The G‐box (CACGTG) and H‐box (CCTACC) cis elements function in the activation of phenylpropanoid biosynthetic genes involved in the elaboration of lignin precursors, phytoalexins and the secondary signal salicylic acid as early responses to pathogen attack. We have isolated a soybean cDNA encoding a novel bZIP protein, G/HBF‐1, which binds to both the G‐box and adjacent H‐box in the proximal region of the chalcone synthase chs15 promoter. While G/HBF‐1 transcript and protein levels do not increase during the induction of phenylpropanoid biosynthetic genes, G/HBF‐1 is phosphorylated rapidly in elicited soybean cells, almost exclusively on serine residues. Using recombinant G/HBF‐1 as a substrate, we identified a cytosolic protein‐serine kinase that is rapidly and transiently stimulated in cells elicited with either glutathione or an avirulent strain of the soybean pathogen Pseudomonas syringae pv. glycinea. Phosphorylation of G/HBF‐1 in vitro enhances binding to the chs15 promoter and we conclude that stimulation of G/HBF‐1 kinase activity and G/HBF‐1 phosphorylation are terminal events in a signal pathway for activation of early transcription‐dependent plant defense responses.
Plants respond to pathogen avirulence signals by the elaboration of inducible defenses including oxidants, phytoalexins, cell wall modifications and deployment of pathogenesis‐related (PR) proteins, such as chitinase, with antimicrobial activities (Briggs, 1995). Defense induction associated with the expression of localized hypersensitive resistance is observed in the early stages of incompatible interactions following attempted infection by a non‐pathogen or avirulent strain of a pathogen (Lamb et al., 1989). In some cases, these responses are also observed in distant tissue, associated with the expression of systemic acquired resistance in which immunity to a broad range of normally virulent pathogens gradually develops throughout the plant (Ryals et al., 1994). Defense responses can also be induced by defined molecules including microbial glycan and peptide elicitors, or metabolites such as arachidonic acid, glutathione and salicylic acid (Ebel and Cosio, 1994).
With the exception of callose production and oxidative responses including cross‐linking of cell wall structural proteins, induction of this battery of defenses involves a massive switch in host gene expression (Dixon et al., 1994), and there is a temporal and spatial hierarchy of defense gene activation with some genes exhibiting rapid, localized activation, whereas many, but not all, PR protein genes undergo slower activation, both locally and then systemically throughout the plant (Ryals et al., 1994; Hahlbrock et al., 1995). Prominent among the class of rapidly induced defense genes are those encoding enzymes of phenylpropanoid biosynthesis involved in the synthesis of lignin precursors and a number of phytoalexins (Dixon and Paiva, 1995). In addition, the phenylpropanoid‐derived metabolite salicylic acid is a signal potentiating the expression of both localized hypersensitive resistance and systemic acquired resistance (Gaffney et al., 1993; Delaney et al., 1994; Mauch‐Mani and Slusarenko, 1996). The early activation of phenylpropanoid biosynthetic genes is correlated with the expression of hypersensitive resistance in tissues challenged with an avirulent pathogen or non‐pathogen (Dixon and Paiva, 1995; Hahlbrock et al., 1995), and in parsley and bean cell suspension cultures transcription of these genes is transiently activated within 10–20 min of elicitor treatment, with maximum rates of transcription after ∼1 h and maximum accumulation of transcripts after 3–4 h (Chappell et al., 1984; Cramer et al., 1985; Lawton and Lamb, 1987).
Delineation of the cis elements and cognate trans factors underlying the rapid activation of phenylpropanoid biosynthetic genes provides the basis for characterizing the terminal stages of a signal transduction pathway involved in the deployment of early transcription‐dependent defenses and the generation of salicylic acid as a secondary signal. Two cis elements, the G‐box (CACGTG) and H‐box (CCTACC), are found in the proximal region of the promoters of a number of genes encoding phenylpropanoid biosynthetic enzymes including phenylalanine ammonia‐lyase (pal) and 4‐coumarate:CoA ligase, which catalyze the first and third steps respectively in the central, common pathway of phenylpropanoid biosynthesis, and chalcone synthase (chs), which catalyzes the first step in a branch pathway specific for the synthesis of flavonoids and isoflavonoid‐derived pterocarpan phytoalexins (Lois et al., 1989; Ohl et al., 1990). These cis elements are the sites of elicitor‐inducible in vivo footprints in the parsley pal1 promoter (Lois et al., 1989) and likewise an H‐box in the chs15 promoter is the location of an elicitor‐inducible DNase I‐hypersensitive site in chromatin from bean cells (Lawton et al., 1990). G‐ and H‐box functions in chs15 expression have been confirmed by functional analysis of the effects of promoter mutations on reporter gene expression in electroporated protoplasts (Dron et al., 1988; Loake et al., 1992), stably transformed cells and transgenic plants (O.Faktor, R.A.Dixon and C.Lamb, unpublished data), as well as by in vitro transcription assays (Arias et al., 1992; W.P.Lindsay, R.A.Dixon and C.Lamb, unpublished data).
The G‐box functions in the regulation of diverse genes by developmental cues, abscisic acid, light, UV irradiation and wounding as well as pathogen signals, and a family of bZIP proteins that bind G‐box sequences has been described (Foster et al., 1994; Menkens et al., 1995). Functional specificity appears to be determined by DNA‐binding affinities governed by nucleotides immediately flanking the G‐box ACGT core and by combinatorial interactions with other cis element–trans factor systems (Williams et al., 1992; Izawa et al., 1993; Menkens et al., 1995). However, only the in vivo function of opaque2, as a regulator of zein expression, has been fully established (Schmidt et al., 1992), and while transgenic manipulation of CPRF‐1 disrupts light induction of parsley chs (Feldbrügge et al., 1994), there is no information on the function of G‐box factors in elicitor or pathogen activation of phenylpropanoid biosynthetic genes. Likewise, a flower‐specific Myb transcription factor, which stimulates pal2 transcription through binding to the H‐box, appears to function in developmental regulation and there is no report of its involvement in elicitor or pathogen induction (Sablowski et al., 1994, 1995). A novel factor from parsley, BFP‐1, binds to an AC‐rich element related to the H‐box that is also in vivo footprinted in elicited parsley cells (da Costa e Silva et al., 1993). BFP‐1 transcripts are induced in elicited cells and infected tissues, consistent with a function in defense gene regulation, although it is not clear whether the activity of this factor is directly modulated by elicitor or pathogen signals for the rapid initial stimulation of early defense genes. Polypeptides of 97 and 56 kDa, designated KAP‐1 and KAP‐2 respectively, which bind to H‐boxes in the bean chs15 promoter, have been purified by DNA affinity chromatography (Yu et al., 1992). Elicitation with glutathione does not affect the total cellular activities of KAP‐1 or KAP‐2, but causes a rapid increase in the specific activities of both factors in the nuclear fraction, consistent with a role in the induction of chs15 and related defense genes.
While these studies have identified a number of trans factors which might function in the elicitor or pathogen induction of phenylpropanoid biosynthetic genes, no picture has yet emerged of how the functional activities of factors interacting with the G‐box and H‐box cis elements are regulated in the initial stimulation of early defense gene transcription. In the present study, we describe a novel bZIP protein, G/HBF‐1, which binds to the G‐box and the adjacent H‐box in the proximal region of the chs15 promoter. While G/HBF‐1 transcript and protein levels do not increase during the induction of pal and chs transcription, G/HBF‐1 is phosphorylated rapidly in elicited cells. Using recombinant G/HBF‐1 as a substrate, we identify a cytosolic protein‐serine kinase activity that is rapidly and transiently stimulated in elicited cells and demonstrate that G/HBF‐1 phosphorylation enhances binding to the chs15 promoter. These observations delineate a terminal event in a signal pathway for activation of early transcription‐dependent defense responses and indicate that in plants, as in animals, stimulus‐dependent transcription factor phosphorylation contributes to the selective regulation of gene expression.
Isolation of a soybean cDNA encoding G/HBF‐1, a novel bZIP DNA‐binding protein
Two soybean (Glycine max) cDNA libraries were constructed in λgt11 and λZAP‐II vectors using respectively poly(A)+ RNA from control soybean cell suspension cultures and a mixture of poly(A)+ RNA from control cells and cells 4 h after elicitor treatment. A total of 3×105 plaques from each library were probed with the −80 to −42 fragment of the bean chs15 elicitor‐inducible promoter. This fragment contains both the G‐box and adjacent, TATA‐proximal H‐box (H‐box III). A clone, designated λG/HBF‐1, expressing a protein that strongly bound this promoter fragment, was isolated from the λgt11 library. λG/HBF‐1 contained a 1.4 kb insert, and repeated screening of the λgt11 and λZAP‐II cDNA libraries with this insert as a probe failed to identify hybridizing clones with larger inserts.
The λG/HBF‐1 cDNA contains a single long open reading frame of 1134 bp encoding a protein of 41 kDa (Figure 1). The putative ATG start codon is flanked by a nucleotide sequence optimal for translation initiation in plants (Lütke et al., 1987). The deduced protein product shows characteristic features of a bZIP transcription factor including a highly basic, putative DNA‐binding domain and a leucine zipper domain in which every seventh amino acid residue is leucine or another small hydrophobic residue (Figure 1). G/HBF‐1 also contains two domains rich in proline and acidic amino acids respectively.
Protein sequence alignments reveal substantial similarities in the bZIP region to the equivalent regions of other plant bZIP proteins such as RITA‐1 (Izawa et al., 1994) or opaque2 (Schmidt et al., 1992). However, while the C‐terminal half of the basic region is highly conserved among G/HBF‐1 and many other plant bZIP factors binding G‐box or related motifs, the N‐terminal part of the basic region of G/HBF‐1 only exhibits a high degree of similarity to the common plant regulatory factor 2 (CPRF‐2) from parsley (Weisshaar et al., 1991). Four other domains, designated D1–D4, are highly conserved among G/HBF‐1, CPRF‐2 (Weisshaar et al., 1991) and the maize opaque2 heterodimerizing proteins OHP1 and OHP2 (Pysh et al., 1993). D1, which is located half way between the N‐terminus and the basic domain, comprises a peptide predicted to form a helix, and D2, which is located just N‐terminal of the basic domain, is relatively rich in acidic residues. D3, which is adjacent to the bZIP region, is similar to a domain also present in some animal transcription factors including the helix–loop–helix protein MyoD (Scales et al., 1990). D4, located near the C‐terminus, shows no distinctive structural features. Overall, G/HBF‐1 showed a high degree of similarity to CPRF‐2 (67% identity in 361 amino acids) and OHP1 and OHP2 (48% identity in 339 amino acids), and these sequence comparisons define a sub‐family of plant G‐box‐binding bZIP proteins. Southern blots of soybean genomic DNA probed with G/HBF‐1 sequences at high stringency showed only one or two hybridizing bands in a range of restriction endonuclease digestions (data not shown), indicating that G/HBF‐1 is likely to be encoded by a single copy gene.
G/HBF‐1 binding to the chs15 G‐box and H‐box III
DNA binding by G/HBF‐1 was examined in experiments using radiolabeled oligonucleotides to probe plaque lawns of λG/HBF‐1 (Table I) and by gel retardation assays with purified recombinant G/HBF‐1 (Table I and Figure 3). G/HBF‐1 cDNA was fused to the T7 promoter in the vector pET‐28a (Novagene), and the recombinant G/HBF‐1 carrying an N‐terminal hexameric histidine peptide tag, G/HBF‐1(His6), was purified from Escherichia coli extracts by immobilized Ni affinity chromatography. Gel retardation assays with the recombinant factor demonstrated binding to the −80 to −42 chs15 promoter sequence containing both the G‐box (−72 to −67) and TATA‐proximal H‐box (H box III, −59 to −53) and to each of these two cis elements when tested separately (Figure 3A). The major binding complex formed with the −80 to −42 sequence had a similar electrophoretic mobility to the complexes formed with either the chs15 G‐box sequence CACGTG (−74 to −69) or the extended H‐box III sequence TCACCTACCCTA (−65 to −53) when tested separately, suggesting that G/HBF‐1 binding was mainly at one location when both cis elements were in close proximity. Incubation of G/HBF‐1 with the −80 to −42 sequence also generated a second, low abundance complex with increased electrophoretic mobility not observed with either cis element alone.
Nucleotide sequence requirements for G/HBF‐1 binding were analyzed by direct binding of test oligonucleotides to λG/HBF‐1 plaque lawns and by competition with the −80 to −42 chs15 promoter fragment for binding to recombinant G/HBF‐1(His6) in gel retardation assays. Mutation of the G‐box sequence GCACGTGA to GCgtacGA abolished binding when tested in isolation from H‐box III. However, the core ACGT was not sufficient for G/HBF‐1 binding since the as‐1 sequence TGACGTT was not recognized. Likewise, the parsley chs G‐box, CCACGTGG, involved in light regulation (Weisshaar et al., 1991), was not recognized, indicating that not only the ACGT core but also immediately flanking nucleotides appear to be important for G/HBF‐1 binding. Interestingly, G/HBF‐1 bound to a larger fragment of the parsley chs promoter containing not only the G‐box but an immediately downstream sequence with an AC‐rich cis element resembling the H‐box.
Mutation of the H‐box core motif CCTACC to CCatCC or aaTAaa severely impaired binding to the H‐box III cis element when tested in isolation from the G‐box, whereas binding was observed with the sequence CCTAtt. However, no binding was observed with upstream fragments of the chs15 promoter‐containing H‐box I (−159 to −135) or H‐box II (−139 to −113) in which the CCTACC motif is embedded in different flanking sequences compared with the TATA‐proximal H‐box III (Yu et al., 1992). Thus, sequences in addition to the core H‐box motif are important for G/HBF‐1 binding to H‐box III and, when immediately flanking sequences are taken into account, a single mutation in H‐box III, CACCTACC to CACgTACC, generates an almost perfect second version of the chs15 G‐box (CACGTG). The CACGTACC sequence, intermediate between the G‐box and H‐box III, was also bound by G/HBF‐1.
G/HBF‐1 regulation during chs induction
G/HBF‐1 sequences hybridized at high stringency to a single transcript in total cellular RNA isolated from uninduced soybean cells (Figure 4). The size of the transcript was 1.4 kb, consistent with the size of the full‐length G/HBF‐1 cDNA. Elicitation of soybean cells with reduced glutathione caused little change in the level of G/HBF‐1 transcripts, whereas chs transcripts rapidly accumulated from low basal levels (Figure 4A).
To monitor G/HBF‐1 regulation at the protein level, we generated a panel of polyclonal antisera, including two peptide antibodies, α‐G/HBF‐1(P2) and α‐G/HBF‐1(P4), to two internal peptides showing little similarity to the peptide sequences found in the corresponding regions of otherwise closely related bZIP transcription factors such as CPRF‐2 (Figure 2). α‐G/HBF‐1(P2) bound to a single protein in Western blots of total cellular protein (Figures 4 and 5A), whereas α‐G/HBF‐1(P4) and a cross‐reacting antibody, α‐OHP, raised against the maize OHP C‐terminal region, which shows very strong peptide sequence identity to the corresponding region of G/HBF‐1, bound to two protein species exhibiting almost identical electrophoretic mobilities (Figure 5A). The protein species recognized by these antibodies were found in both cytosolic and nuclear fractions and neither their overall abundance nor distribution between cytosol and nucleus changed appreciably during glutathione induction of chs transcription. However, antibody supershift gel retardation experiments indicated that G/HBF‐1 was involved in a chs15 promoter–nuclear protein binding complex induced in elicited cells (Figure 5B). The major binding complex formed by incubation of soybean nuclear extracts with the −80 to −42 region of the chs15 promoter was neither elicitor regulated nor supershifted by incubation with α‐G/HBF‐1(P4) prior to gel retardation analysis. In contrast, the other prominent complex was observed only with nuclear extracts from elicited cells, and pre‐incubation with α‐G/HBF‐1(P4) caused a marked reduction in the electrophoretic mobility of this complex. This supershifted DNA‐binding complex co‐migrated with the DNA‐binding complex formed with recombinant G/HBF‐1 in the presence of α‐G/HBF‐1(P4).
Phosphorylation of G/HBF‐1
Incorporation of G/HBF‐1 into a chs15 promoter‐binding complex in elicited cells implied post‐translational regulation of G/HBF‐1, and we next investigated whether G/HBF‐1 was phosphorylated in vivo by labeling soybean cells with [32P]phosphate. Immunoprecipitation of G/HBF‐1 from extracts prepared after exposure of cells to [32P]phosphate resulted in the incorporation of radioactivity into a single protein of the appropriate electrophoretic mobility (Figure 6B). Moreover, several fold greater incorporation of 32P into immunoprecipitable G/HBF‐1 was observed in [32P]phosphate‐labeled cells after treatment with 0.5 mM glutathione than in equivalent pulse‐labeled control cells. Phosphoamino acid analysis of 32P‐labeled, immunoprecipitable G/HBF‐1 demonstrated that phosphorylation was almost exclusively on serine residues, with little detectable phosphothreonine (1–5%) and no phosphotyrosine (Figure 6C).
α‐G/HBF‐1(P4) gave strong immunoreactivity with recombinant G/HBF‐1, whereas α‐G/HBF‐1(P2) did not react with the bacterially expressed protein (Figure 6E). However, incubation of recombinant G/HBF‐1 with soybean whole cell extracts in the presence of ATP resulted in the appearance of strong α‐G/HBF‐1(P2) immunoreactivity in vitro (Figure 6E). Inhibition of the appearance of α‐G/HBF‐1(P2) immunoreactivity by inclusion of either EDTA or K252A and staurosporine in the protein kinase reactions closely followed the effects of these protein kinase inhibitors on the labeling of recombinant G/HBF‐1 with 32P from [γ‐32P]ATP in the same reactions (Figure 6E). Moreover, alkaline phosphatase treatment of soybean cytosolic extracts resulted in the loss of α‐G/HBF‐1(P2) immunoreactivity with native G/HBF‐1, indicating that α‐G/HBF‐1(P2) recognizes a phosphorylation‐dependent conformation of G/HBF‐1 (Figure 6A).
Elicitor and pathogen activate G/HBF‐1 kinase
To investigate further the phosphorylation control of G/HBF‐1, Ni affinity‐purified, recombinant factor was incubated with [γ‐32P]ATP and cytosolic or nuclear extracts prepared from soybean cells at various times after elicitation with glutathione (Figure 7A). No in vitro phosphorylation of recombinant G/HBF‐1 was observed following incubation with nuclear extracts from control or elicited cells, although the nuclear extract from cells 30 min after elicitation gave substantial phosphorylation of an endogenous substrate of apparent Mr ∼33 kDa. In contrast, in vitro phosphorylation of recombinant G/HBF‐1 was observed following incubation with the cytosolic fraction, together with phosphorylation of an endogenous substrate of the same electrophoretic mobility as the endogenous nuclear substrate.
Glutathione induction of soybean cells caused a rapid, transient increase in extractable G/HBF‐1 kinase activity as measured by the incorporation of 32P from [γ‐32P]ATP into recombinant G/HBF‐1 substrate (Figure 7A). As an internal control, no elicitor induction of protein kinase activity was observed using exogenous histone H1 as a substrate. There was little or no apparent lag for stimulation of G/HBF‐1 kinase activity, with maximum activity, 5‐fold above uninduced controls, in extracts prepared from cells ∼30 min after elicitation, followed by rapid decay to basal levels after 60 min (Figure 7B). This transient increase in extractable G/HBF‐1 kinase activity was concomitant with the onset of rapid accumulation of chs transcripts (Ryder et al., 1984 and Figure 4). Inoculation of soybean cells with avirulent Pseudomonas syringae pv. glycinea also caused a marked stimulation in extractable G/HBF‐1 kinase activity in the cytosolic fraction, as measured either by the incorporation of 32P from [γ‐32P]ATP into recombinant G/HBF‐1 substrate or by probing Western blots of the products of equivalent non‐radioactive assay reactions with α‐G/HBF‐1(P2), which is specific for phosphorylated G/HBF‐1 (Figure 7C).
Phosphorylation promotes G/HBF‐1 binding to DNA
As was demonstrated in vivo, phosphorylation of G/HBF‐1 in vitro was also specific for serine (data not shown), and analysis of the pattern of 32P‐labeled peptides generated by trypsin digestion of recombinant G/HBF‐1 immunoprecipitated from the in vitro protein kinase reaction with extracts from elicited cells revealed four labeled peptides, indicating phosphorylation at multiple sites (Figure 6D). To determine whether phosphorylation modulates the functional properties of G/HBF‐1, we examined the effect of in vitro phosphorylation on factor binding to the chs15 H‐box III, monitored by Southwestern blot analysis of the protein kinase reaction products. In vitro phosphorylation of G/HBF‐1 by cytosolic extracts from cells elicited with either glutathione or inoculation with avirulent P.syringae pv. glycinea resulted in markedly enhanced cis element‐binding activity compared with that resulting from protein kinase reactions with extracts from equivalent unstimulated cells (Figure 8A).
The effect of in vitro phosphorylation on G/HBF‐1 binding to H‐box III was confirmed in gel retardation experiments. Incubation of recombinant G/HBF‐1(His6) with ATP and cytosolic extracts from cells isolated 20–30 min after elicitation with glutathione resulted in the generation of a high molecular weight DNA‐binding complex substantially larger than that formed by non‐phosphorylated recombinant factor and cis element in the absence of active cytosolic extracts (Figure 8B). Formation of this large complex was dependent on the addition of both recombinant factor and ATP (Figure 8B). Extracts from control cells only weakly supported formation of this complex, but elicitation caused a rapid, transient increase in the extractable activity driving complex formation, with similar kinetics to those for stimulation of extractable G/HBF‐1 kinase activity.
λG/HBF‐1 cDNA encodes a 41 kDa protein with the characteristic features of bZIP transcription factors, including a basic domain involved in DNA binding adjacent to a leucine zipper involved in factor dimerization. G/HBF‐1 is most similar to three other plant bZIP proteins, CPRF‐2 from parsley (Weisshaar et al., 1991) and OHP1 and OHP2 from maize (Pysh et al., 1993). Four conserved domains, D1–D4, can be discerned in these four trans factors that are not found in other plant bZIP proteins. While D4 does not exhibit any distinctive structural features, the predicted helical structure of D1 may promote specific intermolecular interactions and, by analogy to the role of acidic domains in other transcription factors (Hope and Struhl, 1986), D2 may contribute to transcriptional activation. D3 is adjacent to the leucine zipper and deletion of the equivalent domain in CPRF‐2 reduces DNA‐binding affinity (Armstrong et al., 1992). This domain is also found in certain helix–loop–helix transcription factors such as MyoD (Scales et al., 1990) and may function in dimer stabilization. The conservation of these four domains in CPRF‐2, OHP1, OHP2 and G/HBF‐1 suggests that this group of bZIP proteins share distinctive functional attributes. While opaque2 has been implicated in the expression of zein storage protein genes (Schmidt et al., 1992), the biological functions of OHP1 and OHP2, which were isolated through their interactions with opaque2, have not been established (Pysh, 1994). Likewise, while CPRF‐1 appears to contribute to the light regulation of parsley chs, no function has yet been ascribed to CPRF‐2 (Weisshaar et al., 1991; Feldbrügge et al., 1994). It is unlikely that G/HBF‐1 is the soybean functional ortholog of parsley CPRF‐2 since we have isolated sequences from tobacco and Arabidopsis much more closely related to G/HBF‐1 (95% identity) than is CPRF‐2 (our unpublished data).
One group of plant bZIP proteins including GBF, CPRF‐1 and Taf‐1 preferentially bind to the G‐box CACGTG, while another group, the TGA1‐like family, preferentially bind to the C‐box, TGACGTC (Williams et al., 1992; Izawa et al., 1993). G/HBF‐1 appears to fall in a third class, which also includes CPRF‐2, opaque2 and RITA‐1, exhibiting relaxed binding specificity. For example, opaque2 binds an imperfect C‐box, ATCAGTCAT, lacking a complete ACGT core (de Pater et al., 1994). Likewise, G/HBF‐1 binds to the extended H‐box III sequence, CACCTACC, which does not contain a perfect palindromic ACGT core. While disruption of the ACGT core in the chs15 G‐box GCACGTGA can prevent G/HBF‐1 binding, this sequence is not sufficient for binding since G/HBF‐1 does not recognize either the parsley chs G‐box CCACGTGG or the as‐1 element TGACGTGG, and G/HBF‐1 binding is clearly influenced by the nucleotides flanking the ACGT core. Moreover, G/HBF‐1 does not bind to H‐boxes I or II, which share with H‐box III the core CCTACC motif, and G/HBF‐1 probably binds to the H‐box III region by virtue of its resemblance to an ACGT‐like cis element when nucleotides immediately 5′ of this H‐box are taken into account, rather than binding to the H‐box core CCTACC, which is a canonical Myb‐binding site (Sablowski et al., 1994, 1995).
Binding to two adjacent ACGT‐related cis elements implicated in the induction of phenylpropanoid biosynthetic genes in elicited cells is consistent with a role for G/HBF‐1 in the activation of early transcription‐dependent defenses, and several lines of evidence indicate that rapid phosphorylation of G/HBF‐1 represents a terminal step in an elicitor‐activated signal pathway. Thus, while G/HBF‐1 transcript and protein levels do not change during the induction of chs transcription, antibody supershift analysis implicates G/HBF‐1 in the formation of a binding complex with the −80 to −42 region of the chs15 promoter specifically in elicited cells and there is a rapid, marked stimulation of G/HBF‐1 phosphorylation in vivo. Moreover, recombinant G/HBF‐1 is a substrate in vitro for a cytosolic protein‐serine kinase that is stimulated rapidly in cells treated with glutathione, which closely mimics the effects of microbial elicitors (Wingate et al., 1988), or in cells inoculated with avirulent P.syringae pv. glycinea, which induces a hypersensitive response (Levine et al., 1994). Glutathione stimulation of extractable G/HBF‐1 kinase activity occurs without detectable lag and hence precedes the stimulation of chs transcription first observed 5–10 min after elicitation, and maximal levels of extractable G/HBF‐1 kinase activity are attained ∼30 min after elicitation, concomitant with the onset of rapid accumulation of chs transcripts. Likewise, stimulation of extractable G/HBF‐1 kinase activity is an early event in the hypersensitive response to avirulent P.syringae pv. glycinea.
α‐G/HBF‐1(P2) was generated by immunization with a synthetic peptide hapten corresponding to a region spanning the C‐terminus of the D1 domain. The loss of reactivity of native plant G/HBF‐1 with this antibody following treatment of plant extracts with alkaline phosphatase implies that dephosphorylation causes a major conformational change in which the P2 peptide becomes inaccessible to the antibody. Moreover, α‐G/HBF‐1(P2) does not react with G/HBF‐1(His6) expressed in E.coli, and the observation that incubation of the recombinant factor with ATP and cytosolic extracts from elicited cells generates α‐G/HBF‐1(P2)‐reactive G/HBF‐1(His6) in parallel with factor phosphorylation indicates that the in vitro kinase reaction faithfully reproduces a key effect of phosphorylation of native G/HBF‐1 in vivo and that the phosphorylation‐dependent conformational change is reversible.
Reversible phosphorylation controls the functional activity of many animal transcription factors by modulation of one or more of the following attributes: DNA‐binding affinity, transactivation, interactions with other regulatory proteins and cellular localization (Hunter and Karin, 1992). Incubation with ATP and active cytosolic extracts promotes the binding of recombinant G/HBF‐1(His6) to the cognate cis elements in the proximal region of the chs15 promoter as monitored by both gel retardation and Southwestern blot analysis of the kinase reaction products. The latter experiment, in which enhanced DNA binding to the recombinant substrate was monitored following electrophoretic fractionation of the reaction products, implies that phosphorylation of G/HBF‐1 contributes to the enhanced DNA binding, although we do not rule out secondary effects, mediated by endogenous accessory factors in the cytosolic extracts, in the generation of the high molecular weight DNA‐binding complex observed in gel retardation assays. In animals, phosphorylation control of the DNA‐binding affinities of transcription factors is usually negative, e.g. c‐Myb and c‐Jun, and examples of phosphorylation directly promoting DNA binding are rare. However, G/HBF‐1 regulation is reminiscent of the stimulation of serum response factor binding to the c‐fos promoter following factor phosphorylation in response to epidermal growth factor (Janknecht et al., 1992). The mechanism whereby phosphorylation affects DNA binding is unknown, but phosphorylation induces a conformational change in serum response factor (Manak and Prywes, 1991). In G/HBF‐1, the phosphorylation‐induced exposure of the D1 domain may reflect a conformational change, possibly activated by changes in charge distribution following phosphorylation, that concomitantly exposes the basic domain thereby promoting DNA binding. Such a phosphorylation‐induced conformational change could also enhance dimerization, mediated by the adjacent leucine zipper region, or other protein–protein interactions, and indeed preliminary evidence indicates that G/HBF‐1 undergoes conformational‐sensitive interactions with the H‐box‐binding factor KAP‐2, which also binds to the proximal region of the chs15 promoter (W.P.Lindsay and W.Dröge‐Laser, unpublished data).
Elicitor‐stimulated G/HBF‐1 kinase activity is exclusively cytosolic, with no corresponding activity detectable in nuclear extracts. Hence, unlike for example growth factor‐stimulated ERK MAP kinase (Chen et al., 1992), elicitation does not result in migration of the G/HBF‐1 kinase to the nucleus. However, the phosphorylation‐dependent conformational change that exposes the D1 domain of G/HBF‐1 might also uncover nuclear localization sequences and hence promote migration of the trans factor substrate to the nucleus following elicitor stimulation of the cytosolic kinase. Strikingly, three serine residues are embedded within the cluster of basic amino acids constituting the putative nuclear localization signal within the basic domain (Varagona et al., 1992), raising the alternative possibility that phosphorylation governs its interaction with the nuclear pore complex (Moll et al., 1991). Western blot analysis failed to reveal major changes in the distribution of either total G/HBF‐1 or the phosphorylated fraction between cytosol and nucleus during chs induction, and hence a second factor may be limiting for G/HBF‐1 movement to the nucleus. KAP‐2, which interacts with the activated conformation of G/HBF‐1, is a likely candidate since this trans factor is present at very low abundance and, based on measurements of DNA‐binding activity, appears to migrate from cytosol to nucleus in elicited cells (Yu et al., 1992).
The involvement of protein phosphorylation in the activation of inducible defense mechanisms, inferred from physiological and pharmacological studies (Dietrich et al., 1989; Felix et al., 1991; Levine et al., 1994; Suzuki and Shinshi, 1995), was confirmed recently by reports that the tomato Pto and rice Xa21 disease resistance genes encode protein‐serine/threonine kinases (Martin et al., 1993; Loh and Martin, 1995; Song et al., 1995). Moreover, the Pto kinase phosphorylates a second protein‐serine kinase, Pti1, that is also involved in the hypersensitive response, demonstrating the operation of a phosphorylation signal cascade in the expression of disease resistance (Zhou et al., 1995). Elicitation of potato tuber discs with arachidonic acid is correlated with a staurosporine‐sensitive increase in the extractable activity of a trans factor designated PBF‐1, which binds to the promoter of the PR‐protein PR‐10a gene (Després et al., 1995). While these data suggest that induction of this defense gene involves protein phosphorylation, PRF‐1 has not been isolated and it remains to be established whether phosphorylation control of PR‐10a induction is exerted at the transcription factor level.
The protein kinase inhibitor K252A blocks pal and chs induction in elicited soybean cells (Levine et al., 1994), and the present study demonstrated rapid phosphorylation of a specific transcription factor implicated in the activation of these immediate/early defense genes. This phosphorylation, which modifies the functional activity of G/HBF‐1, thus represents the terminal step in a signal pathway activated by elicitors or pathogen avirulence signals, and the rapid stimulation of the cytosolic G/HBF‐1 kinase without detectable lag suggests that the signal pathway between receptor and transcription factor is short. Transduction through a kinetically compressed pre‐existing signal pathway and the potential signal amplification inherent in phosphorylation‐mediated cascades may be crucial for rapid, massive deployment of transcription‐dependent defenses and effective expression of disease resistance following perception of pathogen attack. Experiments are in progress to clone the G/HBF‐1 kinase and determine whether this gene is the soybean ortholog of the rice Xa21 or tomato Pto resistance genes, the tomato Pti1 gene which functions immediately downstream of Pto, or possibly an ortholog of a substrate of one of these protein‐serine kinases.
Reversible phosphorylation of transcription factors is a common mechanism for selective regulation of gene expression in animals (Hunter and Karin, 1992). However, in plants, although various transcription factors have been implicated in the selective regulation of gene expression during development and in response to environmental cues, there is little information on the possible contribution of phosphorylation control exerted at the transcription factor level (Hunter and Karin, 1992). In a number of developmental processes, such as trichome formation, flower induction and determination of floral organ identity, selective gene expression appears to reflect changes in the abundance of specific transcription factors rather than post‐translational modulation of the functional activities of pre‐existing transcription factors (Meshi and Iwabuchi, 1995). The binding activity of the Arabidopsis bZIP protein GBF‐1 to a G‐box cis element implicated in light regulation of gene expression can be modulated in vitro by the nuclear protein kinase casein kinase II (Klimczak et al., 1992, 1995). However, phosphorylation of this specific factor in vivo, regulation of its phosphorylation by light, or light activation of casein kinase II have not been reported. Experiments with antisera to Arabidopsis GBF‐1 reveal a pool of cytosolic cross‐reacting proteins in parsley cells and suggest that light may stimulate phosphorylation and relocalization from cytosol to nucleus (Harter et al., 1994). While the specific G‐box‐binding factors and cytosolic protein kinases involved have not been characterized, this putative mechanism resembles that delineated here for elicitor and pathogen stimulation of the cytosolic G/HBF‐1 kinase in soybean. Phosphorylation control of the functional activities of specific transcription factors by cytosolic protein kinases may thus prove to be a common mechanism in plants for rapid, flexible regulation of selective gene expression by environmental stimuli.
Materials and methods
Cell suspension cultures of soybean (Glycine max L.) cv. Williams 82 or cv. Harasoy 63 were elicited 3 days after transfer to fresh medium, either by treatment with 0.5 mM reduced glutathione (Dron et al., 1988) or inoculation of 5×107 c.f.u./ml of P.syringae pv. glycinea (Psg) race 4 bacteria (Levine et al., 1994) with a plasmid carrying the avrA gene (Keen and Buzzell, 1991).
Nucleic acid analysis
Total cellular RNA was isolated using the Tri‐Reagent method (Molecular Research Center, Inc.). Northern blot hybridization was performed according to Levine et al. (1994), using chs1 cDNA (Ryder et al., 1984) and a PCR amplicon of nucleotides 248–637 of the G/HBF‐1 cDNA (Figure 1) as probes. Soybean genomic DNA was isolated as described (Ausubel et al., 1987). Southern blot hybridization and other standard molecular biology techniques were performed as described by Sambrook et al. (1989). Oligonucleotides were synthesized on a Cyclone Plus DNA synthesizer (Millipore).
Total cellular RNA was isolated from uninduced soybean cells (cv. Harasoy 63) and cDNA prepared using the Pharmacia cDNA synthesis kit. The cDNAs were cloned into the dephosphorylated arms of λgt11 using EcoRI linkers and packaged with commercially available extracts (Gigapack, Stratagene). The titer of the non‐amplified expression library was 105 p.f.u./ml. For a second library, cDNAs from an RNA population 50% from control soybean (cv. Williams 82) cells and 50% from equivalent cells elicited with 0.5 mM reduced glutathione and 30 mg/ml oligogalacturonide fragments for 4 h were cloned into λZAP‐II (Stratagene). This library contained 1.3×106 independent clones. For Southwestern screening of the λ phage libraries, plaque lifts were probed with an oligonucleotide corresponding to the −80 to −42 region of chs15, containing both the G‐box and H‐box III sequences (GTGTTGCACGTGATACTCACCTACCCTACTTCCTATCCA), end‐labeled by T4 polynucleotide kinase (Yu et al., 1992). G/HBF‐1 cDNA sequences were subcloned in pKSII Bluescript (Stratagene) using the NotI site and sequenced by dideoxy chain termination (Sanger et al., 1977).
A KpnI–NsiI fragment containing the G/HBF‐1‐coding sequence was inserted into pQE30 (Qiagen) restricted with KpnI and SalI to give pWD11.24, carrying a translational fusion of the His6 epitope tag at the N‐terminus of the G/HBF‐1‐coding region. pWD16.1 was constructed by inserting the G/HBF‐1 sequence, obtained as a SacI fragment from pWD11.24, into pET‐28a (Novagen). E.coli BL21 (pWD16.1) cells were grown in 0.8 mM IPTG for 4 h at 20°C to induce the expression of G/HBF‐1(His6) and the extracted fusion protein was purified under non‐denaturing conditions by Ni‐NTA affinity chromatography (Qiagen).
Peptide synthesis was performed with an Applied Biosciences peptide synthesizer. Rabbit antisera to synthetic peptides corresponding to the G/HBF‐1 amino acid sequences (C)SLNPQDSGSTAHD (P2) and (C)QDDPKHHYYQQ (P4) coupled to activated keyhole limpet hemocyanin (KLH) carrier protein, designated α‐G/HBF‐1(P2) and α‐G/HBF‐1(P4) respectively, were generated using standard immunization protocols (Harlow and Lane, 1988). An antibody to the C‐terminal domain of the OHP1 transcription factor was kindly provided by R.Schmidt (University of California, San Diego).
All manipulations were performed at 4°C. Snap‐frozen soybean cells were extracted into RIPA buffer [50 mM Tris, pH 7.4, 100 mM NaCl, 5 mM EDTA, 20 mM Na pyrophosphate, 0.1% bovine serum albumin (BSA), 10 mM NaF, 1% Triton X‐100, 0.1% dithiothreitol (DTT), 0.25% deoxycholate (DOC), 1 mM phenylmethylsulfonyl fluoride (PMSF), 1.5 mg/ml aprotinin, 0.5 mg/ml leupeptin, 1 mM Na3VO4], and the extract cleared by centrifugation. Nuclear and cytosolic protein extracts were obtained by extraction into 300 ml of NH1 buffer (50 mM HEPES, pH 7.9, 1.1 M sucrose, 25 mM NaCl, 25 mM EDTA, 1.1 mM spermine, 1.1 mM spermidine, 5 mM DTT, 2 mM PMSF, 0.5 mg/ml leupeptin, 1.5 mg/ml aprotinin, 0.2% Triton X‐100, 3.2% dextran T500, 0.6% polyvinylpolypyrolidine). After centrifugation for 3 min at 3000 g, the supernatant containing the cytosolic proteins was clarified by centrifugation. The pellet was resuspended in NH2 buffer (NH1 lacking Triton X‐100, dextran T500 and PVPP) and the nuclei washed twice by centrifugation at 3000 g and resuspension. The washed nuclei were resuspended in an equal volume of 2× NE buffer (40 mM HEPES, pH 7.9, 1 M NaCl, 3 mM MgCl2, 0.4 mM EDTA, 1 mM DTT, 0.1 mM PMSF, 40% glycerol) and stirred on ice for 45 min, clarified by centrifugation at 16 000 g for 5 min twice, and the nuclear protein solution dialyzed against 20 mM HEPES, pH 7.9, 25 mM NaCl, 1 mM EDTA, 0.1 mM PMSF, 20% glycerol.
An 8% SDS–PAGE system (Laemmli, 1970) was used to separate 50–100 mg of protein which was then blotted on PVDF membranes (Immobilon‐P, Millipore), probed with the polyclonal rabbit antibodies described above (diluted 1:1000 with phosphate‐buffered saline), and antibody binding detected with anti‐rabbit horseradish peroxidase‐linked antibody using the ECL detection kit (Amersham). Dephosphorylation of proteins in soybean cell extracts by alkaline phosphatase treatment was performed as described by Yu et al. (1992).
DNA binding assays
Gel retardation assays were performed with nuclear protein extracts or recombinant G/HBF‐1 and 32P‐labeled cognate chs15 sequences according to Yu et al. (1992). In antibody supershift experiments, DNA‐binding reactions were performed in the presence of a 1:1000 dilution of α‐G/HBF‐1(P4) or the corresponding pre‐immune serum prior to analysis of electrophoretic mobility. Southwestern blotting was performed as described by Miskimins et al. (1985) except that incubation for renaturation and DNA binding was at 4°C for 8 h.
G/HBF‐1 phosphorylation in vivo
Soybean cell suspensions (3 ml) were labeled with 1 mCi (3.7×107 Bq) of [32P]phosphate for 15 min and then elicited with 0.5 mM reduced glutathione for 30 min. Harvested cells were frozen in liquid N2, extracted into RIPA buffer and the extract clarified by centrifugation. The supernatant was diluted 5‐fold and incubated with protein A/G (Pharmacia) for 1 h at 4°C. After centrifugation, antiserum was added and the immunoreaction incubated for 1.5 h at 4°C, followed by addition of protein A/G and further incubation for 1 h. The immunocomplex was collected by centrifugation, washed sequentially by centrifugation and resuspension in RIPA buffer, twice in TSA buffer (10 mM Tris, pH 8.0, 140 mM NaCl, 0.025% NaN3, 0.1% BSA, 0.1% Triton X‐100) and finally in 0.05 mM Tris, pH 6.8. The immunocomplex was boiled for 5 min in Laemmli buffer and then analyzed by SDS–PAGE, blotting on Immobilon‐P membranes and autoradiography.
In vitro phosphorylation of recombinant G/HBF‐1
Plant nuclear or cytosolic extracts were dialyzed against protein kinase buffer (20 mM MOPS, pH 7.0, 50 mM NaCl, 5 mM MgCl2, 5 mM MnCl2, 0.1 mM Na3VO4). Plant extracts (40–60 μg protein) were incubated with ∼10 μg of G/HBF‐1 recombinant protein in the protein kinase buffer supplemented with protease inhibitors (1 mM PMSF; 0.5 mg/ml leupeptin; 1.5 mg/ml aprotinin), 0.1 mM ATP, 1 mCi (3.7×104 Bq) of [γ‐32P]ATP in a total reaction volume of 50 μl for 30 min at 25°C. Unreacted [γ‐32P]ATP was removed by passage through Sephadex G25 (Pharmacia) spin columns and 32P‐labeled G/HBF‐1 analyzed by SDS–PAGE, transfer to PVDF membranes and autoradiography. Alternatively, the protein kinase reaction was performed with 5 mM ATP in the absence of [γ‐32P]ATP and G/HBF‐1 phosphorylation analyzed by probing Western blots of the reaction products with α‐G/HBF‐1(P2), which reacts only with phosphorylated G/HBF‐1. The effect of in vitro phosphorylation on the DNA‐binding activity of recombinant G/HBF‐1 was monitored by gel retardation assays and by Southwestern blotting following SDS–PAGE fractionation of the kinase reaction products as described above.
Phosphoamino acid analysis
Phosphoamino acid analysis and phosphopeptide mapping were performed as described by van der Geer et al. (1994).
The DDBJ/EMBL/GenBank accession number for the G‐max mRNA for G/HBF‐1 described in this paper is Y10685.
We thank Cindy Doane for help in preparation of the manuscript, R.Schmidt (University of California, San Diego) for α‐OHP and Tony Hunter and Jill Meisenhelder (Salk Institute) for their help with peptide synthesis and mapping. This research was supported by grants to C.L. from the U.S. Department of Agriculture (NRI‐CGP 94‐3703‐0764) and Samuel Roberts Noble Foundation. W.D.‐L. and A.K. were fellows of the Deutsche Forschungsgemeinschaft, G.A.L. and W.P.L. were Noble Foundation/Salk Institute Postdoctoral Fellows in Plant Biology, and B.A.H. was a post‐doctoral fellow funded by the Royal Veterinary and Agricultural University, Denmark. W.D.‐L. thanks A.Pühler (Bielefeld) for allowing completion of this work in his laboratory.
- Copyright © 1997 European Molecular Biology Organization