Knockdown of a cellulose synthase gene BoiCesA affects the leaf anatomy, cellulose content and salt tolerance in broccoli

Cellulose is the major component of cell wall materials. A 300 bp specific fragment from the cDNA fragment was chosen to insert into vector pFGC1008 at forward and reverse orientations to construct the recombinant RNAi vector. Knockdown of BoiCesA caused “dwarf” phenotype with smaller leaves and a loss of the content of cellulose. Moreover, RT-PCR analysis confirmed that the expression of the RNAi apparatus could repress expression of the CesA gene. Meanwhile, examination of the leaves from the T3 of RNAi transformants indicated reduction of cell expansion in vascular bundles, particularly on their abaxial surface. The proline and soluble sugar content increased contrarily. Under the salt stress, the T3 of RNAi plants showed significant higher resistance. The expression levels of some salt tolerance related genes (BoiProH, BoiPIP2;2, BoiPIP2;3) were significantly changed in T3 of RNAi plants. The results showed that the hairpin structure of CesA specific fragment inhibited the endogenous gene expression and it was proved that the cDNA fragment was relevant to the cellulose biosynthesis. Moreover, modulation cellulose synthesis probably was an important influencing factor in polysaccharide metabolism and adaptations of plants to stresses. This will provide technological possibilities for the further study of modulation of the cellulose content of crops.

By antisense expression of different potato CesA clones, the cellulose content of tuber cell walls dropped to 40% of the control plant and the recombined constructs are efficient to control the cellulose synthesis 23 . Currently, we focused on using RNAi to changes cellulose levels and anatomic characteristics of broccoli. The correlation of anatomic changes and plant physiological character was also discussed. The aim of present study is mainly regulate the content of cellulose, so as to improve the quality of vegetables.

Materials and Method
Amplification of the broccoli CesA fragment. Total RNA was extracted from young leaves of broccoli by using the procedure of phenol-guanidine isothiocyanate (Trizol, invitrogen). The RNA was then used as template to synthesize the cDNA.

Construction of RNAi vector.
A 300-bp class-specific region was amplified to construct the recombined RNAi vector pFGCCesA which used the primers RiF containing BamHI and SpeI site s(F1-AAAGGATCCAAAGATGGAACTCAGT; R1-GAAACTAGTGCACAGAT TGTGTATCAG) a n d R i R c o nt a i n i n g A s c I a n d Sw a I s i t e s ( F 2 -A AG G C G C G C C AT T G T G TAT C AG G C ; R2-GGGATTTAAATGAGAGGAAAGATGG). The BoiCesA sense and antisense fragments were inserted into pFGC1008 to construct the recombined vector. The selection of specific cDNA fragment referred to the method that usedvirus-induced gene silencing which published in the Plant Cell 25 . The vector was transferred into Agrobacterium tumefaciensstain EHA105 by the freeze-thaw method 26 .
Genetic transformation of broccoli. The broccoli variety 05-33-105 was used for transformation. It was implemented with Agrobacterium tumefaciens stain EHA105 harboring pFGCCesA constructs and using plasmid pFGC1008 as control. The recombinant plasmid includes the HPTII coding region which was used as a selectable marker (conferring hygromycin resistance).
A hygromycin sensitivity test was performed using cotyledon and hypocotyl explants from seven-day-old seedling 27 . Hypocotyl and cotyledon explants were pre-incubated on the shoot induction medium (MS medium containing 2 mg/L ZT and 0.01 mg/L IAA) for two days in darkness. The incubated explants were immersed into the Agrobacterium tumefaciens solution for 4-8 min (to the hypocotyls and cotyledons) with gentle shaking. The explants were then transferred on the co-cultivation media (MS medium containing 2 mg/L ZT, 0.01 mg/L IAA and 100 μ M AS). After co-cultivation for two days in darkness, the explants were transferred to the same basal medium which was supplemented with 350 mg/L carbenicillinand cultured for seven days. Then the explants were transferred on the selection medium which was supplemented with hygromycin at 4 mg/L and carbencillin at 200 mg/L for another 4-6 weeks to induce shoots. When the shoots emerged, they were subjected to transfer to another medium (MS medium containing 0.5 mg/L NAA and 5 mg/L hygromycin) for root induction. Finally the regenerated plants were transferred to soil. After being vernalized, the seeds of progeny were obtained. The transgenic lines (T3 lines) were used for further experiments.
Southern blotting. The genomic DNA was extracted from control and transgenic plants, using BamHI to digest the DNA. The DNA was transferred and cross-linked onto a nylon membrane. The selectable hygromycin phosphotransferase gene (HPTII) was labeled by PCR for hybridization (Dig Easy Hyb). Then the membrane was washed with different concentration of SSC. At last the membrane was exposed to X-ray 28 . The primers of HPTII gene is 5′ -CGTGTTGAAGGAGAT-GGAGA-3′ and 5′ -AGATTGTGTATCAGGCGTGC-3′ .

Microscopy analysis.
For light microscopy, developed leaves were used to prepare sections by microtome and it were stained with safranin-fast green, then observed and photographed under a light microscope 29 . For scanning electron microscopy (SEM), the methods are detailed by Yu et al. 30 . Small leaf tissues were fixed with 2.5% buffered glutaraldehyde. Then, it was transferred to 1% osmium tetroxide fixative and dehydrated in an ethyl alcohol series from 30 to 100%. The important step was critical point dried and gold coated transmission electron microscopy (TEM) sampling and preparation were carried out as described in the standard procedure 31 .
Scientific RepoRts | 7:41397 | DOI: 10.1038/srep41397 Measurement of carbohydrate. Cell walls were prepared based on previous methods 32,33 . Briefly, using the phenol-methanol to eliminate lipid and protein from the sample and extracting with ethanol and drying, the dried cell wall materials were used to analyze the cellulose content 34 .
The measurement of pectin content was operated as described in papers 35,36 . Shortlythe sample powder with hot absolute ethanol was heated and then centrifuged at 10,000 rpm for 10 min. Alcohol insoluble solids (AIS) were obtained and the concentrated sulfuric acid was used to dissolve AIS. The mixture was transferred into a 25 ml volumetric flask. Then sample solution was added to sodium tetraborate. Color development following addition of m-hydroxydiphenyl, the galacturonic acid was gained that was equal to total pectin content.
Reverse Transcription PCR method. Leaves were picked from T3 of RNAi plants and ground to the fine powder in liquid nitrogen, total RNA was extracted according to the method described by the scription of Trizol. Ten μ g of RNA was used for cDNA synthesis with oligo (DT) 18 as the primer and 1 μ L of cDNA was applied in the PCR reaction. The cycle numbers and transcript levels were optimized.
Proline and soluble sugar content determination. Two independent transgenic lines were selected to measure the proline and soluble sugar content. The measurement of proline content in leaves was prepared according to the method reported by Troll and Lindsley 37 . The content of soluble sugar was then measured 38 .
Evaluation of NaCl to tolerance for T3 of RNAi plants. The

Statistical analysis.
Statistical procedures were carried out with the software package SPSS11.0, Differences among treatments were analyzed taking P < 0.05 as significance according to Duncan's multiple range test. The relative estimate of the amount of cDNA in broccoli leaves was obtained by Image J software.

Results
Molecular cloning and comparative sequences analysis of BoiCesA cDNA. Five cDNA fragments from the CesA gene of broccoli were amplified by standard RT-PCR. Their positions based on the cell wall cellulose biosynthesis gene were shown ( Fig. 1a) 4 , as described by Delmer 5 . The nucleotide sequences of cDNAs CesA-a, CesA-b, CesA-c, CesA-d and CesA-e were identical where they overlap each other. The cDNAs described the sequences of the same CesA gene. Based on the results of sequencing and assembly, a 3252 bp of the CesA cDNA fragment was identified from broccoli, its corresponding deduced amino acid contained D, D, D (aspartic acid residues) and QXXRW motif which was located at the catalytic site. The sequence of the cDNA was compared with the corresponding sequences of the Populus tremuloides PtrCesA4 gene, Acacia mangium AmCesA1 gene and the Arabidopsis AtCesA1 (rsw1) gene (Fig. 1b). Sequence analysis shows that the cDNA fragment designated BoiCesAis a member of CesA superfamily. It shared 90% identity with AtCesA1 at the nucleotide level and 94% identity at the protein level.
Organ-specific expression of BoiCesA gene. To evaluate the transcript accumulation of BoiCesA gene in different organ, the results of quantitative real-time PCR revealed that BoiCesA was expressed in various organs of broccoli, including roots, stems and leaves (Fig. 2). Our results showed that the expression level of BoiCesA was the highest in leaf organs. There were significant differences (P < 0.05) in the expression level of BoiCesA between different organs.

Regulation of CesA gene expression in broccoli.
A recombined RNAi construct pFGCCesA was applied to regulate cellulose biosynthesis in broccoli (Fig. 3a). The 300 bp-length sense BoiCesA sequence and antisense BoiCesA sequence was amplified and inserted into pFGC1008 vector. The T-DNA region of pFGCCesA harbored the selectable hygromycin phosphotransferase gene (HPTII) for hygromycin resistance. Expression of the hairpin structure was driven by the constitutive CaMV 35S promoter.
The Southern blotting analysis for transgenic plants. The pFGC1008-CesA plasmid was transformed into broccoli and 65 plantlets from hypocotyls and 40 plantlets from cotyledons resistant to hygromycin were obtained. To further confirm that the phenotype of transgenic plants is due to the introduction of RNAi construct, southern blot was done in the control and transgenic plants (Fig. 3b). The wild type plants was used as control that is no band was detected. However, transgenic plants had different hybridization bands which were different size. It suggested that the RNAi construct contained BoiCesA gene was random integrated into Brassica oleracea. We selected two independent RNAi lines RNAi-2 and RNAi-8 to do further research.
Transcriptional activity of the CesA gene in broccoli. In order to evaluate the effect of RNAi on CesA gene expression, RT-PCR experiments were performed to study these changed cellulose contents whether related to the expression of BoiCesA gene. The constitutive Actin gene 43 applied as the control in this study. It showed that the RT-PCR results with the BoiCesA and Actin primers in the transgenic plants and control plants (Fig. 4). The amplified product revealed that a reduction of the BoiCesA expression in the RNAi plants in relation to the control plants.

Phenotypes of the RNAi transformed plants.
In order to observe the growth of control and knockdown plants, the germination performance of seeds was observed. There is no significant distinction between control and transgenic seeds (Fig. 5a). However transformed plants showed a typical dwarf phenotype (Fig. 5b) and had obvious change in plant height (Fig. 5c). Furthermore, it was evident to find that some surface lumps presented on the abaxial surfaces of the leaves, and the texture was crisp in the RNAi plants (Fig. 5d). The pFGC-CesA plants were shorter in stature than the control plants (transformed with pFGC1008) (Fig. 5e). Compared with the control plants, the RNAi plants had similar internode length but with less nodes (Fig. 5f). The leaves of the transgenic plants were smaller than those of the control plants, meanwhile the fresh weight decreased relative to that of control plants (Fig. 5g). These phenotypic characteristics had a good agreement with the corresponding observation in tobacco which had been silenced by a plant cellulose synthase gene 25 .  Anatomic and ultrastructural changes. The difference of tissue structure between the T3 of RNAi plants and the control broccoli was surveyed by the light microscopy. The vascular bundles was reduced on the transverse sections located in the elongation zone of leaf veins. Compared with the control plant, the development of lateral veins was not observed in the T3 plant (Fig. 6). Meanwhile, all cells of the vascular bundles of the RNAi plants reduced expansion or elongation but the control plant had the normal leaf veins, compared to control plant, the content of vascular bundles of RNAi plants was approximately 56% (Fig. 6).
Scanning electron microscopy of the leaves from the control plants showed that the abaxial surface were generally smooth, and the epidermal cells were arrayed orderly (Fig. 7a). The stomata of the control plantlets displayed the normal morphology with kidney-shaped guard cells (Fig. 7c). On the contrary, there were many clumps of the epidermal cells along the abaxial surface, especially adjoin to leaf vein in RNAi plants (Fig. 7b). The RNAi and the control leaves also differed in the stomata morphological specificity, the T3 of RNAi leaves presented the abnormal stomata, with guard cells drastically deformed due to the swollen epidermal cells (Fig. 7d). The deformation of guard cells could possibly affect the stomatal function.
Moreover, some significant differences were also observed between the ultrastructure of chloroplasts in the T3 of RNAi and the control leaves. The results of transmission electron microscopy showed that the control cells chloroplasts of mesophyll cells contained the entire double membranes, the regular and inseparable layer of chloroplast grana and stroma, which overflow with starch grains (Fig. 7e). Whereas in the RNAi plants the layer of the slender and spindle-shaped chloroplast grana and stroma were irregular and even disaggregated, but most of all, there was a great difference between the numbers and types of starch grain from the RNAi and the control leaves, furthermore numerous osmiophilic globules appeared in theRNAi plants (Fig. 7f).

The T3 of RNAi plants have altered cellulose and pectin content. It showed the cellulose and pectin
content of the two transgenic lines (RNAi-2 and RNAi-8) and the control plants (Fig. 8). The result showed adout 40% decline in the cellulose content and about 19% reduction in the pectin content of the RNAi plants with that in the control plants. There were significant differences (P < 0.05) in cell wall materials between CesA T3 plants and control plants. It implied that the hairpin structure could affect cellulose biosynthesis.

Proline and soluble sugar contents in the RNAi plants. Accumulation of proline and soluble sugar
is often related to plant adaptation to environmental stresses. Then in order to investigate the correlation of cellulose synthesis and plant physiological characters, proline and soluble sugar contents in T3 and control plants were measured under normal conditions. The two transgenic lines (RNAi-2, RNAi-8) respectively accumulated approximately 3 times higher proline contents than the control plants (Fig. 9a). At the same time, we found that the soluble sugar content of RNAi lines was higher (P < 0.01) than that in control plants (Fig. 9b).

RNAi plants has higher salt resistance capability.
In normal conditions, compared with control plants, the T3 of RNAi plants were dwarf phenotype with smaller dark green leaves ( Fig. 10a and b). Under 250 mM NaCl treatment, the leaves of T3 were still green with thick waxy on the surface (Fig. 10d) whereas the control plants became bleached (Fig. 10c). Moreover under the NaCl treatment, the dry weight of control plants significant reduced relative to that of RNAi plants (Fig. 10e).
Higher plants have developed an antioxidant defense system that includes the antioxidant enzymes SOD, POD, CAT, and APX to deal with adversity stress 44 . The enzymatic activity analysis of antioxidant system was conducted in transgenic and control broccoli under the NaCl treatment (Fig. 11). After 3 weeks of NaCl treatment, the SOD activity of RNAi plants was about 3-fold higher than that of control plants (Fig. 11a). The activities of POD, CAT and APX of RNAi lines showed similar trends (Fig. 11b, c and d). BoiCesA affects the expression of genes related to salt tolerance. To further investigate the NaCl resistant mechanism of T3 of RNAi plants, we analyzed the expression of genes related to salt tolerance. Due to the lack of broccoli genome information, this brings some difficulties in our study. Based on the results of previous   studies [45][46][47] , we found that BoiProDH, BoiPIP2;2 and BoiPIP;-3 genes are associated with the salt tolerance of plants (Fig. 12). The expression level of BoiProDH was significantly reduced in T3 of RNAi plants, it was about 0.5 time that of WT, while the expressions level of BoiPIP2;2 and BoiPIP2;3 up-regulated in T3 of RNAi plants, it were 6-7 times than those of WT, these results might explain the altered salt tolerance of T3 of RNAi plants.

Discussion
The function of the cDNA corresponding to putative cellulose synthase gene from Brassica oleracea L. was analyzed by RNAi. In our study, in attempt to verify the function of the given BoiCesA gene, we constructed the special RNAi vector using the specific region of CesA gene which could be the basis of the multiple alignments. Based  on the alignments for CesA gene from rice previously described, the phylogenetic relationship resulting from the analysis whether based on the alignments from complete amino acid sequences or the hypervariable region (HVR) sequences was the same, and the sub-class identify was primarily defined by the HVR. The sequence in this region does not vary among members of the same sub-class, these experiments already considered that the region was termed 'class-specific region (CSR)' 48 . Hence, we chose the cDNA fragment located in the HVR region to inhibit the specific BoiCesA gene exclusively (see domain structure for plant CesA in Fig. 1a). It could void the lethal phenotype if several endogenous CesA genes were silenced by interference of the homologous region of all CesA genes; this phenomenon had been indicated 22 .
On the other hand, examination of the leaves from the T3 of RNAi plants by light and electron microscopy indicated extensive reduction of cell expansion in vascular bundles, particularly on their abaxial surface, the wider stomatal aperture and the changes of chloroplast ultrastructure due to the down-regulation of BoiCesA gene.
Previous theory has confirmed that microtubules and actin filaments form highly organized arrays in stomatal cells that play key roles in the morphogenesis of stomatal complexes 49 . Moreover, the cellulose fibrils and microtubules as well as actin filaments are radially distributed in guard cells 50 . The depositing cellulose microfibril affects the pattern of local wall thickenings and the mechanical properties of the walls of stomatal cells, thus regulates accurately their shape 49 . In our study, the modulation of cellulose content caused by the RNAi-induced silencing of the specific BoiCesA gene might be the reason of the wider stomatal aperture compared with the control, or, the reduction of the depositing cellulose micro-fibrils weakened support to the guard cells' shape. In addition, we also observed the degradation of starch and the appearance of numerous osmiophilic globules in chloroplasts. The ultrastructural changes of leaves have been reported in wheat and corn 51 . The chloroplasts in sugarcane cultivar YT57/423 were collapsed, presenting many osmiophilic granules 52 . In addition under drought stress conditions, the chloroplasts' lengths decreased and their widths increased, rendering them round in shape. Drought stress also significantly changed the internal structure of the chloroplasts. Membrane systems were damaged, starch grains disappeared, the chloroplasts became deformed and vacuolized 53 . Therefore, under stress condition, in chloroplasts the rapid degradation of starch and soluble sugars accumulation were occurred. Previous studies showed that drought stress can improve the content of some sugars, which may be suppressed some enzyme activities related to cellulose synthesis 54 , the leaf wilting2 mutants which are new alleles of the AtCesA8/IRX1 gene revealed that cellulose synthesis is important for stress responses containing drought induction of gene expression 55 . Our results suggest that regulation cell wall cellulose synthesis are significant influencing factors in polysaccharide metabolism and adaptations of plants to salt stress.
We know that the proline and soluble sugar are important osmotic protective substance that involved in osmotic adjustment 56 . It reported that proline acted as a compatible solute in the cytoplasm. The accumulation of proline can stabilize the macromolecules 57,58 . A central role of soluble sugars depends not only on their direct involvement in the synthesis of other compounds and energy provision, but also on stabilization of membranes 59 . The increment of soluble sugar and proline content associated with reductions in cellulose confirmed that the cell wall could not only perceive, but also adjust for physical changes in its structure.
Under the biotic and abiotic stress, the reduction of cellulose content was observed 60 . In the experiment the cellulose synthesis is important for NaCl stresses response. Compared with control plants, the T3 of RNAi plants were more tolerant to NaCl treatment. Cellulose is the main load-bearing component of cell wall. The change of cellulose content often generates obvious consequences either compensatory or integrity responses. For example, plants may be made more tolerant to stresses as drought, salt and osmotic stress by mutations at AtCESA8 gene which encodes a subunit of a cellulose synthesis complex 55 . The eli1 mutants of CESA3 in Arabidopsis thaliana cause reduced cellulose synthesis, activating defense response through jasmonate and ethylene signaling pathways 61 . Those researches indicate that this is probably a general consequence of CESA loss-of-function.
In this study, we found the obvious starch degradation and soluble sugar accumulation in RNAi plants. It is possible that starchcould be converted into soluble sugar, which might resulted in the increase of soluble sugar content, and enhanced the salt tolerance ability of plants. On the other hand, knockdown of a cellulose synthase gene BoiCesA inhibited the synthesis of cellulose, might lead to the substrate of cellulose synthesis into other metabolic pathways, resulted increased soluble sugar content, which is involved in the intracellular osmotic potential. These inferences are still pending for further experimental verification.
Proline accumulation is a common response to osmotic stress in many plants. Proline dehydrogenase (ProDH) is a key enzyme that catalyzes the degradation of proline in the mitochondria 45 . Our previous results proved that it was effective to increase the accumulation of free proline by silencing BoiProDH gene under salt stress 46 . We found that the BoiProDH expression in RNAi plants was significantly decreased, which might reveal the proline accumulation in RNAi plants.
Aquaporins (AQPs), which play a regulatory role in cellular water transport also called membrane protein family MIP (major intrinsic protein). The function of aquaporin proteinabundant, such as water transport, osmotic adjustment, abiotic stress response. Our previous study confirmed that SlPIPs through improving plant water content and maintaining osmotic balanceto improve the ability of tomato drought resistance 62 . BoiPIP2s plays important roles in the response of broccoli to salinity 47 . The expressions of BoiPIP2;2 and BoiPIP2;3 in T3 of RNAi plants were significantly higher than in WT plants, this suggested that the excessive expression of genes related to salt tolerance enhanced the salt tolerance of RNAi plants.
Cellulose content has an influence on the vegetable quality and resistance. BoiCesA may play a crucial role in the control of cellulose biosynthesis, and it is the first report of the CesA gene cloned from broccoli, according to the genetic relationship of CesA genes in different plants. Based on the different plant CesA genes dendrogram, we predict that the BoiCesA clone is a member of CesA family and the BoiCesA clone may share functional similarity with the Arabidopsis genes in the same cluster. Without the function of the CesA1, CesA2, CesA3 and CesA6 genes, the cellulose of primary wall are difficultly biosynthesized from mutants 17,23,63,18 . So we suggest that the BoiCesA clone may play a role in primary cell wall biosynthesis.