JOURNAL OF BACTERIOLOGY, Aug. 2004, p. 5442–5449
Copyright © 2004, American Society for Microbiology. All Rights Reserved.
Vol. 186, No. 16
DNA Microarray-Based Genome Comparison of a Pathogenic and a
Nonpathogenic Strain of Xylella fastidiosa Delineates Genes
Important for Bacterial Virulence†
Tie Koide,1‡ Paulo A. Zaini,1‡ Leandro M. Moreira,1Ricardo Z. N. Ve ˆncio,2Adriana Y. Matsukuma,1
Alan M. Durham,3Diva C. Teixeira,4Hamza El-Dorry,1Patrícia B. Monteiro,4§
Ana Claudia R. da Silva,1Sergio Verjovski-Almeida,1Aline M. da Silva,1*
and Suely L. Gomes1*
Departamento de Bioquímica, Instituto de Química,1and Departamento de Estatística2and Cie ˆncia da
Computac ¸a ˜o,3Instituto de Matema ´tica e Estatística, Universidade de Sa ˜o Paulo,
Sa ˜o Paulo, and Fundo de Defesa da Citricultura, Araraquara,4Brasil
Received 20 February 2004/Accepted 22 March 2004
Xylella fastidiosa is a phytopathogenic bacterium that causes serious diseases in a wide range of economically
important crops. Despite extensive comparative analyses of genome sequences of Xylella pathogenic strains
from different plant hosts, nonpathogenic strains have not been studied. In this report, we show that X.
fastidiosa strain J1a12, associated with citrus variegated chlorosis (CVC), is nonpathogenic when injected into
citrus and tobacco plants. Furthermore, a DNA microarray-based comparison of J1a12 with 9a5c, a CVC strain
that is highly pathogenic and had its genome completely sequenced, revealed that 14 coding sequences of strain
9a5c are absent or highly divergent in strain J1a12. Among them, we found an arginase and a fimbrial adhesin
precursor of type III pilus, which were confirmed to be absent in the nonpathogenic strain by PCR and DNA
sequencing. The absence of arginase can be correlated to the inability of J1a12 to multiply in host plants. This
enzyme has been recently shown to act as a bacterial survival mechanism by down-regulating host nitric oxide
production. The lack of the adhesin precursor gene is in accordance with the less aggregated phenotype
observed for J1a12 cells growing in vitro. Thus, the absence of both genes can be associated with the failure of
the J1a12 strain to establish and spread in citrus and tobacco plants. These results provide the first detailed
comparison between a nonpathogenic strain and a pathogenic strain of X. fastidiosa, constituting an important
step towards understanding the molecular basis of the disease.
Xylella fastidiosa is a gram-negative bacterium, limited to the
plant xylem vessels, which is responsible for worldwide eco-
nomic losses due to diseases caused in a variety of plants of
agricultural relevance. This bacterium is transmitted to new
host plants during xylem sap feeding by insect vectors and
spreads from the site of infection to colonize the xylem, a water
transport network of vessels composed of lignified dead cells.
Bacterial cells attach to the vessel wall, forming biofilm-like
colonies that, depending on the size, can occlude the xylem
vessels, blocking water transport and causing water stress
symptoms (22, 35).
Different strains of X. fastidiosa have been reported to infect
a wide range of plants, including grapevines and citrus, al-
mond, and pear trees, among others (26). In the United States
for instance, Pierce’s disease prevents profitable viticulture
if leafhopper vectors are present at high densities (1, 14). In
Brazil, citrus variegated chlorosis (CVC) is responsible for
great financial losses to the citrus agroindustry, being detected
in one-third of the citrus trees. Orange production quickly
decreases in orchards affected by CVC, as fruits become hard-
ened and of no commercial value. Interestingly, within the
majority of host plants, X. fastidiosa behaves as a harmless
Several X. fastidiosa strains have had their genomes com-
pletely or partially sequenced, and genome comparative anal-
ysis with different pathogenic strains of X. fastidiosa pointed to
common candidate virulence determinants as well as strain-
specific genomic signatures (4, 24, 32, 37). However, no infor-
mation is available about the genome composition of non-
pathogenic Xylella strains, which would contribute to more
direct insights on pathogenicity mechanisms.
Genome-wide comparison between pathogenic and non-
pathogenic strains within a species is a useful strategy for
identifying candidate genes important for virulence. DNA
microarray-based genome composition analysis is a good al-
ternative to full genome sequencing and has been used in
comparative studies to analyze various bacterial pathogens in-
cluding Mycobacterium tuberculosis (3), Helicobacter pylori (29),
Pseudomonas aeruginosa (38), Bacillus anthracis (28), Yersinia
pestis, and Yersinia pseudotuberculosis (12).
In this report, we show that X. fastidiosa strain J1a12, which
was isolated from citrus and is suitable for genetic transforma-
tion (6, 23), elicits few or no CVC symptoms when inoculated
into citrus and tobacco plants. Furthermore, a DNA micro-
* Corresponding author. Mailing address: Departamento de Bio-
química, Instituto de Química, Universidade de Sa ˜o Paulo, Av. Prof.
Lineu Prestes 748, 05508-900 Sa ˜o Paulo, SP, Brasil. Phone, fax, and
e-mail for Suely L. Gomes: 5511 3091 3826, 5511 3815 5579, sulgomes
@iq.usp.br. Phone fax, and e-mail for Aline M. da Silva: 5511 3091
2182, 5511 3815 5579, firstname.lastname@example.org.
† Supplemental material for this article may be found at http://jb
‡ These authors contributed equally to this work.
§ Present address: Alellyx Applied Genomics, Campinas, SP, Brasil.
array-based genome composition analysis was performed by
comparing J1a12 with strain 9a5c, which produces typical CVC
symptoms (20) but is resistant to transformation with DNA in
vitro (23), a drawback for its genetic manipulation. Our mi-
croarray data revealed that the great majority of the coding
sequences (CDS) are highly conserved on both strains. How-
ever, 14 CDS were shown to be absent or highly divergent in
the nonpathogenic strain. Expression profiling of both strains,
PCR and reverse transcription (RT)-PCR with CDS-specific
primers, and DNA sequence analysis were used to validate the
genomic differences observed.
MATERIALS AND METHODS
Bacterial strains, growth conditions, and pathogenicity tests. Triply cloned X.
fastidiosa strains 9a5c (20) and J1a12 (23) isolated from CVC symptomatic Citrus
sinensis (L.) Osbeck trees (sweet orange) were grown in periwinkle wilt broth
medium (7) at 28°C in the dark with 100 rpm rotatory agitation. As shown in
Table 1, strain J1a12 is nonpathogenic, despite its isolation from a citrus plant
with CVC symptoms. This is possibly due to a mixed population of bacteria
infecting the plant (see “Final remarks” below). A culture started from a single
colony was weekly passaged through serial transfers at a 1/100 dilution in peri-
winkle wilt medium. For citrus plant experiments, 9a5c and J1a12 strains with 9
and 24 weekly passages, respectively, were used; for tobacco plant experiments,
11 weekly passages were used for both strains. Mechanical inoculation of plants
was performed essentially as described in reference 20 for C. sinensis and as
described in reference 21 for Nicotiana tabacum (accession clevelandii). Further
details are found in the supplemental material. The detection of X. fastidiosa in
host plants followed the procedure described in reference 25, which consists of
PCR experiments with a pair of primers (CVC-1 and 272-int) designed to
identify Xylella strains isolated from citrus plants.
Microarray construction and hybridization. A 6,152-element DNA microar-
ray was printed containing unique internal fragments of 2,692 CDS spotted at
least in duplicate, representing 94.5% of all of the 2,848 CDS annotated by
Simpson et al. (32). DNA fragments ranging from 200 to 1,000 bp were PCR
amplified with CDS-specific primers (18- to 23-mers) designed with PRIMER3
(http://www-genome.wi.mit.edu/genome_software/other/primer3.html) and based
on the annotated genome sequence of X. fastidiosa strain 9a5c (http://aeg.lbi.ic
.unicamp.br/xf). A full list of primers, PCR product sizes, and their nucleotide
sequences are available at the project site (http://verjo19.iq.usp.br/xylella
/microarray/). The arrays were hybridized with DNA fragments from strain 9a5c
in combination with itself, J1a12, or grapevine strain Temecula (kindly provided
by Marie-Anne Van Sluys, University of Sa ˜o Paulo) labeled separately with
either Cy3- or Cy5-dCTP analogs (see the supplemental material). Expression
profiling studies were carried out by labeling total RNA with the Cy-Scribe post
labeling kit (Amersham Biosciences) according to the manufacturer’s instruc-
Data acquisition and normalization. Microarray slides were scanned by using
a Generation III DNA scanner (Amersham Biosciences), and fluorescence in-
tensity values (ICy3and ICy5) from each spot were extracted by using Array-
Vision, version 6.0, software (Imaging Research, Inc.). Raw fluorescence inten-
sity and normalized data are available at the project site (http://verjo19.iq.usp.br
/xylella/microarray/). Data normalization was carried out by LOWESS fitting on
an M versus A plot (39), where M is the ratio of fluorescence intensities of the
two measurements for each spot [defined as M ? log2(ICy5/ICy3)] and A is the
geometric mean of the fluorescence intensities [defined as A ? 1/2 ? log2(ICy5?
ICy3)]. The normalization script is available at the project site.
CDS classification process. To determine hybridization noise and to estimate
dynamic cutoff values for classifying a CDS as equally present in both strains
(9a5c and J1a12) we used the hybridization data collected from three indepen-
dent homotypic experiments (9a5c versus 9a5c). For this kind of experiment, also
called self-self hybridization, the microarrays were cohybridized with strain 9a5c
DNA separately labeled with either Cy3- or Cy5-dCTP analogs. As verified in the
M versus A plot, there is a dependence of the hybridization intensity log ratio of
each spot (M) with the mean log intensity of each spot (A). Thus, we have
determined a cutoff value for each interval in the A axis by using kernel density
estimators. We chose kernel density estimators (31) instead of the normal prob-
ability density function (16) because we experimentally derived the null distri-
bution as the result of homotypic experiments and verified that it does not pre-
sent a Gaussian behavior (further information is available at the project site
[http://verjo19.iq.usp.br/xylella/microarray/]). The density distribution was inte-
grated around the mode peak until 0.995 probability was reached, defining
intensity-dependent noise threshold cutoff values (credibility intervals) based on
experimental data from 9a5c versus 9a5c homotypic experiments, thus setting an
interval where the hybridization ratio is considered to be 1:1 (e.g., ?2.5 ? M ?
1.8, for the lowest accepted intensities at A ? 2). These credibility intervals were
subsequently used in the analysis of replicas of 9a5c versus J1a12 hybridization
experiments to nonparametrically estimate the null distribution of the statistical
test H0: CDS is present in both J1a12 and 9a5c strains. Spots outside the
credibility intervals present strong evidence against a 1:1 ratio. Using these
criteria, four categories were defined for the CDS in the J1a12 genome based on
its orthologous 9a5c counterpart: (i) equally present in both strains, (ii) divergent
in strain J1a12, (iii) highly divergent or absent in J1a12, and (iv) higher copy
number in J1a12. Category i includes all of the CDS for which ?60% of the
replicas were inside the credibility intervals.
CDS presenting negative log ratio values outside the credibility intervals could
be classified as divergent in strain J1a12 (category ii) or highly divergent or
absent in J1a12 (category iii). To distinguish between these categories, we per-
formed four control experiments where the microarrays were cohybridized with
DNA from the sequenced grapevine strain Temecula (37) and strain 9a5c labeled
with either Cy3- or Cy5-dCTP analogs. Next, we derived a correspondence
between the hybridization intensity log ratio for each CDS amplicon and its
respective nucleotide sequence identity in both strains. Amplicon sequences
exhibiting nucleotide identities smaller than 20% were defined as category iii,
highly divergent or absent. Their respective log ratio M was taken as the cutoff
threshold (Mcutoff) of category iii. The log ratio value with the least false callings
was an Mcutoffvalue of ?1.7. At this threshold, 19 false positives and 23 false
negatives were observed (0.76 and 0.92%, respectively). Category iii includes
CDS with an M value of ??1.7 and a P value smaller than 0.05 in a t test against
the null hypothesis H0: M ? ?1.7. Sequencing some amplicons from J1a12 that
were outside the credibility intervals and checking the divergence between 9a5c
and J1a12 sequences further supported the adequacy of the threshold. The
remaining CDS that have a negative log ratio value, with an M value of ??1.7,
were considered divergent (category ii). Those CDS with positive log ratio values
were considered to be present at higher copy numbers in J1a12 (category iv).
CDS with low reproducibility (less than 60% of replicas in a single category) were
excluded from the analysis.
Validation of microarray data. PCR and RT-PCR experiments were per-
formed with CDS-specific primers to further investigate the status of CDS in the
genome of strain J1a12 classified by using the microarray data. The reactions
were carried out with genomic DNA or cDNA from strain 9a5c or J1a12 with 35
cycles of amplification. A sample of CDS presenting log ratios outside the
credibility intervals, as determined by homotypic hybridization experiments
(9a5c versus 9a5c), was chosen to perform the validation. Among the 64 CDS
with negative log ratios (categories ii and iii), the following 33 CDS were ran-
domly chosen for PCR validation: XF0077, XF0496, XF0497, XF0500, XF0663,
XF0684, XF0696, XF0890, XF1250, XF1306, XF1581, XF1588, XF1589, XF1646,
XF1663, XF1664, XF1686, XF1708, XF1709, XF1860, XF1863, XF1874, XF1877,
XF1878, XF1884, XF1968, XF2193, XF2195, XF2307, XF2722, XF2768, XF2772,
and XFb0001. In addition, the following CDS with positive log ratios (category
iv) were also tested by PCR: XF0513, XF0514, XF0515, XF0516, XF0518,
XF0519, XF0521, XF1933, XF1934, XF1935, XF1936, and XF1937. A 4-?l
sample of each reaction mixture was electrophoresed in agarose gels, and DNA
was stained with ethidium bromide. The amplicons were then classified by visual
inspection as absent, same copy number, or more abundant in strain J1a12 in
relation to strain 9a5c. In addition, DNA sequence determination was carried out
TABLE 1. Evaluation of C. sinensis and N. tabacum plants
inoculated with X. fastidiosa CVC strains 9a5c and J1a12
No. of infected plants/total no. of plants
at time (mo) postinoculation
C. sinensis N. tabacum
58 15 1.53
aPCR experiments were performed on samples drawn from the plant xylem by
using primers designed to identify Xylella citrus isolates (25).
bOne of the plants died 15 months after inoculation.
VOL. 186, 2004XYLELLA PATHOGENICITY ANALYSIS BY DNA MICROARRAYS5443
for a few CDS. For that, PCR products obtained with primers based on neigh-
boring CDS were cloned in pGEM-TEasy vector (Promega) and dideoxy se-
quencing reactions were performed with 100 ng of plasmid DNA in Big Dye
Terminator sequencing reactions (Applied Biosystems) according to the manu-
facturer’s instructions. Sequencing reactions were carried out with either CDS-
specific or T7 promoter primers.
RESULTS AND DISCUSSION
X. fastidiosa CVC strain J1a12 is nonpathogenic. In planta
pathogenicity tests were carried out with strains 9a5c and
J1a12, and results are presented in Table 1. Inoculation of
citrus plants with strain 9a5c resulted in multiplication of X.
fastidiosa in all plants analyzed 8 months after infection. Plants
were also evaluated visually for characteristic CVC symptoms,
and positive symptoms were observed in the leaves of 77% of
the citrus plants 15 months after inoculation. In contrast, no
symptoms were observed in the leaves of plants inoculated with
strain J1a12 up to 15 months after infection.
Similar results were observed with tobacco plants as the
experimental host. As shown in Table 1, none of the plants
inoculated with J1a12 presented symptoms or were colonized
by the bacteria. In contrast, all tobacco plants inoculated with
9a5c presented the lesions characteristic of X. fastidiosa CVC
infection, as previously described (21).
These results indicate that strain J1a12 shows a nonpatho-
genic phenotype, in contrast to the highly pathogenic behavior
of strain 9a5c. In addition, Table 1 shows that plant coloniza-
tion by strain J1a12 is very inefficient, as no bacteria were
detected in the plant xylem by PCR experiments with a pair of
primers (CVC-1 and 272-int) which are specific for Xylella
isolates from citrus plants (25). This pair of primers amplifies
a genomic region of approximately 500 bp, from chromosome
position 1051239 to 1051745 of the 9a5c genome, encompass-
ing 195 bp of CDS XF1100 and an intergenic region upstream
of this CDS. It is important to stress that these primers can
amplify the correct DNA fragment when directly tested in
J1a12 in vitro cultures (6, 23).
Genotyping by DNA microarray analysis. To investigate
whether the differences in phenotype observed between strains
9a5c and J1a12 could be associated with differences at the
DNA level, we have constructed a DNA microarray encom-
passing 2,692 CDS, which represents 94.5% of all CDS de-
scribed in the reference strain 9a5c (32). Total DNAs isolated
from strains 9a5c and J1a12 were separately labeled with either
Cy3- or Cy5-dCTP fluorescent analogs, and competitive micro-
array hybridizations were performed. Raw and normalized hy-
bridization data are available at the project site (http://verjo19
.iq.usp.br/xylella/microarray/). An initial screening revealed
that 292 CDS presented either low signal intensity or poor
reproducibility and were excluded from further analysis. As
detailed in Material and Methods, the remaining 2,400 CDS
were classified into four categories according to the normalized
hybridization fluorescence intensity ratios of J1a12 over 9a5c
DNA samples determined for each CDS. Among the 2,400
CDS, approximately 96% were found to have a log ratio (M)
around 0 and were classified as equally present in both strains
(category i). One example is shown in Fig. 1A. This figure
shows an M versus A plot, i.e., normalized intensities log ratios
(M) versus the average of log intensities (A) of all the replicas
for a given CDS. This kind of graph shows the dependence of
the ratio on the overall intensity of each spot, indicating that,
for genes with low hybridization intensity signals (A values
below 2), the observed ratios have a higher intrinsic dispersion,
as determined by homotypic hybridizations. As a result, differ-
ent cutoff values for M were used for different ranges of inten-
sity (A) when classifying a CDS as equally present in both
strains. M versus A plots showing the reproducibility of the
data for each CDS are available at the project site.
Fifty CDS were found to have a hybridization intensity log
ratio of ?1.7 ? M ? ?0.3 and were classified as divergent in
strain J1a12 (category ii, an example is shown in Fig. 1B).
Table 2 lists only the CDS with hybridization intensity log
ratios between ?0.5 and ?1.7. For the complete list of CDS in
this category see Table S1 in the supplemental material. Four-
teen CDS were found to have the log ratios of ??1.7 and were
classified as absent or highly divergent in strain J1a12 (Table
3). A typical example is shown in Fig. 1C. Within this group,
three CDS, namely XF0077, XF1250, and XF1646, were espe-
cially interesting due to their putative involvement in patho-
genesis and will be discussed in more detail later. Most of the
CDS in this category belong to the previously defined flexible
gene pool of Xylella, which includes integrated prophages and
genomic islands (4, 24). For instance, CDS XF1860, XF1874,
XF1878, and XF1884 belong to genomic island 4 (24). This
region has a different GC content and altered codon bias,
which are common features of laterally transferred elements
(15). However, CDS XF0077, XF1250, XF1646, XF1707, and
XF1708 are not mapped within any genomic island and do not
show altered GC content or codon bias.
In addition, 40 CDS (see Table S2 in the supplemental
material) were found to have a log ratio of ?0 and were
considered as possibly presenting a higher number of copies in
J1a12 (an example is shown in Fig. 1D). The majority of the
CDS in this category are of unknown function or have phage-
related functions. Among them, there are 2 groups of contig-
uous CDS (XF0512 to XF0523 and XF1932 to XF1937). The
first set is within genomic island 1 (24). The other group in-
cludes genes that belong to different functional categories such
as DNA metabolism and transport, suggesting them as possible
newborn paralogs in the J1a12 genome.
Despite all the information obtained with DNA microarray
genotyping, it is necessary to stress that frameshifts and point
mutations cannot be identified by this method. In addition, our
DNA microarray experiments will not detect genes present
exclusively in the unsequenced strain J1a12, impairing detec-
tion of genes that would eventually attenuate pathogenicity.
Validation of CDS classification. To further validate the
distinction between CDS potentially absent or highly divergent
and those classified as divergent, a sample of 33 CDS (listed in
Materials and Methods) were analyzed by PCR with CDS-
specific primers. All CDS tested were PCR negative for strain
J1a12, including those with log hybridization intensity ratios
between ?0.5 and ?1.7.
Figure 2A shows an example of a CDS (XF0077, encoding a
fimbrial adhesin precursor) classified as absent or highly diver-
gent by microarray analysis. PCR and RT-PCR experiments
with CDS-specific primers corroborated its classification in this
category. In addition, a PCR experiment with a pair of specific
primers flanking XF0077 encompassing CDS XF0075 to
XF0079 produced a smaller amplicon in strain J1a12 than in
5444KOIDE ET AL. J. BACTERIOL.
strain 9a5c (Fig. 2A). The exact position of the deleted region
was determined by DNA sequence analysis, as described in the
Data for XF1968, which encodes a putative methyltrans-
ferase classified as divergent by microarray analysis (mean log
ratio of ?1.4), is shown in Fig. 2B. The PCR and RT-PCR
results with CDS-specific primers were negative, suggesting the
absence of the CDS. However, PCR experiments with a pair of
specific primers flanking XF1968 (encompassing XF1967 to
XF1969) showed amplicons with the same size in both strains
(Fig. 2B). Nucleotide sequencing of both amplicons confirmed
that the methyltransferase encoded by XF1968 is divergent
between 9a5c and J1a12 (Table 4).
Three other examples of CDS classified as divergent accord-
ing to microarray experiments and found to be PCR negative
are also depicted in Table 4. Sequence analysis of these CDS
cloned from strain J1a12 has confirmed their divergence, as
they present nucleotide identities between 92 and 55% when
compared to strain 9a5c (Table 4). These results show that
PCR validation can be misleading, given that the primers used
were based on the genome sequence of strain 9a5c. A few mis-
matches in the regions where the primers should anneal in the
J1a12 DNA template can give PCR-negative results, leading to
a wrong conclusion about the presence or absence of a gene.
RNA expression studies by microarray hybridization com-
paring strains J1a12 and 9a5c were performed as an additional
validation of CDS classification. RNA hybridization data are
available at the project site (http://verjo19.iq.usp.br/cagexylella
/private/). Hybridization signals were found to be at the back-
ground level for the 14 CDS classified as absent or highly
divergent in strain J1a12. On the other hand, all of these CDS
showed significant expression levels in strain 9a5c.
FIG. 1. Xylella strain J1a12 CDS classification based on DNA microarray hybridization ratios. Examples of CDS classified in each of the four
categories are shown. (A) XF1621, classified as equally present in both 9a5c and J1a12 strains, category i; (B) XF0262, classified as divergent in
J1a12, category ii; (C) XF0496, classified as absent or highly divergent in strain J1a12, category iii; (D) XF1937, classified as higher copy number
in J1a12, category iv. Orange dots represent the results for all CDS in the microarray from three homotypic control experiments (9a5c labeled with
Cy5 versus 9a5c labeled with Cy3). Graphs show the fluorescence intensity ratio M ? log2(ICy5/ICy3) versus the fluorescence intensity mean A ?
1/2 ? log2(ICy5? ICy3). Green dots represent the hybridization data (9a5c labeled with Cy3 versus J1a12 labeled with Cy5 or vice versa) from
multiple replicas for the indicated CDS, where the fluorescence intensity ratio M ? log2(IJ1a12/I9a5c). Similar graphs for each of the CDS are
available at the project site (http://verjo19.iq.usp.br/cagexylella/private/).
VOL. 186, 2004XYLELLA PATHOGENICITY ANALYSIS BY DNA MICROARRAYS5445
The expression levels of CDS that were classified as equally
present in both strains (category i, see Materials and Methods)
were studied under standard bacterial growth conditions. This
class of CDS was chosen to eliminate possible hybridization
artifacts due to sequence divergence. We found that among the
2,296 CDS in this category, which presented detectable hybrid-
ization intensity values, approximately 97% exhibited compa-
rable RNA expression levels on both strains, i.e., differences in
expression were smaller than twofold, with P values smaller
than 0.05 in a t test. However, about 1% of the CDS presented
a higher RNA expression level (twofold or more) in strain 9a5c
and about 2% had higher expression in strain J1a12 (see Ta-
bles S3 and S4 in the supplemental material). No obvious
correlation could be made between the CDS presenting differ-
ential expression and the phenotypes of each strain.
Functional characteristics of genes absent or highly diver-
gent in strain J1a12. Among the 14 CDS classified as absent or
highly divergent (Table 3), 10 have no similarity to known
genes and no function could be assigned. Therefore, they will
not be further discussed, although their involvement in patho-
genesis cannot be excluded.
The X. fastidiosa strain 9a5c genome encodes three fimbrial
adhesin subunits from type III pilus (XF0077, XF0078, and
XF0080). Our microarray data classified XF0077 as absent or
highly divergent, and we have confirmed its deletion in strain
J1a12 by DNA sequence analysis. The deleted region, encom-
passing 1,050 nucleotides (Fig. 2A), extends from position
76845 to 77895 (numbers from the strain 9a5c main chromo-
some). XF0078 was classified as divergent in J1a12, and DNA
sequence analysis has shown a 92.2% nucleotide identity with
its ortholog in 9a5c (Table 4). The third fimbrial adhesin para-
log (XF0080) was classified as equally present in both strains
(category i). The three paralogs from 9a5c are similar to mrkD
from Klebsiella pneumoniae and share high similarity with each
other (XF0077 and XF0078 share 70% amino acid sequence
identity and both display around 60% amino acid identity to
XF0080). As reported for the adhesins of K. pneumoniae (30),
the N-terminal region exhibits a greater degree of variability,
probably conferring on strain 9a5c the ability to adhere to
different bacterial and/or host cell components or even pro-
ducing an extracellular matrix with greater bonding capacity.
In K. pneumoniae, mrkD null mutants are fimbriate but non-
adhesive (34) and the mrkD gene product is not required for
biofilm formation (18). Proteomic and mass spectrometric
analyses of whole-cell lysates and extracellular components
have demonstrated that structural and adhesive subunits of
fimbriae are ubiquitous in cultures of Xylella strain 9a5c (33).
Despite the presence of two CDS encoding the adhesion sub-
unit precursors XF0078 and XF0080 in J1a12, we have ob-
served that this strain displays a much less aggregated pheno-
type in vitro than 9a5c cells, as shown in Fig. 3. One possible
explanation for this phenotype is that the presence of the
TABLE 3. CDS absent or highly divergent in strain J1a12
Fimbrial adhesion precursor
Conserved hypothetical protein
Conserved hypothetical protein
Conserved hypothetical protein
aCDS shown here have a P value smaller than 0.05 in a t test for the null
hypothesis H0: M ? ?1.7.
bDNA hybridization intensity ratios [M ? log2(IJ1a12/I9a5c)] were calculated.
For each gene, the values shown are the means of the results from at least eight
TABLE 2. CDS divergent in strain J1a12
XF0078.................Fimbrial adhesin precursor
XF0262.................Colicin V precursor
XF0263.................Colicin V precursor
XF0500.................Phage-related repressor protein
XF0501.................Conserved hypothetical protein
XF0668.................Hemolysin-type calcium binding protein
XF0696.................Phage-related repressor protein
XF1589.................Plasmid stabilization protein
XF1590.................Plasmid stabilization protein
XF1862.................Conserved hypothetical protein
XF1873.................Conserved hypothetical protein
phosphate phosphatase/bifunctional enzyme
XF2722.................Type I restriction-modification system specificity
XF2726.................Type I restriction-modification system specificity
aOnly 44 CDS with log hybridization intensity ratios between ?1.7 and ?0.5
are shown in this table.
5446KOIDE ET AL.J. BACTERIOL.
adhesin encoded by XF0077 is important for the adhesion of
X. fastidiosa cells.
Interestingly, another CDS confirmed to be deleted from
strain J1a12 is XF1250, which encodes a putative arginase. The
deleted region of 963 nucleotides extends from position
1203200 to 1204163 (numbers from the strain 9a5c main chro-
mosome). This enzyme catalyzes the first step of arginine deg-
radation in the urea cycle, which may thus be incomplete in
strain J1a12. Besides being a substrate for arginases, arginine is
also a substrate for nitric oxide (NO) synthase, which converts
L-arginine to L-citrulline, releasing NO. In H. pylori, it has
recently been shown that the arginase (rocF) encoded by this
bacterium inhibits NO production by macrophages at physio-
logic concentrations of L-arginine. H. pylori rocF mutants
cocultured with macrophages were killed due to the restora-
tion of normal levels of NO. These results indicate that bac-
terial arginase down-regulates NO production, acting as a sur-
vival mechanism that contributes to the successful infection of
the host (10). Thus, it is tempting to speculate that the absence
of arginase in X. fastidiosa strain J1a12 is linked to its reduced
growth in planta and its incapacity to colonize the xylem ves-
sels; J1a12 cells would not be able to inhibit NO production by
the plant, analogous to the rocF mutant of H. pylori (10). In
fact, it is known that in plants, NO has an immune protective
role, mediating the plant defense against pathogens and serv-
ing as a signal in hormonal responses. Indeed, two NO-synthe-
sizing enzymes have recently been found in plants, one of
which is pathogen inducible (5, 11).
XF1646, classified as absent or highly divergent in J1a12, was
annotated as a putative UDP-3-O-(R-3-hydroxymyristoyl)-glu-
cosamine N-acyltransferase, which is similar to the lpxD gene
from Rickettsia rickettsii. This enzyme catalyzes the third step in
lipid A biosynthesis, a constituent of lipopolysaccharides from
the outer membrane. In Escherichia coli, an lpxD mutant had
its susceptibility to various antibiotics increased as high as
512-fold, indicating alterations in the outer membrane perme-
ability barrier (36). An altered outer membrane structure due
to the incomplete lipid A biosynthesis may render J1a12 more
sensitive to antimicrobial agents eventually produced by the
plant host. The increased permeability of its outer membrane
may also explain why J1a12 is amenable to DNA transforma-
tion, whereas 9a5c is not (6, 17, 23).
The two highest DNA hybridization intensity ratios obtained
when comparing 9a5c versus J1a12 were derived from the two
CDS encoded by the mini plasmid pXF1.3 (XFb0001 and
XFb0002). This result reflects the presence of multiple copies
of this plasmid in strain 9a5c (32) and its absence in strain
J1a12, as previously reported (6).
Final remarks. Among the 14 genes classified as absent or
highly divergent, three CDS encoding a fimbrial adhesin pre-
cursor, an arginase, and a UDP-3-O-(R-3-hydroxymyristoyl)-
glucosamine N-acyltransferase are conspicuously absent in the
nonpathogenic strain J1a12. Due to their putative role in bac-
terial survival in infected hosts, they emerge here as important
players in Xylella pathogenicity. The observation that several
other genes are missing in J1a12 gives support to the hypoth-
esis that bacterial pathogenesis is a multifactorial process and
that each of these factors may contribute somewhat quantita-
tively to the development of disease. In fact, inactivation of a
single fimbrial adhesin gene (PD0058) or a single fimbriae
protein gene (PD0062) in X. fastidiosa grapevine strain Te-
mecula was not sufficient to decrease bacterial pathogenicity,
causing only a slight reduction in the bacterial population (9).
Recent microarray expression studies comparing X. fastid-
iosa 9a5c cells freshly isolated from citrus with bacteria atten-
uated after several passages in axenic culture have shown that
most genes found to be induced in the freshly isolated condi-
tion were associated with adhesion and with possible adapta-
tion to the host environment (8). However, the set of genes
observed in that study is different from the genes found to be
absent in the nonpathogenic strain analyzed in the present
report, reinforcing the multifactorial hypothesis of bacterial
pathogenesis. Furthermore, our results obtained with J1a12
are independent of the number of passages in culture, since the
microarray experiments were performed with DNA from bac-
terial cells obtained after either 14 or 24 passages and no
differences were observed (data not shown).
Our plant colonization assays showing that strain J1a12 is
unable to induce CVC symptoms or even sustain itself in host
plants raise the question of how strain J1a12 was originally
isolated from a citrus tree. A recent report about the diversity
of the endophytic bacterial community in citrus trees (2) can
shed light upon this intriguing question. Possibly, the presence
of strain J1a12 in symptomatic plants is dependent on other
microorganisms and/or other X. fastidiosa pathogenic strains
FIG. 2. Validation of CDS classification. CDS-specific primers for
XF0077 (A) and XF1968 (B) were employed to perform PCR ampli-
fications with DNA from strains 9a5c and J1a12 (left panels) or to
perform RT-PCR amplifications with total RNA (central panels).
PCRs were also carried out with DNA of both strains and primers
based on the sequence of the CDS flanking XF0077 (amplicon XF0075
to XF0079) or XF1968 (amplicon XF1967 to XF1969) (right panels).
The sizes of the amplicons are shown in base pairs.
TABLE 4. Nucleotide and amino acid identity of selected divergent
CDS of strain J1a12 compared to strain 9a5c
Fimbrial adhesin precursor
Type I restriction-modification system
aComparison was carried out with the complete sequence of each CDS.
VOL. 186, 2004XYLELLA PATHOGENICITY ANALYSIS BY DNA MICROARRAYS5447
eventually present in the biofilms formed by the aggregated
cells and clogging the xylem vessels of infected plant hosts (13,
Among the CDS found to be absent in J1a12 and proposed
here to play a role in disease development, four CDS do have
orthologs in the three other pathogenic Xylella strains that
have been sequenced (see Table S5 in the supplemental ma-
terial). Thus, despite the diversity of hosts, geographical loca-
tion, and disease symptoms, different Xylella strains may
present similar mechanisms of pathogenesis. Our results sug-
gest that common strategies could be undertaken to control
the diseases that are caused by different X. fastidiosa strains
infecting various host plants. Given the importance of the set
of genes found here, further functional characterization is war-
ranted. Towards this end, we are currently trying to comple-
ment strain J1a12 with the CDS shown to be absent from its
genome. Different from 9a5c, the nonpathogenic J1a12 strain
is amenable to DNA transformation with plasmid vectors and
the transposome system (6, 17, 23). Thus, we believe that with
complementation studies it will be possible to evaluate the role
of these CDS in bacterial virulence and the ability to colonize
This work was funded by Fundac ¸a ˜o de Amparo a ` Pesquisa do Es-
tado de Sa ˜o Paulo (FAPESP).
We are greatly indebted to Hugo A. Armelin for coordinating the
Cooperation for Analysis of Gene Expression (CAGE) Project and for
strongly supporting this work. We thank Joa ˜o Carlos Setubal and Joa ˜o
Kitajima for providing information about reannotation of the Xylella
genomic sequence and Apua ˜ C. M. Paquola, Milton Y. Nishiyama, Jr.,
and Abimael A. Machado for providing bioinformatic tools. We thank
Jesus Ferro for coordinating our PCR amplification data bank and for
the Xylella cosmid library. We also thank Sanvai R. P. Rocha, Mateus
de Almeida Santos, and Anelise G. Mariano for help in pathogenicity
tests and Helder Nakaya and Ari J. S. Ferreira for valuable help in the
beginning of this work.
A.M.d.S., H.E.-D., S.L.G., and S.V.-A. were partially supported by
Conselho Nacional de Desenvolvimento Científico e Tecnolo ´gico,
(CNPq). T.K., L.M.M., R.Z.N.V., and P.A.Z. are FAPESP doctoral
1. Almeida, R. P. P., and A. H. Purcell. 2003. Transmission of Xylella fastidiosa
to grapevines by Homalodisca coagulata (Hemiptera: Cicadellidae). J. Econ.
2. Araujo, W. L., J. Marcon, W. Maccheroni, Jr., J. D. Van Elsas, J. W. Van
Vuurde, and J. L. Azevedo. 2002. Diversity of endophytic bacterial popula-
tions and their interaction with Xylella fastidiosa in citrus plants. Appl.
Environ. Microbiol. 68:4906–4914.
3. Behr, M. A., M. A. Wilson, W. P. Gill, H. Salamon, G. K. Schoolnik, S. Rane,
and P. M. Small. 1999. Comparative genomics of BCG vaccines by whole-
genome DNA microarray. Science 284:1520–1523.
4. Bhattacharyya, A., S. Stilwagen, G. Reznik, H. Feil, W. S. Feil, I. Anderson,
A. Bernal, M. D’Souza, N. Ivanova, V. Kapatral, N. Larsen, T. Los, A.
Lykidis, E. Selkov, Jr., T. L. Walunas, A. Purcell, R. A. Edwards, T. Hawkins,
R. Haselkorn, R. Overbeek, N. C. Kyrpides, and P. F. Predki. 2002. Draft
sequencing and comparative genomics of Xylella fastidiosa strains reveal
novel biological insights. Genome Res. 12:1556–1563.
5. Chandok, M. R., A. J. Ytterberg, K. J. van Wijk, and D. F. Klessig. 2003. The
pathogen-inducible nitric oxide synthase (iNOS) in plants is a variant of the
P protein of the glycine decarboxylase complex. Cell 113:469–482.
6. da Silva Neto, J. F., T. Koide, S. L. Gomes, and M. V. Marques. 2002.
Site-directed gene disruption in Xylella fastidiosa. FEMS Microbiol. Lett.
7. Davis, M. J., W. J. French, and N. W. Schaad. 1981. Axenic culture of the
bacteria associated with phony disease of peach and plum leaf scald. Curr.
8. de Souza, A. A., M. A. Takita, H. D. Coletta, C. Caldana, G. H. Goldman,
G. M. Yanai, N. H. Muto, R. C. de Oliveira, L. R. Nunes, and M. A.
Machado. 2003. Analysis of gene expression in two growth states of Xylella
fastidiosa and its relationship with pathogenicity. Mol. Plant-Microbe Inter-
9. Feil, H., W. S. Feil, J. C. Detter, A. H. Purcell, and S. E. Lindow. 2003.
Site-directed disruption of the fimA and fimF fimbrial genes of Xylella fas-
tidiosa. Phytopathology 93:675–682.
10. Gobert, A. P., D. J. McGee, M. Akhtar, G. L. Mendz, J. C. Newton, Y. Cheng,
H. L. Mobley, and K. T. Wilson. 2001. Helicobacter pylori arginase inhibits
nitric oxide production by eukaryotic cells: a strategy for bacterial survival.
Proc. Natl. Acad. Sci. USA 98:13844–13849.
11. Guo, F. Q., M. Okamoto, and N. M. Crawford. 2003. Identification of a plant
nitric oxide synthase gene involved in hormonal signaling. Science 302:100–
12. Hinchliffe, S. J., K. E. Isherwood, R. A. Stabler, M. B. Prentice, A. Rakin,
R. A. Nichols, P. C. F. Oyston, J. Hinds, R. W. Titball, and B. W. Wren. 2003.
Application of DNA microarrays to study the evolutionary genomics of
Yersinia pestis and Yersinia pseudotuberculosis. Genome Res. 13:2018–2029.
13. Hopkins, D. L. 1989. Xylella fastidiosa-xylem-limited bacterial pathogen of
plants. Annu. Rev. Phytopathol. 27:271–290.
14. Hopkins, D. L., and A. H. Purcell. 2002. Xylella fastidiosa: cause of Pierce’s
disease of grapevine and other emergent diseases. Plant Dis. 86:1056–1066.
15. Karlin, S. 2001. Detecting anomalous gene clusters and pathogenicity islands
in diverse bacterial genomes. Trends Microbiol. 9:335–343.
16. Kim, C. C., E. A. Joyce, K. Chan, and S. Falkow. 2002. Improved analytical
methods for microarray-based genome-composition analysis. Genome Biol.
17. Koide, T., J. F. da Silva Neto, S. L. Gomes, and M. V. Marques. 2004.
Insertional transposon mutagenesis in the Xylella fastidiosa citrus variegated
chlorosis strain with transposome. Curr. Microbiol. 48:247–250.
18. Langstraat, J., M. Bohse, and S. Clegg. 2001. Type 3 fimbrial shaft (MrkA)
of Klebsiella pneumoniae, but not the fimbrial adhesin (MrkD), facilitates
biofilm formation. Infect. Immun. 69:5805–5812.
19. Leite, B., M. L. Ishida, E. Alves, H. Carrer, S. F. Pascholati, and E. W.
FIG. 3. X. fastidiosa strain J1a12 has lost the ability to aggregate. Optical microscopy of X. fastidiosa strains 9a5c (A) and J1a12 (B).
5448 KOIDE ET AL.J. BACTERIOL.
Kitajima. 2002. Genomics and X-ray microanalysis indicate that Ca2? and Download full-text
thiols mediate the aggregation and adhesion of Xylella fastidiosa. Braz.
J. Med. Biol. Res. 35:645–650.
20. Li, W. B., L. Zreik, N. G. Fernandes, V. S. Miranda, D. C. Teixeira, A. J.
Ayres, M. Garnier, and J. M. Bove. 1999. A triply cloned strain of Xylella
fastidiosa multiplies and induces symptoms of citrus variegated chlorosis in
sweet orange. Curr. Microbiol. 39:106–108.
21. Lopes, S. A., D. M. Ribeiro, P. G. Roberto, S. C. Franca, and J. M. Santos.
2000. Nicotiana tabacum as an experimental host for the study of plant-
Xylella fastidiosa interactions. Plant Dis. 84:827–830.
22. Machado, M. A., A. A. Souza, H. D. Coletta-Filho, E. E. Kuramae, and M. A.
Takita. 2001. Genome and pathogenicity of Xylella fastidiosa. Mol. Biol.
23. Monteiro, P. B., D. C. Teixeira, R. R. Palma, M. Garnier, J. M. Bove, and J.
Renaudin. 2001. Stable transformation of the Xylella fastidiosa citrus varie-
gated chlorosis strain with oriC plasmids. Appl. Environ. Microbiol. 67:2263–
24. Nunes, L. R., Y. B. Rosato, N. H. Muto, G. M. Yanai, V. S. da Silva, D. B.
Leite, E. R. Goncalves, A. A. de Souza, H. D. Coletta-Filho, M. A. Machado,
S. A. Lopes, and R. C. de Oliveira. 2003. Microarray analyses of Xylella
fastidiosa provide evidence of coordinated transcription control of laterally
transferred elements. Genome Res. 13:570–578.
25. Pooler, M. R., and J. S. Hartung. 1995. Specific PCR detection and identi-
fication of Xylella fastidiosa strains causing citrus variegated chlorosis. Curr.
26. Purcell, A. H., and D. L. Hopkins. 1996. Fastidious xylem-limited bacterial
plant pathogens. Annu. Rev. Phytopathol. 34:131–151.
27. Purcell, A. H., and S. R. Saunders. 1999. Fate of Pierce’s disease strains of
Xylella fastidiosa in common riparian plants in California. Plant Dis. 83:825–
28. Read, T. D., S. N. Peterson, N. Tourasse, L. W. Baillie, I. T. Paulsen, K. E.
Nelson, H. Tettelin, D. E. Fouts, J. A. Eisen, S. R. Gill, E. K. Holtzapple,
O. A. Okstad, E. Helgason, J. Rilstone, M. Wu, J. F. Kolonay, M. J. Beanan,
R. J. Dodson, L. M. Brinkac, M. Gwinn, R. T. DeBoy, R. Madpu, S. C.
Daugherty, A. S. Durkin, D. H. Haft, W. C. Nelson, J. D. Peterson, M. Pop,
H. M. Khouri, D. Radune, J. L. Benton, Y. Mahamoud, L. X. Jiang, I. R.
Hance, J. F. Weidman, K. J. Berry, R. D. Plaut, A. M. Wolf, K. L. Watkins,
W. C. Nierman, A. Hazen, R. Cline, C. Redmond, J. E. Thwaite, O. White,
S. L. Salzberg, B. Thomason, A. M. Friedlander, T. M. Koehler, P. C. Hanna,
A. B. Kolsto, and C. M. Fraser. 2003. The genome sequence of Bacillus
anthracis Ames and comparison to closely related bacteria. Nature 423:81–
29. Salama, N., K. Guillemin, T. K. McDaniel, G. Sherlock, L. Tompkins, and S.
Falkow. 2000. A whole-genome microarray reveals genetic diversity among
Helicobacter pylori strains. Proc. Natl. Acad. Sci. USA 97:14668–14673.
30. Sebghati, T. A. S., T. K. Korhonen, D. B. Hornick, and S. Clegg. 1998.
Characterization of the type 3 fimbrial adhesins of Klebsiella strains. Infect.
31. Silverman, B. W. 1986. Density estimation. Chapman and Hall, London,
32. Simpson, A. J., F. C. Reinach, P. Arruda, F. A. Abreu, M. Acencio, R.
Alvarenga, L. M. Alves, J. E. Araya, G. S. Baia, C. S. Baptista, M. H. Barros,
E. D. Bonaccorsi, S. Bordin, J. M. Bove, M. R. Briones, M. R. Bueno, A. A.
Camargo, L. E. Camargo, D. M. Carraro, H. Carrer, N. B. Colauto, C.
Colombo, F. F. Costa, M. C. Costa, C. M. Costa-Neto, L. L. Coutinho, M.
Cristofani, E. Dias-Neto, C. Docena, H. El-Dorry, A. P. Facincani, A. J.
Ferreira, V. C. Ferreira, J. A. Ferro, J. S. Fraga, S. C. Franca, M. C. Franco,
M. Frohme, L. R. Furlan, M. Garnier, G. H. Goldman, M. H. Goldman, S. L.
Gomes, A. Gruber, P. L. Ho, J. D. Hoheisel, M. L. Junqueira, E. L. Kemper,
J. P. Kitajima, J. E. Krieger, E. E. Kuramae, F. Laigret, M. R. Lambais, L. C.
Leite, E. G. Lemos, M. V. Lemos, S. A. Lopes, C. R. Lopes, J. A. Machado,
M. A. Machado, A. M. Madeira, H. M. Madeira, C. L. Marino, M. V.
Marques, E. A. Martins, E. M. Martins, A. Y. Matsukuma, C. F. Menck,
E. C. Miracca, C. Y. Miyaki, C. B. Monteriro-Vitorello, D. H. Moon, M. A.
Nagai, A. L. Nascimento, L. E. Netto, A. Nhani, Jr., F. G. Nobrega, L. R.
Nunes, M. A. Oliveira, M. C. de Oliveira, R. C. de Oliveira, D. A. Palmieri,
A. Paris, B. R. Peixoto, G. A. Pereira, H. A. Pereira, Jr., J. B. Pesquero, R. B.
Quaggio, P. G. Roberto, V. Rodrigues, A. J. de M. Rosa, V. E. de Rosa, Jr.,
R. G. de Sa, R. V. Santelli, H. E. Sawasaki, A. C. da Silva, A. M. da Silva,
F. R. da Silva, W. A. da Silva, Jr., J. F. da Silveira, M. L. Silvestri, W. J.
Siqueira, A. A. de Souza, A. P. de Souza, M. F. Terenzi, D. Truffi, S. M. Tsai,
M. H. Tsuhako, H. Vallada, M. A. Van Sluys, S. Verjovski-Almeida, A. L.
Vettore, M. A. Zago, M. Zatz, J. Meidanis, and J. C. Setubal. 2000. The
genome sequence of the plant pathogen Xylella fastidiosa. Nature 406:151–
33. Smolka, M. B., D. Martins, F. V. Winck, C. E. Santoro, R. R. Castellari, F.
Ferrari, I. J. Brum, E. Galembeck, H. Della Coletta Filho, M. A. Machado,
S. Marangoni, and J. C. Novello. 2003. Proteome analysis of the plant
pathogen Xylella fastidiosa reveals major cellular and extracellular proteins
and a peculiar codon bias distribution. Proteomics 3:224–237.
34. Tarkkanen, A. M., R. Virkola, S. Clegg, and T. K. Korhonen. 1997. Binding
of the type 3 fimbriae of Klebsiella pneumoniae to human endothelial and
urinary bladder cells. Infect. Immun. 65:1546–1549.
35. Tyson, G. E., B. J. Stojanovic, R. F. Kuklinski, T. J. Divittorio, and M. L.
Sullivan. 1985. Scanning electron-microscopy of Pierce’s disease bacterium
in petiolar xylem of grape leaves. Phytopathology 75:264–269.
36. Vaara, M., and M. Nurminen. 1999. Outer membrane permeability barrier in
Escherichia coli mutants that are defective in the late acyltransferases of lipid
A biosynthesis. Antimicrob. Agents Chemother. 43:1459–1462.
37. Van Sluys, M. A., M. C. de Oliveira, C. B. Monteiro-Vitorello, C. Y. Miyaki,
L. R. Furlan, L. E. Camargo, A. C. da Silva, D. H. Moon, M. A. Takita, E. G.
Lemos, M. A. Machado, M. I. Ferro, F. R. da Silva, M. H. Goldman, G. H.
Goldman, M. V. Lemos, H. El-Dorry, S. M. Tsai, H. Carrer, D. M. Carraro,
R. C. de Oliveira, L. R. Nunes, W. J. Siqueira, L. L. Coutinho, E. T. Kimura,
E. S. Ferro, R. Harakava, E. E. Kuramae, C. L. Marino, E. Giglioti, I. L.
Abreu, L. M. Alves, A. M. do Amaral, G. S. Baia, S. R. Blanco, M. S. Brito,
F. S. Cannavan, A. V. Celestino, A. F. da Cunha, R. C. Fenille, J. A. Ferro,
E. F. Formighieri, L. T. Kishi, S. G. Leoni, A. R. Oliveira, V. E. Rosa, Jr.,
F. T. Sassaki, J. A. Sena, A. A. de Souza, D. Truffi, F. Tsukumo, G. M. Yanai,
L. G. Zaros, E. L. Civerolo, A. J. Simpson, N. F. Almeida, Jr., J. C. Setubal,
and J. P. Kitajima. 2003. Comparative analyses of the complete genome
sequences of Pierce’s disease and citrus variegated chlorosis strains of Xylella
fastidiosa. J. Bacteriol. 185:1018–1026.
38. Wolfgang, M. C., B. R. Kulasekara, X. Y. Liang, D. Boyd, K. Wu, Q. Yang,
C. G. Miyada, and S. Lory. 2003. Conservation of genome content and
virulence determinants among clinical and environmental isolates of Pseudo-
monas aeruginosa. Proc. Natl. Acad. Sci. USA 100:8484–8489.
39. Yang, Y. H., S. Dudoit, P. Luu, D. M. Lin, V. Peng, J. Ngai, and T. P. Speed.
2002. Normalization for cDNA microarray data: a robust composite method
addressing single and multiple slide systematic variation. Nucleic Acids Res.
VOL. 186, 2004 XYLELLA PATHOGENICITY ANALYSIS BY DNA MICROARRAYS5449