Introduction

Sugarcane (Saccharum spp.) is produced on approximately 26 million hectares in more than 90 tropical and subtropical regions and countries (FAOSTAT 2022). It is an important crop responsible for much of the world’s sugar and bioethanol production. In many of these regions, water deficiency and drought are among the main weather events limiting sugarcane production; thus, proper water management and breeding of drought-tolerant sugarcane varieties are important research goals for sustainable production in these regions (Ferreira et al. 2017; Singels et al. 2019; Dlamini 2021). In sugarcane, drought stress not only attenuates physiological functions such as photosynthesis (Du et al. 1996; Zhao et al. 2013; Dinh et al. 2017; Marchiori et al. 2017) but it also significantly reduces biomass production and the final yield (Robertson et al. 1999; Basnayake et al. 2012). Therefore, improving the crop production capacity per amount of water resource applied to the field or used by the plant, namely the water-use efficiency (WUE), is important for the yield and profitability of sugarcane production, not only under rainfed conditions but also under irrigated conditions (Basnayake et al. 2012; Natarajan et al. 2020). Introducing new varieties via drought-tolerant breeding is the most effective strategy for this purpose. However, drought-tolerant varieties that have been produced are accidental products resulting from conventional selection trials in drought-prone regions and cropping seasons; efficient breeding and selection has not been conducted (Acreche 2017).

In sugarcane, the limitations of crossbreeding between varieties have been highlighted for many years, warranting the need to expand the genetic base using closely related genetic resources and to introduce their promising traits in sugarcane (Wei and Jackson 2016; Jackson 2019). In particular, Erianthus arundinaceus, a species closely related to sugarcane, is considered a genetic resource that can contribute to the improvement of drought tolerance in sugarcane through intergeneric hybridization (Jackson and Henry 2011). However, the physiological determinants of drought tolerance in Erianthus have not been thoroughly compared with those in sugarcane (Jackson and Henry 2011), and no progress has been made in searching for promising traits as selection indicators or in improving the efficiency of drought-tolerance breeding. Understanding the physiological mechanisms, as well as other important traits, is important for efficient drought-tolerance breeding in sugarcane (Jackson et al. 1996).

In crops that are large and cultivated for long periods of time, such as sugarcane, WUE is considerably impacted by environmental factors, and its determinants are complex (Basnayake et al. 2012; Liu et al. 2016; Natarajan et al. 2020). Therefore, the physiological mechanism of WUE is commonly studied by scaling it down along both the time/phenology and size axes. Intrinsic WUE (iWUE), also known as transpiration efficiency, which is the individual leaf photosynthetic rate (An) divided by its transpiration indicator, namely stomatal conductance (gs), is recognized as the minimum unit of WUE and has been proposed as an important target in crop breeding (Condon et al. 2004).

$$iWUE\, = \,Assimilation \, rate\left( {A_{n} } \right)/Stomatal \, conductance(g_{s} ).$$

A joint research group comprising scientists from Australia and China investigated WUE in sugarcane based on individual leaf gas exchange characteristics under various soil-moisture conditions in diverse genotypes, including wild germplasm (Jackson et al. 2016; Li et al. 2017; Zhao et al. 2017; Natarajan et al. 2021). This group noted variations in iWUE among the genetic resources and suggested iWUE as a promising trait that could be correlated with WUE. Erianthus, which had a high iWUE, showed high potential as a promising genetic resource. It is, however, unclear as to which physiological factors contribute to a high iWUE. Other physiological functional traits, such as leaf morphology and metabolite concentration, should be considered as targets for physiological understanding of high iWUE and in the search for efficient selection indices.

In recent years, with the development of various analytical techniques, increasing attempts have been made to comprehensively understand the physiological and genetic functions of plants through “omics” analysis (Hong et al. 2016; Ali et al. 2019; Singh et al. 2021). Metabolomics is the comprehensive analysis of metabolites and is considered a method of phenotypic analysis that most closely reflects genomic information (Watt et al. 2010; Horgan and Kenny 2011). In addition, these techniques have potential applications in growth and disease diagnostics and in the identification of selection markers (i.e., biomarkers) of stress tolerance for use in breeding (Kumar et al. 2017; Villate et al. 2021). As of date, in sugarcane, “targeted or selective” analyses of sugars and inorganic ions in the juice (e.g., Watanabe et al. 2016) and C4 photosynthesis-related metabolites in the leaf blades (e.g., Du et al. 1996) have been performed. In recent years, metabolomics has been attempted for a “comprehensive” analysis of a wide range of metabolites (sugars, amino acids, organic acids, etc.) to investigate the mechanisms of sugar accumulation in stems (Glassop et al. 2007; Perlo et al. 2022), and to identify functional components in leaf blades and cane juice (Coutinho et al. 2016; Date et al. 2022) and biomarkers for breeding and selection (Kumar et al. 2017; Villate et al. 2021). However, metabolomics studies in sugarcane have been scarce compared with those in other crops (Langridge and Fleury 2011). Although metabolomic analyses related to drought tolerance in sugarcane were performed in a few studies, investigations have been limited to comparisons among varieties (Vital et al. 2017; Budzinski et al. 2019). No metabolome analysis has focused on interspecific differences in iWUE of sugarcane and closely related genetic resources (e.g., Erianthus).

As with many other crops, studies on the effects of soil drought on sugarcane metabolites have mainly focused on the youngest fully expanded leaf (Du et al. 1996; Budzinski et al. 2019), which is the most active site for gas exchange, and other individual organs and leaf ages have not been considered. Gas exchange rates at different leaf ages were determined in a few studies (Meinzer and Grantz 1989; Marchiori et al. 2014). The existence of a stress-sensitive gradation of organs from roots to the aboveground tip of plants exposed to drought conditions has been suggested, based on the corresponding variations in metabolites (some osmolytes, such as proline) and water potential among some organs (Westgate and Boyer 1985; Choluj et al. 2008; De Souza et al. 2015; Adnan et al. 2021; Sapes and Sala 2021). In addition, metabolites that accumulate in leaves, which are the sites of gas exchange, are not necessarily biosynthesized in leaves. To understand the stress-response mechanisms of plants, it is necessary to determine the trends in the accumulation of metabolites and understand the relationship among them. However, whole-plant metabolomic analysis to identify the organs/parts most responsive to soil drying and the metabolites associated with iWUE has never been attempted in sugarcane and related germplasm. We surmised that a comprehensive analysis of metabolites accumulated at various sites in Erianthus, a promising genetic resource for drought-tolerance breeding, would facilitate a better understanding of drought tolerance and iWUE physiology in sugarcane. The present study was conducted to investigate the responses of leaf gas exchange characteristics and primary metabolites in various parts of sugarcane and Erianthus under wet and dry soil conditions, with the aim of identifying useful traits related to drought tolerance.

Materials and methods

Plant materials

A pot trial was performed in a temperature- and humidity-controlled glasshouse (Day/night temperature at 30/27 °C, relative humidity 70%) at the Tropical Agriculture Research Front, Japan International Research Center for Agricultural Sciences (24°22′43′′ N, 124°11′4′′ E). Natural radiation was introduced in the glasshouse without supplemental lighting. The average of daily accumulated solar radiation in the greenhouse during the experiment was 12.4 ± 5.1 mol m−2 day−1. Sunrise and sunset times during the experiment were 6:20 am/7:13 pm on Aug. 20 and 6:39 am/6:20 pm on Oct. 10. Sugarcane variety ‘NiF8’ and Erianthus arundinaceus accession ‘JW630’ were used for the analyses. One-node seedlings of Erianthus (May 14, 2021) and sugarcane (June 2, 2021) were planted in 518 mL volumetric polyethylene pots (10.5 cm top diameter × 7.0 cm bottom diameter × 8.5 cm height) filled with potting soil mix (pH, 5.6; electrical conductivity, 527 µS cm−1) to start seedling growth. Because germination of Erianthus was slow in the preliminary test, we started raising Erianthus seedlings half a month earlier for aligning the growth stages with those of sugarcane. When two to three leaves were fully developed (July 19, 2021), the seedlings were transplanted to 1/2000a Wagner pots (25.6 cm top diameter × 23.4 cm bottom diameter × 29.7 cm height) filled with 10 kg of the same soil, with four replicates per treatment. A soil-moisture sensor (EC-5, METER Group, Inc., Pulman, WA, USA) was inserted at a soil depth of 15 cm, and data were recorded at 15-min intervals using a data logger (ZL6, METER). Solid fertilizer (N:P:K = 2.4:0.7:1.0 g pot−1) was applied at transplantation. Until the start of the irrigation treatment, the plants were irrigated using automatic drip irrigation three times a day. With a sufficient amount of water to avoid drought stress, the water was allowed to drain freely from the bottom of the pots. The irrigation treatment was initiated 43 days after transplanting (August 30) by plugging the bottom of the pot. According to Dinh et al. (2017), wet treatment was performed by manually adding the daily consuming amount of water, while dry treatment was performed by gradually reducing the volume water content (VWC) 1% daily, depending on weather conditions for approximately 3 weeks. Finally, the plants were exposed to intense drought stress (pF > 4.0) for approximately another 3 weeks. Water consumption was estimated on the basis of soil-moisture sensor observations, soil volume, and soil bulk density.

Leaf gas exchange measurements

The gas exchange characteristics were determined at the middle part of leaves for each leaf age—the upper (youngest fully developed leaf), middle (third leaf), and lower (fifth leaf) leaves (Fig. 1)—using a portable gas exchange measuring device (LI-6400, LI-COR BioSciences, Lincoln, NE, USA) from 10 am to 2 pm for 3 days (October 7 to October 9, 2021) before harvesting. Measurements were performed with four plants under the following conditions: high light (2000 µmol m−2 s−1) and low-light (500 µmol m−2 s−1) intensity provided using an LED light source (6400-02B, LI-COR), a flow rate of 400 µmol s−1, and a reference CO2 concentration of 400 µmol mol−1. The leaf temperature was maintained at 31.1 ± 1.1 °C by controlling block temperatures attached to the chamber. Vapor pressure deficit was controlled at 2.0 ± 0.3 kPa using Drierite® (W. A. Hammond Drierite Co., Xenia, OH, USA).

Fig. 1
figure 1

Outline of examined plant organs in the present study. Gas exchange measurements were conducted only for middle part of each leaf (blue rectangles). Each leaf tip, middle, and base was collected for metabolomics (red and blue rectangles). The upper, middle, and lower leaves indicate the first (youngest), third, and fifth fully expanded leaves, respectively. Nodes were sampled between dewlaps of first and second youngest leaves (upper node), third and fourth leaves (middle node), and fifth and six leaves (middle node). Each node contained stem, outer and inner leaf sheaths. In Erianthus, the upper and middle nodes contained undeveloped tissues rather than stem. Only the main stem was sampled. The total number of parts sampled per plant was 19 (leaf: 9; stem: 3; leaf sheath: 6; root: 1)

Full size image

Visualization of leaf anatomy

Stomatal replicas of the adaxial and abaxial surfaces of the leaves used for gas exchange measurements (n = 4) were collected according to Wu and Thao (2017). A nail polish was brushed on the leaf surface. After it dried, the thin layer of nail polish was collected using double-sided tape. The sample of stomatal replica on a glass slide was observed by light microscopy. Transverse cross-sections (approximately 40 µm thick) of the tested leaves were prepared manually using a hand microtome (THK, Kenis Ltd., Osaka, Japan). Stomatal replicas and cross-sections were observed under an optical microscope (Eclipse E800, Nikon, Tokyo, Japan) and the stomatal density and interveinal distance were evaluated using an image analysis software (NIS-elements, Nikon) (Fig. S1).

Metabolome analysis

Sample preparation

Based on the results of gas exchange measurements, one moderate-grown individual plant was selected. At 85 days after transplanting (42 days after the start of treatment), the plant was harvested for metabolome analysis. Total fresh leaf numbers were 6 under dry condition and 10 under wet condition for sugarcane, while 7 under dry condition and 8 under wet condition for Erianthus. Individual sample variations could lead to potential erroneous data interpretation under big data analysis of numerous metabolite concentrations. Because variations among individuals might be widely influenced by external factors such as growth conditions, time of sampling, and weather on the day of sampling, a single individual plant was examined. The plantlets were immediately separated into upper leaves and upper nodes, middle leaves and middle nodes, lower leaves and lower nodes, and roots (Fig. 1). Nodes were sampled between dewlaps (conjunction between leaf and leaf sheath) of first and second youngest leaves (upper node), third and fourth leaves (middle node), and fifth and six leaves (middle node). Leaves were separated into tip, middle, and basal parts, and nodes were separated into outer and inner leaf sheaths, and stem (a total of 19 parts for each plantlet). The roots were thoroughly washed and a part of it was collected. Although Erianthus vigorously produced tillers, only the main stem was targeted. For the upper and middle nodes of Erianthus, undeveloped tissues, excluding the external and internal leaf sheaths, were sampled because no clear stem was observed. Each part was packed in a labeled non-woven fabric bag (TOPVALU Dashi Pack, 12.0 cm × 9.5 cm), frozen in liquid nitrogen to stop enzymatic activity, and stored frozen at − 80 °C. For temperature-stabilized analysis, completely frozen materials were freeze-dried in a freeze-dryer (freeze trap: VA-800R, Taitec Corp., Saitama, Japan; vacuum pump: GLD-137CC, Ulvac., Kanagawa, Japan) in a cooling trap at − 70 °C for 7 days until the sample was depressurized below 10 Pa (Kim and Verpoorte 2010; Hamid et al. 2018). The frozen dry matter weight was measured for calculating the dry matter percentage, and the samples were vacuum-packed and frozen until analysis. Freeze-dried samples were placed in a 50 mL crush tube (ST-5010PCR, Yasui Kikai, Co., Osaka, Japan) with a special metal cone (MC-5012R, Yasui Kikai, Co., Osaka, Japan), precooled in liquid nitrogen, and ground using a crusher (Multi-Beads Shocker® MB-3000, Yasui Kikai, Co.) at a rotational speed of 2500 rpm for 30 s in two sets.

Sample extraction

Five mg of freeze-dried sample powder was weighed into a 2 mL microtube. To it, 400 µL of MeOH containing 200 µM internal standard solution [l-methionine sulfone, d-camphor-10-sulfonic acid (CSA), 13C6-glucose, 2-(N-morpholino) ethane sulfonic acid (MES)] was added, and the mixture was stirred (2500 rpm, 3 min) and centrifuged (9100 g, 5 min). Milli-Q water (400 µL) was added to the mixture, and the mixture was stirred again (2500 rpm, 3 min) and centrifuged (4 °C, 9100 g, 10 min) using a cooling centrifuge (MX-307, Tomy Seiko Co., Ltd., Tokyo, Japan). The supernatant (400 µL) was filtered through an ultrafiltration filter (MW 5000 kDa, HMT, Inc., Tsuruoka, Japan) using a cooling centrifugation (4 °C, 9100 g, 3 h). The filtrate was dispensed into 30 µL aliquots for liquid chromatography-mass spectrometry (LC–MS). For capillary electrophoresis-mass spectrometry (CE-MS), 100 µL filtrate was dried using a vacuum centrifuge (CentriVap Concentrator, Labconco Corp., Kansas City, MO, USA) at 4 °C for 4 h. The dried sample was diluted with 20 µL of Milli-Q water containing 200 µM of internal standard for analytical time correction for CE-MS.

Conditions for free sugar analysis using liquid chromatography-tandem mass spectrometry (LC–MS/MS)

LC–MS/MS analysis of the extracted samples was performed using an API 3000 Quadrupole LC/MS/MS tandem mass spectrophotometer system (Sciex, Framingham, MA, USA) equipped with an Agilent 1100 series column oven (Agilent Technologies, Santa Clara, CA, USA), an LC binary pump, and an autosampler (Hamid et al. 2018, 2019; Ogura et al. 2020). The extract (0.3 µL) was analyzed using a hydrophilic interaction chromatography (HILIC) amino column (Asahipak NH2P-50E; 4.6 mm inner diameter (i.d.) × 250 mm length; 5 µm; Showa Denko K.K., Tokyo, Japan) and detected using multiple reaction monitoring in the negative ion mode (Hamid et al. 2019). Cationic metabolites, including amino acids, were analyzed using forward electrophoresis (1 M formic acid) on fused silica capillaries in the positive-ion mode. Anionic metabolites, including sugar phosphates and organic acids, were analyzed using reverse directional electrophoresis (50 mM aqueous acetate-ammonium solution, pH 8.5) on coated capillaries with suppressed electroosmotic flow. Quantification by molecular weight was performed by mass spectrometry using an Agilent capillary electrophoresis system consisting of an Agilent G6220A LC/MSD TOF, an Agilent 1100 series isocratic HPLC pump, a G1603A Agilent CE-MS Adapter Kit, and a G1607A Agilent CE-ESI–MS Sprayer Kit (Agilent Technologies) in full scan mode from m/z 50–1000 (Wakayama et al. 2015). Standard mixtures were prepared and measured for each analysis to identify the peak of each metabolite in the sample solution and to quantify the amount of each metabolite. Finally, the metabolite concentration per dry matter weight was calculated and used for various analyses including multivariate analysis.

Statistical analysis

Multivariate analysis of all metabolite data was performed using principal component analysis (PCA) and hierarchical clustering analysis (HCA) with JMP ver. 17.0.0 (SAS Institute Inc., Cary, NC, USA), as described previously (Hamid et al. 2020; Ogura et al. 2020). HCA was performed using the Ward’s method for two-way clustering of metabolite concentrations and analyzed samples. To elucidate any trend in concentration of metabolites, PCA was performed on all samples and leaf blades only.

To determine the significant differences in the concentration of each metabolite between genotypes and treatments, P values between genotypes and treatments for each metabolite were calculated and volcano plots were created for each organ or part, except the root, for each case.

Analysis of variance (ANOVA) was performed to determine the effects of genotype, water regime, and leaf age (1st, 3rd, and 5th; Fig. 1), and their interactions. Differences in the mean values of leaf gas exchange, anatomical features (n = 4), and concentrations of targeted metabolites between the genotypes or water regimes were determined using the t-test (n = 3). Differences in the mean values for the analyzed leaf ages (1st, 3rd, and 5th) or leaf parts (tip, middle, and base; Fig. 1) between the genotypes or water regimes were determined using the Tukey’s test (P < 0.05). These analyses and tests were performed using the statistical analysis software (Bell Curve for Excel; Social Survey Research Information Co., Ltd., Tokyo, Japan).

Results

Leaf gas exchange and anatomical characteristics in response to drought stress

The trends of soil moisture are shown in Fig. 2. The average VWC in the wet treatments was 34.5 ± 4.6% for the two species (Fig. 2, yellow and blue lines). In the dry treatments, the VWC reached ~ 10% in approximately 4 weeks after the start of the soil drying treatment, and the drought stress was continued for approximately 3 weeks until the end of the experiment (Fig. 2, red and green lines). The average VWC in the dry treatment from September 24 to October 12 was 10.6 ± 1.7% for the two species.

Fig. 2
figure 2

Changes of soil volume water content (VWC) during the experiment. Soil water treatment was started at 43 days after transplanting based on VWC readings (black arrow). Number of x axis in parenthesis indicates days after transplanting. VWC for sugarcane pot under dry and wet conditions was indicated by red and orange lines, respectively. VWC for Erianthus pot under dry and wet conditions was indicated by green and blue lines, respectively

Full size image

The leaf gas-exchange characteristics for each leaf age are listed in Table 1. ANOVA based on the means and significance of differences between genotypes, treatments, and leaf ages are also shown in Tables S1 and S2. Photosynthetic rate (A) and stomatal conductance (gs) were lower in the lower leaves than in the upper and middle leaves. This trend was particularly apparent for sugarcane in the wet treatment and for Erianthus in the dry treatment. A tended to be lower in Erianthus than in sugarcane as a whole. gs of Erianthus was significantly lower than that of sugarcane, regardless of the light conditions, except for the lower leaves in the wet treatment and for the upper leaves in the dry treatment. iWUE was significantly higher in Erianthus than in sugarcane, except for the lower leaves in the wet treatment. Such genotypic differences in gs and iWUE were more remarkable under low-light conditions than under high-light conditions regardless of soil water conditions.

Table 1 Gas exchange parameters measured under different water regime and light intensity conditions
Full size table

Stomatal density and interveinal distance were measured in the sugarcane and Erianthus leaves. No significant difference in stomatal density on the adaxial leaf surface was noted between Erianthus and sugarcane but Erianthus had significantly lower stomatal density on the abaxial leaf surface (Tables 2 and S2). Therefore, the total stomatal density in Erianthus was significantly lower, except in the upper leaves in the dry treatment, and the ratio of stomatal distribution on the adaxial and abaxial surfaces was significantly higher. The interveinal distance was significantly greater in the Erianthus. ANOVA showed that the abaxial stomatal density and stomatal-distribution ratio were less affected by leaf age and treatment and were characteristic for the two species (Tables S1 and S2).

Table 2 Stomatal density and interveinal distance
Full size table

Profiling of plant metabolites in response to drought stress using multivariate analysis

A total of 188 metabolites were detected and quantified. Hierarchical clustering analysis (HCA) was performed using the concentration of each metabolite as an explanatory variable (Fig. 3). The organs were broadly classified from source-capable leaf parts (tip and middle parts of leaf in the two genotypes and base part of sugarcane leaf) to sink organs (base part of the Erianthus leaf, leaf sheath, and stem/root of both genotypes). The basal part of the Erianthus leaf was classified as a sink organ, not a source organ, because the dewlap (conjunction between leaf and leaf sheath) was indistinct, and the basal part of leaf showed a morphology similar to that of the leaf sheath. Furthermore, there was a tendency for clustering according to the genotype and adjacent parts from the leaf blade to the leaf sheath. For each cluster of leaf parts, the effect of the treatments (dry and wet) was moderate within each cluster (Fig. 3, arrows).

Fig. 3
figure 3

Hierarchical clustering analysis (HCA) and heatmap of dry weight-based metabolite concentrations after freeze-drying. All quantified concentrations were normalized to Z-scores for metabolite concentration. Blue and red indicate low and high concentrations, respectively. The legends for the plant parts are shown in. Color-changing arrows indicate soil water conditions (dry: yellow; wet: blue)

Full size image

PCA was performed using the metabolite concentrations in all plant parts as explanatory variables. Despite the integration analysis of the two genotypes, principal component (PC) 1 (32% contribution) showed a difference between the stem and other organs, whereas PC2 (15.3%) showed a difference between the leaves and other organs. The difference between the two genotypes was confirmed as PC3 (6.74%). Although PC1–PC5 were analyzed, the effects of the treatment were not detected.

To determine the effects of genotypes and treatments, PCA was performed using only the metabolite data for the leaves (Fig. 4), in which these effects were evident in the cluster analysis (Fig. 3). PC1 (24.6% contribution) showed a difference between the two genotypes, PC2 (14.6%) showed a difference between the leaf parts, and PC3 (8.63%) showed the effect of the treatment (Fig. 4). In contrast, the effect of leaf age was weak in PC5 (6.29%). Therefore, a loading plot of PC1, which represented the difference between the two genotypes, and PC3, which represented the effect of the treatments, allowed us to capture the metabolites that specifically accumulated in each genotype under each soil-moisture condition (Fig. 5). This type of loading plot was used to qualitatively analyze the concentrations of metabolites. For example, in the negative region of PC1, which indicates that the metabolite concentration is likely to be high in Erianthus, we found metabolites, such as betaine, GABA, glutamine, proline, and raffinose, whereas in the positive region of PC1, which indicates that the metabolite concentration is likely to be higher in sugarcane, we found metabolites, such as isocitrate, pyruvate, glutamate, mannitol, spermine, and putrescine. In the positive region of PC3, which indicates high concentration under dry conditions, sugars and polyols (sucrose, glucose, and raffinose), amino acids (tyrosine and tryptophan), and polyamines were found.

Fig. 4
figure 4

Principal component analysis (PCA) score plots for leaf parts. The metabolite concentrations used for the PCA were calculated based on the dry weight after freeze-drying. Percentage in parentheses indicates the contribution ratio of each principal component. The PCA loading scores are presented in Table S4

Full size image
Fig. 5
figure 5

Principal component analysis (PCA) loading plots for leaf parts. The metabolite concentrations used for the PCA were calculated based on the dry weight after freeze-drying. Percentage in parentheses indicates the contribution ratio of each principal component. The PCA loading scores are listed in. Red, blue, green, and pink symbols indicate selected polyols and saccharides, amino acids, organic acids, and nitrogen-related metabolites other than amino acids, respectively

Full size image

Changes in metabolite concentrations in different organs and parts under different water conditions

The metabolites were classified into different chemical component groups, and the concentrations and compositions in the plant organs and parts are shown using stacked graphs (Fig. 6). The stacking graph for the accumulated concentrations shows that polyols/sugars and amino acids were markedly affected by soil drying, regardless of the genotype (Fig. 6a, c). In contrast, N metabolites other than amino acids showed little change in concentrations with soil drying, and only the concentration of betaine in Erianthus tended to increase with soil drying (Fig. 6e). The concentrations of organic acids decreased slightly in the sugarcane stalks after soil drying; however, the effect of drying was generally small in both species (Fig. 6g). It is clear from part-specific data that each component group accumulated more in the stem than in the other sites (Fig. 6a, c, e, g). To determine the metabolite composition of each plant part, the metabolite gradient between adjacent parts from the root to the leaf tip was determined (Fig. 6b, d, f, h).

Fig. 6
figure 6

Profiling of selected metabolites in continuous whole plant parts under different water regimes. Metabolite concentrations were calculated based on the dry weight after freeze-drying. a, c, e, g Metabolite concentrations of polyols and saccharides, amino acids, other N-related metabolites, and organic acids. b, d, f, h Composition ratios of polyols and saccharides, amino acids, other N-related metabolites, and organic acids

Full size image

Regarding the polyol/sugar composition, the percentages of fructose in sugarcane and glucose in Erianthus were particularly high under dry conditions (Fig. 6b). Genotypic differences in amino acid composition were small, with a high composition of glutamate and alanine in the leaves, and a high composition of aspartate and asparagine in the leaf sheath, which decreased toward the stem and roots (Fig. 6d). N metabolites, other than amino acids, were distinctly genotype-specific, with higher proportions of choline and polyamines in sugarcane, and betaine and GABA in Erianthus (Fig. 6f). Remarkable genotypic differences in the composition of organic acids were observed in the leaves, with a higher proportion of isocitrate in sugarcane, and higher proportions of malate, shikimate, and quinate in Erianthus (Fig. 6h).

To identify the components that showed significant differences between genotypes or treatments, volcano plots were prepared for each plant part other than the root, plotting P values for each metabolite for genotypic and treatment differences. The number of metabolites with significant differences (P < 0.01) between the species and between the treatments tended to be higher for the leaves and lower in the parts that were closer to the stem. Metabolites with significant differences in concentrations between treatments tended to be fewer in each plant part than those with significant differences between genotypes. The concentrations of betaine and GABA tended to be higher in Erianthus in many parts, especially in the leaves, whereas isocitrate concentration was markedly higher in sugarcane leaves. The differences between treatments showed a tendency for raffinose to increase consistently with drying.

For precise identification of the metabolites associated with stress response, we displayed the changes in the concentration of 28 metabolites, selected based on the PCA factor loadings for the leaf, in the plant parts (Fig. 5). Methionine, glutamine, GABA, betaine, and S-adenosylmethionine (SAM) accumulated more in Erianthus leaves than in sugarcane leaves under both dry and wet conditions, whereas their accumulated in the sugarcane stem was higher than in other plant parts (Fig. 7). In contrast, the accumulation of raffinose was less in sugarcane in leaves or stems, regardless of the part. GABA showed contrasting part-specific gradients between the genotypes, with sugarcane showing higher concentrations near the root, and Erianthus showing higher concentrations near the root and leaf tips.

Fig. 7
figure 7

Selected metabolite concentrations in leaf parts under different water regimes. Metabolite concentrations were calculated based on the dry weight after freeze-drying. Significant difference was analyzed using the mean concentrations of the three leaf ages

Full size image

Changes in specific leaf metabolites in response to drought stress

As mentioned above, the effect of leaf position on metabolite variation was low, as determined using PCA. Therefore, the concentrations of the 28 metabolites mentioned above was averaged without considering the leaf age and examined as described below (Fig. 7). ANOVA results and the significance of differences in metabolite concentrations between the genotypes, treatments, and leaf parts are presented in Tables S5 and S6.

Analysis of genotypic differences in the accumulation of polyols and sugars revealed that sugarcane had higher concentrations of mannitol, inositol, and fructose, whereas Erianthus had higher concentrations of raffinose, glucose, and sucrose (Fig. 7). Among the amino acids, glutamate concentrations were higher in sugarcane, whereas the concentrations of alanine, glutamine, histidine, methionine, proline, tyrosine, and tryptophan were higher in Erianthus. Among the N metabolites other than amino acids, the concentrations of putrescine and spermine tended to be higher in sugarcane, whereas those of betaine, GABA, SAM, and cyclohexylamine were higher in Erianthus. Among the organic acids, C4 photosynthesis-related metabolites (citrate, isocitrate, malate, and pyruvate) were highly accumulated in the sugarcane.

Regarding differences between treatments, the concentrations of all polyols and sugars increased with soil drying (Fig. 7). In Erianthus, the concentrations of alanine, histidine, proline, tyrosine, and tryptophan increased with soil drying. In contrast, in sugarcane, the concentrations of histidine, tyrosine, and tryptophan decreased with soil drying. Accumulation of polyamines tended to increase with soil drying, particularly in Erianthus. The pyruvate concentration tended to increase with drying. Shikimate and quinate concentrations were higher in Erianthus, particularly under dry conditions. These metabolites are generated from the shikimate pathway, which biosynthesizes aromatic amino acids, followed by the biosynthesis of secondary metabolites, such as antioxidants and lignin. Among these, the concentrations of raffinose, glutamine, betaine, GABA, and SAM were particularly high in Erianthus, regardless of the dry or wet conditions. These trends were generally consistent with those in the PCA loading plot (Fig. 5).

Metabolite concentrations in different leaf parts exhibited significant part-specific differences, with the middle parts showing values that were intermediate between those for the tip and basal parts for many selected metabolites.

Discussion

Effects of leaf age and light conditions on genotypic differences in gas exchange and WUE

Erianthus maintained low gs regardless of leaf age (Table 1) and thus maintained high iWUE without being affected by differences in stomatal responses due to aging. In general, low gs is associated with low stomatal density (Franks et al. 2015; Caine et al. 2019; Pitaloka et al. 2022), and the low gs of Erianthus in the present study was consistent with low stomatal density, especially on the abaxial surface, and high stomatal-distribution ratio (adaxial/abaxial) (Table 2). High photosynthetic activity, even at low gs (i.e., high iWUE), may involve metabolites, such as photosynthetic enzymes and osmolytes involved in maintaining the water potential (see discussion below).

The differences in iWUE between the two genotypes were greater under low-light than under high-light conditions. In general, under low-light conditions, photosynthetic capacity is reduced, and iWUE decreases because of excessive transpiration (Ye et al. 2020; Ozeki et al. 2022). Although, in this study, young plants were grown under pot conditions with low self-shading, the low stomatal responsiveness of sugarcane under low-light conditions (Table 1) can significantly suppress photosynthesis in self-shading canopies grown under field conditions (Almeida et al. 2022), leading to a further decline in iWUE. Because C4 leaves with low-light acclimation have special anatomical characteristics (e.g., cell wall suberization and aggregated arrangement of related cells), which causes less bundle-sheath leakiness (von Caemmerer and Furbank 2003), it is necessary to focus more on the ultrastructural anatomical characteristics of Erianthus leaves in future research.

The results of ANOVA showed that the effect of leaf age on A and gs, but not on iWUE, was significant and there was no effect of interactions, including leaf age, especially under unsaturated-light conditions (Table S1). Therefore, when screening iWUE, it was necessary to select the first to the fifth leaf and to evaluate iWUE under unsaturated light with uniform soil-moisture conditions. This result should draw our attention to the fact that many previous studies on sugarcane iWUE have been conducted under saturated light conditions (photosynthetic photon flux density (PPFD) ≈ 2000 µmol m−2 s−1) using the youngest fully expanded leaf to achieve maximum photosynthetic rates (e.g., Jackson et al. 2016; Natarajan et al. 2021). In addition to physiological traits, such as gas-exchange rate and metabolite concentrations, leaf anatomical traits varied less with environmental and sampling conditions (Tables S1), which may be more suitable for genotypic comparison of iWUE and, consequently, for screening applications if the measurement throughput is improved.

The high iWUE-related leaf metabolites responding to drought are common in the two genotypes

Plants close their stomata in response to soil drought and stress and regulate plant water potential by accumulating osmolytes and other metabolites in response to the deteriorating plant water status (Boyer 1995; Hare et al. 1998). The responses of each parameter to soil drying are summarized in Table 3. In both genotypes, soil drying decreased A and gs and increased iWUE, stomatal density, and intervascular distance, whereas the stomatal density ratio was not changed much (Table 3). In addition, the concentrations of all the analyzed polyols and sugars were increased by soil drying for both genotypes. Polyols/sugars are known to accumulate under drought and salt stress and function as osmolytes in a wide range of species (Ozturk et al. 2021). Among amino acids, alanine is commonly increased with soil drying. Alanine is produced via glutamate and GABA in a GABA biosynthesis pathway called the GABA shunt (Miyashita and Good 2008) and its concentrations are commonly known to be increased under drought stress conditions (Rhodes et al. 1986). Among N metabolites other than amino acids, soil drying increased the concentrations of SAM, spermine, and spermidine, whereas little change was noticed for choline, GABA, and cyclohexylamine, which responded to drought and were present in similar amounts in the two species. Polyamines (spermine and spermidine) and their precursor, SAM, are stress-responsive compounds that accumulate under stress (Bouchereau et al. 1999). GABA is involved in stomatal closure (Kinnersley and Turano 2000; Hasan et al. 2021), and choline is a precursor of betaine, a stress-response compound (Sakamoto and Murata 2002), that may accumulate with or without stress to reduce the damage caused by the stress. Among the organic acids, the concentration of pyruvate increased, that of citrate decreased, and that of isocitrate changed slightly under drought; the responses were similar for the two genotypes (Table 3). Pyruvate is an important metabolic component of the C4 photosynthetic cycle as it is a source of PEP (Sage et al. 2013); its concentrations increase to enable alanine synthesis for stress adaptation (Rhodes et al. 1986). Both citrate and isocitrate are intermediate metabolites in the tricarboxylic acid cycle and osmotic stress has been reported to decrease the citrate concentrations (Dabravolski and Isayenkov 2023). Thus, we were able to classify metabolites that commonly respond to drought in sugarcane and Erianthus.

Table 3 List of selected leaf metabolites affected by soil water regime in each genotype
Full size table

Responses of iWUE-related leaf metabolites to drought in Erianthus

In addition to polyols and sugars, the concentrations of amino acids, such as proline and betaine, shikimate pathway metabolites, polyamines, and N-related metabolites were increased by drought in Erianthus (Table 3). These diverse metabolites act as osmolytes and stress defense molecules (Raggi 1994; Weaver and Herrmann 1997), and their concentrations are suggested to increase in Erianthus under drought conditions, resulting in high iWUE. The metabolomics profiles of sugarcane leaves under drought stress have been compared between drought-tolerant and drought-susceptible varieties (Vital et al. 2017). Raffinose, glucose, betaine, proline, and tryptophan, which were commonly found at high concentrations in drought-tolerant varieties in a previous study and in Erianthus in this study, may be highly concentrated metabolites common to drought-tolerant genotypes. In addition, high concentrations of stress-responsive substances, such as GABA and SAM, were present in Erianthus under dry conditions, but not in sugarcane (Fig. 5). GABA is a non-protein amino acid biosynthesized from polyamines via proline and glutamate and is also implicated in plant stress responses (Kinnersley and Turano 2000; Hasan et al. 2021). In particular, its involvement in maintaining turgor pressure and low gs has been noted (Xu et al. 2021), which is consistent with the low gs in Erianthus (Table 1).

The iWUE of Erianthus was higher than that of sugarcane, even under wet conditions, except for the lower leaves (Table 1). The stomatal-distribution ratio and methionine, glutamine, and GABA concentrations exhibited minor changes upon drying, and were higher than those in sugarcane (Table 3). In addition, the concentrations of raffinose, betaine, and SAM were higher in Erianthus than in sugarcane under both dry and wet conditions, although they increased under conditions of drought (Fig. 7). Therefore, the stomatal-distribution ratio and concentrations of these six metabolites are likely involved in establishing a potentially high iWUE for Erianthus.

Responses of iWUE-related leaf metabolites to drought in sugarcane

In sugarcane leaves, osmolytes, such as mannitol, inositol, fructose, and polyamines (spermine and putrescine), as stress-response substances, glutamate as a precursor of GABA, and isocitrate as a precursor of glutamate, alanine, and pyruvate, which are used in alanine synthesis, were highly concentrated under dry conditions (Figs. 5 and 7). Proline, GABA, SAM, glutamine, polyamines, and methionine are closely involved in metabolic pathways and have been implicated in drought tolerance in plants (Ravanel et al. 1998; Silvente et al. 2012; Hasan et al. 2021). In particular, sugarcane leaves showed higher glutamate concentrations, but lower concentrations of glutamine, proline, and GABA, which are produced from glutamate, than the respective concentrations in Erianthus (Figs. 5 and 7). Moreover, polyamines are believed to be the precursors to GABA; however, the high polyamine concentration in sugarcane was not linked to increased GABA concentration in this study. This suggests that biochemical and physiologic characteristics, such as the abundance and degree of activity of polyamine dehydrogenase (polyamine oxidase) (Prabu et al. 2011) and the negative feedback between GABA and polyamine biosynthesis (Wang et al. 2014), may be related to genotypic differences in iWUE. Glutamate, alanine, and pyruvate are components of the GABA biosynthesis pathway, known as the GABA shunt (Miyashita and Good 2008; Hassan et al. 2021). Moreover, isocitrate, a metabolite in the citric acid cycle, is consumed in the synthesis of glutamate via 2-oxaglutarate (Popova and de Carvalho 1998). Because sugarcane produces large amounts of isocitrate, it was suggested that large amounts of glutamate were synthesized from the abundant isocitrate, and large amounts of alanine were synthesized from the abundant glutamate and pyruvate (Figs. 6 and 7). Sugarcane may be unique in that the abundant glutamate is consumed in alanine synthesis under abundant pyruvate conditions, and therefore, does not lead to an increase in GABA concentration.

Sugarcane accumulated less amounts of the osmolyte betaine and more of choline, which is a betaine precursor, regardless of the dry or wet condition (Figs. 6 and 7). The low activity of the catalytic enzyme choline monooxygenase, which oxidizes choline (Sakamoto and Murata 2002), was considered as the reason why the increase in choline concentrations was not linked to the increase in betaine concentrations in sugarcane. In terms of the trends in part-specific accumulation, betaine tended to accumulate more in the stem, although in smaller amounts, in sugarcane, whereas it tended to accumulate in both the stem and leaf blades in Erianthus. This difference in accumulation between sites implies a preferential target for protection against stress (Hattori et al. 2009), and suggests that in sugarcane, the priority is to maintain the moisture status in the stem.

Challenges in biomarker identification and the need for a detailed investigation of the causal relationship of metabolites with iWUE

HCA suggested that leaves were associated with interspecific differences in iWUE (Fig. 3). Furthermore, the results of the PCA for the leaf (Fig. 4) suggested that leaf sampling should be carefully performed for metabolome analysis, and that the effect of leaf age from the first to the fifth youngest leaf was small. The small effect of leaf age on metabolite concentrations was consistent with the small effect of leaf age on iWUE (Table S1), supporting the lesser need to consider the effect of leaf age, at least at the 4-month-old growth stage, when assessing iWUE-related metabolites (i.e., for creating biomarkers).

The results of this study could not completely clarify the causal relationship between the various metabolites, iWUE, and drought tolerance. Therefore, with regard to GABA, it is necessary to conduct transcriptome analysis in future and to analyze the activity of each degradative enzyme in combination with metabolomics to clarify the causal relationship between GABA biosynthesis and stress response in the two genotypes. If these specific metabolites are to be used as biomarkers for high-throughput phenotyping of WUE and drought tolerance, it is desirable to screen them in the very early growth stages when they are relatively unaffected by the environment (i.e., when the genotype effect is stronger). However, the phenotype and gene expression may vary with growth stage and time (Boyes et al. 2001), and there may be discrepancies between plants at the very early growth stage in greenhouses and mature plants established in the field (Strock et al. 2022). Regarding the effects of growth stage on metabolome sampling, it is also necessary to investigate the variation among individuals. Therefore, a more precise verification, such as expression analysis on a timescale using several replicates, is required in future. In addition, it is necessary to further investigate and analyze the population of intergeneric hybrids between sugarcane and Erianthus for quantitative trait locus analysis to create genetic markers.