ABI4 downregulates expression of the sodium transporter HKT1;1 in Arabidopsis roots and affects salt tolerance


  • Doron Shkolnik-Inbar,

    1. Department of Life Sciences and Doris, Bertie Center for Bioenergetics in Life Sciences, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Search for more papers by this author
  • Guy Adler,

    1. Department of Life Sciences and Doris, Bertie Center for Bioenergetics in Life Sciences, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Search for more papers by this author
  • Dudy Bar-Zvi

    Corresponding author
    • Department of Life Sciences and Doris, Bertie Center for Bioenergetics in Life Sciences, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Search for more papers by this author

(e-mail barzvi@bgu.ac.il).


A plant's ability to cope with salt stress is highly correlated with their ability to reduce the accumulation of sodium ions in the shoot. Arabidopsis mutants affected in the ABSCISIC ACID INSENSITIVE (ABI) 4 gene display increased salt tolerance, whereas ABI4-overexpressors are hypersensitive to salinity from seed germination to late vegetative developmental stages. In this study we demonstrate that abi4 mutant plants accumulate lower levels of sodium ions and higher levels of proline than wild-type plants following salt stress. We show higher HKT1;1 expression in abi4 mutant plants and lower levels of expression in ABI4-overexpressing plants, resulting in reduced accumulation of sodium ions in the shoot of abi4 mutants. HKT1;1 encodes a sodium transporter which is known to unload sodium ions from the root xylem stream into the xylem parenchyma stele cells. We have shown recently that ABI4 is expressed in the root stele at various developmental stages and that it plays a key role in determining root architecture. Thus ABI4 and HKT1;1 are expressed in the same cells, which suggests the possibility of direct binding of ABI4 to the HKT1;1 promoter. In planta chromatin immunoprecipitation and in vitro electrophoresis mobility shift assays demonstrated that ABI4 binds two highly related sites within the HKT1;1 promoter. These sites, GC(C/G)GCTT(T), termed ABI4-binding element (ABE), have also been identified in other ABI4-repressed genes. We therefore suggest that ABI4 is a major modulator of root development and function.


The plant transcription factor ABA-INSENSITIVE 4 (ABI4) was first isolated in mutant plants that displayed insensitivity to ABA during germination (Finkelstein, 1994). Mutations in the ABI4 gene were shown to promote seed germination in the presence of inhibitory concentrations of sugars and salt (Arenas-Huertero et al., 2000; Huijser et al., 2000; Laby et al., 2000; Quesada et al., 2000). ABI4 was shown to be involved in the regulation of photosynthesis-associated nuclear-encoded genes (Acevedo-Hernandez et al., 2005), ascorbate-dependent growth regulation (Kerchev et al., 2011) and lipid metabolism regulation (Yang et al., 2011), and to function in retrograde signaling from both chloroplasts and mitochondria to nuclei (Koussevitzky et al., 2007; Giraud et al., 2009). Early studies have focused on the role of ABI4 in seed maturation, germination and early seedling growth, developmental stages at which its transcript levels are highest (Soderman et al., 2000; Shkolnik-Inbar and Bar-Zvi, 2011). Other studies have shown that ABI4 plays a role at later developmental stages, when it is involved in root branching (Signora et al., 2001; Shkolnik-Inbar and Bar-Zvi, 2010, 2011), salt tolerance (Quesada et al., 2000), and pathogen resistance (Ton and Mauch-Mani, 2004; Adie et al., 2007; Kaliff et al., 2007), in part due to regulation of jasmonate signaling (Kerchev et al., 2011). Interestingly, ABI4 protein was found to be expressed mainly in the roots and at much lower levels in germinating seedlings, a finding that suggested that there is no correlation between the steady-state levels of ABI4 transcript and the protein it encodes (Finkelstein et al., 2011).

Expression of ABI4 is induced by a variety of factors, which includes high sugar concentrations (Arroyo et al., 2003; Price et al., 2003; Shkolnik and Bar-Zvi, 2008; Bossi et al., 2009). ABI4 expression is also modulated by phytohormones: it has been shown to be induced by ABA treatment of imbibed seeds but not of germinating seedlings (Price et al., 2003; Munoz-Bertomeu et al., 2010). We recently showed that ABI4 expression in roots is induced by ABA and cytokinin and repressed by auxin (Shkolnik-Inbar and Bar-Zvi, 2010).

A large number of genes whose expression is affected by ABI4 have been identified by comparing gene expression in seeds and seedlings of wild-type (WT) plants with that of abi4 mutants or ABI4 overexpressors (OE) (Soderman et al., 2000; Nakabayashi et al., 2005; Penfield et al., 2006; Kerchev et al., 2011; Reeves et al., 2011). ABI4 appears to bind a number of DNA sequences. The sequence of Coupling Element 1 (CE1) for example, CACCG, was determined in vitro using polymerase chain reaction (PCR)-based selection of the DNA sequence bound to maize ABI4 (Niu et al., 2002). Ectopically expressed tomato ABSCISIC ACID RIPENING 1 (ASR1) in Arabidopsis was shown to bind the CE1 element and to outcompete ABI4 binding to its own promoter, resulting in an abi4-like phenotype (Shkolnik and Bar-Zvi, 2008). ABI4 was also shown to bind S-box and S-box similar sequences that play role in the expression of sugar-inducible genes (Acevedo-Hernandez et al., 2005; Bossi et al., 2009). These motifs are missing in the promoters of a large number of ABI4-regulated genes, suggesting that ABI4 binds additional motifs (Koussevitzky et al., 2007; Reeves et al., 2011); therefore its DNA binding might be less stringent, or alternatively it might function by interaction with other transcription factors. Many of the ABI4 binding sites are tightly associated with the classic G-box motif (Shen and Ho, 1997; Reeves et al., 2011), involved in responses to ABA, light and pathogens (Jakoby et al., 2002). Double-mutant analyses demonstrated synergy between the action of ABI4 and several bZIP ABA response factors such as ABI5 (Reeves et al., 2011).

In this study, we further explored the role of ABI4 in the plant's response to salinity. The salobreño5 (sañ5) mutant, isolated by screening for germination of Arabidopsis in the presence of salt, is mutated in the ABI4 gene (Quesada et al., 2000). On the other hand, vegetative growth of sañ5 mutants has been reported to be similar to that of WT plants (Quesada et al., 2000). Sodium transporters play a central role in the plant's response to salinity. These include a number of H+/Na+ antiporters (NHXs and SOS1) and the sodium-specific transporter HKT1;1 which plays a major role in maintaining cytosolic sodium levels and thus salt tolerance in Arabidopsis (for a recent review see Conde et al., 2011). HKT1;1 retrieves sodium ions from the root xylem, in the roots thus limiting the accumulation of sodium in the shoots (Mäser et al., 2002; Moller et al., 2009). Arabidopis hkt1;1 mutants are hypersensitive to salt (Mäser et al., 2002). Accordingly, Arabidopsis plants overexpressing HKT1;1 in the root stele display enhanced salt tolerance (Moller et al., 2009).

In this study, using three previously isolated abi4 mutant alleles (abi4-1, abi4-102 and abi4-103) and plants that inducibly overexpress ABI4 (ABI4-OE2 and ABI4-OE4), we found that the abi4 mutants display increased salt tolerance at all developmental stages, whereas ABI4 overexpression increases plant sensitivity to salt treatments. Using quantitative reverse transcription polymerase chain reaction (qRT-PCR) and plants expressing ABI4::GUS, we showed that ABI4 is induced by salt. Upon exposure to NaCl, abi4 mutants accumulated less Na+ and more proline in the shoots than WT plants. Concomitantly, expression of the Na+ transporter HKT1;1 was enhanced in abi4 mutants and reduced in ABI4-OE plants. Interaction between ABI4 and HKT1;1 promoter sequences was demonstrated in planta by chromatin immunoprecipitation (ChIP) and in vitro by electrophoresis mobility shift assay (EMSA). Two copies of an ABI4-binding site, termed ABI4 binding element (ABE), were identified in the HKT1;1 promoter, by ABI4 binding to truncated and mutated DNA sequences. We suggest that ABI4 represses HKT1;1 expression in the root stele. The increased salt tolerance of abi4 mutants thus results from increased HKT1;1 activity, leading to increased Na+ unloading from the xylem vessels to the xylem parenchyma cells, thus reducing shoot accumulation of Na+.


Screening Arabidopsis ethane methylsulfonate (EMS)-mutagenized seedlings for germination in the presence of NaCl resulted in isolation of salt-tolerant (SALOBREÑO, SAÑ) mutants (Quesada et al., 2000), one of which, sañ5, was shown to be allelic to abi4, and was termed abi4-2. We tested the effects of NaCl on radicle emergence and cotyledon opening in plants carrying three other mutated alleles of ABI4: the abi4-1 mutant conferring a deletion in codon 157 (Finkelstein et al., 1998) and abi4-102 and abi4-103 which are nonsense mutants at codons 80 and 39, respectively (Laby et al., 2000). These mutants displayed reduced sensitivity to salt inhibition of seed germination and cotyledon opening compared with WT plants (Figure 1).

Figure 1.

Germination and cotyledon opening of abi4 mutant seedlings in the presence of NaCl. Wild-type (WT) (white bars), abi4-1 (gray), abi4-102 (hatched) and abi4-103 (black) seeds were plated on solid medium that contained 0.5 × Murashige and Skoog (MS) medium, 0.5% sucrose, 0.7% agar and 0.2 m NaCl. Radicle emergence (a) and cotyledon opening (b) were scored daily. Parallel experiments performed on germination medium without NaCl yielded 97–100% radicle emergence and cotyledon opening on days 2 and 3, respectively. Data represent mean ± SE of three independent experiments, each containing 50 plants per treatment.

We also challenged 1-week-old seedlings of abi4-mutant plants with salt treatment. We used the agar-block transfer method (Shkolnik-Inbar and Bar-Zvi, 2012), which provides a more physiologically relevant salt treatment since the aerial organs never come into contact with the salt-containing medium. All tested abi4 alleles led to improved survival compared to the WT when the seedlings were placed on top of agar-solidified medium containing 0.3 m NaCl (Figure 2a–c). As both agar layers were similar in volume, the final NaCl concentration after equilibrium between the two layers was 0.15 m (Shkolnik-Inbar and Bar-Zvi, 2012). To determine whether overexpression of ABI4 affects seedling survival after transfer to salt conditions, we used inducible expression of ABI4, since constitutive expression of this gene is lethal (Shkolnik-Inbar and Bar-Zvi, 2010). In this experiment, shown in Figure 2(d–f), we applied a more moderate stress for a shorter duration than that applied in the parallel experiment with the abi4 mutants (Figure 2a–c). Overexpression of ABI4 induced by dexamethasone (DEX) treatment of ABI4-OE seedlings resulted in increased salt sensitivity. DEX treatment did not have any effect on the salt sensitivity of WT plants or plants transformed with empty vector (Figure 2d–f).

Figure 2.

Effect of NaCl on abi4 mutant and ABI4-overexpressor (OE) seedlings. Seedlings of the indicated plants [wild-type (WT) and abi4 mutants (a–c); WT, ABI4-OE or plants transformed with empty vector (VEC) (d–f)] were grown for 7 days on medium that contained 0.5 × Murashige and Skoog (MS) medium and 0.7% agar. Slices corresponding to one-quarter of the plate area were transferred to the top of a second medium that contained 0.5 × MS, 0.5% agar without (a, c, f white bars) or with 0.3 m NaCl (b, c gray bars), 4 μm dexamethasone (DEX) (d, f dashed bars), 0.25 m NaCl (gray bars in f), or 4  μm DEX + 0.25 m NaCl (e, black bars in f). Seedling survival was scored 10 days (a–c) or 5 days (d–f) after transfer. Pictures of representative plates are shown (a, b, d, e), and quantitative data (c, f) representing mean ± standard error (SE) of three biological repeats, each consisting of 50 seedlings of each line.

To determine whether steady-state levels of ABI4 transcript are affected by NaCl, we used transgenic plants expressing ABI4::GUS constructs and WT plants. Histochemical staining of ABI4::GUS plants showed that ABI4 promoter activity is enhanced by NaCl treatment (Figure 3a–d). Detailed analysis showed that ABI4 expression is enhanced in mature regions of the root but not in developing lateral roots (Figure 3a,c). Promoter activity was also observed in the root meristem region of NaCl-treated plants (Figure 3d) but not in that of plants grown under control conditions (Figure 3b). Transverse sections of the upper root region showed that phloem and parenchyma cells express ABI4 under both conditions (Figure 3e,f). Quantitative RT-PCR analysis of ABI4 mRNA levels in roots of WT plants confirmed the salt-induction of ABI4 (Figure 3g).

Figure 3.

ABI4 expression is induced by salinity. Arabidopsis plants expressing the ABI4::GUS construct were treated with 0 (control) or 0.3 m NaCl (salt treated). (a–f) GUS-stained roots were imaged using a light microscope. Cell types shown in (e) and (f): Ep, epidermis; C, cortex; En, endodermis; P, pericycle; x, xylem; *, parenchyma; •, metaphloem and protophloem; +, companion cells. (g) RNA was isolated from root tissue and used for quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis of ABI4 transcripts. Data represent mean ± SE of three biological repeats with 50 seedlings each.

Salt tolerance of abi4 mutants was also observed in pot-grown plants (Figure 4). All three studied abi4 alleles led to plant survival under extended salt treatment which was fatal to WT plants (Figure 4e–h). Rosette leaves of pot-grown plants exposed to more moderate salt stress than that used in the survival test (Figure 4) were analyzed for ion content. Under non-stress growth conditions, the levels of Na+, K+ and Ca2+ were similar in leaves of abi4-mutant and WT pot-grown plants (Figure 5a–c). Upon irrigation with NaCl solution, the abi4 mutant accumulated lower levels of Na+ than the WT plants (Figure 5a). The levels of K+ did not differ between the studied genotypes (Figure 5b). Ca2+ levels in the leaves of abi4 mutants were consistently but statistically non-significantly higher than in WT plants (Figure 5c). Hydroponically grown abi4 mutant plants also accumulated less Na+ in the shoots than WT plants (Figure 5d). Na+ levels in the roots, on the other hand, did not differ between abi4 mutant and WT plants (Figure 5e). These results suggest that abi4 selectively affects Na+ content in the leaves.

Figure 4.

Effect of NaCl treatment on pot-grown abi4 mutants. Wild-type (WT) (a, e), abi4-1 (b, f), abi4-102 (c, g) or abi4-103 (d, h) plants were grown in pots under optimal conditions for 1 month. Plants were irrigated with water (a–d) or 0.2 M NaCl solution containing a 15:1 molar ratio of CaCl2 (e–h). Plants were photographed 3 weeks later. Experiments were repeated three times using over 20 plants per treatment.

Figure 5.

Ion accumulation in salt-treated abi4 mutant plants. (a–c), One-month-old pot-grown wild-type (WT) and abi4 mutant plants were irrigated with the indicated NaCl solutions. Rosette leaves were harvested 48 h later and assayed for sodium (a), potassium (b) and calcium (c). (d and e), One-month-old hydroponically grown plants were transferred to fresh aerated mineral growth solution containing 0 or 100 mm NaCl. Plants were harvested 8 h later, and sodium content was determined in shoots (d) and roots (e). Data represent mean ± SE of three biological repeats with 50 seedlings each. WT, white bars; abi4-1, gray bars; abi4-102 hatched bars; abi4-103, black bars.

Proline is known to be accumulated in tissues of salt-stressed plants (Szabados and Savoure, 2010). Figure 6(a) shows that although under non-stress conditions, proline levels of abi4 mutants were equivalent to those of WT plants, the mutant leaves accumulated higher levels of proline than WT plants upon exposure to salt stress. These levels were proportional to the extent of the stress. The steady-state transcript levels of genes that encode enzymes that catalyze proline biosynthesis and catabolism were compared between WT and abi4 mutant plants (Figure 6b,c). The steady-state levels of P5CS1 and PDH2, encoding Δ1-pyrroline-5-carboxylate synthase 1 and proline dehydrogenase 1, respectively, were twice as high in shoots of abi4 mutants than in WT shoots, whereas the expression of P5CS2, P5CR and PDH1 was not altered in the abi4 mutants (Figure 6b). In contrast, only PDH2 was induced in roots of abi4 mutants (Figure 6c).

Figure 6.

Proline accumulation in pot-grown salt-treated abi4 mutant plants and expression of genes encoding enzymes catalyzing proline biosynthesis or catabolism. (a) One-month-old pot-grown wild-type (WT) and abi4 mutant plants were irrigated with solutions containing 0 (white bars), 50 (gray bars) or 100 (black bars) mM NaCl. Rosette leaves were harvested 48 h later and assayed for proline. Data represent mean ± SE of three biological repeats with 50 seedlings each. (b, c) RNA was isolated from shoots and roots of plate-grown WT and abi4 mutant plants. Expression of the following genes was assayed by quantitative reverse transcription polymerase chain reaction (qRT-PCR): P5CS1, Δ1-pyrroline-5-carboxylate synthase 1; P5CS2, Δ1-pyrroline-5-carboxylate synthase 2; P5CR, pyrroline-5-carboxylate (P5C) reductase; PDH1, proline dehydrogenase 1; PDH2, proline dehydrogenase 2.

ABI4 is expressed in companion cells and parenchyma of the mature root stele (Shkolnik-Inbar and Bar-Zvi, 2010). Similar tissue-specific expression has been reported for the Na+ transporter HKT1;1 (Mäser et al., 2002; Moller et al., 2009). Moreover, hkt1;1 mutants have been shown to be hypersensitive to salt, whereas overexpression of HKT1;1 in the root stele increases the salt tolerance of Arabidopsis plants, and reduces Na+ translocation to the leaves. We therefore used qRT-PCR analysis to determine whether HKT1;1 expression is affected by ABI4. Steady-state levels of HKT1;1 were higher in roots of abi4 mutant plants than in WT roots (Figure 7a). Accordingly, transcript levels of HKT1;1 were markedly reduced in ABI4-OE plants (Figure 7b), supporting ABI4 repression of HKT1;1 expression. Type-B cytokinin response regulators ARR1 and ARR12 regulate the expression of Arabidopsis HKT1;1 and the accumulation of Na+ (Mason et al., 2010). We thus looked at the expression of ABI4 in arr1-3 and arr1-3/arr12-1 mutants, and of ARR1 and ARR12 genes in abi4 mutants (Figure 8). Expression of ARR1 and ARR12 was not significantly altered in roots of abi4 mutants (Figure 8a,b), and ABI4 expression was enhanced by salt treatment to similar extents in arr1-3 and arr1-3/arr12-1 mutants (Figure 8c). Furthermore, we could not detect any interactions in planta between ABI4 and ARR1 or ARR12 using the split-YFP (yellow fluorescent protein) approach (Bracha-Drori et al., 2004) (Figure S1).

Figure 7.

HKT1;1 expression is regulated by ABI4. (a) Wild-type (WT) plants and the indicated abi4 mutants were grown for 7 days on solid medium (0.5 × Murashige and Skoog (MS) medium, 0.5% sucrose and 0.5% agar). Total root RNA was extracted and subjected to quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis for HKT1;1 transcript. Data represent average values and standard error (SE) of triplicates performed in three biological repeats. (b) WT and ABI4-overexpressors (OE) were grown for 7 days on solid medium as in (a). Seedlings were transferred to fresh plates containing the same medium without (white bars) or with (gray bars) 4 μm dexamethasone (DEX). Roots were harvested 16 h later, total RNA was extracted and HKT1;1 was analyzed as in (a). Data represent mean ± SE of three independent experiments, each containing 50 plants per treatment.

Figure 8.

Expression of ARR1, ARR12 and ABI4 in roots of seedlings of the genetic background of the reciprocal mutant. Transcript levels of ARR1 (a), ARR12 (b) and ABI4 (c) were determined in roots of 10-day-old wild-type (WT) seedlings and the indicated abi4 and arr mutants, treated for 2 h with growth medium supplemented with 0 (gray bars) or 0.2 (black bars) M NaCl. Data represent mean ± SE of three independent experiments, each containing 50 plants per treatment.

To determine whether ABI4 interacts directly with the HKT1;1 promoter or represses its expression indirectly, we constructed Arabidopsis plants that overexpressed HA-FLAG-ABI4 fusion (TAP-ABI4) protein. Expression of TAP-ABI4 had minor effects on the morphology and growth parameters of the plants. Overexpression was confirmed by qRT-PCR (Figure 9a) and western blot (Figure 9b) analyses. ChIP analysis confirmed that the TAP-ABI4 protein binds the HKT1;1 promoter in vivo (Figure 9c). Controls suggested that TAP-ABI4's binding selectivity resembles that of non-tagged ABI4, as the tagged protein bound the ABI4 promoter but not the HsfA9 promoter (Figure 9c); (Shkolnik and Bar-Zvi, 2008).

Figure 9.

ABI4 binds the HKT1;1 promoter in planta. (a, b) Overexpression of TAP-ABI4 in Arabidopsis. Total RNA (a) or protein (b) extracted from 1-week-old seedlings of wild-type (WT) and two TAP-ABI4-overexpressing lines (TOE1 and TOE2). (a) Transcript levels were assayed by quantitative reverse transcription polymerase chain reaction (qRT-PCR). (b) Protein levels were assayed by western blot using anti-FLAG primary antibody. (c) Chromatin immunoprecipitation: sheared cross-linked chromatin was prepared from seedlings overexpressing TAP-ABI4. I, input sample; M, mock, without primary antibody; IP, immunoprecipitated with anti-FLAG primary antibody. The respective promoter sequences were analyzed in the obtained samples by PCR using the promoter-specific primers listed in Table S4.

Expression of Arabidopsis HKT1;1 is affected by DNA sequences that are approximately 5 kbp and 4 kbp upstream of the translation start codon (Rus et al., 2006; Baek et al., 2011). However, analysis of the nucleotide sequence of HKT1;1 promoter for cis-acting elements which have the potential to be bound by ABI4 suggested that such sequences are found in the proximal promoter sequence: putative DRE and ABRE core sequences at nt −13 to −18 and −71 to −76 respectively, upstream of the translation start codon. In addition, we identified three imperfect sequences resembling that of CE1 (nt −279 to −283 and −596 to −600) and S-box (nt −898 to −903). We used EMSA with purified recombinant ABI4 protein and probes based on different proximal HKT1;1 promoter sequences to detect the ABI4 DNA-binding site(s). The results are summarized in Figure 10. An initial screen showed that ABI4 does not bind any sequences in the −1 to −600 region (Figure 10a, probes A1 to A4), thus ABI4 does not bind promoter sequences comprising either ABRE-like or CE1-like sequences (Figure S2). Binding was detected using probes A5 and A6 representing upstream sequences. Binding of the HKT1;1 A6 promoter fragment was specific since it could be competed off with unlabeled DNA sequence of the ABI4 promoter containing CE1, but not by the ABI4 3′ untranslated region (Figure 10b). Furthermore, ABI4 appears to bind the A6 fragment at lower affinity than CE1 (Figure 10c). Detailed analyses of the A6 promoter region, using sub-fragments as probes, showed that two fragments, A6-1 and A6-5, are bound by ABI4 (Figure 10a,f, lanes 1 and 3), whereas other fragments are not (Figure 10a). The S-box-like sequence was in the A6-2 fragment that did not interact with ABI4. A comparison of the nucleotide sequences of the A6-1 and A6-5 promoter fragments showed some homology, which appears to be in two separate blocks, GCCGCTTT and TAACC in A6-1, along the center of these 50-bp fragments (Figure 10d). We used a number of mutated A6-1 50-bp-long probes to further characterize the ABI4 DNA-binding site. The cores of the sequences are presented in Figure 10(e). Mutating the second block (mutant M1, Figure 10e) did not affect ABI4 binding (Figure 10f, lane 5), suggesting that this sequence is not part of the ABI4 binding site. On the other hand, different combinations of mutations within the GCCGCTTT sequence affected ABI4 binding (Figure 10e,f). Mutating the first seven nucleotides abolished binding (mutants M2 to M7, Figure 10f, lanes 7–18), whereas mutating the last nucleotide (T) resulted in reduced binding of ABI4 (mutant M8, Figure 10f, lane 15). We searched promoter sequences of Arabidopsis genes whose expression in the root is affected by salt (Dinneny et al., 2008) for a truncated ABE motif [GC(C/G)GCTT]. Approximately 200 genes were identified (Table S1). Gene ontology analysis of the biological processes in which these root-expressed genes are involved is shown in Figure S3(a). We then compared, for each biological process, the fraction of genes containing the ABE motif out of the entire population of salt-modulated root-expressed genes assigned to that particular process; we found that ABE is highly frequent in genes involved in transport processes, transcription and DNA or RNA metabolism (Figure S3b). The ABE motif was less abundant in promoters of genes involved in signaling and electron transport or energy pathways (Figure S3b).

Figure 10.

Binding of HKT1;1 promoter DNA fragments by recombinant His-ABI4. His-ABI4 was expressed in Escherichia coli, purified, and used for electrophoresis mobility shift assay (EMSA). (a) Black bar, scheme of the promoter sequence: putative cis-acting elements and the ABI4 binding elements (ABE) are shown. White and gray bars represent probes used for the EMSA. Probes A1–A6 and A6-2 were prepared by polymerase chain reaction (PCR) using the respective 5′-[32P]-labeled primers. Probes A1, A6-1 and A6-3 to A6-6 were prepared by annealing 5′-[32P]-labeled 50-bp oligonucleotides. Probes that did not bind ABI4 are in white, probes that bound ABI4 are in gray. (b, c) Radiolabeled probes prepared from DNA sequences of HKT1;1 promoter sub-fragment A6 (b) or ABI4 promoter (c) were analyzed for ABI4 binding (lane 1). Mixtures also contained the specified quantity of unlabeled competitive DNA (lanes 2–9). Lane 10 contains a mixture without ABI4 protein. (d) Alignment of the DNA sequences of HKT1;1 promoter fragments A6-5 and A6-1. ABE sequence is shown in bold. (e, f) EMSA of HKT1;1 promoter 50-bp fragments A6-5 and A6-1 (lanes 1–4) and 50-bp fragments containing eight mutated A6-1 sequences (M1–M8). Mutations are shown in bold in (e). (f) Odd- and even-numbered lanes are mixtures with (+) and without (–) ABI4, respectively. Free and protein-bound probes are marked by F and B, respectively.


We suggest that ABI4 is a key modulator of root activities. ABI4 has been shown to be involved in the effect of nitrate on root branching (Signora et al., 2001). We have shown recently that ABI4 modulates the initiation of lateral roots by mediating ABA and cytokinin inhibition of lateral root formation (Shkolnik-Inbar and Bar-Zvi, 2010), and we show here that it decreases the expression of the Na+ transporter HKT1;1. Finkelstein et al. (2011) showed that ABI4 protein is preferentially accumulated in roots, and argued that there is little similarity between ABI4 transcript and protein levels between different tissues.

The sañ5 mutant was isolated by screening for germination in the presence of salt (Quesada et al., 2000), and was shown to be allelic to abi4. Here, a salt-tolerant germination phenotype was displayed by other abi4 alleles as well (Figure 1). Both radicle emergence and cotyledon opening were less inhibited by NaCl in germinating abi4 mutants compared with germinating WT plants. abi4 mutants also displayed increased salt tolerance when exposed to salt for 1 week (Figure 2) or 1 month (Figure 4) after germination. These results do not agree with the results obtained for the sañ5 mutant, which did not differ from WT plants when exposed to 100 or 150 mm NaCl for 4–10 days after germination (Quesada et al., 2000). The latter study also claimed that sañ5 mutants are hypersensitive to salt compared with WT plants during the first 3 weeks after germination. These differences might result from different experimental conditions. We did not see significant differences between 7-day-old seedlings of abi4 mutants and WT plants which were transferred directly onto growth medium that contained 100 mm NaCl. Transfer onto medium that contained higher NaCl concentrations resulted in necrosis of the seedling, probably due to direct contact of the salt with the apical meristem. We could only demonstrate the differential salt sensitivity of abi4 mutants and ABI4-OE plants when using the agar-layer transfer method, in which seedlings are transferred while embedded in agar, such that only the roots are in contact with the second, NaCl-containing medium.

NaCl treatment induced ABI4 expression in the roots (Figure 3). ABI4 has been shown previously to be induced by ABA and by high concentrations of glucose and other sugars (Arroyo et al., 2003; Price et al., 2003; Bossi et al., 2009; Munoz-Bertomeu et al., 2010; Shkolnik-Inbar and Bar-Zvi, 2010, 2011). These treatments are related, as high sugar and NaCl concentrations share osmotic components, and ABA levels are known to be elevated following exposure of plants to salt or water (osmotic) stresses.

abi4 mutants accumulated a lower level of Na+ in the shoots than WT plants following exposure to NaCl (Figure 5a,d), whereas no significant differences were observed for K+ or Ca2+ accumulation (Figure 5b,c). On the other hand, abi4 mutants and WT plants accumulated similar Na+ concentrations in their roots (Figure 5e), suggesting that the decrease in shoot Na+ content might result from a reduction in root-to-shoot transfer of Na+ rather than from altered balance between uptake and efflux of Na+ by the roots. The plasma membrane Na+/H+ antiporter SOS1 mediates Na+ efflux at the root epidermis (Shi et al., 2002). Expression levels of SOS1 and its modulator SOS3 were not significantly altered in abi4 mutants or in ABI4-OE plants (Figure S4), supporting our suggestion that salt tolerance of abi4 mutants and hypersensitivity of ABI4-OE plants probably do not result from reduced Na+ uptake, but rather from its reduced transport to the shoots. This suggestion is also supported by the increased expression of HKT1;1 in the abi4 mutant, and its complementary reduction in expression in ABI4-OE plants (Figure 7). The Na+ transporter HKT1;1 gene is expressed mainly in the root xylem parenchyma cells, and is suggested to pump Na+ from xylem flow, thus reducing the levels of Na+ transported to the shoot so that less Na+ accumulates in the leaves (Mäser et al., 2002; Sunarpi et al., 2005; Davenport et al., 2007). Overexpression of HKT1;1 in the root stele results in exclusion of shoot Na+ and increased tolerance of Arabidopsis plants to salt stress (Moller et al., 2009). Moreover, HKT1;1 has been suggested to play a role in whole-plant Na+ recirculation by loading Na+ into the phloem sap in the shoot and unloading it in the root, thus enhancing the removal of Na+ from aerial tissues (Berthomieu et al., 2003). The overlap in expression profiles of HKT1;1 and ABI4 in the root phloem and xylem parenchyma (Sunarpi et al., 2005; Shkolnik-Inbar and Bar-Zvi, 2010) is in agreement with our suggestion of ABI4 modulation of HKT1;1 gene expression.

Interestingly, although HKT1;1 expression was elevated in all three studied abi4 mutants, its steady-state levels were higher in abi4-102 and abi4-103 than in abi4-1 (Figure 7a). However, upon exposure to salt, the abi4-1 mutant's response was similar to that of the other two mutants (Figures 2, 4 and 5). This seeming discrepancy might result from the nature of these mutants: abi4-1 is a deletion mutant (codon 157) (Finkelstein et al., 1998), whereas abi4-102 and abi4-103 possess nonsense point mutations at codons 80 and 39, respectively (Laby et al., 2000). As a result, the abi4-1 mutant expresses the AP2-DNA-binding domain and can function as a dominant negative mutant, whereas abi4-102 and abi4-103 are null mutants.

The arr1/arr12 double mutant is less sensitive to salt treatment than WT plants, accumulates less Na+ in its shoots, and displays increased expression of HKT1;1 in its roots (Mason et al., 2010). This phenotype was not observed in the single mutants. The salinity-related arr1/arr12 phenotype thus resembles that of abi4 mutants (Figures 1, 2, 4, 5 and 7). ABI4 does not seem to affect the expression of ARR1 or ARR12 (Figure 8a,b). ABI4 expression is increased following salt treatment of WT, arr1 mutant and arr1/arr12 double-mutant seedlings (Figure 8c). The ABI4 promoter contains multiple ARR binding sites (Sakai et al., 2000), suggesting the ARR1 and probably also ARR12 interact directly with the ABI4 promoter. However, the effect on ABI4 transcript levels in normal growth conditions is marginal. Furthermore, ABI4 maintains its salt-induction in the arr1-3 and arr1-3/arr12-1 mutants. Nevertheless, as arr1/arr12 and abi4 plants display similar improved salt tolerance, we suggest that the arr1/arr12 double mutant is affected in HKT1;1 expression and its cytokinin-mediated salt response occurs via a pathway that is distinct from that mediated by ABI4. The HKT1;1 promoter is most likely highly regulated by a number of unrelated or independent pathways. HKT1;1 expression has been shown to be regulated by distal sequences (Rus et al., 2006; Baek et al., 2011) which seems to be distinct from modulation by ABI4 via the proximal promoter region (Figure 10).

Proline is known to be accumulated in response to salt and water stress (Szabados and Savoure, 2010). Shoots of salt-stressed abi4 mutants accumulated higher proline concentrations than those of WT plants (Figure 6a). qRT-PCR analyses of genes involved in proline biosynthesis and degradation indicated that P5CS1 and PDH2 are induced in shoots of abi4 mutant plants (Figure 6b). P5CS1 encodes the rate-limiting step in proline biosynthesis (Yoshiba et al., 1999), and thus its overexpression is expected to result in increased proline content. On the other hand, induction of PDH2 is expected to result in decreased proline content, contrasting with our observation in the abi4 mutants. This apparent discrepancy can be resolved by the fact that PDH1 has been shown to be the dominant isoform in most plant tissues under most conditions (Deuschle et al., 2004; Funck et al., 2010). Moreover, in contrast with PDH1 loss of function, loss of function of PDH2 does not result in hypersensitivity to proline (Deuschle et al., 2004; Funck et al., 2010). Finally, PDH2 is expressed at much lower levels than PDH1. Taken together, it is suggested that increased expression of PDH2 in abi4 mutants will have only a minor effect on proline levels.

ABI4 has been suggested to act as both transcription activator and repressor (Acevedo-Hernandez et al., 2005; Bossi et al., 2009; Giraud et al., 2009; Munoz-Bertomeu et al., 2010; Finkelstein et al., 2011; Kerchev et al., 2011; Reeves et al., 2011; Yang et al., 2011). We suggest that ABI4 is a major player in determining root physiology and response to abiotic stress. In the root, ABI4 represses lateral root formation (Shkolnik-Inbar and Bar-Zvi, 2010) and expression of the Na+ transporter HKT1;1 (this study). Interestingly, in both processes, ABI4 acts as an inhibitory factor. The determination of ABI4's activity as activator or repressor might result from its interaction with other proteins, protein modification or binding to discrete DNA-binding sites leading to either activation or repression of target genes. Dual functionality of transcription factors has been previously reported. For example, the maize VIVIPAROUS-1 (VP1) transcription factor was shown to function as both an ABA-mediated activator and gibberellic acid-dependent repressor of gene expression in developing aleurone tissue (Carson et al., 1997; Hoecker et al., 1999). WUSCHEL (WUS), which regulates the maintenance of shoot meristems, functions mainly as a repressor, but also as an activator of AGAMOUS (AG) gene expression (Lohmann et al., 2001; Ikeda et al., 2009). Arabidopsis WRKY53 may act as a transcription activator or repressor depending on DNA sequences adjacent to its W-box-binding sites (Miao et al., 2004).

ABI4 binds the HKT1;1 promoter in planta (Figure 9). This binding results in reduced target gene expression (Figure 7). ABI4 has been shown to bind a number of GC-enriched DNA sequences (Niu et al., 2002; Koussevitzky et al., 2007; Shkolnik and Bar-Zvi, 2008; Bossi et al., 2009; Finkelstein et al., 2011; Reeves et al., 2011). CE1 was the first identified ABI4 DNA-binding site (Niu et al., 2002). The binding affinity of ABI4 to CE1 is higher than that observed with other DNA-binding sites (Finkelstein et al., 2011). Although expression of Arabidopsis HKT1;1 is affected by distal promoter sequences (Rus et al., 2006; Baek et al., 2011), potential ABI4-binding sequences are found in the proximal promoter region, a finding that suggested that the modulation of gene activity via distal promoter sequences does not involve ABI4. We identified two highly related ABI4-binding sites in the HKT1;1 promoter (Figure 10), GC(C/G)GCTT(T), termed ABEs. The third T in the nucleotide sequence of the binding site might be less conserved, as mutating it reduced, but did not abolish ABI4 binding (Figure 10f, lane 19). As with other ABI4-binding sites, the ABE is rich in GC. However, this motif also contains TTT at its 3′ end: the first two of these have been shown to be essential for binding, and the third T has been found to enhance ABI4 binding affinity.

The ABE motif is present in 3.6% of salt-modulated root-expressed genes listed in Dinneny et al. (2008). Gene ontology analysis for biological processes suggested that the fraction of the majority of salt-modulated root genes whose promoters contain this motif resembles the fraction of ABE-containing genes in the entire gene population (Figure S3). This suggests that ABI4 modulates genes involved in a large functional spectrum of activities, in agreement with published microarray analyses of abi4 mutants (Kerchev et al., 2011; Reeves et al., 2011). Interestingly, this group is enriched in genes involved in transport and in DNA-dependent transcription activities (Figure S3b), which indicated that ABI4 affects the expression of additional transporters and downstream transcription factors.

We also analyzed promoter sequences of genes whose expression levels are higher in abi4 mutants than in WT plants, for the existence of ABE-related sequences. The P5CS1 promoter has one truncated ABE sequence, GCGGCTT, in agreement with its modulation by ABI4 (Figure 6). ABE is missing in the promoter sequences of the other genes involved in proline metabolism, shown in Figure 6. The PDH2 promoter contains CE-1 and DRE-like sequences that may function as ABI4 binding sites. The promoters of the genes encoding WRKY40 transcription factor (At1g80840), NAC domain-containing protein transcription factor (At2g17040) and PR1-encoding pathogen related protein 1 (At2g14610), which are overexpressed in abi4 mutants compared with WT plants (Kerchev et al., 2011), contain GACCGCTTT, GCCTGCTTT, and GCGGCATA, respectively. Thus we suggest that binding of ABI4 to ABE might result in repression of gene expression in the root.

Experimental procedures

Plant material and growth conditions

All plants used in this study were Arabidopsis thaliana (Columbia ecotype): abi4 mutants were obtained from the Arabidopsis Stock Center (http://abrc.osu.edu/), and plants expressing ABI4::GUSpOp6::ABI4, and cauliflower mosaic virus 35S::ABI4 were obtained as previously described (Shkolnik-Inbar and Bar-Zvi, 2010). Seed sterilization and plant growth on solid 0.5 × Murashige and Skoog (MS) medium + 0.5% (w/v) sucrose or in pots were as described previously (Shkolnik and Bar-Zvi, 2008). Hydroponic growth of Arabidopsis was performed as described (Huttner and Bar-Zvi, 2003).

Constructs and Arabidopsis plant transformation

DNA corresponding to the ABI4 coding sequence was isolated by PCR using genomic DNA as the template and the primers listed in Table S2. These primers also contained recognition sites for the restriction enzymes XhoI and SpeI. The amplified sequence was digested with XhoI and SpeI and subcloned in frame into the respective sites of pJIM19 vector downstream of sequences encoding a 3 × HA-3 × FLAG polypeptide. Constructs were introduced into Agrobacterium tumefaciens GV3101 and used in the transformation of wild-type (WT) Arabidopsis plants (Clough and Bent, 1998). Transgenic plants were selected on plates containing hygromycin. Plants were selfed twice, and T3 homozygous plants were used in this study.

Split-YFP assay was performed according to Bracha-Drori et al. (2004): the ABI4-encoding DNA fragment was subcloned into the YFP–C-terminal-harboring plasmid pRTL2-HAYC, and the ARR1- and ARR12-encoding DNA fragments were subcloned into the YFP–N-terminal-harboring plasmid, pRTL2-EEYN. The split-YFP constructs were isolated from the plasmids by digestion with HindIII and subcloned into the pCAMBIA-2300 binary vector followed by introduction into Agrobacterium. Logarithmic Agrobacterium cultures were harvested, resuspended in transformation buffer [0.25 × MS salt mixture, 1% sucrose, 0.005% (v/v) silwet, 100 mm acetosyringone (3′,5′-dimethoxy-4′-hydroxyacetophenone) adjusted to pH 6.0] to OD 600 nm = 0.6, and were imaged using a Zeiss LSM510 confocal laser scanning microscope.

Salt treatment


Cold-treated surface-sterilized seeds were plated on agar-solidified medium containing 0.5 × MS, 0.5% sucrose, 0.7% (w/v) agar and 0.2 m NaCl. Radicle emergence and cotyledon opening were scored daily for 5 days. Each assay was performed with 50 seedlings from each line.


Cold-treated surface-sterilized seeds were plated on agar-solidified medium containing 0.5 × MS, 0.5% sucrose and 0.7% agar, and plants were grown as described above. A week later, solid medium was sliced into quarters, and slices with the embedded seedlings were placed on a second medium containing 0.5 × MS and 0.5% agar with or without 0.3 m NaCl. Plates were incubated horizontally in the growth room and seedling survival was scored 10 days after the transfer. Gene expression was assayed after 1 h of stress. Each assay was performed with 50 seedlings from each line.

Pot-grown plants

Plants were grown in pots as described (Shkolnik and Bar-Zvi, 2008). One-month-old plants were irrigated with a 0.2 m NaCl solution containing an additional 1/15th concentration of CaCl2. Control plants were irrigated with water. Plant survival was scored 3 weeks later.

Hydroponically grown plants

One-month-old plants were transferred to fresh aerated growth medium (Huttner and Bar-Zvi, 2003) supplemented with 0 or 100 mm NaCl. Plants were harvested 8 h later.

qRT-PCR analysis

RNA isolation and cDNA synthesis were performed as described previously (Shkolnik-Inbar and Bar-Zvi, 2010). Relative transcript levels were assayed by real-time PCR analysis using the Applied Biosystems 7300 real-time PCR system (http://www.appliedbiosystems.com). Primer design and reaction conditions were as in Shkolnik-Inbar and Bar-Zvi (2010), and the primers are listed in Table S3. Data represent mean ± SE of n = 3 independent experiments. Each data point was determined in triplicate in each of the three biological replicates and presented as mean ± SE.

Western blot analysis

Seven-day-old seedlings were harvested and homogenized in 2:1 (v/w) 2× Laemmli sample buffer (Laemmli, 1970). Samples were heated for 10 min at 80°C and proteins were resolved by SDS-PAGE (10% acrylamide; Laemmli, 1970). Proteins were blotted onto nitrocellulose membranes. TAP-ABI4 protein was detected using anti-FLAG antibody (Cayman Chemicals Company, http://www.caymanchem.com).

GUS staining

Plant tissues were fixed in acetone for GUS staining as described previously (Weigel and Glazebrook, 2002). Transverse root sections were prepared as described previously (Shkolnik-Inbar and Bar-Zvi, 2010). Pictures were taken with a Nikon camera (DXM1200F) using a stereoscope and microscope. Each treatment was performed using three biological replicates.

Ion and proline contents

One-month-old pot-grown plants were treated with the indicated salt concentrations. Rosette leaves were harvested 48 h later, and ion and proline contents were determined as described previously (Kalifa et al., 2004). Hydroponically grown plants were harvested 8 h after treatment and rinsed three times in 50 ml deionized water. Roots and shoots were separated and dried for 48 h at 75°C. Dried shoots were ground to a fine powder, whereas root tissues were transferred without further grinding to corked glass tubes. A 2.5-ml aliquot of HNO3:HClO4 (1:2, v/v) mix was added, and the samples were incubated overnight at room temperature. Mineral oil (0.2 ml; Sigma-Aldrich Co., http://www.sigmaaldrich.com) was added, and the tubes were further incubated for 5 min in a boiling water bath. The samples were cooled to room temperature, and 10 ml of deionized water was added. Samples were then filtered through Whatman No. 4 filter paper, and Na+ content was determined by Varian AA240FS flame atomic absorption spectrometer (www.varianinc.com).

Chromatin immunoprecipitation

ChIP was performed essentially as in (Shkolnik and Bar-Zvi, 2008). Chromatin was isolated from 1-week-old seedlings constitutively expressing HA-FLAG-ABI4, grown on agar-solidified medium containing 0.5 × MS and 0.5% sucrose, cross-linked and sheared as in (Shkolnik and Bar-Zvi, 2008). Immunoprecipitation was performed using anti-FLAG (Cayman Chemicals Company, http://www.caymanchem.com). PCR analyses were carried out as described (Shkolnik and Bar-Zvi, 2008) using primers presented in Table S3. PCR products were analyzed by agarose gel electrophoresis. DNA was visualized by ethidium bromide staining.

Electrophoresis mobility shift assay

ABI4-coding sequence was isolated by PCR using ABI4 FOR SalI and ABI4 REV SalI primers (Table S2) and cloned into pGEM T Easy vector (Promega Biotech, http://www.promega.com). The resulting plasmid was sequenced, digested with SalI, and the ABI4-encoding sequence was subcloned in frame into the compatible XhoI site in pRSET A vector (Invitrogen, http://www.invitrogen.com). The resulting plasmid was introduced into Escherichia coli BL-21 and bacteria were cultured at 25°C. His-tagged recombinant ABI4 was expressed following induction by the addition of isopropyl β-d-1-thiogalactopyranoside (IPTG) to culture in the logarithmic growth phase. Recombinant His-ABI4 was purified on a Ni-NTA column (Qiagen, http://www.qiagen.com/), dialyzed twice against cold buffer containing 10 mm Tris–HCl pH 8.0 and 150 mm NaCl, aliquoted and stored at −80°C. EMSA reaction mixture (final volume of 20 μl) contained 100 mm Tris–HCl pH 7.4, 250 mm KCl, 15 mm MgCl2, 5 mm EDTA, 10% (w/v) glycerol, 50 ng poly(dIdC):poly(dIdC) (Pharmacia GE Health Life Sciences http://www.gelifesciences.com), 32P-labeled probe (5000 cpm) and 150 ng of His-ABI4 protein. The mixture was incubated for 20 min at room temperature, 1 μl bromophenol blue solution was added, and it was electrophoresed in a 5% polyacrylamide gel in 0.25 × Tris-borate (TB) buffer (Shkolnik and Bar-Zvi, 2008). Gels were dried and exposed to phosphorimager or X-ray film.

5′-[32P]-labeled oligonucleotides were prepared using a reaction mixture (20 μl) containing 5 μm each of forward and reverse primers (for primers used for promoter sequence amplification), or 25 μm each of forward and reverse 50 bp-long oligonucleotides used directly for EMSA, 2 μl of [γ-32P]ATP (specific activity 3000 Ci mmol−1, Perkin Elmer, http://www.perkinelmer.com), T4 polynucleotide kinase (New England Biolabs, http://www.neb.com), and a buffer supplied by the enzyme manufacturer. Mixtures were incubated for 60 min at 37°C followed by denaturation of the enzyme (10 min at 70°C). Labeled primers were used in a PCR mixture containing genomic DNA template, 0.2 mm dNTP, Ex-Taq (TaKaRa Bio. Inc., http://www.takara-bio.com) and buffer provided by the manufacturer, and the following reaction was run: 1 min at 96°C, two cycles of 30 sec each at 96°C, 47°C and 72°C followed by 30 cycles of 30 sec each at 96°C, 57°C and 72°C, and three min at 72°C. The resulting products were electrophoresed on a 2% agarose gel, bands corresponding to amplified DNA were excised, and the DNA was eluted using a DNA isolation kit (Biological Industries, http://www.bioind.com). Labeled 50-bp oligonucleotides were placed in a microcentrifuge tube containing 100 μl of a mixture containing 10 mm Tris–HCl pH 8.0 and 50 mm NaCl. Tubes were placed in an ice bucket containing 3 L of boiling water, and slowly allowed to cool overnight to anneal the oligonucleotides. Primers and oligonucleotides are listed in Table S4.

Promoter sequence analysis

The 3000-bp upstream sequences of the Tair10 release of the Arabidopsis genome were scanned for putative motifs using the Pathmatch algorithm (http://www.arabidopsis.org/cgi-bin/patmatch/nph-patmatch.pl). The resulting data were further screened, and hits residing within the coding sequence of upstream genes were discarded. Common groups of genes whose promoter sequences contained the putative box (such as ABE box) and given lists of expressed genes from published microarray analyses (as specifically indicated in the text) were combined, and their annotation analyzed (http://www.arabidopsis.org/tools/bulk/go/index.jsp). We used the published list of salt-regulated root-expressed genes identified by microarray analyses (Dinneny et al., 2008).

Accession numbers

ABI4, AT2G40220; HKT1;1, AT4G10310.


We Thank Rina Yeger for her help with specimen sectioning. This study was supported in part by a fellowship awarded to DSI by the Kessel Salinity Center at Ben-Gurion University, and in part by the United States–Israel Binational Science Foundation. DBZ is the incumbent of The Israel and Bernard Nichunsky Chair in Desert Agriculture, Ben-Gurion University.