Phytochrome monomers Light-induced Changes in the Dimerization Interface of Bacteriophytochromes*

Phytochromes are dimeric photoreceptor proteins that sense red light levels in plants, fungi and bacteria. The proteins are structurally divided into a light-sensing photosensory module, consisting of PAS, GAF and PHY domains, and a signaling output module, which in bacteriophytochromes typically is a histidine kinase (HK) domain. Existing structural data suggest that two dimerization interfaces exist between the GAF and HK domains, but their functional roles remain unclear. Using mutational, biochemical and computational analyses of the Deinococcus radiodurans phytochrome, we demonstrate that two dimerization interfaces between sister GAF and HK domains, respectively, stabilize the dimer with approximately equal contributions. The existence of both dimerization interfaces is critical for thermal reversion back to the resting state. We also find that a mutant, in which the interactions between the GAF domains were removed, monomerizes under red light. This implies that the interactions between the HK domains are significantly altered by photoconversion. The results suggest functional importance of the dimerization interfaces in bacteriophytochromes. INTRODUCTION Phytochromes are photoreceptors found in plants, fungi and various microorganisms, like cyanobacteria and proteobacteria (1, 2). They modulate their biochemical activity in response to the light environment. Phytochromes exist in two states that absorb red light (the Pr state) and farred light (the Pfr state). Depending on the species either Pr or Pfr is the resting state and the proteins are classed as ‘prototypical’ and ‘bathy’ phytochromes, respectively (3-5). Generally, the proteins have a photosensory module and an output module. The photosensory module usually contains PAS (PER, ARNT, SIM) and GAF (cGMP phosphodiesterase, adenylate cyclase, FhlA) domains, which bind a chromophore, and a C-terminal PHY (phytochrome-specific GAFrelated) domain. These domains convert an incident light signal into a conformational signal, which is then relayed to control the activity of the output module. The output module varies between species but is a histidine kinase (HK) domain in many bacterial phytochromes (6, 7). 1 http://www.jbc.org/cgi/doi/10.1074/jbc.M115.650127 The latest version is at JBC Papers in Press. Published on May 13, 2015 as Manuscript M115.650127 Copyright 2015 by The American Society for Biochemistry and Molecular Biology, Inc. by gest on N ovem er 9, 2017 hp://w w w .jb.org/ D ow nladed from Phytochrome monomers The first step of the photocycle is the photoisomerization of the linear tetrapyrrole chromophore. In bacteriophytochromes, this is a biliverdin that is attached to the PAS-GAF domains. The signal is transmitted to the PHY domain by the so-called ‘tongue’ or ‘hairpin’ extension (8-13). Crystal structures of prototypical and bathy phytochromes in their respective Pr and Pfr resting states (8, 9, 12, 14) and crystal structures of the same phytochrome in the dark and after red light illumination show that the PHYtongue refolds (11). This leads to a shortening of the tongue and a large-scale opening of the entire phytochrome photosensory module dimer (11, 12). The overall arrangement of PAS-GAF-PHY is conserved in all structures published to date (810, 14-16). Bacterial phytochromes are generally considered as parallel (head-to-head) dimers (9, 11, 12, 17), even though some phytochrome fragments crystallize as antiparallel (head-to-tail) dimers (8, 10, 15) and head-to-tail arrangement was proposed to be functionally relevant for BphP1 from Rhodopseudomonas palustris (18). The head-to-head arrangement is supported by electron micrographs (12, 19) and small-angle Xray solution scattering data (20). From crystal structures, dimer contacts were identified between the GAF domains (9, 11, 12, 14, 16, 17, 21) and possibly also between the PHY domains (9, 22, 23). Since the structure of a full-length phytochrome is yet to be solved, it is not fully established whether the HK domains have a buried dimer interface. However, electron micrographs (12, 19) and other sensor histidine kinase structures (24-26) suggest that such an interaction exists, at least in one of the photochemical states. In solution, the strength of the dimerization interfaces varies between different phytochromes. The complete photosensory module (PAS-GAFPHY) and the PAS-GAF fragment of D. radiodurans form stable dimers in solution (11, 12, 17, 27, 28) but this is not the case in some other species. The photosensory modules of the Agrobacterium tumefaciens Agp1 form a mixture with mainly dimers (29). This is also true for cyanobacterial Cph1 from Synechocystis sp. (8, 15, 30), whereas the photosensory module of Cph2 is predominantly monomeric in solution (10, 31). Plant and cyanobacterial phytochromes are commonly thought to have the principal dimerization interface at the C-terminal region, whereas the N-terminal photosensory module region does not seem to form dimers (14, 32-34). Indeed, the photosensory module of Arabidopsis thaliana phyB behaves as a monomer in solution but crystallizes as a parallel dimer (14). In summary, it is likely that most phytochromes form parallel dimers by interactions between the sister GAF and HK domains. However, the number of exceptions described above makes it worthwhile to put this hypothesis to the test. Dimerization is an intrinsic feature of phytochromes, but what is its functional role? In Arabidopsis thaliana, dimerization of the output modules was demonstrated to be necessary for the proper function of the plant phytochrome B (35, 36). Moreover, phytochromes B to E form heterodimers, but the functional role of this flexible quaternary arrangement has yet to be discovered (37, 38). Considering that many bacterial histidine kinases phosphorylate in trans, dimer arrangement of the HK domains may be functionally important for bacterial phytochromes (39-41). However, the role of the dimerization interface in the photosensory module is less clear. While searching for an efficient fluorescent protein, Auldridge et al. found a three-point mutation that completely monomerized the D. radiodurans PAS-GAF fragment (42). Surprisingly, this also restored photochromicity (42). This hints at that even the dimerization interface in the GAF domain could influence phytochrome function. Phytochromes can be switched between the Pr and Pfr states by light, but in most cases they thermally dark-revert into one of the two states. The dark reversion rate is likely important for regulating the activity of phytochromes in vivo. Factors that influence this rate include truncation of the HK domain (12, 28, 43). By extensive mutational analysis it has emerged that no single residue alone, but rather a set of residues controls dark reversion (10, 12, 14, 44). Together, this leads us to hypothesize that the tertiary or quaternary organization of the protein is important for dark reversion. We test this in this study. It is currently unclear how the activity of the output domain of phytochromes is controlled structurally. Electron microscopy analysis of the full-length bacteriophytochrome from D. radiodurans suggests that the opening in between the PHY domains (11) impacts the relative 2 by gest on N ovem er 9, 2017 hp://w w w .jb.org/ D ow nladed from Phytochrome monomers orientation of the HK output domains (12). Crystallographic snapshots of related dimeric sensor histidine kinases have yielded a number of proposed mechanisms, which often involve asymmetric kinking or bending of the histidine kinase structures (41, 45-48). In these proposals, the histidine kinase domains remain dimeric, i.e. the dimer interface is not broken by the light stimulus, neither fully nor partially. This seems reasonable, because the hydrophobic interactions that tie together the histidine kinase monomers are presumed to be very stable. However, the opening movement of the PAS-GAF-PHY protein observed by us (11) and the electron micrographs reported by Burgie et al. (12) suggest that a mechanism in which the HK domains separated fully or partially should be tested. Indeed, we find in this study that the dimer interface between the HK domains can be broken by light and conclude that such a mechanism should be considered. Here, we apply the set of mutations that monomerize PAS-GAF fragment to PAS-GAFPHY (photosensory module) and the full-length phytochrome from D. radiodurans. Our data show that the phytochrome dimer is stabilized across two interfaces: one between neighboring GAF domains and one between the HK domains, but not between the PHY domains. Surprisingly, the HK interface can be broken by red light illumination in the construct with mutations at the GAF/GAF interface. The integrity of both dimerization interfaces is needed for normal thermal (dark) reversion from the Pfr state to the Pr state. The absence of the HK domain (with its dimerization interface) in the PAS-GAF-PHY construct slows down dark reversion whereas the loss of GAF dimerization completely abolishes it. We conclude that both dimerization interfaces are important for phytochrome function. EXPERIMENTAL PROCEDURES Cloning and protein purification. The expression plasmids coding for wild-type D. radiodurans fragments (PAS-GAF, PAS-GAFPHY, full-length) were kindly provided by the labs of Prof. R. D. Vierstra and Prof. K. T. Forest and are described elsewhere (12, 17, 27). The three monomerizing mutations were introduced by using the QuikChange Lightning Multi Site-Directed Mutagenesis Kit (Agilent Technologies, La Jolla, CA) with the following primers: F145S, 5’– CTGCGCAACGCGATGTCAGCGCTCGAAAG TGC–3’; and L311E/L314E, 5'– CTCGAATACCTGGGCCGCGAGCTGAGCGA GCAAGTTCAGGTC–3' (mutated nucleotides underlined). All sequences were verified using the sequencing facility at the University of Jyväskylä. The phytochrome constructs were expressed and purified as described previously (11, 28, 49). The purified phytochrome fragments in the final SEC buffer of (30 mM Tris•HCl, pH 8.0) were concentrated to 20–40 mg/ml, flash-frozen, and stored at -80°C. The samples were thawed, and filtered with 0.22 μm centrifugal filter units (Amicon Ultrafree, Millipore) immediately before experimental characterization. Sample illumination. The samples were illuminated or kept in dark just prior each measurement. The illuminated samples were illuminated with 780 nm (far-red, 9 mW output power) or 655 nm (red, 7mW output power) lightemitting diode lights until photoequilibrium was reached. The samples were kept dark, unless otherwise indicated. UV-Vis spectroscopy. The UV-Vis spectra were measured with a Perkin Elmer LAMBDA 850 UV-Vis spectrophotometer as previously described (11, 28, 49). The samples were diluted with buffer (30 mM Tris•HCl, pH 8.0) to obtain approximate A280 value of 0.1. All measurements were carried out at ambient conditions (room temperature) and in complete darkness. The dark reversion was measured with Hitachi U-2910 UV/VIS spectrophotometer (Hitachi, Japan) by sequentially recording the absorption spectrum in the wavelength range of 500–850 nm after the photoequilibrium had been reached under red light illumination. The samples were diluted with buffer (30 mM Tris•HCl, pH 8.0) to obtain approximate A700 value of 1. PAS-GAF and PAS-GAFmon reversion data were recorded at 3 min intervals; longer fragments (PAS-GAF-PHY, PAS-GAFPHYmon, FL, and FLmon) were recorded also at 15 min intervals. The UV-Vis detection light did not affect the steady-state spectra of the constructs, but the rate of the dark reversion was slightly increased by the detection system (data not shown). However, this effect was minimized by recording data at long time intervals. The effect can be considered identical in all measurements and therefore the results are comparable. The 3 by gest on N ovem er 9, 2017 hp://w w w .jb.org/ D ow nladed from Phytochrome monomers exponential fits from normalized dark reversion data were calculated using equation (1).