PR gene families of citrus : Their organ specific-biotic and abiotic inducible expression profiles based on ESTs approach

In silico expression profiles, of the discovered 3,103 citrus ESTs putatively encoding for PR protein families (PR-1 to PR-17), were evaluated using the Brazil citrus genome EST CitEST/database. Hierarchical clustering was displayed to identify similarities in expression patterns among citrus PR-like gene families (PRlgf) in 33 selected cDNA libraries. In this way, PRlgf preferentially expressed by organ and citrus species, and library conditions were highlighted. Changes in expression profiles of clusters for each of the 17 PRlgf expressed in organs infected by pathogens or drought-stressed citrus species were displayed for relative suppression or induction gene expression in relation to the counterpart control. Overall, few PRlgf showed expression 2-fold higher in pathogen-infected than in uninfected organs, even though the differential expression profiles displayed have been quite diverse among studied species and organs. Furthermore, an insight into some contigs from four PRlgf pointed out putative members of multigene families. They appear to be evolutionarily conserved within citrus species and/or organor stress-specifically expressed. Our results represent a starting point regarding the extent of expression pattern differences underlying PRlgf expression and reveal genes that may prove to be useful in studies regarding biotechnological approaches or citrus resistance markers.


Introduction
Plants evolved mechanisms that enable them not only to resist drought and wounding but also to oppose attacks by pathogenic microorganisms.One of the ways that plants respond to biotic and/or abiotic stress factors is in the accumulation of pathogenesis-related proteins (PR proteins) (van Loon and van Strien, 1999).Plant PR proteins are defined as proteins encoded by host plants that are induced in pathological or related situations, and represent major quantitative changes in soluble protein during the defense response (van Loon et al., 1994;Stintzi et al., 1993).Originally described in tobacco leaves upon virus infection, PR proteins were first classified into PR-1 to PR-5 families, based on serological properties and later on sequence data.They generally have two subclasses: an acidic subclass, secreted to cellular space, and a vacuolar basic subclass (van Loon and van Kammen, 1970;Stintzi et al., 1993;Kitajima and Sato, 1999).Subsequently, several other protein groups were recommended for inclusion into this class.Now plant PR proteins comprise a large group of 17 protein families, even though PR-15 to PR-17 families have been recognized only recently (van Loon et al., 1994;van Loon and van Strien, 1999;van Loon et al., 2006).
Plant protection against nematode and herbivorous insect attacks has been associated with members of the PR-6 protein family, which are proteinase inhibitors (Ryan, 1990).The PR-7 proteins are endoproteinases which act as an accessory to antifungal activity in cell wall dissolution (van Loon and van Strien, 1999).A very interesting role has been ascribed to members of the ribonuclease-like PR-10 protein family, apparently the only PR protein family possessing antiviral activity (Somssich et al., 1986;Zhou et al., 2002).The PR-9 proteins are lignin-forming peroxidase with peroxidase activity, implicated in the oxidative crosslinking of plant cell wall components in order to prevent pathogen penetration (Lagrimini et al., 1987(Lagrimini et al., , 1997)).The remaining PR-like proteins, classified as PR-15 and PR-16 protein families, are pathogen-induced germin-like oxalate oxidases and germin-like oxalate oxidase-like with superoxide dismutase activity.These are thought to be involved in signal transduction pathway for the regulation of HR (Hypersensitive Response) as the members of PR-17 protein family that are possible proteolytic enzymes presenting sequences for a putative active site of zinc-metalloproteinases, pathogen-induced transcript and protein accumulation patterns (Zhang et al., 1995;Wei et al., 1998;Okushima et al., 2000;Zhou et al., 1998;Christensen et al., 2002;Park et al., 2004;van Loon et al., 2006).
High levels of PR gene expression during local HR and systemic plant defense (Systemic acquired resistance, SAR) have been suggested as markers for both defense responses (Ahl Goy et al., 1982;Lawrence et al., 1996;Tornero et al., 1997).Signaling molecules mediate induction of PR proteins in plants during pathogen infection including SA (salicylic acid) for acidic PR genes as well as ethylene and methyl jasmonate for basic PR genes (Kitajima and Sato, 1999).In addition, PR genes (basic in gen-eral) also are present constitutively in some plant organs or tissues, including roots, leaves and floral tissues.Citrus PR gene families have been poorly reported so far.Citrus chitinase and glucanase proteins were associated with fruit development (McCollum et al., 1997) and pathogen response (Porat et al., 2001;Porat et al., 2002;Fanta et al., 2003;Porat et al., 2003) as well as with constitutive expression (Recupero et al., 1997).PR gene expression in citrus was shown to be promoted by hot water (Pavoncello et al., 2001), UV irradiation and wounding (Porat et al., 1999) and ß-aminobutyric acid (Porat et al., 2003).Moreover, Fagoaga et al. (2001) has reported that a tobacco PR-5 protein, constitutively overexpressed within Citrus sinensis transgenic plants, confers enhanced resistance towards the Phytophthora citrophthora pathogen.This suggests that PR proteins can also be used successfully in citrus genetic engineering approaches.
Citrus is the main fruit crop in the world and, as such, an important commodity.For this reason, the sequencing in large scale of expressed sequence tags (ESTs) from citrus organs was performed as an approach to fill some of the gaps in knowledge concerning the genetic and molecular factors involved in several citrus diseases and fruit development.cDNA libraries were constructed using tissues from different plant organs, such as leaf, flower, fruit, bark, seed and root, from twelve citrus species (Citrus sinensis, C. limonia, C. reticulata, C. aurantium, C. limettiodes, C. aurantifolia, C. sinensis x C. reticulata, C. sunki, C. latifolia, C. reshni, Citrumelo swingle, Fortunella margarita, Poncirus trifoliata).These plants were submitted to diverse situations of biotic stresses caused either by bacteria (Xylella fastidiosa), viruses (Citrus leprosis virus, Citrus tristeza virus) or Phytophthora parasitica, or abiotic stress caused by environmental factors, such as drought.Therefore, Brazilian citrus genome EST database (CitEST) covers a wide diversity of gene sequence information for the study of components of citrus defense response pathways to pathogen, wounding and other abiotic stresses, which have often been used as marker resistance onset.In addition to providing an efficient method for gene discovery, the ESTs data set can also provide information about gene expression.However, the challenge is to extract biological knowledge from large amounts of gene expression data deposited in databases.
Tools for in silico analysis of the gene expression allow comparison of expression profiles of specific genes in plant tissues, based on the frequency of sequence tags in cDNA libraries.According to Ohlrogge and Benning (2000), gene expression analysis is based on the rationale that an abundance of mRNAs synthesized from a particular gene highly expressed in a given tissue can be estimated by the counting of the number of ESTs corresponding to the cDNA of this gene, which is present in cDNA library constructed from the tissue.A natural basis for organizing gene expression data is to group together genes with similar pat-terns of expression (Eisen et al., 1998).In plants, hierarchical clustering has been used routinely to identify genes highly expressed from rice (Ewing et al., 1999), sugarcane (Lambais, 2001), soybean (Shoemaker et al., 2002), barley (Zhang et al., 2002), wheat (Ogihara et al., 2003) and eucalypt (Domingues et al., 2005) transcriptome; and it is expected that this number will continue to increase.A combination of clustering methods with a graphical representation of the primary data, by representing each data point with a color, may quantitatively and qualitatively reflect the original experimental observations and allow an understanding of the data in a naturally intuitive manner (Eisen et al., 1998).
By using this strategy, we explored the CitEST database to analyze the expression of the putative PR gene families in citrus plants, since the proteins encoded by them have been described as associated with all of the different conditions cited for citrus library construction.Albeit in silico, the results here presented may provide insights about the expression differences underlying PR gene family expression patterns and reveal genes that may prove to be useful in studies regarding biotechnological approaches or citrus resistance markers.Furthermore, a similarly compiled study concerning expression profiles from all of the recognized 17 PR-like gene families has never been reported for citrus before.

Identification and in silico expression analysis of citrus PR-like homologous ESTs
Search analyses by comparison were performed within the CitEST database (http://citest.centrodecitricultura.br)against amino acid sequences of members (preferentially 'type member') of each pathogenesis-related (PR) protein family (http://www.bio.uu.nl/~fytopath/PRfamilies.htm), which were found in public databases, in attempts to identify homologous sequences in citrus.Additionally, searches using keywords and more than one amino acid sequence (from different PR isoforms or classes) in BLAST were also used for each PR family.Afterwards, by using the Basic Local Alignment Tool (blastx) program (Altschul et al., 1997) with the cut-off value of e-10 -05 and BLOSUM62 matrix criteria, a total of 3,103 ESTs-reads were selected from CitEST in order to analyze the PR-like gene expression profiles in citrus cDNA libraries, which cover 173,967 useful reads (Table 1).
As an initial consideration in constructing expression profiles, the frequency of reads in the selected libraries was computed and normalized by the whole number of useful reads from each library, corrected to 1,000 ESTs.Hierarchical clustering was used to group EST-contigs and librar-PR gene families of citrus 919 ies by similarities, displayed in a Cluster program and Tree View software (http://rana.lbl.gov/EisenSoftware.htm).To obtain the hierarchical clustering, the calculation of the distance between all pairs of objects was performed using an un-centered correlation matrix and the pairwise averagelinkage method (Eisen et al., 1998).The reordering of data matrix was performed according to similarities in the pattern of gene expression and graphically displayed as color arrays of EST-contigs, using a color scale representing the number of reads from a particular library in each ESTcontig.
Computer subtraction analyses were also performed by subtracting the stress libraries from non-stress libraries, thus getting the positive, negative and co-regulation of each PR gene family in both abiotic and biotic stress situations.EST-contigs, expressed in non-stressed organs compared to the stressed organs, were calculated as described in Lambais (2001).Graphical representation of each data was highlighted with a color, using a color scale representing the range between suppression and induction gene expression, in relation to the counterpart control library.

Phylogenetic analysis
EST clusters were built only for each PR-2, PR-3, PR-5 and PR-7 gene families separately (Table S1), by alignment using the Contig Assembly program (CAP3) (Huang and Madan, 1999).A consensus sequence from each cluster used in phylogenetic analysis was compared with the amino acid sequences from PR-3 protein homologous sequences deposited in the public GenBank database, using the TBlastN and BlastX algorithms (Altschul et al., 1997).Final alignment was obtained with Clustal X 1.83 (Thompson et al., 1997) and afterwards used for phylogenetic analysis.Phylogenetic analysis of amino acid sequences was performed with the Neighbor joining method and Dayhoff Matrix Model with the substitution method, with 1,000 bootstrap replicates, using the MEGA 3 program for construction and visualization of trees (Kumar et al., 2004).

Description and identification of CitEST cDNA libraries
All citrus sequences used in this work correspond to sequenced EST-reads.Cluster consensi were obtained from the Genetic Breeding, Functional Genome and Comparative of Citrus project (http://citest.centrodecitricultura.br)and derived from cDNA libraries specific for different citrus species, organs or growth and stress conditions, as described in Table 2 and by Targon et al. (in this issue).

Selection of CitEST PR-like gene homologous ESTs and expression profiles in citrus
The use of in silico methods to search homologous sequences of known genes is an important approach for the discovery of new genes.Comparison searches using the blast program (Altschul et al., 1997) and preferentially 'type member' amino acid sequences of each PR protein family as query led to identification of a total of 3,538 citrus PR-like EST-reads with a cut-off value of e -05 in CitEST database (Table 1).This number represents around 2% of the whole CitEST database, which contains about 173,967 ESTs obtained from all citrus plant libraries.
In this study, we reported, for the first time, the presence of homologous ESTs encoding for all of the recognized 17 PR gene families in citrus plants (van Loon and van Strien, 1999;van Loon et al., 2006).A graphical representation of expression profiles of the 17 gene families encoding PR proteins was generated by using 3,103 PR-like EST-reads (Figure 1).The cDNA library construction conditions in which the PR-like EST-reads were isolated are described in Table 2.In a general view, transcripts coding for members of eight PR protein families (PR-2, PR-3, PR-6, PR-7, PR-9, PR-14, PR-15 and PR-16) were found to be highly expressed within 24 to 30 of the 33 citrus libraries studied, whereas members of the PR-13 and PR-17 families were present in only one library each (Figure 1).Even though PR-15 and PR-16 gene families display similar nomenclature and activities, the expression profiles were notably distinct.These in silico expression profiles that were preferentially induced under the different situations indicate conserved functions in citrus species.
Within the libraries, the majority of putative PR protein families encoded by highly expressed citrus transcripts were found in those constructed from healthy organs during normal plant growth, such as in non-drought-stressed roots and flowers, and fruit peel.Transcripts coding for members of eleven of the 17 PR-like protein families (PR-1, PR-2, PR-3, PR-4, PR-6, PR-7, PR-8, PR-9, PR-10, PR-15, and PR-16) seem to be highly expressed within the CL06-C4-500 library, which was constructed from roots of Citrus limonia 'Cravo' non-drought-stressed.Likewise, transcripts coding for members of twelve PR-like protein families (the highly expressed PR-2, PR-3, PR-5, PR-6, PR-7, PR-9, PR-14, PR-15, PR-16 and the moderately expressed PR-1, PR-8, PR-12 families) were found in the healthy flowers of the Citrus sinensis 'Pêra IAC' (CS00-C5-003) library.PR proteins present in apparently healthy organs during normal plant growth are thought to play additional unsuspected roles in morphogenesis or in symbiosis (Datta and Muthukrishnan, 1999).The reason why transcripts encoding members of PR-11, PR-13 and PR-17 protein families were not expressed in these libraries needs to be investigated.
Only members for the putative PR-3, PR-10 and PR-14 gene families were highly expressed in stem bark tissues of Citrus sunki BAG cv.200 RG 23 (TS27-C2-300).Similarly, only PR-3, PR-9 and PR-14 gene families were strongly expressed in leaves of Citrus sinensis cv.Pêra IAC induced by the Citrus leprosis virus (CiLV) (CS00-C1-401).In the case of the former, despite expression of members for only few (i.e., three) PR gene families, this expression pattern may be associated with wounding, given that wounding was mimicked here as a mock control library for the P. parasitica infection library.Interestingly, a very similar experiment carried out with Poncirus trifoliata plants, using the same organs and conditions (PT11-C2-300 library), was able to induce not only transcripts coding for members of these same three PR-families (PR-3, PR-10 and PR-14) but also for members of seven PR protein families extra within Citrus sunki.It is noteworthy that in addition to differences of environmental stimuli, differences between P. trifoliata and C. sunki could also be noted in the constitutive expression pattern of PR genes.This is an indication of an effective defense response to stresses, which appears to be more elaborated in P. trifoliata than in C. sunki organs.Since P. trifoliata is a citrus rootstock plant known to be resistant to several citrus diseases, including P. parasitica and CTV (Citrus tristeza virus) among others (Yang et al., 2003;Siviero et al., 2006), in contrast to C. sunki which is known to be susceptible, these results contribute to an understanding of the defense responses to stresses involved in both plant species.Using real time-PCR, it has already been reported by our group that high quantitative levels of constitutive and timing induced expression of PR genes in P. trifoliata leaves and stem bark tissues, respectively, were associated with resistance to P. parasitica (Teixeira et al., 2005).
In the latter case, the number of members of the three PR gene families expressed in C. sinensis CiLV-induced PR gene families of citrus 921 leaves contrasts with the amount of members for unexpected fifteen PR gene families found within the counterpart uninfected control library (CS00-C1-100).One explanation for this fact could be that C. sinensis plants used to construct these libraries were pre-immunized with a non-virulent CTV strain, about 48 h before CiLV infection, as an attempt to mimic in vivo conditions occurring in the field groves, since that is a standard procedure carried out in our nurseries.Because of this, a constitutive expression profile in leaves could not be inferred for this CiLV susceptible citrus species.It was apparent, however, that C. sinensis susceptible plants have genetic potential for resistance, but the success of the defense may also depend on the pathogen's ability to overcome it.It is possible that preimmunized leaves either provided high PR gene expression profiles, which were suppressed by presence of CiLV, or that a large number of PR gene family transcripts could not be detected within C. sinensis leaves at 48 h.a.i.(hours after inoculation) with CiLV.Further analysis could confirm these hypotheses, using time points larger than 48 h.a.i., and this is the subject of our current research projects.
Interestingly, within P. trifoliata leaves infected with a virulent CTV strain (PT11-C1-901) or uninfected control leaves (PT11-C1-900), members of the same groups of ten PR gene families were apparently expressed in both libraries (Figure 1), even though changes in the relative amount of gene expression can be noted.In this case, a high level of PR transcripts constitutively expressed in the uninfected leaves apparently suggests a possible role preformed barrier.High levels of constitutive expression of PR transcripts have been associated with high levels of natural nonspecific quantitative resistance to pathogens (Ahl Goy et al., 1992;Vleeshouwers et al., 2000).Likewise, within C. reticulata uninfected leaves (CR05-C1-100), high levels of gene expression for members of PR-2, PR-3, PR-7, PR-9 and PR-14 gene families could also represent a putative constitutive expression pattern; however, members of these same five PR gene families were also highly expressed in C. reticulata leaves at 30 and 60 days after infection with X. fastidiosa.Detailed in silico cluster analysis (as discussed from Table S1) accomplished by quantitative experimental methods could provide concrete insights to elucidate if the same or different members, i.e., PR protein isoforms belonging to each of the five PR gene families, could be playing a role in both situations.These analyses could contribute to understanding unclear molecular mechanisms of the resistance of P. trifoliata and C. reticulata to CTV and X. fastidiosa pathogens, respectively.

Differential expression profiles of PR-like gene families within citrus organs upon pathogen infection and drought stresses
The recruitment of different PR genes for conserved functions in response to pathogen and drought stresses in citrus was analyzed based on expression profiles of all of PR-like gene families.For all situations studied, citrus plant samples were collected in specific non-stressed and stressed organs as an effort to determine, by using gene expression patterns, if a relevant putative local response is occurring.This is an important approach for determining if a particular response to infection is truly a defense against pathogens, at the time and location of the stress.Thus, in an attempt to gather more evidence for changes of PR gene activity upon biotic and abiotic stress conditions, the differential expression profiles of pathogen and drought-induced citrus PR ESTs were subtracted from uninfected or nondrought-stressed mock controls, respectively (Figure 2).As a result, unexpectedly, few PR gene families showed relative expression 2-fold higher in pathogen-induced than in uninfected organs, even though the differential expression profiles displayed have been quite diverse among studied species and organs.
It appears that transcripts of PR gene families found have converged from different defense responses triggered against virus pathogens in leaves of different citrus species.In C. sinensis leaves, putative suppression of expression for  most of the PR gene families was observed in the presence of the CiLV, when only the PR-3 gene family showed high expression patterns.Here, the supposed effect of the preimmunization with the non-virulent CTV strain, previously discussed, is thought to have been eliminated when the subtractive analysis was performed, since both uninfected (mock control) and CiLV-infected C. sinensis leaves have been subjected to the same pre-immunization conditions.In P. trifoliata, PR-2, PR-3, PR-15 and PR-16 gene families were highly expressed within leaves after infection by a virulent CTV strain, whereas no expression was found for seven PR gene families.Although PR-10 protein family has been associated with antiviral activity, no expression of PR-10 transcripts was found in P. trifoliata leaves.On the other hand, the differential expression pattern of PR genes, assembled within C. sinensis leaves as part of the defense responses against X. fastidiosa pathogen, include the high expression of PR-7, PR-9, PR-10 and PR-14 gene families at 30 d.a.i (days after inoculation) and of PR-7, PR-10 and PR-11 gene families at 270 d.a.i.These data indicate that the same citrus species appears to trigger different defense responses in leaves against different pathogens.The differential expression pattern of PR genes upon X. fastidiosa infection of C. reticulata leaves was different from these of C. sinensis, including the expression of PR-2, PR-6 and PR-17 gene families at 30 d.a.i.Within the same species, the differential PR gene expression profiles vary between infected and healthy organs as well as varying between different infections caused by different pathogens.For instance, the high expression of the PR-3, PR-15 and PR-16 gene families within P. trifoliata leaves upon CTV inoculation was found to be putatively suppressed in stem bark after P. parasitica infection.It is also possible that PR gene expression profiles may vary among organs and/or different pathogens may lead to induction of different PR protein set.Citrus PR-17 gene family expression was found only in Citrus reticulata leaves induced by X. fastidiosa.This data suggests that the PR-17 gene family might be involved in a defense response of C. reticulata species to X. fastidiosa infection.Molecular mechanisms involved in C. reticulata resistance to X. fastidiosa were discussed by Souza et al. (in this issue).Moreover, high expression of the PR-7 family was observed only within leaves of both C. reticulata and C. sinensis upon X. fastidiosa infection, whereas the PR-11 gene family was observed within leaves of both species upon X. fastidiosa infection and also in C. limonia drought-stressed roots.Taken together, members of PR gene families identified with altered expression by the presence of a particular pathogen may participate in an effective response against these pathogens, as a component of a highly specialized signaling pathway against pathogen infection.Induction of PR-1 genes is typical of the onset of SAR.However, PR-1 gene family expression observed here was not significantly found within any pathogen infection condition.No expression of PR-13 gene family members was observed within any stressed or non-stressed organ.
Members of PR-3, PR-11, PR-12, PR-15 and PR-16 gene families showed expression 2-fold higher in droughtinduced roots than in non-stressed roots of C. limonia.Suppression in expression for members of six PR gene families, which were preferentially expressed in non-stressed roots, indicates that more PR gene families were repressed than induced in artificial drought-stressed roots.Additional experiments to show the difference in expression between the drought sensitive and the drought resistant cultivars will contribute to elucidate the dynamic response capacity to stress in citrus roots with the participation of PR gene families.

Expression pattern of PR-like gene families within citrus fruits
The expression patterns of PR-like ESTs from CitEST fruit libraries were also analyzed.These fruit librar-PR gene families of citrus 923 ies were constructed from pericarp, using fruits of different diameters (1, 2.5, 5, 7, 8, 9 cm).Interestingly, at least members of nine PR gene families were expressed in fruits of different sizes from two studied citrus species, C. reticulata and C. sinensis, and also, apparently, the same PR gene families were found within both plant species (Figure 3A).With the exception of the PR-1, PR-4, PR-13 and PR-17 gene families, which were not expressed in fruits, members of other PR gene families were present throughout all the different fruit size stages in both species.However, it was possible to observe changes in the relative amount of expression patterns among different fruit stages and also between citrus species.An overview of changes in expression profiles of PR gene families in fruits of C. reticulata and C. sinensis is shown in Figure 3B.The greatest number of PR-like ESTs was found within C. sinensis fruits of 2.5 cm of diameter.Accumulation of defense-related mRNAs in citrus fruits has been studied in the context of fungal and ethylene perception (Marcos et al., 2005).Therefore, the high expression level of different PR gene families observed in fruits of citrus might indicate constitutive or developmental defense responses, as a preformed barrier to pathogen infection or hormone-induced putatively playing unsuspected roles in fruit development.

Organ-specific expression profiles of citrus PR-like gene families
Genome wide transcriptome analysis with histological information can provide insights into candidate genes that are differentially expressed in certain organs.At the transcriptome level, differential expression of PR genes may play key roles in maintaining resistance functions in plants during the development or, similarly, in normal developmental processes.It is worth mentioning that devel-  opmental-induced PRs are accumulated in an organ and tissue-specific manner (van Loon and van Strien, 1999;Edreva, 2005).In order to speculate whether the preferential expression of some PR gene family could be associated with particular organ-specificity within citrus plants, we performed an expression profiles analysis for different organs (Figure 4).The results show that, with the exception of the PR-13 gene family, which was found only in seeds of P. trifoliata, members of the other PR gene families seem to be expressed in citrus leaf tissue.Even though present in all plant organs, PR proteins are particularly abundant in leaves, where they can amount to 5%-10% of total leaf proteins (Edreva, 2005).In this organ, PRs were reported present both in epidermal and mesophyll cells, as well as in the vascular bundles (van Loon et al., 2006).PR-17 gene family expression was observed exclusively within leaves, whereas PR-13 gene family was presented only within seeds among the studied citrus organs.Further molecular and biochemical characterization of P. trifoliata PR-13 EST-contigs will provide information about a putative seed-specific expression pattern.In addition, construction of libraries using seeds of other citrus species will provide the answer as to whether this P. trifoliata PR-13 gene family is present in other citrus seeds.Organ-specific expression of certain PR genes suggests that the proteins also play roles in normal developmental process (Edreva, 2005).However, the PR-17 gene family expressed in C. reticulata was induced only upon X. fastidiosa infection, as shown in Figure 2. It has already been demonstrated for two barley (Hordeum vulgare L.) proteins belonging to the PR-17 family that they accumulated in the mesophyll apoplast following inoculation with Blumeria graminis f.sp.hordei, as well as in the leaf epidermis, the only tissue to be invaded by the fungus (Christensen et al., 2002).Thus, whether the PR-17 gene family expression in leaves could to be associated more closely to the pathogen-induced than an organ-specific pattern needs to be tested, focusing on the C. Reticulata-X.fastidiosa interaction.
Transcripts of PR-4 gene family were found only in leaves and roots.Members of PR-1 gene families were not expressed in citrus fruit peel and seed, whereas PR-8 transcritpts were not found in stem bark and seed organs.Members of four PR gene families were expressed in most of the studied organs, except in root for PR-5 and PR-14, in flower for PR-10 and in seed for PR-15 and PR-16 transcripts.Finally, members of six PR gene families (PR-2, PR-3, PR-6, PR-7, PR-9 and PR-12) were expressed in all of the studied citrus organs: leaf, stem bark, fruit peel, root, flower and seed.

Occurrence of PR-like gene families in citrus species
The occurrence of transcripts putatively encoding PR protein families in all of the studied citrus species was also investigated (Figure 5).Our first question was whether it was possible that all of the recognized PR gene families could be present in a single plant genome, but here we were not able to show this.Regulation of the different PR transcripts requires different stimuli within the same plant; however, the analyzed libraries had not been constructed using several different organs and stimuli for a single citrus species.In addition, the analyzed libraries also had not been constructed using the same organs and conditions varied in a considerable manner for each species, as well as the whole number of ESTs among citrus species libraries (Table 2).Therefore, the data shown in Figure 5 is not supposed to reflect the representative number of PR gene family members within the genome of each citrus species.Furthermore, comparisons among species could not be made.Nevertheless, it can bring to light which types of PR gene families were found in the CitEST database expressed within each species under analyzed conditions.A total of sixteen out of seventeen PR gene families were found in the C. sinensis species, a plant that had the greatest number of ESTs sequenced from four different organ libraries (leaf, fruit, bark and flower).Hence, it is possible that different EST-contigs from the same PR gene family have been expressed in all of the four organs; and also it is expected that some EST-contigs of different PR gene families have had co-regulated expression within the same C. sinensis organ.The lowest number the PR gene families (3) was observed in a C. sunki stem bark library, wich also contained the lowest number of sequenced ESTs.In summary, 3 to 15 PR gene families/species were found expressed in analyzed citrus species, and their expression profiles were similar to those found for PR-3, PR-6, PR-7, PR-9 and PR-16 gene families among all of citrus species.
Members of the PR-14 gene family were found to be expressed in all of the studied species with exception for C. limonia, whereas the PR-13 gene family was found only in P. trifoliata species.Nevertheless, it is noteworthy that the CitEST libraries using root or seed tissues were constructed only for C. limonia and P. trifoliata, respectively.Members of the PR-17 gene family were found only in the C. reticulata species and it was also observed that the PR-1 gene family was found only within C. sinensis and C. limonia species.

Citrus PR multigene families: an insight into the clusters
In this work, we have studied expression profiles of citrus PR gene families.Additionally, in order to speculate on the diversity of PR genes in a particular PR gene family, we have analyzed the different PR ESTs and their grouping into clusters generated for four PR gene families separately.Distribution of the total number of citrus PR-like ESTs within the 17 PR gene families was indicated in Table 1.Here, we have focused only in contigs belonging to the PR-2, PR-3 (largely expressed), PR-5 (poorly expressed) and PR-7 (moderately expressed) gene families (Table S1).Two of the best represented contigs in the PR-2 gene family (contigs 2 and 8) were found to be expressed in leaf, fruit and stem bark tissues from Citrus aurantium (CA), C. aurantifoilia (CG), C. latifolia (LT), C. reticulata (CR) and C. sinensis (CS) species.Similar patterns of gene expression were found for two contigs (contigs 1 and 2) of the PR-3 gene family as well as for two contigs (contigs 3 and 6) of the PR-7 gene family, which comprise ESTs expressed in leaf, fruit, stem bark, flower and seed tissues from CA, CG, LT, CR, CS and P. trifoliata (PT) species.In other words, several putative PR-3 genes were clustered within chimerical contigs comprising EST-reads isolated from various cDNA libraries of several different organs, species and conditions.This may suggest the occurrence of the same PR genes with evolutionarily conserved functions among different citrus species genomes.The report that the origin of Nicotiana tabacum PR genes was confirmed from their wild progenitors N. sylvestris and N. tomentosiformis, based on PR gene patterns (Ahl Goy et al., 1982), represents strong evidence that PR genes can be distinguished by species specificity, thus allowing their application as general markers in taxonomic, phylogenetic and evolutionary studies (Edreva, 2005).Moreover, it has been proposed that genes with similar functions, or cDNA libraries expected to share similar patterns of gene expression, cluster together (Ewing et al., 1999).Hence, a related function could be implicated in each particular group of PR genes that was clustered together suggesting a common mechanism controlling their regulation.
Categorizing PR genes into clusters or regulons based on the similarity of PR gene expression profiles in organs from a particular citrus species under several conditions could also point to PR gene members of a multigene family.For instance, clustering of PR-3 gene family ESTs provided a total of 24 contigs and 13 singlets (Table 1), among them 6 contigs and 2 singlets possess ESTs derived from Poncirus trifoliata species.In theory, these cited 6 contigs and 2 singlets may be pointing to the occurrence of at least 8 PR-3-like genes within the P. trifoliata genome.However, this hypothesis needs to be confirmed by Southern genomic hybridization analysis, for example, using a PR-3 gene as probe.The occurrence of PR genes organized in plant genomes as multigene families has already been demon-  strated for the PR-5 gene family in Solanum species (Zhu et al., 1995;Vleeshouwers et al., 2000;Campos et al., 2002) and oat (Lin et al., 1996), among other PR gene families.Therefore, the citrus PR gene families putatively comprise several members.
Likewise, PR genes belonging either to the same or to different PR gene families that share similar pattern of gene expression within the same plant organ library may possibly indicate a coordinated expression control of multiple PR genes playing roles together in a given biochemical pathway.In this context, two PR-3 contigs (contigs 22 and 24) were found to be co-expressed only in drought-stressed C. limonia roots as well as the PR-3 contigs 3 and 11 comprise ESTs isolated only from C. reticulata fruits (Table S1).Nevertheless, more detailed analyses regarding the expression pattern of these clusters/genes will be necessary in order to gain evidence for a possible organ and/or species expression specificity.Most of the different PR contigs, coordinately expressing all of analyzed PR-2, PR-3, PR-5 and PR-7 gene families, were found in both C. reticulata and C. sinensis fruit libraries (Table S1).It has been postulated that the co-regulation is associated with the presence of the same promoter in cis elements, such as SAresponsive element (SARE), GCC box, G-box, W-box, and MRE-like sequence (Zhou, 1999;Chakravarthy et al., 2003;Edreva, 2005), leading to coordinated expression control of multiple PR genes.In this context, coordinated expression for multiple PR genes was correlated with the onset of SAR (Ward et al., 1991).Additionally, enhanced defense actions have already been demonstrated by the synergistic effect of the combinatorial expression of PR protein classes in transgenic plants (Zhu et al., 1994, Jach et al., 1995).
A BLAST search was performed for proteins with amino acid sequences similar to the deduced citrus chitinase proteins from selected contigs; the best hits were found to be PR-3 homologues from Citrus jambhiri, Gossypium hirsutum and Sambucus nigra plants, which were used in phylogenetic analysis.The neighbor-joining tree (Figure 6) shows that the ten studied citrus PR-3 gene sequences were grouped into three major clusters containing, interestingly, different members.Four citrus PR-3 contigs (3, 11, 12 and 16 contigs) comprise the first cluster which covers ESTs that were expressed within several organs (fruit, root, stem bark, leaf and seed) from C. reticulata, C. limonia and P. trifoliata species.They share 95%, 84%, 88% and 86% of amino acids identity with Citrus jambhiri acidic class I chitinase (gi|23496445), respectively.In the second cluster, there are four citrus PR-3 contigs (1, 2, 10 and 21 contigs) that share sequence identity (73%, 71%, 73% and 75%, respectively) with Gossypium hirsutum basic class VII chitinase (gi|32401255).All of these contain P. trifoliata ESTs, which are 1 and 2 chimeric contigs.The remaining two contigs (contigs 22 and 24) belonged to a third cluster, in which only ESTs from C. limonia drought-stressed roots that share sequence identity (only 62% and 53%, respectively) with Sambucus nigra class II chitinase (gi|603884) were placed.These findings correlate with the presence of three different PR-3 chitinase classes with citrus species, two of which (I and VII classes) can be found within the P. trifoliata species, while the class II chitinase was found within the C. limonia species, based on sequence similarities.Interestingly, PR-3 contigs 16 and 21, which were co-expressed within P. trifoliata stem bark tissue, were grouped into two different major clusters.This is an indication of the presence of the two PR-3 chitinase I and VII classes co-expressed within a same organ of a single plant, putatively in response to an identical signal.
Members of the PR-3 family belong to family 19 of glycoside hydrolases (EC 3.2.1.14),which catalyses the hydrolysis of beta-1,4-N-acetyl-D-glucosamine linkages in chitin polymers, a major component of the cell wall of most fungi.It has been demonstrated that the Citrus jambhiri acidic class I chitinase transcripts were not constitutive but accumulated within leaves after wounding or inoculation with non-pathogenic or pathogenic isolates of the Alternaria alternata fungus (Gomi et al., 2002).The putative citrus class I chitinase genes studied herein that make up the major first cluster were expressed under fruit development (from C. reticulata), CTV-infection (from P. trifoliata), stem bark wounding or wounding/P.parasitica-infection (from P. trifoliata), root drought-stress (from C. limonia) and in seed tissue (from P. trifoliata) conditions.Likewise, the Gossypium hirsutum basic class VII chitinase gene was inducible by salicylic acid in seedlings, with transcript accumulation in root and cotton fibers, associated with the cotton's resistance to diseases (Li and Liu, 2003).
Whether the studied putative citrus class VII chitinase genes were induced by salicylic acid remains to be investigated.Similarly, the Sambucus nigra class II chitinase gene was found to be expressed during ethylene-promoted leaflet abscission (Coupe et al., 1997), but the putative citrus class II chitinase genes analyzed herein were associated with root drought-stress.Whether they are involved in ethylene pathways needs to be studied.

Concluding Remarks
In this paper, we present a large-scale analysis of gene expression profiles to identify citrus PR candidate genes that may participate in environmental and developmental responses.This can now be examined further in experimental studies regarding biotechnological approaches or citrus resistance markers.Albeit in silico, the data presented in this work represent a starting point to elucidate the complex responses of citrus plants to biotic and abiotic stresses.The identified PR-gene families may be useful to verify innate immunity mechanisms in citrus species possessing basal or induced/nonhost or host resistances to certain pathogens taking place with a participation of PR proteins, and this is the subject of current research.

Figure 1 -
Figure 1 -Expression patterns of 3,103 citrus ESTs encoding 17 putative PR protein families (PR-1 to PR-17) in 33 selected cDNA libraries from the CitEST database.Each library is represented by a single row of colored boxes, whereas each PR-like gene family is represented by a single column.Data represent the relative number of PR reads from a specific library in each EST-contig per 1,000 reads.

Figure 2 -
Figure 2 -In silico differential expression profiles of putative PR gene families (PR-1 to PR-17) in cDNA CitEST libraries from citrus organs induced by pathogen or drought stresses.Each subtracted stress situation (treatment vs. control) is represented by a single row; each PR-like gene family is represented by a single column.The color bar represents the color scales between relative induction (intense red for larger than 2) or suppression (intense green for lower than -2) ranges in relation to the counterpart control library.Black means no change (ratio 0) and blue no expression.CiLV, citrus leprosis virus; CTV, citrus tristeza virus; Xylella, X. fastidiosa; Phytophthora, P. parasitica; CS, C. sinensis; CR, C. reticulata; CL, C. limonia; PT, Poncirus trifoliata.D.A.I., days after inoculation.

PR gene families of citrus 925 Figure 4 -
Figure 4 -Graphical representation of the expression profiles of putative PR gene families (PR-1 to PR-17) in different citrus organ cDNA libraries.Data represent the relative number of PR reads from a specific library in each EST-contig per 1,000 reads.Each organ library is represented by a single row; each PR gene family is represented by a single column.

Figure 5 -
Figure 5 -PR-like gene families expression patterns present in each citrus species.Each citrus species is represented by a single row; each PR gene family (PR-1 to PR-17) is represented by a single column.Data represent the relative number of PR reads from a specific library in each EST-contig per 1,000 reads.CA, Citrus aurantium; CG, C. aurantifolia; CL, C. limonia; CM, C. limettiodes; CR, C. reticulata; CS, C. sinensis; LT, C. latifolia; PT, Poncirus trifoliata; TS, C. sunki.

PR gene families of citrus 927 Figure 6
Figure 6 -A neighbor-joining tree of PR-3 homologues from Citrus constructed and visualized in MEGA program from deduced amino acids sequences alignment in Clustalx.The numbers indicate in percentage bootstrap analysis (1,000 replicates) of the data supporting the branches.The following sequences were taken from the EMBL GeneBank: Citrus jambhiri acid class I chitinase (gi|23496445); Sambucus nigra class II chitinase (gi|603884) and Gossypium hirsutum basic class VII chitinase (gi|32401255).

Table 1 -
PR gene sequences used as query for searching the homologous PR genes within the citrus genome EST data bank -CitEST and distribution of the number of citrus PR-like ESTs within the 17 PR gene families.

Table 2 -
EST libraries used in this study.