Xylella fastidiosa gene expression analysis by DNA microarrays

Xylella fastidiosa genome sequencing has generated valuable data by identifying genes acting either on metabolic pathways or in associated pathogenicity and virulence. Based on available information on these genes, new strategies for studying their expression patterns, such as microarray technology, were employed. A total of 2,600 primer pairs were synthesized and then used to generate fragments using the PCR technique. The arrays were hybridized against cDNAs labeled during reverse transcription reactions and which were obtained from bacteria grown under two different conditions (liquid XDM2 and liquid BCYE). All data were statistically analyzed to verify which genes were differentially expressed. In addition to exploring conditions for X. fastidiosa genome-wide transcriptome analysis, the present work observed the differential expression of several classes of genes (energy, protein, amino acid and nucleotide metabolism, transport, degradation of substances, toxins and hypothetical proteins, among others). The understanding of expressed genes in these two different media will be useful in comprehending the metabolic characteristics of X. fastidiosa, and in evaluating how important certain genes are for the functioning and survival of these bacteria in plants.


Introduction
X. fastidiosa (Wells et al., 1987) belongs to the gram-negative group and is restricted to the xylem vessels of host plants. It has been associated with diseases that affect diverse plant species, some of which economically important, these including alfalfa, almonds, blackberries, coffee, citrus fruits, grapes, peaches, pears, plums and certain ornamental plants (Hopkins, 1989). Citrus Variegated Chlorosis (CVC) was first detected in Brazil in 1987, and currently constitutes a serious threat to the Brazilian orange juice industry, since it is present in the main cultivation areas, being responsible for significant losses in orange production (Rossetti et al., 1990). In 2000, a consortium of laboratories in São Paulo State published the X. fastidiosa isolate 9a5c sequenced genome. A main chromosome (2,679,305 base pairs) and two other plasmids (51,158 and 1,285 base pairs) were sequenced, these presenting a total of 2,905 genes, from which half presented similarity with unknown protein functions (Simpson et al., 2000).
Understanding the complete genome sequence was a substantial advance towards comprehension of metabolic and replicate characteristics, and for starting the first approach in determining pathogenicity mechanisms. Papers published recently have explored the information generated by genomic sequencing, highlighting a series of hypotheses related to the functioning of energy metabolism, nutrient transport, adherence, aggregation, toxicity, the secretion of pathogenicity factors, intercellular interactions, iron homeostasis, antioxidant responses and other important pathogenicity mechanisms (Simpson et al., 2000;Keen et al., 2000;Dow and Daniels, 2000;Lambais et al., 2000;Silva et al., 2001;Leite et al., 2002;Meidanis et al., 2002).
It is known that X. fastidiosa demands a complex medium for its in vitro development (Holt et al., 1994). With X. fastidiosa genome sequencing, the possible genes involved in bacterial metabolism have become known, and as a result, a defined and adequate medium for cultivating X. fastidiosa, known as XDM 2 (Xylella defined medium), was set up . The components of this medium have been included based on metabolic pathways found with the help of information obtained from the X. fastidiosa genome. XDM 2 contains glucose, vitamins (biotin, thiamine, pyridoxine hydrochloride and nicotinic acid) and amino acids (serine, methionine, asparagine and glutamine), as well as iron, phosphate, sulfate and myoinositol. The XDM 2 medium has made it possible to cultivate X. fastidiosa more successfully than by using the complex BCYE modified media , which presents in its formulation only yeast extract and an ACES buffer. Furthermore, differences among X. fastidiosa isolates, obtained from various host plants, have been observed. These are related to their cultivation in media of different compositions, thereby indicating the existence of genetic variability within this group of bacteria (Hopkins, 1989).
The present accumulation of information with the sequencing of genomes from various organisms has offered an enormous opportunity to understand the biological functions of many genes, previously described as unknown (Lashkari et al., 1997). Furthermore, microarray technology (Shalon et al., 1996) provides a simultaneous way for immediately monitoring the expression of several genes. In practice, it is possible to arrange about 6,000 elements (genes) in an area of less than 1.8 cm 2 . A nearly complete collection of 4,290 Escherichia coli open reading frames (ORFs) was obtained for analyzing the expression ratio of this bacterium when cultivated in two different media (minimal and rich). Bacteria cultivated in a rich medium presented accelerated multiplication, the higher number of genes significantly expressed being related to the translation apparatus. On the other hand, bacteria cultivated in the minimal medium showed elevated expression of many of those genes involved in biosynthetic pathways, mainly in the amino acids (Tao et al., 1999).
The aim of this work was to develop a DNA microarray analysis set, and undertake a transcriptional study of those genes related to the metabolism of the X. fastidiosa 9a5c strain isolated from citrus fruits, when cultivated under two distinct conditions, BCYE (complex media) and XDM 2 (defined media).

Cultivation conditions
For genomic DNA extraction, X. fastidiosa isolate 9a5c was cultivated in 'Petri dishes' containing BCYE medium (Wells et al., 1981) at 28°C for six days. For RNA extraction, the bacteria were cultivated for four days at 30°C in a 250 mL Erlenmeyer flask containing 30 mL of either liquid XDM 2 medium  or liquid modified BCYE  under shaking conditions (140 rpm).

Isolation of genomic DNA and total RNA
Genomic DNA was extracted according to the described methodology (Ausubel et al., 1987) with a modification that includes a step with RNAse treatment as follows: 200 mg/mL, 1.5 h at 37°C. The methodology used for RNA extraction (Chomczynski and Sacchi, 1987) involved a monophasic solution of phenol and guanidine isothiocyanate (Trizol -Invitrogen). The RNA samples treated with DNAse I were purified through the NucleoSpin ® RNA II BD Bioscience kit (Clontech), resuspended in H 2 O DEPC and stored at -80°C.

Synthesis of fluorescent cDNA from total RNA
The construction of fluorescent cDNA for hybridization reactions was by means of a CyScribe cDNA Post Labeling kit (Amersham Biosciences) with 30 mg of RNA and 15 mg of random primers (Amersham Bioscience). The reverse transcriptase reaction occurred at 37°C for 3 h in a programmable thermocycler (PTC-100 Programmable Thermal Controller -MJ Research, Inc.). As the control of the transcriptase reaction, 1 mL of the synthetic RNA from the Lucidea Universal ScoreCard kit (Amersham Biosciences) was used. The reaction was neutralized with 20 mL of 2 M of HEPES, and the cDNAs were purified through precipitation with 3 M of sodium acetate and 75 mL of 100% v/v ethanol, and kept at -20°C overnight. After centrifuging and washing with 70% ethanol, the cDNA was resuspended in 30 mL of CyDye diluted in 0.1 M of bicarbonate of soda, pH 9.0. The sample was kept in the dark at 25°C for 1 h, and the labeling reactions were stopped by the addition of 15 mL of 4 M of hydroxylamine for 15 min at 25°C. The sample was then resuspended in 400 mL of a TE buffer (10 mM of Tris-HCl pH 8.0, 1 mM of EDTA) and concentrated in a Microcon-type column -YM30 (Millipore). The efficiency of the reading was monitored by measuring absorption at 260 nm (for DNA concentration), 550 nm (for Cy3) and 650 nm (for Cy5).

Amplification of X. fastidiosa genes
Primers were used to amplify the 2,600 ORFs of the X. fastidiosa genome. These primers were built for both forward and reverse directioning, ranging from 16 to 19 nucleotides in length and from 48°C to 57°C in Tm (melting temperature). PCR reactions were done in a total volume of 100 mL containing a PCR buffer (50 mM of KCl, 200 mM of Tris-HCl pH 8.4), 2 mM of MgCl 2 , 10 mM of dNTP, 2 U of Taq DNA polymerase (all from Invitrogen), 5 pmol of each primer, 60 ng of genomic DNA and pure sterile water to complete the volume. The reactions were performed in a programmable thermocycler (PTC-100 Programmable Thermal Controller-MJ Research, Inc.), the following programs being used: 94°C for 2 min, 35 cycles (94°C for 1 min, 58°C for 1 min and 72°C for 1 min and 30 s) and a final step at 72°C for 5 min. All the products were analyzed by electrophoresis in agarose gels of 1.5% w/v in TEB containing 0.5 mg/mL ethidium bromide. Amplifications were considered successful when only one product was visualized and when the size of the expected product varied from 300 to 1000 bp.

Microarray construction
The amplified products were suspended in 50% v/v DMSO in a final concentration of 100 to 300 ng/mL and arranged in duplicate at a distance of 250 mm in glass slides treated with aminosilane (Corning). Printing of microarrays was done by a robot model GMS 417 Arrayer (Affymetrix Inc.). After printing, the DNAs were rehydrated (42°C for 10 s), dried (70°C for 1 min) and fixed in a UV camera cross-link (1300 x 100 mJ cm 2 ). The slides were kept at 70°C for 2 h and then stored under vacuum at room temperature. Genetically distant negative controls were also included in this array, these consisting of human genes (pHUM1 and pHUM7) and plant genes (707050B11 -Rubisco), as well as synthetic controls from various species (human, mouse, Arabidopsis spp., Archaeabacteria and E. coli) obtained by the Lucidea Universal ScoreCard kit (Amersham Biosciences).

Hybridization and washing
Hybridization and washing were carried out in a GeneTac Hybridization (Genetic MicroSystems) device. Initially, slides containing the microarrays were denatured at 65°C for 5 min. A solution containing 8 mL of blocking liquid (Amersham Biosciences RPN 3601), 19 mL of SSC 20x, 5.5 mL SDS 2% w/v and 100 pmol of cDNA marked with the fluorescent dyes Cy3 and Cy5, totaling 110 mL, was denatured at 95°C for 2 min, deposited on the slide and kept for 12 h at 42°C. After hybridization, the slides were washed at 25°C in the following solutions: 2x SSC/SDS 0.5% w/w, 0.5x SSC and 0.05x SSC. All washing-steps consisted of 10 cycles with 10 s of solution flux and 20 s of incubation. The slides were then dried for 15 min and submitted to fluorescence detection.

Image acquisition and data analyses
The slides were submitted to fluorescence reading in a model GMS 418 Arrayer Scanner (Affymetrix Inc.) under different wavelengths -550 nm (Cy3) and 650 nm (Cy5). The location and identity of each gene on the slide was defined in a text file, created with the aid of the Clone-Tracker2 program (BioDiscovery). The signal was quantified through ImaGene software (v. 4.1, BioDiscovery), in which two images from the Cy3 and Cy5 fluorescent dyes were overlapped and the spots classified according to morphology and intensity. The computer displays an electronic symbol as a false-color image where a red or green spot corresponds to expression of a gene in sample 1 or 2, respectively, while a yellow-orange spot indicates that the gene was expressed at similar levels in both samples. For transformation of data, the background signal was discounted from the signal of each spot using the local background obtained by the GeneSight (BioDiscovery) program. The transformation sequence included background correction, omitted flagged spots, combined repli-cates and floor, by adding a shifted Log transformation and ratio.
Data obtained from the intensity ratio of the signal measured by the Cy5 (experiment) over the Cy3 (control) were normalized according to the average intensity of the total signal. We used all the genes in our dataset to calculate normalization since this assumes that the majority of the measured genes were not differentially regulated. The normalization procedure is a suitable approach for minimizing variations so that a common base for comparison is established. There are a number of reasons that justify data normalization, these including unequal quantities of starting RNA, differences in labeling or detection efficiency among the different fluorescent dyes used, and systematic bias in measured expression levels (Quackenbush, 2002). However, current normalization methods are not applicable to all conditions. Normalization can be carried out in several different ways, such as within the slide in order to adjust dye incorporation efficiency, between two slides for dye swap experiments and across slides for repetition of the same experiments (Yang et al., 2001). In the latter case, application would be to the entire data set (overall normalization), instead of to a particular physical data subset or sub grid (local normalization).
Final intensity of hybridization was determined in all the experiments from six replicates per data point, and is representative for three independent determinations (slides) from each media culture. Replicates in duplicate within each slide were combined by the median of their values, whereupon statistical analysis was carried out using the SAM method (Significance Analysis of Microarrays). This method is based on t-test statistics and is employed to calculate the false discovery rate (FDR) and gene error chance (q-value) (Tusher et al., 2001). Significant variations in expression of those genes related to X. fastidiosa metabolism were compared when cultivated in liquid modified BCYE and liquid XDM 2 median.

Detection of cDNA by micro-chip electrophoresis
The reverse-transcription step for generation of cDNA was performed in a final volume of 20 mL using 1 mg of total RNA digested with 0.5 mM DNAse I. The random primer (1 mM) and digested RNA were denatured for 5 min at 70°C, then immediately cooled on ice for another 5 min and added to a 15 mL RT mix containing 2.0 mM of dNTPs, 3 mM of MgCl 2 , 1x RT buffer and 1 mL of ImProm-II RT (all from Promega, Madison, WI, USA). The mixture was incubated for 5 min at 25°C, 60 min at 40°C, and 15 min at 70°C. PCR reactions were set up in 10 mL total reaction volume containing 1x PCR buffer, 2 mM of MgCl 2 , 10 mM of dNTP, 2 U of Taq DNA polymerase, 5 pmol/L of each primer (Table 1) and 1.5 mL of cDNA. The reactions were performed in a programmable thermocycler (PTC-100 Programmable Thermal Controller-MJ Research, Inc.) under the following conditions: 94°C for 2 min, 35 cycles (94°C for 1 min, 58°C for 1 min and 72°C for 1 min and 30 s) and a final step at 72°C for 5 min. All products were analyzed by using the Bioanalyzer 2100 (Agilent Technologies, Waldbronn, Germany) in conjunction with the LabChip DNA 500 kit, according to manufacturer's instructions.

Synthesis of fluorescent labeled cDNA
In order to verify gene expression differently, we analyzed the growth of X. fastidiosa from four days after culture in two different liquid media, modified BCYE and XDM 2 (Table 2). Based on the genetic analysis of the X. fastidiosa genome, Lemos et al. (2003) developed certain media with a defined composition, whereby the growth abilities of these bacteria were evaluated in both liquid media and on solid plates. X. fastidiosa growth was compared in XDM 2 defined media as well as in BCYE by measuring cell turbidity and protein content for 14 days at 28°C under shaking conditions. The authors observed that, after 14 days, the growth rate of bacteria on complex media, such as BCYE was substantially lower than in XDM 2 , and that, in the latter, the strains grew equally well both in liquid and on solidified media. However, after four days (96 h) cell turbidity and protein content were similar for both.
Preparation of fluorescent labeled cDNAs was carried out by total RNA extraction from X. fastidiosa growth after four days in liquid modified BCYE and liquid XDM 2 media, its concentration being determined by absorbance measurement at 260 nm (A 260 ). RNA integrity was checked by formaldehyde agarose gel electrophoresis, where the occurrence of two ribosomal subunit bands (23S and 16S containing of 2.9 and 1.5 kb, respectively) was examined (data not shown).
Fluorescent labeled cDNAs were prepared from total X. fastidiosa RNA by reverse transcription. Total RNA was used since most of the mRNAs produced by bacteria do not have a poly (A) tail and are difficult to separate. Labeling efficiency by reverse transcription depends on incorporation efficiency and on the amount of specific nucleotides present in a particular mRNA species. The labeling kit used was developed as a two-step procedure. The first step in- Travensolo et al. 343 Simpson et al. (2000); b Rate of fluorescence (RFU) obtained from replicate media; c Log expression ratios of measured transcript levels determined for the two cultures. The log expression ratio is positive for genes that were more highly expressed in BCYE medium. The log is negative for genes that were more highly expressed in XDM 2 medium; d Genes that expressed only one condition and e Coefficient of Variation (CV). Glucose ( volves the incorporation of amino allyl-dUTP (AA-dUTP) during cDNA synthesis by using an optimized nucleotide mixture. The second step involves chemically labeled amino allyl-modified cDNA using CyDye NHS-esters. Coupling reactions of amino allyl-modified cDNA were performed separately with Cy3 and Cy5 and both targets were combined in the hybridization solution. The amount of target used for hybridization depends on array format and labeling method. Targets containing 100 pmol of incorporated fluorescent dye were employed. Such an amount was calculated from the formulas, (OD 550 x dilution factor x total volume)/0.15 for Cy3 and (OD 650 x dilution factor x total volume)/0.25 for Cy5, where the obtained values are in pmol.
Arraying amplified X. fastidiosa genes DNA arrays were developed through the synthesis of 2,600 amplicons using pairs of primers related to each of the X. fastidiosa genome ORFs. Amplified ORFs were set in a concentration varying from 100 to 300 ng/mL, with fragment-size also varying from 300 to 1,000 bp. The DNA arrays were composed of amplicons that did not need to be purified. The spots were printed in duplicate with a 250 mm distance between each, to a total of 5,200 spots, including positive and negative controls. Studies in microarray gene expression analysis, using unpurified amplified products, emphasized non-significant differences between purified and unpurified PCR products, showing a low alteration level in the hybridization signal (6%) in the latter, when compared to the purified version (Diehl et al., 2002).
The comparison of the expression of X. fastidiosa 9a5c genes when isolated from bacteria cultivated in the two different media (liquid modified BCYE and liquid XDM 2 ) was carried out by using SAM software which develops a statistical evaluation of probes' hybridization patterns. The significance of gene expression differences was calculated by the ratio of median fluorescence intensities for each condition, having as parameters a ratio difference of at least 1.5x together with a threshold D of 0.49514. Missing data points were estimated with a K-Nearest-Neighbor imputator equal to 10. The test was undertaken with response format type paired data with a false discovery rate (FDR) of less than 0.5%. Data analysis resulted in a 0.42 false positive rate (FSN -False Significant Number) and 0.31 of false discovery genes, thus demonstrating that 99.69% of our experiments present positive results and only 0.31% are false. Among the analyzed genes, approximately 5.15% (134) were detected as differentially expressed in both studied conditions. 30 of these (22.4%) showed higher expression in the BCYE medium and 104 (77.6%) in the XDM 2 (Table 3). All gene-chip data can be found in the Gene Expression Omnibus (GEO) Repository under accession number GPL7554.
According to the results obtained through microarray analysis, bacteria cultivated in XDM 2 medium expressed a higher number of significant genes than those cultivated in BCYE modified medium. This was expected, since the XDM 2 defined medium offers a smaller variety of nutrients than the BCYE complex medium. These differences in gene expression patterns were analyzed in detail, as described below.
The functionality of the glycolytic pathway in X. fastidiosa was evaluated (Facincani et al., 2003) by studying the enzymes phosphoglucose isomerase, aldolase, glyceraldehyde-3-phosphate dehydrogenase and pyruvate kinase from the glycolytic pathway, and glucose-6-phosphate dehydrogenase from the Entner-Doudoroff, followed by cloning and expression studies of the enolase gene and determination of its activity. These studies showed that X. fastidiosa does not use the glycolytic pathway to metabolize carbohydrates. As a result, no enzymatic activity was detected for enolase, aldolase and glyceraldehyde-3-phosphate dehydrogenase, this suggesting that X. fastidiosa may be using the Entner-Doudoroff pathway to produce pyruvate as an alternative. Nevertheless, an increase in gene expression of those enzymes related to the glycolytic pathway in the cultivated cells was detected through microarray analysis, this regardless of the supporting XDM 2 medium. This set of genes codes for 6-phosphofrutokinase, glyceraldehyde-3-phosphate dehydrogenase and triosephosphate isomerase (data collected from the cells raised on XDM 2 ), as well as for pyruvate kinase (data collected from cells raised in a BCYE modified medium).
In this work we observed the expression of dihydrolipoamide acetyltransferase, malate dehydrogenase, oxoglutarate dehydrogenase, fumarate hydratase, aconitate hydratase 2, electron transfer protein azurin I and ATP synthase, which act in the reduction of pyruvate, the citric acid cycle, electron transportation, and the production of ATP. The presence of certain TCA cycle enzymes and of the respiratory chain contributes to the hypothesis that X.

344
Xylella fastidiosa gene expression fastidiosa uses cell respiration to obtain energy from glucose.

Transport related genes
A total of 140 genes that code for proteins related to the transport of a number of biological molecules were identified in the X. fastidiosa genome, these representing 4.8% of all ORFs (Simpson et al., 2000). Among these genes, 19 were considered differentially expressed in this comparison, 14 being detected as higher expressed in XDM 2 medium conditions and five in BCYE. These genes refer to the transport of anions, cations, carbohydrates, peptides, proteins and substances related to secretory pathways. Genes involved with the secretion of peptides and proteins (xpsH and secY in XDM 2 medium, xpsD and secA in BCYE) were expressed in both media, these four being related to general secretion and Sec systems. The xpsH and xpsD genes expressed code for external membrane proteins that act in the General Secretory Pathway (GSP) Type II. The secreted enzymes in this pathway include polygalacturonate lyase, endoglucans, aamylase and proteases (Gough et al., 1988;Hu et al., 1992). On the other hand, in gram-negative bacteria such as X. fastidiosa, macromolecules, which include excreted enzymes, toxins and structures from the surface of the cell, need to pass through both the internal and external membranes before reaching the surface of the cell (Hu et al., 1998;Fekkes and Driessen, 1999). The Sec system involves an integral membrane heterotrimer, SecYEG, also known as the translocation complex, which acts together with a homodimeric protein, SecA, which is ATP-dependent. A characteristic of this mechanism is that proteins are translocated in extended conformation, and are frequently bound to SecB or another cytoplasmic chaperonin for proper folding (Berkes et al., 2000).
Most bacteria have other secretory pathways that are distinct from the Sec apparatus (Weiner et al., 1998). One of these Sec-independent pathways was named the TAT system (Twin Arginine translocation system) (Sargent et al., 1998), due to precursors activating the pathway through a signal peptide that includes two consecutive arginine residues. The characteristic of the TAT pathway is that it works to transport folded proteins of various sizes through the cytoplasmatic membrane (Berkes et al., 2000). The tatD gene, which is cotranscribed with tatA, tatB and tatC, was expressed under BCYE medium conditions, but apparently does not have any effect on translocation of those proteins containing arginine residues, since it codes for a cytoplasmatic protein with DNAse activity with no discernible role in tat translocation (Wexler et al., 2000).
Other more expressed genes in the XDM 2 medium were malG, ynhE, yecS, algS, yheS and ccmA, which code for proteins belonging to the ABC transport system. This secretory system depends on the mediation of ABC proteins, consisting of three cell wall proteins, two internal membrane proteins and an external membrane polypeptide (Binet et al., 1997). The malG and algS genes are related to the ABC sugar-transportation system. These two genes were significantly expressed in cells cultivated in the XDM 2 medium, in which glucose was found at a concentration of 10 g/L, as compared to cells cultivated in BCYE medium, with no glucose at all . Glucose is the only carbon source found in XDM 2 which is transported and used for energy production within X. fastidiosa cells, and whose intermediary compounds are involved in glycolysis, the citric acid cycle and the electron-transportation chain, since this microorganism has all the genes related to such energy-associated cycles.

Genes involved in membrane components and surface structures
Six genes that code for proteins related to fimbriae were expressed in both analyzed cultivation media: (XF2542, mrkD, pilU, pilP and pilY-1 in the XDM 2 medium and pilQ in the modified BCYE medium). The pilP, pilY-1 and pilQ genes are related to type IV fimbriae involved in fimbriae biogenesis, whereas pilU and XF2542 are supposedly responsible for fimbriae retraction and extension, a mechanism known as twitching motility. XF2542 is similar to the subunits of Xanthomonas spp. and Pseudomonas spp. type IV fimbriae. A specific gene expression regulation mechanism of type IV fimbriae was observed in different cultivation conditions, suggesting that this is an important factor for X. fastidiosa survival (Smolka et al., 2003).
In the XDM 2 medium, transcripts of mrkD (fimA family) genes involved in the adherence of X. fastidiosa bacteria were detected. This protein is similar to others found in a number of bacterial species that infect plants, animals and human beings (Ojanen-Reuhs et al., 1997). It is considered to be a key mediator for adhesion and mobility, being an important virulence factor. However, isolated specimens with mutations in fimAand fimFbecame pathogenic when inoculated into vine plants (Feil et al., 2003). Hitherto, these two genes have been discarded from involvement in the mechanisms of pathogenicity.
As far as membrane components are concerned, four genes were found expressed in the XDM 2 medium (dc-14, XF0881, mreB and murD) and only one in the BCYE modified medium (mopB). These genes were related to proteins of the internal and external membrane, besides cell wall biogenesis. The mreB and murD genes code for proteins linked to the production of peptidoglycan, which is the main component of bacterial cell walls, and consists of the heteropolymers of N-acetylglucosamine and N-acetylmuramic acid.
The X. fastidiosa mopB gene is very similar to the OprF porin protein that belongs to the OmpA family from Pseudomonas spp., and which is involved in xylem endophytic bacteria growth and survival ability in low osmo-larity niches (Rawling et al., 1998). The mopB gene can be pinpointed as an interesting target, since X. fastidiosa survives in a low osmolarity environment when inside xylem vessels.

Genes involved in RNA, DNA and nucleotide metabolism
The analyses revealed higher gene expression in bacteria cultivated under XDM 2 conditions for those genes related to RNA and DNA metabolism (vacC, metG, holA, holB, recG and mutY). On the other hand, for bacteria cultivated in a modified BCYE medium, only gltX, ilaIIA and tyrS genes were considered as showing significant and high expression levels (Table 3). Furthermore, X. fastidiosa showed a higher rate of cell multiplication, when grown in XDM 2 medium than in, modified BCYE medium . This is in accordance with the levels of expression of those genes related to nucleic acid metabolism, since a larger number of these genes were expressed under XDM 2 conditions. Thus a larger number of ribosomes and a higher speed of protein synthesis were observed for accelerated cell division cycles (Grunberg-Manago, 1996), as happens in the XDM 2 medium. The significant expression difference for X. fastidiosa genes, when cultivated in the XDM media series, was essentially related to the production of ribosomal proteins (Nunes et al., 2003). This defined medium presents only glycerol and glutamic acid in its composition, or rather the XDM 2 precursors used in this paper. The authors suggest that the majority of X. fastidiosa genes may be under the control of constitutive promoters, which are induced under nutrient limiting conditions, this representing an important step towards the adaptation of such a bacterium to the adverse conditions found within the xylem vessels of infected plants. 346 Xylella fastidiosa gene expression   Three ORFs related to nucleotide biosynthesis were expressed in the XDM 2 and modified BCYE media: phosphoribosylaminoimidazole carboxylase (purE), gluconolactonase precursor (SCF 11.04) and 5-phosphoribosyl-5-aminoimidazole synthetase (purM). The purE and purM genes are responsible for synthesis of purine ribonucleotides, while SCF11.04 acts on the biosynthesis of nucleosides. This reaction is part of purine biosynthesis, starting with the metabolic precursors, ribose-5-phosphate, CO 2 and NH 3 . All the pathways for synthesis of purinic and pyrimidinic nucleotides have already been described for X. fastidiosa (Simpson et al., 2000).

Genes involved in the biosynthesis of amino acids and proteins
Through the analysis of those genes related to amino acid biosynthesis, it was possible to observe that X. fastidiosa is able to synthesize certain amino acids such as aspartate, cysteine, glutamate, histidine and metionine. Most microorganisms can uptake amino acids from their cultivation medium and oxidize them to sustain energy levels, as required by metabolic conditions (Nelson and Cox, 2002). X. fastidiosa presents high biosynthetic capacity, this probably resulting from its success in colonizing the xylem vessels of a number of host plants (Simpson et al., 2000). However, xylem fluid contains a low concentration Travensolo et al. 349 (Tusher et al., 2001), b Log expression ratios of measured transcript levels determined for the two cultures; c The log expression ratio is positive for genes that were more highly expressed in BCYE medium and d The log is negative for genes that were more highly expressed in XDM 2 medium. of organic composts (available energy sources), although it presents a high concentration of amino acids such as glutamine and asparagine (Raven, 1984). Glutamine and arginine are important in the composition of the XDM 2 medium, as sources of nitrogen and in helping X. fastidiosa cells to reach the end of their exponential growth phase in less generation-time . Genes related to amino acid biosynthesis were found in both culture media at various expressed levels. Two implications arise from this analysis. The first confirms that the TCA cycle is active, since it generates the intermediaries for amino acid biosynthesis from the glucose oxidative degradation pathway. The second implication is that the source of amino acids in both media, mainly in XDM 2 (which contains arginine, glutamine, metionine and serine), can be used in protein synthesis as well as for supplying the carbon skeleton: a) to replace intermediaries of TCA cycle components in anaplerotic reactions and b) for synthesis of the other amino acids.
The operon sspA-sspB expression in the XDM 2 medium was similar to that observed in E. coli during the stationary phase of the growth curve and under carbon, amino acids and phosphate limiting conditions (Willians et al., 1994). This operon expression level during the four days of X. fastidiosa cultivation shows that the active metabolism of the bacterial cells in the XDM 2 medium results in the consumption of nutrients up to cells entering the stationary growth phase.
Analyses revealed higher gene expression for the pspB gene which codes serine protease, in bacteria cultivated in the XDM 2 medium (Table 3). Serine protease is not secreted via a type I pathway, but belongs to the autotransporter family of secreted proteins (Chabeaud et al., 2001). Many proteins belonging to the autotransporter family are involved in adhesion or auto-aggregation, even though several of them possess the active-site motif of serine protease (Henderson et al., 1998).

Genes involved in cofactors, prosthetic groups and regulatory functions
Through genomic analysis, it was suggested that vitamins like thiamin, biotin, nicotinic acid and pyridoxine are synthesized by X. fastidiosa cells (Simpson et al., 2000). The significant expression of genes related to the prosthetics groups, cofactors and vitamins, such as those involved with riboflavin (ribA and ribD), biotin (bioA and bioB), pantothenate (panC), porphyrin (hemB and hemF), folic acid (folK) and thiamin (thiL), may suggest that the concentrations used in media composition were insufficient to sustain growth, and that the decrease in these levels was responsible for their synthesis. On the other hand, the XDM 2 medium containing only biotin as a vitamin source and in a 0.2 mg/L concentration was efficient enough to maintain bacterial cell growth (data not shown). The regulatory functions category presented six highly expressed genes in the XDM 2 medium (sspA, sspB, tctD, algR, colR, pilR) in comparison to only one gene (phoR) in the BCYE modified medium. Since the functions of most of these systems are still unknown, it is believed that the organisms in which they are expressed may show a higher level of adaptive answers to certain environmental changes, situations in which the two component systems are induced (Stock et al., 1989).
The phoR gene is induced by phosphate limitation (Hullet, 1996), being significantly expressed only in the BCYE modified medium which does not include ferric pyrophosphate, a possible source of phosphate, in its composition. The significant expression of pho genes clearly indicates the need to include other sources, for this bacterium to make use of phosphorus for growth.

Genes involved in molecule degradation
Four expressed genes were related to the category of degradation, with emphasis on est (esterase), lipP (lipases) and dhaA (haloalkane dehalogenase). A correlationp between lipase concentration and the production of biofilms might be linked to adhesion and construction of the latter (Smolka et al., 2003). Lipases hydrolyze ester bonds between the insoluble triacilglycerides interface and the aqueous phase where the enzyme is dissolved (Anthonsen et al., 1995). In Candida albicans, LIP family lipases are expressed and eliminated during the infection cycle, and it is believed that they contribute to the survival and virulence of this organism in human tissues (Hube et al., 2000).

Genes related to hypothetical and conserved proteins
Approximately 30% (40 genes) of the 134 differentially expressed genes did not present homology with sequences deposited in GenBank. The involvement of hypothetical and conserved proteins, for which functions in other organisms have not yet been described, should be taken as an indication of significant differences in the metabolism of this phytopathogen.

Categories of genes expressed only under XDM 2 cultivation conditions
The expression of genes related to the toxin categories (frpC, gaa and cvaB), functions related to plasmids (taxC), phago (int), cell division (ftsW) and others (bcp), were only observed in XDM 2 medium growth conditions. Among these categories, toxins that can be depicted belong to the RTX family (gene frpC) and the production of colicin V (gene cvaB). The cvaB gene from E. coli, together with two other genes, cvaA and tolC, mediates the elimination of colicin V (ColV) to the outer part, so as to act within target cells, thereby destroying them by membrane rupture (Zhong and Tai, 1998). 350 Xylella fastidiosa gene expression

Categories of genes expressed only in BCYE modified cultivation conditions
The expression of the fadG gene can be assigned to biosynthesis of fatty acids and was observed to be high under the BCYE modified medium conditions. The E. coli fab genes presented higher expression levels in a rich medium, thereby suggesting that the regulation of phospholipid biosynthesis genes might be dependent on the speed of growth, since these genes need a higher number of membrane compounds (Tao et al., 1999). However, studies with X. fastidiosa demonstrated that growth was higher in the XDM 2 medium than in the BCYE during the 14-day period . Thus, it is possible that fab gene expression is mediated by one or more signal molecules found in the modified BCYE medium.

Detection of cDNA by microchip electrophoresis
In order to validate the results from microarrays, six ORFs (XF0358, XF0671, XF0785, XF1426, XF1937 and XF2688) were analyzed for RT-PCR by using microchip electrophoresis. The cDNAs from two different conditions (XDM 2 and BCYE media) were synthesized and used in PCR with specific primers (Table 1). The Agilent 2100 Bioanalyzer separated the RT-PCR amplicons and quantified the expression of each gene. As expected, the ratios observed in this experiment were similar to those obtained through the microarray technique. Under XDM 2 conditions, the ORFs XF0358, XF0785 and XF2688 showed a higher expression in this medium. Similar results were shown in the BCYE medium for ORFs XF1937, XF1426 and XF0671 (Figure 1). It is important to stress that the coefficient of variance between the replicates for each gene under the conditions studied was seen to be between 2.68 and 5.19%, demonstrating the high level of reliability of the results (Table 1).