Trees constitute the majority of lignocellulosic biomass existing on our planet. Trees also serve as important feedstock materials for various industrial products. However, little is known about the regulatory mechanisms of cellulose synthase (CesA) genes of trees. Here, the cloning and characterization of three CesA genes (EgraCesA1, EgraCesA2, and EgraCesA3) from an economically important tree species, Eucalyptus grandis, are reported. All three genes were specifically expressed in xylem cells of eucalyptus undergoing secondary cell wall biosynthesis. The GUS gene, expressed under the control of the EgraCesA2 or EgraCesA3 promoter, was also localized in the secondary xylem in transgenic tobacco stems. However, the EgraCesA1 promoter alone or along with its 5′-UTR introns was insufficient to direct appropriate GUS expression. EgraCesA2 and EgraCesA3 gene expression was up-regulated in tension-stressed eucalyptus xylem cells. Accordingly, GUS expression directed by the EgraCesA2 or EgraCesA3 promoter was also up-regulated. EgraCesA1 had no such response. Thus, it is most unlikely that EgraCesA1 is a subunit of the EgraCesA2–EgraCesA3 complex. The presence of at least two types of cellulose biosynthesis machinery in wood formation is an important clue in deciphering the underpinnings of the perennial growth of trees in various environmental conditions. By analysing GUS gene expression directed by the EgraCesA3 promoter or its deletions, several negative and positive regulatory regions controlling gene expression in xylem or phloem were identified. Also a region which is likely to contain mechanical stress-responsive elements was deduced. These results will guide further studies on identifying cis-regulatory elements directing CesA gene transcription and wood formation regulatory networks.
Cellulose, commonly found in plant cell walls, is the most abundant form of living terrestrial biomass (Crawford, 1981). It has been estimated that annually plants produce approximately 1.5×1015 kg of cellulose on earth (Deguchi et al., 2006). Chemically, cellulose is a simple and linear chain polymer composed of 500–14 000 glucose molecules (Somerville, 2006). These glucoses are linked together through 1,4-β-glycosidic bonds. In plant cell walls, 36 glucan chains are linked by hydrogen bonds formed among hydroxyl groups on the glucose residues (Nishiyama et al., 2002, 2003). The aggregated glucan chains form microfibrils which are strong, cable-like structures. In microfibrils, the glucan chains are oriented in parallel and form highly ordered and crystalline regions that are interspersed by disordered and amorphous regions (Beguin and Aubert, 1994). The crystalline regions are resistant to chemical or enzymatic hydrolysis, while the amorphous regions are more susceptible to this type of degradation.
Cellulose is synthesized at the plasma membrane by the coordination of multiple proteins (Delmer, 1987, 1999; Doblin et al., 2002; Somerville, 2006; Joshi and Mansfield, 2007). These proteins are assembled into terminal complexes and use uridine diphospho-glucose as a substrate for glucan chain elongation (Delmer, 1999; Howles et al., 2006). Terminal complexes can be observed as particle arrays using freeze-fracture electron microscopy (Kimura et al., 1999). In bacteria and some algae, terminal complexes are arranged in single or multiple rows. In higher plants, however, terminal complexes are hexagonal rosette structures with 6-fold symmetry (Delmer, 1999; Doblin et al., 2002).
Several proteins have been associated with cellulose biosynthesis. Sucrose synthase, an enzyme catalysing the formation of uridine diphospho-glucose from sucrose, has been postulated to play a role in providing uridine diphospho-glucose to the cellulose synthase complex for cellulose chain elongation (Haigler et al., 2001). Suppression of sucrose synthase gene expression repressed initiation and elongation of cotton fibre cells and development of seed coat fibres (Ruan et al., 2003). Korrigan has been characterized as a membrane-localized β-1,4-glucanase (Nicol et al., 1998; Lane et al., 2001; Molhoj et al., 2001, 2002; Sato et al., 2001; Master et al., 2004; Szyjanowicz et al., 2004; Bhandari et al., 2006; Paredez et al., 2006) and may be involved in removal of non-crystalline glucan chains and/or reduction of tensional stress (Molhoj et al., 2002; Somerville, 2006).
Cellulose synthases (CesAs) are the only identified components of rosettes. Genes encoding plant cellulose synthases were first identified in cellulose-enriched cotton fibres (Pear et al., 1996). Expression of the cloned cotton cellulose synthase genes GhCesA1 and GhCesA2 was highly correlated with secondary wall synthesis. Characterization of Arabidopsis cellulose-deficient mutants led to the identification of 10 CesA genes: AtCesA1 to AtCesA10 (Richmond and Somerville, 2000). AtCesA1, AtCesA3, and AtCesA6 are required in primary wall synthesis (Arioli et al., 1998; Scheible et al., 2001; Beeckman et al., 2002; Burn et al., 2002; Desprez et al., 2002; Doblin et al., 2002; Zhong et al., 2003) while AtCesA4, AtCesA7, and AtCesA8 appear to coordinate cellulose biosynthesis in the secondary walls (Taylor et al., 1999, 2000, 2003; Gardiner et al., 2003). The functions of AtCesA2, AtCesA5, AtCesA9, and AtCesA10 are currently unknown. In maize, 12 CesA genes have been found (Appenzeller et al., 2004). ZmCesA10, ZmCesA11, and ZmCesA12, being highly homologous to AtCesA4, AtCesA7, and AtCesA8, are suggested to be involved in secondary wall formation. Additionally, three rice CesA genes are required for cellulose biosynthesis in the secondary walls (Tanaka et al., 2003). In aspen, PtrCesA1, PtrCesA2, and PtrCesA3 are involved in secondary wall formation, while PtrCesA4, PtrCesA5, PtrCesA6, and PtrCesA7 are involved in primary wall formation (Joshi et al., 2004).
Wood is a major lignocellulosic biomass and it provides raw materials for various industrial products. Because wood, by dry weight, is ∼50% cellulose (Fengel and Wegener, 1984), understanding its biosynthesis is of vital importance to improving its production with biotechnology. Several full-length CesA cDNAs have been cloned from aspen (Wu et al., 2000; Samuga and Joshi, 2002, 2004; Kalluri and Joshi, 2003, 2004; Liang and Joshi, 2004), hybrid poplar (Djerbi et al., 2004), eucalyptus (Ranik and Myburg, 2006), and loblolly pine (Nairn and Haselkorn, 2005). In Populus trichocarpa, 18 CesA gene sequences have been identified of which seven appear to be xylem specific (Djerbi et al., 2005; Suzuki et al., 2006).
The CesA sequence information provided a basis for investigating the molecular regulation mechanism of cellulose biosynthesis in tree species. Compared with herbaceous species, woody plants are unique in their cellulose biosynthesis: they grow for multiple years and the characteristics of their cellulose fibres change with age. Inherent and environmental factors regulate cellulose synthesis. Under tension stress, for example, angiosperm trees boost cellulose synthesis and develop cellulose-rich woody tissues, termed ‘tension wood’ (Sinnott, 1952; Barnett, 1981; Timell, 1986). The formation of tension stress-induced cellulose may be associated with the expression of specific CesA genes in quaking aspen; the regulatory mechanisms of this process, however, remain to be elucidated (Wu et al., 2000; Bhandari et al., 2006).
Here, the isolation and characterization of three CesA genes (EgraCesAs) from an economically important tree species, Eucalyptus grandis, are reported. The three EgraCesAs were co-expressed in developing secondary xylem tissues, where they also exhibited differential responses to tension stress. Promoter deletion analysis further revealed a mechanical stress responsive element (MSRE)-containing DNA fragment in the EgraCesA3 promoter.
Materials and methods
cDNA library construction and screening
Total RNA was extracted from developing xylem tissues of 3-year-old, field-grown E. grandis trees by the CTAB RNA extraction method (Chang et al., 1993). The mRNA was isolated using a Poly(A) Quik mRNA Isolation Kit (Stratagene, La Jolla, CA, USA). About 7 μg of mRNA was applied to construct a cDNA library using the ZAP-cDNA Synthesis Kit and the Gigapack III Gold Packaging Extract according to the manufacturers suggestions (Stratagene). The constructed cDNA library consisted of 3.3×106 pfu with an insert size range of 1.0–4.0 kb.
The amplified eucalyptus developing xylem cDNA library was screened at 48 °C for ∼60 000 pfu using a 32P-labelled probe derived from the 0.5 kb aspen cellulose synthase cDNA (Wu et al., 2000) fragment surrounding the conserved QVLRW motif. About 20 positives were selected after primary screening. Further purification, in vivo excision, and sequencing produced three full-length cellulose synthase cDNAs which were designated EgraCesA1, EgraCesA2, and EgraCesA3.
Southern and northern blot analyses
Genomic DNA was extracted from leaves of 3-year-old, field-grown E. grandis trees by the CTAB method (Song et al., 2006). Briefly, 1 g of ground leaf tissue was added to 5 ml of extraction buffer [3% CTAB, 100 mM TRIS (pH 8.0), 20 mM EDTA (pH 8.0), 1.4 M NaCl, 1.2% PVP, and 100 μl of β-mercaptoethanol] and incubated at 60 °C for 1 h. Nucleic acids were extracted with chloroform–isoamyl alcohol (24:1) once and precipitated with isopropanol at –20 °C for 20 min, followed by washing with 70% ethanol. The isolated nucleic acid sample was digested with RNase A and then extracted with phenol–chloroform–isoamyl alcohol (25:24:1) once and chloroform–isoamyl alcohol (24:1) once. Genomic DNA was precipitated in ethanol and dissolved in sterile water. Aliquots of 10 μg of genomic DNA were digested with EcoRI, HindIII, or XbaI. The digested DNA was fractionated on a 0.8% agarose gel and blotted onto nylon membranes. Total RNA was isolated from eucalyptus leaves and developing xylem tissues by the CTAB RNA extraction method (Chang et al., 1993). Aliquots of 10 μg of total RNA were electrophoresed and transferred onto nylon membranes. Blots were hybridized at 60 °C with randomly primed, 32P-labelled probes from the 5′ regions of EgraCesA1 (0.9 kb), EgraCesA2 (0.5 kb), and EgraCesA3 (0.7 kb). Hybridization was carried out overnight at 60 °C in a hybridization buffer containing 5× SSC, 5× Dehardt's solution, and 0.5% SDS. Blots were washed at 60 °C in washing buffer 1 (2× SSC and 0.1% SDS) for 15 min and in washing buffer 2 (0.5× SSC and 0.1% SDS) for 10–40 min and then exposed to X-ray film at –80 °C.
Genomic DNA library construction and screening
The genomic DNA library was constructed using the Lambda DASH II/EcoRI Vector system (Stratagene). Genomic DNA isolated from leaves of 3-year-old, field-grown E. grandis trees was digested with EcoRI and dephosphorylated by calf intestinal phosphatase. DNA fragments were size fractionated by a Sepharose CL-2B gel filtration column and then ligated with T4 DNA ligase into the Lambda DASH II vector as per the manufacturer's protocol. The ligation products were packaged using Gigapack III XL packaging extract (Stratagene). Titration of the packaging reaction mixture revealed that the primary genomic library consists of 1.4×106 pfu.
The genomic DNA library was screened at 62 °C for about 40 000 pfu using 32P-labelled probes derived from the 5′ regions of EgraCesA1 (0.9 kb), EgraCesA2 (0.5 kb), or EgraCesA3 (0.7 kb). Genomic DNA clones harbouring EgraCesA1, EgraCesA2, or EgraCesA3 were isolated after two rounds of phage screening and hybridization. The largest isolated DNA fragments were mapped, subcloned, and then sequenced.
In situ mRNA localization
In situ hybridization was performed as previously described (Wu et al., 2000; Harding et al., 2002; Lu et al., 2006). To generate DIG-labelled probes, the 5′ regions of EgraCesA1 (667 bp), EgraCesA2 (506 bp), or EgraCesA3 (631 bp) cDNAs were firstly subcloned into the pBluescript SK- (Stratagene) vector. DNA fragments between the T3 and T7 primer sequences in constructs were then amplified by PCR. DIG-labelled sense and antisense RNA probes were prepared using an in vitro transcription kit (Roche, Indianapolis, IN, USA). Stems of several 1-year-old, greenhouse-grown eucalyptus plants, with or without 3 d of bending, were harvested and used for in situ hybridization. Sections were viewed using a Nikon microscope system with image capture by a Nikon digital imaging system.
To generate GUS-specific probes, pBI121 plasmid was double-digested by XbaI and EcoRI. The 2.2 kb XbaI–EcoRI fragment containing the polylinker, GUS open reading frame, and NOS terminator was recovered and ligated into the XbaI–EcoRI gap of pBluescript SK- (Stratagene) to produce pSKGUS-1. The pSKGUS-1 plasmid was then digested by EcoRV and the largest fragment was recovered and self-ligated to produce pSKGUS-RV-3. The 0.6 kb XbaI–EcoRV fragment in the 5′ region of the GUS open reading frame was PCR amplified using T3 and T7 primers, and used as an in vitro transcription template. The transgenic tobacco plants grown in a greenhouse were bent for 3 d to induce tension stress before being harvested for in situ hybridization.
Construction of EgraCesA promoter::GUS reporter gene chimeras
To construct A1p-bsm, PCR was first performed on eucalyptus genomic DNA using A1PF (5′-GGAAGCTTCCCGGGTGAGCTCAAGCCTCATCAACGC-3’) as a forward primer and A1PR2 (5′-GGTCTAGAGTGGCATTAGCAGAAGCTGTAGAG-3′) as a reverse primer. Fragment A1PR2F2 (accession no. EU165714) was recovered and sequenced, double-digested with HindIII and XbaI, and then ligated into the pBI121 vector (Clontech, Mountain View, CA, USA) to replace the CaMV 35S promoter.
To construct A1p-xho/bsm, A1p-bsg/bsm, and A1p-dde/bsm, the A1PR2F2 fragments in pCR2.1 vector (Invitrogen, Carlsbad, CA, USA) were first digested by BsmI and then partially digested by XhoI, BsgI, or DdeI to remove the corresponding DNA fragments. The ends of digested fragments were blunted by T4 DNA polymerase and self-ligated by T4 DNA ligase. The resulting EgraCesA1 promoter DNA fragments were then released from the pCR2.1 vector by double-digesting with HindIII and XbaI and subsequently ligated into pBI121 to replace the CaMV 35S promoter.
To construct A2p, PCR was first performed on eucalyptus genomic DNA using A2PF (5′-GGAAGCTTAGGCCTACGCCAGTTTTGAGATGCAGAT-3′) as a forward primer and A2PR (5′-GGTCTAGAGTGTCGGGGAGGCCACGGGCCGA-3′) as a reverse primer. A fragment designated A2PRFD3 (accession no. EU165715) was recovered and sequenced and then digested with HindIII and XbaI and ligated into the HindIII–XbaI gap of pBI121 to replace the CaMV 35S promoter.
The EgraCesA3 promoter was PCR-amplified using A3P-HS5 (5′-GGAAGCTTCCCGGGGAATTCAATCTGGTGCGATTT-3′) as a forward primer and A3P-XB3 (5′-GGTCTAGAATGAGTTCTTGGCGATACCCAAG-3′) as a reverse primer, and lambda DNA of the EgraCesA3 gene clone as a template. The PCR product was ligated with pCR2.1. A clone designated A3P18 was chosen for generation of A3p and deletion constructs. To construct A3p, the EgraCesA3 promoter was double-digested from A3P18 using HindIII and XbaI, and then used to replace the CaMV 35S promoter in pBI121.
To construct A3pssp/hpa, A3pssp, A3phpa, A3pbsg, A3phinc, A3pacc, and A3pmfe, A3P18 was double-digested with SspI/HpaI, SmaI/SspI, SmaI/HpaI, SmaI/BsgI, SmaI/HincII, SmaI/AccI, or SmaI/MfeI, respectively, followed by blunt-end ligation. EgraCesA3 promoter fragments were double-digested from these constructs using HindIII and XbaI, and then ligated into the HindIII–XbaI gap of pBI121 to replace the CaMV 35S promoter.
A3pssp-hpa/bsg and A3pssp-hinc were constructed by digesting A3pssp (in pCR2.1 vector) with HpaI/BsgI or HincII, followed by blunt-end ligation. The resulting EgraCesA3 promoter fragments replaced the CaMV 35S promoter in pBI121. Similarly, A3phpa-bsg/hinc was constructed by digesting A3phpa (in pCR2.1 vector) with HincII/BsgI, followed by blunt-end ligation. The constructed plasmid was further digested with HindIII and XbaI and ligated into the HindIII–XbaI gap of pBI121.
Agrobacterium-mediated transformation and GUS activity assay
EgraCesAp::GUS constructs were mobilized into Agrobacterium tumefaciens strain C58 by the freeze and thaw method (Holsters et al., 1978). Leaf disc transformation of tobacco (Nicotiana tabacum cv. Havana) was carried out as previously described (Hu et al., 1998). Half of the transgenic internodes were hand-sectioned and stained for GUS activity (Hu et al., 1998). The other half were stored in liquid nitrogen for future quantitative analysis of GUS enzyme activity (Lu et al., 2004).
Molecular cloning of eucalyptus cellulose synthase cDNAs and genes involved in wood formation
In order to identify CesA genes involved in wood formation, a cDNA library was constructed using mRNAs isolated from developing xylem of E. grandis. This cDNA library was screened using a probe derived from aspen CesA cDNA (Wu et al., 2000). The screening and subsequent phage purification yielded three full-length CesA cDNAs, which were designated EgraCesA1 (Eucalyptus grandis CesA1) (accession no. EU165708), EgraCesA2 (accession no. EU165709), and EgraCesA3 (accession no. EU165710). Like other plant cellulose synthases, all the predicted EgraCesA proteins contain the conserved D, D, D, QXXRW motif which is a signature of processive β-glycosyltransferases (Saxena et al., 1995; Delmer, 1999; Joshi et al., 2004; Somerville, 2006). Additionally, there are eight highly conserved cysteine residues in four pairs of CX2C in the N-terminal region of each EgraCesA protein. These cysteine residues appear to form the putative LIM-like zinc-binding domain/RING finger domain (Arioli et al. 1998), which can bind two atoms of Zn2+ under reduced conditions. It has been suggested that through this domain, cellulose synthases can interact to form homo- or heterodimers under oxidative conditions (Delmer, 1999; Doblin et al., 2002; Kurek et al., 2002).
Next the genomic organization of the EgraCesA genes was investigated using Southern hybridization. DNA gel blots containing genomic DNA digested with EcoRI, HindIII, and XbaI were hybridized with probes derived from EgraCesA1, EgraCesA2, and EgraCesA3, respectively. These probes shared 47–51% identity and exhibited no cross-hybridization under the experimental conditions used (data not shown). As shown in Fig. 1, each probe yielded a discrete and simple hybridization pattern, suggesting that each EgraCesA belongs to a small gene subfamily or exists as a single copy gene in the eucalyptus genome.
In order to clone EgraCesA genes, a genomic DNA library of eucalyptus was constructed and screened using the same probes employed in the above Southern blot analysis. After two rounds of plaque purification and hybridization, the corresponding genomic clones for EgraCesA1, EgraCesA2, and EgraCesA3 were isolated. As shown in Fig. 2, EgraCesA1 (accession no. EU165711) contains 15 introns, two of which are located in the 5′ untranslated region (UTR). The number of introns in EgraCesA2 (accession no. EU165712) and EgraCesA3 (accession no. EU165713) is 11 and 13, respectively. Additionally, the genomic clones of the EgraCesA1, EgraCesA2, or EgraCesA3 contain a 0.2, 2, and 1.8 kb promoter sequence, respectively.
Sequence comparison of eucalyptus cellulose synthase cDNAs
The three eucalyptus CesA genes share a 54–63% nucleotide sequence identity and a 65–72% amino acid sequence identity. Multiple sequence alignment of proteins was performed for three EgraCesAs, 10 Arabidopsis CesAs (AtCesAs; Richmond and Somerville, 2000), three cotton CesAs (GhCesAs, Pear et al., 1996; Laosinchai et al., 2000; Kim and Triplett, 2001), seven quaking aspen CesAs (PtrCesAs; Wu et al., 2000; Samuga and Joshi, 2002, 2004; Kalluri and Joshi, 2003, 2004; Liang and Joshi, 2004), and 18 P. trichocarpa CesAs (PtCesAs; Suzuki et al., 2006) (Fig. 3). EgraCesA1 was classified into a clade with AtCesA3, GhCesA3, PtrCesA5, and four PtCesAs (Fig. 3). AtCesA3 was highly expressed in the stems, flowers, and roots of Arabidopsis (Scheible et al., 2001) and is required for deposition of primary wall cellulose (Burn et al., 2002). However, it is unknown whether AtCesA3 is also required for secondary wall deposition. Similarly, GhCesA3, the other member in this clade, was found expressed in the fibre cells which form the primary and secondary cell walls (Laosinchai et al., 2000). PtrCesA5 was highly expressed in developing xylem undergoing significant secondary cell wall biosynthesis and weakly expressed in leaves, young internodes, and apices undergoing significant primary cell wall biosynthesis (Kalluri and Joshi, 2003). In accordance with these results, one of the four PtCesAs in this clade, PtCesA13, was strongly expressed in developing xylem cells while only weakly expressed in leaves, shoot tips, and phloem tissues (Suzuki et al., 2006).
EgraCesA2 is grouped with AtCesA4, PtrCesA3, and PtCesA4, while EgraCesA3 is grouped with AtCesA8, GhCesA1, GhCesA4, PtrCesA1, PtCesA8, and PtCesA18 (Fig. 3). All CesAs in these two clades may be involved in secondary cell wall deposition, indicating that EgraCesA2 and EgraCesA3 may be important in the biosynthesis of cellulose during eucalyptus wood formation.
Expression of EgraCesA1, EgraCesA2, and EgraCesA3 in eucalyptus
To examine the expression pattern of EgraCesAs, total RNAs were extracted from leaves and developing xylem tissues of eucalyptus and northern blot analysis was performed (Fig. 4). Transcripts of EgraCesA1 and EgraCesA2 were detected in both leaves and developing xylem tissues; however, these genes were more highly expressed in developing xylem tissues than leaves. This is particularly true for EgraCesA1 whose signal in leaf tissue was extremely low. EgraCesA3 was strongly expressed in developing xylem, but no signal was detected in leaves.
Expression patterns of the cloned EgraCesAs in eucalyptus stems were further investigated by in situ localization. In the second (Fig. 5A, I, Q) and the third (Fig. 5B, J, R) stem internodes, which represent early-stage primary growth with the development of primary xylem, no transcripts of EgraCesA1, EgraCesA2, or EgraCesA3 were detected in the epidermis, cortex, or pith. In the xylem of these internodes, only weak signals were detected in the developing vessel cells. With the development of secondary xylem, transcripts of these three EgraCesAs began accumulating in developing xylem cells of the fourth internode (Fig. 5C, K, S). In the sixth (Fig. 5D, E, M, N, U, V) and the eighth (Fig. 5F, O, W) internodes where the secondary xylem was undergoing active development, EgraCesA transcripts were abundant in developing xylem cells. In the older tissues (Fig. 5G, H, O, P, W, X), EgraCesAs showed continuous expression in developing xylem cells and some phloem fibre cells. Thus, expression of EgraCesAs apparently concurred with secondary xylem development in eucalyptus.
Expression of EgraCesA1, EgraCesA2, and EgraCesA3 in response to mechanical stress in eucalyptus
To support the crown structure and the mechanical loads generated by gravity and wind, inclined stems of angiosperm trees form tension wood in the upper part of the bent stem to counteract the tension stress. Opposite wood forms in the lower part of the inclined stem in response to compressive forces (Sinnott, 1952; Barnett, 1981; Timell, 1986). The gelatinous or G layer of tension wood contains more cellulose and less lignin than normal wood. Conversely, opposite wood contains more lignin and less polysaccharide than normal wood (Timell, 1986). Because of these unique characteristics, this specialized wood system provides an excellent platform for the study of cellulose biosynthesis.
In order to understand the regulation mechanisms of EgraCesAs in wood formation, greenhouse-grown eucalyptus plants were bent at the tenth internode at an angle of 45° for 3 d. Cross-sections from the bent stem were then hybridized with probes derived from EgraCesA1, EgraCesA2, or EgraCesA3 (Fig. 6). In tension-stressed xylem tissues, EgraCesA2 and EgraCesA3 produced intensified transcript signals in cell layers close to the cambium zone (Fig. 6C, E). The opposite sides, however, produced much weaker transcript signals (Fig. 6D, F). The up-regulation of EgraCesAs may be attributable to cellulose accumulation in tension wood. Moreover, the similar expression patterns of EgraCesA2 and EgraCesA3 suggest they are co-regulated in tension-stressed xylem tissues. Unlike EgraCesA2 and EgraCesA3, EgraCesA1 was less intensely expressed in tension-stressed xylem tissues (Fig. 6A), indicating that the signalling pathways regulating EgraCesA1 expression differ from those controlling EgraCesA2 and EgraCesA3. The EgraCesAs are up-regulated in phloem fibre cells of the opposite side (Fig. 6B, D, F) with EgraCesA2 responding most sensitively (Fig. 6D). Consistently, the expression level of EgraCesA2 in the xylem of the opposite side (Fig. 6D) is higher than that of EgraCesA1 and EgraCesA3 (Fig. 6B, F).
Isolation and dissection of EgraCesA promoters
To investigate further the regulatory mechanisms of EgraCesA expression, promoter regions of the EgraCesA genes were cloned. The 2.0 kb and 1.8 kb upstream flanking regions of EgraCesA2 and EgraCesA3 cDNAs, respectively, were isolated by screening a genomic DNA library (Fig. 2). The isolation of the EgraCesA1 promoter was difficult because the EgraCesA1 gene clone contained only a short (about 0.2 kb) upstream flanking region of cDNA (Fig. 2). In order to obtain a longer EgraCesA1 promoter region, two rounds of genome walking were performed using eucalyptus genomic DNA as a template. The first round used gene-specific primers complementary to the 5′ end region of the EgraCesA1 gene clone and produced a 0.8 kb DNA fragment clone. The second round used gene-specific primers complementary to the 5′ end region of the 0.8 kb fragment isolated in the first round. Through genome walking, 1.4 kb of EgraCesA1 cDNA upstream flanking sequence was obtained (Fig. 2). Sequence comparison revealed that the upstream flanking regions of EgraCesA cDNAs were divergent and shared only 42, 43, and 52% identity between EgraCesA2 and EgraCesA3, EgraCesA1 and EgraCesA2, and EgraCesA1 and EgraCesA3, respectively, suggesting the transcriptional regulation mechanisms of each EgraCesA gene could be different.
Then a series of entire or deleted EgraCesA promoter::GUS reporter gene chimeras were constructed (Figs 7, 8) and introduced into tobacco using the Agrobacterium tumefaciens-mediated leaf disc transformation method (Hu et al., 1998). Transgene integration in tobacco was validated by PCR (data not shown). Transgenic tobaccos were assayed for GUS activity at various growth stages from young to mature flowering plants.
In order to dissect the EgraCesA1 promoter, initially a construct A1p-dde/bsm with the 1.4 kb upstream flanking sequence of EgraCesA1 cDNA but lacking any introns was created (Fig. 7A). Unexpectedly, nine of the 10 analysed transgenic lines showed no GUS staining in stem sections (Fig. 9A, B). Only one line showed weak GUS staining in the inner phloem of older internodes, such as internode 16 (Fig. 9C, D). This suggests the 1.4 kb upstream flanking sequence was insufficient to direct appropriate expression of this gene in planta. The EgraCesA1 upstream region contains two 5′-UTR introns (Fig. 2, 7A), the first of which is near the 5′ end of the cloned cDNA and is 119 bp in length, and the second of which is near the translation start site and is greater in length (295 bp). It has been reported that the 5′-UTR introns of EF1α-A1 and EF1α-A3 genes in Arabidopsis enhanced the expression of GUS and firefly luciferase reporter genes in transgenic plants (Curie et al., 1992, 1993; Chung et al., 2006). In order to assess whether the 5′-UTR introns of EgraCesA1 were involved in the regulation of its expression, three additional constructs were made containing (i) the 1.4 kb upstream flanking sequence and two introns (A1p-bsm), (ii) the entire first intron and partial second intron (A1p-xho/bsm), or (iii) the entire first intron but no second intron (A1p-bsg/bsm) (Fig. 7A). Observed GUS activity in stem sections of tobacco plants carrying these constructs produced results similar to those in A1p-dde/bsm transgenic plants. This suggests that other regulatory sequences may be required to direct appropriate EgraCesA1 expression in planta.
Unlike the EgraCesA1 promoter, the EgraCesA2 promoter (A2p in Fig. 7B) was able to drive GUS gene expression appropriately in transgenic tobacco plants. All A2p transgenic tobacco lines showed intense GUS staining in xylem cells of stem internodes 9–14 (Fig. 9G, H), where xylem cells were undergoing strong secondary cell wall thickening. Younger and older internodes exhibited relatively low GUS activity (Fig. 9E, F). No GUS staining was shown in the xylem cells of stem internodes 1–6. These results are consistent with the results of in situ hybridization of EgraCesA2 transcripts in eucalyptus (Fig. 5I–P) which revealed EgraCesA2 to be specifically expressed in developing xylem.
EgraCesA3 showed the strongest xylem-specific expression (Fig. 4) and up-regulation in response to tension stress (Fig. 6E, F). In order to analyse the EgraCesA3 promoter in detail, 11 constructs with the entire or deleted EgraCesA3 promoter fused to the GUS reporter gene were generated (Fig. 8). Analysis of the resultant transgenics suggested that the EgraCesA3 promoter fragments analysed were able to drive GUS gene expression. The expression patterns produced were similar to those in the A2p transgenic plants. Strong GUS activity was found in the xylem cells of stem internodes 9–14 (Fig. 9I–T), but no activity was found in the xylem cells of stem internodes 1–6. Based on GUS staining intensity in xylem and phloem tissues (Figs 8, 9I–T), several cis-regulatory regions in the EgraCesA3 promoter were deduced that were probably involved in either enhancing or repressing gene transcription (Fig. 10). Deletion of the –1787 to –1188 region (A3pssp, A3pssp-hpa/bsg, and A3pssp-hinc) from the entire EgraCesA3 promoter (A3p) increased GUS activity in the xylem cells of stem internodes, indicating that this region contains a negative regulation sequence controlling gene expression in xylem. Deletion of the –1188 to –749 region (A3p-ssp/hpa or A3phpa) from A3p or A3pssp decreased GUS activity in stem xylem cells, indicating that this sequence in the –1188 to –749 region positively regulated gene expression in xylem. The –749 to –616 region appeared to regulate gene expression negatively in the phloem given that deletion of this region (A3pssp-hpa/bsg) from A3pssp increased GUS activity in the phloem cells. Deletion of the –616 to –481 region (A3pssp-hinc, A3phpa-bsg/hinc, or A3phinc) from A3pssp-hpa/bsg, A3phpa, and A3pbsg decreased GUS activity in the xylem and phloem cells, indicating that this region could positively regulate gene expression in both the xylem and phloem.
Mechanical stress-responsive elements in EgraCesA promoters
Using in situ hybridization, both EgraCesA2 and EgraCesA3 were found to be up-regulated under tension stress in developing xylem tissues of eucalyptus stem (Fig. 6). In order to test whether this up-regulation occurred at the transcriptional level, EgraCesA promoter-directed GUS expression was analysed in response to mechanical stress. Transgenic tobacco plants carrying A2p or A3p (Figs 7B, 8) were bent for 3 d to induce stress. These stressed portions were then sectioned and stained for GUS activity. Surprisingly, GUS activity was similar between the tension-stressed and the opposite xylem tissues of the A2p and A3p transgenics (Fig. 11A, B). This similarity was likely to be due to the long half-life of GUS protein (>50 h) in planta (Jefferson et al., 1987) that led to GUS staining indiscernible in a short period of stress treatment. To test the validity of such a hypothesis, the accumulation of GUS transcripts in bent transgenic tobacco was analysed using in situ hybridization. GUS in situ results supported our hypothesis. GUS transcripts in tension-stressed xylem cells (Fig. 11At, Bt) were found to be significantly more abundant than those in the opposite xylem cells (Fig. 11Ac, Bc). These results were consistent with the observation of in situ EgraCesA2 and EgraCesA3 transcript hybridization in eucalyptus (Fig. 6C–F). Overall, this suggests that tension stress-related up-regulation of EgraCesA transcripts in xylem tissues is due to transactional regulation imparted by the regulatory elements existing in the promoter regions.
In order to find possible cis-elements responsive to the mechanical stress signal, GUS expression was analysed in tension-stressed transgenic tobacco plants carrying the deleted EgraCesA3 promoter::GUS reporter gene chimeras (Fig. 8). Similar to the A2p and A3p transgenics (Fig. 11A, B), tension-stressed xylem tissues of tobacco plants carrying these chimeras did not exhibit differential GUS staining in the opposite xylem tissues (Fig. 11). In situ hybridization revealed GUS transcript accumulation in tension-stressed xylem cells of transgenic plants carrying A3p-ssp/hpa (Fig. 11Ct), A3pssp (Fig. 11Dt), A3pssp-hpa/bsg (Fig. 11Et), A3pssp-hinc (Fig. 11Ft), A3phpa (Fig. 11Gt), A3phpa-bsg/hinc (Fig. 11Ht), A3pbsg (Fig. 11It), or A3phinc (Fig. 11Jt). However, GUS gene expression was similar in both tension-stressed and the opposite xylem cells of transgenics carrying A3pacc (Fig. 11Kt, Kc) or A3pmfe (Fig. 11Lt, Lc). Thus, the –481 to –331 region in the EgraCesA3 promoter is likely to contain mechanical stress-responsive elements (MSREs) (Fig. 10). This segment had a length of 155 bp and was named the HincII–AccI fragment.
Searching the PLACE database (Higo et al., 1999; http://www.dna.affrc.go.jp/PLACE/index.html), a number of cis-element sequences were found in the HincII–AccI DNA fragment, including W-box (Yu et al., 2001), W box (Nishiuchi et al., 2004), MYC recognition site (also known as E-box or RRE; Stalberg et al., 1996; Abe et al., 2003; Hartmann et al., 2005), EEC element/CCA1-binding site (Wang et al., 1997; Kucho et al., 2003; Yoshioka et al., 2004), and ARR1-binding element (Sakai et al., 2000; Ross et al., 2004) that were specifically recognized by WRKY, ERF3, bHLH, MYB, and ARR1, respectively. These elements were stress-responsive and may be involved in plant response to mechanical stress (Fig. 10). Further deletion analysis of the HincII–AccI fragment may help characterize the MSREs.
From the eucalyptus developing xylem cDNA library, three cellulose synthase cDNAs, EgraCesA1, EgraCesA2, and EgraCesA3, were cloned. EgraCesA1 shared low nucleotide homology (56–63%) and amino acid identity (67–72%) with the other two EgraCesAs. The deduced EgraCesA1 protein was classified into a clade with AtCesA3, GhCesA3, PtrCesA5, and four PtCesAs (Fig. 3). Members in this clade are distinct in their roles in the different plant species although their sequences share high homology. AtCesA3 is highly expressed in stems, flowers, and roots of Arabidopsis (Scheible et al., 2001) and is required for deposition of primary wall cellulose (Burn et al., 2002). In contrast, EgraCesA1 was expressed only in secondary cell wall-depositing cells, such as developing vessel cells in the second and third internodes and developing fibre cells in the fourth to tenth internodes (Fig. 5A–H). Its transcripts were also detected in xylem fibre cells at the late stage of cell wall thickening (Fig. 5H). Based on this evidence, EgraCesA1 appears to be involved in secondary cell wall deposition. The other members in this clade were also highly expressed in stems, except PtCesA3 and PtCesA12 which were undetectable in the tissues tested (Laosinchai et al., 2000; Kalluri and Joshi, 2003; Suzuki et al., 2006). The four tree CesAs in this clade, EgraCesA1, PtrCesA5, PtCesA13, and PtCesA14, accumulated transcripts specifically in developing xylem cells undergoing secondary cell wall biosynthesis (Fig. 5; Kalluri and Joshi, 2003; Suzuki et al., 2006), suggesting the tree CesAs in this clade are significant to wood formation.
EgraCesA2 and EgraCesA3 are grouped in two distinct clades with CesAs involved in secondary cell wall deposition (Fig. 3). EgraCesA2 and EgraCesA3 are similar to EgraCesA1 in that they are also secondary xylem specific (Figs 4, 5). EgraCesA2 and EgraCesA3 transcripts were detectable only in eucalyptus stem cells undergoing secondary cell wall thickening and the EgraCesA2 and EgraCesA3 promoter-driven GUS expression in the developing xylem of transgenic tobacco stems (Fig. 9). Thus, the three EgraCesA genes were involved in secondary cell wall deposition, which is consistent with the concept that multiple cellulose synthase genes are required for cellulose biosynthesis (Doblin et al., 2002; Joshi et al., 2004; Somerville, 2006).
It is suggested that three secondary cell wall-related CesA genes are present in the genome of Arabidopsis (AtCesA4, AtCesA7, and AtCesA8) (Taylor et al., 1999, 2000, 2003; Gardiner et al., 2003, maize (ZmCesA10, ZmCesA11, and ZmCesA12) (Appenzeller et al., 2004), aspen (Joshi et al., 2004; Bhandari et al., 2006), and rice (OsCesA4, OsCesA7, and OsCesA9) (Tanaka et al., 2003). However, the number of secondary cell wall-related CesA genes in some tree species may be higher, indicating the cellulose synthase families of some tree species are more complex than those of other plant species. In P. trichocarpa, the only sequenced tree species (http://genome.jgi-psf.org/Poptr1_1/Poptr1_1.home.html), 18 cellulose synthase genes (PtCesAs) have been annotated (Suzuki et al., 2006). Transcripts of 15 PtCesAs were detected in developing xylem tissues, seven of which were secondary xylem specific. There are likely to be other secondary cell wall-related CesAs that have not yet been identified in eucalyptus. The existence of multiple secondary cell wall-related CesAs in tree species suggests that cellulose biosynthesis is controlled by a complex regulatory network. It has also been suggested that these tree CesAs may show differential responses to specific growth conditions (Wu et al., 2000). This hypothesis was tested by analysing expression patterns of EgraCesAs in response to mechanical stress.
During the development of angiosperm xylem, cell wall morphology and composition are regulated by various environmental conditions such as mechanical stress generated by wind, snow, or bending. Consistent with known stress reaction mechanisms, aspen xylem-specific PtrCesA1, PtrCesA2, and PtrCesA3 genes are up-regulated under tension stress in the developing secondary xylem (Wu et al., 2000; Bhandari et al., 2006). The up-regulation of EgraCesA2 and EgraCesA3 was observed in tension-stressed eucalyptus xylem cells using in situ hybridization (Fig. 6). The up-regulation of EgraCesA2 and EgraCesA3 was further confirmed to be controlled by promoter sequences using promoter-GUS analyses (Fig. 11). Thus, in tension wood, both EgraCesA2 and EgraCesA3 are involved in pure cellulose biosynthesis. This cellulose has a high degree of crystallinity although the regulatory mechanisms of cellulose crystallinity are currently unknown. EgraCesA2 and EgraCesA3 proteins may be subunits of the same cellulose biosynthesis machinery as evidenced by their co-expression in developing eucalyptus secondary xylem and their co-regulation in response to tension stress. Gene co-expression and co-regulation have been previously found for AtCesA4, AtCesA7, and AtCesA8, which are subunits of a complex essential for cellulose synthesis in secondary cell walls (Taylor et al., 2003).
Sequence comparison among the three isolated EgraCesAs showed that EgraCesA1 had the longest cDNA and gene (Fig. 2) sequences and the greatest number of introns (Fig. 2). EgraCesA1 contains two 5′-UTR introns (Figs 2, 7A) which could enhance gene expression in plants (Curie et al. 1992, 1993; Chung et al. 2006). The EgraCesA1 promoter alone or with the 5′-UTR introns was not sufficient to direct the appropriate expression of GUS in transgenic tobacco plants (Fig. 9), indicating a complex regulatory network for EgraCesA1 transcription. EgraCesA2 and EgraCesA3, but not EgraCesA1, were up-regulated in tension-stressed xylem cells (Fig. 6), indicating that EgraCesA1, although associated with secondary cell wall deposition, is not likely to be a subunit of the EgraCesA2–EgraCesA3 complex. The presence of two types of cellulose biosynthesis machinery in wood formation is consistent with the finding of multiple secondary cell wall-related CesAs in tree species (Bhandari et al., 2006; Suzuki et al., 2006). Gene suppression analysis of EgraCesAs will provide important new insights into the function of cellulose synthases in wood formation.
Although cellulose synthase genes have been cloned from various plant species, their regulatory mechanisms are poorly understood and no cis-regulatory element has yet been identified. Because EgraCesA3 showed the strongest xylem-specific expression (Fig. 4) and up-regulation in response to tension stress (Fig. 6E, F), its promoter sequence was chosen in order to elucidate the regulatory mechanisms of CesA transcription in wood formation. Eleven constructs containing full, partial, or internal deletion of the EgraCesA3 promoter region fused to the GUS reporter gene were introduced into tobacco. Based on GUS staining (Figs 8, 9), negative (the –1787 to –1188 region) and positive (the –1188 to –749 region) regulatory regions controlling gene expression in the xylem were deduced (Fig. 10). Also a negative regulation region was found at –749 to –616 which directed gene expression in phloem and a positive regulation region at –616 to –481 which controlled gene expression in both xylem and phloem. These preliminary results will guide further studies on identifying cis-regulatory elements directing CesA gene transcription and regulatory networks of wood formation.
The distribution of EgraCesA3 promoter-directed GUS mRNAs was analysed in tension stressed stems of transgenic tobacco using in situ hybridization (Fig. 11). It was found that the –481 to –331 region in the EgraCesA3 promoter is likely to contain MSREs (Fig. 10). This region is enriched in previously characterized cis-elements such as the W-box, W box, MYC recognition site, EEC element, CCA1-binding site, and ARR1-binding element. The W-box (TTGAC) is specifically recognized by WRKY transcription factors which are associated with plant defence responses (Eulgem et al., 2000; Yu et al., 2001; Dong et al., 2003; Miao et al., 2004; Xu et al., 2006). The MYC recognition site (CANNTG), also known as E-box or R response element (RRE), is a binding site of bHLH transcription factors involved in abscisic acid signalling and cold and freezing tolerance in plants (Stalberg et al., 1996; Abe et al., 2003; Chinnusamy et al., 2003; Hartmann et al., 2005; Agarwal et al., 2006). This DNA fragment (–481 to –331) contains an EEC element (CANTTNC) and a CCA1-binding site (AAMAATCT) (Fig. 10). Both these elements are MYB transcription factor-binding sites and are related to CO2 or phytochrome regulation (Wang et al., 1997; Kucho et al., 2003; Yoshioka et al., 2004). The two ARR1-binding elements (NGATT) in this DNA fragment were specifically recognized by the cytokinin-regulated transcription factor, ARR1 (Sakai et al., 2000; Ross et al., 2004). The W box (TGACY) in this DNA fragment is of interest because it is involved in wounding activation of the ERF3 gene (Nishiuchi et al., 2004). Bending-stressed plant stems suffer multiple abiotic stresses including wounding, gravity, and light. These stress-responsive elements probably contribute to EgraCesA3 up-regulation in response to tension stress. It is also possible that some unknown cis-elements in the –481 to –331 region of the EgraCesA3 promoter are responsible for the tension stress-induced up-regulation of the EgraCesA3 gene. Further deletion and mutation analysis of the –481 to –331 region of the EgraCesA3 promoter and sequence comparison with the EgraCesA2 will help to identify the MSREs.
In the production of cellulose fibre materials, it is highly desirable to engineer trees with more cellulose and controllable cellulose properties such as degree of polymerization and crystallinity. However, little is known about the genes controlling wood cellulose formation. The discovery of differential expression of three secondary cell wall-related cellulose synthase genes in response to tension stress and the identification of an MSRE-containing DNA fragment in the EgraCesA3 promoter, however, provide an important clue for the future improvement of cellulosic material production in trees.
The project was supported by a grant from the ArborGen, LLC.