Development of a Nested-MultiLocus Sequence Typing approach for a highly sensitive and specific identification of Xylella fastidiosa subspecies directly from plant samples

Different sequence types (ST) of Xylella fastidiosa were already identified in France and Spain based on direct MultiLocus Sequence Typing (MLST) of plant DNA samples. However, direct typing of plant DNA is partly efficient. In order to improve the sensitivity of X. fastidiosa identification, we developed a direct nested-MLST assay on plant extracted DNA. This method was performed based on a largely used scheme targeting seven housekeeping gene (HKG) loci (cysG, gltT, holC, leuA, malF, nuoL, petC). Nested primers were designed from multi-sequence alignments of 38 genomes representing all subspecies and one genome of Xylella taiwanensis. Sequences obtained were long enough to be used for BLAST comparison in PubMLST database. No nonspecific amplification products were observed in these samples. Efficiency of the nested-MLST was tested on extracted DNA from 106 samples proven positive (Cq<35) or equivocal (35≤Cq≤40) using the Harper’s qPCR test. Samples analyzed included 49 plant species and two insect species (Philaenus spumarius, Neophilaenus campestris) that were collected in 2017 (106 plant samples in France), in 2018 (162 plant samples in France, 40 plant samples and 26 insect samples in Spain), and in 2019 (30 plant samples in Spain). With the conventional-MLST assay, no complete MLST profile was obtained for any of the samples from France and for most samples (59/66) from Spain. Conversely, with the nested approach, complete profiles were obtained for six French plant samples, 55 Spanish plant samples and nine Spanish insect samples. The threshold was improved by 100 to 1000 times compared to conventional PCR and was between 22 pg.mL−1 to 2.2 pg.mL−1 depending on the HKG. Using nested-MLST assay, plants that were not yet considered hosts tested positive and revealed novel alleles in France, whereas for Spanish samples it was possible to assign the subspecies or ST to samples considered as new hosts in Europe. Direct typing by nested-MLST from plant material has an increased sensitivity and may be useful for epidemiological purposes.

also been successfully transferred to be used in digital PCR (Dupas et al. 2019a). Some of these tests were designed to detect only one subspecies. This is the case of the nested PCR test 69 proposed by Ciapina et al (2004) for detecting CVC strains (subspecies pauca) in 70 sharpshooters and citrus plants and also of the qPCR test targeting oleander leaf scorch strains 71 (that are included in the subspecies fastidiosa) (Guan et al. 2013). Other tests were designed 72 to detect and discriminate two or more subspecies (Burbank & Ortega 2018;Dupas et al. 73 2019b). 74 Precise identification of Xf at an infraspecific level is essential for epidemiological and 75 surveillance analyses, and to allow a proper description of the population structure and their 76 dynamics. The widely used multilocus sequence typing (MLST) scheme designed for Xf 77 (Scally et al. 2005, Yuan et al. 2010) is based on amplification by conventional PCR and 78 designed inside the gene fragments used in the conventional-MLST scheme and hence some informative sites are lost. MLST with nested PCRs has also been developed in medical field 94 to enable the direct typing of samples infected by Leptospira or Trichomonas, for example 95 (Weiss et al. 2016; Van der Veer et al. 2016). 96 The objective of this study was to develop a Xf detection assay based on the largely used 97 suspensions were prepared from fresh cultures in sterile distilled water, adjusted at 118 OD 600 nm = 0.1 and used as templates for amplification after boiling for 20 minutes, thermal 119 shock on ice and centrifugation 10 000g, 10 min. 120

Nested-MLST primers and reactions 152
The seven HKG sequences (cysG, gltT, holC, leuA, malF, nuoL, petC) were extracted from 39 Xylella and 1292 Xanthomonas, and in vitro on target and non-target bacterial strains (Table  168 1). 169 PCRs were performed in 25 μ l reaction buffer (Promega) with MgCl 2 at 1.5mM final, 200µM 170 dNTP, 300µM each of the forward and reverse primers, 0.6 U GoTaq G2 (Promega) and 2µl 171 of sample DNA. The first-round PCR program consisted of an initial denaturation step of 3 172 min at 95°C followed by 35 cycles of 30 s denaturation at 95°C, 30 s annealing at the relevant 173 temperature according to each gene (determined by gradient PCR) and 60s elongation at 72°C 174 followed by a final extension step of 10 min at 72°C (

Nested-MLST proved to be specific 201
The specificity of the outer and inner primer pairs was tested in silico and in vitro. In silico,202 all primers pairs showed the best scores of alignment with Xf genomic sequences. Some non-203 target organisms showed sequences nearly identical at outer primer locations with only one 204 mismatch and a similar expected fragment size, but sequences of inner primers were more 205 different indicating that there will be no amplification. This was the case for various 206 Xanthomonas strains that contained one mismatch at position 15 of the petC forward outer 207 primer and an identical sequence for the outer reverse primer. X. taiwanensis holC sequence 208 corresponding to inner primers contained also only one mismatch. The fragment size 209 predicted was as expected for Xf. Other predictions with one mismatch located in primers did 210 not end in fragment amplifications of the same expected size. Then, the specificity of the 211 outer and inner primer pairs (Table 2) was validated in vitro on five target strains and 33 non-212 target strains (Table 1). Specificity of the nested-MLST assay could not been tested in vitro 213 on X. taiwanensis as no strain was available. Amplifications were obtained for all Xf strains. the first round of the nested PCR for the petC outer primers, providing a product of the 217 expected size. However, these products were not amplified in the second round of the nested 218 PCR and no false positive signal was finally obtained. 219

Nested-MLST Limit of Detection is comparable to that of qPCR 220
The sensitivity of each primer combination was evaluated on serial dilutions of a genomic 221 DNA solution calibrated (Qubit fluorimeter, Invitrogen) at 220 ng.mL -1 (Figure 1). First round 222 PCRs gave a signal more or less intense for concentrations up to 2.2 ng.mL -1 (0.8 x 10 6 223 copies.mL -1 ) for all HKG except malF and cysG (220 pg.mL -1 ). The second round of PCRs 224 allowed a sufficiently strong signal for sequencing for concentrations up to 22 pg.mL -1 (0.8 x 225 10 4 copies.mL -1 ) for gltT, holC, petC, leuA, cysG, and up to 2.2 pg.mL -1 (0.8 x 10 3 copies.mL -226 1 ) for nuoL and malF. The same range of genomic DNA solutions was tested with the 227 Harper's qPCR test to compare sensitivity of these two tests (S3 Table). The latest signal 228 (LoD) for the Harper's qPCR test (Cq =37.64) was obtained with the concentration of 0.8.10 3 229 copies.mL -1 and no amplification was obtained for lower concentrations. was obtained as all signals remained around 0.8 x 10 6 bacteria.mL -1 (S1 Figure).  Table 3 and S4 Table). Positive and equivocal 240 samples were tested using the first round of PCR of the MLST assay: five samples from 2017 241 (one Spartium junceum, three Polygala myrtifolia, and one Genista corsica) gave a signal for 242 at least one gene, but no complete typing was obtained for any sample.

Nested-MLST improved successful HKG typing by increasing sensitivity level 247
were run in the first and the second PCR and were always negative. The negative control 286 coming from the first reaction always tested negative in the second one, confirming the 287 absence of contamination during the entire process.. Positive control was a suspension of 288 strain CFBP 8084 (ST29) from the subspecies morus or strain CO33 (ST72) as this STs were 289 not previously found in Corsica, France or Spain, respetively. 290 291

Nested-MLST allowed identification of new alleles among French samples 292
Incomplete profiles were obtained for various French samples due to variable amplification 293 efficiencies varying according to the HKG. From 9% (with gltT) to 55% (with holC) of 294 French samples gave a signal applying the nested-MLST assay. Alleles that were not yet 295 described in plant samples in France were detected in 2017. This was the case for holC_1 and 296 holC_2 alleles known to occur in ST from ST1 to ST5 and ST75 that cluster in the subspecies 297 fastidiosa (https://pubmlst.org/xfastidiosa/). These alleles were sequenced in samples of 298

Recombinants or mixed infections were identified by nested-MLST 314
Some French samples were further sequenced for several loci and these sequencing confirmed 315 the presence of alleles occurring in the subspecies fastidiosa, multiplex and pauca (S4 Table). 316 All alleles were previously described but were detected in combinations that were not 317 previously described, suggesting the presence of recombinants or of mix infections (S4 318   Table). This is the case for Cistus monspeliensis 7 showing an unknown combination of 319 The objective of this study was to improve the published MLST scheme supported by a public 410 website (https://pubmlst.org/xfastidiosa/) by designing nested primers to lower the limit of 411 detection and help in Xf diagnosis and typing. Thus, this improved MLST assay enables a 412 higher sensitivity and specific typing of Xf directly from plant and insects samples without the 413 need of isolating the strain and at an affordable cost. 414