GUS reporter expression from 11 basal promoters (CaMV –90) with G-box cores (CACGTG) was analysed to evaluate the regulatory roles of G-box flanking sequences. While most G-box motifs exhibited some tissue preference of gene expression, a distinct tissue-specific expression was not apparent. However, one of 11 G-box sequences, the G-box 10 (GCCACGTGCC) tetramer, conferred a high-level constitutive expression in seed, root, leaf, axillary bud, almost all parts of flower buds and pollen of transgenic tobacco plants. Furthermore, the G-box 10 tetramer promoter exhibited high-level expression in transgenic dicot carrot and monocot rice. This is apparently the first report of a G-box motif conferring a high-level constitutive expression in a non-tissue-specific manner.
Most of G-box binding proteins are bZip proteins with proline-rich activation, basic DNA-binding, and leucine zipper dimerization domains. Comparison of other G-box binding proteins, including wheat EmBP1 (Niu & Guiltinan 1994), TAF1 (Oeda et al. 1991), HBP1a (Tabata et al. 1991), and CPRF1, 2 and 3 (Weisshaar et al. 1991), revealed that these G-box binding proteins had different binding preferences and affinities against G-box flanking sequences (Izawa et al. 1993), even though the G-box binding protein family has a conserved basic DNA binding region. Most G-box binding proteins bind G-box motifs as homodimers. However, Arabidopsis GBFs (Schindler et al. 1992) and parsley CPRFs (Armstrong et al. 1992) heterodimerize promiscuously, the result being an increase in functional dimers of G-box binding proteins with heterologous activator domains (Armstrong et al. 1992; Schindler et al. 1992).
Based on in vitro binding studies, G-boxes were classified into class I, class II and hybrid class I/II (Williams et al. 1992). How in vitro classification reflects in vivo regulation of gene expression remains unknown. The only available in vivo study revealed that the palindromic G-box (CCACGTGG) and G-box-related motif I (GTACGTGG) from rice rab16A gene represent a different regulation of gene expression in tobacco plants (Salinas et al. 1992), which means that G-box flanking sequences may affect in vivo regulation of gene expression.
To investigate in vivo properties of G-box sequences with different flanking sequences in the regulation of gene expression, we examined GUS gene expression driven by 11 different G-box tetramers fused to a CaMV –90/35S promoter in transgenic tobacco plants. Each G-box sequence directed different modes of gene expressions. The G-box 10 motif conferred high-level constitutive expression in seeds, leaves and roots. We also confirmed the stable, strong expression of the G-box 10 promoter in transgenic dicot carrot and monocot rice plants. Roles of flanking sequences of G-box motifs on gene regulation are also discussed.
G-box 10 tetramer strongly enhances gene expression in tobacco
Eleven different G-box tetramers were cloned upstream of a CaMV –90/35S promoter and the resulting constructs were introduced into tobacco plants to examine expression patterns (Fig. 1a). Nucleotide sequences of G-box 1 to G-box 6 contain palindromic G-box sequences, whereas the remaining G-box 7 to G-box 11 have non-palindromic ones. G-box 1, 3 and 10 have GC-rich flanking sequences and G-box 2, 4 and 6 contain flanking AT-rich sequences. About 40 independent transgenic plants were generated for each construct and their R1 seeds were harvested. To evaluate tissue specificity and strength of gene expression directed by each G-box promoter construct, seeds and 10-day-old seedlings of each construct were stained with X-Gluc (Fig. 1b).
Transgenic tobacco plants containing pBI-90GUS as a negative control showed no significant gene expression in leaf and stem, but weak expression was observed around the radicle pole at the tip of the embryo (Fig. 1c) (Benfey et al. 1989). G-box 2, 6, 7 and 11 significantly enhanced gene expression in seeds, cotyledons and roots, G-box 3 specifically increased gene expression in cotyledons and roots, and G-box 9 in cotyledons. G-box 5 may be an inactive motif as it does not affect gene regulation. G-box 10 appears to act as a constitutive enhancer in that it confers a strong expression in seeds, cotyledons and roots (Fig. 1b).
Microscopically (Fig. 1c), GUS staining was evident in cotyledons, roots and seeds of transgenic plants with the CaMV 35S promoter. Compared to the CaMV 35S promoter, the G-box 10 promoter produced a much higher expression level in cotyledons, roots and seeds. The expression level of the G-box 10 promoter in seeds was extremely high and more than a 30-fold increase in activity was evident compared with the CaMV 35S promoter. GUS activity in the cotyledons was higher in G-box 10 plants relative to plants containing the CaMV 35S promoter. On average, G-box 10 plants contained at least a fourfold higher GUS activity in roots compared to plants containing the CaMV 35S promoter. Thus, our results demonstrate that the G-box 10 promoter confers a higher expression level than the 35S promoter in almost all tissues, including seeds, cotyledons and roots of tobacco plants. In five independent transgenic tobacco lines, the G-box 10 tetramer strongly enhanced gene expression, on average, by 390-fold in seeds, by 550-fold in cotyledons and by 42-fold in roots, when compared with the CaMV –90/35S construct.
Histochemical analysis of expression patterns driven by the G-box 10 promoter in transgenic tobacco plants
To examine gene expression patterns conferred by the G-box 10 promoter, histochemical assays were performed on thin sections of leaves, stems, roots and floral organs of transgenic R1 tobacco plants. In cross-sections of leaves, strong expression was detected in the palisade and spongy tissues, and stomata (Fig. 2a). In cross-sections of the stem, G-box 10 plants showed a high-level expression in epidermis and cells in the outer cortial layers. In longitudinal sections of roots, the G-box 10 conferred high expression in vascular bundle, root cortex and tip, but not in root cap. Strong expression was also found in vascular bundles and axillary buds (data not shown) of stems in G-box 10 plants. We found no GUS staining in leaf and stem of plants containing the CaMV –90/35S promoter, although light staining was occasionally evident in the root cortex.
Strong GUS staining was observed in longitudinal sections of flower buds from G-box 10 plants (Fig. 2b). In mature flowers, the upper regions of styles just below the stigma and the outer integuments of the ovary showed GUS staining. The strong staining observed in anthers was restricted to the inside pollen sacs and an exclusively high expression was detected in the pollen grains in G-box 10 plants (Fig. 2b). In plants containing the CaMV –90/35S construct, no staining was seen in sections of flowers, but sometimes a weak staining was seen in pollen grains, and the upper region of styles.
Enhanced and constitutive expression patterns of the G-box 10 promoter in transgenic carrots and rice plants
Figure 3(a) shows that the G-box 10 tetramer also conferred a strong expression in transgenic carrot seedlings and seeds. In cross-sections of petioles, the vascular bundles, surrounding tissues and collenchyma exhibited GUS staining. High-level expression was also detected in the secondary phloem and cambium of auxetic roots. GUS activity conferred by the G-box 10 promoter was approximately 13-fold higher than that of the CaMV 35S promoter in cotyledons of transgenic carrot plants.
A strong expression pattern by the G-box 10 enhancer was also observed in transgenic monocot rice plants. Strong staining was detected in the outer aleurone layer of seeds (Fig. 3b). High-level expression was also found in leaves, roots and basal region of shoots of G-box 10 seedlings. G-box 10 rice seedlings showed clear increases in GUS activity of approximately 90-fold in seeds, 8-fold in cotyledons and 440-fold in roots, as compared to findings in CaMV –90/35S seedlings (Fig. 3b). We suggest that the G-box 10 promoter strongly enhances gene expression in leaves, roots and seeds in a variety of plants, including both dicot and monocot plants. We also confirmed in transient expression assays strong activity of the G-box 10 promoter in rapeseed, soybean, wheat and corn (data not shown).
Among 11 chimeric promoters containing different G-boxes, the G-box 10 motif provided high-level, constitutive expression. Compared with tetramer of the G-box 10 motif, the monomer exhibited a somewhat weaker enhancement and the octamer approximately a two-fold higher one. Increasing the copy number of the G-box 10 sequence appears to enhance gene expression from the CaMV –90/35S promoter. Only one nucleotide at the flanking nucleotides was different in G-box 1, 3 and 9 compared with G-box 10. This means that the flanking sequence of the G-box core has important roles for determining the tissue preference and the level of gene expression in vivo. Tissue-specific modifications of G-box binding proteins, tissue-specific expression of binding proteins or interactions with additional factors expressed preferentially in certain tissues may be relevant for in vivo regulation of gene expression.
Several ABA-responsive elements are G-box-related sequences (Marcotte et al. 1989; Ono et al. 1996). To determine whether the 11 different G-box elements would respond to different signalling compounds such as ABA and methyljasmonate, seedlings of G-box plants were treated with these compounds. Enhancement of gene expression was never evident in response to ABA and methyljasmonate (data not shown).
The four tandemly repeated G-box 10 motifs contain three repeated GCCGCC sequences at junctions of G-box 10 motifs. The GCCGCC sequence (GCC-box) in the chitinase gene and the PRB-I gene from tobacco contains TAAGAGCCGCC as an essential sequence for binding of the GCC-box binding factor and enhancement of gene expression in response to ethylene (Shinshi et al. 1995). The flanking sequence of GCCGCC differed between the G-box 10 repeated sequence and the GCC-box. Furthermore, the GCC-box motif placed upstream of the CaMV –90/35S region did not confer a high-level, constitutive expression (Shinshi et al. 1995). It is unlikely that GCC-binding factors are directly involved in high-level expression conferred by the G-box 10 tetramer.
Co-operative interaction of two or three regulatory elements placed in close proximity controlled gene expression at the transcriptional level (Faktor et al. 1997; Sessa et al. 1995). In addition, the ocs binding protein that binds to ACGT elements interacted with an ethylene-inducible GCC binding protein (Büttner & Singh 1997). Hence, the G-box motif shows co-operative interaction with other motifs, and G-box binding factors might form a protein–protein interaction with other transacting factors to confer concerted gene expression. The –46/35S construct, which contains only a TATA binding site of the 35S promoter, can be used as a minimal promoter to form the constructs, including the G-box motifs. However, these G-box motifs/–46 constructs are not expected to confer high-level expression because of loss of other motif sequences, which can be recognized by other trans-acting factors. Actually, the G-box 10/–46 construct significantly enhanced gene expression in tobacco, but expression was extremely low compared with that of the G-box 10/–90 construct. The G-box 10 is a potent enhancer element and elevated effects of gene expression might be determined by interactions with other motifs in a promoter context. The preliminary gel shift analysis suggested that one candidate binding factor exists in tobacco nuclear extracts. The bZip proteins, TGA1a and TGA1b, which are major trans-acting factors to the as-1 element of –90/35S promoter (Katagiri et al. 1989), might interact with the binding factor of the G-box 10 motif. Characterization of the binding protein will elucidate the method of gene expression through the G-box 10 tetramer enhancer.
Construction of chimeric genes
The CaMV –90/35S promoter region (–90 to +8) was cloned into a BamHI site of the binary plasmid pBI101.3 (Clonetech, Palo Alto, California, USA). The constructed plasmid was named pBI-90GUS. Tetramer sequences of 11 different G-boxes, referred to as G-box 1 to G-box 11 (Fig. 1A), which consist of a common six nucleotide-core sequence CACGTG and two different flanking nucleotide sequences at their 5′-and 3′-ends, and their complementary sequences with HindIII and XbaI overhangs at the 5′ and 3′ ends, respectively, were synthesized using a DNA synthesizer (Applied Biosystems, Foster City, California, USA). Complementary oligonucleotides were annealed and directly introduced into the HindIII and XbaI restriction site upstream of the CaMV –90/35S promoter.
Transformation of tobacco, carrot and rice plants
Nicotiana tabacum cv. SR1 was transformed by co-cultivation of leaf-discs with Agrobacterium tumefaciens LBA4404 containing the appropriate binary plasmids (Horsch et al. 1985).
Transgenic carrots were generated according to the procedure of Pawlicki et al. (1992). Hypocotyl segments of Daucus carota L. cv. Nantes Scarlet were infected by A. tumefaciens LBA4404 containing the appropriate binary plasmid.
Genomic DNAs for PCR experiments were prepared using Isoplant (Nippongene, Japan).
Assay of GUS activity and histochemical analysis
Enzymatic assays of β-glucuronidase activity and staining with X-Gluc (5-bromo-4-chloro-3-indolyl glucuronide) were carried out according to Jefferson et al. (1987).
We thank Dr T. Shimada for production of transgenic rice and M. Iwai and S. Sakamoto for technical assistance. M. Ohara provided helpful comments on the manuscript.