A gain-of-function mutation of Arabidopsis cryptochrome1 promotes flowering.

Plants use different classes of photoreceptors to collect information about their light environment. Cryptochromes are blue light photoreceptors that control deetiolation, entrain the circadian clock, and are involved in flowering time control. Here, we describe the cry1-L407F allele of Arabidopsis (Arabidopsis thaliana), which encodes a hypersensitive cryptochrome1 (cry1) protein. Plants carrying the cry1-L407F point mutation have elevated expression of CONSTANS and FLOWERING LOCUS T under short-day conditions, leading to very early flowering. These results demonstrate that not only the well-studied cry2, with an unequivocal role in flowering promotion, but also cry1 can function as an activator of the floral transition. The cry1-L407F mutants are also hypersensitive toward blue, red, and far-red light in hypocotyl growth inhibition. In addition, cry1-L407F seeds are hypersensitive to germination-inducing red light pulses, but the far-red reversibility of this response is not compromised. This demonstrates that the cry1-L407F photoreceptor can increase the sensitivity of phytochrome signaling cascades. Molecular dynamics simulation of wild-type and mutant cry1 proteins indicated that the L407F mutation considerably reduces the structural flexibility of two solvent-exposed regions of the protein, suggesting that the hypersensitivity might result from a reduced entropic penalty of binding events during downstream signal transduction. Other nonmutually exclusive potential reasons for the cry1-L407F gain of function are the location of phenylalanine-407 close to three conserved tryptophans, which could change cry1’s photochemical properties, and stabilization of ATP binding, which could extend the lifetime of the signaling state of cry1.

Light determines the plant's life, because light is the essential energy source for plant metabolism. The spatial, temporal, and spectral variability of light provides cues about the time of day, the season, and the presence of competitors for light. Sensitive and precise light perception, therefore, is essential to properly adjust plant development for maximal photosynthetic efficiency, to correlate vegetative and reproductive growth with favorable seasons, and eventually to maximize fitness. To cope with this task, plants have evolved several types of photoreceptors, including the phytochromes and cryptochromes (for review, see Banerjee and Batschauer, 2005;Josse et al., 2008;Mü ller and Carell, 2009). Phytochromes are red and far-red light receptors and regulate different aspects of plant development, such as hypocotyl elongation in red and far-red light and shade avoidance responses (Franklin et al., 2005). In addition, the two major phytochromes in Arabidopsis (Arabidopsis thaliana), phytochrome A (phyA) and phyB, are involved in flowering time control: phyA promotes flowering under short-day (SD) and long-day (LD) photoperiods (Johnson et al., 1994), while phyB acts as a floral inhibitor (Reed et al., 1993;Mockler et al., 1999).
Cryptochromes are flavoproteins with two chromophores that sense blue and UV-A light in plants (Lin and Todo, 2005). The essential chromophore is a FAD, and the second chromophore is supposed to function in light harvesting and is a pterine (methenyltetrahydrofolate; Mü ller and Carell, 2009). Cryptochromes have an N-terminal photolyase-related (PHR) domain that is similar to photolyases, but they are distinguished from the latter by a variable C-terminal domain. Furthermore, crystallization of the photolyase-like domain of Arabidopsis cryptochrome1 (cry1) has revealed additional structural differences between photolyase and cryptochrome that explain the lack of DNA-binding activity of the cryptochromes (Brautigam et al., 2004). Moreover, the crystal structure confirmed the previ-ously described ATP binding of cry1 (Bouly et al., 2003) by soaking cry1 crystals with the nonhydrolyzable ATP analog adenosine 5'- [b,g-imido]triphosphate (AMP-PNP) and showing AMP-PNP located in the FAD access cavity (Brautigam et al., 2004). Despite the available protein structure, cryptochrome's mode of action still remains to be determined. Interestingly, the C terminus of plant cryptochromes as well as a C-terminal 80-residue motif (NC80) confer constitutive cryptochrome signaling when overexpressed in plants even in the absence of light (Yang et al., 2000;Yu et al., 2007). For plant cryptochromes, it has been proposed that the PHR domain and the C terminus form a "closed" conformation to mask the NC80 motif in the absence of light. Blue light would trigger phosphorylation of the C-terminal tail and its electrostatic repulsion from the surface of the PHR domain to form an "open" conformation, exposing the NC80 motif and initiating signal transduction (Yu et al., 2007).
The Arabidopsis genome harbors two cryptochrome genes: CRY1 and CRY2. A third member of this family, CRY3, belongs to the DASH-type subgroup with repair activity for cyclobutane pyrimidine dimer lesions in single-stranded DNA (Selby and Sancar, 2006) and loop structures of duplex DNA (Pokorny et al., 2008), but with so far unproven photoreceptor function. Mutations in both CRY1 and CRY2 interfere with the inhibition of hypocotyl elongation under blue light conditions (Ahmad and Cashmore, 1993;Guo et al., 1998). Current data suggest that cry1 is the major blue light receptor for seedling photomorphogenesis, while cry2 is more important for the control of flowering time Mockler et al., 1999Mockler et al., , 2003El-Din El-Assal et al., 2003;Endo et al., 2007). Nevertheless, several studies also reported cry1 as a floral regulator: some cry1 mutant alleles conferred late flowering under certain growth conditions (Bagnall et al., 1996;Blázquez et al., 2003), but others did not (Zagotta et al., 1996;Mockler et al., 1999).
Even though the functions of the different photoreceptors are assigned to specific segments of the spectrum of light, physiological and mutant analyses have revealed extensive cross talk between blue and red light photoreceptors (Casal, 2000): phyA and phyB display antagonistic and synergistic effects on the action of each other, depending on which responses are studied (Reed et al., 1994;Casal and Boccalandro, 1995), and several of the phytochromes functionally interact with the cryptochromes (Ahmad and Cashmore, 1997;Casal and Mazzella, 1998;Neff and Chory, 1998;Hennig et al., 1999;Mockler et al., 1999;Devlin and Kay, 2000). In addition, a physical interaction of phyA with cry1 (Ahmad et al., 1998b) and of phyB with cry2 (Más et al., 2000) has been reported. While these cross talks are of minor importance under controlled monochromatic light conditions, they are probably essential for fine-tuning of developmental programs in natural environments.
Here, we provide evidence that a hyperactive cry1 allele confers hypersensitivity not only to blue light but also to red light and strongly shortens flowering time under SD.

Isolation of a New cry1 Allele
For a suppressor screen, seeds of the late-flowering msi1-tap1 transgenic line (Bouveret et al., 2006) were mutagenized with ethyl methane sulfonate (EMS; Exner et al., 2009). Flowering time was scored under LD photoperiods for 1,045 M2 families. Among these, 11 families segregated plants that shortened the vegetative phase of msi1-tap1. For six of them, we confirmed the phenotype in subsequent generations. One of these six contained a mutation in LIKE HETERO-CHROMATIN PROTEIN1 (Exner et al., 2009); the others belong to at least four complementation groups (data not shown). All these mutants still reacted to changes in daylength but exhibited different responses as tested by flowering time experiments under LD and SD (data not shown). The mutant with ID 0.3-457 was chosen for further characterization (Table I).
The mutation in 0.3-457 was localized between markers CER446440 (bacterial artificial chromosome T3H13) and CER460528 (bacterial artificial chromosome T3H13) on the lower arm of chromosome IV. This region contains nine genes including CRY1. Sequencing of the CRY1 locus revealed a C-to-T transition in the third exon, 1,469 bp after the ATG start codon (Fig.  1A), which caused a change of Leu-407 to Phe (Fig. 1B). Therefore, 0.3-457 is henceforth called cry1-L407F.
Comparison of cryptochrome and photolyase sequences from different organisms revealed that Leu-407, which is located in the photolyase-like PHR domain, is conserved among nearly all plant cryptochromes. Among the few exceptions are Arabidopsis, Arabidopsis lyrata and Fragaria vesca (strawberry) CRY2, which all have an Ile instead of the Leu at this position ( Fig. 1B). In contrast, photolyases and animal cryptochromes have amino acids with small side chains (Ala, Ser, Thr, or Gly) at this position but rarely if ever a bulky hydrophobic residue such as Leu. We did not find any cryptochrome or photolyase sequence with a Phe at this position. Furthermore, Leu-407 is located in a block of 12 amino acids that are highly conserved in plant cryptochromes but not in photolyases and animal cryptochromes. A structure of the cry1 PHR domain has been reported (Brautigam et al., 2004), and according to this structure, Leu-407 is buried inside the protein without solvent contact (Fig. 1C).

cry1-L407F Strongly Accelerates Flowering
The cry1-L407F mutant did not only flower earlier in the msi1-tap1 background but also when backcrossed into the Columbia (Col) wild type both under LD and SD conditions (Fig. 2, A and B). Under LD, the acceleration of flowering in cry1-L407F was caused by a shortened adult phase (0.8 versus 2.6 leaves), while the duration of the juvenile phase (5.3 leaves) was not affected. The early-flowering phenotype was even more dramatic under SD. While wild-type plants produced 61 leaves before bolting, cry1-L407F produced only 12 leaves. Tests for genetic interaction between cry1-L407F and msi1-tap1 revealed additivity of both mutants (Supplemental Fig. S1), suggesting that CRY1 and MSI1 function in separate genetic pathways in the control of flowering. Besides the early-flowering phenotype, cry1-L407F featured a small and compact rosette (Supplemental Fig. S1), which is partly caused by shortened petioles (data not shown). Such a reduction of petiole length has also been described for plants overexpressing cry1 (Lin et al., 1996).
These results, together with the observation that cry1-L407F behaved semidominantly (data not shown), suggested that we identified a gain-of-function CRY1 allele. Because genetic complementation tests and transgenic complementation are difficult with hypermorphic Cryptochrome and photolyase protein sequences of several organisms were aligned, and a segment of the generated sequence alignment is shown; numbers at the end of each line indicate the amino acid position within the respective protein. The arrowhead marks the Leu that is exchanged for a Phe in CRY1-L407F. Note that this Leu is usually conserved in plant cryptochromes but not in photolyases. The arrows indicate conserved Trp residues involved in electron transfer to FAD. The gray box marks a plant cryptochrome-specific 12-amino acid motif that includes Leu-407. C, Left, x-ray structure of the complex of cry1, FAD, and ATP taken from Brautigam et al. (2004). Conserved Trp residues are in black, and Leu-407 is in magenta. The red, yellow, and orange sequence regions (labeled 1, 2, and 3, respectively) correspond to the peaks in the RMSF plot (see text). Right, structural model of cry1-L407F. The mutation L407F is in magenta. The first 20 residues, for which the reduction of flexibility caused by the L407F mutation is highest, are in yellow. and neomorphic alleles, the ability of cry1-L407F to accelerate flowering was tested by transgenic phenocopy experiments. When wild-type cry1 was introduced and overexpressed in Col wild-type plants, the four tested transgenic lines all flowered as late as or even slightly later than the Col wild type under SD (Fig.  2C). Similarly, Lin et al. (1996) had previously reported that increased cry1 dosage caused slightly delayed flowering. In contrast, when the cry1-L407F mutant gene under the control of the same cauliflower mosaic virus 35S promoter was introduced into Col, the four tested transgenic lines all flowered earlier than the Col wild type under SD (Fig. 2C).
Together, these results show that cry1-L407F is a gain-of-function CRY1 allele that strongly accelerates flowering.

cry1-L407F Causes FLOWERING LOCUS T Overexpression
Cryptochromes can affect flowering in at least three different ways. First, cryptochromes control the phase of the circadian clock, which in turn controls diurnal expression patterns of CONSTANS (CO). Second, cryptochromes stabilize CO protein in the light, and CO then activates flowering by inducing the expression of FLOWERING LOCUS T (FT), which encodes the mobile flowering signal FT (for review, see Kobayashi and Weigel, 2007). Third, cry2 can directly induce FT expression (Liu et al., 2008a). To elucidate whether one of these mechanisms is involved in the accelerated flowering of cry1-L407F, we measured the gene expression of EARLY FLOWERING4 (ELF4), GIGANTEA (GI), CO, and FT under SD conditions. ELF4 and GI participate in signaling from the circadian clock to downstream processes such as CO expression (Park et al., 1999;Doyle et al., 2002). In SD, CO is usually expressed only after the end of the light phase, and because CO protein is rapidly degraded in the dark, FT remains inactive. We found that ELF4 expression was not significantly affected in cry1-L407F, maintaining its typical evening peak (Fig. 3), suggesting that the accelerated flowering was not caused by a malfunction of the circadian clock. Likewise, GI expression was very similar between the wild type and cry1-L407F (Fig. 3). In contrast, expression of CO and FT was considerably increased in cry1-L407F (Fig. 3). However, FT expression was strongest during the light period (zeitgeber time = 3 h), while CO expression was strongest during the dark period (zeitgeber time = 14 h), suggesting that cry1-L407F does not establish aberrant CO protein stabilization in the dark.
Together, these results indicate that cry1-L407F causes the untimely expression of CO during the light period and thus allows for the accumulation of CO under SD photoperiods. Increased CO levels then strongly activate FT and cause the very early flowering of wild-type plants and the suppression of msi1-tap1's late-flowering phenotype.

cry1-L407F Causes Hypersensitivity toward Blue and Red Light
The cry1 photoreceptor is known to control hypocotyl growth in response to blue light (Koornneef et al., 1980;Ahmad and Cashmore, 1993). To investigate the effects of the amino acid substitution on the function of cry1-L407F in further detail, seedlings were grown under different fluence rates of blue, red, and far-red light and hypocotyl elongation was measured. Under blue light, inhibition of hypocotyl elongation was observed under much lower fluence rates in cry1-L407F than in the wild type (Fig. 4). Thus, cry1-L407F is a hypersensitive blue light photoreceptor, and cry1-L407F is a hypermorphic allele. The hypersensitivity of cry1-L407F toward blue light is a dominant trait. Heterozygous cry1-L407F seedlings uniformly displayed short hypocotyls when grown under low fluence rates of blue light (data not shown). Surprisingly, cry1-L407F seedlings were not only strongly hypersensitive to blue light but also to red light (Fig. 4), which should not efficiently activate cry1 (Lin et al., 1995;Ahmad et al., 2002). It is unlikely that this phenotype was caused by a "contamination" of the red light by photons from the blue part of the spectrum, because there is extremely little if any blue light emitted by the light source used in these experiments (Supplemental Fig. S2). Similar to the situation under red light, cry1-L407F seedlings were also hypersensitive to far-red light, which is believed to be predominantly sensed by phyA.
Repression of hypocotyl elongation under blue light is a normal function of wild-type Arabidopsis cry1 and cry2 (Koornneef et al., 1980;Ahmad and Cashmore, 1993;Lin et al., 1998). In order to test whether cry1-L407F does also affect the red light sensitivity of a process normally not controlled by cryptochromes, we compared light-dependent germination of the wild type and cry1-L407F. It is commonly thought that germination of Arabidopsis seeds is exclusively controlled by phytochromes but not by cryptochromes (Shinomura et al., 1996;Oh et al., 2007). Under continuous white light, both the wild type and cry1-L407F show a similarly high frequency of germination, while in the dark or after a far-red pulse, almost no germination occurred (Fig. 5). Pulses of red or white light that caused a germination rate of about 30% in wildtype seeds caused a germination rate of 80% in cry1-L407F seeds. Thus, cry1-L407F strongly increased the sensitivity of red light-induced germination. Red lightinduced germination is usually a function of phyB. However, the results shown in Figure 5 could indicate that cry1-L407F itself can induce germination. Because photoreversibility is a hallmark of phyB function, we tested whether the red light-induced induction of germination could be reverted by a pulse of far-red light. Indeed, a far-red light pulse could completely prevent red light-induced germination in both the wild type and cry1-L407F (Fig. 5). This strongly suggests that cry1-L407F can increase the sensitivity to red light or signaling of photoreversible phytochromes such as phyB in the low fluence response.
The development of cry1-L407F seedlings differed from the wild type not only in the light but also in the dark. For up to 3 d of growth in the dark, hypocotyls were of similar length and cotyledons were folded into a protective hook in Col and cry1-L407F. In cry1-L407F, cotyledons started to unfold from day 2 on, and hypocotyl elongation was slightly reduced from day 3 on (Fig. 4 S3). In addition, hypocotyls of dark-grown cry1-L407F were often bent while hypocotyls of the wild type grew straight (Supplemental Table S1; Supplemental  Fig. S3). The hypocotyl-bending phenotype was temperature dependent: growth at 21°C led to strong bending in most of the seedlings, but growth at 26°C led to only mild bending (Supplemental Fig. S3). Thus, cry1-L407F can at least partially function even without light activation. Nevertheless, no strong constitutive activation of photomorphogenesis in the dark, such as was observed in plants overexpressing the CCT domain (Yang et al., 2000), was observed in cry1-L407F. For Relative expression values are shown as mean 6 SE (n $ 4). White and gray bars on top of the diagrams represent periods of light and darkness, respectively. Values were normalized to a PP2A gene (At1g13320). WT, Wild type.
example, there was no induction of cell division in the shoot apical meristem in the dark (data not shown).
Phytochromes regulate the transcription of many genes; EARLY LIGHT INDUCED2 (ELIP2), for instance, is rapidly up-regulated after exposure to red or far-red light, and this up-regulation requires phyB and phyA, respectively (Harari-Steinberg et al., 2001). We tested whether the cry1-L407F mutation would affect the red light-induced expression of ELIP2. In darkness and during the investigated time course up to 6 h after transfer to red light, ELIP2 expression was consistently higher in cry1-L407F than in the wild type, but with very similar kinetics for both genotypes (Fig.   6). Thus, the effect of cry1-L407F on red light signaling is not restricted to germination and hypocotyl growth inhibition.
Together, these results show that cry1-L407F causes strong hypersensitivity to both blue and red light in a dominant manner.
The Hypersensitivity of cry1-L407F Is Not Caused by Elevated cry1 Levels Overexpression of CRY1 under the control of the constitutive 35S promoter results in hypersensitivity of the transgenic plants toward blue light (Lin et al.,Figure 4. Inhibition of hypocotyl elongation in cry1-L407F mutant seedlings is hypersensitive to blue, red, and far-red light. Fluence rate response curves of hypocotyl growth inhibition under continuous light treatments are shown. Diagrams show means 6 SE of three replicates with at least 15 seedlings analyzed in each experiment. Black circles, the wild type (WT); white circles, cry1-L407F. 1996). Thus, we reasoned that the observed hypersensitivity of the cry1-L407F mutant could be caused by increased cry1 protein levels, although flowering time was not affected in overexpressors of wild-type cry1 (Fig. 2). Quantitative immunoblots, however, revealed unchanged cry1 levels in cry1-L407F mutants ( Fig. 7; data not shown). These results demonstrate that the blue and red light hypersensitivity and early flowering of cry1-L407F are not caused by increased expression of cry1 but most likely by increased activity of the cry1-L407F protein. This conclusion is supported by the observation that blue fluence rates of up to 50 mmol m 22 s 21 caused a stronger shift of the cry1 band in the cry1-L497F mutant than in the wild type (Fig. 7). This blue light-induced shift reflects phosphorylation of the cry1 protein that is associated with photoreceptor activation (Shalitin et al., 2002(Shalitin et al., , 2003Bouly et al., 2003). We thus conclude that the L407F mutation in cry1 increases the fraction of active (phosphorylated) photoreceptor over a broad range of blue fluence rates but has no effect in darkness, as seen from the absence of a shifted cry1 band in both the wild type and the cry1-L407F mutant (Fig. 7).

The Hypersensitivity of cry1-L407F Could Be Caused by Reduced Structural Flexibility of the Photoreceptor
To understand the potential consequences of the L407F mutation on cry1 structure and function, we carried out three independent molecular dynamics simulations for each of the four following systems: wild-type protein and the L407F mutant, both with ATP and without ATP. The time series of root mean square deviation (RMSD) are useful to visualize the spatial deviation of the structure during the simulation with respect to the energy-minimized x-ray conformation (Supplemental Fig. S4). RMSD plots of the Ca atoms show that wild-type and mutant protein are  . cry1-L407F is stronger phosphorylated in blue light than in wild-type cry1. Immunoblot analysis of wild-type cry1 (cry1-WT) and cry1-L407F protein levels (both in the msi1-tap1 background). Seedlings were grown for 96 h in complete darkness before being transferred to monochromatic blue light (l max = 471 nm) of given fluence rates. A, Representative immunoblot of samples kept in darkness for 96 h (0), then irradiated with blue light for 30 or 120 min, or kept in darkness for another 120 min (120d). The blot shows samples from irradiation with 25 mmol m 22 s 21 blue light. The cry1 and tubulin signals are indicated with arrows. Note the shifted cry1 bands appearing in the light-treated samples that correspond to phosphorylated forms of cry1. B, Ratios of shifted (phosphorylated) cry1 to the total amount of cry1 in the respective genotypes. Seedlings were irradiated with the given fluence rates of blue light for 30 min. C, Same as B but seedlings were treated for 120 min with blue light. Quantification of the bands was done with the LI-COR Odyssey infrared imaging system and software in the linear range of cry1 and tubulin signals. stable in all the simulated systems (RMSD = 2-3 Å ). Generally, simulated RMSD were smaller for proteins with ATP than for proteins without ATP (Supplemental Fig. S4A). Interestingly, the crystallographic ATPbinding mode was unstable in the simulations of both wild-type and mutant proteins (Supplemental Fig. S5). During the simulations of the wild-type and mutated proteins, the distance between the N6 atom of ATP and the Cg of Asp-409 was higher than the crystallographic distance of 3.5 Å (Supplemental Fig. S4B, top), indicating that this interaction is not maintained. Similarly, the RMSD plots of the adenine moiety of ATP showed instability of the ATP-binding mode in the wild-type and mutant proteins (Supplemental Fig. S4B, bottom; RMSD . 5 Å ). To identify potential regions of altered structural flexibility, we plotted root mean square fluctuation (RMSF) values of Ca atoms, which illustrate structural plasticity along the protein sequence. The RMSF plots of the wild-type protein show that ATP reduces the atomic fluctuations of three segments (Fig. 8A, top). The segments 1, 2, and 3 are spatially close to FAD and the adenine moiety of ATP, respectively ( Fig. 1, left) but distant in sequence. A qualitative representation of the backbone flexibility where the thickness of the backbone is proportional to the RMSF is shown in Figure 8B. The RMSF plots of the mutant protein show that ATP reduces the atomic fluctuations only of segment 1, because the L407F mutation already has a stabilizing influence on segments 2 and 3 (Fig. 8A). To assess the effect of the mutation on the protein backbone flexibility, the residue-wise RMSF difference between the wild-type and mutant simulations without ATP was calculated (Fig. 8A,  bottom). The 20 residues affected by the highest flexibility reduction include the mutation site; they are located in segments 2 and 3 and are close to the three conserved Trp residues (Fig. 1C, right).
Together, the simulations indicate that ATP binding stabilizes three regions of wild-type cry1 and that the L407F mutation partially mimics the effect of ATP binding by stabilizing two of three ATP bindingresponsive regions even in the absence of ATP. Therefore, the differences in flexibility suggest that the L407F mutation reduces the conformational entropy penalty of ATP binding and thus might promote ATP binding, autophosphorylation, and eventually cry1 signaling.

DISCUSSION cry1 Controls Flowering Time
Blue light promotes flowering , and this effect was attributed mainly to cry2, as cry2 mutations delay flowering under LD conditions in a phyB-dependent manner (Koornneef et al., 1991;Guo et al., 1998Guo et al., , 1999El-Din El-Assal et al., 2001Endo et al., 2007;Liu et al., 2008b). Some studies had reported that cry1 functions in flowering time regulation as well, but others failed to find such evidence (Bagnall et al., 1996;Zagotta et al., 1996;Mockler et al., 1999;Blázquez et al., 2003). While cry2 mainly affects flowering time under LD conditions, the effects on flowering time reported for certain cry1 alleles and the C-terminal domain of cry1 were prominent under SD conditions (Bagnall et al., 1996;Yang et al., 2000). Since both cry1 and cry2 are involved in blue light-mediated repression of hypocotyl elongation (Ahmad and Cashmore, 1993;Lin et al., 1998), it is also possible that both act to some extent in flowering time control but that the effect of cry1 on flowering time control is often masked by other floral regulators. In fact, it was reported that cry1 cry2 double mutants flower significantly later than cry2 single mutants when grown under monochromatic blue light . Here, we describe the new CRY1 allele cry1-L407F, which supports previous findings showing that cry1 can act as a positive regulator of the floral transition.
The cry1-L407F allele was isolated in a mutant screen for suppression of the late-flowering phenotype of msi1-tap1 plants. Because cry1-L407F and msi1-tap1 showed an additive genetic interaction (Supplemental Fig. S1), MSI1 and CRY1 possibly function in independent genetic pathways of flower induction.

cry1-L407F Causes Increased Expression of CO and FT
The cry1-L407F photoreceptor caused increased expression of FT (Fig. 3), and this can explain the observed early-flowering phenotype. FT plays only a minor role in flowering time regulation under SD, but high levels of FT expression cause early flowering even in SD. FT expression is affected by light in several ways. First, light controls the phase of the circadian clock to establish the correct diurnal expression of CO. Second, light stabilizes CO and allows the accumulation of CO specifically under LD, when expression peak and light coincide. This coincidence of the diurnal CO expression peak and external light stimulus is the main mechanism of photoperiodic acceleration of flowering by LD in Arabidopsis (Kobayashi and Weigel, 2007). Cryptochromes appear to play a dominant role in this process because they were proposed to be needed for CO degradation by CONSTITUTIVE PHOTOMORPHOGENIC1 (COP1) in the dark under SD (Jang et al., 2008;Liu et al., 2008b). Third, cry2 can also directly induce FT expression (Liu et al., 2008a).
Because increased FT expression in cry1-L407F remained constrained to the light period, it is unlikely that stabilization of CO by cry1-L407F in the dark caused the increased FT expression. In contrast, loss of COP1 leads to the stabilization of CO in the dark (Jang et al., 2008;Liu et al., 2008b) and to an FT transcript peak in the middle of the night (Jang et al., 2008). In cry1-L407F, FT expression peaked during the light period and was similarly low as in the wild type during the dark period, suggesting that cry1-L407F does not interfere with the degradation of CO protein by the COP1 pathway. In cry1-L407F, the increased FT expression during the light period is probably caused by the expression of CO during the light period. It is not clear how cry1-L407F caused increased CO expression during the light period. An indirect effect via altered function of the circadian clock seems unlikely given the unchanged expression of ELF4 and GI (Fig.  3). It is conceivable, however, that cry1-L407F directly affects CO expression, a possibility we are currently testing.

A Conserved Leu Is Important for cry1 Function
Cryptochromes are flavoproteins with two chromophores and high sequence similarity to photolyases. However, cryptochromes lack several of the characteristics of the DNA-repairing photolyases, most prominently binding to DNA, which is explained by a negative electrostatic potential of the surface around the flavinbinding pocket of DNA-photolyase (Brautigam et al., 2004;Mees et al., 2004). In addition to the N-terminal photolyase-like PHR domain, cryptochromes contain a characteristic C-terminal domain, termed CCT, which is not present in the photolyases. Expression of the CCT domain in transgenic Arabidopsis led to constitutive photomorphogenesis and mimicked the phenotype of mutations in COP1 (Yang et al., 2000). COP1 is involved in the regulation of hypocotyl elongation, anthocyanin production, and chloroplast development and binds to cry1 and cry2 via their CCT domains independent of light (Yang et al., 2000;Wang et al., 2001). It is possible that normally the CCT domain is kept inactive by an interaction with the PHR domain. Absorption of light would then cause conformational changes of the PHR domain, leading to release of the CCT domain, which could eventually activate the signaling chain (for review, see Lin and Todo, 2005). Because cry1 functions at least partially by affecting the ability of COP1 to ubiquitinate target proteins, which will then be degraded by the proteasome (Wang et al., 2001;Yang et al., 2001;Jang et al., 2008;Liu et al., 2008b;Kang et al., 2009), it is possible that cry1-L407F attenuates COP1 activity. This would lead to increased accumulation of HY5, causing light hypersensitivity of seedlings.
In addition to increased light sensitivity, the cry1-L407F mutants have some defects in skotomorphogenesis: cotyledons partially unfold, hypocotyls elongate less, and the light-induced ELIP2 gene has a slightly increased basal expression in extended darkness. It is possible that the observed hypocotyl bending of dark-grown cry1-L407F seedlings was caused by an autonomous activation of phototropism. Notably, Figure 8. The L407F mutation reduces the structural flexibility of cry1. A, Comparison of backbone flexibility. Each curve is the average of the RMSF calculated over three trajectories. The segments corresponding to the peaks labeled 1, 2, and 3 are shown in Figure 1. WT, Wild type. B, The tube-like rendering of the backbone flexibility of the wild type (left) and cry1-L407F (right) was generated using values from A.
light-activated wild-type cryptochrome has been implicated in phototropism (Ahmad et al., 1998a;Lascève et al., 1999;Kang et al., 2008;Tsuchida-Mayama et al., 2010). Because the hypocotyl bending of cry1-L407F mutants was much weaker at elevated growth temperatures, it is possible that the interaction with a cry1 signaling partner in the dark was enabled by the L407F mutation at 21°C but was not stable enough to remain effective at 26°C. Because cry1-L407F mutants differ from the wild type also under light conditions that do not activate cry1 (i.e. darkness and red and far-red light), cry1-L407F can partially function without the blue light requirement that is typical for wild-type cryptochromes. Thus, Leu-407 of cry1 seems to be essential to lock the photoreceptor in an inactive form and to prevent precocious activation of signaling cascades.
Leu-407 is located in a region of the protein that is conserved in all plant crytochromes but not in photolyases or animal cryptochromes (Fig. 1). Despite the strong conservation of Leu-407, it is not immediately obvious why the change to Phe, which is of similar size and hydrophobicity like Leu, would increase the light sensitivity of cry1. The mutated Leu-407 is close to the phosphate residues of AMP-PNP, sticking out of the flavin-binding pocket in the cocrystal structure of cry1 with AMP-PNP (Brautigam et al., 2004). To test whether the L407F mutation could modulate the ATP binding to cry1, molecular dynamics simulations of cry1 wild type and the L407F mutant in complex with FAD and with or without ATP were run. ATP binding reduced the Ca flexibility of three sequence segments, which are distant in sequence (segments 1 and 2, more than 50 amino acids; segments 2 and 3, more than 100 amino acids) but close in space. In contrast to the wild type, ATP binding did not reduce the flexibility of segments 2 and 3 in the L407F mutant because their flexibility is already diminished by the single point mutation.
To explain the hyperactivity of the cry1-L407F mutant, four not mutually exclusive hypotheses can be formulated. First, the novel Phe of the mutant is close to three conserved Trp residues, which are involved in electron transport from the surface to the FAD at least in vitro, as extensively studied in Escherichia coli photolyase, and considered to have the same function in plant cryptochromes (Park et al., 1995;Giovani et al., 2003;Banerjee et al., 2007). Thus, it is possible that cry1-L407F has altered photochemical properties. Second, partial prestabilization of the ATP-binding pocket in cry1-L407F could stabilize ATP binding, extending the lifetime of the signaling state of cry1. This conclusion is at least consistent with the increased level of shifted and phosphorylated cry1-L407F compared with wild-type cry1 (Fig. 7). Third, the reduced flexibility of cry1-L407F could favor binding to a signaling partner, because of a reduced conformational entropy penalty upon binding. Such an effect was recently observed for a mutant of a PDZ domain (Petit et al., 2009). Fourth, it is possible that conformational changes at the surface of cry1 induced by the L407F mutation altered cry1 binding specificity and led to the activation of signaling events that normally are not under cry1 control. The increased germination frequency of cry1-L407F mutants, for instance, could be explained by cry1-L407F having acquired a novel function in the control of germination, which is usually restricted to phytochromes. Discrimination between these possibilities will be addressed in future studies.
It is possible that the L407F exchange affects binding to interacting proteins. The fact that Leu-407 is in the N-terminal domain of cry1 suggests that partner recognition occurs, at least in part, via the N-terminal domain of cry and not exclusively via the C-terminal tail. The L407F exchange may be the first identified mutation that alters the downstream target specificity or affinity of cry.

Hypersensitivity of cry1-L407F to Various Light Qualities Reveals Tight Integration of Several Light Signaling Pathways
The cry1-L407F allele was not only hypersensitive to blue light but also to red and far-red light. Normally, red and far red light are not sensed by cryptochromes but by phytochromes. This raises the question of which photoreceptor is then responsible for the increased sensitivity to red light. Because the hypersensitivity of cry1-L407F to red pulses could be fully reverted by far-red pulses, at least in this case it was clearly phytochrome signaling that was affected in the cry1-L407F mutant. Interactions of red light-absorbing phytochrome and blue light-absorbing cryptochrome signaling cascades have been reported (Casal and Boccalandro, 1995;Ahmad and Cashmore, 1997;Hennig et al., 1999). Furthermore, nuclear import of phyB was initiated by blue light, but not by light of 695 nm, which establishes a similar phytochrome photoequilibrium as blue light (Gil et al., 2000). Finally, cryptochromes were found to be required for phytochrome signaling to the circadian clock . On the molecular level, these effects could potentially be based on an interaction of the cry1 C-terminal domain with phyA (Ahmad et al., 1998b) or of cry2 with phyB (Más et al., 2000). Further work will establish whether the L407F mutation in cry1 affects direct interactions with phytochrome or whether the photochemical properties of cry1-L407F are changed. We propose that the increased sensitivity of cry1-L407F plants to blue, red, and far-red light reveals the intimate cross talk between cryptochrome and phytochrome light signaling cascades, which has been suggested to be important for concerted plant development under natural light conditions (Casal, 2000).

Plant Material and Growth Conditions
Seeds of Col and Landsberg erecta Arabidopsis (Arabidopsis thaliana) wildtype accessions were obtained from the Nottingham Arabidopsis Seed Stock Centre. The line msi1-tap1 (accession Col) has been described before (Bouveret et al., 2006). The EMS allele cry1-L407F (accession Col) was isolated from a mutant screen (this study). To construct plants that ectopically overexpress cry1 or cry1-L407F (35S::CRY1 and 35S::CRY1-L407F), the full-length coding sequences were inserted into the binary destination vector pK7WG2 (Karimi et al., 2002) downstream of the cauliflower mosaic virus 35S promoter. Constructs were transformed into Col wild-type plants.
Seeds were usually germinated on sterile basal salts Murashige and Skoog medium (Duchefa) after 2 or 3 d of stratification treatment of the imbibed seeds at 4°C, and plants were analyzed on plates or transferred to soil (Einheitserde; H. Gilgen Optima-Werke) 10 d after germination. Alternatively, seeds were directly sown on soil. Plants were kept in Conviron growth chambers with mixed cold fluorescent and incandescent light ( For immunoblot analyses, seeds were plated on half-strength Murashige and Skoog plates and stratified at 4°C for 4 d in darkness. Germination was induced by white light illumination for 4 h. Plants were grown at 22°C for 4 d, and seedlings were harvested after treatment with blue light emitted from light-emitting diodes (l max = 471 nm; CLF Plant Climatics) for 30 or 120 min with the fluence rates indicated and measured with a P-2000 optometer (Gigahertz-Optik).

Flowering Time Analysis
Flowering time was scored as the length of time between the end of stratification and the development of a primary shoot of 5 mm height (=bolting). The number of rosette leaves was determined at bolting. For phase transition, all formed rosette leaves were inspected for the presence of abaxial trichomes at bolting.

RNA Isolation, Reverse Transcription-PCR, and Quantitative PCR
RNA was extracted from plant tissue as described previously (Hennig et al., 2003). For reverse transcription (RT)-PCR analysis, 1 mg of total RNA was treated with DNase I (Promega). The DNA-free RNA was reverse transcribed using a RevertAid First-Strand cDNA Synthesis Kit (Fermentas) according to the manufacturer's instructions. For quantitative PCR analysis, the Universal ProbeLibrary system (Roche Diagnostics) was used on a 7500 Fast Real-Time PCR instrument (Applied Biosystems). Quantitative PCR was performed with three replicates, and the results were analyzed as described (Exner et al., 2009). Details of the assays used are given in Supplemental  Table S2.

Analysis of Hypocotyl Length
Seeds were plated on two layers of water-soaked 3MM chromatography paper (Whatman Schleicher & Schuell), which were placed into clear plastic boxes. A 48-to 96-h dark treatment at 4°C was followed by induction of germination by white light for 10 h (24 h for far-red studies) at 23°C and further incubation of the seedlings under specific light conditions, which were as follows: blue light, Philips TLD 18W/18 Blue E003, continuous light, 21°C; red light, Philips TLD 18W/18 Red, continuous light, 21°C; far-red light, as described (Sperling et al., 1997), continuous light, 26°C. The hypocotyl length was measured by spreading the seedlings on millimeter paper and reading the length.

Quantitative Immunoblots
Per sample, approximately 50 seedlings were collected, frozen in liquid nitrogen, and ground to a fine powder with a cell mill (MM200; Retsch). Protein was extracted by TCA-acetone precipitation according to Shultz et al. (2005) with the following modifications: after the washing steps, samples were dried in a SpeedVac and then dissolved in SDS sample buffer (45 mM Tris-HCl, pH 6.8, 10% glycerol, 1% SDS, 0.01% bromphenol blue, and 50 mM dithiothreitol). Samples were incubated at 95°C for 10 min followed by a centrifugation step (10 min, 20,000g) to remove cell debris. For SDS-PAGE, 15 mg of total protein was loaded per lane on 10% SDS minigels (Shultz et al., 2005). PageRuler (Fermentas) was used as marker. Separated proteins were transferred to nitrocellulose membranes (porablot NCP; Macherey-Nagel). Membranes were blocked with 7% milk powder in Tris-buffered saline (TBS; 20 mM Tris-HCl, pH 7.5, and 150 mM NaCl). Incubation with the two primary antibodies was done step-wise with monoclonal antibody against a-tubulin (anti-a-tubulin, produced in mouse; clone B-5-1-2; Sigma) and then with anti-cry1 antibody (raised in rabbits and provided by M. Ahmad, Universite Paris VI). Both antibodies were used in a 1:2,000 dilution in TBS-T (TBS with 0.1% [v/v] Tween 20). Fluorescence-labeled secondary antibodies against rabbit (donkey anti-rabbit IRDye800CW; LI-COR Biosciences) and mouse (donkey anti-mouse IRDye 700DX; Rockland) were incubated simultaneously for 1 h (each diluted 1:10,000 in TBS-T). The membranes were scanned and analyzed with the LI-COR Odyssey Infrared Imaging System. Bands detected in the 700-nm channel correspond to a-tubulin, and bands detected in the 800-nm channel correspond to cry1. The system was calibrated to ensure measurements in the linear range for both a-tubulin and cry1. The cry1 signal was normalized against the a-tubulin signal. In addition, the percentage of shifted cry1 bands compared with the total cry1 signal was determined.

Sequence Alignment
The Arabidopsis CRY1 (At4g08920), CRY2 (At1g04400), and photolyase (At3g15620) protein sequences were obtained from The Arabidopsis Information Resource (http://www.arabidopsis.org/) and blasted against the nonredundant protein databases at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=Search&db= pubmed) and at the Department of Energy Joint Genome Institute (http:// genome.jgi-psf.org/). The obtained sequences were aligned using ClustalX 2.0. The identifiers of the protein sequences included in this analysis are listed in Supplemental Table S3. The nomenclature of phytochrome as well as cryptochrome photoreceptors and their genes is according to Quail et al. (1994).

Molecular Dynamics Simulations
The crystal structure of the PHR domain of Arabidopsis cry1 with AMP-PNP bound (Protein Data Bank accession code 1U3D) was used for modeling and molecular dynamics simulations. The L407F mutation was introduced with PyMOL (The PyMOL Molecular Graphics System, version 1.2r1; Schrödinger), and the most common rotamer was selected. To generate ATP from AMP-PNP, the nitrogen atom between phosphate groups of AMP-PNP was replaced by oxygen. Ions and crystallization water were kept for further calculations. All the simulations were carried out using CHARMM version c35b2 (Brooks et al., 1983(Brooks et al., , 2009 and the PARAM22 force field (Mackerell et al., 1998(Mackerell et al., , 2004) with the TIP3P model of water (Jorgensen et al., 1983;Mackerell et al., 1998). To effectively compare simulations with experiments, pH 7.4 was considered. The side chains of Asp and Glu residues were negatively charged, those of Lys and Arg residues were positively charged, His residues were considered neutral, the N terminus was positively charged, and the C terminus was negatively charged. First, structures were minimized in vacuo using a dielectric constant « = 4r (where r is the distance in Å between atoms/ partial charges) to an energy gradient of 0.01 kcal mol 21 Å 21 . The minimized protein was then inserted into a water box, where each atom of the protein had a distance of at least 14 Å from the boundary. Water molecules within 2.8 Å from any atom of the protein were removed. Chloride and sodium ions were added to neutralize the total charge of the system at a concentration of 200 mM. The final system consisted of around 96,000 atoms, approximately 7,900 of which belong to the solute. To avoid finite-size effects, periodic boundary conditions were applied. Long-range electrostatic effects were taken into account by the Particle Mesh Ewald summation method (Darden et al., 1993). The temperature was kept constant at 300 K by the Nosé-Hoover thermostat (Nose, 1984;Hoover, 1985), while the pressure was held constant at 1 atm by applying the Langevin piston pressostat. Lookup tables (Nilsson, 2009) for the calculation of water-water nonbonded interactions (van der Waals and Coulomb) were used to increase efficiency. SHAKE was applied to the hydrogens, allowing an integration step of 2 fs. Different initial random velocities were assigned to every simulation. Four systems were simulated: the wild-type protein with only FAD bound; the wild-type protein with FAD and ATP bound; the mutant with only FAD bound; and the mutant with FAD and ATP bound. Three independent 30-ns-long simulations were carried out for each system.

Trajectory Analyses
RMSD and RMSF were calculated with CHARMM, and their formulae are as follows: where N is the number of atoms; x i , y i , and z i are the coordinates of the atom i after best superposition on a reference structure; and x i,ref , y i,ref , and z i,ref are the coordinates of the atom i in the reference structure. The coordinates x i , y i , and z i refer to the average structure; N f is the number of frames in the trajectory segment analyzed for RMSF calculations; the coordinates x i,ave , y i,ave , and z i,ave refer to the average structure. The reference structure for RMSD analyses was the starting structure used in the dynamics (i.e. the energyminimized x-ray structure). The average structures and RMSF were calculated along 2-ns segments of trajectory, skipping the first 2 ns and the last incomplete segment shorter than 2 ns. For the first 30 ns of simulation time, 13 values of RMSF were calculated and then averaged. RMSD expresses how different an object is with respect to another after the best superposition of the two. A RMSD value of zero means perfect superposition. RMSF is a measure of atomic flexibility and can be related to the crystallographic B-factor, B = 8p 2 /3(RMSF) 2 . The distance between the N6 atom of ATP and the Cg of Asp-409 was calculated with the program Wordom (Seeber et al., 2007). Structures were plotted with PyMOL.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S2. Emission spectrum of the red light source.
Supplemental Figure S3. Phenotype of cry1-L407F seedlings in the dark.
Supplemental Figure S4. Structural stability of the PHR domain of cry1 and stability of the ATP-binding mode.
Supplemental Figure S5. Binding mode displacement of ATP.
Supplemental Table S1. Phenotype of cry1-L407F seedlings in the dark.
Supplemental Table S2. Primers used for quantitative RT-PCR.
Supplemental Table S3. Protein sequences used for the alignment.