From 2 October 2022 to 28 June 2023, experiments were conducted at Zhejiang Academy of Agriculture Sciences (Latitude: 30.30°N, Longitude: 120.19°E), Hangzhou, China, within an artificial climate chamber. Roman lettuce seeds (Lactuca sativa L., “Ideal-205”; Ideal Agriculture Technology, Nanjing, China) were planted in a small flowerpot (7 cm diameter and 5 cm height) containing substrate soil (substrate: vermiculite: perlite = 3:1:1), and positioned inside an artificial climate chamber. The climate chamber measured 1,130 mm (L) x 795 mm (W) x 1920 mm (H), consisting of three microclimate chambers and a control system that regulated temperature, relative humidity and LED lighting. The growth conditions were set as a photosynthetic photon flux density (PPFD) of 204.41 ± 4.97 μmol m⁻² s⁻¹ at the upper part of the canopy, while maintaining an air temperature of 25 ± 2°C and a relative humidity of 85%.

After the second true leaf had fully unfolded, each seedling was transplanted separately into a small flowerpot with a diameter of 7 cm. The pot contains substrate soil as described above. The seedlings were then exposed to three diel scenarios. The three diel scenarios were as follows: (1) 16 hours of light and 8 hours of darkness (L16/D8); (2) 12 hours of light and 6 hours of darkness (L12/D6); (3) 8 hours of light and 4 hours of darkness (L8/D4). The light intensity for three scenarios was set to 207.88 ± 4.92 μmol m⁻² s⁻¹. The LED light devices were bought from Kesheng Experimental Instrument Co., LTD in Ningbo, China. The spectral radiometer (PLA-30, Everfine Optoelectronic Information Co., LTD., Hangzhou, China) was utilized to measure the photon flux ( Supplementary Figure S1 ). Each scenario was tested three times. As shown in Figure 1 , plants were exposed to three different light treatments for 30 days (720 hours). The control was 16 h light/8 h dark (L16/D8), which resulted in totally 30 L/D cycles in the tested period (720 h). Similarly, 12 h light/6 h dark (L12/D6) and 8 h light/4 h dark (L8/D4), which had totally 40 and 60 L/D cycles in the tested period. Following a 30-day period of transplantation, lettuce plants were sampled at a fixed initial planting time point to measure the corresponding parameters. Additional samples were promptly placed in liquid nitrogen and stored at a temperature of -80°C.

After 30 days transplanting, phenotype measurements were conducted within 30 minutes of the lighting switch on. For each scenario, 18 plants randomly selected were regarded as a repetition, and there were 3 repetition in each treatment. The plant height was the linear distance between the highest point of the canopy leaf and the substrate soil, and the maximum canopy diameter was the maximum linear distance between the canopy leaf tip. The leaf was photographed with a Nikon camera, and its area was calculated using the image segmentation approach. The root shoot ratio was computed as follows:

For analysis, plant leaves that were in good health and fully developed were selected randomly from each scenario. The determinations were performed within 30 minutes of the lighting switch on. The analysis was repeated 3 times for each physiological index.

The same part of mature leaves of lettuce in different light cycle were collected for analysis. For each scenario, three biological samples were chosen and designated as L16/D8_1, L16/D8_2, L16/D8_3, L12/D6_1, L12/D6_2, L12/D6_3, L8/D4_1, L8/D4_2, and L8/D4_3, correspondingly. All samples were harvested within 30 minutes of the lighting switch on and stored in -80°C. TRIzol reagent (Thermo Fisher, Waltham, USA) was used to extract the entire RNA according to the manufacturer’s protocol. Nanodrop2000 (Thermo Fisher, Waltham, USA) was used to determine the quality and strength of the extracted RNA. A total of nine RNA samples were sent to Hangzhou Zhijun Technology Co., Ltd. for library preparation and RNA sequencing. RNA sequencing was performed using the illumina novaseq 6000. Sequencing data quality control involves analyzing statistics of the sequencing data, raw data and quality control data. Trimmomatic (version 0.39) (Bolger et al., 2014) software was used to filter the pair-end reads. Bwa (version 0.7.17) software is used to align the filtered FASTQ sequence to the reference genome of lettuce downloaded from NCBI (https://www.ncbi.nlm.nih.gov/genome/?term=Lactuca+sativa), and output the transcription.

CoverM (version 0.6.1) (https://github.com/wwood/CoverM) software was employed to filter transcription. Using the reads_per_base and method counts functions, the fragments per kilobase of transcript sequence per millions base pairs (FPKM) and feature count were calculated from the filtered transcripts. Gene function and pathway were screened based on Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) database (Kanehisa and Goto, 2000). By input feature count, the analysis of differentially expressed genes (DEGs) was carried out using DESeq2 (Love et al., 2014). The differential gene expression volcano plots and correlation coefficient heatmap for comparison were analyzed using ggplot2 package and corrplot package in R, respectively. The gene ontology (GO) annotations were assigned to each gene by GOseq package (Young et al., 2010). The software KO-Based Annotation system was used to analyze the statistical enrichment of DEGs in the KEGG pathway.

For the detection of co-expression network and module, the R package (version 3.6) WGCNA (Weighted gene co-expression network analysis) was employed (Langfelder and Horvath, 2008). Pearson’s correlations were conducted to detected the correlations of the overall gene expression levels between biological replicates in the RNA-seq assay (R²>0.8). Clusters and heatmaps from the RNA-seq data under different light cycle were analyzed using ggplot2 package according to the FPKM value. The Cytoscape-V3.7.2 and its plugin cytohubba were used to construct the network for the modules. The top 30 core genes in connectivity were visually analyzed.

The plants in each scenario will return to the same time point to enter the light cycle after going through the 72 h cycle ( Figure 2 ). So we sampled the lettuce 2 hours before entering this light period, which is d1; When the lettuces entering this light period, we sampled the materials and remarked as l1; When the lettuces entering this light period 2 hours later, we sampled the materials and remarked as l2; And 4 hours later remarked as l3. Finally, When the lettuces of each scenario entering the subsequent dark phase 2 hours later, we sampled the materials and remarked as d2. 1.0 g fresh plant leaves of different phases were ground into powder in liquid nitrogen. The first-strand cDNA was generated by reverse transcribing total RNA extracted with TRIzol reagent and transcriptase (Takara, Kyoto, Japan). In qRT-PCR, the gene β-tubulin (LOC111880334) was used as a control. Supplementary Table S2 showed a list of the primers used in this study. The 2^-ΔCt method was used to determine the relative gene expression (Livak and Schmittgen, 2001; Thomas et al., 2008). Each scenario had three replicates, and the experiments were conducted at least twice.

For phenotypic and physiological data, statistical analysis were conducted by one-way analysis of variance (ANOVA), using SPSS 24.0 (SPSS, Inc., Chicago, USA). 18 samples and 3 samples were collected as one repeat for phenotypic and physiological measurement, respectively, and each treatment was repeated three times. The values represented the mean ± standard error. Tukey’s multiple range test was used to determine the differences between mean values at the 0.05 level.

To investigate the sensitivity of lettuce to different L/D cycle, we measured plant shoot, canopy diameter and fresh and dry weight of lettuce grown at L16/D8, L12/D6 and L8/D4. We found that the phenotype of lettuce changed significantly under different L/D cycle ( Figure 3A ). Lettuce that experienced diel change twice a day (L8/D4) were significantly taller than lettuce that experienced diel change only once (L16/D8) ( Figure 3B ), indicating that different L/D cycle could affect lettuce growth (p<0.05). Canopy diameter plays a crucial role in plant growth and fruit bearing. From the results, it was found that the diameter of the canopy of lettuce in L8/D4 scenario was significantly larger than that of lettuce in L12/D6 scenario ( Figure 3C ) (p<0.05). In addition, in terms of fresh weight and dry weight, lettuce in L8/D4 scenario was not much different from lettuce in L16/D8 scenario, but both were heavier than lettuce in L12/D6 scenario ( Figures 3D, E ) (p<0.05). By measuring the root shoot ratio and leaf area, it was found that the underground part of lettuce growing under L16/D18 was more developed ( Supplementary Figure S2 ). These results suggested that different L/D cycle had significant impacts on the lettuce phenotype. Moreover, contrary to previous understanding of the L/D cycle of plants, lettuce grown in L8/D4 scenario grew taller.

To investigate the impact of different L/D cycle on lettuce quality, some physiological indexes of lettuce leaves were measured. In the lettuce leaves, the chlorophyll a, b, and a + b exhibited a greater rise in the L16/D8 and L12/D6 scenarios compared to the L8/D4 scenario ( Figure 4A ) (p<0.05). On the contrary, the lettuce leaves exhibited a greater rise in carotenoid content under the L8/D4 scenario compared to the L16/D8 and L12/D6 scenarios ( Figure 4B ) (p<0.05). Carotenoid, as auxiliary pigments in plant photosynthesis, restored the growth of lettuce mediated by chlorophyll reduction in the L8/D4 scenario by compensating for the use of visible light radiation. In addition, we found that L/D cycle increased the quality of lettuce. The contents of soluble protein and soluble sugar in lettuce leaves were notably higher under the L8/D4 scenario compared to the L12/D6 scenario, indicating an increase of 3% and 5%, respectively ( Figures 4C, D ) (p<0.05). Moreover, it experienced a slight decrease under the L16/D8 scenario compared to the L8/D4 scenario, yet it did not attain statistical significance. There was no significant difference in total free amino acid content among the three scenarios ( Figure 4E ). Nevertheless, when comparing the nitrate levels in lettuce leaves, we found that there was a significant decrease of approximately 50% in L8/D4 scenario comparing that under L16/D8 and L12/D6 scenarios (p<0.05) ( Figure 4F ).

RNA-seq analysis was conducted on the lettuce leaves to investigate the underlying mechanism behind the different phenotypes observed in lettuce grown at L16/D8, L12/D6, and L8/D4. For transcriptome sequencing, a total of 9 RNA samples were sent, with each experiment comprising three biological replicates. Biological replicates displaying a strong correlation (R²>0.8) in gene expression levels ( Supplementary Figure S3 ). The RNA-Seq data of 9 cDNA libraries (L16/D8_1, L16/D8_2, L16/D8_3, L12/D6_1, L12/D6_2, L12/D6 _3, L8/D4_1, L8/D4_2, and L8/D4_3) were summarized in Supplementary Table S1 . Following the completion of quality control, we obtained a total of 66.94 Gb of clean data. The content of Q30 reached 95.83% throughout the 9 samples. A total of 123,569 transcripts were obtained in the filtered assembly. Most of the sequence lengths were concentrated below 20000bp ( Supplementary Figure S4 ).

There were 42,796 non-redundant high-quality unigenes annotated by searching a database of common functions. From the COG, EC, GO, KEGG, and Pfam databases, a total of 40,332 (94.24%), 9097 (21.26%), 20,882 (48.79%), 20,052 (46.85%), and 39,527 (92.36%) annotated unigenes were obtained, respectively ( Supplementary Figure S5 ). The PCA analysis showed an obvious differentiation between the three scenarios, suggesting that the differences among the nine sample bioreplicates aligned with the anticipated experimental design ( Supplementary Figure S6 ).

The fragments per kilobase of transcript per million reads mapped (FPKM) method were used to predict gene transcriptional accumulation. By comparing the expression level under different L/D cycle (L16/D8 vs L12/D6; L16/D8 vs L8/D4; L12/D6 vs L8/D4), genes that differentially up-regulated and down-regulated were screened out (Log2FoldChange > 1, adjust padj< 0.05). The results showed that 2270 DEGs were found between the L16/D8 and L12/D6 scenarios. And there were 2131 upregulated genes and 139 downregulated genes ( Figure 5A ). Furthermore, 3284 DEGs were found between the L16/D8 and L8/D4 scenarios; among these genes, 1338 were found to be upregulated, while 1946 were downregulated ( Figure 5B ). The numbers of DEGs between L12/D6 and L8/D4 scenarios were 5159. The upregulated genes were 2523, and downregulated genes were 2636 ( Figure 5C ). In addition, venn diagram were used to summarize DEG counts between all combinations of L16/D8, L12/D6, and L8/D4 scenarios. And there were 167 common DEGs between the three comparison scenarios ( Figure 5D )

By Go annotations, we discovered that the DEGs exhibited enrichment in the categories of “cellular process”, “cellular metabolic process”, and “metabolic process” within the GO classification of “biological process”. The DEGs were found to be enriched in terms such as “chloroplast”, “plastid”, and “intracellular membrane-bound organelle” within the “cellular component” categories. In the “molecular function” categories, the DEGs were enriched in “binding” and “catalytic activity” terms ( Figure 6A ). By conducting KEGG pathway analysis, we enriched the top 10 DEGs and observed their significant enrichment in various signal transduction and metabolic pathways, such as “photosynthesis”, “plant hormone signal transduction”, and “carbon metabolism” ( Figure 6B ). These results laid a solid foundation for the study of different L/D cycle adaptation mechanisms of lettuce.

In order to identify the genes associated with the L/D cycle in lettuce, we performed a WGCNA analysis. When the scale-free fit index is 0.8, the minimum soft threshold for constructing scale-free networks is 9. Hence, the value of 9 can be chosen as the most suitable soft threshold for further analysis ( Supplementary Figure S7 ). The co-expression network was constructed using the optimal soft threshold, dividing the genes into various modules, and conducting the cluster dendrogram( Figure 7A ). From the results of WGCNA, 36 gene modules were identified, and marked with different colors. The member number in modules ranged from 47 to 18221 ( Figure 7B ). By constructing cluster tree and correlation heatmap, the interaction between modules was explored. These 36 modules could be aggregated into two clusters with a high degree of interactive connectivity ( Figure 7C ).

In order to identify the module that related to the treatments of L16/D8, L12/D6, and L8/D4, we conducted a correlation analysis between treatments and modules. From the results, we found that module in turquoise was positively correlated with L8/D4, and negatively correlated with L16/D8 and L12/D6. Module in green was positively correlated with L12/D6, and negatively correlated with L16/D8 and L8/D4. Furthermore, module in greenyellow was positively correlated with L16/D8, and negatively correlated with L12/D6 and L8/D4 ( Figure 8A ).

By Go annotation, we found that the DEGs of three modules in green, greenyellow, and turquoise exhibited enrichment in the categories of “regulation of transcription”, “DNA-templated” and “translation” within the GO “biological process” terms. The DEGs were found to be enriched in the terms “cytosol” and “membrane” within the “cellular component” categories. The DEGs in the GO categories of “molecular function” were found to be enriched in terms such as “ribosome structural constituent” and “protein kinase activity” ( Figure 8B ). The KEGG enrichment analysis revealed that the green module was enriched in the pathway of “sphingolipid metabolism”, the greenyellow module was enriched in the pathway of “plant hormone signal transduction”, and the turquoise module was enriched in the pathway of “nucleocytoplasmic transport”( Figure 8C ). These three pathways directly related to plant cell growth, differentiation and signal transduction. Overall, the eigengenes of these three modules played important roles in promoting the growth and improving the quality of lettuce.

In order to explore the key gene involved in lettuce’s L/D cycle, we analyzed the top 30 nodes in the green, greenyellow, and turquoise modules with the highest connectivity through cytoscape 3.7.2 version. From the results, it was found that CIP1 and SCL34 were hub genes in green module. ROPGEF1 and ACD6 were located at the center of the greenyellow network. Furthermore, CcmB and Rps4 were hub genes in turquoise module ( Figure 9 ). As previously reported, CIP1 can be ubiquitinated by COP1, resulting in suppression of light signal transduction and light morphogenesis (Han et al., 2020). ROPGEF1 is associated with ABA-induced stomatal closure in plants (Li and Liu, 2012). In Arabidopsis, CcmB has the ability to form an ABC transporter within the mitochondria, playing a role in the maturation of cytochrome c (Rayapuram et al., 2007). Therefore, it is probable that these genes functioned as crucial switches in the alterations of plant growth and development induced by different L/D cycles.

By qRT-PCR, we examined the expression levels of LsCIP1, LsSCL34, LsROPGEF1, LsACD6, LsCcmB, and LsRps4 at different time points throughout the diurnal cycle. From the results, we have found that the diurnal response patterns of the 6 hub genes in different scenarios were significant. In L12/D6 scenario, all the hub genes showed down-regulated expression after turning on the light, and the down-regulated expression trend slowed down 2 hours after the light was turned off. On the contrary, most of the hub genes (except LsCcmB in L16/D8) of lettuce grown in L16/D8 and L8/D4 scenarios were upregulated after the light was turned on, and the upregulation trend slowed down two hours after the light was turned off, while the expression trend of L8/D4 was more flat than that of L16/D8 ( Figure 2 ). The difference of expression patterns in different scenarios indicates the difference of lettuce phenotype, and also illustrates that these six hub genes play important roles in regulating the growth and development of lettuce.