Emerging Themes of Plant Signal Transduction

Transcription

1 The Plant Cell, Vol. 6, , November American Society of Plant Physiologists REVIEW ARTICLE Emerging Themes of Plant Signal Transduction Chris Bowler a and Nam-Hai Chua bi' a Stazione Zoologica, Villa Comunale 1, Naples, ltaly Laboratory of Plant Molecular Biology, The Rockefeller University, 1230 York Avenue, New York, New York INTRODUCTION The successful existence of all higher organisms is dependent upon their ability to coordinate complex developmental changes and to sense and respond to fluctuations in their surroundings. Responses to developmental and environmental cues occur by stimulus-response coupling: a stimulus is perceived by the cell, a signal is generated and transmitted (signal transduction), and a biochemical change is instigated (the response). This process most often requires the recognition of the stimulus by a receptor and the subsequent use of chemical second messengers and/or effector proteins to transmit a signal that will then trigger the appropriate response. Key properties of signal transduction are speed, sensitivity (achieved by amplification), and specificity, all of which are controlled by a network of positively and negatively acting elements. Whereas the positively acting intermediates are essential to drive signal transduction, the negative elements are responsible for ensuring a response that is quantitatively appropriate, correctly timed, and highly coordinated with other activities of the cell. Specifically, negative control can (1) allow changes in sensitivity to a particular stimulus, permitting a signaling pathway to work over a broad dynamic range of stimulus intensities; (2) terminate a response when it is completed, even if the stimulus may remain (sometimes referred to as desensitization or adaptation); and (3) allow a signaling pathway to be reused. In animal cells, the signal transduction events that couple avariety of stimuli with their responses have been well characterized, as have the mechanisms for their negative control. Plants, however, are challenged in different ways from animals because their developmental programs are generally more flexible and because they are unable to escape at will from unfavorable environmental situations. These differences indicate the increased adaptability of plants over animals and lead us to speculate about the nature and control of the sensing systems plants employ to regulate their development and behavior. Have these different constraints necessitated the evolution of mechanisms altogether different from those found in animals? To some extent the answer appears to be yes; both To whom correspondence should be addressed. genetic and biochemical approaches have now yielded sufficient information for us to realize that, whereas the basic mechanisms may often be conserved (e.g., the use of the same second messengers and similar protein phosphorylationl dephosphorylation reactions), there are often very real differences in the components of the signaling machinery and in their regulation. In some cases, plant signaling appears more analogous to bacterial signaling systems (Hughes, 1994), whereas in others it may be altogether novel. Studies of signal transduction pathways are facilitated by some knowledge about the specificity of the stimulus, the biochemical nature of the receptor, and the specificity of the responses. Specificity of the response for the stimulus is particularly important for genetic studies because in many cases it is desirable to generate mutants that are constitutive for a response in the absence of the stimulus, for example, the cop and det mutants, which are constitutive for lighi responses in darkness, and the cif mutants, which display constitutive ethylene responses in the absence of exogenous ethylene (see below). In these cases, to be sure that the mutant is blocked in the signal transduction pathway of interest, it is important to be able to show that all of the phenotypes of the mutant are consistent with all or at least a subset of the normally observed phenotype of wild-type plants exposed to the stimulus. This is particularly difficult to demonstrate in plants because many responses can be mediated by different stimuli. For example, the effects of light on plant morphology (photomorphogenesis) and gene expression can be recapitulated to a large extent by cytokinin, and the response to pests and pathogens may involve the actions of (at least) ethylene, jasmonic acid, salicylic acid, systemin, and electrical signals. Furthermore, genetic screens for constitutive mutants are largely dependent on the assumption that there are negative regulators that normally repress signal transduction in the absence of the stimulus and that the stimulus acts (at least at this step) by overcoming the repression. For some kinds of signal transduction pathways, for example, in which the negative regulator is one step in a linear pathway, loss of function mutations in such a negative regulator would produce a constitutive phenotype in the absence of any stimulus. More frequently, however, individual steps in signal transduction (at

2 1530 The Plant Cell least those upstream of transcription factors) are controlled by counteracting positive and negative regulators. In such cases, loss of function mutations in a negative regulator would not result in a constitutive phenotype unless the signal transduction pathway had a basal level flux passing through it in the absence of the stimulus. This flux could be a result of a basal level activity of the signaling intermediates. For example, although a kinase may be highly stimulated in the presence of inducer, it may also have a low constitutive activity in its absence, as may the complementary phosphatase that counteracts the kinase activity. In the case of second messenger pathways, the flux is known as a current and relates to a nearly equal hydrolysis and synthesis of the second messenger (e.g., cgmp) in the absence of inducer. Stimulation of the pathway then disturbs this cycle to result in a net production of the active signal transducing molecule (see Koch, 1992). Whereas genetic approaches can yield information about the role of particular molecules during plant growth, biochemical methods can allow the activities of signal transducing intermediates to be identified. However, as with genetic approaches, biochemical studies of plant signal transduction can be hampered by insufficient knowledge of stimulus-response specificity. Consequently, although several signal transducing molecules (both second messengers and proteins) have been identified biochemically, their roles in specific responses are often obscure. Notwithstanding the limitations of genetic and biochemical approaches, intensive research is now yielding information about several signal transduction pathways in plants, and some consistem features have emerged. In this review, we attempt to highlight these themes by discussing recently published work on signaling mediated by phytochrome, ethylene, abscisic acid (ABA), and auxin. Particular emphasis is given to the negative control of these pathways, because it is one recurring theme that has emerged. We have illustrated a variety of ideas by presenting models of signal transduction that are consistent with the currently available experimental data. These schemes are intended to be thought provoking and may not necessarily be correct. This review is by no means exhaustive, and the reader is referred to the original articles as well as to other recent reviews on various other aspects of plant signaling involved in these and other responses (Trewavas and Gilroy, 1991; Estelle, 1992; Bush, 1993; Deng, 1994; Jones, 1994; Lamb, 1994). PHYTOCHROME Phytochrome is the only plant protein that has been unequivocally identified as a receptor (for review, see Quail, 1991; Furuya, 1993). Phytochrome apoproteins are usually encoded by small multigene families-for example, five phytochrome genes (PHYA-PHYE) have been identified in Arabidopsis (Sharrock and Qyail, 1989; Clack et al., 1994)-and each phytochrome is thought to have a different physiological role (Whitelam and Harberd, 1994). Holophytochrome is a dimer containing a linear, covalently bound chromophore that mediates light perception. The photoreceptor is synthesized in the redabsorbing Pr form, and absorption of red light alters its conformation to the far-red-absorbing form, Pfr, which is generally considered to be the physiologically active form. Pfr responses can be reversed by far-red light, which converts Pfr back to Pr. Of all the photoreceptors so far characterized, this property, that is, to behave as a molecular light switch, is unique to phytochrome. Its molecular structure is also unique; although we know that it is a receptor, it does not look like any other receptor so far characterized. Most conspicuously, phytochrome is cytoplasmic, and as yet there is no convincing evidence that it ever uses membranes as a platform to mount its biochemical reactions (Quail, 1991). How the phytochrome molecule transduces the light stimulus is still unknown. Structure-function studies have not answered many of the crucial questions about phytochrome activity, largely because the studies have been confined to overexpression of mutant phytochrome gene sequences in wild-type plants, which also contain the endogenous phytochromes. Although much of the N-terminal region, which contains the chromophore binding domain, is involved in light detection, it appears that the C-terminal region is more likely to be involved in mediating biological activity, that is, signal transduction (Cherry et al., 1993; Boylan et al., 1994). For example, removal of the C-terminal 35 residues from oat PHYA results in a protein that is physicochemically normal in vitro but has no activity in vivo (Cherry et al., 1993). Nonetheless, the first N-terminal 69 residues are probably also important in this regard (Cherry et al., 1992; Stockhaus et al., 1992). In addition to genetic approaches (see below), phytochrome signal transduction has been studied by a variety of biochemical approaches: (1) using pharmacological agonists and antagonists to interfere with phytochrome-mediated events, (2) studying phosphorylation reactions or examining the effects of antibodies that bind to signal-transducing components of animal cells, and most recently, (3) using a microinjection-based approach to deliver putative signaling intermediates directly into the cells of a phytochrome-deficient tomato mutant. These approaches (reviewed in Millar et al., 1994) have identified heterotrimeric Ggroteins as the most upstream component of the phytochrome signaling machinery, although this has not been definitively proven (because the G-protein(s) has not yet been purified and characterized). If true, this would indicate that phytochrome either localizes transiently to membranes or acts through an intermediary transducer to activate this membranebound complex. Alternatively, it is possible that the component with which phytochrome interacts does not have all the characteristics of animal heterotrimeric G-proteins, which at any rate are not integral membrane proteins, being localized to membranes via lipid tags (Casey, 1994). Downstream of G-protein activation, phytochrome utilizes Ca2+ and cgmp (Neuhaus et al., 1993; Bowler et al., 1994a). A role for Ca2+ in phytochrome signaling was first postulated

3 Plant Signal Transduction years ago, and circumstantial evidence has accumulated since then to support this hypothesis (reviewed in Tretyn et al., 1991; Millar et al., 1994). However, it is only recently that a red/far-red-reversible Ca 2+ transient was directly observed in plant cells (Shacklock et al., 1992). Subsequently, microinjection of Ca 2+ and Ca 2+ -activated calmodulin (Ca 2+ /CaM) into phytochrome-deficient tomato cells provided direct evidence that these molecules are involved in PHYA signal transduction, specifically in stimulating chloroplast development (Neuhaus et al., 1993). Using the same microinjection methodology, cgmp was found to mediate PHYA-dependent anthocyanin biosynthesis and also to participate, together with Ca 2+ and Ca 2+ /CaM, in coordinating chloroplast development (Bowler et al., 1994a). The use by the plant of Ca 2+ and cgmp to mediate phototransduction has an interesting evolutionary parallel with the photoperception mechanisms found in retinal cells, even though the photoreceptors, phytochrome in the plant and rhodopsin in the retina, are structurally completely unrelated (Koutalos and Yau, 1993). Figure 1A summarizes our current biochemical knowledge of PHYA signal transduction pathways, based on the results of microinjection experiments in tomato and studies of endogenous gene expression in soybean cells (Neuhaus et al., 1993; Bowler et al., 1994a, 1994b). Most interestingly, the targets of the three different positively acting pathways are divided very clearly along functional lines. The cgmp pathway appears, thus far, to be dedicated to the production of the photoprotective anthocyanin pigments, whereas the Ca 2+ - and Ca 2+ / cgmp-dependent pathways control chloroplast development by stimulating genes encoding subunits of the photosynthetic complexes. Obviously, with these divisions along functional lines, it is very important that the final responses of the different pathways be carefully coordinated to allow, for example, for photoprotectant production before photosynthesis (known as juvenile anthocyanin; Drumm-Herrel, 1984) and to ensure proper stoichiometry of the different photosynthetic complexes. We have begun to understand how coordination is achieved by specifically manipulating signal flow through the different pathways, either by microinjecting large or small amounts of signal intermediates or by using pharmacological agonists and antagonists (Bowler et al., 1994b). Coordination appears to be controlled by negative regulatory mechanisms acting at the level of individual pathways (e.g., desensitization of the cgmp pathway to terminate light-induced expression of the genes encoding anthocyanin biosynthetic enzymes, such as chalcone synthase [CHS], in continuous light) and by cross-talk mechanisms between the pathways (Bowler et al., 1994b). The negative interactions between the pathways (summarized in Figures 1B and 1C), which we have collectively termed reciprocal control, are presumably important at different times. We believe that the negative regulation of the Ca 2+ and Ca 2+ /cgmp pathways by high levels of cgmp (Figure 1C) may act during the early stages of light exposure to suppress synthesis of photosynthetic complexes at a time when there are not sufficient photoprotectants, whereas negative regulation Pfr' Pfr' Pfr' r iu] iu ^ Negative IRegulation * chs» cab chs ^ cab Figure 1. Summary of Current Biochemical Models of PHYA Signal Transduction. (A) Design of the three basic signaling pathways dependent on cgmp and/or Ca 2+. The products of these pathways (anthocyanin pigment formation within the vacuole and chloroplast development within the cytoplasm) are shown in highly schematized cells. Examples of gene targets for each pathway (chs, fnr, and cab) are also indicated. Inhibitors specifically blocking the different pathways (Gen. and Trif.) are shown and their block points indicated by two parallel lines drawn across the arrows within the main pathways. (B) Negative regulation of the cgmp pathway by the Ca 2+ pathway. (C) Negative regulation of the Ca 2+ - and Ca 2+ /cgmp-dependent pathways by the cgmp pathway. In (B) and (C), the blue lines indicate the negative interactions, and brackets indicate the positions of the components that current information suggests are responsible for these negative interactions. In (B), negative regulation does not affect the Ca 2+ /cgmp-dependent pathway, whereas in (C) it does. Go, a subunit of heterotrimeric G-protein; CaM, Ca 2+ -activated calmodulin; Gen., Genistein; Trif., Trifluoperazine; chs, chalcone synthase; fnr, ferredoxin NADP+ oxidoreductase; cab, chlorophyll alb binding protein.

4 1532 The Plant Cell of the cgmp pathway by the Ca2+ pathway (Figure 16) suppresses the production of anthocyanins when they are no longer required. These hypotheses are generally supported by the timing of expression of genes encoding anthocyanin biosynthetic enzymes and chloroplast components in soybean cells (Bowler et al., 1994b) and in seedlings exposed to red or white light (Drumm-Herrel, 1984; Brodenfeldt and Mohr, 1988; Wenng et al., 1990; Ehmann et al., 1991). Furthermore, because we have found that the Ca*+/cGMP-dependent pathway has a lower concentration requirement (probably by nearly an order of magnitude) for cgmp than does the cgmpdependent pathway (Bowler et al., 1994b), it would appear that full chloroplast development and synthesis of all the photosynthetic components can proceed in the absence of anthocyanin pigment biosynthesis. Such a scenario clearly highlights the importance of negative regulation for coordinating phytochrome signal transduction. Although we can only speculate about the precise mechanisms involved, the experimental results so far obtained (Bowler et al., 1994b) indicate that the most obviously apparent forms of negative regulation involve interactions between downstream components of the signaling pathways (Figures 1B and 1C). In addition to these interactions, the sheer number of examples from animal cells would tend to suggest that phytochrome signaling will also be attenuated at the leve1 of the photoreceptor itself. One possibility is that phytochrome is inactivated by phosphorylation, like rhodopsin (Kawamura, 1993) and many other animal receptors. Biochemical data have indicated that phytochrome is phosphorylated in vivo (Hunt and Pratt, 1980). Furthermore, mutation of PHYA N-terminal serine residues to alanine residues results in a molecule that has increased biological activity (Stockhaus et al., 1992), suggesting that one desensitization mechanism for phytochrome signaling may normally be the phosphorylation of these residues. Wong et al. (1986) have purified a polycation-dependent kinase that may be responsible for phosphorylating phytochrome in vivo, and its targets indeed appear to be N-terminal serine residues, at least in vitro (McMichael and Lagarias, 1990). A different phytochrome-associated kinase has also been purified (Grimm et al., 1989). Interestingly, unlike the polycation-stimulated kinase, this kinase has been found to be stimulated by Ca2+ and cgmp (Grimm et al., 1989). This observation suggests that the two signaling molecules that PHYA uses for its responses may also regulate photoreceptor desensitization, providing a mechanism for feedback regulation to control the activity of the photoreceptor. It may also be significant that the Pfr form of PHYA is known to be very unstable (perhaps because of ubiquitination; Jabben et al., 1989) and is rapidly sequestered into large aggregates in the cytoplasm (Quail, 1991). Controlled removal of physiologically active PHYA from its signaling machinery by sequestration andlor degradation would clearly constitute a form of desensitization, and such mechanisms have now been found in severa1 other signaling systems (see below). Although the other phytochromes are not sequestered, sequestration may well prove to be one mechanism for attenuating PHYA signal transduction. The importance of negative regulation for light responses is underlined by the types of mutant that have been isolated in screens for mutants with defective photomorphogenesis. In Arabidopsis, two types of mutant have been studied (Chory, 1993): (1) those that have a partially etiolated morphology when grown in the light (e.g., a long hypocotyl), and (2) those that show characteristics of light-grown plants when grown in the dark (e.g., short hypocotyls, open cotyledons, and expression of light-regulated genes). In simple terms, the former class can be thought of as light-insensitive mutants, whereas the latter can be thought of as mutants constitutive for light responses. The light-insensitive mutants are recessive, indicating that the mutations most likely affect positively acting components of the light signaling machinery. For example, one would expect that some are deficient in photoperception and may contain mutated photoreceptors. This has been borne out by the characterization of the genetic lesions of mutants within this class, denoted hy for long hypocotyl. hyl, hy2, and hy6 mutants are believed to have defects in phytochrome chromophore biosynthesis or attachment (Parks and Quail, 1991; Nagatani et al., 1993); hy3 mutants (now denoted phyb; Quail et al., 1994) have mutations in their PHYB genes (Reed et al., 1993); hy8 mutants (now phya; Quail et al., 1994) have defective PHYA genes (Dehesh et al., 1993; Nagatani et al., 1993; Whitelam et al., 1993); and hy4 mutants are probably blue-light photoreceptor mutants (Ahmad and Cashmore, 1993). One would also expect some hy mutants to be defective in positively acting light signal transduction intermediates; however, with the possible exceptions of hy5 (Chory, 1992), and fhyl and fhy3 (Whitelam et al., 1993), such mutants have not been found. The paucityof mutants that are likely to bedefective in positive elements downstream of phytochrome suggests that such mutants are extremely difficult to find in these screens. If signa1 transduction from each of the phytochromes were to converge in the same pathway, such mutations might even be lethal, because too many processes would be affected. Observations that PHYA and PHYB do have some overlapping functions (Reed et al., 1994) would argue that at least some signaling pathways are indeed shared. However, the fhyl and fhy3 mutants do appear to affect PHYA responses specifically (Whitelam et al., 1993; Johnson et al., 1994). Whether the cgmp and Ca2+ pathways that have been defined biochemically (Figure 1) are used specifically by PHYA or are shared by different phytochromes is an interesting question. Although these pathways have been linked to PHYA (Neuhaus et al., 1993; Bowler et al., 1994a), it is clearly possible that they may also be used by the other phytochromes or even by the blueluva photoreceptors. The constitutive mutants, called der (for deetiolated) and cop (for constitutive photomorphogenic), were isolated in

5 Plant Signal Transduction 1533 screens for seedlings showing light-grown morphologies (e.g., short hypocotyls and open cotyledons) in darkness (discussed in Chory, 1993; Deng, 1994). The recessive nature of these mutants suggests that the mutations lie within negative regulatory molecules. Epistasis tests using the hy mutants have indicated that the gene products function downstream of both the phytochromes and the blueluva photoreceptors (Chory, 1993; Deng, 1994; but see Millar et al., 1994). It is difficult, however, to conclude that they are light perception and phototransduction mutants because their phenotypes are not restricted to those we would expect if photoreceptor signaling was specifically deregulated (discussed in Millar et al., 1994). The problem lies in the complexity of the photomorphogenic developmental program, because photomorphogenic traits are influenced not only by light but also by (at least) cytokinin and ABA (Millar et al., 1994). For example, it has been shown that der7 can be phenocopied in wild-type seedlings by addition of cytokinin (Chory et al., 1991, 1994). Furthermore, the fact that six of the 11 constitutive mutants characterized to date (deu, cop7, cop8, cop9, cop70, and cop77) are allelic to previously isolated seedling lethal mutants known as fusca (fus) mutants (Castle and Meinke, 1994; McNellis et al., 1994; Misbra et al., 1994; Wei et al., 1994a) also indicates that the mutant phenotypes are more pleiotropic than was originally suspected. With the realization that morphological screens do not allow the specific isolation of mutants with altered phytochrome responses (except for the elegant utilization of responses to constant red or far-red light irradiation to discriminate between different phytochromes; Parks and Quail, 1993), other screens have been developed in an attempt to improve specificity. These screens have been based on the aberrant expression of a light-regulated hygromycin resistance gene (Susek et al., 1993; Li et al., 1994). Although the mutants isolated have not yet been characterized to determine whether they are defective specifically in the response to phytochrome, the type of screens used (i.e., designed to isolate mutants constitutive for a response in the absence of the normal stirnulus) and the fact that all but one of the mutations are recessive indicate that these mutants are blocked in some aspect of negative regulation. Although genetic screens for constitutive mutants select predominantly for mutants with defects in negative regulators, the number of mutants now isolated indicates that negative regulation may be a common mechanism for controlling plant photomorphogenesis and is perhaps as important as regulation by positive elements. Even though the cop and der mutants do not appear to have defects specific to phytochrome signaling, their phenotypes imply that the mutations lie within negatively acting global regulators for plant development that are able to receive signals from multiple stimuli and integrate them into a morphological (i.e., developmental) response. Because many of the cop and der mutants are allelic to seedling lethal mutants, it has been suggested that this class of molecule is critically important for controlling certain aspects of plant development (Castle and Meinke, 1994). It is clearly of great importance, therefore, to understand how such molecules work. We now have knowledge of the structure of four of these negative regulators: FUSG (COPll), COP9 (FUS7), DETl (FUS2), and COPl (FUS1). FUSG (Castle and Meinke, 1994) and COP9 (Wei et al., 1994b) have entirely novel sequences but are probably cytoplasmically localized. COP9 exists as a large complex whose formation or stability requires COP8 and COP11, because it is absent in cop8 and cop77 mutants (Wei et al., 1994b). This is some of the first evidence for molecular interactions among different COPlDETlFUS gene products and indicates that at least COP8, COP9, and COP11 may act together to control a particular biochemical process. The sequence of DETl is also novel, although the protein is nuclear (Pepper et al., 1994). Unlike these novel proteins, COPl contains a combination of recognizable structural motifs: a zine finger domain at the N terminus (similar to a newly characterized zinc binding domain called a RlNG finger; Boddy et al., 1994), a putative coiled-coil region, and a domain at the C terminus with multiple WD-40 repeats homologous to the p subunits of heterotrimeric G-proteins (Deng et al., 1992; McNellis et al., 1994). In spite of an extensive analysis of the COP7 gene in an allelic series of cop7 mutants, no domain has clearly emerged as being more important than the others (McNellis et al., 1994). However, the homology of COPl (particularly the Gghomologous domain) with the Drosophila TAF1180 protein, a subunit of the TFllD complex that is required for RNA polymerase II transcription initiation, may suggest a role for COPl in repressing general transcription (McNellis et al., 1994), which would be consistent with its expected function as a negatively acting global regulator of plant responses. If so, COPl would be expected to be a nuclear protein. The nuclear localization of DETl may hint that its function is also to repress transcription. Although DETl does not appear to bind DNA directly (Pepper et al., 1994), its repressor activity may be manifested ata much more general level, for example, by remodeling the nucleosomes to organize repressive regions of chromatin (Cooper et al., 1994; lmbalzano et al., 1994; Kwon et al., 1994). COPl may have a similar function, because Gp-homologous domains are present within at least one such global repressor, yeast TUPl (Deng et al., 1992; Cooper et al., 1994). Biochemical experiments are clearly required to examine this possibility. The current lack of overlap between the biochemical (Figure 1) and the genetic (Chory, 1992; Ang and Deng, 1994) models of phytochrome signaling deserves some comment and is a major challenge to address in the future. Severa1 points are relevant. First, the biochemical pathways have been defined mainly by their positive elements, whereas all the gene products inferred from mutant studies are likely to be repressors (with the possible exceptions of hy5, fhy7, and fhy3). This is simply a result of the different experimental approaches: the microinjection experiments were designed to identify positively

6 1534 The Plant Cell acting signal intermediates, whereas the genetic screens were designed to yield predominantly negative regulators. Recent biochemical studies of the negative regulatory mechanisms involved in phytochrome signaling (Bowler et al., 1994b) can, however, allow us to predict the phenotypes of mutants altered in specific negative regulatory aspects of phytochrome signaling, for example, in CHSdesensitization or in reciproca1 control. Recessive mutants of this kind might not be expected to be constitutive for a light response in darkness (as are cop and detmutants) because at this leve1 in the signal transduction pathway, there does not appear to be a dark current (i.e., turnover; see Introduction) of Ca2+ or cgmp (Bowler et al., ). With respect to their responses in the light (a condition under which we do believe there is a turnover of second messengers; Bowler et al., 1994b), phenotypes would be more apparent. For example, mutant seedlings unable to repress the Ca2+ and Ca2+/cGMP-dependent pathways (Figure 1C) might be expected to display a light fluence-dependent bleaching due to chloroplast development before sufficient photoprotectants have been synthesized. Because the screens used so far to isolate photomorphogenic mutants have been based largely on aberrant phenotypes in darkness rather than aberrant phenotypes in dynamic responses to light signals, such mutants have not yet been studied. Currently, the simplest way to reconcile the apparent absence of a dark current with the constitutive phenotypes of cop and det mutants is if COP and DET act downstream of the second messengers. For example, they could be general transcriptional repressors normally inactivated by Ca2+ and cgmp, among other things. In their absence (i.e., in mutants), light-responsive and other genes would no longer be repressed in the dark and constitutive transcription would ensue, even though Ca2+ and cgmp levels would remain unaltered. Current information about the structure of COP1 and DET1 does indeed suggest that they are somehow involved in regulating transcription (see above). The possibility that COP and DET act downstream of Ca2+ and cgmp can be tested by examining whether or not the pharmacological inhibitors of the phytochrome pathways (Bowler et al., 1994b) are able to reverse the constitutive phenotypes of the mutants in the dark. A second important difference between the biochemical and genetic data is that the biochemical pathways have defined very specific cellular reactions (leading to chloroplast development and anthocyanin biosynthesis) and have not even begun to address the developmental inputs that may affect the stimulation of these responses. Anthocyanin biosynthesis, for example, is known to be under exquisite photobiological control during development (Drumm-Herrel, 1984; Frohnmeyer et al., 1992; Nick et al., 1993). The majority of mutants have, by contrast, been isolated based on their developmental aberrations. Most likely, therefore, the gene products affected in these mutants are involved in coordinating global developmental switches to a wide range of different signals, including light, rather than being specific to some aspect of the phytochrome cellular signaling system that has so far been defined biochemically. ETHY LENE Gaseous ethylene is an important growth regulator, influencing a variety of plant responses that include germination, leaf abscission, and fruit ripening, as well as responses to environmental stresses such as wounding, chilling, and pathogen attack. In contrast to phytochrome, biochemical approaches have contributed little to our current knowledge of ethylene signal transduction, and recent interest has arisen largely as a result of genetic approaches that have identified gene products involved in regulating ethylene responses. These approaches have exploited a morphological response to ethylene observed in etiolated seedlings known as the triple response: which comprises three distinct changes in seedling morphology: inhibition of hypocotyl elongation, hypocotyl thickening, and an exaggerated tightening of the apical hook. As previously described for the photomorphogenesis mutants, both insensitive and %onstitutive mutants have been screened for and isolated from Arabidopsis (Guzmán and Ecker, 1990). The success of these morphological screens is a result of the fact that the simultaneous appearance of these three responses is highly specific for ethylene; hence, the mutants isolated are very likely to have defects in the same signal transduction pathway. This greatly facilitates the interpretation of results from epistasis tests and contrasts with the mutants so far isolated in screens for photomorphogenic phenotypes, which may have defects in any one of severa1 signaling pathways. For the ethylene-insensitive mutants, three complementation groups have been reported, designated etrl (Bleecker et al., 1988) and ein2 and ein3 (Guzmán and Ecker, 1990; Kieber et al., 1993). etrl mutations are dominant and are allelic to the dominant einl mutants isolated by Guzmdn and Ecker (1990; Chang et al., 1993). ein2 and ein3, on the other hand, are recessive mutations. From screens for constitutive triple response mutants, two mutant classes have been identified that can be distinguished based on whether or not they overproduce ethylene (Kieber et al., 1993). Ethylene overproducers are not likely to be mutated in ethylene perception or signal transduction, whereas those that have normal levels of ethylene are. Of this latter class, only one complementation group has currently been reported, designated ctr7. The recessive nature of the constitutive phenotype indicates that CTRl is a negative regulator of ethylene responses. Epistatic relationships among these mutants have been examined, and the results suggest the order of action as being ETRl, CTRl, and EIN3 (Kieber et al., 1993). ETR7 has now been cloned, and the strong similarity of the C terminus to domains of proteins involved in the twocomponent signal-transducing mechanisms of bacteria allows some speculation about its possible mode of action (Chang et al., 1993). The two-component pathway of prokaryotes, as its name suggests, usually contains two proteins (Parkinson, 1993). One is called the sensor and contains both an input domain (which senses the stimulus) and a transmitter. The

7 Plant Signal Transduction 1535 other, called the "response regulator," consists of both a "receiver" module (which receives the signal from the transmitter, via phosphorylation) and an "output" domain, which triggers the response. ETR1 contains a "transmitter-homologous domain and a "received-homologous domain downstream of an N-terminal region of 313 amino acids whose sequence is novel. It has been postulated that this N-terminal domain may be the ethylene-sensing region, whereas the C terminus may be involved in signal transmission (perhaps like phytochrome). ETR1 may therefore be an ethylene receptor, as is also supported by the fact that etrl mutants have a lower capacity to bind ethylene compared with wild-type plants (Bleecker et al., 1988). The N-terminal region of ETR1 also contains three potential membrane-spanning domains, indicating that the protein is membrane localized, even though ethylene is able to diffuse through membranes. All the etrl mutations so far characterized lie within these putative membrane-spanning domains (Chang et al., 1993). Based on the homology of the C terminus of ETR1 with the prokaryotic proteins, it is possible that perception is signaled via phosphorylation reactions. The prokaryotic transmitters are phosphorylated on a histidine residue, and the conservation of this residue in ETR1 suggests that histidine may also be phosphorylated in ETRL Histidine phosphorylation was once thought to be restricted to prokaryotes, but with the isolation of not only Arabidopsis ETR1, but also yeast SLN1, an osmolarity sensor that also shows homology with the "transmitter" and "receiver" domains of bacterial two-component systems (Ota and Varshavsky, 1993), the interesting possibility now arises that phosphorylation of histidine residues may prove to be an important regulatory mechanism in eukaryotic cells. CTR1 has also been cloned recently (Kieber et al., 1993). It contains no obvious membrane-spanning domains, but its C terminus has considerable homology with protein kinases, particularly with members of the Raf family of serine/threonine kinases, which activate MAP kinase cascade pathways (Blumer and Johnson, 1994). All five mutations sequenced are in the putative kinase catalytic domain. The utilization by the plant of ETR1 and CTR1 in ethylene signal transduction therefore appears to be an interesting combination of a prokaryotic signaling system with a typical eukaryotic one. A similar system has also been found in yeast for controlling adaptation to osmotic stress (Maeda et al., 1994). How does ethylene signal transduction work? One possible model, based on current information, is presented in Figure 2. It is reasonable to speculate that ETR1 is an ethylene receptor and that it controls ethylene responses via a phosphorylation cascade that involves CTR1 and EIN3. Because CTR1 appears to be a kinase, it is unlikely that it acts by itself, and the phosphorylation status of its substrate probably is also controlled by a "complementary" protein phosphatase. However, the nature of the phenotype conferred by Ctrl and the fact that the mutations are recessive suggest that CTR1 is a negative regulator of ethylene responses. Furthermore, the fact that the phenotype is manifested in the absence of exogenous ethylene would therefore indicate that there is always a flux through Response Figure 2. Possible Model for Ethylene Signal Transduction for Controlling the Triple Response. ETR1 is shown as a dimeric plasma membrane-localized ethylene receptor whose phosphorylation status, indicated by 0, is altered by ethylene binding. Possible interactions with the CTR1-regulated step are indicated by gray arrows. CTR1 is shown as being activated by phosphorylation and as repressing the activity of its target by phosphorylating it. A phosphatase is proposed to work together with CTR1 to control the phosphorylation status of the CTR1 target. EIN3 is shown as subsequently being activated by phosphorylation mediated by the target of CTRL It is also possible that EIN3 is the direct downstream target of CTRL A thick arrow indicates the unknown remainder of the pathway leading to the triple response. the pathway (perhaps as a result of low constitutive activities of the phosphatases and kinases; see Introduction) and that one role of CTR1 is to attenuate this signal flow in the absence of ethylene by phosphorylation of its target substrate(s). Mutation of CTR1 would then prevent phosphorylation of its target, allowing the basal level signal to pass this regulatory step unhindered, thus leading to the triple response in the absence of ethylene. In this model, dephosphorylation would be the forward reaction, at least for the CTR1-controlled step. The triple response elicited by exogenous ethylene probably occurs as a result of increased flux through the pathway initiated by a phosphorylation cascade from ethylene-stimulated ETRL If ETR1 were to interact directly with the CTR1-regulated step in the presence of ethylene, it could overcome CTR1 repression by inactivating CTR1, by activating the phosphatase partner of CTRL or by both (Figure 2). In either case, the ethylene-insensitive phenotype of etrl mutants can be most simply explained if the mutations were to result in an inactive protein that can no longer carry out its subsequent interaction(s). Dominance could then be explained if ETR1

8 1536 The Plant Cell functions as a multisubunit complex, because the presence of mutant subunits would disrupt its activity (Chang et al., 1993). Some of the two-component regulators of bacteria are known to function as dimers (Parkinson, 1993). If ETRl were to interact with the CTR1-controlled step in the absence of ethylene (e.g., to activate CTRl), other, more complicated schemes could be proposed to account for the phenotypes of the mutants (discussed in Chang et al., 1993). However, with the current absence of any biochemical information about an interaction between ETRl and CTRl, all models remain highly speculative. First and foremost, kinase activity must be demonstrated for each protein. Subsequently, important questions that can be answered include whether ETRl works as a multimer, how the activities of ETRl and CTR1 change in response to ethylene, and how their activities are altered in etrl and ctrl mutants. If there is a basal leve1 flux through the pathway in the absence of ethylene, ectopic expression of CTRl would be predicted to result in ethylene insensitivity, because the equilibrium of the phosphorylation reaction would have shifted toward the production of the inactive form of the target of CTRl. However, ectopic expression of a modified (i.e., deregulated) CTRl molecule should be even more effective in causing ethylene insensitivity. Overexpression of the gene that encodes the protein phosphatase partner of CTRl (Figure 2) would cause a constitutive triple response unless its activity were also regulated by ethylene, in which case only a deregulated form might show a constitutive triple response. In analogy with these predictions, pharmacological inhibitors of CTRl activity might cause a triple response, whereas an inhibitor of the phosphatase might prevent it. Some experiments detailing the effects of protein kinase and phosphatase inhibitors on ethylene responses have been reported, although these have been studied with respect to their effects on the ethylene-mediated induction of pathogenesis-related genes (Raz and Fluhr, 1993). The results indicate that protein kinase inhibitors can block the response, whereas protein phosphatase inhibitors can stimulate pathogenesis-related gene induction in the absence of ethylene, the opposite of what we would predict if CTRl and its phosphatase partner were, respectively, the targets of these inhibitors. In similar experiments, Ca2+ has also been implicated as a second messenger in ethylene-mediated pathogenesis responses (Raz and Fluhr, 1992), although its potential to regulate the phosphorylation cascade involved in the triple response remains to be critically evaluated. ABSClSlC AClD Among other things, ABA regulates embryo maturation and seed dormancy and inhibits mitotic activity in root meristems. During stress conditions such as water shortage, it also induces stomatal closure (Zeevaart and Creelman, 1988). Mutants defective in these responses have been isolated in screens for seedlings that grow on normally inhibitory concentrations of ABA. Much like the hy mutants, these mutants can therefore be thought of as insensitives, and they have been named abi mutants, for abscisic acid insensitive (Koornneef et al., 1984). These mutants are severely impaired in a wide spectrum of ABA responses, and they show phenotypes including excessive water loss (caused by an inability to regulate stomatal function) and reduced seed dormancy. These phenotypes would therefore suggest that the mutations affect ABA responses specifically. One such mutant, abil, which is dominant, has recently been cloned (Leung et al., 1994; Meyer et al., 1994). The gene product, ABI1, has a nove1 structure containing a domain highly homologous to the serineahreonine protein phosphatase type 2C, together with an N-terminal Ca2+ binding site. Functionally, therefore, it may be the plant equivalent of calcineurin, the Ca2+-dependent phosphatase present in animal and yeast cells, even though it is structurally very different: calcineurin is a heterodimeric protein composed of a Ca2+ binding regulatory domain and a catalytic subunit that has no homology with protein phosphatase type 2C (Cohen, 1989). There is very good evidence for the involvement of Ca2+ in ABA responses, at least in controlling stomatal closure (Gilroy et al., 1991). Logically, therefore, the structure of ABI1 would indicate that it is a downstream target of ABA-stimulated Ca2+. This can be tested by examining ABA-mediated Ca2+ increases in wild-type and mutant cells and by examining whether responses to artificially increased Ca2+ (e.g., stomatal closure induced by release of caged Ca2+; Gilroy et al., 1990) are affected in the mutant. Based on current information, models of ABA signaling should incorporate a plasma membrane-localized receptor (Anderson et al., 1994; Gilroy and Jones, 1994), Ca2+, and ABI1 and should account for the dominant nature of the abil mutation (Koornneef et al., 1984). In simple terms, two functional possibilities exist, dependent upon whether phosphorylation or dephosphorylation is the forward reaction for the ABI1- regulated step in ABA responses (Figure 3). If dephosphorylation is the forward reaction, then ABA-mediated increases in cytosolic Ca2+ may activate ABI1, which would then dephosphorylate its target (Figure 3A). If phosphorylation is the forward reaction, ABll would be predicted to function as a negative regulator, constitutively suppressing forward flow through the signal transduction pathway in the absence of ABA (Figure 36). In the first scenario (Figure 3A), dominance could be explained most simply if ABll were part of a multisubunit complex, because the mutant subunits would disrupt the activity of the complex. In the latter scheme (Figure 3B), the dominance of the abil mutation could be explained if the mutation were a result of a loss of sensitivity to Ca2+, resulting in phosphatase activity that was no longer repressible by Ca2+ and hence causing a permanent switching off of the pathway. Once phosphatase activity of ABll has been demonstrated,

9 Plant Signal Transduction 1537 B ABA ^H - -I- Response Response Figure 3. Two Possible Models for the Role of ABU in ABA Signaling. The ABA receptor is shown to be localized to the plasma membrane and to receive the ABA signal externally (see Anderson et al., 1994; Gilroy and Jones, 1994). In both models, an increase in cytosolic Ca 2+ is shown to precede ABU action, and the putative protein kinase partner of ABU is shown as being regulated by Ca 2+ in the opposite manner to ABU. (A) A model in which dephosphorylation, mediated by Ca 2+ -activated ABU, leads to activation of the target of ABU and consequently to ABA responses. (B) An alternative model in which phosphorylation of the ABU target leads to its activation and to subsequent ABA responses. In this case, Ca 2+ would be predicted to repress ABU activity. biochemical experiments should be able to resolve these different possibilities. AUXIN Auxin has been implicated in regulating almost every aspect of plant development. Although we now have some understanding of its biosynthesis, biochemical responses, and physiology, molecular studies of its action have proven very difficult (Estelle, 1992). As may also be the case with other hormones, one likely way that auxin responses are regulated is by controlling the amount of free hormone that is available for signal transduction (Szerszen et al., 1994). The conjugation of free hormone to form inactive pools can potentially attenuate signal transduction, whereas its release from these conjugates could activate it. In addition to this type of control, two recent articles have now provided other clues about auxin signaling mechanisms (Leyser et al., 1993; Abel et al., 1994). The first article details the cloning of the auxin resistance gene AXR1 from Arabidopsis (Leyser et al., 1993). To date, 20 axrl mutants have been isolated in screens for mutants able to grow on normally inhibitory concentrations of auxin, and all are recessive (Estelle and Somerville, 1987; Lincoln et al., 1990). Other phenotypes of these mutants include decreases in height, in apical dominance, in root gravitropism, in hypocotyl length, and in fertility. Cloning and sequencing of the AXR1 gene revealed significant similarity to the ubiquitin-activating enzyme E1 (Leyser et at., 1993), although there are sufficient important differences between the sequences to make it clear thataxri is not an E1 homolog. Nonetheless, the similarities do suggest that the ubiquitination pathway for protein degradation may somehow play an important role in auxin responses. This possibility has also been inferred from an independent series of experiments. Abel et al. (1994) have found that a family of conserved genes induced rapidly and strongly by auxin in excised stem sections encodes nuclear proteins that have very short half-lives. Each of these proteins contains a motif similar to that found in the DNA binding domain of prokaryotic represser proteins such as Arc, suggesting that they are DNA binding proteins that may regulate auxin responses. Their short half-lives have been proposed to be a result of selective degradation by ubiquitination pathways that possibly involve the AXR1 gene product (Abel et al., 1994), a possibility that can now be tested by examining the half-lives of these proteins in axrl mutants. Overall, this would provide a mechanism for executing rapid responses to auxin that could be equally rapidly attenuated. A proposed role for AXR1 must account for the hormoneinsensitive, recessive nature of axrl mutants. Most simply, there are two formal possibilities: (1) that it degrades a represser (i.e., that it negatively regulates a negative regulator) or (2) that it is involved in positively regulating auxin responses. The second possibility is difficult to reconcile with a role in protein degradation, although auxin responses may involve the repression of other pathways. AXR1 could achieve this by degrading key positive regulators within these pathways. Further genetic and biochemical characterization is necessary to examine these possibilities. Unfortunately, a major obstacle for assigning an auxin-specific function to AXR1 will be the known pleiotropy of auxin responses. Although the function of AXR1 in auxin signal transduction is unclear, its identification as a possible participant in ubiquitinmediated protein degradation is interesting because selective protein degradation is an emerging theme for desensitizing some signal transduction pathways in animal cells. Examples include regulation of cyclins, the human growth regulator p53, Drosophila Ftz, and yeast MATa2 (Gottesman and Maurizi, 1992). Compared with phosphorylation and other reversible modifications, proteolysis is a drastic way to inactivate a protein, but it is especially important for controlling proteins that are required either for only limited periods of time or under specific developmental or metabolic conditions and whose presence may be detrimental under other conditions. For example, cyclin acts to stimulate cell division, but its degradation is an essential regulatory mechanism for allowing the cell to advance to the next phase of the cell cycle.

10 1538 The Plant Cell PERSPECTIVES This brief overview of recent advances in the understanding of plant signal transduction mechanisms serves to illustrate severa1 emerging features: (1) the involvement of nove1 proteins containing more than one functional domain, (2) the concerted utilization of signaling systems previously thought to be specific for prokaryotes or eukaryotes, (3) the presence of basal leve1 fluxes through signaling pathways in the absence of a stimulus, (4) the central importance of negative regulation, and (5) the significance of cross-talk between pathways. The mechanism by which hybrid molecules such as COP1, ETR1, and A811 receive and transduce signals will be important to understand, because it is possible that they are central players in transmitting cross-talk information, for example, between different hormones. Cross-talk is likely to be of extreme importance in plants because of their developmental plasticity. The importance of cross-talk has indeed been inferred from the phenotypes of many of the mutants isolated in screens designed initially to be specific for one stimulus. Witness, for example, the pleiotropy of the cop and det mutants and the resistance of auxin-resistant mutants to other hormones (Pickett et al., 1990; Wilson et al., 1990). Such mechanisms are likely to be elucidated largely by genetic approaches; in fact, the information already available gives us some clues about possible mechanisms. For example, A611 may be regulated by Ca2+ signals generated not only from ABA, but also from other hormones or from phytochrome, because all are known to stimulate increases in cytosolic Ca2+ (see Bush, 1993, and references therein). Similarly, the COP and DET proteins may be targets for more than one stimulus, thus allowing access by different stimuli to the same signal transduction pathway or at least the same target. In addition, it is possible that different signals activate the same target in different cell types or that the same protein can behave as either an activator or a repressor, depending on the context (Lehming et al., 1994). The emergence of negative regulation as a central theme in plant signaling may to some extent be a biased interpretation of the true situation because of the type of genetic screens used to date. Nonetheless, there are clearly negative regulatory molecules of extreme importance in plant development, of which the COR DET, and CTR1 proteins are obvious examples. Recessive mutations in some of the ACD and LSD genes result in plants that spontaneously form necrotic lesions on leaves even in the absence of pathogens (Dietrich et al., 1994; Greenberg et al., 1994); these genes are also likely to encode negative regulators, implying that the defense response to pathogens may be mediated, to some extent, by release of repression mechanisms. The presence of signaling pathways primed to respond but blocked by repressors (such as the ethylene pathway and the defense response pathway) is a conceptually simple as well as an efficient mechanism for permitting cross-talk. Moreover, the lack of rigid developmental constraints during plant development may indicate that such negative regulatory mechanisms will be found to be prevalent in plant signal transduction. Future work must confirm or identify by biochemical methods the functions of gene products identified in genetic screens. Activities proposed for different proteins-e.g., ETRI, CTR1, and A811 -must be demonstrated, their regulation understood, and their substrates defined. The development of new methods may also be required to open up a signal transduction pathway biochemically. For example, the extension of microinjection and related methodologies to Arabidopsis should prove to be extremely powerful, as has been recently shown for tomato (Neuhaus et al., 1993; 8owler et al., 1994a). Combined with biochemical approaches, we believe new genetic screens should be designed to isolate mutants more likely to be affected in specific signaling pathways. Rather than designing screens aimed at isolating mutants that are constitutively switched on or off, it is now both feasible (Millar et al., 1992) and worthwhile to design screens for isolating mutants that are affected in their quantitative responses to different stimulus strengths. By screening for aberrant phenotypes in dynamic responses to a particular stimulus, mutants with defects specific to the stimulus are more likely to be obtained. Furthermore, the nature of the phenotype (e.g., a hypersensitive or an amplitude response) can give an indication of whether the mutation affects the activity of the receptor or downstream signal-transducing steps within the signaling pathway. In summary, both the genetic and biochemical approaches that have been applied to the study of plant signal transduction are clearly beginning to reveal the details of a variety of signaling networks. The union of these approaches should now be seen as the next major objective, because it will allow both the elucidation of the underlying mechanisms and the identification of the molecules responsible. Ultimately, it should permit a thorough understanding of how plants perceive, respond, and adapt to changing developmental and environmental stimuli. ACKNOWLEDGMENTS We would like to thank our colleagues in the laboratory for many stimulating discussions and especially Drs. Diana Horvath, Raphael Mayer, and Diane Shevell for their comments on the manuscript. We are also grateful to Dr. Joe Ecker for sharing his insights into ethylene responses with us. C.B. was supported in New York by a Science and Engineering Research Council (North Atlantic Treaty Organization) postdoctoral fellowship and later by afellowship from the Norman and Rosita Winston Foundation. Research in New York was supported by National lnstitutes of Health Grant No to N.-H.C. Received July 15, 1994; accepted September 16, REFERENCES Abel, S., Oeller, P.W., and Theologis, A. (1994). Early auxin-induced genes encode short-lived nuclear proteins. Proc. Natl. Acad. Sci. USA 91,

11 Plant Signal Transduction 1539 Ahmad, M., and Cashmore, A.R. (1993). HY4 gene of A. thaliana encodes a protein with characteristics of a blue-light photoreceptor. Nature 366, Anderson, B.E., Ward, J.M., and Schroeder, J.I. (1994). Evidence for an extracellular reception site for abscisic acid in Commelina guard cells. Plant Physiol. 104, Ang, L.-H., and Deng, X.-W. (1994). Regulatory hierarchy of photomorphogenic loci: Allele-specific and light-dependent interaction between the HY5 and COP7 loci. Plant Cell 6, Bleecker, A.B., Estelle, M.A., Somenrille, C., and Kende, H. (1988). lnsensitivity to ethylene conferred by a dominant mutation in Arabidopsis thaliana. Science 241, Blumer, K.J., and Johnson, G.L. (1994). Diversity in function and regulation of MAP kinase pathways. Trends Biol. Sci. 19, Boddy, M.N., Fleemont, P.S., and Borden, K.L.B. (1994). The p53- associated protein MDM2 contains a newly characterized zincbinding domain called the RlNG finger. Trends Biol. Sci. 19, Bowler, C., Neuhaus, G., Yamagata, H., and Chua, N.-H. (1994a). Cyclic GMP and calcium mediate phytochrome phototransduction. Cell 77, Bowler, C., Yamagata, H., Neuhaus, G., and Chua, N.-H. (1994b). Phytochrome signal transduction pathways are regulated by reciprocal control mechanisms. Genes Dev. 8, Boylan, M., Douglas, N., and Quail, P.H. (1994). Dominant negative suppression of Arabidopsis photoresponses by mutant phytochrome A sequences identifies spatially discrete regulatory domains in the photoreceptor. Plant Cell 6, Brijdenfeldt, R., and Mohr, H. (1988). Time courses for phytochromeinduced enzyme levels in phenylpropanoid metabolism (phenylalanine ammonia-lyase, naringenin-chalcone synthase) compared with time courses for phytochrome-mediated end-product accumulation (anthocyan i n, que rcet i n). PI anta 176, Bush, D.S. (1993). Regulation of cytosolic calcium in plants. Plant Physiol. 103, Casey, P.J. (1994). Lipid modifications of G proteins. Curr. Opin. Cell Biol. 6, Castle, L.A., and Meinke, D.W. (1994). A FUSCA gene of Arabidopsis encodes a nove1 protein essential for plant development. Plant Cell 6, Chang, C., Kwok, S.F., Bleecker, A.B., and Meyerowitz, E.M. (1993). Arabidopsis ethylene-response gene ETR7: Similarity of product to two-component regulators. Science 262, Cherry, J.R., Hondred, D., Walker, J.M., and Vierstra, R.D. (1992). Phytochrome requires the 6-kDa N-terminal domain for full biological activity. Proc. Natl. Acad. Sci. USA 89, Cherry, J.R., Hondred, D., Walker, J.M., Keller, J.M., Hershey, H.P., and Vierstra, R.D. (1993). Carboxy-terminal deletion analysis of oat phytochrome A reveals the presence of separate domains required for structure and biological activity. Plant Cell 5, Chory, J. (1992). A genetic model for light-regulated seedling development in Arabidopsis. Development 115, Chory, J. (1993). Out of darkness: Mutants reveal pathways controlling light-regulated development in plants. Trends Genet. 9, Chory, J., Aguilar, N., and Peto, C.A. (1991). The phenotype of Arabidopsis thaliana detl mutants suggests a role for cytokinins in greening. Symp. SOC. Exp. Biol. 45, Chory, J., Reinecke, D., Sim, S., Washburn, T., and Brenner, M. (1994). A role for cytokinins in deetiolation in Arabidopsis. Plant Physiol. 104, Clack, T., Mathews, S., and Sharrock, R.A. (1994). The phytochrome apoprotein family in Arabidopsis is encoded by five genes: The sequences and expression of PHYD and PHYE. Plant MOI. Biol. 25, Cohen, P. (1989). The structure and regulation of protein phosphatases. Annu. Rev. Biochem. 58, Cooper, J.P., Roth, S.Y., and Simpson, R.T. (1994). The global transcriptional regulators, SSN6 and TUPI, play distinct roles in the establishment of a repressive chromatin structure. Genes Dev. 8, Dehesh, K., Franci, C., Parks, B.M., Seeley, K.A., Short, T.W., Tepperman, J.M., and Quail, P.H. (1993). Arabidopsis HY8 locus encodes phytochrome A. Plant Cell 5, Deng, X.-W. (1994). Fresh view of light signal transduction in plants. Cell 76, Deng, X.-W., Matsui, M., Wei, N., Wagner, D., Chu, A.M., Feldmann, K.A., and Quail, P.H. (1992). COP7, an Arabidopsis regulatory gene, encodes a protein with both a zinc-binding motif and a GB homologous domain. Cell 7l, Dietrich, R.A., Delaney, T.P., Uknes, S.J., Ward, E.R., Ryals, J.A., and Dangl, J.L. (1994). Arabidopsis mutants simulating disease resistance response. Cell 77, Drumm-Herrel, H. (1984). Blue/UV light effects on anthocyanin synthesis. In Blue Light Effects in Biological Systems, H. Sanger, ed (Berlin: Springer-Verlag), pp Ehmann, B., Ocker, B., and Schafer, E. (1991). Development- and light-dependent regulation of the expression of two different chalcone synthase transcripts in mustard cotyledons. Planta 183, Estelle, M. (1992). The plant hormone auxin: lnsight in sight. BioEssays 14, Estelle, M.A., and Somenrille, C. (1987). Auxin-resistant mutants of Arabidopsis thaliana with an altered morphology. MOI. Gen. Genet. 206, Fmhnmeyer, H., Ehmann, B., Kretsch, T., Rocholl, M., Harter, K., Nagatanl, A., Furuya, M., Batschauer, A., Hahlbrock, K., and Schgfer, E. (1992). Differential usage of photoreceptors for chalcone synthase gene expression during plant development. Plant J. 2, Furuya, M. (1993). Phytochromes: Their molecular species, gene families, and functions. Annu. Rev. Plant Physiol. Plant MOI. Biol. 44, Gilroy, S., and Jones, R.L. (1994). Perception of gibberellin and abscisic acid at the externa1 face of the plasma membrane of barley (Hordeum vulgare L.) aleurone protoplasts. Plant Physiol. 104, Gilroy, S., Read, N.D., and Trewavas, A.J. (1990). Elevation of cytoplasmic calcium by caged calcium or caged inositol trisphosphate initiates stomatal closure. Nature 346, Gilroy, S., Fricker, M.D., Read, N.D., and Trewavas, A.J. (1991). Role of calcium in signal transduction of Commelina guard cells. Plant Cell 3, Gottesman, S., and Maurizi, M.R. (1992). Regulation by proteolysis: Energy-dependent proteases and their targets. Microbiol. Rev. 56,

12 1540 The Plant Cell Greenberg, J.T., Guo, A., Klessig, D.F., and Ausubel, F.M. (1994). Programmed cell death in plants: A pathogen-triggered response activated coordinately with multiple defense functions. Cell 77, Grimm, R., Gast, D., and Rüdiger, W. (1989). Characterization of a protein-kinase activity associated with phytochrome from etiolated oat (Avena sativa L.) seedlings. Planta 178, Guzmln, P., and Ecker, J.R. (1990). Exploiting the triple response of Arabidopsis to identify ethylene-related mutants. Plant Cell 2, Hughes, D.A. (1994). Histidine kinases hog the limelight. Nature 369, Hunt, R.E., and Pratt, L.H. (1980). Partia1 characterization of undegraded oat phytochrome. Biochemistry 19, Imbalzano, A.N., Kwon, H., Green, M.R., and Kingston, R.E. (1994). Facilitated binding of TATA-binding protein to nucleosomal DNA. Nature 370, Jabben, M., Shanklin, J., and Vierstra, R.D. (1989). Ubiquitinphytochrome conjugates. J. Biol. Chem. 264, Johnson, E., Bradley, M., Harberd, N.P., and Whitelam, G.C. (1994). Photoresponses of light-grown phya mutants of Arabidopsis. Plant Physiol. 105, Jones, J.D.G. (1994). Resistance crumbles? Curr. Biol. 4, Kawamura, S. (1993). Rhodopsin phosphorylation as a mechanism of cyclic GMP phosphodiesterase regulation by S-modulin. Nature 362, Kieber, J.J., Rothenberg, M., Roman, G., Feldmann, K.A., and Ecker, J.R. (1993). CTR7, a negative regulator of the ethylene response pathway in Arabidopsis, encodes a member of the Raf family of protein kinases. Cell 72, Koch, K.-W. (1992). Biochemical mechanism of light adaptation in vertebrate photoreceptors. Trends Biol. Sci. 17, Koornneef, M., Reuling, G., and Karssen, C.M. (1984). The isolation and characterization of abscisic acid-insensitive mutants of Arabidopsis thaliana. Physiol. Plant. 61, Koutalos, Y., and Yau, K.-W. (1993). A rich complexity emerges in phototransduction. Curr. Opin. Neurobiol. 3, Kwon, H., Imbalzano, A.N., Khavari, P.A., Kingston, R.E., and Green, M.R. (1994). Nucleosome disruption and enhancement of activator binding by a human SWllSNF complex. Nature 370, Lamb, C.J. (1994). Plant disease resistance genes in signal perception and transduction. Cell 76, Lehming, N., Thanos, D., Brickman, J.M., Ma, J., Maniatis, T., and Ptashne, M. (1994). An HMG-like protein that can switch a transcriptional activator to a repressor. Nature 371, Leung, J., Bouvier-Durand, M., Morris, P.X., Guerrier, D., Chefdor, F., and Giraudat, J. (1994). Arabidopsis ABA response gene ABl7: Features of a calcium-modulated protein phosphatase. Science 264, Leyser, H.M.O., Lincoln, C.A., Timpte, C., Lammer, D., Turner, J., and Estelle, M. (1993). Arabidopsis auxin-resistance gene AXR7 encodes a protein related to ubiquitin-activating enzyme El. Nature 364, Li, H.-m., Altschmied, L., and Chory, J. (1994). Arabidopsis mutants define downstream branches in the phototransduction pathway. Genes Dev. 8, Lincoln, C., Britton, J.H., and Estelle, M. (1990). Growth and development of the axr7 mutants of Arabidopsis. Plant Cell2, Maeda, T., Wurgler-Murphy, S.M., and Saito, H. (1994). A twocomponent system that regulates an osmosensing MAP kinase cascade in yeast. Nature 369, McMichael, R.W.J., and Lagarias, J.C. (1990). Phosphopeptide mapping of Avena phytochrome phosphorylated by protein kinases in vitro. Biochemistry 29, McNellis, T.W., von Arnim, A.G., Araki, T., Komeda, Y., Misbra, S., and Deng, X.-W. (1994). Genetic and molecular analysis of an allelic series of cop7 mutants suggests functional roles for the multiple protein domains. Plant Cell 6, Meyer, K., Leube, M.P., and Grill, E. (1994). A protein phosphatase 2C involved in ABA signal transduction in Arabidopsis thaliana. Science 264, Millar, A.J., Short, S.R., Hiratsuka, K., Chua, N.-H., and Kay, S.A. (1992). Firefly luciferase as a reporter of regulated gene expression in higher plants. Plant MOI. Biol. Rep. 10, Millar, A.J., McGrath, R.B., and Chua, N.4. (1994). Phytochrome phototransduction pathways. Annu. Rev. Genet. 28, in press. Misbra, S., Müller, A.J., Weiland-Heidecker, U., and Jürgens, G. (1994). The FUSCA genes of Arabidopsis: Negative regulators of light responses. MOI. Gen. Genet. 244, Nagatani, A., Reed, J.W., and Chory, J. (1993). lsolation and initial characterization of Arabidopsis mutants that are deficient in phytochrome A. Plant Physiol. 102, Neuhaus, G., Bowler, C., Kern, R., and Chua, N.-H. (1993). Calcium/calmodulin-dependent and -independent phytochrome signal transduction pathways. Cell 73, Nick, P., Ehmann, E., Furuya, M., and Schafer, E. (1993). Cell communication, stochastic cell responses, and anthocyanin pattern in mustard cotyledons. Plant Cell 5, Ota, IA., and Varshavsky, A. (1993). A yeast protein similar to bacteria1 two-component regulators. Science 262, Parkinson, JS(1993). Signal transduction schemes of bacteria. Cell 73, Parks, B.M., and Quail, P.H. (1991). Phytochrome-deficient hyl and hy2 long hypocotyl mutants of Arabidopsis are defective in phytochrome chromophore biosynthesis. Plant Cell 3, Parks, B.M., and Quail, P.H. (1993). hy8, a new class of Arabidopsis long hypocotyl mutants deficient in functional phytochrome A. Plant Cell 5, Pepper, A., Delaney, T., Washburn, T., bole, D., and Chory, J. (1994). DE T7, a negative regulator of light-mediated development and gene expression in Arabidopsis, encodes a nove1 nuclear-localized protein. Cell 78, Plckett, F.B., Wilaon, A.K., and Estelle, M. (1990). The aux7 mutation of Arabidopsis confers both auxin and ethylene resistance. Plant Physiol. 94, Quail, P.H. (1991). Phytochrome: A light-activated molecular switch that regulates plant gene expression. Annu. Rev. Genet. 25, Quail, P.H., Brlggs, W.R., Chory, J., Hangarter, R.P., Harberd, N.P., Kendrick, R.E., Koornneef, M., Parks, E., Sharrock, R.A., Schafer, E., Thompson, W.F., and Whitelam, G.C. (1994). Spotlight on phytochrome nomenclature. Plant Cell 6, Raz, V., and Fluhr, R. (1992). Calcium requirement for ethylenedependent responses. Plant Cell 4,

13 Plant Signal Transduction 1541 Raz, V., and Fluhr, R. (1993). Ethylene signal is transduced via protein phosphorylation events in plant cells. Plant Cell 5, Reed, J.W., Nagpal, P., Poole, D.S., Furuya, M., and Chory, J. (1993). Mutations in the gene for the red/far-red light receptor phytochrome B alter cell elongation and physiological responses throughout Arabidopsis development. Plant Cell 5, Reed, J.W., Nagatani, A., Ellch, T.D., Fagan, M., and Chory, J. (1994). Phytochrome A and phytochrome B have overlapping but distinct functions in Arabidopis development. Plant Physiol. 104, Shacklock, P.S., Read, N.D., and Trewavas, A.J. (1992). Cytosolic free calcium mediates red light-induced photomorphogenesis. Nature 358, Sharrock, R.A., and Quail, RH. (7989). Nove1 phytochrome sequences in Ara6i&psis hdiana: Structure, evolution, and differential expression of a plant regulatory photoreceptor family. Genes Dev. 3, Stockhaus, J., Nagatanl, A., Halfter, U., Kay, S., Furuya, M., and Chua, N.-H. (1992). Serine-to-alanine substitutions at the aminoterminal region of phytochrome A result in an increase in biological activity. Genes Dev. 6, Susek, R.E., Ausubel, F.M., and Chory, J. (1993). Signal transduction mutants of Arabidopsis uncouple nuclear CAB and RBCS gene expression from chloroplast development. Cell 74, Szerszen, J.B., Szczyglowski, K., and Bandunki, R.S. (1994). iaglu, a gene from Zea mays involved in conjugation of growth hormone indole-3-acetic acid. Science 265, Tretyn, A., Kendrick, R.E., and Wagner, G. (1991). The role(s) of calcium ions in phytochrome action. Photochem. Photobiol. 54, Trewavas, A., and Gilroy, S. (1991). Signal transduction in plant cells. Trends Genet. 7, Wel, N., Kwok, S.F., von Arnim, A.G., Lee, A., McNellis, T.W., Piekos, B., and Deng, X.-W, (1994a). Arabidopsis COP8, COP70, and COP77 genes are involved in repression of photomorphogenic development in darkness. Plant Cell 6, Wei, N., Chamovitz, D.A., and Deng, X.-W. (1994b). Arabidopsis COP9 is a component of a nove1 signaling complex mediating light control of development. Cell 78, Wenng, A., Batschauer, A., Ehmann, B., and Schafer, E. (1990). Temporal pattern of gene expression in cotyledons of mustard (Sinapis alba L.) seedlings. Bot. Acta 103, Whitelam, G.C., and Harberd, N.P. (1994). Action and function of phytochrome family members revealed through the study of mutant and transgenic plants. Plant Cell Environ. 17, Whitelam, G.C., Johnson, E., Peng, J., Carol, P., Andenon, M.L., Cowl, J.S., and Harberd, N.P. (1993). Phytochrome A null mutants of Arabidopsis display a wild-type phenotype in white light. Plant Cell 5, Wilson, A.K., Pickett, F.B., Turner, J.C., and Estelle, M. (1990). A dominant mutation in Arabidopsis confers resistance to auxin, ethylene and abscisic acid. MOI. Gen. Genet. 222, Wong, Y.-S., Cheng, H.-C., Walsh, D.A., and Lagarias, J.C. (1986). Phosphorylation of Avena phytochrome in vitro as a probe of lightinduced conformational changes. J. Biol. Chem. 261, Zeevaart, J.A.D., and Creelman, R.A. (1988). Metabolism and physiology of abscisic acid. Annu. Rev. Plant Physiol. Plant MOI. Biol. 39,