Wild-type G. lemaneiformis was harvested from Zhanshan Bay, Qingdao, China (36° 03′ N, 120° 22′ E) in October 2021 and transported to the laboratory on ice. In the laboratory, the thalli were washed with sterile seawater to remove sediment and epiphytes. Healthy algae were then selected and cultured for 1 week in sterile seawater enriched with Provasoli medium under the following conditions: 23°C, salinity 25, a photoperiod of 12 h light:12 h dark and a light intensity of 30 μmol photons m^-2 s^-1. The development of tetrasporophytes on the epidermis of the thalli was observed under light microscope and the different developmental phase of the thalli were selected for subsequent experiments according to the delimitation method of Wang et al. (2016).

The levels of phycoerythrin (PE) and phycocyanin (PC) were determined with reference to the method of Chen et al. (2022). Approximately 0.1 g of fresh algae was ground with liquid nitrogen and extracted with 5 ml of 50 mM phosphate-buffered saline (PBS, pH 6.8). After 2 h of reaction, the mixture was centrifuged at 8,000 ×g for 20 min, and the supernatant was analyzed using a spectrophotometer.

Chlorophyll fluorescence was measured using a Photon Systems Instruments (PSI) fluorometer (AquaPen-C, Photo System Instruments, Drasov, Czechia). The actual photochemical efficiency (Fv'/Fm') and photochemical quenching coefficient (Qp) were measured after 15 min of dark adaptation. The relative electron transfer rate (rETR) was measured at different photosynthetically active radiation (PAR) levels (10, 20, 50, 100, 300, and 500 μmol photons m⁻² s⁻¹) and then the light use efficiency (LEU, α) was calculated using OriginPro 9 (OriginLab) software according to the method of Jiang et al. (2022).

For determination of superoxide dismutase (SOD), peroxidase (POD), and catalase (CAT) enzymatic activities, approximately 0.1 g of fresh seaweed was ground in liquid nitrogen, extracted in 50 mM PBS buffer (pH 7.8), and centrifuged at 8,000 ×g for 10 min and the supernatant was used to determine SOD, POD, and CAT activities by assay kits (Comin Bio, Suzhou, China) according to the manufacturer’s instructions. For the determination of malondialdehyde (MDA) content, approximately 0.1 g of fresh seaweed was ground in liquid nitrogen and extracted with trichloroacetic acid, centrifuged at 8,000 ×g for 10 min and then measured using the MDA assay kit (Jiancheng Bio, Nanjing, China).

Total RNA was extracted using an RNeasy Plant Mini Kit (Qiagen, Shanghai, China) in accordance with the manufacturer’s recommendations. The quality of RNA was assessed with the Agilent Bioanalyzer 2100 (Agilent, Santa Clara, CA, United States) and quantified with the NanoDrop ND 1000 Spectrophotometer (Agilent, Santa Clara, CA, United States). Using the mRNA library preparation kit (MGIEasy™, MGI, Shenzhen, China), 12 complementary DNA (cDNA) libraries were constructed. The poly-A mRNA was isolated from total RNA by using OligodT magnetic beads, which were then cleaved into fragments with interrupted buffer. Reverse transcription was performed using random primers N6 on target RNA fragments to generate double-stranded cDNA (ds-cDNA). Then the ds-cDNA was ligated with adaptors and amplified with PCR. A single-stranded DNA library was produced by cyclizing single-stranded DNA with a bridge primer after PCR products had been heat-denatured into single strands.

RNA-seq was performed on the DNA nanoball sequencing (DNBSEQ) platform at the Wuhan Genomic Institution (BGI, Shenzhen, China). Read filtering and unigene annotations were performed using SOAPnuke software (BGI, Shenzhen, China). To acquire clean reads, adaptor sequences, reads with an unknown base N content larger than 10%, and low-quality reads (bases with quality values below 15 are higher than 50% of the total number of bases) were all eliminated (Cock et al., 2010). Clean reads were mapped with the reference transcript using Bowtie2 (Langmead and Salzberg, 2012) and gene and transcript expression values were determined using RNA-Seq by Expectation Maximization (RSEM) (Dewey and Bo, 2011). With the DESeq method, differentially expressed genes (DEGs) were identified, and genes with adjusted p values for false discovery rate (FDR, q-values) ≤0.001 and Log₂ (fold change) ≥1 were identified. Kyoto Encyclopedia of Genes and Genomes (KEGG) Orthology-Based Annotation System (KOBAS) 2.0 was used to produce a corrected p value, which was the Q value (Mao et al., 2005). All DEGs were analyzed using Gene Ontology (GO) and the KEGG databases. Significant enrichment was determined for the GO terms and enrichment analysis with a Q value of 0.05.

To validate the RNA-seq results, eight genes were randomly chosen for quantitative real-time PCR (qRT‐PCR). Following the manufacturer’s instructions, single-stranded cDNA was synthesized by reverse transcription using the Perfectstart Uni RT & qPCR Kit (TransGen, Beijing, China). On an ABI7500 qRT‐PCR System (Applied Biosystems, Foster City, CA, United States), qPCR was carried out using a reaction system of 20 μl of PerfectStart Green qPCR SuperMix (TransGen, Beijing, China). The relative expression levels of genes were calculated using the 2^−ΔΔCT method with the β-actin gene as the respective reference gene using three separate biological replicates. The qRT-PCR primers are listed in Supplementary Table 1 .

The data are provided as the mean and standard deviation and were analyzed by Statistical Product and Service Solutions (SPSS) 22.0 software. Additionally, the data were assessed using one-way analysis of variance (ANOVA) and Duncan’s multiple range test, with a significance threshold of 0.05.

Previous studies have divided the G. lemaneiformis process of tetraspore formation and release into four phases based on the state of the tetrasporangium (Wang et al., 2016). In this study, microscopic observation of epidermal cells was used to determine the developmental stage of the thalli ( Figure 1 ). Phase I was the immature phase, where the epidermal cells were closely arranged, the tetrasporangium had not yet begun to form, and all thalli that had not developed reproductively were in this stage ( Figure 1A ). Phase II was the prematuration phase, when a large number of tetrasporangia cells began to form but had not yet started to released tetraspores, and was a process that lasts for approximately two weeks ( Figure 1B ). Phase III was the full maturation phase, when tetraspores began to be released, leaving a large number of holes on the surface of the thalli; this was a process that lasts for approximately two weeks ( Figure 1C ). Phase IV was the postmaturation phase, when the release of the tetraspores was complete and a large number of holes remained on the surface of the thalli, this stage lasted for approximately four weeks ( Figure 1D ). The thalli whose surfaces matched the characteristics from Phase I (no tetrasporangium), Phase II (tetrasporangium formed but not tetraspores released), Phase III (approximately 50% of tetraspores released and a large number of holes left behind) and Phase IV (completion of tetraspore release and a large number of holes left behind) were selected for subsequent experiments.

To investigate the changes in photosynthesis during tetraspore formation, we measured phycobiliprotein levels and chlorophyll fluorescence parameters during the four phases. The results showed that the PE content was higher in Phase IV than in Phases I, II and III ( Figure 2A ), and that the PC content was not different among the four phases ( Figure 2B ). The actual electron photosynthetic efficiency (Fv’/Fm’) was markebly increased in Phases II and IV compared to the Fv’/Fm’ in Phase I, yet it did not change in Phase III ( Figure 2C ). This indicates enhanced photosynthesis with the development of tetrasporophyte formation. The trends in Qp and Fv’/Fm’ were consistent across the four phases, but the differences were not significant in Phases II and IV ( Figure 2D ). It is possible that increased light loss during photosynthesis was the cause of the opposite trend in α, which was decreased in Phases II and IV ( Figure 2E ).

To investigate the thallus response to reactive oxygen species (ROS) during tetraspore formation, the changes in the activities of three antioxidant enzymes were examined. The results showed that SOD activity was activated, with 63.1%, 24.3% and 34.9% higher activity in Phases II, III and IV, respectively, than in Phase I ( Figure 3A ). POD activity was increased by 39.1% and 49.0% in Phases III and IV, respectively ( Figure 3B ). The CAT activity significantly increased in Phase IV, while there was no change from Phases I to III ( Figure 3C ). Meanwhile, compared to Phase I, Phases II to IV had significantly increased MDA content ( Figure 3D ). These results suggest that tetraspore formation and release in G. lemaneiformis are related to oxidative stress.

To elucidate the molecular mechanism of tetraspore formation in thalli, RNA-seq analysis was carried out on the thalli of four phases of G. lemaneiformis tetraspore formation. Using the Beijing Genomics Institute Sequencing (BGISEQ) platform to analyze the RNA samples, it was found that a total of 12 samples generated clean reads between 21.61 and 21.69 after data filtering. Additionally, 41,099 unigenes were obtained in total. Supplementary Table S2 provides details on the transcriptome sequencing and de novo assembly of the 12 samples. All the assembled unigenes were searched by the Basic Local Alignment Search Tool (BLAST) against the public KEGG, GO, Non-Redundant (NR), Nucleotide (NT), SwissProt, Protein Family (Pfam), and Eukaryotic Orthologous Groups (KOG) databases with an E-value threshold of 10^-5. Among the 40,997 unigenes annotated in at least one database, 25,510 unigenes were annotated throughout all databases ( Supplementary Table S3 ).

Principal component analysis (PCA) was performed to describe transcriptome differences at different developmental phases, and Phases I, II, and IV could be clearly separated from each other, while there was partial overlap between Phases II and III, indicating that Phase II and III were relatively close in development ( Figure 4A ). The four phases of G. lemaneiformis development were examined using the transcriptome. The results revealed that a total of 10947 unigenes, including 6069 upregulated and 4878 downregulated unigenes, were differentially expressed between Phase I and II. Between Phase II and III, 7124 unigenes showed differential expression, including 3928 upregulated and 3196 downregulated unigenes. There were 12154 unigenes between Phases III and IV, including 6547 upregulated and 5607 downregulated genes ( Figures 4B-D ). In addition, between Phase I and Phase III/IV, there were 11386/14744 unigenes ( Supplementary Figure 1 ). Eight unigenes were randomly chosen from the DEGs for RT‐PCR analysis in order to confirm the accuracy of the RNA-seq results. The expression patterns of these eight unigenes were largely in agreement with the RNA-seq results ( Supplementary Figure 2 ).

GO analysis showed that these differentially expressed genes were annotated into three main functional categories—molecular function, cellular component and biological process—with the majority classified as cellular anatomical entity, catalytic activity, binding, cellular process and metabolic process. The KEGG classification showed that DEGs were mainly classified into global and overview maps, carbohydrate metabolism, translation, folding, sorting and degradation, amino acid metabolism, transcription and transport and catabolism ( Supplementary Figure 3 ).

GO and KEGG enrichment analyses were carried out to analyze the function of the DEGs, and GO enrichment showed that in the comparison between Phases I and II, the processes in which the DEGs were significantly enriched were those related to photosynthesis light harvesting, carbohydrate metabolic process and the membrane. In the comparison between Phases II and III, the processes in which the DEGs were significantly enriched were response to oxidative stress, reactive oxygen species metabolic process and cytochrome complex assembly. In the Phase III and IV comparison, the processes in which the DEGs were significantly enriched were photosynthesis, photosynthesis light harvesting, photosynthesis light reaction, and signaling. Overall, the GO enrichment analysis of the DEGs showed that they were primarily associated with photosynthesis, carbon metabolism and oxidative stress. Correspondingly, KEGG enrichment analysis was also carried out, and between Phases I and II, the processes in which the DEGs were significantly enriched were those related to photosynthesis-antenna proteins, carbon fixation in photosynthetic organisms, carbon metabolism and starch and sucrose metabolism pathways. In Phase III and IV, the processes in which the DEGs were significantly enriched were those involved in photosynthesis-antenna proteins, alpha-linolenic acid metabolism, linoleic acid metabolism and starch and sucrose metabolism. There was no pathway of enrichment between Phase II and III ( Figure 5 ).

Photosynthesis is the process by which green plants, algae and cyanobacteria use solar energy to produce carbohydrates from carbon dioxide and water (Rehman et al., 2016). In photosynthesis, antenna proteins are responsible for capturing light energy and transferring its excitation to the photosystem II (PS II) core complex, where it is converted to chemical energy with the release of oxygen (Lv et al., 2015; Li et al., 2020). Phycobilisomes are the main light-harvesting antennae in red algae, which mainly transmit light energy to PSII and can also be coupled with PSI (Pagliano et al., 2013; Li et al., 2021). In addition, LHCI also exists in red algae to transmit light energy to PSI (Tian et al., 2017; Bai et al., 2021). In the present study, phycobilisomes and LHCI-related genes were highly expressed in Phases II and IV ( Figure 6A ), which suggests that G. lemaneiformis absorbed more light energy during these phases. Additionally, Phase IV had the highest expression of related genes, indicating that this phase had the greatest absorption of light energy. Similar results were also found in chlorophyll fluorescence parameters, where the Fv’/Fm’ was higher in Phases II and IV than that in Phase I ( Figure 2 ).

The light energy captured by the antenna protein undergoes a series of reactions in PSII, PSI, the cytochrome b6/f complex and F-type ATP synthase to synthesize ATP (Bielczynski et al., 2020). Transcriptome analysis revealed that genes involved in encoding these complexes were significantly upregulated in Phases II to IV, suggesting increased photosynthetic electron transport in these phases and ultimately increased photosynthesis ( Figure 6B ). In Porphyra yezoensis, it provides energy for sperm formation and release by enhancing the photosynthetic efficiency of PSII at the four-cell phase (Yang et al., 2012). The increased photosynthesis in Phases II and III of this study may be used for the energy supply for the process of tetraspore formation and release. Similarly, the photosynthesis enhanced in the reproductive development of the green algae Ulva prolifera was found to provide the energy for sporogenesis (Jiang et al., 2022). Here, the highest expression of genes encoding proteins in Phase IV indicated that photosynthesis is further enhanced in this phase, which is consistent with the results of Wang et al. (2016). The enhanced photosynthesis in Phase IV may be related to the repair of wounds after tetraspore release.

Carbon fixation is the primary source of carbon in plants, and carbon metabolism plays a crucial role in both the structural makeup and energy generation of plants (Duarte et al., 2013). It is possible that carbon assimilation increases during the formation and release of tetraspores (Phases II to IV), based on the facts that the genes involved in the calvin cycle were significantly upregulated in these phases. Rubisco, a key rate-limiting enzyme in the calvin cycle, was also upregulated, which is consistent with the findings of increased light energy uptake and photosynthesis in Phase II to IV. Interestingly, the expression of most genes in Phase IV was slightly lower than that in Phase II and III, suggesting that carbon assimilation may be less efficient in this phase than in Phase II and III, but light energy uptake and photosynthesis were higher in Phase IV than in Phase II and III, possibly due to increased energy consumption in this phase resulting in lower carbon assimilation efficiency ( Figure 7 ). The increase in carbon fixation flux ultimately leads to an increase in starch and trehalose synthesis, which accumulates more energy for the formation and release of tetraspores. It is worth mentioning that trehalose-6-phosphate in the trehalose biosynthesis pathway has an important role in plant signaling and is associated with plant embryo development, seed germination and the timing of flowering (Gazzarrini and Tsai, 2014). For instance, delayed flowering was observed in plants with defective trehalose-6-phosphate synthesis (Ponnu et al., 2011), and mutations in TPS1 in Arabidopsis resulted in slow growth and stalled reproductive development, demonstrating that TPS1 plays an important role in the transition from nutritional to reproductive development in Arabidopsis (Dijken et al., 2004). The significant up-regulation of genes encoding trehalose-6-phosphate in Phases II to IV suggests that there may be a strong link between tetraspore formation and increased synthesis of trehalose-6-phosphate during tetrasporophyte development in G. lemaneiformis, but the exact function of trehalose-6-phosphate requires further investigation.

Carbon metabolism also provides the energy and carbon backbone for amino acid biosynthesis (Zhao et al., 2021). In the present study, the synthesis of amino acids including Asp, Asn, Gly, Ser, Glu, Gln, His, and Orn, was promoted from Phase II to IV; notably, these amino acids are required for reproductive development in G. lemaneiformis ( Supplementary Table 6 ). In addition to serving as precursors for the synthesis of proteins and peptides, these amino acids are also used by plants to make a variety of other natural products (Joshi et al., 2007). Asp is involved in the synthesis of pyrimidine and purine nucleotides in plants, providing the carbon backbone for the de novo synthesis of purines and pyrimidines (Sullivan et al., 2015). Glutamate is not only a precursor for the synthesis of chlorophyll, γ-aminobutyric acid, etc., but also regulates plant development by participating in physiological processes such as seed germination and pollen germination (Qiu et al., 2020). In addition, the glycine-rich binding proteins GR-RBPs have important roles in seed germination and in promoting fruit ripening (Ma et al., 2021). Overall, G. lemaneiformis meets its material requirements during reproductive development by upregulating the synthesis of amino acids such as Asp, Gly and Glu.

As byproducts of cellular metabolism, ROS can function as signaling molecules to control a variety of physiological and developmental processes in plants, such as stomatal opening, biotic stress responses, and reproductive development (Lodde et al., 2021; Shim et al., 2022). Increasing evidence suggests that ROS are involved in the reproductive and developmental processes of red algae. H₂O₂ can mediate fertilization and postfertilization development of the red alga Bostrychia moritziana, and can regulate sperm mitosis, gamete fusion, and zygotic development (Shim et al., 2022). Exogenous H₂O₂ in Pyropia yezoensis can promote the production and release of monospores, indicating that oxidative stress can regulate the life cycle of P. yezoensis by inducing the transition from vegetative cells to monospores (Takahashi and Mikami, 2017). In the present study, SOD and POD activities and MDA levels were increased during spore formation and release ( Figure 3 ), and genes involved in encoding antioxidant enzymes (APX, TRX, GR, TR, PRX, and CAT) were significantly upregulated from Phase II to IV ( Figure 8 ), suggesting that oxidative stress is associated with the tetraspore formation and release process in G. lemaneiformis, but its specific regulatory mechanism still requires further study.

Riboflavin and its derivatives flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN) are involved in important physiological processes such as energy metabolism and redox metabolism are essential for normal plant growth and development (Sa et al., 2016). Recent studies have shown that riboflavin can be involved in the regulation of maize seed germination and development through the integration of mitochondrial energy metabolism and the cell cycle (Tian et al., 2022). In the present study, the synthesis of riboflavin was induced in Phases II to IV, suggesting that riboflavin may have an important role in tetraspore formation and release ( Figure 9 ). In land plants, riboflavin can regulate ROS concentrations to improve drought tolerance in tobacco plants and can also affect reactive oxygen balance and chloroplast development in rice (Deng et al., 2014; Hu et al., 2021). In the present study, the changes in riboflavin biosynthesis and photosynthesis as well as in antioxidant enzymes were consistent with these findings. Thus, riboflavin may be involved in the reproductive processes of G. lemaneiformis through its participation in oxidative stress and photosynthesis.