3. RESULTS

3.1 Accelerated flowering and increased NFFs in LR treatment

Plants treated with SW treatment flowered at 31 days after terminal spikelet stage (DAT) with an average of 78 NFFs; LW treatment accelerated flowering by four days but significantly reduced NFFs as compared to SW treatment; surprisingly, LR treatment identified earlier flowering by three days and significantly increased NFFs by 14.0% when compared to LW treatment (Figure 2a, 2b). Therefore, LR treatment was an effective light regimen for accelerating flowering and increasing NFFs simultaneously. Furthermore, there was no difference between treatments in the number of spikelets per spike (Figure 2c), and the difference in NFFs was caused only by fertile florets (Figure 2b). So, we focused on investigating the growth and development dynamics of floret primordia.
Spike samples with 20 spikelets were chosen to eliminate the interference of spikelet differences on floret development (Figure S3a). Floret primordia (FP) within spikelets were numbered from F1 to Fn, from the closest to the most distal positions with respect to the spike rachis, respectively (Figure 2d). At the preceding times (3 to 12 DAT), the number of total floret primordia per spike (NTFPs) increased rapidly at a rate of 4.6 FP, 5.4 FP, and 5.5 FP per day in SW, LW, and LR treatment, respectively (Figure 2e), suggesting a faster differentiation rate in LW and LR treatments. From 12 to 18 DAT, NTFPs of LR treatment remained largely unchanged, while NTFPs of SW and LW treatments continued to increase at a rate of 3.1 FP and 2.1 FP per day, respectively. At 18 DAT, NTFPs in three treatments was similar. During the early stage, floret primordia in the spike underwent a differentiation process characterized by an increase in NTFPs (including a rapid increase in the first 12 days and a slow increase in the last 6 days). Of note, LR treatment accelerated and concentrated this process in the first 12 days, shortening the time required for primordia differentiation.
As the number of primordia grew, preferentially initiated primordia began to develop floral organs, reaching the scale of fertile floret primordia (Figure S1b). Before 12 DAT, the number of fertile floret primordia per spike (NFFPs) increased at a rate of 3.2 FP, 4.3 FP, and 4.6 FP per day in SW, LW, and LR treatment, respectively (Figure 2f), indicating a faster morphology development in LW and LR treatments. From 12 to 18 DAT, NFFPs increased at a rate of 3.3 FP and 2.8 FP per day in SW and LW treatment, respectively, which was similar to the increase rate of NTFPs; in addition, NFFPs in LR treatment increased faster, at a rate of 3.9 FP per day (Figure 2f). Subsequently, from 18 to 21 DAT, when NTFPs approached the maximum (~ 200), NFFPs increased at a rate of 3.9 FP per day in SW treatment but did not increase in LW or LR treatment (Figure 2e, 2f). At 24 DAT, NFFPs of SW and LR was significantly higher than that of LW (Figure 2f), with SW treatment mainly increasing fertile floret primordia located at the central and apical portions of the spike and LR treatment mainly increasing fertile floret primordia located at the basal and central portions of the spike (Figure S3b). During the middle stage, floret primordia in the spike underwent a morphological differentiation and development process characterized by an increase in NFFPs (including a simultaneous increase of NFFPs with NTFPs and a continued increase of NFFPs after maximal NTFPs).
At later times (21 DAT of LR and LW and 24 DAT of SW), some floret primordia that did not reach W7.5 began to degenerate and eventually died at or after flowering, earning the name infertile floret primordia (Figure S1b). The number of infertile floret primordia per spike (NIFPs) decreased sharply at an average rate of 10.0 FP per day (Figure 2g), but it had no effect on NFFPs regardless of treatments (Figure 2f), suggesting the certainty of fertile floret primordia for future fertile fate. During the last stage, floret primordia in the spike underwent a polarization process, with two morphologically distinct floret primordia (fertile and infertile floret primordia) proceeding independently toward opposing fates (fertility and infertility at flowering).
Together, changes in the number and morphology of floret primordia occurred along with floret development in wheat, and light regimens had a significant effect. LW treatment might advance the initiation of degeneration and shorten the duration of differentiation and morphological development of floret primordia, leading to early flowering and decreased NFFs, whereas LR treatment greatly accelerated the differentiation and morphological development of floret primordia in a shortened duration, ensuring NFFs.

3.2 Assimilate accumulation in the wheat spike for floret development

Assimilate supply to the spike affects the fate of floret primordia and NFFs (González et al. , 2011; Shen et al. , 2018), but its relationship with the growth and development of floret primordia under different light regimens is unclear. Therefore, we investigated the dynamics of assimilate accumulation in wheat plants during floret development (Figure 3). Compared with SW, LW treatment significantly increased dry plant weight, dry spike weight, and dry matter partitioning to spike from 12 to 24 DAT. At 27 and 30 DAT, dry matter accumulation in the plant and spike in SW treatment were comparable to those in LW treatment. Furthermore, compared with LW, LR treatment significantly increased dry spike weight and dry matter partitioning to spike from 9 to 30 DAT. The dry spike weight at flowering was 0.50g, 0.45g, and 0.47g in SW, LW, and LR treatment, respectively. In terms of soluble sugar content, compared with SW, LW treatment significantly increased glucose and fructose content from 12 to 15 DAT in the spike. Compared with LW, LR treatment further significantly increased soluble sugar content from 9 to 12 DAT and significantly increased glucose and fructose content from 9 to 15 DAT in the spike. Collectively, although LW treatment increased assimilate accumulation and sugar content prior to floret primordia degeneration, it was inadequate compared to LR treatment in enhancing the growth and development of more floret primordia.

3.3 Transcriptome atlas along floret development in wheat

To investigate the gene activity dynamics of all floret primordia in a spike, we performed RNA-seq libraries of developing wheat spikes from the terminal spikelet stage to flowering stage, with an interval of six days (3 to 15 DAT) or three days (15 to 27 DAT; Figure 4a). Some biological replicates, including the replicate 1 at 24 DAT of SW (i.e., SW24.1), SW27.2, LW3.1, LR15.2, and LR27.2, were filtered out because they didn’t get together with the other two biological replicates (data not shown). PCA (Figure 4b) and hierarchical clustering (Figure 4c) were performed to identify the similarity of gene expression at different times. In SW treatment, the first group consisted of samples from the earliest 3 DAT and 9 DAT, when the increase rate of NTFPs was 43.7% greater than that of NFFPs, and was thus identified as the differentiation stage (Stage I, Figure 4d); the second group included samples from 15 to 18 DAT, which represented the stage of differentiation and morphology development concurrently when the increase rate of NFFPs was comparable to that of NTFPs (Stage II, Figure 4d); samples at 21 DAT fell into the third group, when floret primordia had reached its quantitative maximum and the focus was on the morphological development of floret primordia (Stage III, Figure 4d); the fourth group consisted of samples from 24 to 27 DAT and corresponded to the polarization process (Stage IV, Figure 4d). A comprehensive cluster analysis of all transcriptomes from three treatments showed that samples of LW and LR at 3 DAT and 9 DAT clustered with samples of SW in Stage I; samples of LW at 15 DAT and 18 DAT corresponded to Stage II; samples of LR from 15 DAT and 18 DAT clustered with samples of SW at 21 DAT, corresponding to Stage III; and samples of LW and LR at 21 DAT, 24 DAT, and 27 DAT were classified into Stage IV (Figure 4e). LW and LR treatment, respectively, greatly accelerated the growth and development of floret primordia in Stage III and Stage II so that they were not visible at our sampling interval.
Together, these results confirmed that the transcriptome atlas successfully captured the characteristics of floret development and objectively divided it into four distinct stages corresponding to floret primordia differentiation (Stage I), differentiation and morphology development concurrently (Stage II), morphology development (Stage III), and polarization (Stage IV).

3.4 LW versus SW—Altered transcriptional signatures induced early flowering

To identify the regulatory mechanism responsible for the observed early flowering phenotype, we investigated transcriptional differences between LW and SW and discovered 14035 DEGs (Dataset S1). The number of DEGs and specific DEGs (expressed at a single sampling time) increased sequentially in the first three stages, with 21 DAT having the most DEGs (more than 60% of total DEGs) (Figure 5a, 5b).
According to the conserved GI -CO -FT (Fornara et al. , 2010), we screened 19 key genes or homologs involved in the photoperiodic flowering pathway (Figure 5c, Dataset S1) and analyzed their relationships with 2370 functional DEGs involved in flower development, light response, hormone metabolism, and carbohydrate metabolism (Figure S4-S8). In Stage I, compared to SW, three GIgenes were significantly up-regulated by LW treatment (Figure 5c). ThreeIAA16 , which act as repressors in the auxin signaling pathway (Liscum & Reed, 2002), were found to interact directly with GI(Figure 6d) and were significantly inhibited by long photoperiod from Stage I to II (Figure S9a). As a result, IAA content in LW treatment was significantly higher than that in SW treatment (Figure S9b).
Then, three CO3 genes were up-regulated by an average of 6.4-fold from 18 to 21 DAT; three COL2 , three COL5 , and threeGRAIN NUMBER PLANT HEIGHT AND HEADING DATE 7 (GHD7 ) were up-regulated by 2.3~8.9-fold at 21 DAT; also, aCOL3 gene, which functions as a floral repressor (Datta et al. , 2006), was found to be significantly down-regulated at 21 DAT (Figure 5c). Meanwhile, 102 functional DEGs, including numerous MYB transcription factors, were predicted to interact directly withCO -like. These functional DEGs were mainly involved in jasmonic acid, salicylic acid, and auxin responses (e.g., Zm38 ,MYB1 , MYB61 , MYB308 , Zm1 , MYB4 , andMYB16 ; Li, X et al. , 2021; Zhou et al. , 2022) and anther wall tapetum development (e.g., MYB80 and MYB35 ; Zhu et al. , 2008; Phan et al. , 2011) (Figure 5d, 5e). In addition, other genes, such as CRYD , encoding light-sensing cryptochrome DASH in chloroplastic/mitochondrial (Castrillo et al. , 2013), and SWEET2B , a member of the bidirectional SWEET sugar transporter family (Ji et al. , 2022), also cooperated with CO -like genes to accelerate flower development under LW treatment (Figure 5d).
In Stage IV, three FT homologous genes, HEADING DATE 3A(HD3A ), were up-regulated by an average of 3.5-fold in LW treatment compared to SW treatment (Figure 5c). Functional DEGs directly associated with HD3A were significantly enriched in pollen tube growth (such as At2g41970 and ANX1 ; Miyazaki et al. , 2009; Boisson-Dernier et al. , 2015) and abscisic acid activated signaling pathway (such as MPK16 and MPK17 ; Goyal et al. , 2018) (Figure 5d, 5f). Also, PRR1, a component that coordinates the circadian oscillator and the long photoperiod (Boss, 2004), was found to interact directly withHD3A and be up-regulated by LW treatment (Figure 5d).
Overall, stepwise regulation of GI , CO -like, andHD3A and their directly related functional DEGs during floret development might be responsible for early flowering.

3.5 LR versus LW—Altered transcriptional signatures promoted flowering and yield

Despite having similar GI and some CO -like expression in treatments with the same long photoperiod, LR treatment had significantly higher HD3A expression than LW treatment, implying the existence of another early-flowering pathway in modulated light quality (Figure S9c). Since we were interested in identifying the underlying causes of the simultaneous flowering acceleration and NFFs increase in LR treatment, we focused on the transcriptional differences between LR and LW (Dataset S2). The numbers of DEGs were 382, 152, 289, 1875, and 1497 at 3, 9, 21, 24, and 27 DAT, respectively, with 15 DAT and 18 DAT having far more DEGs (8558 and 3641, respectively) (Figure 6a). Similarly, there were 5657 DEGs specifically expressed at 15 DAT, which was a lot more than at other sampling times (Figure 6b).
K -means clustering of 12382 DEGs between LR and LW defined seven coexpression modules (Figure 6c). In the results, 293 DEGs in Module i were significantly up-regulated at 3 DAT and were enriched in the hormone signaling pathways (ethylene and salicylic acid) and phloem or xylem histogenesis (Figure 6d, 6e). Of these, 226 DEGs were highly expressed in Stage I (genes with expression levels in one stage that were 1-fold or higher than those in the remaining stages, Figure S10a), with the top five having the most connections being UBQ11 ,ubqB , KIC , CML16 , and ZAT18 (Figure 7, Table S3). UBQ4 is a key regulator of flower responses to developmental and environmental cues in Arabidopsis (Sun & Callis, 1997); similarly, our results suggested that UBQ11 and ubqB might be active regulators of wheat spikes in response to LR treatment. Many calcium (Ca2+) sensors play important roles in ethylene response, flowering, and pollen development (Reddy et al. , 2004; Ding et al. , 2018), including CM16 and KIC, which occupied the hub nodes in our study. In addition, ZAT18 , a zinc finger transcription factor gene, affects floral organ morphology (Yin et al. , 2017). These results suggested that LR treatment accelerated the environmental adaptation and differentiation rate of floret primordia in Stage I by regulating a gene network centered on ubiquitin, calcium signaling, and zinc finger protein.
Afterwards, 3458 DEGs in Module ii were significantly up-regulated only at 15 DAT, 2818 DEGs in Module iii were significantly up-regulated from 15 to 18 DAT, and 3174 DEGs in Module iv were significantly down-regulated at 15 DAT (Figure 6d, 6e). These up-regulated DEGs were responsible for light response and photosynthesis, stamen morphogenesis, pollen exine formation, and sporopollenin biosynthesis, as well as phenylpropanoid, carbohydrate, and flavonoid metabolic processes; in contrast, these down-regulated DEGs were involved in nucleosome assembly, microtubule-based movement, chromosome organization, regulation of cell cycle, and DNA replication initiation, suggesting that new floret primordia were no longer or infrequently differentiated in LR-treated spikes at this time. Among the total DEGs, 1847 DEGs were highly expressed in Stage III (Figure S10a), and the top five DEGs with the most connections were RPS27AA , RPL40A , ALDH2B7 ,ALDH2C4 , and ALDH3I1 (Figure 7, Table S3). RPS27AAand RPL40A , which encode ubiquitin-ribosomal proteins, were predicted to interact directly with UBQ11 . ALDH family genes affect many aspects of plant growth and development. For example, the homolog of ALDH2B7 , rf2 , is required for anther development in maize (Liu et al. , 2001); the homolog ofALDH2C4 , REF1 , is involved in the phenylpropanoid pathway and plays a major role in cell wall construction and strength in Arabidopsis (Nair et al. , 2004). These findings revealed that, in Stage III of LR treatment, the growth center of the spike had completely shifted from the number increase to morphology development of floret primordia, and that ubiquitin and ALDH genes were the key regulators responsible for enhanced secondary metabolite production, cell wall building, and anther development.
During Stage IV, DEGs in Module v (1322) and Module vi (748) were significantly up-regulated at 24 DAT and 27 DAT, respectively (Figure 6d, 6e). These DEGs were represented by genes related to floral organ development and regulation (including pollen tube growth, stamen filament development, and floral organ abscission), pollination and fertilization (peptidyl-serine phosphorylation, pectin catabolism, cell wall loosening, autophagy, and sterol), and hormone signaling pathways (abscisic acid and jasmonic acid). Conversely, 569 DEGs in Module vii were significantly down-regulated at 27 DAT and were enriched into regulation of cellular respiration, cellular response to red light, and primary metabolism (fatty acid, lipid, and carbohydrate) (Figure 6d, 6e). Among them, there were 1309 highly expressed DEGs in Stage IV (Figure S10a), and the top five DEGs with the most connections wereZF1 and ZF2 encoding zinc finger proteins andHSP70 , HSP90-2A , and HSP90-2D encoding heat shock proteins (Figure 7, Table S3). ZF1 and ZF2 are involved in plant abiotic stress response via abscisic acid and jasmonic acid signaling pathways (Kodaira et al. , 2011). Heat shock proteins also help plants recover from stressful situations (Chaudhary et al. , 2019). In addition, HSP70 and HSP90 are required for pollen maturation and successful fertilization (Margaritopoulou et al. , 2016; Nixon et al. , 2017). Overall, LR treatment regulated a gene network centered on ZF s and HSP s in Stage IV, which coordinated the development and stressful adaptation of fertile floret primordia and maintained the survivability of fertile floret primordia without necessarily upregulating genes involved in carbohydrate metabolism.

3.6 A functional gene network confirming the dual effects of LR treatment

To verify the direct effect of LR treatment on floret development, 122 highly expressed DEGs involved in flower development between LR and LW were investigated (Figure S4, Figure S10a, Dataset S2). The results showed that these genes were enriched in biological processes such as sporopollenin biosynthesis, pollen exine formation, and pollen development (Figure S10b). And, PTC1 , CYP704B1 ,TKPR1 , and 4CLL9 , which control programmed tapetal development and functional pollen formation (Lallemand et al. , 2013; Wang et al. , 2018; Yang et al. , 2019; Peng et al. , 2020), were predicted to be the most important regulators (Figure S10c, Table S4). Among them, PTC1 expression was significantly up-regulated by 3.9-fold at 15 DAT by LR treatment (Figure S10e). Meanwhile,UDT1 , a gene that is expressed upstream of PTC1 (Li et al. , 2011), was found to be significantly up-regulated by 4.8-fold at 9 DAT (Figure S10d). These results suggested that timely and sufficient expression of UDT1 and PTC1 in LR treatment induced rapid and concentrated morphological development of floret primordia, promoting more primordia to acquire fertile potential during Stage III, and thus improving NFFPs and NFFs. Following that,COL5 , a key gene in the above-mentioned photoperiodic flowering pathway, interacted directly with PTC1 and was significantly up-regulated at 24 DAT, contributing to increased HD3A expression and earlier flowering under LR treatment (Figure S10f, S10g).
In addition, in our experiment, the expression levels of all phytochrome members (three PHYA , three PHYB , and three PHYC ) that mediate the effect of red/far-red ratios on FT (Cerdan & Chory, 2003) were not different across three treatments (Figure S11). In three light regimens, red/far-red ratios were much higher than that in sunlight (about 1.0) (Table S2), which were sufficient to activate similar phytochrome levels and inhibit downstream reactions (Gururani et al. , 2015; Galvāo et al. , 2019).