In plant cryptochromes cry1 and cry2, the dark state of the flavin is in the oxidized form in vivo (B). To determine the nature of the dark state of flavin in animal-type cryptochromes, we measured a classic action spectrum for Dmcry activity in living flies. An action spectrum is a dose-response curve for photoreceptor sensitivity in which the response of an organism is determined at multiple wavelengths of light and at multiple light intensities at each wavelength. In this way, the response will depend on how well the photoreceptor absorbs light at the given wavelength. The wavelength at which peak activity can be observed in the living organism indicates the absorption maximum (in which the light is absorbed at highest efficiency) of the responsible photoreceptor. If performed to sufficient resolution, such action spectra can be compared to the absorption spectrum of a purified pigment or photoreceptor and in this way identify the nature of the photoactive pigment implicated in a given biological response [6
As a possible assay for Dmcry function, we investigated its characteristic property of degradation that followed upon photoreceptor activation. Levels of Dmcry protein decrease rapidly in flies subsequent to blue-light irradiation, likely due to conformational change in the photoreceptor leading to targeting to the proteasome [11
]. This degradation can be quantitatively monitored by western blot analysis with Dmcry antibody. However, in order to be useful for action spectroscopy, the amplitude of the response (decline in Dmcry concentration) must be proportional to the number of photons of light energy absorbed by the photoreceptor, and not simply a delayed response with little direct relation to the light input signal. To test this property, we irradiated living flies for fixed time intervals with blue light (450 nm) and observed decrease in levels of Dmcry protein over time as previously described [11
] (A). Importantly, when blue-light irradiation was performed at different blue-light intensities, the time required to reach a given decline in Dmcry protein levels was proportional to the given light intensity. For example, to obtain a decrease to 50% of the original Dmcry protein concentration requires 6.2 min at irradiance of 200 μmol m−2
, 9.3 min at 150 μmol m−2
, and 12.5 min at 100 μmol m−2
(from A), respectively. Therefore, Dmcry degradation obeys the Bunsen-Roscoe law of reciprocity, indicating that it is a response to the total number of photons, independent of irradiance time, and so represents an accurate measure of photoreceptor light responsivity [32
Redox State of Animal Cryptochromes and Photoreduction In Vivo
For generation of this Dmcry action spectrum, living flies were first dark adapted to accumulate maximum levels of dark state cryptochrome. Flies were then subjected to continuous irradiation at a set photon fluence rate of 17 μmol m−2
at wavelengths between 380 and 502 nm. Shorter wavelengths are impractical because of the increasing absorption from cell components and therefore increasing errors. Levels of Dmcry protein were monitored by western blot analysis. Illumination time was varied to provide the different irradiance, as permitted by reciprocity. Shorter or longer wavelengths of light proved to be ineffective at eliciting significant response (unpublished data). Dose-response plots of the time course of Dmcry protein degradation at different wavelengths of light showed linear decay as a function of the log irradiation time for points between 20% and 90% of dark levels of protein accumulation (Figure S1
). The action spectrum is plotted from these dose-response curves using the total irradiance required to reduce Dmcry protein levels to 50% of dark controls at each wavelength (B). The curve is inverted to give a visual image whereby the peak efficiency (the wavelength that required the shortest time to elicit 50% Dmcry degradation) represents the absorption maximum of the responsible photoreceptor. Peak wavelength sensitivity was at 450 nm, with defined shoulders around 420 and 480 nm, matching the spectrum of protein-bound oxidized flavin [33
]. Oxidized flavin is therefore the likely photoactive pigment of Dmcry in living whole flies, similar to plant cryptochromes [14
] and in marked contrast to DNA photolyases in which flavin is fully reduced [2
We next determined the nature of the chemical reaction induced by light in animal cryptochromes in living cells. We performed baculovirus-driven expression of Dmcry and Hscry1 cryptochromes in Sf21 insect cells, where photoreceptor protein accumulates to sufficiently high levels for direct application of spectroscopic and biophysical techniques in vivo [18
]. To verify whether cryptochrome-bound flavin can be directly observed, expressing whole Sf21 cells were harvested and placed intact inside a fluorimeter. Fluorescence emission was measured at 525 nm (characteristic of oxidized flavin) over an excitation range of 400–500 nm. Despite the substantial scatter due to measurements of these living intact cells, there was clearly observable signal increase peaking for excitation at 450 nm in cells overexpressing both Dmcry and Hscry1 as compared to uninfected control cells. These results are consistent with oxidized flavin bound to the dark state of the photoreceptors (Figure S2
). These data showing increased oxidized flavin in cryptochrome-expressing cells are in agreement with the resting state of Dmcry determined from action spectroscopy (B). Interestingly, mammalian cryptochrome also accumulates in the oxidized form and thereby shows functional similarity to Dmcry and not to DNA photolyases. Similar results have been previously obtained for Atcry1 [18
To initiate the photochemical reaction, Dmcry- or Hscry1-expressing cells were irradiated with blue light and returned to the fluorimeter at intervals for measurement of excitation spectra. This assay detects change in levels of oxidized flavin in these living cells. For both Dmcry and Hscry1, peak excitation at 450 nm showed a progressive decrease over time that matches the spectra for photoreduction of oxidized flavin (C and D). This decrease was not due to protein degradation since both Dmcry and Hscry1 protein levels remain stable throughout the time course of illumination in Sf21 cells (Figure S3
). Furthermore, flavin reoxidation is observed when illuminated cell cultures are returned to darkness (unpublished data), indicating that no cryptochrome degradation has occurred. Therefore, both tested cryptochromes had undergone a photoreaction in vivo, leading to change in redox state of protein-bound flavin (D).
A similar reaction, known as photoactivation, occurs in DNA photolyases, wherein the flavin chromophore is converted to the fully reduced form by an electron transfer reaction ultimately fed by an extrinsic reductant. An intraprotein electron transfer pathway from the protein surface to the buried flavin has been derived for this light-driven reaction in Escherichia coli
DNA photolyase (EcPl) based on crystallographic structural information and on a combination of site-directed mutagenesis and spectroscopy [35
]. This pathway comprises a chain of three tryptophan residues (W382–W359–W306) that are highly conserved throughout the photolyase/cryptochrome family. Recently, a study with purified Atcry1 has demonstrated the functional relevance of this reaction to cryptochrome photoreceptor activity [16
] by substitution of redox-inactive phenylalanines for two tryptophan residues, W400 and W324, which are found in the Atcry1 sequence and crystal structure [39
] at the homologous positions to W382 and W306 of EcPl, respectively. These mutant proteins (W400F and W324F) lack the predicted electron donor proximal to the flavin (W400) or exposed to the protein surface (W324). Both proteins were found to have impaired electron transfer activity in vitro and reduced biological activity in living plants. To determine whether flavin photoreduction may occur by a similar pathway of intermolecular electron transfer in animal-type cryptochromes, we have made point mutations in Dmcry of two conserved tryptophan residues. One mutation is distal to the flavin in this pathway (W342F), which corresponds to W306 in EcPl and W328F in Dpcry1 [26
], respectively. Second, we have introduced a substitution into the middle member of the electron transfer chain of Dmcry, corresponding to W359 of EcPl. The mutant Dmcry proteins were expressed in Sf21 insect cells to high levels and subjected to in vivo fluorescence spectroscopy to follow flavin photoreduction.
For determination of in vivo photoreduction, cryptochrome-expressing Sf21 cells were irradiated with blue light and returned to the fluorimeter periodically to determine remaining levels of oxidized flavins. Unexpectedly, both wild-type and mutant Dmcry proteins showed similar rates of photoreduction in these living cells at high light intensity (150 μmol m−2
white light), as did the W400F mutant of Atcry1 (A). This result is surprising since purified preparations of Atcry1 W400F protein and of the W328F Dpcry1 (homolog to Dmcry W342F) showed greatly reduced photoreduction in vitro at even higher light intensities, and there was no significant radical accumulation after this time period [16
]. Therefore, the efficiency of cryptochrome photoconversion in vivo is much higher than that of the purified, isolated protein in vitro, perhaps due to a more conducive redox environment and the presence of relevant electron donors/acceptors in vivo. Nevertheless, at lower photon fluence (10 μmol m−2
blue light), a significant decline in the rate of photoreduction is observed in the phenylalanine mutants of both Dmcry and Atcry1 as compared to wild-type proteins (B—see also Figure 4
SA and 4
SB for experiments performed with further reduced photon fluence). This phenomenon provides a consistent explanation for the observed biological activity of amino acid–substitution mutants in Dpcry and Atcry1. In Arabidopsis
, biological activity of the mutant proteins (W400F and W324F) was determined at only low blue-light intensity and found impaired at this irradiance in vivo.
Disruption of the Putative Electron Transfer Chain Alters Photoresponse of Dmcry at Low Photon Fluence Rates
To determine the state of the photoreceptor (radical or fully reduced) in the activated cryptochrome, fluorescence emission techniques in whole cells are not sufficient as they can identify only the oxidized form of the flavin chromophore. It cannot, therefore, be concluded from the above studies whether photoreduction in vivo leads to accumulation of a semiquinone intermediate as for plant cryptochromes [16
], or whether the fully reduced form of flavin accumulates in animal cryptochromes as for DNA photolyases [1
]. To directly monitor for radical accumulation in response to light in vivo, whole-cell EPR spectra were recorded as previously described [17
]. Intact Sf21 insect cells with overexpressed Dmcry or Hscry1 protein were irradiated in parallel with nonexpressing control cells at the identical intensities of blue light and rapidly frozen for EPR analysis. A paramagnetic species that does not accumulate in control cells was induced by blue-light irradiation of Dmcry- (A, traces B and C) and Hscry1-expressing cells (A, traces E and F). This species appears with similar kinetics to both plant cryptochromes [17
]. Interestingly, there was detectable amount of a radical present even in the dark in some samples (see trace D), although not in all trials, perhaps due to concentrations below our level of detection. This result is in marked contrast to previous experiments, for which radical accumulation was never observed in unilluminated cells [17
]. Finally, we have examined the Drosophila
mutant proteins W397F and W342F for radical accumulation in vivo (A, traces H–K). Saturating illumination (40 μmol m−2
blue light) leads to accumulation of a radical intermediate form.
Animal Cryptochrome Photocycle Includes Light-Dependent Radical Accumulation In Vivo
To further characterize these signals, X-band–pulsed ENDOR spectroscopy was applied to illuminated whole cells expressing Dmcry (B). The observed spectrum in the expressing cells (trace A) is very similar to that obtained from the purified Dmcry protein (trace B). Both spectra differ from those of neutral flavin radicals as seen in plant cryptochromes and corroborate the assignment to an anionic radical species as given previously for the purified protein [22
]. Taken together, the in vivo spectroscopic data conclusively indicate that the photocycle for both Dmcry and Hscry1 involves light-dependent flavin reduction and accumulation of the radical state.
Finally, it is necessary to establish the biological relevance of the observed in vivo photoconversion of animal cryptochromes. In plant cryptochromes, the radical state has been demonstrated to be the biologically active signaling state for both Atcry1 and Atcry2 [17
]. This conclusion resulted from the observation that green light reversed the effect of blue light in the course of cryptochrome activation, due to photoconversion of the active, radical form to the fully reduced, inactive species [17
] (see also A). A simple means to determine whether light-induced radical accumulation also has biological relevance for animal cryptochromes in vivo is therefore to measure whether green light (above 525 nm) affects both Dmcry protein accumulation and the kinetics of cryptochrome photoreduction.
To test this prediction, we performed bichromatic irradiation of flies simultaneously with blue and green light (B+G) and compared the response to that obtained with the identical intensity of blue light by itself (B) (A). Green-light irradiation by itself resulted in no change in Dmcry protein levels (unpublished data). In each of three independent trials, we observed more rapid decline in Dmcry protein levels in blue light (B) as compared to coirradiation with blue and green light (B+G). This antagonistic effect can only be explained by photoconversion of the (green-light absorbing) radical signaling state to an inactive redox form. In the case of Dmcry-expressing cell cultures, an effect of green light on cryptochrome photoreduction was directly monitored. Cell cultures irradiated with blue and green light (B+G) show accelerated photoreduction of Dmcry as compared to blue light (B) alone (B). Although the accumulation of fully reduced flavin can not be directly monitored by this technique, these data are consistent with a shift in the overall flavin photoequilibrium subsequent to formation of the radical, and thereby consistent with the effect of green light on biological activity observed in living flies.
Green Light Alters Cryptochrome Photoconversion In Vivo That Correlates with Function
Although the present study so far has shown that mammalian cryptochromes undergo similar photoreactions to those of insect and plant, a functional role for light in biological activation remains to be demonstrated. To address this question, we have assayed for a form of activation of Hscry1 in response to light in living flies, where endogenous cryptochrome (Dmcry) is known to undergo light-dependent changes resulting in proteolysis (). Transgenic flies expressing full-length Hscry1 under the control of the UAS promoter element were obtained by established procedures (see Materials and Methods
). Expression of the recombinant Hscry1 was verified by western blot analysis in two independent transformed lines (A and B). Expressing flies were then dark adapted to accumulate maximal cryptochrome protein and subsequently irradiated with white light. Levels of Hscry1 were assayed during the course of the irradiation. Interestingly, as is the case for Dmcry, significant decrease in Hscry1 protein levels were observed shortly after transfer to white light (). These results indicate that Hscry1 undergoes light-dependent proteolysis as does Dmcry in living flies. Since degradation of Dmcry correlates with activation by light and biologically relevant radical formation, a similar mechanism of biological activation is also likely for Hscry1.
Light Alters Biochemical Properties of Hscry1 in an In Vivo Context