1. Introduction

The rapid growth of plants holds substantial importance, which helps plants to occupy their ecological niche, adapt to varying environmental conditions, and can also generate economic value for human beings (Rubin, 2008). In recent years, cultivation of fast-growing plants has emerged as a possible way to meet the increasing global demand for renewable biomass. For example, paulownias timber can be harvested in six years (Cao et al., 2021) with the Eucalyptus species at even a younger age (Zhang and Wang, 2021). Polars can grow very fast under suitable conditions (Bunder et al., 2020), enabling efficient and sufficient timber supply. In practice, the rapid growth and size regulation of crops have a profound impact on agricultural performance and productivity (Sasaki and Ashikari, 2018).

Previous studies suggested that rapid growth and size regulation in plants is affected by several factors including cell proliferation, cell elongation and phytohormones (Zhang and Wang, 2021; Chen et al., 2022). In particular, a large number of genes in controlling plant size have been identified in cereals (e.g., rice, wheat, barley) (Sasaki and Ashikari, 2018) and Arabidopsis thaliana (Regnault et al., 2014), and most of them functioning in the biosynthesis or signal transduction of gibberellins (GAs). Phytohormones, particularly GAs, also play a significant role in culm growth and internode elongation of moso bamboo, Phyllostachys edulis (Tao et al., 2018). The previous studies mainly focused on P. edulis showed that GAs and auxin (IAA) content were positively correlated with the growth rate, while cell division rate during the internode elongation was positively correlated with zeatin content and negatively correlated with abscisic acid (ABA) content (Ding, 1997; Cui et al., 2012; Chen et al., 2022). The application of exogenous GAs (GA3 and GA4) on P. edulis seedlings led to a significant increase in internode length, culm size, and shoot growth rate (Zhang et al., 2018; Chen et al., 2022).

Belonging to the grass family (Poaceae), the bamboos (Bambusoideae) exhibit a high degree of diversity and possess many unique traits compared to other grasses (Liese and Köhl, 2015). The unique traits in woody bamboos include rapid shoot growth, highly lignified mature culms, and massive and synchronous flowering (Wang et al., 2024b), which are absent in herbaceous bamboos. The distinct traits make woody bamboos as important non-timber forest plants due to their ecological, and economic value, as they are extensively distributed across subtropical and tropical regions (Liese and Köhl, 2015). To date, researchers have documented approximately 1700 species (Soreng et al., 2022) of bamboos into one minor diploid herbaceous lineage (HBs, 2n = 2x, 20–24, HH) and three major polyploid woody lineages (Guo et al., 2019). These woody lineages encompass tetraploid temperate woody bamboos (TWBs, 2n = 4x = 46–48, CCDD), tetraploid neotropical woody bamboos (NWBs, 2n = 4x = 40–48, BBCC), and hexaploid paleotropical woody bamboos (PWBs, 2n = 6x = 70–72, AABBCC) (Guo et al., 2019; Ma et al., 2024). Although some PWBs may bear higher chromosome number (2n = 9x = 104), it was revealed that they possessed three haplotype genomes, each including the subgenomes A, B, and C (Wang et al., 2024b). The bamboos represent an exceptional polyploid system and have undergone complex reticulate evolution with multiple hybridizations and polyploidizations, which may trigger widespread species diversification due to adaptation (Alger and Edger, 2020; Ma et al., 2024). The subgenomes from different progenitors can regulate traits asymmetrically, for instance, in allotetraploid cotton Gossypium hirsutum, fiber quality-related homoeologs show phenotypically favorable allele aggregation in different cultivars (You et al., 2023), and salt responsiveness in both Go. hirsutum (AD1) and Go. mustelinum (AD4) was strongly biased toward the D-genome progenitor (Dong et al., 2022). However, a recent study in Nepenthes gracilis illustrated that recessive subgenomes contribute most to the adaptive evolution of tissue-specific genes (Saul et al., 2023).

Woody bamboos as an essential alternative to timber are important sources of lignocellulosic biomass energy (Liese and Köhl, 2015). Previous studies indicated that growth of bamboo shoots exhibit a "slow-fast-slow" pattern (Chen et al., 2022; Niu et al., 2022). Moreover, the fastest growth rate of bamboo shoots was suggested to be positively correlated to culm size of bamboos, and that this was achieved through the simultaneous growth of multiple culm internodes (Wang et al., 2012a; Huang et al., 2016; Tan and Wu, 2018; Xu, 2018). In recent years, high-throughput sequencing technology has been extensively employed in the investigation of the rapid growth mechanisms in woody bamboos. Through transcriptome and proteome analyses, numerous differentially expressed genes (DEGs) were identified to be related to phytohormone biosynthesis and signal transduction, lignin and cellulose biosynthesis, carbohydrate metabolism and key transcription factors (Cui et al., 2012; Peng et al., 2013; Wei et al., 2018). For example, the Snf1 (Wei et al., 2019) and COMT genes may have an effect on the lignin content of woody bamboos (Ma et al., 2024), while ABA and mechanical pressure stimulate rapid secondary cell wall (SCW) thickening by upregulating the MYB transcription factor PeMYB83L in P. edulis (Chen et al., 2022). The microRNAs (miRNAs)-mediated regulatory pathways also play a vital role in rapid growth of P. edulis (Wang et al., 2021a; Yang et al., 2021). Furthermore, DNA methylation could also play a critical role in the regulation of the rapid shoot growth and culm development in Bonia amplexicaulis and P. edulis, and the nanopore direct RNA sequencing (DRS) revealed a dynamic N6-methyladenine (m6A) modification rate in shoots of different height (Niu et al., 2022; Li et al., 2023). And the new genes and whole-genome duplicates have underlined the genetic basis for the evolving rapid growth of woody bamboos (Jin et al., 2021).

In addition, the auxin and cytokinin signaling crosstalk was suggested to regulate bamboo growth by PheARF52 and PheRR3 (Bai et al., 2023), and auxin signaling could directly modulate lignin biosynthesis genes to alter culm lignin content through auxin response factors ARF3 and ARF6 during shoot development (Wang et al., 2024a). By analyzing extensive interspecific culm variations among 77 woody bamboo species, researchers found the heritable legacy playing a key role in the regulation of functional traits of culms and the climate acting as a relative moderating factor (Liu et al., 2024b). The internode diameter, thickness, length, and volume were correlated to one another and the precipitation may affect the primary thickening growth and, consequently, the culm size (Zhang et al., 2024). And the decreased woody bamboo internode length may be due to a decrease in cell number and cell length (Zha et al., 2023). Spatiotemporal transcriptome atlas supports the hypotheses that the intercalary meristem originates from surrounded parenchyma cells through dedifferentiation (Guo et al., 2024). Taken together, the extant studies on rapid growth of bamboos have mainly focused on a few economically important bamboos, particularly the tetraploid P. edulis. Notably, woody bamboos show a great of diversity with culm size. For example, the height of Shibataea chiangshanensis is only 0.5 m, while that of P. edulis can reach 20 m (Shi et al., 2021). The largest known bamboo, hexaploid Dendrocalamus sinicus can reach 37.5 m in height with culm diameter to 28.7 cm (Liu et al., 2024a; Ma et al., 2024).

The woody bamboos have undergone multiple whole-genome duplications (WGDs), as well as small-scale duplications (Ma et al., 2024). These events would lead to the divergence of gene across different tissues through processes such as neo- or sub-functionalization (Guo et al., 2024). Built on our previous work on chromosome-level genome sequences of 11 representative bamboos (Ma et al., 2024), the current study aimed to further investigate the molecular mechanism and evolution of rapid shoot growth in woody bamboos. Importantly, the 11 representative bamboos covered different ploidal levels, genome and culm size diversity. And we adopted an integrated approach with morphological, physiological, transcriptomic and genomic analyses to address the questions.

The height of well-growing woody bamboos and the growth patterns of key bamboos were determined, and levels of representative phytohormones were measured in rapid growing shoots. Comprehensive transcriptomes of different bamboo species were analyzed with key candidate genes regulating the culm growth and internode elongation. We also investigated the expression patterns of those genes in representative bamboos using available bamboo genome and transcriptome data, and compared the gene structure of key genes located in different subgenomes in individual bamboo species. Our study aims to determine the genetic basis of rapid shoot growth in the context of species diversification following hybridizations and polyploidizations of woody bamboos.

2. Materials and methods 2.1. Morphological analysis of representative woody bamboos

To investigate the differences in culm size among woody bamboos, we selected well-growing population of seven representative woody bamboos for measurements. Including three tetraploid TWBs (Ampelocalamus luodianensis, Hsuehochloa calcarea and P. edulis), one tetraploid NWB (Guadua angustifolia), and three hexaploid PWBs (Melocanna baccifera, B. amplexicaulis and D. sinicus).

2.2. Investigation of the growth pattern of shoots and internodes of Phyllostachys edulis

We continuously measured the shoot height and 9th, 10th, 11th, 12th internode length of young shoots of P. edulis until the end of the target internode elongation in April to May 2021 at Zhongchang Village, Yiliang County, Yunnan Province, China (27°53′35″N, 104°42′59″E, alt. 1120 m). A total of 115 well-growing shoots was selected, with the smallest one being 5 cm.

2.3. Quantification of phytohormones in Dendrocalamus sinicus internodes during rapid shoot growth

The samples of shoots used to quantify contents was consistent with transcriptome analyses and anatomical observation in our previous work (Ma et al., 2024). The contents of endogenous IAA, ABA, and cytokinin (CK) were determined using previously published methods with slight modifications (Cao et al., 2021; Song et al., 2022), the levels of GAs were measured by the previously described method (Liu et al., 2019).

2.4. Plant material and transcriptome sequencing

The culms of the herbaceous bamboo, Raddia guianensis were divided into three stages on the basis of different developmental statuses (Fig. S1). We peeled off leaves and leaf sheaths, then culms were collected for transcriptome sequencing and analyses. The sequencing and analyses methods followed previous studies (Tao et al., 2020; Cao et al., 2021; Yang et al., 2021; Chen et al., 2022; Niu et al., 2022).

2.5. Transcriptome analyses

Transcriptome sequence processing, assembly, and quantification of Ra. guianensis culms were performed using methods described (Ma et al., 2024). All of the quantification results above were used to do the clustering analysis among samples in species, and the checked samples were kept for subsequent analyses. Then, DEGs during rapid shoot growth in three woody bamboos (D. sinicus as a large-sized PWB, P. edulis as a large-sized TWB, and B. amplexicaulis as a small-sized PWB) and regular growth in a herbaceous bamboo Ra. guianensis at different developmental stages and tissues were identified. After that, weighted correlation network analysis of DEGs in D. sinicus, P. edulis, and B. amplexicaulis were performed with WGCNA (Langfelder and Horvath, 2008), as well as KEGG and GO enrichment were performed in parallel. And gene interaction networks in important modules were visualized using Cytoscape (Shannon et al., 2003).

2.6. Identification of homoeolog genes in representative bamboos

In order to further investigate the molecular mechanism of rapid shoot growth in woody bamboos, we identified the gene copies related to GAs in 11 bamboo species. We used known genes involved in GAs biosynthesis and signal transduction pathway (https://rapdb.dna.affrc.go.jp/) from Oryza sativa as seed sequence to identify homologous gene copies in two herbaceous species (HBs, Olyra latifolia and Ra. guianensis), and nine woody bamboos: TWBs (Am. luodianensis, H. calcarea and P. edulis), NWBs (Rhipidocladum racemiflorum, Otatea glauca and Gu. angustifolia) and PWBs (M. baccifera, B. amplexicaulis and D. sinicus). The protein sequences of O. sativa (http://www.genobank.org/grass) and 11 bamboo genomes (https://bamboo.genobank.org/) were downloaded, and the GAs-related genes in O. sativa was extracted for seed sequences. Identification methods were described in our previous work (Ma et al., 2024). Moreover, MCScan was used with default parameters to identify the homoeologs among bamboo subgenomes and rice genome (Wang et al., 2012b).

2.7. Gene expression analysis

We further investigated the expression patterns of homoeolog genes involved in GAs during rapid shoot growth in eight woody bamboos with the exclusion of Rh. racemiflorum due to the lack of related shoot samplings, the transcripts per million (TPM) value of shoots at different developmental stages in woody bamboos were employed for gene expression analysis (Ma et al., 2024).

Based on the identified GAs-related DEGs during rapid growth, we further integrated the transcriptome data from young shoot of eight representative woody bamboos to check the expression patterns of GA-related genes among species and subgenomes. In D. sinicus, P. edulis and B. amplexicaulis, we defined the DEGs identified from their own transcriptome data during rapid shoot growth. The remaining bamboo species without rapid growing shoot transcriptome data were calculated using homologous genes that were identified as DEGs in the three species. Initially, we compared the expression proportion of GAs-related DEGs in total GAs-related genes within young shoots across eight woody bamboos with different culm sizes, and obtained the expression patterns of GAs-related DEGs across different woody bamboos. Furthermore, we analyzed the expression levels of GA-related DEGs on different subgenomes in individual bamboo species, including rapid growing shoots of D. sinicus, P. edulis and B. amplexicaulis and young shoots of eight woody bamboos (except Rh. racemiflorum). Through this process, we were able to identify the expression patterns of key GA-related genes across different subgenomes, and the results were exhibited using R package.

2.8. Gene structure analyses of hub genes involved in GAs

To study structure differences of hub genes involved in GAs in bamboos, phylogenetic trees were constructed by IQ-TREE 2 with the sequence alignment trimmed as described in our previous work (Ma et al., 2024). The gene structures of hub genes were analyzed using TBtools (Chen et al., 2020a). The Multiple Expectation Maximization for Motif Elicitation (http://meme-suite.org/tools/meme) program was used to identify conserved motifs using multiple alignment analysis with default parameters. And the domains of the genes were identified using the NCBI Conserved Domain Database (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi). To identify the cis-acting elements, we extracted the 2000 bp upstream sequences of the transcription start site of the genes and the cis-acting elements(CREs) in the putative promoter regions were predicted using PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/). The results were visualized by R package and TBtools (Chen et al., 2020a).

3. Results 3.1. Morphological observation of representative woody bamboos

To explore the variation in culm size among woody bamboos with differing ploidal levels and genome diversity, we measured the height of seven representative woody bamboos for a total of 783 mature culms (Fig. 1a). Results showed that D. sinicus was the largest bamboo, followed by P. edulis. Notably, the culm size of bamboos with identical ploidy also varied widely. Among PWBs, the mean height of D. sinicus (28.4 m) and B. amplexicaulis (5.0 m) was in sharp contrast, and this phenomenon was also observed in different tetraploid TWBs (1.4 m in H. calcarea versus 16.6 m in P. edulis), indicating that in addition to ploidy (Sattler et al., 2016), there are more complex regulation mechanisms affecting culm growth and size remained to be explored.

After measuring heights of different woody bamboos, we focused on the hexaploid D. sinicus and B. amplexicaulis, both of PWBs which have consistent subgenomes composition but contrasting in culm size, as well as P. edulis as one of the largest in tetraploid bamboos. To accurately perform the rapid growth transcriptome sampling, we investigated the growth pattern of shoots and internodes in P. edulis as described in the D. sinicus and B. amplexicaulis (Niu et al., 2022; Ma et al., 2024). There was a similar shoot growth pattern of P. edulis with other woody bamboos (Wang et al., 2012a; Tan and Wu, 2018; Xu, 2018). The shoot growth stages for P. edulis followed a similar pattern as in B. amplexicaulis (Niu et al., 2022) and D. sinicus (Ma et al., 2024). Parallel to B. amplexicaulis and D. sinicus, we selected the 10th internode for analysis on P. edulis. When the shoot height is less than 50 cm, the growth rate of the shoot is notably slow, with daily height growth less than 5 cm; this period is referred to as the dormant stage (ST1). Subsequently, the growth rate of the shoot gradually increases with the height of 150 cm and we refer to this period as the initial rapid growth stage (ST2). Following that, both the shoot growth and the internode elongation are further accelerated. When the shoot height reaches 550 cm and the length of the internode is approximately 25 cm, it is defined as the rapid growth stage (ST3). Afterward, the elongation rate of the internode decelerates when the shoot height is about 650 cm; this is the end of rapid growth stage (ST4). Finally, the internode elongation ceases when reaching about 27 cm, termed as the mature stage (ST5).

3.2. Phytohormone changes in internode of Dendrocalamus sinicus during rapid shoot growth

Hormonal regulation was recognized as a key process in plant growth and development, and we quantified content of phytohormones (ABA, IAA, GAs, CK) in the 10th internode of D. sinicus shoots at different growth stages of ST1 to ST4. Measurements of total GAs showed that the content of total GAs in the ST2 stage was higher than other stages (especially ST3 and ST4). Moreover, the content of CK in the ST2 stage was also the highest (Fig. 1b). The content of auxin (IAA) in the ST1 stage was significantly lower than other stages, which was consistent with recent studies in P. edulis (Tao et al., 2018). In contrast to IAA, the ABA content in ST1 was significantly higher than the remaining stages, which is a well-known growth inhibitor by inhibiting cell division, elongation and metabolism (Chen et al., 2020b). These results indicated that cell division and metabolism in internodes become more active in D. sinicus from dormant to rapid growth stages. In short, the phytohormones were closely related to the rapid shoot growth of D. sinicus.

3.3. A key role of GAs in rapid growth of bamboos

GAs are phytohormones that play essential roles in internode elongation through enhancement of cell elongation and, in some cases, cell division (Nagai et al., 2020; Chen et al., 2022). To date, more than 100 kinds of GAs have been identified with GA1 and GA4 being the major bioactive ones in higher plants, and other GAs were non-bioactive and act as precursors of the bioactive forms (Hedden and Thomas, 2012). We thus identified the contents of various GAs separately. GA1, GA6, GA19, GA20, GA44 and GA53 were detected in GA1 pathway, but only GA4, GA34 and GA51 were detected in GA4 pathway (Fig. 1c). Furthermore, the level of GAs in GA1 pathway was higher than those in GA4 pathway, and the content of bioactive GA1 in the ST2 was significantly higher than other stages. This indicated that GA1 may be more important than GA4 in internode elongation during rapid shoot growth of D. sinicus.

To further investigate which phytohormone (IAA, GAs, CK, and ABA) was more closely related to the rapid shoot growth of D. sinicus, we tested correlations between growth rate and level of endogenous phytohormones in internodes (Fig. 1d). Growth rate showed significant correlations with ABA, GA19 and GA51. Furthermore, the total GAs, GA19, GA20 and GA51 were all significantly correlated with the ABA, as also observed between CK and the total GAs, GA19, GA53. Additionally, the correlation between IAA and ABA was significant. Our results confirmed that endogenous phytohormones and GAs in particular were closely correlated with internode elongation of D. sinicus.

Meanwhile, we calculated and compared the expression proportion of GAs-related DEGs in total GAs-related genes in representative woody bamboos with different culm sizes. The results showed that GAs-related DEGs were more highly expressed in the large-sized woody bamboos than the small-sized ones (Figs. 1a, f and S2). In addition, a significant positive correlation was found between the expression proportion of GA-related DEGs expression levels in total GAs-related genes and culm size of woody bamboos (Fig. 1f). These results further indicated that GAs was crucial in the rapid shoot growth and culm size regulation of woody bamboos.

3.4. Transcriptome analyses in woody and herbaceous bamboos

We further identified potential candidate genes of rapid shoot growth by WGCNA in D. sinicus, as well as in P. edulis (Fig. 1e) and B. amplexicaulis. To compare with an herbaceous bamboo (HB), we also performed transcriptome sequencing and analyses using data acquired from Ra. guianensis at different developmental stages (Fig. S1).

By estimating the module-trait relationships using the correlation between MEs and traits, modules that were significantly associated with growth rate, content of endogenous phytohormones in rapid growing shoots were identified in D. sinicus. The yellow module, comprising 1226 DEGs, exhibited a significant positive correlation with the growth rate, total GAs, GA6, GA44, GA51, and GA53 (Figs. 2a and S3). It was suggested that this module may regulate rapid shoot growth through gibberellins driving other factors. KEGG and GO enrichment analyses implicated many DEGs in this module involved in various pivotal biological processes during plant development (Figs. 2b and S4). In P. edulis, a total of 12, 589 genes were identified as DEGs, which clustered into 16 modules ranging in size from 12 to 4581 genes by WGCNA (Fig. S5), we found that the blue module was significantly associated with growth rate. Furthermore, we discovered a positive correlation between the trend of gene expressions within this module and the growth rate (Fig. S6). Therefore, we speculated that these DEGs in blue module may positively regulate rapid shoot growth in P. edulis. Similar to P. edulis, a total of 10, 174 genes were identified as DEGs, which clustered into 26 modules ranging in size from 24 to 2176 genes by WGCNA during rapid shoot growth in B. amplexicaulis (Figs. S7 and S8), the red module was most significantly associated with growth rate.

We conducted an in-depth comparison of the interaction network among key genes involved in key processes across three growth-related modules in three woody bamboos (Fig. 2c–e). In general, the number of key genes identified in the yellow module of D. sinicus and the blue module of P. edulis were similar, with 52 and 59, respectively. However, the red module from small-sized bamboo B. amplexicaulis displayed lesser number of key genes (38). Based on the above results, it seemed that the biosynthesis of GAs, IAA and lignin was essential during rapid shoot growth of three woody bamboos. And the metabolic processes shared by two large-sized woody bamboos (D. sinicus and P. edulis) also included transcription, cellulose and sugar metabolism. This suggested that the biosynthesis of cellulose and sugar were vital for rapid shoot growth of the woody bamboos. However, the metabolic processes in small-sized bamboo B. amplexicaulis was simpler than those in other two large-sized bamboos. We further analyzed the genes involved in metabolic pathways commonly found in three woody bamboos and thus identified the SLRL1 gene shared by all three bamboos, which plays an important role in GAs signal transduction in model plants (Bao et al., 2020).

To identify which genes contributed to culms growth of herbaceous bamboos, we performed transcriptomic analysis of Ra. guianensis at three different developmental stages defined here (Fig. S1). In the initial stage (S1), the shoots were young with culm heights of approximating 2.5 cm. Subsequently, at S2, the shoots grew to 15 cm and continued to grow. By the final stage (S3), the culms were mature and elongation stopped while culms grew up to 29 cm. A total of 2692 DEGs were identified, KEGG and GO enrichment analysis showed that 111 DEGs were assigned to key biological processes (Fig. S9; Table S1). Nevertheless, no gene was enriched in lignin biosynthesis despite there were 34 genes involved in hormone pathways, making the largest proportion among the enriched DEGs (Fig. 2f). Remarkably, by comparing the transcriptomic results with woody bamboos, we found that KAO homologous to woody bamboos was also DEGs in the herbaceous Ra. guianensis. The results indicated that the KAO genes are essential for the biosynthesis of GAs in both herbaceous and woody bamboos, which is consistent with previous studies in other plants (Binenbaum et al., 2018). Interestingly, the SLRL1 gene were not differentially expressed in culm development of herbaceous bamboo, it is thus inferred that SLRL1 gene may be vital for rapid shoot growth and would have contributed to the origin of the innovative trait in woody bamboos.

Taken together, we identified two candidate genes associated with the rapid shoot growth of woody bamboos. The KAO and SLRL1 might be responsible for GAs biosynthesis and signal transduction in bamboos, respectively.

3.5. Expression divergence of GAs-Related genes among subgenomes

We further identified the homoeologs of GA-related genes in 11 bamboos and investigated their expression patterns in young shoots. As expected, the largest number of GAs-related genes were identified in three hexaploid PWBs, followed by six tetraploid woody bamboos, and the least in two herbaceous bamboos (Fig. 3e; Table S2). More gene copies may provide a genetic basis for complex biological processes in woody bamboos compared with herbaceous bamboos.

We next traced homoeolog expression patterns for GAs-related DEGs in D. sinicus, B. amplexicaulis and P. edulis during rapid shoot growth. In D. sinicus, 30 DEGs were identified (Fig. 3a). These genes encode major protein families known to be involved in GAs biosynthesis and signal transduction, and their expression levels varied substantially at different developmental stages (ST1, ST2, ST3 and ST4) and among the subgenomes A, B and C. Overall, we found the total gene expression of DEGs was dominated by the subgenomes A and C, except for ent-kaurene synthase (OsKS1). In P. edulis, we detected 22 GAs-related DEGs (Fig. 3b). They varied substantially both in the number and expression levels between subgenomes C and D. There were 15 copies of DEGs on subgenome C, in sharp contrast to only seven copies on subgenome D. Furthermore, the expression levels of most DEGs on subgenome C were higher than homoeolog DEGs on D, indicating the dominance of subgenome C during rapid shoot growth. In B. amplexicaulis, we identified 16 copies of GAs-related DEGs (Fig. 3c). Unlike the observations in D. sinicus, the expression levels of DEGs on the subgenome B were higher in B. amplexicaulis, and it could be called the subgenome B dominance here.

Enrichment analysis of GAs-related DEGs expression to subgenome showed that in two hexaploid PWBs, DEGs with high expression levels were significantly enriched on subgenome A in D. sinicus while instead on subgenome B in B. amplexicaulis. In tetraploid TWB P. edulis, GAs-related DEGs on subgenome C was preferentially expressed (Table S3). These results reflecting the greater contribution of individual gene copies on subgenomes A and C than B and D to rapid shoot growth of D. sinicus and P. edulis, respectively. However, in small-sized B. amplexicaulis (Fig. 1a), GAs-related genes on subgenome B may be dominant over both subgenomes A and C.

We examined the expression patterns of homoeolog GAs-related DEGs among different subgenomes in individual bamboo species using young shoots of eight woody bamboos of different culm size. In TWBs, the highly expressed homoeolog copies were on subgenome C in P. edulis, which is in accordance with expression pattern during rapid shoot growth, while the highly expressed homoeolog copies in Am. luodianensis and H. calcarea were on subgenome D. In PWBs, the expression patterns in B. amplexicaulis and D. sinicus were consistent with observations during rapid shoot growth, and gene copies on subgenome A were dominant over both subgenomes B and C in M. baccifera shoots. The expression levels of gene copies on subgenome C were higher than subgenome B in two tetraploid NWBs (Fig. 3d; Table S4).

Enrichment analysis of GAs-related DEGs expression with respect to subgenome in young shoots suggested that these results were significant and solid (Table S3). In all, GAs-related genes on the subgenomes A and C were predominant over B or D in large-sized species, whereas the trend is converse in small-sized woody bamboos.

3.6. Subgenome-specific gene structure

To explore whether the gene structure had an effect on the transcriptional expression of the genes from different subgenomes, we identified all KAO and SLRL1 gene members in 11 bamboos.

A total 20 KAO genes were identified in 11 bamboos genomes using rice KAO sequences as queries, with one gene on each subgenome except for subgenome C of Am. luodianensis, subgenome B of Rh. racemiflorum and B. amplexicaulis (Figs. 5a and S10a). KAO genes were located in a conserved syntenic region of chr6 between rice and bamboo subgenomes. However, an inter-chromosomal translocation of a segment within KAO located occurred between chrs 3 and 6 of subgenome C in six NWB and PWB species (Fig. S10a). To explore the evolutionary relationship of the KAO genes in bamboos, a phylogenetic tree was constructed and 20 copies were clearly grouped based on subgenomes A, B, C, D, and H. Notably, the total expression of KAO in shoots was all dominated by the C copy in woody bamboos, except for D. sinicus (Fig. 4a). And the expression of genes on subgenomes B or D was suppressed in all woody bamboos.

As UTR-exon-intron distribution patterns, motifs and domains of KAO genes were very conserved (Fig. 5), the sequences of their DNA and protein are too similar to have effect on homoeolog expression divergence between subgenomes. However, UTR was variable among these 20 genes. In detail, almost all of genes on subgenomes A, C and H contained UTRs, but all genes on the subgenomes B and D did not have (Fig. 5). Remarkably, this pattern was consistent with the expression trend of the KAO genes in woody bamboo shoots. These results demonstrate that differences of UTRs may result in homoeolog expression divergence among subgenomes, which is the basis for the functional divergence of genes in bamboos, providing a subgenome-specific gene structure for studying the unique traits in woody bamboos.

A total 22 SLRL1 genes were identified in 11 bamboos genomes, with one gene on each subgenome except for subgenome B of Rh. racemiflorum (Figs. 6a and S10b). The SLRL1 genes were located in a well-conserved syntenic region of chr1 between rice and bamboo subgenomes, and the 22 gene copies were also clustered based on subgenomes. Interestingly, the total expression of SLRL1 in shoots was generally dominated by A and C copies in PWBs, and all dominated by C in NWBs and TWBs. The expression of genes on subgenomes B and D was suppressed in all woody bamboos (Fig. 4b).

Investigation of gene structure and proteins uncovered that UTR-exon-intron distribution patterns, motifs and domains of SLRL1 genes were conserved among subgenomes (Figs. S11 and S12). We thus further analyzed the promoter region of these genes and identified numerous CREs distributed in the promoter regions, which could be classified into nine fundamental functional categories (Fig. 6d). Furthermore, the number of these CREs also varies among gene copies, particularly for the core promoter, which represents a high proportion in CREs in most SLRL1 genes. In individual species, the numbers of core promoter of genes from different subgenomes were positively correlated with the transcriptional expression level (Fig. 6d and Table S5). For example, genes on subgenomes A and C have more core promoters while the gene expression is higher in M. baccifera. In P. edulis and Am. luodianensis, the number of the core promoters on subgenome D was low, so as to the gene expression. We also found that core promoters were more distributed in non-conserved regions among different genes across rice and bamboos. These results implied that the differences (in number and position) of the core promoter may contribute to the gene expression, resulting in homoeolog expression divergence among subgenomes.

4. Discussion

Bamboos have experienced multiple rounds of intricate hybridizations and polyploidizations, resulting in a range of ploidy levels and subgenome compositions (Ma et al., 2024). The significance of subgenome dominance in plants has been increasingly recognized and believed to have adaptive plasticity, thereby shaping the evolution and diversifications of plants (Alger and Edger, 2020). This plasticity has facilitated the domestication and adaptation of several major domesticated crops, for example, hexaploid bread wheat (Triticum aestivum) (Pfeifer et al., 2014; Wang et al., 2016; Ramirez-Gonzalez et al., 2018), allotetraploid cotton (Go. hirsutum) (Wang et al., 2017; Bao et al., 2019; Dong et al., 2022; You et al., 2023), and the resynthesized allotetraploid Brassica napus (Bird et al., 2021). Our previous work on genome assemblies across 11 bamboo species revealed an empirical case that large scale species diversification of woody bamboos could be induced by dynamic subgenome dominance (Ma et al., 2024).

The rapid shoot growth of woody bamboos was a key innovative and distinguishing trait within the grass family. Recent studies found that monthly precipitation may regulate the internode size by affecting the primary thickening growth of P. edulis (Zhang et al., 2024), and the 'PeSAPK4-PeMYB99-PeTIP4-3' regulatory model might be involved in water transport during its shoot growth (Zhu et al., 2024). Genome-wide analysis in this bamboo identified the KNOX (KNOTTED1-like homeobox), GRF (Growth-regulating factors) and aspartic protease gene families in promoting the rapid shoot growth (Jiao et al., 2024; Wang et al., 2021c; Zhou et al., 2023). And analyses of 77 woody bamboos found that the heritable legacy dominates development of culms functional traits (Liu et al., 2024b). Our study suggests that the genetic basis of adaptation and subsequent diversification in bamboos would further benefit from in-depth investigation using multiple bamboo species with different ploidal levels and culm sizes.

4.1. Importance of GAs in rapid growth and culm size of bamboos

By summarizing a large number of previous studies on the growth patterns of bamboo culms, we found that woody bamboos share a similar growth pattern, and large bamboos, particularly the tetraploid P. edulis, usually grow faster than small ones during the rapid growth stages (Wang et al., 2012a; Huang et al., 2016; Tan and Wu, 2018; Xu, 2018; Chen et al., 2022; Niu et al., 2022).

Quantification of the phytohormones in D. sinicus showed that the changes in content of GAs and CK had similar trends with P. edulis during the rapid growth with the highest level found in the ST2. The ST2 was the initiation period of rapid internode elongation, which might undergo vigorous cell division and metabolisms for synthesis of materials synchronously (Niu et al., 2022; Ma et al., 2024). Furthermore, the level of GAs was higher in the internode cell division zone with a relatively higher cell division intensity, closely correlated with the CK content in the internode of P. edulis during the rapid growth (Chen et al., 2022). So, we hypothesized that GAs and CK also cooperated with each other to regulate cell division and metabolisms of internode in D. sinicus.

Notably, GAs was significantly correlated with other important phytohormones (Fig. 1d). One possible explanation for these observations might be that GAs was the hub phytohormone that can interact with other phytohormones to regulate rapid shoot growth in D. sinicus, as observed in P. edulis (Tao et al., 2018; Chen et al., 2022). The proportion of GAs-related DEGs expression levels in representative woody bamboos with different culm sizes also indicated that GAs was crucial in the rapid shoot growth and size regulation of bamboos (Fig. 1a and f). Indeed, GAs can antagonistic crosstalk with ABA by distinct TF regulators and the direct molecular interaction between core signaling components, thereby regulating plant growth and development (Liu and Hou, 2018). GAs can also crosstalk with other phytohormones through DELLA protein, which together regulate various developmental processes in plants (Daviere and Achard, 2016; Bao et al., 2020).

Analyses of various kinds of GAs found that there were two pathways to synthesize bioactive GAs in D. sinicus (GA1 and GA4 pathway) (Fig. 1c), which is consistent with previous study in P. edulis (Gamuyao et al., 2017). Meanwhile, some non-bioactive GAs would also be generated (Hedden and Thomas, 2012). By comparing various GAs contents on these two pathways, we found that the type of GAs that actually regulates rapid shoot growth in D. sinicus might be GA1. On the contrary, previous studies suggested that GA4 may be more important in P. edulis (Chen et al., 2022).

By integrating transcriptome analysis results of rapid shoot growth from three woody bamboos (D. sinicus, P. edulis and B. amplexicaulis), we identified a candidate gene SLRL1 that were associated with the rapid growth in all of them. SLRL1 gene encodes a DELLA protein that belongs to a subfamily of the GRAS protein superfamily and that functions as a repressor of GA signaling (Fukao and Bailey-Serres, 2008), DELLA proteins were the master negative regulators of GA signaling, and can integrate multiple hormones signaling pathways, such as cross-talking of GAs with IAA, CK, BR, ABA and JA, and thus are involved in multiple biological processes in plants (Itoh et al., 2005; Claeys et al., 2014; Daviere and Achard, 2016; Bao et al., 2020). Three DELLA genes, SLR1, SLRL1 and SLRL2, were identified in rice (Itoh et al., 2005; Thomas et al., 2016). Interestingly, SLRL1 and SLRL2 lack the N-terminal DELLA and TVHYNP domains, and the existence of SLRL-type GRAS proteins appears to be restricted to monocots (Hauvermale et al., 2012). The SLRL1 was found to be expressed after submergence in deepwater rice, but not for SLR1 (Fukao and Bailey-Serres, 2008) and it is inferred that SLRL1 may be more sensitive to environmental stimulus than SLR1 (Minami et al., 2018). Consistent with the findings in rice, we identified SLR1, SLRL1 and SLRL2 genes in the genomes of 11 representative bamboo species (Fig. S12). Through transcriptome data from rapid growing shoots of three woody bamboos, we further found that the SLRL1 gene was shared by all three woody bamboos during rapid growth, but not differentially expressed in culm development of herbaceous bamboos. The SLRL1 gene was thus more important for rapid growth and may contribute to the origin of the innovative trait in woody bamboos.

In D. sinicus and P. edulis, SLRL1 gene might interact with IAA, BR, JAZ, ABA and ethylene by transcription factor, then regulating the lignin, cellulose, sugar and formin metabolism. However, in B. amplexicaulis, the crosstalk of phytohormones with GAs via the SLRL1 gene was reduced, with only IAA identified. Simultaneously, the biosynthesis processes were reduced, only one lignin-associated gene and a flavonoid-associated gene were identified (Fig. 2d). The lignin content in the mature culms of B. amplexicaulis was significantly lower than that in D. sinicus and P. edulis (Ma et al., 2024), and its size was significantly smaller than D. sinicus and P. edulis, too (Fig. 1a). It indicates that the larger the bamboo, the more complex the biological processes were involved during rapid growth.

By comparing the transcriptomic results of the largest woody bamboo (D. sinicus) and a small herbaceous bamboo (Ra. guianensis), we identified the KAO gene essential for the biosynthesis of GAs in both species. KAO genes can regulate plant growth, in Ar. thaliana, the level of GAs and plant size in kao mutants were substantially reduced compared with the wild-type (Regnault et al., 2014). Lesions in KAO have been shown to be the cause of severe dwarfism in several model species, including rice (d35 mutant) (Sakamoto et al., 2004), maize (dwarf3) (Helliwell et al., 2001), pea (na) (Davidson et al., 2003) and sunflower (dwarf2) (Fambrini et al., 2011).

4.2. Importance of dominant subgenome in rapid growth of woody bamboos

In our previous work, it was revealed that subgenome dominance could be established after polyploidization and inherited by their descendants, thereby contributing to the evolution of unique traits across woody bamboos (Ma et al., 2024). In this study, by analyzing the transcriptome of young shoots from eight representative woody bamboos, we investigated homoeolog expression patterns for GAs-related DEGs during rapid shoot growth in the woody bamboos. Generally, GAs-related genes on subgenomes A and C were predominant over subgenomes B and D in large-sized woody bamboos, conversely in small-sized ones. It was revealed that subgenomes C and A were dominant subgenomes in woody bamboos (Ma et al., 2024). Our findings indicate that dominant subgenomes positively regulates rapid shoot growth and the submissive subgenome also play important roles in regulation of culm size in woody bamboos. Similarly, submissive subgenomes were crucial in adaptation of radiated polyploid lineages in Nepenthes (Saul et al., 2023). In fact, the phenomenon that the patterns of transcriptome asymmetry were associated with their morphological traits were observed in tetraploid wheats (Wang et al., 2016) and cotton (You et al., 2023). The subgenomic or transcriptomic functional asymmetry has been shown to play critical roles in allopolyploid speciation and adaptation to the environments (Wang et al., 2021b). Specially, the patterns and mechanisms of homeolog biased expression pattern related to rapid growth and culm sizes in woody bamboos are not yet fully understood, calling for further investigation.

4.3. Genetic basis driving species diversification of woody bamboos

By analyzing the homoeolog expression pattern, gene structure, and conserved non-coding regulatory of key genes in 11 bamboos, we found that the homoeolog expression divergence among subgenomes in KAO genes may be closely related to UTRs. UTRs can play a crucial role in post-transcriptional control by regulating various mRNA properties like stability, transport, translation efficiency and functioning and subcellular localization of the translated proteins (Srivastava et al., 2018). The variation in UTRs of KAO genes on different subgenomes in woody bamboos may regulate the expression level by affecting properties of mRNAs.

The homoeolog expression divergence among subgenomes of SLRL1 genes may be related to the core promoter in 2000 bp upstream of the gene body. Within the same bamboo species, the number of core promoters was positively correlated with the transcriptional expression level. A recent study in cotton suggests that diversity in the transcriptional regulation region for homoeolog genes may result in homoeolog expression divergence (You et al., 2023). Transcriptional regulation is mainly mediated by transcription factors interacting with cis-regulatory elements like core promoters (Shafee and Lowe, 2017; Bottani et al., 2018).

Our results from homoeolog expression patterns of KAO and SLRL1 genes revealed that gene copies which exhibited subgenome-biased expression in woody bamboos tend to have subgenome-specific cis-regulatory sequences, similar to the results in cotton (Wang et al., 2017; You et al., 2023). This phenomenon will help us to understand the molecular basis for species diversification in woody bamboos following polyploidizations.

5. Conclusion

In this study, we found GAs may regulate rapid shoot growth and thus the mature culm size of woody bamboos, and GA1 could be the main type of bioactive in the largest bamboo D. sinicus, a representative of the PWB clade (Liu et al., 2024a). Meanwhile, we also identified both KAO and SLRL1 involved in the GAs biosynthesis as important candidate genes for regulating the rapid shoot growth. By comparing the genomic and transcriptomic data of 11 representative bamboos, we found that expression level of GAs-related genes at subgenomes A and C were predominant over subgenomes B and D in rapid shoot growth of large-sized woody bamboos. This subgenome asymmetry in woody bamboos may be closely related to their adaptation and species diversification after multiple rounds of hybridizations and polyploidizations. Furthermore, we found intriguing correlations between subgenome asymmetry and subgenome-specific gene structure, e.g., UTRs and core promoters (Fig. 7). These findings provide new insights into the genetic mechanism of innovative traits regulation in polyploid woody bamboos and the subsequent species diversification.

Acknowledgements

We thank W. Dai and K.-Y. Liu for help in the experiments. This work was supported by the Key Project of Natural Science Foundation of Yunnan Province (202401AS070082) and the National Natural Science Foundation of China (grants 32120103003 and 31970355), and facilitated by the Germplasm Bank of Wild Species.

CRediT authorship contribution statement

Ling Mao: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Cen Guo: Software, Methodology, Investigation, Data curation. Liang-Zhong Niu: Methodology, Investigation, Formal analysis. Yu-Jiao Wang: Software, Methodology, Investigation. Guihua Jin: Methodology, Investigation. Yi-Zhou Yang: Software, Investigation. Ke-Cheng Qian: Software, Investigation. Yang Yang: Investigation. Xuemei Zhang: Investigation. Peng-Fei Ma: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing. De-Zhu Li: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Methodology, Funding acquisition, Conceptualization. Zhen-Hua Guo: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Funding acquisition, Conceptualization.

Data accessibility

RNA-seq data are available at NCBI (accession: PRJNA948693).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2024.10.004.