Identification of gene products involved in plant colonization by Pantoea sp. YR343 using a diguanylate cyclase expressed in the presence of plants

Microbial colonization of plant roots is a highly complex process that requires the coordination and regulation of many gene networks, yet the functions of many of these gene products remain poorly understood. Pantoea sp. YR343, a gamma-proteobacterium isolated from the rhizosphere of Populus deltoides, forms robust biofilms along the root surfaces of Populus and possesses plant growth-promoting characteristics. The mechanisms governing biofilm formation along plant roots by bacteria, including Pantoea sp. YR343, are not fully understood and many genes involved in this process have yet to be discovered. In this work, we identified three diguanylate cyclases in the plant-associated microbe Pantoea sp. YR343 that are expressed in the presence of plant roots, One of these diguanylate cyclases, DGC2884 localizes to discrete sites in the cells and its overexpression results in reduced motility and increased EPS production and biofilm formation. We then performed a genetic screen by expressing this diguanylate cyclase from an inducible promoter in order to identify candidate downstream effectors of c-di-GMP signaling which may be involved in root colonization by Pantoea sp. YR343. Further, we demonstrate the importance of other domains in DGC2884 to its activity, which in combination with the genes identified by transposon mutagenesis, may yield insights into activity and regulation of homologous enzymes in medically and agriculturally relevant microbes.

quorum-sensing as a mechanism in the rhizosphere for influencing changes in gene expression that 56 can lead to root colonization and biofilm formation (6-9). Indeed, genome analyses showed that 57 acyl-homoserine lactone (AHL)-based signaling systems are prevalent in the microbiome of 58 Populus deltoides (10). Additionally, plant colonization involves the second messenger signaling 59 molecule, cyclic diguanylate monophosphate (c-di-GMP), which is known to affect motility, 60 virulence, exopolysaccharide (EPS) production, and biofilm formation in many bacterial species 61 (11-15). 62 The levels of c-di-GMP within cells are regulated by two different enzymes: diguanylate 63 cyclases, which catalyze the production of c-di-GMP from two molecules of guanosine 64 triphosphate (GTP), and phosphodiesterases, which degrade c-di-GMP to guanosine 6 cells were grown in M9 minimal media with 0.4% glucose, we found that twelve diguanylate 118 cyclase reporters showed an average fluorescence intensity below 2.00 (weak or no expression), 119 making them suitable candidates for further study in terms of expression in biofilms, pellicles, and 120 during root colonization (Table 1). To test for expression during biofilm formation, the cells were 121 grown statically in M9 minimal medium with 0.4% glucose for 72 hours in 12-well dishes 122 containing a vinyl coverslip as described in Materials and Methods. These data show that eleven 123 diguanylate cyclases showed increased expression under these conditions, with DGC2884 and 124 DGC2242 showing the highest levels (Table 1 and Fig 1). Interestingly, we found that each of the 125 strains showed an increase in expression during biofilm formation based on GFP fluorescence, but 126 images showed that GFP levels driven from the DGC2884 promoter were not uniform within the 127 biofilm (Fig 1). Instead, we found that GFP was highly expressed in specific patches throughout 128 the biofilm, but expressed at low or undetectable levels in other regions. This expression pattern 129 was also observed in some of the other promoter constructs and is reflected, in part, by the higher  Table 1. We also tested for expression during pellicle formation and found 131 that most strains only exhibited a modest increase in expression (Table 1). 132 Next, we tested the activity of these 12 promoters during root colonization of T. aestivum and P. 133 trichocarpa. Bacteria associated with roots were examined for the presence or absence of 134 fluorescence, since quantification of expression levels was difficult due to plant autofluorescence 135 (Table 1). After one week of growth post-inoculation, we found that DGC2884, DGC3006, and 136 DGC3134 were expressed on T. aestivum and P. trichocarpa roots (Fig 1 and Table 1). We cannot 137 exclude the possibility that the eight untested diguanylate cyclases may also be expressed during   carrying an empty vector (Fig 2). Growth curves were compared in both minimal and rich media 194 ( Fig S2). Notably, expression of wild type DGC2884, but not any of the variants, resulted in  (YR343 (pSRK (Km)-DGC2884)) resulted in red, wrinkly colony formation (Fig 2A). In contrast, suggesting that expression of DGC2884 in the absence of enzymatic activity may still retain some 204 function (Fig 2A). We observed that Congo Red binding by strains expressing the DGC2884ΔTM 205 variant was less than that of the DGC2884 expressing strain and we no longer observed wrinkly 206 colony morphology, supporting the hypothesis that the TM domain of DGC2884 is critical to its 207 function (Fig 2A). Since increased levels of c-di-GMP are typically associated with decreased motility (11, 237 43, 44), we next tested whether overexpression of these diguanylate cyclases affected motility 238 using a swim plate agar assay. As expected, overexpression of DGC2884 resulted in impaired 239 motility compared to the control strain, which was partially restored in the DGC2884 AADEF 240 variant ( Fig 2B). We found that, in comparison to strains overexpressing DGC2884, expression 241 of DGC2884ΔTM resulted in partial restoration of motility behavior reminiscent of that observed 242 for strains expressing the DGC2884 AADEF mutant ( Fig 2B). Together, these data suggest that a 243 fully functional DGC2884 is required to modulate motility.

244
Next, we examined whether overexpression of these diguanylate cyclases influenced 245 biofilm formation (Fig 2C). While each of these strains showed formation of biofilms on vinyl 246 coverslips, the most robust biofilms were formed during expression of the wild type DGC2884, domain. 250 We also tested the effect of overexpression of each diguanylate cyclase on pellicle 251 formation and calculated the percentage of cells in pellicles and found that overexpression of 252 DGC2884 resulted in significantly increased pellicle formation when compared to the empty 253 vector control (p < 0.005, t-test) ( Fig 2D). While expression of DGC2884 AADEF and 254 DGC2884ΔTM also resulted in more pellicle formation than the control (significantly more by 255 DGC2884ΔTM, p < 0.05, t-test), they produced significantly less pellicle than that of wild type 256 cells expressing the full-length DGC2884 (p < 0.05, t-test) ( Fig 2D). that the AADEF mutation indeed affected enzyme activity (Fig 2E, 2F). We also found that 266 expressing DGC2884ΔTM resulted in little to no activity (Fig 2E, 2F). To verify that the genes 267 encoding these diguanylate cyclases were expressed in these cells, we examined transcript levels 268 using RT-PCR ( Fig S3). Taken together, results from each of these assays confirm that both  To gain further insight into the function of DGC2884, we performed a simple Protter 272 analysis using the amino acid sequence of DGC2884 (46) and found that the sequence for 273 DGC2884 is predicted to have two transmembrane domains at its N-terminus that make up a 274 CHASE8 domain, followed by the GGDEF domain ( Fig 3A).  We next examined localization of wild type DGC2884 and DGC2884ΔTM in a wild type 289 background by expressing it fused to either a 3HA or 13Myc tag ( Fig 3B). These data show that 290 DGC2884 was found to primarily localize in discrete foci at the cell pole or towards the mid-cell.

291
In the absence of the N-terminal transmembrane domain, however, DGC2884 no longer localized 292 as discrete foci, but rather the localization pattern became more diffuse with fewer visible foci ( Fig   293   3B and Table 2). To verify that the tag did not alter the expression or function of these enzymes, we performed a motility assay ( Fig 3C) and western blot ( Fig S5)   Identification of c-di-GMP responsive genes using transposon mutagenesis 305 Overexpression of DGC2884 resulted in a number of phenotypes (shown in Fig 2) Behavioral defects observed in selected mutants 331 Using the list of genes found in the genetic screen (Table 3) examining EPS production (by observing phenotypes on media with Congo Red) (Fig 4A, 4B).

343
Next, we used the cured transposon mutants to observe pellicle formation (Fig 4C), and measure 344 biofilm production with a crystal violet assay (Fig 4D). Compared to the wild type control, each 345 mutant had a different growth phenotype on media with Congo Red, some of which were more 346 noticeable on one media type over the other (Fig 4A, 4B). These phenotypes were further 347 influenced based on whether the mutant expressed DGC2884 (pSRK (gm)-DGC2884) or an empty 348 vector (pSRK (gm)). We next examined the effects of these mutations on pellicle formation and 349 found that the UDP::Tn5, FliR::Tn5, and GlpF::Tn5 mutants produced significantly less pellicle 350 than the wild type strain (Fig 4C). We also examined biofilms attached to vinyl coverslips and 351 found that while some mutants appear to produce more biofilm, such as FliR::Tn5 and GlpF::Tn5,

352
there were no statistically significant differences measured by quantifying Crystal Violet staining.

353
Interestingly, we did find that the UDP::Tn5 and Ndk::Tn5 mutants produced significantly more 354 biofilm than the wild type strain in this assay (Fig 4C). Ndk::Tn5 mutants showed a slight, though significant, increase in colonization (Fig 5A; 377 statistically significant differences with p < 0.005, t-test). Comparisons of growth rates between 378 transposon mutants and the wild type strain showed no significant differences for most strains, 379 except for growth with UDP::Tn5 ( Fig S4); however, based on growth curves, the maximum OD YfiN has been shown to modulate production of Psl polysaccharides, whose operon possesses 447 genes also found to regulate amylovoran biosynthesis in Erwinia amylovora (33,34,55  insights into the roles of multiple diguanylate cyclases in coordinating these behaviors.

535
Bacterial strains and growth conditions. We sequenced each plasmid from the transposon outwards using the following primers, tpnRL17-633 1 and tpnRL13-1 (79). All resulting sequences were analyzed using BlastX from NCBI in order 634 to identify the region of DNA flanking each transposon.

635
Individual transposon mutants were grown three to four times sequentially on rich media 636 without selection in order to remove the pSRK (Gm)-DGC2884 plasmid. Removal of the plasmid 637 was verified by growth on kanamycin at 50 µg mL -1 , but not on gentamycin at 10 µg mL -1 .

638
Construction of fluorescent strains. We generated fluorescent strains that were also resistant to  Microimaging, Thornwood, NY). Images were processed using Zen2012 software (Zeiss). Cell 663 fluorescence intensity measurements were performed using Fiji ImageJ for assays with promoter-664 reporter fusions for DGCs and for the Vc2 Spinach aptamer following the protocol described by 665 Kellenberger, et al (45). Briefly, images were initially collected using the same parameters and 666 then collectively processed so that brightness and contrast was adjusted and normalized across the