Sugar and ABA responsiveness of a minimal RBCS light-responsive unit is mediated by direct binding of ABI4
Gustavo Javier Acevedo-Hernández,
Departamento de Ingeniería Genética de Plantas, Centro de Investigación y de Estudios Avanzados del Instituto Politécnico Nacional – Unidad Irapuato, Apartado Postal 629, Irapuato, Guanajuato 36500, Mexico, and
Departamento de Biología Molecular de Plantas, Instituto de Biotecnología, Universidad Nacional Autónoma de México, Avenida Universidad 2001 Chamilpa, Apartado Postal 510-3 Cuernavaca, Morelos 62271, Mexico
Departamento de Ingeniería Genética de Plantas, Centro de Investigación y de Estudios Avanzados del Instituto Politécnico Nacional – Unidad Irapuato, Apartado Postal 629, Irapuato, Guanajuato 36500, Mexico, and
Photosynthesis-associated nuclear genes (PhANGs) are able to respond to multiple environmental and developmental signals, including light, sugars and abscisic acid (ABA). PhANGs have been extensively studied at the level of transcriptional regulation and several cis-acting elements important for light responsiveness have been identified in their promoter sequences. However, the regulatory elements involved in sugar and ABA regulation of PhANGs have not been completely characterized. Using conserved modular arrangement 5 (CMA5), a previously characterized minimal light-responsive unit, we show that in Arabidopsis thaliana this unit responds not only positively to light signals, but also negatively to sugars and ABA. The latter responses were found to be impaired in the abi4 mutant, indicating that ABSCISIC ACID INSENSITIVE-4 (ABI4) is a regulator involved in sugar and ABA repression of this minimal regulatory unit. Furthermore, we report a new sequence element conserved in several rbcS promoters, herewith named S-box, which is important for the sugar and ABA responsiveness of CMA5. This sequence corresponds to a putative ABI4-binding site, which is in fact bound by the Arabidopsis ABI4 protein in vitro. The S-box is closely associated with the G-box present in CMA5, and this association is conserved in the promoters of several RBCS genes. This phylogenetically conserved promoter feature probably reflects a common regulatory mechanism and identifies a point of convergence between light- and sugar-signaling pathways.
Photosynthesis is a process of fundamental importance not only for plants but also for all organisms, being essentially the only mechanism of energy input into the living world. Because of this central role in plant metabolism, photosynthesis has been the subject of extensive and intensive research efforts. The pathways of photosynthesis are well established and the properties of the enzymes involved are well understood (Salisbury and Ross, 1992). However, the regulatory mechanisms that allow plants to adjust the rate of carbon fixation in response to changes in environmental, metabolic or developmental cues are not completely known. Although this process may be partly regulated at the enzyme or translation level, bulk regulation has been shown to take place at the transcriptional level (Kuhlemeier, 1992).
One of the most important environmental factors affecting photosynthesis is light, which is used by plants not only as a source of energy for CO2 reduction, but also as a signal that controls many aspects of plant development (Kendrick and Kronenberg, 1994). Plants respond to light signals adopting a growth pattern, known as photomorphogenesis, which allows them to increase their access and exposure to light radiation (Nemhauser and Chory, 2002). In addition, light acts as a signal that has a profound effect on the transcriptional regulation of many plant genes. In particular, light induces the expression of many nuclear-encoded photosynthetic genes, such as RBCS (encoding the small subunit of ribulose-1,5-bisphosphate carboxylase), CAB (encoding chlorophyll a/b binding proteins), and PC (encoding plastocyanin) (Thompson and White, 1991; Vorst et al., 1993). Effects caused by light depend on its perception by a sophisticated system of photoreceptors (phytochromes, cryptochromes and UV-light receptors) that allow plants to detect distinct light wavelengths within a wide spectral range (reviewed by Briggs and Olney, 2001). Furthermore, the presence of a chloroplast-derived signal has proven to be necessary for light-regulation of these genes (Kusnetsov et al., 1996).
Light is not the only signal affecting expression of photosynthetic genes. Several studies have shown that high levels of metabolizable sugars exert a negative feedback regulation that directly represses photosynthetic gene expression (Krapp et al., 1991; Sheen, 1994). This regulation is at least in part dependent on a hexokinase-mediated sugar-signaling pathway, which is independent of the catalytic activity of this enzyme (Moore et al., 2003). Isolation of mutants that are defective in sugar responses allowed the identification of several components of the transduction pathway for this metabolic signal (reviewed in León and Sheen, 2003; Rolland et al., 2002). One of these components, the ABSCISIC ACID INSENSITIVE-4 (ABI4) gene, encodes a putative AP2 domain transcription factor that is proposed to be involved in abscisic acid (ABA) signal transduction (Finkelstein, 1994; Finkelstein et al., 1998). Thus, the participation of the ABI4 gene in sugar responses reveals a central role for ABA in this process (Arenas-Huertero et al., 2000; Huijser et al., 2000; Laby et al., 2000). Despite the growing information about the components of the sugar-signaling pathway, the precise mechanism that links sensing and transmission of sugar signals with the regulation of gene transcription still remains to be determined.
Promoters of nuclear-encoded photosynthetic genes, particularly RBCS and CAB, have been extensively studied to identify the cis-regulatory elements and the trans-acting factors involved in their regulation, especially by light. Sequence comparisons, deletion and mutagenesis analyses of such promoter regions led to the identification of a number of regulatory elements implicated in the control of transcription by light (Terzaghi and Cashmore, 1995; Tobin and Kehoe, 1994). Three classes of conserved sequences, the G-, I- and GT-boxes, are present in the promoter region of many light-regulated genes and have been shown to be important in their light responsiveness (Donald and Cashmore, 1990; Puente et al., 1996). Most of the DNA motifs involved in light-responsiveness are present in phylogenetically conserved modular arrangements (CMAs) that constitute minimal light-responsive units (Argüello-Astorga and Herrera-Estrella, 1996). However, little has been done to define the cis-elements involved in sugar repression of these promoters. Previous analyses have failed to identify any consensus sequence responsible for sugar regulation of photosynthetic genes (Sheen, 1990). There is only one report defining a small promoter region involved in sucrose repression of a photosynthetic gene, namely the RBCS2 gene of Phaseolus vulgaris (Urwin and Jenkins, 1997). This region is located close to an I- and G-box arrangement that has been shown to be important for the light-regulated expression displayed by other Rubisco promoters (Argüello-Astorga and Herrera-Estrella, 1996; Donald and Cashmore, 1990). Interestingly, a sugar-response sequence in the promoter of a rice α-amylase gene also contains an I–G combination, which suggests a role for these regulatory elements in transcriptional regulation not only by light, but also by sugars (Lu et al., 1998; Toyofuku et al., 1998).
The functional characterization of the conserved modular arrangement 5 (CMA5), the shortest native light-responsive element of a photosynthetic gene promoter characterized to date, was previously reported by Martínez-Hernández et al. (2002). The CMA5 arrangement corresponds to a 52-bp fragment of the Nicotiana plumbaginifolia rbcS 8B promoter. CMA5 contains an I- and a G-box, both of which are essential for the activation of a minimal heterologous promoter in a phytochrome-, cryptochrome- and plastid signal-dependent manner (Martínez-Hernández et al., 2002).
Here we show that CMA5 is able to respond not only to light and chloroplast signals, but also to sugar signals in a pathway involving ABA. Moreover, we demonstrate that a putative ABI4-binding site closely associated with the G-box present in CMA5, named S-box in this work, is important for sugar and ABA responses. We also show that the ABI4 gene is involved in the regulation of CMA5 responses to sugar and ABA, probably through the direct binding of the ABI4 protein to the S-box. An arrangement containing the I, G and S boxes is shown to be present in the promoter region of many sugar-repressed photosynthesis genes, probably defining a common mechanism for sugar repression in these genes.
CMA5 is similar to sugar-responsive sequences
A sequence comparison among CMA5 and previously reported sugar-responsive sequences in bean rbcS2 and rice α-amylase promoters (Lu et al., 1998; Morita et al., 1998; Urwin and Jenkins, 1997) revealed obvious similarities among them (Figure 1a). The most evident were the presence and position of the I- and G-boxes and the presence in CMA5 of a sequence (CACCTCCA) quite similar to the bean rbcS2 sugar-responsive element (CACTTCCA), which is also conserved in other CMA5-containing rbcS promoters (Figure 1b). This conserved sequence was named S-box.
CMA5 is a sugar-regulated element
Because the bean rbcS2 sugar-repression element was identified by means of a deletion approach, it was interesting to define whether the CACCTCCA sequence present in CMA5 was able to confer sugar-regulated expression in a gain-of-function approach. To test whether CMA5 was able to confer sugar responsiveness to a heterologous minimal promoter, we determined the effect of different concentrations of glucose on the level of β-glucuronidase (GUS) expression in previously reported CMA5-GUS Arabidopsis transgenic lines (Martínez-Hernández et al., 2002). The CMA5-GUS construct consists of one copy of CMA5 fused to the −46 truncated version of the 35S promoter of the cauliflower mosaic virus (CaMV) and the GUS reporter gene. As controls, Arabidopsis transgenic lines harboring the GUS reporter gene under the control of the CaMV35S promoter (35S-GUS) and the complete Nicotiana tabacum rbcS1 promoter (rbcS-GUS) were used. At least three independent transgenic lines of each construct were utilized for all assays. Because the expression pattern of the reporter gene and its responses were similar for all lines tested, we present results for one representative line in each case.
Seven-day-old Arabidopsis seedlings were treated for 4 days in MS liquid medium containing different concentrations of glucose, harvested and subjected to fluorometric GUS analyses. Seedlings harboring the 35S-GUS construct did not show a significant change in GUS activity in response to increasing concentrations of glucose (Figure 2a). In contrast, GUS activity in CMA5-GUS and rbcS-GUS lines decreased concomitantly with an increase in glucose concentrations up to 50 mm, with a maximal repression at 25 mm (Figure 2a). The repression effect was not observed with concentrations of glucose lower than 25 mm (data not shown). Interestingly, the observed reduction in GUS activity in the CMA5-GUS and rbcS-GUS lines was lost when higher glucose concentrations were used. Semi-quantitative RT-PCR (Marone et al., 2001) for the GUS transcript showed that GUS activity properly reflects the level of GUS mRNA in both cases (data not shown).
Similar results were obtained in transgenic tobacco lines harboring the same constructs (data not shown) or when glucose was substituted by sucrose or fructose (data not shown). In all cases, the repression/de-repression effect of increasing concentrations of glucose was observed. However, the concentration of glucose at which maximum repression occurred varied between 25 and 50 mm glucose in distinct experiments, probably because of small variations in the growth conditions or to subtle differences in the developmental stage of the seedlings used in each experiment. Similar results were observed using histochemical GUS assays. Seedlings harboring the 35S-GUS construct did not show a significant change in GUS activity under the different glucose treatments. In contrast, GUS staining in rbcS-GUS and CMA5-GUS lines clearly decreased at 25 mm glucose (see Figure S1). Although glucose was found to modulate the level of GUS activity levels, no change was observed in the tissue-specific pattern of expression for any of the constructs tested (see Figure S1).
To determine whether the expression of the endogenous RBCS gene family in Arabidopsis is also subjected to the repression/de-repression effect of increasing concentrations of glucose, semi-quantitative RT-PCR analysis was also carried out using PCR primers that amplify the transcripts of all four Arabidopsis genes (rbcS1A, 1B, 2B, and 3B). It was observed that the steady-state level of the endogenous RBCS transcripts decreased in 25 and 50 mm glucose and this repression decreased in higher concentrations (Figure 2b,c). These results are similar to those observed for the effect of glucose on the expression of CMA5-GUS and rbcS-GUS, except that for the endogenous RBCS transcripts a slight induction was observed in 200 mm glucose. This difference could be the result of distinct stability for GUS and RBCS transcripts or the presence of cis-acting regulatory elements in at least one of the Arabidopsis RBCS genes different to those present in the tobacco rbcS1 promoter.
The response observed for CMA5-GUS and rbcS-GUS constructs, and the Arabidopsis RBCS genes was somewhat intriguing, because to date, only negative responses to the addition of high concentrations of glucose had been reported for photosynthesis-associated nuclear genes (PhANGs) (Arenas-Huertero et al., 2000; Krapp et al., 1993; Sheen, 1994). However, our experimental conditions differed from those reported previously in which protoplasts, solid media or non-isosmotic conditions were used. It is also evident from previous work that sugar regulation depends on the general growth conditions as well as on the developmental status of the plant (Moore et al., 2003). In order to corroborate that this discrepancy was due to differences in the experimental conditions used in each case, we tested our lines using those reported by Arenas-Huertero et al. (2000). Thus, plants were grown in solid media containing 2% (w/v) glucose, 7% (w/v) glucose or 7% (w/v) mannitol, which in molar concentrations correspond to 111 and 389 mm, respectively. Under these conditions, CMA5-GUS and rbcS-GUS lines showed a significant decrease (to about 50%) in GUS activity when growing in medium containing 389 mm glucose, compared with the levels observed in either 111 mm glucose or 389 mm mannitol (see Figure S2). This demonstrated that the effect of increasing concentrations of glucose on GUS activity in both CMA5-GUS and rbcS-GUS lines was dependent on the assay conditions used and not the result of an osmotic effect. It is possible that the differences observed between liquid and solid media were the result of the fact that in liquid media the photosynthetic tissue was directly in contact with different concentrations of glucose, whereas in solid media only roots were in contact with the sugar. Differential responses depending on the availability of sugars in the root or the aerial part of the plant have been previously reported (Roldán et al., 1999).
Altogether, these results showed that CMA5 behaves as a sugar-responsive element in different experimental conditions and that its repression/de-repression behavior in liquid media was dependent on the concentration of sugar added. To the best of our knowledge, CMA5 is the shortest native sequence from a nuclear-encoded photosynthesis-associated gene promoter that has been shown to be able to respond to both light and sugar signals.
CMA5 sugar response is affected by ABA
Previous reports proposed a central role for ABA in plant glucose responses (Arenas-Huertero et al., 2000; Huijser et al., 2000; Laby et al., 2000; Rook et al., 2001). Therefore, we decided to test the effect of ABA on the expression directed by CMA5 and its ability to respond to sugars. For this purpose, 7-day-old seedlings were grown for 4 days in liquid media with or without glucose and containing different concentrations of ABA.
The sole presence of ABA, without exogenous addition of sugar, repressed GUS expression driven either by CMA5 or the complete rbcS1 promoter (Figure 3c,e). A similar ABA treatment had no effect on the expression of the 35S promoter (Figure 3a). In contrast, when seedlings were treated with increasing concentrations of ABA in media containing a glucose concentration known to be repressive (50 mm in this assay), an increase in GUS activity in CMA5 and rbcS lines was observed. Increasing ABA concentrations had no effect on 35S promoter activity (Figure 3b,d,f). Semi-quantitative RT-PCR analysis showed that the steady-state level of the RBCS mRNA was reduced by addition of 100 μm ABA, but this effect was gradually lost by the addition of increasing concentrations of glucose (Figure 4).
These results showed that ABA, when added alone, has a negative effect on the expression of RBCS gene, and on the gene expression directed by CMA5 and the tobacco rbcS1 promoter, although this effect can be positive when ABA is in combination with high concentrations of glucose.
The S-box is important for the sugar responsiveness of CMA5
With the purpose of testing the participation of the S-box in CMA5 sugar responsiveness, we generated a new reporter construct, consisting of a truncated version of CMA5, which lacks the S-box, fused to the minimal −46 version of the 35S promoter driving the expression of the GUS reporter gene. This construct was designated IGΔS-GUS (Figure 5a). The IGΔS-GUS construct was used to transform Arabidopsis and three homozygous transgenic lines were used in the experiments reported. The results for one representative line are shown.
Initial analysis of the expression directed by IGΔS-GUS, showed that deletion of the S-box in CMA5 did not affect the activity of this regulatory unit nor its light-regulated or chloroplast-dependent activity (see Figure S3). However, the level of GUS activity directed by IGΔS-GUS was not negatively affected by the addition of glucose (Figure 5b). In contrast, a 30–40% increase in GUS activity was observed in IGΔS-GUS lines treated with 100 and 200 mm glucose. Similarly, no significant effect was observed in the GUS expression driven by IGΔS when ABA was added, except a slight but reproducible repression at a concentration of 100 μm, which strongly represses CMA5 and rbcS activity (Figure 5c).
These results showed that the S-box does not participate in determining the enhancer activity or light responsiveness of CMA5, but appears to be essential for the ABA and glucose repression-responses of this minimal light-responsive unit.
Putative ABI4-binding sites are closely associated with the G-box of several light-regulated promoters
The transcriptional factor ABI4 has been shown to be an important component in the sugar signal transduction pathway leading to the transcriptional regulation of photosynthetic genes (Arenas-Huertero et al., 2000; Huijser et al., 2000; Laby et al., 2000). Recently, Niu et al. (2002) reported several ABI4-binding sequences determined by PCR-assisted binding site selection (Figure 6a). A search of putative ABI4-binding sites in CMA5, revealed that the S-box was very similar to an ABI4-binding site (Figure 6b). The S-box of CMA5 is an octamer with the sequence CACCTCCA, which differs only in one nucleotide from one of the ABI4-binding sites (CACCGCCA) reported by Niu et al. (2002). In this search, another putative ABI4-binding site was found in the complementary strand of CMA5, also in close association with the G-box (Figure 6b).
To determine whether the presence of ABI4-binding sites could be a phylogenetically conserved feature of PhANG promoters, we searched for putative ABI4-binding sites in other sugar-repressed photosynthetic promoters having an I–G-box arrangement, including those regions that were previously reported as harboring sugar-responsive sequences (from the bean rbcS2 and the rice Amy3D promoters). ABI4-binding sites were consistently found in close association with the G-box in all rbcS promoters investigated (Figure 6b). Although in Figure 6 we present data for only one member of the tomato and Arabidopsis rbcS gene family, several other members of this gene family contain an I–G–S-box arrangement (data not shown).
For cab promoters, putative ABI4-binding sites were found in a conserved position relative to elements previously defined as important for light-regulated expression (Figure 6c; Argüello-Astorga and Herrera-Estrella, 1996). Furthermore, putative ABI4-binding sites were found closely associated with G-boxes in the promoter of previously reported sugar-responsive genes such as the rice Amy3D gene and in the Arabidopsis pc promoter (Figure 6d; Dijkwel et al., 1996; Lu et al., 1998; Morita et al., 1998; Toyofuku et al., 1998). The presence of an arrangement containing the I, G and S-boxes in several photosynthetic genes, where the S-box corresponds to a putative ABI4-binding site, suggest a common mechanism for sugar repression in these promoters.
The ABI4 gene is involved in sugar and ABA responses of CMA5
In order to determine the participation of the ABI4 transcription factor in sugar and ABA responsiveness of CMA5, we introduced the rbcS-GUS and CMA5-GUS gene constructs into the Arabidopsis abi4 mutant background. This was achieved through genetic crosses of the wild-type transgenic lines with the abi4-1 mutant. At least three lines homozygous for both the abi4 mutation and the different constructs were obtained. Results for one representative line of each construct are shown.
As can be observed in Figure 7, the level of GUS activity driven by the CMA5-GUS and rbcS-GUS constructs in the abi4 mutant background was not significantly decreased by glucose or ABA treatments at concentrations known to repress their activity in the wild-type background. In contrast, the light-regulated and chloroplast-dependent activity of CMA5 was not affected by the abi4 mutation (see Figure S3). These results demonstrated that the ABI4 gene is important for the regulation of ABA and sugar responsiveness of CMA5 and the complete rbcS1 promoter.
ABI4 binds to the S-box of CMA5
The finding that deletions of the S-box and mutations in ABI4 impaired the response of CMA5 to both glucose and ABA, suggested the possibility that the sugar and ABA responsiveness of CMA5 could be mediated by direct binding of this transcription factor to the S-box. To determine whether ABI4 could directly bind to the S-box, the Arabidopsis ABI4 protein was expressed in Escherichia coli, purified, and used for gel-shift assays employing the CMA5 S-box as a probe (Figure 8a). We used the native S-box sequence from CMA5, except that the G-box was eliminated. Because of this fact, the S-box probe contained the putative ABI4-binding sites found both in the coding and in the complementary strand of CMA5. It was found that the recombinant ABI4 protein produced a clear S-box retardation complex (Figure 8b). Moreover, competition assays showed that this binding was specific as it was competed by an excess of cold S-box but not by two distinct unrelated competitor sequences (Figure 8). Additionally, we used CMA5 S-box oligonucleotides with one (S2m1) or four altered nucleotides (S2m4) as competitors in gel-shift assays (Figure 8). These sequences showed a gradually reduced ability to compete with the native S-box as the number of mutations increased, showing that the altered sequence is involved in the binding of ABI4. These results demonstrate, at least under in vitro conditions, that ABI4 can bind directly to the S-box present in CMA5.
Every aspect of plant development and metabolism is regulated by the input from a number of signaling pathways. These pathways include those involved in responses to external or environmental stimuli (such as light) and to internal signals (such as hormones or metabolites). The idea that plants respond to each of these signals via separate, linear pathways has become insufficient to explain the ability of these organisms to adapt their responses to simultaneous changes in these stimuli. Many researchers consider plant response pathways can be more usefully conceived as forming part of an inter-connected web, which allows the optimization of a non-cognitive behavior in response to various combinations of stimuli (Genoud and Métraux, 1999; Gibson, 2000).
Interaction between light, sugar and ABA signal transduction pathways
It has been long recognized that both light and metabolizable sugars affect the expression of photosynthetic genes (Harter et al., 1993; Krapp et al., 1993; Sheen, 1990). However, little has been done to determine whether or not a direct connection between these signaling pathways exists. Short (1999) reported that sugars affect phyA-light signaling through an inhibitory mechanism mediated by phyB, revealing an interaction between these stimuli at the level of light receptors. Some approaches have been directed to ascertain the components through which both signals interact. Dijkwel et al. (1997) determined that sucrose represses several physiological responses that are mediated by phyA-signal transduction pathways. They isolated a set of mutants in which the sugar-repression of the light-regulated plastocyanin promoter was altered. One of such mutants, allelic to abi4, provided evidence for a close interaction between light, sucrose and ABA signaling pathways (Dijkwel et al., 1997; Huijser et al., 2000, Van Oosten et al., 1997).
One of the significant novel findings of this work is the observation that high (50 mm) glucose represses gene expression, but that higher (200 mm) glucose concentrations restores a normal level of expression. This is an interesting observation and is reminiscent of a high and low fluence light response first observed with promoter:GUS fusions and later confirmed for CAB genes (White et al., 1995). It has been proposed that this light intensity regulation of CAB gene transcription is coupled to photosynthetic electron transport (PET), and specifically to the redox status of the plastoquinone pool (Escoubas et al., 1995). It is noteworthy that PET is also associated with sugar-dependent gene suppression of photosynthetic genes, through a mechanism involving the ABI4 gene (Oswald et al., 2001), showing an interactive relationship among light signals, sugars, PET, and ABA.
In this work, we demonstrated that CMA5, a 52-bp Nicotiana plumbaginifolia rbcS8B promoter region, contains all the cis-acting elements required for light-inducibility and sugar repression, exhibiting similar transcriptional regulatory properties to those of the complete tobacco rbcS1 promoter. Our results also showed that the transduction pathways for these stimuli converge into a compact regulatory unit. This finding could be explained by two alternative mechanisms: (i) sugar effects occur at any point in the light signal transduction cascade, making CMA5 responsive to integrated stimuli through common transcription factors or (ii) both transduction pathways act separately and the integration of the signals occurs at the level of cis-regulatory elements, through the differential binding of transcription factors responding to each pathway. The finding that a CMA5 version lacking the S-box, is impaired in glucose responsiveness but maintains its light-regulated properties, suggests that the latter mechanism is most probably the correct one.
Interestingly, we also found that treatment with 100 and 200 mm glucose resulted in the induction of rbcS and CMA5 activity in the abi4 mutant background and for the chimeric promoter containing IGΔS in the wild-type background. This suggests that when the ABI4 signaling pathway is not functioning, sugars can have a positive effect on the expression of PhANGs. It has been previously reported that the expression directed by a pc promoter, also containing a putative ABI4-binding site, is not repressed but rather induced by sugars in the abi4/sun6 mutant background (Oswald et al., 2001). However, the signaling pathways by which sugar-repressed promoters become sugar-inducible remains to be determined.
The S-box is involved in sugar and ABA repression of CMA5
Previous analyses of several photosynthetic promoters failed to reveal a common element responsible for sugar repression (Sheen, 1990). Deletion analysis of maize rbcS promoters, suggested that sugar responses were due to the presence of positive, rather than negative regulatory elements, thus activating transcription under sugar starvation (Sheen, 1990). This possibility was proposed in order to explain the failure of the deletion approach utilized to produce a non-repressible promoter version. However, a more refined deletion analysis of the bean rbcS2 promoter allowed the identification of a negative element important for sugar regulation downstream of the G-box present in this promoter (Urwin and Jenkins, 1997).
Analysis of the promoter region of several RBCS genes that harbor a combination of I- and G-boxes showed that they contain a conserved DNA motif similar to the CACCTCCA sequence (S-box) present in CMA5 and previously identified in the bean rbcS2 promoter as important for glucose repression (Figure 1b). Moreover, deletion of the S-box present in CMA5 resulted in an almost complete loss of glucose and ABA repression of this regulatory element. As deletion of the S-box affected glucose and ABA responsiveness, but had no effect on the level of expression of CMA5, we conclude that the S-box acts as a negative regulatory element, potentially bound by a cognate repressor.
As ABI4 has been shown to be an important link between ABA and glucose signaling pathways, a comparison of the S-box with the ABI4-binding sites was carried out. We found that the S-box closely resembles one of the ABI4-binding sites identified by PCR-assisted binding site selection (Niu et al., 2002). A non-exhaustive search of the different ABI4-binding sites in the promoter region of photosynthesis-related genes, showed that at least 14 rbcS, six cab and one pc promoters from different plant species contain putative ABI4-binding sites closely linked to a light-responsive region. In particular, these sites were consistently found in close association with the G-box in all rbcS promoters investigated (Figure 6b). Even though most of the putative ABI4-binding sites found in this search are located in the complementary strand, at this point we have no evidence for an orientation-dependent functionality, although it has previously been shown that CMA5 is able to activate a heterologous minimal promoter in an orientation-independent manner (Argüello-Astorga and Herrera-Estrella, 1995). However, we cannot rule out that the relative orientation of other sites associated with this modular arrangement is important for a particular regulated expression. For instance, it has been previously reported that a G-box (ABRE) and an ABI4-binding site (CE1) are important for the ABA induction of several genes (Niu et al., 2002). Recently, it has been demonstrated that these elements need to be separated by multiples of 10 bp in order to confer high ABA induction, suggesting that they have to be located on the same side of the DNA double helix (Shen et al., 2004). On the other hand, in this work we show that an ABI4-binding site almost overlapping the G-box in CMA5 is important for the ABA repression of this unit. This suggests that the relative position of an ABI4-binding site with respect to the G-box is relevant for the expression of ABA responsive genes.
ABI4 as a putative transcription factor for sugar and ABA responses of CMA5
The transcription factor ABI4 has been implicated in the sugar signal transduction pathway that affects expression of photosynthetic genes (Arenas-Huertero et al., 2000; Huijser et al., 2000). ABI4 seems to be a point of convergence between the ABA signal transduction cascade and a hexokinase-dependent sugar response pathway (Arenas-Huertero et al., 2000). Analysis of the effect of glucose and ABA on the CMA5-directed gene expression in the abi4 mutant background showed that CMA5 responsiveness to both stimuli is impaired in the absence of ABI4. These results suggest that ABI4 acts as a repressor of CMA5 activity that is produced or activated as a result of glucose or ABA treatments. The finding that ABI4 is able to specifically bind to the S-box in vitro, provides strong evidence that this transcription factor could repress the activity of CMA5 by direct binding to the S-box. The fact that even with four mutations we were not able to completely abolish the capacity of S-box to compete for ABI4 binding, is probably the result of a higher stability of protein binding to the probe used, due to the presence of sites on both strands. However, further studies will be necessary to determine the relative importance of each S-box for the binding of the ABI4 protein and the ABA and sugar responsiveness of CMA5. As the S-box is present in the promoter of many PhANGs, it is quite possible that ABI4 mediates sugar and ABA repression of these genes by binding directly to their promoters.
Based on our results, we propose a hypothetical model for the participation of ABI4 in the control of CMA5 activity in response to glucose and ABA (Figure 9). In this model, ABI4 is able to bind directly to the S-box sequence, repressing the transcription activity of CMA5, probably through the interaction or interference with factors binding to the G-box, because of the conserved association with this element. Previous reports have shown the importance of the G-box sequence as a positive element necessary for high expression levels of cab and rbcS promoters (Donald and Cashmore, 1990; Martínez-Hernández et al., 2002). The close association, and in some cases overlap, of the ABI4-binding sites with the G-box sequence could partly account for a negative regulatory mechanism involving competition for binding among activators and repressors. ABI4 abundance or activity could be regulated by glucose through a pathway involving ABA or through an ABA-independent pathway. This notion is supported by findings from Arroyo et al. (2003), who reported that the expression of the ABI4 gene is induced by both glucose and ABA.
The observations that high levels of glucose (over 100 mm) had no negative effect on the expression of CMA5 and the rbcS1 promoter, and that treatment with ABA decreases or eliminates sugar repression and vice versa, were somewhat puzzling (Figures 2–4). Although ABI4 is induced in Arabidopsis seedlings treated with glucose or ABA (Arroyo et al., 2003), it is possible that either a high concentration of exogenous glucose or the simultaneous application of glucose and ABA lead to a decrease in the transcription of the ABI4 gene or in the abundance or activity of the ABI4 protein. Because of the presence of an ABI4-binding site located downstream to the TATA-box of the ABI4 gene, Niu et al. (2002) proposed that ABI4 could repress its own expression. Protein binding to TATA-proximal regions has been shown to repress transcription (Heins et al., 1992). Therefore, to explain all the effects of glucose and ABA on the expression of CMA5, our model also proposes that a negative regulatory loop is involved in the transcriptional regulation of ABI4, resulting in a lack of repression for PhANGs at high sugar concentrations or when both sugar and ABA signals are simultaneously present. However, we cannot exclude that alternative control mechanisms responding to higher glucose or ABA levels could be responsible for the proposed negative regulation of ABI4. For instance, Rook et al. (1998) reported that low levels of exogenous sucrose stimulated transcription whereas high concentrations decreased translation of the Arabidopsis ATB2 gene, a transcription factor involved in sugar-regulated responses. To confirm our model, we are currently analyzing the effect of glucose and ABA on the pattern of expression of the ABI4 gene under our experimental conditions, and the effect of other components of the ABA and glucose signaling pathways on the transcriptional activity of CMA5.
Physiological implications of sugar and ABA regulation of PhANGs.
Several mutants affected in sugar responses have been isolated by different groups taking advantage of the inhibitory effects of high sugar concentrations on Arabidopsis seedling growth (reviewed in Smeekens, 2000). It has previously been reported that these effects are dependent on ABA biosynthesis and are confined to approximately 2-day post-germination (denominated state 1). Once the seedling becomes photosynthetically competent (state 2), its development is relatively insensitive to sugar-induced ABA synthesis (Gazzarrini and McCourt, 2001). In this work, we have used Arabidopsis seedlings that are clearly in state 2, which could be more reflective of the effect of high sugar concentrations on the expression of photosynthesis-associated genes during autotrophic metabolism, with no obvious effect on development.
It has been previously shown that ABA negatively controls the expression of light-regulated photosynthetic genes, such as RBCS and CAB (Medford and Sussex, 1989). This effect has been proposed to be associated with physiological processes in which an increase in ABA levels is observed, such as embryo development (Medford and Sussex, 1989), water stress (Bartholomew et al., 1991), seed maturation (Chang and Walling, 1991) and dark treatments (Weatherwax et al., 1996). Our results indicate that, at least for RBCS genes, ABA could regulate the expression of photosynthetic genes by promoting direct binding of ABI4 to S-box-related sequences under the above physiological conditions. Interestingly, it has been reported that during cold acclimation, a condition in which the level of ABA increases and soluble sugars accumulate and act as cryoprotectants, transcripts of RBCS and CAB genes accumulate to normal levels (Chen and Gusta, 1983; Chen et al., 1983; Strand et al., 1997). The finding that ABA can overcome the repression of photosynthetic genes caused by high levels of glucose, and the proposed model for ABI4 regulation, could explain why in plants grown under low temperatures the expression of photosynthetic genes occurs at normal levels. Moreover, it has been observed that the inhibitory effects of ABA on radicle emergence during germination are dramatically suppressed in the presence of sugars (Finkelstein and Lynch, 2000). It will be interesting to determine the mechanism through which these effects occur and whether the proposed model of ABI4 response to glucose and ABA operates under different physiological conditions.
Plant material and growth conditions
The 35S-GUS and CMA5-GUS Arabidopsis transgenic lines were described previously (Martínez-Hernández et al., 2002). The rbcS-GUS Arabidopsis transgenic line, containing the same reporter gene fused to the Nicotiana tabacumrbcS1 promoter, was kindly provided by Dr Luisa López (unpublished results from our laboratory). The Arabidopsis thalianaabi4-1 mutant was kindly provided by Dr Bas Dekkers (Utrecht University, Utrecht, The Netherlands).
Arabidopsis seeds were surface sterilized, imbibed in sterile water for 4 days at 4°C to break dormancy, and then plated on media containing 1× Murashige and Skoog (MS) basal salt mixture (Gibco-BRL, Grand Island, NY, USA) supplemented with 2% (w/v) sucrose and 6.5 g l−1 phytagar (Gibco-BRL). Seeds were plated on cellophane circles fitting the Petri dish, to facilitate collecting or transferring plants and avoid tissue damage. Plants were grown under sterile conditions in controlled growth chambers (MOD AR-32L; Percival Scientific Inc., Boone, IA, USA) at 24°C, under a 16 h light/8 h dark photoperiod with a light fluence of 65 μmol m−2 sec−1.
Light, sugar and ABA treatments
For different sugar conditions, seedlings were grown for 7 days as described above, and transferred to 0.5× MS liquid media supplemented with 0, 25, 50, 100, or 200 mm glucose. Mannitol was added in the necessary amount to maintain isosmotic conditions for all treatments (to 100 or 200 mm). Plants were grown with gentle orbital agitation in a growth room at 25°C, under a 16 h light/8 h dark photoperiod with a light fluence of 65 μmol m−2 sec−1, and collected 4 days after transfer to glucose medium. For the experiments on solid media, the procedure was as described by Arenas-Huertero et al. (2000). For ABA treatments, seedlings were grown as described, and transferred to 0.5× MS liquid media supplemented with 0, 1, 10, or 100 μm ABA. Plants were grown under the same conditions as described for sugar treatments.
Reporter construct and plant transformation
Oligonucleotides corresponding to CMA5 (Martínez-Hernández et al., 2002) lacking the S-box (namely IGΔS) were synthesized and fused to the Δ46 CaMV35S promoter upstream of the coding sequence of the uidA (GUS) reporter gene in pBI146S (Martínez-Hernández et al., 2002). Arabidopsis plants were transformed by Agrobacterium tumefaciens containing this construct through the floral dip method reported by Clough and Bent (1998). Three independent transgenic lines homozygous for the construct were analyzed. One representative line (IGΔS-1) was selected for subsequent experiments.
Reporter constructs were introduced into mutant abi4-1 background by crossing these mutant plants with wild-type rbcS-GUS and CMA5-GUS transgenic lines. After crossing, lines homozygous for both the construct and the mutation were selected on the basis of the segregation of resistance to 30 mg l−1 kanamycin for those harboring the construct and the ability to germinate in presence of 3 μm ABA for those homozygous for the abi4 mutation. At least three independent homozygous lines were analyzed in all the reported experiments.
GUS histochemical and fluorometric analyses
GUS histochemical analysis was performed as described previously (Stomp, 1992). Staining was performed without previous tissue fixation for 12 h at 37°C. Clearing was accomplished by incubating tissue with a methanol:acetone (3:1, v/v) solution for 1 h. Plants were observed and the images captured using a Hitachi KP-D51 video camera (Hitachi Denshi, Ltd, Tokyo, Japan) adapted to a Olympus SZH10 microscope (Olympus Optical Co. Ltd, Tokyo, Japan). Seedlings for fluorometric assays were collected and ground in GUS extraction buffer (Gallagher, 1992). Total protein extracts were centrifuged at 12 000 g for 10 min at 4°C, and protein content was quantified using Bradford reagent (Bio-Rad Laboratories, Hercules, CA, USA). For each assay, 5 μg of protein were used, and the analysis was performed according to Gallagher (1992). Fluorescence was measured using a TKO 100 fluorometer (Hoefer Scientific Instruments, San Francisco, CA, USA). GUS activity is reported as picomoles of 4-methylumbelliferone per microgram of protein per minute, although for means of comparison, it is expressed as relative activity in several figures.
Total RNA was extracted from Arabidopsis seedlings grown in liquid media with the sugar and ABA treatments described above. Tissue was frozen and ground, and RNA extraction was performed as described by Urwin and Jenkins (1997). Samples were treated with RNase-free, DNase I (Roche, Basel, Switzerland). RT-PCR reactions were performed with 100 ng of total RNA using the SuperScript One-Step RT-PCR with Platinum Taq (Invitrogen, Carlsbad, CA, USA), following the protocol suggested by the manufacturers on a PTC-200 thermal cycler (MJ Research Inc., Waltham, MA, USA). The optimal number of cycles for semi-quantitative analysis was 25, which was determined as reported by Marone et al. (2001). The primers used were: rbcs-fw, 5′-ACCTTCTCCGCAACAAGTGG-3′; rbcs-rv, 5′-GAAGCTTGGTGGCTTGTAGG-3′; tub-fw, 5′-CTCAAGAGGTTCTCAGCAGTA-3′; tub-rv, 5′-TCACCTTCTTCATCCGCAGTT-3′. Product size was 275 bp for RBCS and 483 bp for TUB. The four primers were added to all reactions, and no obvious competition among them was detected. PCR products were separated in 2% agarose gels, and images of the gels were digitally captured and analyzed with the Scion Image program version Beta 4.0.2 (Scion Corporation, Frederick, MA, USA). Relative RT-PCR values are presented as a ratio of the RBCS signal in the selected treatment divided by the TUB signal. RT-PCR signals were averaged from at least three replicates and mean and standard deviation of all experiments performed were calculated after normalization to TUB.
Sequences and sequence alignments
The sequences used for alignments were obtained from the GenBank/EMBL nucleotide sequence database. The GenBank/EMBL accession numbers for these sequences are as follows. rbcS promoters: Arabidopsis thalianarbcS2B, X14564; Coffea arabicarbcS1, AJ419827; Gossypium hirsutumrbcS1, X54091; Helianthus annuus rbcS1, Y00431; Lycopersicon esculentumrbcS1, X66068 and rbcS2, X66069; Nicotiana plumbaginifolia rbcS8B, X13711; Nicotiana tabacumrbcS1, M32419; Petunia hybrida rbcS1, X12986; Phaseolus vulgarisrbcS2, AF028707; Pisum sativum rbcS, X00806; Solanum tuberosumrbcS1, X69759; Zea maysrbcSZm1, S42508 and rbcSZm3, S42568. cab promoters: Arabidopsis thalianacab2, X15221 and cab3, X15222; Petunia hybrida cab91R, X02356; Sinapis alba cab, X16436; Spinacia oleracea cab1, X64350; Zea mays cab1, X14794. pc promoters: Arabidopsis thalianapc, S67901. α-amylase promoters: Oryza sativaRAmy3D, M59351.
Sequence searches and alignments were performed with the Lasergene software package (DNASTAR Inc., Madison, WI, USA), with manual adjustment when necessary.
Nearly full-length Arabidopsis ABI4 protein (aminoacids 1–325), was expressed in E. coli as a polyhistidine-tagged (6× His) fusion protein and purified according to established protocols (Invitrogen Xpress purification manual). The purified protein was used for gel-shift assays as described by Niu et al. (2002). To probe S2 DNA, 35-bp complementary oligonucleotides corresponding to a dimer of the CMA5 S-box were synthesized, and phosphorylated with 32P using T4-polynucleotide kinase (Invitrogen). For the competition assays using mutant versions of the S-box, two pairs of 35-bp complementary oligonucleotides corresponding to the S-box with one (S2m1) or four (S2m4) mutations were synthesized (sequences are shown in Figure 8a). For the non-specific competition assays, polylinker regions of plasmids pBluescript and pUC18 were obtained by digestion with restriction endonucleases HindIII–BamHI and purified. All competitors were added in a 50× molar excess. Images of gel-shift assays were processed using Adobe photoshop 7 and contrasted with Gene-Spring software.
This work was supported in part by the Consejo Nacional de Ciencia y Tecnologia, México (grant no. 31628-B), the European Commission (grant no. ICA-4-CT2000-30017) and by the Howard Hughes Medical Institute (grant no. Nbr55003677). G. A.-H. is grateful to CONACYT and CONCYTEG for a PhD fellowship. We also thank Massive Attack for musical support during the preparation of this manuscript.