Plant buffering against the high-light stress-induced accumulation of CsGA2ox8 transcripts via alternative splicing to finely tune gibberellin levels and maintain hypocotyl elongation

In plants, alternative splicing (AS) is markedly induced in response to environmental stresses, but it is unclear why plants generate multiple transcripts under stress conditions. In this study, RNA-seq was performed to identify AS events in cucumber seedlings grown under different light intensities. We identified a novel transcript of the gibberellin (GA)-deactivating enzyme Gibberellin 2-beta-dioxygenase 8 (CsGA2ox8). Compared with canonical CsGA2ox8.1, the CsGA2ox8.2 isoform presented intron retention between the second and third exons. Functional analysis proved that the transcript of CsGA2ox8.1 but not CsGA2ox8.2 played a role in the deactivation of bioactive GAs. Moreover, expression analysis demonstrated that both transcripts were upregulated by increased light intensity, but the expression level of CsGA2ox8.1 increased slowly when the light intensity was >400 µmol·m−2·s−1 PPFD (photosynthetic photon flux density), while the CsGA2ox8.2 transcript levels increased rapidly when the light intensity was >200 µmol·m−2·s−1 PPFD. Our findings provide evidence that plants might finely tune their GA levels by buffering against the normal transcripts of CsGA2ox8 through AS.


Introduction
Gibberellins (GAs) are known to be the dominant hormone governing hypocotyl elongation 1,2 . GAs regulate cell expansion, thus causing changes in hypocotyl length or plant height via the GA signaling pathway. GA-GID1-DELLA is a widely accepted signaling module through which bioactive GAs regulate plant growth and development 2 . For example, GIBBERELLIN INSENSITIVE DWARF 1 (GID1)-inactivation mutants show reduced plant height 3,4 , and substitution of proline to leucine in DS-3, which encodes a DELLA protein, results in semidwarf oilseed rape plants 5 .
Light is among the most essential factors for plant growth and development 6 . Seedlings grown under dark conditions are characterized by relatively long hypocotyls compared with those grown under light conditions 7,8 . One of the reasons for this is that under light conditions, the GA level decreases, while under dark or shaded conditions, the GA level increases [9][10][11] . Most related studies have focused mainly on low-light (shading)-induced GA biosynthesis for signaling 12 , and only a few studies have shown high-light-induced excessive GA deactivation. However, it remains unknown how plants respond to high-light stress and finely tune their GA levels to maintain growth.
Alternative splicing (AS) is a widespread mechanism in eukaryotes in which different protein variants are generated from a single gene via different combinations of splicing sites. AS is a prevalent approach in which plants respond to environmental stress, but it is counterintuitive why plants would invest in protein synthesis under conditions of declining energy supplies 13,14 . Chaudhary et al. proposed that, under stress conditions, plants buffer against normal protein synthesis levels via AS to decrease the translation of a significant portion of the transcriptome and produce the protein isoforms necessary for adaptation to stresses; however, the mechanisms of this process are not yet known 14 .
In this study, we performed RNA-seq to identify differentially expressed genes and AS events in cucumber (Cucumis sativus L.) seedlings grown under different light intensities. We focused on GA-related genes and found a novel transcript of Gibberellin 2-beta-dioxygenase (CsGA2ox8), CsGA2ox8.2, with intron retention (IR) between the second and third exons. Gibberellin 2-betadioxygenases compose a family whose members can catalyze the deactivation of bioactive GAs or their precursors 15 . Most studies have shown that increasing transcripts of GA2oxs can reduce plant height by regulating endogenous bioactive GA levels [15][16][17][18][19][20] . We found that under high-light stress, cucumber seedlings tended to generate more nonfunctional CsGA2ox8.2 but not functional canonical CsGA2ox8.1. In other words, the plants buffered against normal transcript levels of CsGA2ox8 via AS to limit the inactivation process of GAs under highlight stress, and this strategy may allow seedlings to maintain a certain GA level and sustain hypocotyl elongation.

Results and discussion
CsGA2ox8 expression is upregulated in response to increasing light intensity We recorded the changes in cucumber hypocotyl lengths after the cotyledons emerged from the soil under treatments involving two light intensities: X1 (180 µmol·m −2 ·s −1 PPFD) and X2 (40 µmol·m −2 ·s −1 PPFD). The hypocotyl lengths of seedlings grown under treatment X1 were shorter than those grown under treatment X2 (Fig. 1a), and a significant difference was observed after 40 h of light treatment (Fig. 1b). To investigate the molecular differences of seedlings grown under the two light conditions, a transcriptome analysis was performed. In total, 279 upregulated and 321 downregulated genes were identified between the two treatments ( Fig. 1c, Supplementary Table S1). In this study, we focused on GA-related genes because bioactive GAs have a large effect on hypocotyl elongation in cucumber 1 . We found that CsGA2ox8, a gene whose product catalyzes the deactivation of bioactive GAs, was significantly upregulated in treatment X1 compared to treatment X2. This expression difference was confirmed by quantitative reverse-transcription PCR (qRT-PCR) (Fig. 1d). These results suggested that the expression of CsGA2ox8 might be affected by light intensity. To test this hypothesis, we treated young cucumber seedlings with seven different light intensities (40,80,120,160,200,240, and 280 µmol·m −2 ·s −1 PPFD) and quantified the expression of CsGA2ox8 in seedlings grown under these conditions. As shown in Fig. 1e, the transcript level of CsGA2ox8 increased with increasing light intensity. Pioneering studies have shown that light modulates photomorphogenesis-related hypocotyl elongation mainly by interacting with GAs, and many genes are involved in this light-GA system 10 . Our results suggested that the expression of the GA deactivation factor CsGA2ox8 was upregulated by increased light levels, which might be involved in light-regulated hypocotyl elongation.

Differential AS (DAS) analysis for identifying novel transcripts of CsGA2ox8
AS is a prevalent approach used by plants in response to stress and involves the generation of multiple transcripts from the same gene 21 . In this study, five types of AS events, namely, RIs (retained introns), A5SSs (alternative 5′ splice sites), A3SS (alternative 3′ splice sites), MXEs (mutually exclusive exons), and SEs (skipped exons), were identified by RNA-seq analysis (Fig. 2a, Supplementary  Table S2). Among them, SEs and RIs accounted for more than 25% of all AS events (Fig. 2b), which is similar to that reported in Arabidopsis 22,23 .
Among all RI events, we found a novel transcript of CsGA2ox8 with an IR between the second and third exons. The original transcript of CsGA2ox8 was defined as CsGA2ox8.1, and the isoform with IR was defined as CsGA2ox8.2 (Fig. 2c). To validate the existence of CsGA2ox8.2, the coding regions of CsGA2ox8 of plants under treatment X2 were amplified and then sequenced. Two distinct bands were observed after RT-PCR-based amplification (Fig. 2c). After sequencing, the two amplicons were found to be 1056 and 1655 bp in length (Fig.  2c). Sequence comparison showed the same result as that after RNA-seq, and a schematic diagram of this event is illustrated in Fig. 2c. RIs promoted the disruption of the open reading frame of CsGA2ox8, resulting in an mRNA with a premature stop codon (TGA), and leading to a truncated protein (Fig. 2c). Compared with other GA2ox proteins in Arabidopsis, the AtGA2ox7 and AtGA2ox8 proteins have a unique region 15 . Alignment of the amino acid sequences of AtGA2ox8 and CsGA2ox8.1 revealed that the CsGA2ox8.1 protein had a conserved motif (Fig.  2d, denoted by a red box), which may define the specificity of the reactions catalyzed by these enzymes 15 . The absence of amino acids in CsGA2ox8.2 is indicated using a purple line in Fig. 2d. The asterisks indicate significant differences in hypocotyl lengths between X1 and X2 (P < 0.05). c Volcano plot of the transcriptome between X1-and X2-treated cucumber seedlings. GA2ox genes function in reducing GAs [15][16][17][18][19][20]24 . Given that CsGA2ox8.2 lost the third and fourth exons from CsGA2ox8.1 (Fig. 2c, d), we questioned whether CsGA2ox8.2 still maintains the same function as that of CsGA2ox8.1. To verify this, CsGA2ox8.1 and CsGA2ox8.2 were overexpressed in Arabidopsis Col-0. CsGA2ox8.1 overexpression lines (OE8.1) had shorter hypocotyl lengths than Col-0 under both light and dark conditions (Fig. 3a-c). The RT-PCR results suggested that GA2ox8.1 was overexpressed in OE8.1 but not in Col-0 (Fig. 3d), suggesting that GA2ox8.1 was functional for hypocotyl elongation. However, the hypocotyl length of CsGA2ox8.2 overexpression lines (OE8.2) was not significantly different from that of Col-0 ( Fig. 3e-g), even though the transcript level of GA2ox8.2 was as high as that of GA2ox8.1 (Figs. 3d, h). Next, the GA levels were quantified in the Col-0, OE8.1, and OE8.2 lines. Based on the GA synthetic pathway, GA2ox genes encoding enzymes that convert GA 12 , GA 9 , GA 53 , GA 20 , GA 1 , and GA 4 to GA 110 , GA 51 , GA 97 , GA 29 , GA 8 , and GA 34 , respectively 25,26 (Fig. 3i). In this study, we found that GA 12 , GA 9 , GA 53 , GA 20 , GA 1 , and GA 4 significantly decreased in OE8.1 compared with Col-0, and no significant differences in GA levels were detected between OE8.2 and Col-0 (Fig. 3i), suggesting that CsGA2ox8.1 is a functional transcript but that CsGA2ox8.2 is not.
AS is a prevalent approach in which plants respond to environmental stress. One hypothesis is that plants buffer against normal protein synthesis levels via AS to decrease the translation of a significant portion of the transcriptome under stress conditions 14 . Therefore, for future studies, it is important to know how CsGA2ox8.1 and CsGA2ox8.2 work at the protein level.
The homolog of CsGA2ox8 in Arabidopsis is AtGA2ox8, whose product functions in hydroxylating the C 20 -GA precursors GA 12 and GA 53 15 . In this study, consistent with the results from overexpression of AtGA2ox8 15 , GA 12 and GA 53 levels dramatically decreased in OE8.1, suggesting that CsGA2ox8 has a function similar to that of AtGA2ox8. However, we found that the content of GA 9 , GA 20 , GA 1 , and GA 4 also decreased, which is inconsistent with the results of Schomburg et al. (2003), who reported that GA 9 , GA 20 , and GA 4 levels increased in AtGA2ox8 overexpression lines 15 . GA 12 is the precursor of GA 9 and GA 4 , and GA 53 is the precursor of GA 20 and GA 1 25 . Therefore, it is possible that decreasing GA 12 and GA 53 levels would affect the synthesis of GA 9 , GA 4 , GA 20 , and GA 1 . Taken together, these results suggest that CsGA2ox8.1, rather than its CsGA2ox8.2 isoform, regulated hypocotyl elongation, and this regulation was achieved mostly via decreasing endogenous GA levels.
Plants tend to generate nonfunctional CsGA2ox8.2 but not functional CsGA2ox8.1 under high-light stress As one type of AS, IR mostly results in the production of nonfunctional transcripts carrying premature termination codons targeted for degradation via the nonsensemediated mRNA decay pathway 27,28 . However, it is unclear why plants perform AS under stress conditions. Given that CsGA2ox8.2 was detected under relatively high-light conditions and that CsGA2ox8.2 seemed to be a nonfunctional transcript, we hypothesized that CsGA2ox8.2 was generated for buffering against normal CsGA2ox8.1 transcripts under high-light stress to finely tune GA levels and maintain plant growth. Chaudhary et al. posited the same idea and proposed that, under stress conditions, plants buffer against normal protein synthesis levels via AS to decrease translation of a significant portion of the transcriptome, but the mechanisms of this process have not yet been elucidated 14 .
To test our hypothesis, the relative abundance of the two mRNA forms of CsGA2ox8 in plants under 50, 100, 150, 200, 400, 800, and 1200 µmol·m −2 ·s −1 PPFD was analyzed by RT-PCR. The results showed that both transcripts were positively regulated by light, and CsGA2ox8.1 was the dominant transcript under all light conditions (Fig. 4a). However, the expression level of CsGA2ox8.1 increased slowly when the light intensity was >400 µmol·m −2 ·s −1 PPFD, while that of CsGA2ox8.2 increased rapidly when the light intensity was >200 µmol·m −2 ·s −1 PPFD (Fig. 4a). Furthermore, the inclusion/exclusion (In/Ex) ratio (CsGA2ox8.2/ CsGA2ox8.1) significantly increased when the light intensity was >400 µmol·m −2 ·s −1 PPFD (Fig. 4b). In other words, the CsGA2ox8.2 expression level was negligible under low-light conditions, but under increasing light intensity, especially that corresponding to high-light stress, plants tended to generate more nonfunctional CsGA2ox8.2 transcripts and limit the increase in functional CsGA2ox8.1 levels (Fig. 4c). In this way, plants could produce less functional GA2ox and could maintain a certain level of GA for hypocotyl elongation under highlight conditions. AS is known to promote plant stress tolerance 13 . For example, stress imposed by high salinity promotes the use of noncanonical splice sites 29 . In addition, AS events finely tune gene expression by regulating the ratio of active and nonactive isoforms 30 . Similarly, we showed that high-light stress increased the CsGA2ox8.2/CsGA2ox8.1 ratio in this study, suggesting a new mechanism for the GAdeactivating-enzyme response to high-light stress. However, it is not clear how high light modulates AS events. To our knowledge, AS events are mediated by core spliceosomal components (small nuclear ribonucleoproteins) as well as RNA-binding proteins (serine/arginine-rich proteins and heterogeneous nuclear ribonucleoproteins) 30 . One hypothesis is that high-light stress affects the expression or protein structure of spliceosome genes, thereby modulating AS events at the whole-genome level.
Our results suggest that IR may have been employed to finely tune gene expression by altering the ratio between functional and nonfunctional variants. Following a BLAST search of The Arabidopsis Information Resource (TAIR, https://www.arabidopsis.org/) and the MSU Rice Genome Annotation Project Database (http://rice. plantbiology.msu.edu/index.shtml), AT4G21200 and LOC_Os02g41954 were identified has sharing high sequence homology (57.42% and 49.03%, respectively) with that of CsGA2ox8. As shown in TAIR, AT4G21200 has one splice variant with an IR between the second and third exons, which is the same as CsGA2ox8.2, with the RI promoting the disruption of the open reading frame of AT4G21200, resulting in the generation of an mRNA with a premature stop codon (TGA), leading to a truncated protein 23 . From the rice database, LOC_Os02g41954 also has one splice variant, which has a premature stop codon (TGA) between the second and third exons 30 . These findings suggest that AS of GA2ox8 commonly occurs in plants, and this strategy might also be used by other plant species to buffer against the stress-responsive transcriptome and promote efficient adaptation to stress conditions.

Plant materials and growth conditions
Cucumber inbred line 9930 was used in this study. Seeds were surface sterilized and soaked in water for 3 h, after which they were then transferred to Petri dishes lined with wet filter papers and germinated at 28°C.
The germinated seeds were then sown in autoclaved soil at 24°C. Single cucumber seedlings with cotyledons emerging from the soil were chosen for light treatment. A. thaliana Col-0 was used in this study. The seeds were disinfected and germinated on Murashige-Skoog (MS) media consisting of 1% sucrose and 0.2% Phytagar at 4°C for 3 days before being moved into a growth chamber at 22°C 31 .

Light treatment and hypocotyl measurements
Light intensities were measured using a Lighting Passport TM ALP-01 Handheld Spectrometer (HESON, Shanghai, China). Cucumber seedlings were grown in a greenhouse with "DE-HPS 1000 W" lights. To maintain the light conditions within an appropriate range, the plants were placed at different heights below the fixtures. For example, to generate a high-light condition, the plants were moved closer to the lamps; to generate low-light conditions, the plants were moved farther from the lamps or were covered with film. For Arabidopsis, the plates with disinfected seeds were treated at 4°C for 72 h and then placed under white light for 10 h to induce uniform germination. Afterward, the plates were transferred to dark or light conditions and then incubated at 22°C for 3-6 days for hypocotyl measurements. The hypocotyl lengths of the seedlings were measured using a tapeline.

GA level analysis
The GA level was analyzed as previously reported, with moderate changes 32 . Briefly, for product purification, 0.5 g of the sample was quickly immersed in liquid nitrogen, homogenized, and extracted with 5 mL of precooled 80% methanol for 20 h at 4°C. Afterward, the samples were shaken at 4°C for 30 min and then centrifuged at 18,000 × g for 20 min at 4°C. The supernatant was collected and immersed in liquid nitrogen. Then, the samples were dissolved in 100 µL of chromatographygrade methanol supplemented with 0.1 M glacial acetic acid and filtered through a 0.22 µm filter. Highperformance liquid chromatography-based analysis was performed using an Agilent (CA, USA) Zorbax SB-C 18 column (5 µm, 4.6 × 250 mm). The mobile phase A solvent consisted of 100% methanol, and mobile phase B was 0.1 M acetic acid. The gradient elution was performed as follows: 0-40 min, 3% A and 97% B; after 40 min, 67.6% A and 32.2% B. The flow rate was 1 mL·min −1 , and the column temperature was set at 30°C. Chromatograms were quantified using the Agilent chromatography workstation. Each sample involved three replicates from independent experiments. To validate the GA2ox8 AS isoforms, seedlings were treated with the same light conditions as those applied for RNA-seq (X1, 180 µmol·m −2 ·s −1 PPFD; X2, 40 µmol·m −2 ·s −1 PPFD) for 56 h, after which the hypocotyls were harvested for RNA extraction, cDNA synthesis, and PCR-based analysis. The primers used are listed in Supplementary Table 3.

Sequence alignment
Sequence alignments were performed based on the amino acid sequences of CsGA2ox8.1, CsGA2ox8.2, and AtGA2ox8. Multiple sequence alignment was performed using DNAMAN for Windows (Lynnon Corporation, Quebec).

Gene expression analysis
Gene expression analysis methods were used as previously reported 31 . Total RNA was extracted from the hypocotyls using an SV Total RNA Isolation System (Promega), and cDNA was synthesized using a Multi-Scribe TM Reverse Transcriptase Kit (Applied Biosystems) according to the manufacturer's instructions. Quantitative RT-PCR was performed using a SYBR ® Premix Ex Taq TM Kit (TaKaRa, China) in conjunction with an Applied Biosystems 7500 Real-time PCR system. For RT-PCR, PCR primers were designed to amplify two isoforms of different sizes, and their RNA products were quantified and reported as an In/Ex ratio 38 . Cucumber α-TUBULIN (TUA) and Arabidopsis ACTIN 2 were used as internal controls 39 . The primers used are listed in Supplementary  Table S3.

Transformation of A. thaliana
Transformation assays were performed as previously described 31 . To generate CsGA2ox8.1 and CsGA2ox8.2 overexpression constructs, full-length CsGA2ox8.1 and CsGA2ox8.2 cDNAs were cloned and inserted into a pBI121 vector between the XbaI and SmaI restriction sites 40 . The constructs were then introduced into Agrobacterium tumefaciens strain LBA4404 by electroporation, which were then transformed into wild-type A. thaliana (Col-0), as previously described 41 . The transgenic plants were screened on MS media supplemented with 50 mg/L kanamycin and then confirmed via PCR. The primers used are listed in Supplementary Table S3.

Statistical analysis
All the treatments mentioned in this study involved at least three independent biological and technical replicates. The results were analyzed using analysis of variance. All the analyses were carried out using Statistics Analysis System 15.1 for Windows.