characterization of elements involved in allosteric light ... · munication pathway. here we show...

Article

Andreas Winkl

0022-2836/$ - see front m

Characterization of Elements Involved inAllosteric Light Regulation ofPhosphodiesterase Activity by Comparisonof Different Functional BlrP1 States

er1, Anikó Udvarhelyi 1, El
isabeth Hartmann1, Jochen Reinstein1,Andreas Menzel 2, Robert L. Shoeman1 and Ilme Schlichting1
1 - Department of Biomolecular Mechanisms, Max Planck Institute for Medical Research, Jahnstrasse 29,69120 Heidelberg, Germany2 - Paul Scherrer Institut, 5232 Villigen PSI, Switzerland

Correspondence to Andreas Winkler and Ilme Schlichting: [email protected];[email protected]://dx.doi.org/10.1016/j.jmb.2013.11.018Edited by C. Kalodimos

Abstract

Bacteria have evolved dedicated signaling mechanisms that enable the integration of a range of environmentalstimuli and the accordant modulation of metabolic pathways. One central signaling molecule in bacteria is thesecond messenger cyclic dimeric GMP (c-di-GMP). Complex regulatory mechanisms for modulating c-di-GMPconcentrations have evolved, in line with its importance for maintaining bacterial fitness under changingenvironmental conditions. One interesting example in this context is the blue-light-regulated phosphodiesterase1 (BlrP1) ofKlebsiella pneumoniae. This covalently linked system of a sensor of blue light using FAD (BLUF) andan EAL phosphodiesterase domain orchestrates the light-dependent down-regulation of c-di-GMP levels.To reveal details of light-induced structural changes involved in EAL activity regulation, we extended previouscrystallographic studies with hydrogen–deuterium exchange experiments and small-angle X-ray scatteringanalysis of different functional BlrP1 states. The combination of hydrogen–deuterium exchange and small-angleX-ray scattering allows the integration of local and global structural changes and provides an improvedunderstanding of light signaling via an allosteric communication pathway between the BLUF and EAL domains.This model is supported by results from a mutational analysis of the EAL dimerization region and the analysis ofmetal-coordination effects of the EAL active site on the dark-state recovery kinetics of the BLUF domain. Incombination with structural information from other EAL domains, the observed bidirectional communicationpoints to a general mechanism of EAL activity regulation and suggests that a similar allosteric coupling ismaintained in catalytically inactive EAL domains that retain a regulatory function.

© 2013 The Authors. Published by Elsevier Ltd. All rights reserved.

Introduction

Allostery [1,2] is a widely used concept to describevarious biomolecular processes ranging from proteindynamics, activation of membrane receptors orchannels, chaperon function and virus assembly toallosteric enzymes (recently reviewed in Ref. [3]).The family of cyclic dimeric GMP (c-di-GMP)-inter-acting EAL domains [4] represents an interestingexample. C-di-GMP was originally discovered as anallosteric activator of cellulose synthase in Glucona-cetobacter xylinus [5] and has subsequentlyemerged as a central bacterial second messengerinvolved in regulating a wealth of cellular functions

atter © 2013 The Authors. Published by Els

(reviewed in Refs. [6–8]). C-di-GMP's importance forbacterial homeostasis is reflected in the evolution ofcomplex regulatory mechanisms that use c-di-GMPas an allosteric effector and in the modulation of itssynthesis and degradation, which is controlled byvarious environmental sensor modules. EAL domainsare involved in bacterial c-di-GMP signaling due totheir phosphodiesterase (PDE) activity resulting inasymmetric cleavage of one substrate phosphodies-ter bond forming the linear 5′-pGpG product [4,9–13].Additionally, however, sequence analysis [9,11] andmutational studies [13] revealed EAL subfamilies withmutations in otherwise highly conserved regions thatlack PDE activity. It was suggested that such EAL

evier Ltd. All rights reserved. J. Mol. Biol. (2014) 426, 853–868

mailto:[email protected]

mailto:[email protected]

http://dx.doi.org/[email protected]�Udvarhelyi1Aniko.Udvarhelyi@mpimf-heidelberg.mpg.deElisabethHartmann1Elisabeth.Hartmann@mpimf-heidelberg.mpg.deJochenReinstein1Jochen.Reinstein@[email protected] L.Shoeman1Robert.Shoeman@mpimf-heidelberg.mpg.deIlmeSchlichting1NIlme.Schlichting@mpimf-heidelberg.mpg.de1Department of Biomolecular Mechanisms, Max Planck Institute for Medical Research, Jahnstrasse 29, 69120 Heidelberg, GermanyDepartment of Biomolecular Mechanisms, Max Planck Institute for Medical ResearchJahnstrasse 29Heidelberg69120Germany2Paul Scherrer Institut, 5232 Villigen PSI, SwitzerlandPaul Scherrer InstitutVilligen PSI5232SwitzerlandNCorresponding authors.

http://dx.doi.org/

http://dx.doi.org/10.1016/j.jmb.2013.11.018

854 Light Regulation of Phosphodiesterase Activity

domains cancommunicate thebindingof c-di-GMP toavariety of effector domains via allosteric modulation[14–17]. Interestingly, EAL activity was also shownto be influenced in an allosteric manner upon GTPbinding inGGDEF-EAL systems (reviewed inRef. [18])with degenerate GGDEF domains (reviewed in Ref.[19]) that lack their endogenous diguanylate cyclaseactivity [20,21]. Notably, EAL domains can also beregulated by a variety of other stimuli that are sensedbyreceiver (REC) domains [11]; Per-ARNT-Sim (PAS)domains [22–24]; domains found in cGMP-specificPDEs, adenylyl cyclases and FhlA (GAF) [25]; helix–turn–helix motifs [4]; phytochromes (PHY) [26]; orBLUF [10,12,15] modules.The latter domain, acronym for sensor of blue light

using FAD [27], enables the light-mediated control ofdifferent biological processes including gene tran-scription [28–30], phototaxis [31], adenylyl cyclaseactivity [32,33] and, as mentioned above, PDEreactivity in EAL domains. However, despite a greatdeal of information on these diverse systems, mo-lecular and mechanistic details of light-inducedchanges in the vicinity of the flavin cofactor uponillumination are still under debate (recently summa-rized in Refs. [34] and [35]). Moreover, the confor-mational changes responsible for communicatingthe light signal to various effector domains regulatedby BLUF are also not well understood. Therefore, theblue-light-regulated phosphodiesterase 1 [BlrP1; Uni-Prot (UNP) ID: A6T8V8] from Klebsiella pneumoniae,which represents a covalently linked BLUF sensorand EAL effector system exhibiting a characteristicBLUF photocycle [36] and PDE activity [10], providesan intriguing system to improve our understanding ofboth allosteric regulation of EAL activity and lightsignaling by BLUF domains.Previous crystallographic studies of BlrP1 have

revealed a dimeric arrangement of two EAL domainsfeaturing an evolutionary conserved interface madeup of two dimerization helices and one compoundhelix that is formed by two short helices provided byeach protomer (Fig. 1a). Importantly, the light-sensingBLUF domains (Fig. 1b) are positioned close to thisdimerization region [10]. PDE activity is also stimulat-ed by an increase in pH that affects the coordination ofthe catalytically relevant metal ions in the active site.Based on this observation, a mechanism suggestingthat structural changes induced by flavin excitation arepassed on via the C-terminal BLUF capping helices tothe EAL dimerization interface ultimately leading to achangeof the EALactive-site geometrywas proposed[10]. Despite the knowledge on structural details ofdark-adapted BlrP1 and mechanistic insight into EALregulation [10], there are still open questions related tostructural aspects of the light-activated BlrP1 stateand details of elements involved in the allosteric com-munication pathway.Here we show for the first time structural details

of the light-adapted BlrP1 state by integrating the

analysis of global and local structural changesobtainedfrom small-angle X-ray scattering (SAXS) studies andhydrogen–deuterium exchange (HDX) experimentsanalyzed by mass spectrometry (MS), respectively.We describe molecular details of EAL activity regula-tion by the BLUF sensor, providing evidence for anallosteric bidirectional communication between theflavin environment and the metal coordination in theactive site of EAL. The compound helix, positioned atthe EAL dimerization interface, plays a key role in thisregulation, which is supported by complementarymutational and functional studies. In addition, SAXSexperiments and normal mode analysis (NMA) sug-gest inter-domain rearrangements in BlrP1 that appearto be functionally conserved in other EAL dimers. Incombination, our data provide new insight into molec-ular details involved in light sensing by BLUF domainsand into regulatory aspects of EAL activity.

Results

TheEALdimer interface is influencedbysubstrateand calcium binding

As shown previously, c-di-GMP hydrolysis in BlrP1is affected by changes in pH, by binding of divalentmetal ions to the active site and by blue-light illu-mination [10]. It was proposed that absorption of ablue photon by the BLUF domain eventually resultsin changes in coordination of two metal ions that arecritical for EAL activity (Fig. 1c). To obtain a betterunderstanding of which structural elements are in-volved in the regulation of PDE activity, we probedfour functionally relevant states of BlrP1 by HDX–MS:the substrate-free states of dark- and light-adaptedBlrP1 in the presence of magnesium (Mgd and Mgl,respectively; i.e., Magnesiumdark andMagnesiumlight);in addition, we addressed the influence of substratebinding by forming the inhibited EAL–Ca2+–c-di-GMPcomplex also under dark and blue-light conditions(Ccd and Ccl, respectively; i.e., Calcium–c-di-GMPdarkand Calcium–c-di-GMPlight). Representative deuteri-um incorporation plots are shown in Fig. 2, anddifferences in relative deuterium levels (ΔDrel) for allassigned peptides in various states are summarizedin Fig. 3. The localization of the most importantstructural elements in the quaternary BlrP1 assem-bly is illustrated in Fig. 4, and close-up views ofimportant regions are provided in Fig. S1. Fulldetails of all evaluated peptides are provided inFigs. S2–S5.

We initially compared Ccd to Mgd (Fig. 3a) to identifystructural features that are affected by substrate andcalcium binding in the dark. As expected, this com-parison shows several elements that, based on theavailable crystal structures [10], are involved in sub-strate binding. The region containing the conserved

Fig. 1. Overview of the K. pneumoniae BlrP1 structure. (a) Schematic representation of two different orientations of theBlrP1 structure reported previously [10]. The N-terminal BLUF domains are colored violet with protomer B in light color andtransparent mode to show structural elements of the EAL domains in the background. EAL protomers A and B are coloreddark and light green, respectively, and their overlap is in cyan. The compound helix formed by two short helices ofprotomers A and B in this region is illustrated in dark and light blue, respectively. The dimerization helices, one of eachprotomer, are colored dark and light red. (b) Cartoon representation of the BLUF domain to illustrate the arrangement ofsecondary structure elements (protomer B, PDB ID: 3GG0 [10]). The flavin cofactor is shown as yellow stick model. (c)Stereo view of the EAL active site and the dimerization region of BlrP1. The compound helix and the dimerization helices ofthe two protomers are colored according to (a). Regions involved in metal coordination and substrate binding are coloredgreen, with important residues shown as stick models. The centrally coordinated metal centers are shown as purplespheres with labels for metal sites 1 and 2. The catalytic water molecule is represented by a small red sphere andc-di-GMP is shown as orange stick model. The conformation corresponds to the activated state of BlrP1 in the presence ofmanganese and high pH, PDB ID: 3GG0 [10].

855Light Regulation of Phosphodiesterase Activity

Arg192, which is important for coordination of c-di-GMP via one of its phosphodiester bridges (Fig. 1c),shows the most pronounced stabilization (Fig. 2a). In

the crystal structure, this loop region provides severalresidues contacting c-di-GMP while only few interac-tions are observed with the rest of the protein, thus

Fig. 2. Deuterium incorporation plots of BlrP1 regions at four different experimental conditions addressed by HDX–MS.Labeling time dependence of relative deuterium incorporation of peptides indicated on top and the upper right corner ofeach panel are shown for Mgd, Mgl, Ccd and Ccl in blue, red, green and orange, respectively. The estimated abundancedistribution of individual deuterated species is presented in the lower sub-panels on a scale from undeuterated to allexchangeable amides deuterated. (a) EAL region involved in substrate binding via Arg192 to one c-di-GMPphosphodiester bridge. (b) EAL element extending from the compound helix end to β6E including conserved residuesinvolved in, for example, water activation (Lys323). (c) α1-β2B region of the BLUF domain responding to both lightactivation and metal and substrate binding. (d) Loop α3-α4B in the BLUF capping helix region positioned in proximity to theEAL dimerization region. Drel values are shown as the mean of three independent measurements with error barscorresponding to their standard deviation.


explaining the pronounced protection of deuteriumincorporation upon substrate binding (Fig. S1a).Additional elements stabilized by c-di-GMP bindinginvolve residues Asp215, Gln379 and Asn239, all ofwhich show a high degree of conservation amongactive EAL domains and are directly involved in sub-strate binding as judged from the crystal structure(Fig. 1c and Fig. S1a). Interestingly, one structuralelement shows an increase in deuterium incorporationdeviating from the expected stabilization due to thepresence of c-di-GMP (Fig. 2b). This element corre-

sponds to the compound helix and the subsequentβ-strand (β6E, subscript E or B indicates the secondarystructure elements of the EAL or BLUF domains,respectively) projecting into the active site (Fig. S1b)without interacting directly with c-di-GMP. However,β6E includes the conserved residues Asp325 andLys323, of which Lys323 was previously proposedto be involved in orientation and activation of thecatalytically active water molecule [10,13]. Consid-ering the crystal structure of BlrP1 in its metal-freeform (PDB ID: 3GFY [10]) and the disorder of the

Fig. 3. Overview of HDX experiments. Each box reflects one peptide and contains five different colors for the incubationtimes of 15, 60, 300, 1200 and 3600 s (bottom up), respectively. Individual colors correspond to the difference in relativedeuteration (ΔDrel) of two compared states according to the legend on the left. MS/MS confirmed peptides are marked withdiamonds and arrowheads at box termini indicate continuation of the peptide in the previous or following line. Secondarystructure elements are taken from DSSP (Define Secondary Structure of Proteins) analysis of BlrP1, PDB ID: 3GG0 [10].Numbering corresponds to the wild-type protein (UNP ID: A6T8V8) and negative values originate from the purification tag[10]. Secondary structure elements are indicated above the sequence and their labels are colored purple and green forBLUF and EAL domains, respectively. Zooming in on the electronic version allows viewing full details of all comparisons.(a) Ccd–Mgd. (b) Ccl–Ccd. (c) Mgl–Mgd. (d) Ccl–Mgl. Animations of the time course of ΔDrel for all four comparisons withcorresponding coloration of the BlrP1 crystal structure are shown in Movies S1–S4.


Fig. 4. Summary of structural elements of BlrP1 involved in the light regulation of EAL activity. (a) BlrP1 structurecolored according to changes in deuterium incorporation. A single time point (5 min) of the Ccl–Mgl comparison is shown toillustrate the structural arrangement of BlrP1 elements involved in inter-domain communication based on PDB ID: 3GG0[10]. Colors correspond to the differences in Drel according to the bar legend. Red colors reflect an increased deuteriumuptake upon substrate binding and calcium coordination, while blue colors indicate a stabilization of structural elements.Flavin mononucleotide (FMN) and c-di-GMP are shown as yellow and orange stick models, respectively, and metal ions aspurple spheres. (b) Proposed allosteric signaling pathway between the BLUF and EAL domains. Based on the changes indeuterium incorporation in different functional states, a coupling of structural elements that supports the highlightedpathway was observed (see Discussion). BLUF-specific interactions are shown in blue and range from the place of photonabsorption at the flavin cofactor (yellow sticks) to α3-α4B and the C-terminal part of the β4B that interact with thedimerization region of the EAL domains. The central role of this dimerization element is illustrated by the red arrows thatindicate the coupling with the BLUF domains and the signaling to the EAL active sites where c-di-GMP (orange stickmodel) is bound. Cartoon representations of protomers A and B are colored light and dark gray, respectively.


EAL dimerization region induced by the absence ofthe metal ions [10], it is suggestive that the observeddecrease in HDX protection of the compound helix is

due to the differences in the metal coordination(calcium instead of magnesium in Ccd and Mgd,respectively).


Interestingly, substrate and calcium binding in thedark affected regions not only in the EAL domain butalso in the BLUF-EAL linker and the α1-β2B element(Fig. 2c and Fig. S1c). The latter region includesAsn31 that interacts with N3 and C4=O of theisoalloxazine ring system and close by residues,such as Arg26 and Lys30, interacting with theribitylphosphate chain of the flavin cofactor. Sincechemical shift perturbations upon illumination of theisolated BlrP1 BLUF domain [37] and NMR andHDX–MS studies of other BLUF proteins [30,38]showed that the α1-β2B element is affected byblue-light activation, our data suggest a bidirectionalcommunication between the BLUF and EAL do-mains. While the effect of flavin excitation on EALactivity in BlrP1 is well established [10], so far, noinfluence of ligands bound to an effector domain onthe environment of the flavin cofactor has beendescribed for any BLUF protein. Such a couplingof elements involved in substrate and metal coordi-nation with the flavin environment indicates anallosteric signaling mechanism.

Illumination affects EAL elements in the absenceof calcium and substrate

In order to test the involvement of the α1-β2B regionalso in the light response of full-length BlrP1, we firstcompared Ccl to Ccd (Fig. 3b) again because of thestructural information available for substrate-bound,inhibited BlrP1. Indeed, the most pronounced light-induced changes affected peptides of the α1-β2Bregion and evaluation of overlapping peptidessuggests that the pronounced destabilization ismost likely due to a reduction of α1B stability uponblue-light illumination. Additional elements of theBLUF domain affected by illumination include β4Bthat interacts with the EAL domain and β5B thatcontains functionally relevant amino acids such asMet92. The latter, strictly conserved BLUF residueis especially interesting since two overlapping pep-tides in the β4-β5B region allow detailed information ofthis residue to be extracted from HDX experiments(Fig. S6). Analysis of the overlap showed that theMet92 amide proton forms a relatively stable hydro-gen bond in the dark that is significantly destabilized inthe presence of blue light. In terms of the dark-adaptedcrystal structures, this can be rationalized by thehydrogen bond between the hydroxyl group of Ser28and the Met92 amide group (cf. Fig. S1c; O–Ndistances of 3.2 and 2.7 Å for chains A and B ofPDB ID: 3GFX, respectively). Importantly, Ser28 isalso part of the abovementioned α1B region that ispronouncedly destabilized upon illumination suggest-ing a correlation between the two structural elements.The destabilization of β4B (residues 62–77; Fig. 3b)involves relatively slowly exchanging amides, wherechanges in deuteration are usually interpreted aschanges in conformational dynamics. Therefore,

illumination causes no distinct structural rearrange-ments in this region, which interacts with both theflavin cofactor and the EAL domain.The lack of light-induced changes in the EAL

domain complexed with calcium and c-di-GMPprompted the comparison of light- and dark-adaptedstates in substrate-free BlrP1 in the presence ofmagnesium (Mgl–Mgd). As shown in Fig. 3c, thesame elements of the BLUF domain describedabove are affected. However, additional regionsincluding the C-terminal BLUF helices, the linkerregion to the EAL domain and the PDE domain itselfshow characteristic changes in deuterium incorpo-ration. Interestingly, similarly to α1-β2B and β5B, asubtle destabilizing effect is observed for theC-terminal part of α3B including residues of theloop to α4B (residues 108–115; Fig. 3c). In thethree-dimensional structure, this loop region is indirect contact with the previously mentioned β5Bregion and the α4B helix contacts the EAL dimer-ization interface (cf. Fig. S1d). Similar to theseregions, the linker between BLUF and EAL (resi-dues 133–150; Fig. 2d) also shows an increase indeuterium incorporation that was not observed inthe Ca2+-inhibited EAL domain upon illumination.Most importantly, however, light-induced changesobserved in the Mgl–Mgd comparison are transmit-ted to elements of the EAL domain that are involvedin substrate binding and metal coordination (resi-dues 238–255; Fig. 3c). This region contains theconserved Asn239 residue that is required forpositioning of the metal 1 ion (Fig. 1c) [10]. Notably,this region is one of the structural elements that alsoshows altered deuterium incorporation upon sub-strate and calcium binding. In addition, a subtleeffect of illumination is also observed for the com-pound helix described above (Fig. 2b). These, albeitmoderate, changes in deuterium incorporation ofadditional elements identified in the Mgl–Mgdcomparison might explain light-induced changes incatalytic activity. However, this comparison does notprovide detailed information of a specific chain ofstructural changes involved in transmission of theblue-light signal from the BLUF domain to the activesite. Considering the Ccl–Ccd comparison, it is alsodifficult to rationalize the contribution of metalcoordination or substrate binding in preventing thechanges of the compound helix and the active siteobserved in Mgl–Mgd. Importantly, both compari-sons addressing the effect of BlrP1 illuminationshow no indications of an alteration of the oligomer-ization state of the characteristic EAL dimer, whichwould be expected to result in pronounced destabi-lization of the dimerization helix, the compound helixand elements involved in intermolecular BLUF-EALcontacts (Fig. 3b and c). The observation that lightdoes not alter the oligomerization state is furthersupported by the solution scattering studies dis-cussed below.


Evolutionary conserved regions are involved inBLUF-EAL signaling

The observed changes in deuterium incorporationof the compound helix region upon both illuminationand substrate/calcium binding suggest a centralinvolvement of this element in the bidirectionalcommunication of BLUF and EAL. Interestingly,this inter-subunit coupling is more pronounced inlight-adapted BlrP1 (Ccl–Mgl comparison; Fig. 3d).While the substrate- and calcium-induced effects onthe EAL domain are similar to the Ccd–Mgd compar-ison, the cross-talk with the BLUF domain is morepronounced in the presence of light. This is evident inincreased changes in relative deuteration levels(ΔDrel) of previously identified elements such as thecompound helix, the domain linker and the α1-β2Bregion but, interestingly, also in additionally observedregions such as the loop between the BLUF cappinghelices (α3-α4B) and the C-terminal part of the β4Belement (Fig. 3d and Fig. S1d). Both regions contactthe EAL domains near their dimerization and com-pound helices. Notably, the observed destabilizationof α1-β2B is additive as indicated by the enhanceddestabilization due to calcium and substrate binding inthe presence of blue light (Fig. 3d, residues 21–40).In summary, the results of our HDX analysis reveal

elements involved in light regulation and metalcoordination of the EAL domain and the combinationof all comparisons enables the mapping of apotential signaling pathway between the sensorand effector domains (Fig. 4b). They provide mo-lecular details of regions involved in inter-domaincommunication supporting the critical role of previ-ously proposed elements important for signal trans-duction. This includes the α1-β2B, β4B, β5B andα3-α4B (capping helices) regions and elementsclose to the compound helix involved in substratebinding and metal coordination in the EAL domain(Fig. 3d). In combination with the observation thatc-di-GMP binding does not induce significant struc-tural rearrangements of the EAL domain [16,24], theobserved light- and metal-induced changes of theBLUF-EAL linker region and of interface elementsinvolving the BLUF C-terminus and the EAL dimer-ization interface suggest inter-domain rearrange-ments accompanying the coupling of receptor andeffector domains.To test this hypothesis, we performedSAXSstudies

ofBlrP1 under conditions resemblingMgd andMgl andused NMA for structural interpretation (Supplementa-ry Data and Figs. S7 and S8). Importantly, the radialdensity distributions of BlrP1 in the dark and lightstates overlap at small scattering angles (Fig. S7d).This confirms that illumination does not result inchanges of the oligomerization state, which is aprerequisite for interpreting the HDX data in terms ofan allosteric signaling pathway. A closer comparisonbetween the experimental data and the theoretical

scattering curves calculated from the BlrP1 crystalstructure and its computed normal modes suggestedthat the observed differences correspond to inter-do-main rearrangements (cf. Supplementary Data). Inter-estingly, the structural difference between thedark-state BlrP1 dimer in solution and in crystallocorresponds to a clam-shell opening of the EALdomains (Fig. S7c). This structural movement resem-bles different EAL dimer arrangements observed invarious EAL structures [10,13,24,39,40] (Fig. 5), whichimplies a functional relevance of the opening–closingmovement of the EAL dimer. The light-induceddifferences in the scattering curves can be explainedby a twistingmotion that results in a subtle reorientationof the BLUF domains relative to the EAL domains,which also affects the opening and closing of the EALdimer (Fig. S7f). Considering the coupling of the twointer-domain rearrangements, these are the globalstructural changes that are responsible for lightregulation of EAL activity. Importantly, the contactsites between the BLUF and EAL domains thatcommunicate the quaternary rearrangements be-tween the two domains (Fig. S7f) correspond toelements identified by the HDX measurements. Ananalysis of the evolutionary conservation of BlrP1residues further highlights the functional relevanceof the BLUF-EAL interface involving the β4B andcapping helix (α3-α4B) regions of BLUF andelements close to the compound helix of EAL (Fig.S9). These regions show a comparable evolutionaryconservation to residues lining the c-di-GMPbindingpocket and the EAL dimerization interface, whichfurther supports the functional relevance of theidentified structural elements in conformationalcoupling between the EAL and BLUF domains.

The compound helix environment is involved ininter-domain communication

Based on the observation of a bidirectional com-munication betweenBLUF and EAL described above,we tested the influence of metal coordination of theEAL domain on the dark-state recovery kinetics of theBLUF photocycle. As summarized in Table 1, weaddressed catalytically active forms of full-lengthBlrP1 in the presence of magnesium or manganeseand inactive calcium-bound or metal-free states in thepresence or absence of the substrate c-di-GMP. Inaddition, we included the BlrP1 BLUF domain as acontrol to probe any potential metal effect on thedark-state recovery kinetics of the isolated photore-ceptor domain and to dissect the additional influenceof the EAL domain. While no significant metal- orprotein-construct-dependent differences in the char-acteristic BLUF dark-state spectrum or the ~10-nmred-shifted spectrum of the light-activated state ofBlrP1 were observed (Fig. S10), the dark-staterecovery rates significantly differed for the variousconditions (Table 1).


The influence of the EAL domain alone can beinferred from the ethylenediaminetetraacetic acid(EDTA)-treated full-length sample versus that of theisolated BLUF domain. Similar to recent reports fora non-covalently linked BLUF-EAL system fromRhodopseudomonas palustris [12], we observedan effect of the EAL domain on the BLUF photocyclesupporting a cross-talk between the two domains.EDTA chelation successfully removes metal ions asindicated by the consistent dark-state recoverykinetics with two EDTA-treated BlrP1 variants thatotherwise show different metal-induced effects(Table 1, Y308F and R316M; cf. Fig. 1c and details

below). Interestingly, the observed slower dark-staterecovery in full-length BlrP1 compared to the isolatedBLUF domain is even more pronounced in themetal-activated states with magnesium or manga-nese bound to the EAL domain. Importantly, theinfluence of calcium, which inhibits EAL catalyticactivity, is opposite to that of either magnesium ormanganese. In fact, the dark-state recovery isaccelerated and resembles that of the BLUF domainalone. This, however, does not indicate uncouplingbetween the two domains as reflected by the small butreproducible additional decrease in the lifetime of thered-shifted BLUF spectrum upon addition of c-di-GMPto either EDTA-treated or calcium-inhibited BlrP1(Table 1). Importantly, the more pronounced effect ofmetal coordination compared to substrate bindingobserved in dark-state recovery experiments alsoindicates that the observed differences in deuteriumincorporation for elements along the signaling path-way are dominated by metal coordination. Althoughthe effects on the recovery kinetics are relativelysmall, they are highly accurate and the precisionindicated by the standard deviation is also supportedby control measurements using different wild-typeBlrP1 batches in the presence of Mg2+ or EDTA.Importantly, the trends described here support the

conclusions drawn from HDX–MS and SAXS mea-surements concerning the bidirectional communica-tion between the light-sensing BLUF domain and theresidues involved in metal and substrate coordinationin the EAL domain. Therefore, the measurements ofdark-state recoveries were used to additionally char-acterize the effect of perturbations in the EALdimerization interface on PDE activity. Previously, itwas observed that a variant containing two pointmutations in this region (BlrP1 S309C S312C) showssignificantly reduced enzymatic activity [10]. Since thecatalytic activity of this variant is below the detectionlevel, no detailed insight into the role of these amino

Fig. 5. Structural superposition of various EAL domaindimers. Structural alignments with respect to protomers Aof the characteristic EAL dimers of YkuI (UNP ID: O35014)[24], BlrP1 [10], TdEAL (UNP ID: Q3SJE6) [13], CcEAL(UNP ID: Q9A310 and PDB ID: 3U2E; unpublished), RocR(UNP ID: Q9HX69) [40] and DosP (UNP ID: P76129) [39]are shown in cartoon representation. The orientations ofthe upper two panels correspond to those of Fig. 1a andcolors reflect the amplitude of the clam-shell opening of theEAL dimer. Dark blue corresponds to YkuI and the mostpronounced closed state and red belongs to DosP with thelargest opening. Transitions over blue, light blue, gray andlight red correspond to BlrP1, TdEAL, CcEAL and RocR,respectively. Representative c-di-GMP molecules of YkuIare shown as black stick models to indicate the substratebinding sites. The preferred orientation of EAL protomersalong one trajectory and the similarity to normal modemovements (Fig. S7c) further support the functionalrelevance of the EAL dimer assembly.

image of Fig.�5

Table 2. Enzymatic activity of c-di-GMP hydrolysis to5′-pGpG for BlrP1 and two variants measured understandard conditions as described in Ref. [10].

Protein kcat (s−1) (dark) kcat (s

−1) (light)

BlrP1 wild type [10] 0.13 ± 0.02 0.54 ± 0.02BlrP1 Y308F 0.014 ± 0.007 0.06 ± 0.03BlrP1 R316M 0.029 ± 0.007 0.10 ± 0.06

Table 1. Dark-state recovery of BlrP1 and its isolated BLUFdomain in the presence or absence of various divalentmetals and the substrate c-di-GMP at 10 °C.

Experiment Mean lifetime,τ (s)

BlrP1–Mg2+ 326 ± 4BlrP1–Mn2+ 351 ± 8BlrP1–Ca2+ 208 ± 5BlrP1–Ca2+–c-di-GMP 182 ± 7BlrP1–EDTA 258 ± 4BlrP1–EDTA–c-di-GMP 230 ± 5BlrP1 Y308F–Mg2+ 264 ± 4BlrP1 Y308F–Ca2+ 177 ± 3BlrP1 Y308F–EDTA 252 ± 6BlrP1 R316M–Mg2+ 281 ± 7BlrP1 R316M–Ca2+ 185 ± 5BlrP1 R316M–EDTA 259 ± 6BlrP1 BLUF–Mg2+ 212 ± 4BlrP1 BLUF–Ca2+ 210 ± 4BlrP1 BLUF–EDTA 226 ± 4

Mean lifetimes are stated as the mean of four repetitive light–darkcycle measurements ± standard deviation.


acids for the light-activation pathway is possible.Therefore, we made two additional single pointmutations in the loop regions preceding and followingthe compound helix, Y308F and R316M, respectively.Arg316 is involved in stabilizing the dimeric EALarrangement by inter-protomer contacts with theGly307 carbonyl group in the presence of c-di-GMPand manganese (PDB ID: 3GG0), whereas no suchinteraction isobserved in theotherBlrP1structures [10].The Tyr308 side chains of both monomers are locatedat the center of the compound helix, and they arepositionedsuch that they can interferewith theArg316–Gly307 interaction. In contrast to Arg316, Tyr308 onlyadopts a well-defined conformation in the low pH,Ca2+–c-di-GMP structure (PDB ID: 3GFX [10]). Thestructural element containing Tyr308 is interesting alsobecause it shows a high degree of asymmetry betweenthe two protomers compared to the overall symmetricEAL dimer arrangement.The enzymatic activity of Y308F and R316M

variants was tested as described previously [10]and found to be reduced by a factor of ~9 and ~5,respectively (Table 2). Interestingly, however, theblue-light-induced ca 4-fold stimulation of EALactivity is retained by both variants. While theEDTA-treated samples of both variants show thesame dark-state recovery kinetics as wild-typeBlrP1, the distinct stabilization of the light-activatedstate due to magnesium binding is missing. Theacceleration of the dark-state recovery induced byCa2+ coordination, however, is more pronouncedthan in the wild type, suggesting that either metalbinding to the active site or structural consequencesthereof are affected.The results from dark-state recoveries and cata-

lytic activity measurements support the conclusion

that EAL activity is regulated via the EAL-EAL di-merization interface. As indicated by the dark-staterecovery kinetics of two variants of this region,changes at the interface likely affect metal coordi-nation in the active site and are coupled to the BLUFdomain, where they influence the flavin cofactorenvironment, leading to either stabilization or desta-bilization of the light-activated state under conditionsresembling Mgl and Ccl, respectively. Especially thesubstitution of Tyr308 that is positioned at the contactsite of the two short helices forming the compoundhelix has a pronouncedeffect on thePDEactivity. Thisfurther confirms the central role of the dimeric EALassembly for regulation of c-di-GMP hydrolysis andthat the modulation of the quaternary arrangementis exploited by the light-sensing BLUF domain in anallosteric manner (Fig. 4b).

Discussion

Our studies of the blue-light-regulated PDE BlrP1provide new structural and functional insights intoregulation of EAL domains with implications not onlyfor allosteric control of PDE activity but also forcontrol of other c-di-GMP responsive processes thatare mediated by inactive EAL domains. In addition,the observed cross-talk of the EAL domains with theflavin-based BLUF photoreceptor domains presentsnew aspects of structural elements involved in blue-light sensing of BLUF domains.In the context of BLUF signaling, open questions

concern the mechanism of signal transduction tovarious effector domains (reviewed in Ref. [41]) andthe controversial interpretation of molecular detailsof hydrogen bonding in the flavin binding pocket withimplications for the photoactivation mechanism(summarized in Ref. [35]). While the spatial resolu-tion of HDX–MS limits its use for addressing differ-ences in the hydrogen bonding network of individualresidues, HDX–MS has the advantage of probinglight-induced changes in the context of full-lengthproteins providing both global and local structuralinformation. The observed changes in deuteriumuptake of BLUF elements are in good agreementwith previous NMR studies of the isolated BlrP1BLUF domain [37] and demonstrate the importanceof the α1-β2B region, the β4-β5B loop, β5B and theα3-α4B loop in the C-terminal capping helices alsofor the holo protein. In addition, our experiments


including the EAL domain in different functionallyrelevant states further suggest a light-inducedsignaling cascade that originates in the α1-β2Bregion and β5B. Interestingly, α1-β2B appears to bea central element of the blue-light response of BLUFdomains as also observed in similar studies withAppA (UNP ID: Q53119) [30]. Depending on themetal-coordination state of the EAL domain, thisstructural change can be further communicated tothe β4-β5B and α3-α4B loops and to β4B that directlyinteract with elements of the EAL dimerizationregion. These observations reveal new moleculardetails of the initially proposed signaling pathway[10] and support the central role of the C-terminalcapping helices in communicating the signal to thecompound helix environment of the EAL dimer. Thiscrucial role of the BLUF capping element wasrecently also shown for photoreceptor chimerasfeaturing capping helices from different BLUFsystems [42] and HDX–MS studies on the prototypicBLUF member AppA in complex with its non-cova-lently linked effector PpsR [30]. However, given themarkedly different structures of the capping helices(Fig. S11; Refs. [10,30,43] and [44]), these elementsappear to have evolved as system-specific featuresthat relay the light signal from the BLUF core to theircorresponding effector regions.With respect to signal transmission toEAL domains,

it is interesting to note that the EAL conformationdominates the communication with the BLUF domain.In the EAL inhibited state with substrate and calciumpresent, light only triggers changes in the close vicinityof the flavin cofactor (α1-β2B and β5B; cf. Ccl–Ccd).In contrast, substrate and calcium binding in thelight-activated state (cf. Ccl–Mgl) shows that substratebinding and metal coordination induce changes allalong the signaling pathway and even affect theα1-β2B region. Illumination, however, also affectsstructural elements in the substrate binding site andthe EAL dimerization region in the presence ofmagnesium and in the absence of c-di-GMP (cf.Mgl–Mgd). This bidirectional cross-talk of EAL andBLUF domains supports an allosteric communicationpathway that involves the EAL dimer interface ascentral communication platform. Light signals fromboth BLUF domains are integrated at the conservedEAL dimerization region and communicated to theEAL active sites. Such an inter-domain coupling in theBlrP1 dimer satisfies allosteric concepts such asglobal dyad symmetry, and typically, communicationin a co-operative dimeric assembly (Fig. 4b) results inthe amplification of the light response [45]. A similarsymmetry as observed for the BLUF-EAL couple inBlrP1 is also present in the crystal structure of YkuIthat features a PAS-EAL couple [24]. The positioningof the individual modules appears to be very similar inboth cases [10], and the BLUF and PAS domainsare placed close to the EAL dimerization regions.Importantly, in both cases, the overall symmetry is

governed by the evolutionary conserved EAL di-merization interface that is observed for all catalyt-ically active EALs and some inactive domainstructures [10,13,24,39,40]. A superposition of allstructures with respect to protomer A shows that themajor differences in positioning of the second EALdomain correspond to a specific opening–closingtransition of the EAL dimer (Fig. 5) that alsoresembles the different BlrP1 conformations ob-served under different experimental conditions bySAXS in solution and by X-ray crystallography.Interestingly, the light-induced structural changealso modulates this specific movement of the EALdomains. Therefore, the clam-shell-like opening andclosing of the EAL dimer appears to be a centralregulatory mechanism that seems governed by theenvironment and stability of the compound helixpositioned at the center of this transition. Recently,an unusual quaternary arrangement was observedfor RocR that resembles the in-solution conforma-tion [40] of this REC-regulated EAL protein andreveals an intertwined tetrameric assembly thatcombines two EAL dimers with one EAL active siteeach blocked by a REC domain. In contrast, theRocR EAL domains that appear competent forcatalysis seem to retain a similar regulation mech-anismwith sensory modules in close proximity to thecompound helices and elements involved in thisinteraction featuring changes in deuterium incorpo-ration upon activation of the REC domains [11]. Thiscentral role of the EAL compound helix is furthersupported by our HDX experiments and mutationalstudies on BlrP1 and by similar experimental ap-proaches for RocRwhere this region was identified asa key regulatory element [11]. In terms of controllingPDE activity, this makes sense due to the directconnection of the compound helix and both β5E andβ6E that contain conserved residues involved in metalcoordination (Asp302 and Asp303) and water activa-tion (Lys323). The reduced activity observed forprotein variants with amino acid substitutions in thecompound helix is likely due to altered metalcoordination, which is indicated by the significantchanges in their magnesium effects and calciumeffects on the allosteric communication pathway asreflected in changes in their BLUF dark-state recov-eries. Importantly, the compound helix of catalyticallyactive EAL domains is highly conserved [10,11],further supporting its central role for EAL functioningeven though minor modifications do not completelyabolishPDEactivity [13]. However, other substitutionsin the compound helix can reduce activity below thedetection limit as shown for BlrP1 and RocR [9,10].Interestingly, EAL subfamilies without c-di-GMP

degrading activity frequently have substitutions notonly of key residues involved in catalysis but they alsohave degenerate compound helices [11]. While thisobservation supports the catalytic relevance of thecompound helix due to co-evolution with key catalytic


residues, it suggests that this element might haveeither lost its function or evolved toward a new role inthese EAL families. The recently solved crystalstructure of a LapD (UNP ID: Q3KK31) fragmentincluding an EAL, a GGDEF and a signaling helixshows that, in this case, the EAL “dimerization” regionhas actually retained part of its function [46]. Theimportant signaling helix that is involved in inside-outsignaling of LapD binds directly to the dimerizationhelix and the truncated compound helix of a singleEALdomain. Additional structural and biochemical datashow that c-di-GMP binding to the LapD EAL domainleads to dimerization of the EAL domain and preventsthe associationwith the signaling helix due to structuralchanges induced in compound helix residues [46].Importantly, the dimeric LapD EAL assembly in thepresence of c-di-GMP resembles the arrangement ofcatalytically active EAL domains shown above. Thisexample nicely illustrates how allosteric changesinduced by binding of the second messenger toinactive EAL domains can be used for the regulationof cellular processes employing mechanisms thatwere originally important for regulating PDE activity.Another degenerate EAL system that is evolutionaryrelated to BlrP1 is YcgF fromEscherichia coli (UNP ID:P75990). In this case, a BLUF sensor is connected toan EAL domain that has lost both its c-di-GMP bindingand its c-di-GMP degrading activity but gained thefunction as an anti-repressor [15]. This new role ismediated via direct protein–protein interaction with atranscriptional repressor and enables a blue-lightregulation of processes that are indirectly involved inc-di-GMP associated processes such as biofilmformation [47]. Considering the evolutionary relation-ship with BlrP1, it is suggestive that the light signal isstill transmitted to the EAL dimerization regionmodulating the affinity of the interaction partner thatmight bind in this region similar to the situation in LapD.In line with the important role of c-di-GMP in

bacterial signaling, different protein–protein interac-tions involved in various biological processes haveevolved. For many degenerate EAL systems, molec-ular details of their interaction interface are not known.However, recently, a study of the Xanthomonascampestris FimX (UNP ID: Q8P8F1) protein revealeda unique interaction of its degenerate EAL domain witha noncanonical type II PilZ domain mediated by theEAL-coordinated c-di-GMP [48]. Thus, c-di-GMPbinding can directly control protein–protein interaction,but its interaction with the FimX homologue fromPseudomonas aeruginosa (UNP ID: Q9HUK6) wasshown to also induce long-range allostericmodulationsof distant REC domain [17]. In this case, both the RECdomain and the EAL domain might be involved inhomo- and hetero-oligomerization, which ultimatelyaffects localization of FimX to a single pole for correctpilus assembly [16,17]. Importantly, HDX analysis ofc-di-GMP binding showed also an effect on thecompound helix element in this case [17], and

considering the central role of this region in thepreviously discussed examples, it might also beinvolved in propagating the long-range structuralchanges in P. aeruginosa FimX.In conclusion,wehave shown that regulation of EAL

activity in BlrP1 is centrally coordinated by the EALdimerization interface. The compound helix plays acrucial role for the correct assembly of the EAL dimerwith consequences formetal coordination in the activesite and hence PDE activity. Positioning of sensorymodules close to this dimerization region enables theintegration of different signals into c-di-GMP metab-olism, which, due to the importance of this bacterialsecond messenger, is reflected in various environ-mental stimuli-controlled EAL domains. Commonfeatures of the corresponding signaling pathwaysappear to be conserved even in non-catalytic EALdomains. Thus, we provide insight into the functioningof EAL domains that appears to be directly related totheir central position in regulation of various cellularprocesses mediated by the bacterial secondmessen-ger c-di-GMP.

Materials and Methods

Protein expression and purification

Two single point mutations in BlrP1 were introduced bysite-directed mutagenesis using PCR amplification of thepET_MBP_BlrP1 vector described in Ref. [10] according tothe QuikChange protocol (Stratagene) with the followingprimer pairs. All sequences are shown in 5′–3′ orientationand the mutated codon is underlined: BlrP1 Y308F,forward GACTTTGGCGCAGGTTTCTCCGGCCTGTCGTTA and reverse TAACGACAGGCCGGAGAAACCTGCGCCAAAGTC; BlrP1 R316M, forward CTGTCGTTACTGACCATGTTTCAGCCTGATAAAATC and reverseGATTTTATCAGGCTGAAACATGGTCAGTAACGACAG.Verification of successful mutagenesis was performed byDNA sequencing.Expression and purification of wild-type BlrP1 and of the

two variants followed the published protocol for full-lengthBlrP1 [10], while the BlrP1 BLUF domain was isolatedaccording to Ref. [36]. All proteins were prepared undersafe-light conditions and finally concentrated in storagebuffer [25 mM Tris–Cl, 40 mM NaCl, 5 mM MgCl2, 2 mMEDTA, 2 mM dithioerythritol and 5% (w/v) glycerol]. Smallaliquots were flash frozen in liquid nitrogen and subse-quently stored at −80 °C.

EAL activity assays

PDE activity was assayed as described previously[10]. This includes the bovine serum albumin pretreat-ment of reaction vessels, the conditions for illumination,the quenching step and the HPLC analysis. Standardconditions of this assay refer to 1.1 μM protein, 100 μMc-di-GMP, 50 mM Tris–HCl (pH 7.5), 50 mM NaCl and10 mM MgCl2 at 20 °C.


UV–Vis spectroscopy

Photocycle properties of BlrP1 and the isolated BlrP1BLUF domain were analyzed with an experimental setupdesigned to minimize spectral artifacts originating from themeasuring light. To this end, we used a balanced deuteriumtungsten lamp (DH-2000-BAL; Mikropack), which wasintensity reduced with a neutral density OD10 filter, asmeasuring light source and a royal blue light-emitting diode(LED) (λmax = 455 nm; Doric Lenses) as excitation lightsource. The sample was kept at 10 °C throughout the timecourse of the experiment using a Peltier controlled cuvetteholder connected to a TC 125 temperature controller(QuantumNorthwest). Blue-light illumination was performedfor 10 s with an intensity of 0.6 mW cm−2 at 450 nm at thesample position. Data were continuously acquired in“kinetic”mode for 15 min employing a spectrograph coupledto an electron multiplying charge-coupled device detector(Andor Technology). Recorded spectra were averaged from6accumulations of 0.15 s integration time each. The changein absorbance at 502 nmwas fittedwith a single exponentialdecay function. For addressing different metal and substratestates, we diluted the proteins to concentrations between 50and 100 μM in the appropriate buffer systems that were allset to pH 8 at 10 °C accounting for their temperaturedependence. Dilutions for the photocycle experiments inthe Mg2+-bound state were performed using buffer A[25 mM Tris–Cl, 40 mM NaCl, 5 mM MgCl2, 2 mM EDTAand 5% (w/v) glycerol]. The effect of Mn2+ was tested inbuffer B [25 mMTris–Cl, 40 mMNaCl, 10 mMMnCl2, 2 mMEDTA and 5% (w/v) glycerol], and dilutions for Ca2+ wereprepared with buffer C [25 mM Tris–Cl, 40 mM NaCl,30 mM CaCl2, 2 mM EDTA and 5% (w/v) glycerol]. Forremoving the metal ions from the active site, we diluted inMg2+-free buffer A and incubated the sample in thepresence of 20 mM EDTA for 15 min prior to dataacquisition. The effect of c-di-GMP binding was tested byincluding the substrate at a final concentration of 47 μM ineither the Ca2+-treated sample or the EDTA-treated sampleof 50 μM wild-type BlrP1.

Hydrogen–deuterium exchange–mass spectrometry

All samples for HDX were prepared under safe-lightconditions except where noted otherwise. For addressinglight- and/or substrate-induced structural changes, weperformed four labeling experiments: BlrP1 in the pres-ence of Mg2+ (Mg) or inhibiting concentrations of Ca2+ andc-di-GMP present (Cc) under dark (d) or blue-light (l)conditions. The corresponding experimental setups arereferred to as Mgd, Mgl, Ccd and Ccl, respectively, in themain text. To this end, we pre-incubated BlrP1 (108 μM) inbuffer A at 10 °C for 1 min in the dark or with parallelillumination from a royal blue (455 nm) collimated LED lamp(Thorlabs) providing a light intensity of 1 mW cm−2 at thesample position (Mgd or Mgl, respectively). Ca

2+-inhibitedsamples were prepared by pre-incubation of BlrP1 at 4 °C inbuffer C for 5 min followed by addition of c-di-GMP andanother 5 min incubation step. Final concentrations were108 μM BlrP1 and 611 μM c-di-GMP and 30 mM Ca2+.Equilibration of the samples for either dark-state orlight-state measurements (Ccd or Ccl, respectively) wasperformed as described above for Mgd or Mgl, respec-

tively. The corresponding light conditions were main-tained throughout the labeling procedure. For thispurpose, 2-μL aliquots of the equilibrated samples werediluted with 38 μL buffer AD or buffer CD [corresponding tobuffers A and C, however, prepared with D2O andglycer(d3-ol) including the temperature and D2O correc-tions for pD 8.0 at 10 °C] and 5-μL samples wereremoved after 15 s, 1 min, 5 min, 20 min and 60 min.Deuterium incorporation was terminated by quenchingthe samples in 55 μL of 200 mM ice-cold NH4 formic acid(pH 2.6) of which 50 μL was then injected into a cooledHPLC setup.Deuterated samples were digested on a pepsin column

(Applied Biosystems) kept at 10 °C. All subsequentchromatographic steps were carried out in a water bathat 0.5 ± 0.1 °C. Peptic fragments were buffer-exchangedon a 2-cm C18 guard column (Discovery Bio C18 packing)and separated in the presence of 0.6% formic acid with a15-min acetonitrile gradient (15–50%) on a reversedphase column (Discovery Bio Wide Pore C1810 cm × 1 mm − 3 μm; Supelco). Eluting peptides wereinfused into a maXis electrospray ionization–ultra highresolution–time-of-flight mass spectrometer (Bruker) formeasuring the extent of deuterium incorporation. Weanalyzed deuterium incorporation with an improvedversion of the automated software package Hexicon [49].This in-house version (manuscript in preparation) is basedon the previously published NITPICK algorithm for featuredetection [50] and Hexicon for the analysis of deuteriumincorporation. Deuterium incorporation was quantified intriplicate measurements by the mean shift of a peptide'smass centroid. Deuterium incorporation plots provided inFig. 2 and Figs. S2–S6 show the mean relative deuteriumincorporation at each exchange time point with error barsreflecting the standard deviation of triplicate data points.The absolute difference of a peptide's relative deuteriumincorporation (ΔDrel) between two states (i.e., light adaptedversus dark adapted) was used to evaluate the changes instructural dynamics of the corresponding protein region.Assignment of colors according to the legends in Figs. 3and 4 accounts for the calculated standard deviation, andtherefore, only statistically significant changes in deuteri-um incorporation are highlighted. As an additional qualitycontrol of data processing, estimated abundance distribu-tions are provided in the sub-panels of Fig. 2 and Figs. S2–S6. Briefly, this deuterium incorporation distribution corre-sponds to a deconvolution of the average extractedisotope pattern taking into account the natural isotopedistribution. Therefore, it provides an estimate of theabundance of species with a given number of deuteronsincorporated that are plotted on a scale from undeuterated(0) to all exchangeable amides deuterated [number ofamino acids except prolines minus 1 (due to rapidback-exchange of the N-terminal amine group)].

SAXS and NMA

SAXS data were collected at the X12SA cSAXSbeamline at the Swiss Light Source (Villigen, Switzerland).A series of protein concentrations between 2 and20 mg mL−1 were measured in buffer containing 25 mMHepes, 100 mM NaCl, 5 mM MgCl2, 2.5 mM EDTA, 2 mMdithioerythritol and 10% glycerol at pH 7.5. The sampleswere mounted in Ø 1-mm-quartz capillaries under dimmed


red light and kept at 11 °C throughout the experiment. Forlight activation of the BLUF domain, two LEDs (λmax =455 nm; Doric Lenses) were mounted on two sides of thecapillary and each focused with a cylindrical lens toprovide a light power of 10 mW cm−2 each. Dataacquisition using 12.4 keV photons and an X-ray beamdiameter of 300 μm was performed in 330 μm steps alongthe capillary with 4 × 0.25 s exposures at each position.Scattered X-rays were recorded with a Pilatus 2M detectorpositioned at a distance of 2.2 m from the sample. Datawere collected from the buffer alone and from the protein inthe dark and under constant illumination with both LEDs.The order of dark- and light-state data acquisition wasvaried to check for the effect of radiation damage and thereversibility of structural changes upon light activation. Alldata were azimuthally integrated and averaged. Thescattering vector q is defined as q = 4πsin(θ)λ−1.For data analysis, the buffer signal was subtracted from

that of the buffered protein solution. Guinier plots of thelowest protein concentration data are shown in Fig. S8a.To reduce possible aggregation artifacts in the lowq-region of high concentration samples, data from thelowest concentration sample were merged with data fromthe highest concentration at q = 0.09 Å−1 for both thedark state and the light state. Kratky plots correspondingto the globular two domain architecture of BlrP1 areshown in Fig. S8b. CRYSOL 2.6 [51] with defaultparameters was used to fit the crystal structure and thestructures obtained by NMA with the NOMAD-Ref Webserver [52] to the scattering data. The initial input modelfor the analysis of normal modes was the publishedcrystal structure of BlrP1 PDB ID: 3GG0 [10]. Out of the30 generated substructures that correspond to one fullcycle of the normal mode transition, structures #8 and#23 correspond to the maximal amplitude of structuralchanges. As structure #23 of normal mode 8 bestexplained the dark-state data, it was taken as a startingstructure for an additional NOMAD-Ref analysis tocompare the results with light-induced structural rear-rangements observed by SAXS.Supplementary data to this article can be found online at

http://dx.doi.org/10.1016/j.jmb.2013.11.018.

Acknowledgments

Weare grateful toRobert Lindner for input regardingtheHDXanalysis software andwe thankChrisRoomefor excellent IT support. We are grateful to T.R.M.Barends for stimulating discussions and comments onthe manuscript. We acknowledge financial support bytheMaxPlanck Society, EMBO:ALTF1309-2009 andtheAustrianScienceFund (FWF): J 3242-B09 toA.W.and the German Research Foundation (DFG):FOR526 to I.S.

Received 17 September 2013;Received in revised form 18 November 2013;

Accepted 20 November 2013Available online 27 November 2013

Keywords:BLUF-photoreceptor;

c-di-GMP;EAL;

allostery;HDX

This is an open-access article distributed under the termsof the Creative Commons Attribution License, which

permits unrestricted use, distribution, and reproduction inany medium, provided the original author and source are

credited.

Abbreviations used:BLUF, Sensor of blue light using FAD; BlrP1, blue-light-regulated phosphodiesterase 1; c-di-GMP, cyclic dimeric

GMP; EDTA, ethylenediaminetetraacetic acid; HDX,hydrogen–deuterium exchange; LED, light-emitting diode;

MS, mass spectrometry; NMA, normal mode analysis;PDE, phosphodiesterase; SAXS, small-angle

X-ray scattering.

References

[1] Monod J, Changeux JP, Jacob F. Allosteric proteins andcellular control systems. J Mol Biol 1963;6:306–29.

[2] Koshland DE, Nemethy G, Filmer D. Comparison ofexperimental binding data and theoretical models in proteinscontaining subunits. Biochemistry 1966;5:365–85.

[3] Changeux JP. Allostery and the Monod-Wyman-Changeuxmodel after 50 years. Annu Rev Biophys 2012;41:103–33.

[4] Schmidt AJ, Ryjenkov DA, Gomelsky M. The ubiquitousprotein domain EAL is a cyclic diguanylate-specific phos-phodiesterase: enzymatically active and inactive EAL do-mains. J Bacteriol 2005;187:4774–81.

[5] Ross P, Weinhouse H, Aloni Y, Michaeli D, Weinberger-Ohana P, Mayer R, et al. Regulation of cellulose synthesis inAcetobacter xylinum by cyclic diguanylic acid. Nature1987;325:279–81.

[6] Romling U, Amikam D. Cyclic di-GMP as a secondmessenger. Curr Opin Microbiol 2006;9:218–28.

[7] Krasteva PV, Giglio KM, Sondermann H. Sensing themessenger: the diverse ways that bacteria signal throughc-di-GMP. Protein Sci 2012;21:929–48.

[8] Römling U, Galperin MY, Gomelsky M. Cyclic di-GMP: thefirst 25 years of a universal bacterial second messenger.Microbiol Mol Biol Rev 2013;77:1–52.

[9] Rao F, Yang Y, Qi Y, Liang ZX. Catalytic mechanism of cyclicdi-GMP-specific phosphodiesterase: a study of the EALdomain-containing RocR from Pseudomonas aeruginosa. JBacteriol 2008;190:3622–31.

[10] Barends TRM, Hartmann E, Griese JJ, Beitlich T, KirienkoNV, Ryjenkov DA, et al. Structure and mechanism of abacterial light-regulated cyclic nucleotide phosphodiester-ase. Nature 2009;459:1015-1018.

[11] Rao F, Qi Y, Chong HS, Kotaka M, Li B, Li J, et al. Thefunctional role of a conserved loop in EAL domain-basedcyclic di-GMP-specific phosphodiesterase. J Bacteriol2009;191:4722–31.

[12] Kanazawa T, Ren S, Maekawa M, Hasegawa K, Arisaka F,Hyodo M, et al . Biochemical and physiological

http://dx.doi.org/10.1016/j.jmb.2013.11.018

http://refhub.elsevier.com/S0022-2836(13)00729-8/rf0005






































characterization of a BLUF protein–EAL protein complexinvolved in blue light-dependent degradation of cyclicdiguanylate in the purple bacterium Rhodopseudomonaspalustris. Biochemistry 2010;49:10647–55.

[13] Tchigvintsev A, Xu X, Singer A, Chang C, Brown G,Proudfoot M, et al. Structural insight into the mechanism ofc-di-GMP hydrolysis by EAL domain phosphodiesterases.J Mol Biol 2010;402:524–38.

[14] Suzuki K, Babitzke P, Kushner SR, Romeo T. Identification ofa novel regulatory protein (CsrD) that targets the globalregulatory RNAsCsrB and CsrC for degradation by RNase E.Genes Dev 2006;20:2605–17.

[15] Tschowri N, Busse S, Hengge R. The BLUF-EAL proteinYcgF acts as a direct anti-repressor in a blue-light responseof Escherichia coli. Genes Dev 2009;23:522–34.

[16] Navarro MV, De N, Bae N, Wang Q, Sondermann H.Structural analysis of the GGDEF-EAL domain-containingc-di-GMP receptor FimX. Structure 2009;17:1104–16.

[17] Qi Y, Chuah ML, Dong X, Xie K, Luo Z, Tang K, et al. Bindingof cyclic diguanylate in the non-catalytic EAL domain of FimXinduces a long-range conformational change. J Biol Chem2011;286:2910–7.

[18] Schirmer T, Jenal U. Structural and mechanistic determi-nants of c-di-GMP signalling. Nat Rev Microbiol2009;7:724–35.

[19] Hengge R. Principles of c-di-GMP signalling in bacteria. NatRev Microbiol 2009;7:263–73.

[20] Christen M, Christen B, Folcher M, Schauerte A, Jenal U.Identification and characterization of a cyclic di-GMP-specificphosphodiesterase and its allosteric control by GTP. J BiolChem 2005;280:30829–37.

[21] Kazmierczak BI, Lebron MB, Murray TS. Analysis of FimX,a phosphodiesterase that governs twitching motility inPseudomonas aeruginosa. Mol Microbiol 2006;60:1026–43.

[22] Chang AL, Tuckerman JR, Gonzalez G, Mayer R,Weinhouse H, Volman G, et al. Phosphodiesterase A1,a regulator of cellulose synthesis in Acetobacter xylinum,is a heme-based sensor. Biochemistry 2001;40:3420–6.

[23] Tanaka A, Takahashi H, Shimizu T. Critical role of the hemeaxial ligand, Met95, in locking catalysis of the phosphodies-terase from Escherichia coli (Ec DOS) toward cyclic diGMP.J Biol Chem 2007;282:21301–7.

[24] Minasov G, Padavattan S, Shuvalova L, Brunzelle JS, MillerDJ, Basle A, et al. Crystal structures of YkuI and its complexwith second messenger cyclic di-GMP suggest catalyticmechanism of phosphodiester bond cleavage by EALdomains. J Biol Chem 2009;284:13174–84.

[25] Kumar M, Chatterji D. Cyclic di-GMP: a second messengerrequired for long-term survival, but not for biofilm formation, inMycobacterium smegmatis. Microbiology 2008;154:2942–55.

[26] Tarutina M, Ryjenkov DA, Gomelsky M. An unorthodoxbacteriophytochrome from Rhodobacter sphaeroides in-volved in turnover of the second messenger c-di-GMP. JBiol Chem 2006;281:34751–8.

[27] Gomelsky M, Klug G. BLUF: a novel FAD-binding domaininvolved in sensory transduction in microorganisms. TrendsBiochem Sci 2002;27:497–500.

[28] Gomelsky M, Kaplan S. appA, a novel gene encoding atrans-acting factor involved in the regulation of photosynthe-sis gene expression in Rhodobacter sphaeroides 2.4.1. JBacteriol 1995;177:4609–18.

[29] Masuda S, Bauer CE. AppA is a blue light photoreceptor thatantirepresses photosynthesis geneexpression inRhodobactersphaeroides. Cell 2002;110:613–23.

[30] Winkler A, Heintz U, Lindner R, Reinstein J, Shoeman RL,Schlichting I. A ternary AppA–PpsR–DNA complex mediateslight regulation of photosynthesis-related gene expression.Nat Struct Mol Biol 2013;20:859–67.

[31] Yuan H, Bauer CE. PixE promotes dark oligomerization ofthe BLUF photoreceptor PixD. Proc Natl Acad Sci USA2008;105:11715–9.

[32] Iseki M, Matsunaga S, Murakami A, Ohno K, Shiga K,Yoshida K, et al. A blue-light-activated adenylyl cyclasemediates photoavoidance in Euglena gracilis. Nature2002;415:1047–51.

[33] Stierl M, Stumpf P, Udwari D, Gueta R, Hagedorn R, Losi A,et al. Light modulation of cellular cAMP by a small bacterialphotoactivated adenylyl cyclase, bPAC, of the soil bacteriumBeggiatoa. J Biol Chem 2011;286:1181–8.

[34] Masuda S. Light detection and signal transduction in theBLUF photoreceptors. Plant Cell Physiol 2013;54:171–9.

[35] Udvarhelyi A, Domratcheva T. Glutamine rotamers in BLUFphotoreceptors: a mechanistic reappraisal. J Phys Chem B2013;117:2888–97.

[36] Tyagi A, Penzkofer A, Griese J, Schlichting I, Kirienko NV,Gomelsky M. Photodynamics of blue-light-regulated phos-phodiesterase BlrP1 protein from Klebsiella pneumoniaeand its photoreceptor BLUF domain. Chem Phys 2008;354:130–41.

[37] Wu Q, Gardner KH. Structure and insight into blue light-induced changes in the BlrP1 BLUF domain. Biochemistry2009;48:2620–9.

[38] Grinstead JS, Hsu STD, Laan W, Bonvin AMJJ, HellingwerfKJ, Boelens R, et al. The solution structure of the AppA BLUFdomain: insight into the mechanism of light-induced signal-ing. ChemBioChem 2006;7:187–93.

[39] Tarnawski M, Barends TR, Hartmann E, Schlichting I.Structures of the catalytic EAL domain of the Escherichiacoli direct oxygen sensor. Acta Crystallogr Sect D BiolCrystallogr 2013;69:1045–53.

[40] Chen MW, Kotaka M, Vonrhein C, Bricogne G, Rao F,Chuah ML, et al. Structural insights into the regulatorymechanismof the response regulatorRocR fromPseudomonasaeruginosa in cyclic Di-GMP signaling. J Bacteriol2012;194:4837–46.

[41] Zoltowski BD, Gardner KH. Tripping the light fantastic: blue-light photoreceptors as examples of environmentally modu-lated protein–protein interact ions. Biochemistry2011;50:4–16.

[42] Ren S, Sawada M, Hasegawa K, Hayakawa Y, Ohta H,Masuda S. A PixD–PapB chimeric protein reveals the functionof the BLUF domain C-terminal alpha-helices for light signaltransduction. Plant Cell Physiol 2012;53:1638–47.

[43] Jung A, Domratcheva T, Tarutina M, Wu Q, Ko WH,Shoeman RL, et al. Structure of a bacterial BLUFphotoreceptor: insights into blue light-mediated signaltransduction. Proc Natl Acad Sci USA 2005;102:12350–5.

[44] Yuan H, Anderson S, Masuda S, Dragnea V, Moffat K, BauerC. Crystal structures of the Synechocystis photoreceptorSlr1694 reveal distinct structural states related to signaling.Biochemistry 2006;45:12687–94.

[45] Crick FH, Wyman J. A footnote on allostery. J Mol Biol2013;425:1500–8.

[46] Navarro MV, Newell PD, Krasteva PV, Chatterjee D, MaddenDR, O'Toole GA, et al. Structural basis for c-di-GMP-mediated inside-out signaling controlling periplasmic prote-olysis. PLoS Biol 2011;9:e1000588.


































































































































[47] Tschowri N, Lindenberg S, Hengge R. Molecular function andpotential evolution of the biofilm-modulating blue light-signallingpathway of Escherichia coli. Mol Microbiol 2012;85:893–906.

[48] Chin KH, Kuo WT, Yu YJ, Liao YT, Yang MT, Chou SH.Structural polymorphism of c-di-GMP bound to an EAL domainand in complex with a type II PilZ-domain protein. ActaCrystallogr Sect D Biol Crystallogr 2012;68:1380–92.

[49] Lou X, Kirchner M, Renard BY, Kothe U, Boppel S, Graf C,et al. Deuteration distribution estimation with improvedsequence coverage for HX/MS experiments. Bioinformatics2010;26:1535–41.

[50] Renard BY, Kirchner M, Steen H, Steen JA, Hamprecht FA.NITPICK: peak identification for mass spectrometry data.BMC Bioinformatics 2008;9:355.

[51] Svergun D, Barberato C, Koch MHJ. CRYSOL: a programto evaluate X-ray solution scattering of biological macro-molecules from atomic coordinates. J Appl Crystallogr1995;28:768–73.

[52] Lindahl E, Azuara C, Koehl P, Delarue M. NOMAD-Ref:visualization, deformation and refinement of macromolecularstructures based on all-atom normal mode analysis. NucleicAcids Res 2006;34:W52–6.























characterization of elements involved in allosteric light ... · munication pathway. here we show...

Documents