Differentiation of Strains of Xylella fastidiosa by a Variable Number of Tandem Repeat Analysis

Abstract

Short sequence repeats (SSRs) with a potential variable number of tandem repeat (VNTR) loci were identified in the genome of the citrus pathogen Xylella fastidiosa and used for typing studies. Although mono- and dinucleotide repeats were absent, we found several intermediate-length 7-, 8-, and 9-nucleotide repeats, which we examined for allelic polymorphisms using PCR. Five genuine VNTR loci were highly polymorphic within a set of 27 X. fastidiosa strains from different hosts. The highest average Nei's measure of genetic diversity (H) estimated for VNTR loci was 0.51, compared to 0.17 derived from randomly amplified polymorphic DNA (RAPD) analysis. For citrus X. fastidiosa strains, some specific VNTR loci had a H value of 0.83, while the maximum value given by specific RAPD loci was 0.12. Our approach using VNTR markers provides a high-resolution tool for epidemiological, genetic, and ecological analysis of citrus-specific X. fastidiosa strains.

Xylella fastidiosa has been associated with diseases in economically important crops such as grapevine, plum, almond, peach, citrus, and more recently, coffee (13, 17) as well as some diseases in ornamental plants (3). Reciprocal transmission tests including X. fastidiosa from several hosts evidenced the occurrence of host infection groups (for details, see reference 10). In Brazil, X. fastidiosa has been responsible for causing diseases in important crops such as citrus and coffee (5, 13), although it has also been observed in plum (7) and Hibiscus schizopetalus (E. W. Kitajima, H. D. Coletta-Filho, M. A. Machado, and Q. S. Novaes, 33th Congr. Brazil Phytopathol. Soc., abstr. 323, 2000). The major economic damage is in the sweet orange crop (Citrus sinensis Osb.) which has suffered an annual loss of about $100 million. However, the total cost to Brazilian agriculture is probably higher because the economic damage caused to the coffee crop by X. fastidiosa has not been estimated.

Methods for distinguishing between bacterial strains are important for detecting disease outbreaks for epidemiological analysis, and for understanding the genetic structure of microbial populations. Molecular techniques of DNA profiling based on PCR of randomly amplified polymorphic DNA (RAPD) and repetitive element polymorphism PCR (rep-PCR) have been used with great success for studies of genetic variation and the relationships between X. fastidiosa strains (2, 6, 8, 16). However, the data produced by these methods are biallelic, which limits the amount of genetic information per locus and thus the use of these methods in estimating genetic diversity. Interlaboratory reproducibility is also a weakness of RAPD and rep-PCR analysis (20). Short sequence repeats (SSRs) with a potential variable number of repeats (VNTR) within prokaryotic DNA have been used as markers for discrimination between strains (1, 9, 12, 23). Some of these repetitive regions are located within or near DNA coding regions and so could have the potential to affect gene expression (22).

The complete sequencing of the genome of the pathogenic X. fastidiosa strain 9a5c (19) has allowed the identification of repetitive DNA motifs. In the present study, we conducted a search for SSR size variation in different strains of X. fastidiosa, comparing the results with the frequently used RAPD method.

MATERIALS AND METHODS

Strains and DNA isolation.

Table 1 shows the X. fastidiosa strains used, all of which were cultured on BCYE agar medium (24) for 7 to 10 days at 28°C. The hexadecyltrimethylammonium bromide (CTAB) miniprep method (25) was used for DNA preparation from each strain. Briefly, cells were harvested, washed twice in washing buffer (pH 8.0) containing 20 mM Tris HCl and 10 mM EDTA, and treated with proteinase K and sodium dodecyl sulfate. Proteins and other cellular components were removed by using CTAB and by two chloroform extractions. The DNA present in the aqueous phase was precipitated with absolute ethanol followed by washing with 70% ethanol. The pellet obtained was dissolved in TE buffer (10 mM Tris HCl, 1 mM EDTA) (pH 8.0) containing 20 mM RNase. The DNA was subjected to electrophoresis in a 0.8% agarose gel. After staining with ethidium bromide, the samples were visualized under ultraviolet light and the DNA concentration was estimated. The DNA solution was diluted to 5 ng/μl and stored at −20°C.

TABLE 1.

X. fastidiosa strains used in this study

Strain	Host	Geographic origin	Source
9a5c	Citrus sinensis	São Paulo, Brazil	INRAa
B14	Citrus sinensis	São Paulo, Brazil	CCSMb
11779	Citrus sinensis	Santa Catarina, Brazil	IAPARc
M2-1	Citrus sinensis	São Paulo, Brazil	CCSM
ITAa	Citrus sinensis	São Paulo, Brazil	CCSM
11834	Citrus sinensis	Paraná, Brazil	IAPAR
NP3.2	Citrus sinensis	São Paulo, Brazil	CCSM
11775	Citrus sinensis	Rio Grande do Sul, Brazil	IAPAR
11066	Citrus sinensis	Paraná, Brazil	IAPAR
11067	Citrus sinensis	Paraná, Brazil	IAPAR
NP2.3	Citrus sinensis	São Paulo, Brazil	CCSM
10438	Citrus sinensis	Paraná, Brazil	IAPAR
11038	Citrus sinensis	São Paulo, Brazil	IAPAR
11347	Citrus sinensis	Paraná, Brazil	IAPAR
SR9.2	Citrus sinensis	São Paulo, Brazil	CCSM
11037	Citrus sinensis	São Paulo, Brazil	IAPAR
11399	Citrus sinensis	Sergipe, Brazil	IAPAR
11380	Citrus sinensis	Santa Catarina, Brazil	IAPAR
SL4	Citrus sinensis	Minas Gerais, Brazil	CCSM
C3	Coffea arabica	São Paulo, Brazil	CCSM
C6	Coffea arabica	São Paulo, Brazil	CCSM
C13	Coffea arabica	São Paulo, Brazil	CCSM
12288	Coffea arabica	Paraná, Brazil	IAPAR
8935	Vitis vinifera	United States	ATCC 35879d
Hib. 5	Hibiscus schizopetalus	Brasilia, Brazil	CCSM
9746	Prunus salicina	Paraná, Brazil	IAPAR
12319	Catharantus sp.	São Paulo, Brazil	IAPAR

^aINRA, Institut National de la Recherche Agronomique et Université Victor Ségale, Bordeaux, France. 

^bCCSM, Centro de Citricultura Sylvio Moreira, Instituto Agronômico, Cordeiropolis, São Paulo, Brazil. 

^cIAPAR, Instituto Agronômico do Paraná, Paraná, Brazil. 

^dATCC, American Type Culture Collection, Manassas, Va. 

Computer analysis of repetitive DNA.

The genomic DNA sequence of X. fastidiosa strain 9a5c (available on the UNICAMP website at http://onsona.lbi.ic.unicamp.br/xf) was screened for repetitive DNA with the Tandem Repeat Finder version 2.02 software (4), freely available at http://c3.biomath.mssm.edu /trf/. This software scores all DNA motif categories, although in our study, only perfect repeats were selected. BLASTN searches of the regions upstream and downstream of the repeat motif were conducted on the complete X. fastidiosa 9a5c genome to localize the repeats and related genes.

Repeat analysis.

For some 7-, 8-, and 9-nucleotide repeats, sets of primers (Table 2) with the potential for locus-specific amplification were designed using Lasergene 99 software (DNASTAR, Inc). Most of the primers were deduced from sequences bordering the repeat, 5 nucleotides upstream and downstream of the locus. The other primers (SSR26, SSR30, and SSR32) were designed up to 30 nucleotides distant from the repeat locus. Amplifications were conducted in a volume of 25 μl containing 10% of 10× PCR buffer (200 mM Tris-HCl, pH 8.4, 500 mM KCl), 2.5 mM MgCl₂, 0.25 mM (each) deoxynucleoside triphosphate (GIBCO BRL), 50 ng (each) primer, 50 ng of template DNA, and 1.5 U of Taq DNA polymerase (GIBCO BRL). An initial denaturation step at 94°C for 3 min was followed by a touchdown amplification program. DNA was denatured at 94°C for 1 min, and primers were annealed for 30 s and extended at 72°C for 1 min. The initial annealing temperature was 64°C for 1 cycle. The temperature was subsequently dropped 1°C every cycle until a final annealing temperature of 55°C was reached. For the remaining 22 cycles, the annealing temperature was 55°C. A final extension step of 4 min at 72°C was followed by 10°C soak. PCR products were separated by gel electrophoresis in 3% agarose gels and stained with ethidium bromide, with the fragment size estimated based on migration relative to that of a 100-bp size marker (GIBCO BRL) loaded together with the samples in the gel.

TABLE 2.

SSR regions monitored for polymorphism in 27 strains of X. fastidiosa isolated from different hosts^a

Potential VNTR name	Repeat motif	Repeat position in genome	Primer sequence for VNTR (5′ → 3′ amplification)
SSR20	(ATTGCTG)13	18527–18619	ATGAAGAAGCCAGGATACAT
			GCTACACGTGCAACAAC
SSR21	(TGTTATC)21	52902–53049	AACACGGATCAAGCTCATG
			GGAACACGCAATAGTAAGAC
SSR26	(GTGTGTGA)37	49439–49733	CTGTGATCGGTGAATTGA
			TCAAGCACACTTCCTACG
SSR28	(GTGTGCCT)11	1046415–1046503	GCAACGCTGTTATCTCAAT
			ATTACGCTTCTTATCGCTGT
SSR30	(TGATCCTG)15	1512303–1512422	TACGCTGCACCTGTCTG
			CTGTGAACTTCCATCAATCC
SSR32	(CTGATGTG)9	2050614–2050688	AGATGAACCTCGCCAC
			GTACTCATCTGCGATGG
SSR34	(TTGGGTAG)22/(TTGGGTAA)35	2289266–2289727	TGATAGAACTGTTTGACGCATTTG
			TCGGGAAGTTTGGGGTGAC
SSR36	(TGTTGGGG)10	2639987–2640070	ATGTCACTCAGGTCAGG
			CAGAACCACCGACTG
SSR40	(GAAGGCGTA)27	1220096–1220340	ACCTTGACGACGGATG
			TAGGAACTGCTGCTACTGAT

^aAll data are based on the genome sequence of X. fastidiosa strain 9a5c. 

RAPD analysis.

RAPD analysis was performed in a volume of 13 μl containing 10% 10× PCR buffer, 2.0 mM MgCl₂, 0.25 mM (each) deoxynucleoside triphosphate (GIBCO BRL), 15 ng of template DNA, 1.5 U of Taq DNA polymerase (GIBCO BRL), and 15 ng of primer (Operon kit of 10-mer primers; OPG10, OPG17, OPG19, OPH03, OPH07, OPH12, OPN04, OPQ05, and OPW07). Amplification was performed using a temperature profile, with initial denaturation at 94°C for 3 min followed by 36 cycles of denaturation (1 min at 94°C), annealing (1 min at 36°C), and extension (2 min at 72°C) with a final 10-min extension step at 72°C. The RAPD reaction products were separated by electrophoresis in 1.3% agarose gels and stained with ethidium bromide, with the RAPD band size estimated by comparison with a 1-kb DNA ladder marker (GIBCO BRL).

Data analysis.

All genotypes used for SSR and RAPD were included in the analysis. For the RAPD methodology, only high-intensity markers of unambiguous interpretation and good reproducibility were scored. Nei's measure of genetic diversity was calculated as H = [1 − Σpi²], where p_i is the frequency of allele i at the locus (15).

RESULTS

Database searches for SSRs.

There were no 1- or 2-base SSRs, but there were 67 perfect SSRs with 3 to 33 nucleotides per unit (Table 3). Most common were 8-nucleotide SSRs, which were present at twelve loci at copy numbers varying from 3 to 37 copies per repeat. In one of these loci, we encountered mixed repeats harboring the (TTGGGTAG)₂₂/(TTGGGTAA)₃₅ motif.

TABLE 3.

Perfect SSR regions encountered in the genome of X. fastidiosa strain 9a5c by computerized screening

Repeat unit length	No. of copiesa
3	8, 10
4	7, 8, 13
5	11, 12, 21, 22
6	(5)b 4, 7, 8, 10, 15, 16
7	4, 5, (2) 7, 13, 21
8	(3) 3, 4, 5, (2) 9, 10, 11, 15, 22, 35, 37
9	(2) 3, 27
10	3, 4
11	2
12	(7) 2, 3
13	(2) 2
14	2
15	(6) 2
17	2
18	4
21	2
23	2
29	2
33	2

^aThe number of copies in different positions in the genome is shown. 

^bNumber of times that the number of copies of repeat were encountered through the genome. 

SSR amplification by PCR.

For DNA polymorphism analysis of the SSR regions, we selected two 7-nucleotide SSRs, six 8-nucleotide SSRs, and one 9-nucleotide SSR (Table 2). These SSRs were selected as potential VNTRs because they were of intermediate size, uncommon for prokaryotic SSRs (22), and presented a large number of copies per repeat unit. Amplifications using all these potential VNTR primers were performed in X. fastidiosa strains 9a5c and B14 (Fig. 1), although for two SSR primers (SSR32 and SSR34), the amplification step failed even though variations in the amplification program and reaction mixtures were tried. Although the SSR26 and SSR36 primers generated the same product for both strains 9a5c and B14, they were still used for typing studies together with primers SSR20, -21, -28, -30, and -40.

FIG. 1.

Gel electrophoresis of seven SSR primers for two X. fastidiosa strains. X. fastidiosa strains 9a5c (lanes 1) and B14 (lanes 2) and SSR primers SSR36 (A lanes) SSR20 (B lanes), SSR26 (C lanes), SSR21 (D lanes), SSR28 (E lanes), SSR40 (F lanes), and SSR30 (G lanes) were used. The intensely stained band represents a 600-bp fragment (arrow). Lanes M, 100-bp ladder marker (GIBCO BRL).

VNTR analysis.

Table 4 shows the SSR and RAPD primers used and the genetic diversity values obtained. For the SSR26 and SSR36 primers, no variation in the number of repeats was observed in the strains. Five pairs of primers (SSR20, -21, -28, -30, and -40) showed definite length polymorphism in their respective SSR regions and thus appear to represent genuine VNTRs. Even though the reactions were repeated, doublet bands were sometimes observed when the SSR20 and SSR40 primers were used with strain 12319 and when the SSR30 primer was used in some other strains. This indicates the presence of two alleles for the same SSR locus in some strains, although no doublet bands were observed in strain 9a5c.

TABLE 4.

Mean H values for the SSR and RAPD loci in 27 strains of X. fastidiosa isolated from different hosts

Locusa	H value for X. fastidiosa strain isolated from:
	All hosts	Citrus	Coffee	Other hostsb
SSR
SSR20	0.77	0.70	0.50	0.72
SSR21	0.44	0.28	0.63	0.38
SSR26	0.00	0.00	0.00	0.00
SSR28	0.81	0.80	0.75	0.63
SSR30	0.69	0.64	0.63	0.48
SSR36	0.00	0.00	0.00	0.00
SSR40	0.83	0.83	0.63	0.56
Avg	0.51	0.46	0.45	0.39
RAPD
OPG10	0.23	0.12	0.09	0.36
OPG17	0.16	0.02	0.15	0.33
OPG19	0.22	0.16	0.05	0.27
OPH03	0.14	0.07	0.06	0.39
OPH07	0.16	0.04	0.08	0.43
OPH12	0.13	0.09	0.06	0.21
OPN04	0.19	0.11	0.06	0.35
OPQ05	0.07	0.00	0.00	0.28
OPW07	0.20	0.00	0.13	0.48
Avg	0.17	0.07	0.07	0.34

^aSSR and RAPD primers. 

^bData from strains isolated from plum, grapevine, hibiscus, and periwinkle. 

A minimum of six and a maximum of eight alleles per locus were observed with the VNTR primers, generating a high level of polymorphism. The discriminatory power of each VNTR locus and RAPD analysis was estimated from the genetic diversity (H) values based on the number of alleles and their frequency. In the analysis of citrus and coffee X. fastidiosa strains, the average H value was higher for VNTR markers than for RAPD markers (Table 4), with VNTR primers SSR28 and SSR40 showing the highest H values. For citrus strains, these values reached 0.80 (SSR28) and 0.83 (SSR40). It has previously been shown that VNTR typing of Bacillus anthracis and Yersinia pestis has produced H values of 0.80 and 0.82, respectively (1, 11). Table 4 also shows that in the analysis of X. fastidiosa strains isolated from other hosts (plum, grapevine, hibiscus, and periwinkle), the average VNTR marker H value (0.39) was only slightly higher than the average RAPD marker H value (0.34).

DISCUSSION

The presence of SSRs and their usefulness for typing have previously been shown for both human and animal pathogens (22), but this is the first time that these molecular markers have been used to type phytopathogenic bacteria.

The characteristics of the SSR regions found in the X. fastidiosa genome were quite different from the SSRs already known in other bacteria. No mono-or dinucleotide SSRs were detected in X. fastidiosa, but there was a large number of 7- and 8-nucleotide repeats (Table 3), a fact which led us to use intermediate-sized repeats in our typing studies. Shorter, 1- or 2-nucleotide repeats are abundant in other species of pathogenic bacteria such as Helicobacter, Neisseria, Mycobacterium, and Escherichia coli, while intermediate-sized repeats have rarely been found (22). This phenomenon appears to be independent of genome size, because in the 4.2-Mb chromosome of Bacillus subtilis, there are only repeats with short sequences, while in the 0.58-Mb chromosome of Mycoplasma genitalium, there are other classes of repeats (21).

The VNTR typing method presented in this paper is a powerful method for the genetic characterization of X. fastidiosa isolates. The potential VNTR regions selected for typing were conserved in all X. fastidiosa strains tested, irrespective of host (Table 4 and Fig. 2). The genetic diversity (H) values based on VNTR primers were extremely high, especially for citrus- and coffee-specific X. fastidiosa strains. This contrasts with the RAPD technique, which although it has allowed the construction of well-defined X. fastidiosa host groups (5, 16) has shown limited discriminatory capacity between X. fastidiosa strains isolated from citrus species (18). Although the RAPD technique has the potential for genome-wide analysis, SSRs have the advantage that they can link to the genomic hypervariable regions produced as a result of variation in sequence composition and DNA polymerase activity. It seems that the high discriminatory power of VNTR locus analysis is a result of the hypervariability of this molecular marker rather than the occurrence of artifacts due to serial subculture. The VNTR assay is also fairly reproducible (1, 14), and VNTR mutation rates, as seen in Bacillus anthracis, are less than 10⁻⁵ (11).

FIG. 2.

Gel electrophoresis of SSR20, SSR21, SSR26, SSR28, SSR30, and SSR40 primers from 12 X. fastidiosa strains. X. fastidiosa strains 9a5c, B14, 11779, M2.1, ITAa, 11834, 9713, C13, C3, Hib 5, 9746, and NP3.2 are shown in lanes 1 to 12, respectively (see Table 1 for host identification). The intensely stained bands represent a 600-bp fragment (arrows). Lanes M, 100-bp ladder marker (GIBCO BRL).

The growing numbers of prokaryotic DNA sequences (including those from plant pathogens) in databases and computer programs available for the detection of SSR loci have facilitated the evaluation of SSR within DNA sequences. Not only are the hypervariability and reproducibility of SSRs markers useful, but if the primers are X. fastidiosa specific, there is also the possibility of in situ analysis of a known gene without isolating the bacteria. This approach would greatly facilitate epidemiological, genetic, and ecological studies of fastidious bacteria, such as X. fastidiosa, which are normally difficult to isolate and grow.