Phylogenetic and Variable-Number Tandem-Repeat Analyses Identify Nonpathogenic Xanthomonas arboricola Lineages Lacking the Canonical Type III Secretion System

Abstract

Xanthomonas arboricola is conventionally known as a taxon of plant-pathogenic bacteria that includes seven pathovars. This study showed that X. arboricola also encompasses nonpathogenic bacteria that cause no apparent disease symptoms on their hosts. The aim of this study was to assess the X. arboricola population structure associated with walnut, including nonpathogenic strains, in order to gain a better understanding of the role of nonpathogenic xanthomonads in walnut microbiota. A multilocus sequence analysis (MLSA) was performed on a collection of 100 X. arboricola strains, including 27 nonpathogenic strains isolated from walnut. Nonpathogenic strains grouped outside clusters defined by pathovars and formed separate genetic lineages. A multilocus variable-number tandem-repeat analysis (MLVA) conducted on a collection of X. arboricola strains isolated from walnut showed that nonpathogenic strains clustered separately from clonal complexes containing Xanthomonas arboricola pv. juglandis strains. Some nonpathogenic strains of X. arboricola did not contain the canonical type III secretion system (T3SS) and harbored only one to three type III effector (T3E) genes. In the nonpathogenic strains CFBP 7640 and CFBP 7653, neither T3SS genes nor any of the analyzed T3E genes were detected. This finding raises a question about the origin of nonpathogenic strains and the evolution of plant pathogenicity in X. arboricola. T3E genes that were not detected in any nonpathogenic isolates studied represent excellent candidates to be those responsible for pathogenicity in X. arboricola.

INTRODUCTION

Eubacteria constitute a major component of the commensal microbiota, and the nature of their interaction with plants is still unknown (1). Xanthomonas strains living in close association with plants but causing no apparent disease symptoms on their host have been reported (2, 3). On the basis of amplified fragment length polymorphism (AFLP) analysis, Gonzalez et al. (2) showed that nonpathogenic Xanthomonas strains colonizing cassava were clearly distinguished from Xanthomonas axonopodis pv. manihotis strains that cause cassava bacterial blight. Although some of these nonpathogenic strains have been characterized genetically and phenotypically, little is known about their epidemiological or ecological importance.

In previous studies, the genetic diversity and population structure of Xanthomonas have been investigated using DNA-DNA hybridization (4, 5), repetitive-sequence PCR (rep-PCR) (4–7), AFLP (4, 8), and fluorescent AFLP (9, 10). As an alternative to these methods, the comparative sequence analysis of protein-encoding genes has also been widely explored. For example, Parkinson et al. (11) used the gyrB gene, which encodes the subunit B protein of DNA gyrase, for establishing a phylogenetic relationship among 203 Xanthomonas pathotype strains. Young et al. (12, 13) used a multilocus sequence analysis (MLSA) based on four genes (dnaK, encoding the chaperone protein; fyuA, encoding one tonB-dependent transporter; gyrB and rpoD, encoding the RNA polymerase sigma factor) to study the phylogenetic and taxonomic relationships within the genus Xanthomonas. MLSA is a powerful technique for inferring phylogenetic relationships at the interspecific and intraspecific levels, as well as for evolutionary studies and systematics, and it can be useful in bacterial taxonomy as a complementary tool for defining species and for identification of new strains (14, 15). MLSA provides a robust method for the differentiation of most Xanthomonas spp. One of its most important contributions, applied to Xanthomonas, is that it allows strains to be allocated to known species or to be identified as members of new species more easily. Moreover, MLSA generally mimics grouping generated by DNA-DNA hybridization within Xanthomonas, AFLP, and rep-PCR and may therefore offer a refined method for differentiation of species (13). In some cases, MLSA is insufficient for discriminating closely related isolates and studying intraspecies genetic diversity. Thus, highly discriminative typing methods are needed for surveillance and outbreak studies. Multilocus variable-number tandem-repeat (VNTR) analysis (MLVA) has been successfully developed for many bacterial species (16–21). It is a bacterial typing method involving amplification and fragment size analysis of polymorphic regions of DNA containing variable numbers of tandem repeat sequences. VNTRs can be rapidly characterized by PCRs with specific primers based on the flanking regions of the tandem repeats. MLVA based on a few highly variable VNTRs usually displays a high level of discriminatory power in distinguishing closely related isolates for the investigation of disease outbreaks and epidemiological studies (16–21).

Vauterin et al. (3) suggested the investigation of hrp (hypersensitive reaction and pathogenicity) genes to distinguish the nonpathogenic Xanthomonas strains from the pathogenic ones. hrp genes are known to be involved in induction of hypersensitive response (HR) in resistant host and nonhost plants and pathogenicity in susceptible host plants (22, 23); hrc (hrp conserved) genes are considered to be critical for pathogenicity and initiation of disease and encode the type III secretion system (T3SS), a highly conserved protein secretion system (22, 24). Previous studies reported that the distribution of type III effectors (T3Es) within Xanthomonas strains may suggest a basic role in host specificity (25). T3Es are candidate determinants of host specificity of pathogenic bacteria, since it has been shown that many T3Es can act as molecular double agents that betray the pathogen to plant defenses in some interactions and suppress host defenses in others (26). More recently, Hajri et al. (27) investigated the variability of T3E repertoires in the species X. arboricola and their potential role in structuring its populations according to host range and confirmed that T3SS is an essential virulence mechanism in X. arboricola.

Walnut blight (WB), caused by Xanthomonas arboricola pv. juglandis (5), is a major disease of walnut in France and the most important one in all walnut-growing areas (28). Common symptoms include stem, fruit, and leaf spots, catkin necrosis, and fruit drop. Previous studies showed that X. arboricola pv. juglandis has been also isolated from tissues affected by brown apical necrosis (29–31). Hajri et al. (10) reported the association of X. arboricola pv. juglandis with vertical oozing canker (VOC) and clarified the taxonomic position of VOC strains as belonging to a singular lineage within X. arboricola pv. juglandis. During surveys in the two main production areas of walnut in France (Grenoble in the southeast and Périgord in the southwest), we noticed that nonpathogenic strains of X. arboricola were isolated from healthy and diseased walnuts. These strains were characterized by pathogenicity tests on walnut seedlings and a range of other plants. The main aim of the present study was to assess X. arboricola population structure associated with walnut, including nonpathogenic strains, in order to gain a better understanding of the role of nonpathogenic xanthomonads in walnut microflora. Knowledge pertaining to the population structure of X. arboricola isolated from walnut should shed light on the epidemiology of diseases associated with X. arboricola pv. juglandis, with the final aim of helping in development of reliable identification and specific detection tools that will facilitate ecological and epidemiological studies. Hence, the genetic structure of nonpathogenic strains was investigated using MLSA and MLVA approaches, and the type of interaction that this group of bacteria develops with host and nonhost plants was characterized. In this context, we investigated the distribution of T3Es and T3SS-coding genes in nonpathogenic strains.

MATERIALS AND METHODS

Bacterial strains.

The bacterial strains used in this study are listed in Table 1. Strains of X. arboricola were obtained from the International Center for Microbial Resources, French Collection for Plant Associated Bacteria (CIRM-CFBP), INRA, Angers, France (http://www.angers.inra.fr/cfbp/), or isolated from buds of healthy walnuts in the two main walnut-growing areas in France (Rhône-Alpes region in the southeast and Périgord in the southwest). Bacterial strains were routinely cultured at 27°C on YPGA medium (7 g liter⁻¹ yeast extract, 7 g liter⁻¹ peptone, 7 g liter⁻¹ glucose, 15 g liter⁻¹ agar) for 24 to 48 h.

TABLE 1.

Bacterial strains used in this study^a

Taxon	Strain	Other collection accession no. and/or strain no.	Host plant	Geographic origin	Yr of isolation
X. arboricola pv. celebensis	CFBP 3523PT	LMG 677, NCPPB 1832, ATCC 19045	Musa acuminata	New Zealand	1960
	CFBP 7150	LMG 676, NCPPB 1630, ICMP 1484	Musa acuminata	New Zealand	1960
X. arboricola pv. corylina	CFBP 1159PT	LMG 689, NCPPB 935, ATCC 19313	Corylus maxima	United States	1939
	CFBP1846		Corylus avellana	France	1975
	CFBP1847		C. avellana	Algeria	1977
	CFBP1848		C. avellana	United Kingdom	1977
	CFBP 2565	ICMP 11956	C. avellana	France	1985
	CFBP 5956		C. avellana	France	1979
	CFBP 6101		C. avellana	France	1979
Xanthomonas arboricola pv. fragariae	CFBP 6771PT	LMG 19145, PD 2780	Fragaria × ananassa	Italy	1993
	CFBP 6762	PD 2694	Fragaria × ananassa	Italy	NA
	CFBP 6763	PD 2697	Fragaria × ananassa	Italy	NA
	CFBP 6770	LMG 19144, PD 2696	Fragaria × ananassa	Italy	1994
	CFBP 6772	PD 2803	Fragaria × ananassa	Italy	NA
	CFBP 3548	LMG 19146, PD 3164	Fragaria sp.	France	1986
	CFBP 3549	PD 3160	Fragaria sp.	France	1986
X. arboricola pv. pruni	CFBP 3894PT	NCPPB 416, ATCC 19316, ICMP 51, CFBP 2535	Prunus salicina	New Zealand	1953
	CFBP 3893		Prunus persica	Italy	1989
	CFBP 3898		Prunus domestica	United States	1989
	CFBP 3900		P. persica	United States	1987
	CFBP 3901		Prunus armeniaca	United States	1987
	CFBP 3921		P. persica	Italy	1996
	CFBP 411	ATCC 10016, ICMP 12475	P. persica	United States	1963
	CFBP 5229		Prunus sp.	Argentina	1996
	CFBP 5529	NCPPB 1607	P. persica	Australia	1964
	CFBP 5530		Prunus persica	Italy	1989
	CFBP 5580		Prunus japonica	France	2000
	CFBP 5722		P. persica	Brazil	1991
	CFBP 5723		Prunus sp.	Uruguay	NA
	CFBP 5724		Prunus amygdalus	United States	NA
	CFBP 6653		P. persica	France	2000
	CFBP 7098		P. domestica	Spain	2002
	CFBP 7099		P.domestica	Spain	2003
	CFBP 7100		Prunus dulcis	Spain	2006
X. arboricola pv. populi	CFBP 3123PT		Populus × canadensis	Netherlands	1979
	CFBP 2113		Populus × interamericana	Netherlands	1980
	CFBP 2666		Populus × interamericana	France	1983
	CFBP 2669		Populus × canadensis	France	1987
	CFBP 2983		Populus × canadensis	Italy	1989
	CFBP 2985		Populus × interamericana	Belgium	1989
	CFBP 2986		Populus × interamericana	Belgium	1989
	CFBP 3004		Populus × interamericana	France	1989
	CFBP 3121		Salix alba	Netherlands	1980
	CFBP 3122	ICMP 9140	Salix alba	Netherlands	1980
	CFBP 3124	LMG 9713, ICMP 9367	Populus × generosa	New Zealand	1986
	CFBP 3338		Populus × interamericana	France	1991
	CFBP 3342		Salix sp.	New Zealand	1988
	CFBP 3343		Populus sp.	New Zealand	1988
	CFBP 3344		Salix sp.	New Zealand	1988
	CFBP 3839		Populus deltoides	Belgium	1984
X. arboricola pv. juglandis	CFBP 2528T	LMG 747, NCPPB 411, ATCC 49083, ICMP 35	Juglans regia	New Zealand	1956
	CFBP 176		J. regia	France	1961
	CFBP 878		J. regia	France	1966
	CFBP 2564	ICMP 11955	J. regia	Italy	1985
	CFBP 2568		J. regia	Italy	1985
	CFBP 2632	ICMP 11963	J. regia	Spain	1984
	CFBP 6557		J. regia	Italy	1999
	CFBP 7071		Juglans sp.	Spain	1993
	CFBP 7072		Juglans sp.	Spain	1993
	CFBP 7244		J. regia	France	1978
	12572		J. regia	France	2001
	12573		J. regia	France	2001
	12575		J. regia	France	2001
	12576		J. regia	France	2001
	12577		J. regia	France	2001
	12579		J. regia	France	2001
	12581		J. regia	France	2001
	12586		J. regia	France	2001
	12707		J. regia cv. Fernor	France	2002
	12710		J. regia cv. Franquette	France	2002
	12680		J. regia	France	2002
	12783		J. regia	France	2003
	12763*	CFBP 7179	J. regia cv. Fernor	France	2002
	12574*		J. regia	France	2001
	12578*		J. regia	France	2001
	12580*		J. regia	France	2001
	12582*		J. regia	France	2001
	12583*		J. regia	France	2001
	12584*		J. regia	France	2001
	12585*		J. regia	France	2001
	12587*		J. regia	France	2001
	12588*		J. regia	France	2001
	12589*		J. regia	France	2001
	12591*		J. regia	France	2001
	12592*		J. regia	France	2001
	12681*		J. regia	France	2002
	12708*		J. regia cv. Fernor	France	2002
	12709*		J. regia cv. Fernor	France	2002
	12711*		J. regia cv. Fernor	France	2002
	12712*		J. regia cv. lara	France	2002
	12713*		J. regia cv. Fernor	France	2002
	12714*		J. regia cv. Fernor	France	2002
	12715*		J. regia cv. Fernor	France	2002
	12762*		J. regia cv. Fernor	France	2002
	12765*		J. regia cv. Fernor	France	2003
	12766*		J. regia	France	2003
	12768*		J. regia	France	2003
	12769*		J. regia	France	2003
	12770*		J. regia cv. Fernor	France	2003
	12772*		J. regia cv. Fernor	France	2003
	12774*		J. regia	France	2003
	12775*		J. regia	France	2003
	12776*		J. regia	France	2003
	12777*		J. regia	France	2003
	12778*		J. regia	France	2003
	12779*		J. regia	France	2003
	12780*		J. regia cv. Fernor	France	2003
	12781*		J. regia	France	2003
	12782*		J. regia	France	2003
	12784*		J. regia cv. Vina	France	2003
	12785*		J. regia cv. Franquette	France	2003
	CFBP 7643		J. regia	France	2009
X. arboricola	CFBP 1022		J. regia	France	1967
	CFBP 7654	CB1	J. regia	Périgord, France	2008
	CFBP 7653	CS2	J. regia	Périgord, France	2008
	CFBP 7651	CS5F	J. regia	Isère, France	2008
	CFBP 7652	SPS1	J. regia	Périgord, France	2008
	CFBP 7647	P2-4	J. regia cv. Franquette	Isère, France	2009
	CFBP 7641	P2-7	J. regia cv. Franquette	Isère, France	2009
	CFBP 7635	P2-21	J. regia cv. Franquette	Isère, France	2009
	CFBP 7637	P3-6	J. regia cv. Franquette	Isère, France	2009
	CFBP 7638	P3-24	J. regia cv. Franquette	Isère, France	2009
	CFBP 7645	P7-4	J. regia	Loire, France	2009
	CFBP 7640	P7-18	J. regia	Loire, France	2009
	CFBP 7636	P7-27	J. regia	Loire, France	2009
	CFBP 7633	P8-6	J. regia	Rhône, France	2009
	CFBP 7634	P8-10	J. regia	Rhône, France	2009
	CFBP 7639	P8-14	J. regia	Rhône, France	2009
	CFBP 7632	P8-15	J. regia	Rhône, France	2009
	CFBP 7629	P8-20	J. regia	Rhône, France	2009
	CFBP 7630	P9-12	J. regia	Isère, France	2009
	CFBP 7656	P9-21	J. regia	Isère, France	2009
	CFBP 7646	P10-8	J. regia	Rhône, France	2009
	CFBP 7644	P10-14	J. regia	Rhône, France	2009
	CFBP 7650	P10-19	J. regia	Rhône, France	2009
	CFBP 7649	P10-25	J. regia	Rhône, France	2009
	CFBP 7648	P11-12	J. regia	Isère, France	2009
	CFBP 7631	P11-21	J. regia	Isère, France	2009
	CFBP 7655	P16-11	J. regia	Isère, France	2009
X. campestris pv. campestris	CFBP 5241		Brassica oleracea	United Kingdom	1957

^aCIRM/CFBP, Collection Française de Bactéries associées aux Plantes, INRA, Angers, France; ICMP, International Collection of Microorganisms from Plants, Auckland, New Zealand; LMG, BCCM/LMG Bacteria Collection, University of Ghent, Ghent, Belgium; NCPPB, National Collection of Plant Pathogenic Bacteria, York, United Kingdom; ATCC, American Type Culture Collection, Manassas, VA; PD, Culture Collection of Plant Pathogenic Bacteria, Plant Protection Service, Wageningen, Netherlands. Superscript T and PT indicate type strain of a species and pathotype strain of a pathovar respectively. *, Xanthomonas arboricola pv. juglandis strain isolated from vertical oozing canker symptoms on trunks and branches (10). NA, not available.

Plant material.

Seedlings of walnut (Juglans regia cv. Fernor and cv. Franquette), peach (Prunus persica cv. Dixired), radish (Rhaphanus sativus var. Kocto), tomato (Lycopersicon esculentum cv. marmande), and pepper (Capsicum annum cv. ECW) were used for pathogenicity tests. A hypersensitivity reaction was induced on leaves of Nicotiana benthamiana. Plants were grown in a greenhouse under 18°C at night and 24°C during day with a 12-h photoperiod. For negative controls, plants were inoculated with sterile distilled water. For positive controls, plants were inoculated with X. arboricola pv. juglandis strains CFBP 2528^PT and CFBP 7179 for walnut, X. arboricola pv. pruni strain CFBP 3894^PT for peach, Xanthomonas vesicatoria CFBP 1941 for tomato, Xanthomonas euvesicatoria CFBP 5618 for pepper, and X. campestris pv. campestris strain CFBP 5241 for tests on R. sativus and N. benthamiana plants.

Pathogenicity tests.

Walnut seedlings were grown in a greenhouse until four to six young leaves were evident. Bacterial suspensions (1 × 10⁸ CFU ml⁻¹) were sprayed onto the foliage, and plants were maintained for 2 days under plastic bags and incubated in growth chambers. The plastic bags were then removed, and the plants were maintained in the growth chamber under the same climatic conditions. Plants were checked for symptoms weekly for up to 30 days after inoculation.

Two-year-old peach seedlings were planted in 30-cm-diameter pots. Young leaves (third to sixth leaves from shoot tip) were detached from peach seedlings, collected, and inoculated using detached-leaf assays as described by Randhawa and Civerolo (32). Detached leaves were disinfected for 40 to 60 s with 70% ethanol and then rinsed in sterile water. These leaves were then immersed in a bacterial suspension (1 × 10⁷ CFU ml⁻¹), and a 10-kPa vacuum was applied for 2 min. Inoculated leaves were placed in a sterile tube with the leaf upright, and the petiole was immersed in 6% water agar. Symptom development was recorded daily for 3 weeks after inoculation. For a positive result, after 6 to 9 days, all inoculated sites had to exhibit confluent water soaking, become dark brown, and exhibit brittle necrotic spots, often surrounded by a greyish white or purple margin.

On radish, two leaves per plant at the stage of four fully expanded leaves were inoculated by the leaf-clipping method. The last completely expanded leaf was cut with scissors dipped in bacterial suspensions (1 × 10⁸ CFU ml⁻¹). Ten leaves were inoculated for each strain. A positive result was obtained if V-shaped lesions appeared at the leaf margin approximately 2 weeks after inoculation.

Leaves of tomato and pepper were punctured at four locations with a sterile needle and 1 ml of bacterial suspension (1 × 10⁶ CFU ml⁻¹) was infiltrated through wounds. A positive reaction was obtained when inoculated leaves exhibited dark brown irregularly shaped splotches with chlorosis surrounding the lesions, while inoculated leaves of pepper had small, yellow-green lesions that became deformed and twisted.

Leaves of N. benthamiana were punctured at four locations with a sterile needle, and 1 ml of bacterial suspension (1 × 10⁶ CFU ml⁻¹) was infiltrated through wounds. Necrosis of the infiltrated area after 24 to 48 h was considered a hypersensitive response (HR).

The disease occurrence was monitored by quantification of symptoms and bacterial populations at 2, 7, 14, and 21 days after inoculation into plant tissues. Inoculated plants were maintained in growth chambers under 18°C at night and 20°C during the day with a 15-h photoperiod (light intensity of 85 μE m⁻² s⁻¹) and a high relative humidity.

MLSA.

Gene fragments of atpD, dnaK, efp, fyuA, glnA, gyrB, and rpoD were amplified from genomic DNA of the 27 nonpathogenic strains by using primers described by Fargier et al. (14). A list of genes and primer sequences used for PCR amplification and sequencing is provided in Table 2. PCR amplifications were carried out in a total volume of 25 μl containing 1× GoTaq buffer (Promega, Fitchburg, WI, USA), a 200 μM concentration of each deoxynucleoside triphosphate (dNTP), a 0.5 μM concentration of each primer, 0.4 U of GoTaq polymerase (16 U ml⁻¹ final concentration), and 5 μl of boiled bacterial cells (3 × 10⁸ CFU ml⁻¹). The PCR cycling conditions consisted of an initial denaturation step at 94°C for 5 min, followed by 35 cycles of denaturation at 94°C for 30 s, annealing at 60°C for all loci (except 62°C for efp) for 60 s, and extension at 72°C for 30 s, and a final extension step at 72°C for 7 min.

TABLE 2.

Primers used in MLSA for amplification and sequencing

Gene	Primer sequence		PCR fragment size (bp)
	Forward	Reverse
atpD	GGGCAAGATCGTTCAGAT	GCTCTTGGTCGAGGTGAT	750
dnaK	GGTATTGACCTCGGCACCAC	ACCTTCGGCATACGGGTCT	759
efp	TCATCACCGAGACCGAATA	TCCTGGTTGACGAACAGC	339
fyuA	ACCATCGACATGGACTGGACC	GTCGCCGAACAGGTTCACC	753
glnA	ATCAAGGACAACAAGGTCG	GCGGTGAAGGTCAGGTAG	675
gyrB	ACGAGTACAACCCGGACAA	CCCATCARGGTGCTGAAGAT	735
rpoD	ATGGCCAACGAACGTCCTGC	AACTTGTAACCGCGACGGTATTCG	609

PCR amplicons sequencing was performed by the Biogenouest platform (Nantes, France). Nucleotide sequences were corrected using Geneious v. 4.8.4 (33) and edited using BioEdit (34). These sequences were aligned together with sequences of 73 representative strains of Xanthomonas arboricola available from the CIRM-CFBP sequence data (http://www.angers.inra.fr/cfbp/) using Clustal W (35). A neighbor-joining tree was generated with MEGA v. 5.0.3 (36) using the Kimura two-parameter model (37) and 1,000 bootstrap replicates. X. campestris pv. campestris strain CFBP 5241 was used as an outgroup.

MLVA.

Seventeen loci of 6- to 15-bp tandem repeat (TR) units previously constructed (38) were used in the multilocus VNTR analysis (MLVA) scheme (Table 3). PCR and capillary electrophoresis were conducted as described in reference 38. Output data from capillary electrophoreses were managed with BioNumerics v.6.5 (Applied Maths, St-Martens-Latem, Belgium), and chromatograms were also checked with Peakscanner software v. 1.0 (Life Technologies). The allele scores based on the fragment sizes were converted into repeat numbers and used as character data for cluster analysis. A minimum spanning tree (MST) was generated using BioNumerics v. 6.5 (Applied Maths, St-Martens-Latem, Belgium) using the categorical coefficient and the maximum number of single-locus variants (SLVs) as a priority rule (39).

TABLE 3.

Primers used for amplification of VNTR loci and their corresponding positions on the genome of X. arboricola pv. pruni CFBP 5530

Multiplex	VNTR locus	Primer sequence		Position
		Forwarda	Reverse	Start	End
A	TR50I	V-CGTGCATCAGACGCTTGCGT	GTTGCGAGATCGGGCGCTTC	3413992	3414042
	TR33I	P-CTCGCAAAACCCTTGCCATC	CGAGTGGATGTTATGGCGTGG	2022442	2022490
	TR68I	F-AAATCATCGGCGCCTGAAAC	CTTGCGGTACTGGCTGTTCA	4140738	4140827
B	TR19I	N-GATTGACGGCACCCACACAG	CCAGGACGTTGTGCCGTGGT	1523902	1523949
	TR36I	P-CGATCGCATCTGTGTGGGTTAG	GCAGGAGAAGGAAAGCGCCAG	2284172	2284214
	TR58I	V-ACCAACACCGAGCTTGCCTC	ATCTGTTGCTGGCCGAGAGC	3538491	3538528
C	TR3I	N-GGTTGCTTGGTCGTTGATCG	GACATTCGCCGGGAGTGCAG	150355	150401
	TR40I	P-TGGAATGTGGAGGCTGTTCG	TATCAGGCAGCGCACCAGCT	2426642	2426699
D	TR15I	N-TCGAGCGGTTCCTGCGGTTGT	GCCATGTCGCCGGGAAACGA	1310624	1310662
	TR37I	P-CCAACAGAACCCCGCACCCA	ATGGAGGATGCGGTTGCGGCT	2348164	2348197
E	TR05II	F-CAGATGCTGTCCCGATTCCCG	GTCGACGGGTTCGCGGAAGGT	340287	340353
	TR39II	P-GGTACGGAAGGTGGTGGTCTGC	CCCGCATACTGATGCAGTTCG	1940044	1940160
	TR06II	F-GTGCAGCACCAGCCAAAGGCA	TCATAGGCTGGGGATTGGGGA	340223	340303
F	TR21II	V-ACACGGACGTACTTGCGGCGT	GGAGCGTATTGCTTGAACGGGA	1114440	1114597
	TR58II	P-TCTGATCGGTGCTGAGCGTCT	GGAAGAGTACCCGGCAATTCT	3391120	3391216
G	TR38II	N-CCCGTAGCTGTATCAGTGCCT	TCTCGGTATCGATGTGGGTGC	1930537	1930595
	TR67II	P-AGCTCGCAACTGCTTTTCCCGA	GATACAAGGCGAACGCGATGA	3641116	3641204

^aForward primer pairs were marked with one of the following fluorescent dyes: F, FAM; V, VIC; N, NED; or P, PET.

Amplification of T3E and T3SS genes.

Nonpathogenic strains used in this study were tested for the presence or absence of hrp genes by PCR amplification of genomic DNA using primers described by Hajri et al. (27) (Tables 4 and 5). For detection of T3E genes, PCRs were carried out in a total volume of 20 μl containing 1× GoTaq buffer (Promega), a 200 μM concentration of each dNTP, a 0.5 μM concentration of each primers, 0.4 U GoTaq polymerase (16 U ml⁻¹ final concentration), and 5 μl of boiled bacterial cells (3 × 10⁸ CFU ml⁻¹). All PCRs were performed with the following cycling conditions: initial denaturation step at 94°C for 2 min, 30 cycles of denaturation at 94°C for 1 min, annealing at 60°C for 1 min, and extension at 72°C for 2 min, and a final extension step at 72°C for 10 min.

TABLE 4.

PCR primers used to amplify T3Es genes

T3E gene	Forward primer	Reverse primer	PCR fragment Size (bp)
avrBs2	ACCGCGCTGGCCACACCG	TCACTCGCCCGGCTCGATC	2,118
	TGCCGGTGTTGATGCACGA	TCGGTCAGCAGGCTTTC	850
xopF1	TGAAACTCACCAGCAATATCG	CTAGCGAAGCGCCTCGCTC	1,996
	AGGCCATCGACCCCAAGATCC	GTTCTTGGCCTTGAGCGCATTCC	779
xopA	ATGGACTCATCTATCGGAAACTT	GCCGGTGATGCTCGACAG	381
	TGCAGACGATGGGCATCG	CTGCATCAGCTGCATCACGATC	239
hrpW	ATGCAACGCATGCTCAGCGACAT	GTCTTCAGGTTCGCCAGCTTCAC	905
	AAGGTCGTCACCGCGC	GTCCTGCACGACCTTGTCT	399
hpaA	ATGATCCGGCGCATTTCGCCAG	TCATGCACGAATCTCCTGAGCGGC	816
	CGCTGGATGGCATGGACGACG	CGTCTGAGCGTCTGGTCGGCGGC	292
xopR	ATGCGCCTGAGTCAGTTGTTT	GTAGCCGTTGTCGATTGCCTCTT	1,230
	CGTGCGGCCCTGATCGC	GTAGCCCTGCATCATGCGTT	303
xopN	ATGAAGTCATCCGCATCCGTCGAT	CTCGATCGGTTCGGGCTACTCG	2,092
	GTCATGACCCAGGGCGC	GGTGATGGCGGTGTGCTG	864
xopX	ATGGAGATCAAGAAACAGCAAACCGC	GGCGACAGGCTTTGCACATATCTGG	1,865
	GTGGAAAACAACCTGGG	CCCCAGTTCATCGCC	827
xopZ	GCACTTGCGGATACTAATGCGG	GTCGACGAAGTCCTGCAATTGG	2,868
	TTCGGCCGCGGCTCGGC	GCACGGCATGGCGCGCTCC	1,012
xopQ	GTGCCCGCAGGCGCTCATGCAA	CCTTGGCGTGAACAGCATGCC	1,224
	ACCCCGACGATGT	TTGTTGTAGGCGCG	484
xopK	GACGCCCTTGCTTCAGCGAAC	TTCGGTGGCCAGCAACGTGCC	2,454
	CTCGGCATCCAGGGC	GACAAAGCCCTTGTTCCA	357
xopV	ATGAAAGTCTCCGCAACCCTT	TCAGGTTGCGAAAGGTGAGG	1,023
	ACACGCCTGTTCGTCTC	GCGATGTTCCATTTGTA	236
xopL	ATGCGACGCGTCGATCAACCG	CTACTGATGGCCTGAGGGTTCCG	1,863
	CCACCGACCGTGGGCGCTTCATCATTA	ACATCTGCACTGCCTTGGCCAGC	1,324
xopAI	ATGACTTCGGTAAGCCAGCGCGAATC	TCGATCTGGCTTTGATAAATCCTCAGAC	950
	AGAGCAGACCACGCCCTCTACG	GAATATTCTTCGGGAAGCGAGTGC	507
avrXccA1	GTGGTTCGCTGCGATGGC	TCACCCAGCCAGCGGG	813
	GATGGGCGGCACCG	ATCGCCACGCACCTG	163
avrXccA2	GCCGATGGCTGCCGCCGGCGCTA	TTGGTGTTCCAGTTCCGATCCAGG	1,442
	ACGGCCCGTTCTTTCCGCAAAGCC	CAACGGGCGCTCCGGCGACG	371
xopAH	ATGAAGAACACGTCTGTCCCT	CTACTTCTGCGTGGGAGGC	1,002
	ATTGTGGTATGGGCCTAGGC	TGCTTGGCGTACTCGTAGAAT	220
xopB	ATGAAGGCAGAGCTCACACGA	TCAGGCGCGGGTTGGTGCGAAGTA	1,835
	AGCATTTGGCCCAAGCGCTTT	CGCTTCGGTTGTCGTCATATTGG	574

TABLE 5.

PCR primers used to amplify T3SS genes

Gene	Forward primer	Reverse primer	PCR fragment Size (bp)
hrcR	GCTGGTGGTCATCATGCTGG	GTGTTTGAGGAGGAATTGC	292
hrcN	ATGTCAACGTGATCGTGC	CTGGCTCATCACCCGGCTC	524
hrcT	GTCGTTCTACGCGCTGG	GTTGGCGGCATCGTGCAA	376
hrcC	ACCGAAGTGCAGGTGTTTC	ATCTCGATGATGGTGGCATCGAT	575
hrcV	GCGCCATGAAATTCGTCAAGG	GCCAGCAGCAGGAACAGC	367
hrcU	GGCGTGGTGCTGTGG	GGTTGACCACCATCACCTTG	340
hrcJ	CTCGGCGAGATGTTCAAG	GCCACCAATACAGCGC	436
hrpB1	CTGATCACGGTCGG	TCGGCATCGGCGTC	287
hrpF	ACGCTGGACACCATC	TTCTTGTAGCCGGTGAT	188

Nucleotide sequence accession numbers.

The nucleotide sequences obtained in this work were deposited in GenBank under accession numbers KF904342 to KF904442 for atpD, KF904443 to KF904543 for dnaK, KF904544 to KF904644 for efp, KF904645 to KF904745 for fyuA, KF904746 to KF904846 for glnA, KF904847 to KF904947 for gyrB, and KF904948 to KF905048 for rpoD.

RESULTS

Identification of nonpathogenic strains of X. arboricola on walnut.

Xanthomonas-like strains isolated from walnut buds were characterized by pathogenicity tests on walnut seedlings and on a range of other plants. Strains were identified as X. arboricola based upon phenotypic characteristics and biochemical tests as described by Schaad (40). and amplification of a specific PCR fragment using the PCR test with XarbQ-F and XarbQ-R primers developed by Pothier et al. (41). A set of 27 strains formed convex, yellow-pigmented colonies, which were characterized as Gram-negative rods able to perform oxidative metabolism of glucose, galactose, mannose, cellobiose, trehalose, and arabinose and hydrolysis of gelatin, esculin, starch, and Tween 20 (except for strains CFBP 7645 and CFBP 7636, which did not hydrolyze Tween 20). A single amplicon with the correct size of 432 bp was obtained for each of these 27 strains using the PCR test (41). None of the 27 strains induced symptoms on walnut (Juglans regia), the plant from which they were isolated, after inoculation on walnut seedlings. Dynamics of the bacterial population sizes revealed that these strains were not able to reach bacterial population sizes higher than 1 × 10⁶ CFU per g leaf tissue, the typical level observed for pathogenic interactions on walnut, whereas X. arboricola pv. juglandis CFBP 7179 induced typical necrotic leaf spots (Fig. 1) and reached about 1 × 10⁷ CFU per g 7 days after inoculation (see Fig. S1 in the supplemental material).

FIG 1.

Illustration of walnut seedling responses to inoculations of X. arboricola strains CFBP 7634, CFBP 7645, CFBP 7653, and CFBP 7179. Observations were made 14 days after spraying bacterial suspension on walnut leaves. No reaction was induced on walnut leaves inoculated with X. arboricola strains CFBP 7634, CFBP 7645, and CFBP 7653 (A, B, and C, respectively). Dark brown spots and necrosis were observed on walnut leaves inoculated with X. arboricola pv. juglandis strain CFBP 7179 (D).

The pathogenicity of five strains representing each clonal group identified by MLSA and MLVA (CFBP 7634, CFBP 7645, CFBP 7651, CFBP 7652, and CFBP 7653) was evaluated on plants known to be hosts of other X. arboricola pathovars: P. persica, R. sativus, L. esculentum, and C. annum. While positive-control strains developed characteristic symptoms on their respective hosts, none of the five X. arboricola strains isolated from walnut and nonpathogenic on walnut (CFBP 7634, CFBP 7645, CFBP 7651, CFBP 7652, and CFBP 7653) induced any disease symptom on these plants (Fig. 2). Cell death in the inoculated area was observed on tomato and pepper plants following inoculation with strains CFBP 7651 and CFBP 7652 (Fig. 3). The dynamics of bacterial population sizes showed that the five strains were not able to reach population sizes equal to 1 × 10⁸ CFU per g, the typical level of pathogenic interactions on the tested plants (see Fig. S1 in the supplemental material).

FIG 2.

Plant reaction observed on peach (A and B) and radish (C and D) leaves 14 days after inoculation. Compatible interactions cause typical symptoms when X. arboricola pv. pruni strain CFBP 3894^PT was inoculated onto peach leaves (B) and when X. campestris pv. campestris CFBP 5241 was inoculated onto radish leaves (D). No reaction was observed when peach leaves were inoculated with X. arboricola strain CFBP 7634 (A) and when radish leaves were inoculated with X. arboricola strain CFBP 7652 (C).

FIG 3.

Tomato and pepper reactions to inoculation with Xanthomonas strains, 7 days after inoculation. No reaction is induced on tomato and pepper leaves inoculated with X. arboricola strain CFBP 7634 isolated from healthy walnut (A and D). Hypersensitive reactions were observed on tomato and pepper leaves inoculated with X. arboricola strain CFBP 7651 (B and E). Dark brown, irregularly shaped splotches with chlorosis surrounding lesions were observed on tomato leaves inoculated with Xanthomonas vesicatoria strain CFBP 1941 (C). Brown lesions were observed on pepper leaves inoculated with X. axonopodis pv. vesicatoria strain CFBP 5618 (F).

The ability of these strains to induce cell death was also tested on N. benthamiana. CFBP 7651 and CFBP 7652 cause cell death on N. benthamiana within 48 to 72 h after inoculation, while CFBP 7634, CFBP 7653, and CFBP 7645 did not elicit any plant reaction on inoculated N. benthamiana leaves (Fig. 4).

FIG 4.

Plant reactions when Nicotiana benthamiana leaves were inoculated with X. arboricola strain CFBP 7634 (A), X. arboricola strain CFBP 7651 (B), and X. campestris pv. campestris strain CFBP 5241 (C). Images were taken 72 h after bacterial infiltration.

MLSA confirmed that nonpathogenic strains isolated from walnut belong to X. arboricola.

Partial sequences of atpD, dnaK, efp, fyuA, glnA, gyrB, and rpoD genes were used in the present study to investigate the phylogenetic relationships between pathogenic and nonpathogenic strains of X. arboricola isolated from walnut. A phylogenetic tree was constructed based on the neighbor-joining method using the concatenated nucleotide sequences of the seven gene fragments of 27 nonpathogenic strains and 73 representative strains of X. arboricola. The sizes of the seven gene fragments were 750 bp (atpD), 759 bp (dnaK), 339 bp (efp), 753 bp (fyuA), 675 bp (glnA), 735 bp (gyrB), and 609 bp (rpoD), leading to a total of 4,620 bp for the concatenated data set. Except for the strain CFBP 7653, all strains of X. arboricola used in the MLSA scheme clustered in a large clade separately from X. campestris pv. campestris, used as an outgroup in the phylogenetic analysis (Fig. 5). Phylogenetic analyses distinguished several groups with high bootstrap values corresponding to different pathovars of X. arboricola. This clear correspondence between phylogenetic clustering and pathovar classification was not supported by phylogenetic trees based on individual loci (see Fig. S2 in the supplemental material). These observed incongruences might be explained by recombination events that shuffle the phylogenetic signal and by the fact that each individual locus does not harbor enough phylogenetic information. Thus, the addition of nonpathogenic strains in the phylogenetic tree does not modify the phylogenetic relationships between pathovars of X. arboricola (pathovars pruni, corylina, fragariae, populi, celebensis, and juglandis). Based on the phylogenetic position of nonpathogenic strains in the neighbor-joining tree, we confirmed that nonpathogenic strains definitely belonged to X. arboricola, grouped outside clusters defined by pathovars, and formed separate genetic lineages. Nonpathogenic strains isolated from walnut were polymorphic as they were distributed into three separated clusters within X. arboricola (Fig. 5). Thus, three main groups, termed NP1, NP2, and NP3, and four single branches were revealed by MLSA. The high bootstrap values of nonpathogenic strains depicted the robustness of these lineages; NP2 and NP3 with a bootstrap value equal to 100%, and NP1 with a bootstrap value equal to 98%. For NP3, CFBP 7640 presents differences in nucleotide sequences of the seven genes used in the phylogenetic analysis, whereas CFBP 7636 and CFBP 7645 are identical. The remaining strains, CFBP 7630, CFBP 1022, and CFBP 7652, did not cluster with either NP1 or NP2 with high bootstrap support.

FIG 5.

Phylogenetic tree constructed based on the neighbor-joining method using concatenated nucleotide sequences from 100 X. arboricola strains of partial regions of the atpD, dnaK, efp, fyuA, glnA, gyrB, and rpoD genes, representing a total of 4,620 bp. Bootstrap support as a percentage was based on 1,000 replicates. X. campestris pv. campestris CFBP 5241 was used as the outgroup.

MLVA distinguished several clonal complexes within X. arboricola strains isolated from walnut.

Ninety-three X. arboricola strains isolated from walnut, including 27 nonpathogenic strains, were typed using multilocus variable-number tandem-repeat analysis (MLVA). A minimum spanning tree (MST) was constructed using the highest number of single-locus variants (SLVs) as the priority. The minimum spanning tree is an undirected network in which all the samples within the population studied are linked together with the smallest possible linkages between nearest neighbors. MLVA types were distinguished to define clonal complexes that grouped strains that differ from each other by at most three locus variants (Fig. 6). Strains causing VOC grouped separately from WB strains and from nonpathogenic strains isolated from walnut. VOC-causing strains were grouped into one clonal complex and two singletons, whereas WB strains were more heterogeneous; they were separated into two clonal complexes and 16 singletons. Nonpathogenic strains were grouped separately from clonal complexes containing the WB- and VOC-causing strains and confirmed to be heterogeneous, as they were divided into six clonal complexes and four singletons. Nonpathogenic strains belonging to NP1 clustered in two clonal complexes and one singleton (CFBP 7638), while NP2 strains clustered into three clonal complexes and two singletons, NP3 strains, were divided into one clonal complex and one singleton (Fig. 6).

FIG 6.

Minimum spanning tree of 93 X. arboricola pv. juglandis and nonpathogenic strains based on 17 VNTR loci. Each circle represents an MLVA type with a size corresponding to the number of strains that share an identical MLVA type. MLVA types connected by a thick solid line differ by at most one VNTR locus, while MLVA types connected by a thin solid line differ by two or three VNTR loci. MLVA types that differ from each other by four VNTR loci or more are connected by dashed and dotted lines. A maximum distance between nodes in the same partition is 3. Branch distances above 5 are in blue. Clonal complexes defined in the gray zone, grouped strains that differed from each other's by at most three locus variants. NP, nonpathogenic; WB, walnut blight; VOC, vertical oozing canker.

In order to better understand the correlation between the population structure of nonpathogenic strains and their geographical origin, we performed an MST on the 27 nonpathogenic strains. Some clonal complexes grouped only strains belonging to the same geographic origin, i.e., isolated from the same field, such as orchard P10, located at Saint-Romains (CFBP 7644, CFBP 7646, CFBP 7649, and CFBP 7650), or orchard P8, located at Laval (CFBP 7629, CFBP 7632, CFBP 7633, and CFBP 7634). However, other clonal complexes grouped strains isolated in different geographic locations, such as clonal complexes grouping CFBP 7631, CFBP 7637, CFBP 7648 and CFBP 7656 (Fig. 7). These clonal complexes grouped strains belonging to different geographic locations from both eastern and southwestern areas, and other nonpathogenic strains belonging to different clonal complexes populations were recovered in the same orchards. Thus, the nonpathogenic strains were not structured according to their geographical origins based on this MLVA.

FIG 7.

Minimum spanning tree of the 27 nonpathogenic X. arboricola strains isolated from walnut, based on 17 VNTR loci. MLVA types connected by a thick solid line differ by at most one VNTR locus, while MLVA types connected by a thin solid line differ by two or three VNTR loci. MLVA types that differ from each other by four VNTR loci or more are connected by dashed and dotted lines. Clonal complexes defined in the gray zones are grouped strains that differed from each other's by at most three locus variants.

Several nonpathogenic strains lack the hrp-hrc cluster, coding for the type III secretion system.

Most nonpathogenic strains isolated from walnut were unable to elicit an HR on N. benthamiana and to cause disease on walnut and on other plant species tested. These observations suggest that these strains may lack the T3SS. We monitored the distribution of 9 genes coding the highly conserved genes of this secretion apparatus (27). Based on PCR results, no T3SS was detected in NP1 strains, NP3 strains, or CFBP 7653, since the 9 hrp-hrc gene primer pairs failed to amplify any DNA fragment for these strains. However, the NP2 strains CFBP 1022, CFBP 7630, and CFBP 7652 harbored genes of a typical T3SS of the Hrp2 family based on our PCR results (Fig. 8).

FIG 8.

Distribution of 18 T3Es and nine genes (two being hrp and seven being hrc) coding for the structural and regulatory components of the T3SS of the Hrp2 family among 27 nonpathogenic strains isolated from walnut, in comparison to X. arboricola pv. juglandis CFBP 2528^T and CFBP 7179. The presence or absence of an orthologue of each selected gene was determined by PCR. Black squares represent presence of the corresponding gene, whereas white squares represent absence of the gene.

The composition of T3E repertoires differs between nonpathogenic strains.

In this study, we investigated the distribution of 18 T3Es present in X. arboricola pv. juglandis (27). Many differences between nonpathogenic strains in the size and composition of their T3E repertoires were noticed. Strains that belonged to NP2 and contain T3SS have more effector-encoding genes (seven in total) than strains of NP1 that do not contain T3SS, which harbor only three effectors (xopR, avrBs2, and avrXccA1). In addition, strains belonging to NP2 and CFBP 7630, CFBP 1022, and CFBP 7652 showed a homogeneous pattern of T3Es genes. All these strains harbored seven T3E-encoding genes, six of them being orthologues of avrBs2, xopF1, xopA, hrpW, hpaA, and xopR (Fig. 8). These genes are considered to be the ubiquitous set of T3E genes for X. arboricola strains (27). The XopB-encoding gene, which was detected previously only in VOC strains, and the XopAH-encoding gene, which is present only in WB isolates of X. arboricola pv. juglandis, were not detected in any of the nonpathogenic strains tested. Two strains from NP3 (CFBP 7636 and CFBP 7645) have only the XopR-encoding gene, while two strains (CFBP 7640 from NP3 and the singleton CFBP 7653) do not have any of the 18 T3E-encoding genes tested. A total of 11 of the 18 T3E genes studied were not detected in any nonpathogenic isolates studied.

DISCUSSION

X. arboricola is conventionally known as a taxon of plant-pathogenic bacteria that includes including seven pathovars. However, this study showed that X. arboricola also encompasses nonpathogenic bacteria that do not cause disease on the plants from which they were isolated or on a panel of plants representative of species usually used for plant-bacterium interaction studies. Knowledge pertaining to the population structure of X. arboricola isolated from walnut should shed light on the epidemiology of diseases associated with X. arboricola pv. juglandis, the evolutionary mechanism of this pathogen, the taxonomy and ecology of nonpathogenic xanthomonads and possible ways of increasing the effectiveness of detection and management of plant diseases associated with Xanthomonas taxa. During epidemiological surveys conducted in the two main production areas of walnut in France (Grenoble in the southeast and Périgord in the southwest), 27 bacterial strains isolated from asymptomatic buds were identified as nonpathogenic based on phenotypic and genotypic characteristics and pathogenicity tests on walnut seedlings. Phylogenetic analyses, performed by MLSA, distinguished several groups with high bootstrap values corresponding to different pathovars of X. arboricola (pathovars pruni, corylina, fragariae, populi, celebensis, poinsettiicola, and juglandis). Nonpathogenic strains definitely belong to X. arboricola and grouped outside clusters defined by pathovars that form separate genetic lineages. Nonpathogenic strains were polymorphic, as they were distributed into three separate clusters within X. arboricola, termed NP1, NP2, and NP3, and four single branches. Clustering of phylogenetic positions of nonpathogenic strains is not correlated to their geographical origins, indicating the absence of genotype-geographic structure. NP2 showed a high percentage of similarity and clustered strains collected in both eastern and southwestern regions in France. It is remarkable that all nonpathogenic xanthomonad strains isolated from walnut belong to the same species, X. arboricola, which is the only pathogenic bacterium known to affect walnut so far. It suggests that a link exists between these lineages.

MLVA distinguished several clonal complexes within X. arboricola strains isolated from walnut. Strains causing VOC grouped separately from WB strains. VOC strains are divided into one clonal complex and two singletons, whereas WB strains are more heterogeneous and are divided into two clonal complexes and 16 singletons. Thus, strains causing VOC might be the result of a natural selection of some WB strains that gain additional features necessary to cause canker in woody parts. One VOC strain (12714) was not grouped with other VOC strains and had already been found to belong to a separate lineage by MLSA studies (10). We could hypothesize that the gain of features could occur several times on the WB populations. This hypothesis of gaining features is supported by the ability of VOC strains to cause cankers on trunks and necrotic spots on leaves and fruits as well, whereas WB strains cause only necrotic symptoms on leaves and fruits. Nonpathogenic strains were grouped separately from clonal complexes grouping WB and VOC strains and revealed to be genetically heterogeneous, as they were divided into six clonal complexes and four singletons. There is a concordance between MLSA and MLVA results. All genetic lineages identified by MLSA were distinct from each other in the MLVA scheme as well. However, MLVA was confirmed to be a more discriminative typing method than MLSA, given that some genetic lineages defined by MLSA were divided into several clonal complexes and singletons in the population structure defined by MLVA. Strains that appear identical by MLSA may present different VNTR profiles, such as strains belonging to NP2 that formed a single genetic lineage with a 100% internal similarity in the MLSA scheme and are divided into three clonal complexes and two singletons by MLVA. In addition, population structure of clonal complexes defined by VNTR analysis is not correlated to their geographical origins, given that some clonal complexes grouped strains belonging to different geographic locations from both the Rhône-Alpes and Périgord regions. The occurrence of the same genotype in different geographic areas supports the fact that nonpathogenic strains have been spread all over walnut-growing areas in France. It would be useful to check the occurrence of these bacterial lineages on other cultivated plants and weeds in both areas to gain a better understanding of the ecology of these bacteria and to highlight the role of walnut in the ecology of X. arboricola strains. VNTR markers proved to be relatively easy and rapid to use and provide informative data for subtyping bacterial strains. VNTR analysis will gain more attention in the future because of the availability of more Xanthomonas genomes sequenced, since VNTRs can rapidly be characterized by PCRs with specific primers based on the flanking regions of the tandem repeats.

Based on the population structure of X. arboricola described in this study, we can presume that pathovars result from a selection of host-specialized strains that have been further developed as a single clonal lineage. Nonpathogenic X. arboricola strains were more polymorphic than pathovars and were spread across different geographic locations, suggesting that the plant-bacterium interaction of these nonpathogenic strains occurs differently from plant-pathogen interactions. In this context, the plant-bacterium interactions of representative strains of nonpathogenic X. arboricola defined by MLSA and MLVA (CFBP 7634, CFBP 7645, CFBP 7651, CFBP 7652, and CFBP 7653) were assessed on a range of plants, including P. persica, R. sativus, L. esculentum, C. annum, and N. benthamiana. None of the xanthomonad strains tested induced disease symptoms when inoculated into Prunus and Raphanus leaves. However, L. esculentum, C. annum and N. benthamiana were resistant to strains CFBP 7651 and CFBP 7652 as local cell death was noticed at the point of inoculation, while CFBP 7634, CFBP 7653, and CFBP 7645 did not induce any disease symptoms on these plants. According to the results obtained from pathogenicity and HR assays, the following questions arise: (i) why do some nonpathogenic strains induce cell death when inoculated into tomato, pepper, and tobacco leaves whereas others do not? and (ii) what are the molecular pathogenicity determinants that differ between pathogenic and nonpathogenic strains and induce the inability of nonpathogenic strains to cause disease on plant species tested and especially on walnut?

Previous studies reported that the repertoires of T3Es within Xanthomonas strains play a basic role in aggressiveness and host specificity (25), and more recently, Hajri et al. (27) investigated the variability of T3E repertoires in the species X. arboricola, showed that T3SS is an essential virulence mechanism in X. arboricola, and suggested the use of nonpathogenic strains to test whether a modification in T3E repertoire would lead to changes in the pathogenic behavior of the bacterium. Thus, the distribution of 18 T3Es, which are the core sets of T3Es present in X. arboricola pv. juglandis (27), was investigated together with the presence or absence of genes coding for the highly conserved T3SS of the Hrp2 family. We noticed congruence between the composition of T3E repertoires and phylogenetic structure of the nonpathogenic strains in three major groups. Interestingly, no T3SS was detected in NP1, NP3, and CFBP 7653, although these strains clearly belong to X. arboricola. The groups of strains defined by MLSA, i.e., NP1, NP3, and the strain CFBP 7653, lack an hrp-hrc cluster, whereas strains CFBP 1022, 7630, 7652, and NP2 encoded a typical hrp-type T3SS. The major common attribute of strains belonging to NP1 and NP3 is the lack of T3SS genes and the inability to elicit any HR or cause disease symptoms in any of the plants tested. Hence, we can hypothesize that nonpathogenic strains lacking T3SS and containing the T3Es xopR, avrXccA1, and avrBs2 are unable to translocate effectors into plant cells, which may explain their inability to cause disease on plant species tested and to elicit an HR.

Given that hrpF functions as a translocon of effector proteins into the host cell (42, 43), we can assume that T3Es present in NP1 strains are not translocated into plants inoculated due also to the absence of hrpF, which might explain their inability to elicit an HR on nonhost plants. In fact, previous studies showed that mutation of the hrpF locus of X. oryzae pv. oryzicola strain resulted in the loss of pathogenicity in rice and the ability to induce HR in nonhost tobacco (44). Similarly, mutations in hrpF of X. campestris pv. vesicatoria strain or X. axonopodis pv. glycines strain resulted in strains that were nonpathogenic in host plants and unable to elicit race-specific HRs (44, 45).When corresponding R and avr genes are present in the host and pathogen, respectively, the result is disease resistance (46). AvrBs2 is a functional protein reporter for avrBs2-dependent HR activity in plant cells (47). Transient expression of AvrBs2 in BS2 pepper leaves induced a strong HR response. The AvrBs2/Bs2 reporter system has been previously used as a tool to identify translocated effectors in bacterial pathogens that infect other naturally occurring or transgenic BS2 plant lines (48). In addition, we showed that within the species X. arboricola, two isolates, CFBP 7640 and CFBP 7653, do not contain either an hrp-hrc cluster coding for a T3SS or known Xanthomonas T3E genes. Given that the successful establishment of a disease relies on the presence of a T3SS and on the translocation of T3Es (26, 49, 50), it appears that these strains may depend on an entirely nonpathogenic lifestyle. Previous work reported that nonpathogenic Pseudomonas isolates lacking a T3SS are common leaf colonizers of healthy plants and grow as well as or better than other Pseudomonas syringae strains on nonhost species without causing disease (51). Clarke et al. (50) showed later that these strains contain an unusual hrp-hrc cluster that is only distantly related to the canonical P. syringae hrp-hrc cluster.

This study reports on the occurrence of nonpathogenic isolates within the species X. arboricola that do not contain an hrp-hrc cluster coding for a T3SS and of strains that harbor some T3Es but no T3SS-encoding genes. The present finding raises questions about the origin of these nonpathogenic strains and the evolution of plant pathogenicity in X. arboricola. Mohr et al. (51) suggest that loss of the T3SS in one pathogenic strain was the initial event in the evolution of T3SS lacking isolates. They assumed that nonpathogenic Pseudomonas strains most likely evolved from a pathogenic strain ancestor through the loss of its T3SS. The hrp-hrc cluster and most of the effector genes were deleted during the evolution of nonpathogenic strains.

In contrast, pathogenic strains might have evolved from their nonpathogenic ancestors after (i) acquisition of pathogenesis-associated gene clusters (in this context, Lu et al. [52] suggested that acquisition of the hrp clusters was a critical step in the evolution of plant pathogenicity in Xanthomonas), (ii) mutations in genes related to virulence or avirulence function, and (iii) horizontal gene transfer. Nonpathogenic strains were present together with pathogenic ones. They grow or at least survive epiphytically on plants without causing disease. Consequently, these strains have been largely overlooked because of their lower economic importance.

The identified nonpathogenic strains provide excellent tools to elucidate the difference in HR and pathogenicity reactions between these strains and pathogenic ones in relation to their T3E repertoires. T3E genes that were not detected in any nonpathogenic isolates studied are excellent candidates for being responsible for pathogenicity in X. arboricola. Furthermore, these strains could be used to study pathogenicity factors and molecular determinants (particularly effectors), to better understand the basis of host range in Xanthomonas, and to gain insight into molecular determinants of plant resistance and how bacterial pathogenicity works. In fact, it would be interesting to engineer these isolates back into pathogens with different host ranges by adding a functional T3SS or different assortment of effector genes to see whether a modification of the interaction between these strains and plants tested would be observed.

Because of the practical obstacle of extensive pathogenicity tests, the possibility cannot be excluded that some of these nonpathogenic isolates could be pathogenic Xanthomonas strains with an unknown host plant(s), particularly strains belonging to NP2, CFBP 7630, CFBP 1022, and CFBP 7652, which harbor genes of a typical T3SS of the Hrp2 family and seven T3E-encoding genes. Therefore, the use of the term “nonpathogenic” for these strains might not always be correct. In this context, Mohr et al. (51) assumed that nonpathogenic P. syringae strains can live without causing disease on plants for extended periods of time but can cause disease when they find themselves on a susceptible plant under favorable environmental conditions.

Further work is under way to conduct a complete genomic comparison of pathogenic and nonpathogenic strains sequenced, in order to find bacterial functions involved in the epiphytic colonization of plants and to determine common and differential genes within these strains (and other xanthomonads) that could be linked to pathogenicity. Further analysis of sequenced genomes would aid in understanding the function of the missing components of a T3SS in nonpathogenic strains and to provide novel insights into other pathogenicity determinants that may play a role in the plant-bacterium interaction. Among them, particular attention should be paid to genes involved in adhesion, biofilm formation, flagellum synthesis, motility, lipopolysaccharide synthesis, quorum sensing, and finally type IV and VI secretion systems to answer the question regarding the possibility of involvement of other pathogenicity determinants like T4SS and T6SS in NP1 strains that carry T3Es but do not harbor any gene of T3SS. Understanding the acquisition and evolution of type III effectors in X. arboricola may help us to deduce the roles of these proteins in pathogenicity and will help in understanding functions of most T3Es identified in X. arboricola.