The GA biosynthetic pathway can be divided into three parts. In the first part, geranylgeranylpyrophosphate (GGPP) is converted into ent-copalyl diphosphate by ent-copalyl diphosphate synthase (CPS or GA1) (Sun and Kamiya, 1994) and then into ent-kaurene by ent-kaurene synthase (KS or GA2) (Yamaguchi et al., 1998; Figure 1). Neither Arabidopsis GA1 nor GA2 orthologs were identified in the sampled dinoflagellates, including Symbiodinium microadriaticum, Breviolum minutum, Cladocopium goreaui, and Fugacium kawaguti (Figure 1A). Unlike vascular plants, fungi catalyze ent-kaurene formation from GGPP using a single bifunctional CPS/KS enzyme (Tudzynski et al., 1998). Such a bifunctional enzyme has not been identified in these algae. However, genes encoding ent-kaurene oxidase (KO or GA3) and ent-kaurenoic acid oxidase (KAO) (Helliwell et al., 1999), the enzymes from the second part of the GA-biosynthetic pathway, were identified in every species examined here (Figure 1A). In the third part of the pathway, GA₁₂ and GA₅₃ are further oxidized to other C₂₀- and C₁₉-GAs by a series of 2-oxoglutarate-dependent dioxygenases (2ODDs) (Lange et al., 1994; Israelsson et al., 2004; Figure 1B). In Arabidopsis, the 2ODDs in each of the three families (17 genes in total) act as biosynthetic and catabolic enzymes, respectively. However, far fewer putative 2ODD genes were found in the dinoflagellates. Thus the GA biosynthetic pathway is moderately conserved in evolutionary terms and occur in dinoflagellate species.

To validate the genome-based metabolic reconstruction, we then selected B. minutum as a model to exam the presence of GA metabolites in the dinoflagellates. This species is an emerging research model for study on the mutualistic partnerships between dinoflagellate endosymbionts and cnidarian hosts (Davy et al., 2012; Lu et al., 2020; Zhou et al., 2022). Ultra-performance liquid chromatography-electrospray ionization tandem mass spectrometry (UPLC-ESI-MS/MS) revealed that B. minutum contained significant quantities of GAs, with GA51 being the most abundant (Figure 1B and Supplementary Table 1). It thus seems that the GA-synthesis pathway in microalgae involves as-yet-unidentified CPSs and KSs (or may involve an entirely novel route). This is supported by the finding that Arabidopsis ga1 null mutants (GA1 is the single-copy gene that encodes ent-copalyl diphosphate synthase; i.e., CPS), although very severely dwarfed, contain trace quantities of various GAs (Sponsel and Hedden, 2010). GA₄ and GA₁ are both present in B. minutum (Figure 1B and Supplementary Table 1) and are known to be bioactive in many species including Arabidopsis. GA₄ and GA₁ are formed from GA₉ and GA₂₀, respectively, in plants by the action of GA3 hydroxylase (GA3ox). GA₁₂ and GA₅₃ are the precursors of GA₉ and GA₂₀, respectively, and are formed by the action of GA20ox (Figure 1B). In B. minutum, GA₅₃ and GA₂₀ were identified whereas GA₁₂ and GA₉ were absent (Figure 1B and Supplementary Table 1).

In many plants, GA₁₂ is converted to GA₅₃ by a GA13ox-catalyzed hydroxylation at C-13. In most plants, the 13-hydroxylation pathway predominates, although in Arabidopsis and some other species, the non-13-OH-pathway is more prevalent (Zanewich and Rood, 1995). In the cytosol, GA₅₃ is converted into various other GAs via parallel pathways. This conversion involves a series of oxidations at C-20, resulting in the eventual loss of C-20 and the formation of C₁₉-GAs. In the non-13-hydroxylation pathway, GA₁₂ is oxidized to produce GA₉ (both are absent in B. minutum), via the intermediacy of GA₁₅ and GA₂₄ (both are absent) (Figure 1B and Supplementary Table 1). GA₉ is then oxidized to the bioactive compound GA₄ (present) by a 3β-hydroxylation reaction (Figure 1B and Supplementary Table 1). In the 13-hydroxylation pathway, GA₅₃ is sequentially oxidized at C-20 to produce GA₂₀ (present), which is then 3β-hydroxylated to yield bioactive GA₁ (present) (Figure 1B and Supplementary Table 1). Finally, hydroxylation at C-2 converts GA₄ and GA₁ into their inactive forms, GA₃₄ (present), GA₅₁ (present), GA₂₉ (absent) and GA₈ (present), respectively (Figure 1B and Supplementary Table 1). Therefore, most intermediates from the 13-OH pathway for the synthesis of bioactive GA₁ are present in B. minutum, along with intermediates from their deactivation pathways. In contrast, the intermediates from the non-13-OH pathway are absent, but bioactive GA₄ and its deactivation forms (GA₃₄ and GA₅₁) are present (Figure 1B). Moreover, the amount of GA₅₁ is relatively higher than the remaining GA forms (Supplementary Table 1). These results may suggest that the non-13-hydroxylation pathway may predominate in B. minutum and the multifunctional GA dioxygenase in B. minutum has a distinct substrate preference from those in Arabidopsis.

Little was known about the functional roles of GAs in microalgae, particularly at the molecular level. Although the application of exogenous GAs to microalgal cultures suggests that they may function as stress signals (Park et al., 2013; González-Garcinuño et al., 2016; Yu et al., 2016), the evidence is still inconclusive. We thus probed the possible effects of abiotic stresses on the synthesis of GAs in dinoflagellate endosymbionts. For reef-building corals, the bleaching process is induced by environmental stimuli, such as high light, high temperature, and acidification (Lu et al., 2020; Jiang et al., 2021). Therefore, to probe the physiological relevance of GAs in microalgae, we monitored the transcriptional activities of the committed GA biosynthetic gene GA 20-oxidase (GA₂₀ox1) and GA profiles under the high light, high temperature, and acidification conditions. These time-series datasets unveiled the active involvement and the dynamics of GAs and GA-related transcripts during stress responses in B. minutum (Figure 2).

In higher plants, GAs play direct or indirect roles in plant responses to abiotic stress (Peleg and Blumwald, 2011). In B. minutum, under the thermal stress, the transcription of GA₂₀ox1 was largely stable where it declined somewhat after the stress onset (p > 0.05; 6 h) and then return to the control level (24 h) (Figure 2A). In contrast, either high light or acidification increased the abundance of GA₂₀ox1 transcripts somehow (Figures 2B,C). Upregulation of GA₂₀ox1 occurred rapidly within the first 6 h upon acidification, with its transcript levels at a relatively higher level at 24 h (Figure 2B). Nevertheless, a depressed level of GA₂₀ox1 transcript was observed following long-term acidification (72 h; Figure 2B). Following high light, the transcript levels of GA₂₀ox1 follow a trend that declined at the early stage and increased slowly and continuously afterward, peaking at 24 h (Figure 2C). This suggested that, upon high light and acidification, relative to high temperature, the GA₂₀ox1 gene may contribute more to stress response at the transcriptional levels. The rate of GA biosynthesis is thus regulated as part of the response to the bleaching-inducing environmental stimuli. This strongly suggests that GA-based processes are important for at least the explored stress conditions in B. minutum, which is consistent with the alleviation stress symptoms of green microalga Chlorella vulgaris by exogenously applied GA₃ (Piotrowskaniczyporuk et al., 2012).

The increased transcription of GA biosynthesis genes in B. minutum does not always equal to increased biosynthesis of GAs. We thus probed the dynamics of intracellular GA levels under the three selected stresses. To assess the dynamic equilibrium of 13-OH and non-13-OH pathways for GA biosynthesis, we further determined the detailed GA profiling. Despite of the fluctuation, with long-term stress treatments (72 h), levels of GA₅₃ increased under the high-light and acidification conditions while it decreased under high temperature (Figure 3A). The levels of the known bioactive species GA₁ generated through the 13-OH pathway were not stable (could be detected at limited time points). In contrast, although the accumulation of GA₁₂ could not be detected under either control or stressed conditions, the bioactive product of the non-13-OH pathway (i.e., GA₄) was remarkably decreased following the onset of high light treatment (P ≤ 0.001) and returned to a level higher than the control after 72 h treatment (Figure 3B). As for the acidification conditions, the level of GA₄ was rapidly elevated within the first 6 h and remained at a relatively higher level than the control thereafter (Figure 3B). This was consistent with the transcriptional abundance of GA₂₀ox1 under the high light and acidification conditions where the GA₂₀ox1 transcript increased at the early stage upon acidification while it declined at the early stage and increased under prolonged high light conditions (Figure 2B,C). For the inactive GA forms in the non-13-OH pathways, GA₅₁ was rapidly decreased following the onset of all three stresses (Figure 3C) while the level of GA₃₄ was relatively similar between the stressed algal cells and the controls (Supplementary Table 1). Meanwhile, inactive forms in the 13-OH pathways (i.e., GA₈) remained at relative stable levels under high-light and acidification treatments (Figure 3D). In contrast, under thermal stress, the content of GA₈ increased at the end of detection (Figure 3D). Therefore, both the biosynthesis and deactivation pathways are actively involved in the stress responses and the non-13-hydroxylation pathway played predominant roles in GA homeostasis in B. minutum. Moreover, despite fluctuations and wide discrepancy in the levels of various GA species, the content of the active GA species (i.e., GA₄ at late phase) increased upon the high-light and acidification stresses (Figure 3B). In contrast, GA₄ decreased at the end of detection under the thermal stress (Figure 3B). These observations suggest a general function and a stress-dependent manner of GAs in the dinoflagellate endosymbiont.

More than 130 GAs have been identified in plants while only a handful of them are biologically active in flowering plants which include GA₁, GA₃, GA₇, and GA₄ (∼1000 fold higher activity than GA₁) (Kapoore et al., 2021). Among the active forms, GA₃ and GA₄₊₇ (a mixture of GA₄ and GA₇) are commercially available, we thus select them to further probe the potential functions of GAs in B. minutum. The levels of GAs were perturbed by supplementing GA₃ and GA₄₊₇ to algal cultures at concentrations of 0.008, 0.08, and 0.8 μg⋅L^–1. For GA-treated and untreated cultures, algal cells at the linear growth phase were inoculated into the fresh medium under the indicated conditions, and the cell growth was tracked for three days. Although subtle difference was observed between the algal cells treated with or without GAs under the control conditions (pH 8.2, 25°C, and 50 μmol⋅photons⋅m^–2⋅s^–1 light intensity), the growth was higher than the control under the thermal stress in either GA₃- (Figure 4A) or GA₄₊₇- (Figure 4B) treated cells (p ≤ 0.05). Exogenously applied GAs may compensate for the decreased endogenous GAs under the thermal stress (i.e., GA₄; Figure 3B) which subsequently improved the dinoflagellate’s growth. To assess whether GAs plays a role in the response to environmental stresses other than high temperature, the algal cells in the linear growth phase were inoculated into the fresh medium with the presence or absence of GAs under the high-light or the acidification conditions. However, a bare difference was observed in cell numbers between the GA-treated and the untreated cells till the end of detection. This might attribute to that the amount of GA₄ under these stresses is already higher than that of the controls (Figure 3B). Therefore, GAs could act as a stress hormone in B. minutum where their effects are stress- and time-dependent.

Breviolum minutum was purchased from the National Institute for Environmental Studies (Japan). They were cultured routinely in 250-ml conical flasks with L1 medium containing kanamycin (50 mg⋅L^–1), and streptomycin (50 mg⋅L^–1) to minimize the growth of bacteria (Lu et al., 2020). When culturing dinoflagellate cells, the control group was maintained under the organism’s preferred physiological conditions (pH 8.2, 25°C, and continuous 50 μmol⋅photons⋅m^–2⋅s^–1 light intensity) as described in earlier reports (Jiang and Lu, 2019). Under otherwise identical high-light, high-temperature and acidification stress conditions, the light intensity, temperature, and pH were set at 200 μmol⋅photons⋅m^–2⋅s^–1, 34°C, and pH 7.6, respectively, based on our previous tests (Jiang and Lu, 2019; Lu et al., 2020). Algal cells at a concentration of approximately 1 × 10⁶ cells⋅ml^–1 were harvested, washed with sterile seawater, and inoculated into the fresh medium in triplicate. Cultures were started with the same initial cell concentration of 2 × 10⁵ cells⋅ml^–1, acclimated for 12 h under 50 μmol⋅photons⋅m^–2⋅s^–1 light, and then exposed to designated stress conditions. Growth was monitored by measuring the turbidity (Biochrom GeneQuant 1300, GE Healthcare) and the cell number (LUNA-FL Dual Fluorescence Cell Counter, Logos Biosystems) at indicated intervals (Gan et al., 2017; Lu et al., 2020).

The cDNA preparation and reverse transcription were performed as previously (Lu et al., 2021a). In brief, the threshold cycle (2-^–ΔΔCt) method was used to quantify relative changes in transcript levels from the quantitative PCR (qPCR) data. Levels of the transcripts under each set of treatment conditions at each time point were first normalized to actin expression levels. The values obtained for each gene were then normalized to the values in the control treatments (specimens that were not subjected to the stresses) at the corresponding time point. Values are means and standard errors obtained from three experiments (Han et al., 2020). The primers used are listed in Supplementary Table 2.

The sample preparation and analysis of GAs was performed according to the previous report (Urbanová et al., 2013) with some modifications. Briefly, the lyophilized algal materials (100 mg dry weight) were sonicated and extracted with 1 mL 80% acetonitrile containing 5% formic acid and 19 internal gibberellin standards ([²H₂]GA₁, [²H₂]GA₃, [²H₂]GA₄, [²H₂]GA₅, [²H₂]GA₆, [²H₂]GA₇, [²H₂]GA₈, [²H₂]GA₉, [²H₂]GA₁₂, [²H₂]GA₁₂ald, [²H₂]GA₁₅, [²H₂]GA₁₉, [²H₂]GA₂₀, [²H₂]GA₂₄, [²H₂]GA₂₉, [²H₂]GA₃₄, [²H₂]GA₄₄, [²H₂]GA₅₁, [²H₂]GA₅₃). The supernatant was collected and purified using mixed mode reversed-phase and ion-exchange cartridges (Waters, Milford, MA, United States) and analyzed by UPLC-MS/MS (Micromass, Manchester, United Kingdom). GAs were detected using a multiple-reaction monitoring mode of the transition of the ion [M–H]^– to the appropriate product ion. Masslynx 4.2 software (Waters, Milford, MA, United States) was used to analyze the data.

Exogenous application experiments were conducted according to the early report (Lu et al., 2014). Cells were synchronized by alternating light/dark (12/12 h) cycles with a final extended dark period. Synchronous cells were added with GA₃ or GA₄₊₇ at a gradient concentration (0.008, 0.08, 0.8, 8 μg⋅L^–1) or an equivalent amount of ethanol (not to exceed a final concentration of 0.1%). Samples were taken at the indicated periods for further analysis.

Genome sequences for dinoflagellates were retrieved from http://reefgenomics.org. Proteins from all genomes were blasted with the validated Arabidopsis proteins. The sequences that were ultimately selected are listed in Supplementary Dataset 1. All the experiments were carried out in triplicate, statistical analyses were done using SPSS 18.0 software, standard deviation analysis was used to compare mean values of replicate data sets. Significant difference analysis was undertaken by using Tukey’s test in Graph Prism Two-Way ANOVA.