The distribution of onion virulence gene clusters among Pantoea spp

Pantoea ananatis is a gram-negative bacterium and the primary causal agent of center rot of onions in Georgia. Previous genomic studies identified two virulence gene clusters, HiVir and alt, associated with center rot. The HiVir gene cluster is required to induce necrosis on onion tissues via synthesis of a predicted small molecule toxin. The alt gene cluster aids in tolerance to thiosulfinates generated during onion tissue damage. Whole genome sequencing of other Pantoea species suggest that these gene clusters are present outside of P. ananatis. To assess the distribution of these gene clusters, two PCR primer sets were designed to detect the presence of HiVir and alt. Two hundred fifty-two strains of Pantoea spp. were phenotyped using the red onion scale necrosis (RSN) assay and were assayed using PCR for the presence of these virulence genes. A diverse panel of strains from three distinct culture collections comprised of 24 Pantoea species, 41 isolation sources, and 23 countries, collected from 1946-2019, were tested. There is a significant association between the alt PCR assay and Pantoea strains recovered from symptomatic onion (P<0.001). There is also a significant association of a positive HiVir PCR and RSN assay among P. ananatis strains but not among Pantoea spp., congeners. This may indicate a divergent HiVir cluster or different pathogenicity and virulence mechanisms. Last, we describe natural alt positive [RSN+/HiVir+/alt+] P. ananatis strains, which cause extensive bulb necrosis in a neck-to-bulb infection assay compared to alt negative [RSN+/HiVir+/alt-] P. ananatis strains. A combination of assays that include PCR of virulence genes [HiVir and alt] and an RSN assay can potentially aid in identification of onion-bulb-rotting pathogenic P. ananatis strains.

Pantoea ananatis is a gram-negative bacterium and the primary causal agent of center rot of onions 18 in Georgia. Previous genomic studies identified two virulence gene clusters, HiVir and alt, associated 19 with center rot. The HiVir gene cluster is required to induce necrosis on onion tissues via synthesis of 20 a predicted small molecule toxin. The alt gene cluster aids in tolerance to thiosulfinates generated 21 during onion tissue damage. Whole genome sequencing of other Pantoea species suggest that these 22 gene clusters are present outside of P. ananatis. To assess the distribution of these gene clusters, two 23 PCR primer sets were designed to detect the presence of HiVir and alt. Two hundred fifty-two strains 24 of Pantoea spp. were phenotyped using the red onion scale necrosis (RSN) assay and were assayed 25 using PCR for the presence of these virulence genes. A diverse panel of strains from three distinct 26 culture collections comprised of 24 Pantoea species, 41 isolation sources, and 23 countries, collected 27 from 1946-2019, were tested. There is a significant association between the alt PCR assay and 28 Pantoea strains recovered from symptomatic onion (P<0.001). There is also a significant association 29 of a positive HiVir PCR and RSN assay among P. ananatis strains but not among Pantoea spp., 30 congeners. This may indicate a divergent HiVir cluster or different pathogenicity and virulence 31 mechanisms. Last, we describe natural alt positive [RSN + /HiVir + /alt + ] P. ananatis strains, which 32 cause extensive bulb necrosis in a neck-to-bulb infection assay compared to alt negative 33 [RSN + /HiVir + /alt -] P. ananatis strains. A combination of assays that include PCR of virulence genes 34 [HiVir and alt] and an RSN assay can potentially aid in identification of onion-bulb-rotting 35 pathogenic P. ananatis strains. 36

Introduction 38
Onion center rot is an economically impactful disease that routinely results in significant losses to the 39 yield and marketability of onion (Allium cepa L.). Symptoms of the disease include necrotized leaves 40 and rotted bulbs (Gitaitis and Gay, 1997 and functional genetic studies have determined that the HiVir and alt gene clusters function together 47 to drive necrotrophic infection of onion by virulent P. ananatis strains (Stice et al., 2020). 48 The HiVir gene cluster is a chromosomal gene cluster in P. ananatis that is hypothesized to code for 49 the synthesis of a yet undescribed phosphonate secondary metabolite that acts as a plant toxin 50 (Asselin et al., 2018;Takikawa and Kubota, 2018 Each institution selected strains from their respective collections to include diverse Pantoea spp. and 108 isolation sources, with specific enrichment of P. ananatis strains and Pantoea spp. strains isolated 109 from diseased onion. Biases in strain selection were unavoidable; however, we employed relevant 110 statistical methods to account for biases in our data sets. 111

alt PCR assay design 169
The alt1p_F/R primer pair was designed to detect P. ananatis and Enterobacter ludwigii strains  negative results and dividing by the total number of strains tested in this manner. A true positive is a 217 strain that had a positive PCR amplicon and a positive in-silico primer bind. A true negative is a 218 strain that had a negative PCR amplicon and a negative in-silico primer bind. 219

High-throughput red scale necrosis assay (RSN) 220
The red scale necrosis assay (RSN) was conducted as a high-throughput phenotypic assay to assess 221 the onion pathogenic potential of strains in each culture collection. The assay was conducted as 222 previously described with minor modifications (Stice et al., 2018). At UGA-CPES and UP-BCC, 223 consumer produce red onions (Allium cepa. L.) were purchased, whereas at UR-MAI stored red 224 onions cv. "Naqué" from postharvest experimental plots were taken, cut to approximately 3 cm wide 225 scales, sterilized in a 3% household bleach solution for 1 m, promptly removed and rinsed in dH 2 O. 226 Scales with a healthy unmarred appearance were used. Simple humidity chambers were used to 227 encourage disease and prevent the drying of onion scales (approximately 80% relative humidity). 228 Chambers consisted of a potting tray (27 × 52 cm) or similar plastic storage container, two layers of 229 paper towels pre-wetted with distilled water, and the plastic removable portion of pipette trays or a 230 similar item to prevent direct contact between the paper towels and the scales (Supplementary Figure  231 1). Disinfested onion scales were spaced evenly with approximately 1 cm of buffer between each 232 scale in the humidity chamber (Supplementary Figure 1). Individual onion scales were wounded 233 cleanly through the scale with a sterile pipette tip or needle and inoculated with a 10 μ L drop of 234 bacterial overnight LB or NA culture (Supplementary Figure 1). Sterile LB or NA culture was used 235 as a negative control and strain PNA 97-1R was used as a positive control. The tray was covered with 236 a plastic humidity dome or a loosely sealed container lid and incubated at room temperature for 72 h. 237 Following incubation, strains that induced a distinct necrotic lesion with a defined border and 238 clearing of the anthocyanin pigment on the onion scales were recorded as RSN + , while strains 239 exhibiting no clearing were recorded as RSN -( Figure 1A, picture inset). 240

Mini-Tn7Lux labeling of PNA 97-1R and LMG 2665 T 241
To determine the colonization of the putative non-bulb rotting type strain P. Sterile toothpicks were soaked in the inoculum suspension for 20 m. 271 inoculum-soaked toothpick horizontally through the onion neck just below the leaf fan (a favored 280 thrips feeding site). Toothpicks were left in the plants. Colored tags were used to mark each 281 treatment. Following inoculation, onions were randomized into blocks representing each replication. 282

Plant growth conditions and inoculation
At 20 d post-inoculation, the onions were harvested for imaging. For harvesting, onion bulbs were 283 removed from soil (which was sterilized and discarded following the experiment), rinsed with water, 284 and cut twice transversely at the center of the bulb to produce a 1.5 cm section of the center of the 285 onion (Supplementary Figure 2). The remaining portion of the top of the bulb had foliage removed 286 and was cut longitudinally (Supplementary Figure 2). 287

Imaging 288
Onions inoculated with non-labeled strains were imaged with a color camera and center rot symptom 289 incidence was recorded. Onion inoculated with Tn7Lux auto-bioluminescent reporter strains (LMG 290 2665 T and PNA 97-1R) were imaged with a color camera followed by bright-field and long exposure 291 imaging with the ccd imager (Analytik Jena UVP ChemStudio, Upland, CA). Within the 292 VisionWorks software manual, long-exposure imaging was selected with the following settings: 293 capture time 2 m, 70% focus, and 100% brightness (aperture), stack image, and saved to TIFF 294 This is a provisional file, not the final typeset article format. Brightfield images were captured with the following settings: capture time 40 mS, 70% 295 focus, 60% brightness (aperture), and saved to TIFF format. After imaging, bright-field and long-296 exposure images were merged using ImageJ (Fiji release). The incidence of center rot symptoms and 297 presence of a bioluminescent signal in long exposure images were recorded. 298

Statistical analysis 299
To determine whether a significant association exists between the HiVir PCR assay (positive vs. 300 negative) and the RSN phenotype (positive vs. negative) we conducted a two-tailed Fisher exact test 301 on 2×2 contingency tables in Microsoft Excel. The accuracy of a positive HiVir PCR assay as a 302 predictor of the RSN phenotype was calculated by summing the true positive and true negative 303 results and dividing by the total number of strains tested. 304 We tested whether a significant association exists between the alt PCR assay (positive vs. negative) 305 and the source of isolation (onion vs. non-onion) using the two-tailed Fisher exact test previously 306 described. 307 To determine whether source of isolation (onion or non-onion) or Pantoea group (P. ananatis or 308 Pantoea spp.) was associated with an alt + genotype and an RSN + phenotype, we used a Z-proportions 309 test to compare the proportion of four groups (RSN + alt -, RSN + alt + , RSNalt + , and RSNalt -). The 310 two-tailed (α < 0.05) Z-proportions test assumes a normal distribution and considers the number of 311 samples of each proportion being tested to determine whether the two proportions statistically differ 312 according to a Z-statistic (Ramsey and Schafer, 2002). 313 To determine whether the percent center rot incidence at the onion midline differed among the three 314 bulb rotting and non-bulb rotting strains, we conducted an ANOVA and Tukey-HSD test using R-315 Studio v1.2.1335 (package agricolae). 316

3
Results 317 An objective of this study was to assess the utility of two PCR assays and the phenotypic RSN assay 318 for identifying Pantoea onion virulence genes. To accomplish this objective, we worked among three 319 separate institutions to test the three assays. The assays were originally developed for use with P. 320 ananatis strains; however, to investigate the breadth of their utility and potential limitations, we 321 included a wider selection of Pantoea spp. 322

HiVir PCR assay 323
The HiVir cluster encodes predicted biosynthetic enzymes for the production of a small molecule 324 phosphonate phytotoxin. The HiVir2p_F/R primer pair amplified an 857 bp portion of DNA 325 including pepM, the intergenic region, and pavC while overlapping with an SNP associated with the 326 RSNphenotype ( Figure 1A, C)

alt PCR assay 338
The alt cluster encodes a cohort of putative disulfide exchange redox enzymes that confers tolerance 339 to antimicrobial thiosulfinates produced in disrupted onion tissues. The alt1p_F/R primer pair 340 amplified a 414 bp conserved portion of DNA encompassing the end of the altB, the intergenic 341 region, and the beginning of altC ( Figure 1B, C). Amplification of the alt1p_F/R amplicon is 342 depicted in Supplementary Table 1 (1=alt + , 0=alt -). 343 The accuracy of a positive alt PCR assay corresponding to an in-silico primer bind was calculated 344 among the 89 sequenced strains to be 95.51% with a significant association (P<0.001). There was a 345 statistically significant association of the alt PCR assay (positive vs. negative) with isolation source 346 (onion vs. non-onion) among P. ananatis (n=115; P<0.001) and Pantoea spp. (n=137; P<0.001) 347 (Supplementary Table 1). 348

Discussion 386
Pantoea ananatis is a broad host range pathogen with strains inducing disease symptoms ranging 387 from fruitlet rots to leaf blights in a variety of hosts, including pineapple, eucalyptus, rice, maize, 388 melon, and onion (Coutinho and Venter, 2009 sequencing (Stice et al., 2020). By leveraging these simple assays against the UGA-CPES (focused 402 on diseased onion strains, primarily of P. ananatis, or isolated from the onion-associated weeds and 403 thrips) and UP-BCC (focused on many Pantoea spp. including pathogenic rice, maize, and 404 eucalyptus strains, among others), we observed a significant enrichment of the alt gene cluster among 405 strains recovered from diseased onions, and a strong statistical association between the HiVir PCR 406 and RSN phenotypic assays (Figure 2, Supplementary Table 1). The distribution of these onion 407 virulence genetic clusters will be useful for future studies of Pantoea strains involved in the onion 408 center rot complex. In addition, defining genetic features of bulb-rotting P. ananatis strains, based on 409 the presence or absence of the alt cluster, will help distinguish strains of P. ananatis that are a greater 410 threat to onion production ( Figure 3). 411

Utility of HiVir PCR assay 412
We designed the HiVir PCR assay to detect P. ananatis strains that induce red onion scale necrosis. 413 To do this, we compared three strains in our collection that have the HiVir cluster in their sequenced 414 draft genomes but lack the RSN phenotype. The forward primer, HiVir2p_F, overlaps with a 415 conserved SNP among the RSN -/HiVir + strains. Our results indicate the HiVir2p_F/R primers work 416 as expected with no amplification in the RSN + /HiVirstrains such as PNA 98-11 ( Figure 1A,C). The 417 HiVir PCR assay is sensitive and accurate in predicting the RSN ability of P. ananatis strains with a 418 significant association between the PCR result and the RSN phenotype.  Figure 2). 448

UGA-CPES 450
The UGA-CPES collection included mostly P. ananatis strains isolated from symptomatic onions 451 along with strains isolated from onion-associated weeds and thrips. P. ananatis strains originating 452 from symptomatic onion typically contained the RSN + /HiVir + /alt + result; however, strains with only 453 HiVir, only alt, or neither gene cluster were also recovered. We would expect center-rot-causing 454 strains to harbor both traits, as these virulence determinants have been demonstrated experimentally 455 to strongly contribute to neck-to-bulb infection. Among P. ananatis strains isolated from sources 456 other than onion, the RSN + /altresult was more pronounced (Table 1). 457

UP-BCC 458
The UP-BCC collection included Pantoea strains from diverse sources with specific enrichment in 459 maize, eucalyptus, and onion. Most strains tested RSNalt - (Table 1). P. ananatis species had This is a provisional file, not the final typeset article enrichment of the RSN + and alt + results when isolated from onion. There was a larger proportion of 461 P. ananatis strains in the UP-BCC with the RSN + altphenotype, which supports our interpretation 462 that RSN + /HiVir + may be broadly associated with P. ananatis plant pathogenicity but that the alt 463 cluster is common among strains associated with onion. 464

UR-MAI 465
The UR-MAI collection was entirely comprised of strains originating from diseased onion. The 466 RSN + /HiVir + /alt + result was obtained from the two strains belonging to the species P. ananatis 467 within this collection. Interestingly, the RSN + /alt + result was also found in one P. eucalypti strain 468 (MAI 6036) and one P. agglomerans strain (MAI6045) recovered from symptomatic onion leaves 469 and seed stalks. The HiVir -/RSN -/alt + result was nearly universally distributed among recovered 470 strains from this survey (Supplementary Table 1, Figure 2, Table 1). This suggests that alternative 471 onion necrosis mechanisms exist among Pantoea spp. 472

Proportional analysis of RSN and alt results among strains 473
Considering the selection of strains was biased in regard to isolation source (mostly of onion origin), 474 we grouped the RSN/alt results of our strains based on two metrics for a proportional analysis: 1.) 475 onion vs. non-onion strains, 2.) P. ananatis vs. Pantoea spp. strains. We then conducted a Z-476 proportions test. This test allowed us to reveal broad differences by comparing proportions of strains 477 in defined groups. A significantly greater portion of strains of onion origin have the RSN + /alt + and 478 RSN -/alt + results compared to those of non-onion origin for both P. ananatis and Pantoea spp. strains 479 (Figure 2, Supplementary Table 2). This offers support for our conceptual model of alt cluster 480 enrichment among onion pathogenic strains. Conversely, a significantly greater proportion of P. 481 ananatis strains of non-onion origin has the RSN + /altresult compared to those of onion origin 482 (Figure 2, Supplementary Table 2). This suggests that strains causing foliar symptoms on crops such 483 as rice and maize may utilize the HiVir encoded toxin but do not require the alt cluster, as their hosts 484 do not produce the thiosulfinate defensive chemistry found in onions. 485

The association of HiVir and alt gene clusters with P. ananatis bulb rot symptoms 486
We previously demonstrated that both the HiVir and alt gene clusters were required by P.

Conclusion 499
The three culture collections tested in this study have allowed us to gather a preliminary picture of 500 the distribution of the [HiVir + /alt + ] genotypes and RSN + phenotype associated with onion center rot. 501 The HiVir and alt PCR assays which were respectively designed to be specific and inclusive were 502 validated on a panel of sequenced genomes with 95.15% accuracy (Figure 1). The RSN assay was 503