Four full-length candidate CesA cDNA sequences were isolated from leaves and stems of B. platyphylla. A search for conserved domains or functional motifs in the CDD revealed that the four encoded proteins possessed two separate conserved domains: a glycosyltransferase domain (E value = 2 × 10⁻¹¹) and a cellulose synthase domain, which are typical of CESAs (E value < 1.0 × 10⁻¹⁸⁰) [2,4]. Therefore, the four genes were confirmed as CesAs and designated BplCesA3, −4, −7 and −8, using the three-letter prefix nomenclature (Bpl) for cellulose synthase genes reported in Populus [23].

The four full-length CesAs ranged from 3255 to 3968 bp in length, containing ORF lengths ranging from 2958 to 3255 bp, encoding 985 to 1084 amino acids (Table 1). The four putative BplCESAs differed at residues 11 to 33 of the N terminus and 16 residues at the C terminus (Figure 1) and shared only 63.8% to 70.5% identity (Table 1), which was similar to that shared by the seven aspen CESAs (64%–76%) [18,22]. However, the putative BplCESAs shared the highest identity (85%–98%) with orthologs in other plants. Importantly, the four BplCESAs contained all the typical motifs of plant cellulose synthases described by Joshi [19], including the plant conserved region (CRP), eight predicted transmembrane regions, a conserved zinc finger motif (CX₂CXnCX₂CXnCX₂C) (Figure 2) and the conserved D, D, D, QxxRW signature sequence (D, D, D, QVLRW) involved in substrate binding and catalysis during cellulose synthesis (Figure 1). The predicted plant-specific conserved and hypervariable regions (HVRs) shared only approximately 16% to 35% identity. Thus, all these genes are predicted to be true CesA genes rather than cellulose synthase-like (CSL) genes as defined by Richmond and Somerville [10].

It is estimated that the Populus genome has 18 CesA genes [24,28,29], which can be grouped into eight sets of clearly defined paralogs, plus a single copy of PtiCesA4 [23,28]. We used the 39 predicted full-lengh CESA amino acid sequences from five plant species, Physcomitrella patens (PpCESA8 as outgroup), Arabidopsis thaliana (10 AtCESAs), P. trichocarpa (17 PtiCESAs), P. tremuloides (7 PtdCESAs) and B. platyphylla (4 BplCESAs) to generate a phylogenetic tree (Figure 3). The multiple sequence alignment was shown in Figure S1. The two hypervariable regions (HVRI and HVRII) of the selected CesA gene products were highly divergent and excluded for the construction of the phylogenetic tree. The four Betula CESAs clustered with specific Arabidopsis and Populus CESA homologs (>80% identity) to form discrete clades (Figure 3). For example, BplCESA8 clustered with PtiCESA8-A/PtiCESA8-B (88.8%–89.2% amino acid sequence identity), AtCESA8 (86.6%) and PtdCESA1 (87.4%). BplCESA3 was considered orthologous to the PtiCESA3-C/PtiCESA3-D paralogs (92.8%–93.1% identity), while AtCESA3 (91.3%) and PtdCESA5 (89.6%). BplCESA7 shared 89.7% to 89.3% identity with PtiCESA7-A/PtiCESA7-B, 85.0% with AtCESA7 and 89.2% with PtdCESA2. BplCESA4 clustered with the single set comprising PtiCESA4 (89.9% identity), AtCESA4 (86.1%) and PtdCESA3 (88.0%). Genes in such clusters presumably share a common evolutionary history in which an ancestral CESA existed before the divergence of Betula, Arabidopsis and Populus. This parallel numbering model suggests that the clustered CESAs share similar functions in the three species. In addition, three predicted protein sequences of the four BplCESAs shared a maximum identity of 97.8% to 99.2% with three CESA isoforms from B. luminifera, whereas the other two CESA homologs, BplCESA3 and BluCESA5, shared only 63.6% to 71.9% identity with other CESAs of the two Betula species (Data not shown). Further investigations are required to confirm that the diverged CesA genes in Betula species perform distinct roles in cellulose biosynthesis.

In order to research the transcript abundance and characteristics of the four novel BplCesAs in cell wall formation in distinct tissues, transcript expression profiles were generated in 11 different tissues of Betula by quantitative RT-PCR using gene-specific primers (Table 2). In our study, the abundance of all BplCesA mRNAs was detected and normalized to constitutively expressed ACTIN mRNA which has been improved to be qualified as a reference gene by Chao Dai et al. [30]. The results showed that BplCesA8 was highly expressed in mature tissue, including some tissues with secondary thickening, such as male and female inflorescences (FI and MI), mature petioles as well as young and mid-development stem tissues rich in secondary cell wall (Figure 4). BplCesA3 was predominantly expressed in tissues either typical of primary cell walls (young leaves in mid July) (Figure 4) or consistent with the expression pattern in early development (in July) of leaves as shown in Figure 5a and seeds (Figure 4). BplCesA7 and BplCesA4 were strictly co-expressed and expressed predominantly in developing and mature vascular tissues or tissues with secondary thickening, such as young and mature stems and male inflorescences. Another interesting result is that the four BplCesAs were expressed in mature pollen at an extremely low level compared with other tissues (Figure 4).

In addition, expression profiles for the four BplCesAs in developing leaves and stems were determined by quantitative RT-PCR using the same gene-specific primers (Table 2). Expression of BplCesA in stems was high during most developmental periods, with abundant expression of BplCesA7 and BplCesA4 (Figure 5). With the exception of BplCesA3, all BplCesAs were expressed at higher levels in stems than in leaves at most stages (Figure 5). BplCesA8 mRNA was more abundant in stems than in leaves at earlier developmental stages, whereas BplCesA3 mRNA was notably more abundant in leaves, which are richer in primary cell wall than stems at some time-points, especially earlier and later development stages (in July and September) (Figure 5). Both the BplCesA7 and BplCesA4 mRNAs were more abundant in stems (which are rich in secondary cell walls) than in leaves (which are typical of primary cell walls) during most of the examined periods (Figures 5).

No other CESA homologs have been identified in B. platyphylla in this study or elsewhere, indicating that the CesA gene family of B. platyphylla is comprised of at least four genes. It has been reported that the CesA gene family is comprised of 10 or more homologs in plants. For example, 93 cellulose synthesis-related genes have been identified in the P. trichocarpa genome, containing 17 CesA genes [23,24,28], 10 CESA homologs in Arabidopsis [12,23], and 11 CesA genes in rice [14]. There are 12 CesA gene members in Pinus [27]. The CesA gene family contains 11 CesA members in the moss (Physcomitrella patens) [31]. Here, only four BplCesAs were identified, which may be due to very low mRNA levels of other BplCesAs in certain tissues or developmental stages. Further studies are required to identify other BplCesAs in distinct tissues.

Each of the CESAs may have specific roles related to the formation of primary and secondary walls in plants. The distinct transcript expression profiles of the four BplCesAs indicated their different involvement in growth of Betula. Almost all of the four BplCesAs were highly expressed in the tissues we examined, indicating their involvement in the biosynthesis of cellulose which is the main component of plant cell walls. BplCesA8 was more highly expressed in earlier growth stages of the stem (Figure 5) and there were more abundances of BplCesA8 transcript in some tissues with secondary thickening (female and male inflorescences with abundant bracts and some mature tissues) or undergoing active secondary cell wall formation (in earlier growth as young stems) than other members (Figure 4). This indicated similarity to its homolog AtCESA8, which participates in secondary cell wall biosynthesis in Arabidopsis [11–13]. Similarly, PtiCESA8-A and PtiCESA8-B, which clustered with BplCESA8, also exhibited apparently higher xylem specificity [24]. This indicates BplCESA8 might be involved in secondary cell wall biosynthesis.

Notably we found high transcript expression levels of BplCesA8 in male and female inflorescences (Figure 4). In Arabidopsis, CESA1 appears to be possibly required for either embryo or male gametophyte growth [32,33]. CESA3 is coexpressed with CESA1, and homozygous cesa3 alleles are also male gametophyte lethal [13]. By comparison, in rice, OsCESA2,-9 were highly expressed in young panicle and the OsCSLF genes (OsCSLF2 & −7) were preferentially expressed in the hull of rice, while several CSL (cellulose synthase-like) genes (CSLG2,3 and CSLB2 in sepals, CSLG1, CSLD6 and CSLA1,2,10,11 in Carpels) are specifically expressed in flower organs in Arabidopsis [14]. The homologs of BplCESA8 were not reported to be related to the floral development either in rice or in Arabidopsis, suggesting BplCesA8 may participate in floral growth in Betula unlike the other two plant species. Despite the involvement of BplCesA8 in floral growth, there is no evidence for its involvement in male or female gametophyte growth because male and female inflorescences contain abundant bracts.

Also, we could assume BplCesA8 to be a partially redundant candidate gene with BplCesA3. BplCesA8 was detected with high transcript levels in specific tissues (male and female inflorescences, young stems) where the expression of BplCesA3 is relatively low, while BplCesA8 was lowly expressed in seed where BplCesA3 showed much higher expression (Figure 4). In other words, BplCesA3 may be partially redundant with BplCesA8 in those specific tissues. Furthermore, the expression level of BplCesA8 in female inflorescences very rich in bracts is much higher than that of BplCesA3, while BplCesA3 was much more highly expressed than BplCesA8 in developing seeds which were isolated from female inflorescences (Figure 4). The bracts in inflorescences are typical of secondary thickening cell walls. Thus, BplCesA8 might be involved in the development of bracts and be related to secondary cell wall formation in Betula as with the result mentioned above.

Both BplCesA7 and BplCesA4 were expressed at apparently higher levels in the stem at most time-points, which are rich in secondary cell walls (Figures 4 and 5), indicating that the two genes may be involved in the biosynthesis of the secondary cell wall throughout the development of the stem. In addition, they were also more highly expressed in old leaves (in September) than young leaves. It is proposed that the changes in expression may reflect a role in the synthesis of homogalacturonan, which accumulates to a high level in old leaves [14]. But in Arabidopsis, it is cellulose synthase-like (CSL) genes (AtCSLD2 and AtCSLE1) which show sequence similarity to CESA apparently exhibiting strong increases in expression in old leaves versus young leaves instead of CesA genes [34]. This suggests the differences in both transcript expression and specific importance in different tissues of the members in the CES/CSL superfamily in the plant. More importantly, BplCesA7 and BplCesA4 were strictly co-expressed. OsCESA4, −7 (homologs of BplCESA4, −7, respectively) and -9 in rice are thought to be organized as a cellulose synthase complex involved in secondary cell wall synthesis [14], suggesting that the BplCESA7 and BplCESA4 may be related to the formation of a cellulose synthase complex for secondary cell wall biosynthesis. Similarly, the secondary cell wall development-related genes PtdCESA1, PtdCESA2 and PtdCESA3 in P. tremuloides [17–19] are all homologs of BplCESA8, BplCESA7, BplCESA4, respectively (Figure 3), suggesting that these three genes might be associated with secondary cell wall development in Betula. This is also consistent with the xylem-specific expression pattern of PtiCESA7-A or PtiCESA7-B (clustered with BplCESA7) and PtiCESA4 (clustered with BplCESA4) (Figure 3) in P. trichocarpa [24]. Redundant xylem-specific expression may suggest their involvement in the massive production of cellulose in xylem secondary cell walls for wood formation [25]. However, it has been reported that CesAs related to secondary cell walls are not functionally redundant, which suggests that AtCESA4 (a homolog of BplCESA4), AtCESA7 (of BplCESA7) and AtCESA8 (of BplCESA8) are the only CESAs involved in cellulose synthesis and form a complex in the secondary cell wall in Arabidopsis [35]. In spite of this, there is insufficient evidence in this research to support the participation of BplCESA8 (homolog of AtCESA8) complexed with BplCESA7 and BplCESA 4 (homologs of AtCESA7 and AtCESA4, respectively) for secondary cell wall formation.

BplCesA3 exhibited a unique and complex expression pattern (Figures 4 and 5). Although this gene was expressed at lower levels in most tissues compared with the other three BplCesAs, higher expression was detected at the earlier development in young leaves rich in primary cell wall (Figure 4 and Figure 5a). OsCESA1, −3 (homolog of BplCESA3), −5, −6, −8 in rice and AtCESA1, −2, −3 (homolog of BplCESA3), −5, −6 were detected at higher transcript expression levels in young seedlings mainly comprised of young leaves and stem. Also, OsCESA5/OsCESA6 is likely to be partially redundant with OsCESA3 for OsCESA complex organization in the young tissues, such as plumule and radicle [14]. AtCESA3 was reported to function predominantly in the primary cell wall. Furthermore, its homolog PtdCESA5 in P. tremuloides is involved in primary cell wall development together with PtdCESA4, PtdCESA6 and PtdCESA7 [21,22]. Usually, three or four CESAs are responsible for primary cell wall biosynthesis in higher plants. AtCESA1 [36], AtCESA 3 and AtCESA6 in Arabidopsis, OsCESA1, OsCESA3 and OsCESA8 in rice [14] and PtdCESA4, PtdCESA5, PtdCESA6 and PtdCESA7 in P. tremuloides [21,22] may form a cellulose synthase complex for primary cell wall biosynthesis in the three species. Therefore, it can be speculated that other primary cell wall related CESAs exist in Betula in addition to BplCESA3. In addition, the higher expression level in the late-development stem reflects a potential relation to the formation of secondary cell wall in Betula. Finally, BplCesA3 was observed at highest expression level in old leaves (Figure 5a), which indicated that BplCesA3 may be related to the synthesis of homogalacturonan similar to BplCesA4, −7 in Betula.

Moreover, the transcript level of BplCesA3 in developing seed was notably higher than in other tissues examined (Figure 4). During the development of seed, the secondary cell walls in the seed protect the embryo. Cellulose is proposed to play a major role in reinforcing the secondary cell wall in the seed coat epidermal cells [37] and to be a component of seed mucilage [38,39]. In Arabidopsis thaliana, AtCESA9 is required for normal secondary wall synthesis in epidermal seed coat cells [37]. AtCESA2, AtCESA5, and AtCESA9 subunits contribute to secondary wall synthesis in epidermal seed coat cells. AtCESA2 and AtCESA9 serve in radial wall reinforcement, as does CESA5, but CESA5 also functions in mucilage biosynthesis in the seed coat epidermis and is indispensable for mucilage attachment to the seed coat These data suggest unique roles for different CESA subunits in one cell type [40]. Though the homolog of BplCESA3, AtCESA3, has not been reported about its participation in the development of seed, AtCESA4, −7, −8, −9, −10 were detected highly expression levels in seed and silique [14]. The high transcript level of BplCesA3 in developing seed reflects its possible involvement in development of seed in Betula. To sum up, the unique expression pattern of BplCesA3 mentioned above indicated that it is possibly participating in diverse developments in Betula compared with the other three CESAs examined. Also, BplCESA3 might possess different transcript expression mechanisms in development of Betula from other plants.

We also found an unexpected result that almost each of the four BplCesA3, −4, −7 and −8 was observed at extremely low transcript level in mature pollen in Betula (Figure 4). It has been reported that AtCESA2, AtCESA6 and AtCESA9 are coexpressed during pollen development [11]. Triple cesa2cesa6cesa9 mutant plants are pollen lethal, indicating that they function redundantly in pollen development. CESA9 has been proved to function redundantly with CESA6 during pollen development [13]. The extremely low transcript abundances of the four BplCesAs reflected their lower involvement in mature pollen development and that there would be other CesAs related to the maturity of pollen in Betula.

Young and mature leaves and stem tissues (after bark removal, the cambium and stem tissues were scraped with clean blades free of RNase contamination) were harvested from annual branches of mature birch trees growing in forests under natural conditions in the Northeast of China. Total RNA was extracted using the CTAB method and was treated with RNase-free DNase I (Promega, USA) to remove DNA contamination. The cDNA was synthesized from 2 μg total RNA using Oligo(dT) as the reverse primer in the reverse transcription-PCR (RT-PCR) system (Invitrogen, USA). The cDNA was used as a template for amplification in 50 μL PCR buffer (TaKaRa, China). Two upstream (U1 and U2) and two downstream (D1 and D2) degenerate primers (Table 2) were designed with Primer Premier 5.0 software based on conserved motif sequences of eight CESAs mRNAs from other plants [Populus tremuloides CesA2 (AY095297), Populus trichocarpa CesA6-E (Pti806784), Eucalyptus grandis CesA3 (DQ014507), Arabidopsis thaliana CesA3 (At5g05170), Bambusa oldhamii CesA5 (DQ020213), Pinus taeda CesA3 (AY789652), Pinus radiata CesA1 (AY639654), Zea mays CesA12 (AY372246)]. PCR products were separated by agarose gel electrophoresis, and amplified products (approximately 2000-bp) were isolated and ligated into the pGEM-T easy vector (Promega, USA). 5′- and 3′-RACE were carried out with the SMART RACE cDNA amplification kit (Clontech, USA) employing gene-specific primers inferred from the PCR fragments (Table 2). The full-length sequences of BplCesAs were submitted to GenBank with the accession numbers EU591529, EU591530, EU591531 and EU591532.

Open reading frames were identified by the ORF Finder tool [41,42]. Putative protein sequences were aligned using Clustal W2 with default parameters (BioEdit version 7.0.0.0; Tom Hall: Raleigh, NC, USA, 2005) [43] and MEGA 4.0 [44]. The alignments were edited with the BioEdit program. Transmembrane domains were identified and analyzed by TMpred [45]. The Populus genome assembly version 1.1 [29,46] was searched for CESA homologs. The phylogenetic tree was constructed based on the CLUSTAL W algorithm of MegAlign (version 5.01; DNAStar: Madison, WI, USA, 2001). For statistical analysis, 100 bootstrap replications were analyzed. The two hypervariable regions (HVRI and HVRII) of the selected CesA gene products were highly divergent and excluded for the construction of the multiple sequence alignment and phylogenetic tree. The largest part of the sequences containing the conserved cellulose synthase domain was selected to generate the tree, including CRP, eight predicted transmembrane regions, a conserved zinc finger motif and the conserved D, D, D, QxxRW signature sequence involved in substrate binding and catalysis during cellulose synthesis. Domain conservation was assessed by the conserved domain database (CDD Version 2.18; National Center for Biotechnology Information: Bethesda, MD, USA) in the NCBI website [47,48].

Real-time RT-PCR was carried out on RNA derived from young and old leaves and stems (eight developmental stages) of three 20-year-old mature B. platyphylla grown under natural conditions in the Northeast of China in 2007. Each sample was comprised of a pool of identical quantities of RNA from three birch plants. The cDNA corresponding to the HVRs (hypervariable regions) can be used as gene-specific primers or probes for molecular analyses [18,49,50]. Therefore, specific primer pairs (18–23 bp) (Table 2) were designed based on the two HVRs of the four BplCesA mRNAs and the specificity of the primers was tested by RT-PCR (Figure S2) and sequencing (Data not shown). The B. platyphylla Actin gene (GenBank accession number: EU588981) [51] was used as an internal control (reference gene) to normalize the amount of total RNA present in each reaction, which has been improved by Chao Dai et al. in 2011 [30]. The validity of the Bplactin gene as a control gene has been tested and proved by pre-experiment before the real-time RT-PCR (Figure S2). Experiments were carried out on a MJ OpticonTM2 machine (Bio-Rad, Hercules, CA) using the QuantiTect SYBR-green PCR Master Mix (TOYOBO, Osaka, Japan). All reactions were carried out in triplicate for technical and biological repetitions. PCR was performed in a 20 μL mixture consisting of 10 μL SYBR Premix Ex Taq, 2 μL cDNA, 1 μL of each gene specific (Table 1), and 6 μL dd H₂O. The PCR amplification conditions were 95 °C for 5 min, followed by 40 cycles at 95 °C for 30 s, 60 °C for 30 s, 72 °C for 30 s, 79 °C for 1 s. Data from qRT-PCR experiments were analyzed by relative quantification according to [52]. The mRNA levels were normalized by the reference gene Bplactin mRNA. A melting curve was generated for each sample at the end of each run to assess the purity of the amplified products.