Novel tetraplex qPCR assays for simultaneous detection and identification of Xylella fastidiosa subspecies in plant tissues

Xylella fastidiosa is an insect-borne bacterium confined to the xylem vessels of plants. This plant pathogen has a broad host range estimated to 560 plant species. Five subspecies of the pathogen with different but overlapping host ranges have been described, but only three subspecies are widely accepted, namely subspecies fastidiosa, multiplex and pauca. Initially limited to the Americas, Xf has been detected in Europe since 2013. As management of X. fastidiosa outbreaks in Europe depends on the identification of the subspecies, accurate determination of the subspecies in infected plants as early as possible is of major interest. Thus, we developed various tetraplex and triplex qPCR assays for in planta X. fastidiosa detection and subspecies identification in a single reaction. We designed primers and probes using SkIf, a bioinformatics tool based on k-mers, to detect specific signatures of the species and subspecies from a dataset of 58 genome sequences representative of X. fastidiosa diversity. We tested the qPCR assays on 39 target and 30 non-target strains, as well as on 13 different plant species spiked with strains of the different subspecies of X. fastidiosa, and on samples from various environmental and inoculated host plants. Sensitivity of simplex assays was equal or slightly better than the reference protocol on purified DNA. Tetraplex qPCR assays had the same sensitivity than the reference protocol and allowed X. fastidiosa detection in all spiked matrices up to 103 cells.mL−1. Moreover, mix infections of two to three subspecies could be detected in the same sample with tetraplex assays. In environmental plant samples, the tetraplex qPCR assays allowed subspecies identification when the current method based on multilocus sequence typing failed. The qPCR assays described here are robust and modular tools that are efficient for differentiating X. fastidiosa subspecies directly in plant samples.

Xylella fastidiosa is an insect-borne bacterium confined to the xylem vessels of plants. This plant 20 pathogen has a broad host range estimated to 560 plant species. Five subspecies of the pathogen with 21 different but overlapping host ranges have been described, but only three subspecies are widely 22 accepted, namely subspecies fastidiosa, multiplex and pauca. Initially limited to the Americas, Xf has 23 been detected in Europe since 2013. As management of X. fastidiosa outbreaks in Europe depends on 24 the identification of the subspecies, accurate determination of the subspecies in infected plants as early 25 as possible is of major interest. Thus, we developed various tetraplex and triplex qPCR assays for in 26 planta X. fastidiosa detection and subspecies identification in a single reaction. We designed primers 27 and probes using SkIf, a bioinformatics tool based on k-mers, to detect specific signatures of the species 28 and subspecies from a dataset of 58 genome sequences representative of X. fastidiosa diversity. We 29 tested the qPCR assays on 39 target and 30 non-target strains, as well as on 13 different plant species 30 spiked with strains of the different subspecies of X. fastidiosa, and on samples from various 31 environmental and inoculated host plants. Sensitivity of simplex assays was equal or slightly better 32 than the reference protocol on purified DNA. Tetraplex qPCR assays had the same sensitivity than the 33 reference protocol and allowed X. fastidiosa detection in all spiked matrices up to 10 3 cells.mL -1 . 34 Moreover, mix infections of two to three subspecies could be detected in the same sample with tetraplex 35 assays. In environmental plant samples, the tetraplex qPCR assays allowed subspecies identification 36 when the current method based on multilocus sequence typing failed. The qPCR assays described here 37 are robust and modular tools that are efficient for differentiating X. fastidiosa subspecies directly in 38 plant samples. Sequence Typing (MLST) scheme (Yuan et al., 2010). 58 In Europe, Xf has been reported for the first time in Apulia area, Italy, in olive trees (Saponari et al.,59 2013). Then, Xf was detected in 2015 in France, more precisely in Corsica and in the French Riviera 60 region, mainly on Polygala myrtifolia and other ornamentals (Denancé et al., 2017). Two years later, 61 Xf has been reported in the Balearic Islands mostly in olive tree, grapevine and sweet cherry and in 62 continental Spain in almond trees (Landa, 2017 (Wells et al., 1981) or modified PWG media (agar 12 g.L -1 ; soytone 4 g.L -1 ; bacto tryptone 1 g.L -125 1 ; MgSO4.7H2O 0,4 g.L -1 ; K2HPO4 1.2 g.L -1 ; KH2PO4 1 g.L -1 ; hemin chloride (0.1% in NaOH 0.05 126 M) 10 ml.L -1 ; BSA (7.5%) 24 ml.L -1 ; L-glutamine 4 g.L -1 ) at 28°C for one to two weeks. Other strains 127 were grown at 25°C for one to two days on: MG media (Mougel et al., 2001) for Agrobacterium and 128 Rhizobium, TSA (tryptone soy broth 30 g.L -1 ; agar 15 g.L -1 ) for Clavibacter, Ensifer, 129 Stenotrophomonas, Xanthomonas and Xylophilus and King's B medium (KH2PO4 1.5 g.L -1 ; MgSO4, 130 7H2O 1.5 g.L -1 ; protease peptone 20 g.L -1 , glycerol 10 mL.L -1 ; agar 15 g.L -1 ) for Dickeya, Erwinia, 131 Pantoea and Pseudomonas. For qPCR assays, bacterial suspensions were prepared from fresh cultures 132 in sterile distilled water, adjusted at OD600 nm = 0.1. To evaluate assay specificity bacterial suspensions 133 were boiled for 20 min, followed by a thermal shock on ice and a centrifugation at 10,000 g during 10 134 min. 135

Spiking of samples and DNA extraction 165
Prior to DNA extraction, plant samples were inoculated by mixing 1 g of healthy plant material with 166 0.5 mL of a bacterial suspension, at a known concentration, and ground with 4.5 mL of sterile distilled 167 water. Each matrix was spiked in order to end up with concentrations ranging from 1x10 6 CFU.mL -1 168 to 10 CFU.mL -1 . Spiking with more than one strain was done in equal amounts to end up with final 169 concentrations ranging from 1x10 6 CFU.mL -1 to 1x10 CFU. pauca (XFP primers) ( Table 3). The parameters were set up with an optimal size of 20 bp (sizing 201 between 18-27 bp), an optimal product size of 85 to 150 bp; a Tm of 60°C (± 3°C) and 70°C (± 3°C) 202 for primers and probes, respectively. Then, the individual primer and probe combinations and the six 203 sets of four combinations were tested using Amplify to check the absence of dimer and cross-204 amplification (Engels, 1993 Four others primer and probe combinations previously published were used in this study. The first 210 targets the rimM gene of Xf (Harper et al., 2010) and was used as reference protocol. The second targets 211 the eukaryotic rRNA18S gene (Ioos et al., 2012) and was used as internal control. The remaining two 212 tests target fastidiosa or multiplex subspecies (Burbank and Ortega, 2018). 213

Optimization of qPCR assays and tetraplexing 214
The tetraplex qPCR assays designed in this study were optimized for: i) primer and probe hybridization 215 temperature that was checked individually by PCR using a gradient ranging from 57.5 to 61. results were analyzed, with expert verification, using Bio-Rad CFX Manager 3.1 software and its 230 regression mode. The reaction efficiency was calculated using serial dilutions with the formula: E = 231 10 (-1/slope) . 232

qPCR assay specificity, efficiency and limit of detection 233
The specificity of the newly designed primer and probe combinations was validated using the 234 optimized protocol on the boiled bacterial suspensions of the 69 strains listed in the Primers and probes were designed within specific long-mers (Table 3). Specific amplifications were 261 obtained in silico on XF genome sequences and WGS bacterial sequences from NCBI at the expected 262 amplification size, without any mismatch for the five primer and probe combinations (XFF, XFFSL, 263 XFM, XFMO and XFP). Only two mismatches were observed and concerned the XF primer and probe 264 combination. One mismatch was on the eighth nucleotide on the XF probe for the Xfm Dixon, Griffin1, 265 M12, Sycamore, CFBP 8416, CFBP 8417, CFBP 8418 strains and the second one was on the sixth 266 nucleotide of the forward XF primer of the Ann-1 Xfs strain. As there were not many possible 267 combinations of primers and probes for the XF set, this combination was nevertheless retained, and 268 subsequent in silico checks proved the specificity of all primer and probe combinations. 269

Specificity and sensitivity of simplex and tetraplex qPCR assays on strains 270
The specificity of each newly designed primer and probe combination was validated in simplex qPCR 271 assays on 39 Xf strains and on 30 plant associated-bacterial strains (Table 1). These strains were 272 selected as they potentially share the same niche than Xf or for being phylogenetically closely related. 273 No amplification was detected on non-target strains or healthy host plant species and the primer and 274 probe combinations allowed amplification of all strains or subspecies of Xf, for which they were 275 designed (XF: 39/39, XFF: 10/10, XFM: 16/16, XFMO: 1/1, XFP: 7/7, XFFSL: 16/16). 276 In simplex qPCR assays, the LODs of the new primer and probe combinations designed in this study 277 were as good as the LODs obtained with the Harper's qPCR assay or 10 times better for XFM primers 278 ( primers (XFMO) the LOD of tetraplex qPCR assays was usually 10 times higher than the LOD of the 291 simplex test on DNA (Table 4 and Supplemental data 2). In addition, it should be noted that the closer 292 the Ct value was to the detection limit, the higher the SEM was. In tetraplex qPCR assays set n°1, XF, 293 XFM and XFP primers allowed a detection up to 100 pg.mL -1 . The XFFSL primers allowed the 294 detection of Xff up to 10 pg.mL -1 and of Xfmo up to 100 pg.mL -1 . The set n°2 allowed detection up to 295 100 pg.mL -1 using XFF and XFM primers and up to 10 pg.mL -1 with XFP primers. The XF primers 296 allowed the detection of Xff and Xfm up to 100 pg.mL -1 and of Xfp up to 10 pg.mL -1 . The set n°3, 297 allowed a detection up to 100 pg.mL -1 with XF, XFF and XFM primers and up to 10 pg. sequencing was feasible to confirm the identification. 319

Identification of Xf subspecies in spiked samples with tetraplex qPCR assays 320
After validation of the efficiency and specificity of the primers and probe, the three sets of tetraplex 321 qPCR assays n°1, 2 and 3, were tested on spiked samples. As the three sets gave similar results, this 322 section is focused on the tetraplex set n°1: XF -XFFSL -XFM -XFP, which covers the full known 323 diversity of Xf (Table 5). The results of the other two tetraplex assays are provided in Supplemental 324 Data 5 and Supplemental data 6. This tetraplex qPCR assay (set n°1) was tested on 29 combinations of 325 plant petioles and midribs spiked with one to three strains of the different subspecies. (The full results  326 of the dilution ranges are available in Supplemental data 7). This tetraplex allowed the detection and 327 correct identification of all subspecies in all combinations without false positive result. Although the 328 detection limit was expected to be similar for all plants, since they were all enriched with the same 329 bacterial suspensions, different LODs were observed ranging from 1x10 3 to 1x10 5 CFU.mL -1 (5 to 330 5x10 3 CFU/reaction) depending on the matrix for plants spiked with only one strain. An independent 331 repetition of this test was performed two months after the first one. For O. europaea, P. myrtifolia, P. 332 cerasus, P. dulcis and Q. ilex the LOD was either identical between the two assays or 10 time higher. 333 The LOD of Xf in V. vinifera was 100 times higher in the second assay highlighting a potential 334 accumulation of qPCR inhibitors between the two experiments. Moreover, on 11 combinations out of 335 46, XF primers had a LOD 10 times higher in planta than the one obtained for the subspecies. Xf 336 subspecies could be identified until a Ct value of 35.08 using Harper's qPCR assay in a spiked sample 337 of P. dulcis. In other matrices the LOD of the tetraplex qPCR assay corresponded usually to a Ct value 338 ranging from 30 to 34 using Harper's qPCR. 339 Moreover, the tetraplex qPCR assay set n°1 allowed the detection and identification of mix infections 340 with two to three subspecies simultaneously. On N. oleander, O. europaea, P. myrtifolia and P. dulcis 341 the LOD for the two or three inoculated subspecies is similar of the one obtained for single inoculations 342 (Table 5). 343 To demonstrate that our multiplex qPCR assays are modular tools, which can be adapted to one's needs, 344 three other primer and probe sets were evaluated. In one set, we removed the primers and probe 345 targeting the species (set n°4: XFFSL-XFM-XFP). In a second one, we replaced it by the Harper's 346 primers and probe as this test is known to be highly sensitive (set n°5: Harper-XFFSL-XFM-XFP), and 347 we also tested the use of primers and probes targeting the 18S rRNA as an internal control (set n°6: 348 18S-XFFSL-XFM-XFP). Evaluation of these three sets on calibrated DNA suspensions of the Xff strain 349 CFBP 7970 indicated that the LOD for the XFFSL primers was the same than the one found previously 350 for the sets n°1, 4, 5 and 6 (10 pg.mL -1 ) (Supplemental data 8). In Q. robur and C. monspeliensis 351 samples spiked with the Xfm strain CFBP 8416, the LOD obtained for the primers detecting the 352 multiplex subspecies (XFM) was the same for the three sets (1x10 5 CFU.mL -1 ) (Supplemental data 9). 353 The use of Harper's primers and probe in set n°5 allowed the detection of Xf strain at the same LOD 354 than for XF primers and probe in spiked Q. robur samples, but the detection was slightly better (a gain 355 of one Log unit) in the spiked C. monspeliensis samples. A Ct value was obtained for all spiked samples 356 with the 18s rRNA primers, highlighting that these primers and probe were reliable internal 357 amplification controls. 358

Identification of Xf subspecies in environmental plant samples and inoculated 359
plants by tetraplex qPCR assays 360 Ten plant samples from Corsica, France (Table 6) and ten samples from inoculated plants (Table 7)  361 were tested using the tetraplex set n°1. Our tetraplex qPCR assay was able to detect the bacterium in 362 samples declared contaminated with Harper's qPCR assay up to Ct =34.97. However, this LOD was 363 variable depending on the matrices (Table 7). While the bacterium was detected at the subspecies level 364 with one or the other primer and probe combinations in eight environmental plant samples, the XF 365 primers and probe was less efficient and allowed the detection in only five samples (  Tested on a large collection of target and non-target strains, the primers and probes showed high 399 specificity for Xf and its subspecies and no cross-reactions. In vitro, the specificity was tested in two 400 steps. Inclusivity was evaluated on strains of Xf subspecies and exclusivity on a range of strains chosen 401 to be present in the same plant and insect niches as Xf (Rogers, 2016)  pg.mL -1 (i.e. 4x10 3 to 4x10 4 copies.mL -1 ), these multiplex qPCR assays still present a sensitivity that 431 is similar to the one of the reference protocol, on single bacterial suspensions (Harper et al., 2010). 432 In spiked and environmental plant samples, the benefit from the use of our tetraplex assays is obvious. 433 The tetraplex qPCR assays developed here are able to identify Xf subspecies up to 10 3 CFU. 2008). These variations could explain the 10 to 100 fold higher LOD obtain for the second repetition 460 that was performed with grapevine and olive tree sampled two months after the first sample set. 461 While we added a sonication step to improve DNA extraction, we did not test here other ways to 462 improve per se the DNA extraction step and improve the LOD of our assays. Various options are 463 available. A phenol-chloroform step could be added to the DNA extraction method to reduce the level 464 of extracted proteins (Schrader et al., 2012). Reagents such as Tween 20, DMSO, polyethylene glycol 465 or active carbon could be used to precipitate the polysaccharides before DNA precipitation (Schrader 466 et al., 2012). Phenol levels may be reduced with the use of polyvinyl-pyrrolidone or the addition of 467 borate (Wilkins and Smart, 1996). Drying plant samples at 65°C for 2 days, prior to DNA extraction, 468 could also help to cancel out the effect of phenolic inhibitors (Sipahioglu et al., 2006). 469 One of the great advantages of the multiplex qPCR assays we developed is that they are modular and 470 reliable. Combinations of primers and probe can be adapted to include sets aiming at detecting 471 infections at the species and/or only at the subspecies level, and having internal controls for each 472 reaction. We showed here as proofs of concept, that replacing our XF primers and probe with the ones 473 from Harper's test is feasible and leads to highly susceptible test, as using 18S rRNA primers and probe 474 as internal control is efficient. 475 In addition, unlike with identification relying on MLST scheme, the qPCR tetraplex assays allow the 476 simultaneous identification of several subspecies in one sample, as demonstrated with spiked samples. 477 In When a new assay is developed, the time and cost difference with current protocols must be taken into 485 account. The tetraplex qPCR assays are much faster and cheaper than using a test for detection and 486 then a reduced MLST scheme for subspecies assignation. The current protocol costs are for Harper's 487 qPCR detection at the writing time ~0.52€ for reagents, (for a volume of 10 µL) ~1.62€ for the 488 amplification of two housekeeping genes (~0.81€/gene for a volume of 20 µL) and ~10.2€ for their 489 sequencing (~5.1€/gene in both directions), hence totalizing ~12.35€ per sample. In comparison a 490 single tetraplex qPCR assay costs ~0.37€ per sample (for a volume of 10 µL). None of these costs 491 includes the cost of plastic materials or specialized equipment such as a qPCR thermocycler. 492 To conclude, we developed specific, effective, fast, cost-efficient and easy to set up tools allowing in 493 one step to detect and identify at the subspecies level Xf infection directly in plant samples. Compared 494 to current protocols, the LOD of our tetraplex assays allowed subspecies identification at levels where 495 regular amplifications such as the one used for MLST failed. Tetraplex qPCR assays are also easily to 496 perform in a routine lab and as such should be easily transferable to laboratories and are modular 497 according to the user's needs. The authors declare that the research was conducted in the absence of any commercial or financial 532 relationships that could be construed as a potential conflict of interest. The present work reflects only 533 the authors' view and no analysis has been made in the French Reference Lab; in particular ED is not 534 authorized to perform any official tests at Anses.  List of strains used in this study and signals obtained with the primers and probe 737 combinations in simplex qPCR assays on DNA suspensions calibrated at OD600nm = 0.1. 738 Table 2: Description and composition of the longest specific long-mers obtained using SkIf for 739 the various targets. 740 Table 3: Primers and probes designed in this study for Xf detection at the species and subspecies 741 level. 742