Overexpression of HMG-CoA synthase promotes Arabidopsis root growth and adversely affects glucosinolate biosynthesis

Inhibition of 3-hydroxy-3-methylglutaryl-CoA synthase (HMGS) retarded root growth in Arabidopsis seedlings by reducing stigmasterol, auxin, and cytokinin levels. Differentially expressed proteins and metabolite changes revealed a relationship between HMGS and glucosinolate biosynthesis.

Abstract

3-Hydroxy-3-methylglutaryl-CoA synthase (HMGS) catalyses the second step of the mevalonate (MVA) pathway. An HMGS inhibitor (F-244) has been reported to retard growth in wheat, tobacco, and Brassica juncea, but the mechanism remains unknown. Although the effects of HMGS on downstream isoprenoid metabolites have been extensively reported, not much is known on how it might affect non-isoprenoid metabolic pathways. Here, the mechanism of F-244-mediated inhibition of primary root growth in Arabidopsis and the relationship between HMGS and non-isoprenoid metabolic pathways were investigated by untargeted SWATH-MS quantitative proteomics, quantitative real-time PCR, and target metabolite analysis. Our results revealed that the inhibition of primary root growth caused by F-244 was a consequence of reduced stigmasterol, auxin, and cytokinin levels. Interestingly, proteomic analyses identified a relationship between HMGS and glucosinolate biosynthesis. Inhibition of HMGS activated glucosinolate biosynthesis, resulting from the induction of glucosinolate biosynthesis-related genes, suppression of sterol biosynthesis-related genes, and reduction in sterol levels. In contrast, HMGS overexpression inhibited glucosinolate biosynthesis, due to down-regulation of glucosinolate biosynthesis-related genes, up-regulation of sterol biosynthesis-related genes, and increase in sterol content. Thus, HMGS might represent a target for the manipulation of glucosinolate biosynthesis, given the regulatory relationship between HMGS in the MVA pathway and glucosinolate biosynthesis.

Introduction

Many natural products are isoprenoids or isoprenoid derivatives, including chlorophylls and carotenoids in photosynthesis, ubiquinone in oxidative phosphorylation, sterols as membrane constituents, plant hormones including cytokinin (CK), brassinosteroids (BRs), auxin, gibberellic acid (GA), and precursors of abscisic acid regulating growth and development. Monoterpenes and sesquiterpenes are active in pathogen defense and signaling. Isoprenylated proteins are also involved in regulatory processes (see Rodríguez-Concepción 2006; Orlova et al., 2009; Hemmerlin et al., 2012; Dudareva et al., 2013; Wölwer-Rieck et al., 2014; Tholl, 2015 and references therein). The universal biosynthesis intermediates (isopentenyl diphosphate, IPP; geranyl diphosphate, GPP; farnesyl diphosphate, FPP; or geranylgeranyl diphosphate, GGPP) are generally synthesized via both cytosolic mevalonate (MVA) (Goodwin, 1958; Modi and Patwa, 1961) and plastidial 2C-methyl-d-erythritol 4-phosphate (MEP) pathways (Rohmer, 1999). Many studies have reported on a crosstalk between these two pathways (Hemmerlin et al., 2012 and references therein; Liao et al., 2016 and references therein; Huchelmann et al., 2017).

3-Hydroxy-3-methylglutaryl-CoA (HMG-CoA) synthase (HMGS), which catalyses the second step in the MVA pathway, is responsible for the condensation of acetoacetyl-CoA and acetyl-CoA into HMG-CoA (Ferguson and Rudney, 1959; Rudney and Ferguson, 1959; Stewart and Rudney, 1966; Lynen, 1967; Miziorko et al., 1975). HMG-CoA is converted to MVA by HMG-CoA reductase (HMGR), a key enzyme in the pathway (Bach, 1986; Chye et al., 1991, 1992; Chappell et al., 1995; Schaller et al., 1995) and further down the pathway into IPP by other enzymes. Many studies on eukaryotic organisms showed that HMGR and HMGS are co-regulated (Balasubramaniam et al., 1977; Goldstein and Brown, 1990; Alex et al., 2000; Wang et al., 2012; Suwanmanee et al., 2013), and together they regulate the biosynthesis of mammalian cholesterol (Goldstein and Brown, 1990).

In plants, HMGS overexpression significantly enhances germination, reproduction, seed production and stress tolerance as shown in studies using Brassica juncea HMGS1 (Wang et al., 2012; Liao et al., 2014b). Studies on Arabidopsis hmgs knock-out mutants and transgenic Arabidopsis, tobacco (Nicotiana tabacum), and tomato (Solanum lycopersicum) overexpressing BjHMGS1 and its mutant forms indicated a regulatory role for HMGS in isoprenoid biosynthesis and sterol production (Ishiguro et al., 2010; Wang et al., 2012; Liao et al., 2014a, 2018; Lange et al., 2015). HMGS overexpression in tomato not only up-regulated MVA-derived isoprenoids, but also MEP-derived carotenoids and vitamin E through putative interactions between the MVA and MEP pathways (Liao et al., 2018). The inhibitory effect of a specific inhibitor (hymeglusin, 1233A, F-244, or L-659-699) on HMGS (Greenspan et al., 1987; Tomoda et al., 1988) has been demonstrated in B. juncea seedlings, etiolated wheat coleoptiles, and tobacco cell culture with different efficacies (Jacyno et al., 1991; Liao et al., 2014a), but the mechanism is still unknown.

Inhibitors of specific enzymes from the MVA and MEP pathways, namely mevinolin and fosmidomycin, which inhibit HMGR and 1-deoxy-d-xylulose 5-phosphate reductoisomerase (DXR), respectively, have been widely tested on plants (Hemmerlin et al., 2012 and references therein). F-244 was shown to be very effective on His₆-BjHMGS1 (Nagegowda et al., 2004). Furthermore, F-244, which simulates a substrate molecule and irreversibly binds to Cys117 at the HMGS reaction center, halts condensation (Pojer et al., 2006). This study yielded insights on the potential application of HMGS for the production of herbicides and cholesterol-lowering drugs (Pojer et al., 2006). As Cys117 is conserved between BjHMGS1 and Arabidopsis HMGS, F-244 is expected to portray a similar effect in Arabidopsis. Here we show that F-244 treatment on Arabidopsis seedlings retarded germination, cotyledon emergence, and primary root growth. Furthermore, an interesting relationship between the MVA pathway and glucosinolate biosynthesis has emerged. This prompted us to investigate further to seek an explanation for these phenomena by metabolomic, genomic, and proteomic approaches. Although the effects of HMGS on downstream isoprenoid metabolites have been extensively reported (Ishiguro et al., 2010; Wang et al., 2012; Liao et al., 2014a, 2018; Lange et al., 2015), not much is known on how it might affect non-isoprenoid metabolic pathways (Ishiguro et al., 2010).

Materials and methods

Plant material and growth conditions

Seeds of wild-type (WT) Arabidopsis (Col-0), the vector-transformed control (pSa13), HMGS-OEs (Wang et al., 2012) (OE-wtBjHMGS1-1 (401) and OE-wtBjHMGS1-2 (402)), and hmgs mutants (Wang, 2010) were surface-sterilized and germinated on MS medium (Murashige and Skoog, 1962) with 2% (w/v) sucrose. The Petri dishes were incubated at 4 °C for 2 d and then transferred to a tissue culture room under 16 h light (23 °C)–8 h dark (21 °C) cycles, with dishes placed vertically. Three-day-old HMGS-OE and pSa13 seedlings were transferred to fresh MS medium containing 2% (w/v) sucrose and roots from 7-day-old HMGS-OE and pSa13 seedlings were collected and used for qRT-PCR and glucosinolate analyses.

Inhibitor treatment

Inhibitor treatment was performed as previously reported (Bach and Lichtenthaler, 1982, 1983; Liao et al., 2014a). The stock of HMGS inhibitor (F-244) was prepared in dimethyl sulfoxide (DMSO). For germination assays, seeds were stratified on MS medium containing 2% (w/v) sucrose with or without F-244 at different concentrations at 4 °C for 2 d before incubation in a tissue culture room (16 h light (23 °C)–8 h dark (21 °C) cycles). Germination rates were recorded daily. For each concentration of F-244 treatment or DMSO control, about 20 seeds were germinated in a small Petri dish and a total of three Petri dishes were used for each concentration. This experiment was performed independently three times. The values presented in Fig. 1 represent the average from three experiments. To test the effect of F-244 on seedling development and/or root development, 3-day-old WT seedlings with similar primary root length were transferred to fresh MS medium containing 2% (w/v) sucrose plus different F-244 concentrations (500 nM, 1 µM, 2 µM, or 5 µM). As controls, 3-day-old WT seedlings were transferred to fresh MS medium with 2% (w/v) sucrose plus DMSO. For each concentration of F-244 treatment or DMSO control, about 15 seedlings were grown in a small Petri dish and a total of four Petri dishes were used for each concentration. DMSO did not affect primary root growth at any concentration tested (Fig. 1C). Thus, the DMSO control was used in subsequent experiments (Perelló et al., 2014).

Fig. 1.

Effects of F-244 on Arabidopsis (Col-0). (A) Seed germination. (B) Cotyledon emergence. (C) Primary root growth. Arabidopsis seeds were germinated on MS control, DMSO control, or MS medium containing F-244 (500 nM, 1 µM, or 5 µM) solubilized in DMSO. Seed germination rate, cotyledon emergence rate and primary root length were determined daily. L, germination rate lower than the DMSO/MS control. Values are means ±SD (n=3); Bars are SD. **P<0.05; ***P<0.01.

Seedlings were photographed daily and their root lengths measured manually with a ruler, and confirmed by ImageJ software (Abràmoff et al., 2004). This experiment was performed independently four times. For recovery at day 7 after F-244 treatment, seedlings were transferred to new square plates (12.5 cm×12.5 cm) containing fresh MS medium or fresh MS medium containing the original concentration of F-244 (1 µM or 2 µM) as controls to examine whether the inhibitor-treated roots could recover. RNA was extracted from roots of 7-day-old seedlings (4 d after F-244 treatment) and subsequently used to synthesize cDNA for quantitative real-time PCR (qRT-PCR) analysis. Proteins, sterols, glucosinolates, or plant hormones were extracted from roots of 7-day-old seedlings (4 d after F-244 treatment) for proteomic or metabolite analysis.

SWATH-MS measurement and analysis

Preparation of protein extracts from Arabidopsis roots and mass spectrometry analysis of peptides were performed as previously described (Zhu et al., 2016; Panthapulakkal Narayanan et al., 2019). A global false discovery rate (FDR) (<1%) was used as the criterion to accept peptide assignments and protein identifications (Zhu et al., 2016; Panthapulakkal Narayanan et al., 2019). Four biological repeats of 7-day-old Arabidopsis roots treated with F-244 (2 µM) or DMSO (control) for 4 d were used. After global normalization of the total peak area of four biological repeats, the relative quantification of proteins between F-244-treated Arabidopsis roots and the DMSO control was generated and expressed in the form of fold change. Proteins with a fold change >1.2 (up-regulated) or <0.8 (down-regulated) with q value <0.05, Student’s t-test corrected for FDR (<5%) through the Benjamini–Hochberg procedure (Benjamini and Hochberg, 1995), were considered as differentially expressed proteins (DEPs) (Zhu et al., 2016). DEPs were distributed to different functional categories by the MapMAN BIN system (Thimm et al., 2004).

qRT-PCR

The RNeasy Plant Mini Kit (Qiagen) was used to extract total RNA (5 µg) from roots of 7-day-old Arabidopsis seedlings. The SuperScript First-Strand Synthesis System (Invitrogen) was subsequently used to reverse-transcribe the total RNA for first-strand cDNA. Then qRT-PCR was performed with FastStart Universal SYBR Green Master (Roche) and a StepOne Plus Real-Time PCR System (Applied Biosystems) under the following conditions: 95 °C denaturation for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 30 s. Three technical replicates for each reaction using gene-specific primers were performed, and Arabidopsis UBIQUITIN EXTENSION PROTEIN2 (UBQ2) was used as an internal control. There was no significant difference on UBQ2 expression in F-244-treated roots in comparison with the DMSO control or in OE-wtBjHMGS1 over the vector control (see Supplementary Figs S1, S2 at JXB online). qRT-PCR data were analysed by the 2^−ΔΔCt method (Schmittgen and Livak, 2008). This experiment was performed independently three times. The relative changes in gene expression were analysed using data from three independent experiments. Primers for qRT-PCR are listed in Supplementary Table S1.

Sterol analysis by GC-MS and glucosinolate profiling by HPLC

For sterol and glucosinolate extraction, 10 mg of freeze-dried Arabidopsis roots from 7-day-old seedlings (55–60 seedlings for control and 75–80 seedlings for F-244 treatment) were used in each sample. Sterols were analysed by GC-MS as previously described (Liao et al., 2014b; Lung et al., 2017, 2018). Glucosinolate analysis was conducted following Mugford et al. (2009), Burow et al. (2006) and Doheny‑Adams et al. (2017) with minor modifications. Briefly, each sample was extracted in 1 ml of pre-cooled (−20 ^oC) 80% methanol containing 0.05 µM sinigrin (Sigma-Aldrich) as an internal standard at room temperature for 30 min, followed by gentle shaking at room temperature for 30 min. The sample was cooled to 4 °C and centrifuged at 4000 g for 10 min. The supernatant was transferred to an anion-exchange column (DEAE-Sephadex A-25) and washed with 67% (v/v) (aqueous) methanol and deionized water. Following capping of the column, the glucosinolates were converted to their desulfated derivatives by overnight incubation with 50 µl of 2.8% (w/v) sulfatase according to Graser et al. (2001). The desulfoglucosinolates were then eluted twice with 800 µl of 60% (v/v) (aqueous) methanol and the eluates were pooled. Subsequently, the eluate was dried under nitrogen and dissolved in water.

Desulfoglucosinolate samples were analysed by HPLC (Agilent 1260 Infinity equipped with diode array detector) according to Reintanz et al. (2001). An ACE reverse-phase C18, 5 µm column (250 mm×4.6 mm) was used with gradient mobile phases consisting of water (solvent A) and acetonitrile (solvent B) at a flow rate of 1 ml min⁻¹ at 25 °C. The injection volume was 10 µl. The gradient conditions of mobile phases are as follows: 1.5–5.0% (v/v) solvent B (6 min), 5.0–7.0% (v/v) solvent B (2 min), 7.0–21.0% (v/v) solvent B (10 min), 21.0–29.0% (v/v) solvent B (5 min), 29.0–57.0% (v/v) solvent B (14 min), 57.0–93.0% (v/v) solvent B (2 min), 5 min of hold at 93.0% (v/v) solvent B, 93.0–1.5% (v/v) solvent B (3 min), and 6 min of hold at 1.5% (v/v) solvent B. The eluent was monitored at 229 nm. Desulfoglucosinolates were identified by comparing retention times and UV spectra against known standards (Reichelt et al., 2002). Their concentrations were quantified relative to the peak area of an internal standard using response factors (Brown et al., 2003).

Measurement of endogenous plant hormones

The extraction and analysis of plant hormones including BRs, CKs, indole-3-acetic acid (IAA), 1-aminocyclopropane-1-carboxylic acid (ACC) and GAs were performed at the National Centre for Plant Gene Research (Beijing), Institute of Genetics and Developmental Biology, Chinese Academy of Sciences (Beijing, China). Extraction and analysis of BRs, CKs, IAA, ACC, and GAs were performed as described previously (Müller and Munné-Bosch, 2011; Xin et al., 2013; Wang et al., 2015; Ma et al., 2015; Du et al., 2017). Three biological repeats were used in analysis of each hormone.

Prediction of protein subcellular localization

Prediction of protein subcellular localization was conducted using the Subcellular Localization database for Arabidopsis proteins (SUBA) (Tanz et al., 2013).

Results

F-244 reduced seed germination and cotyledon emergence rates

As transgenic Arabidopsis and tobacco overexpressing HMGS (HMGS-OEs) had displayed an increase in seed germination (Wang et al., 2012; Liao et al., 2014b) and F-244 inhibited seed germination in B. juncea (Liao et al., 2014a), the effect of varying concentrations of F-244 on Arabidopsis seed germination was examined. At day 2 after imbibition, the germination rate of F-244 (500 nM)-treated seeds was 57% lower than in the control (Fig. 1A). For F-244 (1 µM)-treated seeds, it was 26% lower than in the control, although the difference was not significant (Fig. 1A). From day 3 to day 7, the germination rates in F-244 (5 µM)-treated seeds were 22%, 15%, 12%, 15%, and 15% lower, respectively, in comparison with the control (Fig. 1A). These results indicated that F-244 partially inhibited Arabidopsis seed germination. As expected, F-244 (5 μM) was more effective in reducing seed germination than F-244 at 500 nM and 1 µM (Fig. 1A).

Arabidopsis cotyledon emergence rates after F-244 (1 μM) treatment were significantly (P<0.05) reduced by 19%, 13%, 12%, and 12% in 4-, 5-, 6-, and 7-day-old seedlings, respectively, in comparison with the control (Fig. 1B). Furthermore, cotyledon emergence rates after F-244 (5 μM) treatment were 33%, 17%, 13%, and 18% lower in 4-, 5-, 6-, and 7-day-old seedlings, respectively, than in the control (Fig. 1B).

F-244 retarded primary root growth in a dose-dependent manner

Primary root length of 4- to 7-day-old Arabidopsis seedlings was significantly inhibited by F-244 (500 nM, 1 µM, and 5 µM) (see Supplementary Fig. S3A–D) in a dose-dependent manner (Fig. 1C). The presence of DMSO in the control did not yield significant differences in primary root length (Fig. 1C). It appears that in the presence of a low concentration of F-244 (500 nM), root growth recovered after 6 d (Supplementary Fig. S3C, D), whilst a higher concentration of F-244 (5 µM) resulted in shorter primary roots or inhibition of root growth (Supplementary Fig. S3A–D). Thus two more concentrations (1 and 2 µM) between 500 nM and 5 µM were further tested. Similarly, 1 and 2 µM F-244 inhibited primary root growth, and as expected the higher concentration (2 µM) was more effective (Fig. 2A, B).

Fig. 2.

Effects of F-244 on Arabidopsis primary root growth and phytosterol and cholesterol biosynthesis and profiles of sterols in Arabidopsis hmgs mutant roots. (A) Phenotypic changes in primary root elongation. Root tips of 5- and 7-day-old F-244-treated seedlings are indicated with black bars. (B) Primary root length measurement. Seeds were germinated on MS medium for 3 d before transfer to fresh MS medium containing different concentrations of F-244 (1 or 2 µM) or DMSO (control). Scale bar: 1 cm. L, primary root length lower than control. Values are means ±SD (n=60); Bars are SD. a, significant difference between control and treated samples (P<0.05, Student’s t-test); b, significant difference between F-244 (1 μM)- and F-244 (2 μM)-treated samples (P<0.05, Student’s t-test). (C) Expression of MVA- and phytosterol-related genes in F-244-treated roots by quantitative RT-PCR analysis. Total RNA was extracted from 7-day-old roots, which were subject to 4-d treatment with F-244 (2 µM) or DMSO (control). Expression levels were normalized by UBQ2. AACT, ACETOACETYL-COA THIOLASE; DWF1, Δ²⁴STEROL REDUCTASE; FPPS, FARNESYL DIPHOSPHATE SYNTHASE; HMGR, 3-HYDROXY-3-METHYLGLUTARYL-COA REDUCTASE; HMGS, 3-HYDROXY-3-METHYLGLUTARYL-COA SYNTHASE; IDI, ISOPENTENYL DIPHOSPHATE ISOMERASE; MPDC, DIPHOSPHATE-MEVALONATE DECARBOXYLASE; MVK, MEVALONATE KINASE; PMK, PHOSPHOMEVALONATE KINASE; SMT, STEROL METHYLTRANSFERASE. L, expression lower than the control; H, expression higher than the control. Values are means ±SD (n=3); bars are SD. **P<0.01, Student’s t-test. (D) GC-MS analysis of campesterol, stigmasterol, β-sitosterol and cholesterol content (µg g⁻¹ dry weight (DW)) in F-244-treated roots. Sterols were extracted from 7-day-old roots, treated with F-244 (2 µM) or DMSO (control). L, value lower than the control. Values are means ±SD (n=4); bars are SD; **significant difference between F-244-treated roots and the DMSO control (P<0.01, Student’s t-test). (E) GC-MS analysis of campesterol, stigmasterol, and β-sitosterol content (µg g⁻¹ dry weight (DW)) in hmgs mutant and wild-type (WT) roots. Sterols were extracted from 7-day-old roots. L, value lower than the control. Values are means ±SD (n=4); bars are SD; *, **significant difference between hmgs roots and the WT (P<0.05 and P<0.01, respectively, Student’s t-test).

Inhibition of primary root growth was reversible

The F-244-mediated inhibition of Arabidopsis primary root growth was reversible (Fig. 3 with statistical analysis in Fig. 4), since primary root growth recovered gradually after removal of F-244 (Figs 3, 4). Average primary root length of F-244 (1 or 2 µM)-treated seedlings at day 7 (3.4 and 2.4 cm, respectively) was significantly lower than in the control (4.3 cm) (Fig. 4). However, after removal of F-244, the differences in primary root length between F-244 (1 or 2 µM)-treated seedlings and the control diminished (Fig. 4). For example, primary root length of F-244 (1 or 2 µM)-treated seedlings at day 7 after transfer to fresh MS medium was 5.2 and 4.4 cm, respectively, in comparison with 6.6 cm in the control (Fig. 4). Even without inhibitor removal, root elongation sped up in inhibitor-treated seedlings from day 5 to day 7 after F-244 treatment (Fig. 4).

Fig. 3.

Representative images of Arabidopsis seedlings exhibiting recovery of primary root growth after removal of F-244. Arabidopsis seedlings germinated on MS medium for 3 d were subsequently moved to fresh MS medium containing different concentrations of F-244 (1 or 2 µM) or DMSO for 7 d. Then 10-day-old Arabidopsis seedlings after 7-d F-244 treatment or DMSO control seedlings were transferred to new square plates containing fresh MS medium for recovery or fresh MS medium containing the original concentration of F-244 (1 or 2 µM) as controls. Photographs were taken daily up to 8 d after transfer. Scale bar: 1 cm.

Fig. 4.

Primary root length measurement during F-244 treatment and recovery period after removal of F-244. Arabidopsis seedlings germinated on MS medium for 3 d were subsequently moved to fresh MS medium containing different concentrations of F-244 (0, 1 or 2 µM) or DMSO for 7 d (days 0–7). Then the seedlings were grown on new square plates containing fresh MS medium for recovery or fresh MS medium containing the original concentration of F-244 (1 or 2 µM) as controls for 8 d (days 7–15). Values are mean ±SD (n=60); Bars are SD. a, significant difference between F-244 (1 µM) and the DMSO control; b, significant difference between F-244 (2 µM) and the DMSO control; c, significant difference between F-244 (1 µM) and F-244 (2 µM); d, significant difference between 0 and 1 µM F-244 during recovery; e, significant difference between 0 and 2 µM F-244 during recovery; H, value higher than the control; L, value lower than the control (P<0.001 by Student’s t-test).

F-244 treatment lowered stigmasterol and cholesterol content in Arabidopsis roots

In order to explore which processes were affected by F-244 and to further investigate the relationship between HMGS and other metabolic pathways, F-244-treated Arabidopsis roots and controls were studied by SWATH-MS quantitative proteomic analysis, which identified 1571 proteins, of which 214 were found to be differentially expressed. The expression of 85 of these 214 proteins was significantly (q<0.05) down-regulated, while the remaining 129 proteins were significantly (q<0.05) up-regulated. The proteomics data were specifically analysed with respect to the MVA pathway. HMGS, isopentenyl diphosphate isomerases IDI1 and IDI2, FARNESYL DIPHOSPHATE SYNTHASE1 (FPPS1), and Δ ²⁴-sterol reductase (DWF1) were identified from 1571 proteins but showed no significant difference in expression levels (see Supplementary Table S2). Other proteins downstream of HMGS, namely HMGR, squalene synthase (SQS), cycloartenol synthase (CAS), sterol methyltransferase (SMT), geranylgeranyl diphosphate synthase (GGPPS), and BRASSINOSTEROID-6-OXIDASE2 (BR60X2) remained undetected, most likely due to their low abundance.

Quantitative RT-PCR analysis on HMGS expression of F-244-treated Arabidopsis roots did not show any differences from the control (Fig. 2C). However, the expression of HMGR1, HMGR2, MEVALONATE KINASE (MVK), PHOSPHOMEVALONATE KINASE (PMK), DIPHOSPHO-MEVALONATE DECARBOXYLASE1 (MPDC1), FPPS1, FPPS2, and SMT1 was significantly (P<0.01) reduced (Fig. 2C), while MPDC2 was significantly (P<0.01) elevated (Fig. 2C). The GC-MS analysis of the major phytosterols (campesterol, stigmasterol, and β-sitosterol) and cholesterol in F-244-treated roots revealed that stigmasterol and cholesterol were 29.0% and 55.9% lower (P<0.01), respectively, than in the controls (Fig. 2D). However, such significant differences were not seen for campesterol and β-sitosterol, which may not be synthesized de novo but rather mobilized from seed stores (Fig. 2D). A complementation test with MVA indicated that MVA only partially complemented the inhibitory effect of F-244 on primary root growth after 6 d of treatment (9-day-old seedlings) (see Supplementary Fig. S4). Furthermore, a complementation test with stigmasterol indicated that it also partially complemented the inhibitory effect of F-244 on primary root growth after 3 d of treatment (on 6-day-old seedlings) (Supplementary Fig. S5). These results indicated that other factors exist for the primary root growth inhibition by F-244. When the content of the major phytosterols in the previously characterized hmgs mutant (Wang, 2010; Liao et al., 2014a) roots was analysed, campesterol, stigmasterol, β-sitosterol, and total phytosterols were 25.5%, 15.8%, 20.1%, and 19.9% lower (P<0.01), respectively, than the WT (Fig. 2E).

F-244 treatment reduced CK and auxin levels but enhanced BR and ethylene content in Arabidopsis roots

When hormone-related proteins were analysed to unravel the possible molecular mechanisms of F-244 inhibition of primary root growth, some were found to be ‘differentially expressed proteins' (DEPs) as shown in Table 1. When plant hormone (BR, CK, auxin, ethylene, or GA) levels and hormone marker gene (Nemhauser et al., 2006) expression were analysed in F-244-treated roots, it was found that the levels of two BRs (castasterone (CS) and typhasterol (TY)) significantly (P<0.05) increased by 72.2% and 16.7%, respectively (Fig. 5A). They corresponded to the up-regulation of BR biosynthesis-related genes (3-EPI-6-DEOXOCATHASTERONE 23-MONOOXYGENASE (CYP90D1), CYTOCHROME P450 90B1 (DWF4), and 7-DEHYDROCHOLESTEROL REDUCTASE (DWF5); Fig. 5B), and a BR-related protein, TRIP-1 (Table 1). The content of five CKs, namely isopentenyladenine riboside (iPR), isopentenyladenine (iP), trans-zeatin 7-glucoside (tZ7G), trans-zeatin riboside (tZR), and trans-zeatin 9-glucoside (tZ9G), significantly (P<0.01) declined by 45.7%, 22.1%, 39.0%, 44.3%, and 29.6%, respectively (Fig. 5A). This result is consistent with the down-regulation of 10 CK-related genes (ARABIDOPSIS TWO-COMPONENT RESPONSE REGULATORS (ARR1, ARR4, ARR5, ARR6, ARR7, and ARR9), ISOPENTENYLTRANSFERASES (IPT3, IPT5, and IPT7) and CYTOKININ OXIDASE (CKX4); Fig. 5C). However, the levels of two CKs (isopentenyladenine 9-glucoside (iP9G) and isopentenyladenine 7-glucoside (iP7G)) significantly (P<0.01) increased by 48.2% and 37.8%, respectively (Fig. 5A), corresponding to the up-regulation of CKX5 and HISTIDINE KINASE4 (AHK4) (Fig. 5C). IAA content was significantly (P<0.01) reduced by 26.9% (Fig. 5A), in agreement with the down-regulation of SAUR-LIKE AUXIN-RESPONSIVE PROTEINS (SAUR23, SAUR24 and SAUR62; Fig. 5D). Interestingly, the level of 1-aminocyclopropane-1-carboxylic acid (ACC) dramatically increased by 398% in F-244-treated roots (Fig. 5A), corresponding to the up-regulation of ETHYLENE RECEPTOR2 (ETR2) and ETHYLENE-RESPONSIVE TRANSCRIPTION FACTOR (ERF2) (Fig. 5E). There were no significant differences in GA levels between F-244-treated roots and the control, except for GA7, which was detected in control but not F-244-treated roots (see Supplementary Fig. S6A). The expression of two GA-related genes (GA REQUIRING4 (GA4) and RGA-LIKE1 (RGL1)) was lower in F-244-treated roots than in the control (Supplementary Fig. S6B). Taken together, these results indicate that reduction in primary root growth after F-244 treatment positively correlates with CK and auxin levels, and negatively with ACC and BR contents.

Table 1.

Representative differentially expressed proteins (DEPs) in F-244-treated Arabidopsis roots in comparison with the DMSO control

ID in Arabidopsis	Protein name	Function	Fold change
Secondary metabolism-related DEPs
AT3G25830	TERPENE SYNTHASE-LIKE SEQUENCE-1,8-CINEOLE (TPS-CIN)	Monoterpene synthesis	0.39
AT3G19450	CINNAMYL-ALCOHOL DEHYDROGENASE (CAD4)	Phenylpropanoid/lignin synthesis	0.39
AT4G33360	Terpene cyclase/mutase-related, farnesol dehydrogenase (FLDH)	Farnesol synthesis	1.29
AT1G75280	Isoflavone reductase (IFR)	flavonoid/isoflavone synthesis	1.53
AT2G25450	2-Oxoglutarate-dependent dioxygenase (GS-OH)	Glucosinolate synthesis	0.59
AT1G47600	BETA GLUCOSIDASE34 (BGLU34)	Glucosinolate degradation	0.38
AT5G23020	2-ISOPROPYLMALATE SYNTHASE2 (MAM3)	Glucosinolate synthesis	3.11
AT4G13430	ISOPROPYL MALATE ISOMERASE LARGE SUBUNIT1 (IPMI LSU1)	Glucosinolate synthesis	1.44
AT2G43100	ISOPROPYLMALATE ISOMERASE2 (IPMI SSU2)	Glucosinolate synthesis	2.89
AT3G44300	NITRILASE2 (NIT2)	Glucosinolate synthesis	2.08
Hormone metabolism-related DEPs
AT2G46280	TGF-BETA RECEPTOR INTERACTING PROTEIN1 (TRIP-1)	Brassinosteroid (BR) signal	1.62
AT2G25980	Jacalin lectin family protein	Jasmonic acid (JA) induced	1.54
AT3G16450	Jacalin lectin family protein	JA induced	0.52
AT2G43820	UDP-GLUCOSYLTRANSFERASE 74F2 (UGT74F2)	Salicylic acid (SA) synthesis	0.71
Cell organization-related DEPs
AT3G18780	ACTIN2 (ACT2)	Cell organization	0.75
AT4G30160	ARABIDOPSIS THALIANA VILLIN 4 (VLN4)	Cell organization	0.67
AT5G62700	TUBULIN BETA CHAIN3 (TUB3)	Cell organization	1.29
AT4G34450	GAMMA-2 COAT PROTEIN, PUTATIVE	Cell vesicle transport	1.21
Development-related DEPs
AT5G24780	VEGETATIVE STORAGE PROTEIN 1 (VSP1)	Development	0.42
AT5G44020	Acid phosphatase class B family protein	Development	0.65
AT3G18060	WD-40 REPEAT FAMILY PROTEIN	Development	0.43
AT1G18080	RECEPTOR FOR ACTIVATED C KINASE 1 A (RACK1A)	Development	1.45
AT2G34040	APOPTOSIS INHIBITORY5 (API5) family protein	Development	2.69
AT3G13300	VARICOSE (VCS)	Development	1.35
AT3G50670	U1 SMALL NUCLEAR RIBONUCLEOPROTEIN-70K (U1-70K)	Development	2.29
AT4G21150	HAPLESS 6 (HAP6)	Development	1.31

DEPs: differentially expressed proteins (fold change >1.2 (up-regulated) or <0.8 (down-regulated), q<0.05). DEPs were distributed to different functional categories by the MapMAN BIN system.

Fig. 5.

Analysis of plant hormone levels and quantitative RT-PCR analysis on the expression of hormone-related genes in F-244-treated Arabidopsis roots. (A) Levels of BRs, CKs, auxin, and ethylene. BRs, CKs, auxin, and ACC was extracted, respectively, from 7-day-old roots, after treatment by F-244 (2 µM) or DMSO (control). (B) Expression of BR-related genes. (C) Expression of CK-related genes. (D) Expression of auxin-related genes. (E) Expression of ethylene-related genes. Total RNA was extracted from 7-day-old roots, after treatment by F-244 (2 µM) or DMSO (control). Values are means ±SD (n=3). Bars are SD. L, value lower than in control; H, value higher than in control. *P<0.05; **P<0.01, Student’s t-test. ACC, 1-aminocyclopropane-1-carboxylic acid; AHK, HISTIDINE KINASE; ARR, ARABIDOPSIS TWO-COMPONENT RESPONSE REGULATOR; BAS1, CYTOCHROME P450 MONOOXYGENASE 734A1; BR60X2, BRASSINOSTEROID-6-OXIDASE2; BRI1, BRASSINOSTEROID INSENSITIVE1; BZR1, BRASSINAZOLE RESISTANT1; CKX, CYTOKININ OXIDASE; CS, castasterone; CYP90D1, 3-EPI-6-DEOXOCATHASTERONE 23-MONOOXYGENASE; 6-deoxoCS, 6-deoxocastasterone; DFL2, GH3-LIKE PROTEIN; DHZR, dihydrozeatin riboside; DWF4, CYTOCHROME P450 90B1; DWF5, 7-DEHYDROCHOLESTEROL REDUCTASE; EBF2, EIN3-BINDING F-BOX PROTEIN1; ERF, ETHYLENE-RESPONSIVE TRANSCRIPTION FACTOR; ETR2, ETHYLENE RECEPTOR2; FW, fresh weight; IAA, indole-3-acetic acid; IAA5, INDOLE-3-ACETIC ACID INDUCIBLE5; iP, isopentenyladenine; iPR, isopentenyladenine riboside; IPT, isopentenyltransferase; iP9G, isopentenyladenine 9-glucoside; iP7G, isopentenyladenine 7-glucoside; PBP1, PINOID-BINDING PROTEIN1; ROT3, 3-EPI-6-DEOXOCATHASTERONE 23-MONOOXYGENASE; SAUR, SAUR-LIKE AUXIN-RESPONSIVE PROTEIN; TY, typhasterol; tZ, trans-zeatin; tZ7G, trans-zeatin 7-glucoside; tZ9G, trans-zeatin 9-glucoside; tZOG, trans-zeatin-O-glucoside; tZR, trans-zeatin riboside.

Proteomic analysis of F-244-treated roots reveals DEPs in secondary metabolism

F-244 treatment of Arabidopsis roots identified DEPs in other secondary metabolic pathways including monoterpene, phenylpropanoid/lignin, farnesol, isoflavonoid, and glucosinolate biosynthesis. For example, TERPENE SYNTHASE-LIKE SEQUENCE-1,8-CINEOLE (TPS-CIN) in monoterpene biosynthesis and CINNAMYL-ALCOHOL DEHYDROGENASE (CAD4) in phenylpropanoid/lignin biosynthesis were both down-regulated, by 0.39-fold (Table 1). However, an NAD⁺-dependent farnesol dehydrogenase (FLDH) that oxidizes farnesol to farnesal and an isoflavone reductase (IFR) in isoflavonoid biosynthesis were up-regulated, by 1.29- and 1.53-fold, respectively (Table 1). The expression of two glucosinolate-related proteins, namely 2-oxogutarate-dependent dioxygenase (GS-OH) and BETA GLUCOSIDASE34 (BGLU34), decreased, by 0.59- and 0.38-fold, respectively (Table 1). However, three proteins in two consecutive reactions in glucosinolate biosynthesis, namely the rate-limiting enzyme 2-ISOPROPYLMALATE SYNTHASE2 (MAM3) (Textor et al., 2007), ISOPROPYLMALATE ISOMERASE LARGE SUBUNIT1 (IPMI LSU1), and ISOPROPYLMALATE ISOMERASE2 (IPMI SSU2) (De Kraker et al., 2007), and another protein in glucosinolate biosynthesis, NITRILASE2 (NIT2), were enhanced, by 3.11-, 1.44-, 1.89-, and 2.08-fold, respectively (Table 1).

Correlation between proteomics data and gene expression by qRT-PCR

To test whether changes in protein expression correlated with alterations in gene expression, selected genes encoding monoterpene-, phenylpropanoid/lignin-, farnesol-, isoflavonoid-, auxin- and glucosinolate-related DEPs were analysed. The changes in expression of 8 out of 10 selected genes were consistent with the proteomics data (Fig. 6A). For example, the expression of monoterpene-related TPS-CIN and phenylpropanoid/lignin-related CAD4 in F-244-treated roots was 75.2% and 76.8% lower than in the control, respectively, while the expression of farnesol-related FLDH and isoflavonoid-related IFR was 171% and 83.0% higher, respectively, than in the control (Fig. 6A). Also, the expression of glucosinolate-related BGLU34 was 67.1% lower than in the control (Fig. 6A), but the expression of MAM3, IPMI LSU1, and IMPI SSU2 was 391%, 455%, and 440% higher, respectively, than in the control (Fig. 6A). In contrast to proteomics data for GS-OH and NIT2 (Table 1), GS-OH expression was elevated, while NIT2 expression declined (Fig. 6A).

Fig. 6.

Correlation analysis of proteomics data and gene expression by quantitative RT-PCR analysis, effects of F-244 on glucosinolate biosynthesis in F-244-treated Arabidopsis roots, and profiles of glucosinolates in Arabidopsis hmgs mutant roots. (A) Expression of genes encoding selected DEPs identified by SWATH-MS. (B) Expression of another five genes in glucosinolate biosynthesis. (C) HPLC analysis of glucosinolate content in F-244-treated roots. (D) HPLC analysis of glucosinolate content in hmgs mutant roots. Total RNA was extracted from 7-day-old roots, treated with F-244 (2 µM) or DMSO (control). Glucosinolates were extracted from 7-day-old roots. Glucosinolate content is shown in µmol g⁻¹ DW. Values are means ±SD; n=4. Bars are SD. L, value lower than in control; H, value higher than in control. *P<0.05; **P<0.01, Student’s t-test. BACT4, BRANCHED-CHAIN AMINO ACID AMINOTRANSFERASE4; BGLU34, BETA GLUCISIDASE34; CAD4, CINNAMYL-ALCOHOL DEHYDROGENASE; DW, dry weight; FLDH, FARNESOL DEHYDROGENASE; GS-OH, 2-OXOGLUTARATE-DEPENDENT DIOXYGENASE; I3M, indol-3-ylmethyl; IFR, ISOFLAVONE REDUCTASE; IPMDH, 3-ISOPROPYLMALATE DEHYDROGENASE; IPMI LSU1, ISOPROPYLMALATE ISOMERASE LARGE SUBUNIT1; IPMI SSU2, ISOPROPYLMALATE ISOMERASE2; IPMI SSU3, ISOPROPYLMALATE ISOMERASE3; MAM1, METHYLTHIOALKYLMALATE SYNTHASE1; MAM3, 2-ISOPROPYLMALATE SYNTHASE2; 1MOI3M, 1-methoxyindole-3-ylmethyl; 4MSOB, 4-methylsulfinylbutyl; 7MSOH, methylsulfinylheptyl; 8MSOO, 8-methylsulfinyloctyl; 4MOI3M, 4-methoxyindole-3-ylmethyl; 4MTB, 4-methylthiobutyl; 7MTH, 7-methylthioheptyl; 8MTO, 8-methylthiooctyl; NIT2, NITRILASE2; TPS-CIN, TERPENE SYNTHASE-LIKE SEQUENCE-1,8-CINEOLE.

Inhibition of HMGS activates glucosinolate biosynthesis

With five DEPs related to glucosinolate biosynthesis/degradation identified by SWATH-MS proteomic analysis (Table 1), an interaction between the HMGS/MVA pathway and glucosinolate biosynthesis/degradation has emerged. When the expression of five other genes in glucosinolate biosynthesis, namely BRANCHED-CHAIN AMINO ACID AMINOTRANSFERASE4 (BACT4), METHYLTHIOALKYLMALATE SYNTHASE1 (MAM1), IPMI SSU3, 3-ISOPROPYLMALATE DEHYDROGENASE1 (IPMDH1), and IPMDH3, in F-244-treated roots were analysed by qRT-PCR, BCAT4, MAM1, IPMI SSU3, IPMDH1, and IPMDH3 appeared to be induced (Fig. 6B). For example, the expression of BCAT4, MAM1, IPMI SSU3, IPMDH1, and IPMDH3 was 105%, 249%, 1172%, 1064%, and 121% higher, respectively, than in the control (Fig. 6B). When the content of glucosinolates in F-244-treated roots was analysed by HPLC, 4-methylthiobutyl (4MTB), 8-methylsulfinyloctyl (8MSOO), 7-methylsulfinylheptyl (7MSOH), indol-3-ylmethyl (I3M), and 4-methoxyindole-3-ylmethyl (4MOI3M) content was 88.1%, 35.7%, 24.2%, 74.3%, and 92.3% higher, respectively, than in the control (Fig. 6C). However, 8-methylthiooctyl (8MTO) content was 28.1% lower than in the control (Fig. 6C). When the glucosinolate content in hmgs mutant roots was analysed, 4-methylsulfinylbutyl (4MSOB), 4MTB, 7-methylthioheptyl (7MTH), 8MTO, I3M, 4MOI3M, and IMOI3M had increased 114%, 45.5%, 71.7%, 66.7%, 67.5%, 59.3%, and 16.9%, respectively, over the WT (Fig. 6D). When the total glucosinolate content in F-244-treated and hmgs mutant roots was analysed, it was enhanced by 20.2% and 28.0%, over the DMSO control and the WT, respectively (see Supplementary Fig. S7A, B).

The overexpression of HMGS enhanced primary root growth and inhibited glucosinolate biosynthesis

Arabidopsis HMGS-OEs showed increased primary root growth in 3-day and 7-day-old seedlings (Fig. 7A, B). For 3-day-old seedlings, the average primary root length in HMGS-OEs was 44.1% higher than in the vector control (Fig. 7B); for 7-day-old seedlings, it was 15.4% higher (Fig. 7B). The analysis of Arabidopsis HMGS-OE roots by qRT-PCR and HPLC revealed that the glucosinolate biosynthesis-related genes BCAT4, MAM1, MAM3, IPMI LSUI1, IMPI SSU2, IPMI SSU3, IPMDH1, and IPMDH3 were significantly down-regulated (Fig. 7C). The content of some glucosinolates, namely 4MSOB, 8MTO, I3M, 4MOI3M, and 1-methoxyindole-3-ylmethyl (1MOI3M), had decreased (Fig. 7D). For example, the average content of 4MSOB, 8MTO, I3M, 4MOI3M, and 1MOI3M in HMGS-OE roots was 26.9%, 16.3%, 17.3%, 29.4% and 14.0% lower, respectively, than in the control (Fig. 7D). When the total glucosinolate content in HMGS-OE roots was analysed, the average total amount was 11.1% lower than in the control (see Supplementary Fig. S7C).

Fig. 7.

The overexpression of HMGS enhanced primary root growth and inhibited glucosinolate biosynthesis in Arabidopsis HMGS-OE roots. (a) Phenotype of primary roots of 3- and 7-day-old Arabidopsis HMGS-OE seedlings (OE-wtBjHMGS1-1 and OE-wtBjHMGS1-2) and vector-transformed control (pSa13). (b) Root length measurements of 3- and 7-day-old HMGS-OEs and vector-transformed control (pSa13). Arabidopsis seeds germinated on MS medium for 3 d were subsequently moved to fresh MS medium for 4 d. Photographs were taken and root length measured on 3- and 7-day-old seedlings. Scale bar: 1 cm. L, primary root length lower than control. Values are means ±SD (n=30); bars are SD. (c) Quantitative RT-PCR analysis on the expression of genes in glucosinolate biosynthesis. Total RNA was extracted from 7-day-old roots. (d) HPLC analysis of glucosinolate content in 7-day-old roots. Glucosinolate content is shown in µmol g⁻¹ dry weight (DW). Values are means ±SD; n=4. Bars are SD. L, value lower than in control; H, value higher than in control. *P<0.05; **P<0.01, Student’s t-test. BACT4, BRANCHED-CHAIN AMINO ACID AMINOTRANSFERASE4; BGLU34, BETA GLUCISIDASE34; DW, dry weight; I3M, indol-3-ylmethyl; IPMDH, 3-ISOPROPYLMALATE DEHYDROGENASE; IPMI LSU1, ISOPROPYLMALATE ISOMERASE LARGE SUBUNIT1; IPMI SSU2, ISOPROPYLMALATE ISOMERASE2; IPMI SSU3, ISOPROPYLMALATE ISOMERASE3; MAM1, METHYLTHIOALKYLMALATE SYNTHASE1; MAM3, 2-ISOPROPYLMALATE SYNTHASE2; 1MOI3M, 1-methoxyindole-3-ylmethyl; 4MSOB, 4-methylsulfinylbutyl; 7MSOH, methylsulfinylheptyl; 4MTB, 4-methylthiobutyl; 8MSOO, 8-methylsulfinyloctyl; 4MOI3M, 4-methoxyindole-3-ylmethyl; 7MTH, 7-methylthioheptyl; 8MTO, 8-methylthiooctyl.

Discussion

Reduced primary root length after F-244 treatment may have resulted from lower sterol, IAA, and CK contents

In this study, the function of Arabidopsis HMGS in primary root development was demonstrated by the use of F-244 or by the overexpression of BjHMGS1. Similar phenotypic changes in root growth inhibition have been reported with the HMGR inhibitor mevinolin on radish and Arabidopsis seedlings (Bach and Lichtenthaler, 1983; Suzuki et al., 2004). At a slightly higher concentration of mevinolin (1 µM), radish hypocotyl growth decreased, accompanied by a drop in sterol accumulation (Bach and Lichtenthaler, 1983). The results from SWATH-MS quantitative proteomics, qRT-PCR, GC-MS, and HPLC suggested several reasons for F-244-induced reduction in primary root growth: down-regulation of isoprenoid-related genes such as HMGR1, HMGR2, MVK, PMK, MPDC1, FPPS1, FPPS2, and SMT1 (Fig. 2C) followed by a reduction in stigmasterol and cholesterol content (Fig. 2D); down-regulation of hormone-related genes, namely CK-related (ARR1, ARR4, ARR5, ARR6, ARR7, ARR9, IPT3, IPT5, IPT7, and CKX4), and auxin-related (SAUR23, SAUR24, and SAUR62) genes, followed by a reduction in some CKs and IAA (Fig. 5); and reduction in primary root growth corresponding to a decline in proteins involved in cell organization (ACT2 and VLN4) and development (VSP1, acid phosphatase class B family protein and WD-40 repeat family protein) (Table 1).

Also stigmasterol and cholesterol contents were reduced in F-244-treated roots (Fig. 2D), similar to the situation in Arabidopsis hmgs mutant roots (Fig. 2E). In contrast, the stigmasterol and β-sitosterol content of Arabidopsis OE-wtBjHMGS1 seedlings had increased by 118.8% and 8.8%, respectively (Wang et al., 2012). The total sterol content was significantly increased in Arabidopsis rosette leaves overexpressing AtHMGS or AtHMGR1 (Lange et al., 2015). Furthermore, many studies have reported that mutation in sterol biosynthesis genes such as SMT and sterol C4-METHYL OXIDASE2 (SMO2) resulted in lower β-sitosterol and stigmasterol content, accompanied by retarded root development (Diener et al., 2000; Willemsen et al., 2003; Schrick et al., 2004; Carland et al., 2010; Nakamoto et al., 2015; Zhang et al., 2016). The defects in lateral root development in the smt2 smt3 double mutant were partially restored by exogenously supplemented β-sitosterol or stigmasterol (Nakamoto et al., 2015). Taken together, stigmasterol and β-sitosterol appear important for root development in Arabidopsis.

IAA content was significantly reduced in F-244-treated roots (Fig. 5A), implying that the reduced root growth may be related to auxin as well. It has been previously observed that normal sterol composition is deemed important for auxin biosynthesis and signalling in Arabidopsis smt1, smt2 smt3, and smo2 mutants (Willemsen et al., 2003; Carland et al., 2010; Nakamoto et al., 2015; Zhang et al., 2016).

The function of CK in root development has been reported previously (Higuchi et al., 2004; Dello Ioio et al., 2007, 2008; Ruzicka et al., 2009; Schaller et al., 2015; Liu et al., 2017 and references cited therein). The active forms of CKs such as trans-zeatin (tZ)- and iP-type CKs are mainly dependent on the plastidial MEP pathway, while the cis-zeatin-type CKs are largely derived from the MVA pathway in Arabidopsis (Kasahara et al., 2004). In F-244-treated roots, the levels of two iP-type CKs (IPR and IP) and all tZ-type CKs were significantly reduced, while other CKs (IP9G and IP7G) significantly increased (Fig. 5A). Hence, changes in HMGS activity appear to affect homeostasis of CKs and the levels of tZ- and iP-type CKs. Changes in CK levels have also been detected in transgenic Arabidopsis overexpressing FPPS, and reduction in tZ- and iP-type CKs (Manzano et al., 2006) could have resulted from redirecting IPP units away from CK synthesis, favoring sterol biosynthesis instead. The Arabidopsis hmg1-1 mutant showed shorter primary root length accompanied by enhanced tZ-type CK levels (Suzuki et al., 2004). Inhibition of HMGR caused a decrease in Z-type CK in Arabidopsis seedlings and tobacco BY-2 cells (Laureys et al., 1998; Åstot et al., 2000). Thus, the reduced root growth after F-244 treatment may be also related to CKs. There is also evidence for some crosstalk between different hormones (Skoog and Miller, 1957; Dello Ioio et al., 2007, 2008; Ruzicka et al., 2009; Schaller et al., 2015). Taken together, the decline in primary root growth by F-244 treatment seems to have resulted from a lower content in stigmasterol, IAA, and some CKs.

Although extensive studies have reported that BRs play an important role in regulating root development including primary root growth (Wei and Li, 2016 and references cited therein), F-244 treatment caused up-regulation of several BR biosynthesis-related genes and thereby an increase in CS and TY levels (Fig. 5A, B), suggesting that the reduced primary root growth by F-244 treatment is independent of BRs. This is supported by the observation that SMTs regulate Arabidopsis development including primary root growth that was caused by greatly decreased sterol but not BR levels in smt mutants (Carland et al., 2010).

Given the proposed signaling functions for MVA and MEP intermediates such as methylerythritol cyclodiphosphate (Xiao et al., 2012), the results from qRT-PCR indicated that F-244 might cause some accumulation of downstream intermediates, thereby down-regulating most genes located downstream from HMGS (Fig. 2C), and indirectly affecting metabolism and development. F-244 was shown to irreversibly bind to Cys-117 at the HMGS reaction center (Pojer et al., 2006). In this study, HMGS activity apparently recovered after removal of F-244 (Figs 3, 4) suggesting that recovery of pathway activity relies on de novo expression of HMGS molecules, as supported by the observation that after F-244 treatment HMGS transcripts and their proteins were maintained at levels similar to the control (Fig. 2C; Supplementary Table S2).

A possible relationship between the MVA pathway and glucosinolate biosynthesis

In our study, SWATH-MS quantitative proteomic analysis on F-244-treated Arabidopsis roots revealed a connection between HMGS and glucosinolate biosynthesis. Aliphatic glucosinolates are derived from Met (Sønderby et al., 2010). MAM3 is the committed enzyme in Met side chain elongation in glucosinolate biosynthesis (Textor et al., 2007). It catalyses the condensation of acetyl-CoA with ω-methylthio-2-oxo acid that is derived from Met to form ω-methylthioalkylmalate (Campos de Quiros et al., 2000; Kroymann et al., 2001, 2003; Field et al., 2004; Textor et al., 2007). Acetyl-CoA is one of the substrates for MAM3 in glucosinolate biosynthesis. Given that F-244 specifically inhibits the condensation reaction of acetyl-CoA and AcAc-CoA to form HMG-CoA catalysed by HMGS (Pojer et al., 2006), the accumulated acetyl-CoA could be available as an activator for an independent pathway, such as glucosinolate biosynthesis as reported here (Fig. 6). Considering the up-regulated MAM3 expression, it might also somehow be used as a substrate of MAM3, participating in glucosinolate biosynthesis (Table 1; Fig. 6A). The converse is to be expected if the accumulation of glucosinolates were diminished when the MVA pathway is enhanced through HMGS overexpression (Fig. 7). As Arabidopsis MAM, IPMI, and isopropylmalate dehydrogenase (IPMDH) are localized in plastids (Tanz et al., 2013; Aarabi et al., 2016), cytosolic acetyl-CoA in F-244-treated roots needs to be transferred to the plastids for Met side chain elongation and glucosinolate biosynthesis, and likewise plastidial acetyl-CoA in HMGS-OE roots to be mobilized to the cytosol for sterol biosynthesis. As acetyl-CoA cannot be directly transported from the cytosol to the plastids or vice versa (Oliver et al., 2009; Balcke et al., 2017), some acetyl-CoA-related transporters such as malate, citrate, or pyruvate may facilitate this process (Oliver et al., 2009; Hemmerlin et al., 2012; Xing et al., 2014; Balcke et al., 2017). Overall, these results implicate a relationship between HMGS activity and glucosinolate biosynthesis through acetyl-CoA, and suggest that HMGS serves as another target for manipulation of glucosinolate biosynthesis (Fig. 8). However, as acetyl-CoA is also produced and used by other pathways such as carbohydrate metabolism, fatty acid biosynthesis and metabolism, and polyketide biosynthesis, the possible effects of F-244 treatment on these pathways cannot be excluded.

Fig. 8.

Modulation of HMGS affects glucosinolate biosynthesis in Arabidopsis roots. Inhibition of HMGS activates glucosinolate biosynthesis in Arabidopsis roots, while HMGS overexpression hinders glucosinolate biosynthesis. Inhibition of HMGS down-regulated genes in sterol biosynthesis and reduced stigmasterol and cholesterol, but up-regulated genes/proteins in glucosinolate biosynthesis and elevated glucosinolate content. The overexpression of HMGS up-regulated genes in sterol biosynthesis and enhanced β-sitosterol and stigmasterol (Wang et al., 2012), but down-regulated genes in glucosinolate biosynthesis and reduced glucosinolate content. The MVA pathway occurs in the cytosol (light yellow), and Met side chain elongation occurs in plastids (green). Other steps in glucosinolate biosynthesis occurs in the cytosol. The cytosolic MVA pathway interacts with plastidial Met side chain elongation, possibly through acetyl-CoA-related compounds that can be transferred from the cytosol to the plastids or from the plastids to the cytosol, e.g. malate, pyruvate, or citrate (pink), when HMGS is inhibited or overexpressed. Up-regulated genes/proteins or compounds are marked in red, down-regulated genes/proteins or reduced compounds in blue. Ac-CoA, acetyl-CoA; AcAc-CoA, acetoacetyl-CoA; BCAT4, BRANCHED-CHAIN AMINO ACID AMINOTRANSFERASE4; CYP710A1, C-22 STEROL DESATURASE; DMAPP, dimethylallyl diphosphate; FPP, farnesyl diphosphate; FPPS, FPP SYNTHASE; HMG-CoA, 3-hydroxy-3-methylglutaryl-coenzyme A; HMGR, HMG-COA REDUCTASE; HMGS, HMG-COA SYNTHASE; IDI, ISOPENTENYL DIPHOSPHATE ISOMERASE; IPP, isopentenyl diphosphate; IPMDH, ISOPROPYLMALATE DEHYDROGENASE; IPMI LSU1, ISOPROPYLMALATE ISOMERASE LARGE SUBUNIT1; IPMI SSU, ISOPROPYLMALATE ISOMERASE SMALL SUBUNIT; MAM, METHYLTHIOALKYLMALATE SYNTHASE; Met, methionine; MPDC1, MEVALONATE DIPHOSPHATE DECARBOXYLASE1; MVK, MEVALONATE KINASE; PMK, PHOSPHOMEVALONATE KINASE; Trp, tryptophan; SMT, STEROL METHYLTRANSFERASE.

It should be noted that not only the content of aliphatic glucosinolates, but changes in indolic glucosinolates such as I3M, 4MOI3M and/or 1MOI3M occurred in F-244-treated roots, hmgs mutant roots and HMGS-OE roots (Figs 6C, 6D, 7D). Given that indolic glucosinolates are derived from Trp but not Met, the results suggest that other mechanisms may be involved in the changes of indolic glucosinolate content after F-244 treatment. The biosynthesis of indolic glucosinolates and IAA can be linked through their common intermediate indole-3-acetaldoxime (IAOx) generated from Trp (Zhao et al., 2002; Frerigmann and Gigolashvili, 2014; Malka and Cheng, 2017). Indole glucosinolate metabolism had been reported to be involved in auxin homeostasis (Delarue et al., 1998; Barlier et al., 2000; Bak et al., 2001; Salehin et al., 2019). Interestingly, IAA content declined (Fig. 5A), while indole glucosinolates (I3M and 4MOI3M) increased, in F-244-treated roots (Fig. 6C), indicating that the increase in indolic glucosinolate may arise from the decrease in IAA content of F-244-treated roots, due to the larger availability of IAOx routed away from IAA formation in Arabidopsis roots. Furthermore, the accumulation of some glucosinolates has been reported to inhibit root growth and development (Katz et al., 2015; Francisco et al., 2016; Malinovsky et al., 2017; Urbancsok et al., 2017), consistent with the observations that indole glucosinolates increased (Fig. 6C) and root growth reduced after F-244 treatment (Fig. 2). Taken together, both the MVA pathway and glucosinolate biosynthesis are associated with root growth regulation and the plant hormone auxin, supporting a regulatory link between these pathways.

Inconsistency at protein and mRNA levels for GS-OH and NIT2 (Fig. 6A; Table 1) suggests some post-transcriptional modifications may have occurred. Indeed, uncoupling of transcript and protein level is well known for glucosinolate-related genes (Svozil et al., 2015). Although a contradictory change occurred for 8MTO levels in F-244-treated and hmgs roots (Fig. 6C, D), changes for the four other glucosinolates 4MTB, I3M, 4MOI3M, and IMOI3M were similar (Fig. 6C, D) and the total glucosinolate content increased in both samples (see Supplementary Fig. S7A, B), suggesting that alteration in HMGS activity results in changes in total glucosinolate amount.

However, the identified proteins with differential expression patterns from SWATH-MS quantitative proteomic analysis did not include several key enzymes of glucosinolate biosynthesis (i.e. CYTOCHROME P450, FAMILY 79 (CYP79), CYTOCHROME P450, FAMILY 83 (CYP83), and UDP-glycosyltransferase (UGT)). This may be due to their low abundance in Arabidopsis roots. Similarly, a key enzyme of the MVA pathway, HMGR, was also not identified from proteomic analysis. An alteration in glucosinolate metabolism may be dependent on a pool of pre-existing intermediates. Also, the possibility of changes in glucosinolate content depending on de novo biosynthesis cannot be excluded at this point.

Supplementary data

Supplementary data are available at JXB online.

Fig. S1. Expression of UBIQUITIN EXTENSION PROTEIN2 (UBQ2) in F-244-treated roots by quantitative RT-PCR analysis.

Fig. S2. Expression of UBIQUITIN EXTENSION PROTEIN2 (UBQ2) in OE-wtBjHMGS1 roots by quantitative RT-PCR analysis.

Fig. S3. Phenotypic changes of Arabidopsis (Col-0) primary root growth after F-244 treatment.

Fig. S4. The effect of F-244 on Arabidopsis primary root growth was partially complemented by mevalonate.

Fig. S5. The effect of F-244 on Arabidopsis primary root growth was partially complemented by stigmasterol.

Fig. S6. The effect of F-244 on GA biosynthesis and GA-related gene expression.

Fig. S7. HPLC analysis of total glucosinolate content in 7-day-old roots.

Table S1. Sequences of primers used in this study.

Table S2. Secondary metabolism-related proteins identified from SWATH-MS proteomic analysis without significant differences in F-244-treated Arabidopsis roots in comparison with the DMSO control.

Data deposition

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (Vizcaíno et al., 2014) via the PRIDE (Vizcaíno et al., 2016) partner repository with the dataset identifier PXD009776.

Author contributions

PL, MLC, and TJB designed the research. PL performed inhibitor treatment, phenotypic analysis, RNA extraction, qRT-PCR, GC-MS analysis of sterols, and HPLC analysis of glucosinolates. PL, WLC, and CL performed protein extraction and proteomic analysis. SCL repeated inhibitor treatment and recovery experiments. PL, SCL, MLC, and TJB analysed data. PL, SCL, and MLC wrote the manuscript. All authors have read and approved the manuscript. There is no conflict of interest.