6
A scissor blade-like closing mechanism implicated in transmembrane signaling in a Bacteroides hybrid two-component system Elisabeth C. Lowe a , Arnaud Baslé a , Mirjam Czjzek b,c , Susan J. Firbank a,1 , and David N. Bolam a,1 a Institute for Cell and Molecular Biosciences, Newcastle University, The Medical School, Newcastle upon Tyne NE2 4HH, United Kingdom; b Université Pierre et Marie Curie, Université Paris 6, F-29682 Roscoff, France; and c Centre National de la Recherche Scientique, Unité Mixte de Recherche 7139 Marine Plants and Biomolecules, Station Biologique de Roscoff, F-29682 Roscoff, France Edited by John Kuriyan, University of California, Berkeley, CA, and approved March 23, 2012 (received for review January 11, 2012) Signaling across the membrane in response to extracellular stimuli is essential for survival of all cells. In bacteria, responses to environ- mental changes are predominantly mediated by two-component systems, which are typically composed of a membrane-spanning sensor histidine kinase and a cytoplasmic response regulator. In the human gut symbiont Bacteroides thetaiotaomicron, hybrid two- component systems are a key part of the bacteriums ability to sense and degrade complex carbohydrates in the gut. Here, we identify the activating ligand of the hybrid two-component system, BT4663, which controls heparin and heparan sulfate acquisition and degrada- tion in this prominent gut microbe, and report the crystal structure of the extracellular sensor domain in both apo and ligand-bound forms. Current models for signal transduction across the membrane involve either a piston-like or rotational displacement of the trans- membrane helices to modulate activity of the linked cytoplasmic kinases. The structures of the BT4663 sensor domain reveal a signif- icant conformational change in the homodimer on ligand binding, which results in a scissor-like closing of the C-termini of each proto- mer. We propose this movement activates the attached intracellular kinase domains and represents an allosteric mechanism for bacterial transmembrane signaling distinct from previously described mod- els, thus expanding our understanding of signal transduction across the membrane, a fundamental requirement in many important biological processes. gut microbiota | glycan sensing | structural biology T wo-component systems (TCSs) are the dominant mechanism by which perception of extracellular stimuli is coupled to adaptive responses in prokaryotes (13). The prototypical TCS consists of two core elements: a membrane-bound sensor histidine kinase (HK) and a cytoplasmic response regulator (RR), which, together, perceive and respond to a specic environmental cue, usually a dened chemical ligand (1, 2). A typical sensor HK is composed of an N-terminal extracellular sensory domain anked by two transmembrane (TM) helices, with the C-terminal helix linked directly to a cytoplasmic HK domain (1, 2, 4). Recognition of signal by the sensory domain activates the HK, resulting in auto- phosphorylation of a conserved histidine (2). The phosphoryl group is subsequently transferred to an aspartate in the N-terminal re- ceiver domain of its cognate RR, which leads to activation of the associated output module that mediates the appropriate cellular response, typically by modulating transcription (1, 2). Modications to the basic scheme include hybrid HKs, which comprise additional receiver and phosphotransfer domains, and the recently described hybrid two-component systems (HTCSs), which contain all the domains of a classic TCS, including the output domain, but in a single polypeptide (2, 5) (Fig. 1). Over 77,000 sensor HKs and the closely related chemoreceptors have been identied to date, and although sequence analysis suggests there is great diversity among extracellular sensor domains, structural studies have revealed only a few different folds, predominantly mixed α-β or all α-helical (4, 6, 7). Of these, the mixed α-β proteins of the so-called PDC(PhoQ- DcuS-CitA) class predominate (4). These are domains of 150 amino acids that display a similar ve-stranded β-sheet core to the PAS(Per/ARNT/Sim) fold, but with distinctive structural fea- turesmost importantly, the presence of a functionally relevant N-terminal helix that is involved in homodimerization. Unlike many eukaryotic extracellular signaling systems, such as receptor tyrosine kinases (8), the mechanism of signal transduc- tion across the membrane in sensor HKs and chemoreceptors is not through ligand-mediated dimerization. Instead, perception of signal induces a conformational change in a preformed sensor homodimer that is transmitted across the membrane to activate the cytoplasmic phosphorelay (4). The precise nature of this conformational change has been best studied in the Escherichia coli chemoreceptor Tar, where ligand binding has been shown to generate an intramolecular piston-like displacement of one of the C-terminal membrane-spanning helices in the homodimer relative to the N-terminal helix of the same subunit (913). Structural studies have also implicated piston-like displacements in the sig- naling mechanisms of the CitA, TorS, and NarX sensor HKs, al- though precisely how this movement modulates kinase activity is not known (1416). An alternative mechanism for TM signaling has been demonstrated in the Vibrio harveyi quorum-sensing HK, LuxPQ, where signal binding induces formation of an asymmet- rical dimer that is predicted to be transmitted across the mem- brane to the cytoplasmic domains via rotation of the TM helices (17). A key feature of all currently studied sensor HKs is that they are anchored in the membrane via four discrete TMs, one from each terminus of the sensor protomers, that are expected to form a four-helix bundle similar to that observed in the functionally related sensory rhodopsin IItransducer complex (15, 1719). The functional relevance of this four-helix TM bundle is unclear, but its conservation suggests it plays an important role in transmission of signal across the membrane in TCSs. Our resident gut microbiota plays a key role in maintaining normal health and nutrition (20, 21). It is important therefore to understand the mechanisms used by these organisms to survive and ourish in this densely populated and highly competitive en- vironment, where they are faced with a constant ux of nutrient availability (22, 23). Recent studies have shown that bacteria from one of the dominant genera in the gut, the Bacteroides, use HTCSs to sense and respond to the presence of host and dietary poly- saccharides, which are the major source of carbon and energy in their environment (2224). Recognition of these complex sugar molecules is via direct binding of signature glycan fragments to the extracellular sensory domain of the HTCS (24). These oligo- saccharide fragments are produced by the action of the specic Author contributions: E.C.L., A.B., M.C., S.J.F., and D.N.B. designed research; E.C.L., A.B., M.C., and S.J.F. performed research; E.C.L., A.B., M.C., S.J.F., and D.N.B. analyzed data; and E.C.L., A.B., M.C., S.J.F., and D.N.B. wrote the paper. The authors declare no conict of interest. This article is a PNAS Direct Submission. Data deposition: The atomic coordinates and structure factors reported in this paper have been deposited in the Protein Data Bank, www.pdb.org (PDB ID codes 4a2l and 4a2m). 1 To whom correspondence may be addressed. E-mail: s[email protected] or david. [email protected]. This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1200479109/-/DCSupplemental. 72987303 | PNAS | May 8, 2012 | vol. 109 | no. 19 www.pnas.org/cgi/doi/10.1073/pnas.1200479109 Downloaded by guest on June 27, 2020

A scissor blade-like closing mechanism implicated in ... · A scissor blade-like closing mechanism implicated in transmembrane signaling in a Bacteroides hybrid two-component system

  • Upload
    others

  • View
    7

  • Download
    0

Embed Size (px)

Citation preview

Page 1: A scissor blade-like closing mechanism implicated in ... · A scissor blade-like closing mechanism implicated in transmembrane signaling in a Bacteroides hybrid two-component system

A scissor blade-like closing mechanism implicated intransmembrane signaling in a Bacteroides hybridtwo-component systemElisabeth C. Lowea, Arnaud Basléa, Mirjam Czjzekb,c, Susan J. Firbanka,1, and David N. Bolama,1

aInstitute for Cell and Molecular Biosciences, Newcastle University, The Medical School, Newcastle upon Tyne NE2 4HH, United Kingdom; bUniversité Pierre etMarie Curie, Université Paris 6, F-29682 Roscoff, France; and cCentre National de la Recherche Scientifique, Unité Mixte de Recherche 7139 Marine Plants andBiomolecules, Station Biologique de Roscoff, F-29682 Roscoff, France

Edited by John Kuriyan, University of California, Berkeley, CA, and approved March 23, 2012 (received for review January 11, 2012)

Signaling across themembrane in response to extracellular stimuli isessential for survival of all cells. In bacteria, responses to environ-mental changes are predominantly mediated by two-componentsystems, which are typically composed of a membrane-spanningsensor histidine kinase and a cytoplasmic response regulator. In thehuman gut symbiont Bacteroides thetaiotaomicron, hybrid two-component systems are a key part of the bacterium’s ability to senseand degrade complex carbohydrates in the gut. Here, we identifythe activating ligand of the hybrid two-component system, BT4663,which controls heparin and heparan sulfate acquisition and degrada-tion in this prominent gut microbe, and report the crystal structureof the extracellular sensor domain in both apo and ligand-boundforms. Current models for signal transduction across the membraneinvolve either a piston-like or rotational displacement of the trans-membrane helices to modulate activity of the linked cytoplasmickinases. The structures of the BT4663 sensor domain reveal a signif-icant conformational change in the homodimer on ligand binding,which results in a scissor-like closing of the C-termini of each proto-mer.We propose this movement activates the attached intracellularkinase domains and represents an allostericmechanism for bacterialtransmembrane signaling distinct from previously described mod-els, thus expanding our understanding of signal transduction acrossthe membrane, a fundamental requirement in many importantbiological processes.

gut microbiota | glycan sensing | structural biology

Two-component systems (TCSs) are the dominant mechanismby which perception of extracellular stimuli is coupled to

adaptive responses in prokaryotes (1–3). The prototypical TCSconsists of two core elements: a membrane-bound sensor histidinekinase (HK) and a cytoplasmic response regulator (RR), which,together, perceive and respond to a specific environmental cue,usually a defined chemical ligand (1, 2). A typical sensor HK iscomposed of an N-terminal extracellular sensory domain flankedby two transmembrane (TM) helices, with the C-terminal helixlinked directly to a cytoplasmicHK domain (1, 2, 4). Recognition ofsignal by the sensory domain activates the HK, resulting in auto-phosphorylation of a conserved histidine (2). The phosphoryl groupis subsequently transferred to an aspartate in the N-terminal re-ceiver domain of its cognate RR, which leads to activation of theassociated output module that mediates the appropriate cellularresponse, typically by modulating transcription (1, 2). Modificationsto the basic scheme include hybrid HKs, which comprise additionalreceiver and phosphotransfer domains, and the recently describedhybrid two-component systems (HTCSs), which contain all thedomains of a classic TCS, including the output domain, but in asingle polypeptide (2, 5) (Fig. 1). Over 77,000 sensor HKs and theclosely related chemoreceptors have been identified to date, andalthough sequence analysis suggests there is great diversity amongextracellular sensor domains, structural studies have revealed onlya few different folds, predominantly mixed α-β or all α-helical (4, 6,7). Of these, the mixed α-β proteins of the so-called “PDC” (PhoQ-DcuS-CitA) class predominate (4). These are domains of ∼150amino acids that display a similar five-stranded β-sheet core to the

“PAS” (Per/ARNT/Sim) fold, but with distinctive structural fea-tures—most importantly, the presence of a functionally relevantN-terminal helix that is involved in homodimerization.Unlike many eukaryotic extracellular signaling systems, such as

receptor tyrosine kinases (8), the mechanism of signal transduc-tion across the membrane in sensor HKs and chemoreceptors isnot through ligand-mediated dimerization. Instead, perception ofsignal induces a conformational change in a preformed sensorhomodimer that is transmitted across the membrane to activatethe cytoplasmic phosphorelay (4). The precise nature of thisconformational change has been best studied in the Escherichiacoli chemoreceptor Tar, where ligand binding has been shown togenerate an intramolecular piston-like displacement of one of theC-terminal membrane-spanning helices in the homodimer relativeto the N-terminal helix of the same subunit (9–13). Structuralstudies have also implicated piston-like displacements in the sig-naling mechanisms of the CitA, TorS, and NarX sensor HKs, al-though precisely how this movement modulates kinase activity isnot known (14–16). An alternative mechanism for TM signalinghas been demonstrated in the Vibrio harveyi quorum-sensing HK,LuxPQ, where signal binding induces formation of an asymmet-rical dimer that is predicted to be transmitted across the mem-brane to the cytoplasmic domains via rotation of the TM helices(17). A key feature of all currently studied sensor HKs is that theyare anchored in the membrane via four discrete TMs, one fromeach terminus of the sensor protomers, that are expected to forma four-helix bundle similar to that observed in the functionallyrelated sensory rhodopsin II–transducer complex (15, 17–19). Thefunctional relevance of this four-helix TMbundle is unclear, but itsconservation suggests it plays an important role in transmission ofsignal across the membrane in TCSs.Our resident gut microbiota plays a key role in maintaining

normal health and nutrition (20, 21). It is important therefore tounderstand the mechanisms used by these organisms to surviveand flourish in this densely populated and highly competitive en-vironment, where they are faced with a constant flux of nutrientavailability (22, 23). Recent studies have shown that bacteria fromone of the dominant genera in the gut, the Bacteroides, use HTCSsto sense and respond to the presence of host and dietary poly-saccharides, which are the major source of carbon and energy intheir environment (22–24). Recognition of these complex sugarmolecules is via direct binding of signature glycan fragments tothe extracellular sensory domain of the HTCS (24). These oligo-saccharide fragments are produced by the action of the specific

Author contributions: E.C.L., A.B., M.C., S.J.F., and D.N.B. designed research; E.C.L., A.B.,M.C., and S.J.F. performed research; E.C.L., A.B., M.C., S.J.F., and D.N.B. analyzed data; andE.C.L., A.B., M.C., S.J.F., and D.N.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The atomic coordinates and structure factors reported in this paper havebeen deposited in the Protein Data Bank, www.pdb.org (PDB ID codes 4a2l and 4a2m).1To whom correspondence may be addressed. E-mail: [email protected] or [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1200479109/-/DCSupplemental.

7298–7303 | PNAS | May 8, 2012 | vol. 109 | no. 19 www.pnas.org/cgi/doi/10.1073/pnas.1200479109

Dow

nloa

ded

by g

uest

on

June

27,

202

0

Page 2: A scissor blade-like closing mechanism implicated in ... · A scissor blade-like closing mechanism implicated in transmembrane signaling in a Bacteroides hybrid two-component system

locus-encoded polysaccharidases on their cognate substrate. Forexample, β1,4-linked xylotetraose, a breakdown product ofendoxylanase action on xylan, binds to and activates the HTCSthat controls expression of the Bacteroides ovatus xylan locus (24).TheN-terminal sensor domain of most HTCSs is much larger thanpreviously characterized HK sensor domains at ∼750–800 aminoacids and comprises 14 repeats that likely adopt a β-propeller fold(PfamReg_prop; PF07494), followed by a domain termedY_Y_Yof no known function (Pfam Y_Y_Y; PF07495), immediately be-fore the TM domain (Fig. 1). Here, we present the structure of theperiplasmic sensory domain of BT4663, a Reg_prop class HTCSfrom Bacteroides thetaiotaomicron (Bt) that controls expression ofthe heparin/heparan sulfate (HS) utilization locus in this prom-inent gut bacterium (Fig. 1 and Fig. S1), in both ligand-bound andapo forms (23). The structures reveal a significant scissor-likeclosing movement of the C-terminal Y_Y_Y domains of thehomodimer on ligand binding and indicate that each HTCS pro-tomer contains only a single membrane-spanning domain, pre-cluding the formation of a four-helix TM bundle in this class ofsensor regulators. These data suggest that signal transductionacross the membrane in HTCSs is driven by a scissor mechanismthat is distinct from previously described piston and rotationmodels for bacterial TM signaling.

ResultsIdentification of the Signal Molecules Recognized by BT4663. Toidentify the signal molecule that activates the Bt heparin/HS lo-cus, we initially assessed the ability of the N-terminal Reg_propclass periplasmic domain of BT4663 (the region after the signalpeptide and before the predicted internal TM) to bind glycansderived from heparin or HS, focusing on disaccharides with a 4,5-unsaturated terminal uronic acid, because these are the likely

products of heparin/HS cleavage by the locus-encoded poly-saccharide lyases (Fig. 1 and Fig. S2). Isothermal titration calo-rimetry (ITC) analysis revealed that the sensor domain ofBT4663 bound specifically to unsaturated heparin and HS-de-rived disaccharides, indicating these molecules are the activatingsignal (Ka values ranged from 4.3 × 104 to 1.6 × 105 M−1 witha stoichiometry of 1:1; Fig. 1 and Table S1). The protein dis-played plasticity in its ability to recognize variably sulfated formsof the glucosamine component of the disaccharide, but was pre-cluded from binding if the uronic acid was sulfated.

Overall Structure of the BT4663 Sensor Domain. To provide insightinto the mechanism of signal perception and transduction acrossthe membrane in this family of bacterial signaling proteins, wedetermined the crystal structure of the 764-amino acid periplasmicdomain (from residues 24–787 of the full-length protein) ofBT4663 in both apo and ligand-bound forms (Fig. 1 and Table S2).The structure of the selenomethionine (SeMet)-derivatized pro-tein was determined by single-wavelength anomalous dispersion(SAD), allowing refinement of the native structure to 2.5 Å(Materials and Methods). Each sensor molecule contains threesubdomains: the N-terminal 640 amino acids are arranged in two7-blade β-propellers, consistent with the previous predictions (25),whereas the C-terminal ∼120 residues display an all β-sheet Ig-type structure, revealing the fold of the Y_Y_Y domain and thetrue domain boundaries (Fig. 2). The Pfam Reg_prop motif cor-responds to only 24 of the ∼40 amino acids in a blade, with themost conserved residues forming a hydrogen-bonding networkbetween the innermost two β-strands of the blade (Fig. S3A). Thetwo propellers are positioned adjacent to each other, with blades 1and 7 of each propeller close to the interface and theN terminus ofthe protein positioned between them, but the planes of the two

Fig. 1. Periplasmic domain of BT4663 HTCS binds unsaturated disaccharides that are derived from heparin and HS. (A) (Upper) Bt heparin/HS locus structurewith genes of known function labeled. GH, glycoside hydrolase; PL, polysaccharide lyase [Carbohydrate-Active EnZymes database (34)]. (Lower) Domainorganization of BT4663. Domain predictions, including the type I signal peptide (SP1), Reg_prop repeats, Y_Y_Y, TM, phosphoacceptor (Pfam HisKA), ATPase(Pfam HATPase_c), receiver (Pfam REC), and DNA binding (Pfam HTH_AraC) domains, are from SMART (35). (B) ITC data showing binding of the unsaturateddisaccharide ΔUA-GlcNAc6S to the periplasmic sensor domain of BT4663 and the structure of the sugar ligand. The upper part of the ITC panel shows the rawheats of binding and the lower part the integrated heats fit to a single-site binding model.

Lowe et al. PNAS | May 8, 2012 | vol. 109 | no. 19 | 7299

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

27,

202

0

Page 3: A scissor blade-like closing mechanism implicated in ... · A scissor blade-like closing mechanism implicated in transmembrane signaling in a Bacteroides hybrid two-component system

propellers are offset from one another by a rotation of roughly 52°about the intersection axis of the planes of the propellers (Mate-rials and Methods) (Fig. 2C). The Y_Y_Y domain with eightβ-strands is found below, and almost central to, the two propellers.The C-terminus is distal to the propeller domains and ∼70 Å fromthe N-terminal residue observed in the structure (the sixth residueafter signal sequence cleavage); significantly, the two termini areorientated in essentially opposite directions, precluding the for-mation of a four-helix bundle in the full-length protein (Fig. 2A).The sensor domain crystallized with six molecules in the

asymmetrical unit, arranged as three dimers. In each dimer, thetop face of the N-terminal propeller (N-prop) of one protomer isin contact with the top face of the C-terminal propeller (C-prop)of the other protomer, although the planes of opposing propellersare not parallel (offset from one another by a rotation of ∼26°about the axis perpendicular to both planes of the propellers)(Fig. 2C). The Y_Y_Y domains also contribute to the dimer in-terface, with the predominant interactions formed between theloop immediately C-terminal to strand 1 of the Y_Y_Y and theloop between strands 4 and 5, along with contacts between strand6 from each protomer (Fig. 2B and Fig. S3B). Thus, the dimer hasan approximately twofold symmetry and is believed to be physi-ologically relevant (see below) with a dimer interface of 1,900 Å2.The distance between the Cα atoms of Ile-779, the last residuethat can be definitively assigned to the secondary structure of theY_Y_Y, is ∼31 Å apart (chains A and B), and a further fiveresidues are visible in the structure, leaving only four residuesbetween the last visible amino acid in the electron density and thepredicted start of the TM helix (Fig. 2A).

Structure of Ligand-Bound Sensor Domain. The crystal structure ofthe BT4663 sensor domain in complex with disaccharide ligandΔUA-GlcNAc6S revealed that the sugar binding site is located atthe dimer interface between the N-prop of one protomer and theC-prop of the opposing protomer, such that the protomer/ligandstoichiometry is 2:2, consistent with the ITC data (Fig. 3). Thesugar binds close to the central core of the propeller domains andmakes a number of hydrophobic and polar interactions with res-idues from the top face of each propeller (Fig. 3C and Fig. S4).Significantly, binding of ligand alters the dimer structure: Thedimer interface area increases to ∼2,140 Å2, and the relative ori-entation of the two protomers changes slightly with respect to eachother, bringing the faces of the propellers closer together by arotation about the axis of the planes of ∼13° (Movie S1 andMovieS2). This change dramatically alters the interaction between theY_Y_Y domains, translating to a movement of 28° and a 15-Ådisplacement of the two C-termini (chain B 779Cα apo to chain B779Cα ligand, after superimposition of the apo and ligand chain A

within the dimer) that brings the two Y_Y_Y domains into a moreparallel orientation (Fig. 4 and Movies S1 and S2). Recognitionof signal by the sensor domain induces a change in the relativepositions of the Y_Y_Y domains in the dimer reminiscent ofscissor blades closing, although there is also a rotational elementto the movement (Fig. 4 and Movies S1 and S2).

Residues from both Protomers Contribute to the Ligand Binding Site.The binding site and interactions between the ligand and side chainsfrom both propellers are shown in Fig. 3C and Fig. S4. The majorityof interactions with the ligand are polar, but Tyr-328 also formsa face-to-face hydrophobic stacking interaction with the ring of theuronic acid. Likely polar interactions between amino acid side chainsand sugar are shown (black dashed lines in Fig. 3C), although theresolution of the ligand-bound structure (3.4 Å) means that precisedistances cannot be determined. The majority of point mutationsof amino acids surrounding the ligand abrogate binding to ΔUA-GlcNAc6S entirely (Fig. 3C; R91A, Q216A, Q218A, R266A,S311A, R313A, Y328A, N375A, K409A, H425A, and Y455A asdetermined by ITC), confirming that residues from both protomersare required to form a functional binding site [i.e., binding will onlyoccur if BT4663 is in a dimeric form (NB. E233A mutant was madebut did not express in a recombinant form)]. Interestingly, mutationof amino acids surrounding the 6-sulfate group of the sugar had littleor no effect on affinity (Ka of Y61A 1.4 ± 0.1 × 104 M−1 and Ka ofQ499A 4.3 ± 0.2 × 104 M−1), indicating that sulfation at this pointis not essential for recognition, an observation supported by theligand specificity studies (Table S1).

Role of the Y_Y_Y Domain. The Y_Y_Y domain is so called becauseof the presence of three conserved tyrosine residues in the family.In BT4663, the first of these is actually a phenylalanine (Phe-726)and is buried in the core of the Y_Y_Y; the other two residues aretyrosines (Tyr-728 and Tyr-756) and are closer to the surface withtheir hydroxyl groups solvent-exposed (Fig. S3B). None are foundat the dimer interface in either the apo or ligand-bound structures,thus, their significance is unclear. Y_Y_Y domains are, however,observed C-terminal to all known Reg_prop domain HTCSs, in-dicating that they are fundamental to the signaling mechanism ofthis type of sensor HK. The Y_Y_Y domain, although covalentlyconnected to the C-prop, has a larger contact area with the N-prop,and comparison of the two structures indicates that the N-prop andthe Y_Y_Y act as an almost rigid body, mostly retaining their rel-ative orientation and contacts (Movies S1 and S2). As ligand bindsbetween the propellers, reorientation of the C-prop to close thesugar binding site appears to pull on the N-prop-Y_Y_Y domainsof each protomer, thereby bringing the two Y_Y_Y domains closertogether, and the C-termini into a more parallel orientation.

Fig. 2. Crystal structure of the apo form of BT4663 periplasmic sensor domain dimer. (A) One protomer is color-ramped from N-terminus–C-terminus (blue-red) with a pale blue surface; the surface of the second protomer is shown in green. The gray bar represents the cytoplasmic membrane. (B) Enlarged view ofthe Y_Y_Y domains, showing the strand labels and Y_Y_Y dimer interface. (C) Side view of the BT4663 sensor domain dimer and cartoon of the relativedomain arrangements (Inset). The solid lines symbolize the planes of the propellers (Materials and Methods). The axes of rotation are the intersection be-tween two planes and perpendicular to the plane of the paper.

7300 | www.pnas.org/cgi/doi/10.1073/pnas.1200479109 Lowe et al.

Dow

nloa

ded

by g

uest

on

June

27,

202

0

Page 4: A scissor blade-like closing mechanism implicated in ... · A scissor blade-like closing mechanism implicated in transmembrane signaling in a Bacteroides hybrid two-component system

Evidence for Dimer Formation in Solution. The oligomeric state ofthe sensor domain in solution was investigated using small-angleX-ray scattering (SAXS) both in the presence and absence ofligand (Fig. 4C, Fig. S5, and SI Results). In both cases, theresulting data can only be satisfactorily modeled when the pro-tein is considered as a dimer. The envelopes generated corre-spond well with the crystal structure dimers, with the best fit with

Crysol leading to χ-values of 16 (apo) and 10 (ligand), re-spectively (Table S3), and show a similar difference in shapebetween the apo and ligand-bound forms, suggestive of a ligand-induced change in quaternary structure in solution (Fig. 4C).Binding studies using ITC showed that neither the individualpropeller domains, the N- and C-propellers in a 1:1 ratio, nor thetruncated form of the sensor lacking the Y_Y_Y domain exhibit

Fig. 3. Location of the ligand binding sites within BT4663 sensor dimer. (A) Electron density difference map (Fobs-Fcalc) into which the ligand (ΔUA-GlcNAc6S;yellow sticks) was built (final model shown). (B) Side and top views of the ligand-bound BT4663 sensor domain dimer. The ligand is represented as spheres. (C)Binding site with disaccharide ligand (ΔUA-GlcNAc6S; carbons shown as yellow sticks, sulfur shown in light green) indicating potential polar interactions withthe two BT4663 protomers (blue and green).

Fig. 4. Conformational changes in the BT4663dimer on ligand binding. (A) Apo dimer (Left)and ligand-bound dimer (Right, ligand shownas spheres). In each case, one protomer is color-ramped N-terminus–C-terminus (blue-red), withthe other protomer shown in green. (B) View ofthe Y_Y_Y domains looking up through the mem-brane to illustrate the movement of the C-terminion ligand binding. The Y_Y_Y domain of chain Aof the apo (gray) and liganded (red) dimers issuperimposed to show the relative movement ofchain B. The Cα of residue Ile-779, the last residuefor which secondary structure can be unambig-uously defined, is represented as a blue sphere.(C) Overall shape envelope representing the so-lution structure of the apo (gray envelope) andliganded (yellow envelope) BT4663, respectively,giving the lowest χ-values (Table S3) and the fitto the experimental data represented in Fig. S5.The front view (Upper) and side view (Lower) ofthe dimer are shown.

Lowe et al. PNAS | May 8, 2012 | vol. 109 | no. 19 | 7301

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

27,

202

0

Page 5: A scissor blade-like closing mechanism implicated in ... · A scissor blade-like closing mechanism implicated in transmembrane signaling in a Bacteroides hybrid two-component system

any affinity for ligand. These studies, together with the sensitivityof the ligand binding site to point mutations in either propeller,support the structural data showing that the complete apo dimeris necessary to form the binding site(s) (Fig. 3). The ability of theBT4663 sensor domain to dimerize in the absence of TM orcytoplasmic domains also provides further support for the bi-ological relevance of the dimer structure observed in the apoform. Many extracellular domains of sensor kinases do not in-teract when dissociated from the membrane, leading to problemsinterpreting the dimer structures formed in crystal (2, 4).

DiscussionHKs are the dominant TM signaling proteins in bacteria andregulate a host of key biological processes; however, the mecha-nisms by which extracellular stimuli are transduced to the intra-cellular kinase domains are one of the least understood aspectsof TCS activation (6, 10, 26).The observed differences in the relative positions of the proto-

mers in the apo and ligand-bound structures of the BT4663 sensordomain homodimer provide a likely mechanism for TM signalingin the Reg_prop family of bacterial sensor HKs. Ligand bindingalters the dimerization interface, particularly between the Y_Y_Ydomains, such that the two C-termini are brought closer togetherby ∼15 Å, with a movement similar to that of scissor blades closing(Figs. 3 and 5 and Movies S1 and S2). This closing movement isexpected to be transmitted via the attached TM helices to thecytoplasmic HK domains to trigger autophosphorylation (Fig. 5).This scissor-like mechanism is considerably different from modelsfor signal transduction described previously for sensor HKs andchemoreceptors, where ligand binding triggers either a small pis-ton-like vectorial displacement of ∼1.0–1.5 Å or a significant ro-tational movement of ∼140° as observed in LuxPQ. Interestingly,in all previously characterized TCSs, the receptor homodimer isanchored in the membrane via four discrete TM helices, one fromeach terminal of the protomers, that are expected to form a four-helix bundle in themembrane in both the active and inactive statesof the system (10, 17, 18). However, in BT4663, the N-terminus ofeach protomer is ∼70 Å from the C-terminus and pointing in theopposite direction; thus, even when taking into consideration thefour amino acids not visible in the electron density, the N-terminicannot be envisaged to enter the membrane. Furthermore,BT4663, in common with all other Reg_prop HTCSs, containsa typical type I signal peptide that has a strongly predicted cleavagesite (Fig. 1), supporting the structural evidence indicating that theHTCS is not anchored to the membrane via its N-terminus. Thecrystal structures presented here suggest that in theReg_prop classof HTCSs, only two TM helices span the membrane per homo-dimer. The implications of this finding for signal transduction areunclear, but it is worth noting that in the open, unbound form, theC-termini of theBT4663 sensor protomers are oriented such that itis unlikely the attached TM domains would be able to interact,suggesting that the TM helices in BT4663 only interact when thesensor is in the ligand-bound state. This requirement of the scissormechanism for a relatively loose association of the TM domains,such that they can easily move from an interacting (on) to non-interacting (off) form, may explain why the more stable four-helixTM bundle is not found in HTCSs.The crystal structure of the apo form of the sensor domain of

another Reg_prop class HTCS, BT4673, has been recently re-leased in the Protein Data Bank (PDB ID 3OTT) but not pub-lished. Although the BT4673 is also dimeric in crystal and has asimilar overall domain structure to BT4663, the relative positionsof the protomers in the BT4673 dimer are somewhat differentfrom those observed in BT4663, suggesting caution in drawinggeneral conclusions regarding the mechanism of TM signaling inall Reg_prop HTCSs based on the data from BT4663 alone.Precisely how the scissor-like closing movement of the Y_Y_Y

domains observed in this study modulates kinase activity is notknown. The simplest model involves the cytoplasmic kinasedomains being held apart in the “off” state of the sensor dimer,such that the catalytic domains are unable to phosphorylate theconserved histidine on the opposing protomer (Fig. 5A). Ligand

binding to the sensor dimer drives a closing movement of the C-termini of the Y_Y_Y domains that is directly transduced acrossthe membrane to the attached kinase domains, bringing theminto close enough proximity to autophosphorylate. Although thismodel is seductive in its simplicity, it should be noted that theHisKA and HATPase cytoplasmic domains of both HTCSs andclassic sensor HKs display significant sequence similarity and inthe latter, it is known that the cytoplasmic domains are consti-tutively dimeric, suggesting this is likely the case with HTCSs aswell (1, 2, 27). If so, a mechanism of kinase activation inReg_prop HTCSs that involves a rearrangement of a preformedcytoplasmic dimer interface seems more plausible (Fig. 5B). Ascissor-like displacement has previously been proposed as a pos-sible model for signaling across a membrane (13) but, so far, hasonly been reported for the erythropoietin cytokine receptor (28,29). The data presented here indicate that the scissor mechanismfor TM signaling is more widespread than previously realizedand is shared among cells from all domains of life.Although Reg_prop class sensor domains are the main type

associated with HTCSs, a recent report revealed that the sensordomain from the Bt fructose sensing HTCS BT1754 adopts aperiplasmic binding protein (PBP) fold with a unique C-terminalhelical extension that forms the dimer interface (30). This is verydifferent from the Reg_prop fold and unlike BT4663, ligand bind-ing occurs at independent sites on each protomer. Interestingly,

Fig. 5. Scissor model for TM signaling in HTCSs. Cartoon of proposed modelfor HTCS activation. Ligand binding drives a scissor-like closing movement ofthe C-termini of the Y_Y_Y domains in the dimer, leading to activation of theassociated locus by either by bringing the attached cytoplasmic HK domainsclose enough together to allow autophosphorylation (A) or resulting in re-arrangement of an existing kinase dimer in the cytoplasm to drive autophos-phorylation (B). Chains A and B are colored blue and green, respectively. Thered arrows indicate the direction of movement of the C-termini of the Y_Y_Ydomains on ligand binding. Note that the attached RR domains of the HTCSare omitted for clarity.

7302 | www.pnas.org/cgi/doi/10.1073/pnas.1200479109 Lowe et al.

Dow

nloa

ded

by g

uest

on

June

27,

202

0

Page 6: A scissor blade-like closing mechanism implicated in ... · A scissor blade-like closing mechanism implicated in transmembrane signaling in a Bacteroides hybrid two-component system

BT1754 and orthologs from other Bacteroides species appear topossess an N-terminal lipidation motif rather than a TM domain,suggesting that, in common with Reg_prop HTCSs, the PBP-classHTCSs comprise only two membrane-spanning helices in thehomodimer. Furthermore, unlike sensor domains from TCSs,where the nonsignaling N-terminal helices form the dimer in-terface, the dimer interface in BT1754 is between the C-terminalhelices that connect directly to the intracellular domains (10).These differences suggest that the mechanism of TM signaling inPBP-class HTCSs is also distinct from that seen in classic TCSs,and possibly Reg_prop HTCSs, although the lack of an apostructure of BT1754 precludes further speculation as to theprecise mode of signal transmission.Pfam analysis indicates that over 3,600 sequences contain

Reg_prop-Y_Y_Y sensor domains, almost all of which are frombacteria with a ∼85:15 split between Bacteroidetes and Proteo-bacteria. Although the majority of these proteins are HTCSs withcomplete phosphorelay (e.g., BT4663), a significant number con-tain only the AraC family DNA binding domain after the TM re-gion or, alternatively, GGDEF/EAL second messenger domains,suggesting that the scissor mechanism for signaling across themembrane is able to drive other modes of cellular output.The data presented here provide compelling evidence that sig-

naling across the membrane in Reg_prop class HTCSs occurs viaa scissor-like closing movement distinct from the previously pro-posed piston or rotation models. This raises interesting questionsabout precisely how the cytoplasmic kinase domains are activatedby this movement and what further conformational changes, if any,are necessary to drive up-regulation of the cognate locus by theassociated AraC type DNA binding domain. Although the HisKAand HATPase domains of HTCSs show significant similarity tothose of typical TCSs, it is intriguing to speculate on whetherhaving all the TCS domains in a single polypeptide, including theeffector DNA binding domain, means that the mode of HTCSactivation is necessarily distinct from that of TCSs.

Materials and MethodsBacteroides Growth and Gene Expression. Bacteroides cells were grown asdescribed previously (30). Quantitative RT-PCR was performed using gene-

specific primers as described previously with SYBR Green (ABgene) utilizinga Roche Lightcycler (30).

Cloning and Recombinant Protein Expression. The sensor domain of BT4663was amplified from Bt VPI-5482 genomic DNA using the primers listed inTable S4. All constructs were cloned into pET28b (Novagen) and expressed inE. coli BL21 as described previously (31). His-tagged proteins were purifiedby immobilized metal affinity chromatography followed by gel filtration asdescribed previously (32).

Binding Studies. ITC was carried out at 25 °C essentially as described by Bolamet al. (31), except a Microcal VP-ITC system was used and the buffer was 20mM Hepes, pH 8.0. Protein concentrations were between 50 and 100 μM,and ligands were between 0.5 and 7.5 mM. All glycans were purchased fromDextra Laboratories Ltd.

Structural Studies. X-ray diffraction data were collected at the Diamond LightSource, Didcot, United Kingdom (beamlines I02 and I04). The structure of theSeMet derivative of BT4663 was solved using SAD. Domains of the apostructure were used as molecular replacement models for the ligand-boundform. The planes for the propellers are the least-square fit to the Cα ofresidues 22, 39, 67, 84, 108, 127, 170, 193, 211, 242, 259, 288, and 306 for theN-prop and of residues 336, 353, 385, 403, 431, 448, 477, 494, 522, 539, 565,582, 608, and 627 for the C-prop. Further details of crystallization andstructure determination are provided in SI Materials and Methods.

SAXS Experiments. Synchrotron X-ray scattering data from solutions of theBT4663 periplasmic domain in the presence and absence of ligand werecollected at the ID14-EH3 beamline of the European Synchrotron RadiationFacilities, Grenoble, France, using the X-ray detector Pilatus 1M from Dectriscoupled to a CCD camera. Data were analyzed using the ATSAS suite ofprograms (http://www.embl-hamburg.de/biosaxs/software.html), and crystalstructures were modeled into envelopes generated by GASBOR (33). Furtherdetails of data collection and analysis are provided in SI Results.

ACKNOWLEDGMENTS. We thank Carl Morland for expert technical assis-tance. This work was funded by the Biotechnology and Biological SciencesResearch Council (Grant BB/F014163/1).

1. Stock AM, Robinson VL, Goudreau PN (2000) Two-component signal transduction.Annu Rev Biochem 69:183–215.

2. Gao R, Stock AM (2009) Biological insights from structures of two-component pro-teins. Annu Rev Microbiol 63:133–154.

3. Perry J, Koteva K, Wright G (2011) Receptor domains of two-component signaltransduction systems. Mol Biosyst 7:1388–1398.

4. Cheung J, Hendrickson WA (2010) Sensor domains of two-component regulatorysystems. Curr Opin Microbiol 13:116–123.

5. Xu J, Chiang HC, Bjursell MK, Gordon JI (2004) Message from a human gut symbiont:Sensitivity is a prerequisite for sharing. Trends Microbiol 12:21–28.

6. Mascher T, Helmann JD, Unden G (2006) Stimulus perception in bacterial signal-transducing histidine kinases. Microbiol Mol Biol Rev 70:910–938.

7. Hunter S, et al. (2009) InterPro: The integrative protein signature database. NucleicAcids Res 37(Database issue):D211–D215.

8. Lemmon MA, Schlessinger J (2010) Cell signaling by receptor tyrosine kinases. Cell141:1117–1134.

9. Ottemann KM, Xiao W, Shin YK, Koshland DE, Jr. (1999) A piston model for trans-membrane signaling of the aspartate receptor. Science 285:1751–1754.

10. Falke JJ, Erbse AH (2009) The piston rises again. Structure 17:1149–1151.11. Falke JJ, Hazelbauer GL (2001) Transmembrane signaling in bacterial chemoreceptors.

Trends Biochem Sci 26:257–265.12. Milburn MV, et al. (1991) Three-dimensional structures of the ligand-binding domain

of the bacterial aspartate receptor with and without a ligand. Science 254:1342–1347.13. Chervitz SA, Falke JJ (1996) Molecular mechanism of transmembrane signaling by the

aspartate receptor: A model. Proc Natl Acad Sci USA 93:2545–2550.14. Sevvana M, et al. (2008) A ligand-induced switch in the periplasmic domain of sensor

histidine kinase CitA. J Mol Biol 377:512–523.15. Cheung J, Hendrickson WA (2009) Structural analysis of ligand stimulation of the

histidine kinase NarX. Structure 17:190–201.16. Moore JO, Hendrickson WA (2009) Structural analysis of sensor domains from the

TMAO-responsive histidine kinase receptor TorS. Structure 17:1195–1204.17. Neiditch MB, et al. (2006) Ligand-induced asymmetry in histidine sensor kinase

complex regulates quorum sensing. Cell 126:1095–1108.18. Gordeliy VI, et al. (2002) Molecular basis of transmembrane signalling by sensory

rhodopsin II-transducer complex. Nature 419:484–487.19. Cheung J, HendricksonWA (2008) Crystal structures of C4-dicarboxylate ligand complexes

with sensor domains of histidine kinases DcuS and DctB. J Biol Chem 283:30256–30265.

20. Hooper LV, Gordon JI (2001) Commensal host-bacterial relationships in the gut. Sci-

ence 292:1115–1118.21. McNeil NI (1984) The contribution of the large intestine to energy supplies in man.

Am J Clin Nutr 39:338–342.22. Hooper LV, Xu J, Falk PG, Midtvedt T, Gordon JI (1999) A molecular sensor that allows

a gut commensal to control its nutrient foundation in a competitive ecosystem. Proc

Natl Acad Sci USA 96:9833–9838.23. Martens EC, Chiang HC, Gordon JI (2008) Mucosal glycan foraging enhances fitness

and transmission of a saccharolytic human gut bacterial symbiont. Cell Host Microbe

4:447–457.24. Martens EC, et al. (2011) Recognition and degradation of plant cell wall poly-

saccharides by two human gut symbionts. PLoS Biol 9:e1001221.25. Menke M, Berger B, Cowen L (2010) Markov random fields reveal an N-terminal

double beta-propeller motif as part of a bacterial hybrid two-component sensor

system. Proc Natl Acad Sci USA 107:4069–4074.26. Casino P, Rubio V, Marina A (2010) The mechanism of signal transduction by two-

component systems. Curr Opin Struct Biol 20:763–771.27. Marina A, Waldburger CD, HendricksonWA (2005) Structure of the entire cytoplasmic

portion of a sensor histidine-kinase protein. EMBO J 24:4247–4259.28. Livnah O, et al. (1999) Crystallographic evidence for preformed dimers of erythro-

poietin receptor before ligand activation. Science 283:987–990.29. Remy I, Wilson IA, Michnick SW (1999) Erythropoietin receptor activation by a ligand-

induced conformation change. Science 283:990–993.30. Sonnenburg ED, et al. (2010) Specificity of polysaccharide use in intestinal bacteroides

species determines diet-induced microbiota alterations. Cell 141:1241–1252.31. Bolam DN, et al. (2004) X4 modules represent a new family of carbohydrate-binding

modules that display novel properties. J Biol Chem 279:22953–22963.32. Montanier C, et al. (2009) The active site of a carbohydrate esterase displays divergent

catalytic and noncatalytic binding functions. PLoS Biol 7:e71.33. Svergun DI, Petoukhov MV, Koch MH (2001) Determination of domain structure of

proteins from X-ray solution scattering. Biophys J 80:2946–2953.34. Cantarel BL, et al. (2009) The Carbohydrate-Active EnZymes database (CAZy): An

expert resource for Glycogenomics. Nucleic Acids Res 37(Database issue):D233–D238.35. Letunic I, Doerks T, Bork P (2009) SMART 6: Recent updates and new developments.

Nucleic Acids Res 37(Database issue):D229–D232.

Lowe et al. PNAS | May 8, 2012 | vol. 109 | no. 19 | 7303

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

June

27,

202

0