Introduction

Environmental stress conditions are major factors limiting growth, productivity, and the distribution of plants in the world (Boyer 1982). One third of the earth surface is classified as arid or semiarid, and in most of the humid regions where much of the world food is produced, the crops are subjected to periods of severe drought and extreme temperatures. Moreover, nearly 40% of the world land surface can be categorized as having potential high salinity problems (Boyer 1982). As a major horticultural crop worldwide, citrus production is not safe against these environmental threats. Citrus cultivation is mainly located in semiarid zones, where water supply is restricted, and primarily limited to the region between 40°N and 40°S latitudes, where low temperatures occasionally affect production. In addition, the use of low-quality water for irrigation has increased the salt concentration in citrus-producing areas (Storey and Walker 1999).

Responses to abiotic stresses require the production of important functional proteins, such as those involved in the synthesis of osmoprotectants, and regulatory proteins operating in the signal transduction pathways, such as kinases and transcription factors (TFs) (Saibo et al. 2009). Because most of these responses imply the control of gene expression, TFs play a critical role in the abiotic stress responses (Chaves and Oliveira 2004). At least four different TF families, designed according to their DNA-binding domain (Riechmann et al. 2000), were identified to be involved in the plant response to abiotic stresses: (1) CBF/DREB (C-repeat binding factor/DRE binding protein), (2) NAC (NAM, ATAF, and CUC) and ZF-HD (zinc-finger homeodomain), (3) AREB/ABF (ABA-responsive element-binding protein/ABA-binding factor), and (4) MYC (myelocytomatosis oncogene)/MYB (myeloblastosis oncogene). Some of these TFs are controlled by ABA, while others are not, indicating the involvement of both ABA-dependent and ABA-independent regulatory systems controlling stress-inducible gene expression (Yamaguchi-Shinozaki and Shinozaki 2006).

NAC domain-containing proteins comprise one of the largest plant-specific TF families, with more than 100 members identified in both Arabidopsis and rice (Ooka et al. 2003; Fang et al. 2008), which are expressed in several tissues and developmental stages. Proteins of this family are characterized by a highly conserved N-terminal domain involved in DNA binding, known as NAC domain, and a C-terminal region highly divergent in sequence and length containing the transcriptional activation domain (Ooka et al. 2003). The NAC domain comprises about 160 amino acid residues, divided into five subdomains (A–E) of conserved blocks intercalated with heterogeneous blocks or gaps.

NACs play important roles in plant development, including the formation and maintenance of shoot apical meristems (Souer et al. 1996), floral development and morphogenesis (Sablowski and Meyerowitz 1998), embryo development (Duval et al. 2002), hormone signaling (Fujita et al. 2004), and formation of secondary walls (Zhong et al. 2006). NACs have also been implicated in the response to abiotic and biotic stresses, and some of them have been used as targets for engineering stress tolerance in plants. Several NAC genes from Brassica napus (Hegedus et al. 2003) and the SsNAC23 gene from Saccharum sp. (Nogueira et al. 2005) were induced by cold, dehydration, and herbivorous insect attack. OsNAC6 from Oryza sativa was induced by abscisic acid (ABA), cold, dehydration, and salt stress (Ohnishi et al. 2005). Transgenic rice plants overexpressing OsNAC6 constitutively exhibited tolerance to dehydration, high-salt stress, and blast disease (Nakashima et al. 2007). anac019, anac055, and anac072 from Arabidopsis were also induced by ABA, drought, and salt stress, and the transgenic lines overexpressing these genes displayed an increased tolerance to drought (Tran et al. 2004).

While a portion of the NAC genes has been primarily characterized in model plants such as Arabidopsis and rice, NAC genes remain poorly characterized in woody fruit species like citrus. Recently, Fan et al. (2007) reported for the first time the molecular cloning of a citrus NAC-like gene involved in the peel pitting stress response. This gene was induced under several postharvest abiotic stresses, including wounding, anoxia, low temperature, and ethylene. More recently, Liu et al. (2009) isolated another citrus NAC-like gene that was found to be involved in fruit development and senescence. In the present study, the identification of 45 citrus NAC genes belonging to 13 different subfamilies is reported as part of an effort to identify novel genes involved in abiotic stress responses, especially drought stress, in citrus. One of them, CsNAC1, was showed to be a member of the stress-NAC subfamily. Given its potential as target for engineering tolerance to environmental stresses in citrus, this gene was further characterized. It is suggested that citrus CsNAC1 is involved in the response to multiple abiotic stresses and in ABA signaling.

Materials and methods

Plant materials and stress treatments

Seeds of ‘Cleopatra’ mandarin (Citrus reshni hort. ex Tanaka) and ‘Rangpur’ lime (Citrus limonia Osbeck) were collected at the Citrus Germplasm Collection of Embrapa Cassava & Fruits (Cruz das Almas, Bahia, Brazil). For the dry-down experiments (drought stress treatment), citrus seeds were germinated in plastic dibble tubes (115 cm3 and 180 cm3) containing a mixture of Plantmax (Eucatex Agro., Brazil) substrate and coconut fiber, at 1:1 proportion, under greenhouse conditions. Four months after germination, the plants were transplanted to 3-l pots containing a mixture of soil, sand, and Plantmax substrate, at 2:1:1 proportion. Eight months after germination, plants of nucellar origin were selected based on their uniformity and divided into two subsets: (1) a set of 10 plants of each genotype that were daily irrigated (well-watered plants—WW) and (2) a set of 10 plants of each genotype that were drought stressed (DS). Prior to the stress initiation, the pots of both treatments were saturated with water and left overnight to drain the excess water. On the next morning, all the pots were closed with aluminum foil in order to allow the water loss only by transpiration. The water content of the soil in each pot was thereafter measured with a time-domain reflectometry (TDR) probe. The normalized transpiration rate (NTR) was calculated for each DS plant by dividing their transpiration rate determined from the differences in daily soil moisture by the mean transpiration rate of the WW plants (Wahbi and Sinclair 2007). The explored soil volume (Vs) and the total leaf area (m2) of plants were taken into account in order to allow comparisons among plants of different sizes. The fraction of transpirable soil water (FTSW) was calculated for each pot on every day. This was performed by subtracting for each pot the lower limit soil moisture from the soil moisture measured each day, and dividing it by the total transpirable soil water for the pot (Sinclair and Ludlow 1986). Data collection for each plant was stopped when the NTR value decreased to less than 0.1. Leaf tissue samples were collected at different FTSWs, following the initiation of the experiments.

Values of NTR and FTSW obtained during the dry-down experiments for each plant on each day were all combined in order to obtain the NTR response curve as a function of FTSW (Wahbi and Sinclair 2007). A linear-plateau analysis was used to calculate the FTSW threshold value where NTR begins to decline. A plateau region, where NTR was approximately 1.0, was defined for the region where the soil medium was wet. The region of the linear decline in NTR as the soil dried further was fitted by linear regression. The intersection of the linear region with NTR = 1.0 was calculated as the plateau intercept where drying induced a response in plant gas exchange. This model has been one that has been shown to fit well the response of plants grown on mineral soil (Wahbi and Sinclair 2007). All regressions were carried out using GraphPad Prism ver. 4.0 (GraphPad Software, USA).

For salt, cold, and ABA treatments, citrus seeds were germinated in vitro (Oliveira et al. 2009) and the seedlings were maintained at 27 ± 2°C on a 16:8 h light/dark regime with a photon flux density of 250 μmol m−2 s−1. Eight weeks after germination, seedlings of nucellar origin were selected based on their uniformity, divided into three subsets (six replicated plants), and transferred from the basal MS (Murashige and Skoog 1962) solution to the MS solution containing 200 mM NaCl or 100 μM ABA. For the cold treatment, seedlings were transferred to and kept at 4°C. Leaves and roots were harvested after 3, 6, and 12 h of treatment. Control plantlets were transferred to basal MS solution and kept at 27 ± 2°C. All tissue samples were immediately frozen in liquid nitrogen and stored at −80°C until use.

Identification and sequence analysis of NAC domain-containing proteins of citrus

The NAC domain-containing protein sequences of citrus were identified in the HarvEST: Citrus database, which contains more than 200,000 citrus expressed sequence tags (ESTs), using the NAC domain of proteins from Arabidopsis thaliana and O. sativa (Ooka et al. 2003) as driver sequences for searching. The set of non-redundant citrus assembled sequences with an E value ≤10−20 were considered to represent putative citrus NAC domain-containing genes. Only the sequences that contained at least four out of five conserved N-terminal NAC subdomains (A through E) were considered and used for the multiple alignments and phylogenetic analysis.

Multiple sequence alignments were performed using the default parameters of ClustalW (Thompson et al. 1994) and the dendrogram was constructed by the MEGA4 program (Tamura et al. 2007) that used the neighbor-joining method (Saitou and Nei 1987) and bootstrap re-sampling analysis (1,000 replications). The identification of the sequences used is provided by the accession number in the corresponding figure.

Digital gene expression profiling

Digital gene expression profile of the CsNAC1 gene was calculated as previously reported (Chen et al. 2010). Briefly, tissue specificity and treatment were summarized, respectively, from the tissue sources and treatments of all the 89 EST libraries contained in the HarvEST: Citrus database. Approximate expression intensity in each tissue/treatment was inferred from the ratio of the count of CsNAC1 ESTs to the total number of clones in non-normalized EST libraries of the same tissue/treatment. For comparison, the ratio was converted into transcripts per million (TPM) in each tissue/treatment. The ESTs from those normalized, subtracted, or otherwise biased libraries were not counted or used for TPM calculation.

Total RNA isolation and cDNA synthesis

Total RNA was isolated from leaf and root tissues with the RNAqueous® kit (Applied Biosystems, USA), treated with 100 U of RNase-free DNase (Amersham Pharmacia Biotech, USA) and further purified with RNeasy Mini kit (Qiagen, USA). The purity and concentration of the isolated RNA samples were checked on 1% (w/v) agarose gels and in a spectrophotometer. Reverse transcription was carried out using the RETROscript kit (Ambion, USA) and 2 μg of total RNA, according to the manufacturer’s instructions. The cDNA concentration was determined spectrophotometrically.

Quantitative real-time reverse transcription PCR (qPCR)

All the procedures for qPCR, including tests, validations, and experiments, were carried out according to the recommendations of Applied Biosystems. The qPCR reactions were performed on an ABI Prism 7500 instrument and Sequence Detection System software, version 1.6.3 (Applied Biosystems). Based on the expression stability observed under the experimental conditions analyzed, actin was amplified along with the target gene as an endogenous control to normalize expression among different samples. The control primers were: forward 5′-CATCCCTCAGCACCTTCC-3′ and reverse 5′-CCAACCTTAGCACTTCTCC-3′. The primers for the CsNAC1 gene were: forward 5′-GGCACCGACAAGATTATCACAA-3′ and reverse 5′-GCTGCCATTTTTGCGAGAAG-3′. Reactions were performed in triplicate, containing 100 ng of cDNA, 0.4 μl of each primer (5 pmol), 10 μl Power SYBR Green Master Mix (Applied Biosystems), and sterile Milli-Q water for a final volume of 25 μl. The amplification reactions were performed under the following conditions: (1) activation of Taq DNA polymerase at 50°C for 2 min, (2) initial denaturation at 95°C for 10 min, (3) denaturation at 95°C for 15 s, (4) annealing at 60°C for 30 s, and (5) extension at 60°C for 1 min. Steps 3–5 were repeated for 40 cycles. To verify that only a single PCR product was generated for the amplified transcript, the multicomponent data for each sample was subsequently analyzed using the Dissociation Curve 1.0 program (Applied Biosystems). Gene expression was quantified using the comparative methods Ct: 2−ΔCt and 2−ΔΔCt, with data obtained from a pool of at least three biological replicates that were individually validated.

Results

Identification of citrus NAC genes and sequence analysis

NAC domain protein sequences from A. thaliana and O. sativa were used as queries to identify citrus NAC domain-containing sequences encoded by the assembly of 229,570 ESTs of Citrus and Poncirus available at the HarvEST: Citrus database. Forty-five open reading frames (ORFs) encoding proteins with at least four out of five conserved NAC subdomains (A through E) were identified in this study and sequentially named as Citrus sp. NACs (CsNACs). These included sequences from both Citrus and Poncirus. Each gene was further confirmed to contain a NAC domain through the BLASTP program against the non-redundant database. The CsNAC proteins identified in the present study are listed in Table 1. To our knowledge, this is the first report describing the NAC gene family in citrus.

Table 1 CsNAC proteins identified in the HarvEST: Citrus database
Full size table

The deduced amino acid sequences of the identified CsNACs [see electronic supplementary material (ESM) S1], including the two previously characterized CsNACs together with their respective orthologs from other plant species, were aligned for sequence comparison and phylogenetic analysis (Fig. 1). The phylogenetic analysis classified the CsNACs into 13 different subfamilies with an internal stability supported by bootstrapping testing. These results indicate the existence of a diversified family of NAC domain-containing protein in citrus, with distinct roles in stress response and/or development. One of these proteins, CsNAC1, showed to be a member of the stress-NAC subfamily. This group also encompasses GhNAC4 from cotton, GmNAC3 and GmNAC4 from soybean, and anac019, anac055, and anac072 from Arabidopsis, which have been shown to be induced by ABA, drought, cold, and salt stresses and to enhance tolerance to drought stress in transgenic plants (Tran et al. 2004, 2009; Meng et al. 2009; Pinheiro et al. 2009).

Fig. 1
figure 1

Phylogenetic tree of NAC domain-containing proteins from citrus. Amino acid sequences were aligned using ClustalW and a neighbor-joining tree was constructed with a 1,000-bootstrap replication support. The subfamilies within the NAC family, as designated by Ooka et al. (2003), are grouped as indicated. The NAC domain-containing proteins of citrus are identified as listed in Table 1. CsNAC1 protein is highlighted by box. Arrows indicate the previously described CsNAC proteins of citrus. Abbreviations for the name of the subfamilies are as follows—stress NAC stress NAC transcription factor family, ATAF1 Arabidopsis transcription factor 1-like family, ATAF2 Arabidopsis transcription factor 2-like family, NAM-B1 no apical meristem B1 transcription factor-like family, NAP NAC-like activated by APETALA 3/PISTILLATA family, SENU5 tomato senescence up-regulated 5-like family, NAC2 Arabidopsis thaliana NAC protein 2-like family, NAM no apical meristem transcription factor-like family, NAC1 Arabidopsis thaliana NAC protein 1-like family, VND-2 vascular-related NAC domain 2 family, ANAC011 Arabidopsis NAC protein 011 transcription factor-like family, TERN tobacco elicitor-responsive NAC protein-like family, TIP turnip crinkle virus interacting protein-like family. Accession numbers of NAC sequences from other plant species used in the analysis are in parentheses. anac011 (NP_174529.2); anac013 (NP_001117401.1); anac016 (NP_564439.1); anac017 (NP_564440.1); anac019 (NP_175697.1); anac036 (NP_565404.2); anac041 (NP_001118435.1); anac046 (NP_187056.1); anac052 (NP_566375.1); anac053 (NP_566376.1); anac055 (NP_188169.1); anac062 (NP_190522.1); anac071 (NP_193532.1); anac072 (NP_001078452.1); anac074 (NP_567811.1); anac076 (NP_195339.1); anac080 (NP_568182.2); anac082 (NP_196495.1); anac083 (NP_196822.1); anac089 (NP_568414.1); anac096 (NP_199471.1); anac100 (NP_200951.1); anac103 (NP_201211.1); ATAF1 (NP_171677); ATAF2 (NP_680161); AtCUC2 (BAA19529.1); AtNAC2 (NP_188170.1); AtNAM (AAM65392.1); AtNAP (NP_564966.1); AtNTL9 (NP_001119122.1); AtTIP (NP_197847.3); AtVND1 (NP_179397.1); CsNAC (ABM67699.1; ABQ96643.1); GhNAC2 (ACI15342.1); GhNAC4 (ACI15345.1); GhNAC6 (ACI15343.1); GmNAC3 (AAX85980.1); GmNAC4 (AAX85981.1); GmNAC5 (AAX85982.1); GmNAC16 (ACD39374.1); NAM-B1 (ABI15896 .1); PhNAM (CAA63101.1); PtNAC (XP002300972.1; XP002311983.1; XP002325400.1; XP002329110.1); PtrifoliataNAC (ACM90162.1); RcNAC (XP_002520341.1); Senu5 (CAA99760.1); StNAC (CAC42087.1); TERN (BAA78417.1); ZmCUC2 (ACG42560.1)

Full size image

Since members of the stress-NAC subfamily are involved in general responses to abiotic stresses and, thereby, they may constitute potential targets for engineering plant tolerance to abiotic stresses, we have selected the CsNAC1 gene for further characterization. The CsNAC1 gene contains a 1,044-bp ORF encoding a 347 amino acid protein (38.8 kDa) that shared 76% identity and 84% similarity at the amino acid level to GhNAC4 from cotton (Gossypium hirsutum). Sequence analysis of the deduced CsNAC1 protein showed that it contains features typical of known NAC domain proteins, including the five N-terminal A through E NAC subdomains (amino acid 14–164) and a C-terminal region containing three transcriptional activation motifs according to Ooka et al. (2003) (Fig. 2a). Sequence alignment of NAC domains from different plant species showed that CsNAC1 contains all the conserved residues from A through E subdomains present in other plants (Fig. 2b). A degenerate bipartite nuclear localization signal (NLS) conserved in several NAC domain proteins among different plant species (Greve et al. 2003) was identified between the amino acids 113–131 (Fig. 2b).

Fig. 2
figure 2

Structural analysis of CsNAC1 protein. a Diagrammatic representation of CsNAC1. NAC domain is identified by white box while transcriptional activation motifs, as reported by Ooka et al. (2003), are identified by black boxes. Amino acid sequences of motifs are indicated above each black box. b Sequence alignment of NAC domains from different plant species. Only amino acids belonging to the NAC domain sequence were used in the alignment. Subdomains A through E are shown by lines below the sequences. Identical amino acid residues in a given position are in white text on a black background, while those conserved residues are in white text in a gray background. Dashes denote a gap in the amino acid sequence. Arrow above the sequences denotes the degenerate bipartite nuclear localization signal (NLS) that is conserved in several NAC domain proteins (Greve et al. 2003). Accession numbers for protein sequences of other plant species are in parentheses. GmNAC4 (AAX85981.1); ATAF1 (NP_171677); StNAC (CAC42087.1); ATAF2 (NP_680161); AtNAP (NP_564966.1); AtNAC2 (NP_188170.1); NAM-B1 (ABI15896.1); AtNAM (AAM65392.1); SENU5 (CAA99760.1); TERN (BAA78417.1). CsNAC1 protein is in bold

Full size image

Digital gene expression profile of CsNAC1

The inferred transcripts per million (TPM) in each tissue or stress treatment provided a comparable basis to reflect the tendency of the CsNAC1 expression, though not representing the true-to-type expression level. CsNAC1 ESTs were distributed among 1,897 non-normalized EST clones of shoot meristem (one library), 1,950 (three libraries) of leaves, 10,922 (four libraries) of floral organs, and 24,463 (12 libraries) of fruits, obtained from different Citrus species (Table 2). Leaves (2,051.3 TPM) and shoot meristems (1,581.4 TPM) were the tissues where CsNAC1 was more highly expressed (Table 2). However, it is noteworthy that the absence of CsNAC1 ESTs in any other tissue does not necessarily imply that they are not expressed in those tissues, but due to the lack of non-normalized libraries from other tissue sources that could be considered for TPM calculation.

Table 2 CsNAC1 EST distribution and digital gene expression profile (TPM) in different tissues/organs
Full size table

CsNAC1 ESTs were distributed in libraries of different Citrus species subjected to seven distinct stress treatments, including cold, salt, and iron deficiency (Table 3). TPM calculation indicated that CsNAC1 was highly expressed under salt (2,656 TPM), iron deficiency (1,763.7 TPM), exocortis (1,587.3 TPM), and cold (1,307.8 TPM) treatments. Gibberellic acid (956.9 TPM), hot water (351.2 TPM), and thrips (541.7 TPM) were also treatments in which CsNAC1 was expressed. Thus, the digital expression data suggest an involvement of CsNAC1 in general stress responses, providing useful information for further experimental validation.

Table 3 CsNAC1 EST distribution and digital gene expression profile (TPM) under different treatments
Full size table

The CsNAC1 gene is induced by drought stress

To investigate whether the CsNAC1 gene is induced by drought stress, plants of two important citrus rootstocks (C. reshni and C. limonia) were subjected to dry-down experiments, which mimic the different phases that a plant experiences under drought in natural conditions, and gene expression was monitored by qPCR. The dry-down experiments showed that the onset of transpiration decline occurred at FTSW of 0.31 for C. reshni and 0.36 for C. limonia, as indicated by the linear-plateau model (Fig. 3a). These FTSW values represent the transition from stage I to stage II dehydration according to Sinclair and Ludlow (1986). The results of the qPCR analysis revealed that CsNAC1 expression was significantly induced by drought stress in both citrus rootstocks (Fig. 3b). CsNAC1 expression was induced in leaves at an early stage of soil drying (FTSW of 0.6), and enhanced and maintained following further soil drying (FTSWs of 0.5–0.2). It is interesting to note that the induction of CsNAC1 expression in both citrus rootstocks has occurred prior to the transition between the dehydration stages I and II. It suggests a modulation of CsNAC1 expression by early signal(s) triggered by the initial decline in the soil water potential, before the drought stress took place.

Fig. 3
figure 3

Expression of CsNAC1 in response to drought. a Soil dry-down experiments. Eight-month-old Citrus reshni and C. limonia plants were subjected to progressive soil drying and a normalized transpiration rate (NTR) response curve was obtained as function of the fraction of transpirable soil water (FTSW). Arrow indicates the fraction of extractable water whereby the transpiration rates began to decline in an approximately linear manner in each genotype. b quantitative RT–PCR analysis of CsNAC1 expression during the soil dry-down experiments. The transcript levels were measured in leaves and standard errors were calculated from three biological replicates in which β-actin transcripts were used as internal controls. The fold induction by each treatment as related to the control (FTSW = 1) is indicated

Full size image

The CsNAC1 gene is induced by salt, cold, and ABA

We further analyzed the effects of salt stress, cold, and ABA, a hormone involved in abiotic stress signaling (Mahajan and Tuteja 2005), on the expression of the CsNAC1 gene by qPCR analysis. C. reshni was selected for these assays based on the high level of CsNAC1 expression (Fig. 3b). Figure 4 shows that the CsNAC1 gene was strongly induced within 3, 6, and/or 12 h after salt, cold, and ABA treatments in both leaves and roots of C. reshni. Thus, these results also suggest the involvement of CsNAC1 in salt and cold stress response pathways, in a regulation that may be dependent on ABA. Taken together, the results indicate that CsNAC1 is involved in general responses to abiotic stresses, as it is induced by drought stress, salt, cold, and ABA, and therefore may constitute a potential target for engineering tolerance to multiple abiotic stresses in citrus.

Fig. 4
figure 4

Expression of CsNAC1 in response to salt, cold, and ABA treatments. Eight-week-old Citrus reshni plants were treated with 200 mM NaCl (salt), 100 μM ABA, or exposed to 4°C (cold) for 3, 6, and 12 h. The transcript levels were measured in leaves and roots through quantitative RT–PCR and standard errors were calculated from three biological replicates in which β-actin transcripts were used as internal controls. The fold induction by each treatment as related to the untreated control is indicated

Full size image

Discussion

The diversified NAC domain-containing protein family of citrus

Although several members of the NAC domain gene family have been characterized in Arabidopsis and rice (O. sativa), the characterization of NAC genes in woody fruit species remains scarce. Aiming to identify novel genes involved in abiotic stress responses in citrus, we carried out a comprehensive analysis that has identified 45 members of the NAC citrus family using Arabidopsis and rice sequences as drivers to screen the HarvEST: Citrus database. This number, however, can be considered a conservative estimate for the number of NAC domain genes in citrus because we did not include in the analysis assembled contig sets or singlets containing less than four out of five conserved NAC subdomains (A through E). Only 26 members of the NAC citrus family contained the five conserved NAC subdomains, which is in accordance with the number of full-length NAC domain genes reported for Citrus sinensis in the Plant Transcription Factor Database (PlantTFDB 2.0; http://planttfdb.cbi.pku.edu.cn/family.php?sp=Csi&fam=NAC).

The phylogenetic analysis placed the 45 identified citrus NAC-deduced proteins into 13 different subfamilies according to the Ooka et al. (2003) classification. Members of the different subfamilies have been implicated in functionally diverse roles, including response to abiotic stresses, defense against biotic stresses, and regulation of developmental programs (Olsen et al. 2005). Members of the stress-NAC subfamily have been demonstrated to be induced by ABA, drought, and salt stress (Tran et al. 2004, 2009; Meng et al. 2009; Pinheiro et al. 2009), to function as positive regulators in the drought stress response (Tran et al. 2004), and to increase drought tolerance in transgenic plants (Tran et al. 2004). ATAF1 and ATAF2 clades encompass proteins that are strongly induced by wounding (Collinge and Boller 2001; Delessert et al. 2004) and associated with response to several pathogens (Hauck et al. 2003). TERN and TIP members have been also associated with defense responses against pathogens (Ren et al. 2000). NAM-B1 has been shown to stimulate leaf senescence and increase the remobilization of nutrients from the leaves to the developing grains in wheat (Uauy et al. 2006). NAP and SENU5 members have been demonstrated to be also involved in leaf senescence (Guo and Gan 2006). Members of the NAM subfamily are involved in embryogenesis and shoot apical meristem formation and development (Souer et al. 1996; Olsen et al. 2005), while NAC1 and NAC2 of Arabidopsis are involved in the regulation of lateral root development and auxin signaling (Xie et al. 2000; He et al. 2005).

The results suggest that the 45 citrus NAC proteins described may constitute functionally non-redundant transcription factors, which could be involved in the response to distinct types of stresses and/or events of development and morphogenesis. This functional diversity is supported by the recent observations that sequence similarity-based clustering of the members of the NAC superfamily correlates with their function (Olsen et al. 2005; Pinheiro et al. 2009).

The CsNAC1 gene is involved in abiotic stress response pathways and in ABA signaling

CsNAC1 shares an extensive similarity with GhNAC4, GmNAC3, GmNAC4, anac019, anac055, and anac072, all of which have been shown to function in pathways of abiotic stress response and tolerance (Tran et al. 2004, 2009; Meng et al. 2009; Pinheiro et al. 2009). Consistent with its putative role as a stress-NAC transcription factor containing all the five NAC subdomains (A through E), the CsNAC1-deduced protein also possesses a predicted NAC nuclear localization signal (Xie et al. 2000; Greve et al. 2003) and three transcriptional activation motifs (Ooka et al. 2003) on its primary structure. Several NAC proteins containing these conserved sequences in the primary structure have been demonstrated to be addressed to the nucleus and to show transactivation activity (Xie et al. 2000; Greve et al. 2003; Fujita et al. 2004; Nogueira et al. 2005; Nakashima et al. 2007; Hu et al. 2008; Pinheiro et al. 2009; Tran et al. 2009; Zheng et al. 2009).

A number of reports have suggested that the tissue specificity of NACs provides clues about their biological function (Xie et al. 2000; Sharp et al. 2004; He et al. 2005; Mitsuda et al. 2007; Yoo et al. 2007; Tran et al. 2009). We have therefore explored the HarvEST: Citrus database to provide the first approximation of CsNAC1 transcript abundance in different tissues and under different treatments. The digital gene expression profile analysis demonstrated that CsNAC1 was primarily expressed in leaves and shoot meristems and was involved in general stress responses. The tissue specificity of some stress-NAC members has been previously determined. GhNAC4 of cotton was shown to be abundantly expressed in leaves, but it has little or no expression in stems and roots (Meng et al. 2009). On the other hand, GmNAC3 and GmNAC4 of soybean were shown to be primarily expressed in roots and flowers (Tran et al. 2009). In our in silico analysis, we were not able to observe CsNAC1 expression in roots due to the absence of non-normalized libraries from this tissue source in the HarvEST: Citrus database. Yet, it was then confirmed through qPCR analysis (Fig. 4).

We have further demonstrated by qPCR analysis that CsNAC1 is not only structurally similar to the stress-NAC subgroup but it also shares a conserved stress-inducible expression profile as it was strongly induced by drought stress, salt, cold, and ABA. GhNAC4 was shown to be induced by drought, salt, cold, and ABA (Meng et al. 2009). GmNAC3 and GmNAC4 were induced by osmotic stress, salt, cold, ABA, and jasmonic acid (JA) (Pinheiro et al. 2009; Tran et al. 2009). anac019, anac055, and anac072 of Arabidopsis were mainly induced by drought stress, salt, and ABA, and some of them also by JA (Tran et al. 2004). These findings suggest that CsNAC1 likely plays a role in stress responses similar to that of its counterparts in other plant species.

It has been shown that NACs can regulate abiotic stress responses through both ABA-dependent and ABA-independent pathways (Fujita et al. 2004; Tran et al. 2004). Since the expression of CsNAC1 was induced under exogenous ABA treatment, its participation in abiotic stress responses through an ABA-dependent signaling pathway is suggested, as demonstrated for its closest orthologs (Tran et al. 2004, 2009; Meng et al. 2009; Pinheiro et al. 2009). This hypothesis is additionally supported by the fact that the cis-acting element ABRE (ABA-responsive element) has been identified in the promoter region of several NAC genes (Fujita et al. 2004; Tran et al. 2004; Nakashima et al. 2007; Fang et al. 2008). Nevertheless, the induction of CsNAC1 as well through an ABA-independent signaling pathway cannot be ruled out since the cis-element DRE (dehydration-responsive element)/CRT (C-RepeaT) that functions in ABA-independent gene expression (Yamaguchi-Shinozaki and Shinozaki 2006) has also been identified in the promoter of some NACs (Fujita et al. 2004; Fang et al. 2008). DRE/CRT has been shown to function in the early process of stress-responsive gene expression whereas ABRE functions after the accumulation of ABA (Yamaguchi-Shinozaki and Shinozaki 2006). The presence of both cis-elements in the promoter region of CsNAC1 could explain its induction observed in leaves since an early stage of soil drying (FTSW of 0.6), before the threshold soil moisture level for decreasing transpiration rate and establishing drought stress are to be reached (Fig. 3). However, further analysis of the promoter elements of CsNAC1 and their relation with the induction of its expression is required to determine the regulatory systems controlling the stress-inducible expression of CsNAC1.

In conclusion, our data provide the first analysis of the previously uncharacterized CsNAC family of citrus and reveal a novel member involved in the response to multiple abiotic stresses. Our results also suggest that the CsNAC1 gene may represent a potential target for engineering stress tolerance in citrus. We are currently testing this hypothesis through the ectopic expression of CsNAC1 in tobacco and citrus.