Exploring key cellular processes and candidate genes regulating the primary thickening growth of Moso underground shoots

Transcription

1 Research Exploring key cellular processes and candidate genes regulating the primary thickening growth of Moso underground shoots Qiang Wei 1,2 *, Chen Jiao 3 *, Lin Guo 1, Yulong Ding 1,2, Junjie Cao 1, Jianyuan Feng 1, Xiaobo Dong 1, Linyong Mao 3, Honghe Sun 3, Fen Yu 4, Guangyao Yang 4, Peijian Shi 1,2, Guodong Ren 5 and Zhangjun Fei 3 1 Co-Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, Nanjing, Jiangsu 2137, China; 2 Bamboo Research Institute, Nanjing Forestry University, Nanjing, Jiangsu 2137, China; 3 Boyce Thompson Institute, Cornell University, Ithaca, NY 14853, USA; 4 Jiangxi Provincial Key Laboratory for Bamboo Germplasm Resources and Utilization, Jiangxi Agriculture University, Nanchang, Jiangxi 3345, China; 5 State Key Laboratory of Genetic Engineering, Institute of Plant Biology, School of Life Sciences, Fudan University, Shanghai 2433, China Authors for correspondence: Zhangjun Fei Tel: zf25@cornell.edu Qiang Wei Tel: weiqiang@njfu.edu.cn Received: 24 August 216 Accepted: 13 September 216 doi: /nph Key words: Moso bamboo, pith development, shoot apical meristem (SAM) differentiation, spiral growth, transcriptome profiling, underground shoot growth. Summary The primary thickening growth of Moso (Phyllostachys edulis) underground shoots largely determines the culm circumference. However, its developmental mechanisms remain largely unknown. Using an integrated anatomy, mathematics and genomics approach, we systematically studied cellular and molecular mechanisms underlying the growth of Moso underground shoots. We discovered that the growth displayed a spiral pattern and pith played an important role in promoting the primary thickening process of Moso underground shoots and driving the evolution of culms with different sizes among different bamboo species. Different with model plants, the shoot apical meristem (SAM) of Moso is composed of six layers of cells. Comparative transcriptome analysis identified a large number of genes related to the vascular tissue formation that were significantly upregulated in a thick wall variant with narrow pith cavity, mildly spiral growth, and flat and enlarged SAM, including those related to plant hormones and those involved in cell wall development. These results provide a systematic perspective on the primary thickening growth of Moso underground shoots, and support a plausible mechanism resulting in the narrow pith cavity, weak spiral growth but increased vascular bundle of the thick wall Moso. Introduction Bamboo is an important nontimber forest plant with an annual trade value of > 2.5 billion US dollars and c. 2.5 billion people depend on it economically (Peng et al., 213a,b). Moso is valued as a source of food, building and industrial material. As the most valuable bamboo species in China, it has been studied mainly regarding its cultivation and downstream applications. Little attention has been paid to its fundamental biological processes such as short and long parenchyma cells interspreading in the ground tissues of its culm in contrast to a single type of parenchyma cells in rice, maize and other major crop species (He et al., 22), long and unpredictable flowering period, and fast elongation of internodes during its shooting period. Understanding these developmental processes is necessary for the better and sustainable management of the bamboo forest. One of the important factors that restrict the deeper understanding of the unique developmental characteristics of Moso at *These authors contributed equally to this manuscript. the molecular level is the lack of efficient resources and tools. However, the rapid advancement of sequencing technologies gives bamboo researchers a suitable platform that is helpful to identify genes involved in regulating its unique developmental processes. Several groups have reported the transcriptome profiles of the fast growth process of Moso using the RNA-Seq approach (He et al., 213; Peng et al., 213b). RNA-Seq also was used to help elucidate the molecular mechanisms underlying the regulation of bamboo flowering (Liu et al., 212; Zhang et al., 212; Gao et al., 214; Shih et al., 214). In addition, using the highthroughput sequencing technology, the draft genome sequence of Moso was generated (Peng et al., 213a), providing a valuable resource for bamboo researches. Over a thousand years, of the many contributions regarding the culm morphology of Moso, from the poem written by Lei Song who lived in the Song Dynasty of China, which described the fast growth of Moso shoots after a spring rain, to today s molecular researches using transcriptome sequencing technology (He et al., 213; Peng et al., 213b), almost all have focused on the fast internode elongation process which determines the length 81

2 82 Research of internodes. Little attention has been paid to the primary thickening growth of underground shoots, which mainly determines the circumferences of internodes (Dong, 27). So far, the cellular and molecular mechanisms of the primary thickening growth of Moso underground shoots have remained largely unknown. In the present study, we systematically investigated the morphological and cellular processes of the primary growth of underground shoots of Moso. We found that the growth displayed a spiral pattern and the pith tissue played an important role in promoting the growth. Furthermore, through transcriptome analysis we identified key candidate genes regulating the pith development using a thick wall variant, Ph. edulis Pachyloen, with abnormal proportion of pith and culm wall tissues. Materials and Methods Plant materials Moso (Phyllostachys edulis (Carr.) H. de Lehaie) plants used for investigating the primary growth of underground shoots were grown in the Bamboo Garden of Nanjing Forestry University. The wild-type (WT) Moso and the thick wall variant used for transcriptome sequencing were collected at Daao village, Yichun City, Jiangxi Province, China. Morphology analysis In order to investigate morphological changes during the primary growth of underground Moso shoots, over 1 buds were collected from three 2 3-yr-old rhizomes at each time point. The number of buds at different developmental stages was recorded at each time point (from 28 August 28 to 28 February 29), and the diameter of the biggest bamboo shoot on each rhizome was measured. Light microscopy For young bamboo shoots or rhizome shoots, c..5-cm 3 samples at different developmental stages were fixed in the formalinacetic-7% alcohol (FAA, v/v) buffer and exhausted with an aspirator pump. For safranin and fast green staining, serial transverse and longitudinal sections (8 lm thick) from paraffin-embedded tissue were stained sequentially with safranin and fast green. For haematoxylin staining, serial transverse and longitudinal sections (7 lm thick) were made from paraffin-embedded tissues which were soaked in the haematoxylin solution for c. 7 d. Subsequently, these sections were observed under an Olympus QG2-32 (Olympus, Tokyo, Japan) light microscope. Investigation of the growth pattern of bamboo shoots Bamboo shoots at the S-6 stage were harvested and the sheaths were carefully removed. The resulting shoots were scanned with an HP Scanjet 485 (Hewlett Packard, Palo Alto, CA, USA). The images of node rings were extracted using the magnetic lasso tool of PHOTOSHOP (Adobe, San Jose, CA, USA). The scanned images of culm cross-sections were collected for the following 35 bamboo species: Bambusa comigera, B. emeiensis viridiflavus, B. multiplex Silverstripe, B. oldhamii, B. pervariabilis, B. rigida, B. stenostachya, B. suavis, B. textilis, B. tulda, B. tuldoides, Bashania fargesii, Chimonobambusa quadrangularis, Dendrocalamus barbatus, D. barbatus internodiradicatus, D. giganteus, D. latiflorus, D. sinicus, Dendrocalamopsis beecheyana pubescens, Indosasa crassiflora, Melocanna baccifera, Phyllostachys aureosuleata Spectabilis, Ph. bambusoides, Ph. bambusoides shouzhu, Ph. bissetii, Ph. dulcis, Ph. edulis, Ph. heteroclada, Ph. makinoi, Ph. nigra, Pleioblastus juxianensis, Pseudosasa amabilis, Qiongzhuea tumidinoda, Sasaella kongosanensis Aureostriaus and Sinobambusa tootsik laeta. The MATLAB and R functions were used to describe the node rings, the outside contours and pith cavity rings of bamboo culm cross-sections as described in Shi et al. (215). Calculation of cross-sectional areas of pith, epidermis, cortex, vascular tissue, ground tissue and culm wall One stage S-5 and one stage S-6 bamboo shoot (Fig. 1a) were selected randomly. After removing the shoot sheath leaves, a series of paraffin cross-sections were made for parts from top to bottom of the two bamboo shoots, which were used to represent the primary thickening process before the S-5 stage and the entire primary thickening growth process of underground bamboo shoots, respectively. The paraffin sections were observed under an Olympus QG2-32 light microscope. The areas of pith were calculated using the formula: (average of the biggest and smallest diameters of pith circle) 2 /4. Because the shapes of cross-sections of cortex and epidermis of bamboo shoot are each close to a circle ring, their total areas were obtained using the standard formula for calculating circle ring area. Vascular tissue was calculated using the Qwin software coming with the Olympus QG2-32 light microscope. Ground tissue area was obtained by subtracting the pith area, vascular tissue area and epidermis and cortex area from the total crosssectional area. For the calculation of pith and wall areas of different parts of the Moso culm, a 3-m-high Moso shoot with 7-cm ground diameter, and six Moso culms with different ground diameters in the Bamboo Museum of Nanjing Forestry University were collected. A series of cross-sections were made, and the total cross-sectional area and the pith area were then calculated using the formula: (average of the biggest and smallest diameters) 2 /4. The culm wall area was obtained by subtracting the pith area from the total crosssectional area. We further calculated the pith and wall areas of culms from 5 different bamboo species. Of these bamboo species, 3 (B. beecheyana pubescens, B. blumeana, B. chungii, B. comigera, B. pervariabilis, B. remotiflora, B. rigida, B. stenostachya, B. suavis, B. textilis, B. tulda, B. tuldoides, B. vulgaris striata, Ch. quadrangularis, D. barbatus internodiradicatus, D. fumingensis, D. giganteus, D. latiflorus, D. sinicus, Fargesia dracocephala, I. crassiflora, Ph. bambusoides shouzhu, Ph. edulis, Ph. glauca,

3 Research 83 Developmental process S-6 Ground S-5 S-4 17 cm S-3 S-2 2 cm Daily increment (cm d 1) S-1 (b) The biggest diameter (cm) (a) 15 cm Date (d) (c) (e) (h) (g) (f) (k) (j) (i) 1 μm 2 μm 1 μm 2 μm 25 μm (l) 2 μm (m) 1 μm 2 μm (n) (o) 25 μm 2 μm 1 mm 2 μm 2 μm (r) (q) (p) 1 μm 2 μm 2 μm Fig. 1 Morphological and cellular analysis of the primary growth of Moso bamboo underground shoots. (a) Morphology of underground bamboo shoots at different developmental stages. (b) Diameter (black line) and daily increment (green bars) of underground bamboo shoots at different time points of growth. Data are means SD. (c) Longitudinal view of the S-1 stage bamboo shoot. (d) Vascular bundle in the bottom part of the S-1 bamboo shoot. (e) Longitudinal view of the S-2 stage bamboo shoot. (f) Horizontal rib meristems (red arrows) in the bottom part of the S-2 bamboo shoot. (g) Axillary bud primordium-like cell (red arrow) first emerged in the S-2 bamboo shoot. (h) Longitudinal view of the S-3 bamboo shoot. (i) Longitudinal view of shoot tips of the S-4 bamboo shoot. (j) Obvious node (stars) and internode (arrow) differentiation found in the S-4 bamboo shoot. (k) Axillary bud (stars) and phloem ganglion (arrow) first appeared in the S-4 bamboo shoot. (l) Longitudinal view of shoot tips of the S-5 bamboo shoot. (m) Numerous horizontal rib meristems (indicated with red arrows) emerged in the centre of the pith zone. (n) Root primordium started to appear in the bottom part of the S-5 bamboo shoot. (o) Longitudinal view of developing axillary buds (red arrow) of S-6 bamboo shoot. (p) Fully differentiated node (star) and internode (arrow) found in the S-6 bamboo shoot. (q) Broken pith zone (red arrow) in the bottom internode of the S-6 bamboo shoot. (r) Long (stars) and short (red arrows) parenchyma cells in the bottom internode of the S-6 bamboo shoot. Ph. heteroclada, Ph. makinoi, Ph. nigra, Ph. sulphurea f. houzeauana, Ps. amabilis and S. tootsik) were collected from the Bamboo Museum of Nanjing Forestry University, and the other 2 (Arundinaria fortunei, B. emeiensis Viridiflavus, B. multiplex, Hibanobambus tranguillans shiroshima, Ph. angusta, Ph. aureosulcata, Ph. bissetii, Ph. carnea, Ph. fimbriligula, Ph. iridescins, Ph. meyeri, Ph. praecox natata, Ph. prominens, Ph. varioauriculata, Pl. gozadakensis, Pl. maculatus, Ps. japonica, Sa. kongosanensis Aureostriaus, Semiarundinaria densiflora and Se. sinica) were grown in the Bamboo Garden of Nanjing Forestry University. Investigation of SAM shape using the superellipse equation In order to investigate the shape of shoot apical meristem (SAM), images of SAM from bamboo shoots at different stages were collected using a Leica DM25 light microscope (Leica, Wetzlar, Germany). Semi-ellipses were extracted from the images using the magnetic lasso tool of the PHOTOSHOP CS6 software (Adobe, San Jose, CA, USA), and two of them were combined to form a new ellipse image. The outside contours of ellipses were then extracted. The MATLAB and R functions were used to describe the outside contour of bamboo SAM as described in Shi et al. (215).

4 84 Research RNA extraction and transcriptome sequencing Five bamboo shoots at the S-2 stage of the thick wall variant and five corresponding WT shoots were harvested, representing five biological replicates. After careful excision of bamboo sheaths, the remaining parts were immediately frozen in liquid nitrogen and stored at 7 C. Total RNA was extracted using the HF19-1 Kit (Yuanpinghao Biotechnology Company, Beijing, China) that can efficiently remove genomic DNA and polysaccharide. The ratio of OD 26 and OD 28 of the extracted RNA was determined with a NanoDrop 1 spectrophotometer (Thermo Scientific, Waltham, MA, USA). Electrophoresis was performed to examine the RNA quality. RNA samples with OD 26 /OD 28 values between 1.9 and 2., and with sharp and clear electrophoretic bands, were further checked for the RNA Integrity Number (RIN) value. Samples with RIN > 8. were selected for transcriptome sequencing. RNA-Seq library preparation and sequencing were both performed at Novogene Biotech (Beijing, China) using standard Illumina protocols. The RNA- Seq libraries were sequenced on an Illumina HiSeq 25 system with the paired-end mode. Raw sequence reads have been deposited into the NCBI sequence read archive under accession SRP Transcriptome sequence processing and analysis Raw RNA-Seq reads were first processed to remove adaptor and low-quality sequences using Trimmomatic (Bolger et al., 214). The resulting reads shorter than 4 bp were discarded. RNA-Seq reads were then aligned to the ribosomal RNA (rrna) database (Quast et al., 213) using bowtie (Langmead et al., 29), and those that could be aligned were discarded. The resulting high-quality cleaned reads were aligned to the Moso genome (Peng et al., 213a) using HISAT (Kim et al., 215) allowing up to five mismatches. Following alignments, raw read counts for each Moso gene were derived and then normalized to fragments per kilobase of exon model per million mapped fragments (FPKM). Raw counts of Moso genes were fed to the edger package (Robinson et al., 21) for differential expression analysis. Genes with adjusted P-values <.5 and fold changes no less than two were identified as differentially expressed genes (DEGs). Gene ontology (GO) term enrichment analysis of DEGs was performed using GO::TermFinder (Boyle et al., 24) with GO annotations downloaded from BambooGDB (Zhao et al., 214). Raw P-values of multiple tests were corrected using false discovery rate (FDR; Benjamini and Hochberg, 1995). We also used MapMan (v.3.5.1r2) for visualization of expression patterns of DEGs (Thimm et al., 24). The mapping file of Ph. edulis was generated using Mercator (Lohse et al., 214). Quantitative real time PCR analysis Same total RNA used for transcriptome sequencing was used for quantitative real-time polymerase chain reaction (qrt-pcr) analysis. 1 lg total RNA was transcribed in a total volume of 2 ll solution as described in the operation manual of the PrimeScript TM RT reagent Kit (code no. RR47A; Takara, Japan). The qrt-pcr was performed using the TransStart Tip Green qpcr SuperMix Kit (Transgene Biotech, Beijing, China) on an ABI StepOne Plus Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) according to the manufacturer s instructions. Fold changes of RNA transcripts were calculated using the 2 MMCt method (Livak & Schmittgen, 21). The experiments were repeated technically at least three times. Tonoplast intrinsic protein (TIP41) was used as the internal control (Fan et al., 213). Gene-specific primers used for qrt-pcr are provided in Supporting Information Table S1. Transmission electron microscopy Apical parts of bamboo shoots at the S-2 stage of the thick wall variant and the WT were collected and fixed in 2% paraformaldehyde and 1% glutaraldehyde for 4 h at 2 C. After washing in NaOH-PIPES for five times, samples were fixed in 1% osmic acid for another 1 h. The fixed samples were washed five times in phosphate buffer and then dehydrated by a series of ethanol solution with pure acetone for the final step. Serial sections of.8 lm thickness were made from Spurr-embedded tissues by LKB-V microtome (LKB, Sweden). Sections were mounted on copper grids and stained with uranyl acetate and lead citrate. Observations were made using an H-6 transmission electron microscope (Hitachi, Tokyo, Japan). Quantification of cellulose contents Quantification of cellulose was performed according to the method described in Schindelman et al. (21) and Wang et al. (211) with minor modifications. Three bamboo shoots at the S- 4 stage of the WT and the thick wall variant were ground into fine powder, and the resulting product was incubated in 95% ethanol for 3 min at 65 C. After being cooled to room temperature, the mixture was centrifuged. The collected pellets then were washed twice by 95% ethanol and cultivated with methanolchloroform (2 : 3, v/v) overnight. After centrifugation, the extracted material was washed five times in 95% ethanol and dried for 12 h at 65 C. One milligram of dried material was hydrolysed in 1 ml 2 M trifluoroacetic acid for 9 min at 12 C. After heating at 1 C for 3 min in 1 ml updegraff reagent (acetic acid nitric acid water, 8 : 1 : 2, v/v), the trifluoroacetic acid pellet was cooled to room temperature. The air-dried pellet was then incubated with 175 ll 72% (v/v) H 2 SO 4 at room temperature for 3 min. The mixture was diluted by 825 ll H 2 O, and was centrifuged to wipe off the sediment. The resulting supernatant was assayed by the anthrone colorimetric method to identify glucose content as the following steps. A series of diluted pure glucose solutions were reacted with the freshly prepared anthrone reagent (2 mg ml 1 anthrone H 2 SO 4, w/v) at 8 C for 3 min. After cooling to room temperature, absorption of the treated samples at 625 nm was collected to make a standard curve for calculating the next cellulose content of samples needing to be detected.

5 Research 85 Results The primary growth of Moso underground shoots is a continuous, nonuniform, complex but well-organized developmental process According to our analysis, before becoming a mature bamboo shoot, Moso shoots generally go through six different morphological changes (Fig. 1a). The S-1 stage bud, termed the dormant bud, is close to the groove of rhizome, with a small and flat body. Most of the buds on rhizome belonged to this type (Table S2). In the season we investigated (from 28 August 28 to 28 February 29), certain dormant buds became active in mid-august. The S-2 stage bud, termed the awakening bud, with a plump body could be found easily in late August (Fig. 1a; Table S2). The bud started to grow actively after the S-2 stage. According to the shoot size, outside sheath number and morphology, in general it had four distinctive morphology transformations (S-3 to S-6) before becoming a mature bamboo shoot (Fig. 1a). The whole growth process took c. 6 months (Table S2) and is nonuniform (Fig. 1b). The fastest growth happened between early and late November; over half of the total growth happened in these 2 d, and the daily increment during this period was dramatically higher than in the other periods (Fig. 1b). In addition, the cross-sectional area of the active apical meristem of the Moso underground shoot is c cm 2 ( (144 lm/2) 2 /1 8 ) and eventually it becomes a 14.4-cm-diameter wide internode (Table 1). This represents an increase of the cross-sectional shoot area of c. 1 million fold during its 6-month growth period. Therefore, it is an extremely fast developmental process. As shown in Fig. 1(c,d), although the whole bud at the S-1 stage was in an undifferentiated status, mature vascular bundles could be found in the bottom part of the bud. At the S-2 stage (Fig. 1e), horizontal rib meristem, which develops into the vascular bundle of nodes, started to emerge in the bottom part of the bud (Fig. 1f); axillary bud primordium-like cells could be found in the leaf axils of young bamboo shoot sheath (Fig. 1g). In the bud at the S-3 stage (Fig. 1h), various tissues such as pith and rib meristems were developing actively. In the bud at the S-4 stage (Fig. 1i), pith and rib meristems continued to develop actively, Table 1 Morphological parameters of shoot apical meristem of the thick wall variant (Phyllostachys edulis Pachyloen ) and the wild-type Moso (Phyllostachys edulis) Parameters Wild-type Moso Thick wall variant Width (lm) b a Height (lm) a b Volume (mm 3 ) a a Number of cell layers 6 6 Average cell number 14 18b 175 6a in the vertical section Cell length (lm) a a Cell width (lm) a a Cell area (lm 2 ) a a Different letters in the same line represent P <.5, whereas the same letters indicate P >.5. Data are means SD. and the prototypes of nodes and internodes started to form (Fig. 1j). Another typical developmental characteristic at this stage was the appearance of lateral bud primordium and immature phloem ganglion (Fig. 1k; Ding et al., 2). The pith zone of bud at the S-5 stage was more apparent (Fig. 1l), and horizontal rib meristems started to appear in the part that would eventually develop into nodes (Fig. 1m); typical root primordium also could be found at this stage (Fig. 1n). In the final, S-6 stage, various axillary buds at different developmental stages could be found in the leaf axils of bamboo shoot sheaths (Fig. 1o). Nodes and internodes were fully differentiated (Fig. 1p), and the pith tissue had started to break in the bottom internodes (Fig. 1q). Long and short parenchyma cells could be found in the ground tissues of bottom internodes of the S-6 stage bamboo shoots (Fig. 1r). Characteristics of SAM of Moso underground shoots By checking the cross-sections of a bamboo shoot at the S-5 stage from top to bottom, we found a clear differentiation process of SAM cells. The SAM cells differentiated into all different tissues including sheath leaves, shoot wall tissues, pith tissues (Fig. S1a g), internode interval tissues (Fig. S1h,i) and adventitious roots (Fig. 1n). Investigation of the apical meristem of bamboo shoots at different stages indicated that the apical meristem structures of bamboo shoots at the S-1, S-2, S-3, S-4 and S-5 stages were similar (Fig. 2a). However, the number of apical meristem cells of bamboo shoots at the S-6 stage decreased dramatically (Fig. 2a). It contained no more than three layers of cells (Fig. 2a) and the cell number in each layer also decreased, whereas the apical meristem in the bamboo shoots before the S-6 stage usually contained of six layers of cells (Fig. 2a). Longitudinal section of a shoot tip of bamboo shoots at the S-2 stage revealed that the inner layers of apical meristem cells first differentiated into primary meristem-like cells (Fig. 2b), which further differentiated into pith and peripheral meristem cells (Fig. 2c). Based on these findings, a model of the structure and development of the SAM of Moso was proposed (Fig. 2d,e). The active SAM of Moso had six layers of cells (Fig. 2d,e). The outer two layers of cells differentiate into epidermis, cortex and bamboo shoot sheath, the middle two layers of cells develop into root, axillary bud, vascular and ground tissues of shoot walls, and the inner two layers of cells eventually develop into pith and internode interval tissues (Fig. 2d,e). Based on the investigation of the apical meristem of shoot tips at the S-1 to S-5 stages, we found that the shape of apical meristem zone on the vertical section was semi-elliptical with a c. 144-lm-long axis and c. 81-lm-long minor axis (Fig. 2e). The primary thickening growth of Moso underground shoots displays a spiral pattern Using mathematical modelling, we found that the outlines of node rings (Fig. 3a e) of Moso underground shoots could be fitted by the superellipse equation (Gielis, 23; Shi et al., 215; Fig. 3f,g). The angles between the long axis and the horizontal axis of the fitted ellipses progressively decreased from the outer to

6 86 Research (b) (a) S-1 S-2 S-3 SAM 5 μm 5 μm 5 μm S-4 S-6 S-5 (c) PM 5 μm PiM 5 μm 5 μm 5 μm 5 μm (d) (e) 5 μm Cell number: 92 ± ± 11 μm 144± 24 μm and 2 layers, epidermis, cortex and bamboo shoot sheath 3 and 4 layers, vascular tissue, ground tissue, axillary bud and root 5 and 6 layers, pith tissue and internodal interval Fig. 2 Structure of Moso bamboo shoot apical meristem (SAM). (a) Structure of apical meristems of bamboo shoot at different developmental stages. (b, c) Longitudinal section of a shoot tip of the S-2 bamboo shoot. PM, primary meristem; PeM, peripheral meristem; PiM, pith meristem. (d) Six-cell-layer structure of the S-5 stage bamboo SAM. Red arrows indicate the anticlinal division. (e) Model of the active SAM of bamboo shoots. Data are means SD. the inner rings, and were highly correlated with the area changes (Fig. 3h,i). We also found that the superellipse equation could precisely describe the cross-sectional outlines of bamboo culms, even those of C. quadrangularis whose shape is nearly square (Fig. 3j l). Clear differences of the angles of the rings of culm wall and pith cavity were discovered among the 35 different bamboo species (Fig. 3m,n). Some bamboo species including Moso (under certain growth conditions) and Pl. graminens manstopiralis even displayed a clear spiral pattern, like tree trunks (Fig. 3o,p). plays an important role in promoting the primary growth of Moso underground shoots and driving the evolution of various sizes of culms among bamboo species In Moso shoots, the stem structure in cross-section was organized into (from outer to inner) epidermis, cortex, vascular bundle cells, ground tissues and pith tissues (Fig. 4a). We investigated the area changes of these tissues in an S-6 stage underground shoot from apex to the bottom, which could represent different developmental stages of the entire primary growth process of an underground bamboo shoot. We found that pith tissue has the fastest growth rate (Fig. 4b). A similar result was found in an S-5 stage bamboo shoot from top to bottom (Fig. 4c). A series of cross-sections were made from a culm of 3-m high bamboo shoot from top to bottom, and in addition six culms were collected from different locations in China (Fig. 4d), which represented the variations of the culm circumference among different Moso species. We found that the increase of both culm wall and pith areas were positively correlated with the increase of culm diameter (Fig. 4e). However, the increased rate of pith area was much higher than that of the wall area (Fig. 4e). By investigating the culm wall and pith areas in 5 bamboo species with different diameters, we

7 Research 87 (a) (b) (c) (d) (e) (f) (g) (h) Angle Ring number (from outside to inside) (i) Area(cm 2 ) Ring number (from outside to inside) 8 (j) (k) (l) cavity (m) (o) (p) Wall ring ring (n) Angle difference Species Fig. 3 The primary growth of bamboo shoots displays a spiral pattern. An S-6 stage Moso bamboo shoot with (a) and without bamboo sheaths (b). (c) Shoot in (b) viewed from above. (d) Part of node rings extracted from the bamboo shoot shown in picture (c). (e) Merged picture of (c) and (d). (f) Merge pictures of the observed node rings (black solid lines) and the predicted rings (red solid lines). Numbers at the top indicate the ring numbers from outside to inside. The dark dashed line represents the direction of major axis and the red dashed line represents the direction of x-axis. (g) Twenty five predicted node rings using the superellipse equation. (h) Angles between major axis and x-axis of the 25 node rings. (i) Areas of the 25 node rings. (j) Cross-sections of eight bamboo species. Numbers from 1 to 8 represent the following eight bamboo species, respectively: Phyllostachys bissetii, Chimonobambusa quadrangularis, Phyllostachys makinoi, Phyllostachys edulis, Bambusa polgmorpha, Dendrocalamus barbatus internodiradicatus, Dendrocalamus latiflorus and Dendrocalamus sinicus. (k) Outlines of cross-sections of randomly selected 15 bamboo species with different sizes. (l) Predicted rings of (k). (m) Merged pictures of the observed culm wall rings and pith cavity rings (black solid lines) and the predicted rings (red solid lines) of eight bamboo species shown in (j). The dark dashed line represents the direction of major axis and the red dashed line represents the direction of x-axis. (n) Differences of angles between the major axis and x-axis of culm wall rings and pith cavity rings of 35 bamboo species. (o) Culm of Pleioblastus graminens manstopiralis with obvious spiral growth. (p) Culm from Moso showing an apparent spiral growth pattern under certain conditions.

8 88 Research (a) (b) 3 Cortex and epidermis Vascular tissue Ground tissue Epidermis Ground tissue CortexVascular tissue 1 μm (d) Area (cm 2 ) 2 1 (c) Area (cm 2 ) (e) Shoot diameter (cm) Cortex and epidermis Ground tissue Vascular tissue Shoot diameter (cm) Wall 6.5 cm Area (cm 2 ) (f) 3 Wall Diameter Shoot or culm diameter (cm) 25 A rea (cm 2 ) Culm diameter (cm) Bamboo species Fig. 4 plays an important role in determining the circumference of bamboo culm. (a) Cross area of the Moso bamboo shoot. (b, c) Areas of epidermis and cortex, vascular tissue, ground tissue and pith tissue of different sections (from top to bottom) of an S-6 stage (b) and an S-5 stage (c) Moso bamboo shoot. (d) Cross-sections of a culm (from top to bottom) of 3-m high Moso bamboo shoot. (e) Areas of pith and culm wall of different sections (from top to bottom, x-axis labels from.4 to 7) of the bamboo culm shown in (d), and six Moso bamboo culms with different diameters collected from different places (x-axis labels from 7.5 to 17.1). (f) Areas of pith and culm wall of 5 bamboo species with different culm diameters. found that both parameters increased with the increase of culm diameters (Fig. 4f). Again, the increased rate of pith area was much higher than that of the wall area (Fig. 4f). In addition, our analysis revealed that the pith area was highly correlated with the culm area (Fig. S2). Thick wall Moso is a variant with abnormal differentiation of pith and vascular tissues caused by its aberrant SAM Morphology characterization indicated that the thick wall Moso was a variant with significantly smaller pith cavity and fasciated

9 Research 89 stem (Fig. 5a c). Further anatomical analysis revealed that the apical meristem of the thick wall variant did not differentiate into any obvious pith tissue cells, but instead produced much more (a) vascular tissue during the development of underground shoots (Fig. 5d,e). The SAM of the thick wall variant is flat and enlarged, with more cells but the same number of cell layers as Thick wall variant Wild 6.5 cm (c) Shoot diameter (cm) (d) cm 1 μm 1 Shoot diameter (cm) Wall Area (cm2) Area (cm2) 3 Wall (b) 25 μm 2 μm 1 μm (e) cm Developing vascular tissue 1 μm 25 μm 1 μm Developing vascular tissue 2 μm Fig. 5 Thick wall Moso is a variant with abnormal pith development. (a) Cross-sections of culms of a wild-type (WT) Moso (left) and a thick wall variant (right). (b, c) area and wall area of culms in different sections (from the top to bottom parts) of the WT Moso (b) and thick wall variant (c). (d) development of the S-2 bamboo shoot of WT Moso. d-2, longitudinal section of shoot tip of bamboo shoot in d-1. d-3 to d-5, cross-sections of bamboo shoot in d-1 from top to bottom. Inset indicates developmental status of the vascular tissue around pith tissue in d-5 (red arrow). (e) development of the S-2 bamboo shoot of the thick wall variant. e-2, longitudinal section of shoot tip of bamboo shoot in e-1. e-3 to e-5, cross-sections of bamboo shoot in e-1 from top to bottom. Inset indicates the developmental status of the vascular tissue in the central part in e-5 (red arrow).

10 9 Research that of the WT (Fig. 6a; Table 1). In addition, we discovered that the superellipse equation also could precisely describe the shape of the apical meristem of Moso shoots (Fig. 6b,c). By mathematical modelling, we found that the apical meristem of the variant had a different angle between the major axis and the horizontal axis than in the WT, correlating well with its flat shape (Fig. 6b,c). Transcriptome profiling identified candidate genes influencing the pith development and the spiral growth of the thick wall Moso shoots In order to explore the key candidate genes influencing the pith development of the thick wall variant, we compared transcriptome profiles of bamboo shoots at the S-2 stage of the thick wall variant and its native WT Moso. A total of c. 289 million raw read pairs were generated. After removing adaptor and lowquality sequences and contaminated reads, we obtained a total of 278 million high-quality read pairs (Table S3). The cleaned reads were then mapped to the Moso genome database (Peng et al., 213a). Correlation analysis indicated that gene expression patterns between different biological replicates were highly consistent, indicating the high reproducibility of our RNA-Seq experiment (Table S4). To further validate our RNA-Seq expression profile data, we performed qrt-pcr assays on 1 randomly selected candidate genes, including three in the auxin pathway, one in the ethylene pathway, two transcription factors, and four in cell wall biosynthesis. The results showed that although the exact fold changes of the selected unique transcripts varied between RNA-Seq expression and qrt-pcr analyses, the trend of gene expression change was largely similar (Fig. S3). We identified a total of 15 DEGs between the thick wall variant and the WT, of which 11 were significantly downregulated and 94 upregulated in the variant (Table S5). GO enrichment and MapMan analyses revealed that genes in various metabolism pathways were significantly upregulated in the thick (a) S-2 S-3 S-4 S-5 5 μm 5 μm 5 μm 5 μm (b) S-2 S-3 S-4 S-5 (c) Fig. 6 Shoot apical meristem (SAM) of the thick wall Moso is flat and enlarged. (a) Morphology of apical meristems of the thick wall Moso bamboo shoots at different developmental stages. Red curves indicate the cell layers. (b, c) Outlines of the apical meristem of the wild-type Moso (b) and the thick wall variant (c) (black solid lines) and the predicted outlines using the superellipse equation (red solid lines) at different developmental stages (from S-2 to S-5). Each outline was combined by two corresponding semi-outlines of the apical meristem. The dark dashed line represents the direction of major axis and the red dashed line represents the direction of x-axis.

11 Research 91 wall variant, including those related to cell wall synthesis and hormone metabolism (Fig. 7; Table S6). Interestingly, nearly all DEGs related to cell wall synthesis and the metabolism of plant hormones including auxin, cytokinin, ethylene and gibberellin (GA) were upregulated in the thick wall variant (Tables 2, 3). A total of 73 differentially expressed transcription factors (TFs) were identified, of which 67 were significantly upregulated in the variant including those from the bhlh, bzip and MYB families (Table S7). Because a large number of cell wall genes were upregulated in the thick wall variant, we checked the cell wall thickness and the cellulose content of the variant and the WT, and found that, as expected, values for both parameters were greater in the thick wall variant (Fig. 8a,b). In addition, because cell wall is an important mediator between mechanical force and development (Sassi & Traas, 215; Braybrook & Jonsson, 216), we also checked the spiral growth of the thick wall variant to see whether the growth was influenced by its thicker cell wall. As expected, we found that the spiral growth of the thick wall variant was also different from that of the WT, which had smaller angle changes (Figs 3h, 8c e). Discussion As a quick growing, renewable and multifunctional plant, Moso has attracted much interest worldwide. However, the lack of knowledge about its fundamental biological processes has limited a better usage of this economically important plant. As one of the most distinguished developmental processes of Moso, the primary growth of underground shoots determines the diameter of mature culms (Dong, 27). However, its underlying cellular and molecular basis remains largely unknown. In this study, we systematically investigate the development of underground bamboo shoots, and explore its key cellular and molecular mechanisms. Morphological and cellular characteristics of the primary growth of Moso underground shoots Although extensive studies have been performed to characterize the development of shoot apical meristems (SAMs) in model plants such as rice, maize and Arabidopsis, the structure of the bamboo SAM and its developmental characteristics are largely unknown. We found that the apical meristem of the bamboo shoot was roughly semi-oval and consisted of six layers of cells (Fig. 2a). This is different from that in model plants. For example, the SAM of Arabidopsis is a roughly triangular-shaped dome and consists of three layers of cells (Barton, 21). The meristems of maize (Zea mays) and rice (Oryza sativa) are similarly organized with Arabidopsis, but tend to be taller and more fingershaped (Ohtsu et al., 27; Barton, 21; Tsuda et al., 211). However, according to our investigation (Fig. S4), SAMs of early tillers of Moso seedling are similar to those of rice and maize. Therefore, the underlying mechanism regulating the changes of SAM structure during the tilling process is interesting. Further investigation might discover important genes promoting the development of SAM of bamboo. Interestingly, by mathematical modelling, we found that the primary thickening growth displayed an apparently spiral pattern. By checking the angles between the pith cavity and the culm wall, we discovered that the spiral pattern should commonly exist Fig. 7 Overview of differentially expressed genes (DEGs) between the thick wall variant and wild-type Moso. The figure was generated using MapMan software. Red denotes upregulated genes, and green denotes downregulated genes. The log 2 -fold change in the transcript levels was used for the analysis. Each square represents a single gene within the pathways.

12 92 Research Table 2 Differentially expressed genes related to cell wall development Gene ID Description Ratio (variant/ wild-type) Adjusted P-value PH1693G43 Cellulose synthase E-5 PH159G197 Expansin PH11579G33 Expansin PH11624G2 Expansin PH1135G148 Expansin PH1397G93 Expansin PH14229G1 Expansin PH12238G31 Expansin PH1659G19 Expansin PH11155G36 Expansin PH1128G51 Expansin PH1323G43 Fasciclin arabinogalactan protein PH129G25 Fasciclin arabinogalactan protein PH11288G19 Fasciclin arabinogalactan protein PH12598G12 Fasciclin arabinogalactan protein PH11246G58 Fasciclin arabinogalactan protein PH164G94 Pectinacetylesterase PH12448G2 Pectinesterase inhibitor PH1331G7 Pectinesterase inhibitor PH132G81 Pectinesterase PH11692G24 Pectinesterase PH132G82 Pectinesterase PH12992G19 Pectinesterase PH14115G8 Phenylalanine ammonialyase PH11834G29 Long-chain fatty acid CoA ligase PH1435G67 Caffeoyl CoA-O-methyltransferase PH1994G68 Caffeoyl CoA-O-methyltransferase PH11283G36 O-methyltransferase in the primary thickening growth of bamboo plants. The spiral growth pattern is found commonly in the perennial growth of tree trunks (Shi et al., 215). However, so far, such a growth pattern has not been discovered in monocotyledon plants except bamboo, whose stem completes the morphological construction only once according to our investigation. We speculate that such a growth pattern might help Moso shoots grow out from underground better than the straight growth. Further investigation of the relationship between the outside mechanical force and the corresponding genetic mechanism of the spiral growth should be valuable for elucidating the key processes involved in coordinating the mechanical constraints on plant development, which is still poorly understood so far (Bassel et al., 214). As pith cells of bamboo will eventually die and form a cavity which has no value in the utility of bamboo culm, studies on pith development have received little attention. However, in this study we found that the increase of pith tissue area played an important role in promoting the primary thickening growth of Moso underground shoots and driving the development process of culms of bamboo plants with different sizes (Fig. 4). The reduction of pith tissue indeed results in a decreased circumference in the thick wall varient, which has little pith tissue when compared to that of the wild-type (WT) Moso (Guo, 23; Guo et al., 25). These results have rediscovered the potential functions of pith in bamboo, and to some extent explained the causes of the various sizes of bamboo culms. Exploring the cellular and molecular mechanisms underlying the unusual organization and differentiation of the thick wall variant Moso SAM The thick wall Moso variant, Phyllostachys edulis Pachyloen, was first discovered in late 198s in Jiangxi Province of China, and is famous for its thick culm wall (Guo, 23; Guo et al., 25). Fasciated culm is another distinguishable phenotype of the variant. It is the only stable variant of Moso identified so far with narrow pith cavity, making it an ideal system to study pith development. However, over 3 yr since its discovery, the developmental mechanism of its narrow pith cavity and fasciated culm has remained unclear. The results in the present study indicate that the fasciated culm should be caused by its flat SAM which is different from the circular SAM of the WT (Figs 2, 6). This correlation seems very common because a number of mutants in other plant species with flat SAM possess fasciated stems, such as bru1 of Arabidopsis and stf of sunflower (Helianthus annuus) (Laufs et al., 1998; Takeda et al., 24; Williams et al., 25; Fambrini et al., 26; Inagaki et al., 26; Jia et al., 216). In addition, our anatomical results indicate that greater differentiation of vascular tissues by SAM resulted in fewer pith cells in the thick wall variant, which in turn caused its narrow pith cavity (Fig. 5e). The abnormal differentiation highly possibly resulted from the abnormal SAM organization of the thick wall variant, which was larger and had more cells (Table 1; Fig. 6). Several

13 Research 93 Table 3 Differentially expressed genes related to plant hormones Gene ID Description Ratio (variant/ wild-type) Adjusted P-value PH1484G74 Auxin efflux carrier PIN PH146G9 BIG PH11G145 Auxin responsive SAUR gene PH1138G7 Auxin responsive SAUR gene PH157G88 Auxin responsive SAUR gene PH1371G4 Auxin responsive SAUR gene PH152G13 Auxin responsive SAUR gene PH11832G2 Auxin responsive SAUR gene PH111G26 Auxin responsive SAUR gene PH14171G7 Auxin responsive SAUR gene PH14171G2 Auxin responsive SAUR gene PH1314G11 Auxin responsive SAUR gene PH11279G41 Cytokinin dehydrogenase PH1161G3 Cytokinin-N-glucosyltransferase PH19953G1 Cytokinin-O-glucosyltransferase PH125921G1 Cytokinin-O-glucosyltransferase PH1492G3 Cytokinin-O-glucosyltransferase PH143G16 Ethylene insensitive PH1447G83 Ethylene responsive protein PH1812G29 Ethylene responsive protein PH13624G14 Ethylene responsive protein PH118G79 Ethylene responsive transcription factor PH1727G25 GRAS transcription factor PH173G121 GRAS transcription factor E-5 PH1515G23 Gibberellin receptor PH1515G18 Gibberellin receptor PH1515G15 Gibberellin receptor PH1475G3 Gibberellin receptor PH1147G1 Gibberellin receptor PH1165G7 Gibberellin receptor PH1896G44 GRAS family transcription factor PH119544G1 GRAS family transcription factor PH173G1 GRAS family transcription factor PH1246G1 Brassinosteroid responsive ring H mutants in other plant species with similar SAM structure have been found to produce more vascular tissues (Williams et al., 25; Fambrini et al., 26; Jia et al., 216). Larger shoot apical meristem with more cells was considered to produce more auxins (Moon et al., 26). Higher endogenous free auxin levels were indeed found in some mutants with larger SAM size (Tang & Knap, 1998; Fambrini et al., 26; Omar et al., 214). Accumulated evidence has demonstrated that auxin plays a fundamental role in the induction and specification of vascular bundles (Reinhardt, 23; Demura & Fukuda, 27; Ursache et al., 213; Smet & De Rybel, 216). Our comparative transcriptome profiling analysis further supports this hypothesis. A number of genes relating to auxin transport and signalling pathways were upregulated in the thick wall variant (Table 3), which in turn might trigger the expression of a series of regulators that could promote the development of vascular tissue such as AP2/EREBP (Ehlting et al., 25; Etchells et al., 212; Vahala et al., 213; Smet & De Rybel, 216), bhlh (Kubo et al., 25; Zhao et al., 25; Ohashi-Ito & Bergmann, 27), bzip (Paux et al., 24; Ehlting et al., 25), HD (Ko & Han, 24; Ko et al., 24; Paux et al., 24; Ehlting et al., 25; Kubo et al., 25), MYB (Hertzberg et al., 21; Bonke et al., 23; Ohashi-Ito & Fukuda, 21), NAC (Demura et al., 22; Kubo et al., 25), WRKY (Zhao et al., 25; Andersson-Gunneras et al., 26) and Zinc finger (Ko & Han, 24; Kubo et al., 25) transcription factors (Table S7), as well as their downstream functional genes such as various metabolism pathway genes. These include genes related to cellulose synthase and lignin biosynthesis (Ohashi-Ito & Fukuda, 21; Table 2), which have been reported to function together in promoting the development of vascular tissues (Yang & Wang, 216). Auxin might also trigger or coordinate other hormones that have been reported to play important roles in cambia maintenance and specification, such as brassinosteroid (Ohashi-Ito & Fukuda, 21), cytokinin (Milhinhos & Miguel, 213), ethylene (Smet & De Rybel, 216) and gibberellin (Ursache et al., 213); these may have promoted the formation of vascular tissues of the thick wall variant because a number of genes relate to these hormones were also significantly upregulated (Table 3). It is notable that the biomass of the thick wall variant decreased when compared to that of WT Moso (Guo, 23; Guo et al., 25). Therefore, the proportion of pith and vascular

14 94 Research (a) Wild Variant (c) (d) (b) Cellulose (μg g 1 FW) * Thick wall moso Species Wild moso (e) Angle Ring number (from outside to inside) Fig. 8 The thicker cell wall with higher cellulose content possibly weakened the spiral growth of the thick wall Moso. (a) Cell wall thickness of the wild-type (WT) Moso (1, 2) and the variant (3, 4). 2 and 4 were the closer look of 1 and 3, respectively. Red arrows indicate cell wall thickness. (b) Cellulose contents of S-4 bamboo shoots from the WT and the thick wall variant. Data are means SD. *, P value <.5. (c) Twelve node rings extracted from an S-6 stage bamboo shoot of the thick wall variant. (d) Merge pictures of the observed node rings (black solid lines) and the predicted rings (red solid lines). Numbers at the top indicate the ring numbers from outside to inside. The dark dashed line represents the direction of major axis and the red dashed line represents the direction of x-axis. (e) Angles between major axis and x-axis of the 23 node rings. tissue differentiation from the SAM at the very beginning is possibly pivotal for the wood formation of bamboo, and reiterates the important role of pith tissue in determining the wood production of bamboo. The identified differentially expressed genes (DEGs) might also influence the spiral growth of Moso underground shoots Interestingly, we found that the spiral growth of the thick wall variant was different from that of WT Moso. The changes of angles different node rings in the thick wall variant were smaller (Figs 3h, 8c e). Cell wall mechanics have been acknowledged as important for growth in plants (Braybrook & Jonsson, 216). In addition, previous work has reported that the rigid cellulose fibers play an inhibitory role in cell growth (Sassi & Traas, 215). Therefore, we assumed that the thicker cell wall with high cellulose content resulting from upregulated cell wall biosynthesis genes in the variant might have resulted in the decreased sensitivity of the underground shoots to the mechanical pressure from the surrounding environment, which finally caused the weaker spiral growth of the variant. It is notable that DEGs related to vascular tissue development identified in this work, might also play a role in the spiral growth of underground shoots. In summary, the enlarged SAM with more cells of the thick wall variant might trigger a series of genes promoting vascular tissue development but inhibiting spiral growth, and finally result in the narrow pith cavity and weak spiral growth of the thick wall variant Moso. Acknowledgements We thank Jinsheng Huang of Yichun Forestry Institute, Jianping Jiang of Wanzai Forestry Bureau and Jinyuan Forestry Station for helping us to collect samples of the thick wall variant and its native wild-type Moso. This work was supported by grants from the Natural Science Foundation of China (grant nos , and ), a project funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions and a project supported by the Open Research Funds of the State Key Laboratory of Genetic Engineering, Fudan University (grant no. SKLGE-159).