Figures and data
The linker controls RsbU activation through a conserved switch element
A Reaction scheme of the RsbU dependent pathway of the environmental stress response. RsbU is activated by its partner protein RsbT (T, purple) to dephosphorylate its substrate protein RsbV-P (V-P). This initiates transcription by displacing σB from inhibition by the anti-sigma factor RsbW (W). B Amino acid substitutions in RsbU (Q94L and M166V) cause enhanced σB activity in B. subtilis. Strains carrying a lacZ reporter of σB activity (ctc-lacZ) were plated on X-gal indicator plates in the presence (top) or absence (bottom) of IPTG to induce expression of the rsbU constructs. Strains were deleted for rsbT (–) or not (+). A strain overex-pressing rsbT is indicated by oe. Plates were imaged after 24 hours of growth at 37°C, which was shorter than the time used to visualize lacZ expression from the rsbUM166V strain previously (Ho & Bradshaw 2021). C Domain diagram of RsbU. The N-terminal domain of RsbU (blue) (amino acids 1-81) is joined to the α-helical linker domain (grey) (82-112), which connects it to the PPM phosphatase domain (grey box) (121-335). The regulatory switch element of the PPM phosphatase domain is colored orange (156-201). The location of RsbT-bypass mutations are highlighted in green. D AlphaFold2 model of RsbU dimer. Domains are colored as in C and the Cα atoms of residues Q94 and M166 are shown as spheres.
PAE plots of AlphaFold2 prediction
Predicted aligned error of the AlphaFold2 structure predictions of the following proteins: B. subtilis RsbU, B. subtilis 2RsbT (chains A, B)/2RsbU (chains C,D) complex, B. subtilis 2RsbP (chains A,B)/2RsbQ complex (chains C,D), C. difficile RsbZ, Gammaproteo-bacteria sp. NreB, M. thermoacetica MtG, M. tuberculosis Rv1364c, S. coelicolor OsaC, R. aetherivorans WP_029544106.1, S. coelicolor MYU40396.1, and Synechocystis sp. WP_010874143.1. Blue indicates high confidence (0 Å) and red indicates low confidence (30 Å).
AlphaFold2 structure predictions with mapped pLDDT scores
Predicted local distance difference test scores (pLDDT) mapped onto the AlphaFold2 structure predictions of the following proteins: B. subtilis RsbU, B. subtilis 2RsbT/2RsbU complex, B. subtilis 2RsbP/2RsbQ complex, C. difficile RsbZ, Gammaproteo-bacteria sp. NreB, M. thermoacetica MtG, M. tuberculosis Rv1364c, S. coelicolor OsaC, R. aetherivorans WP_029544106.1, S. coelicolor MYU40396.1, and Synechocystis sp. WP_010874143.1. The score key is above each structure model. Blue corresponds to a pLDDT score of 90-100 (high confidence), orange to yellow corresponds to a pLDDT score of 50-70 (medium confidence), and red corresponds to a pLDDT score of 0 (low confidence).
Q94L stabilizes the packing interface of the coiled-coil of the linker
A Coil diagram of RsbU residues 83-97 generated using DrawCoil. Grey circles indicate non-polar residues, orange indicate polar residues, red indicate negatively charged residues, and blue indicate positively charged residues. The blue dashed line depicts a putative salt bridge. Residue Q94 (circled in green) sits in the interface of the coils and is shown packing with L90. B An allelic series of amino acid substitutions at position 94 of RsbU was generated in an rsbU expression plasmid. Plasmids were introduced into B. subtilis cells containing a LacZ reporter of σB activity that were deleted for rsbT (top) or with rsbT retained on the chromosome (bottom). Cells were plated on IPTG/X-gal plates to induce rsbU expression and to indicate σB activity (blue pigmentation). C α-helical pitch analysis of coiled-coil domains. The AlphaFold2 model of an RsbU dimer is overlayed with crystal structure of P. aeruginosa RssB (PDB 3F7A, purple, left). The helical linker region extending out of the coiled coil (RsbU98-112) is shown (grey) overlaid with T. maritima transducing histidine kinase CheA (PDB 1B3Q, red, middle) and Idiomarina A28L diguanylyl cyclases PadC (PDB 5LLW, yellow, right). α angles from Crick parameterization analysis using CCCP are shown for the overlaid region.
RsbU linker influences RsbT binding and phosphatase activity
A Mean anisotropy of RsbT-TMR (200 nM) is plotted as a function of RsbU concentration. The curves are fits to a quadratic binding equation using non-linear curve fitting (RsbU (black) KD = 4.5 ± 0.96 µM, RsbUQ94L (green) KD = 0.56 ± 0.15 µM). The lower plot shows RsbU concentrations to 10 µM. Below is the reaction scheme of RsbT binding to RsbU with the values for the calculated KD below. B The rate of dephosphorylation of RsbV-P is plotted as a function of concentration of MnCl2 for RsbU (black) and RsbUQ94L (green) in the presence (dark) and absence (light) of RsbT. Curves are fits to the Michaelis-Menten equation using non-linear curve fitting. The kcat of wild-type RsbU is 15 ± 0.35 min-1 with RsbT and 1.4 ± 0.077 min-1 without RsbT and the kcat of RsbUQ94L is 9.0 ± 0.64 min-1 with RsbT and 2.4 ± 0.084 min-1 without RsbT. The KMMnCl2 of wild-type RsbU is 0.98 ± 0.069 mM with RsbT and 77 ± 9.0 mM without RsbT and the KMMnCl2 of RsbUQ94L is 1.4 ± 0.19 mM with RsbT and 1.9 ± 0.56 mM without RsbT. The lower plot shows wild-type RsbU in the absence of RsbT for MnCl2 concentrations to 100 mM. Below is a summary of a reaction scheme of RsbU dephosphorylating RsbV-P (denoted as S) with the KMMnCl2 and kcat values below. Data for RsbU are reproduced from Ho & Bradshaw 2021. C A cartoon model of how the switch helices rotate to activate RsbU. Binding of RsbT or mutation moves the switch helices into place during activation to coordinate metal M2. The residues of RsbU that are hypothesized to coordinate metals are shown as sticks (based on homology to RsbX).
RsbT has a hybrid interface with RsbU N-terminal dimerization domain and linker
A An AlphaFold2 model of a heterotetrameric 2RsbT/2RsbU complex is shown in ribbon representation. RsbT (purple) binds with an interface that spans the N-terminal domains (blue) and α-helical linkers (grey) of both RsbU protomers. The PPM phosphatase domains are colored grey and the regulatory switch elements (α1 and α2) are colored orange. The Cα positions of RsbUY28 (red), rsbUY28I suppressor substitutions (green), RsbUR91 (dark pink), and RsbTD92 (pink) are shown as spheres with the residue numbers indicated. B Strains with a σB LacZ reporter were plated on IPTG/X-gal plates to induce expression of plasmid borne rsbT and rsbUY28I and visualize σB activity (indicated by blue pigmentation after 24 hours of growth at 37°C). Additional amino acid substitutions in the rsbU gene are indicated above the spots, with the region where they are located in the RsbU structure specified. C Strains with indicated changes to the RsbT and RsbU amino acid sequence were plated as in B and plates were imaged after 36 hours of growth at 37°C. D R91E does not decrease the basal phosphatase activity of RsbU. Plots show dephosphorylation of RsbV-P (25 µM) by RsbU (black) and RsbUR91E (pink) (0.5µM) in the presence of 10 mM MnCl2 and the presence or absence of 10 µM RsbT (light colors). Linear fits of the data are shown, with the slopes indicating the observed rates (kobs): RsbU 0.0206 ± 0.00124 min-1, RsbU + RsbT 10.0 ± 0.628 min-1, RsbUR91E is 0.0379 ± 0.00543 min-1, RsbUR91E + RsbT 0.142 ± 0.0161 min-1. The lower plot is RsbU in the presence of RsbT to thirty minutes.
RsbT binds to the linker
A Interface of RsbT and N-terminal domain of RsbU. A ribbon representation is shown of the AlphaFold2 model of the heterote-trameric 2RsbT/2RsbU complex with RsbU N-terminal dimerization domain colored blue, the switch element of the PPM domain (α1 and α2) colored orange, and RsbT colored purple. The Cα of positions of amino acids implicated in the binding interface of RsbT and RsbU N-terminal dimerization domain (Hardwick et al. 2007) are represented on the AlphaFold2 structure as yellow spheres (RsbU) and green spheres (RsbT) with the residue numbers.
RsbU linker is flexible in the absence of RsbT, and is rigidified by RsbT binding.
A I(q) versus q plot of the dimer component of SAXS scattering data from RsbU (blue) overlaid with a FoXS generated profile fit of the AlphaFold2 RsbU dimer structure (red). The upper plot shows I(q) versus q on log-linear axes while the lower plot shows the uncertainty normalized residuals of the fit ((experimental-computed)/error), χ2 of 12.53. The AlphaFold2 model used for the fit is shown inset. B Dimensionless Kratky plot of SAXS scattering data from the dimer component of RsbU experimental data (blue). The experimental profile was logarthmically binned to reduce noise at high qRg. C Normalized P(r) plot from the RsbU dimer (blue) and heterotetrameric RsbT/RsbUQ94L complex (light blue) is shown. D I(q) versus q plot of the dimer component of SAXS scattering data from RsbU (blue) is shown overlaid with a MultiFoXS generated two-state profile fit with flexibility allowed for residues 82-96 (pink). The lower plot shows the uncertainty normalized residuals of the fit ((experimental-computed)/error), χ2 of 1.2. E Dimensionless Kratky plot of the heterotetrameric component of RsbT/RsbUQ94L complex is shown binned logarithmically to reduce noise at high qRg values. F I(q) versus q plot of the heterotetrameric component of SAXS scattering data from RsbT/RsbUQ94L complex (blue) overlaid with a FoXS generated profile fit of the AlphaFold2 RsbT/RsbU heterotetramer structure (red). The upper plot show I(q) versus q on a log-linear axes while the lower plot shows the uncertainty normalized residuals of the fits ((experimental-computed)/error), χ2 of 1.83.G Schematic of the conformational change of RsbU upon RsbT binding. The two best fit flexible models of RsbU generated from MultiFoXS in panel B are shown with the calculated Rg and percentage of each model used in the fit indicated. These two models are shown in exchange with arrows to indicate flexibility. The AlphaFold2 model of 2RsbU/2RsbT heterotetramer used for the FoXS fit in E is shown with the calculated Rg indicated.
MALS estimate of molecular weight of RsbU dimer, RsbT/RsbUQ94L complex, RsbT/RsbU complex, and RsbUQ94L
A Plot of inline SEC-MALS analysis of RsbU with measured differential refractive index (blue, left axis) and the molecular weight calculated from MALS (red, right axis). B Denaturing SDS-PAGE gel of successive fractions from the equivalent range of the elution as shown in A from preparative size exclusion chromatography of the sample of RsbU used for SEC-MALS-SAXS. Black box denotes the C-terminal truncation product present in each lane of the main peak. C Plots from inline SEC-MALS analysis of RsbT/RsbUQ94L complex with measured differential refractive index (blue, left axis) and the molecular weight calculated from MALS (red, right axis). D Denaturing SDS-PAGE gel of successive fractions from the equivalent range of the elution of RsbT/UQ94L as shown in C from preparative size exclusion chromatography of the sample used for SEC-MALS-SAXS. Black box denotes lack of C-terminal truncation product during co-expression and purification of RsbT/RsbUQ94L complex. * Denotes load samples on gel. E Plots from inline SEC-MALS analysis of RsbT/U complex with measured differential refractive index (blue, left axis) and the molecular weight calculated from MALS (red, right axis). F Plot inline SEC-MALS analysis of RsbUQ94L with measured differential refractive index (blue, left axis) and the molecular weight calculated from MALS (red, right axis).
EFA deconvolution and Guinier analysis of RsbU dimer SAXS profiles
A SAXS data summary for RsbU. Series intensity (blue, left axis) versus frames, and Rg versus frame (red, right axis) are plotted. B The selected intensity range for EFA (black) (frame index 1330-1620), and the individual component ranges for deconvolution; blue denotes the range of frames where the scattering component we assigned as the c0, RsbU dimer, was detected (frame index 1330-1603) and orange denotes the range of frames where the scattering component that is c1, presumably from a RsbU heterodimer containing a c-terminal truncation product dimer (frame index 1357-1620). C Area normalized concentration profiles for the c0, RsbU dimer component (blue) and c1, RsbU heterodimer containing c-terminal truncation product dimer (orange). D Mean χ2 values for the fit of the EFA deconvolution to the original scattering data. E Guinier fit for RsbU dimer component on the upper plot and uncertainty normalized residuals of the fit on the lower plot. The scattering data for the dimer component (red) is fit to the Guinier approximation (blue). The blue dashed lines are the qmin and qmax at values 0.00356 Å-1 and 0.0234 Å-1, respectively. F Kratky plot of SAXS scattering data from the dimer component of RsbU experimental data (blue) overlaid with the profile computed by FoXS based on the AlphaFold2 model of an RsbU dimer (red) and the profile of the flexible model computed by MultiFoXS (pink). The experimental profile was logarthmically binned to reduce noise at high q.
Varying constraints and flexibility of RsbU model
A I(q) versus q plot of the dimer component of SAXS scattering data from RsbU (blue) overlaid with a MultiFoXS generated two-state profile fit with flexibility allowed for residues 82-86 (pink). The upper plot shows I(q) versus q on log-linear axes while the lower plot shows the uncertainty normalized residuals of the fit ((experimental-computed)/error), χ2 of 1.26. B I(q) versus q plot of the dimer component of SAXS scattering data from RsbU (blue) overlaid with a MultiFoXS generated two-state profile fit with flexibility allowed for residues 82-92 (pink). The upper plot shows I(q) versus q on log-linear axes while the lower plot shows the uncertainty normalized residuals of the fit ((experimental-computed)/error), χ2 of 1.25. C I(q) versus q plot of the dimer component of SAXS scattering data from RsbU (blue) overlaid with a MultiFoXS generated two-state profile fit with flexibility allowed for residues 92-99 (pink). The upper plot shows I(q) versus q on log-linear axes while the lower plot shows the uncertainty normalized residuals of the fit ((experimental-computed)/error), χ2 of 1.26. D I(q) versus q plot of the dimer component of SAXS scattering data from RsbU (blue) overlaid with a MultiFoXS generated two-state profile fit with flexibility allowed for residues 82-96 (pink) and the C-terminal domains constrained. The upper plot shows I(q) versus q on log-linear axes while the lower plot shows the uncertainty normalized residuals of the fit ((experimental-computed)/error), χ2 of 5.4.
EFA deconvolution and Guinier analysis of RsbT/RsbUQ94L SAXS profiles
A SAXS data summary for RsbT/RsbUQ94L. Series intensity (blue, left axis) versus frames, and Rg versus frame (red, right axis) are plotted. B The selected intensity range for EFA (black) (frame index 1300-1540), and the individual component ranges for deconvolution; blue denotes the range of frames where the scattering component we assigned as c0, ambiguous component, was detected (1300-1461) and orange denotes the range of frames where the scattering component we assigned as c1, the 2RsbT/2RsbUQ94L heterotetrametric complex, was detected (1318-1540). C Area normalized concentration profiles for 2RsbT/2Rs-bUQ94L heterotetrameric complex (orange) and ambiguous component (blue). D Mean χ2 values for the EFA deconvolution fit to the original scattering data. E Guinier fit for 2RsbT/2RsbUQ94L heterotetrameric complex component on the upper plot and uncertainty normalized residuals of the fit on the lower plot. The scattering data for the dimer component (red) is fit to the Guinier approximation (blue). The blue dashed lines are the qmin and qmax at values 0.0033 Å-1 and 0.0336 Å-1, respectively. F I(q) versus q plot of the heterotetrameric complex component of SAXS scattering data from 2RsbT/2RsbU (blue) overlaid with a MultiFoXS generated two-state profile fit with flexibility allowed for residues 82-96 (red). The upper plot shows I(q) versus q on log-linear axes while the lower plot shows the uncertainty normalized residuals of the fit ((experimental-computed)/error), χ2 of 1.96. The one-state flexible model is shown with a RMSD to the AlphaFold2 prediction of 2RsbT/2RsbU of 1.36 Å. G Kratky plot of SAXS scattering data from the heterotetrameric component of 2RsbT/2RsbU experimental data (blue) overlaid with the profile computed by FoXS based on the AlphaFold2 model of an RsbT/RsbU heterotetramer (red). The experimental profile was logarithmically binned to reduce noise at high q.
An exchangeable toolkit for bacterial signal transduction
A An AlphaFold2 model of the B. subtilis energy stress phosphatase RsbP is shown bound to its activating α/β-hydrolase protein RsbQ (purple) with a schematic of the σB activation pathway shown above. RsbP has an N-terminal PAS domain (turquoise) that is connected to the PPM phosphatase domain by a predicted coiled coil linker. The Cα positions of bypass suppressor mutations that render RsbP active in the absence of RsbQ(Brody et al. 2009) are shown as green spheres. The switch element of the PPM phosphatase domain (α1 and α2) is colored orange. B An AlphaFold2 model of the C. difficile GSR phosphatase RsbZ is shown with a schematic of the σB activation pathway shown above. RsbZ has tandem N-terminal GAF domains (blue) that connect to the PPM phosphatase domain by a predicted coiled coil linker. The switch element of the PPM phosphatase domain (α1 and α2) is colored orange. C Experimental structures of the GSR initiating adaptor protein RssB from E. coli (left, 8T85) and P. aeruginosa (right, 3F7A) are shown with a schematic of the σS activation pathway shown above. The Cα positions of a bypass suppressor mutation (W143R) that renders RssB insensitive to anti-adapter proteins is shown as a green sphere. The switch element of the PPM phosphatase domain (α1 and α2) is colored orange. The black outline shows the portion of hte PPM phosphatase domain that is missing in the E. coli RssB adaptor protein, which lacks phosphatase activity. D A cartoon schematic of activation by RsbT of proteins with RsbU NTD through ordering of a flexible linker. The protein is depicted with dimerized RsbU N-terminal dimerization domains (blue) with flexible linkers (black) that extend into two effector domains (grey). During stress conditions, RsbT (purple) activates the protein through binding to hte N-terminal dimerization domains and linkers, rigidifying the linkers to dimerize the effector domains. E An AlphaFold2 model of a dimer of the histidine kinase NreB from a Gammaproteobacteria species (MBI3545564.1) is shown with the RsbU N-terminal domain colored blue and the histidine kinase domain colored pink. F An AlphaFold2 model of a dimer of the GGDEF diguanylate cyclase MtG from M. thermoacetica (WP_011392981.1) is shown with RsbU N-terminal domain colored blue, GAF domain colored lite blue, and GGDEF domain colored green. G Schematic illustrating how various sensor domains (RsbU N-terminal dimerization domain, PAS, response receiver (RR), and GAF) can control various output activities (serine/threonine phosphatase (PPM), histidine kinase (HK), protease adaptor, and diguanylate cyclase (GGDEF). We hypothesize that the shared regulatory mechanism through a dimeric coiled-coil linker makes these regulatory domains modularly exchangeable across effector domains. Known mechanisms for how allosteric regulation is transmitted through the linker are listed and shown with arrow diagrams.
AlphaFold2 predictions of varying signaling PPM phosphatases
A An AlphaFold2 model of the M. tubercolisis GSR phosphatase Rv1364c is shown with a N-terminal PAS domain colored turquoise that connects to the PPM phosphatase domain colored grey, an anti-σ domain (related to RsbW) colored pink and an anti-anti-σ domain (related to RsbV) colored purple. The switch element of the PPM phosphatase domain (α1 and α2) is colored orange. B An AlphaFold2 model of the S. coelicolor OsaC GSR phosphatase is shown with a N-terminal RsbW-like kinase domain colored in green, connected to a PAS domain colored in turquoise, tandem GAF domains colored in blue, and a PPM phosphatase colored in grey. The switch element of the PPM phosphatase domain (α1 and α2) is colored orange. C An AlphaFold2 model of a dimer of the phosphatase from R. aetherivorans (WP_029544106.1) is shown with the GAF domain colored in blue connected to the PPM phosphatase colored in grey. The switch element of the PPM phosphatase domain (α1 and α2) is colored orange. D An AlphaFold2 model of a dimer of the phosphatase from S. coelicolor (MYU40396.1) is shown with the response receiver domain colored in red connected to the PPM phosphatase colored in grey. The switch element of the PPM phosphatase domain (α1 and α 2) is colored orange. E An AlphaFold2 model of a dimer of the phosphatase from Synechocystis species (WP_010874143.1) is shown with the HAMP domain colored in yellow connected to the PPM phosphatase colored in grey. The switch element of the PPM phosphatase domain (α1 and α2) is colored orange. The structure models in all panels are shown aligned by the top PPM domain to highlight the conservation of the PPM dimer in predictions of dimeric phosphatases with diverse regulatory domains.
Coiled-coil analysis of structures of bacterial signaling enzymes
Models of selected phosphatases are shown: A, heterotetramer of RsbPQ from B. subtilis; B, RsbZ from C. difficile; C, RssB from P. aeruginosa (PDB: 3EQ2); D, NreB from Gammaproteobacteria; E, MtG from M. thermoacetica; F, Rv1364c from M. tuberculosis; G, OsaC from S. coelicolor; H, WP_029544106.1 from R. aetherivorans; I, MYU40396.1 from S. coelicolor; J, WP_010874143.1 from Synechocystis sp.. Coiled-coil sequences identified by Socket2 are shown above the model with the non-polar positions of a and d shown in pink above the sequence. Polar residues at the a and d positions are in bold. The coiled-coil region on the model is shown in pink, the switch helices of the PPM domains are shown in orange, and the various regulatory domains are shown in the following colors: RsbU NTD (blue), PAS (green), GAF (light blue), response receiver (red), and HAMP (yellow).
