A genetic model for light-regulated seedling development in Arabidopsis

The mechanisms underlying developmental processses in multicellular eukaryotic organisms are only beginning to be unravelled. In plants, development is controlled by the light environment and as such, distinct morphologies arise from growing plants under dark or light conditions (Mullet 1988; Chory, 1991). Dark-grown (etiolated) dicotyledonous seedlings have elongated hypocotyls, small folded cotyledons, and undeveloped chloroplasts (etioplasts) Conversely, light inhibits hypocotyl elongation, and induces leaf expansion and differentiation and chloroplast development (deetiolation) (Mullet, 1988; Dale, 1988; Gruissem, 1989). The etiolated state is accompanied by little or no expression of several light-regulated genes, including the nuclear genes encoding the light-harvesting chlorophyll a/b binding proteins (cab), the small subunit of the RuBP carboxylase/oxygenase (rbcS), and chaicone synthase (chs) (Silverthorne and Tobin, 1987; Gilmartin et al., 1990). In the light, the expression of these genes is restricted to specific cell types; for instance, chs is expressed in the epidermis, and the rbcS and cab genes are expressed in chloroplast-containing cells (mesophyll) (Chory, 1991). Thus, light and cell-specific factors work in concert to produce specific expression patterns for these genes In addition to light, several phytohormones have been implicated in de-etiolation responses, including cytokinins and gibberellins (Stetler and Laetsch, 1965; Harvey et al., 1974; Flores and Tobin, 1986; Mathis et al., 1989). How light might interact with these hormone signal transduction pathways is not understood.

A complete signal transduction cascade has yet to be elucidated between the steps of photon absorption by a photoreceptor and changes in the expression of light-regulated genes. Several classes of photoreceptors mediate light responses, including protochlorophyllide, blue- and UV-light absorbing receptors, and the red/far-red-hght absorbing receptors, the phytochromes (Laganas, 1985; Colbert, 1988; Senger and Schmidt, 1986) Most efforts to study this process have been directed at the biochemical and molecular charactenzation of phytochrome. Phytochrome is a soluble pigmented protein that exists in two spectrally distinct, photointerconvertible forms P_r, the red-absorbing form, and P_rf, the far-red absorbing form. The spectral qualities of purified P_r and Pfr are the results of the combined properties of the 120 × 10³M_r apoprotein with its thioether-linked bilitriene chromophore (Lagarias, 1985, Colbert, 1988). Photoconversion of P_r to Pfr induces a diverse array of morphogenic responses, whereas reconversion of Pf_r to P_r cancels the induction of the responses. Thus, P_fr is considered the active and P_r the inactive form of the photoreceptor. The molecular mechanism by which Pf_r induces morphological and gene expression responses in the developing seedling is not known. A further complication is that phytochrome is actually a family of photoreceptors For instance, in the small cruciferous plant, Arabidopsis thaliana, there are at least five expressed phytochrome apoprotein genes (Sharrock and Quail, 1989; R. Sharrock, personal communication). The spatial and temporal expression patterns of the different phytochromes and what roles each plays during development are largely unknown.

Blue-light also causes profound changes in the morphology of the developing young seedling (Senger and Schmidt, 1986; Chory, 1991). Though a blue-light receptor has not been chemically defined, recent data suggest the involvement of G-proteins in the signal transduction cascade (Warpeha et al., 1991). Further, the isolation of Arabidopsi mutations that cause defects in just one or a few blue-light responses suggests that the blue-light signal transduction pathways are also likely to be complex (Koomneef et al, 1980; Khurana et al., 1989; Khurana and Poff, 1989; Liscum and Hangarter, 1991).

A combined molecular and classical genetic approach will help dissect what are likely to be very complex photomorphogenetic signal transduction pathways. Arabidopsis thaliana is an ideal organism for saturation mutagenesis experiments aimed at identifying genes involved in directing the downstream light-regulated responses. To isolate mutants affected in early seedling development in response to light, we and others have taken advantage of simple screens based on the characteristic morphologies of light- and dark-grown seedlings (Koomneef et al., 1980; Chory et al., 1989a, 1989b, 1991a; Deng et al., 1991; Liscum and Hangarter, 1991). The mutations identified to date fall into two phenotypic classes that define 14 genes, mutations in 10 of these genes produce plants with impaired responses to red (Koomneef et al., 1980; Chory et al., 1989a) or blue light (Koomneef et al., 1980; Liscum and Hangarter, 1991), and mutations in another four genes result in plants that exhibit light-mediated responses more readily than wild-type plants (Chory et al., 1989b, 1991a; Chory, 1991; Deng et al., 1991). This study focuses on eight of these 14 loci

The first phenotypic class of mutants have a partially etiolated morphology when grown in white light. These mutants were designated, hy, for long hypocotyl because they do not undergo light-induced inhibition of the rate of hypocotyl elongation (Redei and Horono, 1964). Homozygous recessive hy mutations define seven complementation groups, six of which, hyl-hy6 will be considered here (Koomneef et al., 1980; Chory et al., 1991; J. Chory, unpublished data). Mutations in any one of four genes, hy 1, hy2, hy3, or hy6 result in phytochrome deficiencies (Koomneef, 1980; Chory et al., 1989a; Parks et al, 1989; Nagatani et al., 1991). hy4 mutants, in contrast, show reduced inhibition of hypocotyl elongation in blue light while maintaining normal phytochrome levels and responses (Koomneef et al., 1980). Plants homozygous for hy5 alleles are defective in red-light-mediated hypocotyl growth inhibition, though phytochrome activity appears to be normal (Koomneef et al., 1980).

The second phenotypic class of mutants show many characteristics of light-grown plants even when grown in complete darkness. We have designated these mutants det, (deetiolated), because they are de-etiolated in the dark, instead of having the usual etiolated seedling morphology. Recessive mutations in any one of three DET genes result in darkgrown plants that grow as light-grown wild-type seedlings (Chory et al., 1989b, 1991a; J. Chory, unpublished data) More recently, similar mutations in a fourth gene, COP1, were identified (Deng et al., 1991).

We have been studying homozygous mutant det1 and det2 seedlings in the most detail. Dark-grown det1 and det2 plants have inhibited hypocotyl growth rates, expanded cotyledons, and developed leaves. Additionally, several light-regulated genes, including cab, chs, and rbcS, are expressed at high levels in the dark in these mutants (Chory et al., 1989b, 1991a). Further, light-grown plants appear to require DET1 for spatial repression of light-regulated genes, since cab, chs, and rbcS are ectopically expressed in DET1 mutants (Chory and Peto, 1990). In contrast, homozygous det2 mutations affect photoperiodic responses in light-grown plants, including delayed timing of flowering, reduced dark adaption of cab gene expression, and delayed timing of leaf and chloroplast senescence (Chory et al., 1991a). Thus, DET2 may play a negative role in the temporal elaboration of light responses during A rabidopsis development.

The experiments described in this paper explore how the activities of the six H Y and two DET genes interact to control the switch from an etiolated to de-etiolated developmental strategy The results from epistasis studies using doubly mutant lines suggest a hierarchical regulatory network among these genes in the control of the downstream lightregulated responses. This analysis should provide the necessary genetic framework for future studies designed to test the molecular basis of these interactions

General methods for the growth and handling of Arabidopsis were described (Somerville and Ogren, 1982) The following mutant strains were used in this work:

All mutant alleles studied are recessive when back-crossed to wild type. The hy1-hy5 mutations originally identified by Koomneef et al. (1980) are in the Landsberg ecotype carrying the erecta mutation. Ay2-2, Ay6-1, and the det alleles were isolated in the wildtype Columbia background. For double mutant studies involving the det mutants and the hy 1-hy5 reference alleles, the det mutations were first back-crossed four times into the Landsberg erecta background, and the phenotypes of the single and double mutants were compared to a wild-type Landsberg erecta line. For double mutants involving the det mutants with either the hy6-1 or hyl-2 alleles, the wild-type was the Columbia line I did not observe significant differences between the ecotype Landsberg and Columbia controls Wild-type alleles are symbolized by upper case italics, the wildtype gene product is in upper case plain text. Mutant alleles are symbolized by lower case italics Doubly mutant strains were created by manual cross-pollination, using homozygous single mutants as parents Most crosses were performed reciprocally, with the exception of crosses that involved det2 alleles for which det2 was always used as the female parent (see Chory et al, 1991a). The resulting F1 plants were allowed to self-pollmate and double mutants were scored in the F2 generation whenever possible. For most crosses, no new phenotype arose in the F2 plants, so F2 plants with the phenotype of one or the other single mutant were picked and allowed to self-pollmate. Double mutants were scored in the F3 generation as segregants in the progeny of such plants In two crosses, namely the hy3-det1 cross and the hy5-det1 cross, some ambiguity in scoring the double mutant was present in the F3 generation The genotype of the putative double mutant was thus verified by performing crosses to each of the single mutant parents and scoring the F₁ plants

Plants were grown at 20°C under a mixture of fluorescent and incandescent lights at an intensity of 350 μ E/m²/sec (“high-white light conditions”) Methods for the growth of plants in pots, seed harvesting, and cross-pollination have been described (Somerville and Ogren, 1982, Chory et al, 1989a, 1989b) Single and double mutants and wild type were always grown together under the same light and humidity conditions for the times indicated in the text The treatment of plants with specific wavelengths of light were as follows: red, six 40-W GE cool white bulbs and two incandescent bulbs with red Plexiglas filter, blue: six 40-W GE cool white bulbs with a blue Plexiglas filter. The outputs of the various light sources, measured at seedling level with a Biosphencal Instruments Inc QSL-100 light detector were “high” white light. 350 μ E/m²/sec, “low” white light 60 μ E/m²/sec; red 55.5 μ E/m²/sec, and blue-38 p.E/m²/sec. Dark-grown seeds were germinated for 24 hours in the light on synthetic medium plus sucrose and then transferred to total darkness for 7 – 10 days, as previously described (Chory et al., 1989b) A green safelight filter was used during all dark manipulations.

Chlorophyll determinations were performed on leaves and stems from wild-type and mutant plants harvested after 12 days growth in high photon fluence rate white light and immediately frozen in liquid nitrogen The frozen tissue was later ground in liquid nitrogen in a mortar and pestle, and the chlorophyll was extracted repeatedly into 80% acetone in the dark until the pellet appeared colorless. Chlorophyll a and b contents were calculated using Mac Kinney’s specific absorption coefficients (MacKinney, 1941), in which chlorophyll a = 12.7(A₆₆₃) – 2 69(A₆₄₅) and chlorophyll b = 22 9(A₆₄₅) – 4 48(A₆₆₃) The total specific chlorophyll content is expressed as micrograms of chlorophyll per seedling

For anthocyanin determinations on dark-grown tissue, 0 1 g of frozen plant tissue was ground m a 1 5 ml microfuge tube with a disposable pestle, and total plant pigments were extracted overnight in 0 3 ml of 1 % HC1 in methanol. After the addition of 0 2 ml of H2O, chlorophyll was separated from the anthocyanins by extraction with an equal volume of chloroform. The quantity of anthocyanins was determined by spectrophotometric measurements of the aqueous:methanol phase (A₅₃₀ – A₆₅₇) and normalized to the total fresh weight of tissue used in each sample (Beggs et al., 1987, Rabino and Mancinelli, 1986). For tissues grown in the light, similar mesurements were made in duplicate (two seedlings/sample) on seedlings grown for 12 days in high photon fluence rate white light.

Hypocotyl elongation was measured with a ruler after growth for 10 days in the various light regimens indicated in Table 5 Measurements for dark-grown seedlings were made after eight days growth. At least 50 seedlings were measured for each sample.

RNA extraction, separation, and gel blot conditions were previously described (Chory et al, 1991a). The DNA probes for nuclear and chloroplast genes used in these studies were published elsewhere (Chory et al, 1989b). To normalize for RNA loading, filters were stripped and rehybndized with an rDNA probe Autoradiograms for different exposure times were scanned with a densitometer. Relative amounts of mRNAs were determined by peak-height measurements, and relative mRNA levels reported are an average of two separate hybridizations

Arabidopsis is typical of dicotyledonous seedlings in that dark-grown (etiolated) seedlings are developmentally arrested, having extended hypocotyls, no cotyledon expansion, no leaves or chloroplasts, and no detectable chlorophyll and anthocyanin pigments (Chory et al., 1989b). A set of well-characterized genes, the light-regulated genes, is not expressed or is expressed at a very low level (Chory et al., 1989b). In contrast, when seedlings are exposed to light, there are profound differences in the morphology of the plant, including rapid inhibition of stem elongation, expansion of cotyledons, and development of leaves and chloroplasts (de-etiolation). During de-etiolation in Arabidopsis, light-regulated gene expression increases up to 100-fold (Chory et al, 1989b; Chory and Peto, 1990). Upon further exposure to a constant and well-defined light environment, the average Arabidopsis wild-type seedling will make approximately 10 vegetative rosette leaves followed by a transition to floral growth at approximately 21 days post-germination (Table 1, and Chory et al., 1991a). The vegetative to floral transition is characterized by rapid growth of the floral bolt accompanied by leaf and chloroplast senescence. During leaf senescence, expression of light-regulated genes is relatively low (a decrease of about 10-fold from early seedling development; Chory, 1991).

Mutations in three separate loci, hy1, hy2, or hy6 result in seedlings with little to no detectable phytochrome activity in thedark(Koomneef et al., 1980; Chory et al., 1989a; Parkset al., 1989). hy1 maps to position 17.0 cM on chromosome 2, hy2 maps to chromosome 3, position 0.0 cM, and hy6 maps to chromosome 2, position 21 9 cM on the morphological marker map (Koomneef, 1990; Chory et al, 1991a). Though a complete phenotypic analysis of all 19 hy1, hy2, and fty6 alleles has not yet been performed, strong alleles of hy 1 (hy1 21.84N), hy2 (hy2-2), and hy6 (hy6-1) have no detectable phytochrome spectral activity (Koomneef et al., 1980; Parks et al., 1989, Chory et al., 1989a), whereas weak alleles of hy1 (d412) and hy2 (To76) have about 25% of wild-type phytochrome spectroscopic activity (Koomneef et al., 1980; Chory et al., 1989a). It has been proposed that the lesions in the hyl, hy2, and hy6 mutants affect either the synthesis or attachment of the bilitriene chromophore of phytochrome, since these mutants contain both the major light-labile and light-stable phytochrome apoproteins (Parks et al., 1989; Chory et al., 1989a; Nagatani et al., 1991 ; Parks and Quail, 1991). Since these two major phytochromes, and perhaps all phytochromes, share the same chromophore, these mutants are severely deficient in phytochrome activity In addition to the long hypocotyl phenotype for which they were selected, severe alleles of hyl, hy2 and hy6 are pale yellow, make fewer leaves, have increased apical dominance, and flower prematurely when compared to wild-type plants (See Table 1). Molecular, biochemical, and ultrastructural studies indicate that these mutants do not complete the leaf and chloroplast developmental program when grown in white light (Chory et al., 1989a).

hy3 maps to chromosome 2, position 2.2 cM on the morphological marker map (Koomneef, 1990). Studies in our laboratory and by Somers et al. have recently shown that homozygous recessive hy3 alleles have decreased accumulation of the major light-stable phytochrome of Arabidopsis to about 5% of wild-type levels (Nagatam et al., 1991; Somers et al., 1991). Our most recent experiments show that the gene that most likely encodes this phytochrome (gene designation PHYB, as per Sharrock and Quail, 1989) and hy3 mutations are genetically linked (within 0.4 cM on the RFLP map, based on analysis of 220 chromatids) Furthermore, two independent hy3 alleles contain mutations in the PHYB gene (P. Nagpal, J. Reed, and J. Chory, unpublished data) Thus, it seems likely that HY3 encodes the type B phytochrome apoprotein. The hy3 mutants have a stoking phenotype, including a long hypocotyl, very elongated petioles and leaves, and an elongated flowering bolt (Table 1, Koomneef et al., 1980; Chory et al., 1989a). Plants homozygous for hy3 mutations are also slightly defective in greening, accumulating less chlorophyll and having fewer chloroplasts per mesophyll cell than wild-type plants (Table 2, J. C, unpublished data).

hy4 (position 6.2 cM, chromosome 4) mutants show reduced hypocotyl growth inhibition in blue-light while maintaining normal phytochrome levels and red/far-red responses (Koomneef et al., 1980; Koomneef, 1990). hy4 (2.23N) may be a null allele since it was generated by fast neutron mutagenesis which has been shown to generate small deletions in Arabidopsis DNA. Hy4 (2.23N) homozygous recessive mutations affect only a subset of blue-light regulated responses, including inhibition of hypocotyl growth (Tables 1, 5) and they cause a 50% decrease in blue-light induced anthocyanin biosynthesis (Table 3). Other known blue-light responses, including stomatai opening, chloroplast development, and phototropism, are normal in these mutants (J. Chory and S. Assmann, unpublished data). The biochemical defects in hy4 reduced or loss-of-function alleles are unknown.

Only two alleles of hy5 (chromosome 5, position 1.1 cM)are known, and one (hy5 Ci88) was available for this study (Koomneef et al., 1980; Koomneef, 1990). This hy5 allele is defective in red-light induced hypocotyl growth inhibition responses, but has normal levels of phytochrome (Koomneef et al., 1980). In addition, hy5 (Ci88) mutants have a higher specific chlorophyll content, and a lower chlorophyll alb ratio than wild-type Arabidopsis grown under the same incident light conditions (Table 2). Though the molecular lesion in hy5 mutants is not known, a hy5 mutation in combination with either a hy1, hy2, or hy3 mutation, has an additive effect on inhibition of hypocotyl elongation (Koomneef et al., 1980). This indicates the possibility that/-/F5 encodes a component on a unique red-light transduction pathway.

Det1 maps at position 13.4 cM on the fourth chromosome in the Arabidopsis RFLP map of Chang et al., (1988); (T. Delaney and J. Chory, unpublished data). Six alleles of DET1 exist. All six are completely recessive when backcrossed to the wild-type Columbia line (H. Cabrera and J. Chory, unpublished data). Detailed descriptions of the light- and dark-grown phenotypes of det1-1 have been published (Chory et al., 1989b, 1991) and are summarized here. When grown in the dark, det1-1 alleles have the gross morphology of light-grown plants, including hypocotyl growth rate inhibition, anthocyanin production, and the development of chloroplasts and rosette leaves. Phytochrome spectral activity and regulation appear to be similar to wild-type etiolated seedlings though several light-regulated genes, including cab, chs, and rbcS, are expressed in the dark in these mutants (Chory et al., 1989b). Further, light-grown plants require DET1 for tissue-specific repression of light-regulated genes, since cab, chs, and rbcS are ectopically expressed in detl-1 alleles (Chory and Peto, 1990). The dark- and light-grown seedling phenotypes of DET1-3 homozygotes are similar to those of det1-1 (Table 1). Homozygous det1-2 alleles are slightly less severe in that fewer rosette leaves develop in the dark than for either det1-1 or det1-3 mutants (J.C., unpublished data).

det2 maps to position 32.9 cM on chromosome 2 (Chory et al., 1991a) of the morphological map of Koomneef (1990). There are five det2 alleles and each is completely recessive when backcrossed to wild type Since all five alleles have the same phenotype (Chory et al., 1991a), only det2-1 is described here. Like det1 mutants, dark-grown homozygous det2-1 mutants have normal levels of phytochrome, increased expression of light-regulated genes, and show cotyledon expansion and an inhibition of the rate of hypocotyl elongation in the absence of light (Chory et al., 1991a; L Pratt and J. Chory, unpublished data). Unlike det1 mutants, dark-grown det2 seedlings develop only primary leaf buds, but not rosette leaves. Though there is cotyledon development and some leaf differentiation in dark-grown det2 mutants, chloroplasts do not differentiate. Light-grown det2 mutants appear to have defects in photoperiodic timing, characterized by a prolonged vegetative phase, a delay in leaf and chloroplast senescence (accompanied by high levels of cab and rbcS gene expression after flowering), and a failure to repress the accumulation of light-regulated RNAs during dark periods (Chory et al., 1991a).

The phenotypic and genetic analysis of det and hy single mutant lines suggests that DET1 and DET2 proteins are required to couple the red- and blue-light signals to the downstream light-regulated developmental and gene expression responses in Arabidopsis In a variety of genetic systems, epistasis tests have helped define functional relationships between mutationally defined genes (e.g., Ferguson et al., 1987; Ambros, 1989; Bowman et al., 1991; Chant and Herskowitz, 1991). I reasoned that by comparing the phenotypes of doubly mutant lines to single mutants, I could provide a genetic framework for the interactions between the various photomorphogenetic genes. This genetic formalism, which does not presuppose any particular molecular mechanism for the interactions, will be useful for future molecular analysis of these interactions.

To investigate the functional relationships among the activities of hyl-hy6, det1 and det2, doubly mutant strains were constructed carrying mutations in one hy gene and one det gene or in two det genes. The effects of the double mutant combinations on the morphology and gene expression responses in light- and dark-grown plants were examined. In particular, the responses studied included: hypocotyl growth, chlorophyll and anthocyanin accumulation, floral induction, and mRNA accumulation of light-induced genes in dark-grown seedlings. The data for each individual analysis, e.g., chlorophyll accumulation, are presented in a separate table that compares all the single and double mutants with the wild type. For these experiments, I used the most severe allele(s) available for each gene, with the exception of hy5 for which I had only one allele, C188.1 do not know if any of the alleles used were null; however, all the alleles used were fully recessive, with the exception of hy3 alleles which are semi-dominant. For a given experiment, when epistasis of one mutation to a second was observed, 1 assumed an order of gene action could be inferred. However, in a couple of cases, the double mutant phenotype was a supenmposition of the phenotypes associated with each of the single mutant parents. Though the additive results suggest that the genes involved affect pathways that proceed independently of each other, I could not rule out the possibility that the single mutants were simply leaky mutations in genes required for regulating the downstream developmental responses.

We had previously shown that plants homozygous for det1-1 and det2-1 mutations have a simple additive phenotype of the single mutant parents (Chory et al., 1991a). Similar results were observed with det1-1 det2-2 double mutants (Chory et al, 1991a). In this study, a det1 -3 det2-1 double mutant was constructed. The double mutant had an additive phenotype in the dark (Fig. 1) and in the light (Tables 1 – 5). The morphological characteristics scored were: hypocotyl length and leaf development in the dark (Table 5) and pigment synthesis (Tables 2, 3), hypocotyl length (Table 5), germination frequencies (Table 4), leaf and bolt number (Table 1), and flowering time in the light (Table 1). Therefore, all det1 det2 double combinations examined to date have a phenotype that is a superimposition of the phenotype associated with the single mutants. This suggests that det1 and det2 affect independent pathways.

Plants homozygous for hy 1 (21.84N) and det1-1 exhibit the dark-grown and light-grown phenotypes of detl-1, including the elevated expression levels of light-regulated nuclear (e.g, rbcS) and chloroplast (e.g., psbA) mRNAs in darkgrown seedlings (Figs 2, 3, 4, Tables 1-6). By the same criteria, det1 is also epistatic to hy2-2 and hy6-1 (Tables 1-6). A puzzling result was obtained when plants homozygous for the less severe hy1 allele (d412) and detl were constructed.

This double mutant essentially looked like det1, except that the double mutant hypocotyl was slightly longer than a detl hypocotyl (3.9 mm versus 2.4 mm in the det 1 single mutant; data not shown). We originally assessed the severity of the hy1(d412) allele by the amount of phytochrome activity measured in dark-grown seedlings, a measurement that mostly quantifies the activity of the type A phytochrome. However, it is likely that the type B phytochrome is responsible for hypocotyl growth inhibition responses, which cannot be assessed spectrally in green plants Therefore, the less severe hy1 alleles may actually be more severe with respect to the activity of the type B phytochrome.

Plants homozygous for hyl (21.84N) and det2-l exhibit the dark-grown and light-grown phenotypes of der2-l, with one exception: the hy1 det2 double mutant had the pale-green phenotype of hyl mutants (Figs 2,3,5, and Tables 1-6). Similar results were observed when det2-l was put in combination with hy2-2 or hy6-l (Tables 1-6). hyl(d412) det2-l strains showed the same phenotype as the above, but the hypocotyl was slightly longer than the hypocotyl from det2 single mutant lines (2 1 mm versus 0.7 mm), consistent with the hyl (d412) det1-1 results just described. Therefore, det2 is partially, but incompletely epistatic to the phytochromedeficient hy mutants.

The double mutant combination of hy3 (Bo64) and detl-l results in plants that look largely like the detl-l single mutant except that the hypocotyl elongates to a greater degree in dark- or light-grown seedlings, and the floral bolt is slightly longer in the double mutant (Figs 4, 6, 7 and Tables 1-6). For most responses, except stem elongation, detl-l is epistatic to hy3 (Bo64). Likewise, the double mutant, hy3 (Bo64) det2-l, has the phenotype of det2-1by every enterion, except that the hypocotyl elongates to a greater degree in the double mutant (Figs 5, 6, 8, and Tables 1-6). det2-1 is therefore also epistatic to hy3.

Plants homozygous for hy4 (2.23N) and det1-1 have the dark-grown de-etiolated phenotype of det1-1 (Figs 4, 9). When grown in high white-light conditions, the hy4 det1 double mutant has the phenotype of det1-1 seedlings, with one exception. hy4 detl double mutant plants are extremely purple dunng the early stages of seedling growth (Figs 10, 11). This is due to a 40-fold increase in anthocyanin accumulation in the double mutants (Table 3). To examine further the molecular basis for aberrant accumulation in the double mutants, I looked at accumulation of chs mRNA in the single and double mutants (Fig. 11). The condensation of 3 acetate units with 4-coumaroyl-CoA to naringenin chaicone by chaicone synthase (chs) is the first committed step in the anthocyanin biosynthetic pathway (Heller and Hahlbrock, 1980). chs is a single copy gene whose expression is regulated by blue light in Arabidopsis (Feinbaum et al., 1991). Dark-grown det1 single mutants accumulate anthocyanins in the dark and light-grown det1 mutants have 5-fold higher levels of anthocyanins than wild type (Chory et al., 1989b, Table 3). hyA single mutants, in contrast, have reduced anthocyanin accumulation (about 2-fold less than wild type) in light-grown seedlings (Table 3). chs mRNA accumulates approximately 0.5 × in the hy4 single mutant and 5 × in the det1 single mutant compared with wild-type seedlings, as might have been predicted by the anthocyanin results (Fig. 11). However, the hy4 det1 double mutant had only about a 5-fold increase in chs mRNA accumulation over wild type and approximately equal amounts to the det1 single mutant parent (Fig. 11). I conclude from these studies that det1 is epistatic to hy4; however, HY4 may play an additional role in anthocyanin biosynthesis.

hyA (2.23N) det2-1 double mutants have the dark- and light-grown phenotypes of det2-1 single mutant plants, including anthocyanin production (Figs. 5, 8, 9, and Tables 1-6). det2-1 is epistatic to hy4 (2.23N).

Plants homozygous for hy5 (Ci88) and det1-1 mutations are only partially de-etiolated when grown in the dark, making fewer leaves than det1-1 single mutants. In addition, darkgrown hy5 det1-1 double mutants have a very elongated hypocotyl, which is similar in length to the hy5 single mutant parent (Fig 12 and Table 5) There was only about half the accumulation of cab, rbcS, and psbA mRNAs in dark-grown det1 hy5 double mutants as in det1 single mutants (Fig. 4, Table 6). However, these mRNA levels were several-fold higher than those of hy5 single mutants or wild type (Fig. 4, Table 6) Light-grown hy5 det1 double mutant plants had a superimposition of the phenotypes of hy5 and det1 single mutants with respect to hypocotyl elongation (Table 5), leaf size (Table 1 and Fig. 13), bolt length (Fig 13), germination frequencies (Table 4), and accumulation of anthocyanins and chlorophylls (Tables 2 and 3)

Dark-grown hy5 (Ci88) det 2-l double homozygotes have a hypocotyl of intermediate length between hy5 and det2 single mutants (Fig. 12, Table 5). Also, anthocyanins accumulate to an intermediate level between det2 and hy5. In the light, the hy5 det2-1 double mutants have an additive phenoltype with respect to bolt number (Table 1, Fig. 8) and flowering time (Table 1). Significantly, hy5 det2 double mutants made a large number of seeds when self-fertilized, unlike the det2 single mutant, which is almost completely male-sterile. Thus, the light and dark phenotypes of both the hy5 det1 and hy5 det2 double mutant combinations were additive.

Phenotypic analysis of double mutant combinations between the hy and det mutants suggests a simple branched pathway (Fig 14). The nearly additive interactions observed between det1-1 and hy5 (Ci88), det2-1 and hy5 (Ci88), and det1 and det2 alleles suggest that these 3 gene products act on distinct signal transduction pathways to affect the downstream lightregulated responses (Fig. 14A). Plants homozygous for hy5 (Ci88) are defective in red-light controlled hypocotyl growth inhibition responses; however, physiological studies with hy5 and double mutant combinations of hy5 with the phytochrome-deficient hyl or hy2 mutants imply that HY5 defines a red-light action pathway that is separate from phytochrome (Koomneef et al., 1980).

An alternative model to explain the double mutant phenotypes is presented in Fig. 14B. In this model, HY5 is a transduction element on a phytochrome action pathway that is influenced by DETL This model takes info account that I have no available means to assess if the mutant alleles used were null. For instance, the hy5 (Ci88) allele may be leaky, a result that would be consistent with the nearly additive interactions that were also observed between hy5 (Ci88) and hy1 (21.84N) or hy5 (Ci88) and hy2 (To76) (Koomneef et al., 1980). The det1 hy5 double homozygote appeared very similar to a wild-type plant when grown in the light, having a much more dramatic phenotype than light-grown det2 hy5 double mutants. This could be due to the fact that det 1 -1 is itself leaky, which is consistent with our ability to isolate dominant extragenic suppressors of det1-1 (A. Pepper and J. Chory, unpublished data). Still, the phenotype of det1 hy5 double mutants is interesting because it points to the opposite roles that DET1 and HY5 appear to play during early development in response to light. To discriminate between the two models, we are currently isolating new mutant alleles of all the genes in question.

In the absence of leaf and chloroplast development, a small, but measurable, induction of cab, chs, and rbcS gene expression can be seen after pulses of red or blue light are applied to etiolated seedlings (e.g., Karlin-Neumann et al., 1988; Feinbaum et al, 1991). This is indicated by a dotted line in Fig. 14A. The intermediates that regulate gene expression by this probably simple signal transduction pathway are unknown, and may possibly be a distinct set of molecules from those defined by hy5, det1 or det2 mutations. We have recently identified a class of trans-acting mutations in Arabidopsis for which cab gene expression is increased in the dark in the absence of de-etiolation (L. Altschmied and J. Chory, unpublished data). These mutations may define such signal transduction components.

Det1 and det2 are epistatic to all strong alleles of hy1, hy2, and hy6. Since det1 and det2 mutants contain wild-type levels of phytochrome activity, I propose an order of gene action that places hy1, hy2, hy3, and hy6 before det2 or det2 on a phytochrome signal transduction pathway. In this model, formation of the active form of phytochrome, P_fr, results in a decrease in activity of either DET1 or DET2 which in turn leads to derepression of the downstream light-regulated morphological and gene expression responses. DET1 appears to play a role in the negative control of all downstream light-regulated responses during growth in the dark (Chory et al., 1989b). Further, DET1 exerts its negative effects spatially on adult light-grown plants since in det1 -mutants there is ectopic expression of chs, rbcS, and cab, and chloroplasts develop ectopically in detl-roots (Chory and Peto, 1990). Det2 is not fully epistatic to the strong hy1, hy2, and hy6 alleles, in that det2 hy mutant combinations are pale yellow-green in color as are the single hy1, hy2, or hy6 parents. This result suggests that DET2 acts on a branch of a pathway downstream from phytochrome that does not affect chlorophyll biosynthesis per se. Further, the prolonged juvenile phase and delayed senescence phenotypes of det1 single mutants implies a temporal negative regulatory role for DET2 during vegetative growth in Arabidopsis (Chory et al., 1991a). In wild-type plants, it is predicted that the activity of DET2 is high prior to light-induced morphogenesis, low during the early stages of chloroplast differentiation and build-up, and relatively high during the later stages of vegetative growth (senescence) when the light developmental program is turned down (Chory, 1991)

hy3 mutations specifically affect accumulation of the type B phytochrome (Nagatani et al, 1991, Somers et al., 1991). hyi mutants have elongated hypocotyls, petioles, and floral bolts, suggesting that B phytochrome deficiencies alone may be responsible for the long hypocotyl phenotype seen in the hy1, hy2, and hy6 mutants. DET 1 and DET2 appear to be on a branched pathway downstream from formation of Pfr(B), however, since we do not have mutants deficient in any one of the other phytochromes, it is not yet clear whether DET1 and DET2 act in a pathway downstream from the other phytochromes as well. Since the temporal and spatial expression patterns for any of the phytochromes or DET1 or DET2 is not known, it is impossible to predict if the individual DETs will act in concert with a particular phytochrome during development.

Det1 and det2 are also epistatic to hy4, suggesting that blue-light signals are involved in decreasing DET1 and DET2 activities (Fig. 14). This is consistent with data from studies of det1 and det2 single mutants which showed that blue-light regulated gene expression and anthocyanin accumulation were derepressed in the dark (Table 3, and Chory et al., 1989b, 1991a). One puzzling observation with regard to anthocyanin accumulation, though, is that det1-1 hy4 (2.23N) double homozygotes had eight-fold higher levels of anthocyanins than det1 single mutants and 40-fold higher levels than wild type (Table 3). chs mRNA and anthocyanin accumulation is 5-fold higher in the det1 mutant than in wild-type seedlings; however, in the det1 hy4 double homozygote, chs mRNA is 5-fold higher than in wild type (similar to det1) (Fig. 11), while anthocyanins are 40-fold increased (Table 3) A possible explanation for these observations is that DET1 is the primary negative regulator of the anthocyanin biosynthetic gene, chs, when DET1 repression of chs is released (det1 background), some other mRNA or gene product in the anthocyanin biosynthetic pathway becomes limiting. Reduced activity, or loss-of-function of, HY4 alleviates this second block. Thus, HY4 might play a negative regulatory role in anthocyanin biosynthesis that is only revealed in the det1 background. Alternatively, since anthocyanins also accumulate during stress conditions in Arabidopsis, a hy4 det1 double mutant may have high anthocyanin levels due to metabolic stress

DET1, DET2, and HY5 may act directly to affect the downstream de-etiolation responses, or they may act through unidentified intermediate regulators, such as hormones, as implied in Fig. 14A. What might this critical regulator be? One suggestion from the literature, (Stetler and Laetsch, 1965; Flores and Tobin, 1986), and from our studies of det1 and det2 mutants is that cytokinin is a critical growth regulator for de-etiolation responses (Chory et al., 1991 b). The similarity of red-light and cytokinin effects was first noted in 1956 (Miller, 1956). Since then, cytokinins have been shown to promote chloroplast development, shoot development and expression of genes for chloroplast-destined proteins in tissue culture cells (e.g, Teyssendier de la Serve et al., 1985; Harvey et al., 1974; Flores and Tobin, 1986). Increased cytokinin levels also correlate with the delayed senescence phenotype of tobacco plants transformed with a gene that increases cytokinin production (Smart et al., 1991) Work in my lab has shown that many of the phenotypes of det 1 and det2 mutants can be mimicked by the addition of the cytokinin, 2-isopentenyl adenine, to the growth medium of wild-type plants (Chory et al., 1991b). Like det1 mutants, dark-grown wild-type seedlings grown in the presence of cytokinins have short hypocotyls, expanded cotyledons and leaves, contain chloroplasts, and have high levels of expression of genes that are normally light-regulated. This intriguing result implies that an increase in available cytokinins in dark-grown seedlings is sufficient to override a light requirement for leaf and chloroplast development and gene expression in Arabidopsis. DET1, DET2, and HY5 may be involved in regulating the activity or availabilty of cytokinins in Arabidopsis (as indicated in Fig 14A). Alternatively, phytochrome, blue-light, and cytokinins may independently modulate the pool size of common intermediates (DET1, DET2, HY5) that directly regulate gene expression.

Clearly, the model proposed here does not define all the elements in the light signal transduction pathways that affect early seedling development. As mentioned above, there are a minimum of five phytochromes defined by DNA sequence homology in Arabidopsis (Sharrock and Quail, 1989, R. Sharrock, personal communication) Though some information is available for the expression pattern of phytochrome A apoprotein in etiolated seedlings, there is little available data on the cell-type specific expression of the different phytochrome genes or proteins in green plants (Komeda et al., 1991). Further, no information is available for the temporal or spatial regulation of any of the five Arabidopsis phytochrome genes or the DET1 and DET2 genes (which are not yet cloned). New mutants that lack a particular phytochrome would help to further define the red-light controlled signal transduction pathways. The blue-light signal transduction pathways are even less well defined However, a new class of photomorphogenetic mutant with long hypocotyls only in blue-light has recently been defined (Liscum and Hangarter, 1991) and mutants with impaired blue-light phototropism have also been described (Khurana et al., 1989; Khurana and Poff, 1989). These two classes of mutants may shed some light on the complexity of blue-light signal transduction pathway(s) in Arabidopsis. Lastly, mutations in DET3, HY7, (J. Chory, unpublished), COP1, (Deng et al., 1991) and the other photomorphogenetic loci need to be integrated into this pathway.

In summary, these studies define just a few steps in what is likely to be a complex regulatory network that involves a large number of gene products that affect early dicotyledonous seedling development in response to red and blue light signals. Clearly, further genetic studies are required to identify mutations that affect only a subset of the downstream light-regulated responses and additional epistasis experiments on extant mutant alleles need to be performed. Lastly, the phenotypic and genetic properties of the various photomorphogenetic genes and the formal nature of their functional interactions may reflect any of a variety of cellular and molecular mechanisms. Resolution of such issues requires that the genes defined by the various mutations be cloned and characterized. Such studies may determine whether the products of these genes interact directly or whether they participate in processes occurring in distinct cellular compartments or at discrete times during development.

I thank Punita Nagpal and Nathan Aguilar for helping with some of the experiments Marc Lieberman and Tom Gatz, photographers at the Salk Institute, took the photographs of light- and dark-grown mutants I am grateful to the members of my laboratory for valuable discussions and for making useful comments concerning this manuscript. This research was funded by grants from the National Science Foundation, the Competitive Grants Program of the USDA, and by The Samuel Roberts Noble Foundation.