A Substrate-Activated Efflux Pump, DesABC, Confers Zeamine Resistance to Dickeya zeae

Zeamines are a family of newly identified phytotoxins and potent antibiotics produced by D. zeae EC1. Unlike most bacterial organisms, which are highly sensitive, D. zeae EC1 is tolerant to zeamines, but the mechanisms involved are unknown. Our study showed, for the first time, that a new RND efflux pump, DesABC, is indispensable for D. zeae EC1 against zeamines. We found that the DesABC efflux pump was zeamine specific and appeared to be conserved only in the Dickeya species, which may explain the high potency of zeamines against a wide range of bacterial pathogens. We also showed that expression of DesABC efflux system genes was induced by zeamines. These findings not only provide an answer to why D. zeae EC1 is much more tolerant to zeamines than other bacterial pathogens but also document a signaling role of zeamines in modulation of gene expression.

phytotoxins, known as zeamines (4,5). Zeamines are a family of structurally related polyamino compounds that play important roles in the pathogenicity of D. zeae EC1. Inactivation of zmsA, the key gene responsible for the biosynthesis of all zeamine compounds, abrogated the infectivity of D. zeae EC1 on rice, potato, and Chinese cabbage (4).
Zeamines are produced by Dickeya species and Serratia plymuthica strains with the zms gene cluster, including D. zeae EC1 and S. plymuthica RVH1 (4-9). Among them, zeamine, zeamine I, and prezeamines are the derivatives of a polyamino chain zeamine II, with polyketide moiety conjugating at the terminal amino group of zeamine II (4)(5)(6)(7)9). Apart from their important role in the virulence of D. zeae EC1, zeamines are also potent antibiotics with broad-spectrum activity against various organisms, including multidrug-resistant bacteria, fungi, oomycetes, and nematodes (8)(9)(10). Evidence shows that zeamines target the outer membrane of Gram-negative bacteria in a way reminiscent of the cationic antimicrobial peptide polymyxin B (11). Organization of the zms gene clusters is genetically well conserved in D. zeae EC1 and S. plymuthica RVH1, with genes encoding polyketide synthases (PKSs), nonribosomal peptide synthetases (NRPSs), and fatty acid synthases (FASs). In addition, five genes within the zms gene cluster were predicted to encode transporter proteins, including four encoding putative ATPases and permeases associated with the ABC transporter system and one encoding a potential HlyD superfamily protein (6,8,12). One of the predicted ABC transport systems encoded by zmn20 and zmn21 was proposed to be a zeamine transporter and associated with zeamine resistance in S. plymuthica RVH1 (6), but this speculation has not yet been validated experimentally.
Multidrug resistance (MDR) efflux pumps are membrane-associated proteins that can export a wide range of antibiotics and confer intrinsic antibiotic-resistant ability to bacteria. The efflux pumps can be classified into five superfamilies: MFS (major facilitator superfamily), ABC (ATP-binding cassette), SMR (small multidrug resistance), MATE (multidrug and toxic compound extrusion), and RND (resistance-nodulation-cell division) (13). In Gram-negative bacteria, RND efflux pumps play important roles in MDR due to their broad-spectrum substrate profile (14). The RND efflux pump is a tripartite complex system comprised of an outer membrane channel, an adaptor, and an inner membrane protein, all of which are required for the full function of antibiotic transportation (15). The genes responsible for encoding RND efflux pumps are commonly presented as a single operon in bacteria, like the MexAB-OprM efflux pump in Pseudomonas aeruginosa (16), but there are also exceptional cases with the gene encoding outer membrane channel protein placed in another location in the genome (17). In RND efflux pumps, antibiotic specificity is determined by the inner membrane protein. Antibiotics belonging to different families can enter into the inner membrane proteins through three putative entrance channels opening to the central cavity of inner membrane protein, the inner membrane, and periplasmic space of bacterial cells (18). The multiple active binding sites in the porter region of inner membrane proteins make it possible for the RND efflux pumps to transport a variety of structurally unrelated antibiotics produced by bacteria themselves (19) or from the environment (20,21).
How D. zeae EC1 protects itself from the antimicrobial activity of zeamines remains unknown. While the MICs of zeamines for most bacterial pathogens are low, in the range of 0.3 to 10 g/ml (9), our preliminary assay results showed that the zeamine producer D. zeae EC1 could tolerate up to 1,800 g/ml of zeamines, suggesting a high-level resistance mechanism(s) is encoded by the D. zeae EC1 genome. In this study, we tested whether the transporter genes within the zms gene cluster, which were speculated to play roles in zeamine resistance (6), and the adjacent genes encoding RND efflux pump are associated with zeamine resistance in D. zeae EC1. Our results rule out the possible involvement of the ABC transporter genes within the zms gene cluster in zeamine resistance but lead to identification of a RND efflux pump, DesABC, that confers a high level of zeamine tolerance in D. zeae EC1. Substrate specificity assay against a range of antibiotics showed that DesABC appeared to only confer resistance against zeamines. In addition, the DesABC efflux system was found to be functionally conserved in Dickeya species. Interestingly, consistent with the zeamine-specific pattern of DesABC, we found that the transcriptional expression of its coding genes, desAB, was stimulated by the presence of zeamines, suggesting that DesABC co-evolved with the genes encoding zeamine biosynthesis to ensure high-level production of the antibiotics.

RESULTS
The ABC transporter systems encoded by the zms gene cluster are not required for zeamine resistance in D. zeae EC1. As the transporter systems present within the antibiotic biosynthesis gene clusters often confer resistance to the encoded antibiotics, we conducted bioinformatics analysis of the five transporter genes, i.e., zmsP, zmsQ, zmsR, zmsL, and zmsM (NCBI accession no. WP_016943528.1, WP_016943529.1, WP_016943530.1, WP_016943542.1, and WP_016943543.1, respectively) within the zms gene cluster of D. zeae EC1 ( Fig. 1) (12). These genes share similar genetic organization (Fig. 1) and high levels of identity and similarity in amino acids (above 69% and 82%, respectively) (Table S2) to their homologs found in S. plymuthica and other Dickeya species and strains with the zms gene clusters (8). Among them, sequence analysis showed that zmsP encodes a potential HlyD family protein, zmsR and zmsM encode potential ABC transporter permeases, and zmsQ and zmsL encode potential ABC transporter ATPases. The homologs of zmsM and zmsL were hypothesized to encode a zeamine transporter, conferring self-resistance against zeamines in S. plymuthica RVH1 (6). As a previous study indicated that the HlyD family protein could form a functional transport system with ABC transporter proteins (22), ZmsP was considered a part of the ABC transporter systems encoded by the zms gene cluster. To elucidate the potential roles of these genes in zeamine resistance, in-frame deletion was performed to generate the deletion mutants of zmsR and zmsM, respectively, which represent the two putative transport systems encoded by the zms gene cluster of D. zeae EC1. Zeamines were purified from the cell culture of D. zeae EC1 and confirmed by liquid chromatography-mass spectrometry (LC-MS) (see Fig. S1 in the supplemental material) and used for determination of MIC against different bacterial strains in this study. The results showed that inactivation of zmsR or zmsM could not cause any change in the MIC of zeamines compared with those of the wild-type strain EC1 (data FIG 1 Organization of transporter genes within and adjacent to the zms gene cluster in Dickeya species and Serratia plymuthica strains. The organization of genes was drawn using Illustrator for Biological Sequences (34). Data were derived from NCBI and updated to 24 July 2018.
DesABC System Confers Zeamine Resistance ® not shown). These findings preclude the possible association of these ABC transport systems with zeamine resistance, and their roles remain to be further investigated.
DesABC efflux system is required for zeamine resistance. In addition to the five potential transporter genes within the zms gene cluster, further bioinformatics analysis unveiled two genes encoding RND efflux pump proteins located adjacent to the zms gene cluster of D. zeae EC1 and another three Dickeya species and strains ( Fig. 1). One of the gene clusters encodes a putative AcrA-like adaptor protein, and the other encodes a potential AcrB-like inner membrane protein. The AcrAB-TolC RND efflux system has been well characterized as being associated with multiple antibiotic resistance in Escherichia coli, including ␤-lactams, tetracycline, chloramphenicol, and rifampin, with acrAB located together in the genome and tolC at a distant location (15). We proposed to name these two genes desA and desB, for Dickeya efflux system proteins A and B (Fig. 1). Interestingly, the desAB genes are not present within the vicinity of the zms gene cluster in S. plymuthica strains (Fig. 1).
Topological analysis of DesB revealed typical features of inner membrane transporter protein in an RND efflux system with 12 transmembrane helix domains (TM) and 2 large periplasmic loops spanning from TM1 to TM2 and TM7 to TM8 (Fig. S2) (23). In RND efflux systems, a tripartite complex is required for the full function of substrate transportation. To identify the outer channel protein for DesAB, a BLAST search was conducted to identify the homolog of E. coli outer membrane protein TolC. The result showed that only one tolC homolog (74% identity and 86% similarity at the amino acid level) is present in the genome of D. zeae EC1, which was designated desC accordingly. To elucidate the role of the DesABC system in zeamine resistance, three des genes were deleted in-frame separately at the background of the strain defective in zeamine production, i.e., zmsA in-frame deletion mutant. Inactivation of desA and desB led to about an 8-fold decrease in the MIC of zeamines, while deletion of desC led to about a 32-fold decrease in MIC (Table 1). Consistent with the above-described results, in trans expression of desB and desC in the corresponding mutants could increase the zeamine resistance level of the mutants (Table 1). In addition, we found that heterologous expression of desABC under the control of the lac and tetO promoter in E. coli DH5␣ increased the MIC of zeamines by 2-fold (Table 1). These results demonstrate the role of the RND system DesABC in self-protection of D. zeae EC1 against the antimicrobial activity of zeamines.
DesABC efflux system is zeamine specific and functionally conserved in Dickeya species. DesABC belongs to the RND efflux systems, in which the inner membrane proteins associated with recognition and binding have been well characterized to aid in understanding their substrate profiles (14). For example, MexY from P. aeruginosa is required for streptomycin resistance (24), MexB and AcrB from P. aeruginosa and E. coli are associated with chloramphenicol and tetracycline resistance (25,26), CmeB from Campylobacter jejuni plays a role in resistance against ampicillin, chloramphenicol, gentamicin, and tetracycline (27), and AdeB in Acinetobacter baumannii BM4454 is involved in tetracycline, chloramphenicol, gentamicin, and kanamycin resistance (28). To understand the potential substrate profile of the DesABC efflux pump, a phylogenic tree was constructed with DesB of D. zeae EC1 and its homologs (sequence similarity above 93%) found by blastp search in Dickeya species, other proteobacterial species, including the homologs (sequence similarity above 80%) from S. plymuthica strains containing the zms gene cluster, as well as the above-mentioned inner membrane proteins with known functions (Table S3). The DesB phylogeny was largely consistent with known evolutionary relationships among the bacterial genomes. All of the Dickeya DesB proteins formed a monophyletic clade in the tree, whereas the homologs from other genera were more distantly related (Fig. S3). Notably, the inner membrane proteins with known substrate profiles were clustered together on the tree and showed considerable divergence from the DesB homologs from Dickeya species. The result suggests that DesB and its homologs from Dickeya species have a different substrate profile than their counterparts from other bacterial species. The substrate profile of DesABC was then examined experimentally by MIC assay. The results showed that in trans expression of desB from D. zeae EC1 or desB 3937 from D. dadantii 3937, which lacks the zms gene cluster, in the zmsA-desB double deletion mutant of D. zeae EC1 could fully restore the zeamine resistance ( Table 1), suggesting that the desB orthologs from other Dickeya species (Table S3) have a similar function in zeamine resistance. MIC assay was also performed using antibiotics which belong to different classes and have different targets (Table S4). The results showed that neither inactivation of desB nor overexpression of desAB genes in D. zeae EC1 could affect the MICs of ampicillin, tetracycline, kanamycin, gentamicin, streptomycin, and chloramphenicol (Table S5). The above data are consistent with the phylogenetic analysis results (Fig. S3), indicating that the DesABC system has a distinct substrate specificity.
DesABC efflux system is essential for D. zeae survival against zeamines. As the DesABC system was found to be required for zeamine resistance, a survival assay was conducted against zeamines using the zmsA mutant, the zmsA-desB double deletion mutant, and the double mutant complemented with desB. Bacterial cells were added to LS5 salts, without carbon source, supplemented with zeamines at a final concentration of 2-fold the MIC of the zmsA-desB double deletion mutant, and bacterial cell numbers were measured at different time points upon treatment with zeamines to evaluate the role of DesABC in D. zeae EC1 survival. The results showed that inactivation of desB resulted in a sharp decline in survival rate, whereas its parental and complemented strains could maintain upon to a three-log larger amount of survivors than the desB mutant at 30 min after treatment (Fig. 2). These results indicate that the DesABC efflux pump plays an indispensable role in the survival of strain EC1 when the bacterial cells were treated with zeamines.
DesABC efflux system confers bacterial self-resistance against zeamines at the late stage of bacterial growth. To investigate the protective spectrum of DesABC during cell growth, an in-frame deletion mutant of desB was constructed using D. zeae FIG 2 Survival analysis of the desB mutant and its parental and complementation strains treated with zeamines. Bacterial cells were measured at 10, 20, and 30 min after treatment with zeamines. The survival rate was expressed as the percentage of the colony counts of the control not exposed to zeamines. Data in the graph are the means from three repeats, and error bars are standard deviations.
DesABC System Confers Zeamine Resistance ® wild-type strain EC1. The cell growth curves and zeamine production of wild-type strain EC1 and desB mutant were compared in LS5 medium, which was optimized for zeamine production (29). At the early stage after inoculation (12 to 24 h), the growth rate and zeamine production were comparable between strain EC1 and the desB mutant (Fig. 3A). However, the growth of the desB mutant was arrested in the subsequent stages after 24 h (Fig. 3A), and similarly, accumulation of zeamines in bacterial supernatant was also flattened after 24 h (Fig. 3B). These results suggest that D. zeae cells could tolerate a certain level of zeamines in the absence of the DesABC efflux system, but along with bacterial growth and zeamine accumulation, the DesABC efflux pump To determine the relationship between expression of DesABC genes and zeamine resistance, the gfp coding region was placed under the promoter of desAB, and the transcriptional fusion construct pDesAB gfp was prepared. The expression of desAB was evaluated by monitoring the fluorescence of wild-type strain EC1 containing the pDesAB gfp construct grown in LS5 medium by using a CytoFLEX flow cytometer system. The results showed that expression of desAB was bacterial population density dependent, showing a basal-level expression at the early growth stage (12 to 20 h) and rapidly increased expression at 20 h onward after inoculation (Fig. 3C). As the DesABC system was critical to the bacterial growth at the late growth stage (Fig. 3A), the above results indicate that zeamine resistance mediated by the DesABC efflux system is positively related to the expression level of desAB genes.
Expression of desAB is induced by zeamines. Expression of desAB genes was consistent with zeamine production during cell growth (Fig. 3), suggesting that desAB expression is influenced by zeamines. To test this possibility, the pDesAB gfp construct was introduced into the D. zeae ΔzmsA mutant. Our previous results showed that deletion of zmsA abolished production of all the zeamines (4). The gfp expression level driven by the promoter of desAB in wild-type strain EC1 and the zmsA mutant was monitored at different time points during bacterial growth in LS5 medium. The results showed that the growth patterns of both strains were similar (Fig. 4A), but the DesABC System Confers Zeamine Resistance ® expression levels of desAB in these strains were varied substantially (Fig. 4B). The expression of desAB in strain EC1 was increased along with bacterial growth but remained flat in the toxin-minus ΔzmsA mutant (Fig. 4B).
For validation of the findings described above, the transcript levels of the DesABC efflux system genes in D. zeae strain EC1 and the zmsA mutant were determined at an optical density at 600 nm (OD 600 ) of about 1.5 (approximately 20 h after inoculation). The quantitative PCR (qPCR) results showed that although the transcript level of desC was comparable between strain EC1 and the ΔzmsA mutant (fold change of less than 2; data not shown), the desB transcript level in strain EC1 was significantly higher than that in the ΔzmsA mutant, which was hardly detectable (Fig. 4C). The basal level of expression of desAB noticed in the zmsA mutant coupled with their induced expression in wild-type strain EC1 suggest that expression of the desAB genes rely on the product of zmsA, i.e., zeamines.
To further confirm this assumption, expression of desAB was monitored in the ΔzmsA(pDesAB gfp ) strain treated with zeamines. The result showed that a supplement of zeamines at a final concentration of 100 g/ml did not decrease the cell growth of the ΔzmsA(pDesAB gfp ) strain but led to about a 30-fold higher expression of desAB genes compared with that of the control without addition of zeamines (Fig. 4D). Notably, a low concentration of zeamines, 5 g/ml, could significantly induce the expression of desAB genes in D. zeae EC1 and their homologs in D. dadantii 3937. As a control, we also tested whether expression of desAB could be induced by polymyxin B by incubation of the ΔzmsA(pDesAB gfp ) strain with polymyxin B at the same concentration as zeamines. The results showed that unlike zeamines, polymyxin B could not trigger the expression of desAB genes (Fig. 4D). Cumulatively, these results unveil a novel and specific signaling role of zeamines in triggering the expression of the desAB genes in Dickeya species and strains.

DISCUSSION
Unlike most bacterial pathogens, which are highly sensitive to zeamines (9, 10), the zeamine producer D. zeae EC1 can withstand a high level of zeamines. In this study, we identified a novel RND efflux pump, DesABC, that plays a role in resistance against zeamines, especially at the late stage of bacterial growth, when zeamines were accumulated at a high level. Null mutation of the efflux pump could lead to up to about a 32-fold decrease in zeamine resistance. In contrast, the DesABC efflux pump was not functional against a range of conventional antibiotics, including ampicillin, tetracycline, kanamycin, gentamicin, streptomycin, and chloramphenicol. Furthermore, we also showed that desAB expression was growth and zeamine dependent and documented a novel signaling role of zeamines in regulation of desAB transcription. Moreover, we found that deletion of the desB gene could substantially reduce the accumulation of zeamines, suggesting that zeamine biosynthesis and resistance are modulated by coordinated and sophisticated regulatory mechanisms.
The zeamine biosynthesis genes and desAB were clustered in the genomes of D. zeae EC1 and another three Dickeya species and strains according to bioinformatics analysis (Fig. 1). However, despite their functional relevance (Fig. 4), the results from this study suggest that desAB and zeamine biosynthesis genes are not tightly linked during evolution. This is evident as the genomes of multiple Dickeya species and strains, such as D. zeae Ech586 and D. dadantii 3937, contain desAB but not the zeamine biosynthesis genes. It is possible that the common ancestor of Dickeya contains both desAB and zeamine biosynthesis genes, but some Dickeya species and strains subsequently lost the ability to produce zeamines while maintaining DesABC as a defense mechanism. How the genes responsible for zeamine biosynthesis and resistance have originated during evolution remains to be further investigated. Interestingly, the desAB genes were not found in the vicinity of the zms gene cluster in S. plymuthica strains (Fig. 1), which is also known for production of zeamines (6, 7), suggesting that D. zeae and S. plymuthica have different evolutionary origins of the genes encoding zeamine biosynthesis and resistance.
The RND family proteins associated with antibiotic resistance can commonly transport a broad spectrum of substrates, which is determined by the corresponding inner membrane proteins of the efflux pumps (14). Most inner membrane proteins in the RND family found in human bacterial pathogens, including E. coli (15), P. aeruginosa (16), Burkholderia (30), and Acinetobacter (31), are related to transportation of multiple structurally dissimilar antibiotics. In contrast, we found that the DesABC efflux system of D. zeae EC1 was zeamine specific (Table 1; see also Table S5 in the supplemental material). Interestingly, two RND efflux pump systems found in Erwinia amylovora, a plant bacterial pathogen causing fire blight disease, also displayed narrow substrate specificity. Overexpression of these two RND efflux systems, MdtABC and MdtUVW, only resulted in the increment of MIC of several phytochemicals but had no effect on various conventional antibiotics (21), including some antibiotics used in our study (Table S5). Given that both D. zeae and E. amylovora are plant pathogens with little exposure to conventional antibiotics, these findings agree with the previous findings about the linkage between intrinsic resistance and the habitat of environmental bacteria (32). The narrow substrate specificity of MdtABC and MdtUVW in E. amylovora and DesABC in D. zeae EC1 may reflect the ancient role of the corresponding RND efflux pumps for bacteria to survive in hazardous environments generated by bacteria themselves or by other organisms.
Intriguingly, our results showed that inactivation of the desABC genes could cause different levels of decrement in the MIC of zeamines. Mutation of desA or desB led to only about an 8-fold decrease in MIC compared with that of the control strain (Table 1), while inactivation of desC could result in about a 32-fold decrease. A plausible explanation is that similar to its homolog tolC in E. coli (15), other proteins in strain EC1 are able to replace DesAB and form a functional transportation system with DesC to efflux zeamines. In addition, heterologous expression of desABC in E. coli resulted in only about a 2-fold increment in the MIC of zeamines, which was not comparable to the MIC changes when desABC were deleted in strain EC1. We first checked the possibility of whether the differences in GC content and codon usage of these two bacterial species affect the expression efficiency of desABC genes in E. coli. Our previous study showed that the GC content of the D. zeae EC1 genome is 53.43% (12), which is not identical but not substantially different from the GC content (50.8%) of E. coli strain K-12 (https://www.ncbi.nlm.nih.gov/genome/browse#!/prokaryotes/Escherichia%20coli%20 K-12). Strain K-12 is the parental strain of E. coli strain DH5␣ used in this study. At the codon usage level, both bacteria have more or less similar codon usage patterns, except that the rarely used codon CUA in E. coli is a frequently used codon in the coding sequence of desB in D. zeae EC1 (Table S6). We then examined the potential toxic effect of overexpressed DesABC in E. coli, as other overexpressed membrane proteins commonly exhibit detrimental effects on bacterial growth (33). We found that E. coli growth was markedly retarded when DesABC were overexpressed (Fig. S4). Taken together, it is most likely that overexpressed DesABC membrane proteins affect the bacterial normal physiological functions and, hence, compromise the ability to withstand zeamines.
Our data indicate that zeamine biosynthesis and DesABC-mediated resistance mechanisms are well coordinated by a sophisticated mechanism(s). The level of desAB gene expression was increased along with accumulation of zeamines ( Fig. 3B and C and 4B), and deletion of zmsA, the gene essential for zeamine biosynthesis (4), caused an arrest of the transcriptional expression of desAB ( Fig. 4B and C). Significantly, exogenous addition of zeamines to the zmsA mutant could boost desAB expression by more than 30-fold compared with that for water or solvent control (Fig. 4D). The findings thus demonstrated that in addition to their roles as phytotoxins and antibiotics (4,5,(8)(9)(10), zeamines can also act as signals in modulation of gene expression (Fig. 4D). The signal role of zeamines in induction of desAB expression was further confirmed in D. dadantii strain 3937 (Fig. 4D), which does not contain a zms gene cluster. Considering the wild distribution of zms gene clusters and desAB homologs in Dickeya species and strains ( Fig. 1 and Table S3) (8), we hypothesize that Dickeya species and strains have evolved a dedicated pathway to sense extracellular zeamines in self-protection against the detrimental effect of these antibiotics. The key regulators in this pathway might at least include a sensor or receptor protein that detects and responds to zeamines and a transcriptional regulator that modulates the expression of desAB. A gene that encodes a proposed two-component system sensor (NCBI accession number WP_029456608.1) was found near the desAB locus. However, inactivation of this gene did not affect the MIC of zeamines, which seems to preclude its potential link with the regulation of desAB (data not shown). In addition, given that the cellular levels of zeamines are important in induction of the zeamine resistance genes (Fig. 4D), several regulators known to be associated with the regulation of zeamine production and virulence, such as the acylhomoserine lactone (AHL) synthase ExpI (1), transcriptional regulator SlyA (3), and global regulator Fis (46), might also influence the transcriptional expression of desAB through modulating the production of zeamines or even more direct regulatory mechanisms, which demands further investigation.
In summary, this study documented a first resistance mechanism against zeamines, which are a new family of potent antibiotics with a broad spectrum of antimicrobial activities. This resistance mechanism is mediated by a novel and substrate-specific RND efflux pump, DesABC. Interestingly, this study also unveiled a signaling role of zeamines in modulation of desAB expression at the transcriptional level, which further expands our understanding about zeamines. In addition, the findings from this study suggest that D. zeae EC1 contains other mechanisms implicated in zeamine resistance besides the DesABC efflux system. This is evident as the MIC of zeamines for D. zeae EC1 was more than 500-fold higher than that for the zeamine-sensitive E. coli DH5␣, whereas inactivation of desC in D. zeae EC1 led to only about a 32-fold decrease in the MIC of zeamines (Table 1). A thorough understanding of the zeamine resistance mechanisms and the cognate regulatory networks might pave the way for practical application of these potent antibiotics and also could provide new insight on the control and prevention of this important bacterial pathogen.
Construction of deletion and complementation strains. Oligonucleotide primers used in this study are listed in Table S1 in the supplemental material. DNA manipulation was conducted by following methods described previously (5). Briefly, for gene in-frame deletion, fusion fragments containing the downstream and upstream regions of target genes were cloned into pKNG101 and transformed into E. coli CC118 for construction of gene in-frame deletion constructs. Triparental mating was performed by using wild-type strain EC1 or a zmsA in-frame deletion mutant as a recipient strain. Mutants were screened on an MM agar plate supplemented with 5% (wt/vol) sucrose, and desired deletions were confirmed by PCR and DNA sequencing. For complementation, the open reading frames (ORFs) of target genes were cloned into pBBR1-MCS4 and genes were expressed under the control of the lac promoter. The desired expression constructs were confirmed by PCR and DNA sequencing and introduced into corresponding mutants by triparental mating. The complementation strains were screened on MM agar plates containing ampicillin and verified by PCR. For construction of the strain expressing desABC heterologously, desC was cloned and expressed under the control of the tetO promoter in pAmob, while desAB were cloned into pBBR1-MCS4 and expressed under the control of the lac promoter. The resultant constructs, pBB-desAB and pAmob-desC, were cotransformed into E. coli DH5␣ for heterologous expression of the desABC efflux pump genes.
Preparation of zeamines. Overnight starter culture of wild-type strain EC1 grown in LB medium was inoculated (0.1%, vol/vol) into LS5 medium and grown at 28°C with rotation at 100 rpm for 48 h. The cells were then removed by centrifugation at 10,000 rpm at 4°C for 10 min. Approximately 10-liter supernatants were then passed slowly through the column containing 500 g of absorbent resin XAD7 (Sigma) at a flow rate of 1 ml/min by following the method described previously (4). The column was consecutively eluted with 2 liters of double-distilled H 2 O and 1 liter of methanol prior to elution with 2 liters of acetone to obtain the elutes containing zeamines. The acetone in the elutes was evaporated, and the residues were dissolved in methanol to obtain crude zeamine antibiotics. For confirmation, liquid chromatography-mass spectrometry (LC-MS) was performed using an Agilent 1260 infinity system connected to a Phenomenex Luna column (C 18 , 250 by 4.6 mm, 5 m) coupled with a Bruker maxis Q-TOF mass spectrometer to identify three main zeamine antibiotics, zeamine, zeamine I, and zeamine II (Fig. S1). The crude zeamines were eluted with a gradient program of 5% to 95% (CH 3 CN supplemented with 1% formic acid in H 2 O) in 20 min at a flow rate of 1 ml/min. The mass spectrometer was employed in the positive ion mode, scanning from 100 to 2,000 m/z. The source conditions were set as the following: ESI source type, end plate offset at Ϫ500 V, capillary at 4,500 V, nebulizer gas (N 2 ) at 0.8 bar, dry gas at 5.0 liters/min, and dry temperature at 180°C. The ion transfer condition was set as the following: collision cell RF of 800.0 Vpp. The antibiotic activity of zeamines was determined according to the method described below.
Determination of MICs. Determination of MICs of antibiotics in D. zeae and E. coli strains was conducted by following the protocol from the Clinical and Laboratory Standards Institute (35). Briefly, 2-fold dilutions of antibiotics in LB were added to 96-well plates, and fresh bacterial culture in LB medium was added to obtain about 2.0 ϫ 10 5 CFU/ml in each well. The plates were incubated at 28°C or 37°C for 18 h, and the minimum antibiotic concentration with no visible cell growth was defined as the MIC.
Construction of phylogenic tree. A total of 59 amino acid sequences obtained from NCBI (Table S3) were used in construction of the phylogenic tree, including 51 sequences found by blastp search, with the highest total score from Dickeya species and other proteobacterial species, 3 sequences from S. plymuthica strains containing homologs of the zms gene cluster, and 5 sequences with known substrate profiles. The protein sequences were aligned using MAFFT v7.402 (36) in the "einsi" mode, and the multiple-sequence alignment (MSA) was filtered for columns with high proportions of missing data using  (37) with the "-gappyout" option. The filtered MSA was analyzed by IQ-TREE v1.6.5 (38) to first perform a model selection with the "-MF" option (39), and then we carried out a maximum-likelihood tree inference under the best-fit model ("LGϩR5") with 1,000 ultrafast bootstrap support (40). Survival assay. The survival assay was conducted by following the kill curve method described previously, with minor modifications (41). Briefly, fresh bacterial cultures in LB medium at exponential phase were collected and adjusted to an OD 600 of about 1.0 (Ϯ0.05). Cells from 1 ml culture were harvested (4,000 rpm, 4°C, 5 min) and washed twice with LS5 salts (LS5 medium without sucrose). Bacterial cells were then resuspended in LS5 salts and added to 96-well plates with LS5 salts containing zeamines. The final concentration of zeamines in the assay was at 2-fold the MIC of the zmsA-desB mutant. The plates were incubated at 28°C with agitation at 200 rpm, and the survivors were determined at specific time points by plating appropriate bacterial dilutions on LB plates.
Growth kinetics assay measured in the flasks with LS5 medium. Bacterial growth curves in LS5 medium were measured by following the procedures described for zeamine preparation, with minor modifications. Briefly, overnight cell cultures were adjusted to an OD 600 of about 0.5 (Ϯ0.05) before inoculation at a ratio of 0.1%, vol/vol, and flasks were kept at 28°C with shaking at 150 rpm. The optical density at 600 nm was measured at different time points, as indicated, by the NanoDrop 2000c (Thermo Fisher Scientific, MA, USA) with proper dilutions when necessary.
Zeamine production assay. The assay of zeamine production was conducted by following a method described previously, with minor modifications (4). Briefly, 25 ml LB agar was poured in 10-by 10-cm square plates and overlaid with 7.5 ml 1% (wt/vol) agarose containing about 1.5 ϫ 10 8 fresh E. coli DH5␣ cells. The wells, at 4-mm diameter, were punched in the plate, and 30 l of cell-free supernatants (filter sterilized with a 0.22-m pore filter) were added in each well. The plates were incubated at 37°C for 24 h, and the inhibition zone around the wells was measured. For semiquantification, one unit of zeamines was defined as the amount that could generate a 2-mm-diameter inhibitory zone around the well. The number of zeamine units per milliliter was calculated by multiplying the units of zeamines calculated from the bioassay by the fold change of sample volume used in the test (30 l) to the total volume (1 ml).
Construction of transcriptional fusion construct pDesAB gfp and pDesAB 3937gfp and flow cytometry analysis. As there is only a 20-bp interval region located between the ORFs of desA and desB, the desAB genes were considered to be located in the same operon. The promoter region of desAB genes was predicted by using the online tool provided by BPROM (42) (http://www.softberry.com/berry.phtml?topic ϭbprom&groupϭprograms&subgroupϭgfindb). The 204-bp DNA fragment upstream of the ORF of desAB was amplified using the primer pair P-desAB-F and P-desAB-R (Table S1) and ligated into the promoterless gfp-reporter plasmid pPROBE-NT (43) for generation of the construct pDesAB gfp. pDesAB gfp and pPROBE-NT were separately mobilized into wild-type EC1 and the zmsA mutant by triparental mating with the helper strain HB101(pRK2013) to construct EC1(pDesAB gfp ) and ΔzmsA(pDesAB gfp ). 3937(pDesAB 3937gfp ) was constructed by a similar method. Expression of desAB and desAB 3937 was analyzed by monitoring the average fluorescence of 50,000 cells when bacteria were grown in flasks with LS5 medium at different time points by a CytoFLEX flow cytometer (Beckman Coulter, Brea, CA, USA) by following the method previously described (44).
RNA extraction and real-time qPCR analysis. Bacterial cells were cultured and harvested at an OD 600 of about 1.5 (Ϯ0.05). RNA extraction was performed using the RiboPure RNA purification kit, bacteria (Thermo Fisher Scientific, MA, USA), by following the manufacturer's instructions. The purity of RNA was determined by gel electrophoresis, and the A 260 /A 280 and A 260 /A 230 ratios were determined using a NanoDrop 2000c (Thermo Fisher Scientific, MA, USA). In qPCR analysis, an aliquot of 300 ng RNA sample was used for genomic DNA elimination and cDNA synthesis by a FastKing RT kit (with gDNase) (Tiangen Biotech, Co., Ltd., Beijing, China) by following the manufacturer's protocol. Specific primers for the desC, desB, and 16S rRNA genes (Table S1) were designed by AlleleID 6.0 (PRIMER Biosoft). The housekeeping gene 16S rRNA was used as a reference. The PCR efficiency of each gene was determined using five DNA standards at different concentrations (10, 1, 0.1, 0.01, and 0.001 g/ml). The qPCR analysis was conducted on a Quantstudio 6 Flex system using PowerUp SYBR green master mix (Thermo Fisher Scientific) with the following cycle profile: 1 cycle at 50°C for 2 min and 95°C for 2 min, followed by 40 cycles at 95°C for 15 s, 57°C for 15 s, and 72°C for 30 s. The experiment was repeated three times, each time with triplicates. Data were analyzed using the 2 ϪΔΔCT method as previously described (45).
Statistical analysis. Experiments were individually performed at least twice with three replicates each time. Data shown are the means from three replicates, and error bars indicated the standard deviations or standard errors. Statistical comparison was performed by using Student's t test in GraphPad Prism 5.0 software (GraphPad, La Jolla, CA). A P value of less than 0.05 was considered significant.
Data availability. The genome sequence of D. zeae EC1 is accessible in NCBI under accession number NZ_CP006929.1. The amino acid sequence of TolC in E. coli K-12 AG100 is accessible in NCBI under accession number WP_000735278.1.