Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Tandem zinc finger (TZF) proteins are characterized by two zinc-binding CCCH motifs arranged in tandem. Human TZFs such as tristetraproline (TTP) bind to and trigger the degradation of mRNAs encoding cytokines and various regulators. Although the molecular functions of plant TZFs are unknown, recent genetic studies have revealed roles in hormone-mediated growth and environmental responses, as well as in the regulation of gene expression. Here we show that expression of AtTZF1 (AtCTH/AtC3H23) mRNA is repressed by a hexokinase-dependent sugar signaling pathway. However, AtTZF1 acts as a positive regulator of ABA/sugar responses and a negative regulator of GA responses, at least in part by modulating gene expression. RNAi of AtTZF1–3 caused early germination and slightly stress-sensitive phenotypes, whereas plants over-expressing AtTZF1 were compact, late flowering and stress-tolerant. The developmental phenotypes of plants over-expressing AtTZF1 were only partially rescued by exogenous application of GA, implying a reduction in the GA response or defects in other mechanisms. Likewise, the enhanced cold and drought tolerance of plants over-expressing AtTZF1 were not associated with increased ABA accumulation, suggesting that it is mainly ABA responses that are affected. Consistent with this notion, microarray analysis showed that over-expression of AtTZF1 mimics the effects of ABA or GA deficiency on gene expression. Notably, a gene network centered on a GA-inducible and ABA/sugar-repressible putative peptide hormone encoded by GASA6 was severely repressed by AtTZF1 over-expression. Hence AtTZF1 may serve as a regulator connecting sugar, ABA, GA and peptide hormone responses.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
A genome-wide analysis of CCCH zinc finger proteins revealed the presence of 68 genes in Arabidopsis and 67 genes in rice (Wang et al., 2008). Sub-family IX of Arabidopsis and sub-family I of rice are characterized by a plant-unique tandem CCCH zinc finger (TZF) motif (Wang et al., 2008; Pomeranz et al., 2010a,b). The functions of most Arabidopsis TZF (AtTZF) genes are unknown, with the exception of PEI1, AtSZF1/AtSZF2 and SOMNUS, which are involved in embryogenesis (Li and Thomas, 1998), the salt-stress response (Sun et al., 2007) and light-dependent seed germination (Kim et al., 2008), respectively. Loss-of-function of SOMNUS (AtTZF4) resulted in reduced levels of ABA and elevated levels of GA, ascribed to differences in expression of ABA and GA metabolic genes (Kim et al., 2008). The expression of AtSZF1 and AtSZF2 was induced by salt; loss-of-function plants are hypersensitive to salt stress, while gain-of-function plants are salt-tolerant (Sun et al., 2007). In general, the genes of sub-family IX are responsive to salt, osmotic stress, cold and ABA (Wang et al., 2008). OsDOS is an Arabidopsis TZF homolog in rice. OsDOS affects the onset of plant senescence via jasmonic acid (JA) response pathways (Kong et al., 2006).
The Arabidopsis thaliana gene TZF1 (AtTZF1/AtCTH/AtC3H23) was identified as a glucose-responsive gene in a transcriptome analysis (Price et al., 2004). AtTZF1 binds both DNA and RNA in vitro, and traffics between the nucleus and discrete cytoplasmic foci, similar to stress granules or processing bodies (Pomeranz et al., 2010a,b). Here we show that AtTZF1 positively affects ABA/sugar responses and negatively affects GA responses. Ectopic expression of AtTZF1 not only delays plant growth and development, but also enhances stress tolerance, resembling the effects of GA deficiency and ABA over-accumulation. However, neither effect is found in over-expression plants, suggesting that downstream hormone responses are affected. Consistent with this notion, gene expression profiling showed that AtTZF1 mainly affects ABA- and GA-responsive genes. We propose that AtTZF1 plays a role in mediating ABA- and GA-dependent growth and stress responses by affecting gene expression.
Hexokinase-dependent glucose repression of AtTZF1
RNA gel-blot analyses were performed to characterize the response of AtTZF1 to sugar. Seven-day-old seedlings were used, as described in Experimental procedures. In contrast to traditional sugar-responsive genes (Jang et al., 1997), AtTZF1 was rapidly repressed by as little as 0.1% glucose (Figure 1a,b). As hexokinase (HXK) is important for sugar signaling (Jang et al., 1997; Moore et al., 2003), we used a variety of sugars (Xiao et al., 2000) to determine the role of HXK in sugar repression of AtTZF1. The results showed that the preferred substrates of HXK, including glucose and mannose, triggered strong repression, whereas poor substrates of HXK, such as mannitol and 3-O-methyl glucose (3-OMG) and non-permeable l-glucose, did not (Figure 1c). Because ABA acts downstream of sugar signaling (León and Sheen, 2003), we determined whether ABA was also involved. The results showed that ABA alone did not affect AtTZF1 expression, and glucose repression remained unchanged in the presence of ABA (Figure 1d). To further confirm these results, we determined AtTZF1 expression in mutants with abnormal sugar responses (Jang et al., 1997; Arenas-Huertero et al., 2000; Cheng et al., 2002; Moore et al., 2003). Consistent with the idea that HXK can affect AtTZF1 expression, glucose repression of AtTZF1 was reduced in HXK loss-of-function plants, including antisense HXK1, antisense HXK2 and the HXK1 knockout mutant gin2 (Figure 1e). In contrast, glucose repression was unaffected in the ABA signaling mutant abi4-1 and the ABA biosynthetic mutant aba2-1, suggesting that ABA is not directly involved in glucose repression of AtTZF1.
AtTZF1 over-expression and RNAi plants
Reverse genetic analyses were performed to reveal the effects of AtTZF1 on plant growth, development and sugar/hormone responses. We previously showed that an AtTZF1–GFP fusion protein moved between the nucleus and putative stress granules or P-bodies (Pomeranz et al., 2010a,b). However, because the T-DNA knockout mutants had no obvious phenotypes, it was not possible to determine whether the AtTZF1–GFP fusion protein was functional by complementation. Hence both CaMV 35S:AtTZF1-GFP and CaMV 35S:AtTZF1 constructs were used for over-expression analysis. The rosettes of at least six over-expression lines (OEs) were more compact than those of the wild-type (WT). Two transgenic lines for each construct (Figure 2a,b) were selected for subsequent characterization. Although OE4 showed a high level of expression of AtTZF1–GFP, it was not used for further phenotypic analyses due to homozygous lethality with unknown reasons. No discernible phenotypes were found in AtTZF1 T-DNA knockout plants (data not shown), probably due to functional redundancy of the gene family (Pomeranz et al., 2010a,b). To circumvent this problem, multi-gene RNAi plants were generated. Based on the results of meta-profile (http://www.genevestigator.com/gv/index.jsp) and e-Northern (http://bbc.botany.utoronto.ca/) analyses, AtTZF1, 2 and 3 are expressed at the highest levels within AtTZF family, and they clustered together in the baseline developmental series expression analysis. Furthermore, expression of AtTZF2 and 3 was also repressed by glucose (Price et al., 2004; Hruz et al., 2008). Hence, triple AtTZF1, 2 and 3 RNAi plants were generated (see Experimental procedures). These RNAi plants showed strong reduction of expression of AtTZF1–3 (Figure 2c), and are considered as loss-of-function plants.
AtTZF1 affects plant growth and development
When grown on soil, OE plants were compact, with scallop-shaped leaves due to defects in apical–basal leaf expansion (Figure 3a,e). Overall, the cellular organization was similar between WT and OE plants (Figure 3b). However, mesophyll cells in OE plants were smaller and irregular, suggesting defects in cell expansion and cell polarity, respectively (Figure 3c, compare OE1 with WT). The spongy mesophyll cells of OE1 expanded preferentially in the lateral direction, causing the leaf to become wider than WT in later stages of development (Figure 3d). Vascular bundles were smaller and under-developed in 3-week-old OE leaves (Figure 3c), indicating defects and/or delay in cell division and differentiation. By 4 weeks, the cell expansion defects gradually disappeared in OE plants. As a result, OE leaf expansion recovered in both the apical–basal and lateral directions. After 6 weeks, OE leaves were wider than those of the WT due to enhanced lateral expansion (Figure 3d). Compared to the WT, mature OE leaves were dark green, wide and wrinkled. There were strong dosage effects of ectopic expression on rosette vegetative growth. In hemizygous (OE/WT) plants, leaf expansion recovered in approximately 4 weeks and leaves continued to grow until the early flowering stages (Figure 3d,f). Due to late flowering and an extended vegetative phase, the hemizygous (OE/WT) plants weighed twice as much as WT plants by 6 weeks (Figure 3f). The rosette expansion recovered much later in homozygous (OE/OE) plants. Nevertheless, they eventually reached a biomass comparable to or greater than that of WT and hemizygous plants. The RNAi plants did not show any discernable phenotypes in terms of rosette growth (data not shown).
Because GA and brassinosteroid deficiencies are two common causes of compact phenotypes, plants were sprayed twice daily with GA (100 μm) or brassinolide (0.1 μm) for 3 weeks. OE plants were partially rescued by spraying with GA (Figure 3g), but not with brassinolide (not shown). The GA treatment was shown to be effective, as the GA-deficient mutant ga1-3 was fully rescued. These results imply that both GA accumulation and GA responses may be compromised in OE plants. Consistent with this, OE plants had produced small flowers and siliques (Figure 3h,i) and compact inflorescences with short internodes (Figure 3j).
The aerial parts of plants are covered by cuticle, an extracellular matrix composed of cutin and waxes that is important for plant protection, gas exchange and water retention. The cuticle is also important in preventing fusion of plant surfaces (Sieber et al., 2000; Pollard et al., 2008). Rosette leaf fusion occurred occasionally in OE lines (data not shown), but organ fusion between inflorescence stems and cauline leaves took place more frequently (Figure 3j). Analysis of tissue cross-sections revealed duplication of parenchyma rays associated with thickening of the inflorescence stem (Figure 3k). Organ fusion in OE lines may partly be due to down-regulation of glycerol-3-phosphate acyltransferase 8 (GPAT8, At4g00400) and 3-ketoacyl-CoA synthase 1 (KCS1, At1g01120) (Table S1), two key enzymes in cutin biosynthesis (Li et al., 2007). Curiously, mutation of a core subunit of exosome, a component of the 3′→5′ mRNA degradation machinery, also caused defects in cuticle biosynthesis (Hooker et al., 2007).
AtTZF1 over-expression causes late flowering
In addition to growth phenotypes (Figure 3), OE plants were late flowering. Under long-day conditions (16 h light), OE plants flowered much later (weeks 6–7) compared with WT and RNAi plants (week 4) (Figure 4a,b). To determine whether late flowering is associated with altered expression of flowering regulators, we examined the expression of FLOWERING LOCUS C (FLC), FLOWERING LOCUS T (FT), and SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1) (Lee et al., 2000; Baurle and Dean, 2006). Consistent with the role of FT as a florigen (Turck et al., 2008), its level was reduced in OE1 plants (Figure 4c). Similarly, expression of the central flowering promoter SOC1 was also reduced in OE1 plants. By contrast, expression of the major flowering repressor FLC (Michaels and Amasino, 1999; Alexandre and Hennig, 2008) was higher in OE1 plants (Figure 4c). Together, these results indicated that the late-flowering phenotype correlates with changes in expression of flowering regulators. Because GA is an effective inducer of SOC1 (Moon et al., 2003), we tested whether GA or brassinolide treatment could rescue the late-flowering phenotype. GA only partially rescued the late-flowering phenotypes (Figure 4d), indicating that OE1 plants may show reduced GA accumulation and responses.
AtTZF1 affects ABA, sugar and GA responses during seed germination
It is well documented that GA and ABA act antagonistically in the regulation of seed germination. As OE plants showed GA deficiency/insensitive growth phenotypes, assays were performed to determine whether OE and RNAi plants had altered responses to ABA/GA during seed germination. In the absence of exogenous sugars, except for the sugar/ABA-hypersensitive mutant ein2-1, germination rates were similar for WT, OE and RNAi plants (Figure 5a). However, whereas OE plants and ein2-1 germinated more slowly than WT in the presence of 0.1 μm ABA, RNAi plants and the ABA/sugar-insensitive mutant ctr1-1 germinated faster (Figure 5b). Similar trends were found when higher concentrations of ABA were used (data not shown). These results suggest that AtTZF1 enhances ABA sensitivity in seed germination. The altered ABA sensitivities suggested the possibility of abnormal sugar responses, due to intimate cross-talk between ABA and sugar signaling (León and Sheen, 2003; Gibson, 2004; Rolland et al., 2006; Rook et al., 2006). Consistent with this, although OE plants and ein2-1 germinated more slowly than WT in the presence of 2% glucose, RNAi plants and ctr1-1 germinated faster (Figure 5c). The OE plants showed an even more severe delay of in seed germination under higher glucose concentrations (4 and 6%) (data not shown). We then determined whether the glucose-induced delay of seed germination could be rescued by GA application. Intriguingly, although GA was able to promote seed germination in WT and ctr1-1, germination of OE plants and ein2-1 was still delayed within the first 2 days (Figure 5d). These results indicate that the OE plants were either compromised in GA sensing or defective in seed germination for unknown reasons such as phytochrome-mediated responses (Kim et al., 2008).
AtTZF1 affects abiotic stress responses
Other than the growth and germination phenotypes, OE plants showed enhanced tolerance to abiotic stresses. In particular, OE plants were drought-tolerant when water was withheld for 10 days (Figure 6a). The drought tolerance phenotype maybe partly due to abnormal stomatal closure (Figure 6b,c) and lower stomatal conductance (Figure 6d). In addition, OE plants showed reduced water loss (Figure 6e) and RNAi plants showed slightly, but significantly, higher rates of water loss than WT (Figure 6f). Coincidently, we previously found that localization of AtTZF1–GFP in putative stress granules or P-bodies was specifically induced by stress hormone methyl jasmonate in stomata (Pomeranz et al., 2010a,b), although the role of AtTZF1 in regulating stomatal activity remains to be determined. As stomatal closure is an integral response in both drought and cold stress (Assmann and Wang, 2001; Luan, 2002), a series of freezing assays were performed. Under the conditions used, both OE and WT plants tolerated freezing temperatures (−9 to −11°C), and the survival rate was almost 100% if previously acclimatized at 4°C for 48 h (data not shown). However, the survival rate of OE plants was significantly higher than that of WT under non-acclimatized conditions (Figure 6g,h), indicating enhanced cold tolerance. The cold tolerance phenotypes were consistent with enhanced expression of the ABA/cold tolerance marker genes COR15A, KIN1, and RD29A (Figure 6i and Table S2) under non-acclimatized conditions. Furthermore, the stress tolerance phenotypes appeared to be independent of developmental stages, as the results were consistent for plants from 3 to 6 weeks old (data not shown). On the basis of these results, we propose that AtTZF1 can enhance abiotic stress tolerance.
Growth and stress phenotypes in relation to ABA, GA and sugar accumulation
To determine whether phenotypes of OE and RNAi plants were associated with abnormal hormone accumulation, ABA and GA levels were determined in 3-week-old plants. Surprisingly, the levels of both ABA and GA4 (the most active GA compound) were similar in WT, OE and RNAi plants (Figure 7a). Hormone levels were also determined at various developmental stages (Figure 7b). Although OE plants resembled GA-deficient mutants (Figure 3), the GA4 levels were actually higher in OE1 than in WT plants. This was also true for 10- and 14-day-old plants of lines OE1–OE4 (data not shown). The higher GA levels in OE plants correlated with their long vegetative growth phase and the larger biomass of the mature plants (Figure 3d,f). As ABA is a critical stress hormone, its level was also determined under drought and cold stress (Figure 7c). RNAi plants were not included as they only showed mild susceptibility to these stresses (data not shown). The ABA levels were negligible, and were similar between WT and OE1 plants under both well-watered and mild drought conditions. However, ABA accumulation was highly induced under severe drought conditions, and the levels were higher in the WT. This initially appeared to be significant. However, it may be a consequence of differential drought responses, during which WT plants wilted severely and OE1 plants remained relatively fresh (Figure 6a). Hence, quantification of ABA based on fresh weight may be biased toward much higher ABA levels in the WT. To determine whether ABA levels are predictors of plant performance under freezing stresses, ABA levels were determined at the pre-treatment stage (−2°C for 12 h, see Experimental procedures). As shown in Figure 7(c), acclimatization induced ABA accumulation in both WT and OE1 plants. However, the ABA levels were lower in OE1 than in WT plants, independently of acclimatization. In summary, the OE plants did not contain higher levels of ABA under drought or freezing conditions. This implies that the stress tolerance phenotypes are unlikely to be due to higher accumulation of ABA.
Brassinosteroid deficiency may also cause compact rosette phenotypes. Detailed measurements of brassinosteroid levels were performed using GC-MS (Fujioka et al., 2002). Overall, the brassinosteroid levels were similar between OE1 and WT plants (Figure S1). Together, these hormone analysis results indicate that the phenotypes of OE and RNAi plants are not entirely correlated with abnormal accumulation of ABA, brassinosteroids or GA.
As sugars are both metabolites and signaling molecules, their accumulation affects plant growth and stress responses (Rolland et al., 2006; Smeekens et al., 2010). To determine sugar accumulation, both plate- and soil-grown plants were used. For 1-week-old plants grown on MS plates supplemented with 2% sucrose, the sucrose levels were similar across WT, OE and RNAi plants. However, whereas glucose levels were slightly lower, starch levels were significantly higher in OE than in WT and RNAi plants (Figure S2a,b). These assays were then repeated using 3-week-old soil-grown WT and OE plants. Once again, glucose and fructose levels were lower but sucrose and starch levels were slightly higher in OE than in WT plants (Figure S2c,d). To avoid differences resulting from developmental variations, samples from the same leaf positions were analyzed and similar trends were found (data not shown). In summary, although there were some differences in sugar and starch levels, these results did not explain why OE plants are sugar-hypersensitive and stress-tolerant, as they did not accumulate abnormally high levels of soluble sugars.
Ectopic expression of AtTZF1 causes coordinated changes in gene expression
To explore whether distinct gene expression changes are associated with OE plants, a microarray analysis was performed using Affymetrix GeneChips® (http://www.affymetrix.com/). Four independent OE lines (3 weeks old, shown in Figure 2a) were used for pairwise comparison with the WT. Previously established procedures were used for sample preparation and array processing (Price et al., 2004). Affymetrix Microarray Suite 5.0 (http://www.affymetrix.com/) and Venn diagrams were used for data processing and analysis (see Experimental Procedures). Genes with consistent changes for all four comparisons were identified (Figure S3), and stringent statistical cut-offs were applied. The analysis generated a list of 59 down-regulated and 46 up-regulated genes in all four OE lines (Tables S1 and S2). To determine whether these genes were co-regulated, gene network analyses based on correlation co-efficiencies were performed (Obayashi et al., 2007). For the 59 down-regulated genes, three networks were identified, including a large network (Figure 8a) involved in cell growth, and a smaller network (data not shown) that contained four sugar-responsive genes (At1g66760, At2g17880, At4g35770 and At5g57760) (Price et al., 2004). Using a less stringent cut-off, i.e. a ≥ twofold change in three of the four pairs of comparison, a list of 81 down-regulated genes was obtained (Table S3). Using this longer list, gene network analysis identified a sugar-responsive cluster of six genes, including the sugar-sensitive marker ASN1/DIN6 (At3g47340) (Palenchar et al., 2004; Baena-Gonzalez et al., 2007). These results imply that AtTZF1 may also affect sugar responses at gene expression level.
The cell growth gene network (Figure 8a) consisted of 14 genes, including three expansins (EXPA3, 8 and 11), two pectate lyases (At3g07010 and At5g48900), two lipases (At4g18970 and At5g45950) and a cutin biosynthetic acyltransferase (At4g00400), most of which are involved in cell-wall modification (Marin-Rodriguez et al., 2002; Cosgrove, 2005; Li et al., 2007). Interestingly, GA-Stimulated Arabidopsis 6 (GASA6, At1g74670) and brassinosteroid-insensitive mutant suppressor 1 (BRS1, At4g30610) were also found in this cluster. GASA6 is thought to be a secreted peptide hormone precursor (Roxrud et al., 2007), and BRS1 is a secreted endopeptidase that is involved in trimming peptide precursors (Li et al., 2001; Zhou and Li, 2005). As the cell growth gene network contained several genes involved in GA-mediated cell expansion processes (Lee and Kende, 2002), we used GASA6 as a single input to perform co-expression analysis using a public collection of 1388 GeneChip data sets (http://www.atted.jp/). Interestingly, these results also indicated that the same expansins (EXPA3, 8 and 11) and pectate lyases (At3g07010 and At5g48900) are tightly connected with GASA6 (Figure 8b), suggesting that the GASA6 co-expression network is affected by over-expression of AtTZF1. In addition to its involvement in the GA response, GASA6 has also been characterized as an in vivo sugar marker gene, together with ASN1/DIN6, among others (Gonzali et al., 2006).
For the 46 up-regulated genes, two major networks were identified by co-expression analyses (Figure S4). Interestingly, seven of the ten genes in one network encoded chloroplast-localized proteins, many of which were involved starch metabolism. Remarkably, up-regulation of this gene cluster was positively correlated with higher accumulation of starch in OE plants (Figure S2). Another network contained 13 genes, most of which were salicylic acid-inducible defense-related genes. A small cluster containing the cold-tolerance genes COR15a and KIN1 was also identified. Hence the up-regulated genes appear to be involved in both biotic and abiotic stress responses.
Over-expression of AtTZF1 mimics the effects of ABA or GA deficiency on gene expression
RNA gel-blot and quantitative RT-PCR analyses were used to confirm the results of the microarray analyses. As shown in Figures 9(a) and S5, genes in the cell growth gene network (GASA6, AtEXP8, pectate lyase, BRS1 and POT) and the sugar response marker gene network (SEN1 and ASN1) were down-regulated in OE lines. By contrast, genes involved in cold (KIN1) and drought (ERD11) responses were up-regulated in OE lines. The extensive co-expression suggested that genes that are up- or down-regulated in OE lines may be co-regulated by specific internal or external cues. To test this hypothesis, we used Genevestigator meta-profile analysis (Zimmermann et al., 2005) to reveal stimuli response profiles of these genes. Interestingly, genes down-regulated in OE lines were generally repressed by ABA and/or paclobutrazol (an inhibitor of GA biosynthesis). By contrast, genes up-regulated in OE lines were generally induced by ABA (Figure 9b) and salicylic acid (data not shown). Upon examining the expression levels of ABA/GA-response marker genes (Goda et al., 2008), we found that the ABA marker genes RD29A and COR15A were up-regulated, whereas the GA marker genes AtEXP1 and GAI were down-regulated in OE lines (Figure S6), suggesting that AtTZF1 mimicked the effects of ABA or GA deficiency on gene expression. As there were only small differences in ABA and GA levels between the WT and OE lines (Figure 7a, 21-day-old plants), the gene expression changes were probably due to enhanced ABA and reduced GA responses.
Although GASA6 was shown to be a GA-inducible and ABA-repressible gene, high concentrations of hormone were used in previous analyses (Zhang and Wang, 2008). The results of our RNA gel-blot analyses showed that both GASA6 and its closest homolog GASA4 (Roxrud et al., 2007; Kryvych et al., 2008) were induced by GA and repressed by ABA at 10 μm concentration (Figure 9c). We then determined the GASA4/6 expression in both 3- and 6-week-old plants. The expression level of GASA6 decreased with age in WT plants, but it expressed at consistently low levels in OE plants. Similar to GASA6, the expression of GASA4 declined in 6-week-old WT plants. However, GASA4 expression declined to a lesser extent in 6-week-old OE plants, probably due to higher GA levels in older OE plants (Figure 7b). It is currently not known why the two GASA genes were differentially expressed in AtTZF1 OE plants, although both are GA-inducible and ABA-repressible.
Human TZFs regulate the expression of cytokines via control of mRNA half-lives (Blackshear et al., 2005). In plants, although the molecular mechanisms are unknown, multiple genetic studies have shown that plant-unique TZF genes are important for both developmental and environmental responses (Li and Thomas, 1998; Kong et al., 2006; Sun et al., 2007; Kim et al., 2008; Grabowska et al., 2009; Guo et al., 2009). We have shown previously that AtTZF1 traffics between the nucleus and cytoplasmic foci, probably P-bodies and stress granules. As AtTZF1 binds both DNA and RNA in vitro, we proposed that it may be involved in transcription in the nucleus and RNA regulation in the cytoplasm (Pomeranz et al., 2010a,b). Here we report that AtTZF1 (AtCTH/AtC3H23) is involved in multiple developmental processes as well as stress responses. Gain- and loss-of-function analyses have shown that AtTZF1 affects ABA/GA responses partly via the modulation of ABA/GA-responsive genes. The results of hormone analyses indicated that the developmental and stress tolerance phenotypes of AtTZF1 over-expression plants were not correlated with predicted ABA and GA level changes, raising the possibility that other regulatory mechanisms may be affected. We have further demonstrated that genes that are down-regulated in AtTZF1 OE lines are members of several co-expression gene networks. The largest gene network is involved in cell growth and centers on GASA6, a GA-inducible and ABA-repressible peptide hormone precursor. Another network is highly enriched with sugar-repressible genes (six of seven genes), implying that AtTZF1 also plays a role in sugar response. The results shown here, together with previous reports (Kim et al., 2008; Wang et al., 2008), indicate that members of the AtTZF family play important roles in maintaining normal ABA/GA responses via the control of gene expression.
Interestingly, the AtTZF1 homolog SOMNUS (SOM/AtTZF4/AtC3H2) acts as a negative regulator of phytochrome-mediated promotion of seed germination. Loss-of-function of SOM/AtTZF4 causes early germination and a decrease in ABA and an increase in GA accumulation, due to its effects on expression of ABA and GA metabolic genes (Kim et al., 2008). Hence SOM/AtTZF4 appears to act as a positive regulator of ABA accumulation and a negative regulator of GA accumulation. Based on their phenotypes, AtTZF1 OE plants are thought to be similar to SOM OE plants, which contain more ABA and less GA than WT. This is consistent with the observation that the GA anabolic gene GA3ox1 (GA4) is down-regulated and the GA catabolic gene GA2ox2 is up-regulated in 3-week-old AtTZF1 OE plants (Figure S4). Surprisingly, the GA levels are slightly higher in OE than WT plants (Figure 7b), raising the possibility that OE plants are compromised in GA responses, and that defective GA signaling could lead to the reduction of negative feedback of GA accumulation (Sun and Gubler, 2004). By contrast, the ABA levels were similar in WT and AtTZF1 OE plants. On the basis of our microarray results, neither the ABA anabolic genes ABA1, ABA2, AAO3, NCED6 and NCED9 nor the ABA catabolic genes CYP707A1 and CYP707A2 show any significant differences in expression between WT and OE plants (data not shown). Again, these results support the idea that the ABA response, but not ABA accumulation, is mainly affected in OE plants. Despite differences in the effects on ABA/GA accumulation, it appears that AtTZF1 and SOM affect similar downstream processes, because both AtTZF1 and SOM/AtTZF4 (J.-C. Jang and P. Bogamuwa, Ohio State University, Columbus, Ohio, USA, unpublished data) caused compact rosette phenotypes when ectopically expressed under the control of the CaMV 35S promoter. In fact, even human TTP may share common biochemical and molecular functions with AtTZF1 because over-expression of human TTP also causes compact rosette and organ fusion phenotypes (P.-C. Lin and J.-C. Jang, Ohio State University, Columbus, Ohio, USA, unpublished data).
Do GASAs act downstream of ABA/GA signal transduction?
Peptide hormones are small, secreted protein molecules that act as ligands recognized by extracellular receptors of various signal transduction pathways. Plants produce a plethora of peptide hormones, most of which are uncharacterized (Matsubayashi and Sakagami, 2006; Farrokhi et al., 2008). A GA-inducible and ABA-repressible peptide hormone GAST1 was first discovered in tomato (Solanum lycopersicum) (Shi et al., 1992). Genes with similar protein sequences have since been found and characterized in several plant species, including Arabidopsis (Herzog et al., 1995; Kotilainen et al., 1999; Ben-Nissan et al., 2004; de la Fuente et al., 2006; Furukawa et al., 2006). Their main functions are to control cell expansion and flower transition, consistent with the function of GA. Although a total of 15 GA-Stimulated Arabidopsis (GASA) genes have been found (Roxrud et al., 2007; Zhang and Wang, 2008), only a subset of them are actually GA-inducible. Their differential responses to various hormones and temporal–spatial expression patterns imply diverse roles in various developmental processes (Zhang and Wang, 2008). Here we have found that GASA4 and GASA6 expression levels are differentially affected in AtTZF1 OE plants. The protein OsDOS is a member of rice TZF sub-family I (Wang et al., 2008). RNAi knockdown of OsDOS caused early senescence, but its over-expression resulted in a delay of leaf senescence. It was proposed that OsDOS acts as a negative regulator of leaf senescence via suppression of methyl jasmonate (MeJA) biosynthetic genes and MeJA responses (Kong et al., 2006). Although AtTZF1 and OsDOS appear to be involved in distinct hormone responses, over-expression of AtTZF1 also causes significant delay of flowering and senescence (Figures 3 and 4). Interestingly, the closest rice homolog of GASA6, Os06g15620, is up-regulated in OsDOS RNAi plants (Kong et al., 2006). This suggests that both AtTZF1 and OsDOS negatively regulate the expression of GASA family genes. It will be interesting to determine how various hormones regulate rice GASA gene expression, and whether OsDOS affects ABA/GA responses.
As expression of both GASA4 and GASA6 is induced by GA and repressed by ABA (Figure 9c), it is surprising that GASA4 is up-regulated and GASA6 is down-regulated in AtTZF1 OE plants (Figure 9d), given that the GA levels in OE1 are slightly higher than in WT (Figure 7b). One possibility is that GASA6, but not GASA4, contains AU-rich elements at the 3′ UTR that may serve as AtTZF1 binding sites. This interaction could lead to degradation of the GASA6 mRNA. However, we were not able to detect binding between AtTZF1 and the GASA6 3′ UTR in an in vitro assay (Pomeranz et al., 2010a,b). Multiple hormones, other than ABA and GA, also affect the expression of GASA4 and GASA6 (J.-C. Jang and J. Qu, Ohio State University, Columbus, Ohio, USA, unpublished data). The differential expression of GASA4 and GASA6 in AtTZF1 OE plants may be due to differential responses of these two genes to other hormones. Hence GASAs appear to be promising candidates of signaling components at downstream convergent points for the actions of ABA, GA and many other hormones. Our finding of altered expression of GASA4 and GASA6 in AtTZF1 OE plants raised the possibility that AtTZF1 may be an upstream regulator of these putative peptide hormones. Hence it is important to determine the molecular mechanisms underlying the control of GASA4 and GASA6 expression by AtTZF1.
Plant material and growth conditions
Plants for sugar-responsive gene expression analyses were prepared as described previously (Kang et al., 2010). Essentially, 6-day-old seedlings were grown in liquid MS medium with 2% sucrose on a platform shaker at 0.643 g at 24°C under constant fluorescent white light (80 μmol m−2 sec−1). Seedlings were then washed five times for 3 min each with sterile distilled water, and incubated in MS liquid without sugar in the dark for 24 h for sugar depletion. A mock or sugar treatment was then applied to the culture for 3 h before sample collection and RNA extraction. Plants were grown using Pro-MixTM BX or Sunshine™ (PREMIER TECH Horticulture, http://www.premierhort.com/eProMix/index.htm) peat-based potting mixture under either 16 h light/8 h dark or 12 h light/12 h dark cycles at 22°C in Conviron® growth chambers (Conviron, http://www.conviron.com) with a light intensity of approximately 120 μmol m−2 sec−1. Unless specified, plants for hormone sensitivity assays were grown on MS plates (MS salts, B5 vitamins, pH 5.7, 0.7% Phytagar® (Invitrogen, http://www.invitrogen.com/), with or without 2% sucrose) at 24°C in either continuous white light (100 μmol m−2 sec−1) or the dark in a tissue culture room.
The germination assays in response to ABA, glucose and GA were performed using two biological repeats. Each experiment included two replicates with 50 seeds equally spaced on MS plates. Except for ctr1-1 and ein2-1, all plants were grown under the same conditions at the same time. WT and RNAi seeds were collected earlier, due to late flowering of AtTZF1 over-expression plants. For biomass measurements, fresh weights of 12 plants of each genotype were measured. For flowering assays, 16 WT plants and 12 plants from each transgenic line were used to measure time to flowering and leaf number before bolting. For RNA gel-blot analysis of expression of flowering regulators, 2-week-old non-flowering seedlings grown under long-day conditions (16 h light) were used. For GA rescue of rosette expansion and flowering, plants were grown under both short-day (12 h light) and long-day (16 h light) conditions. Ten to fifteen plants were used for each treatment in each genotype. For drought tolerance assays, water withholding was applied to 4-week-old plants for 10 days before photographs were taken. Epidermal replicas were taken at midday from the same leaf positions of well-watered 4-week-old plants. Clear nail polish was applied to the abaxial side of the leaf, and the replica film was peeled off immediately for photography using a Nikon Eclipse™ E600 microscope (http://www.nikon.com/). Stomatal aperture and conductance measurements were performed at midday using well-watered 4-week-old plants. Measurements were recorded using a LiCor LI-1600 steady-state porometer (LI-COR Biosciences, http://www.licor.com/) inside the growth chamber. For water loss assays, six plants of each genotype were used, and the experiment was repeated three times with similar results. Briefly, entire rosettes were removed from the roots and immediately weighed and placed on the bench top for gradual drying. Sample fresh weights were then taken at 2 h intervals for 8 h. For freezing tolerance assays (Pennycooke et al., 2008), both WT and transgenic plants grown under short-day conditions (12 h light) were given either an ‘acclimatization treatment’ (4°C for 48 h in a cold room) or a ‘mock treatment’ (48 h at room temperature). The plants were then cooled to −2°C for 12 h to avoid water supercooling before decreasing the temperature to −9, −10 or −11°C in a Percival™ programmable freezer (Percival Scientific, Inc., http://www.percival-scientific.com) at a rate of 1°C per 30 min. Plants were kept at these temperatures for 30 min, and then kept at 4°C in the dark for 24 h before returning to normal growth conditions. Plant survival rates and phenotypes were scored 1 week later.
Molecular cloning, RNA isolation, RNA gel-blot and quantitative RT-PCR analyses
Unless specified, all cloning procedures were performed using GatewayTM technologies (Invitrogen, http://www.invitrogen.com/). Genes were first inserted into pENTR™ vectors and then transferred to various destination vectors via recombination reactions. The binary vectors used for promoter fusions and translational fusions are listed at http://www.psb.ugent.be/gateway/index.php?_app=vector&_act=construct_list_plant&. cDNAs were amplified from either full-length cDNA clones from Arabidopsis Biological Resource Center (http://www.arabidopsis.org/abrc/catalog/cdna_clone_1.html) or the pFL61 cDNA library (Minet et al., 1992) via PCR reactions. The primers used for cloning coding sequences of AtTZF1 (At2g25900) were 5′-ATGATGATCGGCGAAAATAAAAA-3′ and 5′-ACCGAGTGAGTTCTCTCTAC-3′. The RNAi construct was produced by a tandem connection of 3 fragments unique to AtTZF1 (bp 319–405), AtTZF2 (bp 283–369) and AtTZF3 (bp 286–366) via EcoRI, NotI and XhoI linkers, respectively. The entire fragment was then cloned into pENTR3C and subsequently into the pK7GWIWG2(I) RNAi vector via recombination reactions. The intron cloned between the inverted fragments was as described previously (Karimi et al., 2002). Probes used for RNA gel-blot analyses were amplified from either full-length cDNA clones from the Arabidopsis Biological Resource Center or the pFL61 cDNA library (Minet et al., 1992) via PCR reactions. The primers used for these gene probes were 5′-TACGAAAATCCAAGTCCCACT-3′ and 5′-AAACTCGCGAGTGTTGAAGTT-3′ for FT (At1g65480), 5′-TTCTCCAAACGTCGCAACGGTCTC-3′ and 5′-GATTTGTCCAGCAGGTGACATCTC-3′ for FLC (At5g10140), 5′-TGAGGCATACTAAGGATCGAGTCAG-3′ and 5′-GCGTCTCTACTTCAGAACTTGGGC-3′ for SOC1 (At2g45660), 5′-CTCATAACTTCTTTTCTCT-3′ and 5′-TTTGGTCCACCTTGTT-3′ for GASA6 (At1g74670), 5′-CTTTGATCAATGTTTTATCTACT-3′ and 5′-AGAAACACAAGTATTCATGAAAA-3′ for GASA4 (At5g15230), 5′-GCTTTAAGTACGGCTCT-3′ and 5′-AACTGCCAATTAGAAGGA-3′ for ATEXPA8 (At2g40610), 5′-GCTGTTATGGCTTCTAC-3′ and 5′-CTGCAATTAAGAGCACC-3′ for pectate lyase (At5g48900), 5′-GATCAGTACAGCATTTAC-3′ and 5′-ATGATCTGAAAAGAATGA-3′ for BRS1 (At4g30610), 5′-AGCAATGGCAATTTCCG-3′ and 5′-AAAGGACCATACGACCA-3′ for POT (At1g52190), 5′-CCACTGCTTTTAACACAA-3′ and 5′-TCCAAGCGACGTATCC-3′ for SEN1 (At4g35770), 5′-TTGCTCACTTGTACGAG-3′ and 5′-ATTGCTTAGCCGCCTTA-3′ for ASN1 (At3g47340), 5′-GTCAGAGACCAACAAGA-3′ and 5′-TACTTGTTCAGGCCGG-3′ for KIN1 (At5g15960) and 5′-TTCCACAGCCACTAGAA-3′ and 5′-GAAGTGATGTCAGCAAC-3′ for ERD11 (At1g02920). RNA isolation and RNA gel-blot analyses were performed as described previously (Price et al., 2004). The quantitative RT-PCR analyses were performed as described by Morohashi and Grotewold (2009). Briefly, 1 μg of RNA (extracted using a Qiagen RNeasy mini kit; http://www.qiagen.com/) was treated with 1 unit of RQ1 DNase (Promega, http://www.promega.com/) for 30 min. The reaction was stopped using the stop buffer provided, and the enzyme was heat-inactivated at 65°C for 5 min. A small aliquot of the DNase-treated RNA was saved for later PCR reactions to confirm that the sample was free of DNA contamination. The remaining RNA was used in a reverse transcriptase reaction for 2 h using the protocol described by the manufacturer for SuperScript II reverse transcriptase (Invitrogen). The resulting cDNA mixture was purified using standard phenol/chloroform extraction and ethanol precipitation. The cDNA was then re-suspended in ddH20 to a concentration of 20 ng μl−1. Quantitative PCR reactions were performed in a Bio-Rad CFX96 cycler (http://www.bio-rad.com/). Reactions were performed using 10 μl of Bio-Rad IQ pre-mix solution. Each reaction contained 1 μl of the cDNA preparation and 5 pmol of each primer. Primers used for quantitative RT-PCR analyses are listed in Table S4. Relative expression was obtained by normalization to the expression of reference gene PP2A (Czechowski et al., 2005).
Analysis of levels of endogenous hormones, soluble sugar and starch
Plants for hormone analyses were grown under 12 h light/12 h dark cycles at 22°C in a growth chamber. For drought treatments, water withholding was applied to 4-week-old plants for 0 (well-watered), 5 (mild drought) or 10 (severe drought) days. Samples for freezing treatments were prepared as described for the freezing tolerance assays. Analysis of GA and ABA was performed as described previously (Preston et al., 2009; Yoshimoto et al., 2009). Purification and quantification of BRs were performed as described previously (Fujioka et al., 2002). For sugar and starch analyses, 1-week-old plants grown on MS plates supplemented with 2% sucrose and 3-week-old plants grown under 12 h light/12 h dark cycles at 22°C on soil mixture were used. Soluble sugar and starch extraction and quantification were performed using the protocols described previously (Strand et al., 1999; Zhang et al., 2005). Commercial kits (R-biopharm; Roche, http://www.roche.com; catalogue numbers 10 716 260 035 and 10 207 748 035) and UV methods were used to determine the concentrations of glucose, fructose, sucrose and starch. The experiment was repeated twice, each with two or three replicates.
Tissue processing was performed as described previously (Qi and Ding, 2003). Tissue samples were fixed in 10% formaldehyde, 50% ethanol and 5% acetic acid, dehydrated, and embedded in paraffin. Sections (15–20 μm) were cut using a rotary microtome. Tissue sections were de-waxed and stained with toluidine blue. Stained sections were mounted and examined using a Nikon Eclipse 600 light microscope. Images were captured using a SPOT RT3 Color Slider CCD digital camera and associated software (Diagnostics Instruments, http://www.diaginc.com/). Image processing and figure preparation were performed using Adobe Photoshop (http://www.adobe.com/).
Plants were grown under long-day conditions in potting mix in a growth chamber as described above. Three-week-old plants were used for RNA isolation using Qiagen plant RNA isolation kits. To eliminate developmental variation, leaves from the same position on each plant were collected and pooled, i.e. the 5th and 6th youngest leaves of each rosette. RNA quality control, quantification, labeling, Affymetrix GeneChip® processing, array data processing and analyses were performed as described previously (Price et al., 2004). As the analysis was exploratory and four independent OE lines were used, only one biological repeat was performed. The Affymetrix Microarray Suite 5.0 was used for data processing and analysis. Comparison analyses (experiment versus baseline arrays) were performed to identify genes that were differentially expressed based on the Wilcoxon’s signed rank test. Venn diagrams were then used to identify common changes in all four comparisons (Figure S3). A total of 483 genes with a ‘decrease’ call and 353 genes with an ‘increase’ call were obtained. Selection cut-offs were then applied using signal detection P <0.05, change detection P <0.005, and a ≥ twofold change. The signal detection P value was determined based on the discrimination score (R) against the threshold Tau, as described in the Affymetrix document ‘GeneChip® Expression Analysis: Data Analysis Fundamentals’. ATTED-II (http://atted.jp/top_tool.shtml) was used to search for co-expressed genes and identify gene networks. The data source in ATTE-II contains the results from 1388 publicly available microarray slides collected in AtGenExpress (http://www.arabidopsis.org/portals/expression/microarray/ATGenExpress.jsp). Co-expression was measured using mutual rank based on weighted Pearson’s correlation coefficients. Genevestigator version 3 (http://www.genevestigator.ethz.ch/gv/index.jsp) was used to reveal stimulus response profiles by performing meta-profile analyses. The experimental conditions used for the analyses were: ABA#1, Col-0 treated with 20 μm ABA for 24 h; ABA#2, Col-0 treated with 0.5 μm ABA for 48 h; ABA#3, the ABA-hypersensitive germination mutant ahg1 treated with 0.5 μm ABA for 48 h; ABA#4, ahg3 treated with ABA for 48 h; PAC, Col-0 treated with 20 μm paclobutrazol for 24 h. The plant materials used were germinating seedlings.
This publication is dedicated to the memory of the first author, Ms Pei-Chi Lin (1978–2007), who was a great co-worker and made an indispensable contribution to this work. We thank the Arabidopsis Biological Resource Center for providing DNA clones and seeds, Dr Suguru Takatsuto (Joetsu University of Education, Japan) for supplying deuterium-labeled internal standards for brassinosteroid quantification, Dr Biao Ding for the microscopy facility, Dr Steven St Martin for microarray design and data analysis, Dr Eric Stockinger for advice on freezing tolerance assays, Dr Sumire Fujiwara for flowering assays, Drs Esther van der Knaap and Randy Scholl for reading the manuscript, Jacob Ladd for tissue section, Joe Takayama for greenhouse support, and two anonymous reviewers for their insights and constructive comments. This work was supported by grants for the National Science Foundation (IOB-0543751) and the Ohio Plant Biotech Consortium to J.-C.J. Salaries and research support were also provided by state and federal funds provided to Ohio State University and the Ohio Agricultural Research and Development Center.