Fennel (Foeniculum vulgare Mill.) seeds provided by Asia Seed Company (Asia Seed Co., Seoul, Korea) were germinated in a growth chamber under dark conditions for two weeks at the Sangji University at a mean temperature of 20 °C and 70% humidity. The seedlings were transplanted into rock wool cubes (40 × 40 × 40 mm) for hydroponics, and the plants were grown in a plastic film covered greenhouse (150 × 300 × 150 mm) using a deep flow technique with a mean temperature of 27 °C, mean humidity of 80%, and 12/12-h light/dark photoperiod. We used LED modules (MSIP-REI-3SI-SIF-300, OSRAM, Seoul, Korea) as light sources in the greenhouse. CO₂ was supplied to each greenhouse at concentrations of 400 ppm or 800 ppm. We observed changes in CO₂ temperature and humidity in the plants under the different CO₂ conditions. We measured the temperature and humidity of the greenhouse using an Arduino AM2302 sensor ( Figure S1A ). CO₂ was measured using a CM1107 as dual beam non-dispersive infra-red detector from CUBIC (Wuhan, China). The CO₂ sensor is able to measure CO₂ concentrations in the 0–5,000 ppm range, with an error of ± 30 ppm ( Figure S1A ). A 20-L aluminum cylinder tank with a gas regulator was used to supply CO₂, which had a pressure gauge inside the bottle, and a bubbler counter and a power adapter ( Figure S1B ). A constant CO₂ concentration of 400 ppm or 800 ppm was maintained using an Arduino UNO R2 kit with an ATMEGA4809 microcontroller (Microchip Technology Inc., Shanghai, China), and the constant CO₂ was supplied both day and night. The environmental data detected using the Arduino sensors were transmitted to an IBM 3500 computer in real time at one-minute intervals. The nutrient solution was used in the hydroponics systems at weekly intervals according to previous studies (An and Lee, 1991; Kim and Na, 2018). The nutrient solution comprised 606 mg KNO₃, 115 mg NH₄H₂PO₄, 236 mg Ca (NO₃)₂·4H₂O, 246 mg MgSO₄·7H₂O, 22.6 mg Fe-EDTA, 2.9 mg H₃BO₃, 1.8 mg MnSO₄·4H₂O, 0.2 mg ZnSO₄·7H₂O, 0.08 mg CuSO₄·5H₂O, and 0.03 mg NH₄M₀O₄·2H₂O. All chemical components of the nutrient solution were obtained from DAEJUNG CHEMICAL and METALS company (Gyeonggi-do, Korea). To observe plant growth under different light intensities, fennel was grown in the greenhouses under different light intensities of 12 lx and 24 lx with a mean temperature of 26 °C, mean humidity of 30%, and 12/12-light/dark photoperiod. The light intensity was adjusted in the dark greenhouse by controlling the number of the LED modules. To confirm the experimental light intensities, we measured the light intensities using a DT-92 mini light meter (CEM Co., Shenzhen, China). Humidity and temperature were measured in the greenhouses using an AM2302 sensor (Aosong Electronic Co., Guangzhou, China), which has a measuring error of ± 2% humidity, via a polymer humidity capacitor. To observe the nutrient solution changes, we measured the electrical conductivity (EC), pH, and dissolved oxygen (DO) by Lutron WA-2017SD Multi Water Quality Meter (Lutron electronic enterprise Co., Coopersburg, PA, USA).

To compare plant growth under different CO₂ conditions, we observed plant length, stem diameter, number of tillers, SPAD, and above-ground weight in fennels cultivated under CO₂ concentrations of 400 or 800 ppm five weeks after planting. Plant length was measured from the rock wool cube to the top part, and stem diameter was measured using Vernier calipers. The SPAD value was measured using a SPAD meter (SPAD-502plus, Minolta, Japan). All measurements were performed in triplicates. The above-ground weights were measured using the leaf and stem tissues excluding roots, before and after dehydration, using a drying machine at 60°C for 24 h.

RNA was extracted from leaf tissues subjected to 400 and 800 ppm CO₂ concentrations using the Accuprep^® Universal RNA Extraction Kit with three replicates. RNA quality was estimated using a spectrophotometer (NanoDrop-nabi, MicroDigital, Seongnam-si, Korea). The RNA extracted from the six samples were subjected to library preparation according to the manufacturer’s instructions, and then sequenced using Illumina NovaSeq 6000 (Illumina Inc., San Diego, CA, USA) with paired-end reading. To generate the genome information of fennels, we used merged raw RNA sequencing data for de novo transcriptome assembly using Trinity assembler (https://github.com/trinityrnaseq/trinityrnaseq). The assembled contigs were used to find transcripts, including open reading frames over 100 amino acids, using Trans Decoder (https://github.com/TransDecoder/TransDecoder). The annotation of long open reading frames was performed by BLAST (Altschul et al., 1990) using UniProtKB/Swiss-Prot data (https://www.ebi.ac.uk/uniprot/download-center). The redundant transcripts were removed using the CD-hit program (Li and Godzik, 2006). Gene expression levels were estimated using the Trinity Perl script, and then the differentially expressed genes (DEGs) between plants under 400 ppm and 800 ppm CO₂ conditions were identified using the edgeR method (Robinson et al., 2010) with a threshold false discovery rate (FDR) of 0.05. Functional enrichment analysis (FDR< 0.05) of DEGs among the total genes detected in fennels was conducted using DAVID (https://david.ncifcrt.gov/). Gene set enrichment analysis was conducted using the R package clusterProfiler (Yu et al., 2012), enrichplot (https://bioconductor.org/packages/release/bioc/html/enrichplot.html), and pathview (Luo and Brouwer, 2013), with Arabidopsis thaliana gene sets, because the functional information of many genes in fennel are still unknown.

Plant extracts were collected from fennel leaves after the drying process according to a previously described method (Kao et al., 2011). The dried leaves and stems (0.5 g) were extracted with 25 mL 99% methanol (Daejung, Korea) at 58°C for 24 h using a shaking incubator (ED-SI300R, HYSC, Korea), and the supernatants were then collected from plant extracts using a centrifuge (Allegra X-30R Centrifuge, BECKMAN COULTER, USA) at 1300 RCF (×g) for 10 min. Fennel methanol extract was used to measure antioxidant activity.

TPC was measured using the Folin-Ciocalteu method (Slinkard and Singleton, 1977; Odabasoglu et al., 2004) with some modifications. A mixture of 50 μL fennel extract and 50 μL Folin-Ciocalteu’s phenol reagent (Sigma-Aldrich, Burlington, MA, USA) was incubated for 6 min in the dark at room temperature of around 25 to 27°C. The reactant was mixed with 0.5 mL of 7% sodium carbonate solution (Daejung) and 250 μL of distilled water, and then incubated for 90 min under dark conditions. The absorbance of the solution was measured at 760 nm using a spectrometer (OPTIZEN POP, KLAB, Korea). The standard curve equation was generated using gallic acid (Daejung) as the standard at different concentrations, and used for the quantification of TPC ( Figure S3 ).

TFC was measured using a previously described method (Zhishen et al., 1999) with some modifications. A mixture of 50 μL fennel extract and 100 μL 5% sodium nitrate solution was incubated for 6 min at room temperature of around 25 to 27°C, and 150 μL of 10% aluminum chloride solution (Daejung) was added. After 5 min, 200 μL of 1 M sodium hydroxide solution (Daejung) was added, and the mixture was incubated for 1 h at 37°C. The absorbance of the solution was measured at 510 nm using a spectrophotometer. A TFC standard curve was obtained using quercetin (Sigma-Aldrich) for fennel methanol extract ( Figure S3 ). The calibration curve was generated based on different concentrations of quercetin.

Free radical scavenging (DPPH) was estimated according to a procedure described previously (Brand-Williams et al., 1995) with some modifications. A mixture of 100 μL plant extracts and 900 μL 0.1 mM DPPH solution was incubated for 10 min at room temperature of around 25 to 27°C, and the absorbance of the solution was then measured at 515 nm using a spectrometer. Gallic acid was used as the standard, and the standard curve was generated as described above ( Figure S3 ). The control absorbance was measured using a 0.1 mM DPPH solution. DPPH scavenging activity was calculated for each treatment as follows:

Nitrate scavenging activity was estimated using a method described previously (Kato et al., 1987) with some modifications. A mixture of 50 μL of plant extract, 50 μL of sodium nitrite solution (Daejung), and 300 μL of 0.1 N HCl (pH 1.2) was incubated for 1 h at 37°C, and 1 mL of 2% acetic acid solution and 100 μL of Griess reagent (Sigma-Aldrich) were then added. After incubation for 15 min at room temperature of around 25 to 27°C, the absorbance of the solution was measured at 520 nm using a spectrometer. Nitrate scavenging activity was calculated as follows for each treatment:

where A represents the absorbance of the sodium nitrite solution containing Griess reagent and plant extracts, B represents the absorbance of the sodium nitrite solution containing Griess reagent, and C represents the absorbance of the sodium nitrite solution containing distilled water and plant extracts.

The FRAP assay was performed as described previously (Benzie and Strain, 1996) with some modifications. The FRAP reagent was composed of 300 mM acetate buffer (pH 3.6; Sigma-Aldrich), 10 mM 2,4,6-tripyridyl-s-trianzine (Sigma-Aldrich), and 20 mM ferrous chloride (Sigma-Aldrich) at a 10:1:1 (v/v/v) ratio. Plant extracts (30 μL) were mixed with 900 μL of FRAP reagent shortly before the FRAP assay was performed. After 10 min at 37°C, the solution absorbance was measured at 590 nm using a spectrometer. Gallic acid was used as the standard, and the standard curve was plotted as described above. The FRAP was quantified based on the standard curve equation of Gallic acid ( Figure S3 ).

The reducing power assay was performed as described previously (Oyaizu, 1986), with some modifications. A mixture of 300 μL of plant extract, 300 μL of 200 mM phosphate buffer (pH 6.6; Sigma-Aldrich), and 300 μL of 1% potassium ferricyanide (Sigma-Aldrich) was incubated in a water bath (JSWB-22(T), JSR, Korea) for 20 min at 50°C. The reactants were mixed with 300 μL of 10% trichloroacetic acid (Sigma-Aldrich) solution and then centrifuged at 13,000 rpm using a Centrifuge 5415 C machine (Eppendorf, Germany). Ferric chloride (300 μL; Sigma-Aldrich) was added to 300 μL of the supernatant, and the absorbance of the solution was measured at 700 nm using a spectrometer.

The CO₂-treated groups (400 or 800 ppm) were analyzed to determine significant differences using two-tailed Student’s t-test. The statistical data were analyzed using MS Excel 2016 (Microsoft Corp., Redmond, WA, USA). The results are presented as the mean along with the standard deviation from three independent biological replications. The equations and R-squared value of standard curves were determined using MS Excel 2016 (Microsoft).

To evaluate the effects of eCO2 on plant growth, we observed the phenotypic changes in fennels under 400 or 800 ppm CO2 conditions ( Figure 1 ). The fennels showed distinct growth characteristics five weeks after planting ( Figure 1A ). The relatively high CO₂ concentrations promoted shoot and root development, with fennels exhibiting increased shoot length and root phenotype. There were significant differences in plant length, stem diameter, SPAD value, and tissue weight in fennel plants between the 400 and 800 ppm groups. However, there were no significant differences in number of tillers ( Figure 1B ). In particular, the SPAD value in the 800 ppm group was double that in the 400 ppm group (45.87 vs. 25.85, respectively). The fresh weights of leaves and stem tissues were largely higher in the 800 ppm group, at 43.86 and 77.09 g, respectively, when compared with 12.97 and 11.24 g, respectively, in the 400 ppm group. Plant length was significantly different between the 400 and 800 ppm groups (49.81 vs. 57.64 cm, respectively). Additionally, the dry weights of tissues increased significantly in fennels under the 800 ppm CO₂ condition. However, the differences in plant tissues between the fresh and dry weights decreased, indicating relatively high contents in the 800 ppm group. The results indicate that relatively high CO₂ concentrations promoted plant growth and leaf chlorophyll concentrations.

To determine the effect of CO₂ supply in hydroponics under different light conditions, we observed distinct changes in environmental factors (temperature and humidity) and nutrient solution composition (EC, pH, and DO) in the fennel plants for five weeks after planting ( Figure 2 ). The CO₂ concentrations in each sealed greenhouse were maintained using a customized CO₂ system ( Figure 2 ). In days without light, the temperature and humidity increased in the greenhouse under the 800 ppm conditions when compared with those under the 400 ppm condition. Furthermore, lower temperature and higher humidity were observed in the greenhouses under the dark condition than under the light condition. The EC was relatively high in late dark conditions for hydroponics (light condition: 1.53 vs. 1.54 mS for the 400 and 800 ppm groups, respectively, dark condition 2 vs. 2.2 mS for the 400 and 800 ppm groups, respectively). In addition, the pH in the nutrient solution for the 800 ppm group was higher than that in the 400 ppm group. The DO decreased under light conditions in the 800 ppm group when compared with those in the 400 ppm group. These results indicate that the environment and nutrient solutions were changed considerably in fennel under the dark condition owing to distinct photosynthesis and transpiration characteristics based on differences in light intensity and eCO2.

To survey transcriptional changes in response to different CO₂ concentrations, we conducted RNA sequencing analysis in fennels under the 400 and 800 ppm treatments. We assembled a total of 174,266,880 bases with an average contig length of 1250, and 139,352 transcripts were detected (data not shown). The 31,737 assembled contigs were analyzed to determine the DEGs, and the assembled contigs were highly correlated in each experiment under the 400 or 800 ppm conditions ( Figure S2A ). A total of 4,965 DEGs, including 1,776 downregulated and 3,189 upregulated genes, were detected in fennels cultivated under the different CO₂ concentrations ( Figure S2B ).

To determine the significantly altered gene functions of DEGs, we conducted a functional enrichment analysis of the upregulated and downregulated genes in the 800 ppm group ( Figure 3 ). We detected 17 groups (14 and 3 groups for the upregulated and downregulated DEGs, respectively) of overrepresented Gene Ontology terms (FDR< 0.05), including cellular component, biological process, and molecular function. Many groups were associated with photosynthetic mechanisms in chloroplasts, thylakoid lumen-related cellular component, photosystem-related functions, and response to light. In addition, the DEGs involved in lipid catabolic process, translation-related functions, hydrolase activity, and flavonoid biosynthetic process were significantly upregulated in fennels under the 800 ppm condition. However, 3 ethylene-activated signaling pathway, transcription, and syncytium formation were downregulated in the 800 ppm group.

To detect overrepresented pathways in a priori-defined gene set, we conducted gene set enrichment analysis ( Figure 4 ) and determined the connections and regulation of genes involved in prominent Gene Ontology biological processes (pigment biosynthetic process, response to far-red/red light, response to Karrikin, and syncytium formation) ( Figure 4A ). The DEGs of the four pathways were mainly upregulated in the 800 ppm group, in which 39 DEGs were upregulated and six were downregulated. The eight DEGs related to syncytium formation showed four upregulated and four downregulated genes. To determine whether relatively high CO₂ concentration (800 ppm) affected the biomass increase in fennels, we observed the gene expression patterns under flavonoid biosynthesis ( Figure 4B ). Notably, several genes involved in the flavonoid biosynthetic pathway were upregulated, whereas no genes were downregulated, suggesting flavonoid content increase under 800 ppm.

To confirm the possibility of biomass increase under the 800 ppm condition, we conducted antioxidant activity assays for TPC, TFC, DPPH, nitrite-scavenging, FRAP, and reducing power in the leaves and stems of fennels in the 400 and 800 ppm groups ( Figure 5 ). We confirmed that TFC was significantly increased in both the leaves and stems of fennel, confirming the upregulation of the flavonoid pathway-related DEGs. The TFC was 9.93 vs. 15.92 mg QE/mL for the 400 and 800 ppm groups in the leaves and 2.73 vs. 4.86 mg QE/mL for the 400 and 800 ppm groups, respectively, in the stems. However, the rest of the antioxidant activity assays showed significantly decreased activity in the leaves of the fennels in the 800 ppm group; compared to the 400 ppm group, the antioxidant activities in the 800 ppm group were decreased by 35.29% in the TPC assay (0.17 vs. 0.11 Mg GAE/mL for the 400 and 800 ppm groups, respectively), 50.76% in the reducing power assay (1.32 vs. 0.65 O.D. in the 400 and 800 ppm groups, respectively), 52.8% in the DPPH scavenging activity assay (88.66% vs. 39.48% in 400 and 800 ppm groups, respectively), and 35% in the FRAP assay (2.2 O.D. vs. 1.43 O.D. in the 400 and 800 ppm groups, respectively). Although the plants in the 800 ppm group exhibited slightly increased nitrite-scavenging activity, no significant differences were detected between the stems and leaves of the fennels.

As the plants grew in the hydroponics systems, the fennels were exposed to strong light intensity because of the limited spatial extent between artificial light and growth space. The length of fennel increased under the 800 ppm CO₂ condition, as mentioned above ( Figure 1B ). Therefore, we investigated whether the improved plant traits were more strongly affected by light intensity or eCO₂ concentrations, by observing plant growth and antioxidant activities in fennel under different light intensities (12 lx and 24 lx) ( Figure 6 ). When we observed changes in environmental factors and nutrient solution composition, there were no notable differences between 12 lx and 24 lx ( Figure S4 ). We confirmed increased humidity, EC, and DO, and decreased temperature and pH in the greenhouse under dark condition regardless of light intensity. For a day, the distinct changes between light and dark conditions was similar to eCO₂. Furthermore, we examined six plant traits, including SPAD, plant length, stem thickness, tiller number, and fresh and dry weights of fennel under the same concentrations (230 ppm) of CO2 ( Figure 6A ). We found significantly increased (p< 0.001) SPAD units in fennel under the 24 lx conditions than under the 12 lx conditions. No significant differences in other plant traits were observed in fennel under the two light conditions. Four weeks after planting, the average lengths of the plants were 32 cm and 48 cm in the 400 and 800 ppm groups, respectively (data not shown). However, the fennel had short plant lengths under different light conditions (22.33 cm and 21.43 cm under the 12 lx and 24 lx conditions, respectively) because of the low CO₂ concentrations in the growth room. During the same period of plant growth (four weeks after planting), unlike eCO₂ concentration, the different light conditions did not affect the lengths of the fennel plants or the antioxidant activities of the plant extracts ( Figure 6B ). The results suggest that CO₂ has a stronger effect on plant improvement than light intensity.