Soybean NAC transcription factors promote abiotic stress tolerance and lateral root formation in transgenic plants


(fax +86 10 64873428; e-mail or


NAC transcription factors play important roles in plant growth, development and stress responses. Previously, we identified multiple NAC genes in soybean (Glycine max). Here, we identify the roles of two genes, GmNAC11 and GmNAC20, in stress responses and other processes. The two genes were differentially induced by multiple abiotic stresses and plant hormones, and their transcripts were abundant in roots and cotyledons. Both genes encoded proteins that localized to the nucleus and bound to the core DNA sequence CGT[G/A]. In the protoplast assay system, GmNAC11 acts as a transcriptional activator, whereas GmNAC20 functions as a mild repressor; however, the C-terminal end of GmANC20 has transcriptional activation activity. Over-expression of GmNAC20 enhances salt and freezing tolerance in transgenic Arabidopsis plants; however, GmNAC11 over-expression only improves salt tolerance. Over-expression of GmNAC20 also promotes lateral root formation. GmNAC20 may regulate stress tolerance through activation of the DREB/CBF–COR pathway, and may control lateral root development by altering auxin signaling-related genes. GmNAC11 probably regulates DREB1A and other stress-related genes. The roles of the two GmNAC genes in stress tolerance were further analyzed in soybean transgenic hairy roots. These results provide a basis for genetic manipulation to improve the agronomic traits of important crops.


NAC-type proteins constitute a plant-specific transcription factor family. All proteins in this family share a conserved N-terminal DNA-binding domain called a NAC (NAM, ATAF1/2, CUC1) domain. The C-terminus of the NAC proteins varies in sequence and generally regulates transcriptional activation. There are 105 family members in Arabidopsis and 140 members in rice (Ooka et al., 2003; Fang et al., 2008). In Glycine max, we predicted the existence of 54 NAC genes based on the EST database (Tian et al., 2004), and Pinheiro et al. (2009) predicted 101 NAC-type proteins based on the whole-genome sequence of soybean (Glycine max). NAC proteins play essential roles in many biological processes, including development, senescence, morphogenesis and stress signal transduction (Olsen et al., 2005a,b). The first NAC-type transcription factor was identified in Petunia and named NAM (no apical meristem) (Souer et al., 1996). Since then, many NAC-type genes have been found to play roles in shoot apical meristem formation and organ separation (Takada et al., 2001; Hibara et al., 2003, 2006; Vroemen et al., 2003).

Several NAC genes play important roles in auxin signaling and lateral root formation. Xie et al. (2000) showed that NAC1 is induced by auxin and is an essential gene involved in the auxin signaling pathway, which directs lateral root formation in Arabidopsis. Expression of NAC1 can be regulated by the ubiquitin-mediated protein degradation pathway through the RING-finger homolog protein SINAT5, which has ubiquitin protein ligase activity (Xie et al., 2002). We also identified the NAC gene AtNAC2 (He et al., 2005), which can be induced by multiple plant hormones [including abscisic acid (ABA), auxin and ethylene] and high salinity. Over-expression of the gene promoted lateral root formation in transgenic plants (He et al., 2005).

NAC proteins are also involved in biotic and abiotic stress responses. GRAB1/2 (geminivirus RepA-binding) proteins interact with the wheat (Triticum aestivum) dwarf geminivirus RepA protein to interfere with the dwarf geminivirus life cycle (Xie et al., 1999). TIP (TCV-interacting protein) binds to the turnip crinkle virus (TCV) capsid protein to promote TCV resistance (Ren et al., 2000). BnNAC genes from Brassica napus are induced by wounding, cold and drought (Hegedus et al., 2003). Transgenic rice plants expressing OsNAC6 were tolerant to dehydration, high-salt stress and blast disease (Nakashima et al., 2007). Transgenic rice plants over-expressing SNAC1 and SNAC2 displayed high tolerance to dehydration, high salinity and/or cold stress (Hu et al., 2006, 2008). In Arabidopsis, over-expression of three NAC genes ANAC019, ANAC055 and ANAC072 enhanced the drought tolerance of transgenic Arabidopsis plants (Tran et al., 2004, 2007). ANAC072 (RD26, Responsive to Desiccation 26) was found to be involved in a new ABA-dependent pathway, and enhances expression of GLY1 (glyoxalase I family) (Fujita et al., 2004). Arabidopsis ATAF1 and ATAF2 play roles in both biotic and abiotic stress responses (Delessert et al., 2005; Wu et al., 2009). Recently, Tran et al. (2009) analyzed 31 full-length NAC genes from soybean and found that nine of them were induced by drought and other stresses such as high salinity, cold and ABA treatment. However, the functions of these soybean NACs in stress tolerance is unknown.

Soybean is an important crop, and provides large amounts of protein and oil. Previously, we analyzed the genomic features of the soybean genome based on EST analysis, and we identified more than 1000 transcription factor genes (Tian et al., 2004). Several transcription factor gene families, including GmbZIP, GmMYB, GmWRKY, GmGT and GmPHD, have been studied with respect to their expression patterns and roles in stress tolerance (Liao et al., 2008a,b; Zhou et al., 2008; Wei et al., 2009; Xie et al., 2009). In soybean, 54 putative NAC genes have been found (Tian et al., 2004). Their expression levels were analyzed under drought, salt, cold and ABA treatments, and 15 genes were responsive to at least one of these treatments (Hao, 2008). In the present study, two genes, GmNAC11 and GmNAC20, both of which were induced by multiple treatments, were further investigated to determine their functions in plant stress responses. Transgenic Arabidopsis plants over-expressing the GmNAC20 gene showed enhanced tolerance to salt and freezing stress compared to control plants. GmNAC20 expression also promoted lateral root formation in the transgenic plants. In contrast, over-expression of GmNAC11 conferred better performance under salt stress only. The roles of the two genes were also examined using the soybean transgenic hairy root system. These studies allow elucidation of NAC roles in both stress responses and development.


GmNAC gene expression and protein localization

Our previous analysis identified 54 putative NAC genes from soybean (Tian et al., 2004). Among these, two multiple stress-inducible genes, GmNAC11 (EU440354) and GmNAC20 (EU440353), were chosen for further analysis of their roles in abiotic stress responses. GmNAC11 encoded a protein of 351 amino acids, and GmNAC20 encoded a protein of 268 amino acids. Both proteins had a NAC domain of similar length in the N-terminus and a variable C-terminus. Expression of GmNAC11 was induced by high salinity, dehydration, ABA and 1-naphthylacetic acid (NAA) treatments, but it was not significantly affected by cold stress (Figure 1a). GmNAC20 expression was apparently induced by high salinity, drought, cold (4°C) and NAA treatments, but was only slightly induced by ABA (Figure 1a). After salt, drought and NAA treatments, GmNAC11 expression appeared to precede GmNAC20 induction. The differential expression of the two genes implies that they may function in different stress responses or at different stages of the same stress response.

Figure 1.

GmNAC gene expression and protein localization.
(a) GmNAC11 and GmNAC20 expression levels in response to various stresses and hormones in soybean seedlings as revealed by Northern blot analysis. RNA was stained using methylene blue.
(b) GmNAC expression in various organs of soybean plants.
(c) Subcellular localization of GmNAC11 and GmNAC20 proteins in Arabidopsis protoplasts.

The soybean organ expression patterns of the two genes were examined, and both GmNAC11 and GmNAC20 were highly expressed in the roots and cotyledons (Figure 1b). These genes had relatively low expression levels in the other organs tested. NAC proteins are transcription factors that are usually located in the nucleus. However, some NAC proteins are also membrane-associated (Seo et al., 2010). We determined the subcellular localization of GmNAC11 and GmNAC20. Each gene was fused to the GFP gene in a transient expression vector, and the fusion genes driven by the 35S promoter were transformed into Arabidopsis protoplasts. Both GmNAC11 and GmNAC20 were located in the nucleus, and the GFP control was mainly located in the cytoplasm (Figure 1c). These results indicate that GmNAC11 and GmNAC20 are nuclear proteins.

Transcriptional activation ability of the GmNAC11 and GmNAC20

The abilities of GmNAC11 and GmNAC20 to activate transcription were examined using a dual-luciferase reporter (DLR) assay system in Arabidopsis protoplasts. The coding sequences of the two GmNAC proteins were fused to the DNA sequence encoding the GAL4 DNA-binding domain to generate pBD-GmNAC effector plasmids (Figure 2a). The fusion gene was driven by the 35S promoter plus a translational enhancer Ω. The firefly luciferase gene (LUC) driven by a mini-35S (TATA box) promoter with five copies of the GAL4 binding element was used as a reporter (Figure 2b), and the Renilla luciferase gene driven by the Arabidopsis Ubiquitin3 promoter was used as an internal control. The GAL4 DNA-binding domain in the fusion proteins binds to the GAL4 binding element upstream of the reporter LUC gene, and the activation domain in the tested proteins activates LUC gene transcription. Compared with the GAL4-DBD negative control, GmNAC11 strongly activated the reporter gene, while GmNAC20 showed less activity than the negative control (Figure 2b). The C-terminal regulatory domain (20AD) of GmNAC20 was further tested, and had a strong ability to activate reporter gene expression (Figure 2b). These results indicate that GmNAC11 and the C-terminal domain of GmNAC20 can activate transcription, whereas the full-length version of GmNAC20 appears to lack this activity.

Figure 2.

 Transcriptional activation ability of GmNAC11 and GmNAC20 in protoplasts.
(a) Effector constructs used in the Arabidopsis protoplast transient assay. Each effector contained a GAL4 DNA-binding domain (GAL4DBD), except DREB1A, which contained its coding region. The GAL4DBD effector was used as a negative control, and effector 20AD, which contains the activation domain of GmNAC20, was used as a positive control. GmNAC11 or GmNAC20 was fused with the GAL4DBD and expression was driven by the 35S promoter plus the translation enhancer Ω sequence.
(b) Transcriptional activation ability of GmNAC11 and GmNAC20 as revealed by relative LUC activity of the reporter. The effectors and the GAL4–LUC reporter were co-transfected.
(c) GmNAC20 shows moderate repression activity for transcription. The GAL4–DRE–LUC reporter contains four copies of the DRE element for DREB1A binding and five copies of the GAL4 binding element.

Because GmNAC20 does not have transcriptional activation ability, it may have transcriptional suppression activity. A construct was made that contained the GAL4 binding element plus four copies of DRE elements upstream of the LUC reporter gene (Figure 2c). An effector plasmid harboring Arabidopsis DREB1A driven by a 35S promoter was also constructed. DREB1A binds the DRE element (Figure 2a), and has transcriptional activation activity (Figure 2c). Effectors for the GmNAC genes and DREB1A and the reporter and internal control were co-transfected into protoplasts to evaluate their inter-molecular transcriptional interaction. If GmNAC20 has transcriptional suppression activity, it will inhibit the activity of DREB1A after co-transfection, as in the case of ERF transcriptional repressors (Ohta et al., 2001). Co-transfection of DREB1A and GmNAC11 resulted in higher LUC activity than DREB1A and GAL4DBD co-transfection, indicating that GmNAC11 still functions as a transcriptional activator at the intermolecular level. Conversely, co-transfection of DREB1A and GmNAC20 led to lower LUC activity than that of DREB1A and GAL4DBD co-transfection, suggesting that GmNAC20 may act as a transcriptional repressor at the intermolecular level (Figure 2c). 20AD had transcription activation activity when co-transfected with DREB1A (Figure 2c).

GmNAC11 and GmNAC20 bind to the CGT[A/G] core sequence

Plant NAC proteins bind to various cis-DNA elements. NAC1 and AtNAM have the ability to bind to the 21 bp minimal sequence TGACGTAAGGGATGACGCACC of the 35S promoter (−83 to −63) (Xie et al., 2000; Duval et al., 2002), and probe 1 (P1) was derived from this sequence (Figure 3a). ANAC019/055/072 binds to the core element CACG in TTGAAAACTTCTTCTGTAACACGCATGTG in the ERD1 promoter (Tran et al., 2004). This sequence was used as probe 2 (P2) (Figure 3a). The element GAGATCCGTGCACAGTACGTAACTGTTA for TaNAC69 (Xue, 2005) and the element GAATTCAAGTAGCTTA for the CBNAC protein (Kim et al., 2007) were used as probe 3 (P3) and probe 4 (P4), respectively (Figure 3a). GmNAC11 formed a complex with all of the probes, and the signal was dramatically reduced by addition of unlabeled DNA probe (Figure 3a), indicating that GmNAC11 specifically binds to these elements. Olsen et al. (2005a) reported that the target promoter regions of NAC proteins mostly contained the core sequence CGT[G/A]. The probes P1, P2 and P3, but not P4, used in the present experiments contained the core sequence CGT[G/A] or its reverse complementary sequence. We then tested whether GmNAC11 and GmNAC20 could bind to this core sequence CGT[G/A]. A probe containing four copies of CGT[G/A] was used. Both GmNAC11 and GmNAC20 had the ability to bind to CGT[G/A] (Figure 3a). Addition of unlabeled core probe (competitor) at increasing concentrations apparently reduced the binding of GmNAC11 and GmNAC20 to the labeled probe (Figure 3b, upper panel). Neither protein showed binding to the mutated core element (Figure 3a,b, upper panel). The binding ability of the two proteins was gradually reduced with the decrease in protein levels (Figure 3b, lower panel). All these results indicate that GmNAC11 and GmNAC20 specifically bind to the core element.

Figure 3.

 DNA binding ability of GmNAC11 and GmNAC20.
(a) GmNAC proteins bind to various elements. GmNAC11 and GmNAC20 were expressed and subjected to a gel-shift assay. A core binding site mutant was synthesized for use in (b).
(b) GmNAC binding to a core element in the presence of changing concentrations of competitor or protein. The mutant element was included as a negative control.
For (a) and (b), the arrows indicate the positions of a protein/DNA complex.

Performance of plants over-expressing GmNAC under salt stress

Transgenic Arabidopsis plants over-expressing each of the two GmNAC genes driven by the CaMV 35S promoter were generated. The tobacco mosaic virus Ω sequence was inserted downstream of the 35S promoter to enhance translation efficiency. For each gene, at least 15 transgenic lines were obtained, and two or three independent homozygous lines with relatively high expression of transgenes (Figure 4c, and data not shown) were investigated further.

Figure 4.

 Roles of GmNAC genes in lateral root formation in Arabidopsis transgenic plants and in salt tolerance in soybean transgenic hairy roots.
(a) GmNAC20 over-expression in transgenic Arabidopsis lines 20a, 20b and 20c promotes lateral root formation.
(b) Comparison of lateral root density in Col-0 and transgenic plants. Each data point is the mean of 25–30 seedlings. Error bars indicate SD, and asterisks indicate an significant difference (P < 0.01) compared to the control.
(c) Auxin signaling-related gene expression levels were altered in GmNAC20 transgenic lines 20a, 20b and 20c as determined by Northern blot analysis. Ethidium bromide staining was used to verify RNA loading. The relative expression levels of each gene are shown under each panel.
(d) Performance of soybean GmNAC transgenic hairy roots under salt stress. Upper panel: relative gene expression in transgenic hairy roots compared with the corresponding controls; each column is the mean of four replicates, and the error bars indicate SD. Lower panels: net increase of root length (left) and relative growth rate (right) of transgenic hairy roots under salt stress (100 mm NaCl); each column is the mean of 40 salt-treated transgenic roots, and the error bars indicate SE.

Because GmNAC11 and GmNAC20 were both responsive to multiple stresses, we studied their functions in abiotic stress responses. The seedlings were transferred to medium with 150 mm NaCl and maintained for 10 days. After treatment, all seedlings were transferred into pots containing vermiculite soil for recovery. Under normal conditions, no obvious difference was observed between the control and transgenic plants (Figure 5a). After recovery from the 150 mm NaCl treatment for 7 days, the transgenic plants displayed better performance than Col-0 based on rosette size (Figure 5a). After 15 days recovery, approximately 40% of wild-type plants survived, whereas more than 60% of the transgenic plants over-expressing GmNAC11 or GmNAC20 survived (Figure 5a,b). For GmNAC20 transgenic line 20c, the survival rate was over 80%. The rosette diameters of the transgenic plants were also significantly greater than those of wild-type plants (Figure 5a,c). Electrolyte leakage has been used as an indicator of the damage caused by abiotic stresses (Cao et al., 2007). The levels of ion leakage in the transgenic plants over-expressing GmNAC11 or GmNAC20 were significantly lower than those in Col-0, especially under high-salinity stress (Figure 5d). Together, these results indicate that over-expression of GmNAC11 and GmNAC20 results in enhanced tolerance to salt stress.

Figure 5.

 Over-expression of GmNAC11 and GmNAC20 in Arabidopsis improves salt stress tolerance.
(a) Performance of transgenic Arabidopsis plants over-expressing GmNAC20 or GmNAC11 under salt stress. Seedlings were treated on 150 mm NaCl for 10 days and transferred to soil for recovery.
(b) Survival rate after recovery from salt stress. Each data point is the mean of four experiments, and each experiment comprises 30 plants.
(c) Rosette diameter of transgenic plants after recovery from stress. Each data point is the mean of 20–30 seedlings.
(d) Relative ion leakage in transgenic lines after salt stress. Each data point is the mean of three samples.
For (b–d), error bars indicate SD, and asterisks indicate a significant difference (*P < 0.05; **< 0.01) compared with the corresponding controls.

Over-expression of GmNAC20 improves plant freezing tolerance

Three-week-old potted plants were exposed to various freezing temperatures for 12 h, and then transferred to normal 22°C conditions for recovery. All plants grew well at 22 and −2°C (Figure 6a). At −4 and −6°C, the survival rates of the three GmNAC20 transgenic lines were much higher than those of the wild-type and GmNAC11 transgenic plants (Figure 6a,b). When the temperature was decreased to −8°C, almost all of the plants died (Figure 6a,b). In a further experiment, 3-week-old plants were treated at 4°C for 3 days, and the relative electrolyte leakage was measured. The three GmNAC20 transgenic plant lines had lower levels of electrolyte leakage than Col-0 and the GmNAC11 transgenic lines (Figure 6c), suggesting that the GmNAC20 transgenic plants are more tolerant to cold stress than Col-0 and the GmNAC11 transgenic plants. Acclimation usually improves plant adaptation to cold stress. Three-week-old seedlings were acclimated to 4°C for 4 days before treatment at −7°C. After this treatment, the three lines over-expressing GmNAC20 showed higher survival rates than the Col-0 plants (Figure 6d). However, the survival rates were similar to those obtained without acclimation (Figure 6b). No significant difference was observed between Col-0 and GmNAC11 transgenic lines after the acclimation plus freezing treatments (data not shown). These results indicate that over-expression of GmNAC20, but not GmNAC11, promotes plant tolerance to low temperature stress. However, acclimation treatment does not appear to further improve the freezing tolerance.

Figure 6.

 Over-expression of GmNAC20 in Arabidopsis improves the freezing tolerance.
(a) Performance of Col-0 and GmNAC11 and GmNAC20 transgenic lines after freezing treatments without acclimation.
(b) Survival rate of plants in (a) under freezing temperatures. Each data point is the mean of four experiments, and each experiment comprises 30 plants.
(c) Relative electrolyte leakage of wild-type and transgenic lines under cold (4°C) treatment for 3 days. Each data point is the mean of three samples.
For (b) and (c), error bars indicate SD, and asterisks indicate a significant difference (*P < 0.05; **< 0.01) compared with the corresponding controls.
(d) GmNAC20 transgenic lines and wild-type under freezing conditions with acclimation treatment. For acclimation, seedlings were grown at 4°C for 5 days before the freezing treatment was performed. The survival rate was analyzed after recovery for 5 days.

GmNAC11 and GmNAC20 regulate stress-responsive gene expression

The GmNAC transgenic plants showed a higher tolerance to freezing and/or salt stress compared to Col-0 (Figures 5 and 6). We next determined whether expression of stress-responsive genes (Liu et al., 1998) was altered in these transgenic plants. After cold treatment (4°C), expression of DREB1A/CBF3 and KIN2/cor6.6 was induced to a higher level in the three GmNAC20 transgenic lines than in Col-0, especially after the 1 h treatment (Figure 7a). The induction of DREB1C/CBF2 was lower in the GmNAC20 transgenic plants than in Col-0 (Figure 7a). The induction of DREB1B/CBF1 in the transgenic lines was similar to that in Col-0. cor15A and RD29A/cor78 showed higher transcript levels in GmNAC20 transgenic lines than that in Col-0, especially after the 3 h treatment (Figure 7a). These results suggest that GmNAC20 may regulate cold stress tolerance through activation of the DREB/CBF–COR pathway.

Figure 7.

 GmNAC11 and GmNAC20 regulate stress-responsive gene expression levels in transgenic Arabidopsis.
(a) GmNAC20 regulates the expression levels of DREB and COR in transgenic Arabidopsis lines 20a, 20b and 20c upon cold (4°C) treatment. Expression levels were determined by Northern blot analysis. The rRNAs were stained by methylene blue to verify loading.
(b) GmNAC11 regulates stress-responsive gene expression in transgenic lines 11a and 11b as revealed by quantitative PCR. Each column is the mean of four replicates. Actin2 was amplified as a normalization control.
(c) GmNAC20 enhances DREB1A promoter activity but suppresses DREB1C promoter activity in the protoplast assay. The GmNAC20 effector plasmid and vector control were co-transfected with the LUC reporter gene driven by the DREB1A or DREB1C gene promoter.
(d) GmNAC11 and GmNAC20 specifically bind to the DREB1A promoter region. The underlined sequences indicate the core elements or mutated core elements in the DREB1A probe. The arrowheads indicate the position of the protein/DNA complex.

Real-time quantitative PCR analysis was performed to examine GmNAC11-regulated genes in transgenic plants. The expression of DREB1A, ERD11, cor15A, ERF5, RAB18 and KAT2 was enhanced in GmNAC11 transgenic lines compared to Col-0 (Figure 7b). These results imply that GmNAC11 may confer stress tolerance through regulation of stress-responsive genes.

Because over-expression of GmNAC20 enhances expression of DREB1A/CBF3 but suppresses that of DREB1C/CBF2 (Figure 7a), we tested whether GmNAC20 can regulate their promoter activity using the Arabidopsis protoplast transient expression system. Two constructs harboring a LUC reporter gene driven by either the Arabidopsis DREB1A/CBF3 or DREB1C/CBF2 promoter were produced. The promoter–LUC reporter gene and the GmNAC20 effector gene were co-transfected into protoplasts. Compared to the vector control, GmNAC20 increased the DREB1A promoter–LUC activity but reduced the DREB1C promoter–LUC activity (Figure 7c).

Because GmNAC20 and GmNAC11 can activate DREB1A gene expression and/or enhance DREB1A promoter activity (Figure 7a–c), we determined whether the two proteins can bind to the promoter region of DREB1A. A 51 bp sequence from the DREB1A promoter was identified as containing the two core elements CGTG and CGTA (Figure 7d). Both GmNAC11 and GmNAC20 were found to specifically bind to the 51 bp DREB1A probe, but they did not bind to a mutated version of the probe (Figure 7d). The results indicate that GmNAC11 and GmNAC20 most likely activate DREB1A expression through direct binding to core elements in its promoter region.

Over-expression of GmNAC20 promotes lateral root formation in transgenic plants

NAC proteins have been found to regulate lateral root formation (Xie et al., 2000; He et al., 2005). No significant difference in primary root length was found between GmNAC20 transgenic plants and Col-0. However, the lateral root numbers and densities were significantly enhanced in GmNAC20 transgenic plants compared to Col-0 (Figure 4a,b, left panels). Under salt stress, the lateral root density of the GmNAC20 transgenic plants was slightly reduced compared to non-stressed GmNAC20 transgenic plants, but it was significantly higher than in stressed Col-0 plants (Figure 4a,b, right panels). These results indicate that GmNAC20 promotes lateral root formation in transgenic plants under both normal and salt-stress conditions. However, GmNAC11 does not perform these functions (data not shown).

Auxin plays a central role in lateral root initiation and development, and the expression of GmNAC20 was responsive to auxin treatment (Figure 1a). We investigated whether GmNAC20 regulates the expression of genes in the auxin signaling pathway. The expression levels of AIR3, ARF7, ARF19, AXR3 and LBD12 were increased in the transgenic lines, but the expression levels of ARF2 and AXR1 were clearly down-regulated in the transgenic lines (Figure 4c). These results indicate that GmNAC20 may regulate auxin signaling genes to control lateral root initiation and development.

Performance of GmNAC11 and GmNAC20 transgenic hairy roots in soybean seedlings under salt stress

The roles of GmNAC11 and GmNAC20 in stress tolerance were further studied using Agrobacterium rhizogenes-mediated transformation of soybean roots (Kereszt et al., 2007). The over-expression vectors were introduced into the Agrobacterium rhizogenes K599 strain, and the bacterium was used to infect the hypocotyls of soybean seedlings. After removal of the original main roots, the transgenic hairy roots from the infection sites were subjected to salt stress (100 mm NaCl). The net increase in root length and relative growth rate of these roots (net increase of root length under salt stress/mean net increase of root length in water) were calculated. The GmNAC11 transgenic roots showed higher GmNAC11 expression compared to the control (Figure 4d, upper panel), and this correlated well with the higher relative growth rate of the GmNAC11 transgenic roots under salt stress (Figure 4d, lower right panel). However, after extensive tests, we were only able to obtain GmNAC20 transgenic roots with an approximately 40% increase in GmNAC20 expression (Figure 4d, upper panel). The relative growth rate of GmNAC20 transgenic roots was weakly enhanced under salt stress (Figure 4d, lower right panel). It should be noted that under the water control conditions, the root growth in GmNAC20 transgenic plants was greater than in the vector control, but that in GmNAC11 transgenic plants was lower. This result suggests that the two genes differentially regulate the root growth of soybean plants (Figure 4d, lower left panel). Under salt stress, roots of both transgenic plant lines displayed a similar absolute increase in length, but the vector control showed no increase. This finding indicates that the two GmNAC genes improve salt tolerance (Figure 4d, lower left panel). Together, these results indicate that GmNAC11 and GmNAC20 differentially control root growth and salt tolerance.


We investigated the functions of the two stress-responsive genes GmNAC11 and GmNAC20. Both genes encode nuclear-localized NAC-type DNA-binding proteins. Over-expression of GmNAC20 improved the tolerance to both salt and freezing stresses, probably through the DREB/CBF–COR pathway, while over-expression of GmNAC11 only enhanced salt stress tolerance, possibly through a similar pathway. The roles of the two GmNAC genes in salt tolerance were further corroborated in soybean transgenic hairy roots. Additionally, GmNAC20 promoted lateral root formation through the auxin signaling pathway in transgenic Arabidopsis plants. Therefore, GmNAC20 is a transcription factor that regulates both stress tolerance and root development, whileGmNAC11 primarily regulates salt tolerance.

Various binding elements of NAC proteins have been identified. Olsen et al. (2005a) reported that the consensus binding sequences of ANAC019 and ANAC092 contained a core element CGT[A/G]. Tran et al. (2004) identified the core binding sequence CACG, which is complementary to CGTG. TaNAC69 also showed high binding activity with CGT[A/G][N5][C/T]ACG (Xue, 2005). AtNAM and NAC1 bind to a region containing the CGTA sequence in the 35S promoter (Xie et al., 2000; Duval et al., 2002). GmNAC11 and GmNAC20 also bind to the CGT[A/G] core sequence (Figure 3). However, other sequences can also be bound by NACs. The Arabidopsis transcriptional repressor calmodulin-binding NAC (CBNAC) binds to a GCTT core sequence but not CGT[A/G] (Kim et al., 2007), suggesting that different NAC proteins may bind to different elements depending on NAC protein sequence specificity. GmNAC11 can also bind to the element that is bound by CBNAC but lacks a CGT[A/G] core sequence (Figure 3). Flanking sequences outside the core sequence may also affect the DNA binding affinity of NAC proteins (Xie et al., 2000; Xue, 2005; Kim et al., 2007).

Over-expression of GmNAC20 confers both salt and freezing stress tolerance in transgenic plants, but over-expression of GmNAC11 only improves salt tolerance (Figures 4d, 6 and 7). These results are consistent with the stress-induction patterns of the two genes. Although both genes are induced by salt stress, only GmNAC20 is induced by cold stress; this supports the close relationship between the stress induction of genes and the stress tolerance of the corresponding transgenic plants.

The GmNAC genes may enhance stress tolerance by regulating downstream stress-responsive genes. In the plants over-expressing GmNAC20, expression of DREB1A increased but expression of DREB1C decreased after cold treatment compared to Col-0. Three cold-responsive genes, cor6.6, cor78 and cor15A, were also up-regulated in the transgenic lines at various stages of treatment (Figure 7). DREB1C/CBF2 is a negative regulator of DREB1A/CBF3 expression (Novillo et al., 2004), and DREB1A/CBF3 regulates the expression of COR genes during the stress response (Liu et al., 1998). It is therefore likely that GmNAC20 reduces DREB1C/CBF2 expression, resulting in relatively high expression of DREB1A/CBF3. GmNAC20 may also directly activate DREB1A/CBF3 expression. These two situations have been confirmed in the present study using the protoplast assay; GmNAC20 reduces the activity of the DREB1C promoter, but also enhances the activity of the DREB1A promoter (Figure 7c). Furthermore, GmNAC20 directly binds to the promoter of DREB1A (Figure 7d). Therefore, GmNAC20 activates DREB1A by direct regulation and through suppression of DREB1C to induce COR genes and mediate stress tolerance.

Most NAC proteins have been identified as transcriptional activators; however, NACs such as CBNAC (Kim et al., 2007) have been reported to be transcriptional repressors. Based on the results of this study, GmNAC20 may be a transcriptional repressor (Figure 2). This activity is consistent with its suppression of DREB1C expression (Figure 7a,c). However, GmNAC20 appears to activate DREB1A directly (Figure 7a,c,d), in addition to re-activating DREB1A through the suppression of DREB1C. Therefore, GmNAC20 appears to function as both a transcriptional repressor and a transcriptional activator. The activation activity of GmNAC20 is probably due to the high transcriptional activation activity of its C-terminal domain (Figure 2b). Whether GmNAC20 acts as an activator or a repressor may depend on a conformational change or interactions with other regulatory proteins. In our previous work, we identified a repression domain (NARD) in GmNAC20 (Hao et al., 2010). The ability of the GmNAC20 protein to regulate gene expression may depend on the interplay between the NARD, the C-terminal activation domain and the DNA-binding domain. Because NAC proteins may bind elements as a monomer or a dimer (Figures 3b and 7d), dimerization of NAC proteins may also affect their activity. The high transcriptional activation activity of GmNAC11 (Figure 2b) contrasts with the low induction of DREB1A in GmNAC11 transgenic plants (Figure 7b). This discrepancy may result from the fact that different systems were used for evaluation.

In addition to its roles in stress tolerance, GmNAC20 functions in lateral root development by affecting auxin signaling (Figure 4). In plants over-expressing GmNAC20, ARF2 and AXR1 were down-regulated, whereas ARF7, ARF19, LBD12 and AIR1 were up-regulated (Figure 4). The ARF family of transcriptional factors binds to auxin response elements in the promoters of auxin-responsive genes and regulates the transcription of downstream genes (Guilfoyle and Hagen, 2007). A transcript-null Arabidopsis ARF2 mutant (arf2) and artificial miRNA knockdown mutants of ARF2, ARF3 and ARF4 had longer lateral roots (Marin et al., 2010). Over-expression of mango (Mangifera indica L.) ARF2 inhibits the root growth of Arabidopsis (Wu et al., 2011). An Arabidopsis double mutant lacking ARF7 and ARF19 showed decreased lateral root formation (Okushima et al., 2007). LBD12 was found to be primarily expressed in roots and is involved in the formation of lateral organs (Shuai et al., 2002). The auxin-resistance gene AXR1 encodes an E1 ubiquitin-activating enzyme and can also regulate root growth in Arabidopsis (Leyser et al., 1993). NAC1 regulates AIR1 to control root initiation (Xie et al., 2000). It is thus likely that soybean GmNAC20 promotes lateral root formation by controlling auxin signaling and the regulation of lateral organ-related genes.

Many genes have been previously found to be involved in stress tolerance, but these genes are seldom involved in both stress tolerance and plant development. GmNAC20 regulates stress tolerance and lateral root development through control of the DREB/CBF–COR pathway and the auxin signaling pathway, respectively. However, it is not known how the two functions are achieved, and whether there is any cross-talk between the two pathways. Our previous study showed that ethylene signaling regulates plant growth and stress responses (Cao et al., 2008). Over-expression of the tobacco ethylene receptor gene NTHK1 led to salt sensitivity, enhanced rosette size and a late flowering phenotype (Cao et al., 2006, 2007). The kinase domain and kinase activity of the ethylene receptor differentially regulate these responses (Zhou et al., 2006; Chen et al., 2009). Another NAC protein gene, AtNAC2, can be induced by salt stress, and this induction requires the ethylene signaling pathway. AtNAC2 over-expression promotes lateral root formation (He et al., 2005). The product of this gene also regulates plant senescence, and its mutation leads to a higher seed germination rate under high salinity (Kim et al., 2009; Balazadeh et al., 2010). More recently, the H2O2-responsive NAC protein ORS1 was found to control senescence in Arabidopsis (Balazadeh et al., 2011). Some proteins that play roles in both the auxin and drought response pathways have also been identified (Leymarie et al., 1996).

The role of GmNAC11 in salt tolerance was confirmed in soybean transgenic hairy roots (Figure 4d). However, the role of GmNAC20 in salt tolerance was limited in this over-expression system (Figure 4d, lower panels), probably due to the high endogenous expression of the gene in soybean roots (Figure 1b, right panel). Further analysis using the RNAi approach may reveal the functions of GmNAC20 in stress tolerance, and the roles of the two proteins in cold and drought stresses should also be elucidated. In soybean transgenic roots, the two GmNAC proteins play differential roles in the regulation of root growth under normal conditions (Figure 4d, lower left panel), but only GmNAC20 contributes to lateral root formation in Arabidopsis transgenic plants (Figure 4a,b). This finding suggests a possible difference in the functions of these proteins in the different transgenic systems.

In summary, we have identified two stress-responsive GmNAC genes from soybean. Over-expression of the two genes enhanced tolerance to abiotic stresses and affected lateral root formation or root growth in transgenic plants. Further studies should focus on the roles of these genes in soybean plants and their potential manipulation to improve agronomic traits.

Experimental Procedures

Plant materials, stress treatments and RNA analysis

Seeds of soybean (Glycine max, cultivar NanNong1138-2, kindly provided by Professor Gai Jun-Yi of Nanjing Agricultural University, National Soybean Improvement Center, China) were germinated on moistened gauze at 25°C after immersion in water for 2 days at 37°C, and then grown hydroponically on gauze in Petri dishes at 25°C under continuous light (approximately 2500 lux). Four-week-old seedlings were subjected to various treatments. For salt, ABA and NAA treatments, the seedling roots were immersed in solutions containing 200 mm NaCl, 100 μm ABA or 20 μm 3-NAA, and maintained for the indicated times at 25°C. For cold treatment, seedlings were kept at 4°C for the indicated times. For drought treatment, seedlings were transferred onto filter paper, and dried at 25°C with 60% humidity. Leaves were harvested at the indicated times after initiation of each treatment. Other soybean organs were also collected. RNA extraction and Northern blot analysis were performed as described by Zhang et al. (1996).

Cloning of the GmNAC genes

After screening stress-responsive NAC-like EST sequences, two multiple stress-responsive GmNAC genes, GmNAC11 and GmNAC20, were chosen for further analysis. The full-length opening reading frames of the two genes were obtained by RT-PCR with soybean RNA, and cloned into the pMD18-T vector (Takara, The primers for GmNAC11 were 5′-CATCATTTAGCTAGCTAGCC-3′ and 5′-GTCCGACTTAATCTTTTGATA-3′. The primers for GmNAC20 were 5′-CTTTCCCCCAATTTTCTTTCTC-3′ and 5′-CATTCACTCAGTCTCGTGCTTC-3′.

Plasmid construction for localization analysis

35S-GmNAC-GFP was generated using pBI221, with expression driven by the 35S promoter. GmNAC genes were obtained by PCR using primers without a stop codon. Primers 5′-TAGGGGATCCATGGGAAACCCAGAATCC-3′ and 5′-TATTGTCGACATATCCTTGAAATTGAAG-3′ were used for GmNAC11, and primers 5′-TTTCGGATCCATGGCCGCAGCAACACAACTCC-3′ and 5′-GTGGTCGACACAGAAGGGCCTGGAGAGGTAC-3′ were used for GmNAC20. Then the GmNAC genes were cloned upstream of GFP. A plasmid containing the GFP gene driven by the 35S promoter was used as a control. Each plasmid (10 μg) was transformed into Arabidopsis mesophyll protoplasts. After culturing for 20 h, the GFP fluorescence was visualized under a confocal microscope (Olympus FV500,

Transcriptional activation in protoplast assay

The GAL4 reporter plasmid was generated from pUC19 containing the firefly LUC reporter gene driven by the minimal TATA box of the 35S promoter plus five GAL4 binding elements (Ohta et al., 2001). The GAL4–DRE reporter plasmid contained four DRE binding elements and five GAL4 binding elements upstream of the TATA box. As an internal control, we used the AtUbiquitin3 promoter instead of the 35S promoter to drive expression of the Renilla LUC gene. For effecter gene constructs, coding regions amplified by PCR were fused with or without the GAL4 binding domain. Arabidopsis DREB1A was included to test the suppression activity of the two GmNACs (Figure 2c). For the experiment whose results are presented in Figure 7(c), Arabidopsis DREB1A and DREB1C promoters were used to drive the LUC reporter gene. Other procedures are as described previously (Hao et al., 2010).

Gel-shift assay

The GmNAC11 and GmNAC20 coding regions were amplified and cloned into the BamHI/EcoRI sites of the pGEX4T-1 vector containing a GST tag, using primers 5′-TAGGGGATCCATGGGAAACCCAGAATCC-3′ and 5′-TATTGAATTCTTATCCTTGAAATTGAAG-3′ for GmNAC11, and primers 5′-TTTCGGATCCATGGCCGCAGCAACACAACTCC-3′ and 5′-GCGTGGAATTCTCAGAAGGGCCTGGAGAGGTAC-3′ for GmNAC20. The GST–GmNAC fusion proteins were expressed in Escherichia coli (BL21) and purified using Glutathione Sepharose 4B (GE). Oligonucleotides and their reverse complementary oligonucleotides were synthesized, and the sequences are shown in Figure 3. Double-stranded DNA was obtained by heating oligonucleotides at 70°C for 5 min, and annealing at room temperature in 50 mm NaCl solution. The gel-shift assay was performed as described previously (Wang et al., 2005) using radiolabeled or digoxigenin-labeled probes. For the experiment whose results are presented in Figure 7(d), the Arabidopsis DREB1A promoter sequence was used.

Generation of GmNAC transgenic Arabidopsis plants

The coding sequences of GmNAC11 and GmNAC20 were amplified by RT-PCR using primers 5′-TAGGGGATCCATGGGAAACCCAGAATCC-3′ and 5′-TATTGGTACCTTATCCTTGAAATTGAAG-3′ for GmNAC11 and primers 5′-TTTCGGATCCATGGCCGCAGCAACACAACTCC-3′ and 5′-GCGTGGGTACCTCAGAAGGGCCTGGAGAGGTAC-3′ for GmNAC20, and cloned into the BamHI/KpnI sites of pBIN438 under the control of the CaMV 35S promoter, resulting in pBIN438-GmNAC constructs. The two constructs were confirmed by sequencing and then transformed into Arabidopsis Col-0 plants by the vacuum infiltration method (Bechtold and Pelletier, 1998). Three GmNAC20 transgenic lines (20a, 20b and 20c) and two GmNAC11 transgenic lines (11a and 11b) with various expression levels of the GmNAC genes were selected for further analysis.

Performance of transgenic lines under stress treatments

Homozygous T3 seeds of the transgenic lines were used for phenotypic analysis. For high-salinity treatment, seeds of wild-type and transgenic plants were plated on half-strength MS agar medium. Plates were kept at 4°C for 3 days, and then incubated in a growth chamber under continuous light at 23°C. Six-day-old seedlings were transplanted on half-strength MS agar medium containing 150 mm NaCl. After 10 days, treated and non-treated seedlings were transferred onto vermiculite soil in pots saturated with one-third-strength MS for recovery from the salinity stress. After 15 days, the survival rate and rosette diameter of wild-type and transgenic plants were measured. Each sample contained 20–30 seedlings, and the experiments were repeated four times.

For freezing treatment, the seeds of wild-type and transgenic plants were sown on vermiculite soil in pots. Three-week-old seedlings were kept at various freezing temperatures. For the non-acclimation experiment, all plants were kept at 4°C for 3 h, then at −2, −4, −6 or −8°C, respectively, for 12 h. After treatment, the plants were kept at 4°C for 3 h. For acclimation experiments, seedlings were first kept at 4°C for 4 days, then at −7°C for 12 h, and then at 4°C for 3 h. All treated seedlings were subsequently cultured in a 23°C chamber with uninterrupted light, and the survival rate of all lines was measured 6 days later.

Relative electrolyte leakage was measured as described by Cao et al. (2007).

For lateral root measurements in Figure 4, 4-day-old seedlings were grown on vertical MS plates with or without 100 mm NaCl for 9 days, and the lateral root numbers were counted.

GmNAC-regulated gene expression analysis

Northern hybridizations were performed to assess downstream gene expression in GmNAC20 transgenic Arabidopsis plants (Figures 4 and 7). Primers used for template synthesis are listed in Table S1.

Total RNAs of GmNAC11 transgenic plants and Col-0 were used to examine the expression of six stress-related genes (DREB1A, ERD11, ERF5, RAB18, KAT2 and Cor15A) by real-time quantitative PCR. The corresponding specific primers were 5′-AGGAGACGTTGGTGGAGGCT-3′ and 5′-ACGTCGTCATCATCGCCGTC-3′ for DREB1A, 5′-CCCCTTTGGTAAAGTTCC-3′ and 5′- ATGTCCTTGCCAGTTGAG-3′ for ERD11, 5′-TTGAAGACGGAACAGAGC-3′ and 5′-AGGAGATAACGGCGACAG-3′ for ERF5, 5′-GCATAGACTTTGCTCGGGAGT-3′ and 5′-CCGCCAGACGAACCTTCA-3′ for RAB18, 5′-TGATAATCCTTCCTGCTT-3′ and 5′-CATCATCTATTTCTGCGTTT-3′ for KAT2, and 5′-CAGTTCGTCGTCGTTTCT-3′ and 5′-CCAATGTATCTGCGGTTT-3′ for Cor15A. Primers 5′-TGCAAGGAGGAGCACAAGAGAGC-3′ and 5′-TCCGGCACAGAACCCAGTCGT-3′, and 5′-GGCGGCGGATTGCCTGTACT-3′ and 5′-CCTGGAGCGGCGACATCTGA-3′, respectively, were used to assess GmNAC11 and GmNAC20 expression in various soybean organs. Real-time PCR was performed on an MJ PTC-200 Peltier thermal cycler (MJ Research, now Bio-Rad). Each experiment had four replicates and was repeated twice.

Agrobacterium rhizogenes-mediated transformation of soybean hairy roots

GmNAC11 and GmNAC20 over-expression vectors were introduced into Agrobacterium rhizogenes strain K599, and the bacterium was used to infect soybean (Kefeng No. 1) hypocotyls by injection (Kereszt et al., 2007). After 14 days of growth at 28°C with a photoperiod of 16 h/8 h (light/dark), hairy roots were generated at the infected site and the original main roots were removed. After recovery in water for 3 days, the hairy roots were immersed in 100 mm NaCl or water for 3 days. The root length before and after the treatments was measured for each root, and the net increase of each root in NaCl versus the mean net increase for 40 roots in water was used as a measure of the relative growth rate of each individual root. Each data point is the mean for 40 salt-treated transgenic roots from approximately ten plants. Transgene expression in the transgenic roots was determined by real-time quantitative PCR or checked for the presence of the transgene by PCR.

Statistical analysis

The data were subjected to anova analysis or Student’s t-test analysis using spss 11.5 (SPSS Inc., USA).


We are grateful to Prof. P.M. Gresshoff (The University of Queensland, Australia) for kindly providing the K599 strain. This work was supported by the National Key Basic Research Projects, the National Transgenic Research Project and State Key Lab of Plant Genomics.