Development of A Nested-MultiLocus Sequence Typing Approach for A Highly Sensitive and Specific Identification of Xylella fastidiosa Subspecies Directly from Plant Samples

Abstract

Identification of sequence types (ST) of Xylella fastidiosa based on direct MultiLocus Sequence Typing (MLST) of plant DNA samples 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). 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 nested approach, a significant higher number of samples were amplified. The threshold was improved by 100 to 1000 times compared to conventional PCR. 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.

1. Introduction

Xylella fastidiosa (Xf) is the causal agent of several devastating diseases of plants in the Americas and this pathogen was recently detected in Europe, where it causes a severe disease in olive trees in Italy and is present in several other regions. This species encompasses three well recognized subspecies, namely fastidiosa, multiplex, and pauca [1,2] but other subspecies are currently described [3]. The subspecies fastidiosa occurs in North and Central America and was recently detected in Spain (https://gd.eppo.int/taxon/XYLEFA/). It infects a wide range of trees, ornamentals, and other perennials and includes strains responsible for the well-known Pierce’s disease on grapevine [3,4]. The subspecies multiplex is present in North and South America and in Europe (https://gd.eppo.int/taxon/XYLEFA/) and is associated with scorches and dieback of a wide range of trees and ornamentals [3]. The subspecies pauca is mostly found in South and Central America on Citrus spp. and Coffea spp. [5]), but has been recently detected also in olive trees in Spain (https://gd.eppo.int/taxon/XYLEFA/), Brazil [6], Argentina [7], and Italy [8]. Its host range includes also ornamentals and other trees [3]. Altogether more than 560 plant species are hosts of Xf [3]. This member of the Xanthomonadaceae family inhabits the xylem of its host plants [9] and is naturally transmitted by insects from plants to plants.

A range of detection tests has been proposed for Xf [10]. Several immunological methods are available [10]. However, such methods have high limits of detection (LoDs) that are close to 10⁴ to 10⁵ cells.mL⁻¹ [10]. End point and also quantitative PCR (qPCR) are nowadays widely used, with a better sensivity as the LoD is around 10² cells.mL⁻¹ for several qPCR tests [11,12,13,14]. The Harper’s qPCR test is often used in Europe for its high sensitivity. Several tests based on isothermal amplification have also been reported [12,14,15,16]. The Harper’s test has also been successfully transferred to be used in digital PCR [17]. Some of these tests were designed to detect only one subspecies. This is the case of the nested PCR test proposed by [18] for detecting CVC (Citrus Variegated Chlorosis) strains (subspecies pauca) in sharpshooters and citrus plants and also of the qPCR test targeting oleander leaf scorch strains (that are included in the subspecies fastidiosa) [19]. Other tests were designed to detect and discriminate two or more subspecies [16,20].

Precise identification of Xf at an infraspecific level is essential for epidemiological and surveillance analyses, and to allow a proper description of the population structure and their dynamics. The widely used MultiLocus Sequence Typing (MLST) scheme designed for Xf [21,22] is based on amplification by conventional PCR and sequencing of seven HKG (housekeeping gene) fragments (loci), either from strains or from plant samples [23]. For each locus, the different sequence variants are considered as distinct alleles. The combination of allele numbers defines the sequence type (ST). The MLST-Xf data are stored in a public database (https://pubmlst.org/xfastidiosa/) that can be used to automatically identify and assign new allele variants, and provide tools to analyze the potential origin of the strains. The association of the different subspecies with their host plants is useful to better understand Xf epidemiology.

A reliable and enough informative typing method is particularly relevant in cases of new outbreaks or for the description of new host. Due to the large number of host plants to be analyzed, various types of inhibitors can interfere with reagents of PCR and low bacterial loads compromising PCR efficiency and hence typing. Improving DNA extraction methods can, at least partly, solve the problem of PCR inhibitors, and nested PCR appears a solution to allow the detection of low bacterial population sizes. A nested-MLST was already successfully developed to detect and type Xf in vectors [24]. Primers were 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 to enable the direct typing of samples infected by Leptospira or Trichomonas, for example [25,26].

The objective of this study was to develop a Xf detection assay based on the largely used MLST scheme [22] that lowers the limit of detection (LoD) to enable at least the identification of Xf subspecies and, if possible, provide larger sets of typing data directly from plant samples. We used genomic sequences to improve each PCR efficiency and showed a drastic increase in the sensitivity as compared to that of the conventional-MLST approach.

2. Materials and Methods

2.1. Strains and Media

A collection of target and non-target bacterial strains was used to test in vitro the specificity of the newly designed primers and the nested PCR assays. This set was made of five X. fastidiosa strains from different subspecies and 34 strains representing bacteria phylogenetically close to Xf, i.e., various Xanthomonas, as well as strains of other plant pathogenic bacteria and endosymbionts potentially inhabiting the same niches as Xf (

Table 1. List of target and non-target strains used to verify the specificity of nested-MLST primers.

CFBP Code	Bacterial Species	Host Plant	Origin
6448	Agrobacterium rubi	Rubus ursinus var. loganobaccus	USA (1942)
2413	Agrobacterium tumefaciens	Malus sp.	NA (1935)
5523	Agrobacterium vitis	Vitis vinifera	Australia (1977)
2404	Clavibacter insidiosus	Medicago sativa	USA (1955)
4999	Clavibacter michiganensis	Lycopersicon esculentum	Hungary (1957)
3418	Curtobacterium flaccumfaciens pv. flaccumfaciens	Phaseolus vulgaris	Hungary (1957)
1200	Dickeya dianthicola	Dianthus caryophyllus	United Kingdom (1956)
5561	Ensifer meliloti	Medicago sativa	VA, USA (1984)
1232	Erwinia amylovora	Pyrus communis	United Kingdom (1959)
3845	Pantoea agglomerans	Knee laceration	Zimbabwe (1956)
3167	Pantoea stewartii pv. stewartii	Zea mays var. rugosa	USA (1970)
3205	Pseudomonas amygdali	Prunus amygdalus	Greece (1967)
8305	Pseudomonas cerasi	Prunus cerasus	Poland (2007)
7019	Pseudomonas congelans	na 1	Germany (1994)
1573	Pseudomonas syringae pv. persicae	Prunus persica	France (1974)
1392	Pseudomonas syringae pv. syringae	Syringa vulgaris	United Kingdom (1950)
7436	Rhizobium nepotum	Prunus ceresifera myrobolan	Hungary (1989)
13100	Stenotrophomas maltophilia	Phaseolus vulgaris	Cameroon (2009)
3371	Xanthomonas euvesicatoria pv. citrumelonis	Citrus sp.	USA (1989)
2528	Xanthomonas arboricola pv. juglandis	Juglans regia	New Zealand (1956)
2535	Xanthomonas arboricola pv. pruni	Prunus salicina	New Zealand (1953)
4924	Xanthomonas axonopodis pv. axonopodis	Axonopus scoparius	Colombia (1949)
5241	Xanthomonas campestris pv. campestris	Brassica oleracea var. gemmifera	United Kingdom (1957)
2901	Xanthomonas citri pv. aurantifolii	Citrus limon	Argentina (1988)
2525	Xanthomonas citri pv. citri	Citrus limon	New Zealand (1956)
7660	Xanthomonas citri pv. viticola	Vitis vinifera	India (1969)
2625	Xanthomonas gardneri	Medicago sativa	Reunion Island (1986)
4925	Xanthomonas hortorum pv. hederae	Hedera helix	USA (1944)
2533	Xanthomonas hortorum pv. pelargonii	Pelargonium peltatum	New Zealand (1974)
1156	Xanthomonas hyacinthi	Hyacinthus orientalis	Netherlands (1958)
2532	Xanthomonas oryzae pv. oryzae	Oryza sativa	India (1965)
2054	Xanthomonas translucens	Hordeum vulgare	USA (1933)
2543	Xanthomonas vasicola pv. holcicola	Sorghum vulgare	New Zealand (1969)
7970	Xylella fastidiosa subsp. fastidiosa	Vitis vinifera	USA (1987)
8416	Xylella fastidiosa subsp. multiplex	Polygala myrtifolia	France (2015)
8084	Xylella fastidiosa subsp. morus	Morus alba	USA (na1)
8070	Xylella fastidiosa subsp. multiplex	Prunus spp.	USA (2004)
8402	Xylella fastidiosa subsp. pauca	Olea europea	Italy (2014)
1192	Xylophilus ampelinus	Vitis vinifera	Greece (1966)

¹: not available.

), available at the French Collection of Plant-Associated Bacteria (CIRM-CFBP; https://www6.inra.fr/cirm_eng/CFBP-Plant-Associated-Bacteria).

The Xf strains were grown on modified PWG media (agar 12 g.L⁻¹; soytone 4 g.L⁻¹; bacto tryptone 1 g.L⁻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 M) 10 mL.L-1; BSA (7.5%) 24 mL.L⁻¹; L-glutamine 4 g.L⁻¹) at 28 °C for one week. Agrobacterium and Rhizobium were grown at 25 °C for one to two days on MG medium [27]; TSA was used (tryptone soybroth 30 g.L⁻¹; agar 15 g.L⁻¹) for Clavibacter, Ensifer, Stenotrophomonas, Xanthomonas and Xylophilus; and King’s medium B [28] for Dickeya, Erwinia, Pantoea and Pseudomonas. For PCRs, bacterial suspensions were prepared from fresh cultures in sterile distilled water, adjusted at OD_{600 nm} = 0.1 and used as templates for amplification after boiling for 20 min, thermal shock on ice and centrifugation 10 000 g, 10 min.

2.2. DNA Extraction

Genomic DNA from Xf strain CFBP 8070 was extracted with the Wizard genomic DNA Purification Kit (Promega, France) and used to prepare a 10-fold serial dilutions from 220 ng.mL⁻¹ (corresponding to 0.8 × 10⁸ copies.mL⁻¹ of genomic DNA) to 22 fg.mL⁻¹ (8 copies.mL⁻¹) to evaluate the LoD of the nested-MLST. Copies number were calculated using an estimated genome size of 2 903 976 bp, knowing that 1 pg = 9.78 × 10⁸ bp [29]. A total of 268 plant samples were collected in Corsica, France, based on symptoms compatible with those caused by X. fastidiosa; 106 samples were collected in June 2017 and 162 in September 2018. For each French sample, DNA was extracted as described in [10] using two methods in order to optimize the chances of detection. CTAB-based (Cetyl TrimethylAmmonium Bromide) extraction and robotic QuickpickTM SML kit from Bio-Nobile were used with the following modification: a sonication step (1 min, 42 KHz) was added after the samples (petioles, twigs) were finely cut, and was followed by a 15-min incubation period at room temperature. For initial laboratory diagnosis MLST results were compared with the Harper’s qPCR test [12] as in [10] with following modifications: primers XF-F and XF-R, and probe XF-P [12] were used at a final concentration of 0.6 µM and 0.2 µM respectively, non-acetylated BSA (Bovine serum Albumine) was used at final concentration of 1.5 µg.µL⁻¹, and 2 µL of DNA were used in 10 µL reaction volume. The target of this PCR is located in the gene coding for the 16S rRNA-processing RimM protein. Each DNA sample was tested in triplicates. To validate the nested PCR, DNA samples were provided by the National Reference Laboratory for Phytopathogenic Bacteria, Valencia, Spain, and from the Official Phytosanitary Laboratory of the Balearic Islands for determining Xf subspecies. Those DNA samples correspond to DNA extractions made from symptomatic plants sampled during official monitoring surveys. A total of 70 Xf-infected samples were analyzed from Balearic Islands and mainland Spain during 2018 (40 samples) and 2019 (30 samples), as well as 26 insect samples from both regions. DNA was extracted from petioles of symptomatic leaves as described in [10] using a CTAB-based extraction method for plant samples from Alicante and insect samples from Alicante and Balearic Islands. A Mericon DNeasy Food kit from Qiagen was used for plant samples from Balearic Islands. All DNA extraction methods have been validated; validation data is available in the EPPO (European and Mediterranean Plant Protection Organization) Database on Diagnostic Expertise [10].

2.3. Nested-MLST Primers and Reactions

The seven HKG sequences (cysG, gltT, holC, leuA, malF, nuoL, petC) were extracted from 39 Xf genome sequences (Table S1) [2] to design the nested primers. Alignments were performed with BioEdit sequence alignment editor. The primers designed by [22] were destined to be used as inner primers (PCR2) (

Table 2. Primer sequences used in the X. fastidiosa nested-MLST scheme.

Locus	PCR Round	5′-Forward Primer-3′	5′-Reverse Primer-3′	Position on Xf M12 Genome (CP000941.1)	Annealing Temperature (°C)	Size (pb) of Reaction Product
cysG	1	CCAAACATAGAAGCACGCCG	GCGAGTGTTTTCAGCGTTCC	2111116–2111891	64	776
	2	GCCGAAGCAGTGCTGGAAG 1	GCCATTTTCGATCAGTGCAAAAG 1	2111203–2111844	56	642
gltT	1	GGTGCCATCCAATCCGTTTT	TCAGGATGTCCCAATTCCAACG	1731589–1732504	60	916
	2	TCATGATCCAAATCACTCGCTT 1	TTACTGGACGCTGCCTCG	1731783–1732482	56	700
holC	1	CCGATGGTGAAGAACAGTAGACA	GCTCGAGAAACTSGATTAATGG	133166–133714	62	549
	2	GGTCACATGTCGTGTTTGTTC	CACGCGCCGACTTCTATTT	133269–133692	59	424
leuA	1	CGAAGGTGCAAACAAAGTGA	CGCACTGGCTTCGATAATGTCT	1271664–1272549	58	886
	2	GGTGCACGCCAAATCGAATG 1	ACTGGTCCCTGTACCTTCGT	1271752–1272525	60	774
malF	1	AACGTCGTCACCCCAAGAA	ATGAGGCGGGCTTCTTTGG	1680264–1681108	56	845
	2	AGCAGAAGCACGTCCCAGAT	CTGGTCCTGCGGTGTTGG	1680308–1681074	60	767
nuoL	1	TTGGTACGTTGGCTTTGGTG	GACAAAACCAGATTGCGTGC	325347–326191	60	845
	2	GCGACTTACGGTTACTGGGC	ACCACCGATCCACAACGCAT 1	325454–326050	54	597
petC	1	TCAATGCACGTCCTCCCAAT	GGCTGCCATTCGTTGAAGTA	2020498–2021079	60	582
	2	ACGTCCTCCCAATAAGCCT	CGTTATTCACGTATCGCTGC	2020505–2021055	56	551

¹: primers from [22].

) in our nested assay.

We checked their characteristics with Primer3 V4.1.0 software (http://primer3.ut.ee/). Because of high Tm differences between forward and reverse primers for some primer pairs (gltT) (Table S2), or high hairpin Tm values (holC forward primer), some primers from [22] were redesigned nearly at the same positions to improve their efficiency. Moreover, as primer sequences were already near the locus sequence ends, we also had to relocate some of them to design nested primers inside the sequence alignments without loss of informative sites. Outer primers (PCR1) were designed with Primer3 V4.1.0 software (http://primer3.ut.ee/) in flanking regions targeted by the inner primers. Outer and inner primers were tested In silico using a primer search tool available in the galaxy toolbox of CIRM-CFBP (https://iris.angers.inra.fr/galaxypub-cfbp) on 194,438 bacterial Whole Genome Shotgun (WGS) sequences available in the NCBI database (as on March, 2019) including 58 Xylella and 1292 Xanthomonas (Table S3), and in vitro on target and non-target bacterial strains (Table 1).

PCRs were performed in 25 μL reaction buffer (Promega) with MgCl₂ at 1.5 mM final, 200 µM dNTP, 300 µM each of the forward and reverse primers, 0.6 U GoTaq G2 (Promega) and 2 µL of sample DNA. The first-round PCR program consisted of an initial denaturation step of 3 min at 95 °C followed by 35 cycles of 30 s denaturation at 95 °C, 30 s annealing at the relevant temperature according to each gene (determined by gradient PCR) and 60s elongation at 72 °C followed by a final extension step of 10 min at 72 °C (Table 2). The second round was performed with 30 cycles under same conditions and same concentrations but with a final volume of 50 µL for sequencing purposes and with 4 µL of first-round PCR product. The primer pairs of the second round of each nested PCR were used for sequencing (by Genoscreen, Lille, France for French samples and by Stabvida, Caparica, Portugal, for Spanish samples) the corresponding PCR products after 1.8% agarose gel visualization. To avoid contamination, one sample was opened at a time and stringent cleaning measures were applied after each experiment.

2.4. Statistical Analysis

The sensitivity of detection by conventional- and nested-MLST PCRs were compared in plant and vector samples for the seven HKGs that were analyzed by both approaches, by using a Chi square test using SAS (version 9.4, SAS Institute, Cary, NC, USA). Analysis was performed for the Spanish samples only, as HKG-PCRs were not systematically carried out on the French samples. Results were considered significantly different when p ≤ 0.05.

2.5. Sequence Acquisition, Alignment and Analyses

Forward and reverse nucleotide sequences were assembled, and aligned using Geneious 9.1.8 software (French samples) or Bionumerics V7.6.3 software (Spanish samples) to obtain high quality sequences. ST or loci assignation was performed according to http://pubmlst.org/xfastidiosa/. To reduce the costs of sequencing for French samples, only PCR products obtained for samples showing the highest rate of successful HKG amplifications were sequenced. On the other hand, all positive holC amplifications were sequenced to obtain a larger view of alleles present in Corsica.

A flow chart summarizing the different steps of the nested-MLST method is presented in Figure 1.

3. Results

3.1. Nested-MLST Proved to be Specific

The specificity of the outer and inner primer pairs was tested In silico and in vitro. In silico, all primers pairs showed the best scores of alignment with Xf genomic sequences. Some non-target organisms showed sequences nearly identical at outer primer locations with only one mismatch and a similar expected fragment size, but sequences of inner primers were more different indicating that there will be no amplification. This was the case for various Xanthomonas strains that contained one mismatch at position 15 of the petC forward outer primer and an identical sequence for the outer reverse primer. X. taiwanensis holC sequence corresponding to inner primers contained also only one mismatch. The fragment size predicted was as expected for Xf. Other predictions with one mismatch located in primers did not end in fragment amplifications of the same expected size. Then, the specificity of the outer and inner primer pairs (Table 2) was validated in vitro on five target strains and 34 non-target strains (Table 1). Specificity of the nested-MLST assay could not been tested in vitro on X. taiwanensis as no strain was available. Amplifications were obtained for all Xf strains. No amplification was detected on the non-target strains except for strain CFBP 2532 (Xanthomonas oryzae pv oryzae) and CFBP 2533 (Xanthomonas hortorum pv. pelargonii) in the first round of the nested PCR for the petC outer primers, providing a product of the expected size. However, these products were not amplified in the second round of the nested PCR and no false positive signal was finally obtained.

3.2. Nested-MLST Limit of Detection is Comparable to That of qPCR

The sensitivity of each primer combination was evaluated on serial dilutions of a genomic DNA solution calibrated (Qubit fluorimeter, Invitrogen) at 220 ng.mL⁻¹ (Figure 2). First round PCRs gave a signal more or less intense for concentrations up to 2.2 ng.mL⁻¹ (0.8 × 10⁶ copies.mL⁻¹) for all HKG except malF and cysG (220 pg.mL⁻¹). The second round of PCRs allowed a sufficiently strong signal for sequencing for concentrations up to 22 pg.mL⁻¹ (0.8 × 10⁴ copies.mL⁻¹) for gltT, holC, petC, leuA, cysG, and up to 2.2 pg.mL⁻¹ (0.8 × 10³ copies.mL⁻¹) for nuoL and malF.

The same range of genomic DNA solutions was tested with the Harper’s qPCR test to compare sensitivity of these two tests (Table S4). The latest signal (LoD) for the Harper’s qPCR test (Cq = 37.64) was obtained with the concentration of 0.8.10³ copies.mL⁻¹ and no amplification was obtained for lower concentrations.

Previously, we evaluated the LoD of the conventional PCRs for cysG and malF of the initial MLST scheme [22] on a range of dilutions of CFBP 8070 genomic DNA with the Platinum Taq polymerase (Invitrogen) and tested the effect of adding BSA (final concentration at 0.3µg. µL⁻¹) on the efficiency of the conventional PCRs. No improvement was obtained as all signals remained around 0.8 × 10⁶ bacteria.mL⁻¹ (Figure S1).

3.3. Analysis of Naturally Infected Samples

Using qPCR Harper’s test, 22 samples from 2017 and eight samples from 2018 collected in France were positive (Cq values < 35) with one or both DNA extraction methods; 70 samples from 2017 and 36 samples from 2018 were equivocal (35 ≤ Cq ≤ 40), 14 samples from 2017 and 118 from 2018 were negative (Cq > 40) (

Table 3. Number of samples, positive and equivocal in Harper’s qPCR. Percentage of successful amplifications obtained for each locus in conventional and nested PCR.

						Percentage of Successful Amplifications Obtained for Each Locus in Conventional and Nested MLST-PCR
Sample Type	Country	Year	Number of Samples	qPCR Harper Number of Samples		cysG		gltT		holC		leuA		malF		nuoL		petC		Average per Year
				Cq < 35	Cq ≥ 35	conv	nest	conv	nest	conv	nest	conv	nest	conv	nest	conv	nest	conv	nest	conv	nest
Plant	France	2017	106	22	70	1.1	28.3	2.2	26.1	4.3	55.4	4.3	34.8	1.1	35.9	0	26.1	1.1	46.7	2	36.2
Plant	France	2018	162	8	36	0	11.4	0	9.1	0	27.3	0	27.3	0	15.9	0	27.3	0	25	0	20.5
Plant	Spain	2018	40	38	2	55	90 *	10	77.5 *	15	80 *	12.5	75 *	30	75 *	40	85 *	15	85 *	25.4	81.1
Plant	Spain	2019	30	30	0	30	90 *	13.3	90 *	16.7	93.3 *	16.7	90 *	20	90 *	66.7	90 *	20	90 *	26.2	90.5
Insect	Spain	2018	26	18	8	65.4	80.8	7.7	73.1 *	19.2	69.2 *	11.5	57.7 *	7.7	53.8 *	26.9	57.7 *	15.4	73.1 *	22	66.5

(*) Asterisk indicates a significant (p < 0.05) higher number of successful amplifications for nested-MLST as compared to conventional-MLST [22] according to a Chi-square test. The test was conducted only for the Spanish samples on the number of samples, even if frequencies are indicated in the table for MLST-PCR.

and Table S5). Positive and equivocal samples were tested using the first round of PCR of the MLST assay: five samples from 2017 (one Spartium junceum, three Polygala myrtifolia, and one Genista corsica) gave a signal for at least one gene, but no complete typing was obtained for any sample. No sample from 2018 gave a signal. Most of Spanish samples used to evaluate nested-MLST scheme were positive using Harper’s qPCR (only two out of 40 plant samples were equivocal in 2018 and eight out of 26 vector samples).

3.4. Nested-MLST Improved Successful HKG Typing by Increasing Sensitivity Level

Using nested-MLST for French samples, full allelic profiles were obtained for five samples from 2017 and one from 2018 corresponding to the lowest Cq in Harper’s qPCR test (

Table 4. Allele numbers and STs obtained for fully typed samples in France and Spain for plant and insect samples. The numbers correspond to the names of the samples.

Country	Sample Names	cysG	gltT	holC	leuA	malF	nuoL	petC	Sequence Type (ST)
France	Spartium junceum 2	7	3	3	3	3	3	3	ST7
France	Polygala myrtifolia 3, 4	3	3	3	3	3	3	3	ST6
France	Genista corsica 1	7	3	3	3	3	3	3	ST7
France	Polygala myrtifolia 5, 6	7	3	3	3	3	3	3	ST7
Spain	Cistus albidus 2	31	15	10	7	17	16	6	ST80
Spain	Ficus carica 1	1	1	1	1	1	1	1	ST1
Spain	Helichrysum italicum 1	3	3	3	3	3	3	3	ST6
Spain	Juglans regia 1	1	1	1	1	1	1	1	ST1
Spain	Lavandula angustifolia 1	32	3	3	3	3	3	3	ST81
Spain	Olea europaea 1	3	3	3	3	3	3	3	ST6
Spain	Phagnalon saxatile 1	3	3	3	3	3	3	3	ST6
Spain	Polygala myrtifolia 1	3	3	3	3	3	3	3	ST6
Spain	Prunus armeniaca 1	3	3	3	3	3	3	3	ST6
Spain	Prunus domestica 1	32	3	3	3	3	3	3	ST81
Spain	Prunus domestica 2	3	3	3	3	3	3	3	ST6
Spain	Prunus dulcis 4–8,10,11,15,18–26,30–47	3	3	3	3	3	3	3	ST6
Spain	Prunus dulcis 9	31	15	10	7	17	16	6	ST80
Spain	Prunus dulcis 1,2	32	3	3	3	3	3	3	ST81
Spain	Prunus dulcis 3	7	3	3	3	3	3	3	ST7
Spain	Rhamnus alaternus 1	3	3	3	3	3	3	3	ST6
Spain	Rosmarinus officinalis 4	3	3	3	3	3	3	3	ST6
Spain	Rosmarinus officinalis 1,3	31	15	10	7	17	16	6	ST80
Spain	Prunus domestica 3	3	3	3	3	3	3	3	ST6
Spain	Philaenus spumarius 6,7,8,10,11	1	1	1	1	1	1	1	ST1
Spain	Philaenus spumarius 1	3	3	3	3	3	3	3	ST6
Spain	Philaenus spumarius 22	32	3	3	3	3	3	3	ST81
Spain	Neophilaenus campestris 1,2	3	3	3	3	3	3	3	ST6

and Table S5) Among fully typed samples, four were X. fastidiosa subsp. multiplex ST7 (Genista corsica, Polygala myrtifolia, Spartium junceum), and two were X. fastidiosa subsp. multiplex ST6 (Polygala myrtifolia).

Our scheme was also evaluated on Spanish samples already proved infected by Xf. These samples from different outbreaks showed a wide range of Cq values ranging from 18.8 to 36.0 for plant samples and from 23.29 to 37.0 for insect samples (Table 3 and Table S5). Samples were first analyzed using the conventional-MLST assay [22]. Amplification efficiency was variable and ranged from 10% for gltT to 67% for nuoL with an average of 25% and 26% for the seven HKG in 2018 and 2019, respectively. The nested-MLST assay improved the amplification efficiency that increased to 75% for leuA and up to 93% for holC with an average of 81% and 91% in 2018 and 2019, respectively. In total, full allelic profiles were obtained in seven plant samples using the conventional-MLST assay, whereas a total of 55 samples were fully typed with the improved nested-MLST assay (Table 4). For the 70 plant DNA samples that were tested by both protocols, for all the seven HKGs, conventional-MLST showed a significant (p < 0.0005 for 2018 and p < 0.0283 for 2019) lower number of samples amplified as compared to nested-MLST. Among fully typed plant samples using the nested-MLST, we identified X. fastidiosa subsp. fastidiosa ST1 in Ficus carica and Juglans regia, X. fastidiosa subsp. multiplex ST6 in Helichrysum italicum, Olea europaea, Phagnalon saxatile, Polygala myrtifolia, Prunus armeniaca, Prunus domestica, Prunus dulcis, Rhamnus alaternus, and Rosmarinus officinalis, X. fastidiosa subsp. multiplex ST7 in Prunus dulcis, X. fastidiosa subsp. multiplex ST81 in Lavandula angustifolia and Prunus dulcis, and X. fastidiosa subsp. pauca ST80 in Cistus albidus, Prunus dulcis, and Rosmarinus officinalis.

Not all insect samples could be tested by both protocols due to restrictions in DNA amount. In samples tested only by the original MLST assay [22], the percentages of successful amplifications ranged from 8% (gltT and malF) to 65% (cysG). With the nested-MLST assay, successful amplifications ranged from 54% (malF) to 81% (cysG), with an average efficiency for the seven HKG of 22% to 67% for conventional and nested approach, respectively (Table 3 and Table S5). Nine insect samples were fully typed using a combination of both protocols (Table 4). X. fastidiosa subsp. fastidiosa ST1 was identified in insects from Mallorca (Balearic Islands), X. fastidiosa subsp. multiplex ST6 in insects from Alicante (mainland Spain) and X. fastidiosa subsp. multiplex ST81 in insects from Balearic Islands. For the nine insect samples that were tested by both protocols, conventional-MLST showed a significant (p < 0.0247) lower number of samples amplified as compared to nested-MLST for six of the seven HKGs (excluding cysG). These results indicate that for insect samples it is also better to use directly the improved nested-MLST assay.

No nonspecific amplicons were observed in any of the samples. Negative controls (water) were run in the first and the second PCR and were always negative. The negative control coming from the first reaction always tested negative in the second one, confirming the absence of contamination during the entire process. Positive control was a suspension of strain CFBP 8084 (ST29) from the subspecies morus or strain CO33 (ST72) as this STs were not previously found in Corsica, France or Spain, respectively.

3.5. Nested-MLST Allowed Identification of New Alleles Among French Samples

Incomplete profiles were obtained for various French samples due to variable amplification efficiencies varying according to the HKG. From 9% (with gltT) to 55% (with holC) of French samples gave a signal applying the nested-MLST assay. Alleles that were not yet described in plant samples in France were detected in 2017. This was the case for holC_1 and holC_2 alleles known to occur in ST from ST1 to ST5 and ST75 that cluster in the subspecies fastidiosa (https://pubmlst.org/xfastidiosa/). These alleles were sequenced in samples of Asparagus acutifolius, Eleagnus, Cistus monspeliensis and C. creticus, Quercus ilex, Myrtus myrtifolia, Olea europaea, Platanus, Arbutus unedo (Table S5). Other holC alleles already described in STs clustering in the subspecies fastidiosa (holC_24) were also sequenced from Cistus monspeliensis and Pistaccia lentiscus. HolC_10 alleles described in STs clustering in the subspecies pauca were sequenced from Cistus monspeliensis and C. salicifolius, Cypressus, Metrosideros excelsa, Myrtus communis, Pistaccia lentiscus, Quercus ilex, Rubia peregrina, Smilax aspera samples. Similarly, holC_3 (known in ST6, ST7, ST25, ST34, ST35, ST79, ST81 and ST87 clustering in the subspecies multiplex) were obtained from samples of Acer monspeliensis, Arbutus unedo, Calicotome spinosa, Cistus monspeliensis, Genista corsica, Myrtus communis, Olea europaea, Phyllirea angustifolia, Polygala myrtifolia, Quercus ilex and Quercus pubescens, Spartium junceum. Among samples from 2018, only holC_1 allele was detected in Olea europaea, Quercus ilex, and Platanus sp. samples, and holC_3 allele in Cistus monspeliensis, Acer monspeliensis, Myrtus communis, and Polygala myrtifolia samples.

3.6. Recombinants or Mixed Infections Were Identified by Nested-MLST

Some French samples were further sequenced for several loci and these sequencing confirmed the presence of alleles occurring in the subspecies fastidiosa, multiplex and pauca (Table S5). All alleles were previously described but were detected in combinations that were not previously described, suggesting the presence of recombinants or of mix infections (Table S5). This is the case for Cistus monspeliensis 7 showing an unknown combination of cysG_2/petC_2/nuoL_2/gltT_2 (known in ST5) with malF_4 (known in ST2), both from subspecies fastidiosa; Helichrysum italicum 1 showing leuA_1 (known in subspecies fastidiosa) with petC_3/holC_3 known in subspecies multiplex; Myrtus communis 4 with leuA_3/holC_2 respectively known in subspecies multiplex and fastidiosa; Myrtus communis 8 and Platanus presenting form 1 alleles for five HKG mixed with malF_4 (all known in subspecies fastidiosa) and Q. ilex 10 presenting form 1 alleles for two HKG mixed with malF4); Olea europaea 2 with four multiplex alleles combined with nuoL_1 (subspecies fastidiosa); Olea europaea 5 with four pauca alleles combined with malF_15 (known in ST72 and ST76, subspecies fastidiosa). Two samples gave a double sequence for holC that were impossible to analyze (Table S5). Some sequences were ambiguous with superimposed peaks at some locations in otherwise good quality chromatograms revealing mixed infections. In those 12 samples, the number of potential combinations was too high to detect one probable allelic form, excepted for Prunus dulcis where the superimposed chromatograms corresponded to only two allelic forms (holC_3 or holC_6 which are found in subspecies multiplex). The holC_6 allelic form and the leuA_5 allele obtained for this sample are found in ST10, ST26, ST36, ST46, and ST63.

4. Discussion

A two-step nested procedure for MLST was developed to improve the typing of samples infected with low Xf population sizes that cannot be typed using the conventional protocol. In order not to affect the comparability of the results with the databases, the widely used MLST scheme developed for Xf that is supported by the pubMLST public website [22] was re-used.

The nested-MLST approach proved to be specific and efficient. No nonspecific amplifications were observed in any of the samples. Moreover, the sensitivities of the Harper’s qPCR detection test and the nested-MLST were similar with a LoD ranging from 10³ bacteria.mL⁻¹ to 10⁴ bacteria.mL⁻¹ These LoDs are similar to other nested-MLST approaches such as those developed for Burkholderia cepacia [30] but higher than for the one developed for Neisseria meningitidis (10 copies mL⁻¹) [31]. Consequently, in resource-limited settings where qPCR facilities are not available, the assay may be used as a useful diagnostic tool if applied with all necessary precautions to avoid cross-contamination between samples. The sequencing, which is costly, can be done as a consecutive but separate step to provide information on subspecies present in the sample. Higher bacterial loads (as indicated by lower Cq values) were observed in Spanish samples than in French samples, for which low amplification efficiency and partial profiles were observed. Full allelic profiles (ST6 and ST7 from multiplex subspecies) were obtained for Polygala myrtifolia, Spartium junceum and Genista corsica samples from France probably because they carried a higher bacterial load as shown by the low Cq obtained with the Harper’s qPCR test: five of the six typed samples had a Cq value between 23.4 and 26.5. The use of the nested-MLST assay to type plant Spanish samples allowed a higher number of successful complete typing (55 samples versus seven samples with the conventional approach). Spanish samples generally showed higher Xf titer (i.e, lower Cq values in Harper’s qPCR test) than the French samples but also concerned different plant species.

In our nested-MLST assay as well as in the original MLST assay, the amplification efficiencies were variable among genes, while all primers were designed using the same parameters from the software. For example, the holC gene for French samples tested with the nested-MLST assay was successfully amplified in 55% of samples collected in 2017 while the gltT and nuoL genes gave the lowest rates (around 26%). For samples collected in Spain tested with the original MLST assay, amplification rates among the seven HKGs ranged from10 to 67%. Success rate variations were also observed in medical research using MLST between samples and between loci [25]. When conducted on strains, no differences about amplification rates are observed because of DNA excess. Robustness of a PCR reaction is determined by appropriate primers and it is not always obvious why some primer combinations do not amplify well, even if some parameters such as DNA folding can interfere in PCR efficiency [32]. In this study, even if primer annealing temperature was adjusted, design of primers was limited by their arbitrary localization.

Typing results of French samples were concordant with previously published results [23] but also revealed the presence of alleles not yet described in France. It should be noticed that no unknown sequence was obtained, refraining from evoking contaminations as the origin of these yet undescribed alleles in France. Thanks to the high rate of amplification of holC in nested PCR, it was also possible to obtain sequences for equivocal samples (Cq with the Harper’s qPCR test above 35) to confirm the presence of the bacterium in these samples. Surprisingly, these amplifications led to alleles that correspond to subspecies other than the multiplex subspecies. Thereby, alleles from subspecies pauca (holC_10) and fastidiosa (holC_1, holC_2, holC_24) were sequenced. holC_10 was already reported in Polygala myrtifolia in the south of France in 2015 [23]. holC_1 finding is in agreement with [24], who also reported holC_1 in insects in Corsica. Up to now, no holC_2 was reported in France but it is known in the USA. holC_24 was also reported in Polygala myrtifolia in Corsica in 2015 [23]. Further plant sampling efforts are needed to confirm the establishment of those strains in the environment or to document further the dynamics of alleles revealing sporadic infections.

For French samples only, several samples could not be typed since the chromatograms showed an overlap of two peaks precisely on the polymorphic sites (mainly with leuA and holC genes). This has already been reported by [23], it suggests the simultaneous presence of several strains in the same sample since only one copy of these genes are known in Xf [22]. Moreover, the report of previously unknown combination of alleles belonging to different subspecies can also results from the presence of co-infection or of recombinants. Recombination events are reported in Xf [23,33,34,35] and could have led to host shift [36]. In this study, eight samples presented unknown combinations of alleles from the same or different subspecies which could be explained by intrasubspecies or intersubspecies recombination events. As reported in [37], such events may exist and occur but not with the same frequency. Moreover, natural competence can be variable among Xf strains [38]. These events could also reflect a mechanism of adaptation [39]. Five samples among these eight samples were collected in 2017 and three in 2018, and were different between years. In 2018 the three cases were a similar combination of alleles and were found in three different plants. Future surveys will be necessary to know if some of these recombinants strains are indeed present in Corsica or are the consequence of mixed infections and if they have adapted and survived on different hosts.

The objective of this study was to improve the published MLST scheme supported by a public website (https://pubmlst.org/xfastidiosa/) by designing nested primers to lower the limit of detection and help in Xf diagnosis and typing. Thus, this improved MLST assay enables a higher sensitivity and specific typing of Xf directly from plant and insects samples without the need of isolating the strain and at an affordable cost.