Comparative analysis of two paradigm bacteriophytochromes reveals opposite functionalities in two-component signaling

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ Comparative analysis of two paradigm bacteriophytochromes reveals opposite functionalities in two-component signaling © The Author(s) 2021 Published version Multamäki, Elina; Nanekar, Rahul; Morozov, Dmitry; Lievonen, Topias; Golonka, David; Wahlgren, Weixiao Yuan; Stucki-Buchli, Brigitte; Rossi, Jari; Hytönen, Vesa P.; Westenhoff, Sebastian; Ihalainen, Janne A.; Möglich, Andreas; Takala, Heikki Multamäki, E., Nanekar, R., Morozov, D., Lievonen, T., Golonka, D., Wahlgren, W. Y., StuckiBuchli, B., Rossi, J., Hytönen, V. P., Westenhoff, S., Ihalainen, J. A., Möglich, A., & Takala, H. (2021). Comparative analysis of two paradigm bacteriophytochromes reveals opposite functionalities in two-component signaling. Nature Communications, 12, Article 4394. https://doi.org/10.1038/s41467-021-24676-7 2021 ARTICLE Comparative analysis of two paradigm bacteriophytochromes reveals opposite functionalities in two-component signaling Elina Multamäki 1, Rahul Nanekar2, Dmitry Morozov 3, Topias Lievonen2, David Golonka 4, Weixiao Yuan Wahlgren5, Brigitte Stucki-Buchli 2, Jari Rossi 1, Vesa P. Hytönen 6,7, Sebastian Westenhoff5, Janne A. Ihalainen 2✉, Andreas Möglich 4& Heikki Takala 1,2✉ Bacterial phytochrome photoreceptors usually belong to two-component signaling systems which transmit environmental stimuli to a response regulator through a histidine kinase domain. Phytochromes switch between red light-absorbing and far-red light-absorbing states. Despite exhibiting extensive structural responses during this transition, the model bacteriophytochrome from Deinococcus radiodurans (DrBphP) lacks detectable kinase activity. Here, we resolve this long-standing conundrum by comparatively analyzing the interactions and output activities of DrBphP and a bacteriophytochrome from Agrobacterium fabrum (Agp1). Whereas Agp1 acts as a conventional histidine kinase, we identify DrBphP as a light-sensitive phosphatase. While Agp1 binds its cognate response regulator only transiently, DrBphP does so strongly, which is rationalized at the structural level. Our data pinpoint two key residues affecting the balance between kinase and phosphatase activities, which immediately bears on photoreception and two-component signaling. The opposing output activities in two highly similar bacteriophytochromes suggest the use of light-controllable histidine kinases and phosphatases for optogenetics. https://doi.org/10.1038/s41467-021-24676-7 OPEN 1Faculty of Medicine, Anatomy, University of Helsinki, Helsinki, Finland. 2Department of Biological and Environmental Science, Nanoscience Center, University of Jyvaskyla, Jyvaskyla, Finland. 3Department of Chemistry, Nanoscience Center, University of Jyvaskyla, Jyvaskyla, Finland. 4Lehrstuhl für Biochemie, Universität Bayreuth, Bayreuth, Germany. 5Department of Chemistry and Molecular Biology, University of Gothenburg, Gothenburg, Sweden. 6Faculty of Medicine and Health Technology, BioMediTech, Tampere University, Tampere, Finland. 7Fimlab Laboratories, Tampere, Finland. ✉email: janne. ihalainen@jyu.fi;heikki.p.takala@jyu.fi NATURE COMMUNICATIONS | (2021) 12:4394 | https://doi.org/10.1038/s41467-021-24676-7 | www.nature.com/naturecommunications 1 1234567890():,; Two-component signaling systems are mainly found in prokaryotes and allow cells to respond to environmental signals1. These systems have been under extensive research ever since their discovery, as they control a wide range of cellular mechanisms from enzymatic activity to transcription regulation2. A canonical two-component system consists of a homodimeric sensor histidine kinase (HK) and its cognate response regulator (RR)3. To the extent it has been studied, most HK proteins sense chemical signals and generally reside within the plasma membrane4. The output activity is exerted by an intracellular HK module, consisting of two subdomains: a dimerization histidine phosphotransfer (DHp) domain, and a catalytic ATP-binding (CA) domain. Based on their DHp sequence and according to Pfam, the HK proteins can be divided into five subtypes, called HisKA, HisKA_2, HWE_HK, HisKA_3, and His_kinase4,5. The subjects of the current study, the bacteriophytochromes from Deinococcus radiodurans (DrBphP) and Agrobacterium fabrum (Agp1), both fall within the HisKA family. The HK catalyzes autophosphorylation and subsequent phosphotransfer to the cognate RR. During the autophosphorylation reaction, the eponymous histidine of the DHp domain is phosphorylated6,7, either within the same monomer (cis) or the sister molecule of the homodimer (trans)8. In the phosphotransfer reaction, the phosphate is relayed to a conserved aspartate residue within a receiver (REC) domain of the RR. This reaction entails RR activation and elicits output responses such as altered gene expression3,6,9. HKs may also act as phosphatases that hydrolyze the phosphoaspartyl bond in the phosphorylated response regulator, thus resetting the two-component system10,11. Whereas the kinase activity has been extensively studied12, the importance of the phosphatase activity has been appreciated more recently13–15. In two-component systems, a dynamic balance between kinase and phosphatase activities determines the net output and downstream physiological effects. The underlying kinase-active and phosphatase-active conformational states are necessary for balancing the output activity of the two-component system16,17. In contrast to the typical transmembrane HK receptors, lightsensitive receptors are frequently soluble. This facilitates their structural and mechanistic analyses4,18,19. As a case in point, phytochromes are red/far-red light-sensing photoreceptors that regulate diverse physiological processes in plants, fungi, and bacteria, e.g., chromatic adaptation and phototaxis in prokaryotes20,21.Plant phytochromes exert downstream physiological responses via lightdependent interactions with partner proteins, nucleocytoplasmic shuttling and protein degradation22,23. By contrast, bacterial phytochromes (BphPs) usually belong to two-component signaling systems, with a cognate response regulator commonly encoded in thesameoperon 18,24,25. BphPs contain an N-terminal photosensory module (PSM), divided into PAS (period/ARNT/single-minded), GAF (cGMP phosphodiesterase/adenylyl cyclase/FhlA) and PHY (phytochrome-specific) domains20. The PSM binds a biliverdin IXα chromophore via a thioether linkage to its conserved cysteine within the PAS domain26,27. The PSM is followed by a C-terminal output module, most commonly a HK domain. Photoactivation by red and far-red light drives biliverdin Z/E isomerization, which underlies the phytochrome switch between its red light-absorbing (Pr) and far-red light-absorbing (Pfr) states24. In darkness, phytochromes can thermally revert to their resting state which is the Pr state in canonical phytochromes28. As first demonstrated for the model bacteriophytochrome DrBphP from D. radiodurans, light induces extensive structural changes in the photosensory module that are relayed to the output module29. In the cyanobacterial phytochrome Cph130–33 and several bacteriophytochromes34,35, the dark-adapted Pr state exhibited higher kinase activity than the Pfr state. In particular, the Agp1 bacteriophytochrome from A. fabrum (also known as AtBphP1, based on the former species designation A. tumefaciens) displays histidine kinase activity in its resting Pr state28,36; in the Pfr state, the autophosphorylation and phosphotransfer reactions are downregulated by 2-fold and 10-fold, respectively28. The kinase activity of Agp1 has been shown to control bacterial conjugation37. Although DrBphP has been implicated in the control of carotene production21, no kinase activity has been demonstrated for DrBphP, notwithstanding close sequence homology and the elaborate structural changes this receptor undergoes under light24,29. Despite the eminent role of DrBphP as a paradigm for photoreception, the enzymatic activity and hence the exact physiological role of this model phytochrome have hence remained enigmatic. Here, we unravel this long-standing puzzle by studying the enzymatic activity and interactions of DrBphP and Agp1, as two canonical bacteriophytochromes with HK effector domains. By pursuing an integrated biochemical and structural strategy, we show that despite close homology, Agp1 acts as a histidine kinase whereas DrBphP functions as a light-activated phosphatase. Our biochemical and structural data pinpoint two key residues proximal to the catalytic histidine that affect the balance between the kinase and phosphatase activities. Together, the two phytochromes provide soluble, light-controllable systems with opposite activities for the study and application of two-component signaling. Results The dark reversion of DrBphP is affected by DrRR.We employed UV-vis absorption spectroscopy to investigate whether the cognate response regulators interact with DrBphP and Agp1 and potentially affect the photoactive states of these bacteriophytochromes. For reference, we also generated a hybrid receptor, denoted as Chimera, which comprises the DrBphP PSM and the Agp1 HK domain (Fig. 1a). DrBphP, Agp1, and Chimera all showed typical absorption spectra with Soret and Q-band absorption peaks for both the Pr and Pfr states, which were unaffected by the addition of the cognate RR (Fig. 1b). The thermal reversion of phytochrome samples after applying saturating red light (655 nm) exhibited multiple exponential phases in all cases, irrespective of the presence of the RR (Supplementary Fig. 1a). The recovery of Agp1 was faster than that of DrBphP, while that of the Chimera was between those of DrBphP and Agp1. Earlier studies indicated that the dark reversion in phytochromes is affected by the dimerization interfaces in both the PSM and HK domain38. In line with this notion, the spectral characteristics of the Chimera are evidently governed by both the Agp1 HK and DrBphP PSM. The dark reversion kinetics of Agp1 and Chimera were unaffected by the response regulator from Agrobacterium fabrum (AtRR1), but that of DrBphP was significantly accelerated by Deinococcus radiodurans response regulator (DrRR). This finding indicates that DrRR binds to the DrBphP HK, thereby favoring the Pr state conformation. Interestingly, this contrasts with the Arabidopsis thaliana phytochrome B, where the binding of the phytochrome-interacting factor (PIF) stabilizes the Pfr state39. DrBphP interacts with DrRR more strongly than Agp1 does with AtRR1. To further analyze the interaction between the phytochromes and their RRs, we applied size-exclusion chromatography on fluorescently labeled RR proteins (Fig. 2a). Indicative of binding, the addition of DrBphP caused a shift in the retention of EGFP-labeled DrRR towards lower volumes. The observed interaction was independent of the photoactivation of ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24676-7 2NATURE COMMUNICATIONS | (2021) 12:4394 | https://doi.org/10.1038/s41467-021-24676-7 | www.nature.com/naturecommunications DrBphP. By contrast, the EGFP-AtRR1 retention was not significantly affected by the presence of Agp1, suggesting no or a weak interaction between Agp1 and AtRR1 (Fig. 2a). See Supplementary Fig. 1c, d for additional measurements. To further investigate the BphP/RR interactions, we resorted to surface plasmon resonance (SPR). The changes in the SPR signal were measured for the response regulator immobilized on the SPR chips while flowing the phytochromes across the sensor surface. The binding of DrRR to DrBphP in the Pr state was evaluated from the steady-state saturation signal (Fig. 2c), resulting in a dissociation constant K D of (43 ± 8) µM when using a 1:1 molar binding model (Supplementary Fig. 2i). This value was verified by Langmuir kinetic analysis which yielded an affinity of comparable strength (K D ~ 10 µM). A lower signal amplitude and a K D value of (60 ± 7) µM for the DrBphP/DrRR pair were observed upon red-light application. Notably, the slightly weaker binding in the Pfr state concurs with the above spectroscopic measurements where DrRR binding favors the Pr state. Consistent with the SEC analysis, the interaction between Agp1 and AtRR1 was substantially weaker, and the binding curve did not reach saturation at the highest achievable Agp1 concentration of 154 µM. We hence estimated the affinity to be on the order of hundreds of micromolar. The shape of the SPR response graph indicates that the association and dissociation kinetics of the Agp1/AtRR1 are fast, which precluded the kinetic evaluation. That notwithstanding, the Agp1/AtRR1 interaction was not notably affected by red light (Supplementary Fig. 1b). As a complementary method, we applied isothermal calorimetry (ITC). The DrBphP/DrRR interaction could be described by a 1:1 molar binding model with a K D of (8.1 ± 1.3) µM (Fig. 2b). Unlike in a blue light-regulated HK40, this interaction was not affected by the addition of the ATP analog AMP-PNP (Supplementary Fig. 2b). Agp1 binding to AtRR1 could not be reliably detected by ITC (Fig. 2b), consistent with the SEC data and the fleeting binding seen in SPR. The binding parameters were similar in a different buffer condition (Supplementary Fig. 2a, i). Furthermore, cross-interaction was neither detected between DrBphP and AtRR1 nor between Agp1 and DrRR (Supplementary Fig. 2g). Taken together, the interactions of DrBphP and Agp1 with their cognate response regulators were clearly different. Next we studied whether these differences correlate with enzymatic activity, as we speculated that the function of these phytochromes is reflected in their interactions. Agp1 functions as a histidine kinase but DrBphP acts as a phosphatase. We characterized the kinase activity of the bacteriophytochromes by 32P-γ-ATP autoradiography (Fig. 3a). The autophosphorylation reaction of Agp1 occurred preferably in the Pr state and was reduced under red light illumination, consistent with previous reports36. If AtRR1 was present, it received a phosphate from Agp1 in the phosphotransfer reaction. This reaction occurred preferably in the dark-adapted Pr state but was almost absent under constant red-light illumination (i.e., in the Pfr state). This verifies that Agp1 binds to and transfers its phosphate to AtRR1 in its kinase-active Pr state. Intriguingly, DrBphP lacked autokinase or phosphotransfer activities in both the Pr and Pfr states (Fig. 3a). The absence of kinase activity is surprising as the DrBphP and its PSM evidently undergo light-induced structural changes that seem to be conserved among other phytochromes29,41,42. Moreover, all homologous bacteriophytochrome HKs studied to date exhibited light-dependent kinase activity. The unusual absence of kinase activity in DrBphP could in principle be due to (1) lack of interaction with the DrRR; (2) inability of its PSM to transduce signals to the HK effector; or (3) inactivity of the DrBphP HK module. Scenario 1 can be ruled out according to the above results, which consistently showed interaction between DrBphP and DrRR. To address scenario 2, we assessed the histidine kinase activity of the Chimera and found it to function similarly to the wild-type Agp1 with robust autokinase and phosphotransfer activity in the Pr state, but reduced activity in the Pfr state (Fig. 3a). This result states that DrBphP undergoes productive a CA N knot PSM HK DHp P PAS GAF PHY CA DrBphP Agp1 Chimera PSM HK tongue b Absorbance Absorbance 500100 time [min] D R DrBphP+DrRR D R Chimera+AtRR1 D R DrBphP Chimera +AtRR1 -AtRR1 DrBphP +DrRR -DrRR Absorbance 1 0 0 1 1 0 Agp1 +AtRR1 -AtRR1 D R Agp1+AtRR1 A/A 750 700 A/A 750 700 A/A 750 700 500 700300 900 λ [nm]λ [nm] 500 700300 900 1 0 0 1 1 0 Agp1 Chimera D R D R Fig. 1 Overall architecture and UV-vis spectroscopy of DrBphP, Agp1, and their Chimera with and without their cognate response regulator. a Schematic representation of a canonical bacteriophytochrome with a histidine kinase (HK) effector domain. The site of the phosphorylated histidineis indicated as the letter P. In addition, a schematic presentation of the phytochrome chimera is shown, where the photosensory module (PSM) of DrBphP is combined with the HK domain of Agp1. Abbreviations: Period/ARNT/single-minded (PAS), cGMP phosphodiesterase/adenylyl cyclase/FhlA (GAF), phytochrome-specific (PHY), histidine kinase (HK), dimerization Histidine phosphotransfer domain (DHp), catalytic ATP-binding domain (CA). bThe absorption spectra of the BphP HKs with and without their cognate response regulators (RR) in dark (D) or under red light (R). The right-most panels show their dark reversion kinetics as an A 750 /A 700 ratio over time, where 0 min corresponds to the time the 655-nm illumination ceased. Source data are provided as a Source data file. NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24676-7 ARTICLE NATURE COMMUNICATIONS | (2021) 12:4394 | https://doi.org/10.1038/s41467-021-24676-7 | www.nature.com/naturecommunications 3 structural changes that are conducive to controlling HK activity, thereby ruling out scenario 2. To address scenario 3, Phos-tag gels were applied where unphosphorylated proteins and their phosphorylated counterparts are resolved based on migration through the gel matrix. In this analysis, unphosphorylated and phosphorylated response regulators were clearly separated from another (Fig. 3b). The assay confirmed that the wild-type Agp1 phosphorylates AtRR1 preferably in the Pr state and revealed that it crossphosphorylates DrRR with similar efficiency (Fig. 3b). However, like in the radiolabeling assay (Fig. 3a), DrBphP lacked kinase activity, as it could not produce phosphorylated DrRR (phosphoDrRR). The residue immediately following the catalytic histidine, denoted as H +1, is acidic in the majority of sensor histidine kinases and has been implicated in the autophosphorylation reaction12. Whereas Agp1 has Asp529 in this position and thus conforms to the prevalent sequence motif, DrBphP unusually possesses a histidine in the corresponding position 533 (Fig. 3g). To test the role of the H +1 position, we generated the Agp1 D529H and DrBphP H533D variants. The D529H mutation rendered Agp1 inactive (Fig. 3b), thus verifying the importance of this acidic residue for the kinase activity. Like the wild-type DrBphP, the H533D variant appeared inactive (Fig. 3b, Supplementary Fig. 4c). Therefore, this single mutation in the H +1 position is insufficient to rescue the kinase activity of DrBphP. As sensor histidine kinases may also function as phosphatases10,43,44, we tested the DrBphP and Agp1 HKs in that regard. Of particular advantage, the Phos-tag gels allow to assess the dephosphorylation of phospho-RR proteins. To this end, we generated the phosphorylated response regulators chemically by treatment with acetyl phosphate45. DrRR was ba DrBphP+DrRR (R) Response unit [RU] 0 100 200 300 400 DrBphP+DrRR (D) 0 400 conc. [μM] 2000 [RU] 0 400 conc. [μM] 2000 [RU] 0 50 100 150 200 Time [s] 250 0 50 100 150 200 Time [s] 250 Agp1+AtRR1 (D) 0 400 conc. [μM] 2000 [RU] 0 50 100 150 200 Time [s] 250 5 μM 134 μM 67 μM 33 μM 17 μM 8 μM 267 μM 134 μM 67 μM 33 μM 17 μM 8 μM 267 μM 77 μM 39 μM 19 μM 10 μM 154 μM Time [min] 02040 Molar ratio [N] 123 DrBphP+DrRR -20 0 ΔH [kJ/mol] -40 DP [μW] -4 0 -2 K = 8.1±1.3 μM D 0 Time [min] 20 40 Agp1+AtRR1 1234 Molar ratio [N] K = n.d. D Time [min] 02040 Molar ratio [N] DrRR only c A489 [arb. units] 12.5 0.5 1.0 0 Volume [ml] 21 Volume [ml] 1.5 1.5 22.5 EGFP-AtRR1 (57 kDa) Agp1 (D) Agp1 (R) (~300 kDa) A280 [arb. units] DrBphP (D) DrBphP (R) (~300 kDa) EGFP-DrRR (48 kDa) 0.5 1.0 EGFP-DrRR+DrBphP (D) EGFP-DrRR+DrBphP (R) EGFP-AtRR1+Agp1 (D) EGFP-AtRR1+Agp1 (R) 0 Fig. 2 Quantitative analyses of the BphP/RR interactions. a Size-exclusion chromatography (SEC) of the EGFP-labeled response regulators DrRR and AtRR1 in the absence (top) and presence (bottom) of DrBphP and Agp1, respectively. Vertical dashed lines indicate the retention volume of free RR monomer. Top panels: In isolation, EGFP-DrRR (45.4 kDa) and EGFP-AtRR1 (45.4 kDa) eluted as a monomer and a monomer/dimer mixture, respectively. DrBphP (84.0 kDa) and Agp1 (83.8 kDa) are known to be dimers, and their apparently high molecular weights (~300 kDa) can be explained by the poor resolution of large proteins in the conditions. Bottom panels: When combined with DrBphP in its dark-adapted state (D) or after 655-nm illumination (R), the profile for EGFP-DrRR shifted to shorter retention times, indicative of interactions. By contrast, addition of Agp1 had little effect on the retention of AtRR1. The top panels are plotted at 280 nm and the bottom panels at 489 nm. The A489 signals from BphPs were negligible. bIsothermal titration calorimetry (ITC) measurements. Differential power (DP) resulting from injections of the response regulator to the BphP is plotted against time, and the binding enthalpy (ΔH) is plotted against the molar ratio of the proteins. See Supplementary Fig. 2 for additional data and control measurements. cSurface plasmon resonance (SPR) measurements. The response regulators were coupled on the sensor surface, and varying concentrations of the corresponding phytochrome were applied in darkness (D) or after red-light illumination (R). The sensorgrams (black lines), the kinetic fits (blue lines), and the parts of the data that were used for kinetic analysis (orange) are indicated. For kinetic analyses, DrBphP concentrations of 8–67 µM were used. Green lines mark the R eq values that were used for evaluating the steady-state affinity data, shown in Supplementary Fig. 2i. Inset: Steady-state fit of the concentration series, where R eq values were used for affinity approximation. See Supplementary Fig. 2i for a table of SPR fitting values. Source data are provided as a Source data file. ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24676-7 4NATURE COMMUNICATIONS | (2021) 12:4394 | https://doi.org/10.1038/s41467-021-24676-7 | www.nature.com/naturecommunications phosphorylated robustly, whereas AtRR1 responded to the treatment weakly. Phospho-DrRR was then incubated together with ATP and either DrBphP or Agp1. In the reactions, net phosphatase activity would decrease the amount of phosphoDrRR, whereas net kinase activity would increase it. As expected, addition of Agp1 or Chimera led to an increase in phospho-DrRR when incubated in darkness, indicating kinase activity of these proteins (Fig. 3c). By contrast, the addition of DrBphP decreased a DRDR DrBphP DRDR DRDR Chimera Agp1 p-BphP p-RR BphP RR AtRR DrRR 32 γ-P protein AtRR g A K R F M V I L Q K E R I L S T E R Q M A L I V M V I F L I H N V I L V I L T E G D S N G S A A V L I D E Q R K H A F Y R M A G N S T K N Q R E D M A L K R E S T A L N A Q E K R K Q R A E Q V I A L A R Q K E A K Q R E K L S A F L S E D N K Q E R A R L Q D A E M Q R K L N K Q S R D E N S K A D R E Q V I M Y L F S F T I V L A G H Y S A D A M S T N V I M L A I V T A S H Q D E V M F I L N Q K R Q E A S N T P F M V I L G S R A N T G V T S A M L V I R K L M T N L A S G S A L Y F I V T L S A G R S D Q E A T V M I L A V M I L L K D R E N...N 511 550 641 657 DrBphP HH HDLNRALTQSNAEWRQYGFVIS MQEPVRLISQFAELLTR LLRDLLL LIGNALTFG 507 546 630 646 Agp1 DLTRELQRTNKELEAFSYSVSHDLRAPFRHIVGFAQLLRE LLRQVWYNLIENAIKYS H532 bits 4 3 2 1 0 H648 Kinase ... L..PHD Phosphatase ... .H...X b p-AtRR1 AtRR1 BphP DrRR BphP p-DrRR – D529H WT DrBphP Agp1 WT H533D +–––– – – D529H WT DrBphP Agp1 WT H533D +–––– – – acetyl-P 72 28 17 kDa – - - - - - - E536A RD RDRD – H533D WT DrBphP d +++++++ - - - Agp1 A532E RD RDRD D529H WT BphP AtRR1 p-AtRR1 ––––– acetyl-P – f - - - c Chim. RD RDRD – Agp1 DrBphP BphP DrRR p-DrRR + acetyl-P ++++++ - - - kDa 28 17 72 e +++ E536N RDRD E536A E536D RD E536T RD E536Q RD +++++++ DrBphP RD WT ++ – + 28 - 17 - 72 - kDa 28 17 72 kDa 28 17 72 kDa Fig. 3 Kinase and phosphatase activity of DrBphP and Agp1. a Kinase and phosphotransfer activity of the phytochromes (BphP), detected for radioactive phosphate (γ-32P) and total protein. Each phytochrome sample was incubated with γ-32P-ATP, either with or without the response regulator (RR), DrRR in case of DrBphP, and AtRR1 in case of Agp1 and Chimera. Extended gels with molecular weight marker positions are shown in Supplementary Fig. 3. b Kinase activity in darkness of DrBphP, Agp1, and their variants with the H+1 residue mutated. Each well was loaded with equal amounts of response regulator, all reactions contain ATP, and the total protein amount is visualized by protein staining. The phosphorylated response regulators (denoted p-DrRR and p-AtRR1) migrate more slowly in the gels and are therefore resolved from their unphosphorylated counterparts. cPhos-tag detection of the phosphatase activity of DrBphP (red box), Agp1, and Chimera. Equal amounts of phospho-DrRR were applied to each reaction. The letters D and R denote reactions performed in darkness or under red light, respectively. See Supplementary Fig. 4a for an extended gel. dPhosphatase activity of DrBphP and its variants H533D and E536A. Equal amounts of phospho-DrRR were applied to each reaction. See Supplementary Fig. 4b for an extended gel. ePhosphatase activity of additional DrBphP exchanges at the H+4 position. fKinase activity of Agp1 and its variants D529H and A532E. See also Supplementary Fig. 4h. g. Sequence logo of 250,000 histidine kinase sequences, shown here for the H box around the phospho-accepting histidine (H532 in DrBphP) and the N box in the CA subdomain. The height of each letter indicates the amount of conservation of the corresponding amino acid (one-letter code). The protein sequences of DrBphP and Agp1, and the fingerprint sequence motifs are shown below the graph. An aspartate residue in the H+1 position is deemed important for histidine kinase activity12. In the case of phosphatase activity, the determinant residue at the H+4 position14 varies and is denoted as X. D = dark sample; R =red-illuminated sample. Positions of molecular weight markers are shown in panels (b–f), and all measurements of panels (a–f) have been repeated independently at least three times. Source data are provided as a Source data file, which include full versions of the gels. NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24676-7 ARTICLE NATURE COMMUNICATIONS | (2021) 12:4394 | https://doi.org/10.1038/s41467-021-24676-7 | www.nature.com/naturecommunications 5 the amount of phospho-DrRR, especially upon red-light exposure (red box in Fig. 3c). These findings reveal that DrBphP acts as a phosphatase with higher activity in the Pfr state than in the Pr state. The DrBphP apoprotein was unresponsive to light and appeared similar in activity to Pr-state DrBphP (Supplementary Fig. 4f). The H533D mutation did not alter the phosphatase activity (Fig. 3d), and DrBphP was incapable of dephosphorylating phospho-AtRR1 (Supplementary Fig. 4i). Interestingly, the phosphatase activity in DrBphP depended on ATP addition as phospho-DrRR levels remained unchanged in the absence of ATP (Supplementary Fig. 4a), which is consistent with other studies46. In the presence of ADP, the DrBphP phosphatase activity was greatly decreased, and it was altogether absent if ATP was replaced with GTP (Supplementary Fig. 4f). The residue in the H +4 position is important for phosphatase activity of HisKA family proteins14. Indeed, the corresponding residue E536 in DrBphP appeared to have a role in the reaction as its mutation to alanine reduced the phosphatase activity (Fig. 3d). The E536A variant maintained somewhat higher phospho-DrRR amounts, not only in the Pfr but also in the Pr state, potentially by shielding the phospho-DrRR from spontaneous hydrolysis during the reaction. Exchanges of the same residue to threonine, asparagine, and aspartic acid (E356T, E536N, and E536D) abolished the phosphatase activity completely (Fig. 3e), thus underlining the importance of the H +4 position for the phosphatase activity. By contrast, when changing Glu536 to the structurally similar glutamine (E536Q), the phosphatase activity was preserved. Like the wild-type DrBphP, none of the H +4 variants showed any histidine kinase activity (Supplementary Fig. 4c, e). Likewise, the opposite exchange in Agp1 of the H +4 alanine to glutamic acid (A532E) did not affect the net kinase/ phosphatase activity compared to the wild-type HK (Fig. 3f). DrRR crystal structure reveals a canonical response regulator dimer. As DrBphP and Agp1 strikingly differed in their enzymatic activity and interactions, we next asked whether these differences could be explained by the structure of the interface between the DHp and RR. To model this interface with confidence, we solved the crystal structure of the response regulator from D. radiodurans (DrRR) at 2.1 Å resolution (see Table 1for crystallographic statistics). The protein, which consists only of a receiver (REC) domain, crystallized in the tetragonal P4 1 2 1 2 space group with four monomers in the asymmetric unit. These monomers form two inverted 4-5-5 dimers with a dimerization interface built by the α4–β5–α5 face of each monomer47 (Fig. 4a), similar to most other phytochrome RR structures with a REC domain48–50. However, the homologous AtRR1 assumes an armin-arm REC dimer35, in which the C-terminal extension forms an antiparallel β-strand interface with a sister monomer (Fig. 4a). Overall, the structure of the DrRR is highly similar to other reported response regulators. It contains the structural features and the conserved residues critical for its receiver function in two-component signaling (Fig. 4c). These structural details along with functional results (Fig. 3) verify that DrRR can function as a canonical response regulator in a two-component signaling system. Notably, the crystal structure of DrRR contained Ca2+instead of Mg2+ions found in other response regulator structures35,48–50. The Ca2+ions played a central role in this crystal form, as their replacement with Mg2+did not allow crystal formation. Ca2+ions occupied the active site of the DrRR in a similar way to Mg2+in the AtRR1 structure35. Although Ca2+is chemically similar to Mg2+, its larger size leads to diffuse coordination of the ion in the active sites (Fig. 4b) and 45% higher B-factors compared to Mg2+ ions modeled at the same sites. Consequently, the Ca2+ interactions differ between the four monomers in the asymmetric unit, being most similar to AtRR1 in monomer A35. In each case, the Ca2+ions are hexagonally coordinated to surrounding atoms, which involve water molecules, the side chains of Glu15, Asp16, Asn17, the phospho-accepting Asp66, and the main-chain oxygen of Asn68 (Fig. 4b). Given its presence in the DrRR crystal structure, we tested the effects of Ca2+in the DrBphP activity and DrRR binding. We discovered that although the DrBphP/DrRR interaction was slightly stronger in the presence of Ca2+(Supplementary Fig. 2h), the DrBphP enzymatic activity was lost (Supplementary Fig. 4f, g). Complex models show different interactions in DrBphP and Agp1. To analyze how the interplay of the HK and RR proteins impacts on two-component signaling, we prepared models for the DrBphP/DrRR and Agp1/AtRR1 pairs. Given the lack of high-resolution structural data on phytochrome HK domains, we generated homology models based on the complex structure of Thermotoga maritima HK853 (3DGE)9and the crystal structures of the DrRR and AtRR135 (Fig. 5). To assess the physiological relevance of the structural models and the binding interfaces, we performed a covariance analysis of cognate HK/RR pairs51–53. Prior covariance analyses assigned cognate HK/RR pairs based on genomic proximity. By contrast, we focused on a set of hybrid receptors which comprise HK and RR moieties in a single polypeptide chain, thus allowing to assign interacting, cognate HK/RR pairs with high confidence. The multiple sequence alignment of several thousand such receptors revealed strong residue covariation not only within the HK and RR parts individually but also in between them54,55.Asintheprevious Table 1 Crystal data collection and processing statistics. Data collection Space group P 41 21 2 Cell dimensions a, b, c (Å) 87.65, 87.65, 181.21 α,β,γ(°) 90.00, 90.00, 90.00 Resolution (Å) 49.74–2.0 (2.15–2.10)a R merge 0.172 (2.882) CC 1/2 0.999 (0.503) I/σ(I) 12.01 (1.09) Completeness (%) 100.0 (100.0) Redundancy 4.13 (3.94) Wilson B factor 49.54 Refinement Resolution (Å) 49.74–2.10 (2.15–2.10)a No. of reflections 39,970 (2891)a R work /R free 0.181/0.218b(0.330/0.318) Overall B factor 59.84 No. of atoms Protein 4,389 Heterogenc9 Water 240 Geometry RMSD Bond lengths (Å) 0.013 Bond angles (°) 1.811 Ramachandran Favored (%) 96 Allowed (%) 16 Outliers (%) 5 PDB Code 6XVU aOuter shell values used in the refinement are in parentheses. bTest set for R free calculation constitutes 5% of total reflections that were randomly chosen. cThis includes nine Ca2+atoms. ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24676-7 6NATURE COMMUNICATIONS | (2021) 12:4394 | https://doi.org/10.1038/s41467-021-24676-7 | www.nature.com/naturecommunications analyses, significant inter-domain covariance was observed for pairs of certain residues in the DHp domain and the RR (Supplementary Fig. 7a, b). When mapped on the presently generated structural model of the complex, strong pairwise residue covariation likewise localized to the HK/RR interface (Supplementary Fig. 7), speaking for realistic complex models. To address the stability of the complex models in solution, we conducted classical molecular dynamics (MD) simulations at 300 K, 1 atm. pressure, and 0.1 M NaCl using the Gromacs molecular dynamics package56. Over a 200 ns trajectory, both the DrBphP/ DrRR and Agp1/AtRR1 complexes were stable. The RMSD equilibration times for the protein backbone atoms were around ~60 ns for the DrBphP/DrRR complex and ~80 ns for Agp1/ AtRR1 (Supplementary Fig. 5), suggesting that the interactions are more defined and stronger in the DrBphP/DrRR complex. Starting from the 100 ns time point of the trajectory, we extracted snapshots at 10 ns intervals and analyzed their residue interactions. Representative snapshots are shown in Fig. 5, all snapshots are given in Supplementary Fig. 5a, b. Overall, the interactions between Agp1 and AtRR1 were transient and more variable than the ones in the D. radiodurans pair, as gauged by the larger overall RMSD values between successive time steps of the simulation and by higher mobility of the protein backbone atoms throughout the MD trajectory (Supplementary Fig. 5c). Analysis of the snapshots in the PISA server57 revealed that both complex interfaces have hydrophobic core regions. The average solvation free energy upon formation of the interface indicated this interface to be more extensive in the DrBphP/DrRR complex (−47.3 kJ/mol) than in the Agp1/AtRR1 complex (−24.3 kJ/mol). The simulations suggest that the RRs interact mainly through their α1 helix (aa. 18–32 in DrRR) that aligns with the helical bundle of the four DHp helices. In addition to this main interface, the DrRR showed interactions via a loop region (aa. 119–121) that connects strand β5 and helix α5. Notably, the position of this β5– α5 loop and the length of the α5 helix differed between DrRR and AtRR1, thus allowing DrRR more extended interactions with its phytochrome partner. The complexes contain polar interactions D66 D16 K118 a b D66 D16 N68 A-chain D66 E15 D20 B-chain D66 E15 D16 C-chain D66 Glu15 D20 D16 N17 D-chain DrRR AtRR1 2+ Ca α5 β5 α4 C N C N 2+ Mg 2+ Mg C α5 β5 α4 C N N 2+ Ca c K118 K118 K118K118 75 149 FELLQALRADPHLAHLPAIVLTTSN IPQTYQPQDPSDVKRAYALQANSYLTKPSTLEDFLQLIERLTAYWFGTAA 1 74 MPERASVPLRLLLVEDNAADIFLMEMALEYSSVHTELLVARDGLEALELLEQAKTGGPFPDLILLDLNMPRVDG bits 4 3 2 1 bits 4 3 2 1 α5 β5 α4 α5 β5 α4 D20 0 0 Fig. 4 Crystal structure of the Deinococcus radiodurans response regulator (DrRR). a Cartoon representation of the dimeric DrRR and AtRR1 structures (PDB code 5BRJ for AtRR135). Both response regulators only consist of receiver (REC) domains. The α4–β5–α5 face of each response regulator monomer is shown in orange, and the rest of the protein in blue. Ca2+and Mg2+ions at the active sites, the Nand C-termini, as well as the dimerization helices are marked. In the case of DrRR, the dimer formed by chains A and B is shown. bThe active site of DrRR with its Ca2+ions and interacting residues. The localization of the Ca2+ions (green) and their interactions (black dashed lines) differ between the chains. The omit difference (F o –F c ) map of the Ca2+ions is shown as blue mesh at 5.0σ. The omit maps were calculated for each monomer by repeating the final refinement step without the Ca2+ion and the coordinating water molecules. cSequence logo derived from 50,000 response regulator sequences. The height of each letter indicates the amount of conservation for the corresponding amino acid (one-letter code). The key DrRR residues are shown above the graph, and the full amino acid sequence of DrRR is given below the graph with the same coloring as in panel (a). NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24676-7 ARTICLE NATURE COMMUNICATIONS | (2021) 12:4394 | https://doi.org/10.1038/s41467-021-24676-7 | www.nature.com/naturecommunications 7 and well-defined salt bridges that are more pronounced in the D. radiodurans complex (Supplementary Fig. 6b, d). Notably, a set of interactions between DrBphP and DrRR, coordinated by a Mg2+ ion (Fig. 5c), are absent in the Agp1/AtRR1 complex (Fig. 5f). We observe that in the DrBphP/DrRR complex, inter-chain salt bridges are less fluctuating in comparison to the Agp1/AtRR1 complex (Supplementary Fig. 6b, d). As a whole, the spatially confined and less stable interactions seen in the Agp1/AtRR1 model may account for the weak and transient binding observed experimentally for this complex (see Fig. 2and Supplementary Figs. 1 and 2). The DrBphP/DrRR complex model implies that Glu536 at the H+4 position in the DHp domain coordinates with Mg2+and forms additional interactions with Arg539 and DrRR (Supplementary Fig. 6a, b). The corresponding residue in Agp1 is alanine (Ala532), and therefore these interactions are absent in the Agp1/ AtRR1 complex model. DrBphP residue Arg539 forms a distinctive salt bridge with DrRR residue Asp20. The α5 helix is longer in DrRR than in AtRR1, which enables additional contacts between the β5–α5 loop and DrBphP. In our model, this positioning of the β5–α5 loop guides the side chain of Arg539 into close proximity to Asp20, thus enabling the salt bridge with DrRR (Fig. 5b, c). In the case of Agp1, the corresponding residue Arg535 points away from AtRR1 (Fig. 5e). Taken together, the simulations implicate three central DrBphP residues that interact with the DrRR active site: His532, Glu536, and Arg539. These residues form a defined interaction network that includes a hexagonally coordinated Mg2+ion (Fig. 5c, Supplementary Fig. 6). We assessed the relevance of these residues for RR binding by ITC of selected DrBphP and Agp1 variants (Supplementary Fig. 2). In DrBphP E536A, DrRR interaction was only slightly reduced, consistent with the preservation of lightactivated phosphatase activity in this variant (see Fig. 3d). Likewise, the corresponding A532E exchange in Agp1 did not notably affect the AtRR1 interaction (Supplementary Fig. 2e). These findings imply that the residue in the H +4 position does not play a substantial role in the complex formation. By contrast, the arginine at the H +7 position appeared important for the DrBphP/DrRR interaction, as its replacement by alanine abrogated binding (Supplementary Fig. 2d), as also reflected in a reduced phosphatase activity of the R539A variant (Supplementary Fig. 4d). Discussion The bacteriophytochrome from D. radiodurans is a lightactivated phosphatase. Bacterial phytochromes commonly act as light-regulated histidine kinases in two-component systems32. Here, we introduce biochemical and structural insight into the 2+ Mg A532 H528 R535 S97 D16 D66 E11 K118 e α4 α5 H532 D20 R539 2+ Mg E536 α5 α4 E15 D66 b H528 2+ Mg E11 A532 D16 D66 K118 D12 S97 K68 H532 E536 D20 D66 R539 2+ Mg E15 N68 D16 T97 K118 c DrRR AtRR1 f H528 α4 CN N CC N A532 2+ Mg α1 α5 β5 H528 A532 Agp1 d AtRR1 2+ Mg α1 β5 α4 C N N H532 C C N H532 α5 E536 E536 aDrBphP DrRR 180° 180° Fig. 5 Complex models of the response regulators DrRR and AtRR1 and their interacting DHp domains. a–cModel of the DrBphP/DrRR complex. d–f Model of the Agp1/AtRR1 complex. The overall structure of the complexes are shown in panels (a) and (d), detailed interactions around the active site in panels (b,c) and (e,f). The orientation of the complex is flipped by 180° in panels (c) and (f). Representative structural snapshots from the MD simulations are shown. The surrounding water box with Na+and Cl−ions is omitted for clarity, and only one monomer of the response regulator dimer is shown. ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-24676-7 8NATURE COMMUNICATIONS | (2021) 12:4394 | https://doi.org/10.1038/s41467-021-24676-7 | www.nature.com/naturecommunications