Enhanced growth and productivity of useful metabolites by indole-3-propionic acid treatment in Lemna aequinoctialis culture

Abstract

Background

Lemna aequinoctialis is a floating plant with potential uses as food, feed, biofuel, biomass, and pharmaceutical resources. In this study, we examined the effect of various concentrations of indole-3-propionic acid (IPA) on the growth, metabolomic, and transcriptomic profiles of L. aequinoctialis cultures.

Results

IPA significantly improved the growth of L. aequinoctialis culture and altered its metabolome and transcriptome profiles. It also increased the cellular production of aspartate, glutamate, glutamine, serotonin, tryptophan, and phytosterols. Transcriptomic analysis revealed that 7,490 genes were significantly altered by the 10 µM IPA treatment. Metabolomic and transcriptomic profiling indicated that the IPA treatment notably altered the major pathways of ‘glycerolipid metabolism’, ‘glutathione metabolism’, ‘β-alanine metabolism’, ‘alanine, aspartate, and glutamate metabolism’, and ‘starch and sucrose metabolism’. The highest productivities of serotonin, β-sitosterol, campesterol, and stigmasterol were observed on day 28 following treatment with 10 µM IPA.

Conclusions

The results of this study indicate that the application of IPA to L. aequinoctialis cultures significantly enhanced the growth and production of useful metabolites by affecting various metabolic pathways. This study is the first to investigate the biological effects of IPA on L. aequinoctialis, through an integrated analysis of metabolomics and transcriptomics. This could facilitate the large-scale cultivation of L. aequinoctialis for the application in various fields of agriculture and biotechnology including food, feed, and pharmaceutical industries.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-025-07103-7.

Background

Lemna aequinoctialis, commonly known as duckweed, is a small aquatic plant in the Lemnaceae family, and is known for its global distribution, rapid growth, and adaptability to various environmental conditions [1–4]. It has notable potential for environmental sustainability, biofuel production, and biochemical pathways, its high protein, fat, and fiber content, it could become an important source of human food [5–7]. L. minor is used in Europe for pet food production and consumed in Southeast Asia as a vegetable protein source [8]. Additionally, in the United States, a mixture of Lemna and Wolffia is sold as a powdered ingredient [8]. L. aequinoctialis contains primary and secondary metabolites, which possess various bioactivities for pharmaceutical application [6, 9].

Indole-3-propionic acid (IPA) is an indole-based compound produced by gut and soil microbes [10, 11]. Gut microbes such as Clostridium sporogenes, C. botulinum, Peptostreptococcus anaerobius, and C. cadebaris convert tryptophan to IPA via tryptophan deaminase [10–14]. IPA benefits gut health primarily by interacting with the gut microbiota [15]. It is also absorbed from the gut and translocated to the brain, where it neutralizes hydroxyl radicals and prevents the formation of β-amyloid fibers [15]. IPA has gained attention in research for its potential role in the prevention and treatment of neurological diseases such as Parkinson’s and Alzheimer’s [16]. Similar to melatonin, it scavenges radicals without producing reactive and oxidation-promoting intermediate compounds, unlike other antioxidants [17, 18]. A higher intake of fiber-rich foods has been linked with a decreased risk of type II diabetes as well as higher plasma concentrations of IPA [19, 20].

Compared to the extensive studies on IPA in humans, studies on the activity of IPA in plants are limited. In Arabidopsis thaliana, continuous administration of 10 µM IPA increased lateral root and root hair growth and activated the transcriptional auxin signaling pathway, which regulated physiological processes [21]. IPA also stimulates lateral root formation compared to other auxins in A. thaliana and activates cell division and induces auxin-like responses in tobacco BY-2 [22]. IPA-binding proteins have been identified in mung bean seedling tissue, and its binding sites are concentrated in the endothelial cell membrane [23, 24]. In L. aequinoctialis culture, melatonin treatment or optimization of light condition could enhance the growth and increase the accumulation of starch, phytosterols, and amino acids [25, 26]. However, the effects of IPA administration on the growth and metabolic and transcriptomic profiles of L. aequinoctialis culture remain to be investigated.

In this study, we hypothesized that IPA treatment could affect the growth and useful metabolites and transcripts profile of L. aequinoctialis culture. Metabolic and transcriptomic profiling of L. aequinoctialis culture treated with IPA were performed using gas chromatography-mass spectrometry (GC-MS) and RNA-seq analyses. The main aim of the study is to investigate the effect of exogenous IPA treatment on growth, metabolic, and transcriptomic profiles of L. aequinoctialis culture.

Materials and methods

Cultivation of L. aequinoctialis

L. aequinoctialis (PC-10,605) was cultured in vitro using plants sourced from the Korean Collection for Type Cultures (KCTC; Biological Resource Center, Jeongeup, Republic of Korea). We cultivated the L. aequinoctialis plant according to the methods in previous studies [27, 28]. Subcultures were conducted every 2 weeks. Before administering IPA, the cultures were pre-cultivated in 1 L of liquid medium under same environmental conditions for 3 days for adaptation.

IPA treatment

After adaptation, 30 fronds were inoculated into 100 mL in 200 mL flasks (Diamond, Republic of Korea) of liquid medium, excluding Gelrite. IPA was dissolved in water (HPLC grade, Burdick & Jackson, Muskegon, Michigan, USA), and each flask was treated with 0.5, 1, and 10 µM IPA. Each solution concentration was sterilized for 30 min, and filtration (13CP020AS, 0.2 µM pore size, Advantec, Tokyo, Japan) using 2 mL of syringe. L. aequinoctialis inoculated on the liquid medium was cultured under static conditions with a 16:8 h light/dark cycle at a temperature of 25 ± 1 °C and a light intensity of 81–84 µmol/m²/s. All experiments were performed in three biological replicates and plants were harvested on day 28.

Growth measurement

L. aequinoctialis fronds was counted every 7 days during cultivation. To determine the total dry weight, the harvested plants were placed on Whatman filter paper (Whatman, Maidstone, UK). After being washed with distilled water, the plants were allowed to dry for 15 min to remove any excess moisture. The samples were then transferred into tubes for deep freezing at − 80 °C for 24 h, followed by freeze-drying (Ilshinbiobase, Dongducheon, Republic of Korea) for 48 h. Finally, the total dry weight of the plants was measured.

Metabolite profiling by GC-MS analysis

The samples were stored after freeze-drying and ground into powder using a mortar and pestle. For GC-MS analysis, 20 mg of L. aequinoctialis samples were individually weighed and transferred into 2 mL tubes (Eppendorf, Hamburg, Germany). For the extraction method, according to the previously described [27]. GC-MS analysis and derivatization were performed according to the method described previously [28]. All experiments were performed in three biological, and three technical replicates. Human Metabolome Database (http://www.hmdb.ca/) and Golm Metabolome Database (http://gmd.mpimp/golm.mpg.de/) were used to assign the peaks obtained from GC-MS analysis.

Transcriptome analysis

The NucleoSpin RNA Plant Mini kit for RNA from the plant (Macherey-Nagel, #740949.50) was used for RNA extraction. The powdered sample (100 mg) was placed in a 1.5 mL microcentrifuge tube. Solution Lysis buffer 500 µL and β-mercaptoethanol 5 µL were added to the sample and vortexed for 15 s. The mixture was incubated at room temperature (23–25 °C) for 3 min. The lysate was centrifuged at 16,300 x g for 1 min, and 300 µL of the supernatant was transferred to a new 1.5 mL microcentrifuge tube. Membrane Lysis buffer 300 µL was added to the supernatant, vortexed for 15 s, and transferred to an EzPureTM filter. After centrifuging at 16,300 x g for 1 min, the mixture and 500 µL of the pass-through were transferred to a new 1.5 mL microcentrifuge tube. Absolute ethanol 250 µL was added to the supernatant and vortexed for 1 min. All components were added to the mini-column type F and centrifuged at 16,300 x g for 1 min. RNA Binding Wash buffer 500 µL was added to the mini-column and centrifuged at 16,300 x g for 30 s. DNase I reaction mixture 70 µL was placed onto the center of the mini-column for gDNA digestion, and the column was incubated for 10 min at room temperature. The column was centrifuged at maximum speed for an additional 1 min to remove any residual wash buffer. The mini-column was transferred to a new 1.5 mL microcentrifuge tube. Nuclease-free water 50 µL was added to the center of the membrane in the mini-column and centrifuged at 16,300 x g for 1 min. The RNA was stored at − 70 °C for long-term storage.

RNA sequencing was performed using Next Generation Sequencing-based assays. Using the TruSeq Stranded mRNA Library Prep Kit, the samples were prepared for 150-bp paired-end sequencing (Illumina, CA, USA). A total of 1,000 ng of RNA was fragmented, and single-stranded cDNA was synthesized using random hexamer priming. This cDNA served as the template for synthesizing double-stranded cDNA. After performing end repair, A-tailing, and adapter ligation, cDNA libraries were constructed. The quality of the libraries was evaluated using the TapeStation 4200 instrument and D1000 ScreenTape System (Agilent Technologies, CA, USA), and their concentration was measured with the KAPA Library Quantification Kit (Kapa Biosystems, MA, USA), following the manufacturer’s protocol. The denatured templates were subjected to cluster amplification, and sequencing was conducted in paired-end mode (2 × 150 bp) on the Illumina NovaSeq 6000 (Illumina, CA, USA).

Data processing and statistical analysis of metabolites and transcripts

The GC-MS data were processed with MS-DIAL (4.90 alpha version, RIKEN, Japan) for metabolite identification, following established methods [29–31]. Raw data processing and statistical analysis according to previously reported methods [28]. Metabolite raw data were converted to Abf files using Abf Converter (https://www.reifycs.com/AbfConverter/index.html) for compatibility with MS-DIAL. Tetramethylsilane mass peaks with accurate masses of 73 and 147 were excluded to improve peak detection. Retention time and m/z tolerance were set to 0.5 min and 0.5 Da, respectively. Area intensities of each peak were normalized by dividing those by the sum of total intensities of each peak. Statistical analyses, including the Kruskal–Wallis test, Mann–Whitney test, one-way ANOVA, and t-test, were performed using SPSS (version 28, IBM, Somers, CT). Principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were conducted with SIMCA (version 15.0; Umetrics, Umeå, Sweden), using mean centering, UV scaling, and Pareto scaling. OPLS-DA models were validated with 100 permutation tests.

Quality control of the sequencing reads was conducted using FastQC and Trimmomatic to remove low-quality bases and adapter sequences, filter out reads with a quality score below 20 [32, 33]. High-quality sequencing reads were assembled de novo using Trinity (v2.15.1, Broad Institute, MA, USA) [34], and sequences shorter than 200 bp were removed using SeqKit (v2.8.2, developed by Wei Shen, China) [35]. Redundant sequences were removed, and highly similar sequences were clustered retaining representative sequences using CD-HIT (v4.8.1, Weizhong Li, San Diego State University, CA, USA) [36]. Open Reading Frames (ORFs) were predicted using TransDecoder (v5.5.0, Broad Institute, MA, USA). Reads were aligned to the assembled de novo transcriptome using BWA (v0.7.18, Heng Li, Broad Institute, MA USA), and to count reads mapped to exon regions using featureCounts (v2.0.6, Walter and Eliza Hall Institute, Victoria, Australia) [37, 38]. Quantified reads were normalized and analyzed to detect differentially expressed genes (DEGs) (FDR < 0.05,|logFC| >1) using edgeR (v4.0.16, Walter and Eliza Hall Institute, Victoria, Australia) [39]. Enriched functions of DEGs were analyzed using DAVID (Knowledgebase v2024q1, NIH, MD, USA) [40]. The UniProt IDs required for the analysis were obtained by aligning DEG sequences against SwissProt protein sequences using DIAMOND (v2.0.15, Benjamin Buchfink, Max Planck Institute for Developmental Biology, Tübingen, Germany) [41]. For parameters not specified, the default values were applied. Pathway enrichment analysis employing transcriptome and metabolome data was performed using Metaboanalyst (version 6.0, https://www.metaboanalyst.ca/).

Integrated pathway analysis

Metabolome and transcriptome integrated pathway analysis was performed using Metaboanalyst. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis based on A. thaliana (thale cress) was performed; the transcriptome and metabolome were ID-matched with official gene symbols and compound names, respectively. The pathways with p < 0.05 and impact > 0.7 were selected from the final formed pathways.

Absolute quantification of serotonin, β-sitosterol, campesterol, and stigmasterol

For absolute quantitative analysis, the levels of campesterol, serotonin (5-hydroxytryptamine), β-sitosterol, and stigmasterol levels were measured using calibration curves generated from standards obtained from Sigma-Aldrich (St. Louis, MO, USA). Analytical validation was performed using serotonin and stigmasterol (representative compound for three phytosterols). Validation samples were prepared at concentrations of 10, 50, and 100 mg/L, including post-spike extraction, pre-spike extraction, and standard solution (solvent-spike) samples. Subsequently, matrix effect (%), recovery (%), accuracy (%), and precision (%RSD) were evaluated for method validation. Matrix effect (%) = (Peak area in matrix-post spike) -(Existing peak area) *100/(Peak area in solvent-spike), recovery (%) = (Peak area in matrix-pre spike) *100/(Peak area in matrix-post spike), accuracy (%) = (Measured concentration) *100/(Spiked concentration), precision(%RSD) = (Standard deviation/Average) *100. Precision was assessed using QC samples in terms of intra-day and inter-day variability. The calibration curve was corrected to account for matrix effect, and quantitative analysis was performed. All analyses were conducted in triplicate.

Standard curves were generated with 6.25, 12.5, 25, 50, and 100 mg/L of campesterol; 6.25, 12.5, 25, 50, and 100 mg/L of serotonin; 6.25, 12.5, 25, 50, and 100 mg/L of β-sitosterol; and 12.5, 25, 50, 100, and 200 mg/L of stigmasterol. To generate calibration curve data, each standard compound was dissolved in methanol and diluted to the required concentration. GC-MS analysis and derivatization were performed for complete quantification according to the method described previously [28].

Results

Effect of exogenous IPA on the growth of L. aequinoctialis cultures

Figure 1 shows images of cultured L. aequinoctialis obtained 28 day. The IPA treatment promoted the growth of L. aequinoctialis cultures, particularly 10 µM IPA, which caused the highest growth. Figure 2A showed the number of fronds, and Fig. 2B showed dry weight (g/L) were significantly higher in the 1 µM and 10 µM IPA treatment groups, however no difference was observed in dry weight (g/L) for the 0.5 µM IPA treatment group compared to the control. Figure S1 presents the corresponding time-course growth data, showing the number of fronds measured at each time point.

Fig. 1.

Representative photos of L. aequinoctialis culture under various IPA treatment concentrations on day 28

Fig. 2.

Effects of various IPA treatment concentrations on the growth of L. aequinoctialis culture. Total number of fronds on day 28 (A); total dry weight on day 28 (B). The vertical bars represent the mean, while the error bars indicate the standard deviation (n = 3) for each treatment group. An asterisk indicates significant differences (p < 0.05) between the control and treatment groups, as determined by the t-test

Effect of various concentrations of IPA treatment on the metabolic profiles in L. aequinoctialis cultures

Table S1 lists 55 metabolites from the various L. aequinoctialis culture detected by GC-MS analysis, including 4 alcohols, 20 amino acids, 4 fatty acids, 9 organic acids, 2 phenolics, 3 phytosterols, 5 sugars, and 8 other mixed compounds. Table 1 lists the relative levels (relative intensity/g) in L. aequinoctialis culture treated by 0.5, 1, and 10 µM IPA concentrations at 28 days. The relative levels (relative intensity/g) of aspartate, campesterol, glutamate, glutamine, serotonin, β-sitosterol, stigmasterol, and tryptophan were significantly increased in the IPA treatment groups. The changes in the relative yields (relative intensity/L) of the major metabolites in response to IPA treatment are listed in Table S2. The relative yields of β-alanine, aspartate, citric acid, campesterol, β-sitosterol, and stigmasterol were significantly increased in the IPA-treated compared with those in the control group.

Table 1.

Relative level of metabolites of L. aequinoctialis culture cultivated under various IPA concentration

No.	Compounds	Day 28 Control	Day 28  0.5 μM	Day 28 1 μM	Day 28 10 μM
	Alcohols
1	Glycerol	0.48±0.03a	0.3±0.00b	0.15±0.00c	0.08±0.00d
2	Glycerol 3-phosphate	2.64±0.28a	4.09±0.12b	4.4±0.11b	3.50±0.14ab
3	myo-Inositol	1.27±0.03a	1.17±0.01ab	1.16±0.01ab	0.89±0.01c
4	myo-Inositol phosphate	1.81±0.15a	3.92±0.05b	4.94±0.10c	4.3±0.16b
	Amino acids
5	Alanine	48.21±2.77a	43.67±0.31b	25.7±0.20d	13.31±0.21c
6	β-Alanine	0.28±0.03a	0.36±0.01b	0.40±0.01b	0.47±0.01c
7	Asparagine	12.56±0.51a	13.54±0.46a	16.19±0.31b	26.57±0.76c
8	Aspartate	0.18±0.01a	0.31±0.00b	0.75±0.01c	2.37±0.02c
9	Cysteine	2.18±0.12a	1.71±0.02b	1.02±0.01c	0.98±0.01c
10	Glutamate	1.10±0.06a	2.61±0.01b	5.27±0.03c	5.91±0.04c
11	Glutamine	6.13±0.21a	8.49±0.29b	13.32±0.41c	18.06±0.48c
12	Glycine	0.96±0.08	1.10±0.02	1.05±0.01	0.73±0.01
13	Histidine	1.18±0.11a	1.29±0.03a	1.51±0.04b	1.82±0.02c
14	Isoleucine	1.41±0.10a	1.29±0.01ab	1.05±0.01c	1.39±0.01ab
15	Lysine	3.99±0.25a	6.01±0.11b	5.75±0.04ab	5.51±0.03a
16	Phenylalanine	2.13±0.07a	1.91±0.01a	2.05±0.02a	2.17±0.02a
17	Pipecolic acid	0.27±0.02a	0.25±0.01ab	0.27±0.01ab	0.31±0.01a
18	Proline	1.96±0.13a	1.34±0.01b	1.18±0.02c	1.03±0.01d
19	Pyroglutamic acid	1.82±0.09a	2.49±0.03b	3.89±0.03c	5.97±0.06c
20	Serine	0.75±0.06a	1.24±0.01b	1.87±0.02b	3.23±0.04c
21	Threonine	2.33±0.11a	2.42±0.03a	2.82±0.02b	3.01±0.03b
22	Tryptophan	0.50±0.06a	0.77±0.02b	0.97±0.02b	2.15±0.03c
23	Tyrosine	0.81±0.05a	0.90±0.02ab	0.82±0.01ab	1.10±0.02b
24	Valine	5.08±0.31a	4.66±0.04ab	4.30±0.03bc	4.65±0.05abc
	Fatty acids
25	Linoleic acid	0.72±0.12	0.81±0.04	0.79±0.06	0.86±0.04
26	α-Linoleic acid	1.27±0.18a	1.63±0.34b	1.83±0.15bc	1.96±0.15bc
27	Palmitic acid	3.50±0.24a	3.96±0.16b	3.85±0.18abc	4.18±0.11bc
28	Stearic acid	0.46±0.04a	0.35±0.01ab	0.29±0.01c	0.36±0.01abc
	Organic acids
29	2-Keto-L-gluconic acid	1.08±0.01a	1.14±0.01ab	1.23±0.01b	0.88±0.02c
30	3-Hydroxymethylglutaric acid	0.10±0.01a	0.16±0.01b	0.19±0.00b	0.86±0.00c
31	Citric acid	0.41±0.01a	0.93±0.01b	1.33±0.01c	2.17±0.04d
32	Erythronic acid	1.07±0.03a	1.19±0.01ab	1.42±0.01bc	1.27±0.02bc
33	Fumaric acid	0.03±0.00a	0.04±0.00b	0.04±0.00abc	0.02±0.00abc
34	Lactic acid	0.67±0.06a	0.6±0.01ab	0.38±0.01abc	0.17±0.01c
35	Malic acid	0.25±0.02a	0.31±0.01ab	0.32±0.01b	0.82±0.01c
36	Suberylglycine	12.01±1.10a	12.43±0.57a	15.86±0.64b	23.99±0.94c
37	Succinic acid	1.03±0.02a	0.67±0.01b	0.34±0.01c	0.18±0.01c
	Phenolics
38	Caffeic acid	0.28±0.03a	0.28±0.01ab	0.29±0.02ab	0.25±0.01abc
39	Ferulic acid	1.38±0.03a	0.37±0.01b	0.13±0.00c	0.0008±0.001d
	Phytosterols
40	Campesterol	0.49±0.01a	0.54±0.01ab	0.59±0.01bc	0.62±0.01c
41	β-Sitosterol	0.55±0.01a	0.66±0.01b	0.74±0.01bc	0.79±0.00c
42	Stigmasterol	0.81±0.01a	0.92±0.01ab	1.03±0.01bc	1.08±0.02c
	Sugars
43	Fructose	49.39±0.97a	47.63±0.39ab	53.75±0.44bc	48.82±0.41abc
44	Galactose	0.27±0.01a	0.20±0.01b	0.17±0.01b	0.12±0.01c
45	Glucose	66.58±1.62a	58.75±0.49a	64.37±0.42ab	65.8±0.71ab
46	Maltose	0.42±0.03a	0.66±0.02ab	0.75±0.02c	0.73±0.02abc
47	Sucrose	254.08±10.99	257.06±2.08	257.64±2.61	259.75±0.95
	Others
48	γ-Aminobutyric acid	53.51±2.27a	59.47±0.37a	44.08±0.29ab	33.16±0.29b
49	γ-Hydroxybutyric acid	0.96±0.05a	0.52±0b	0.29±0.01c	0.15±0.000c
50	Ascorbic acid	1.52±0.02a	2.02±0.01b	2.73±0.59c	1.48±0.42abc
51	Neophytadiene	0.46±0.03a	0.63±0.01b	0.70±0.02bc	0.70±0.01bc
52	Phosphoric acid	55.6±3.77a	38.05±0.65b	34.4±0.91b	19.24±0.47c
53	Serotonin	6.06±0.32a	7.30±0.06a	8.10±0.06b	8.63±0.07c
54	Threonic acid	0.20±0.01a	0.19±0.01ab	0.19±0.01ab	0.15±0.00c
55	Tryptamine	2.43±0.11a	2.15±0.02a	1.35±0.01b	0.98±0.01b

Data are mean ± standard deviation values of nine measurements from three biological replicates and three technical replicates of the control (not treated with IPA) and various IPA treatment (0.5, 1, and 10 μM) groups. Harvesting for all groups occurred on day 28. Superscript letters (a, b, c, and d) represent statistically significant differences, determined using the Kruskal-Wallis test followed by the Mann-Whitney test with Bonferroni’s correction for post hoc analysis (p < 0.0083)

Figure S2 (A) shows the PCA-derived score plots generated from the metabolic profiles of L. aequinoctialis incubated with different concentrations of IPA treatments, including a QC sample. The QC sample was at the center of the score plot, indicating high stability and reproducibility of the GC-MS analysis. Figure S2 (B) showed an OPLS-DA-derived score plot demonstrating an apparent separation of the IPA treatment groups at each concentration, indicating that the IPA treatments resulted in metabolite changes in L. aequinoctialis culture. The R²X, R²Y, and Q² values from the OPLS-DA-derived score plots were 0.894, 0.978, and 0.943, respectively, indicating excellent explanatory and predictive power. These values surpass commonly accepted thresholds for model validity, thereby supporting the robustness of the multivariate classification model. The model was further validated through 100 permutation tests, which confirmed that the actual Q² value was significantly higher than those of the randomized models, indicating a low risk of overfitting and reliable group separation.

Figure S3 shows the notably altered metabolic pathways associated with 10 µM IPA treatment. In total, 25 pathways were identified, including ‘linoleic acid metabolism’, ‘alanine, aspartate and glutamate metabolism’, ‘indole alkaloid biosynthesis’, and ‘galactose metabolism’(impact > 0.12, p < 0.05).

Transcriptomic analysis of L. aequinoctialis cultures treated by IPA compared to the control

A total of 89,024 transcripts were profiled in the transcriptome analysis, of which 7,490 exhibited significant differences between the control and IPA treatment groups (FDR < 0.05,|FC| < 2). In the 10 µM IPA treatment group, 3,828 transcripts exhibited increased expression whereas 3,662 transcripts showed decreased expression. Figure S4 showed pathway enrichment analysis was conducted based on the significantly altered transcripts. Fifteen pathways including ‘linoleic acid metabolism’, ‘glycerolipid metabolism’, ‘starch and sucrose metabolism’, and ‘glutathione metabolism’ were identified as major altered pathways by IPA treatment in L. aequinoctialis culture (impact > 0.23, p < 0.05).

Integrated pathway analysis

Figure 3 showed the Metaboanalyst was employed to visualize the integrated metabolome and transcriptome pathways. Twenty-one pathways were revealed as major altered pathways by IPA treatment of L. aequinoctialis culture. Among those pathways, 10 pathways showed significant alterations in both metabolites and transcripts. Then, we focused on ‘glycerolipid metabolism’, ‘glutathione metabolism’, ‘β-alanine metabolism’, ‘alanine aspartate, and glutamate metabolism’, and ‘starch and sucrose metabolism’ for elucidation of the enhanced growth and production of useful metabolites (impact > 0.7, p < 0.05). In ‘glycerolipid metabolism’, glycerol-3-phosphate acyltransferase genes (GPAT) such as GPAT1 (log₂(FC) = 1.04), GPAT4 (log₂(FC) = 2.61), and glycerol-3-phosphate (log₂(FC) = 0.41) were overexpressed and glycerol (log₂(FC) = −2.57) was downregulated by the IPA treatment. In ‘glutathione metabolism’, overexpressed 6-phosphogluconate dehydrogenase gene (PGD1) (log₂(FC) = 3.79) and enhanced glutamate (log₂(FC) = 2.42) were observed by the IPA treatment, whereas cysteine (log₂(FC) = −1.15) was decreased. In ‘β-alanine metabolism’, overexpressed polyamine oxidase gene (PAO) (log₂(FC) = 1.37) and increased β-alanine (log₂(FC) = 0.77) were observed by the IPA treatment. In ‘alanine, aspartate, and glutamate metabolism’, increased aspartate (log₂(FC) = 3.72) and glutamate (log₂(FC) = 2.42), and overexpressed glutamate decarboxylase gene (GAD) (log₂(FC) = 1.17) were observed by the IPA treatment, while γ-aminobutyric acid (GABA) (log₂(FC) = −0.69) was downregulated. In ‘starch and sucrose metabolism’, β-glucosidase 6 gene (BGLU6) (log₂(FC) = 1.11), β-amylase 1 gene (chloroplastic) (BAM1) (log₂(FC) = 3.16), β-amylase 1 gene (BMY1) (log₂(FC) = 3.04), hexokinase gene (HXK); HXK1 (log₂(FC) = 1.06), HXK2 (log₂(FC) = 1.15), endoglucanase 9 gene (GLU9) (log₂(FC) = 2.98) were overexpressed by the IPA treatment. The log₂(FC) values of significantly altered metabolites and transcripts in those metabolisms were listed in Table 2.

Fig. 3.

Integrated pathway analysis of significantly altered metabolites and transcripts. The p-values were calculated using the hypergeometric test and adjusted for false discovery rate (FDR) correction (p < 0.05). The adjusted values were then transformed into–log(p). Pathway impact values were determined based on pathway topology analysis

Table 2.

List of selected pathways exhibiting significant alteration in metabolites and transcripts in L. aequinoctialis culture by 10 µM IPA treatment

Related pathway	Metabolite	log2(FC)	Transcript	log2(FC)
		10 µM /Con		10 µM /Con
Glycerolipid metabolism	Glycerol*	−2.57	GPAT1*	1.04
	Glycerol-3-phosphate*	0.41	GPAT4*	2.61
Glutathione metabolism	Cysteine*	−1.15	PGD1*	3.79
	Glutamate*	2.42
β-Alanine metabolism	Aspartate*	3.72	PAO*	1.37
	β-Alanine*	0.77
Alanine, aspartate, and glutamate metabolism	Aspartate*	3.72	GAD*	1.17
	GABA*	−0.69
	Glutamate*	2.42
Starch and sucrose metabolism	Sucrose	0.03	BGLU6*	1.11
			BAM1*	3.16
			BMY1*	3.04
			HXK1*	1.06
			HXK2*	1.15
			GLU9*	2.98

The log₂FC represents the log₂ ratio of the maximum metabolite intensities between the IPA treated group (10 µM) and the control group of L. aequinoctialis culture. A positive log₂FC indicates an increase in metabolites and transcripts level following IPA treatment, whereas a negative value indicates a decrease. Statistical significance was assessed using the Mann–Whitney U test (n = 9, comprising three biological replicates and three technical replicates).* Significantly altered metabolites or transcripts by 10 µM IPA treatment (p < 0.05)

Absolute quantification of serotonin, β-sitosterol, campesterol, and, and stigmasterol

From the identified metabolites, we selected campesterol, serotonin, β-sitosterol, and stigmasterol as bioactive metabolites for absolute quantification. Table S3 listed the validation results for serotonin and stigmasterol (representative compound for three phytosterols) at concentrations of 10, 50, and 100 mg/L. For serotonin, the matrix effect ranged from 117.09 ± 2.51% to 164.45 ± 4.06%, recovery from 69.09 ± 0.19% to 81.66 ± 1.30%, and accuracy from 98.18 ± 0.90% to 100.95 ± 1.12%. Precision was 5.16% (intra-day) and 4.92% (inter-day) at 10 mg/L. For stigmasterol, the matrix effect ranged from 100.43 ± 1.98% to 119.10 ± 9.56%, recovery from 92.14 ± 6.65% to 97.16 ± 1.80%, and accuracy from 98.13 ± 4.82% to 102.49 ± 6.08%. Precision was 6.26% (intra-day) and 10.41% (inter-day) at 10 mg/L. The regression equations, correlation coefficients, LOD, and LOQ for these four metabolites were listed in Table 3. Such R² values have been considered as acceptable indicators of linearity, particularly in plant-based metabolomics [42]. Table 4 listed that 10 µM IPA treatment resulted in the highest productivity levels of campesterol (20.49 mg/L), serotonin (23.09 mg/L), β-sitosterol (25.59 mg/L), and stigmasterol (50.96 mg/L).

Table 3.

Regression equation, correlation coefficient (r² values), LOD, and LOQ for the standard compounds campesterol, serotonin, β-sitosterol, and stigmasterol

Compound	Regression equation	r2 values	LOD (µg/mL)	LOQ (µg/mL)
Campesterol	y = 0.0016x − 0.0123	0.993	0.29	0.88
Serotonin	y = 0.0181x − 0.1364	0.994	1.55	4.69
β-Sitosterol	y = 0.001x − 0.0073	0.994	0.47	1.41
Stigmasterol	y = 0.0008x − 0.0066	0.998	1.17	3.54

Triplicate measurements were performed for each test. LOD, limit of detection; LOQ, limit of quantification

Table 4.

Productivities (mg/L) of campesterol, serotonin, β-sitosterol, and stigmasterol in L. aequinoctialis cultures on day 28 following treatment with various concentrations of IPA

Compounds	Control	0.5 µM	1 µM	10 µM
Campesterol (mg/L)	9.90 ± 1.17a	12.28 ± 1.51b	14.88 ± 2.5b	20.49 ± 0.76c
Serotonin (mg/L)	10.43 ± 2.45a	13.39 ± 1.13b	16.42 ± 1.31c	23.09 ± 1.36d
β-Sitosterol (mg/L)	13.85 ± 1.43a	18.35 ± 2.05b	23.21 ± 4.03bc	25.59 ± 1.45c
Stigmasterol (mg/L)	22.52 ± 2.93a	28.49 ± 4.27b	36.73 ± 6.41b	50.96 ± 2.89c

Data are mean ± standard deviation values of nine measurements from three biological replicates and three technical replicates control group and 0.5, 1, and 10 µM IPA treatment groups on day 28. Superscript letters (a, b, c, and d) represent statistically significant differences, determined using the Kruskal-Wallis test followed by the Mann-Whitney test with Bonferroni’s correction for post hoc analysis (p < 0.0083)

Discussion

In our study, the IPA treatment increased the growth of L. aequinoctialis culture and altered several metabolic and transcriptomic profiles. Major altered pathways identified by integrated analysis of metabolomic and transcriptomic data were ‘glycerolipid metabolism’, ‘glutathione metabolism’, ‘β-alanine metabolism’, ‘alanine, aspartate, and glutamate metabolism’, and ‘starch and sucrose metabolism’ (impact > 0.7, p < 0.05) (Fig. 3).

‘Glycerolipid metabolism’ was the main upregulated pathway following IPA treatment in L. aequinoctialis cultures. The IPA treatment lowered glycerol levels and increased glycerol-3-phosphate and GPAT levels. In plants, glycerolipid metabolism is an essential process that provides energy storage and signal transduction compounds in cell membranes [43]. In this pathway, the GPAT mediates the initial synthetic steps for the formation of glycerolipids, which are essential components of biological membranes and function as a major form of stored energy [44]. GPAT1 deficiency causes damage to the endoplasmic reticulum membrane in A. thaliana study [45]. Furthermore, GPAT enhances the synthesis of suberin and cutin, which are key cell wall materials in tea plant [46]. In Camellia oleifera, glycerol is transformed into glycerol-3-phosphate by ATP and glycerol kinase, which is subsequently converted to the phospholipid precursor lysophosphatidic acid through the action of GPAT [47]. In Brassica napus, increased glycerol-3-phosphate was observed to increase the synthesis of glycerolipids [48]. Our study suggests that the increase in glycerol-3-phosphate was induced by enhanced usage of glycerol by IPA treatment. These suggests that the increase in glycerol-3-phosphate and GPAT possibly contributed to the formation of glycerolipids, supporting leaf membrane biosynthesis thereby promoting the growth of L. aequinoctialis culture.

The administration of IPA influenced the ‘glutathione metabolism’ in the L. aequinoctialis culture. In this metabolic process, the IPA treatment increased the levels of glutamate and PGD1 and decreased cysteine level. Glutathione is an essential metabolite for plant life and is important in controlling reactive oxygen species [49–51]. Additionally, it functions as an antioxidant to protect plants from oxidative damage in abiotic stress [52, 53]. Glutathione is produced in chloroplasts from glutamate using cysteine [54]. The PGD was used in the regeneration of glutathione in Triticum aestivum [55, 56]. Proteomic pathway analysis of the tea plants Phytosuccion and Tieguanyin showed up-regulation of PGD leading to up-regulation of glutathione [57]. Glutamate plays a central role in nitrogen assimilation and carbon–nitrogen metabolism in rice and has been reported to induce the expression of genes related to metabolism and defense responses upon exogenous application [58]. Additionally, upregulation of PGD1 is known to facilitate glutathione regeneration and mitigate oxidative stress, thereby supporting plant growth and environmental adaptability [59]. In our study, IPA treatment was associated with decreased levels of cysteine and increased levels of PGD1, which may indicate enhanced utilization of these metabolites in glutathione metabolism. This metabolic modulation may have contributed to the alleviation of oxidative stress under artificial cultivation conditions, potentially supporting the improved growth observed in L. aequinoctialis culture.

‘β-Alanine metabolism’ was also the main metabolic pathway, which was altered after IPA treatment in the L. aequinoctialis culture. In β-alanine metabolism, IPA treatment increased the levels of β-alanine, aspartate, and PAO. β-alanine is involved in the synthesis and breakdown of phospholipids and fatty acids and is also important in secondary metabolism, including the biosynthesis of lignin, a plant cell wall component [60]. In a study of A. thaliana, β-alanine levels were significantly elevated after heat shock stress [61]. The biosynthetic pathways for β-alanine include the aspartate and polyamine pathways in plants [62]. Aspartate can be decarbonylated by aspartate 1-decarboxylase to produce β-alanine in the aspartate pathway [63]. In the maize polyamine pathway, spermine converted to β-alanine via 1,3-diaminopropane produced by PAO [64, 65]. In our study, the IPA treatment increased levels of PAO1, aspartate, and β-alanine, indicating that the IPA treatment activated the aspartate pathway and polyamine pathway. The increase in β-alanine level is believed to aid L. aequinoctialis stress regulation and cell wall construction, which may have contributed to enhancing the growth of L. aequinoctialis culture.

The IPA treatment also upregulated ‘alanine, aspartate, and glutamate metabolism’ in the L. aequinoctialis culture. Increased levels of aspartate, glutamate, and GAD and decreased levels of GABA were observed following IPA treatment (Table 2). This pathway is linked to several metabolic processes, including asparagine biosynthesis, TCA cycle, and key metabolic pathways that regulate the material and energy cycle in A. thaliana [66]. Alanine, aspartate, and glutamate metabolism confer tolerance to drought stress tolerance in Salvadora persica and reduce heat stress in S. pusiforme [67, 68]. In plants, aspartate and glutamate are synthesized from oxaloacetic acid and α-ketoglutaric acid, which are intermediates in the TCA cycle [69]. The combined treatment of tomatoes with aspartate and glutamate improves photosynthetic reactions and carbon dioxide assimilation and increases tomato biomass and growth [70]. GAD is responsible for the decarboxylation of glutamate into CO₂ and GABA in plants [71]. GABA is metabolized via ‘GABA shunt pathway’ and is converted to succinic semialdehyde, which is subsequently entered into the TCA cycle as succinic acid [72]. In L. turionifera, GAD overexpression modulated the GABA shunt and promoted growth by activating the TCA cycle [73]. Thus, it is speculated that IPA treatment upregulated ‘alanine, aspartate, and glutamate metabolism’ and resulted in enhanced growth of the L. aequinoctialis culture.

‘Starch and sucrose metabolism’ was one of the main pathway altered by IPA treatment in L. aequinoctialis culture. This metabolism is activated as a carbon store for plants, helping them to growth [74]. The levels of BGLU6, BAM1, BMY1, HXK1, HXK2, and GLU9 were increased by the IPA treatment in our study. Starch and sucrose play a role in plants as energy sources, starting molecules for metabolites and more [75]. BGLU6, BAM1, and BMY1 are a family of β-amylase genes involved in starch degradation, stress response and hormone regulation in Arabidopsis [76]. GLU is involved in the conversion of plant cell wall cellulose or starch into simple sugars [77]. In Arabidopsis, HXK1 has been reported to regulate sucrose-induced cell proliferation and expansion, promoting leaf growth [78]. Upregulation of GLU9 in the roots of 32 plant species activated starch and sucrose metabolism, leading to accumulation of simple sugars in plant roots [79]. HXK phosphorylates hexoses, including glucose, fructose, mannose, and sucrose, and plays a role in the regulation of plant hormones and abiotic stress [80, 81]. In pear and grape, overexpression of HXK was positively correlated with increased growth [82, 83]. Likewise, BAM1 is essential for meristem maintenance and organ development in Arabidopsis [84]. In Tibetan wild barley, the BMY1.b allele increased β-amylase activity under drought conditions, promoting starch degradation and contributing to stress resilience through efficient carbohydrate utilization [85]. In our study, sucrose was administered to the L. aequinoctialis culture. However, the level of sucrose did not show a significant difference between the control and IPA treatment groups. Therefore, the increase in BGLU6, BAM1, BMY1, HXK1,2, and GLU9 by IPA treatment may contribute to an efficient metabolism of sucrose for the enhanced growth of L. aequinoctialis culture. Figure 4 showed a schematic diagram showing the effect of IPA treatment on the growth and the major altered pathways in L. aequinoctialis culture.

Fig. 4.

Schematic diagram of metabolites and transcripts changes in L. aequinoctialis cultures following IPA treatment. Metabolites and gene expression level alterations observed in L. aequinoctialis upon IPA treatment were analyzed relative to untreated control cultures. The black arrow denotes the direction of the reaction. The red arrow denotes the upregulation of expression. The blue arrow denotes the downregulation of expression

In vitro plant culture is a biotechnological method employed for the large-scale production of plants. During plant tissue culture, various stressful conditions, including low ventilation rates and light availability, can impede the growth of the plants [86]. In addition, the rapid proliferation in a confined flask may have resulted in overlapping spaces and insufficient light intensity for proliferation. Under these stressful conditions, we expect that the significantly altered metabolome and transcriptome caused by the IPA treatment may have complemented each other to regulate stress, thus supporting the enhanced growth of the L. aequinoctialis culture.

We focused on serotonin and phytosterols among the various metabolites. Table S3 presents the validation results for accurate quantification, including matrix effect and other parameters. Stigmasterol exhibited a matrix effect of 119.10 ± 9.56% at 10 mg/L and 100.43 ± 1.98% at 100 mg/L. In the case of serotonin, a matrix effect of 164.45 ± 4.06% was observed at 10 mg/L. However, this was effectively normalized and corrected using a stable isotope-labelled internal standard (myristic-d27 acid) during calibration curve preparation. As a result, accuracy was maintained at approximately 100% across all concentrations, confirming the reliability of serotonin quantification.

The 10 µM IPA treatment significantly enhanced serotonin productivity to reach 23.09 mg/L (Table 4). Serotonin is an essential signaling molecule in many processes, including the promotion and regulation of plant growth and the mediation of plant circadian rhythms [87]. Its concentration was increased during the formation of new roots and shoots in plants [88–90]. Based on these findings, we speculated that the increase in serotonin caused by the administration of IPA would stimulate the growth of L. aequinoctialis.

The relative levels of phytosterols (campesterol, β-sitosterol, and stigmasterol) were significantly enhanced after IPA treatment. The 10 µM IPA treatment had the highest productivities of campesterol (20.49 mg/L), β-sitosterol (25.59 mg/L), and stigmasterol (50.96 mg/L). In plants, phytosterols are an essential component of plant cell membrane lipids, notably contributing to membrane fluidity, thereby affecting membrane function and plant structure [91–93]. Previous studies have shown that phytosterol accumulation contributes to improved membrane strength, particularly under abiotic stress such as drought in rice seedlings [94]. Moreover, disruption of sterol biosynthesis by knocking out a key gene squalene synthase in plant sterol biosynthesis increased leakage into the plant cell’s defense barrier apoplast and promoted pathogen growth [95].

Our results showed that SQE1 and CYP710A were upregulated by IPA treatment (Fig. S5). Plant sterols such as campesterol, β-sitosterol, and stigmasterol are synthesized via the mevalonic acid (MVA) pathway in the cytoplasm. In this pathway, squalene is converted to 2,3-oxidosqualene by squalene epoxidase (SQE1), which is then cyclized to cycloartenol, a common precursor for various phytosterols. Subsequent enzymatic steps, including desaturation and methylation reactions catalyzed by CYP710A, lead to the final production of phytosterols. Especially in tomato, the upregulation of the CYP710A11 gene is closely associated with the accumulation of stigmasterol, demonstrating that the activity of sterol C-22 desaturase directly contributes to the regulation of phytosterol composition [96, 97]. Furthermore, the SQE1 (squalene epoxidase) gene, upregulated by IPA treatment, has been shown to be essential for phytosterol biosynthesis and early development processes in Arabidopsis thaliana [98]. Collectively, these results suggest that IPA has the effect of promoting the production of biologically active phytosterols with functional roles in plant metabolism in L. aequinoctialis.

There were no previous reports regarding the effects of IPA on metabolite profiles in plants. In Arabidopsis and tobacco plants, there were only studies on the effect of IPA on growth of those plants [21, 22]. The significance of our study is that it is the first to investigate metabolomic and transcriptomic profiles induced by IPA treatment in L. aequinoctialis culture.

Conclusions

In this study, IPA treatment increased the growth and altered various metabolomic and transcriptomic profiles of L. aequinoctialis culture. Integrated pathway analysis of metabolomic and transcriptomic data revealed that the IPA treatment significantly altered ‘glycerolipid metabolism’, ‘glutathione metabolism’, ‘β-alanine metabolism’, ‘alanine, aspartate, and glutamate metabolism’, and ‘starch and sucrose metabolism’. We also observed that the productivities of campesterol, serotonin, β-sitosterol, and stigmasterol were significantly increased following the IPA treatment. Our study is not without limitations. We showed major altered metabolic pathways by IPA treatment by integrating transcriptomic and metabolic analyses. However, we did not validate relevant enzyme activities or metabolite fluxes involved in the pathway. Further studies assessing enzyme activities or metabolite fluxes should be performed in future. In addition, the efficacy of the IPA treatment strategy could be further validated by testing it on other duckweed species within the genera Spirodela or Wolffia. Identification and structure elucidation of IPA receptor also could be challenged for better understanding of IPA in plant system. Additionally, co-treatment of IPA with other elicitors could be performed in the viewpoint of the growth and production of useful metabolites in L. aequinoctialis culture. Despite these limitations, we suggest that IPA treatment could be used as an efficient strategy to produce biomass or bioactive compounds by large-scale cultivation of L. aequinoctialis for application in various field of biotechnology and agriculture including food, feed, and pharmaceutical industries.

Abbreviations

Lemna aequinoctialis

L. aequinoctialis

IPA

Indole-3-propionic acid

GC-MS

Gas chromatography-mass spectrometry

IS

Internal standard

LOQ

Limit of quantification

LOD

Limit of detection

SPSS

Statistical Package for the Social Sciences

PCA

Principal component analysis

OPLS-DA

Orthogonal partial least squares-discriminant analysis

QC

Quality control

BSTFA

N, O-bis(trimethylsilyl)- trifluoroacetamide

MEOX

Methoxyamine hydrochloride

GABA

γ-aminobutyric acid

GPAT

Glycerol-3-phosphate acyltransferase gene

PGD

Phosphogluconate dehydrogenase gene

PAO

Polyamine oxidase gene

GAD

Glutamate decarboxylase gene

BGLU

β-glucosidase gene

BAM

β-amylase gene (chloroplastic)

BMY

β-amylase gene

HXK

Hexokinase gene

GLU

Endoglucanase gene

Author contributions

YB Lee designed, performed the experiments, analyzed the data, and wrote the manuscript. JY Cho, Y Jeong, SH Park analyzed the data and wrote the manuscript. HK Choi conceptualization, funding acquisition, project administration, supervision, wrote– original draft.

Funding

This work was supported by a grant from the National Research Foundation of Korea (NRF) Ministry of Science and ICT (MSIT) (2022R1A2C1005245) (RS-2023-00224099) and a grant (22183MFDS366) from Ministry of Food and Drug Safety of South Korea in 2022–2025.

Data availability

The datasets generated and/or analyzed during the current study are available in the Gene Expression Omnibus (GEO) repository, GSE289260.

Declarations

Ethics approval and consent to participate

This study does not involve any human participants, animal subjects, or personal data requiring ethical approval.

Consent for publication

All authors approved the submission of the manuscript.

Competing interests

The authors declare no competing interests.