Abstract

Background

   Alkaline stress poses severe threats to sustainable triticale
   production. However, lack of molecular and metabolic data restricts the
   efficient breeding and field management of triticale cultivated in
   alkaline soils. The aim of this study was to explore the genotypical
   effects on modifications of transcriptional and metabolic profile in
   triticale, and find out the prospective genes and metabolites
   responsible for alkaline tolerance.

Results

   In this study, we compared the root development in two triticale
   cultivars with contrasting alkali tolerance, and found that root
   number, length, surface area and biomass were reduced, but the average
   root diameter was increased in triticale subjected to alkaline stress.
   The stress effect was extremely significant in the alkali-sensitive
   cultivar. Comparative transcriptomic and metabolomic analyses revealed
   that the genotype effect on alkaline stress resistance was
   predominantly connected with metabolism of amino acids and flavonoids,
   as well as the featuring biosynthesis of benzoxazinoid and
   brassinosteroid. Simultaneous regulation of genes and metabolites
   involved in betalains, dopa and a group of other pathways in the two
   cultivars were suggested to be basic alkaline stress responses in
   triticale. Additionally, expression of key genes involved in these
   processes and typical alkaline stress responses in plants were examined
   and the subset including AT1-PIP2, RGI1, SAUR215, SCaBP3 indicated
   their implication in alkali tolerance in triticale.

Conclusions

   These gene, metabolite, and pathway resources depict the internal
   responsive atlas of triticale under alkaline stress. Those involved in
   the metabolism of key amino acids, flavonoids, and betalains as
   mentioned above represent prior targets for future genetic studies and
   breeding of stress-tolerant triticale germplasm to cope with alkaline
   stress.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12870-025-06973-1.

   Keywords: Triticale, Alkaline stress, Transcriptome, Metabolome,
   Comparative analysis

Introduction

   Triticale (× Triticosecale Wittmack), developed by crossing wheat
   (Triticum spp.) and rye (Secale cereale), is an allopolyploid, annual
   grass of the Poaceae family. Due to its high yield and extensive
   adaptability, triticale has been grown worldwide for grain and forage
   production. In China, triticale is mainly cultivated in the northern
   area as a substitute of wheat in marginal farmland, winter fallow
   fields, and alpine pastures. However, although triticale displayed
   relative strong resistance to various stresses, its plant growth and
   yield are seriously threatened by harsh soil conditions such as the
   widely distributed saline-alkali soil in northern China, where the
   average pH of cultivated land is above 8.0 [[34]1].

   High concentrations of bicarbonate and carbonate in the soil are the
   main cause of alkaline stress for plants [[35]2]. In addition to the
   osmotic pressure and ion toxicity similar to neutral salts, the
   presence of high pH and bicarbonate also disrupts the absorption and
   utilization of other nutrients such as nitrogen and potassium, and
   transforms various mineral elements such as phosphorus, calcium, and
   iron in the soil into insoluble forms [[36]2, [37]3]. Moreover, high pH
   leads to direct damage of the plasma membrane and other structures of
   plant roots, disrupting the normal metabolic activities of cells
   [[38]4–[39]6]. When grown in compound salt and alkali stress at pH 8.0,
   the biomass, chlorophyll content, photosynthetic efficiency, and
   nitrogen absorption efficiency of common wheat significantly decreased;
   Compared with sensitive varieties, enhanced activity of glutamine
   synthase in the roots of tolerant wheat varieties promotes nitrogen
   metabolism, resulting in more robust plant growth [[40]7–[41]9].
   Besides, alkaline stress also causes osmotic and oxidative stress in
   cells of rye seedlings [[42]10]. Flavonoid synthesis, amino acid
   metabolism, energy metabolism, fatty acid metabolism, alkaloid
   biosynthesis, phenylpropanoid biosynthesis, as well as plant signal
   transduction pathway significantly changed in wheat, rice, maize,
   alfalfa or bermudagrass as responses to alkaline stress [[43]6,
   [44]11–[45]14]. This indicates that metabolite content adjustment in
   the root system is very important for plants to resist alkali salt.

   Considering the complex effects caused by alkaline stress, plants are
   supposed to respond to alkaline stress in a composite molecular way.
   Putatively in Arabidopsis, alkali induced osmotic stress is sensed by
   AtOSCA1 [[46]15], ionic stress by AtGIPC [[47]16], extracellular high
   pH by AtRGF1-RGFRs/RGIs and AtPep1-PEPRs [[48]17], and unknown
   HCO[3]^−/CO[3]^2− sensors. Subsequently, several pathways mediated by a
   couple of genes respond accordingly, among which modules regulate the
   plasma membrane H^+-ATPase was considered to serve as a central
   component conferring alkali tolerance in Arabidopsis. Distribution of
   cytoplastic Ca^2+ changed responding to the upstream signals, and Ca^2+
   binding of AtScaBP3 leads to release of its inhibition to AtAHA2. The
   activity of the AtScaBP3-AtPKS5 complex is also reduced and further
   activates plasma membrane H^+-ATPase AtAHA2 [[49]18]. A chaperone AtJ3
   activates plasma membrane H^+-ATPase activity by physically interacting
   with and repressing AtPKS5 kinase activity [[50]19]. In wheat,
   Ca^2+-dependent TaCCD1 cooperates with TaSAUR215 to enhance plasma
   membrane H^+-ATPase activity and alkali stress tolerance by inhibiting
   TaPP2C.D1/8-mediated dephosphorylation of plasma membrane H^+-ATPase
   TaHA2 [[51]20]. Some distinct pathways are also reported to play roles
   in plant responses to alkaline stress. For example, HCO[3]^− increases
   the transcript level of GsBOR2 in root, thus promoting the alkali
   tolerance of Glycine soja [[52]21, [53]22]; Overexpression of genes
   encoding Glycine soja transcription factor Gshdz4 or GsNAC019 could
   enhance alkaline tolerance in transgenic Arabidopsis [[54]23]; A G
   protein γ subunit SbAT1 regulates crop alkaline tolerance by modulating
   SbPIP2 aquaporins and thus reactive oxygen species (ROS) homeostasis in
   sorghum and several other crops [[55]24, [56]25]; Mutation of SlSCaBP8
   resulted in compromised tolerance of alkaline stress in tomato
   [[57]26].

   Similar with other plants, alkaline stress adversely affects all growth
   stages of triticale, including seed germination, seedling development,
   reproductive growth, and seed formation [[58]27, [59]28]. Triticale
   adaptation to alkaline soils was hypothesized to depend largely on the
   single wheat/rye chromosome 2D(2R) substitution, with the substituted
   types being generally more tolerant [[60]29]. However, unlike the
   diverse resources available for model plants and several staple crops
   summarized above, the molecular and metabolic mechanism about alkaline
   stress resistance in triticale is very limited. Considering root as the
   first organ in plant when facing various soil constrains, in this
   study, we integrated transcriptomic and metabolomic analyses in
   triticale root to uncover the differential adaptive mechanism towards
   alkaline stress in two triticale cultivars with different alkaline
   tolerance at the seedling stage. The molecular and metabolic responses
   of triticale were further compared to previous research on other plants
   to identify universal and specific mechanisms. These findings will
   contribute to our comprehension for breeding smarter triticale
   germplasm with strong alkaline tolerance.

Materials and methods

Plant materials and growth conditions

   Triticale used in this study include an alkali-sensitive cultivar
   Mule3000 (short for ML, China originated, obtained from Clover
   (Beijing) Grass Science and Technology Research Co., Ltd.) and an
   alkali-tolerant cultivar Grenado (short for GR, Poland originated, kept
   in Lab of Grass Germplasm Resources and Abiotic Stress Biology
   (Northwest A&F University)). Healthy and uniform seeds were
   surface-sterilized by soaking in 10% hydrogen peroxide for 30 min and
   followed by rinsing in distilled water for 5 times. The seeds were then
   placed on filter paper in a 15 cm-diameter petri dish and kept in a
   growth chamber to discriminate the uniformly germinated seeds. After
   one day, the radicle protruded seed were carefully transplanted to
   rectangle pot (10 cm × 10 cm × 8.5 cm, 8 seedlings per pot) with
   perlite used as culture substrates. For alkaline treatment (AK), the
   seedlings were watered with 30 mM mixed alkali solution
   (NaHCO[3]:Na[2]CO[3] = 5:1, pH = 9.43) [[61]30, [62]31] and 1/2
   strength Hoagland solution (pH = 7) alternately every other week to
   prevent directly minerals precipitation occurred during direct solution
   mixing. No addition of alkali was set as control (CK). During all
   cultivation, the growth chamber was set to maintain a photoperiod of
   16 h light/8 h dark and a temperature of 25/20 °C (light/dark). After
   three weeks, shoots and roots from regularly growing seedlings were
   collected separately to measure height and biomass, and three
   biological replicates—each containing at least 4 seedlings—were used
   per treatment-cultivar combination. For transcriptomic, qRT-PCR, and
   metabolomic analyses, whole roots from 5–6 seedlings were pooled to
   form one biological replicates. These samples were flash-frozen in
   liquid nitrogen and stored at − 80 °C until RNA and metabolites
   extraction, with three replicates per treatment and cultivar.

Morphological analysis

   Plant samples were cleaned by rinsing via deionized water. Height and
   fresh weight were manually measured directly. For determination of dry
   weight, fresh shoots or roots were dried in an oven at 105 °C for
   20 min immediately after sampling and then kept at 65 °C till constant
   weights. For analysis of root system architecture, intact roots were
   separated and cleaned from substrates by washing using tap water. The
   fresh roots were then manually unraveled in water and scanned using a
   ScanMaker i850 scanner (Microtek, Shanghai, China). Subsequently, the
   obtained images were loaded into RhizoVision Explorer [[63]32] for
   analysis of root length, number, diameter, volume and surface area.
   Three biological replicates, with each one imaging at least three
   independent roots, were performed for each treatment.

RNA extraction, RNA-Seq library construction and transcriptome sequencing

   Total RNA from different samples were extracted and purified using
   ethanol precipitation protocol and CTAB-PBIOZOL reagent as described by
   Li et al. [[64]33]. The purified RNA was subsequently dissolved in
   DEPC-treated water, and quantified using a Qubit 4 Fluorometer (Thermo
   Fisher Scientific, USA) and a Qsep400 high-throughput biofragment
   analyzer (Bioptic, China). The following RNA fragmentation, RNA-Seq
   library construction and qualification processes were performed using
   NEBNext Ultra RNA Library Prep Kit for Illumina (NEB, Ipswich, USA)
   according to the manufacturer’s instructions. Briefly, mRNAs with polyA
   tails were enriched by Oligo(dT) magnetic beads and cleaved into small
   fragments with fragmentation buffer at a suitable temperature.
   First-strand cDNAs were produced by reverse transcription using a
   random hexamer primer, and the second strands were synthesized while
   simultaneously performing end repair and dA-Tailing. Sequencing adapter
   were ligated and followed by DNA magnetic bead purification and
   fragment selection, to yield a library with 250–350 bp insert
   fragments. After that, the ligated products were amplified by PCR and
   purified again using DNA magnetic beads. Qubit 4 Fluorometer was reused
   for concentration detection, followed by fragment size detection via
   Qsep400 high-throughput biofragment analyzer. Finally, the different
   libraries were sequenced on Illumina NovaSeq X Plus platform after
   pooling according to the effective concentration and the target
   sequencing output data volume, yielding 150 bp paired-end reads.

Metabolites extraction and widely targeted metabolomics analysis

   Samples were vacuum freeze-dried in a lyophilizer (Scientz-100F), and
   then ground (30 Hz, 1.5 min) to powder using a grinder (MM 400,
   Retsch). Next, 1,200 μL 20 °C pre-cooled 70% methanolic aqueous
   internal standard extract was added to 50 mg of each sample powder,
   followed by 6 times of a 30 s-vortex every 30 min. Subsequently, the
   suspension was centrifuged at 15,000 xg for 3 min), and the metabolite
   supernatant was collected by aspiration and filtering through a
   microporous membrane (0.22 μm).

   The obtained metabolite solution was loaded onto an ultra-performance
   liquid chromatography (UPLC, ExionLC™ AD, CA, USA) platform coupled
   with ESI-triple quadrupole-linear ion trap tandem mass spectrometry
   (QTRAP)-MS/MS system (AB Sciex, CA, USA) for metabolite separation and
   detection. The Agilent SB-C18 (1.8 µm, 2.1 mm × 100 mm) analysis
   columns were balanced till stable in solvent A (pure water with 0.1%
   formic acid), and 2 μL metabolite solution was then automatically
   loaded onto the columns and separated at a flow rate of 0.35 mL/min
   with a linear gradient of solvent B (acetonitrile with 0.1% formic
   acid) in 14 min. The gradient was started from 5 to 95% over 9 min,
   and then maintained at 95% for another 1 min. At the next 1 min, the
   solvent B was decreased to 5% and kept unchanged till the end. The
   effluent was alternatively connected to MS/MS system. The ESI source
   temperature was set as 500 °C and the ion spray voltage (IS) was 5500 V
   (positive ion mode)/−4500 V (negative ion mode). The ion source gas I
   (GSI), gas II(GSII), and curtain gas (CUR) were set at 50, 60, and 25
   psi, respectively. High collision-activated dissociation (CAD) was
   selected and QQQ scans were acquired as multiple reaction monitoring
   (MRM) experiments with collision gas (nitrogen) set to medium. DP
   (declustering potential) and CE (collision energy) for individual MRM
   transitions was done with further optimization. A specific set of MRM
   transitions were monitored for each period according to the metabolites
   eluted specifically. To monitor the technical repeatability of the
   UPLC-MS/MS detection, a quality control (QC) sample was prepared by
   mixing different sample extracts and tested along every ten samples
   (Fig. S[65]1).

Qualitative and quantitative analyses of transcripts and metabolites

   The genome sequences and annotations of Triticum turgidum L. (cultivar
   ‘Svevo’) and Secale cereale L. (inbred line ‘Lo7’) were downloaded from
   Ensembl Plants (Release-59) and merged to use as reference in this
   study. The original fluorescence image files generated from RNA-Seq
   were transformed to short reads (raw data) by base calling and recorded
   in FASTQ format. Data quality control was performed using fastp
   (version 0.23.2) to remove reads with adapters or of low quality. The
   obtained clean reads from each sample were aligned separately to the
   reference genome using hisat2 (version 2.2.1) with the default settings
   for paired-end reads. The SAM files output by hisat2 were then sorted
   and transformed to BAM format via Samtools (version 1.16.1). Besides,
   StringTie (version 2.1.6) was used to assemble the alignments into
   transcripts and distinguish novel transcripts. Finally, the reads
   mapping to each gene were counted via FeatureCounts (version 2.0.3),
   with “-M –fraction” options being specified to make a fractional count
   of the multi-mapping fragments, considering the high duplicate
   attribute of the triticale genome as an allopolyploid.

   The secondary spectrum information obtained from MRM multi-peak graphs
   were used to perform qualitative and quantitative analysis of
   metabolites based on a local metabolic database MWDB constructed by
   Metware Biotechnology Inc (Wuhan, China) via Analyst (version 1.6.3).
   Signals from isotopes, and repeated signals containing cations such as
   K^+, Na^+, and NH[4]^+ were excluded during the qualitative analyses.
   Chromatographic peaks were further integrated and corrected using
   MultiQuant software.

Differential analysis and functional interpretation of gene and metabolites

   For determination of gene expression alteration, the R package DESeq2
   (version 1.40.2) was used to normalize the read counts over different
   samples and evaluate the gene expression differences between treatments
   based on a two-factor design with interaction strategy. For comparison
   of treatment effects on gene expression in each cultivar (ML.AKvCK and
   GR.AKvCK), a significantly differentially expressed gene (DEG) was
   defined when the normalized read counts changed by at least twofold
   (|log2(fold change)|≥ 1) and the adjusted p-value was less than 0.05.
   For evaluation of interaction effects of genotype and treatment
   (GRvML.AKvCK), a gene with adjusted p-value less than 0.05 was
   considered as a DEG. Gene Ontology (GO) and Kyoto Encyclopedia of Genes
   and Genomes (KEGG) ortholog and pathway annotation was obtained by
   sequence alignment with public databases via Diamond blastx (version
   2.0.9.147). Alignment results from TrEMBL and Swiss-Prot databases were
   integrated for GO annotation, while wheat gene resources were used as
   reference in KEGG database. Gene homologs were predicted by blast
   (version 2.12.0 +) searching of the candidate gene sequences to the
   reference gene CDS sequences obtained in the present study, and a gene
   with identity% > 50%, E-value < 10^–15, and top ranked bitscore was
   assigned as a homolog.

   For the metabolomic data, orthogonal partial least squares-discriminant
   analysis (OPLS-DA) was implemented to discriminate the differences
   between treatments in each genotype. The OPLS-DA was conducted using R
   package MetaboAnalystR (version 1.0.1) and the metabolite data was log2
   transformed and mean centered before analysis. In order to avoid
   overfitting, a permutation test (200 permutations) was performed.
   Significantly differentially accumulated metabolites (DAMs) between
   treatments but in same genotype were determined as variable importance
   in projection (VIP) greater than 1), fold change (FC) of accumulation
   level no less than 1.5, and adjusted p-value below 0.05. To evaluate
   the interaction effects between genotype and treatment, two-way
   analysis of variance (ANOVA) was performed on the metabolite
   concentration value using R package rstatix (version 0.7.2), and a DAM
   was defined as adjusted p-value less than 0.05. Functional annotation
   of the metabolites was retrieved from the KEGG COMPOUND and PATHWAY
   database.

   The assigned annotations for genes and metabolites were further
   enriched using R package clusterProfiler (version 4.12.6), with all the
   genes or metabolites detected and received respective annotation in
   this study set as the background. To simplify GO enrichment results of
   genes, the significantly enriched GO biological process terms (number
   of enriched genes > 5 and adjusted p-value < 0.05) were further
   clustered using GOMCL (version 0.0.1) and visualized via Cytoscape
   (version 3.10.1).

Quantitative real-time PCR (qRT-PCR) analysis

   Qualified total RNA was reverse transcribed into cDNA using Fast Quant
   RT Kit with gDNase (Tiangen Biotech Co., Ltd., China). qRT-PCR
   amplifications were performed using PerfectStart® Green qPCR SuperMix
   (TransGen Biotech, China) on a LightCycler480 II thermal cycler (Roche,
   Switzerland). Target genes were randomly selected from the DEGs listed
   in the figures below, with the TaGAPDH [[66]34] and TaCyc [[67]35] used
   as internal reference. Three replicates were performed for each
   treatment. The primers used in this study is listed in Table S[68]1.

Statistical analysis

   Variance comparison and ANOVA were performed using the R package
   rstatix (version 0.7.2) on evaluation of differences between samples in
   morphological, physiological and individual gene expression studies.
   After ANOVA, least significant difference (LSD) multiple range test
   (p < 0.05) via R package agricolae (version 1.3–6) were performed to
   compare the differences between samples. In differential analysis and
   enrichment analysis, Benjamini & Hochberg method was applied for
   correction of original p values. All results were presented as the
   mean ± SE from at least 3 independent biological replicates.

Results

Phenotype of triticale under alkaline stress

   To study phenotypical responses of triticale under alkaline stress, two
   triticale cultivars with contrasting alkali tolerance were compared in
   this study. As shown in Fig. [69]1, both ML (alkali sensitive) and GR
   (alkali tolerance) displayed negative growth effects under AK
   condition, such as stunted seedling growth and advanced chlorosis or
   wilt of mature leaves, while the average root diameter were increased
   in responses to AK. Overall, ML suffered more severely, and the
   contrasting responses can be found particularly at biomass (shoot dry
   weight reduction by 62.2% in ML compared with 7.1% in GR; root dry
   weight reduction by 49.4% in ML compared with 8.6% in GR), shoot height
   (reduction by 41.5% in ML compared with 13.2% in GR), root number
   (reduction by 55.3% in ML compared with 39.6% in GR), length (reduction
   by 68.4% in ML compared with 53.4% in GR) and surface area (reduction
   by 52.1% in ML compared with 29.4% in GR). Focusing on the different
   root morphology over germplasms, GR developed more thinner roots than
   ML, either in CK or AK treatment (Fig. [70]1C, D).

Fig. 1.

   [71]Fig. 1
   [72]Open in a new tab

   Behaviors of two cultivars of triticale seedlings under alkaline
   stress. A Shoot phenotype. B Root image obtained by a root scanner. C,
   Growth parameters of seedlings. D Distribution of root length, surface
   area and volume upon different root diameter ranges. Seedlings of
   different cultivars were grown in perlite substrates and watered by 1/2
   Hoagland solution for three weeks until imaging and measurement.
   Deionized water (CK) or alkali solution (NaHCO[3]:Na[2]CO[3] = 5:1, AK)
   were added 2 times during cultivation. n = 3 sets of at least 5
   seedlings. Error bars represent the means ± SE. Different letters
   represent significant differences (p < 0.05) analyzed by the least
   significant difference test. ML, Mule3000; GR, Grenado; CK, control;
   AK, alkaline treatment

Transcriptome and metabolome profiling of triticale roots under alkaline
stress

   In order to decipher the internal genetic, physiological and metabolic
   adaptation mechanism of triticale to alkaline stress, comparative study
   of transcriptome and metabolome alteration were performed for roots
   from the two cultivars with contrasting alkali tolerance under CK and
   AK conditions. The RNA-Seq analysis yielded more than 8.5 Gb clean data
   for each sample (Table S2). As displayed in Fig. [73]2A, a total of
   90,824, 92,413, 89,539 and 91,540 genes were detected to have expressed
   in CK and AK treated ML and GR roots, respectively. These include
   19,183 novel genes predicted based on our RNA-Seq data apart from the
   annotated genes in allotetraploid wheat and rye reference genomes (File
   S1). Correlation and principal component analysis demonstrate that the
   data has fine repeatability (Fig. S2A, C). The majority of genes
   (84,172, 87.7% of all genes expressed) were co-expressed in the four
   groups of samples, while 2,502 and 1,329 were exclusively detected in
   ML and GR, respectively. Comparative analysis revealed that 14,518
   (8,344/6,174 up/down regulated (AK/CK)) and 11,230 (7,494/3,736
   up/down) DEGs between AK and CK conditions in ML (DEG_ML.AKvCK) and GR
   (DEG_GR.AKvCK) roots, among which 5,001 and 2,595 DEGs were
   simultaneously up- or down-regulated in the two cultivars, respectively
   (Fig. S3A). As we mainly focus on genotype effect on triticale stress
   responses, we further evaluated the interaction effect of the two
   factors. This identified a total of 1,739 DEGs whose expression in the
   two cultivars displayed different regulatory mode or changing level
   upon AK treatment (DEG_GRvML.AKvCK) (Fig. [74]2A; Fig. S4A; Table S3).
   The GRvML.AKvCK DEGs were dispersed on all chromosomes, with largest
   number located on 2B and 7R, and several hotpots on 2B, 7B and 7R (Fig.
   S4B, C). The genes with most markedly varied expression over genotypes
   and treatments are TRITD1Av1G136070 (putative bifunctional
   inhibitor/plant lipid transfer protein), and three putative dirigent
   protein encoding genes: TRITD0Uv1G112200, novel.9874, and novel.9873
   (Fig. [75]3A). Relative expression level of 17 DEGs with specific
   primers was checked via qRT-PCR, and the overall expression patterns
   displayed high concordance with RNA-Seq results (Fig. S5).

Fig. 2.

   [76]Fig. 2
   [77]Open in a new tab

   Statistics of transcriptome and metabolome data. A Upset plot of total
   genes detected and venn diagram of differentially expressed genes
   (DEGs). B Upset plot of total metabolites detected and venn diagram of
   differentially accumulated metabolites (DAMs). ML.AKvCK,
   Alkaline/control in Mule3000. GR.AKvCK, Alkaline/control in Grenado.
   GRvML.AKvCK, Interaction effect between genotype and alkaline treatment

Fig. 3.

   [78]Fig. 3
   [79]Open in a new tab

   Distribution of differentially expressed genes (DEGs) and
   differentially accumulated metabolites (DAMs). A DEGs. B DAMs. DEGs and
   DAMs that ranked top 5 in fold change (FC) (up and down, respectively)
   were marked, and the original TPM (transcripts per million) for genes
   and normalized concentration data for metabolites were displayed in
   heatmap. ML.AKvCK, Alkaline/control in Mule3000. GR.AKvCK,
   Alkaline/control in Grenado. GRvML.AKvCK, Interaction effect between
   genotype and alkaline treatment. FC of GRvML.AKvCK was calculated as
   FC(GR.AKvCK)/FC(ML.AKvCK)

   Using widely targeted metabolomics analysis, we identified 2,320,
   2,298, 2,290 and 2,319 high confidential metabolites in CK and AK
   treated ML and GR roots, respectively (Fig. [80]2B; Fig. S2B, D).
   Overall, the most abundant metabolites were flavonoids, alkaloids,
   amino acids and derivatives, lipids and phenolic acids (Fig. S2E).
   Similar with the transcriptome, most of the metabolites (2,044, 82.9%
   of all metabolites detected) were shared by the four groups of samples.
   Interestingly, there were more treatment specific metabolites (61 and
   61 in CK and AK) than genotype specific ones (19 and 11 in ML and GR),
   which is different with gene expression profile. Differential analysis
   revealed that a larger number of metabolites changed in alkaline
   stressed GR roots, namely 214 (104/110 up/down regulated (AK/CK)) and
   301 (137/164 up/down) DAMs between AK and CK conditions in ML
   (DAM_ML.AKvCK) and GR (DAM_GR.AKvCK) roots, among which 16 and 31 DAMs
   were simultaneously up- or down-accumulated in the two cultivars,
   respectively (Fig. S3B, C, D). Correspondingly, 166 metabolites were
   differentially accumulated between the two cultivars and upon AK
   treatment (DAM_GRvML.AKvCK), and the most remarkable metabolites are
   portulacanone D glucoside, 7-O-methylaloesin, icariside B1, and
   11-hydroxytabersonine (Fig. [81]2B; Fig. [82]3B; Fig. S4D; Table S4).

Functional dissection of transcriptomic alteration

   To discern the main molecular and metabolic change undergoing in
   triticale root subjected to alkaline stress, the representative
   function clusters of DEGs and DAMs were outlined by enrichment
   analysis. GO enrichment analysis revealed that the alkaline stress
   induced DEGs functions generally similar in the two cultivars. The most
   enriched genes encode components of formate dehydrogenase complex and
   perinuclear endoplasmic reticulum membrane, with functions of
   3″-deamino-3″-oxonicotianamine reductase activity and
   4-hydroxy-3-methylbut-2-en-1-yl diphosphate synthase activity
   (ferredoxin), and involve in cadmium ion transport and one-carbon
   compound transport. These function groups were similar in the
   simultaneously up or down regulated genes (Table S5). Nevertheless, the
   stress responses across genotypes were relatively different. The
   GRvML.AKvCK group DEGs mostly participated in formation of
   senescence-associated vacuole and monolayer-surrounded lipid storage
   body, with phenylalanine ammonia-lyase activity and
   4,8,12-trimethyltrideca-1,3,7,11-tetraene synthase activity, and
   involved in cinnamic acid and L-phenylalanine metabolism (Fig. [83]4A;
   Table S5). To reduce the redundant overlapping GO terms that hindered
   the key feature interpretation, non-redundant associations were
   extracted based on the overlap of enriched gene members between GO
   biological process terms. As shown in Fig. [84]4B, the DEGs from
   different comparisons fell into different clusters. The ML.AKvCK and
   GR.AKvCK groups of DEGs were mainly collected in regulation of cellular
   ion transport, while the former also conglomerated in amino acid
   metabolic process and the latter in diphosphate and lipid metabolic
   process, and glycoside and phenylpropanoid metabolic process. Moreover,
   the GRvML.AKvCK group DEGs formed one predominant cluster, regulation
   of amino acid and organic acid metabolism.

Fig. 4.

   [85]Fig. 4
   [86]Open in a new tab

   Gene Ontology (GO) enrichment analysis of differentially expressed
   genes. A GO enrichment results (top 5 significantly enriched terms in
   each comparison and in the three GO categories, namely molecular
   function, biological process, and cellular component, were displayed).
   B Clustering of the significantly enriched GO Biological Process
   annotations. The nodes represent gene sets within each GO term and the
   width of line (grey edge) is proportional to the number of the
   overlapping genes between the two nodes. Clusters of functionally
   related GO terms are circled and the labels are manually assigned based
   on the word frequency of circled GO term names. ML.AKvCK,
   Alkaline/control in Mule3000. GR.AKvCK, Alkaline/control in Grenado.
   GRvML.AKvCK, Interaction effect between genotype and alkaline treatment

Metabolic pathways revealed by differentially gene expression and metabolite
accumulation

   Gene expression and metabolism are coupled and coordinate with each
   other to ensure the plants responding to changing environment swiftly.
   So, the varied profile of DEGs were further integrated with metabolic
   network recorded in KEGG pathway database. As shown in Fig. [87]5A and
   Table S6, the most enriched KEGG pathways in ML or GR responses to AK
   stress is similar, namely betalain biosynthesis, sesquiterpenoid and
   triterpenoid biosynthesis, photosynthesis-antenna proteins and
   benzoxazinoid biosynthesis. However, when taken into consideration of
   influence of genotype, the enrichment results differed. GRvML.AKvCK
   comparison revealed representative pathways include phenylalanine
   metabolism, tyrosine metabolism, benzoxazinoid biosynthesis,and
   brassinosteroid biosynthesis.

Fig. 5.

   [88]Fig. 5
   [89]Open in a new tab

   Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and metabolite
   category enrichment results of differentially expressed genes and
   differentially accumulated metabolites. A KEGG pathway enrichment
   results of differentially expressed genes (all significant pathways in
   either comparison were displayed). B KEGG pathway enrichment results of
   differentially accumulated metabolites (top 5 enriched pathways in each
   comparison were displayed). C Metabolite category enrichment results
   (top 5 enriched categories in each comparison were displayed).
   ML.AKvCK, Alkaline/control in Mule3000. GR.AKvCK, Alkaline/control in
   Grenado. GRvML.AKvCK, Interaction effect between genotype and alkaline
   treatment

   Due to merely a small part of metabolites (354 out of the total 2,466)
   retrieved KEGG pathway annotations, the pathway enrichment analysis
   resulted generally insignificant terms except for glycerophospholipid
   metabolism, arginine and proline metabolism, and alanine, aspartate and
   glutamate metabolism in comparison group of GRvML.AKvCK (Fig. [90]5B;
   Table S6). In addition, in views of category, the significantly changed
   metabolites could be classified into glycerol ester in ML.AKvCK, lysoPE
   in GR.AKvCK and steroidal saponins in the both two groups. No specific
   category of metabolites was found significantly changed in accumulation
   if genotype and treatment were both taken into account but flavonoids
   were mostly enriched (Fig. [91]5C; Table S7).

   DEGs and DAMs identified from ML.AKvCK and GR.AKvCK may serve as
   candidates in elucidating triticale responses to alkaline stress.
   However, because these groups of genes or metabolites are of too large
   number and are less effective in following mechanism study, the main
   aim of this study was set to dissect the genotype differences of gene
   and metabolic alterations in responses to alkaline stress. We mainly
   focused on the key messages denoted by the comparison of GRvML.AKvCK in
   the follow-up analyses.

Alkali effects on amino acid metabolism

   Significant changes were observed in amino acid metabolism both at
   transcriptional and metabolic levels. Part of the components and
   processes of amino acid metabolism were illustrated in Fig. [92]6. The
   transcriptomic data revealed that expression of a group of genes
   involved in synthesis or catabolism of amino acid such as valine,
   leucine, isoleucine, serine, methionine, alanine, aspartate,
   asparagine, tryptophan, phenylalanine, tyrosine, histidine, glutamate,
   glutamine, arginine, ornithine and proline were significantly changed
   and differently regulated in ML and GR roots. Simultaneously, compound
   contents of aspartate, histidine, glutamate, methionine and ornithine,
   and also components involved in their metabolism like citrate,
   LL-2,6-diaminopimelate, N2-succinyl-L-arginine and 4-aminobutanoate
   varied between genotypes and treatments. Besides, 21 amino acids or
   derivatives were detected without KEGG pathway annotation (Fig. [93]5C;
   Table S7; Table S8).

Fig. 6.

   [94]Fig. 6
   [95]Open in a new tab

   Differentially expressed genes and differentially accumulated
   metabolites involved in amino acid metabolism. Core genes and
   metabolites were displayed. Solid arrows represent direct reactions,
   and dashed arrows represent indirect reactions. ADC, Arginine
   decarboxylase; ADT, Arogenate/prephenate dehydratase; ALT, Alanine
   transaminase; amiE, Amidase; argAB, Amino-acid N-acetyltransferase;
   aroK, Shikimate kinase; ASNS, Asparagine synthase
   (glutamine-hydrolysing); ASRGL1,
   L-asparaginase/beta-aspartyl-peptidase; GFPT,
   Glutamine-fructose-6-phosphate transaminase (isomerizing); glnA,
   Glutamine synthetase; GOT2, Aspartate aminotransferase, mitochondrial;
   hisA, Phosphoribosylformimino-5-aminoimidazole carboxamide ribotide
   isomerase; IDH1, Isocitrate dehydrogenase; ilvE, Branched-chain amino
   acid aminotransferase; metB, Cystathionine gamma-synthase; metE,
   5-methyltetrahydropteroyltriglutamate-homocysteine methyltransferase;
   ODC1, Ornithine decarboxylase; P4HA, Prolyl 4-hydroxylase; PAT,
   Bifunctional aspartate aminotransferase and
   glutamate/aspartate-prephenate aminotransferase; pip, Proline
   iminopeptidase; TPM, transcripts per million; trpE, Anthranilate
   synthase component I; ML, Mule3000; GR, Grenado; CK, control; AK,
   alkaline treatment. The set of differentially expressed genes and
   differentially accumulated metabolites identified in the interaction
   effect between genotype and alkaline treatment were displayed. The full
   list of differentially accumulated amino acids or derivatives can be
   found in Table S8

Alkali effects on flavonoids metabolism

   Although metabolites were not significantly enriched in KEGG pathways
   related to flavonoids, the group of compounds were the predominant
   category with regard to interaction of triticale genotype and alkaline
   treatment (Fig. [96]5C; Table S7). Actually, a total of 20 DAMs belong
   to different subclasses of flavonoids were detected in comparison of
   GRvML.AKvCK, in which only one compounds retrieved KEGG pathway
   annotation (Fig. [97]7; Table S7). Correspondingly, flavonoids related
   pathways were significantly enriched by the GRvML.AKvCK DEGs
   (Fig. [98]5A). As illustrated in Fig. [99]7, these DEGs particularly
   participated in metabolism of flavones, flavonols, anthocyanidins,
   flavanones, and flavanols, with flavanones (naringenin and eriodictyol)
   and flavonols (kaempferol, quercetin, and myricetin) and their
   derivatives at the core. Similar associated clusters could be
   summarized by the DAMs (Table S4; Table S7).

Fig. 7.

   [100]Fig. 7
   [101]Open in a new tab

   Differentially expressed genes and differentially accumulated
   metabolites involved in flavonoids metabolism. Core genes and
   metabolites were displayed. Solid arrows represent direct reactions,
   and dashed arrows represent indirect reactions. ANR, Anthocyanidin
   reductase; CYP73A, Trans-cinnamate 4-monooxygenase; CYP75A, Flavonoid
   3',5'-hydroxylase; CYP75B1, Flavonoid 3'-monooxygenase; FG2,
   Flavonol-3-O-glucoside L-rhamnosyltransferase; FLS, Flavonol synthase;
   HCT, Shikimate O-hydroxycinnamoyltransferase; TPM, transcripts per
   million; UGT73C6, Flavonol-3-O-L-rhamnoside-7-O-glucosyltransferase;
   ML, Mule3000; GR, Grenado; CK, control; AK, alkaline treatment. The set
   of differentially expressed genes and differentially accumulated
   metabolites identified in the interaction effect between genotype and
   alkaline treatment were displayed

Basic alkaline stress responses shared by the two genotypes

   Processes or pathways shared by ML.AKvCK and GR.AKvCK may represent
   basic alkaline stress responses in triticale root, especially those
   denoted by genes or metabolites simultaneously up or down regulated. In
   these responses, betalain biosynthesis, photosynthesis—antenna proteins
   and sesquiterpenoid and triterpenoid biosynthesis were predominantly
   enriched (Fig. [102]5). Betalain biosynthesis pathway included as many
   as 39 DEGs from all comparisons. However, these genes encode protein
   homologs of merely two enzymes involved in dopa transformation to
   dopamine and betalamic acid: DDC and DODA. Correspondingly, dopa
   content was decreased under alkaline stress. Content of another two
   compounds, gomphrenin-I and betanin, were increased in ML and GR,
   respectively (Fig. S6).

Expression analysis of other key genes potentially involved in alkaline
stress responses

   Considering functional conservation of homologous genes among species,
   a total of 43 triticale homologs of AtJ3, AtMOCA1, AtOSCA1, AtPKS5,
   AtRGI1, AtSCaBP3, GsBOR2, SbAT1, SbPIP2, TaCCD1, TaHA2, TaPP2C.D1/D2,
   and TaSAUR215 were predicted through sequence alignment, in which 40
   homologs retrieved reliable expression data from the present
   transcriptomic analysis (Fig. S7; Table S9). Among these genes, BOR2-A,
   CCD1-A/B/R, J3-B/R, MOCA1-B, OSCA1-A1/A2, PIP2.1-A/B/R, PKS5-R,
   PP2C.D1-B/R, PP2C.D2-A, RGI1-A/R, SAUR215-R, and SCaBP3-R displayed
   contrasting regulatory mode between alkaline treatments and triticale
   genotypes.

Discussion

Responses to alkaline stress in triticale are genotype-dependent

   Plant roots bear the brunt of various soil related stresses. In case of
   bicarbonate and carbonate induced alkaline stress, triticale root
   systems developed into new architectures as revealed by the present
   study. The modified root system architecture can be briefly summarized
   as reduced root number and length, but increased average root diameter,
   which leading to compromised root biomass formation (Fig. [103]1). The
   morphological responses were similar with wheat and rye [[104]8,
   [105]9, [106]36]. Moreover, the contrasting behaviors of different
   cultivars under stress suggest a promising application of molecular
   design and gene modification in breeding new smarter triticale
   germplasms.

   Adjustments of the overall transcriptomic and metabolomic profiles were
   disaccord in triticale seedlings of different cultivars subjected to
   alkaline stress. Typically, expression of a larger number of genes was
   induced while accumulation of more metabolites was repressed in
   responses to alkaline treatment in either cultivar. Moreover, less DEGs
   but more DAMs were identified in GR compared with ML, both in the up or
   down regulated ones (Fig. S3A, D). We conjectured that the ML seedling
   suffered more to alkaline stress and displayed more drastic responses
   in gene expression, but it failed to synthesize sufficient effective
   metabolites to cope with the adversity, resulted in severer growth
   reduction.

   At gene expressional and metabolic level, most of the DEGs and DAMs
   from comparisons of either ML.AKvCK or GR.AKvCK shared similar
   function. The mostly enriched processes, including ion and compound
   transport, organic acid metabolism, controlling of cell death, betalain
   accumulation, plasma membrane and cell wall related metabolism, could
   be regarded as universal basic molecular responses of triticale grown
   under alkaline stress. These physiological and molecular changes is in
   accordance with other plants [[107]6, [108]11–[109]14]. Although in
   this study we focus more on the interaction effect of genotype and
   treatment on alkaline tolerance in triticale, it should not be
   neglected that some of the stress induced pathways shared by both
   cultivars might also offer assistance in breeding work and mechanism
   study.

Regulation of amino acid metabolism in triticale responses to alkaline stress

   Besides being building blocks for protein synthesis, many amino acids
   or their derivatives, including some not involved in protein synthesis,
   turned out to participate in the plant's response to environmental
   stresses [[110]37, [111]38]. For example, glutamate serves as an
   N-donor for the biosynthesis of amino acids and other N-containing
   compounds. Glutamate can be converted to 4-aminobutanoate (GABA) in a
   decarboxylation reaction, which plays an important role in balancing
   C/N metabolism, organic acid accumulation, Na^+/K^+ homeostasis and
   antioxidant activities under alkaline stress in plants
   [[112]39–[113]42]. In AK condition, a significant increase of
   concentration in aspartate, proline, glutamate, serine, isoleucine,
   valine, glycine and alanine has been observed in soybean roots
   [[114]43]. Arginine synthesis was promoted by alkali only in the
   tolerant rice cultivar, and a shift to N-rich amino acids such as
   asparagine, glutamine, and glycine was observed under alkaline stress
   irrespective of cultivar [[115]44]. These results are in line with the
   amino acid content or the gene expression in their metabolism pathway
   (Fig. [116]6). Methionine and some of the derived metabolites bring
   broad stress resistance and most of them are extremely sensitive to
   almost all forms of ROS [[117]45]. It was reported that the enhanced
   alkaline tolerance was coupled with elevated methionine content in
   alfalfa [[118]46, [119]47]. However, methionine content was decreased
   in responding to alkali in triticale in both cultivars, raised a query
   about the specific function of methionine for alkaline acclimation in
   different species. Our data together with previous research indicate
   potential roles of the alkali and genotype coordinated metabolite
   accumulation and gene expression during metabolism of amino acids and
   their derivatives in alkali acclimation in plants. Nevertheless, it
   should be noted that the present study merely detected a relatively
   small portion of metabolites (2,470 out of 4,000 to 20,000 metabolites
   estimated to exit in a single plant [[120]48]), and the unmarked
   metabolites were not definitely meant unchanged and wait further
   specific verification in future work.

Flavonoids as candidates in amelioration of alkaline stress in triticale

   Flavonoids are the largest group enriched by DAMs from comparison of
   GRvML.AKvCK (Fig. [121]6C). Flavonoids have strong free radical
   scavenging ability and can act as signaling molecules to regulate
   triticale and other plant acclimation to stresses [[122]49, [123]50].
   It was considered that all environmental stresses promote the
   expression of flavonoid metabolism genes in plants, resulting in the
   synthesis of large quantities of flavonoids and their transport to
   various organelles [[124]49]. Integrated metabolomic and transcriptomic
   studies also revealed significant modification in flavonoids
   biosynthesis in Sorghum bicolor, Glycyrrhiza uralensis, Panicum
   miliaceum and alfalfa under alkaline stress [[125]51–[126]54]. Among
   the level-changed enzymes controlling flavonoid metabolism, FLSs
   (Flavonol synthases) featured the predominant homologs in amount and
   mainly be up regulated under alkaline stress (Fig. [127]7). MsFLS13 in
   alfalfa was proved to play a positive role in improving plant tolerance
   to alkaline stress by enhancing flavonols accumulation, antioxidant
   capacity, osmotic balance, and photosynthetic efficiency [[128]55].
   Sucrose signaling was reported to induce flavonoid biosynthesis by
   activating the expression of MhCYP75B1 to regulate the homeostasis of
   ROS and auxin signaling to enhance the resistance to saline–alkali in
   Malus halliana [[129]56]. The expression of FLS and CYP75B1 in the
   present study is generally in line with previous research. However,
   elevated expression of MhANR (Anthocyanin reductase) reduced tolerance
   of Arabidopsis, tobacco, and apple calli to alkali stress by regulating
   osmoregulatory substances, chlorophyll content, and antioxidant enzyme
   activity [[130]57]. Transcripts of ANR in this study were decreased in
   alkaline treated sensitive genotype ML but increased in tolerant GR,
   indicating a complex mechanism of ANR in different species. These
   findings together indicate the role of flavonoids in alkaline stress
   responses in triticale, while the exact function and underlying
   molecular mechanism need to be further validated via genetic and
   physiological means.

Other key genes, metabolites and processes involved in resistance to alkaline
stress

   Benzoxazinoid and brassinosteroid biosynthesis were another two top
   enriched pathways by DEGs in GRvML.AKvCK group (Fig. [131]5; Table S6).
   Benzoxazinoids are bioactive metabolites with protective properties to
   biotic stresses, but they were also modulated by diverse abiotic
   stresses including water deficit, temperature control, and mechanical
   damage [[132]58, [133]59]. The function of benzoxazinoids in plant
   abiotic stress resistance is to date unclear. Brassinosteroid signaling
   was reported to positively regulate plant tolerance to bicarbonate, and
   brassinosteroid biosynthesis was strongly activated in an
   alkaline-tolerant cultivar of Panicum miliaceum [[134]4, [135]60].
   Furthermore, exogenously application of brassinosteroid’s analog
   brassinolide significantly enhanced the resistance to alkaline stress
   in Malus hupehensis and Panicum miliaceum seedlings [[136]60, [137]61].
   These studies indicating a role of brassinosteroid signaling in
   alkaline stress responses in triticale.

   Part of the DEGs or DAMs not included in comparison result of
   GRvML.AKvCK might also function essentially for alkaline stress
   resistance in triticale. The change of these genes, metabolites and the
   corresponding processes could serve as universal responses in
   alkali-stressed triticale. Betalain biosynthesis was predominant
   enriched by the simultaneously up or down regulated genes or
   metabolites (Fig. [138]5A, B; Table S6). As an intermediate substance
   in betalain biosynthesis, dopa was reported unstable in alkaline
   condition, and this might be cause of declined dopa content in the
   present study [[139]62]. Dopamine, derived from dopa, involved in
   osmotic stress acclimation in plant. Dopamine can regulate sugar
   metabolism and ABA content, and was shown to influence the expression
   of the rice aquaporin gene OsPIP1-3 under salt stress [[140]63].
   Exogenous application of dopamine facilitated Malus hupehensis
   maintaining a low Na^+/K^+ ratio, and improved plant resistance to salt
   stress [[141]64]. Additionally, as a kind of betalain, betanin has
   exceptionally high free radical-scavenging activity and its content in
   plant varied in responses to environmental factors including water
   deficit [[142]65–[143]67]. These results suggest the role of dopa,
   betalain and their derivatives in alkaline stress responses in
   triticale. Besides, a couple of genes encoding antenna proteins were
   surprisingly identified and enriched by the co-regulated genes in the
   present study (Fig. [144]5A; Table S6). Since antenna proteins mainly
   function in light-harvesting, whether these proteins received signals
   transmitted from leaves and function in root responses to alkaline
   stress is an interesting subject in future studies.

   Coupling gene expression with morphology, homologs of typical alkali
   responsive genes AT1-PIP2, RGI1, SAUR215, SCaBP3 were differentially
   regulated in the two triticale cultivars, and were supposed to closely
   linked with the genotypic influence on triticale resistance to alkaline
   stress (Fig. S7). As noted earlier, these genes function as pivotal
   regulators in the alkaline stress responses of other plant species
   [[145]17, [146]18, [147]20, [148]24, [149]25]. It would be more
   efficient to validate the function of these genes in triticale and
   apply them in breeding works. In the meantime, the stress involvement
   of other alkali-responsive genes still needs further investigation, as
   not only transcript abundance but also sequence variation determines
   the function realization of a gene.

Conclusions

   In summary, this study described triticale root responses to alkaline
   stress by comparatively analyzing the morphological, transcriptomic and
   metabolomic profile alterations. We found that when subjected to
   alkaline stress, triticale developed into compromised root number,
   length, surface area and total biomass, but increased average root
   diameter. The growth reduction was genotype dependent as it varied in
   different triticale cultivars. Transcriptomic and metabolomic study
   revealed that the genotypic effects on triticale responses to alkaline
   stress mainly attributed to differences in metabolism of amino acids,
   flavonoids and their derivatives. Other top enriched processes included
   biosynthesis of benzoxazinoid and brassinosteroid. Simultaneous
   modification in betalain biosynthesis and antenna protein metabolism
   were considered to serve as universal alkaline stress responses in
   triticale. These findings provide candidate gene and metabolite
   resources for future studies of molecular mechanisms and breeding
   works, and also proposed the evaluation and potential application of
   some compounds in triticale production.

Supplementary Information

   [150]12870_2025_6973_MOESM1_ESM.zip^ (7.9MB, zip)

   Supplementary Material 1: Fig. S1 Total ions current of the two
   quality-control samples. Fig. S2 Basic statistics of total detected
   genes and metabolites. Fig. S3 Basic statistics of differentially
   expressed genes and differentially accumulated metabolites. Fig. S4
   Expression and distribution of differentially expressed genes (DEGs)
   and differentially accumulated metabolites. Fig. S5 qRT-PCR validation
   of gene expression. Fig. S6 Differentially expressed genes and
   differentially accumulated metabolites involved in betalain
   biosynthesis. Fig. S7 Expression of key genes involved in alkaline
   stress responses in plants. File S1 Novel_genes.gtf. Table S1 Primers
   used in this study. Table S2 Quality information of RNA and RNA-Seq
   data. Table S3 Differentially expressed genes. Table S4 Differentially
   accumulated metabolites. Table S5 Gene Ontology enrichment results of
   differentially expressed genes. Table S6 KEGG pathway enrichment
   results of differentially expressed genes and differentially
   accumulated metabolites. Table S7 Metabolite category enrichment
   results of differentially accumulated metabolites. Table S8
   Differentially accumulated amino acid and derivatives. Table S9
   Homologs of key genes involved in alkaline stress responses.

Acknowledgements