Introduction

Iron is essential for the growth and development of all plants. Iron deficiency induces chlorosis, inhibit plant development, and reduce crop yield and quality. Some species are tolerant to iron starvation, such as Malus xiaojinensis, Prunus cerasifera, Prunus cerasus, Citrus limonia, and Citrus volkameriana (Han et al. 1994a, 1998; Tagliavini and Rombolà 2001). However, the mechanisms of their resistance to iron deficiency in woody plants are not fully understood.

Plants improve iron uptake when suffering from iron deficiency. Non-graminaceous plants excrete protons and phenolic compounds from their roots into the rhizosphere in low-iron conditions, which is thought to help increase the solubility of ferric ions or to support the reducing capacity of Fe on the root surface, and the related genes IRT1, FRO2, and HAs are upregulated significantly (Kobayashi and Nishizawa 2012).

Besides improving iron uptake, plants can also resist iron deficiency by enhancing iron remobilization. When suffering iron deficiency stress, expression of citric acid synthase (CS) and nicotianamine synthase (NAS) encoded genes were upregulated (Han et al. 2013; Higuchi et al. 2001). Compared with iron-sufficient conditions, the NA content became higher in roots and stems in Arabidopsis after iron starvation (Douchkov et al. 2005). Citrate is also accumulated in pea roots and leaves (Kabir et al. 2012). Another iron transporter, NRAMP3, which locates in the vacuolar membrane and can transfer iron from the vacuole to the cytoplasm was upregulated upon iron deficiency (Bereczky et al. 2003).

M. xiaojinensis is a Fe-efficient apple species (Han et al. 1994a) that shows no significant iron deficiency symptoms when the iron concentration is 10 μM in the medium (Han et al. 1994b). Although the leaves become yellow after 9 days iron starvation (4 μM iron) in M. xiaojinensis, this phenomenon occurs later than in the Fe-inefficient species Malus baccata. In M. xiaojinensis, as a non-graminaceous plant, the rhizosphere pH decreases significantly and the activities of ferric reductase and iron transporters are upregulated to cope with iron deficiency, which indicates iron uptake is enhanced (Zhang et al. 2012; Wu et al. 2011). In addition, iron deficiency-induced upregulation of MxNAS1 and MxCS1 expression demonstrated that iron remobilization is also enhanced after iron starvation in M. xiaojinensis (Zhang et al. 2009; Han et al. 2013). However, whether the iron tolerance of M. xiaojinensis is due to the enhancement of iron uptake or remobilization is not clear.

Signal substances, such as plant hormones, nitric oxide (NO), and hydrogen peroxide (H2O2), are key regulators in abiotic stress responses. Among them, NO, auxin, and ethylene are the three most important signaling substances responsive to iron deficiency stress (García et al. 2011). Sodium nitroprusside (SNP), a NO donor, completely prevented maize leaf interveinal chlorosis, resulting in a 70 % increase in chlorophyll content compared with control plants (Graziano et al. 2002). In tomato, the application of the NO-scavenger cPTIO [2-(4-carboxyphenyl)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide] inhibited the expression of iron deficiency-induced genes LeFER, LeFRO1, and LeIRT1 in roots (Graziano and Lamattina 2007). In M. xiaojinensis, the expression of MxFIT, MxFRO2, and MxIRT1 is increased significantly with 50 μM SNP treatment in roots compared with control plants, while cPTIO completely represses MxFIT, MxFRO2 and MxIRT1 in roots after iron starvation compared with controls (unpublished data).

Auxin is one of the most important signal substances in iron deficiency. In Arabidopsis, Fe-deficiency-induced reductase activity is enhanced by 40 % after treatment with a-naphthaleneacetic acid (NAA) compared with Fe-sufficient plants, and is strongly inhibited by growing in 1-naphthylphthalamic acid (NPA) condition compared with Fe-deficient plants. In addition, there is a fourfold increase in indole-3-acetic acid (IAA) accumulation in the roots of Fe-deficient plants compared with those of Fe-sufficient plants (Chen et al. 2010). In M. xiaojinensis, IAA content in roots is increased by 32 and 67 % after 3 and 6 days, respectively, growing in iron-deficient conditions. In addition, MxFRO2 is upregulated after supplying NAA in the medium for 3 or 6 days and inhibited by NPA treatment (Wu et al. 2010).

Ethylene is another signal substance in the iron deficiency response. Ferric reductase activity is enhanced by 2-, 2.7-, 2.2-, and 5.4-folds in pea (Pisum sativum L. “Sparkle”), sugar beet (Beta vulgaris L.), tomato (Lycopersicon esculentum Mill. T3238FER), and Arabidopsis [Arabidopsis thaliana (L.) Heynh “Columbia”] respectively, after 24 h growing in the ethylene precursor 1-aminocyclopropane-1-carboxylic acid (ACC) supplying medium (Romera et al. 1999). The expression of the Fe-deficiency-induced genes OsIRO2, OsNAS1, and OsIRT1 is increased by 3-, 28-, and 3-folds, respectively, in rice after ACC treatment for 5 days. In addition, ethylene accumulates in pumpkin and cucumber roots after 5 and 6 days iron starvation, respectively (Romera et al. 1999; Waters and Blevins 2000).

Ethylene, auxin, and NO can interact at different levels. Auxin can enhance ethylene production by affecting ACC synthase activity and promote NO accumulation in plant roots (Morgan and Hall 1962; Abeldeng et al. 1995; Du et al. 2008). On the other hand, ethylene can affect auxin accumulation and polar auxin transport (Swarup et al. 2002). ACC can also enhance NO emission in senescing pea leaves and NO production in the subapical regions of the roots of several Strategy I plants (García et al. 2011). NO induced upregulation of the expression of many genes involved in ethylene synthesis including SAMS, ACS, and ACO in Arabidopsis and cucumber (García et al. 2011). NO also can increase auxin content by reducing auxin degradation mediated by auxin oxidase (Xu et al. 2010).

Besides auxin, ethylene, and NO, other small molecular substances, such as abscisic acid (ABA), jasmonic acid (JA), cytokinin (CK), and H2O2, also participate in iron deficiency responses to some extent. JA and CK negatively regulate Fe-deficiency-induced genes in Arabidopsis. In both Fe-sufficient and Fe-deficient conditions, AtFIT1, AtIRT1, and AtFRO2 are strongly repressed by growing in 100 μM methyl-jasmonate (MeJA) supplying medium for 6 h compared with control plants (Maurer et al. 2011). Ferric reductase activity decreases to 1/3 and the expression of AtIRT1 and AtFRO2 to 1/7 after growing in 1 μM N6-Benzyladenine (6-BA) supplying medium for 2 days compared with control plants (Séguéla et al. 2008).

ABA and H2O2 might also respond to iron deficiency. ABA content is significantly decreased by growing in iron-deficient conditions (Bacaicoa et al. 2009). However, ferric reductase activity is strongly increased after growing in 0.5 μM ABA-supplying medium for 6 h (Bacaicoa et al. 2011). H2O2 content was increased 28 % in sunflower leaves by growing in iron-deficient conditions (Ranieri et al. 2001). In tomato, ascorbate peroxidase (APX), which participates in H2O2 catabolic processes, is upregulated after 9 days growing in iron-deficient conditions (Herbik et al. 1996).

It is not clear in which tissue signal substances respond firstly during iron deficiency. Auxin and ethylene accumulate in plant roots during iron deficiency, and NO emission is also enhanced; therefore, plant roots might be the first response tissue in iron deficiency stress (Bacaicoa et al. 2009; Wu et al. 2011; Romera et al. 1999; Graziano and Lamattina 2007). Reciprocal grafting experiments with the iron-efficient species M. xiaojinensis and the iron-inefficient species M. baccata showed that iron deficiency induced upregulation of root proton extrusion, and ferric reductase activity only occurs when M. xiaojinensis was used as the stock (Wu et al. 2011).

Some researchers have indicated that signal substances are produced in leaves and then transmitted to roots to regulate downstream functional genes. For example, reciprocal grafting experiments showed that ferric reductase activity was higher when the Fe-hyperaccumulating mutant dgl was used as the shoot in pea (Grusak and Pezeshgi 1996). Split-root experiments showed that 3 days after transfer to culture medium, IRT1 and FRO2 messenger RNA (mRNA) accumulation was strongly decreased in the iron-deficient side and increased in the iron-sufficient side of the roots, which indicates signal substances derived from leaves (Vert et al. 2003).

Although previous researches show that NO, H2O2, and plant hormones are involved, to some extent, in iron deficiency stress responses, it is not clear in which order signal substances respond to iron deficiency in plants and whether all of these signal substances respond to iron starvation in M. xiaojinensis.

In this article, to obtain a general overview of Fe deficiency-induced changes in the transcriptome of M. xiaojinensis, 454 and Illumina RNA-seq were used to identify transcripts at different time points from 12 h to 9 days.

Materials and methods

Plant materials

M. xiaojinensis in vitro shoots were subcultured in MS medium (Murashige and Skoog 1962) containing 0.5 mg/l indole-3-butytric acid (IBA), 0.5 mg/l 6-benzyl aminopurine (6-BA), 3 % (w/v) sucrose, and 0.7 % (w/v) agar and propagated for 1 month. These strong and more than 4-cm-tall plants were transferred to 1/2 MS medium containing 0.5 mg/lIBA, 3 % (w/v) sucrose, and 0.7 % (w/v) agar to take root for another month. The tissue culture plantlets with roots extending nearly 2.0 cm were transplanted into a vessel containing a nutrient solution (1.9 g/l KNO3, 1.65 g/l NH4NO3, 0.37 g/l MgSO4 · 7H2O, 0.17 g/l K2HPO4, 0.0278 g/l FeSO4 · 7H2O, 0.0373 g/l Na2 ethylenediaminetetraacetic acid (EDTA) and the micronutrients HBO3 1.54 mg/l, MnCl2 · 4H2O 0.912 mg/l, ZnSO4 · 7H2O 1.12 mg/l, CuSO4 · 5H2O 0.648 mg/l, and [NH4]2MoO4 · 4H2O 0.0408 mg/l, pH 6.3) for 4 weeks. The nutrient solution was replaced every week. Finally, uniform seedlings were selected for study. The seedlings were washed with distilled water twice, dried by absorbent paper and transferred into a new nutrient solution undergoing different treatments. (Supplementary Table S11 for 454 sequencing and Supplementary Table S12 for Illumina sequencing). At different time points, the white roots and the mature leaves (fully expanded) were sampled about 500 mg. All experiments were conducted in a growth chamber at 25 ± 2 °C day/17 ± 2 °C night with a 16 h photoperiod at a light intensity of 1,500 μmol · m-2 · s-1.

cDNA library construction and 454, Illumina sequencing

Total RNA of the samples for 454 sequencing was isolated according to the instructions of the TRIzol kit (Invitrogen, Shanghai, China) and then purified to exclude the transfer RNA (tRNA) and ribosomal RNA (rRNA) and to enrich the mRNA according to the instructions of the mRNA purification kit (Promega, Shanghai, China). mRNAs were reverse-transcribed by PowerScript™ Reverse Transcriptase (Takara, Dalian, China) with SMART IV™ Oligonucleotide (Takara) and CDS III/3′ PCR Primer (Beijing Genomics Institute). Long-distance PCR for double-strand complementary DNA (cDNA) amplification was performed with LA Taq Polymerase (Takara) for 15 cycles (96 °C for 1 min, 96 °C for 30 s, and 68 °C for 6 min) according to the SMART™ cDNA Library Construction Kit (Takara) User Manual. The poly-T stretch was broken, and the cDNA was randomly sheared in the condition of 30 psi nitrogen for 1 min, then cDNAs were purified using a DNA purification kit (Qiagen, Shanghai, China) according to the manufacturer’s instructions. After verified by 1.0 % agarose gel, the target fragments (500–800 bp) were collected using a Gel Extraction Kit (Qiagen) according to the manufacturer’s instructions, emPCR and 454 sequencing were performed according to GS FLX Titanium series kits (Roche Applied Science, Shanghai China).

Total RNA of the samples for Illumina sequencing was isolated using TruSeq Stranded mRNAs Sample Preparation kit (Illumina, Shanghai, China) according to the protocol. After that, mRNAs were purified using poly-T oligo-attached magnetic beads and fragmented, the first and second cDNA were synthesized and end-repaired. Adaptors were ligated after adenylation at 3′ end. After gel purification, cDNA templates were enriched by PCR. Finally, the short cDNA fragments were prepared for Illumina sequencing on an Hiseq 2000 (Illumina, San Diego, CA). Time points selecting in Illumina sequencing was based on our previous research, MxVHA was induced significantly at 12 h after iron starvation in root which indicated iron uptake was improved (Zhang et al. 2012), activity of ferric reductase improved, and pH value decreased began at 2 days after iron starvation in root, and the two values reached to peak in day 6 after iron starvation, while in day 9, the two values were back to normal levels (Wu et al. 2010). Therefore, roots and leaves samples were harvested at 0 h, 12 h, 2 days, 6 days, and 9 days after iron starvation.

Gene annotation and functional classification

All unigenes were blasted against the NCBI nonredundant nucleotide database (NT, by Nov. 3, 2010; E value ≤1e-10), nonredundant protein database (NR, by Oct. 9, 2010; E value ≤1e-5), and SWISS-PROT (downloaded from European Bioinformatics Institute by Oct. 28, 2010; E value ≤1e-10). The BLAST result accessions against SwissProt were used to assign Gene Ontologies (GO) by WEGO.

Gene expression analysis

For gene expression analysis, the expression level of each gene in each treatment of Illumina sequencing was calculated by quantifying the number reads that mapped to each sequence using the Bowtie program. The raw gene expression counts were normalized using RPKM (reads per kilobits per million) method (Mortazavi et al. 2008). Compared with normal treatment, genes which expressed differentially in iron deficiency stress were identified by edgeR (Robinson et al. 2010) and required to have a twofold change and a p value ≤0.01. Heat maps illustrating expression patterns of various subgroups of differentially expressed genes were generated by TIBCO Spotfire software (https://tryspotfire.tibco.com/).

Quantitative PCR

Total RNA was extracted using CTAB method (Zhang et al. 2005). After DNase1 (Takara, Tokyo, Japan) treatment 30 min in 37 °C, purified RNA was identified by 1.0 % agarose gel, and the concentration was determined by using Nanodrop 2000 instrument (Thermo Scientific), the first strand cDNA for each sample was made from 1 μg RNA using M-MLV RT (Takara). Gene-specific primers were designed using Primer software (Supplementary Table S13). Samples and standards were run in triplicate on each plate using the 1 × SYBR Master Mix (Taraka) on an ABI 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA) following the manufacturer’s recommendations. qPCR was performed in a 20 μl reaction containing 6.8 μl dd H2O, 10 μl 1 × SYBR Master Mix, 0.4 μl forward primer, 0.4 μl reverse primer, 0.4 ROX Reference DyeII, and 2 μl template cDNA. The PCR condition were 30 s of pre-denaturation, 40 cycles of 5 s at 95 °C and 30 s at 60 °C. The expression was calculated by “comparative CT method” (Schmittgen and Livak 2008) using the CT value of the β-actin (Kürkcüoglu et al. 2007), the primer was listed in Supplementary Table S13. The data were processed using Microsoft Excel 2007, and the differences were evaluated using two-sample t tests assuming unequal variances. Differences were defined as significant if p value ≤0.01 and fold change ≥2, and asterisks were labelled on the top of their columns.

Results

RNA-seq

A number of 1,200,975 short reads from 454 and Illumina sequencing were assembled into 80,380 contigs with a mean length of 680 bp. After filtering out homopolymers and short reads (<150 bp), 74,839 unigenes with a mean length of 864 bp were obtained. Illumina reads from each sample were mapped to 74,839 unigenes, 42.5–75.4 % reads were mapped to 74, 839 unigenes in different root and leaf samples, respectively (Supplementary Tables S1 and S2).

Approximately 66–91 % of the unigenes were annotated by BLASTx, with a threshold E value of 10-5, from four public databases [i.e., the NCBI nonredundant (nr), Swiss-Prot protein, Kyoto Encyclopedia of Genes and Genomes (KEGG), and Clusters of Orthologous Groups of proteins (COG) databases; Table 1]. Among the 74,839 M. xiaojinensis transcripts, 70,674 were significantly matched to 22,744 different “Golden Delicious” predicted proteins (BLASTx, E value threshold of 10-5). Reciprocal BLAST searches, in which the Golden Delicious predicted sequences were queried against the M. xiaojinensis sequences (Velasco et al. 2010), revealed that 84,436 of the 95,232 proteins had significant similarity with M. xiaojinensis transcripts (tBLASTn, E value threshold of 10-5), corresponding to 20,561 different M. xiaojinensis transcripts. A comparison of the best BLAST matches from these searches identified 13,704 unambiguous orthologs between M. xiaojinensis and Golden Delicious, based on significant reciprocal best matches (Fig. 1).

Table 1 Malus xiaojinensis unigenes from 454 sequencing annotation to public databases BLASTx
Full size table
Fig. 1
figure 1

Summary of the reciprocal comparisons between Malus xiaojinensis nucleotide sequences and “Golden Delicious” peptidic sequences. A Unambiguous orthologs between M. xiaojinensis transcripts and Golden Delicious peptidic sequences in M. xiaojinensis transcriptome; a unambiguous orthologs between M. xiaojinensis transcripts and Golden Delicious peptidic sequences in Golden Delicious genome. B Transcripts of M. xiaojinensis which are significantly matched by Golden Delicious peptidic sequences; b Golden Delicious peptidic sequences which are significantly matched by M. xiaojinensis transcripts. C Transcripts in M. xiaojinensis significantly match to the Golden Delicious peptidic sequences; c peptidic sequences in Golden Delicious genome which are significantly similarity with M. xiaojinensis transcripts. D Putative M. xiaojinensis-specific genes

Full size image

Expression profile in roots and leaves

To obtain a general overview of the M. xiaojinensis transcriptome during iron starvation, mature leaves and white roots were harvested after 12 h (12 h), 2 days (2d), 6 days (6d), and 9 days (9d) of iron deficiency. There were 3,482,026–17,118,764 reads obtained from the samples, respectively, and assembled into 67,844 unigenes.

Compared with root and leaf samples at 0 h of iron deficiency, a total of 21,037 unigenes were differentially expressed (>twofold change, p value ≤0.01) in root (R) and leaf (L) samples after 12-h, 2-day, 6-day, and 9-day iron-deficient treatment (Supplementary Table S3, Supplementary Table S4).

In root samples, the number of differentially expressed genes decreased gradually; 6,947 unigenes were expressed differentially in the R12h sample, then decreased to 2,743 in the R9d sample. However, there was no such tendency in leaf samples; 5,199, 4,776, 5,826, and 4,809 unigenes were expressed differentially in the L12h, L2d, L6d, and L9d samples, respectively (Supplementary Fig. S1).

Expression profile of iron uptake and remobilization-related genes

M. xiaojinensis uses Strategy I to absorb iron. Many related genes were expressed differentially after iron starvation including iron uptake-related genes MxIRT1, MxFRO2, and MxHAs (Gao et al. 2011; Zhang et al. 2012), and iron remobilization-related genes MxNAS1, MxCS1, and MxNRAMP1 (Zhang et al. 2009; Han et al. 2013). Expression profile showed that MxHA2 and MxFRO2 were upregulated significantly, by 11.35- and 8.72-folds, in the R12h sample, while at the same time point, the iron remobilization-related genes MxNAS1, MxNRAMP1, and MxCS1 did not respond in leaf and root (Fig. 2, Supplementary Table S5). These results indicated that iron uptake in roots was already reinforced after 12 h of iron starvation, while iron remobilization-related genes had not responded.

Fig. 2
figure 2

Expression profile of genes related to iron uptake and transport. To be considered differentially expressed, the transcript must have a fold change ≥2 between iron deficiency stress and R0h sample, and a p value ≤0.01. While red cell means transcripts were induced due to iron starvation, green cell indicates transcripts were repressed after iron starvation and black cell means no significant fold change after iron starvation

Full size image

In the R6d sample, although the expression of MxFRO2 was increased 8.26-fold, MxIRT1 was repressed significantly, which indicated that the enhancement of iron uptake had disappeared (Fig. 2, Supplementary Table S5). In contrast, the iron remobilization-related genes MxNRAMP1 and MxNAS1 were upregulated by 31.41-folds and more than 100-fold, respectively, which means that iron remobilization was reinforced in root samples (Fig. 2, Supplementary Table S5). Moreover, MxNAS1 was also upregulated in the L6d sample, which showed that iron remobilization was also enhanced in leaves.

In the R9d sample, the iron remobilization-related genes MxNAS1 and MxCS1 were upregulated significantly, which indicated that iron transport in roots and iron remobilization in roots and leaves were reinforced.

In summary, iron uptake in roots was enhanced significantly from 12 h to 2 days, but expression of IRT1 was downregulated in R6d sample, while iron remobilization was reinforced from days 6 to 9 in roots and leaves under iron starvation.

Expression profile of signal substances pathway

Ethylene-mediated signaling pathway

A number of genes in the ethylene-mediated signaling pathway were induced during iron starvation in M. xiaojinensis. Compared with the R0h sample, ethylene receptor ERS1 was upregulated by 40-folds in the R12h sample (Fig. 3a, Supplementary Table S6). Moreover, the feedback inhibitor of ethylene-insensitive protein 3(EIN3) EBF1 and three downstream transcription factors, EIN3, ERF105, and RAP2.2, were also upregulated significantly. The ethylene synthesis-related gene ACO was obviously repressed (Fig. 3a).

Fig. 3
figure 3

Expression profile of genes related to plant hormone-related pathway. a Different expressed transcripts in ethylene biosynthetic process and ethylene-mediated signaling pathway. b Different expressed transcripts in auxin transport and auxin-mediated signaling pathway. c Different expressed transcripts in abscisic acid (ABA) biosynthetic process and ABA-mediated signaling pathway. d Different expressed transcripts in jasmonic acid (JA) biosynthetic, metabolic process, and JA-mediated signaling pathway. To be considered differentially expressed, the transcript must have a fold change ≥2 between iron deficiency stress and R0h sample, and a p value ≤0.01. While red cell means transcripts were induced due to iron starvation, green cell indicates transcripts were repressed after iron starvation and black cell means no significant fold change after iron starvation. Numbers from 1 to 8 in figures represent a gene name which is too long to write in the figure directly: 1 Ethylene-overproduction protein 1, 2 Protein XAP5 circadian timekeeper, 3 Protein transport inhibitor response 1, 4 Pyrophosphate-energized vacuolar membrane proton pump 1, 5 Serine/threonine-protein phosphatase 2A 65 kDa regulatory subunit A alpha isoform, 6 Type I inositol-1,4,5-trisphosphate 5-phosphatase 1, 7 4CL5-like, and 8 6,7-dimethyl-8-ribityllumazine synthase

Full size image

In the R2d sample, the expression of ethylene receptors ETR and ETR2 were increased by 12.98- and 11.93-folds, respectively, and EBF1 expression was increased 2.56-folds (Fig. 3a, Supplementary Table S6). The transcription factors ERF4, ERF073, and RAP2.7 were also upregulated, while the ethylene receptor ERS1, which was upregulated in the R12h sample, was repressed significantly (Fig. 3a).

In the R6d sample, the ethylene receptors ETR and ETR2 (which were induced in the R2d sample) were downregulated significantly, and the transcription factors ERF105 and RAP2.2 were also clearly downregulated (Fig. 3a). In the R9d sample, genes in ethylene-mediated signaling pathway did not respond significantly, while the ethylene synthesis-related gene ACO was repressed, with expression of less than 1/100 compared with the R0h sample (Fig. 3a, Supplementary Table S6).

In leaf samples, ERS1 was downregulated significantly at days 2 and 9 during iron starvation, while ERF105 was repressed after 6 days iron starvation (Fig. 3a). The expression of other genes in the ethylene signal pathway showed no significant changes. Moreover, the ethylene synthesis-related gene ACO was upregulated in four leaf samples; its expression was increased by over 100 times in the L2d sample, and ACS was upregulated in the L2d and L9d samples (Fig. 3a, Supplementary Table S6).

Taken together, genes in the ethylene-mediated signaling pathways were enhanced in R12h and R2d samples after iron starvation and the enhancement were disappeared after 6 days, but they did not respond significantly in leaf samples. The ethylene synthesis genes ACO and ACS were only upregulated in leaves and downregulated in the R12h sample.

Auxin-mediated signaling pathway

The auxin-mediated signaling pathway was also activated. In the R12h sample, IAA9 and IAA13 repressors of auxin transcription factor ARF and auxin-repressed protein (ARP) were upregulated significantly; their expression levels were increased by 17.13-, 8.83-, and 10-folds compared with the R0h sample, respectively (Fig. 3b, Supplementary Table S7). A component of the SCF E3 ubiquitin ligase complex that can activate the auxin-mediated signaling pathway SKP1 and the auxin influx transporter LAX3 were also upregulated significantly (Fig. 3b).

In the R2d sample, although the ARF repressor IAA13 was still induced, the fold changed from 8.83 to 2.64, compared with R0h sample, while another ARF repressor IAA9 was not upregulated significantly (Fig. 3b, Supplementary Table S7). Moreover, the auxin receptor TIR1, another component of the SCF E3 ubiquitin ligase complex CUL1, some transcription factor ARFs, the auxin transporter LAX1, and Auxin-induced protein were all upregulated significantly (Fig. 3b, Supplementary Table S7).

In the R6d sample, auxin responsive genes were not upregulated, and the auxin transcription factors ARF1, ARF15, and ARF9 were clearly repressed. In the R9d sample, only two genes responded significantly: ARF1 was upregulated and ARF4 was downregulated (Fig. 3b).

In the L12h sample, the auxin transcription factors ARF1 and ARF108 were repressed significantly, with relative expression values of only 0.07 and 0.01, respectively, compared with the R0h sample (Fig. 3b, Supplementary Table S7).

In the L2d sample, the ARF repressor IAA9 was downregulated significantly, while the expression of auxin transporters ABCB19 and LAX3 were increased by 35.20- and 5.27-folds, respectively, compared with the L0h sample (Fig. 3b, Supplementary Table S7). The auxin transcription factor ARF1 was upregulated, but ARF108 was downregulated (Fig. 3b).

In the R6d sample, ABCB19 and LAX3 were still upregulated, with expression of ABCB19 increased by over 100 times compared with the L0h sample (Fig. 3b, Supplementary Table S7). ARF1 and Auxin-induced protein were also induced, and this tendency continued to 9 days of iron starvation (Fig. 3b, Supplementary Table S7).

In summary, the auxin-mediated signaling pathway responded in both roots and leaves during iron deficiency stress. In root samples, although auxin transport was enhanced at 12 h after iron starvation, the ARF repressors IAA9 and IAA13 were upregulated, which indicated the auxin-signaling pathway might be repressed. The auxin-signaling pathway was activated in R2d sample, accompanied by significant upregulation of ARFs. In leaf samples, auxin transport reinforced from days 2 to 6 during iron starvation.

ABA-mediated signaling pathway

A number of ABA-responsive genes were also expressed differentially during iron starvation. In the R12h sample, genes related to the ABA-mediated signaling pathway and ABA synthesis did not respond significantly (Fig. 3c). In the R2d sample, positive regulators of the ABA-signaling pathway were upregulated significantly, including the ABA receptor PYL9, the histidine protein kinase ATHK1, Mitogen-activated protein kinase3 and Syntaxin-61 (Fig. 3c). However, the transcription factor ABI5 and the serine/threonine-protein kinase SAPK1 were downregulated (Fig. 3c).

In the R6d sample, except for PYL9 and ATHK1, which were upregulated, other genes related to abscisic acid signaling and synthesis were repressed significantly, including the ABA synthesis-related genes AAO and ZEP, the ABA degradation-related gene Abscisic acid 8′-hydroxylase, the ABA receptor PYL4, the transcription factors ABI7, MYB44, and ARAC7, and the negatively regulated ABA signal pathway gene E3 ubiquitin-protein ligase KEG (Fig. 3c).

In the R9d sample, Abscisic acid 8′-hydroxylase encoding gene was upregulated, with an expression value over 100 times greater than in the R0h sample (Fig. 3c, Supplementary Table S8). Although PYL9 was upregulated, its expression was increased by only 3.6-fold, which was much less than in the R6d sample (Fig. 3c, Supplementary Table S8). ATHB-7, which can be upregulated by exogenous ABA, was repressed significantly.

In the L12h sample, genes related to the ABA-mediated signaling pathway were not upregulated significantly (Fig. 3c). The ABA receptor PYL9, ABA transcription factors ABI5, ABI7, and MYB44, and the serine/threonine-protein kinase SAPK1 were all downregulated (Fig. 3c).

The ABA synthesis-related gene ZEP and the ABA transcription factor ABI7 were repressed significantly from days 2 to 9 of iron starvation in leaf samples, while the ABA receptor PYL9 was induced during this period (Fig. 3c).

To conclude, although the ABA-mediated signaling pathway positively regulated genes, ATHK1, MAPK3, and PYL9 were upregulated in the R2d and R6d samples, ABA synthesis-related genes and ABA transcription factors were not, which indicated that the ABA-mediated signaling pathway was not activated after iron starvation. In leaf samples, only the ABA receptors PYL9 and SAPK3 were upregulated from days 2 to 9 during iron starvation, which showed that the ABA-signaling pathway was not activated in leaves.

JA-mediated signaling pathway

In root samples, the jasmonic acid synthesis-related genes LOX2- and 4CL5-like were downregulated significantly in the R12h and R2d samples, while the JA degradation-related gene Jasmonate O-methyltransferase (JMT) was upregulated. In the R6d sample, LOX2 was still repressed, and JA-mediated signaling pathway-related genes were also downregulated (Fig. 3d). In the R9d sample, no genes were expressed differentially except JMT, which was induced upregulation. In leaf samples, JA synthesis and signaling pathway-related genes were downregulated at different times after iron starvation (Fig. 3d, Supplementary Table S9).

Therefore, JA synthesis was inhibited and degradation was enhanced in both root and leaf samples, while the JA-mediated signaling pathway might be repressed in root samples during iron starvation.

ROS genes

Reactive oxygen scavenger (ROS) scavenging protein-encoding genes were also expressed differentially in iron deficiency conditions. In the R12h sample, 24 genes from hydrogen peroxide catabolic process were expressed significantly differentially; 20 unigenes were upregulated including two APX2s, 14 CATs, and four POXs, while the other four genes were downregulated including APX3, APX5, CAT1, and POX10. Based on GO cellular component classification, 16 of the upregulated genes were located in the cytoplasm and the extracellular region; among these, eight were induced significantly. In addition, 9 of 19 genes located in the peroxisome were induced upregulation (Fig. 4).

Fig. 4
figure 4

Significantly different expressed genes in hydrogen peroxide catabolic process. To be considered differentially expressed, the transcript must have a fold change ≥2 between iron deficiency stress and R0h sample, and a p value ≤0.01. While red cell means transcripts were induced due to iron starvation, green cell indicates transcripts were repressed after iron starvation and black cell means no significant fold change after iron starvation

Full size image

In the R2d sample, only four unigenes from hydrogen peroxide catabolic process were upregulated including CAT3, POX12, POX42, and POX64 (Fig. 4). There were 11 unigenes downregulated which comprised of two cationic POX2s, seven CATs and two POXs (Fig. 4). There were also five unigenes expressed differentially in the extracellular region; cationic POX2 and CAT2 were repressed while only POX42 was induced. POX10 located in the cytoplasm and POX15 located in the membrane were also repressed (Fig. 4).

In the R6d sample, among 15 differentially expressed genes from hydrogen peroxide catabolic process, only five were upregulated including cationic POX2, POX42, POX64, and POX73; the expression of POX2 and POX73 were increased by over 100 times compared with the R0h sample. In addition, six CATs were all significantly induced downregulation (Fig. 4, Supplementary Table S10).

In the R9d sample, only nine unigenes from hydrogen peroxide catabolic process were differentially expressed, among which CAT1 and APX5 were repressed significantly, while the other seven were upregulated strongly (Fig. 4).

In the L12h sample, among 19 differentially expressed genes from hydrogen peroxide catabolic process, only three were upregulated including APX5, POX12, and CAT1; the other 16 were repressed significantly including 15 CATs and an APX (Fig. 4). Unigenes located in the cytoplasm, extracellular region, chloroplast envelope, and glyoxysome were repressed or did not respond (Fig. 4, Supplementary Table S10).

As in the L12h sample, there were also 19 unigenes from hydrogen peroxide catabolic process differentially expressed in the L2d sample, three of which were induced including APX3/5 and CAT1 (Fig. 4). The other 16 unigenes comprising ten CATs, two APX2s, one APX3, and three POXs were downregulated. Unigenes located in the cytoplasm, extracellular region, chloroplast envelope, and glyoxysome were downregulated or did not respond (Fig. 4, Supplementary Table S10).

In the L6d sample, only eight unigenes from hydrogen peroxide catabolic process expressed differentially, six of which were induced while two were repressed significantly (Fig. 4).

In the L9d sample, 20 unigenes from hydrogen peroxide catabolic process were downregulated significantly, including two CAT1s, five CAT2s, three CAT3s, five other CATs, and five encoding APX2/3 and POX4/42 (Fig. 4, Supplementary Table S10).

Taken together, the genes upregulated in the root samples were located in the cytoplasm, extracellular region, and peroxisome, which indicated that ROS scavenging had been activated. However, this activation disappeared in the R2d and R6d samples. While in the R9d sample, ROS scavenging was enhanced in the extracellular region and membrane, in leaf samples, ROS scavenging was only activated on day 6 of the iron starvation treatment.

qPCR verification

There were 11 unigenes chosen for quantitative polymerase chain reaction (qPCR) from the ethylene, auxin, ABA and JA synthesis, metabolism, and signaling pathways, and hydrogen peroxide catabolism. Their names and GO annotations were contig1137 (ERS1), contig247 (ERF105), contig2353 (IAA9), contig1260 (ARAC7), H_rep_c22375 (CAT1), H_rep_c2053 (TIR1), H_rep_c15522 (LOX2), 1254557 (ARF1), contig15484 (PYL9), contig16283 (ERF4), and contig12462 (Protein TOPLESS). Of all 88 relative expression time-points (11 genes × 8 treatments), only seven differed from the RNA-seq data (Fig. 5).

Fig. 5
figure 5

Quantitative real time PCR confirmation of 11 significantly expressed genes in different signaling pathway. Histograms represent transcript abundance obtained for Contig1137, Contig247, Contig2353, Contig1260, H_rep_c22375, H_rep_c2053, H_rep_c15522, Contig12462, Contig15484, Contig16283, and 1254557. Asterisks were labeled on the top of the columns which indicate that the gene’s expression had significant changes compared to R0 or L0 sample (fold change ≥2 and p value ≤0.01). Error bars on each column indicates SDs from three replicates

Full size image

Discussion

Iron uptake was enhanced in roots in the earlier period of iron starvation, while iron remobilization was reinforced in the later period. The iron uptake-related genes FRO2, HA2, and IRT1 were upregulated at 12 h and 2 days of iron starvation. Although FRO2 was still upregulated in the R6d and R9d samples, HA2 and IRT1 were not, and IRT1 was even downregulated in the R6d sample. The iron remobilization-related genes NRAMP1 and NAS1 were upregulated in R6d, and NAS1 was also upregulated in R9d sample.

In tomato, LeIRT1 was upregulated after 6 h of iron starvation, and its expression reached a peak after 12 and 24 h, before decreasing a little at 48 h (Wang et al. 2002). FIT and IRT accumulated in Arabidopsis roots after 12 h of iron starvation and continued to accumulate to 5 days (Sivitz et al. 2011). These results indicate that Strategy I plants improve iron uptake in roots in the earlier period, while iron remobilization is enhanced in the later period. In Arabidopsis root samples, AtNRAMP1 was upregulated on the 3rd day of iron deficiency treatment and its expression was further increased on the 5th day (Curie et al. 2000). In peanut, NRAMP1 was upregulated in the roots after 4 days of iron deficiency, and the changes continued to 11 days (Xiong et al. 2012). Our previous studies showed that MbNRAMP1 was upregulated on the 5th day of iron deficiency (Xiao et al. 2008). In M. xiaojinensis, another iron remobilization-related gene NAS1 was also upregulated on the 5th day of iron deficiency and its expression was still elevated on the 15th day. Here, iron uptake was enhanced after 12 h and remained high after 2 days of iron deficiency in M. xiaojinensis roots, while iron remobilization was reinforced after 2 days.

Based on our expression profiling, the ethylene and ROS signal pathways might be more important in the low iron stress response than the other signal pathways. The ethylene-mediated signaling pathway was activated after 12 h of iron deficiency in roots, which was earlier than the other signal pathways. Ethylene accumulated in the roots of cucumber, pea, tomato, and pumpkin after iron deficiency, and in cucumber, the ethylene content increased in the first day of iron deficiency (Romera and Alcántara 2004; Bacaicoa et al. 2009). Moreover, the increased ethylene production by Fe-deficient roots paralleled the enhancement of ferric reductase activity and root acidification process (Romera et al. 1999; Waters and Blevins 2000). While in our results, ethylene synthesis genes ACO and ACS were not upregulated in roots might be because they did not expressed in roots or they were not induced in iron deficiency stress. In Arabidopsis, only four genes for ACC synthase AtACS4, AtACS6, AtACS9, and AtACS11; and three for ACC oxidase, AtACO1, AtACO2, and AtACOS3, were examined in roots. And among all the genes, the expression of AtACS11 and AtACO3 were not increased. The other eight AtACSs and AtACO4 were not examined in roots (García et al. 2010). Therefore, the downregulated ACO in root samples after iron deficiency might be homologous of AtACO3 or AtACO4, no significant expression changes ACS in root samples after iron deficiency might be homologous of AtACS11. This evidence shows that in other plants, ethylene is also accumulated in roots after iron deficiency and is related to the improvement of iron uptake.

ROS scavenging was improved significantly after 12 h of iron starvation in M. xiaojinensis root and was activated on days 2 and 6, which indicated that the ROS signal pathway was activated at 12 h of iron deficiency and repressed on the 2nd day. In Arabidopsis, POX was also upregulated after 24 h iron deficiency and repressed after 3 days (Thimm et al. 2001).

The activation of the IAA-mediated signaling pathway occurred later than the enhancement of iron uptake. Expression profiling showed that iron uptake was upregulated strongly after 12 h of iron deficiency, at which point the ARF repressors IAA9 and IAA13 were upregulated, indicating the IAA signal pathway was suppressed. Many ARFs were upregulated after 2 days of iron deficiency, suggesting that the IAA-mediated signaling pathway was then activated.

In tomato, IAA accumulated significantly on the 6th day of iron starvation (Jin et al. 2011), but LeIRT1 was upregulated in the 6th hour after iron starvation (Wang et al. 2002). Our previous research showed that IAA was accumulated in the M. xiaojinensis root after 3 days of iron deficiency, but did not accumulate before the second day (Wu et al. 2011).

On the other hand, NAA and IAA could enhance ferric reductase activity and iron uptake gene expression in iron-sufficient conditions in cucumber and Arabidopsis, and the IAA transport inhibitors NPA and TIBA repressed the enhancement of ferric reductase activity and iron uptake gene expression (Chen et al. 2010; Bacaicoa et al. 2011). In M. xiaojinensis, IAA also induced MxFRO2 in iron-sufficient roots, while NPA repressed the upregulation of MxFRO2 on the 3rd and 6th day of iron starvation (Wu et al. 2011).

Therefore, IAA was unrelated to the improvement of iron uptake after 12 h of iron starvation, but might be related to iron uptake enhancement at 2 days.

Conclusions

Based on expression profiling by RNA-seq, we found many transcripts that were expressed differentially in M. xiaojinensis during iron starvation including iron uptake and remobilization genes and many plant hormone pathway genes. The enhancement of iron uptake occurred earlier than the improvement of iron remobilization in roots. The expression profiles of plant hormone pathways showed that ethylene and ROS signaling responded earlier than the IAA, ABA, and JA signal pathways.