A bacteriophytochrome Pr/Pfr heterodimer studied through single-particle time-resolved cryo-electron microscopy Authors: Prabin Karki, William C Budell, Rohit Kannachel, David Menendez, Carolina Hernandez, Joshua H Mendez, Shemiran Kanou, Manjeet Singh, Chavdar Slavov, Peter Schwander, Tek Narsingh Malla, Emina A Stojković, Marius Schmidt Published: 2026 ABSTRACT AbstractPhytochromes are dimeric photoreceptors found in bacteria, fungi, and plants that reversibly interconvert between a red-absorbing Pr state and a far-red-absorbing Pfr state. In bacteria, phytochromes (BphPs) regulate diverse responses through a two-component signaling pathway comprising a C-terminal histidine kinase (HK) and a response regulator. A previous cryo-EM study of the wild-type BphP from Stigmatella aurantiaca (SaBphP2) revealed a stable Pr/Pfr heterodimer in which the two protomers adopt distinct Pr and Pfr conformations. Here, using the Spotiton technique, we captured the same heterodimer by illuminating SaBphP2 particles on the cryo-EM grid and vitrifying them 10 ms later. Comparison with the Pr/Pr homodimer reveals a 180° rotation of the HK domain, driven by an unwinding of the coiled-coil helices that connect the photosensory core to the enzymatic domain. The large-scale reorientations provide mechanistic insight into light-triggered signal transduction mediated by bacterial phytochromes.Subject terms: Cryoelectron microscopy, Kinases, Photobiology Introduction Understanding the molecular mechanisms that regulate perception of essential environmental stimuli, such as light and activation of the subsequent signaling cascade, is imperative to comprehend and manipulate organismal responses1. Light affects the behavior of higher organisms with specialized organs (eyes) with various opsins2,3 that sense and react to this important stimulus. Prokaryotes sense light typically through molecular photoreceptors that are found in the membrane or in the cytoplasm4. These photoreceptors contain an enzymatic output module (OPM), which, in turn, activates a downstream effector protein. Bacterial signal transduction is achieved by a two-component signaling (TCS) system consisting of a histidine kinase (HK) and a response regulator (RR)5, whose activation results in various physiological responses6,7. Bacteriophytochromes (BphPs) are part of a TCS system responsible for red and far-red light-sensing and signaling. Phytochromes were originally discovered in plants and can also be found in fungi, but not in animals8–10. BphPs consist of a light-absorbing photosensory core module (PCM) and a C-terminal OPM (Fig. 1a). The PCM consists of three domains: PAS (Per-Arnt-Sim), GAF (cGMP phosphodiesterase/adenylyl cyclase/FhIA), and PHY (phytochrome-specific GAF-related). In most BphPs, the light-absorbing chromophore is a bilin pigment called biliverdin (BV) (Fig. 1b). BV is a heme-derived open-chain tetrapyrrole (consisting of pyrrole rings A–D) covalently bound to a conserved N-terminal cysteine (Cys-13 in the BphP under investigation here, Fig. 1b). In typical BphPs, the OPM is a histidine kinase (HK)11,12 whose activity is controlled by light absorbed by the BV in the PCM. Light-dependent HK activity has been characterized for various BphPs such as those isolated from the photosynthetic Rhodopseudomonas palustris (RpBphP2 and RpBphP3)13,14, from the non-photosynthetic Stigmatella aurantiaca (SaBphP1 and SaBphP2)15,16 and from Pseudomonas aeruginosa (PaBphP)17. Certain BphPs exhibit OPMs with different enzymatic activities: for example, the Idiomarina sp. IsPadC features a diguanylyl cyclase18, while the Deinococcus radiodurans DrBphP19 exhibits phosphatase activity20. The HK, the OPM found in most BphPs, binds a RR, which is phosphorylated, typically at a conserved Asp21,22, and subsequently leaves the HK to initiate physiological changes. To date, the molecular mechanisms that explain how light regulates the HK activity and how the RR subsequently interacts with the OPM are yet to be determined. Photochemistry of a classical BphP. a The sequence and the domain composition of one protomer of the Stigmatella aurantiaca full-length SaBphP2 is shown. b The BphP photocycle with several intermediates on various time scales. The chemical structure of the BV chromophore is shown in the upper part. 700 nm light induces the Z to E isomerization of BV ring-D (blue arrow). The structure changes from the Pr to the Pfr state. Illumination by 750 nm light reverses the reaction. The structures of shorter PAS-GAF-PHY fragments lacking OPM are shown. Left: structure of the Pr/Pr homodimer (protomers A and B) as determined by crystallography16,73. Right: The structure of the Pfr/Pfr homodimer as determined by cryo-EM34,74,75. The dimer interface helices that are crossing in Pr/Pr become parallel (red arrows). Dotted red boxes: the sensory tongues in Pr exhibit 2-stranded β-sheets. In Pfr, the strands separate in both sensory tongues, and one strand in each forms an α-helix. c Cryo-EM structure of the full-length SaBphP2 with PCM, linker, 4-helix bundle and HK domains (marked) in the Pr/Pfr heterodimeric state34–38. Note, the linkers cross the dimer boundary. The kink in the linker helix of protomer A is marked. The yellow oval marks the chromophore positions in both subunits. The characteristic separation of strands is visible in protomer B, but not in protomer A. In photosynthetic bacteria, BphPs are involved in the development of the photosynthetic machinery23–25, whereas their roles in non-photosynthetic bacteria are diverse9 and often related to specific situations prevalent in the natural habitat of the bacteria8,15,26. In the non-photosynthetic, soil-dwelling myxobacteria, the BphPs are intimately involved in the formation of fruiting bodies27, which are multicellular stages in their complex life-cycle15,28–30. Myxobacteria are also investigated as a biocontrol agent against plant disease31 and for the production of secondary metabolites with possible anticancer and antimicrobial properties32. The synthesis of bioactive compounds has been modulated by red light32,33, which suggests the involvement of BphPs. The BphPs from Stigmatella (SaBphP1 and SaBphP2) each exhibit an HK domain15,16. Both BphPs exist as homodimers where each of the two protomers binds the same BV chromophore and has the same amino acid sequence, respectively. The classical SaBphP2 has been critical to determining photocycle intermediates on ps, ns and ms timescales via time-resolved crystallography, and it has been established that its HK activity is light dependent34. BphPs undergo a reversible photoconversion between a red-light absorbing Pr state and a far-red light absorbing Pfr state (Fig. 1b). When the BV chromophore is illuminated by red light, it isomerizes from an all-Z to a ZZE configuration by rotating ring-D about the C15-C16 double bond (Fig. 1b). After the isomerization, the protein undergoes conformational changes towards a stable state called Pfr that absorbs in the far-red. It is the balance between the Pr and Pfr states that is sensed by downstream effector proteins. The signature of the Pr to Pfr transition is the conformational transition of an extension of the PHY domain called the sensory tongue that exists in the dark state as a two-stranded β-sheet. During the Pr to Pfr transition, the two strands separate, and one of the strands forms an α-helix35,36 (Fig. 1b). Our recent characterization of the intact, wild-type SaBphP2 with single particle cryo-EM has shown34 that the Pr to Pfr transition proceeds through a Pr/Pfr heterodimer (Fig. 1c)37,38. One protomer is in the Pr state, with the BV in the all-Z configuration and the sensory tongue a β-sheet. The other is in the Pfr state, with the BV chromophore in the ZZE configuration and with the characteristic α-helix found in the sensory tongue. The transition from the Pr/Pr homodimer to the Pr/Pfr heterodimer is accompanied by large structural changes in the PCM (Fig. 1b) as well as in the linker region of the SaBphP2 (Fig. 1c), as outlined previously34,39. The time-resolved cryo-EM study presented in this manuscript confirms that the Pr/Pfr heterodimer (Fig. 1c) can be generated by actively exciting particles in the homodimeric Pr state by an intense light source. Thus, the photocycle needs to be modified to account for this heterodimer intermediate. Results Time-resolved cryo-EM In our previous cryo-EM study on the SaBphP234, the structure of the Pr/Pfr heterodimer was discovered only after the protein was exposed to ambient white light. While this discovery itself was transformative, it left open the possibility that the heterodimer might represent a freezing or illumination artifact rather than a bona fide intermediate of the photocycle. Here, we demonstrate the formation of the Pr/Pfr heterodimer by exciting SaBphP2 Pr/Pr homodimers by a 75 mW continuous-wave (cw) laser in a single millisecond pump-freeze type40 experiment using the Spotiton technique41. In this time-resolved (TR) cryo-EM experiment, SaBphP2 particles on a cryo-EM grid were illuminated by red (660 nm) laser light for 4 milliseconds (ms), approximately 9.5 ms before vitrification in liquid ethane. The time resolution of the TR cryo-EM experiment is defined by the fastest time between laser excitation and vitrification in liquid ethane possible by the Spotiton setup. It is dependent on the plunger velocity and the shortest distance possible between the laser focus and the ethane reservoir. Here, it is about 5 ms. In addition to the illumination conditions, the relative concentrations of the protein conformers depend on the quantum yield of photoconversion qpc, which is estimated to be 12%. As a result of the photoexcitation, 38% of the molecules are in the Pr/Pfr state (see the Methods and the Supplementary Material, SM, for detailed evaluations). From the TR cryo-EM experiment, 3D density volumes for both the Pr/Pr homodimer as well as the Pr/Pfr heterodimer were reconstructed at rather low resolution (~6.2 Å). The resolution of the Pr/Pfr heterodimer volume was improved to 5.01 Å (Table 1, Supplementary Fig. S1) by using a tighter mask with a 3 Å resolution filter. To obtain an unbiased initial model, the partial (incomplete) model of the Pr/Pr homodimer determined earlier34 was modified in the PCM region to fit the cryo-EM density. The HK region of the density was modeled using the corresponding region of the complete Pr/Pfr model previously determined34. After refinement, a complete structure of the Pr/Pfr heterodimer is obtained and called SaBphP2-Pr/Pfr4ms (for 4 ms illumination). To determine a structural model of a Pr/Pr homodimer, cryo-EM data collected with Spotiton previously39 were reassessed because more particles are available for a reconstruction (Supplementary Fig. S2). This previous experiment did not reveal particles in the Pr/Pfr state, since illumination with a weak LED light source did not photoconvert a detectable amount of molecules. At an overall resolution of 5.85 Å (Table 1), the density in the HK region is meaningful and contingent. The reassessment allowed us to finish the partial structure of the Pr/Pr homodimer34 by adding the missing part of the linker helices, the 4-helix bundle and the HK lobes. We call the structure SaBphP2-Pr/PrLL (LL for low light). Density maps and their respective structures are shown in Fig. 2a, b. Table 1Cryo-EM data collection, refinement, and validation statistics#1 SaBphP2-Pr/PrLL /#2 SaBphP2-Pr/Pfr4ms(EMDB-73380)(EMDB- 73612)(PDB 9YS3)(PDB 9YXN)Data collection and processing No. micrograph10,20714,371 Magnification105,00081,000 Voltage (kV)300300 Electron exposure (e–/Å2)45.4759.37 Defocus range (μm)0.8–2.50.8–1.8 Pixel size (Å)0.82900.8560 Symmetry imposedC1C1 Initial particle images (no.)1,955,8472,003,863 Final particle images (no.)70,90887,111 Map resolution (Å)5.855.01 FSC threshold0.1430.143 Map resolution range (Å)5–165–14Refinement Initial model used (PDB code)8uph,8upk8uph, 8upk Model resolution (Å)5.965.1 FSC threshold0.1430.143 Model resolution range (Å)5.4–7.74.8–7.5Model composition Non-hydrogen atoms10,16610153 Protein residues14781478 Ligands22B factors (Å2) Protein532.26490.84 Ligand387.19333.75R.M.S. deviations Bond lengths (Å)0.0030.003 Bond angles (°)0.7551.132 Validation MolProbity score2.242.27 Clashscore21.4822.89Ramachandran plot Favored (%)93.7693.76 Allowed (%)6.116.04 Disallowed (%)0.140.20 Cryo-EM data collection, refinement, and validation statistics Large-scale conformational changes of SaBphP2. Structures of the SaBphP2-Pr/PrLL homodimer (a) and the Pr/Pfr4ms heterodimer (b) from time-resolved cryo-EM. Cryo-EM maps are shown in transparent blue. The contour level is lowered in (a) to show the density of the HK lobes and is adjusted in (b) to show the strand separation in the sensory tongue of protomer B. View R: The structures are reoriented so that the sensory tongues are lying in the same plane. The Pr/Pr homodimer (a) and the Pr/Pfr heterodimer (b) structures are shown side by side to demonstrate the extensive change in the linker geometry after light activation. Rotations that result after light excitation are indicated by the arrows. The domain composition (PAS, GAF, PHY, linker and HK) is shown on the left. (a) The Pr/PrLL homodimer. The linker helices cross in the PHY domain (black dot marked 1) and again in the four-helix bundle (black dot marked 2). The HK lobes are located on top of their respective PCM protomers. View R: The linker helices cross near the PHY domain (black dot marked 1), a ~180° rotation is necessary to reach the Pr/Pfr4ms heterodimer. b The Pr/Pfr4ms heterodimer. The sensory tongue strands (marked yellow) separate in protomer B only. One strand forms an α-helix. For the other protomer, no strand separation is observed. The linker helices cross near the HK domain (black dot marked 2). The HK lobes are located on top of the PCM of the opposite protomer. In view R, the PCM interface helices (red) and the linker helices are straight and parallel (red arrow). A ~60o rotation further unwinds the helices during the transition to the Pfr/Pfr homodimer34. The Pr/Pr homodimer and the Pr/Pfr heterodimer The Pr/PrLL and the Pr/Pfr4ms structures can be distinguished in the PCM region. Apart from the characteristic β-sheet to α-helix transition of the sensory tongue (Fig. 2a, b, compare Fig. 3a with Fig. 3 b), several other structural relaxations provide clues on the progress of the photoreaction. These include changes in domain orientations and distance changes between the domains in the PCM. They are listed in Table 2 and used as criteria to distinguish Pr/Pr homodimers from Pr/Pfr heterodimers. Angles and distances of the Pr/Pfr4ms structure agree with those extracted from the Pr/Pfr heterodimer structure previously determined34. In addition, distances and angles measured from the Pr/PrLL structure agree with those of the Pr/Pr homodimer determined earlier at high resolution34 (Table 2). Focus on the sensory tongue region. a The Pr/PrLL homodimer, the sensory tongue region is marked in yellow, left protomer A, right protomer B. b The Pr/Pfr4ms heterodimer, the sensory tongue region is marked in yellow, left protomer A, right protomer B, amino acid residues that flank the structural moieties are indicated. The strand separation is marked with an arrow. Table 2Structural characteristics to differentiate Pr/Pfr heterodimers from the Pr/Pr and Pfr/Pfr homodimersHelix angleAngleAngleAngleDistance [Å]AA 292-330PAS/GAFPAS/PHYGAF/PHYResidue 400SaBphP2-Pr/Pfr4msa27.235.340.35.073.0Pr/Pfr heterodimerb25.934.338.96.274.5Pfr homodimerb23.627.126.50.773.7Pr homodimerb42.341.655.113.565.0SaBphP2-Pr/PrLLc39.942.755.913.263.8The angle between the dimer interface helices (amino acids 292–330), the relative angles between the PAS, GAF and PHY domains in the PCM and the distances between residues 400 in protomers A and B, respectively, were used as discriminable features between the distinct species.Species obtained from multiple experiments are compared.aSaBphP2-Pr/Pfr4ms: this experiment.bThe Pr/Pfr heterodimer, the Pfr/Pfr and Pr/Pr homodimers from Malla et al.34.cSaBphP2-Pr/PrLL homodimer as determined by re-processing cryo-EM data collected earlier39. Structural characteristics to differentiate Pr/Pfr heterodimers from the Pr/Pr and Pfr/Pfr homodimers The angle between the dimer interface helices (amino acids 292–330), the relative angles between the PAS, GAF and PHY domains in the PCM and the distances between residues 400 in protomers A and B, respectively, were used as discriminable features between the distinct species. Species obtained from multiple experiments are compared. aSaBphP2-Pr/Pfr4ms: this experiment. bThe Pr/Pfr heterodimer, the Pfr/Pfr and Pr/Pr homodimers from Malla et al.34. cSaBphP2-Pr/PrLL homodimer as determined by re-processing cryo-EM data collected earlier39. Rotation by 180 ° of the HK domain, including the 4-helix bundle and the corresponding lobes The Pr/PrLL and Pr/Pfr4ms structures are quite distinct. Particular attention is drawn to the coiled linker helices that cross each other (black dots marked with 1 and 2 in Fig. 2; see also Fig. 4 and Supplementary movie 1). Another perspective (marked R in Fig. 2) is obtained by reorienting both structures so that the sensory tongues of both protomers are approximately in the same plane. Most peculiar, from the perspective of view R, the linker helices in the Pr/Pfr4ms structure are parallel (red arrow in Fig. 2b), whereas the ones in the Pr/PrLL structure cross two times (marked 1 and 2 in Fig. 2a and in Fig. 4a, b). Figure 4d–f presents mechanical models that consist of two strings in lieu of the linker helices of protomers A and B. Figure 4d demonstrates that the coiled helices in the SaBphP2 Pr/Pr homodimer (Fig. 4a, b) can only unwind to a parallel configuration when the entire OPM, including the 4-helix bundle and the HK lobes, rotates ~180° about the central axis. Supplementary Movie 2 provides an impression of this rotation. Supplementary Movie 3 demonstrates that the two linker helices would unphysically interpenetrate if the rotation is not allowed to take place. We conclude that the rotation is complete by 9.5 ms, corresponding to the time delay probed in this study. Shorter time delays will be required to resolve the rotation itself and the accompanying protein relaxations during the transition. The linker helix topology. a The structure of the SaBphP2-Pr/PrLL. The cryo-EM density is shown in gray, protomer A in red, protomer B in blue. The HK lobes and their densities are removed for better illustration. The PCM, the linker and the 4-helix bundle are marked. The liker helices cross twice, marked with (1) and (2). Black arrow: direction of displacement of protomer B. b Same as in a, side view. c Structure of the PaBphP Pr/Pr homodimer. Linker helices are shorter and cross only once, marked with (1)17. The 4-helix bundle is marked by a green, transparent rectangle. d The above (1) and below (2) linker helix topology in SaBphP2 are represented by thick, overlapping strings (red: protomer A, blue: protomer B). The helices can only become parallel by a rotation (green arrow). The HK lobes are schematically shown by ellipses. The linker and the 4-helix bundle are marked on the left. e, f: two alternative dynamical models for the PaBphP represented schematically by strings. e HK lobe separation by sliding. This requires that the 4-helix bundle separates (red arrow). f A rotation of the HK lobes (green arrow) is required when the 4-helix bundle remains intact (green rectangle). The 4-helix bundle in panels (e) and (f) is depicted shorter to reflect its reduced length in PaBphP. The HK lobes are schematically shown by the ellipses in (e) and (f). Although in view R helix intersection 1 unwinds, an additional intersection (marked 2 in Fig. 2b) remains after the rotation. When the Pr/Pfr state is excited, this intersection unwinds by an additional rotation of the HK lobes by ~60°, as shown earlier by the cryo-EM density of the Pfr/Pfr homodimer34. In summary, a complete Pr to Pfr transition occurs via a double-unwinding mechanism of the coiled-coil linker helices driven by the transition of both subunits to Pfr, that ultimately ends with all-parallel linker helices in the Pfr/Pfr homodimer when viewed from both the front and the side of the molecule34. Transitions from the Pr to the Pfr homodimeric states and back are accompanied by a “swivel-like” back-and-forth rotation of the entire HK domain by about 240°. Discussions As described, the Pr/Pr homodimer to Pr/Pfr heterodimer transition is addressed by explaining structural changes necessary to unwind the linker helices starting from coiled-coil helices in the Pr/Pr homodimer. It does not require near-atomic resolution to resolve these large-scale topological changes. At ~5–6 Å resolution, α-helices and their connectivity are clearly discernible. This is adequate to establish domain rotations, resolve the coiled-coil helix rearrangement, trace secondary-structure elements and follow domain reorganization. The string model (Fig. 4) and Supplementary Movie 2 demonstrate that a large-scale rotation of the entire histidine kinase (HK) domain about the central axis occurs. This observation suggests that the rotational dynamics of the HK domain may warrant additional consideration in phytochrome signaling. Rather than solely modulating HK enzymatic activity as demonstrated for other BphPs14,17, the light-induced OPM dynamics suggest an additional mechanism controlling RR binding and dissociation (see also below) through rotation. Implications for other proteins Due to the lack of structures of the HK domains in BphPs, the prevailing opinion has been that the HK lobe of one protomer is positioned on top of the photosensory module (PCM) of the opposing protomer14,17,20,21,42,43. Our complete SaBphP2 structures in different signaling states demonstrate that this assignment is incorrect for the SaBphP2 Pr/Pr homodimer (Fig. 2a), where each HK domain is positioned above its corresponding PCM protomer (the HK lobe of protomer A is located on top of the PCM of protomer A, and that of protomer B is located on top of the PCM of protomer B). In our previous study34, we showed that this arrangement is reversed in the Pr/Pfr heterodimer (Fig. 2b), but the limited resolution at that time did not allow a definitive structural and mechanistic interpretation. The work presented here resolves this ambiguity by showing that a 180° rotation of the entire HK domain provides a coherent explanation for the observed topological differences between the homodimer and heterodimer. Caused by the rotation, the HK lobe of protomer A (B) is displaced so that it is found on top of the PCM of protomer B (A). Other BphPs with HK activity, such as the bathy PaBphP42,44,45, also display 4-helix bundles in the HK domain (Figs. 4c and 5b), albeit sometimes smaller than that of the SaBphP2 (Fig. 5a). Recently, a Pr/Pfr heterodimer was detected in the thermal back reaction of the PaBphP from the (metastable) Pr/Pr homodimer to the Pfr/Pfr resting state, but the HK domain is not well resolved46–48. Note, a crossover of the linker helices near the PCM can be observed in the PaBphP and SaBphP2 Pr/Pr homodimers (Figs. 4c and 5b), yet the HK lobes in the PaBphP are already positioned at a large angular offset compared to those in SaBphP2 (Fig. 5c). Based on the rotations identified here, a sliding mechanism as proposed for the PaBphP42 (Fig. 4 e) could be re-evaluated, since a rotation would maintain the integrity of the 4-helix bundle (compare Fig. 4c, f). The unwinding of coiled-coil helices might not be restricted to the SaBphP. Other enzymes, in particular motor proteins such as the kinesin49 and others50,51, also display coiled coils. These molecular motors may use an analogous unwinding mechanism with a concomitant “motor-head” oscillation, which allows them to function efficiently. Structural overlay. a SaBphP2-Pr/Pfr4ms (green) overlayed on the Pr/Pfr heterodimer (magenta) determined by Malla et al.34. The BphP architecture is shown on the left. The kink in the linker helix of the inactive protomer is marked. b SaBphP2-Pr/PrLL (red, blue) overlayed on the light-induced, metastable Pr/Pr homodimer (orange, yellow) of the P. aeruginosa bathy BphP17,44,45. The linker helices cross near the PCM in both structures. c Same as (b), but viewed from the top. The HK lobes differ in their orientation in both structures (blue arrow: SaBphP2, red arrow: PaBphP). The bacteriophytochrome photocycle The Pr/Pfr heterodimer is central to the photoreactions in the red and far-red branches of the photocycle shown in Fig. 6b, c, respectively. The absorption of a photon results in an electronically excited state of the BV chromophore in its particular protomer, which can then photoconvert, driven by BV isomerization. In the red-branch of the photocycle (Fig. 6b), the BV isomerizes to the E-configuration, resulting in the formation of the Pfr state. In the far-red branch of the photocycle (Fig. 6 c), isomerization continues towards the Z-configuration and the formation of the Pr state. It requires two absorption events to complete the photoreaction in each branch. The discovery of the cryo-EM structure of a Pr/Pfr heterodimer in an unmodified, wild-type BphP with HK activity34 is relevant for the understanding of the photocycle. Previously, a Pr/Pfr heterodimer has been predicted for plant phytochromes52 and observed in a crystal structure of a mutant BphP with diguanylyl cyclase enzymatic domain53,54. The integration of the heterodimer in the BphP photocycle (Fig. 6a) has now been confirmed by a controlled excitation of single wild-type BphP particles. In addition to the rather large structural changes in the PCM region of SaBphP2 (Table 2), the conversion to the Pr/Pfr heterodimer is accompanied by the rotation of the entire HK domain. Only after this rotation, the kink observed in the linker helix of the opposing protomer (the one that is not photo-activated) appears (Fig. 5a). Structural changes of the inactive protomer, therefore, are allosteric by nature. We propose that these allosteric changes may have an impact on the photoconversion quantum yield of the inactive protomer that might even be time-dependent during the relaxation of the activated protomer to Pfr. In any case, the completion of the half-photocycle from Pr to Pfr requires two sequential absorption events (Fig. 6b). The same should be true for the Pfr to Pr photoconversion with the presence of the Pr/Pfr heterodimer in the photocycle (Fig. 6c). The modified photocycle and energy diagrams describing the photoactivation of prototypical BphPs. a The modified photocycle. The structure of the Pr/Pfr state is displayed in the center. Sequential Z to E isomerizations of the two BV chromophores occur by illumination with red light (660 nm). The Lumi-R states are the first metastable states after isomerization. After illumination with far-red light (750 nm), E to Z isomerizations of the two BV chromophores occur. The ~180o rotations of the HK domains are shown schematically by green arrows. b, c: Energy diagrams of the red and far-red absorbing parts of the photocycle. The Pr/Pr, Pr/Pfr, and Pfr/Pfr states are shown in red, orange, and dark-red, respectively. Electronically excited BV chromophores are denoted by a star. Photon activation is shown by black arrows. Isomerizations are shown by green arrows. b The Pr to Pfr transition requires absorption of red light (e.g., 660 nm). (1) and (2) mark the absorption of a photon by the two BV chromophores, each. c The Pfr to Pr transition requires absorption of far-red light (750 nm), which results in the recovery of the (dark adapted) Pr/Pr homodimer. In b, c, thermally activated relaxations shown by dashed arrows. Physiological implications Our structures provide a hypothesis for the existence of a stabilized Pr/Pfr heterodimer in BphPs from soil-dwelling myxobacteria and maybe also in related BphPs. Depending on the predominant environmental stimulus, red or far-red light, this heterodimer could rapidly shift the signaling output by favoring either the enzymatically less active Pfr state (Fig. 6b) or the more active Pr state (Fig. 6c), respectively. Under low light intensity, prevalent in the myxobacteria’s soil habitat, it is likely advantageous to the bacteria to react quickly using the Pr/Pfr species, sensitive to both red and far-red light. Signal generation also competes with thermal conversion (Fig. 6b) from Pfr to Pr, which interferes with red/far-red light perception. Starting from the Pr/Pfr state rather than Pr/Pr, responses to various wavelengths of light could become quick and finely tuned. The accumulation of the Pr/Pfr heterodimer in the photocycle (Fig. 6a) then depends on the ratio of red and far-red light in the myxobacteria’s habitat. It is important to explore to which of the three states the RR binds and when it dissociates. The large rotation of the HK moiety, as modeled by the complete cryo-EM structures of SaBphP2 in two distinct signaling states, suggests that the heterodimer is intricately involved in the binding (and dissociation) mechanism of the RR. TR cryo-EM using chemical diffusion41,55,56 through mixing the RR with the wild-type SaBphP2 can shine light on the mechanism of this interaction. Methods Sample preparation Full-length (intact) SaBphP2 is overexpressed and purified as previously reported34. The pure BphP is shipped on dry ice to the New York Structural Biology Center (NYSBC), where it is stored at −80 oC. An aliquot is thawed to prepare cryo-EM grids with the Spotiton technique41,57. Time-resolved pump-freeze experiment with Spotition (i), light activation Spotiton was developed at NYSBC41,57 and was made available for our experiments. To avoid unintentional light activation of the SaBphPs, a lightproof enclosure was built around the Spotiton device, and a 540 nm safety light was used within the enclosure during grid preparation. Before the experiment, the SaBphP2 solution (0.5 mg/ml) was irradiated with 740 nm light for 1 min using a light-emitting diode to generate a pure Pr state. 5 μl of the solution (containing the SaBphP2 Pr homodimers) was loaded into the Spotiton sample cups, and 3 µl was aspirated into each dispense head. The dispensers delivered approximately 5 nl of solution to a self-wicking grid, which were applied in overlapping stripes. The grid was then plunged into liquid ethane for vitrification (Figs. 7 and 8). To photoexcite the sample, we installed a 75 mW Coeherent OBIS cw-laser (660 nm) into the Spotiton device immediately above the ethane reservoir. The laser beam was focused to a 3 mm spot centered on the grid plunge axis. The laser light illuminated the grid on its way to the liquid ethane for approximately 4 ms (Supplementary Tables S1 and S2). Geometry of the Spotiton “pump-freeze” experiment. Laser illumination (660 nm, 75 mW) is performed using a fiber that produces a 3 mm spot at the grid plunge axis 9 mm from the fiber tip. The laser beam center is 7 mm above the liquid ethane surface. Approximately 10 nl of SaBphP2 is applied “on-the-fly” (not depicted) and rapidly thins on a self-wicking grid plunging at 1 m/s. The grid is then decelerated linearly (red upside-down triangle) starting just before the laser beam position and 25 mm above the final stop position in ethane (“GRID STOP”). Inset: Illumination of the grid begins at point S1, where the leading edge enters the beam, and concludes at point S4, when the trailing edge exits the beam. Points S2 and S3 indicate entry and exit, respectively, of the grid midpoint from the beam. Grid is under maximum illumination (yellow-bordered ellipse) at point SL. The grid is schematically shown by the brown ellipses; the tweezer is not depicted. Time-displacement curve of a laser-illuminated grid plunged using Spotiton. Based on Eq. 1, grid positions at various locations within the laser spot are calculated. Initial grid velocity is 1 m/s, and the deceleration is 20,000 mm/s2. Deceleration starts at 0 mm displacement as shown in Supplementary Table S2, reference frame 2. Key points along the plunge path are indicated. Time-resolved pump-freeze experiment with Spotiton (ii), timing The pump-freeze time delay depends on the distance of the laser spot from the ethane surface, the velocity of the plunger, and its deceleration (Figs. 7 and 8). The pump-freeze time delay is given by the “illumination to ethane contact” distance, which is 7 mm. The time resolution can be defined as the shortest time between light excitation and freezing, which is, in our case, approximately 5 ms when the laser beam is positioned as close as possible to the ethane reservoir. The pump-freeze time-delay as well as the time-delays to characteristic points on the plunger trajectory (Supplementary Table S2) can be calculated (Eq. 1) from the Spotiton plunger velocity which is u = 1 m/s, the point of the onset of the deceleration (25 mm above the resting point (stop) in the ethane, Figs. 7 and 8, and Supplementary Table S2) and the deceleration constant a which is 20,000 mm/s2. As an example, with the laser located at 7.2 mm from the start of the deceleration, the time when the grid reaches the laser midpoint can be calculated. Equation 1 yields7.2=1000⋅t−1220000⋅t2, with t in seconds. The two roots of the quadratic equation are 7.8 ms and 92 ms. Since the latter is unphysical, 7.8 ms is the correct time-delay. The time delays for the other points of interest are tabulated in Supplementary Table S2 and displayed in Figs. 7 and 8. Referring to Fig. 8 and Supplementary Table S2, a more generous estimate for the illumination time can be given by the time interval between the S4 and S1 positions, which is 7.1 ms. A more conservative estimate would yield 3.6 ms (4 ms for simplicity) between the S3 and S2 positions. The time between the light excitation and the ethane contact is 9.4 ms. We report this as the time delay in this paper. The time resolution is defined as the shortest time possible with this setup. It is dependent on the shortest distance possible between the laser focus and the ethane reservoir. This shortest distance is about 4 mm. Given the velocity of the plunger near the ethane surface, the time resolution is approximately 4–5 ms (Fig. 8). In summary, we measure a 9.4 ms time delay with 5 ms time resolution following 4 ms laser excitation. Phytochrome excitation Illumination conditions are listed in Supplementary Table S1. During the 4 ms exposure, a total of 1 × 1015 660 nm photons reach the grid. The photon density at the grid is 1.42 × 1016 photons/cm2. Using the Pr spectrum of the full-length SaBphP2 (Supplementary Fig. S3), the molar extinction coefficient for SaBphP2 is roughly estimated to be ε700 = 80,000 cm2/mmol (see the SM). Based on the spectral form, at 660 nm, the extinction coefficient is about 52% of that at 700 nm, which amounts to ε600 = 41,600 cm2/mmol. Accordingly, the molecular absorption cross-section of a protomer at 660 nm is ln(10)×41,600/(6.024×1020)cm2=1.6×10−16cm2. With a photon density of 1.42 × 1016 photons/cm2, approximately Np = 2.3 photons are absorbed per protomer during the 4 ms illumination time. Absorption of a photon results in an electronically excited BV chromophore. The excited chromophore relaxes through a conical intersection seam58,59 either back to the dark state or towards the photocycle. The probability to enter the photocycle is given by the quantum yield of photoconversion qpc. With a given qpc, the probability p of a successful photoconversion increases with the number of photon absorption events. It can be calculated as with Np the number of photons absorbed. The term in square brackets denotes the probability that a molecule is not in the photocycle after independently absorbing Np photons. This equation is true for light sources such as our cw-laser and may break down for ultrashort, high-intensity, pulsed laser light sources. Equation 2 is dependent on the quantum yield qpc, which needs to be determined (see below). Determination of the quantum yield of photoconversion With an experimentally set up consisting of a low-power light-emitting diode and a spectrophotometer, the qpc can be estimated by measuring the change in the absorption ΔOD for a period of time t and determining the slope ΔOD/t as listed in Supplementary Table S1. Details are outlined by Slavov et al.60, who specify a formula that makes use of this slope to determine the quantum yield. The formula, though, only applies if the spectral overlap of the reactants involved is negligible. However, the Pfr state that results from the photoconversion has a very strong contribution to the absorbance change at the measuring wavelength (700 nm). The formula needs corrections, which are detailed in the SM (Supplementary Equations S1–S6, and Supplementary Fig. S4). In summary, a correction term (1 − 0.51) is introduced to the denominator of the original equation yielding where the optical density change (ΔOD) is measured at the detection wavelength λdetect (here 700 nm). Vol is the volume of the solution in the cuvette, Na is the Avogadro number, P is the power of the exciting LED at λext = 660 nm wavelength, and A0 is the absorbance at the excitation wavelength. To determine A0, the absorbance at 660 nm can be measured with the spectrometer, but must be multiplied by the height of the solution column that is excited, which is here 3 cm. Na, h, c, and t are Avogadro’s number, the Planck constant, the speed of light and the time (in s), respectively. Supplementary Table S1 provides the parameters for the excitation experiment on SaBphP2. The quantum yield is qpc = 12%. With this quantum yield, the probability p to excite a protomer with 4 ms laser light illumination is 25%. This means that (1−0.25)2⇒57% Pr/Pr homodimers,2⋅0.25⋅(1−0.25)⇒37%Pr/Pfr heterodimers and 0.252⇒6% Pfr/Pfr homodimers are estimated to be present after the 4 ms illumination. Equations 2 and 3 are implemented in a short Python script61. Figure 9 shows species concentrations obtained as a function of the illumination time, which were calculated by this script. No more than 50% Pr/Pfr heterodimer can be generated. With the given Spotiton settings (Fig. 8) and the photon density at the grid, the above heterodimer fraction is obtained, which is sufficient to be detected in the TR cryo-EM data. The script can also be used to demonstrate that the amount of the Pr/Pfr species obtained after light excitation is robust against quite large variations of the extinction coefficients, so that the rough estimate of the absorption coefficients presented here is sufficient. Estimates of species concentrations. Fractions of the Pr/Pr homodimer (black circles), Pr/Pfr heterodimer (red squares), and the Pfr/Pfr homodimer (green triangles) are shown as a function of illumination time, given the laser illumination conditions listed in Supplementary Table S1. The data were globally fit either by a single exponential (black line) or a sum of two exponentials (red and green lines, respectively). A maximum of 50% Pr/Pfr heterodimers is expected at about 9.5 ms illumination (horizontal dashed line). At 4 ms (vertical blue dashed line) with the illumination as achieved here, a substantial fraction of the Pr/Pfr species is generated. Data collection Cryo-EM Movies were collected from the Spotiton prepared grids as reported in detail62 using a Thermo Fisher Scientific Titan Krios electron microscope controlled by the Leginon software system63. The movies were sent from the NYSBC to the UWM High Performance Compute clusters, where they were processed further (Table 1). Processing of the TR-cryo-EM data CryoSPARC v. 4.7.164 was used throughout to process the movies. Some details to analyze cryo-EM data of full-length BphP particles are also reported previously34,62. Data collection statistics for the Pr/Pfr heterodimer is shown in Tab. 1, a flow chart is shown in Supplementary Fig. S1. Initially, over 6.5 M particles were picked from over 7000 micrographs using the blob picker. 8506 of the best particles were used to produce a template that was used to pick about 5 M particles, of which 3.3 M particles were retained. After 2D classification and further curation, 379,306 particles remained. From these, a single-class reconstruction was generated. Non-uniform refinement produced a 6.4 Å map that was used again as a template to pick particles. After curation, 106,437 particles remained, a ab-initio volume was generated. Non-uniform refinement produces a 6.7 Å map. Out of these 106,437 particles, 69,189 particles were used to train a Topaz neural network65. Topaz picking yielded 383,989 particles. The ab-initio volume was further improved by using the Topaz picked particles, resulting in a 5.98 Å map. To address the mixture of Pr/Pr and Pr/Pfr particles in the ensemble, we classified the picked particles into 4 3D classes using a 7 Å resolution filter (Supplementary Fig. S1b). After homogenous and non-uniform refinement, the Pr/Pfr volume was present in class 1 with about 60,294 particles and the Pr/Pr volume was detected in class 0 with 55,977 particles (Table 1). We did not refine the Pr/Pr volume further, since more particles were contributing to the 3D volume in an earlier experiment39,66, and we expected higher resolution and better map quality from these particles. In order to obtain a volume for the Pr/Pfr species with better resolution, we started again from the Topaz-picked particles. 3D classification was achieved with a 3 Å resolution filter into 3 classes. After rejecting one class, down-sampling to a smaller, tighter (350 pixel) box and another round of 3D classification, a volume for the Pr/Pfr species was determined from 87 k particles at 5 Å resolution (Table 1). The resolution in the PCM region and that of the HK domain are listed in Table 1. The Fourier Shell Correlation to determine the resolution limit and the local resolution map is shown in Supplementary Fig. S1d, e. Revisiting cryo-EM data for the SaBphP Pr/Pr homodimer The cryo-EM data of a Pr homodimer reported by Karki et al.39,66 were revisited (Table 1 and Supplementary Fig. S2). The earlier experiment provided more particle images from which we reconstructed a higher resolution structure than from the limited number of Pr images collected in this study. Spotiton was also used to prepare the grids, but the sample was only illuminated with a light-emitting diode, resulting in a fraction of >85% of unexcited Pr/Pr homodimers. Because of the insufficient light illumination, we call the resulting structure SaBphP2-Pr/PrLL (LL for low light). 10207 raw movies were imported and motion corrected. Based on the contrast transfer function (CTF) fit resolution, relative ice thickness, and total full-frame motion distance, 8101 micrographs were selected (Supplementary Fig. S2). We used the previously determined volume39,66 to generate templates for repeating the particle picking. From template-based picking, 1,955,847 particles were extracted with a box size of 512 pixels. After iterative 2D curation, 363,029 particles were retained. Topaz was used for particle picking, yielding 251,896 particles. The particles from Topaz and template-based picking were down-sampled to a 350-pixel box. These particles underwent iterative 2D classification, resulting in 214,432 particles. A single ab initio volume was generated, followed by homogeneous and non-uniform refinement. The particles were then classified into three 3D-classes (Supplementary Fig. S2b). Class 0 with 70,908 particles was used for non-uniform refinement, resulting in a 5.85 Å map. The local resolution map and the Fourier Shell Correlation are shown in Supplementary Fig. S2d. By applying local refinement in CryoSPARC, the resolution in the HK domain region improves to >6 Å (Table 1). Structure determination and structural refinement To determine the structure of the SaBphP2-Pr/Pfr4ms heterodimer, the structural model of the Pr/Pr homodimer determined by Malla et al.34,67,68 was used as the initial model to exclude bias. The PCM of the initial model and part of the linker helices were determined at near atomic resolution and, hence, include side chains. Although the resolution of the SaBphP2-Pr/Pfr4ms density was too low to assign side-chain positions, we kept the side chains in the initial model. The BV chromophore was removed from protomer B. Since there are large structural changes between the homodimer and the heterodimer, the initial model required extensive modifications of the protomer B structure to agree with the cryo-EM density. Judged by the separation of strands and the α-helix formation in the sensory tongue region, protomer B is clearly in the Pfr state with the BV in the E-configuration. Accordingly, we added the BV structure with ring-D in the E-configuration obtained from the Pfr protomer in the Pr/Pfr heterodimer determined earlier34–38 and adjusted the chromophore pocket residues accordingly. To complete the model, the HK region of the Pr/Pfr heterodimer that was also determined in the earlier study was added to the initial model. Note, this part, which includes a fraction of the linker helices, the four-helix bundle and the HK lobes, does not contain side chains but consists only of a poly-alanine protein backbone. Proline residues were also included since they can be considered part of the backbone. The complete model was refined against the Pr/Pfr density map using real-space refinement implemented in the program Phenix69 (Table 1). To obtain a model for the Pr/Pr state, the same incomplete Pr/Pr structure used above was oriented into the new cryo-EM density map of the Pr/Pr homodimer. Modifications were not necessary. To complete the structure, the linker region, the 4-helix bundle and the HK lobes were added, guided by the structures of the corresponding regions in the Pr/Pfr heterodimer. However, this time, the individual lobes of the HK are located on top of the PCM region of the same, corresponding protomer (Fig. 2a). This is in contrast to the model of the Pr/Pfr heterodimer, where the HK lobes are positioned on top of the opposite protomer (Figs. 2b and 5a). The resulting full-length Pr/Pr homodimer (SaBphP2-Pr/PrLL) was also refined with Phenix. Refinement and validation statistics are also provided in Table 1. Online content Any methods, additional references, Nature Portfolio reporting summaries, source data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https (to be concluded after submission). Supplementary information Transparent Peer Review file Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Movie 1 Supplementary Movie 2. Supplementary Movie 3 Acknowledgements The research was funded by NSF-2423601 and NSF-2423602. NYSBC is funded by Simons Foundation Grant SF349247, National Institute of General Medical Sciences, Grant GM103310, and the New York State Assembly Majority. We thank Niko McCracken and Dan Sierks (UWM) for helping with the UWM HPC cluster. Author contributions P.K. processed the cryo-EM data with the help of C.H., J.H.M., T.N.M., and P.S. W.C.B. and R.K. built the setup in Spotiton, prepared the grids and executed the pump-freeze cryo-EM experiments. C.H., R.K., W.C.B., and J.H.M. collected the cryo-EM data. E.A.S., S.K., and D.M. provided purified samples of intact SaBphP2. C.S., M.Si., and M.S. estimated molecular excitations, E.A.S., M.S., and W.C.B. designed the experiment. P.K. and M.S. interpreted the data. M.S., P.K., and E.A.S. wrote the manuscript with input from all other authors. Peer review Peer review information Communications Chemistry thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available. Data availability The SaBphP2-Pr/Pfr4ms and SaBphP2-Pr/PrLL structures are deposited in the protein data bank70 with entries 9YXN and 9YS3, respectively. The structural models and their corresponding cryo-EM density volumes are deposited in the EMDB71 with accession codes EMD-73612 and EMD-73380, respectively. The source data for Fig. 9 are in Supplementary Data 1. Code availability Python code to calculate the qpc, the probability of photoactivation and the relative concentrations of the Pr/Pr, Pr/Pfr, and Pfr/Pfr species using the experimental conditions and variables listed in Supplementary Table S1 can be obtained61 zenodo.org/records/20313451 from Zenodo72. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Emina A. Stojković, Email: e-stojkovic@neiu.edu Marius Schmidt, Email: smarius@uwm.edu. Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s42004-026-02084-6. References Associated Data Supplementary Materials Transparent Peer Review file Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Movie 1 Supplementary Movie 2. Supplementary Movie 3 Data Availability Statement The SaBphP2-Pr/Pfr4ms and SaBphP2-Pr/PrLL structures are deposited in the protein data bank70 with entries 9YXN and 9YS3, respectively. The structural models and their corresponding cryo-EM density volumes are deposited in the EMDB71 with accession codes EMD-73612 and EMD-73380, respectively. The source data for Fig. 9 are in Supplementary Data 1. Python code to calculate the qpc, the probability of photoactivation and the relative concentrations of the Pr/Pr, Pr/Pfr, and Pfr/Pfr species using the experimental conditions and variables listed in Supplementary Table S1 can be obtained61 zenodo.org/records/20313451 from Zenodo72.