Morphological Characteristics and Transcriptome Comparisons of the Shoot Buds from Flowering and Non-Flowering Pleioblastus pygmaeus

: Bamboo plants have a distinctive life cycle with long ﬂowering periodicity. Many species remain in vegetative growth for decades, followed by large-scale ﬂowering and subsequent death. Floral transition is activated while shoot buds are still dormant in bamboo plants. In this study, we performed morphological characterization and transcriptome analysis of the shoot buds at di ﬀ erent growth stages from ﬂowering and non-ﬂowering Pleioblastus pygmaeus . The morphological and anatomical structures of the dormant shoot buds were similar in ﬂowering and non-ﬂowering plants, while there was an obvious di ﬀ erence between the ﬂower buds from ﬂowering plants and the leaf buds from non-ﬂowering plants. The transcriptomes of the dormant shoot buds, germinated shoots, and ﬂower buds from ﬂowering P . pygmaeus , and the dormant shoot buds, germinated shoots, and leaf buds from non-ﬂowering P . pygmaeus were proﬁled and compared by RNA-Seq. The identiﬁed sequences were mostly related to metabolic synthesis, signal transmission, translation, and other functions. A total of 2434 unigenes involved in di ﬀ erent ﬂowering pathways were screened from transcriptome comparisons. The di ﬀ erentially expressed unigenes associated with the photoperiod pathway were related to circadian rhythm and plant hormone signal transduction. Moreover, the relative expression levels of a few key ﬂowering-related genes such as CO , FT , FLC , and SOC1 were quantiﬁed by q RT-PCR, which was in accordance with RNA-Seq. The study revealed morphological di ﬀ erences in the shoot buds at di ﬀ erent growth stages and screened ﬂowering-related genes by transcriptome comparisons of the shoot buds from ﬂowering and non-ﬂowering P . pygmaeus , which will enrich the research on reproductive biology of bamboo plants and shed light on the molecular mechanism of the ﬂoral transition in bamboo plants.


Introduction
As an important forest resource with the advantages of rapid growth, strong adaptability and great reproduction ability, bamboo plants play an irreplaceable role in alleviating wood resource shortages, protecting the ecological environment and promoting ecological civilization construction [1,2]. Bamboo plants are perennial flowering plants with long flowering cycles varying from a few years to several decades [2]. The flowering process may lead to large-scale death of bamboo plants or even the

Quality Control and Transcriptome Assembly
The raw sequencing data was deposited in NCBI SRA with the accession number PRJNA648794. The raw sequencing reads were cleaned by removing adapter sequences, reads with ambiguous N bases, and low quality reads (i.e., the number of bases with quality value Qphred ≤ 20 accounts for more than 50% of the entire reads) with RSeQC [17]. A de novo assembly was performed using Trinity 2.2 with min_kmer_cov set to 2 and all other parameters by default [18]. Different contigs from the same transcript were linked based on double-terminal information, and the transcripts were obtained by further sequence splicing. Then the different transcript sequences of each gene were profiled in a transcript cluster unit by Corset [19]. Finally, the library including irredundant unigenes was constructed.

Gene Annotation and Differential Gene Expression
All unigenes were aligned to seven major databases using Blast2GO with default parameters [20], including NR (NCBI non-redundant protein sequences), NT (NCBI non-redundant nucleotide sequences), Pfam (Protein family), KOG (Eukaryotic ortholog group), SWISS-PROT (A manually annotated and reviewed protein sequence database), KEGG (Kyoto encyclopedia of genes and genomes), and GO (Gene Ontology). The biology functions of the unigenes were annotated based on functional information of the above databases. The e-value distribution map was calculated and drawn according to NR library comparison annotation, and the screening criteria was set as e-value < 1.0 × e −5 . The functional annotation of flowering-related unigenes of P. pygmaeus was referred from the flowering-related genes identified from Arabidopsis thaliana, Oryza sativa, Sorghum bicolor, and Brachypodium distachyon. The gene expression level in each sample was quantified by RSEM (RNA-Seq by Expectation Maximization) with Bowtie alignment program by default [21]. Differential expression analysis in the comparisons was performed using the DESeq R package based on negative binomial distribution, and p-values were adjusted using Benjamini and Hochberg's approach [22]. The unigenes with an adjusted p-value < 0.05 were assigned as differentially expressed unigenes (DEUs). GO enrichment analysis of the DEUs was implemented by GOseq based on Wallenius non-central hyper-geometric distribution [23]. KOBAS was used to test the statistical enrichment of the DEUs in KEGG pathways [24]. The unigenes involved in different flowering pathways were clustered based on FPKM (Fragments per Kilobase Million) by Heatmapper (http://www.heatmapper.ca/expression/) with default parameters.

qRT-PCR Verification
The obtained samples were rapidly frozen in liquid nitrogen and stored at −80 • C for qRT-PCR. The mRNA was extracted by a Column Plant RNAout Kit (CAT#:71203, Tiandz, Beijing, China). The quantity of RNA was determined by Nanodrop 2000c, and the quality of RNA was detected by Agilent 2100. The cDNA was obtained using PrimeScript™ RT reagent Kit with gDNA Eraser (RR047A, Takara, Dalian, China).
The key flowering-related genes, such as CO, FLC, FT, and SOC1, which play important roles in several flowering pathways, were chosen for qRT-PCR verification. Tubulin was used as the internal reference gene [25] and all primers (Supplementary Table S1) were designed using Primer 5.0. An IQ Multicolor real-time PCR automatic amplification apparatus was used for qRT-PCR with three biological repeats for each sample. The relative expression level of each gene was calculated using 2 −∆CT [26]. ANOVA (p < 0.05) was used to identify significant differences of gene expression among different samples.

Morphological and Anatomical Characterization of the Samples
According to the biological characteristics of bamboo plants, the shoot buds during three different developmental stages, including the FE, FM, and FL from flowering P. pygmaeus, and the NE, NM, and NL from non-flowering P. pygmaeus were harvested for morphological observation ( Figure 1). Morphological phenotypes of the dormant shoot buds were similar in flowering and non-flowering plants: the main axis was short and the upper layer was covered with 7-8 layers of shoot sheath (Figure 2A,E), and the appearance of the apical growth cone was conical ( Figure 3A,D). Accompanied with the elongation of the main axis, the germinated shoot buds from most flowering plants began to differentiate, and floret primordia continuously sprouted on both sides of the main axis ( Figures 2C and 3B). The morphological characteristics of the germinated shoot buds from a few flowering plants ( Figure 2B) were similar with those from non-flowering plants ( Figure 2F): with the elongation of principal axis, the tunica cells underwent anticlinal division, the apical growth cones elongated longitudinally ( Figure 3E), and bract primordia continuously formed around the growth cones. The flower buds and leaf buds displayed obvious differences in anatomy. The 1-2 cm flower buds developed into a mixed inflorescence with 1-3 spikelets and 9-20 florets ( Figure 2D). The appearance of the apical meristem was hemispherical ( Figure 3C). However, leaf primordia continuously formed on both sides of the main axis ( Figure 2G), and the appearance of the apical meristem was conical in 1-2 cm leaf buds ( Figure 3F).

Data Filtering and Assembly
Transcriptome sequencing information of FE, FM, FL, NE, NM, and NL is shown in Supplementary Table S2. After filtration, the guanine-cytosine (GC) content of clean reads was more than 50%, and the average value of Q30 was more than 90%, which indicated that the clean data can be used for subsequent assembly. A total of 616,137 transcripts were obtained from all samples.

Data Filtering and Assembly
Transcriptome sequencing information of FE, FM, FL, NE, NM, and NL is shown in Supplementary  Table S2. After filtration, the guanine-cytosine (GC) content of clean reads was more than 50%, and the average value of Q30 was more than 90%, which indicated that the clean data can be used for subsequent assembly.

Data Filtering and Assembly
Transcriptome sequencing information of FE, FM, FL, NE, NM, and NL is shown in Supplementary Table S2. After filtration, the guanine-cytosine (GC) content of clean reads was more than 50%, and the average value of Q30 was more than 90%, which indicated that the clean data can be used for subsequent assembly. A total of 616,137 transcripts were obtained from all samples.

Gene Annotation
A total of 364,840 irredundant unigenes were obtained based on functional annotations of the seven major databases, including NR, NT, Pfam, KOG, SWISS-PROT, KEGG, and GO. Among them, most unigenes were annotated in the NR database with 201,547 accounting for 55.24% of the total unigenes. It was followed by NT and GO databases with 199,699 and 154,573 unigenes, accounting for 54.73% and 42.36% of the total unigenes, respectively. There were 147,740, 144,866, 76,768, and 54,875 unigenes annotated in Pfam, SWISS-PROT, KEGG, and KOG databases, respectively, which accounted for 40.49%, 39.70%, 21.04%, and 15.04% of the total unigenes, respectively. As many as 30,935 unigenes were shared in the seven databases, accounting for 8.47% of the total unigenes (Supplementary Table S3).
Based on the distribution of e-values in the database, 44.70% of matched sequences showed strong homology (e-value < 1.0 × 10 −60 ), and 55.30% of matched sequences showed general homology (between 1.0 × 10 −5 and 1.0 × 10 −60 ) ( Figure 5A). Transcript comparisons indicated that P. pygmaeus had the most homologous sequences with Oryza sativa (22.70%), followed by Brachypodium distachyon (14.73%), Setaria italica (11.29%), Oryza brachyantha (11.14%), and Phyllostachys edulis (3.05%) ( Figure 5B). The GO, KOG, and KEGG functional annotations showed that the unigenes were mostly related to cellular process, material metabolism, and genetic information processes. GO analysis indicated that the unigenes with metabolic and synthetic functions were more abundant in flowering plants than in non-flowering plants, which might be related to continuous metabolism in flower bud differentiation. Based on KEGG enrichment analysis, the genes involved in plant hormone signal transduction and circadian rhythm may play an important role in the flowering process of P. pygmaeus.

Gene Annotation
A total of 364,840 irredundant unigenes were obtained based on functional annotations of the seven major databases, including NR, NT, Pfam, KOG, SWISS-PROT, KEGG, and GO. Among them, most unigenes were annotated in the NR database with 201,547 accounting for 55.24% of the total unigenes. It was followed by NT and GO databases with 199,699 and 154,573 unigenes, accounting for 54.73% and 42.36% of the total unigenes, respectively. There were 147,740, 144,866, 76,768, and 54,875 unigenes annotated in Pfam, SWISS-PROT, KEGG, and KOG databases, respectively, which accounted for 40.49%, 39.70%, 21.04%, and 15.04% of the total unigenes, respectively. As many as 30,935 unigenes were shared in the seven databases, accounting for 8.47% of the total unigenes (Supplementary Table S3).
Based on the distribution of e-values in the database, 44.70% of matched sequences showed strong homology (e-value < 1.0 × 10 −60 ), and 55.30% of matched sequences showed general homology (between 1.0 × 10 −5 and 1.0 × 10 −60 ) ( Figure 5A). Transcript comparisons indicated that P. pygmaeus had the most homologous sequences with Oryza sativa (22.70%), followed by Brachypodium distachyon (14.73%), Setaria italica (11.29%), Oryza brachyantha (11.14%), and Phyllostachys edulis (3.05%) ( Figure  5B). The GO, KOG, and KEGG functional annotations showed that the unigenes were mostly related to cellular process, material metabolism, and genetic information processes. GO analysis indicated that the unigenes with metabolic and synthetic functions were more abundant in flowering plants than in non-flowering plants, which might be related to continuous metabolism in flower bud differentiation. Based on KEGG enrichment analysis, the genes involved in plant hormone signal transduction and circadian rhythm may play an important role in the flowering process of P. pygmaeus.

Differential Gene Enrichment Analysis
To identify differentially expressed flowering-related genes, the transcriptomes of the shoot buds at the same developmental stage from flowering and non-flowering plants were compared ( Figure 6). The DEUs in FE vs. NE were mainly involved in metabolic processes and binding agents ( Figure 6A). The unigenes were enriched in biological processes and binding in FM vs. NM ( Figure  6B), while the unigenes associated with biosynthesis, metabolic processes and catalytic activity were

Differential Gene Enrichment Analysis
To identify differentially expressed flowering-related genes, the transcriptomes of the shoot buds at the same developmental stage from flowering and non-flowering plants were compared ( Figure 6). The DEUs in FE vs. NE were mainly involved in metabolic processes and binding agents ( Figure 6A). The unigenes were enriched in biological processes and binding in FM vs. NM ( Figure 6B), while the unigenes associated with biosynthesis, metabolic processes and catalytic activity were abundant in  Figure 6C). These unigenes may be related to metabolism processes during flower bud differentiation and/or flower organ primordium establishment. Moreover, the DEUs in FE vs. FM and NE vs. NM were most abundant in catalytic and binding activity according to GO enrichment analysis ( Figure 6D,E). The results indicated that mRNA abundance of the unigenes related to catalytic and binding activity was much higher in flowering P. pygmaeus than that in non-flowering plants.
Based on significant enrichment analysis using KEGG, the unigenes in genetic information processing (2675) showed the highest enrichment in FE vs. NE (Supplementary Table S4), which was mainly concentrated in the spliceosome (539), protein processing in the endoplasmic reticulum (502), RNA transport (436), and the secondary enrichment was in metabolism (1650). The highest number of unigenes was also found in genetic information processing (1645) in FM vs. NM (Supplementary  Table S5), which mainly included protein processing in the endoplasmic reticulum (412), spliceosome (382), mRNA surveillance pathway (269), and the secondary enrichment was also metabolism (1281). In FL vs. NL (Supplementary Table S6), most enrichment branches were involved in metabolism (2047), which was mainly composed of oxidative phosphorylation (301), pyruvate metadata (242), and glyoxylate and dicarboxylate metabolism (206). A total of 128 DEUs were obtained in the three comparisons, which were mainly involved in cellular processes, environmental information processing, genetic information processing, metabolism and organismal systems. In  2017) suggested that the floral process of Phyllostachys pubescens was related to plant-pathogen interactions, protein processing in the endoplasmic reticulum, and plant hormone signal transaction [27]. In all comparisons (Supplementary Table S9), the plant-pathogen interaction was only enriched in FE vs. FM, which was not significant yet, indicating the blossoming process of P. pygmaeus had a low relationship with plant-pathogen interactions. Protein processing in the endoplasmic reticulum was not significantly enriched in FE vs. NE and FM vs. NM, but it was highly enriched in FL vs. NL, which indicated that there was more protein processing in flower buds than in rhizome buds. Significant enrichment of plant hormone signal transduction was present in each group, which indicated that plant hormone signal transduction plays an important role in the flowering process of P. pygmaeus.
In addition, the q-values of the plant circadian rhythm pathway in FM vs. NM and FL vs. NL were 0.29 and 0.04, respectively (Supplementary Table S9). The enrichment of the pathway was significantly different in flower buds and leaf buds. The q-values of FE vs. FM and NE vs. NM were 1.00 and 0.11, respectively (Supplementary Table S9), which indicated that the enrichment degree of differentially expressed genes in the circadian rhythm pathway was much greater in flowering P. pygmaeus than that in non-flowering plants. The above results indicated that the plant circadian rhythm pathway was critical in the flower formation process of P. pygmaeus.

Expression Analysis of Key Genes Involved in Floral Development
The flowering process was co-regulated by multiple regulation pathways in plants. A total of 129 flowering-related genes with 2434 unigenes were identified from transcriptome analysis of P. pygmaeus. There were 52 genes with 647 unigenes, 19 genes with 298 unigenes, 6 genes with 93 unigenes, 20 genes with 1023 unigenes, 21 genes with 179 unigenes, 4 genes with 71 unigenes, and 7 genes with 123 unigenes, in the photoperiod pathway, vernalization pathway, autonomous pathway, gibberellins pathway, age pathway, pentose phosphate pathway, and flowering signal integrator, respectively (Supplementary Table S10, Figure 7). The highly expressed unigenes were most abundant in flowering P. pygmaeus. In particular, there were more unigenes involved in the photoperiod pathway expressed in FE and FL, such as CO, PHY, TCP, etc., compared to those in NE and NL. Moreover, a few SPLs, such as SPL3, SPL4, SPL5, etc., and flowering signal integrators, such as FT, FD, SOC1, etc., were highly expressed in flower buds (Figure 7).

Expression Analysis of Key Genes Involved in Floral Development
The flowering process was co-regulated by multiple regulation pathways in plants. A total of 129 flowering-related genes with 2434 unigenes were identified from transcriptome analysis of P. pygmaeus. There were 52 genes with 647 unigenes, 19 genes with 298 unigenes, 6 genes with 93 unigenes, 20 genes with 1023 unigenes, 21 genes with 179 unigenes, 4 genes with 71 unigenes, and 7 genes with 123 unigenes, in the photoperiod pathway, vernalization pathway, autonomous pathway, gibberellins pathway, age pathway, pentose phosphate pathway, and flowering signal integrator, respectively (Supplementary Table S10, Figure 7). The highly expressed unigenes were most abundant in flowering P. pygmaeus. In particular, there were more unigenes involved in the photoperiod pathway expressed in FE and FL, such as CO, PHY, TCP, etc., compared to those in NE and NL. Moreover, a few SPLs, such as SPL3, SPL4, SPL5, etc., and flowering signal integrators, such as FT, FD, SOC1, etc., were highly expressed in flower buds (Figure 7).

RNA-Seq Expression Validation of Key Genes Involved in Floral Development
As shown in Figure 8, the relative expression level of CO, SOC1, and FT was significantly higher in flowering P. pygmaeus than that in non-flowering plants, while the expression of FLC was lower in flowering P. pygmaeus than that in non-flowering plants. The expression of CO was significantly higher in FE, FM, and FL than that in NE, NM, and NL. The expression of SOC1 was significantly higher in FM and FL than that in NM and NL. CO and SOC1 were limited or rarely expressed in dormant shoots, while they were highly induced in germinated shoots and flower/leaf buds. The expression of SOC1 increased more significantly than that of CO. CO and SOC1 were both expressed in flower buds and gradually decreased with the completion of floral organ formation, indicating that they played an important role in flower transition of P. pygmaeus. The expression of the FT gene

RNA-Seq Expression Validation of Key Genes Involved in Floral Development
As shown in Figure 8, the relative expression level of CO, SOC1, and FT was significantly higher in flowering P. pygmaeus than that in non-flowering plants, while the expression of FLC was lower in flowering P. pygmaeus than that in non-flowering plants. The expression of CO was significantly higher in FE, FM, and FL than that in NE, NM, and NL. The expression of SOC1 was significantly higher in FM and FL than that in NM and NL. CO and SOC1 were limited or rarely expressed in dormant shoots, while they were highly induced in germinated shoots and flower/leaf buds. The expression of SOC1 increased more significantly than that of CO. CO and SOC1 were both expressed in flower buds and gradually decreased with the completion of floral organ formation, indicating that they played an important role in flower transition of P. pygmaeus. The expression of the FT gene showed an upward trend throughout the whole flowering process and was significantly higher in flower buds, indicating that FT played an important role in flower vessel maturation of P. pygmaeus. However, the expression of FLC was significantly lower in FE, FM, and FL than that in NE, NM, and NL, indicating that FLC was inhibited in the flower transition of P. pygmaeus. showed an upward trend throughout the whole flowering process and was significantly higher in flower buds, indicating that FT played an important role in flower vessel maturation of P. pygmaeus. However, the expression of FLC was significantly lower in FE, FM, and FL than that in NE, NM, and NL, indicating that FLC was inhibited in the flower transition of P. pygmaeus.

Discussion
Plants accomplish the transition from vegetative growth to reproductive growth by flowering. The flowering process is a complex process in higher plants, which is regulated by different flowering pathways, including the photoperiod pathway, vernalization pathway, autonomous pathway, GA pathway, age pathway, etc. [28]. In this study, a total of 129 flowering-related genes with 2434 unigenes involved in different flowering pathways were screened from transcriptome comparisons of the shoot buds from flowering and non-flowering P. pygmaeus.
The photoperiod pathway is a key genetic mechanism regulating flowering in plants [29]. A total of 52 genes with 647 unigenes involved in the photoperiod pathway were found in the transcriptomes of P. pygmaeus in the study, including CO, FT, SOC1, LFY and other genes. In the photoperiod pathway, the FT gene is a direct target gene of the transcriptional regulator, CO [30,31]. Under long sunlight conditions, sustained expression of CO can activate the expression of FT [32], meanwhile FT can activate SOC1 and LFY to promote flowering [33].
The vernalization pathway is regulated by low-temperature signals. A total of 19 genes with 298 unigenes related to vernalization were found in transcription analysis of P. pygmaeus in the study, including key flowering-related genes, FLC, VRN1, VRN2, etc. [34]. FLC belongs to the MADS-box family, which is responsible for maintaining the nutritional state of bud tips [35]. FLC can control flowering by inhibiting the expression of FT and SOC1 [36]. VRN1 also belongs to the MADS family and is homologous with the AP1 gene from Arabidopsis thaliana (L.) Heynh., 1842, which can be induced by low temperature [37]. VRN2 is a flowering repressor that is down-regulated by

Discussion
Plants accomplish the transition from vegetative growth to reproductive growth by flowering. The flowering process is a complex process in higher plants, which is regulated by different flowering pathways, including the photoperiod pathway, vernalization pathway, autonomous pathway, GA pathway, age pathway, etc. [28]. In this study, a total of 129 flowering-related genes with 2434 unigenes involved in different flowering pathways were screened from transcriptome comparisons of the shoot buds from flowering and non-flowering P. pygmaeus.
The photoperiod pathway is a key genetic mechanism regulating flowering in plants [29]. A total of 52 genes with 647 unigenes involved in the photoperiod pathway were found in the transcriptomes of P. pygmaeus in the study, including CO, FT, SOC1, LFY and other genes. In the photoperiod pathway, the FT gene is a direct target gene of the transcriptional regulator, CO [30,31]. Under long sunlight conditions, sustained expression of CO can activate the expression of FT [32], meanwhile FT can activate SOC1 and LFY to promote flowering [33].
The vernalization pathway is regulated by low-temperature signals. A total of 19 genes with 298 unigenes related to vernalization were found in transcription analysis of P. pygmaeus in the study, including key flowering-related genes, FLC, VRN1, VRN2, etc. [34]. FLC belongs to the MADS-box family, which is responsible for maintaining the nutritional state of bud tips [35]. FLC can control flowering by inhibiting the expression of FT and SOC1 [36]. VRN1 also belongs to the MADS family and is homologous with the AP1 gene from Arabidopsis thaliana (L.) Heynh., 1842, which can be induced by low temperature [37]. VRN2 is a flowering repressor that is down-regulated by vernalization and can inhibit the expression of VRN1 [38]. Additionally, 127 unigenes identified from P. pygmaeus transcriptomes are highly homologous with VIN1, VIN2, and VIN3, which also inhibit FLC expression at low temperature [39].
The autonomous pathway is an independent floral induction pathway. A series of key flowering-related genes, such as FCA, FPA, FLD, FLK, FY, FVE, and LD, have been identified in the pathway [40,41]. There were 93 homologous unigenes related to the autonomous pathway found in the transcriptional analysis of P. pygmaeus in the study, including FCA (12), FPA (15), FLK (39), FY (11), FVE (6), and LD (10). FCA and FPA encode RNA binding proteins containing plant-specific RNA recognition motifs (RRMs), which can inhibit the accumulation of FLC [42,43]. LD is the first independent gene cloned from A. thaliana in the pathway [44]. FLK, FY, and FVE encode RNA binding proteins that inhibit expression of FLC [42,45].
The GA pathway can affect flowering induction and organ development by regulating endogenous hormones. A total of 20 genes with 1023 unigenes involved in the pathway were found in the transcriptome comparisons of P. pygmaeus in the study. GA promotes flowering by increasing expression of the integron genes (e.g., SOC1, LFY and FT) in A. thaliana [46]. GAI and RGA genes from the GRAS family can inhibit the GA response in plants [47]. There were six GAI unigenes and five RGA (RGA1/2/3/4/5) with 892 unigenes, accounting for 87.19% of the total unigenes in the GA pathway of P. pygmaeus. It was reported that the GID receptor can detect GA activity in Oryza sativa [48]. There were two GID1 receptors found in the transcription analysis of P. pygmaeus, including GID1B and GID1C.
The ageing pathway is an endogenous way to control flowering time through plant growth. In the study, 21 SPL genes (SPL1-SPL19, and SPL21) with 179 unigenes were found in transcriptome comparisons of P. pygmaeus. A few SPLs, as key TFs in the ageing pathway, can promote the transformation from vegetative growth to reproductive growth of plants [49]. It was reported that several SPLs can positively induce the FT gene, and its activity increased with age [50]. A few SPLs can further activate the expression of flowering-related genes such as LFY, AP1 and FUL [51]. In addition, some SPLs may be indirectly involved in the photoperiod and GA pathways [52,53].
As a key gene in the photoperiod pathway, CO directly regulates the expression of the FT gene, which promotes flowering [30,31]. FLC is an inhibitory factor in flowering regulation and participates in vernalization and autonomic pathways. FLC negatively regulates the expression of FT and SOC1 in A. thaliana [36]. The FT gene is a conservative activator of flowering and differently affects the flowering of short-day and long-day plants [32,33]. In A. thaliana, the FT protein interacts with the FD protein to activate AP1 expression and initiate the transition from vegetative growth to reproductive growth [54]. The SOC1 gene is a TF of the MADS-box family. The expression of the SOC1 gene in the stem tip meristem increases in the flower formation process [33]. Moreover, the FT gene and FLC gene can affect the expression of SOC1 in A. thaliana [33,36]. To verify RNA-Seq results, the relative expression level of the key flowering-related genes was quantified by qRT-PCR, which was in accordance with RNA-Seq. The expression of CO, SOC1, and FT was significantly higher in flowering P. pygmaeus than that in non-flowering plants, while FLC expression displayed the opposite phenomenon. The results indicated that they play an important role in regulating the flowering process of P. pygmaeus.

Conclusions
In the study, morphological characteristics of the shoot buds at different developmental stages from flowering and non-flowering P. pygmaeus were revealed. The dormant shoot buds had a similar morphological phenotype in flowering and non-flowering plants, while there were obvious differences in germinated shoot buds and flower/leaf buds. Meanwhile, the transcriptomes of the shoot buds were compared to screen flowering-related genes in P. pygmaeus. The identified unigenes mainly function in metabolic synthesis, signal transmission, translation, and others based on gene annotation. A total of 2434 unigenes involved in the flowering process were screened from transcriptome comparisons of the shoot buds from flowering and non-flowering P. pygmaeus, including 647, 298, 93, 1023, 179 unigenes in the photoperiod pathway, vernalization pathway, autonomous pathway, GA pathway, and ageing pathway, respectively. In particular, the DEUs associated with circadian rhythm and plant hormone signal transduction play an important role in flower transition of P. pygmaeus. Moreover, qRT-PCR Forests 2020, 11, 1229 13 of 16 results indicated that the relative expression level of a few key flowering-related genes such as CO, SOC1, and FT was significantly higher in flowering P. pygmaeus than that in non-flowering plants, which was in accordance with RNA-Seq analysis. This study enriches the reproductive biology of small-size bamboo plants and provides a scientific basis for further understanding of floral transition in bamboo plants.

Conflicts of Interest:
The authors declare no conflict of interest.