequencing was performed by Macrogen (Seoul). DNA sequences were analyzed using the BLAST program at the National Center for Biotechnology Information (Gish & States, 1993), MEGALIGN (DNASTAR, Madison, WI, USA), and GENETYX-WIN software (Genetyx).

| Carotenoid genes
Genomic DNA of P. ananatis PA13, a bacterial pathogen of rice, was used as the template to amplify carotenoid biosynthetic genes. The carotenoid genes are located on a plasmid (PAGR_p; CP003086).

| Strategy for carotenoid gene reassembly
The reassembly of the carotenoid genes responsible for synthesizing phytoene, lycopene, and β-carotene was performed as described previously (Horton et al., 1995). The sequences of the eight primers used for reassembly are listed in Table A2. Primers "a" and "d," "a" and "f," and "a" and "h" are the flanking primers for PCR amplification of the final reassembled products. Primers "b" and "c," "d" and "e," and "f" and "g" are the SOE primers. Bases have been added to the 5′ ends of the primers in each pair to render them complementary. All of the complementary sequences have been added between primers b-c, d-e, and f-g. During SOE, the upper strands of AB and the lower strands of CD overlap to act as primers ( Figure A1). Fragment AB was PCR amplified from crtE, and fragment CD from crtB. Fragment EF was PCR amplified from crtI, and fragment GH from crtY. The SOE primers "b" and "c" were used to modify the PCR products of the two sequences to have an identical sequence (Table A2). Figure A1b shows the reassembly of crtE-B genes for phytoene biosynthesis. The upper strands of F I G U R E 1 Genetic map and putative pathway responsible for carotenoid biosynthesis by P. ananatis PA13 (CP003086) and P. agglomerans Eho10 (M87280). (a) The carotenoid gene cluster of P. ananatis consisted of crtE-X-Y-I-B and Z and that for P. agglomerans of crtE-idi-crtX-Y-I-B and Z. Gene numbers were shown on the carotenoid gene map. (b) The putative carotenoid biosynthetic pathway of P. ananatis inferred according to the pathway of Pantoea species (Misawa et al., 1995) and plants (Guerinot, 2000) AB and the lower strands of CD overlap to act as primers when the PCR products are mixed, denatured, and reannealed during PCR.
Fragments of AD are formed when this overlap is extended by a polymerase. Figure A1c shows the crtE-B-I gene reassembly for lycopene biosynthesis in which the upper strands of AD and the lower strands of EF overlap to act as primers when the PCR products are mixed, denatured, and reannealed during PCR. Fragments of AF are formed when this overlap is extended by the polymerase. Similarly, the upper strands of AF and the lower strands of GH overlap to act as primers when the PCR products are mixed, denatured, and reannealed during PCR. Fragments of AH are formed when this overlap is extended by the polymerase ( Figure A1d). The XhoI recognition sequence and lacZ RBS were introduced at the beginning of the SOE-AB products.

| Purification and cloning of SOE fragments
The SOE products for use as templates were purified by electrophoresis in agarose (0.8% agarose, Promega) in the TAE buffer (40 mM Tris-acetate, 1 mM ethylenediaminetetraacetic acid) with 0.5 µg/ml ethidium bromide. DNA was recovered from the gel fragment using a DNA Purification Kit (GeneAll, Seoul, South Korea). The final recombinant products were gel-purified before cloning.
The SOE products were TA-cloned into pGEM-T Easy (Promega) and sequenced by Macrogen Services (Daejeon, South Korea). Error-free clones were digested with Xhol and Sacl and ligated into the corresponding positions in pBBR1MCS5 (Kovach et al., 1995).

| RT-PCR analysis of wild-type P. ananatis carotenoid cluster genes
Wild-type P. ananatis PA13 was grown in LB medium to exponential growth phase (12 h after inoculation); total RNA was isolated using an RNeasy Mini Kit according to the supplier's instructions (Qiagen); the RNA samples were treated with RQ1 DNase (Promega) to remove any contaminating DNA. RT-PCR was performed according to a previous report (Kim et al., 2004) as follows. Total RNA from P. ananatis PA13 was reverse transcribed into cDNA using M-MLV reverse transcriptase as described by the manufacturer (Promega) at 50°C for 1 h, followed by 5 min at 75°C. Next, PCR was performed using a T100 Thermal Cycler (Bio-Rad) under the following conditions: 96°C for 2 min, followed by 40 cycles of 96°C for 1 min, 50°C for 1 min, and 72°C for 1 min. The following primers were used for RT reactions, RT1 (crtB), and RT2 (crtZ). The following PCR primers were used: PCR1f and PCR1r; PCR2f and PCR2r; PCR3f and PCR3r; PCR4f and PCR4r; and PCR5f and PCR5r (Table A3). Southern hybridization and DNA sequencing were carried out to confirm the RT-PCR products. As a positive control, pCOK218 DNA was used. As a negative control, PCR reactions with the same primer sets were performed using RNA samples that had not been reverse transcribed.

| RT-PCR analysis of reassembled carotenoid cluster genes
Each SOE product was TA-cloned into pGEM-T Easy vector (Promega, Madison, WI, USA) and sequenced to confirm the presence of the DNA sequences (Macorgen Inc., Daejeon, South Korea).
Escherichia coli harboring recombinant pYS71, pYS69, or pYS76 was grown in LB medium to exponential growth phase (12 h after inoculation). The following primers were used for RT reactions: RT3 (crtZ); RT4 (crtI); and RT5 (crtY). The following PCR primers were used: PCR6f and PCR6r; PCR7f and PCR7r; and PCR8f and PCR8r (Table A3). Southern hybridization and DNA sequencing were carried out to confirm the RT-PCR products. As positive controls, pYS71, pYS69, and pYS76 were used. As a negative control, PCR reactions with the same primer sets were performed using RNA samples that had not been reverse transcribed.

| HPLC
For carotenoid extraction and HPLC analysis, transformed E. coli DH5α harboring pYS71, pYS69, or pYS76 was cultured in 250 ml flasks containing 50 ml of LB broth with 25 μg/ml gentamicin at 37°C for 24 h. After centrifugation at 10,000 g for 10 min, the cultured cells were repeatedly extracted with 3 ml of acetone for lycopene and β-carotene or ethanol for phytoene until the color was completely lost. The extracted solution was centrifuged and filtered through a GHP membrane (0.45 μm pore size). HPLC was performed using 20 μl of a prepared sample, with solvent A (60% acetonitrile, 38% ethyl acetate, 2% acetic acid) and solvent B (100% methanol) as the mobile phase, on a C 18 Shim-pack GIS-DOS column (4.6 × 250 mm, 5 μm; Shimadzu) as a fixed phase at a flow rate of 1.5 ml/min. β-Carotene was measured at 450 nm using a photodiode array detector. Lycopene was measured at 470 nm and phytoene at 280 nm. Phytoene, lycopene, and β-carotene standards were purchased from Sigma-Aldrich, Inc.

| Generation of lacZY-integrations and nonpolar deletion mutants lacZY transcriptional integration mutagenesis (Campbell insertion)
was performed as previously reported (Xu et al., 2013). An internal DNA fragment of eanI was amplified with EanI-1E-1 and EanI-2 K (Table A3). The partial eanI fragment was purified, cloned into pGEM-T Easy (Promega), and confirmed by sequencing. For recombinational mutagenesis, the EcoRI/KpnI-digested eanI fragment was cloned into the pVIK112 suicide vector (Kalogeraki & Winans, 1997), creating pCOK153. The parent strain PA13 was conjugated with pCOK153, and kanamycin-resistant colonies were selected. The mutants were confirmed by PCR using a primer that anneals upstream of the truncated fragment and the primer LacFuse followed by sequencing. We constructed rpoS and crtE null mutants using the same method as described previously.
Non-polar deletion mutagenesis was performed as previously reported (Xu et al., 2013). We amplified upstream and downstream fragments (approximately 450 bp) of the targeted gene region by PCR using the corresponding primer pairs (Table A3). After purification, the fragments were fused by overlap PCR. The final PCR products were cloned into pGEM-T Easy and confirmed by DNA sequencing. The fragments were excised using appropriate restriction enzymes and ligated into the suicide vector pNPTS138-R6 KT (Lassak et al., 2010). The resulting plasmids were introduced into PA13 by conjugative mating, and mating cells were spread on an LB medium containing kanamycin and rifampicin. Single-crossover integrates were selected on LB plates containing kanamycin and rifampicin. Single colonies were grown overnight in LB with rifampicin (25 μg/ml) and plated on LB containing 5% (w/v) sucrose to select for plasmid excision. We checked kanamycin-sensitive colonies for targeted deletion with colony PCR using primers bracketing the location of the deletion.

| Toxoflavin and UV radiation tolerance
Overnight cultures of the PA13 derivatives were sub-cultured and grown for an additional 12 h. A 100-μl aliquot was removed, and serially diluted 10-fold and 10-μl of each culture was spotted on LB agar plates supplemented with 20 μg/ml toxoflavin. The spotted plates were incubated at 28°C for 36 h.
For the UV radiation tolerance assays, PA13 derivatives were spotted on LB plates using the above procedure and treated as previously described (Mohammadi et al., 2012).

| Carotenoid production
To determine the carotenoid content of cells, P. ananatis strains were grown in 5 ml of LB medium at 28°C. Cells were harvested by centrifugation at 10,000 g for 1 min and suspended in 1 ml of methanol.
The samples were vortexed for 10 min and centrifuged at 10,000 g for 10 min, and the methanol supernatant containing carotenoids was transferred to a new tube. We quantified the carotenoid content of the extracts by measuring the absorbance at 450 nm using a Genesys 10S UV-VIS spectrophotometer (Thermo Fisher Scientific).

| AHL signal assay
The isolation and purification of AHLs were performed as described by Kim et al. (2004). The culture supernatants from time course cultures of P. ananatis PA13 and mutants were extracted with ethyl acetate (1:1). The ethyl acetate layer was dried, and the residue was dissolved in methanol. The ethyl acetate extracts were applied to C 18 reversed-phase TLC plates (Merck) and developed with 60% methanol. The TLC plates were dried in a fume hood and overlaid with soft agar containing Chromobacterium violaceum CV026 cells cultured overnight. The plates were incubated at 28°C overnight.

| β-Galactosidase assay
We generated non-polar deletions of lacZY genes from wild-type PA13 named PA13L and confirmed that all traits were identical in the two strains. Wild-type and mutant backgrounds used in the β-galactosidase assays were PA13L. All of the test strains were grown for 20 h and sub-cultured in LB broth at 28°C. The assays were performed using exponential-phase cultures at an OD 600 of ~0.4 as described previously (Choi et al., 2019).

| Identification of the carotenoid biosynthetic gene cluster in P. ananatis PA13
We previously reported the whole genome sequence of P. ananatis PA13 (Choi, Lim, et al., 2012), which revealed a carotenoid gene cluster on a plasmid (PAGR_p; CP003086). Figure 1 shows the genetic map and putative pathway responsible for carotenoid biosynthesis in P. ananatis PA13. The open reading frames (orfs) in the carotenoid biosynthetic gene cluster were analyzed and annotated as crtE, crtX, crtY, crtI, crtB, and crtZ in sequence. When comparing the crt gene clusters between P. ananatis and the genetically close species P. agglomerans, there is a significant difference in the position of idi, which is located in the chromosome in the former strain (PAGR_g2908) and between crtE and crtX in the latter. The structure of the other genes is identical in the two strains ( Figure 1a).
The putative carotenoid biosynthetic pathway of P. ananatis was inferred from the pathways of Pantoea species (Hundle et al., 1994;Misawa et al., 1995) and plants (Guerinot, 2000). Carotenoid biosynthesis begins with isomerization of isopenthyl diphosphate (IPP) from the mevalonate pathway to produce dimethylallyl diphosphate (DMAPP) in a reaction catalyzed by IPP isomerase encoded by idi. Carotenoids are produced from the common precursor farnesyl diphosphate (FPP).
The addition of a further IPP molecule yields geranylgeranyl diphosphate (GGPP) in a reaction catalyzed by GGPP synthetase (encoded by crtE). The next step in the carotenoid pathway is the head-to-head condensation of two molecules of GGPP to produce phytoene in a reaction catalyzed by phytoene synthase (encoded by crtB). Sequentially, the involved enzymes include phytoene desaturase (encoded by crtI), lycopene β-cyclase (crtY), β-carotene hydroxylase (crtZ), and zeaxanthin glucosyltransferase (crtX) (Figure 1b).

| Carotenoid biosynthetic cluster genes crtEXYIB of P. ananatis are polycistronic
We performed reverse transcription-polymerase chain reaction (RT-PCR) to determine whether the wild-type P. ananatis carotenoid biosynthetic cluster genes are polycistronic. We used five sets of primers to amplify crtE-X, X-Y, Y-I, I-B, and B-Z. RT-PCR followed by Southern hybridization indicated that the P. ananatis carotenoid biosynthetic cluster genes crtEXYIB are transcribed as a single transcript, and crtZ is transcribed as an independent single transcript in the opposite direction ( Figure 2).

| Cloning of SOE fragments and RT-PCR analysis
We used RT-PCR to determine whether the reassembled crtEB, crtEBI, and crtEBIY clones on the plasmids pYS71, pYS69, and pYS76 ( Figure A2) are transcribed as a single transcript. We used three sets of primers to amplify crtE-B, B-I, and I-Y. RT-PCR followed by Southern hybridization indicated that the reassembled crtE-B, crtE-B-I, and crtE-B-I-Y clones on the plasmids pYS71, pYS69, and pYS76 are transcribed as single transcripts ( Figure A3).

| Carotenoid production in E. coli
To determine whether E. coli DH5α transformed with pYS71/

| Carotenoid confers P. ananatis with tolerance to toxoflavin and UV radiation
Toxoflavin is a phytotoxin produced by B. glumae, a rice grain pathogen that shares rice environments with P. ananatis and has antibacterial properties. To determine whether the carotenoid production in P. ananatis is responsible for tolerance to toxoflavin and UV radiation, we generated a polarized crtE::pCOK184 mutant by Campbell insertion (Figure 4a). Complementation plasmid pCOK218 was also generated by cloning the carotenoid biosynthetic genes crtE-Z into pBBR1MCS5 (Figure 4a), which recovered the carotenoid deficiency in the crtE::pCOK184 mutant (Figure 4b). The wild-type is sensitive to toxoflavin concentrations >20 µg/ml (Figure 4b). The crtE::pCOK184 mutant exhibited lower tolerance than the wild-type to 20 µg/ml toxoflavin; however, the wild-type and complementation strain (+) times lower than that of the wild-type ( Figure A4).

| Carotenoid production depends on RpoS, which is positively regulated by Hfq/ArcZ and negatively by ClpXP in P. ananatis
The colonies of ∆rpoS and ∆hfq mutants were white, and neither produced carotenoids ( Figure 5); however, colonies of complementation strains (+) carrying pCOK312 and pCOK335, respectively, were orange and produced carotenoids. Colonies of the ∆arcZ mutant were faint orange and exhibited a slight reduction in carotenoid production ( Figure 5), indicating involvement in carotenoid production. Colonies of the ∆clpP mutant were dark orange and exhibited an approximately twofold increase in carotenoid production, indicating a negative carotenoid regulation via RpoS inhibition ( Figure 5).
Complementation strains (+) of ∆arcZ and ∆clpP mutants carrying pBS28 and pOR78, respectively, produced amounts of carotenoids similar to that of the wild-type. These results suggest that carotenoid production of P. ananatis depends on RpoS, which is positively regulated by Hfq/ArcZ and negatively by ClpP, similar to an important regulatory network of E. coli (Hfq ArcZ →RpoS ʵ ClpXP).

| EanR negatively regulates carotenoid production in P. ananatis
A previous report examined EanR de-repression in the QS system of P. ananatis, which causes center rot disease in onion (Morohoshi et al., 2007). QS of P. ananatis PA13 is also similar to that of P. stewartii (Minogue et al., 2005). Figure A5 shows the QS system of P.
ananatis PA13. The eanR and eanI genes are transcribed in the opposite direction, and the lux box is at the eanR gene promoter region ( Figure A5a). To determine whether eanI expression is under the control of EanR, we constructed a lacZY integration of eanI::pCOK153 (i.e., pVIK112 carrying eanI truncated at both ends) in PA13L and PA13L∆eanR mutant backgrounds using Campbell insertion ( Figure   A5a). QS signal production of the mutants was confirmed using thinlayer chromatography (TLC) and a Chromobacterium indicator strain.
The eanI mutant did not produce QS signals, whereas the eanR The wild-type and crtE::pCOK184 mutant carrying pCOK218 exhibited greater toxoflavin tolerance than the crtE mutant; however, the crtE::pCOK184 mutant was more sensitive than the wild-type to toxoflavin at 20 µg/ml. Pantoea ananatis PA13 is sensitive to toxoflavin concentrations >20 µg/ml mutant did ( Figure A5b). The expression of eanI was not decreased in the ∆eanR mutant background or increased by the addition of 3-oxo-C6AHL or C6AHL ( Figure A5c). These data indicate that the expression of eanI is not under the control of EanR.

| EanR negatively regulates carotenoid production via inhibition of rpoS in P. ananatis
To determine whether rpoS expression is regulated by EanR, we con-  (Figure 6e), there is currently no direct evidence for this. We analyzed the candidate lux box(s) in the crtEXYIB gene cluster or the promoter region of the crtZ gene, but did not find it, but did not find it.

| QS is delayed in the absence of Hfq
To elucidate the relationship between Hfq and QS, we performed QS signal-production assays with wild-type, ∆hfq mutant, and ∆hfq complementation strains. QS signals were extracted in the mid/late log phase (OD 600 = 0.9, 1.5, and 1.8) and developed on C 18 reversedphase TLC plates. QS signaling in the ∆hfq mutant (−) decreased significantly but recovered to the level of the wild-type after transforming with pCOK335 (+; pLAFR3::hfq; Figure 7a,b). These results suggest that Hfq positively regulates QS signal production. We also examined whether the reduction in QS signal in the ∆hfq mutant was due to bacterial growth; our results showed that growth in ∆hfq was not retarded compared with the wild-type strain (data not shown).
Using β-galactosidase activity assays, we found that expression of eanI decreased significantly in the absence of Hfq (−) but recovered by transformation with pCOK335 (+; pLAFR3::hfq) (Figure 7c). This is consistent with the finding that QS signal production in the ∆hfq mutant was significantly lower (Figure 7a,b). These results indicate that Hfq positively regulates QS, which is delayed in the absence of Hfq in P. ananatis.

| DISCUSS ION
Pantoea ananatis is an emerging plant pathogen that causes severe loss of many crops and trees, such as corn, onion, rice, and Eucalyptus, worldwide (Coutinho & Venter, 2009;Weller-Stuart et al., 2017). This bacterium has also been associated with insects and humans. It is considered pathogenic because of its virulence in a wide variety of plant hosts and saprophytic ability in diverse ecological niches (Coutinho & Venter, 2009). In this study, we focused on the carotenoid biosynthesis gene cluster of P. ananatis and performed gene reassembly for carotenoid production. Here, we investigated the ecological and physiological functions of regulatory mechanisms of the carotenoid production of P. ananatis.
Much effort has been focused on the biosynthesis of carotenoids using bacterial carotenoid genes (Guerinot, 2000;Lee et al., 2003;Misawa et al., 1995). Techniques based on recombining DNA sequences rely on restriction sites, so the primer must contain the introduced restriction site, which should not be in the center of the fragment (Higuchi et al., 1988;Horton et al., 1989;Mullis et al., 1986). Moreover, if multiple cloning vectors are to be used, plasmid incompatibility is also F I G U R E 5 Carotenoid production in the wild-type (W), ∆rpoS, ∆hfq, ∆arcZ, ∆clpP, and complementation (+) strains. Values are means ± standard deviation (SD) of three independent experiments. *p < 0.05; **p < 0.01; ***p < 0.001 vs. wild-type a limiting factor. Here, we applied the SOE by PCR technique to recombine DNA sequences without relying on restriction sites. In this report, we describe the reassembly of the genes encoding bacterial We used a DNA template from the rice pathogenic bacterium P.
ananatis in SOE by PCR, which is controllable and independent of restriction sequences, for the carotenoid gene reassembly. Pantoea agglomerans, which causes palea browning of rice, is a genetically close species to P. ananatis with which it shares a biological niche.
The organization of the carotenoid biosynthetic gene clusters of the two strains is identical except idi. Interestingly, P. agglomerans has an idi gene between crtE and crtX, which distinguishes it from P. ananatis, suggesting that idi could be used to distinguish genetically similar bacteria.
SOE is a novel PCR-mediated recombinant DNA technology that does not rely on restriction sites, so its coverage is considerably F I G U R E 6 EanR negatively regulates carotenoid production via inhibition of rpoS. (a) QS signal production of the wild-type and ∆eanI, ∆eanR, and ∆eanI-R mutants as well as the ∆eanI-R mutant carrying pCOK199 on C. violaceum CV026 biosensor-embedded plates. (b) Carotenoid production of the wild-type and ∆eanI, ∆eanR, and ∆eanI-R mutants as well as the ∆eanI-R mutant carrying pCOK199. (c) Quantification of carotenoid production of the PA13 derivatives. Carotenoid production was identical to that shown in (b). Values are means ± standard deviation (SD) of three independent experiments. ***p < 0.001 versus wild-type. (d) β-Galactosidase activity reporting rpoS expression. rpoS expression was induced in the absence of EanR and decreased in the absence of EanI, indicating that EanR negatively regulates rpoS expression and QS signals de-repress EanR. Values are means ± standard deviation (SD) of three independent experiments. ***p < 0.001 versus PA13L. (e) Genetic map of rpoS locus and putative lux box. Inverted repeat sequences are shown in bold wider than standard restriction enzyme-based methods for gene recombination. This enables finer control over recombination for genetic engineering. Besides, the sequence of the overlap region is determined by primer design, allowing simultaneous non-polar mutagenesis, site-directed mutagenesis, and recombination. In this study, we applied this technically simple and rapid recombinant DNA technique to the biosynthesis of three carotenoids. The technique will likely be suitable for the recombination of multiple genes.
In bacteria, carotenoids are closely related to the conditions of the surrounding environment. We found that the UV radiation tolerance of P. ananatis was due to the carotenoids they produce.
These results are consistent with those regarding P. stewartii subsp.
stewartii (Mohammadi et al., 2012). Considering the plant environment (particularly rice) in which P. ananatis lives, UV radiation tolerance is advantageous for survival. Interestingly, these carotenoids are unique in that they also make P. ananatis tolerant to toxoflavin.
Toxoflavin was proposed to produce superoxide (O 2 -) and H 2 O 2 during autorecycling oxidation processes under oxygen and light (Latuasan & Berends, 1961;Nagamatsu et al., 1982). Thus, the carotenoid production in P. ananatis can be considered a survival strategy to reduce oxidative stress caused by toxoflavin. These results were consistent with previous studies showing that carotenoids reduce oxidation stress in bacteria (Oren, 2009;Tian & Hua, 2010). This is the first report on the production and use of carotenoids to overcome toxoflavin, resulting from P. ananatis and B. glumae sharing the same rice environment. Although we evaluated whether the carotenoid production of PA13 confers resistance to additional antibiotics such as ampicillin and tetracycline, the carotenoid production of PA13 did not confer tolerance to each antibiotic (data not shown).
We found that QS and Hfq are directly or indirectly involved in regulating carotenoid production in P. ananatis PA13. Q S regulates an extensive range of functions, including bioluminescence, virulence, biofilm formation, DNA exchange, and sporulation in bacteria (Fuqua et al., 1996;Waters & Bassler, 2005). Hfq is a global RNA chaperone that interacts with sRNAs of diverse functions; it also regulates virulence and environmental stress in many plant and animal bacterial pathogens (Chao & Vogel, 2010;Ding et al., 2004;Shin et al., 2019;Zeng et al., 2013). The hfq mutant in Erwinia amylovora Ea1189 reduces virulence, amylovoran EPS production, biofilm formation, motility, and positive regulation of the type III secretion system (Zeng et al., 2013). In Pectobacterium carotovorum, the hfq mutant exhibits defects in motility, biofilm formation, sedimentation, and virulence (Wang et al., 2018). Hfq is also an important regulator of virulence, motility, and biofilm formation in P. ananatis LMG2665 (Shin et al., 2019). We found that Hfq regulates the expression of eanI encoding the QS signal synthase, which was confirmed by eanI expression and QS signal productivity assays. These results are consistent with the finding that Hfq regulates QS signal production directly via interactions with the AHL receptor ExpR in Sinorhizobium meliloti (Gao et al., 2015). QS systems integrate other global regulators, including noncoding sRNAs. This network is activated through the binding of Hfq and Hfq-dependent sRNA and controls gene expression via post-transcription regulation (Storz et al., 2005). There are several reports that the Hfq-dependent sRNAs Qrr1-4 and RsmY interact with Hfq to directly and indirectly control QS targets in Vibrio cholerae and Pseudomonas aeruginosa (Kay et al., 2006;Lenz et al., 2004). previously reported regulatory network Hfq ArcZ → RpoS ʵ ClpXP and that identified here, in which Hfq-controlled quorum signaling derepresses EanR to activate RpoS expression, thereby initiating carotenoid production. Carotenoid production confers tolerance to toxoflavin and UV radiation pPAR237 and pPAR238 are involved in regulating QS by activating EanI without genetic analyses. Further studies are needed to identify the sRNAs in P. ananatis. It was previously reported that EanR mediated QS regulation by de-repression as in P. stewartii (Beck von Bodman & Farrand, 1995;Morohoshi et al., 2007). In P. ananatis, EanR represses the ean box (lux box-like sequences) in the upstream region of eanR and adding AHL promoted dose-dependent de-repression (Morohoshi et al., 2007). This EanR-mediated QS regulation was similar to that of the close homolog EsaR in P. stewartii (Minogue et al., 2005). Overall, we found that QS signal production in P. coli. Here, we elucidated a regulatory network of carotenoid production involving Hfq-dependent QS-RpoS in P. ananatis. Hfq regulates the full production of QS signals, thereby de-repressing the EanR negative regulator to initiate RpoS expression (Figure 8).

| CON CLUS IONS
Microbial biotechnology allows bacterial carotenoids to be used as alternatives to plant-based carotenoids because of the ease of genetic manipulation of prokaryotes compared with eukaryotes, such as yeasts, fungi, and plants. Here, we used SOE by PCR for gene reassembly to redirect carotenoid synthesis from the plantpathogenic bacterium Pantoea ananatis. Using SOE by PCR, we reassembled crtE-B, crtE-B-I, and crtE-B-I-Y for phytoene, lycopene, and β-carotene production, respectively, using E. coli to express the reassembled operons. We found that carotenoids confer tolerance to the phytotoxin toxoflavin. The carotenoid production of P. ananatis depends on RpoS, which is positively regulated by Hfq/ArcZ and negatively by ClpP, similar to an important regulatory network of E. coli, Hfq ArcZ → RpoS ʵ ClpXP. We also demonstrated that carotenoid production is regulated by Hfq-controlled QS since the EanR negative regulator on RpoS must be expressed in the stationary phase.

CO N FLI C T O F I NTE R E S T
None declared. Writing-review & editing (lead).

E TH I C S S TATEM ENT
None required.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analyzed during this study are included in this paper.

Jinwoo Kim
https://orcid.org/0000-0003-1104-8088 A PPEN D I X F I G U R E A 1 Strategy for generating recombinant carotenoid genes responsible for synthesizing phytoene, lycopene, and β-carotene. PCR products with their overlapping regions aligned and the final rearrangement products are shown. The SOE by PCR products AD, AF, and AH are shown. In each case, the overlapping region between the primers, and the priming region in which each primer recognizes its template, was designed to have the ribosome binding sequence (RBS) of each gene. The XhoI recognition sequence and lacZ RBS were introduced at the beginning of SOE-AB products. Dotted arrows indicate the rearrangements of carotenoid genes F I G U R E A 2 Recombinant plasmids for rearrangement of the carotenoid genes responsible for synthesizing phytoene, lycopene, and β-carotene. The SOE by PCR products were first cloned into pGEM-T Easy, digested with Xhol and Sacl, and ligated into the corresponding position of pBBR1MCS5. Open triangles indicate the lacZ RBS. The SacI site, which is dotted and parenthesized, was from the pGEM-T Easy vector (2.2, 3, and 5 kb XhoI-SacI fragments of pYS71, pYS69, and pYS76, respectively) as probes. Lanes 1-3, 4-6, and 7-9 correspond to the products of PCR1, PCR2, and PCR3, respectively. Lanes 1, 4, and 7: PCR products from the DNA template as positive controls; lanes 2, 5, and 8: PCR products from the RNA template as negative controls; and lanes 3, 6, and 9: RT-PCR products F I G U R E A 4 Carotenoid production confers UV radiation tolerance to P. ananatis. The wild-type and crtE::pCOK184 mutant harboring pCOK218 exhibited greater UV radiation tolerance than the crtE mutant at wavelengths of 320-400 nm for 20 s; however, the crtE::pCOK184 mutant showed lower UV radiation tolerance than the wild-type F I G U R E A 5 QS system of P. ananatis PA13. (a) Genetic map of the eanR and eanI loci of the QS system in P. ananatis PA13 and mutant generation. The putative lux box is upstream of eanR, for comparison, the esaR lux box of P. stewartii subsp. stewartii and lux box of Vibrio fischeri are presented. Campbell insertion and non-polar deletion mutants were generated to determine if eanR regulates the expression of eanI. (b) Characterization and quantification of AHL signals of the PA13 derivatives. (c) β-Galactosidase activity reporting eanI expression.